Approaches to Discourse

Deborah Schiffrin

a Hi

B
BLACKWELL
Oxford UK & Cambridge USA
 To David

Copyright © Deborah Schiffrin 1994

The right of Deborah Schiffrin to be identified as author

of this work has been asserted in accordance with the
Copyright, Designs and Patents Act 1988.

First published 1994

Blackwell Publishers
238 Main Street
Cambridge, Massachusetts 02142
USA

108 Cowley Road

Oxford 0X4 1JF
UK

All rights reserved. Except for the quotation of short passages for the purposes of criticism and
review, no part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or
otherwise, without the prior permission of the publisher.

Except in the United States of America, this book is sold subject to the condition that it shall
not, by way of trade or otherwise, be lent, resold, hired out, or otherwise circulated without
the publisher’s prior consent in any form of binding or cover other than that in which it is
published and without a similar condition including this condition being imposed on the
subsequent purchaser.

Library of Congress Cataloging-in-Publication Data

Schiffrin, Deborah.
Approaches to discourse / Deborah Schiffrin.
p. cm.—(Blackwell textbooks in linguistics)
Includes bibliographical references and index.
ISBN 0-631-16622-X.—ISBN 0-631-16623-8 (pbk.)
1. Discourse analysis. I. Title. II. Series.
P302.S334 1994
410'.41—dc20 93-13359
CIP

British Library Cataloguing in Publication Data

A CIP catalogue record for this book is available from the British Library.

Typeset in 10 on 12 pt Sabon by Pure Tech Corporation, Pondicherry, India

Printed in Great Britain by T. J. Press Ltd, Padstow, Cornwall

This book is printed on acid-free paper

Contents

A it

Preface and Acknowledgments viii

PART I THE SCOPE OF DISCOURSE ANALYSIS 1

1 Overview 5
1 Introduction 5
2 “Core” examples from different approaches to discourse 6
3 Plan of the book 12

2 Definitions of Discourse 20
1 Introduction 20
2 Formal and functionalist paradigms 20
3 Discourse: language above the sentence 23
4 Discourse: language use 31
5 Discourse: utterances 39
6 Summary: definitions, issues, and discourse analysis 41

PART II APPROACHES TO DISCOURSE ANALYSIS 45

3 Speech Act Theory 49

1 Introduction 49
2 Defining speech act theory 49
3 Sample analysis: questions, requests, offers 61
4 Speech act theory as an approach to discourse 89
Exercises 91

4 Interactional Sociolinguistics 97
1 Introduction 97
2 Defining interactional sociolinguistics 98
3 Sample analysis: “speaking for another” 106
VI Contents
4 Interactional sociolinguistics as an approach to discourse 133
Exercises 134

5 The Ethnography of Communication 137

1 Introduction 137
2 Defining the ethnography of communication 137
3 Sample analysis: questions as speech acts in speech events 144
4 An ethnographic approach to discourse 185
Exercises 186

6 Pragmatics 190
1 Introduction 190
2 Defining pragmatics 191
3 Referring terms: pragmatic processes in discourse 197
4 Sample analysis: referring sequences in narrative 203
5 Gricean pragmatics as an approach to discourse 226
Exercises 228

7 Conversation Analysis 232

1 Introduction 232
2 Defining conversation analysis 233
3 Sample analysis: “there + BE + ITEM” 239
4 Conversation analysis as an approach to discourse 273
Exercises 279
8 Variation Analysis 282
1 Introduction 282
2 Defining variation analysis 283
3 Sample analysis: the “list” as text 291
4 The variationist approach to discourse 330
Exercises 333

PART III CONCLUSION 335

9 Structure and Function 339

1 Introduction 339
2 Structure and function in discourse analysis 339
3 Comparisons among approaches: from utterance to discourse 353
4 Conclusion 359

10 Text and Context 362

1 Introduction 362
2 What is context? 365
3 What is text? 373
4 Conclusion 333
11 Discourse and Communication 386
1 Introduction 386
2 Aspects of communication 337
 Contents vii

3 Discourse analysis and models of communication 391

4 Summary 405

12 Conclusion: Language as Social Interaction 406

1 Introduction 406
2 Starting points: differences among approaches 406
3 Principles of discourse analysis: language as social interaction 414
4 At the borders of different disciplines 418

Appendix 1 Collecting Data 420

Appendix 2 Transcription Conventions 422

Appendix 3 Sample Data * * 434

Bibliography 439

Index 459
Preface and Acknowledgments

Although discourse analysis is an increasingly popular and important area of

study - both on its own and for what it can tell us about language, society,
culture, and thought - it still remains a vast and somewhat vague subfield of
linguistics. My goal in this book is to clarify the theories and methods of
discourse analysis in such a way that it can continue to deal with a wide range
of problems and phenomena of interest to linguists, sociologists, anthropolog¬
ists, and psychologists, but can do so in a more systematic and coherent way.
The core of the book is a detailed description, application, and comparison
of six different approaches to discourse analysis: speech act theory, inter¬
actional sociolinguistics, ethnography of communication, pragmatics, conversa¬
tion analysis, and variation analysis. Although these approaches originated in
different disciplines (and are relevant to broader topics within each discipline),
they all attempt to answer some of the same questions: How do we organize
language into units that are larger than sentences? How do we use language
to convey information about the world, ourselves, and our social relation¬
ships?
Part I of the book provides an overview of the different approaches, outlines
the plan of the book, and discusses difficulties in defining discourse - difficul¬
ties that are related to the currency of two different paradigms within linguist¬
ics. Part II provides a detailed discussion of each approach to discourse - one
chapter for each approach. Included in each chapter is discussion of the work
of scholars central to the development of the approach, i.e. the central ideas,
concepts, and methods of each approach. Also included in each chapter is an
extensive sample analysis illustrating how each approach handles specific
phenomena and problems of discourse. Although the sample analyses begin
with some of the same issues, problems, and data (centered on either question-
answer sequences or referring expressions), they diverge sharply once the
concepts and methods of each approach are applied (such that the different
chapters also provide analyses of phenomena as diverse as participation frame¬
works, the organization of interviews, existential there sentences, and thematic
 Preface and Acknowledgments ix

constraints on pronouns). Part III compares approaches to discourse in a more

systematic and abstract way: What does each approach assume about the
relationships between structure and function, text and context, language and
communication? Although these issues are quite different in some ways, they
all require reconciliation of what is often defined as part of “language” with
what is not typically defined as part of language. The final chapter searches
for a possible synthesis among the approaches.
The issues discussed in this book have been ones that have long occupied
my attention. When I was still a graduate student in linguistics at the Uni¬
versity of Pennsylvania, one of the questions in my PhD exam noted the
proliferation of perspectives interested in “language use,” e.g. pragmatics, socio¬
linguistics, ethnolinguistics, and ethnomethodology. The question then asked
us to compare these perspectives, e.g. did they focus ypon the same problems,
use the same methods of analysis, and so on. I passed the exam, but the
question continued to haunt me. When I began teaching at Georgetown
University, I found myself intermittently confronting similar questions through
the courses that I taught. Although we sometimes ended up talking about some
of the same phenomena (e.g. discourse markers) and issues (e.g. meaning) in
different courses, our starting points (and often our final perspectives) were
always strikingly different. In 1987, I was invited to teach a course at the
Linguistic Society of America Institute, held that year at Stanford University:
the Institute organizers suggested a course that would compare approaches to
discourse analysis. This course was an exciting opportunity for me to explicitly
work through some of the theoretical and methodological differences among
approaches to discourse. About a year after the Linguistics Institute, Philip
Carpenter (from Blackwell Publishers) approached me with an idea for a book
comparing current approaches to discourse analysis. His proposal could not
have been more finely timed to coincide with my own thinking. However, it
is important to note that the way each approach to discourse analysis has been
incorporated into linguistics has, not surprisingly, changed a great deal since
I first became interested in discourse analysis. I expect the relationship be¬
tween discourse analysis and the rest of linguistics to continue to change: I
hope that this book will help people think about the way such change can
proceed.
Many people have contributed, both directly and indirectly, to the ideas and
the analyses in this book. Although such contributions are hardly limited to
those who read preliminary drafts of my chapters, the latter include Ralph
Fasold, Esther Figueroa, Michael Geis, John Gumperz, John Searle, Deborah
Tannen, Sandra Thompson, and Peter Trudgill. None is responsible for the
final form or content of the chapters. My students at Georgetown University
have always been, and continue to be, a source of challenge and motivation,
their contribution to this book has been invaluable. More recently, the oppor¬
tunity to use a preliminary version of this book during a course that I taught
at the University of California, Berkeley, and to be part of a reading group
(on narrative analysis) provided me with yet other sources of challenge and
X Preface and Acknowledgments

motivation. Material support for preparation of much of the data and analysis
reported here was provided by NSF grant BNS-8819845. A sabbatical from
Georgetown and a year as a visiting scholar at Berkeley also provided the time
and resources needed to complete the work.
Last, but not least, I want to thank my family, especially my husband Louis
Scavo, for being a continual source of support and encouragement. Our son
David has also contributed in immeasurable ways by giving me good reasons
to take time off and for making that time off so much fun: it is to him that I
dedicate this book.

The author and publisher gratefully acknowledge permission to reproduce

extracts from the following: Foundations of Sociolinguistics (1974) by Dell
Hymes. Reprinted by permission of University of Pennsylvania Press; Prin¬
ciples of Pragmatics (1983) by Geoffrey Leech. Reprinted by permission of
Longman Publishing Group; Structures of Social Action (1984), edited by
J. Maxwell Atkinson and John Heritage. Extract by Gail Jefferson. Reprinted
by permission of Cambridge University Press and Gail Jefferson; Discourse
markers (1987) by Deborah Schiffrin. Reprinted by permission of Cambridge
University Press; Talking Voices (1989) by Deborah Tannen. Reprinted by
permission of Cambridge University Press; Linguistic Framing of Discourse
(1993), edited by Deborah Tannen. Extract by Deborah Schiffrin. Reprinted
by permission of Oxford University Press; Semiotica (80), pp. 121-51 (1990).
Extract by Deborah Schiffrin. Reprinted by permission of the publisher; Prag¬
matics 1/1: 71-106 (1991). Extract by Jack DuBois. Reprinted by permission
of the author; Language and Communication 13/2: 133-6 (1993). Extract by
Deborah Schiffrin. Reprinted by permission of the publisher.
Part I The Scope of Discourse Analysis
'
Since discourse analysis is one of the most vast, but also least defined, areas
in linguistics, it is important to begin this book by commenting on the scope
of discourse analysis. The two chapters in this introductory section do this in
two different ways.
Chapter 1 addresses the vastness of discourse analysis by presenting an over¬
view of the six approaches to be compared in the book: speech act theory,
interactional sociolinguistics, ethnography of communication, pragmatics, con¬
versation analysis, variation analysis. This selection seems to represent the
diversity of approaches (from very different sources) that underlie contempor¬
ary discourse analyses. (Compare Taylor ai},d Cameron’s (1987) review of
social psychological, speech act, exchange structure, Gricean pragmatic, and
ethnomethodological approaches to discourse, or Potter and Wetherell’s (1987)
review of speech act theory, ethnomethodology, and semiology.) Chapter 1
introduces the approaches through examples illustrating issues that prompted
scholars to begin to think about language as “discourse”. It also outlines the
general direction of the book.
Chapter 2 addresses the vastness of discourse analysis from another direc¬
tion: how can we define discourse analysis in a way that captures it as a field
of linguistics and differentiates it from other studies? I suggest in chapter 2
that two prominent definitions of discourse (as a unit of language larger than
a sentence, as language use) are couched within two different paradigms of
linguistics (formalist, functionalist).
■
1 Overview

A *

1 Introduction

Discourse analysis is widely recognized as one of the most vast, but also one
of the least defined, areas in linguistics (e.g. Stubbs 1983: 12; Tannen 1989a:
6-8). One reason for this is that our understanding of discourse is based on
scholarship from a number of academic disciplines that are actually very dif¬
ferent from one another. Included are not just disciplines in which models for
understanding, and methods for analyzing, discourse first developed (i.e. lin¬
guistics, anthropology, sociology, philosophy; see van Dijk 1985), but also
disciplines that have applied (and thus often extended) such models and
methods to problems within their own particular academic domains, e.g. com¬
munication (Craig and Tracy 1983), social psychology (Potter and Wetherell
1987), and artificial intelligence (Reichman 1985).
The goals of this book are to describe and compare several different ap¬
proaches to the linguistic analysis of discourse: speech act theory, interactional
sociolinguistics, ethnography of communication, pragmatics, conversation ana
lysis, and variation analysis. My aim is not to reduce the vastness of discourse
analysis: I believe that at relatively early stages of an endeavor, reduction just
for the sake of simplification can too drastically limit the range of interesting
questions that can and should be asked. Thus, I view the vastness of discourse
analysis not as a weakness, but as a strength, and as a sign of interest and
development. What I hope to do, however, is clarify the scope of discourse
analysis in such a way that it can continue to deal with a wide range of
problems and phenomena - but in a more systematic and theoretically coherent
way.
Before describing the ways that I hope to accomplish this goal, I want to
briefly introduce each approach with the help of a prototypical example
from each perspective (section 2). I then go on to an overview of the book
(section 3).
6 The Scope of Discourse Analysis

2 “Core” examples from different approaches to

discourse

The examples in this section reveal some important similarities, and differ¬
ences, among the approaches to be discussed in the book. I have chosen
examples used by those scholars who were instrumental in developing an
approach; sometimes these examples also reflected, or led to, controversy that
motivated the development of theory and methodology within each approach.
Since other kinds of data and other ways of presenting data (including methods
of transcription and analysis) have developed within each approach, the exam¬
ples here might not seem exactly like those from more recent literature repre¬
senting the different perspectives, e.g. conversation analysts now pay much
more attention to transcription than will be illustrated here. But rather than
try to reflect current diversity within an approach through these examples, I
want to reflect (at least in part) the most salient features, and the conceptual
core, of each approach. Finally, as noted above, the examples here are from
seminal works in each perspective: thus, they also suggest the sort of data that
first prompted scholars to begin to think about language in a different way.
My presentation of approaches to discourse (part II of the book) begins with
the speech act approach to discourse (chapter 3). Two philosophers, John
Austin and John Searle, developed speech act theory from the basic insight
that language is used not just to describe the world, but to perform a range
of other actions that can be indicated in the performance of the utterance itself.
For example, the utterance “I promise to be there tomorrow” performs the
act of “promising.” The utterance “The grass is green” performs the act of
“asserting.” An utterance may also perform more than one act, as illustrated
in (1).

(1) speaker: Can you pass the salt?

hearer: /passes the salt/

S’s utterance Can you pass the salt? can be understood as both a question
(about H’s ability) and a request (for H to pass the salt to S). Although these
two understandings are largely separable by context (the former associated,
for example, with tests of physical ability, the latter with dinner table talk),
this utterance has also been labelled an indirect speech act whose illocutionary
force is an outcome of the relationship between two different speech acts (e.g.
Searle 1975; compare the analyses in Clark 1979; Davison 1975; Ervin-Tripp
1976; Gordon and Lakoff 1975; Green 1975).
The speech act approach to discourse focuses upon knowledge of underlying
conditions for production and interpretation of acts through words. In (1), we
saw that words may perform more than one action at a time and that contexts
 Overview 7
(hypothetical contexts supplied along with hypothetical utterances) may help
to separate multiple functions of utterances from one another. The literal
meanings of words and the contexts in which they occur may interact in our
knowledge of the conditions underlying the realization of acts and the inter¬
pretation of acts. Although speech act theory was not first developed as a
means of analyzing discourse, particular issues in speech act theory (e.g. the
problems of indirect speech acts, multifunctionality and context dependence
illustrated in (1)) lead to discourse analysis. Speech act theory itself also
provides a means by which to segment texts, and thus a framework for defining
units that could then be combined into larger structures.
The approach to discourse that I am calling interactional sociolinguistics
(chapter 4) has very diverse origins, for it stems from anthropology, sociology,
and linguistics, and shares the concerns of all three fields with culture, society,
and language. Some interactional approaches'(especially those influenced by
John Gumperz) focus on how people from different cultures may share gram¬
matical knowledge of a language, but differently contextualize what is said
such that very different messages are produced. Other interactional approaches
(especially those influenced by Erving Goffman) focus on how language is
situated in particular circumstances of social life, and on how it adds (or
reflects) different types of meaning (e.g. expressive, instrumental) and struc¬
ture (e.g. interactional, institutional) to those circumstances. (2) is an example
(from Gumperz 1982a: 30) that illustrates the interactional approach.

(2) Following an informal graduate seminar at a major university, a black

student approached the instructor, who was about to leave the room
accompanied by several other black and white students, and said:
a Could I talk to you for a minute? I’m gonna apply for a fellowship
and I was wondering if I could get a recommendation?
The instructor replied:
b OK. Come along to the office and tell me what you want to do.
As the instructor and the rest of the group left the room, the black student
said, turning his head ever so slightly to the other students:
c Ahma git me a gig! (Rough gloss: “I’m going to get myself some
support.”)

(2) is a report of an actual interchange. In addition to what is said (lines a, b,

and c), the example itself includes the context of the interchange (e.g. the
physical setting, social roles, relationship of speech to other activity) and other
information about what participants are doing (e.g. the physical stance of the
interactants). The example also replicates what is said in a way that reveals
the use of a particular variety of speech. (Many of Gumperz’s own examples,
as well as other interactional analyses, rely upon a more precise transcription
of linguistic, including prosodic, detail.) Gumperz’s analysis of the utterance
Ahma git me a gig! focuses upon how interpretations of the speaker’s intent
are related to different linguistic qualities of the utterance (e.g. phonological
8 The Scope of Discourse Analysis

and lexical variants) as well as the way the utterance is contextually embedded
(e.g. what activities it follows, to whom it is directed). These interpretations
are gathered by asking listeners (including, but not limited to, those present
during the actual interchange) what they thought the speaker meant to convey,
and relating those situated inferences to the means by which the speaker
actually presented the utterance. As (2) thus illustrates, the interactional ap¬
proach relies upon actual utterances in social context: the focus of analysis is
how interpretation and interaction are based upon the interrelationship of
social and linguistic meanings.
The ethnography of communication (chapter 5) is an approach to discourse
that is based in anthropology, and it shares with much traditional anthropo¬
logy a concern for holistic explanations of meaning and behavior. Much of
the impetus for this approach was Dell Hymes’s challenge to Chomsky’s well
known refocusing of linguistic theory on the explanation of competence, i.e. tacit
knowledge of the abstract rules of language. What Hymes proposed instead
was that scholarship focus on communicative competence: the tacit social,
psychological, cultural, and linguistic knowledge governing appropriate use of
language (including, but not limited to, grammar). Communicative competence
includes knowledge of how to engage in everyday conversation as well as other
culturally constructed speech events (e.g. prayer, public oratory). (3) is an
example that illustrates the inclusive thrust of the ethnography of communi¬
cation - so inclusive that cultural interpretation and difference permeates even
what seems to be so basic a notion as what “counts as” communication. (This
example, quoted from Hallowell 1964: 64, is cited in Hymes 1972a.)

(3) An informant told me that many years before he was sitting in a tent
one afternoon during a storm, together with an old man and his
wife. There was one clap of thunder after another. Suddenly the old
man turned to his wife and asked, “Did you hear what was said?”
“No”, she replied, “I didn’t catch it.” My informant, an acculturated
[Ojibwa] Indian, told me he did not know at first what the old man
and his wife referred to. It was, of course, the thunder. The old
man thought that one of the Thunder Birds had said something to
him. He was reacting to this sound in the same way as he would
respond to a human being, whose words he did not understand.

Hymes (following Hallowell’s discussion) uses this example to point out that
even so fundamental a notion as “communication” cannot be assumed to be
constant across cultures. Cultural conceptions of communication are deeply
intertwined with conceptions of person, cultural values, and world knowledge
- such that instances of communication behavior are never free of the cultural
belief and action systems in which they occur. Other ethnographic analyses
focus on how grammar itself reflects cultural knowledge and action systems
(e.g. Ochs 1988; Schieffelin 1990); still others focus on communication
through other verbal media (e.g. Feld 1982 on weeping) or even on the social
 Overview 9
distribution and meaning of silence (Basso 1972; Bauman 1974; Philips 1985;
Saville-Troike 1982). Like the example in (3), such analyses depend upon
extensive familiarity with speakers and with their culture. They also depend
upon the analysis of what is particular about each act of communication -
particular to a people, to a setting, and so on. Finally, they all seek to locate
each particularity within a set of universally available possibilities, but at the
same time, to build those possible generalizations from a representative col¬
lection of particular instances.
A pragmatic approach to discourse (chapter 6) is based primarily on the
philosophical ideas of H. P. Grice. Grice proposed distinctions between dif¬
ferent types of meaning and argued that general maxims of cooperation pro¬
vide inferential routes to a speaker’s communicative intention. Pragmatics is
most concerned with analyzing speaker meaning at $ie level of utterances and
■ this often amounts to a sentence, rather than text, sized unit of language use.
But since an utterance is, by definition, situated in a context (including a lin¬
guistic context, i.e. a text), pragmatics often ends up including discourse
analyses and providing means of analyzing discourse along the way. (4) is an
example that illustrates the interplay between cooperation and inference so
critical to a Gricean approach. (4 is from Grice 1975: 51.)

(4) A: Smith doesn’t seem to have a girlfriend these days.

B: He has been paying a lot of visits to New York.

Like the speech act example in (1), the prototypical pragmatics example is a
constructed utterance in a constructed context. The issue driving the construc¬
tion and analysis of (4) is the lack of obvious connection between A’s Smith
doesn’t seem to have a girlfriend these days and B’s He has been paying a lot
of visits to New York. Grice points out that the lack of connection does not
prevent us from trying to interpret B’s utterance as cooperative at a level of
understanding not readily available from the meanings of the words. What
hearers do is supplement the literal meaning of utterances with an assumption
of human rationality and cooperation: these allow B to infer that A has
implicated that Smith has a girlfriend in New York. In other words, despite
the lack of connection between A’s and B’s remarks, A implicates that which
he must be assumed to believe (Smith has a girlfriend in New York) in order
to maintain the assumption that he is following the maxim of relation (i.e.
being relevant). Thus, what Gricean pragmatics suggests is that human beings
work with very minimal assumptions about one another and their conduct, and
that they use those assumptions as the basis from which to draw highly specific
inferences about one another’s intended meanings.
Conversation analysis (chapter 7) offers an approach to discourse that is also
based in philosophy, but in the perspective known as phenomenology, associ¬
ated with Alfred Schutz. Its underlying concerns were more extensively articu¬
lated by a sociologist, Harold Garfinkel, who developed the approach known
as “ethnomethodology,” and then applied specifically to conversation, most
10 The Scope of Discourse Analysis

notably by Harvey Sacks, Emmanuel Schegloff, and Gail Jefferson. Conversa¬

tion analysis (and ethnomethodology) differs from other branches of sociology
because rather than analyzing social order per se, it seeks to discover the
methods by which members of a society produce a sense of social order. Con¬
versation is a source of much of our sense of social order, e.g. it produces
many of the typifications underlying our notions of social role. Conversation
also exhibits its own kind of order and manifests its own sense of structure.
The example in (5) is an exchange (a telephone call opening) that seems to
violate a telephone rule that the person answering the phone is the one who
talks first. As Schegloff (1972a) makes clear, however, (5) actually illustrates
the workings of a deeper rule of sequencing in talk. (The example is from
Schegloff 1972a: 356.)

(5) (Police make call)

(Receiver is lifted, and there is a one second pause)
police: Hello.
other: American Red Cross.
police: Hello, this is Police Headquarters . . . er, Officer Stratton (etc.).

(5) is the only case in Schegloff’s large corpus of telephone openings in which
the Police (the party who had made the phone call) talk first. Although this
case seems unusual, Schegloff uses it (treats it “seriously,” p. 356) as the basis
from which to search for a deeper formulation about telephone openings and
about sequences in general. He ends up suggesting that (5) is a summons-
answer sequence. The telephone ring resulting from the call made by the Police
in (5) is a summons. A summons opens a conditional relevance for a second
part of a sequence, an answer. Although a called party typically answers the
telephone ring issuing the summons by saying Hello?, there is no such answer
in (5): “Receiver is lifted, and there is a one second pause.” The Police’s
“Hello” is thus a response to the “empty” answer slot: “Hello” redoes the
summons. Far from being an anomaly, then, (5) reflects the regular operation
of adjacency pairs in general and summons-answer sequences in particular:
the sequencing of moves provides for a coordinated entry into the conversa¬
tion, and for an orderly exchange of turns within the conversation.
As (5) illustrates, conversation analysis is like interactional sociolinguistics
in its concern with the problem of social order, and how language both creates,
and is created by, social context. Both approaches also focus on detailed
analysis of particular sequences of utterances that have actually occurred. But
unlike the interactional sociolinguistic willingness to judge participants’ inter¬
pretation and intent with the help of contextual information, conversation
analysts seek generalizations about context - and about social conduct and
social life - within the progression of utterances themselves.
A variationist approach to discourse (chapter 8) stems from studies of
linguistic variation and change. Both the initial methodology and the theory
underlying such studies are those of William Labov. (Labov and Fanshel
 Overview 11

(1977) have also applied a perspective to discourse that is more similar to

speech act theory, i.e. “comprehensive discourse analysis.”) Fundamental as¬
sumptions of variationist studies are that linguistic variation (i.e. heterogen¬
eity) is patterned both socially and linguistically, and that such patterns can
be discovered only through systematic investigation of a speech community.
Although traditional variationist studies have been limited to semantically
equivalent variants (what Labov 1972a calls “alternative ways of saying the
same thing”), such studies have also been extended to texts.
An important part of the variationist approach to discourse is the discovery
of formal patterns in texts (often narratives) and the analysis of how such
patterns are constrained by the text. (6) is an example (from Labov 1972b:
387) that illustrates. (Since Labov’s introduction to this example helps to
illustrate the approach, I include this in 6.) t,

(6) One of the most dramatic danger-of-death stories was told by a

retired postman on the Lower East Side: his brother had stabbed him
in the head with a knife. He concludes:
And the doctor just says, “Just about this much more,” he says,
“and you’d a been dead.”

Labov uses this extract as part of his discussion of the basic structure of
narrative. The utterance being presented is a type of evaluation: the means by
which narrators highlight different aspects of a reported experience as a way
of revealing the point of the story. Although evaluations are sometimes sep¬
arate sections of stories, they are also distributed throughout narrative and
embedded within narrative clauses themselves. (Narrative clauses are typically
event clauses that report “what happened.”) Embedded evaluations rely upon
deviations from the simple syntactic structure typical of a narrative clause.
The evaluation in (6), for example, illustrates an evaluative device that Labov
(1972b: 387) calls a comparator, a functional classification that includes
“negatives, futures, modals, quasimodals, questions, imperatives, or-clauses,
superlatives, and comparatives.”
As illustrated in (6), a variationist approach to discourse utilizes Some of the
basic tools of linguistic analysis: it segments texts into sections, labels those
sections as part of a structure, and assigns functions to those sections. This
approach thus allows more context independence (i.e. a greater degree of auto¬
nomy for “text” in relation to context) than would be allowed, for example,
in interactional sociolinguistics, ethnography of communication, or conversation
analysis. An indication of this in (6) is Labov’s willingness to discuss the evaluat¬
ing clause as separate from the rest of the story - to treat it as an example of
a structural unit and functional type that can be extracted from its story for
comparisons with other evaluative devices. The variationist approach also
integrates traditional linguistic categories into a framework of textual analysis.
In sum, the examples in this section revealed some important features of the
approaches to be discussed in this book: what count as data, what problems
12 The Scope of Discourse Analysis

and questions motivate analysis, how to address or resolve a problem. To

oversimplify a bit, speech act theory focuses on communicative acts performed
through speech: data are typically constructed utterances in hypothetical con¬
texts that are chosen to illustrate the interplay between text and context that
mutually informs production and interpretation of the acts performed through
words. Interactional sociolinguistics focuses on the social and linguistic meanings
created during interaction: although hearers draw inferences about speakers’
intent (as in speech act theory and pragmatics), the inferences are considerably
broader and more varied and they are based on a wide array of verbal and
nonverbal cues that are part of cultural repertoires for signalling meaning (and
can be discovered only through the collection of actual utterances). The eth¬
nography of communication focuses on language and communication as cul¬
tural behavior: the status and significance of any particular act can be discovered
only as part of a matrix of more general meanings, beliefs, and values that extend
far beyond the knowledge of the grammar of one’s language. Since such matrices
pervade and organize a great deal of life within a particular society - and since
they are culturally relative - the ethnography of communication requires extensive
fieldwork within a community as well as comparisons between communities.
Pragmatics focuses on meaning, context, and communication of constructed
utterances in hypothetical contexts. The communicative meaning of a particu¬
lar utterance is derived through general assumptions about human rationality
and conduct; together with the literal meaning of utterances, these assump¬
tions are the basis from which to draw highly specific inferences about in¬
tended meanings. Conversation analysis focuses on sequential structures in
conversation: the mechanics of conversation provide a basis through which
social order (including a sense of “context”) is constructed. Although conver¬
sation analysts are careful to transcribe the details of utterances, they pay little
attention to linguistic categories of sound, structure, or meaning. Finally,
variation theory focuses on structural categories within texts and the way
syntactic structure (and variation) helps to define and realize those structures.
Like the other approaches (except for speech act theory and pragmatics),
variation analysis requires the close analysis of what is actually said. Although
it pays close attention to linguistic structure, it relies less upon non-linguistic
(contextual) detail than interactional sociolinguistics or ethnography of com¬
munication. In the next section, I describe in more detail how the rest of this
book will develop the description of, and comparison among, the approaches
that I have just introduced.

3 Plan of the book

Chapters 3 through 8 present the six approaches to discourse analysis just

introduced - one chapter for each approach. Within each approach, I provide
 Overview 13
a general background with discussion of key issues and critical concepts. My
description of each approach will focus primarily upon the work of scholars
who have been most central to the development of that approach. However,
because the approaches to be described have been so influential not only in
discourse analysis, but also for the other domains that they more directly
address (as noted above), a substantial body of work (supportive and critical,
faithful and revisionist) has developed from the original insights upon which
I mostly depend. I do not mean to slight these later works, and I make reference
to them in the course of discussion of specific points. However, it is because
the original works have spurred so much scholarly discussion and research
that I want to take them as the source of the approaches to discourse to be
developed in my own discussion.
,H &

3.1 Describing different approaches to discourse analysis

My first task in this book is the description of the six different approaches to
discourse analysis noted above. Although these approaches often overlap in
the work of particular scholars, we have already seen that they differ in several
important ways. What underlies my decision to differentiate these six ap¬
proaches is what I believe to be their most significant characteristic: they have
very different origins.2 The origin of an approach provides different theoretical
and metatheoretical premises that continue to influence assumptions, con¬
cepts, and methods. For example, different origins may be responsible for
different assumptions and beliefs about language - assumptions about the
stability of linguistic meaning, the role of speaker intentionality, the degree to
which language is designed for communicative purposes, and the contribution
of linguistic meaning to interactive meaning.
Other differences that can be at least partially traced to different origins
include beliefs about methods for collecting and analyzing data. For example,
some approaches focus intensively on a few fragments of talk (e.g. interac¬
tional sociolinguistics), others focus on distributions of discourse items across
a wide range of texts (e.g. variationists). Some require a great deal of social,
cultural, and personal information about interlocutors and may use interloc¬
utors as informants in analysis of their own talk (e.g. ethnography of commun¬
ication); others assume an idealized speaker/hearer whose specific social, cultural,
or personal characteristics do not enter into participant strategies for building
text at all (e.g. pragmatics). Methodological differences such as these are due,
partially, to different theoretical assumptions - assumptions that are based in
the different origins noted above. If it is assumed, for example, that linguistic
meaning is less important to interactive meaning than are sequential structures
of talk, then an analyst would pay little attention to linguistic form and
structure per se (e.g. ns in conversation analysis). In short, no methodological
preferences are reached in a vacuum: they are all the product of more general
beliefs in what constitutes data and what counts as evidence and proof.
14 The Scope of Discourse Analysis

Finally, how a perspective developed may continue to provide a set of

different practical interests or ultimate goals in addition to (or even, instead
of) the analysis of discourse per se. This may help us understand why particu¬
lar findings are expressed in a certain manner (e.g. through abstract rules),
interpreted in a certain way (e.g. in relation to cross-cultural communication),
or seen as relevant to somewhat different issues (e.g. to semantics rather than
linguistic variation).
Also included within each chapter is an extensive sample analysis showing
how each approach can be applied to a specific problem. The sample analyses
are quite detailed for several reasons. First, the approaches being discussed are
supposed to analyze discourse (thus the label “discourse analysis,” not “dis¬
course theory” or “discourse studies”). It is important, then, to see how they
can be applied to concrete problems and to translate the more general over¬
views within each chapter (and then in a more abstract, comparative mode,
in part III; see below) into empirical tools. Second, on a more personal level,
I believe that the best way to learn about something is to see how it works. It
is in this spirit that I try to provide an account of how each approach proceeds
(not so much what it finds, but how it goes about “doing” its finding) and I
would encourage readers to do the same either through sample problems
suggested at the end of each chapter or through problems of their own making.
For these purposes, I strongly encourage readers who want to use this book
as a “manual” to have available data of some sort (for suggestions, see
appendix 1; some data samples are provided in appendix 3).
We will see that the sample analyses draw upon data in ways not always
typical of the approaches being illustrated, e.g. I have never seen a Gricean
pragmatic analysis of referring terms in a story (although this is what I present
in chapter 6). I am intentionally relying upon the use of transcriptions of
language use (see discussion of sociolinguistic interviews in appendix 1, also
chapters 5 and 8), simply because I believe that the approaches to be discussed
allow (indeed, demand) such data not only if they are to be seen as com¬
parable, but if they are to be examined for signs of synthesis. The data that I
use are primarily from sociolinguistic interviews. Elsewhere (Schiffrin 1987a),
I have argued that sociolinguistic interviews provide data well suited to the
different purposes of discourse analysis. We will have a chance to examine
sociolinguistic interviews themselves as discourse - their status as a kind of
“mixed genre,” i.e. a mix of interview and conversation - in the ethnographic
chapter, chapter 5. That chapter will also draw upon a small set of tape-
recorded library reference interviews. Appendix 1 provides further discussion
of how different corpora can provide data for the analysis of discourse.
In addition to drawing upon data in ways that may differ from the ap¬
proaches being illustrated, I will also use discourse transcription conventions
that may differ from those used by other approaches. Specifically, I will not
adopt the transcription conventions used by most interactional sociolinguists
for my analysis of data illustrating that approach (chapter 4); nor will I use
those conventions used by conversation analysts (in chapter 7). Rather, I will
 Overview 15

use a single set of conventions in all the analyses, the one I use in my own
work (e.g. Schiffrin 1987a), not because I believe that it is best (in fact, it
provides no way to represent many of the qualities that can be represented by
the others), but because I want to use one system throughout, and this is the
one with which I am familiar. Readers interested in selecting a system for their
own use, or in comparing different conventions, will find four different sets
of conventions in appendix 2.
Each approach to be discussed adopts a slightly different view of discourse,
and provides a different (sometimes radically different) way of analyzing
utterances. Indeed, the term “discourse analysis” is not used by all of the
perspectives to be discussed. Although variationist approaches do refer to
discourse (or sometimes, text) analysis, pragmatics and speech act theory refer
instead to analyses of “language in use” or “jp context”; “conversation ana¬
lysis” is a term used by scholars with an ethnomethodological orientation;
interactional approaches refer to “interactional sociolinguistics”; ethnographic
approaches share with pragmatics and speech act theory a focus on “language
in use” or “in context” (although their conception of “use” and “context”
differs from that of the philosophically based approaches). These differences
in terminology do capture slightly different domains of interest (both analytical
and theoretical). Finally, as noted earlier, not all the approaches to discourse
to be reviewed explicitly concern themselves with discourse; even with those
that do, there are often other wider (or overlapping) domains of interest. The
broader topics to which discourse is seen as relevant have important influences
on the way in which its analysis is approached.

3.2 Comparing different approaches to discourse

In addition to describing different approaches to discourse, I will also compare

those approaches. I will try to facilitate a comparison among approaches in
two ways: through the focus of my sample analyses in chapters 3 through 8
(section 3.2.1), and through the discussion of key concepts and assumptions
in chapters 9, 10, and 11 (section 3.2.2).

3.2.1 Ongoing comparisons: comparing sample analyses

To enhance the comparative value of my descriptions of the approaches, I have

decided to orient my sample analyses around two phenomena: (a) questions
(and the sequences they initiated) to be analyzed in terms of speech act theory,
interactional sociolinguistics, and ethnography of communication; (b) refer¬
ring expressions (in referring sequences) to be analyzed in terms of pragmatics,
conversation analysis, and variation analysis. We see not only that the differ¬
ent approaches provide different answers to some of the same questions, but
that they highlight different facets of both questions and referring expressions.
16 The Scope of Discourse Analysis

An analogy may help. Imagine that we are trying out a number of different
tools (a saw, a butter knife, a screwdriver, and a pair of scissors) to accomplish
one particular task (we want to cut a piece of bread). We would probably
learn not only about the tools, but also about the task that we are trying to
accomplish (cutting) and the material that we are handling (bread). This
analogy suggests that different analyses of referring terms (or questions) may
tell us something about the analytical tools provided by the approaches them¬
selves (e.g. pragmatics, speech act theory, etc.), and reveal something interest¬
ing about question-answer sequences and referring expressions. Those who
expect to find exhaustive analyses of either phenomenon, however, will be
disappointed: both are vast topics to which linguists have devoted a great deal
of scholarly attention. Rather, what I try to do is find particular issues
pertaining to either topic that can be understood through the approach being
described in each chapter.
Notice, then, that not exactly the same problems or issues will be addressed
in each chapter. Returning to my “breadcutting” analogy, for a moment, I
noted above that we would learn about three things if we were trying different
tools to accomplish one particular task: the task we are trying to accomplish,
the material we are handling, and the tools we are using. One way of organ¬
izing this book would have been to keep the task completely constant, e.g.
analyze the distribution of “there is” constructions (as in chapter 7), or keep
the material completely constant, e.g. use a single episode for data. I felt,
however, that this would have been difficult since different approaches define
even what seems to be the “same” topic differently. (It also would have
resulted in an overly redundant - and perhaps boring - set of chapters.)
Indeed, even though we will see some of the same sample utterances reappear
in different chapters as they are used and reused to make different points, I
would maintain that even though the utterances might seem like the “same”
examples, they should not necessarily be interpreted as the same “data.” In
the sample analyses, then, I decided to maintain some constancy (the two
general phenomena) and allow some diversity (different aspects of the two gen¬
eral phenomena).
Although there are many other issues and problems that discourse analysts
address in their research, I have chosen questions and referring expressions
for several reasons. First, questions and referring expressions have both been
said to be central concerns of discourse analysis. Question—answer sequences,
for example, are a paradigm example of adjacency pairs (a construct central
to conversation analysis (chapter 7)): adjacency pairs not only illustrate the
sequential foundation of discourse, but play a key role in the view of discourse
as fundamentally organized on a pairwise (two part) basis. Questions are also
puzzling in and of themselves: some sentences that might be syntactically
identified as questions are functionally not questions at all: one could not
respond with just Yes to the syntactic question Can you pass the salt? (our
earlier example 1) without being considered rude, making a joke, or having
misunderstood. Similarly, some sentences that would not be syntactically
 Overview 17
identified as questions act like questions in that they “expect” an answer. A
therapist’s I notice you’re late again, for example, is understood to demand
an explanation (an answer) as certainly as "Why are you late again? Examples
such as these suggest not only that questions are difficult to define, but that
what “counts as” a question (chapter 3) is strongly tied to the interactional
(chapter 4) and institutional (chapter 5) contexts in which it is produced.
The study of referring expressions also plays a central role in forming some
scholars’ views of what discourse analysis is about: “the study of the functions
of syntax and reference (e.g. matters of definiteness/indefiniteness) has come
to represent, for certain linguists, the proper domain of discourse analysis”
(Prince 1988: 166). One reason why discourse analysts are concerned with
referring terms is that the processes by which expressions allow one to refer
to an entity (person, thing, concept, etc.) in the univ^se of discourse involve
not only speakers (their intentions, actions, and knowledge) but also hearers.
Clark and Wilkes-Gibbs (1986), for example, speak of referring as a “collab¬
orative process”: they suggest that although a speaker can propose a referent,
the identification of the referent needs to be seen as an outcome of speaker-
hearer interaction (see also Martinich 1984: 161-2). Establishing a referent
that a hearer can identify also involves what Green (1989: 47) calls “the
cooperative exploitation of supposed mutual knowledge” - knowledge that is
inferrable not just from our world knowledge (our “encyclopedic” knowl¬
edge), but from the information made accessible to us from both text (prior
and current) and context.
There are other, more specific reasons why I have chosen these two particu¬
lar areas of research as a basis for comparison among different approaches
to discourse analysis. Question-answer pairs capture the syntagmatic thrust of
discourse analysis (i.e. questions and answers are sequentially organized) and
referring expressions capture the paradigmatic thrust of discourse analysis (i.e.
speakers refer to an entity by choosing among various expressions). Neverthe¬
less, analyses of both question-answer sequences and referring expressions
also show that sequential location is interdependent with the options provided
within a particular sequential slot. The context of a question has important
bearings on how we ask and answer that question. The way we refer to
something is influenced by where in a discourse we are making that reference,
and whether a particular referring expression is an initial (or subsequent)
mention of an entity.

3.2.2 Concluding comparisons: assumptions and concepts

In addition to the comparisons available through the descriptions and sample

analyses themselves, other comparisons will be made towards the end of the
book, following the specific descriptions of each approach. The main part of
this concluding comparison will be centered on three issues that are central to
discourse analysis and about which discourse analysts must make assumptions.
18 The Scope of Discourse Analysis

the relationship between structure and function, the relationship between text
and context, and the nature of communication. I believe that these issues are
critical to discourse analysis, and that all approaches take a stand (albeit often
implicitly) on the relationship between structure and function, text and con¬
text, and discourse and communication, simply because these conceptual dis¬
tinctions are all variants of the dichotomy between what is considered part of
language and what is not (see chapter 2). We see that the approaches take
surprisingly different positions on issues such as the interplay of structure and
function (chapter 9), the degree to which text and context can penetrate one
another (chapter 10), and the role of intention and intersubjectivity in com¬
munication (chapter 11).
My comparison among approaches also responds to more general issues in
discourse analysis and what seem to be two developing needs. We already
know a great deal about some very basic discourse phenomena, e.g. turn-taking,
repair, topic organization, story telling, discourse markers, conversational
inference and style. There now seems to be a need to move from empirical
studies of how we use language to (a) the development of models and theories
that help us organize our knowledge about how discourse works, and (b) links
between our discourse models/theories and our models/theories of language in
general. I believe that a less fragmented vision of discourse analysis may very
well help us in these endeavors. The debate about structure and function in
discourse (alluded to above), for example, could then be considered as one
variant of the more general debate about the relationship between linguistic
structure (the phonological and syntactic structures making up what is typic¬
ally thought of as “grammar”) and language use (the way people use gram¬
matical resources to communicate with one another).
Given the wide variety of studies that are considered to be discourse analysis,
is there any theoretical or conceptual unity to this inquiry? Are there simil¬
arities among approaches that override their differences? For example, one of
the earliest discourse analysts, Zellig Harris (1951), proposed that the goal of
discourse analysis is to discover how it is that discourse differs from random
sequences. Many more recent analysts propose a surprisingly similar goal.
Michael Stubbs (1983: 15) states: “People are quite able to distinguish be¬
tween a random list of sentences and a coherent text, and it is the principles
which underlie this recognition of coherence which are the topic of study
for discourse analysts.” It is important not only to know whether such a goal
is shared by all discourse analytic approaches, but also to know whether
strategies for accomplishing such a goal are shared: Harris, for example,
confined his analytic methods strictly to formal patterns within the text,
whereas Stubbs includes extensive information from outside of the text.
Chapter 12 is the concluding chapter. In addition to proposing some very
general similarities among the approaches, I also try to utilize the two broad
problems to which my sample analyses pertain in a still more general way.
Questions are considered in chapters on speech act theory, interactional socio¬
linguistics, and ethnography of communication; referring expressions are
 Overview 19

considered in chapters on pragmatics, conversation analysis, and variation

analysis. As I will make clear in chapter 12, the order of chapters, and thus
the type of inquiry for each area of empirical focus, is not random: they reflect
a transition (broadly speaking) from a focus upon the individual (whether the
actions, knowledge, or intentions of a self) to a focus upon interaction (how
self and other together construct what is said, meant, and done) to a focus upon
the semiotic systems shared and used by self and other during their interactions
(language, society, and culture). An ability both to build such transitions (from
self to self/other to shared semiotic systems) into one’s theory, and to allow
and account for them in one’s practice, is a crucial part of a discourse analysis
that seeks to integrate what speech act theory, interactional sociolinguistics,
ethnography of communication, pragmatics, conversation analysis, and vari¬
ation analysis can offer, both individually and together, to the analysis of
utterances.

Notes

1 Van Dijk (1985) shows that discourse analysis really has a rather long history - if
one allows classical rhetoric to be considered as discourse analysis.
2 Despite basing my identification and selection of approaches primarily on historical
factors, I also believe that my selection is representative of much of the work being
done in discourse analysis today. Furthermore, other reviews (that are not as histor¬
ically based) also differentiate largely the same approaches: Taylor and Cameron
(1987), for example, compare social psychological, speech act, exchange structure,
Gricean pragmatic, and ethnomethodological approaches to discourse.
2 Definitions of Discourse

1 Introduction

In chapter 1, I introduced six approaches to discourse analysis and outlined

my plan for describing and comparing these approaches in parts II and III of
the book. In this chapter, I consider three different definitions of discourse
and introduce several basic issues that underlie the descriptions of, and com¬
parisons among, the approaches.
Two paradigms in linguistics provide different assumptions about the general
nature of language and the goals of linguistics (section 2).1 These paradigms
are sometimes differently labelled: what Newmeyer (1983) calls a formalist
paradigm is similar to Hymes’s (1947b) structuralist paradigm and to what
Hopper (1988) calls a priori grammar; the functionalist paradigm is sometimes
also called emergent (Hopper) or interactive (Mey et al. 1992). The two
paradigms make different background assumptions about the goals of a lin¬
guistic theory, the methods for studying language, and the nature of data and
empirical evidence. These differences in paradigm also influence definitions of
discourse: a definition derived from the formalist paradigm views discourse as
“sentences” (section 3), a definition derived from the functionalist paradigm
views discourse as “language use” (section 4). A third definition of discourse
attempts to bridge the formalist-functionalist dichotomy (section 5). The
relationship between structure and function in general is an important issue
that is related to other issues central to discourse analysis (section 6).

2 Formal and functionalist paradigms

Discourse is often defined in two ways: a particular unit of language (above

the sentence), and a particular focus (on language use; see Schiffrin 1987a: 1).
 Definitions of Discourse 21
These two definitions of discourse reflect the difference between formalist and
functionalist paradigms. After briefly reviewing the two paradigms, I discuss
discourse as structure (section 3) and discourse as function (section 4).2
Hymes (1974b: 79) suggests that the following qualities contrast structural
(i.e. formalist) with functional approaches.

“Structural” “Functional”

Structure of language (code) as Structure of speech (act, event)

grammar as ways of speaking

Use merely implements, perhaps Analysis of use prior to analysis

limits, may correlate with, what of code; organization of use
is analyzed as code; analysis of discloses additional features and
code prior to analysis of use relations; shows code and use in
integral (dialectical) relation

Referential function, fully Gamut of stylistic or social

semanticized uses as norm functions

Elements and structures Elements and structures as

analytically arbitrary (in cross- ethnographically appropriate
cultural or historical (“psychiatrically” in Sapir’s
perspective), or universal (in sense)
theoretical perspective)

Functional (adaptive) Functional (adaptive)

equivalence of languages; all differentiation of languages,
languages essentially varieties, styles; these being
(potentially) equal existentially (actually) not
necessarily equivalent

Single homogeneous code and Speech community as matrix of

community (“replication of code-repertoires, or speech styles
uniformity”) (“organization of diversity”)

Fundamental concepts, such as Fundamental concepts taken as

speech community, speech act, problematic and to be
fluent speaker, functions of speech investigated
and of languages, taken for
granted or arbitrarily postulated

Leech (1983: 46) suggests other ways that formalism and functionalism are
“associated with very different views of the nature of language, including the
following:

1 Formalists (e.g. Chomsky) tend to regard language primarily as a

mental phenomenon. Functionalists (e.g. Halliday) tend to regard it
primarily as a societal phenomenon.
22 The Scope of Discourse Analysis

2 Formalists tend to explain linguistic universal as deriving from a com¬

mon genetic linguistic inheritance of the human species. Functionalists
tend to explain them as deriving from the universality of the uses to
which language is put in human society.
3 Formalists are inclined to explain children’s acquisition of language
in terms of a built-in human capacity to learn language. Functionalists
are inclined to explain it in terms of the development of the child’s
communicative needs and abilities in society.
4 Above all, formalists study language as an autonomous system, where¬
as functionalists study it in relation to its social function.

At the risk of great simplification, we can say that functionalism is based on

two general assumptions: (a) language has functions that are external to the
linguistic system itself; (b) external functions influence the internal organiza¬
tion of the linguistic system. These shared assumptions contrast functionalism
with approaches that are not concerned with how external processes impinge
upon language (or view such a relationship as irrelevant to the goals of lin¬
guistic theory). They also contrast functionalism with the views of earlier
linguists who largely restricted their analyses to functions within the linguistic
system (e.g. Sapir’s view that speech sounds are functionally organized did not
go outside of language per se) and with the views of contemporary linguists
who view functions as the role a category may play within a sentence (e.g.
relational grammar: Perlmutter 1983) and/or as mathematical representations
from names to values (e.g. lexical-functional grammar: Kaplan and Bresnan
1982).
Formalist views, on the other hand, argue that although language may very
well have social and cognitive functions, these functions do not impinge upon
the internal organization of language. Newmeyer (1983) captures these qual¬
ities in two defining characteristics: autonomy and modularity. First, auto¬
nomy (p. 2):

the grammar of a language is characterized by a formal autonomous

system. That is, the phonology, syntax, and those aspects of meaning
determined by syntactic configuration form a structural system whose
primitive terms are not artifacts of a system that encompasses both human
language and other human facilities or abilities. (Emphasis in original)

The formal autonomy of the grammar, however, does not prevent intersection
with other modules: surface features of phonology, syntax, and semantics
can result from the interaction of the “formal grammar” module with other
equally autonomous modules, each governed by its own set of principles. Such
modules might include perceptual psychology, physiology, acoustics, conver¬
sational principles, and general principles of learning and concept formation.
(See also Newmeyer 1991, and comments on that paper in Harris and Taylor
1991.)
 Definitions of Discourse 23

Although scholars often articulate formalist/functionalist differences in

terms that are mutually exclusive, Bates and MacWhinney (1982) suggest that
differences within the functionalist paradigm bring functionalism either closer
to, or further from, the formalist assumptions of autonomy and modularity.
The most radical functionalist position, for example, would be that external
functions (such as communicative concerns) define primitive categories, such
that there would be no need to posit independently definable, autonomous
grammatical categories (Bates and MacWhinney 1982: 188), e.g. DuBois’s
(1987) suggestion that ergativity is discourse based (also Hopper and Thom¬
pson 1980). A more conservative position would allow an interaction between
form and function, such that external functions would work in tandem with
the formal organization inherent in the linguistic system - influencing it at
certain points in the system, but not fundamentally defining its basic cat¬
egories. (This actually seems to be the position takefi by Newmeyer (1983),
although he presents his work as a strong defense of formal theory.)
Related to the difference in the degree to which external functions condition
the system is a difference in the degree to which the linguistic system itself is open
to functional influences. An extremely useful way of differentiating degrees of
openness of the system is Bates and MacWhinney’s (1982: 178-90) differen¬
tiation of four levels of correlation between form and function. The weakest
correlation (a diachronic form-function relation) requires minimal assump¬
tions about how open the system is to functional influence (and further
suggests that certain points are only open for limited time periods, i.e. when
in change). The strongest correlation (a form-function relation in adult com¬
petence) entails maximal assumptions about the openness of the system to
functional influence.
The two definitions of discourse prevalent in the field reflect the differences
between formalist and functionalist paradigms. After describing these defini¬
tions in the next two sections, I suggest an alternative definition that attempts
to avoid some of the pitfalls of taking either a strong formalist or strong
functionalist approach to the definition of discourse.

3 Discourse: language above the sentence

The classic definition of discourse as derived from formalist (in Hymes’s

1974b terms, “structural”) assumptions is that discourse is “language above
the sentence or above the clause” (Stubbs 1983: 1). Van Dijk (1985: 4)
observes: “Structural descriptions characterize discourse at several levels or
dimensions of analysis and in terms of many different units, categories, sche¬
matic patterns, or relations.” Despite the diversity of structural approaches
noted by van Dijk, there is a common core: structural analyses focus on
the way different units function in relation to each other (a focus shared
24 The Scope of Discourse Analysis

with structuralism in general (e.g. Levi-Strauss 1967; Piaget 1970), but they
disregard “the functional relations with the context of which discourse is
a part” (van Dijk 1985: 4). Since it is precisely this relationship - between
discourse and the context of which discourse is a part - that characterizes
functional analyses, it might seem that the two approaches have little in
common.
Structurally based analyses of discourse find constituents (smaller linguistic
units) that have particular relationships with one another and that can occur
in a restricted number of (often rule-governed) arrangements (cf. Grimes 1975;
Stubbs 1983: chapter 5). In many structural approaches, discourse is viewed
as a level of structure higher than the sentence, or higher than another unit of
text. Z. Harris (1951) - the first linguist to refer to “discourse analysis” -
claimed explicitly that discourse is the next level in a hierarchy of morphemes,
clauses, and sentences. Harris viewed discourse analysis procedurally as a
formal methodology, derived from structural methods of linguistic analysis: such
a methodology could break a text down into relationships (such as equi¬
valence, substitution) among its lower-level constituents. Structure was so
central to Harris’s view of discourse that he also argued that what opposes
discourse to a random sequence of sentences is precisely the fact that it has
structure: a pattern by which segments of the discourse occur (and recur)
relative to each other.
Harris’s approach sought to be a theoretical and methodological extension
of linguistic structuralism, not only because it extended the notion of linguistic
unit to another level, but also because it was methodologically dependent upon
lower-level structural analyses for the identification of higher-level constitu¬
ents: the constituents of discourse were morphemes and morpheme sequences
(words, phrases) that were themselves identifiable through “any grammatical
analysis” of a sentence (p. 1). In addition, the only type of structure admissible
into analysis was what could be investigated by inspection of the data without
taking into account other data, e.g. speakers, context, meanings. However,
Harris’s intention was that the regular recurrence of constituents would corres¬
pond to a semantic interpretation for the discourse - a hope that was quite
consistent with the structural focus (at lower levels) on morphemes as sound/
meaning correspondences.
Although structural approaches have been modified by Harris himself (e.g.
Harris 1988) and by others (e.g. Grimes 1975; Polanyi 1988), what is still
critical to structural views of discourse is that discourse is comprised of units.
Although Harris’s unit was the morpheme (and their combination into sen¬
tences), more recent approaches have identified the clause (e.g. Linde and
Labov 1985), the proposition (e.g. Grimes 1975; Mann and Thompson 1988),
or the sentence (see below) as the unit of which discourse is comprised. Some
scholars also differentiate sources of “connectedness” within discourse and
assign different roles to different units. Holker (1989), for example, suggests
that the linguistic structures of an expression, including both form-based
(morphological and syntactic) and meaning-based (referential and conjunctive)
 Definitions of Discourse 25

relations, create connexity and cohesion. (Coherence, however, would be a

result of the interpreter’s knowledge about states of affairs mentioned in a text.)
Other structural approaches search for multi-based and/or diversified units:
Polanyi (1988), for example, allows structures to be comprised of units as
varied as sentences, turns, speech actions, and speech events.
Consistent with the definition of discourse as language “above the sentence,”
many comtemporary structural analyses of discourse view the sentence as the
unit of which discourse is comprised. Yet several problems stem from the reli¬
ance of definitions and analyses on the smaller unit of “sentence.”
One immediate problem is that the units in which people speak do not always
seem like sentences. Research by Chafe (1980, 1987, 1992), for example, sug¬
gests that spoken language is produced in units with intonational and semantic
closure - not necessarily syntactic closure. Some scholars also believe that the
grammarian’s focus on sentences stems from the value that we - as members
of a literate culture - place on written language (Harris 1980; Hopper 1988).
If we were to focus solely on spoken language, we would be more likely to
view language in terms of intonation units that reflect not underlying gram¬
matical structures, but underlying focuses of consciousness in which informa¬
tion is organized (Chafe 1987).
Support for this view is often found by examining the transcript of a stretch
of speech and noting that the intonational breaks do not always correspond
to syntactic boundaries. In (1), for example, we find chunks of speech that do
not fit our traditional notions of sentencehood:

(1) You can run a hou- whatcha- now whatcha you can- ran a house- you
can run a house a- and do the job, which is important, y can t y- a man
can’t do it himself, and a woman can’t do it himself w- if y’ want it to be
successful. In most cases.

Of course it is possible to transcribe (1) in a way that would make it look

more like a sentence, e.g. by applying editing rules to the non-fluencies and
false starts (Labov 1966; but see Taylor and Cameron 1987: chapter 7). Never¬
theless, one could still argue that the use of a transcription system that builds
upon graphic punctuation symbols does not really capture the way words and
expressions actually cluster together in spoken language (e.g. G. Brown 1977).
Even more critically, one could argue that the use of such devices forces us to
think of such chunks as sentences, rather than as providing an accurate
representation of how speakers themselves produce language, e.g. as interna¬
tionally packaged foci of consciousness (Chafe 1987), as rhetorical amalga¬
mations of clauses (Cumming 1984), in collaboration with interlocutors
(Goodwin 1979). To reflect such concerns, some discourse analysts try to
exclude from their transcription systems those conventions of punctuation
(e g period, commas, capital letters) that are used in written language to
indicate syntactic structure or closure, or to use such devices to capture aspects
of speech production (see appendix 2).
26 The Scope of Discourse Analysis

The reliance on sentence as the unit of which discourse is comprised is

theoretically problematic in other ways. Bloomfield (1933: 170) defined a sen¬
tence as “an independent linguistic form not included by virtue of any gram¬
matical construction in some larger linguistic form.” However, the view that
discourse is a level of structure higher than sentences - more precisely, that
discourse is a structure within which sentences are embedded - sometimes ends
up challenging the view that sentences have grammatical autonomy and clo¬
sure. Many analysts focus upon how syntactic properties of clauses or senten¬
ces contribute to (or, alternatively, are influenced by) higher-level structures
of a text (e.g. Prince 1986; Ward et al. 1991). Interestingly, however, such
analyses sometimes end up arguing that what, at first, seemed to be relatively
static and stable properties of sentence grammar are really dynamic, emergent
byproducts of the processes by which people organize information and transfer
that information to another (e.g. Fox and Thompson 1990; Givon 1979; see
also Goodwin 1979; Schegloff 1979a for proposals that sentences are interac-
tionally constructed). In brief, some analyses end up challenging the very
notion that sentences are what Bloomfield says they are: they challenge the
notion that sentences are independent linguistic forms (for elaboration of this
idea see Hopper 1987, 1988). However, if sentences have no existence outside
of discourse - if they are created by discourse - then it is confusing (and
perhaps even meaningless) to try to define discourse as something larger than
the very thing that it creates, i.e. sentences.
Another consequence of the view that discourse is language above the sen¬
tence is that we may begin to expect discourse to exhibit a structure analogous
to the sentences of which it is comprised - an expectation that may be un¬
warranted (Stubbs 1983: chapter 5). Take, for example, the sentence grammar¬
ian’s use of the term “well formed” as it applies to structures. To illustrate,
we can take Chomsky’s well known example:

(2a) Colorless green ideas sleep furiously.

Although this sentence is meaningless (in the usual sense of meaning, as sense
and reference), it is syntactically well formed. In addition to being meaning¬
less, however, (2b) is also syntactically deviant. (Examples and discussion are
from Sells 1985: 3-4.)

(2b) ’’'Furiously sleep ideas green colorless.

Note that this distinction reappears even when the words do make conven¬
tional sense. Again, Sells’s (1985: 3) examples:

(3c) Revolutionary new ideas appear infrequently.

(3d) ^Infrequently appear ideas new revolutionary.

Sells (p. 4) makes the following important point:

 Definitions of Discourse 27

the real syntactic truth underlying the contrasts in grammaticality seen

above is that while some sequences of the form “Adjective-adjective-
noun-verb-adverb” are syntactically well formed in English, sequences of
the form “Adverb-verb-noun-adjective-adjective” are not.

Structure in this sense just does not seem to apply to discourse: it is simply
not possible to contrast constituent strings of well-formed versus ill-formed
discourse in the same way. One reason has to do with our inability to identify
units of discourse in a way as clear cut (and mutually exclusive) as our ways
for identifying constituents of sentences (e.g. chapter 3). As I discuss in Schif-
frin (1988a: 257-9), a related point is that the types of structures identified
by discourse analysts have not always been comprised of sentences, or indeed
of language units per se, e.g. analysts have spoken of action structures and
turn structures. Indeed, even units that might be defined linguistically are often
viewed in terms of their consequences for the interaction between speaker and
hearer. For example, even though a question-answer pair may be defined as
a semantic sequence in which an incomplete proposition presented by one
speaker is completed by another, such a definition is less relevant to the
concerns of many discourse analysts than the fact that questions may be used
to replicate (or negotiate) social relationships and status (e.g. chapter 5). A
related point is that there are several different levels at which discourse
analysts have identified structure. Some scholars have observed that entire
encounters are framed and bracketed from each other at both their initiation
(Collett 1983; Corsaro 1979; Godard 1977; Goffman 1974; Schegloff 1972a;
Schiffrin 1977, 1981b) and termination (Schegloff and Sacks 1973). Others
focus on how single turns at talk are sequenced (Sacks et al. 1974). Included
between these extremes is a focus on topic structures (Brown and Yule 1983:
chapter 3; Button and Casey 1984; Jefferson 1984; Keenan and Schieffelin
1976), and dialogic pairs (such as question-answer pairs). Because units such
as encounters, topics, dialogic pairs, and turns are so different from each other
in substance, however, it is not at all clear that they form the sort of hier¬
archical structures to which linguists are accustomed at other levels of analysis.
One way to try to get around some of the problems just noted is to adopt
Lyons’s (1977: 385, 387) distinction between system-sentence and text-
sentence. System-sentences are “the well-formed strings that [are] generated
by the grammar” (p. 387), i.e. they “are abstract theoretical constructs, corre¬
lates of which are generated by the linguist’s model of the language-system
(p. 622). Text-sentences, on the other hand, are context-dependent utterance-
signals (or parts of utterance-signals), tokens of which may occur in particular
texts” (p. 622). Lyons’ distinction allows discourse to be comprised of text-
sentences rather than system-sentences. And as Lyons (1977: 632) notes,
“there is no reason to suppose that system-sentences, as such, play any role
in the production and interpretation of utterances (cf. text-sentences).
Defining discourse as text-sentences helps to capture the sorts of internal
dependencies that we expect texts to have. In (4), for example, we see a variety
28 The Scope of Discourse Analysis

of ways that the information presented in one sentence presupposes informa¬

tion in another (cf. Halliday and Hasan 1976; Lyons 1977).

(4a) ELLIPSIS
Jim: Where’s the milk?
Karen: The milk is on the table.
(4b) NOMINAL ANAPHORA
I saw that cat the other day.
It was still wandering around without a home.
(4c) TEMPORAL ANAPHORA
We moved here in 1982.
We didn’t even have jobs then.
(4d) REFERENCE
I saw a robin the other day.
It was the first one I saw this spring.

Viewing the sentences in (4) as text-sentences captures the idea that they are
situated in a linguistic unit larger than themselves; furthermore, the fact that
they are located in this way allows us to account for some of their internal
properties.
Separating text-sentences from system-sentences also allows us to maintain
a definition of system-sentences that is even more abstract than Bloomfield’s
definition. In 1957, for example, Chomsky viewed a sentence as a string of
words with an abstract representation mapping sound to meaning; such a rep¬
resentation is derived in a certain way, i.e. generated by rules. Such a definition
depends upon a theory of generative grammar: other definitions may be
differently attuned to other theories of grammar. If we define discourse in
terms of text-sentences, neither the theory dependency nor the level of abstrac¬
tion of system-sentences is a problem for our definition of discourse. However,
we may still end up returning to the problem of circularity noted earlier:
defining discourse as comprised of just those constituents (text-sentences)
whose definition actually depends on discourse.
We have seen thus far that viewing discourse as a unit above the sentence is
not just a definition of discourse, but a way of leading to a particular type of
analysis. Although this definition and the analysis to which it leads can be
appealing, it also raises some problems. Let me briefly review these problems
before turning to another definition of discourse. First, the view of discourse
as a unit above the sentence allows one to focus quite easily upon how syntactic
properties of clauses or sentences contribute to (or alternatively, are influenced
by) higher level structures of a text (e.g. Linde and Labov 1975, Matthiessen
and Thompson 1987), e.g. specific properties of sentences, such as word order
or typotactic versus paratactic coordination, can be related to the properties of
texts. Such analyses, however, often end up deriving the syntax of sentences
rom the properties of texts and the communicative goals of speakers. As noted
earlier, it is somewhat circular to define discourse as something larger than the
 Definitions of Discourse 29
lower-level unit that it seems to create. Second, the structural view of discourse
places discourse in a hierarchy of language structures, thus fostering the view
that one can describe language in a unitary way that continues unimpeded
from morpheme to clause to sentence to discourse. This view, however, at¬
tributes to discourse the kind of formal regularity often found at lower levels
of linguistic description. (Taylor and Cameron 1987 suggest that this is one
reason why formalisms may be generally attractive to those searching for
patterns.) Thus, the extension of concepts from one level of linguistic descrip¬
tion to discourse can actually perpetuate the view that discourse is parallel in
“kind” to lower level linguistic constituents. But as we noted earlier, this view
might not be warranted: not only is it difficult to define a text as well-formed
or ill-formed, but discourse structures are not always the sort of hierarchical
structures to which linguists are accustomed at other levels of analysis.
Before turning to a functional view of discourse, let us take a sample discourse
and see how we might find its structure. This will allow us to focus not just
on the conceptual consequences, but also on some of the more specific analyt¬
ical consequences of a view of discourse couched within a formalist paradigm.
Let us consider a hypothetical exchange between two colleagues.

(5) JAN: (a) Are you free for lunch today?

Barbara: (b) I have to advise students all day.

(5) illustrates an easily recognizable and familiar discourse structure that we

will be focusing upon in later chapters of this book: a question-answer pair.
We can identify this as a structure because we know the units of which it is
comprised (“question,” “answer”), we know their relationship to one another
(the question opens a proposition that the answer fills), and we know the order
in which they must occur (question before answer). But how is it that we
actually identify (a) as a question? What is the background knowledge that
allows us to hear (or read) a string of words and come up with a label for
those words? These are important questions: if we cannot come up with a way
of identifying questions — of identifying the initial unit of which the structure
is comprised - then we cannot go much further in our analysis of the relation¬
ship between the unit “question” and the unit answer (e.g. Selting 1992;
also see discussion in chapters 3 and 5).
It seems to be quite easy to define (5a) Are you free for lunch today? as a
question, simply because we can rely on purely formal clues typical of inter¬
rogative sentences, e.g. subject—verb inversion. The examples in (6), however,
show that this process need not always seem so automatic, simply because it
is not always easy to find exactly those criteria that allow utterances to be
identified as questions.

(6) (a) You’re free for lunch today?

(b) Free for lunch today?
(c) Lunch today?
30 The Scope of Discourse Analysis

Utterances (a), (b), and (c) in (6) are not questions by the syntactic criterion
noted above. (6a), for example, is clearly a declarative, rather than interrogat¬
ive, sentence. (6b) and (6c) cannot be unequivocally expanded to interrogatives:
we cannot be certain of the fuller sentences underlying the elliptical forms.
The fuller form for Lunch today? (6c), for example, could be a number of
different interrogatives: Are you free for lunch today?, Are you eating lunch
today?, Do you want to eat lunch today? Although these examples thus
suggest the difficulty of relying upon syntax alone as a cue to utterance
identity, they also suggest that we might be able to rely on intonation: note
that (a), (b), and (c), are all represented as having final rising intonation -
thought to be a frequent prosodic indicator of questions (see discussion in
chapter 3).
The example in (7), however, suggests the reverse problem - for it is intona¬
tion, but not syntax, that fails to provide a consistent clue to the identity of
the utterance as a question. (In keeping with many systems for transcribing
conversation (see appendix 2) I am using a period (.) to indicate final falling
intonation typical of what we might find with a declarative sentence.

(7) Do you want to have lunch.

Thus, in (7), the word order is typical of interrogatives, but the intonation
seems more typical of declarative sentences.
Finally, the utterances in (8) and (9) show neither the syntactic nor the
intonational features typical of questions:

(8) (a) I want to ask if you’re free for lunch.

(b) I was wondering whether you wanted to have lunch.

(8a) and (8b) might be called indirect questions and they do seem to “do”
some of the same things as direct questions (with interrogative syntax), e.g.
they elicit a particular kind of information from a respondent. But other
utterances also seem to be directed toward the same purpose:

(9) (a) I’m hungry.

(b) Come eat.

Although (9a) and (9b) are clearly not questions according to either our syn¬
tactic or intonational criteria, the declarative statement in (9a) and the imper¬
ative in (9b) seem to accomplish some of the same things as the questions cited
earlier, and could possibly be answered through the same range of responses.
These examples suggest that syntax and intonation are neither necessary
nor sufficient criteria for the identification of questions. Although question-
answer is an easily recognizable and familiar discourse structure, it turns out
to be difficult to provide criteria allowing us to identify as questions all of the
many different strings of words that we may intuitively think of as questions.
 Definitions of Discourse 31
The problem is that if we cannot come up with a way of identifying the initial
unit of which the structure is comprised, then we cannot go much further in
our analysis of that structure.
To summarize: we have seen in this section that structurally based definitions
of discourse lead to analyses of constituents (smaller units) that have particular
relationships with one another in a text and that can occur in a restricted set
of text level arrangements. They also try to extend methods of linguistic
analysis that have been useful for other levels of linguistic description and/or
to rely upon linguistic characteristics of clauses (or sentences) as clues to
textual structures. We have also seen, however, that identifying structural
constituents of discourse is often a difficult task. The next definition to be
considered replaces what is basically a formalist thrust with a functionalist
thrust: discourse is language use. 11

4 Discourse: language use

In section 2,1 presented several ways that formalist and functionalist views of
language differ, and in section 3, I discussed a view of discourse largely
compatible with formalist assumptions about language. In this section, I con¬
sider a more functionalist view: “the study of discourse is the study of any
aspect of language use” (Fasold 1990: 65). Another statement of this view of
discourse is Brown and Yule’s (1983: 1):

the analysis of discourse, is necessarily, the analysis of language in use.

As such, it cannot be restricted to the description of linguistic forms
independent of the purposes or functions which these forms are designed
to serve in human affairs.

As these views make clear, the analysis of language use (cf. Saussure’s parole)
cannot be independent of the analysis of the purposes and functions of lan¬
guage in human life. This view reaches an extreme in the work of critical
language scholarship, i.e. the study of language, power, and ideology. Fair-
clough (1989: 23), for example, advocates a dialectical conception of language
and society whereby “language is a part of society; linguistic phenomena are
social phenomena of a special sort, and social phenomena are (in part) lin¬
guistic phenomena” (emphasis in original). In Fairclough s view, language and
society partially constitute one another - such that the analysis of language as
an independent (autonomous) system would be a contradiction in terms (see
also Foucault 1982; Grimshaw 1981). Even in less extreme functionalist views,
however, discourse is assumed to be interdependent with social life, such that
its analysis necessarily intersects with meanings, activities, and systems outside
of itself.
32 The Scope of Discourse Analysis

A definition of discourse as language use is consistent with functionalism in

general: discourse is viewed as a system (a socially and culturally organized
way of speaking) through which particular functions are realized. Although
formal regularities may very well be examined, a functionalist definition of
discourse leads analysts away from the structural basis of such regularities to
focus, instead, on the way patterns of talk are put to use for certain purposes
in particular contexts and/or how they result from the application of com¬
municative strategies. Functionally based approaches tend to draw upon a
variety of methods of analysis, often including not just quantitative methods
drawn from social scientific approaches, but also more humanistically based
interpretive efforts to replicate actors’ own purposes or goals. Not surprising¬
ly, they rely less upon the strictly grammatical characteristics of utterances as
sentences, than upon the way utterances are situated in contexts.
Functional analyses may start from one of two directions, roughly paralleling
the distinction made by linguists and anthropologists (e.g. Pike 1967) between
etic and emic approaches. To take a relatively familiar example, compare
phonetics and phonemics. A phonetic analysis provides all the possible dis¬
tinctions (among means of articulation) irrespective of whether they are ac¬
tually used in a particular language or a particular utterance. A phonemic
analysis, on the other hand, provides only those distinctions that make a
meaningful difference to speakers of a particular language (that produce se¬
mantically different strings of sounds).
Functional analyses of discourse that start from an etic direction delimit the
functions served by a system (such as language or communication) and match
particular units (such as utterances or actions) to those functions. An example
of this kind of functional approach is Goffman’s analysis of system and ritual
conditions. Goffman (1981a: 14-15) proposes that conversational interaction
requires two sets of conditions. System conditions (also called constraints,
because they impose constraints on what can be said and done) center on the
mechanical requirements of talk, e.g. a two-way capability for transmitting
acoustically adequate and readily interpretable messages, feedback capa¬
bilities, contact signals, turnover signals, and so on. Ritual constraints, on the
other hand, center on interpersonal requirements of talk: the management of
oneself and others so as not to violate appropriate standards regarding either
one’s own demeanor or deference for another. Different ways of acting (in¬
cluding, but not limited to, verbal utterances) can serve system and/or ritual
constraints (e.g. Schiffrin 1988b, on turn-initial particles), and when they do
so we can say that they are serving the functional requirements of talk. Note
that, if a particular way of speaking were to have a use undefined by the initial
system, we might still discover it - but, unless we were to argue that we had
missed a function, we would not be able to locate it in our system of functional
requirements.
Many functionally based discourse analyses assume etic schemas more
general than those devised specifically for conversational interaction. One such
framework was proposed by Jakobson (1960). Jakobson differentiates six
 Definitions of Discourse 33
language functions that are identified through the way that utterances (cf.
texts) can be related to different components of the speech situation (cf. con¬
texts). I have adapted Jakobson’s framework below - capitalizing the situ¬
ational component and putting the corresponding function in brackets.

CONTEXT
[referential]

CONTACT
[phatic]

ADDRESSOR MESSAGE ADDRESSEE

[emotive] [poetic] [conative]

CODE ft
[metalinguistic]

Although others have proposed different functions (e.g. Halliday 1973),

Jakobson’s schema most firmly grounds language functions in the speech
situation per se. Note that Jakobson’s view of the speech situation includes
language as just one of the components of a speech situation and as one of
the foci of speech. That is, the basis for a metalinguistic function is the “code”;
the basis for emotive and conative functions are addressor and addressee.
Jakobson also makes the critical point that utterances do not have a single
function: although a particular expression may have a primary function, it is
most typical for it to be used to simultaneously realize different functions. Do
you know the time?, for example, may have a phatic function (it opens
contact), an emotive function (it conveys a need of the addressor), a conative
function (it asks something of the addressee), and a referential function (it
makes reference to the world outside of language).
The second direction available to functional analyses is a more emic direc¬
tion: begin with how particular units (again, utterances, actions) are used and
draw a conclusion about the broader functions of such units from that ana¬
lysis. In other words, one would begin from observation and description of an
utterance itself, and then try to infer from analysis of that utterance and its
context what functions are being served. It is important to note that such
inferences are not totally ad hoc: rather, they can be firmly grounded in
principled schema as to what functions are available. But they do differ from
more etic approaches because they are not as wed to the notion of system, and
because they are more open to the discovery of unanticipated uses of language
(see Hymes 1961).
Let us go on to see how a discourse analysis stemming from the functionalist
view that discourse is language use might actually proceed. We can begin by
returning to (5) from section 3.

(5) jan: (a) Are you free for lunch today?

BARBARA: (b) I have to advise students all day.
34 The Scope of Discourse Analysis

In addition to being identifiable as question and answer, we can recognize

quite easily that the utterances in (5) are being used to realize certain functions,
i.e. to try to accomplish interpersonal goals and to convey social and ex¬
pressive meanings. Although we can suggest such goals and meanings quite
readily, however, we might not be able to verify them without more knowledge
of the context of the exchange, including such information as the relative
status of, and relationship between, the participants, their setting, and their
usual ways of interacting, as well as information about the conventional mean¬
ings of invitations to lunch. (For example, in some groups in American society,
saying Let’s have lunch sometime does not really “count as” an invitation to
lunch.) We might find, for example, that Jan is pursuing a friendship, recipro¬
cating for a prior offer from Barbara, and/or paying deference to Barbara’s
higher status. But in order to decide which (if any, or perhaps all) of these
functions were being realized, we would probably need a good deal of infor¬
mation about the context of the exchange.
We have seen thus far that functional definitions of discourse assume an
interrelationship between language and context. One problem stemming from
this assumption is that it becomes difficult to separate the analysis of discourse
per se from other analyses of language and context - even analyses that may
really belong more in different areas of inquiry. I will illustrate through
another example. (10) is from a sociolinguistic interview (Schiffrin 1987a;
chapters 5 and 8 of this book); (10) also appears in Schiffrin 1988b.

(10) DEBBY: (a) Yeh. Well some people before they go to the doctor, they’ll
talk to a friend, or a neighbor.
(b) Is there anybody that [uh . . .
ZELDA: (c) [Well:: well I guess-
HENRY: (d) [Sometimes it works.
(e) Because there’s this guy Louie Gelman,
(f) he went to a big specialist,
(g) and the guy .. . analyzed it wrong.
(h) In fact his doctor didn’t know.
(i) and the specialist didn’t know.

We can informally characterize this exchange as one in which I am seeking in¬

formation (a, b), and in which Henry tells me a story that provides the
information that Henry thinks I am seeking. (Although space prevents me from
including the entire story in (10), what is included in (e) to (i) is the story
initiation.) The language used during this exchange might then be examined
as to how it serves these functions.
When we focus on different aspects of what is said in (10), however, we find
that there are many different ways to focus upon the language that is used
without saying anything about the functional gloss just noted: a question-
answer exchange in which the answer is a story. Let us focus first on my own
utterances in (a) and (b):
 Definitions of Discourse 35

debby: (a) Yeh. Well some people before they go to the doctor, they’ll talk
to a friend, or a neighbor.
(b) Is there anybody that uh . . .

We are fortunate, in some ways, that we can use my own knowledge (as the
speaker) of what I was trying to accomplish with these utterances. Although
my utterances could have a number of different communicative functions,
what I wanted to find out was who Henry talks to about his problems - to
whom he would complain, turn to for advice, if he didn’t feel well. (As we will
see, Henry acts upon an interpretation different from what I had intended.) I
first describe the general situation in which I was interested (in a), and then
assume that Henry can understand my elliptical question (in b) as asking for
a specific instance of that general situation, i.e,/,whethfer there is anybody that
he (as a member of the larger set of “some people”) would go to.
The function that I have just described is most typically analyzed as part of
speech act theory: the intended communicative force of an utterance can
provide us with a way to categorize it as a particular speech act, e.g. as a
“request”. Performing actions through speech, or, to paraphrase Austin
(1962), doing things with words, is certainly a function of language. The
problem, however, is that speech acts may be (and often are) analyzed at a
sentence level: in other words, an analysis of intended communicative func¬
tions need not say anything at all about discourse.
Let us take another aspect of (10) to illustrate how a focus on language use
can take us in another direction. Although I did not include phonetic detail in
my transcription of (10) above, I will now be more specific about how two
words are pronounced: there, this. In saying there and this in (e), for example,
Henry pronounces the initial consonants not as fricatives, but as affricates.
These possible pronunciations join with a third - a stop - to form what has
been called a sociolinguistic variable (Labov 1972a): an alternative way of
saying the same thing (e.g. pronouncing the same word, producing the same
phoneme), whose variants are used differently by people depending on who
they are (e.g. social class, gender, age, ethnicity), where they are (setting), and
how they are speaking (e.g. carefully, casually). Thus, given the particular
distribution of these variants in the speech community in which Henry lives,
we might interpret Henry’s use of the affricate as a marker (Scherer and Giles
1979) of his social identity (male, lower middle class, who is speaking rela¬
tively casually). Put another way, we could say that the affricates have the
social and interactional functions of displaying Henry as a particular kind of
person engaging in a particular kind of interchange. Thus, these pronunci¬
ations have a great deal to do with the social identity and style of the speaker
as someone who exists (and presents himself) within a particular constellation
of social and cultural meanings — with contextual inferences resulting from
what Henry says (e.g. chapter 4). However, they have little to do with the way
what is said is used to fill the other functions I noted earlier - answering a
question and telling a story.
36 The Scope of Discourse Analysis

I have been suggesting thus far that analyses stemming from a view of
discourse as language use can be too inclusive: they can include sentence-sized
units and phonological variation. I will take one other aspect of (10) to illus¬
trate another way that such analyses can be too inclusive. Recall that, in (10),
Henry complies with my request for information by telling a story about his
friend Louie Gelman. Note that this is not the compliance that I had intended:
rather than telling me who he talks to, Henry provides support for the general
situation that I have described, i.e. he illustrates why talking to one’s friends
can be a better solution to medical problems than seeking professional help.
Clauses (e), (f), and (g) focus on how Henry introduces some of the characters
in his story (see chapters 6, 7, and 8).

henry: (d) Sometimes it works.

(e) Because there’s this guy Louie Gelman,
(f) he went to a big specialist,
(g) and the guy . . . analyzed it wrong.

Note, first, that this guy Louie Gelman is introduced as the object of an
existential predicate “there is.” This structure places a new referent (a referent
about whom Henry cannot expect me to know anything) at the end of a clause,
and predicates very little about the new referent except that he “exists.”
Crucial for our purposes here is that this information structure serves a
communicative function. Of course this is not the function discussed earlier
as “intended communicative force” or “display of social identity.” Rather, we
might say that this structure eases the transmission of information, i.e. it eases
the hearer’s understanding of a new referent.
The order of words in (f) and (g) serves a similar function: it eases the
hearer’s processing of information by placing a new entity in the context of
already assumable information. I show this below:

(e) Because there’s this guy Louie Gelman [NEW],

(f) he [OLD = Louie Gelman] went to a big specialist [NEW],
(g) and the guy [OLD = a big specialist] . . . analyzed it wrong.

We have already seen that in (e), Louie Gelman is new information in clause
final position. “Louie Gelman” is next mentioned as he (in f) in clause initial
position. This old information is then a background for the introduction of
another new referent, “a big specialist”, at the end of the clause. When “a big
specialist” becomes old information, the referring term is definite and less
specific (the guy) and in clause initial position. Thus, the old/new information
order within a clause serves a communicative function, i.e. it eases the hearer’s
understanding of new information by placing it in the context of familiar
information. Again, although this certainly contributes to our understanding
of the functional design of word order within sentences (in fact, it allows us
to view this as a text-sentence; see pp. 27-8), it has little to do with our
 Definitions of Discourse 37

intuitive gloss of the exchange as a question-answer sequence or with these

clauses as a story initiation.
Before we go on, it is important to note that if we adopt a definition of
discourse that is purely functional (e.g. analyze what Fairclough (1989) calls
discursive practices), then what I have just described would certainly fall
within the purview of discourse analysis. But is analysis of the phenomena
noted above - pronunciation, speech acts, word order - really the analysis of
how what Henry says serves the communicative functions I noted earlier -
telling a story to provide an answer to a question? Such analyses seem, instead,
to take us into areas of inquiry that have little to do with the functions just
noted. My point here, then, is that although there are things we can say about
(10) as language use, they do not all contribute to the function of Henry’s
contribution as an “answer” or a “story.” Y^t “giving an answer” and “be¬
ginning a story” seem to be functional glosses that contribute to our under¬
standing of the communicative content of the utterances in (10): they give
Henry’s utterances an internal unity (grouping the utterances in one “turn”
together) as well as an identity through which they are related to what I said
initially in the exchange (they connect “turns” across speakers). In brief, these
functions help us understand relationships “across,” not “within,” utterances.
There are some features and qualities of (10) that seem to bear more heavily
on the way different utterances are related to one another. Observe first that
Henry’s story describes how different medical professionals wrongly diagnosed
Louie until a neighbor was finally able to tell him what was really wrong; also,
at the end of the story, Henry summarizes his point by saying that doctors can
make mistakes. With this information about Henry’s story in mind, let us
re-examine the following section of (10):

(f) he went to a big specialist,

(g) and the guy . . . analyzed it wrong.
(h) In fact his doctor didn’t know,
(i) and the specialist didn’t know.

Several aspects of clauses (f) to (i) (the introduction to the story) prefigure the
actions in the story, and the point being established through the story and
made explicit through its summary. First is the contrastive stress on BIG
specialist and analyzed it WRONG. This establishes a frame in which author¬
ity can be seen as wrong (Henry later states that Doctors are not God!) and
thus makes Henry’s point more salient.
Note, also, that Henry modifies a basic rule of narrative ordering (Labov
1972b): he does not always report events in the order in which they occurred.
People typically visit their own doctors before consulting with a specialist. In
(f) and (g), however, Henry reports his friend’s encounter with the specialist
and then mentions an encounter with his doctor (h). This deviation from
temporal ordering is marked with in fact in (h): in fact his doctor didn t know.
Modifying the reported order of events in this way has the discourse function
38 The Scope of Discourse Analysis

of helping Henry enlarge the size or membership of the group of professionals

who were wrong - thus adding to the overall point of his story.
Finally, the way Zelda begins to answer my question also hints at the overall
point of Henry’s story. I present this section again.

debby: (a) Yeh. Well some people before they go to the doctor, they’ll talk
to a friend, or a neighbor.
(b) Is there anybody that [uh . . .
zelda: (c) [Well: : well I guess-

Zelda’s Well:: had a high pitch and a final rise - an intonation often associ¬
ated with uncertainty and doubt (Ward and Hirschberg 1988). We might
say that this, too, has a discourse function: it conveys the attitudinal frame
through which Zelda treats my proposition about whom people talk to when
they’re sick, and perhaps even Zelda’s doubt about the wisdom of medical
professionals.
As I have just illustrated, there are aspects of (10) that are relevant to key
functions of the utterances (answering a question, opening a story) - functions
that help utterances “fit” with other utterances. If we define discourse just as
language use, however, there may be no way to legitimately separate questions
about the functions that help us understand relationships across utterances
from questions about the functions that are realized within utterances (e.g.
speech acts, word order, phonological variation). Although this may be seen
as a positive outcome if we are searching for ways to unify functionalism in
general, it ends up removing a certain degree of autonomy from discourse
studies as a particular branch of linguistics. Interestingly, it is this very syn¬
thesis that Tannen (1989a: 6) finds appealing about the definition of discourse
as language beyond the sentence:

Discourse - language beyond the sentence - is simply language - as it

occurs, in any context (including the context of linguistic analysis), in
any form (including two made-up sentences in sequence; a tape recorded
conversation, meeting, or interview; a novel or play). The name for the
field “discourse analysis,” then, says nothing more or other than the term
“linguistics”: the study of language.

Despite the value of so integrative an approach, I believe that efforts to

define discourse analysis as the study of language itself (as does Tannen) or
to define discourse analysis as the study of language functions are more appro¬
priate at later stages of theory building (e.g. when one wants to articulate a
theory of language that accommodates discourse analysis) than at early stages
of discourse analysis (e.g. when one wants to define what it is that one wants
to study).
We have seen in this section that defining discourse as language use depends
upon broader assumptions about the relevance of language to meanings,
 Definitions of Discourse 39
activities, and systems outside of itself. A corollary of this definition is that
functionally based approaches view discourse as a socially and culturally
organized way of speaking. Those functions are not limited to tasks that can
be accomplished by language alone; rather they can include tasks such as
maintaining interaction or building social relationships. Thus, functional ana¬
lyses focus on how people use language to different ends: they are typically
concerned less with the way people intend what they say to serve referential
meanings (to convey propositional information), and more with the unin¬
tended social, cultural, and expressive meanings stemming from how their
utterances are situated in contexts.
Although this inclusive view of the scope of discourse analysis is critical to
the articulation of a linguistic theory within which discourse can fit, it also
has the untoward effect of threatening to subrperge discourse analysis within
broader and more general analyses of language functions, without leaving a
space within which discourse analysts can formulate a clear set of principles,
goals, topics, and methods specific to their own enterprise. It makes no
provision for analysis of the way the communicative content of an utterance
contributes to our understanding of relationships across utterances, or, altern¬
atively, for the way relationships across utterances help us understand the
form, function, or meaning of a single utterance. Put most simply, it fails to
make a special place for the analysis of relationships between utterances.
Instead, a functionalist definition of discourse includes within its scope all
language use: it provides no way to define discourse as different from other
levels of language use (e.g. the use of sounds, words, or sentences). What we
need to capture in a definition of discourse is the idea that discourse analysis
imposes its own set of phenomena, its own problems and puzzles — and can
discover its own regularities — in addition to those that it “inherits from
lower-level parts of discourse and those based in the way language is a social
practice “determined by social structures” (Fairclough 1989: 17).

5 Discourse: utterances

In this section, I consider another definition of discourse: discourse is utter¬

ances. This view captures the idea that discourse is above (larger than) other
units of language; however, by saying that utterance (rather than sentence) is
the smaller unit of which discourse is comprised, we can suggest that discourse
arises not as a collection of decontextualized units of language structure, but
as a collection of inherently contextualized units of language use.
The main problem with this definition is that the notion of “utterance” is
not really all that clear. For many linguists, utterances are contextualized
sentences, i.e. they are context bound (as well as text bound). Hurford and
Heasley (1983: 15), for example, make the following distinction:
40 The Scope of Discourse Analysis

A sentence is neither a physical event nor a physical object. It is conceived

abstractly, a string of words put together by the grammatical rules of a
language. A sentence can be thought of as the ideal string of words
behind various realizations in utterances and inscriptions.

In Hurford and Heasley’s framework, “Discourse analysis is fun” in (11a),

(lib), and (11c) would all be considered one sentence, but different utteran¬
ces:

(11a) Discourse analysis is fun.

(lib) Discourse analysis is fun.
(11c) bob: I really like my linguistics courses.
sue: Oh I do too! Discourse analysis is fun.

The three occurrences of “Discourse analysis is fun” are three different utter¬
ances - three different instantiations of a single sentence. (11a) and (lib) are
realizations of the sentence that share many aspects of a context: they both
occur on the same page of the same book, and I am using them merely to
make a point about utterances, without providing any more information about
who, where, when, or how they may be found outside of this text. Despite
this similarity, (11a) and (lib) are two different utterances: (11a) is the first
appearance of the sentence, and (lib) is the second appearance (a repetition)
of the sentence. What this means is a difference in the textual environment of
(11a) and (lib). Finally, (11a) and (lib) both differ from (11c): although
(11c) is also an illustration in a book, I have provided for it a hypothetical
interactive context between two hypothetical interlocutors.
Before I go on to a problem with the view that discourse is “utterances”
(stemming from the sentence-utterance distinction just illustrated), let me
note the advantages of this view. First, defining discourse as utterances forces
us to attend to the contextualization of language structure in a way going
beyond Lyons’s notion of text-sentence (pp. 27-8) to what we might call
(following Lyons) context-sentence. Second, since this definition demands
attention to more than one utterance, extended patterns and sequential ar¬
rangements automatically come under examination. Thus, defining discourse
as utterances seems to balance both the functional emphasis on how language
is used in context and the formal emphasis on extended patterns.
The main problem with this view of discourse is the definition proposed
above of utterances - as realizations of sentences. Some linguists propose that
sentences and utterances are radically different from each other, e.g. Fasold’s
(1990) belief that utterances need have no grammatical backing at all (they
may or may not conform to grammatical principles) and sentences are abstract
objects that may never actually “happen” or be realized. Others reverse the
sentence-to-utterance mapping relationship to propose that sentences are “de-
contextualized” utterances. Figueroa (1990: 284), for example, claims the
following: “all human actions take place within a particular spatio-temporal
 Definitions of Discourse 41

context; therefore whether one is introspecting about a sentence, whether one

is reading a sentence, whether one is speaking, one is performing utterances.”
To extrapolate a bit from Figueroa, sentences are as contextualized as any¬
thing else that is the object of attention and intention; what complicates their
consideration as utterances is that the context of a sentence is that of meta¬
linguistic scientific discourse (see also Goffman 1981b). Thus, “the question
for linguistic theory is . . . whether utterances are decontextualized either into
an utterance type or sentence ... or whether utterances are left contextualized”
(Figueroa 1990: 284).
Regardless of these difficulties, the view that I will take in this book is that
discourse can best be thought of as “utterances.” I will view utterances as
units of language production (whether spoken or written) that are inherently
contextualized; whether (or how) they are related to^sentences (or, in fact, to
other units such as propositions, turns, or tone units) is an issue that will not
explicitly enter into our discussion. A definition of discourse as “utterances”
implies several goals of discourse analysis that underlie much of what follows
in this book. First is what we might call syntactic goals, or more appropriately
for discourse analysis, sequential goals: are there principles underlying the
order in which one utterance, or one type of utterance, follows another?
Second is what might be called semantic and pragmatic goals: how does the
organization of discourse, and the meaning and use of particular expressions
and constructions within certain contexts, allow people to convey and inter¬
pret the communicative content of what is said? how does one utterance (and
the sequential relationship between utterances) influence the communicative
content of another? Consistent with the differences between formalist and
functionalist paradigms discussed earlier, the approaches to discourse analysis
to be compared vary in terms of how deeply they pursue problems of sequen¬
tial structure, and in terms of their willingness to delve into interpretations of
meaning and use.

6 Summary: definitions, issues, and discourse

analysis

I began this chapter with a brief description of two different paradigms

underlying our conception of language (section 2). After comparing two dif¬
ferent definitions of discourse stemming from these two paradigms - discourse
as language above the sentence or clause (section 3), discourse as language use
(section 4) — I proposed a third definition that sits at the intersection of
structure and function — discourse as utterances (section 5).
The definitions of discourse that we discussed raise important issues that
will help us understand the similarities and differences among the approaches
42 The Scope of Discourse Analysis

to be presented in the rest of the book. Although I presented the structural

(formalist) definition of discourse as separate from the functionalist one, I will
suggest later in this book that actual analyses of discourse reveal an inter¬
dependence between structure and function (chapter 9). The distinction be¬
tween structure and function also bears on two other issues that I discuss later.
One is the relationship between text and context: structural definitions focus
upon text and functional definitions upon context (chapter 10). Another is the
way linguists view communication: structural definitions take a narrower view
of communication than do functional definitions, and place a higher priority
on the role of the code (cf. text) in communication (chapter 11). Although I
discuss all three of these issues in general terms in relation to different ap¬
proaches to discourse in part III, the sample analyses in chapters 3-8 (part II)
will also show how different kinds of empirical analyses provide a basis for
understanding these theoretically important issues.
Before closing this chapter, I want to make a slightly different point. I noted
at the outset of chapter 1 that discourse analysis is one of the most vast, but
also least defined, areas in linguistics. The availability of two different per¬
spectives - stemming from two different ways of defining discourse - is
partially responsible for the tremendous scope of discourse analysis. If we
focus on structure, our task is to identify and analyze constituents, determine
procedures for assigning to utterances a constituent status, discover regu¬
larities underlying combinations of constituents (perhaps even formulating
rules for producing those regularities), and make principled decisions about
whether or not particular arrangements are well formed. If we focus on
function, on the other hand, our task is to identify and analyze actions
performed by people for certain purposes, interpret social, cultural, and per¬
sonal meanings, and justify our interpretations of those meanings for the
participants involved. Dealing with either structure or function alone is thus
a hefty task: but dealing with both can take us into two different analytical
worlds that are often difficult to integrate.
The need to consider both text and context also increases the scope of
discourse analysis. One obvious reason is that context can be tremendously
broad and defined in different ways, e.g. mutual knowledge, social situations,
speaker-hearer identities, cultural constructs. Another reason is that the text-
context relationship is not independent of other relationships often assumed
to hold between language and context (context as “culture,” “society,” or
“interaction”). Yet, references to language “in” (or “and”) context are far
from neutral descriptions. Rather, any description of the language-context
relationship veils tacit assumptions about the relationship between two differ¬
ent symbol systems and two different structures, e.g. their relative autonomy,
the precedence of one over another, the way they impinge upon one another.
Speaking of language in society, for example, assumes that language is a
system both smaller than, and dependent upon, a broader matrix of social
interactions and structures - thus making assumptions not only about how
two systems impinge upon one another, but also about the very nature of those
 Definitions of Discourse 43

two systems. Such assumptions bear not only on our understanding of dis¬
course, but on our understanding of language in general.

Notes

1 Compare Hopper’s (1988) distinction between the A Priori Grammar Postulate and
the Emergence of Grammar attitude.
2 I am by no means attempting a comprehensive review of all formal and functional
approaches (a task that would require its own book). Rather, I am trying to take
one or two typical (often classic) examples of each approach in order to show how
they differ. Nor am I claiming that the characteristics that I present are shared equally
by all representatives of structural or functional approaches. One criterion differen-
. tiating functionalists, for example, is the type of external function deemed important:
in addition to (or as alternative to) a concern for social function (e.g. Halliday) and
communicative function (e.g. Givon) is a concern for cognitive and perceptual func¬
tion (e.g. Bates and MacWhinney, Kuno). Although these functions are associated
with different perspectives (e.g. Halliday with sociolinguistics and Bates and Mac¬
Whinney with psycholinguistics) - and the influence they exert may be markedly
different - what the perspectives share is an emphasis on the influence of factors
outside of language on linguistic processes and structures. Thus, what I claim through
the following discussion is that the assumptions made by formal and functional
approaches are different: despite variation within structural approaches and within
functional approaches, there is nevertheless greater variation between (than within)
these two approaches.
3 The view that discourse is similar in kind to lower-level linguistic units leads to a
tendency to view only grammatical units (sentences, clauses) as the building blocks
of discourse, i.e. the hierarchy is comprised of units of “the same kind.” In this sense,
the view that discourse is propositionally structured can thus be quite different in
some ways. Some analysts (e.g. Halliday and Hasan 1976) have proposed that the
task of conveying meanings (semantics) is so different from the task of building
sentences (syntax) that discourse is fundamentally different in kind than sentences.
(Interestingly, however, Halliday and Hasan (1976: 2) still incorporate sentences into
their view of text: although “a text does not consist of sentences; it is realized by,
or encoded in, sentences.”)
4 In addition to saying that a linguistic structure eases the transmission of information,
we might also say that a structure in discourse (i.e. a sequence of referring terms in
which a first-mention is explicit and indefinite, and a next-mention is inexplicit and
definite) emerges because of constraints on the sequential flow of information: if a
first-mention provides explicit information, a next-mention can provide considerably
less information. (See discussion in chapter 5.)
■
Part II Approaches to Discourse
Analysis
This part of the book discusses six different approaches to discourse analysis.
None of the approaches takes as its sole object of inquiry “discourse” per se,
and the broader topics to which discourse is seen as relevant have import¬
ant influences on the way in which they approach the analysis of utterances.
Speech act theory (chapter 3) focuses on communicative acts performed
through speech. Interactional sociolinguistics (chapter 4) focuses on the social
and linguistic meanings created during interaction. The ethnography of com¬
munication (chapter 5) focuses on language and communication as cultural
behavior. Pragmatics (chapter 6) focuses on the meaning of individual utter¬
ances in hypothetical contexts. Conversation analysis (chapter 7) focuses on
how sequential structures in conversation provide a basis through which social
order is constructed. Variation theory (chapter 8) focuses on structural cat¬
egories in texts and how form and meaning in clauses help to define text.
Each chapter in this part of the book provides a general background (critical
concepts and key issues) of the approach, focusing primarily on the work of
scholars central to the development of the approach. Each chapter also pro¬
vides an extensive sample analysis, using data drawn primarily from socio-
linguistic interviews (see appendix 1), to illustrate the methodology of each
approach and its application to one of two concrete problems (question-
answer sequences in chapters 3, 4, and 5; referring expressions in chapters 6,
7, and 8).
■
3 Speech Act Theory

A in

1 Introduction

Two philosophers, John Austin and John Searle, developed speech act theory
from the basic belief that language is used to perform actions: thus, its fun¬
damental insights focus on how meaning and action are related to language.
Although speech act theory was not first developed as a means of analyzing
discourse, some of its basic insights have been used by many scholars (e.g.
Labov and Fanshel 1977; see also chapter 8) to help solve problems basic to
discourse analysis.' In addition, particular issues in speech act theory lead to
discourse analysis, e.g. how an utterance can perform more than one speech
act at a time, and the relationship between context and illocutionary force.
I begin in section 2 with an overview of the critical concepts and ideas
introduced by both Austin (1962) and Searle (1969). I then use a sample
discourse to discuss two different (but related) stages in the application of
speech act theory to discourse analysis: how to identify utterances as speech
acts (section 3.1); how to analyze sequences of speech acts (section 3.2). After
exploring some of the issues raised by this analysis (section 3.3), I summarize
the speech act approach to discourse (section 4).

2 Defining speech act theory

Speech act theory begins with the work of John Austin (section 2.1), whose
ideas are expanded and incorporated into linguistic theory by John Searle
(section 2.2). Searle’s work also raises important questions concerning the
inventory (and classification) of acts about which people know (section 2.3)
and the way that a single utterance can be associated with more than one act
50 Approaches to Discourse Analysis

(section 2.4). Although it is not initially proposed as a framework in which

to analyze discourse, the issues with which speech act theory is concerned
(meaning, use, actions) can lead to such an analysis (section 2.5).

2.1 Austin: from performative to illocutionary act

A series of lectures by John Austin in 1955, compiled in How to Do Things

with Words (1962), is widely acknowledged as the first presentation of what
has come to be called speech act theory. Austin’s presentation seems inten¬
tionally argumentative and provocative: distinctions are proposed in the first
few chapters that are then systematically dismantled in later chapters, such
that the presentation of the theory by the end of the book is dramatically
different from its presentation in the beginning of the book.
Austin begins by noticing that some utterances that seem like statements lack
what is thought to be a necessary property of statements - a truth value. Not
only do such statements not “ ‘describe’ or ‘report’ ” anything (p. 5), but “the
uttering of the sentence is, or is a part of, the doing of an action, which again
would not normally be described as, or as ‘just’, saying something” (p. 5;
emphasis in original). Austin calls these performatives and distinguishes them
from constatives, i.e. declarative statements whose truth or falsity can be
judged. The following are examples of sentences (or utterances - the terms are
initially interchangeable) that are performatives.

I do (take this woman to be my lawful wedded wife) - as uttered in the

course of the marriage ceremony.

I name this ship the Queen Elizabeth - as uttered when smashing the
bottle against the stern.

I give and bequeath my watch to my brother - as occurring in a will.

I bet you sixpence it will rain tomorrow.

The examples given all share several qualities. They all include a particular
type of verb - a performative verb - that realizes a particular action (the action
that the verb “names”) when uttered in a specific context. Such a context can
include setting (a marriage ceremony, writing a will), physical objects (a ship,
legal documents), and institutional identities; it may also require a particular
response (a bet requires what Austin calls “uptake”). Performatives require
not only “the appropriate circumstances” (p. 13), but also the appropriate
language: the performative verb in the above examples is in the present tense,
each sentence has a first person subject, and the adverb hereby may modify
any of the verbs (e.g. one can say “I hereby give . . .”). Thus, performatives
meet certain contextual and textual conditions.
Austin goes on to classify the circumstances (the conditions) that allow
utterances to act as performatives. He does so according to the circumstances
 Speech Act Theory 51
themselves, and according to the consequence for a performative if the cir¬
cumstance does not hold. As noted above, the circumstances allowing an act
are varied: they include the existence of “an accepted conventional procedure
having a certain conventional effect” (p. 26), the presence of “particular
persons and circumstances” (p. 34), “the correct and complete execution of a
procedure” (p. 36), and (when appropriate to the act) “certain thoughts,
feelings, or intentions” (p. 39). An act can either misfire (not go through at
all) or go through but, due to an abuse of the procedure, in a way that is not
totally satisfactory. Appointing someone to office (i.e. saying “I appoint you”)
misfires if that person has already been appointed or if the speaker is not in
the position to appoint anyone (p. 34). Likewise, saying “I bet you the race
won’t be run today” if more than one race was arranged (p. 36) causes the
act of betting to misfire. So, too, saying “I bet you sixpence” will misfire and
not be a “bet” unless you say “I take you on’V'or preside other words to that
effect (p. 37). As noted above, other violations allow an act to go through,
but in an “unhappy” or “infelicitous” (p. 14) way. Saying “I advise you to
[do X] ” without the requisite thought (i.e. if I do not think X is the course most
expedient for you; p. 40), for example, does not void the performative of
advice-giving: the act is still “advising” but it is infelicitous because one of
the procedures has been abused. Thus, there is variation in both the circum¬
stances allowing a felicitous performative and the way a performative can go
wrong.
I noted above that Austin proposes distinctions in the first few lectures/chap¬
ters that are systematically dismantled in later sections. The distinction be¬
tween constatives and performatives is one such distinction. Recall that
constatives are declaratives whose truth could be judged; performatives are
declaratives that “do” an action. By the end of the book, Austin proposes
instead that all utterances have qualities that were initially seen as charac¬
teristic of constatives and performatives. The focus of attention is no longer
sentences, but “the issuing of an utterance in a speech situation” (p. 139).
All utterances perform speech acts that are comprised of a locutionary act (the
production of sounds and words with meanings), an illocutionary act (the
issuing of an utterance with conventional communicative force achieved “in
saying”), and a perlocutionary act (the actual effect achieved siby saying”).
Before discussing these three aspects of a speech act, I want to briefly go
over the way the constative-performative distinction is collapsed. Basically,
what Austin shows is that the conditions defining one type of utterance apply
equally well to the other, and that neither type can be differentiated by formal
clues. Thus, the dismantling of the constative-performative distinction can
help to reveal Austin’s view of two aspects of the conditions underlying speech
acts: context (what makes an utterance “true” and “appropriate”) and text
(how what is said conveys what is done).
Consider the distinction between truth/falsity (applicable to constatives) and
felicitous/infelicitous (applicable to performatives). Austin argues that perfor¬
matives (as well as constatives) involve judgements of truth and falsity. Recall
52 Approaches to Discourse Analysis

that certain conditions are necessary for a performative to be felicitous. These

conditions are like statements that have to be true: saying that “certain things
have to be so . . . commits us to saying that for a certain performative utter¬
ance to be happy, certain statements have to be true” (p. 45; emphasis in
original). “I apologize,” for example, implies the truth of certain conditions,
e.g. it is true (and not false) that I am committed to doing something sub¬
sequently (p. 46). Similarly, implications thought to hold only for constatives
hold for performatives as well. The entailment of one proposition (e.g. “The
mat is under the cat”) by another (e.g. “The cat is on the mat”), for example,
is parallel to the way an underlying condition (“I ought”) is entailed by a
performative (e.g. “I promise”). Similarly, saying “I promise but I ought not”
is as much of a contradiction as saying “it is and it is not” (p. 51). Thus, the
conditions necessary for performatives to be felicitous have the same truth-
based relationship with a performative utterance that constative statements
have with other statements.
Just as performatives can be said to have truth conditions, so constatives
may be said to meet felicity conditions. Take a statement that refers to
something that does not exist: “The present King of France is bald.” Austin
argues that this is similar to someone purporting to bequeath an estate he does
not own: both are void (i.e. they “misfire”), simply because both purport to
refer to an entity (the King of France, an estate) that does not exist (pp. 20,
137). Constatives are also subject to the same specific kinds of infelicities that
result in abuses of performatives. Recall that certain thoughts, feelings, or
intentions may be part of the circumstances that allow a felicitous performat¬
ive - such that an insincere “promise” would abuse the procedure of promis¬
ing. Statements are also liable to sincerity abuses: saying “The cat is on the
mat” if I do not believe that the cat is on the mat abuses a procedure in the
same way as saying “I promise to be there,” but not really intending to
be there (p. 136).
We noted initially that performatives meet certain contextual and textual
conditions. Now we have seen that the contextual conditions for performatives
are not different in kind from those for constatives: both involve truth and
falsity, both involve felicity and infelicity. It turns out that the textual condi¬
tions are not as different as initially suggested either. Although Austin initially
grounded his analysis of performatives with a discussion of performative verbs
(and he maintains this focus, to a certain degree, in his taxonomy; see section
2.3), he also raises the possibility that performatives can be realized without
verbs (contrasting an explicit performative to a primary performative) and that
not all types of performatives need verbs specialized to that task.
Let us briefly consider the distinction between explicit performatives (with
the verb) and primary performatives (without the verb). Austin suggests that
although performative verbs are neither necessary nor sufficient textual con¬
ditions for performative utterances, they do make explicit certain features of
the speech situation (e.g. the actor, the action being undertaken). One outcome
of this is that primary performatives may be ambiguous: saying “It is yours,”
 Speech Act Theory 53
for example, may be either an act of bequeathing (“I give it to you”) or an
acknowledgment that “it (already) belongs to you” (p. 62). A way to get
around this problem is to suggest that primary performatives should be “re¬
ducible or expandible, or analysable ... or reproducible” (pp. 61-2) to a
performative formula (e.g. I hereby [verb] you that. ..).
Relying on the availability of a performative formula, however, does not
solve all the problems of finding grammatical criteria (i.e. textual conditions)
for performatives. The criteria themselves may be individually problematic:
the use of the present tense, for example, need not always convey an action
concurrent with the time of speaking (pp. 64-5). Nor can we be sure of the
equivalence of primary and explicit performatives: “I am sorry” may not really
be exactly like “I apologize” (p. 66). Finally, we may be able to do something
through speech (e.g. insult), yet not have a performative verb by which to do
it (pp. 65-6). t*
We have seen thus far that the constative-performative distinction cannot
be maintained because both constatives and performatives involve truth and
falsity; both are felicitous or infelicitous in relation to the conditions in
which they occur; both are realized through a variety of forms that can be
rewritten in terms of a performative formula. To put this more generally, we
cannot find either contextual or textual conditions that support the constat¬
ive-performative distinction. Thus, “to state is every bit as much to perform
an illocutionary act as, say, to warn or pronounce” (p. 134). “Stating” is
exactly on a par with “betting”: “I state that he did not do it” is as much of
an act as “I bet that he did not do it.” Despite the collapse of the constative-
performative distinction, performative verbs still play a key role in Austin’s
framework. Utterances whose act is not linguistically explicit (e.g. “He did
not do it”) may be made explicit through the use of a formula built upon a
performative verb (“I state that. . .”, “I argue that. . .”, “I bet that. . .”). So
pervasive is the resource for action provided by performative verbs that Austin
suggests the existence of acts (and verbs) “of the order of the third power of
10” (p. 150) that can be discovered by “going through the dictionary” (p.
150). And so important is the speech situation (the context) in which these
verbs are used (and acts are performed) that even the truth and falsity of
statements is contextually bound: “in the case of stating truly or falsely, just
as much as in the case of advising well or badly, the intents and purposes of
the utterance and its context are important; what is judged true in a school
book may not be so judged in a work of historical research” (p. 143).
I noted above that Austin segments the speech act itself into component acts
- only one of which is the “act” typically spoken of in speech act theory. Three
acts underlie the issuing of an utterance. A locutionary act involves the utter¬
ing of an expression with sense and reference, i.e. using sounds and words
with meaning. This seems to capture the properties of the original constative
group: the act “of saying something” (p. 100). An illocutionary act is the act
performed “in saying” the locution (p. 99), such that what was said had the
force (not the meaning) of that illocution. This level captures the acts initially
54 Approaches to Discourse Analysis

viewed as performative: these acts are conventional in that they could be made
explicit by a performative formula (p. 103). A perlocutionary act is the “con¬
sequential effects” (p. 102) of an utterance on an interlocutor, i.e. what is
achieved “by saying” something. Since these three aspects of an utterance
are all actions, they are all subject to the same kinds of failures - to “the ills
that all action is heir to” (p. 105). Together, these acts produce a total speech
act that must be studied in the total speech situation (pp. 52, 148): “the
words used are to some extent to be ‘explained’ by the ‘context’ in which they
are designed to be or have actually been spoken in a linguistic interchange”
(p. 100).

2.2 Searle: from conditions to rules

Searle’s (1969) Speech Acts builds upon Austin’s work to propose a syste¬
matic framework by which to incorporate speech acts into linguistic theory.
Searle also introduces several ideas that provide important ideas for the ap¬
plication of speech act theory to discourse (section 2.3 and 2.4), although he
resists the idea that conversation is governed by constitutive rules (Searle
1989).
Searle (1969: 21) proposes that “the speech act is the basic unit of communi¬
cation.” Far from divorcing speech acts from the study of language, how¬
ever, this view places speech acts at the very crux of the study of language,
meaning, and communication; in fact, speech act rules are argued to be part
of linguistic competence (see below). What allows the integration of speech
act theory into linguistic theory is Searle’s principle of expressibility (pp. 18-
21): what can be meant can be said. This principle establishes that it is
possible (in theory) for a speaker to come to be able to say exactly what she
means either by increasing her knowledge of the language or by enriching the
language (p. 19). Furthermore, all languages “can be regarded as different
conventional realizations of the same underlying rules” (p. 39), comparable
to the way a chess game is played in different countries, but still considered
a game of chess. The principle of expressibility has several different conse¬
quences. Taken broadly, this principle moves nonliteral meaning, vagueness,
ambiguity, and incompleteness out of the theoretical essence of linguistic
communication (p. 20; cf. Gricean pragmatics; also chapter 11). Taken more
narrowly, this principle has the consequence noted above: it brings together
the study of speech acts, meaning, language, and communication. In Searle’s
own words, the principle of expressibility enables us to

equate rules for performing speech acts with rules for uttering certain
linguistic elements, since for any possible speech act there is a possible
linguistic element the meaning of which (given the context of the utter¬
ance) is sufficient to determine that its literal utterance is a performance
of precisely that speech act. (pp. 20-1)
 Speech Act Theory 55
Thus, viewing speech acts as the basic unit of communication allows Searle to
explicitly associate speech acts with the study of language (its production, its
interpretation) and meaning (both speaker meaning and linguistic meaning):
“there are a series of analytic connections between the notion of speech acts,
what the speaker means, what the sentence (or other linguistic element) uttered
means, what the speaker intends, what the hearer understands, and what the
rules governing the linguistic elements are” (p. 21).
I noted above that speech act rules are part of linguistic competence: lan¬
guage can be used for speech acts because people share rules that create the
acts that say what is meant. Before describing how rules are responsible for
the creation of acts (and how they incorporate both textual and contextual
conditions), it is important to comment on the importance of rules within
Searle’s framework and on the type of rules important for speech acts.
.Searle observes that “speaking a language is Engaging in a (highly complex)
rule-governed form of behavior” (p. 12). A methodological consequence of
this is that linguistic characterizations do not report “the behavior of a group.”
Rather, they describe aspects of speakers’ mastery of a rule-governed skill
(p. 12) that can be obtained by relying heavily on the intuitions (and linguistic
characterizations) of native speakers (p. 15). What such intuitions can provide
are “idealized models” (p. 56) of the conditions that are necessary and suffi¬
cient for the utterance of a given sentence to be a successful, non-defective
performance of a given act. The rules that are responsible for speech acts,
however, are a special type of rule that Searle calls constitutive. In contrast to
regulative rules (that regulate independently existing forms of behavior) con¬
stitutive rules “create or define new forms of behavior” (p. 33). The forms of
the two types of rules reflect their different status: regulative rules are ex¬
pressed as (or can be paraphrased as) imperatives, but constitutive rules are
more definitional, e.g. “X counts as Y in context C” (p. 35).
Our discussion of Austin mentioned contextual conditions (“circumstances”)
and textual conditions (e.g. the availability of explicit performative formula)
that allow an utterance to perform a certain illocutionary act. Like Austin,
Searle’s rules and conditions for speech acts draw upon both context and text:
they also elevate intentions and other psychological states as conditions en¬
abling a speech act, by assigning them their own type of rule (see below). Like
Austin, Searle classifies conditions and rules according to their necessity for
the act. But in contrast to Austin, Searle classifies different kinds of conditions
(and rules) according to what aspect of text and context is focused upon in
the condition or rule; the different conditions also overlap (partially) with the
different components of a speech act.
Searle segments utterances into speech acts very similar to those proposed by
Austin. The uttering of words (morphemes and sentences) is an utterance act.
Referring and predicating are propositional acts. Acts like stating, questioning,
commanding, and promising are illocutionary acts. Illocutionary acts are what
is constituted by the rules noted above: in addition to being rule-governed, they
are intentional, they have a name, and they are what the speaker (S) is doing,
56 Approaches to Discourse Analysis

in relation to the hearer (H), with words. The consequences of illocutionary

acts (the effects on actions, thoughts, beliefs of hearers) are perlocutionary acts.
As just noted, it is the illocutionary act that is subject to the conditions and
rules so central to Searle’s framework. The textual and contextual circumstan¬
ces that allow speech acts to have an illocutionary force are categorized as
different kinds of conditions; rules are extracted from the conditions. Because
our sample analysis presents conditions for questions and requests (section 3),
I will not present all the conditions for any one particular speech act here
(Searle 1969: 66-7, presents the rules for requests, assertions, questions,
thanks, advising, warnings, greetings, and congratulating). Rather, I will dis¬
cuss the conditions in general terms, and in relation to the rules extracted from
the conditions.
Propositional content conditions or rules are the most textual: they concern
reference and predication (the propositional act). A propositional content rule
for promises, for example, is the predication of a future act (A) by the speaker.
Preparatory conditions or rules are varied: they seem to involve background
circumstances and knowledge about S and H that must hold prior to (and may
then be altered by) the performance of the act. A preparatory condition for
promises, for example, concerns H’s preference about S’s doing of an act (A).
The sincerity condition or rule concerns S’s psychological state as it is expressed
in the performance of an illocutionary act (e.g. S’s intention, belief, desire).
Finally, the essential condition or rule is what the utterance “counts as,” i.e.
the “point” of the act (Searle p. 59; also Searle 1979: 2-3). As the terminology
“counts as” suggests, the essential rule is most critical to the creation of an
act (i.e. the central constitutive rule). Thus, each rule focuses upon a slightly
different aspect of what is said: the propositional content rule focuses only
upon the textual content, preparatory rules focus upon background circum¬
stances, the sincerity rule upon S’s psychological state, and the essential rule
upon the illocutionary point of what is said.
Just as Austin found some acts to be infelicitous (and to different degrees,
i.e. misfires versus abuses), so Searle finds that different conditions or rules
are more or less crucial to the non-defective performance of an act. As sug¬
gested above, the essential condition is critical. Whereas each condition or rule
is individually necessary for the successful and non-defective performance of
a given act, however, it is the set of conditions or rules that is collectively
sufficient for such a performance (p. 54).
An earlier quote from Searle noted analytic connections between what the
speaker means, what the sentence (or other linguistic element) uttered means,
and what the speaker intends. Although speaker meaning and intention are
sometimes separated from sentence meaning (see discussion of Gricean prag¬
matics, chapter 6), Searle argues that an analysis of illocutionary acts must
capture “both the intentional and the conventional aspects” of meaning, and
crucially, “the relationship between them” (p. 45) - a relationship that is
sensitive to the circumstances of an utterance. Certain linguistic elements are
viewed as illocutionary force indicating devices that provide conventional
 Speech Act Theory 57

procedures (cf. Austin’s mention of conventional procedures) by which to

perform a given act. (Searle (p. 30) lists a variety of forms that function in
this capacity: “word order, stress, intonation contour, punctuation, the mood
of the verb and the so-called performative verbs.”) But both conventional
meaning (cf. text) and speaker intention can contribute to the circumstances
in which saying something realizes a certain kind of doing.
In sum, Searle places the speech act at the center of the study of language,
meaning, and communication: he proposes that “the basic unit of human
linguistic communication is the illocutionary act” (Searle 1979: 1). Speech acts
are performed through the use of conventional procedures and linguistically
realized through illocutionary force indicating devices. They are enabled (i.e.
created) by constitutive rules, the knowledge of which is part of our linguistic
competence. Speech act theory thus analyzes the way meanings and acts are
linguistically communicated. To summarize irf'Searle^s words: “the semantic
structure of a language may be regarded as a conventional realization of a
series of sets of underlying constitutive rules, and . . . speech acts are acts
characteristically performed by uttering expressions in accordance with these
sets of constitutive rules” (p. 37).

2.3 Taxonomies of acts

Discovering the number, and categories, of illocutionary acts is an important

part of speech act theory (Searle 1979: 1). The identity of an act is a product
of the set of constitutive rules by which it is created. Since different acts may
be similar to one another if they share particular rules, categorizing speech
acts and speech act types can reveal relationships between rules, as well as
relationships between acts. It is also important to know which acts can be
created simply because speech acts are central to linguistic communication:
knowing which speech act to perform (and the rules that govern it) is a crucial
part of how speakers use language to communicate; likewise, knowledge of
how to identify that act is critical to hearer understanding.
Although Austin (1962: chapter 12) proposes a classification of speech acts,
Searle (1979) argues that Austin’s taxonomy does not maintain a clear distinc¬
tion between illocutionary verbs and acts; nor are the categories based on
consistently applied principles. Searle relies upon taxonomic principles (that
sometimes reflect the different types of conditions underlying speech acts) to
build a two-tiered classification. He proposes five classes of speech acts: repre¬
sentatives (e.g. asserting), directives (e.g. requesting), commissives (e.g. promis¬
ing), expressives (e.g. thanking), and declarations (e.g. appointing). Three
main principles differentiate these classes. Other speech acts within these
classes follow the same three principles, but are differentiated by less compre¬
hensively applicable principles.
The most important principles are those that differentiate the five major cat
egories of speech acts. The first taxonomic principle concerns the illocutionary
58 Approaches to Discourse Analysis

point of the act: this is derived from the essential condition of an act (the
condition that defines what the act “counts as”). The illocutionary point of
directives (e.g. requests, orders, challenges, and dares), for example, is that
they are attempts by S to get H to do A. The illocutionary point of com¬
missives (e.g. promises, vows) is that they commit the speaker to some future
course of action. The second principle is the way that words are fit to the
world. Both commissives and directives are built upon a world-to-words fit:
in making a promise, S undertakes to create a world first presented in the
words; in making a request, S attempts to get H to create a world first presented
in the words. Representatives (whose point is to commit S to something being
the case), however, are built upon a words-to-world fit: insist, state, boast,
and conclude, for example, are all based upon the way words are fit to a
world that is “pre-existing” (in the sense that it is not being created by those
words). The third principle is the expressed psychological state: this is derived
from the sincerity condition. The psychological state expressed by repre¬
sentatives, for example, is “belief” (e.g. S believes that X). In contrast, the
psychological state of directives is “want”; the psychological state of commis¬
sives is “intention.”
The other principles discussed in Searle (1979) help to differentiate speech
acts within the five broad categories noted above; they also reveal similarities
(and differences) between specific speech acts that are not in the same general
category. One principle concerns different strengths with which the illocution¬
ary point is presented. For example, “insist” and “suggest” are both directives,
just as “swear” and “guess” are both commissives: the former member of each
pair is presented with more strength than the latter. Two additional principles
are derived from the preparatory conditions of speech acts. The status of S
and H bears differently on the illocutionary point: this principle would dif¬
ferentiate “proposal” from “command” even though both are directives. How
the utterance relates to the interests of S and H is also derived from prepara¬
tory conditions: “boasts” (a type of representative) and “requests” (a type of
directive) are similar because both have to do with S’s interest; these would
contrast with “congratulations” (a type of expressive) since that concerns H’s
interest. Another principle is based upon the propositional content condition:
“differences in propositional content that are determined by illocutionary
force indicating devices” (Searle 1979: 5). Thus, “prediction” differs from
“report” because the former must be about the future, where as the latter can
be about the past or present.
Although I will not discuss each principle underlying the taxonomy in
detail, it is important to note that one principle concerns the availability of
an illocutionary verb: for some acts, the corresponding illocutionary verb
has a performative use; for others, it does not, e.g. “state” versus “boast.”
Illocutionary verbs can also mark aspects of an act other than their illocution¬
ary point (e.g. “insist” marks the degree of intensity). These two points
loosen the connection between verbs and acts so important to Austin. In so
doing, they also extend the boundaries of what can count as an act: this makes
 Speech Act Theory 59
it more difficult to define a closed set of language functions that can be
encoded by illocutionary force indicating devices and realized by constitutive
rules.
In sum, speech acts can be classified into groups and subgroups by a prin¬
cipled set of criteria. Communication relies upon shared knowledge of the
name and type of a speech act: speaker and hearer share knowledge of how
to identify and classify an utterance as a particular “type” of act, as a unit of
language that is produced and interpreted according to constitutive rules.

2.4 Multiple functions and indirect acts

At the end of his article on a classification of illocutionary acts, Searle (1979:

23) points out that there are a limited number of things that we do with
language: “we tell people how things are, we try to get them to do things, we
commit ourselves to doing things, we express our feeling and attitudes and we
bring about changes through our utterances.” We have just seen that much of
the 1979 paper is devoted to telling us how we make these functional discrimi¬
nations among utterances. Taken up in other work is an observation made in
the very last sentence of the article: “often, we do more than one of these at
once in the same utterance.”
How we do more than one thing at once with our words (i.e. the multiple
functions of an utterance) is part of the important issue of indirect speech acts
(Searle 1969). Searle’s view of indirectness (like his taxonomy of speech acts)
draws upon his analysis of the conditions underlying speech acts. An indirect
speech act is defined as an utterance in which one illocutionary act (a “pri¬
mary” act) is performed by way of the performance of another act (a “literal”
act). Hearers are able to interpret indirect speech acts by relying upon their
knowledge of speech acts, along with general principles of cooperative con¬
versation (see Gricean pragmatics, chapter 6), mutually shared factual infor¬
mation, and a general ability to draw inferences.
Although I will not go through the inferencing of indirect speech acts in
detail, I will take some brief examples to illustrate how Searle’s view of in¬
directness arises from his speech act theory. Take, for example, the sentences
“I hope you’ll write a letter of recommendation for me” and “Would you be
able to write a letter of recommendation for me?” Although both sentences
are conventionally understood as directives (their “primary” act), they are also
other acts: the former is a statement; the latter, a question. Both sentences,
however, are understood as directives because the “literal” speech acts that
they also perform (i.e. the statement, the question) focus upon a condition that
allows directives to be performed (i.e. a rule that constitutes a directive). What
the phrase “I hope you will do X” states is a speaker-based sincerity condition
for requests, i.e. S wants H to do A. Likewise, what the phrase “Would you
be able to do X?” questions is a hearer-based preparatory condition, i.e. H is
able to perform A. It is by way of stating a sincerity condition, or questioning
60 Approaches to Discourse Analysis

a preparatory condition, of directives that these sentences perform directives

(their primary act).
The question of how we identify particular utterances as specific acts is of
course one that we addressed before. This problem was noted by Austin:
whereas performative verbs make explicit certain features of the speech situ¬
ation, primary performatives do not, and thus may be ambiguous. Like Austin,
Searle suggests a kind of trade-off between the contributions of textual and
contextual information to our identification of speech acts: it is possible to
perform an act “without invoking an explicit illocutionary force-indicating
device where the context and the utterance make it clear that the essential
condition is satisfied” (Searle 1962: 68). Thus, just as text (e.g. illocutionary
force indicating devices) can assign an act a single identity, so too can context.
But the question of how multiple identities are contextually assigned (or con¬
textually separated) is not discussed at length in orthodox speech act theory.
In sum, an utterance can do more than one thing at a time. Some utterances
have multiple functions because one act is being performed by way of another:
these are called “indirect” speech acts. The conditions underlying speech acts
provide an analytical resource for indirectness. That conditions can have this
analytical function is possible because they have a critical role in our knowl¬
edge of speech act types. When more than one act is performed by a single
utterance, the conditions for the two speech acts nevertheless have a systematic
relationship to one another. Thus, it is relationships between underlying con¬
ditions that allow utterances to do more than one thing at a time.

2.5 Summary: meaning, use, and actions

We have seen in this section that speech act theory is basically concerned with
what people “do” with language - with the functions of language. Typically,
however, the functions focused upon are those akin to communicative inten¬
tions (the illocutionary force of an utterance) that can be performed through
a conventional procedure and labelled (cf. that have a performative verb).
Even within this relatively well-defined set of acts, the act performed by a
single utterance may not be easy to discover: some utterances bear little surface
resemblance to their underlying illocutionary force.
Despite the emphasis on language function, speech act theory deals less with
actual utterances than with utterance-types, and less with the ways speakers
and hearers actually build upon inferences in talk, than with the sort of
knowledge that they can be presumed to bring to talk. Language can do things
- can perform acts - because people share constitutive rules that create the
acts and that allow them to label utterances as particular kinds of acts. These
rules are part of linguistic competence, even though they draw upon knowl¬
edge about the world, including an array of “social facts” (e.g. knowledge
about social obligations, institutions, identities), as well as knowledge about
the grammar of language.
 Speech Act Theory 61

3 Sample analysis: questions, requests, offers

Now that we have discussed some of the key insights and concepts critical to
speech act theory, let us see how it applies to discourse analysis. Speech act
theory provides a framework in which to identify the conditions underlying
the production and understanding of an utterance as a particular linguistically
realized action. Utterances perform different acts because of their “circum¬
stances” (Austin) and because of the knowledge that we have of the conditions
and rules that constitute particular acts (Searle). We see in this section that
our knowledge of the constitutive rules for acts provides a systematic frame¬
work in which we can not only identify relationships between different speech
a.cts (e.g. understand how a threat differs from a promise), but also use a single
utterance to perform more than one speech act at a time (section 3.1). We also
see that this knowledge can be put to use as a way to understand sequential
relationships between utterances (section 3.2).
Before beginning, it is important to note that although speech act theory
began in philosophy (and relied upon hypothetical utterances), it has also been
developed extensively in linguistics. We noted above Austin’s (1962: 100) view
that words “are to some extent to be ‘explained’ by the ‘context’ in which they
are designed to be or have actually been spoken in a linguistic interchange.”
Yet many linguists (e.g. the collection in Cole and Morgan 1975) rely upon
constructed utterances and hypothetical context as data by which to analyze
speech acts. (In this sense, they are following Searle’s (1969: 56) view that
abstraction and idealization are crucial to systematization and theory con¬
struction.) Other scholars have relied upon actual utterances to try to answer
the same sorts of questions concerning speech act conditions (e.g. Blum-Kulka
1987; Ervin-Tripp 1976), contexts (e.g. participant identity and relationship:
Cherry 1990; Herbert 1990; Holmes 1989, 1990), modality (Pufahl 1988;
Stubbs 1986), and categories (e.g. Halliday 1973, 1975; Labov and Fanshel
1977). Although these analyses sometimes considered speech acts in connected
discourse, they did not apply their analyses to the sequential relationships
between utterances themselves (but see Ferrera 1985; Labov and Fanshel
1977). Nor did they consider how utterances can define one another’s speech act
functions (but see Clark 1979; Schegloff 1987). If we want to consider speech
act theory as an approach to discourse, however, we need to consider both of
these issues: how speech act function contributes to sequential coherence, and
how the speech act function of one utterance contributes to that of another.
As I noted in chapter 2, one focus of our sample analyses in this book is
question-answer sequences. We begin in this section by analyzing how a single
utterance fulfills the conditions of a question (section 3.1.1); we then go on
to see that this same utterance also acts as a request (section 3.1.2) and an
offer (section 3.1.3). This will help to illustrate how the underlying conditions
of these different acts can be used to try to identify the functions of a specific
62 Approaches to Discourse Analysis

utterance in talk, ^57e then go on to analyze the utterances following the

question/request/offer in terms of their relationship with the multifunctional
utterance (section 3.2). This will allow us to see how the underlying conditions
of acts have an effect on relationships between utterances, and on the sequen¬
tial organization of talk.
The data that I use are an excerpt from an interaction among four people
(Henry, Zelda, Irene, and Debby (myself)). At the time of the conversation,
Henry and Zelda were in their late fifties, Irene was in her mid thirties, and I
was in my mid twenties. Henry and Zelda are married; Irene is their next door
neighbor and close friend. I was visiting Henry and Zelda’s home for a socio-
linguistic interview (see Schiffrin 1987a: 41—7; chapter 5 of this book). (1)
occurs just after Henry, Zelda, and Irene have been discussing a recent funeral
(of a teacher at their neighborhood school) that Irene and Zelda had attended.
(1) also precedes the more formal beginning of our sociolinguistic interview,

(1) henry: (a) Y’want a piece of candy?

IRENE: (b) No.
zelda: (c) She’s on a diet.
DEBBY: (d) Who’s not on [a diet.
IRENE: (e) [I’m on-
I’m on a diet
(f) and my mother: [buys-=
ZELDA: (g) [You’re not!
IRENE: (h) =my [mother buys these mints. =
DEBBY: (i) [Oh yes I amhhhh!
ZELDA: (j) Oh yeh.
IRENE: (k) The Russell Stouffer mints.
(1) I said, “I don’t want any Mom.”
(m) “Well, I don’t wanna eat the whole thing.”
(n) She gives me a little tiny piece,
(o) I eat it.
(P) Then she gives me an[other,=
henry: (q) [Was =
IRENE: (r) =so I threw it out the window=
HENRY: =there a lot of people?=
IRENE: (s) =1 didn’t [tell her. =
henry: (t) [Was there=
IRENE: (u) =She’d kill me.
henry: =a lot of people at the house?
ZELDA: (v) All: the teach[ers.
IRENE: (w) [A lot of teachers will- probably will all
there till late.
HENRY: (X) Je:sus Christ.
ZELDA: (y) All: the teachers.
HENRY: (z) What a heartache
 Speech Act Theory 63
As I noted above, my initial focus is on the very beginning of (1): I use Henry’s
Y’want a piece of candy? to consider how speech acts are identified (3.1). In
section 3.2 I turn to Irene’s no (3.2.1), Zelda’s She’s on a diet (3.2.2), and
Irene’s story about her diet (3.2.3), in order to discuss speech act sequences.
There are two broad issues critical to the application of speech act theory
to discourse; the analysis will be concerned with both. First is the identification
of speech acts per se: how to identify an utterance as a particular speech act
(3.1) . Although identifying speech as action requires knowledge of the consti¬
tuent rules for speech acts, it also depends upon an assumption that what is
said can be “mapped onto” what is done - an assumption sometimes difficult
to uphold given the fact that utterances may have multiple functions (chapter
2: pp. 33-8; also, recall the speech act view of this in section 2.4). The second
issue is the sequential arrangement of speech acts: h^w an initial speech act
creates an environment in which a next speech" act is (or is not) appropriate
(3.2) . This issue bears centrally on discourse analysis simply because discourse
(by definition) is comprised of sequentially arranged units, and because se¬
quential regularities (sequences that fulfill our expectations) are a key ingre¬
dient in our identification of something as text.
Although I discuss these two issues separately, it will become clear that they
are intimately related to each other: we cannot discover whether a particular
string of utterances forms a “well-formed” sequence of acts unless we are
reasonably certain of what actions those utterances are performing. Thus,
despite the difficulty of making secure judgements about utterance-action
correlations, this initial step is critical to the second issue with which we will
be concerned, and ultimately, with the application of speech act theory to
discourse. Put another way, identifying the speech act performed by an utter¬
ance is critical to the application of speech act theory to discourse: we need
to know the units of a discourse before we can seek to discover, and explain,
the principles responsible for their arrangements, i.e. the reason why some
sequences seem coherent, but others do not.

3.1 Identifying utterances as speech acts

In this section, I show how Henry’s Y’want a piece of candy? can be identified
as a question (3.1.1), a request (3.1.2), and an offer (3.1.3); a key part of my
analysis is that these three acts are themselves intertwined (3.1.4).
Before we begin, note that we will be focusing on the process of identifying
utterances as sequences of speech acts: I try to show the sorts of issues and
problems with which a researcher might have to deal while doing a speech act
analysis of discourse. Although this might seem laborious at times, it is import¬
ant for two reasons. First, analyzing the process by which people identify speech
acts is a critical part of speech act theory: thus, although uncovering bits and
pieces of our knowledge (some of which might seem to be just “common
sense”) is tedious, this is exactly what speech act theory is concerned with.
64 Approaches to Discourse Analysis

Second, the way analysts try to resolve methodological problems stemming

from analysis of this process forms a key part of the speech act approach to
discourse analysis.

3.1.1 “Y’want a piece of candy?” as a question

We noted above that speech act theory defines underlying conditions (includ¬
ing considerations of speaker intent, desired outcome, and so on) that must
hold for an utterance to be used to realize a particular speech act. These rules
- and the felicity conditions they create - often require consideration of both
what is said (its form, its meaning, how information is presented) and the
context in which it is said. For example, since a promise is an obligation to
undertake a future act that the hearer is assumed to want, one cannot enact
a promise using a past tense verb; nor can that future act be one deemed
harmful to the recipient - and this is a matter of social and personal value.
To identify y’want a piece of candy? as a question, we need to consider how
particular conditions are both linguistically met and contextually satisfied. We
begin with Searle’s (1969: 66) rules for questions, and then consider how we
might argue that Henry’s particular utterance realizes those rules. Here are
Searle’s rules:

QUESTION

Types of rules

Propositional content Any proposition or propositional function

Preparatory (a) S does not know “the answer,” i.e. does not
know if the proposition is true, or, in the case of
the propositional function, does not know the
information needed to complete the proposition
truly
(b) It is not obvious to both S and H that H
will provide the information at that time
without being asked

Sincerity S wants this information

Essential Counts as an attempt to elicit this information

from H

The rules above show that a question is constituted under the following
conditions: the speaker lacks knowledge of a particular state of affairs (pre¬
paratory rule) and wants to gain that knowledge (sincerity rule) by eliciting
information from the hearer (essential rule). We will go through these rules in
more detail to see if Y’want a piece of candy? counts as a question.
 Speech Act Theory 65
Consider, first, that it would be helpful if Y’want a piece of candy? could
be considered a reduced form of an interrogative: interrogative sentences are
well suited to the function of questions. Because interrogatives are incomplete
propositions, they fulfill Searle’s first preparatory rule (the speaker lacks
knowledge as to how to complete a proposition). The first preparatory rule
also specifies two possible gaps in S’s knowledge: S can lack either knowledge
of the truth of a proposition or information needed to complete a true prop¬
osition. These two sources of propositional incompleteness parallel the syn¬
tactic and semantic difference between closed and open questions. Closed
questions are those in which the subject noun phrase and verb auxiliary are
inverted (e.g. Does Andy want to go?, Is Betty there?). Closed questions are
propositionally incomplete due to a lack of knowledge as to whether a prop¬
osition is true (we don’t know if “Andy wants to go” is true or not). What
provides their completion is drawn from a chased s^t of options (e.g. yes or
no) that fixes a positive or negative polarity, or allows an inference of confir¬
mation or disconfirmation. Open questions also have subject/auxiliary inver¬
sion, but are initiated with a WH word [who, what, when, which, how, where,
why) that specifies the source of their incompleteness. A question such as who
was there?, for example, presents a proposition whose argument is not speci¬
fied. Open questions are incomplete because the speaker lacks information
sufficient for completion of a true proposition; in contrast to closed questions,
their completion is relatively open-ended.
The form-function correlation between interrogative sentences and Searle’s
preparatory rule often leads to the view that interrogatives are the unmarked
syntax for question asking (e.g. Geis 1989). This is important for our purposes:
if Y’want a piece of candy? is an interrogative, it would then be fairly easy to
say that it fulfills the preparatory rule for questions. Unfortunately, we cannot
be sure that Y’want a piece of candy? is an interrogative. Although it may very
well be a reduction of Do you want a piece of candy? (and thus an incomplete
proposition), its surface syntactic form is a declarative sentence - a complete
proposition that seems not to fit the preparatory condition of questions at all.
Note, however, that I have transcribed Y’want a piece of candy? with a
question mark: this reflects its final rising intonation. In prior work (Schiffrin
1987a), I followed the practice of many scholars (e.g. Quirk et al. 1972: 386;
Stenstrom 1984) and identified declarative sentences with final rises as ques¬
tions. (See also Selting (1992), who argues that prosody is one system of
constitutive cues for questions - sometimes the only distinctive cue.) Still other
scholars have identified sentence fragments as questions even when a fuller
interrogative form of those fragments is not unequivocally recoverable. Mer¬
ritt (1976), for example, identifies the elliptical Coffee to go? and Cream and
sugar? as questions, even though they are not clear reductions of any single
fuller form (i.e. they can be expanded into a number of different interrogatives,
e.g. Do you have/can I buy coffee to go?). Instead of depending on syntax as
criterial for questionhood, such analyses seem to place more weight on final
rising intonation.
66 Approaches to Discourse Analysis

Research on intonation suggests that final rising intonation may very well
be an illocutionary force indicating device for some kinds of questions. Bol-
inger (1982), for example, suggests that final rises convey incompleteness (the
first preparatory rule). Others suggest that final rises convey uncertainty in
domains other than propositional knowledge, e.g. uncertainty about listener
comprehension (Guy et al. 1986) or adequacy of a contribution to conversa¬
tion (Lakoff 1975). Still others (e.g. Brown et al. 1980; Stenstrom 1984)
suggest that final rises demand a response (the essential rule). Thus, some of
the meanings that have been associated with final rises are compatible with
the rules underlying questions.
Here I present some examples that support the idea that final rising intona¬
tion is an indicator of questions (cf. Geluykens 1988, 1989). All the examples
suggest that final rises convey uncertainty (not propositional incompleteness)
that may be interactionally motivated (e.g. by a prior question from H) and/or
resolved (e.g. by an implicit or explicit acknowledgment from H). First is one
of Guy et al.’s (1986) examples, along with a parallel interchange from one
of my own sociolinguistic interviews:

(2) interviewer/H: (a) What’s your name?

informant/S: (b) Maria Martinetti?
(3) irene/H: (a) What’s your name.
debby/S: (b) Debby Schiffrin?

In (2), the interviewer asks for the informant’s name; in (3), these identities
are reversed. S is the person upon whose utterance (2b, 3b) we are focusing;
H is the person asking a prior question (What’s your name?) and receiving the
information that he has sought and that is being provided by S.
(2) and (3) both illustrate several important points about final rising intona¬
tion as an illocutionary force indicating device. Because final rises are used
in these examples with information about which S does know (i.e. S’s own
name), the utterances Maria Martinetti? and Debby Schiffrin? do not seem
like questions, i.e. they violate the preparatory rule of questions that S not
know “the answer.” However, there is uncertainty in this exchange - in fact,
two sources of uncertainty. First, H does not know S’s name: it is this uncer¬
tainty to which the propositional content of S’s response is addressed. Second,
although the propositional content of Maria Martinetti? and Debby Schiffrin?
provides an answer to What’s your name?, S cannot be certain how this
information will be taken. In other words, it is not propositional content per
se about which S is uncertain, but the adequacy of propositional content for
H’s needs (cf. the gloss “Is that what you meant?”) or the adequacy of H’s
reception of the information (cf. the gloss “Did you get that?”). Note, finally,
that S’s response opens a third part of the exchange: it is up to H to let S
know whether S’s information is adequate for H’s needs, i.e. to resolve S’s
uncertainty about the sufficiency of the response (Guy et al. 1986: 26). Thus,
(2) and (3) suggest that intonation can serve as an illocutionary force indicat-
 Speech Act Theory 67
ing device based on the role of an utterance in an exchange - a role that
supplements the contribution of propositional content to communicative func¬
tion.3
The proposed function of the final rises in (2) and (3) can be related quite
easily to the conditions underlying questions. We can restate what I just
suggested above in terms of the rules for questions: S wants information about
H’s reception of information (the sincerity rule) that S does not have (the
preparatory rule) and that S is attempting to elicit from H (the essential rule).
Before returning to Y’want a piece of candy?, let me present several other
examples also suggesting that final rising intonation can evoke the constitutive
rules of questions.
(4) is from the opening portion of a phone call.
a ft
(4) Phone rings.
called: (a) Hello?
caller: (b) Yeh, hi. This is Debby, David’s mother?
called: (c) Oh hi . . . how are you . . .

Final rising intonation on Hello? (a) is suited to its dual function as answer
to the summons provided by the phone ring, and, question as to who is on
the other end of the line (Schegloff 1972). Similarly, final rising intonation
with caller self-identification (David’s mother? (b)) suggests its dual function
as answer to the prior Hello? and attempt to elicit recognition of the self-identi¬
fication. (Note that David’s mother? follows a preliminary self-identification
Debby, to which Called did not respond).
(5) is an example from a phone call in which I am providing my social
security number to an insurance company agent. (Social security numbers are
orthographically and conventionally, even when presented orally, broken up
into three segments.)

(5) DEBBY: One two four?

AGENT: Urn.
DEBBY: Three two?
AGENT: Okay.
DEBBY: Nine four six six.

In (5), I segment the sections of the number sequence and show by final rises
on the first two segments that they are preliminary to completion of the full
sequence. Prior to my continuation of the next segment, the recipient acknowl¬
edges receipt of each intermediate segment; the last segment has falling
intonation.
A final example is (6):

(6) zelda: (a) The following year, his son, who ha- was eighteen years old
just graduating from high school.
68 Approaches to Discourse Analysis

(b) Was walking through the em . . . the fountain, Logan Square

Library?
(c) Y’know that fountain?
debby: (d) Yeh.
ZELDA: (e) Bare footed, and stepped on a- a bare wire.

In (6), Zelda is telling me a story about her neighborhood doctor (Schiffrin

1988b) and is fixing a location important to a key event in the story: she
presents that identificatory information in (b) and (c) with final rising intona¬
tion to elicit my recognition of those locations prior to continuing her story.
Although some of the specific meanings of final rises in these examples differ,
what they share is the following: S elicits from H a response concerning
something (a referent, a proposition) that S has put forth in response to an
inferred (or actual) “uncertainty” from H; S then pursues some other goal or
activity that had been dependent upon H’s receipt of the information put forth
by S. Note that when expressed in these terms, the function of final rises comes
very close to fulfilling the sincerity, preparatory, and essential conditions of
questions. Final rising intonation marks S’s uncertainty about how informa¬
tion provided to H will be taken: what S is questioning is not propositional
content per se, but the adequacy of propositional content for H’s needs. Thus,
what S wants is information about H’s reception of information (the sincerity
rule) that S does not have (the preparatory rule) and that S is attempting to
elicit from H (the essential rule).
Let us return now to Y’want a piece of candy? I have just suggested that the
intonation of this utterance realizes the preparatory, sincerity, and essential
rules of questions. The propositional content of Y’want a piece of candy? also
helps realize these same conditions: in speech act terms, we may say that both
are illocutionary force indicating devices. Consider, also, the meaning of the
verb want. Since verbs like want (like, feel) describe a state internal to the
person of whom that state is predicated, being in a state of “wanting” cannot
be verified by an examination of external evidence and cannot be something
about which another person is certain. Furthermore, since a state like “want¬
ing” is inherently subjective (and thus “knowable” only by its experiencer),
assertions about that state are something to which its experiencer should
respond. When S questions H’s wants, then, it is up to H to either confirm or
disconfirm the accuracy of those wants (cf. Labov and Fanshel 1977). Thus,
Y’want a piece of candy? conveys S’s lack of information (the first preparatory
rule) and counts as an attempt to get H to provide that information (essential
rule).
Note that we have moved from a discussion of S’s uncertainty to ways to
resolve that uncertainty. Once we begin to talk about H’s responsibility to
provide information to S, we have moved from the preparatory rule to the
essential rule of questions - arguing, in effect, that Y’want a piece of candy?
counts as S’s attempt to elicit information from H. Thus, an expression of
uncertainty per se does not tell us that Y’want a piece of candy? counts as an
 Speech Act Theory 69
attempt to elicit information from H: people often express uncertainty about
a state of affairs without trying to elicit information from another that will
resolve that uncertainty. Rather, it is because it is up to H to confirm (or
disconfirm) what S has said about her that Y’want a piece of candy ? realizes
both the first preparatory and the essential rules of questions.
We began our analysis of Y’want a piece of candy? by noting that we cannot
be sure that this utterance is a reduced form of an interrogative sentence - an
important consideration since it would be that form (an incomplete proposi¬
tion) that would most easily allow us to argue that this utterance conveys
lack of speaker knowledge. We then focused on other linguistic qualities of
Y’want a piece of candy? - intonation, meaning - that seemed to converge as
indicators of the same underlying rules. We argued that the final rising into¬
nation of this utterance could be considered an expression of lack of speaker
knowledge; this interpretation fit with the inherent uncertainty of predicating
another’s wants. We also argued that final rising intonation conveyed the
speaker’s desire to resolve uncertainty by appeal to the hearer; this interpre¬
tation fit with the essential rule for questions. Thus, our consideration of
the linguistic qualities of Y’want a piece of candy? has ended up showing
that Henry’s utterance realizes one of the preparatory rules of a question (S
does not know “the answer”), the sincerity rule (S wants this information),
and the essential rule (it counts as an attempt to elicit this information
from H).
One preparatory rule has not yet been considered: it is not obvious to both
S and H that H will provide the information at that time without being asked
(Searle 1969: 66). We really have no way of knowing for sure that Irene would
not tell Henry that she wants a piece of candy if Henry did not ask her about
this. We can guess that Irene might not tell Henry directly that she wants
candy: she might not say I want candy simply because of the asymmetrical
status or intimacy that such a statement implies. But we cannot be sure that
Irene would not request that Henry give her candy (“Can I have a piece?”) or
hint that she would like candy (“That looks good”). Thus, all we can say is
that Y’want a piece of candy? conveys Henry’s desire (the sincerity condition,
see below) that Irene provide information that she might not otherwise provide
(the second preparatory condition).
Finally, we could also construct contextual arguments for the fulfillment of
the felicity conditions. Recall that the sincerity condition underlying questions
is that S wants this information (Searle 1969: 66). There are social and
interactional reasons for arguing that Henry wants this information. First, it
is considered impolite to eat candy without finding out if others also want
some: candy is considered “a treat” in many middle-class American house¬
holds that should be shared with others. Second, Henry and Irene are good
friends who (despite open forms of competition: Schiffrin 1984a) build on each
other’s sense of well-being by not offending one another (e.g. not criticizing
too harshly: Schiffrin 1985b), and by defending one another. Such a relation¬
ship is built upon a continued display of sensitivity to what the other wants
70 Approaches to Discourse Analysis

and needs; thus, it is in the long-term interest of their relationship for Henry
to continue to find out what Irene wants - in this case, whether she wants
candy that he is eating.
We have now examined Henry’s Y’want a piece of candy? from several
directions, and found linguistic clues that support our identification of this
utterance as a question. Thus, we have relied primarily upon how something
was said as a clue to the sincerity, preparatory, and essential conditions under¬
lying questions.

3.1.2 “Y’want a piece of candy?” as a request for

information

In addition to identifying Y’want a piece of candy? as a question, we might

also identify it as a directive, specifically a request for information. Before we
do so, however, it is important to note that requests and questions have a long
and complicated relationship in speech act theory. Focusing first on their
differences, note that directives (the larger speech act type of which requests
are a subtype) differ from questions in an important formal way: the syntactic
structure assumed to most directly manifest a directive is the imperative (e.g.
Come here) whereas the basic syntactic structure for questions is assumed to
be the interrogative. This form-function correlate, however, is not absolute.
We saw in section 2.4 that many speech acts are performed by way of sentence
structures other than those assumed to provide their most direct form: either
declaratives (e.g. I need water) or interrogatives (e.g. Can you pass the salt?),
for example, can be used with an intended directive force.6 The problematic
relationship between questions and requests is illustrated by the fact that one
of the key puzzles in analysis of requests like Can you pass the salt? is how
they are understood: does interpreting Can you pass the salt? as a request for
salt require prior inference from an understanding of its linguistic meaning as
a question about ability (2.4) or is it so conventionally used as a request that
its interpretation bypasses identifying (and understanding) it as a question
(Morgan 1975).
Many studies of directives have also shown that they are typically performed
indirectly. Scholars often allow directness to be indicated not just by unmarked
syntax (interrogatives for questions, imperatives for directives, declaratives for
assertives), but also by the use of performative verbs. Given the latter criterion,
it is quite clear that some directives do have a direct form: one can say I request
that you arrive by 9:00, I order you to pay your taxes, I warn you to stay
away from there. Regardless of which criteria of directness we apply, however,
directives are realized as imperatives and through performative verbs only
under fairly limited conditions, e.g. intimacy and/or expediency for the former
(Brown and Levinson 1987; Ervin-Tripp 1976) in relatively formal, institu¬
tional, written modes of communication for the latter (Pufahl 1988). Thus,
requests are typically performed in a number of very different (but also quite
 Speech Act Theory 71

regular) ways (Ervin-Tripp 1976; Gordon and Lakoff 1975; Pufahl 1988;
Searle 1975).
Let us now return to Y’want a piece of candyf to see how this question can
also be a request. Again, let us start with Searle’s (1969: 66) felicity conditions:

REQUEST

Type of rule

Propositional content Future act A of H

Preparatory (a) H is able to do A. S believes H is able to do

A.
(b) It is not obvious to both S and H that H
will do A in the ftormaP course of events of
his own accord.

Sincerity S wants H to do A.

Essential Counts as an attempt to get H to do A.

The rules for requests should seem familiar: they are very similar to the rules
for questions. Requests and questions share some of the same conditions:

Preparatory
It is not obvious to both S and H that H will:
do A in the normal course of provide the information at that
events of his own accord time without being asked
[request] [question]
Sincerity
S wants:
H to do A this information
[request] [question]
Essential
Counts as an attempt to:
get H to do A elicit this information from H
[request] [question]

This comparison shows that the preparatory, sincerity, and essential condi¬
tions for questions and requests are similar: since it is not obvious that H will
provide information without being asked (preparatory condition for questions),
or that H will do A in the normal course of events of her own accord (prepara¬
tory condition of requests), both questions and requests count as attempts to
get H to do something (their essential conditions) that S wants (their sincerity
conditions). The difference between questions and requests is that what a
speaker wants through a question (“elicit information ) is more specific than
what a speaker wants through a request (“do A”). But what this suggests is
72 Approaches to Discourse Analysis

that questions are one specific type of request: questions are attempts to get a
hearer to do a certain A - to provide information (Searle 1969: 69). It is the
provision of information that is the future act of H (the propositional content
condition of the request). By fulfilling the conditions for a question, then,
Y’want a piece of candy? fulfills the condition for a particular type of request.
In sum, we have seen in this section that Y’want a piece of candy? is both
a question and a request. It enacts a request for Irene to undertake a particular
verbal action, i.e. to provide information about whether she does or does not
want candy. This is a future act (the propositional content condition) that will
resolve Henry’s uncertainty about what Irene wants (the preparatory condition
for questions). We see in the next section that Y’want a piece of candy? has
still another speech act identity: it is an offer.

3.1.3 “Y’want a piece of candy?” as an offer

Although I spent a great deal of time arguing that Y’want a piece of candy? is
a question (3.1.1) and a request for information (3.1.2), I continue our sample
analysis by proposing that it can also be identified as an offer, i.e. Henry is
using the utterance to make something available to Irene. In some ways this
seems to be the action that is most conventionally identified with Y’want a
piece of candy? Finding the speech act labels that people would typically use
to categorize the illocutionary force of an utterance is important. Speech act
analysts assume that interlocutors agree on the speech act performed by a
particular utterance: this intersubjective agreement is a prerequisite to com¬
munication that is assumed to proceed through the reciprocal processes of
producing and interpreting speech acts (Taylor and Cameron 1987: chapter 3;
see chapter 11 of this book). Put most simply, it is by finding the “unit” into
which an act fits - the unit intended by S - that H can present a next act.
But if “offer” seems to be the speech act most conventionally associated with
Y’want a piece of candy?, why have we spent so much time showing that this
utterance could be a question and a request? Here I propose that the function
of Y’want a piece of candy? as an offer is intricately tied to its functions as
question and request - such that we needed to understand this utterance as
both question and request before we could understand it as offer. The rela¬
tionship between question/request and offer also illustrates the application of
important aspects of speech act theory to discourse: both the principles by
which speech acts are classified (2.3), and the view that utterances are multi¬
functional because of relationships between their underlying conditions (2.4),
can be incorporated into speech act applications to discourse. Finally, analyz¬
ing the multifunctionality of utterances helps to reveal the different response
options made available by an utterance, and thus the possibility of underlying
sequences bound by different functional relationships.
Before we begin, it is important to note a general difference between direct¬
ives (the larger class of speech acts encompassing requests) and commissives
 Speech Act Theory 73

(the larger class of speech acts encompassing offers): directives are attempts
by S to have H do A; commissives are commitments from S to do A for H.
Many utterances are requests but not offers: if I ask you to pass the salt, I
have not committed myself to any action beyond using the salt in some way.
But the asymmetry in who does what for whom that divides requests from
offers can also disappear, such that many acts are simultaneously offers and
requests (cf. Wierzbicka 1987: 190-7). Take, for example, the speech act
“invitation.” Although Searle (1979: 11) and Leech (1983: 217) view invita¬
tions as directives, they may also be analyzed as both offer and request: if I
invite you to a party at my house, I am simultaneously offering you access to
an event of which I am a sponsor and requesting access to your company at
a future time (cf. Schiffrin 1981b: 239-40). Note, also, that Searle (1979:
11-12) assigns the same direction of fit (world-to-words) to both commissives
and directives, noting that classifying speech*acts v^buld be simpler if they
were really members of the same category (e.g. promises could be requests to
oneself). (See also Leech’s (1983: 206) suggestion that directives and com¬
missives be merged into a “superclass.”)
We begin our analysis of Y’want a piece of candy? as an offer by noting the
general conditions underlying commissives. Searle (1979: 11) defines com¬
missives as “illocutionary acts whose point is to commit the speaker ( ... in
varying degrees) to some future course of action.” Commissives differ among
themselves in terms of S’s degree of commitment. When making a promise,
for example, S undertakes an obligation to perform A in the future (essential
condition); S also believes that H would prefer A being done to A not being
done (preparatory condition; Searle 1962). When making an offer, however,
the essential and preparatory conditions of promises do not hold. Further¬
more, the reason the essential condition does not hold is because the prepara¬
tory condition does not hold.
The chart below reformulates the difference between promises and offers in
terms of knowledge (as reflected in preparatory conditions) and commitment
(as reflected in essential conditions):

Promise Stage 1 Stage 2 Stage 3 Stage 4

Knowledge S knows H wants A --->
[preparatory]
Commitment S commits to do A --->
[essential]
Offer
Knowledge S does not know S finds out S knows
[preparatory] if H wants A if H wants A H wants A
Commitment S commits
[essential] to do A

This suggests that a key difference between an offer and a promise is the
knowledge that S has about what H wants (the preparatory condition noted
74 Approaches to Discourse Analysis

above). It also suggests that reducing S’s uncertainty about H’s wants can alter
S’s commitment to do A. Recall that uncertainty is a preparatory condition
for questions and requests: thus, we might very well expect S to ask a question
to find out what H wants before committing herself to do A. This is what I
suggest at stage 2: S finds out if H wants A by requesting information from
H. Given affirmation that H does want A, S’s uncertainty is then reduced
(stage 3) and S commits to do A (stage 4). This view of the difference between
promises and offers allows us to see that questions (i.e. requests for informa¬
tion) can play a critical role in offers: they reduce uncertainty about whether
H wants A, thus, potentially leading to a commitment to do A.
The relationship can be summarized in slightly different terms:

Act 1: “S give H candy”

Offer S intends act 1
Question S does not know if H wants act 1
So Request S wants H to do act 2: tell S if H wants act 1

This analysis suggests that we may paraphrase Y’want a piece of candy? as “I

intend to give you candy if you want it.” Note the important role of the
conditional in the paraphrase: it is because S does not know if H wants candy
(a preparatory condition of questions) that S wants H to tell S if H wants
candy (a preparatory condition of requests). Once this uncertainty is reduced
- with an answer to a question - S can undertake a commitment to do A for
H. Thus, asking a question about the preparatory condition for promises (H
prefers A to not A) can lead S to undertake an obligation to do A. We see in
a moment that this relationship is reflected in the way people make offers.
The relationship between offers and requests described above suggests that
people can make offers either with two separate utterances or with one
utterance. The two-utterance possibility would occur if S were to elicit some
go-ahead (e.g. approval, endorsement, confirmation) from H about the bene¬
fits of A for H, prior to committing to do A for H. Thus, the interchange in
(a) and (b) of the hypothetical (7) could precede the offer in (c):

(7) cliff: (a) Do you want help?

diane: (b) Yes.
cliff: (c) Well, here, let me help you then.

It seems, however, that speakers usually do not try to explicitly gain confir¬
mation that H wants A prior to offering A. The reason is what I have been
hinting at above: questions can act as offers when they question (i.e. request
information about) a preparatory condition of promises (H would prefer S’s
doing A to his not doing A and S believes this).
In order to see how the one-utterance possibility for offers works, we need
to note yet another connection among the conditions underlying different
speech acts. The preparatory condition of promises noted above (to simplify:
 Speech Act Theory 75
S believes that H wants A) has an interesting similarity with the sincerity
condition of requests (S wants H to do A) and an important effect on indirect
speech acts. Observe, first, Searle’s (1975: 65, 72) suggestion that S can issue
an indirect directive by stating that the sincerity condition obtains. This
strategy accounts for the use of statements such as those in (8) to make
requests:

(8) (a) I would like you to go now.

(b) I want you to do this for me.
(c) I’d rather you didn’t do that anymore.
(d) I wish you wouldn’t do that.

Each of the utterances in (8) uses a verb that states<,(and thus fulfills) the
sincerity condition for requests, i.e. S wants H to do A. Although a question
about whether the sincerity condition obtains cannot issue a request, it is just
such a question that can issue what we have been calling offers - just so long
as we switch the role of subject and object. (9), for example, contains counter¬
parts to (8) that I believe we would interpret as offers:

(9) (a) Would you like me to go now?

(b) Do you want me to do this for you?
(c) Would you rather I didn’t do that anymore?
(d) Do you wish I wouldn’t do that?

The syntactic reversal between statement-based requests (in 8), and question-
based offers (in 9), reflects the different roles played by S and H in relation
to A - whether it is S or H who wants A. That is, what changes is our
assumption about for whose benefit A is intended and who it is that wants A:
the request seems to benefit S and the offer seems to benefit H. Note that this
is entirely consistent with the different conditions underlying requests and
promises. Whereas the sincerity condition of requests is that S wants H to do
A, a preparatory condition of promises is that H would prefer (cf. want) S’s
doing A to his not doing A and that S believes this. Thus, stating that S would
like A is a request because it states a speaker-based sincerity condition of
requests; asking if H would like A is an offer because it questions a preparatory
condition of promises. As we saw in the chart on page 73, the establishment
of this preparatory condition (S’s reduced uncertainty about H’s wants) paves
the way for S to undertake an obligation to do A.
Returning now to Henry’s Y’want a piece of candy?, we see that this utterance
is exactly parallel to the hypothetical “do you want X?” (8b). Not only does
it question whether H wants A, but (as we see in a minute) it can receive a
response appropriate to either a request for information or an offer. Multiple
response possibilities provide important evidence in speech act theory, the
availability of more than one response type shows that a single utterance
performs more than one speech act. That one can respond to Do you have the
76 Approaches to Discourse Analysis

time? by saying either Yes, ten o’clock, or just Ten o’clock, for example,
indicates its potential understandings as either question or request jClark
1979) (section 3.2 discusses the sequential effects of multifunctionality). (10a)
and (10b) show two different responses to Y’want a piece of candy?

(10a) S: Y’want a piece of candy?

H: Oh yes.
S: Here it is.
H: [takes it] Thanks.
(10b) S: Y’want a piece of candy?
H: [takes it] Thanks.

(10a) and (10b) illustrate that a respondent to Y’want a piece of candy? may
choose between two alternative courses of action. The response in (10a)
depends upon the literal meaning of Y’want a piece of candy? as a question
(and request); the response in (10b) depends upon its primary (indirect) use
as an offer. These two possibilities highlight the multiple speech act identities
of Y’want a piece of candy?
In this section, we have seen that offers are commissives: S proposes a future
A for H; S’s uncertainty as to whether H wants A reduces S’s obligation to do
A. Although one way for S to resolve uncertainty is through a sequentially
prior question that asks whether H wants A, it is also possible for S’s question
itself to act as an offer if the propositional content of the question focuses on
the preparatory condition of a promise. Put another way, an offer can be a
“primary” speech act (2.4) performed by way of a question (the “literal”
speech act (2.4)) about H’s desire for A; it is this question that allows S to
find out whether A is what H wants before undertaking an obligation to do
A. Thus, when Henry says Y’want a piece of candy?, he is indeed making an
offer to give Irene candy. However, his offer simultaneously questions Irene,
requesting that Irene tell him whether she wants the candy.

3.2 Identifying speech act sequences

Built into the felicity conditions of questions, requests, and offers is a need for
hearer response. Recall, for example, the essential condition of questions (an
utterance counts as S’s attempt to elicit information wanted from H) and
requests (an utterance counts as an attempt to get H to do A). Similarly, offers
require H to indicate interest in what A has conveyed a willingness to do.
What this need for response suggests is that an analysis of a single utterance
as a question, request, or offer leads naturally to an analysis of the utterance(s)
that follow. Consistent with the notion that utterances perform speech acts,
an analysis of succeeding utterances becomes an analysis of speech act se¬
quences. As noted by Taylor and Cameron (1987: 58), however, “the question
of how illocutionary acts are sequenced in actual episodes of connected speech
 Speech Act Theory 77
is not one that looms large in the lives of philosophers. . . . For the analyst
working with natural data, however, it is an issue on a par with that of
classification/identification.” Thus, although the identification of speech acts
per se is central to the discourse application of speech act theory, and to speech
act theory itself, the combination of speech acts into well-formed sequences is
important only to discourse analysis.
Despite the sequential focus of this section, I begin again with the problem
of identifying units, this time the second unit in a sequence. We see almost
immediately that the “identification” issue differs dramatically for a unit in
“second place,” simply because the unit in “first place” provides information
central to the identification of the next unit. We have just seen that Y’want a
piece of candy? can be analyzed as three different speech acts: question,
request, offer. This suggests that what follows Y’wattf a piece of candy? can
also be classified in more than one way, such that the act performed in
response to an offer is as multilevelled as the offer itself. I consider three
responses: Irene’s No (3.2.1), Zelda’s She’s on a diet (3.2.2), and Irene’s story
about her diet (3.2.3).11

3.2.1 Irene’s “No”: answer, compliance, and rejection

We saw above that Irene says No immediately after Henry’s Y’want a piece
of candy? It is relatively easy to identify No as an answer: although it is
elliptical, we (as hearers and as analysts) rely on the content of the prior
question to expand No and to understand that it provides the polarity left
open by that question, thereby semantically completing the proposition. In
addition to answering Henry’s question, Irene’s No also complies with Henry’s
request for information: Irene provides the information that she does not want
candy. Finally, Irene’s No also responds to Henry’s offer of candy: she does
not accept (she rejects) the offer of candy.
Note, now, that the sequential consequences of our analysis would differ
tremendously if we analyzed Irene’s No only as an answer. The main difference
is that question-answer sequences are structurally complete after the provision
of their second part: despite the addition of other relevant material, e.g. an
explanation for a negative answer, such material is not necessary for the
sequence and the sequence is coherent on its own. If we analyze the sequence
as having an initial request or offer, however, it is a little more difficult to
argue for structural completion. Although No does comply with a request for
information, either a positive or negative response alone seems incomplete. If
Irene had said Yes, for example, she might then be expected either to wait for
Henry to explicitly offer her candy, or to actually take the candy. And the No
response seems to call either for a conventional marker of politeness (e.g. no
thanks) or an explanation.
But why should Irene’s No - which does supply the requested information
and does respond to the offer - create a third slot in the sequence? Although
78 Approaches to Discourse Analysis

many linguistically oriented speech act analysts have proposed answers to this
question (e.g. Davison 1975; Green 1975), a more thorough account comes
from more socially oriented analyses of politeness (discussed in chapter 4). We
now go on to see what else actually does follow Irene’s No.

3.2.2 Zelda’s “She's on a diet”: expansion and account

After Irene says No to Henry’s Y’want a piece of candy?, Zelda says She’s on
a diet.

(1) henry: (a) Y’want a piece of candy?

IRENE: (b) No.
z
zelda: (c) She’s on a diet.

Our discussion of Zelda’s remark will raise two sets of questions critical to
applications of speech act theory to discourse analysis. First, what is the range
of acts that can be performed through speech? Put another way, what is the
overall inventory of acts for which utterances can be used and onto which
utterances can be mapped? Questions such as these go to the very heart of
speech act theory - for some of the acts that we discuss in this section are not
those that would ordinarily be included in an orthodox speech act taxonomy
that focused solely on those acts whose illocutionary force might be indicated
by performative verbs (2.3). Rather, these are acts that are sequentially emergent:
they arise only in relation to another prior act. The second set of questions
concerns exhaustiveness: is everything that is said a realization of a speech act
in a well-formed sequence? In other words, do we have to pack every utterance
into a speech act that has a sequential relationship with other speech acts?
This set of questions is also central to speech act theory: the assumption that
the speech act is the basic unit of linguistic communication (2.2) leaves little
room for communication by way of utterances that are not speech acts.
Here I will propose two different speech act identities of She’s on a diet -
the first more challenging to speech act theory than the second. Consider, first,
the possibility that She’s on a diet is an expansion: a sequentially dependent
unit which adds information supplementary to a prior unit. Calling She’s on
a diet an expansion seems to capture the way that Zelda’s remark develops,
adds to, or follows up on (Coulthard et al. 1981) what Irene has said. But
giving such a unit a place in a speech act sequence raises several concerns.
Expansions are totally dependent on prior units, such that it would be difficult
to state the rules for expansions independent of sequentially prior speech acts.
Expansions are difficult to differentiate from other units in other ways: not
only are there no clear criteria by which to identify them, but they do not meet
Searle’s necessary and sufficient conditions for a speech act.
Let me more fully explain these problems through some examples. Consider,
first, that expansions need not be tied to answers: they may be expansions of
 Speech Act Theory 79

virtually any other unit in a structure. Also in (1), for example, is a section
where Henry repeats a question:

(q) Was there a lot of people?

(t) Was there a lot of people at the house?

Another way of looking at (t), however, would be to say that the addition of
at the house to the initial question in (q) makes the question in (t) an expansion
of the question in (q). But does this then mean that all additions are expan¬
sions? Does it also mean that only additions are expansions?
(11) suggests that additions are neither necessary nor sufficient criteria for
counting a remark as an expansion. (11) is also from one of my interviews.

(11) debby: (a) Do you think there’s much prejudice between like other
groups-
(b) Other ethnic groups?
(c) Other nationalities?

There are two changes in (11) that we might call expansions. First, if we view
the change from other groups (a) to other ethnic groups (b) as an addition
(simply because I have added ethnic), then we might call other ethnic groups
an expansion. However, the change from (a) to (b) is the sort of addition often
described as a self-repair (e.g. Schegloff et al. 1977). When a speaker repairs
a word or expression, it is usually assumed that what is received by a listener
as the message is the replacement itself (e.g. other ethnic groups), not the
repairable (e.g. other groups) plus an addition (e.g. ethnic). But if it is only
the repair that contributes to the message, we would probably not want to
call other ethnic groups an expansion; instead, we might want to call it a
replacement or even just a continuation (or completion) of the same prior unit
(see e.g. Polanyi 1988). Thus, what the change from (a) to (b) suggests is that
not all additions of lexical information are expansions.
The second change to consider is from other ethnic groups (b) to other
nationalities (c): this suggests that just as additions are not sufficient criteria
for expansions, nor are they necessary criteria. “Nationalities” does not add
lexical information to “ethnic groups”; nor does it supplement the semantic
meaning of “ethnic groups.”13 But note how I intended this change: I was
trying to clarify what I meant by ethnic group. Couldn’t we therefore call the
change from other ethnic groups to other nationalities an expansion of my
meaning, i.e. an expansion of what I intended to convey?
These examples have illustrated several problems. First, at a very general
level, we can identify many different utterances as expansions - expansions of
answers, expansions of questions - and see them all as having the same role.
On a more particular level of analysis, however, this would be very misleading.
The specific identity of an utterance as an expansion of something is intimately
tied to the identity of that of which it is an expansion: a question expansion
80 Approaches to Discourse Analysis

is thus as different from an answer expansion as a question is from an answer.

Second is a methodological repercussion of allowing some units to be so
dependent on, and non-autonomous from, what precedes them: our analysis
of a single discourse unit rests critically on our analysis of its precursors in
the text. Thus, if we are wrong in analyzing the status of an initial utterance,
we cannot help but be wrong in analyzing the status of subsequent utterances.
And our opportunities for being wrong thus multiply with each succeeding
utterance.
Now that we have considered some problems with identifying She’s on a
diet as an expansion of an answer, let us turn to its possible function as an
account - as an explanation for Irene’s rejection of Henry’s offer. Like expan¬
sions, accounts are relational: they provide reasons and/or motivations for
some prior action that has been considered marked (e.g. inappropriate, impol¬
ite, insulting) in some way (Scott and Lyman 1968). Because they provide
explanations, accounts are causally linked to prior actions, as if the person is
saying “I did X because of Y.” Zelda’s She’s on a diet acts as an account for
a very simple reason. Our knowledge about candy and diets tells us that being
on a diet provides a reason for not accepting candy: candy is fattening and
diets don’t allow fattening food; therefore, being on a diet is a reason for not
accepting candy. Note, also, that the concept of “diet” is nicely fit to the
content of Henry’s asking about what Irene “wants”: diets require people to
forgo their more immediate “wants” (fattening food like candy) for the sake
of their more long-term “wants” (losing weight).
To conclude: we can describe the sequential structure of (1) in two differ¬
ent ways: as a two-part sequence of S-question/H-answer with H’s option¬
al answer expansion, or we can see it as a three-part sequence of S-offer/
H-rejection/H-account. Thus, our identification of Y’want a piece of candy?
as multifunctional (question, request, offer) can lead us to see sequential
completion after either a second act or a third act. In the next section, we
show how Irene’s I’m on a diet story provides another account.

3.2.3 Irene’s “I’m on a diet” story: expansion and account

Although Zelda accounts for Irene’s refusal of Henry’s offer of candy, Irene
also provides her own account, first by saying I’m on a diet in (e), and then
by telling a story (e to u) about her diet:

IRENE: (e) =I’m on a diet

(f)
ZELDA: (g)
IRENE: (h)
DEBBY: (i)
ZELDA: (j)
IRENE: (k) The Russell Stouffer mints.
 Speech Act Theory 81

(1) I said, “I don’t want any Mom.”

(m) “Well, I don’t wanna eat the whole thing.
(n) She gives me a little tiny piece,
(o) I eat it.
(P) Then she gives me anrOther,=
HENRY: (q) Was =
IRENE: (r) =so I threw it out the window=
HENRY: =there a lot of people?=
IRENE: (s) =1 didn’t rtell her. =-,
HENRY: (t) Was there=
IRENE: (u) =She’d kill me.

As I noted above, it is not merely saying I’m on a diet tl^jt provides an account,
bdt Irene’s entire story about her diet. Treating a story as a speech act raises
important issues for the application of speech act theory to discourse - since
it implies that an entire discourse unit can perform a speech act. (I consider
consequences of this shift in unit size in section 3.3.) We will proceed by
examining the content of Irene’s story as a sequence of speech acts within the
story world; we will also pay attention to how the story is linguistically
tailored to the conversational world, i.e. as an account for the rejection of an
offer.
Consider, first, how Irene opens her story:

I’m on a diet
and my mother buys- my mother buys these mints.
The Russell Stouffer mints.

The clause my mother buys these mints is not concurrent with speaking time:
it shifts backward in time to locate an event as part of a previous experience
(I comment on the use of the present tense below). Shifts in reference time
often help to initiate a story world, separating it from an ongoing conversa¬
tional world. Despite this shift, my mother buys these mints is syntactically
connected with and to I’m on a diet; it is also presented in the same intonation
unit. These linguistic connections link the first part of Irene’s account (I’m on
a diet) with the story told to expand that account. Note, also, that mention
of mints establishes a cohesive tie with Henry’s prior offer of candy.
The use of the present tense in my mother buys these mints is ambiguous: it
can suggest either that buying mints is a repeated, habitual action or it can be
a use of the historical present tense, a tense typically reserved for narrative
events within the story world itself (Schiffrin 1981a). The latter interpretation
is intruiging: it is a marked use of the historical present (since the HP is
unusual in story abstracts) that could be said to have an evaluative, highlight¬
ing function, showing how critical buying candy is to later events. Introduction
of the mints with the indefinite these can also be considered evaluative, but
interestingly in a pejorative sense (Wright and Givon 1987), thus prefiguring
82 Approaches to Discourse Analysis

the disdain that Irene will show towards her mother’s offer. Finally, the
addition of detail through The Russell Stouffer mints also draws attention to
the candy (Tannen 1989a: chapter 5). Use of a proper name to identify the
mints makes a claim about their “knowability” (i.e. they are recognizable; see
chapter 6). Postponing that name to the initial introduction of the mints
iconically marks that information as supplementary - a way of conveying
increased relevance (see chapter 6). Thus, the story abstract establishes a
cohesive tie with what Henry has just offered and reveals the centrality of the
candy bought by Irene’s mother to the upcoming story events.
Although a story about a past experience can never really tell us what
actually happened during that experience, we can still interpret how a story -
as a version of the past - fits into a present conversational world. In other
words, we can examine Irene’s story to see how Irene constructs a version of
her experience that fits her current conversational needs, i.e. that addresses a
current topic and accomplishes a current purpose. A convenient way to see
how Irene accomplishes these goals is to dissect the story into actions - in
effect, to perform a speech act analysis on the events in the story (cf. Labov
1984). In addition to providing another application of speech act theory to
discourse, this analysis also provides further insight into offers and their
rejections. Thus the actions underlying Irene’s reported experience are:

Actions in the story world

MOTHER: Offer 1: S will do A [unstated]
IRENE: Rejects Offer 1 I said, “I don’t want any Mom.” (1)
MOTHER: Explains Offer 1 “Well, I don’t wanna eat the whole
thing.” (m)
does A She gives me a little tiny piece, (n)
IRENE: Accepts Offer 1 I eat it. (o)
MOTHER: [no Offer 2]
does A Then she gives me another. (P)
IRENE: Rejects A so I threw it out the window. (r)
Reactions to the story world
I didn’t tell her. (s)
She’d kill me. (t)

Note that Irene’s mother makes an offer (although it is not explicitly stated
in the story; see below). Irene rejects the offer (1), although the rejection is
ineffectual (the mother gives Irene the candy anyway (n)). Irene’s mother then
repeats the act (p) without first offering to do so.
We spoke earlier of the conditions underlying offers, suggesting also that
offers contain implicit requests about whether H wants the A that S is making
available. Although Irene does not report her mother’s offer of candy, I believe
we can infer that an offer was made (either verbally or nonverbally, e.g. by
holding out the box of candy). We saw earlier that a precondition for an offer
of goods is the availability of those goods to S: the fact that Irene reports
 Speech Act Theory 83
herself as addressing her mother (I don’t want any, Mom (1)) suggests that her
mother is the one with the available goods (the candy). We also saw that S
cannot be sure that H wants A: we can infer that Irene’s mother is S and that
she is guessing that this is something that Irene might want. (Note the import¬
ance of the negative in I don’t want any, Mom: negative statements typically
imply a contrast with expectation; Givon 1979; Horn 1988.)
Additional evidence that Irene’s mother has offered her candy is that Irene’s
mother then explains why Irene should take candy: Well, I don’t wanna eat
the whole thing (m). Well in (m) helps establish the sequential location of this
utterance. More specifically, it is third in a sequence in which the initial move
set up an option (an offer of candy), the second move deviated from that
option (I don’t want any. Mom (1)), and the third responds to the deviation
(Schiffrin 1987a). Since Irene’s mother does not want the whole thing, it seems
reasonable that she would be trying to give candy awaf by offering it to Irene.
A final reason to infer that an offer was made stems from our earlier obser¬
vation that offers make goods or services available to another. Irene’s mother
actually gives Irene candy (She gives me a little tiny piece (n), Then she gives
me another (p)). These are actions that physically realize exactly what an
earlier offer would have made available.
After Irene’s initial refusal of her mother’s offer, Irene does accept the
candy: I eat it (o). Note that Irene’s mother has forced a compromise by
appealing to Irene for help: Well, I don’t wanna eat the whole thing (m)
solicits Irene’s help in preventing her from doing something that she herself
does not want to do. (Irene’s mother does not want to eat the whole thing
because she does not want to get fat; but if she doesn’t eat the whole thing, she
may have to throw the rest away, thereby wasting the candy.) This appeal
transforms the mother’s offer into an act that will not just benefit Irene (recall
that offers are beneficial to H, e.g. candy is supposed to taste good), but will
also benefit the mother herself. If Irene does accept the candy, then, she can
be seen to be putting aside her own best judgement in an effort to help her
mother. And if Irene helps her mother, she can be seen as a good daughter -
one who ignores her own reservations in order to comply with her mother’s
wishes.15
Because Irene’s initial rejection of A was ineffectual, the mother’s repetition
of A (Then she gives me another (p)) has added social and personal implica¬
tions. Irene reports that her mother just gives her the candy the second time
around: S does A without asking whether H wants A and without trying to
justify A. Irene does not attempt to reject this candy: rather, she avoids the
need to do so by throwing the candy out the window (so I threw it out the
window (r)). This action is important for it allows Irene to maintain the
appearance of deference to her mother. That is, Irene manages to look as if
she is doing what her mother wants, but she also maintains her own desire
not to have her own wants (to stay on her diet) impinged upon. As Irene
reports, there are consequences if her mother discovers this deception: she’d
kill me (u). But it is precisely the fact that Irene is willing to risk these
84 Approaches to Discourse Analysis

consequences that underline her seriousness about her diet and, thus, take us
back into the conversational world in which her story provides an account for
her refusal of Henry’s offer of candy. Irene has gone to such great efforts to
avoid eating candy - even to the extent of deceiving her mother - that she is
not going to undo the product of those efforts just for Henry (see also
Schiffrin, forthcoming).
Earlier I focused on the linguistic details of Irene’s introduction of the candy
into her story - details that established a cohesive tie with what Henry has
just offered and showed the importance of the candy for the upcoming story
events. Now that we have seen how a speech act analysis of the story world
can help us interpret it as a speech act in the conversational world, it is worth
noting again that the story is told in ways that help fit it into the conversa¬
tional world. Note, for example, the alternation of tenses in which the main
events are reported: preterite in (1), shift to historical present in (n), (o), and
(p), then shift back to preterite in (r). The events reported in the historical
present are those that provoke Irene’s drastic action of throwing the candy
out the window:

(n) She gives me a little tiny piece,

(o) I eat it.
(p) Then she gives me another,

Since it is the mother’s persistence that creates the dilemma that Irene faces,
these events are important for the point of Irene’s story. The use of the
historical present in (n), (o), and (p) also highlights these events not only in
the story world, but in relation to the conversational world (Fleischman 1990;
Schiffrin 1981a; Silva-Corvalan 1983): had Irene’s mother not been so persist¬
ent, Irene would not have been forced to defy her mother. Thus, Irene will
refuse an offer of candy despite persistent offers from her mother or from
Henry.
In sum, Irene’s story shows her sincere efforts to diet: the only time she
accepts candy from her mother (even a little tiny piece) is when greater
social and personal damage would be created by refusing candy. Irene’s re¬
ported display of will and sincerity provides an account for why she has
rejected Henry’s offer of candy: Irene feels so strongly about her diet that
she is willing to throw candy out the window, thereby deceiving her mother
and facing the risk incurred by that rejection and deception (she’d kill
me (u)).16

3.3 Summary of sample analysis

In sections 3.1 and 3.2 we considered the performance of several different acts
through four chunks of speech: Henry’s Y’want a piece of candy?, Irene’s No,
Zelda’s She’s on a diet and Irene’s “I’m on a diet” story. To summarize:
 Speech Act Theory 85
Utterances Sequence 1 2 3

Henry: Y’want a piece of candy? Question Request Offer

Irene: No. Answer Compliance Refusal
Zelda: She’s on a diet. {Expansion} Account Account
Irene: I’m on a diet + story {Expansion} Account Account

This section summarizes and makes more general points concerning the utter-
ance-act pairs (3.3.1) and the act sequences (3.3.2).

3.3.1 Utterances and actions

We began our application of speech act theory to discourse by analyzing

relationships between utterances and actions: we showed how Y’want a piece
of candy? is a question, request, and offer (3.1). We saw that the basis for our
identifications of speech acts varied. In identifying questions, we relied largely
upon linguistic clues. Requests and offers required that we try to judge com¬
municative function. Instead of focusing just on the linguistic characteristics
of utterances, we relied more upon our knowledge of the general background
conditions necessary for an utterance to have a particular function (to count
as a particular kind of action) and the applicability of those general conditions
to particular circumstances. We also saw that identifying speech acts is com¬
plicated by the fact that utterances and speech acts need not have a one-to-one
relationship. In analyzing how an utterance could perform more than one act
simultaneously, we suggested that multifunctionality could arise (in part) from
the way that underlying conditions for speech acts are themselves related to
one another. We also saw that more than one utterance could figure in the
performance of a single act, i.e. Irene’s story served as an account. In brief,
rather than “one form for one function” we found “one form for many func¬
tions” and “many forms for one function.”
One-for-many and many-for-one relationships between form and function
raise problems that affect the application of speech act theory to discourse
analysis. Consider, first, problems stemming from “one form for many func¬
tions” relationships. Although the acts that we considered above (question,
request, offer) are typical of those treated by philosophically oriented speech
act theorists, many scholars who have taken a more social interactional ap¬
proach to speech acts (e.g. Ciccourel 1980; Halliday 1975; Labov and Fanshel
1977; Schegloff 1987) have located acts quite unlike those that we have
discussed. The relevance of their discoveries for the “one form for many
functions” relationship is that these interactively embedded acts are often not
performed alone, i.e. their identity is often dependent on the performance of
another, simultaneously performed act. The challenges identified by Labov
and Fanshel (1977: 93-7), for example, are second-order functions of re¬
peated requests. Speech acts such as these raise problems: should we assign
86 Approaches to Discourse Analysis

speech act labels to acts that emerge only as byproducts of other actions? Put
another way, do we want to say that all of the many functions realized through
a single utterance are speech acts? And do we then need to include all of these
functions in the classificatory schema of speech acts assumed to be part of
communicative competence? (Compare the taxonomies of those noted above,
for example, to the taxonomies of Austin (1962) and Searle (1976), discussed
in section 2.3.) How we answer such questions depends, of course, on how
broadly we view speech acts: if we adopt Austin’s relatively narrow view of
speech acts, for example, then we cannot have a speech act without a perform¬
ative verb (such a criterion would allow an act like “challenge” into the
inventory, for example, but not an act like “tease”). Although these problems
are not created by the existence of “one form for many functions” relation¬
ships, they are highlighted by them. Once we start finding multiple functions,
we realize that not all of the many layers of functions that are realized through
speech are as easily codified as those that have been more typically considered
by speech act theorists, i.e. not all are first-order functions associated with
communicative intentions.
“One form for many functions” relationships not only highlight problems
already implicit in speech act theory, they also create new problems. For
example, I suggested that Y’want a piece of candy? enacts three speech acts:
question, request, offer. Questions, requests, and offers are intricately related
to one another. Similar conditions underlie questions and requests: both count
as an attempt to get H to do something that S wants; questions are a more
specific attempt to get H to provide information that S does not have and that
S wants. The question-request interdependence also arises because S’s attempt
to get H to tell him or her something (“I want you to do this: tell me if X”)
is based on the preparatory condition of questions, i.e. that S does not yet
have information about X (“I do not know if X”). And offers may contain
implicit requests that the hearer provide the speaker with information about
the desirability of the act, thereby fulfilling the sincerity condition of requests.
Relationships such as these imply that a “one form for many functions”
relationship may arise (at least partially) because of links among the acts
(functions) themselves:

Utterance ‘ act 1
act 2
. act 3

Are other mappings also possible? For example, can a single utterance also be
used to perform acts that are not themselves related? If so, then the mapping
relationship might be more like this:

Utterance act 1
act 2
act 3
 Speech Act Theory 87
Such possibilities imply very different interpretive processes: the former, a
single link between form and function, with links among related functions;
the latter, multiple links between form and a number of independent func¬
tions. Thus, the existence of “one form for many functions” relationships not
only forces us to consider links among different speech acts themselves, but
also raises the possibility of different interpretive processes linking form and
function.
The opposite mapping problem - “many forms for one function” - also
creates dilemmas for the application of speech act theory to discourse. Com¬
plications can stem simply from the fact that it is discourse, rather than a
single sentence, that is said to have a function. For example, we are used to
the idea that a discourse itself has functions: stories are used for instruction
(Heath 1982), for involvement (Tannen 1984), for self-aggrandizement (Labov
1972b; Schiffrin 1984b), for socialization (Gees*and Michaels 1989). Further¬
more, speakers may very well orient toward discourse level goals (e.g. thematic
goals) even in the most minute details of individual utterances (e.g. Bamberg
1992). However, it need not follow that every linguistic feature or quality of
that discourse serves the same, single function, or that every detail has a speech
act function. Nor is it always easy to decide which particular discourse func¬
tions to assign to small details of utterances. Although we considered the
historical present tense (HP) in Irene’s story to have an evaluative function,
for example, Wolfson (1979) has argued instead that it is the switch between
the HP and the preterite that has a discourse function (to separate episodes).
Finally, there is sometimes a circular quality to arguments about discourse
function. Drawing again from our analysis of the HP in Irene’s story, we
claimed that events reported in the HP were being highlighted to relate the
mother’s persistence in offering Irene candy to Irene’s defiance of her mother.
We then claimed that, because it was this sequence of events that was being
evaluated, it was this sequence that had a conversational relevance. But had
other events been in the HP, we could just as easily have argued that it was
those events that had conversational relevance for the central speech act
function of Irene’s story. I threw it out the window (r), for example, is central
to Irene’s account because it shows her willingness to defy her mother’s
repeated offers of candy.
Another problem stemming from “many forms for one function” relation¬
ships is that language rarely serves just a single function (Jakobson 1960; see
chapters 2, 11). Thus, not only does a discourse itself probably have multiple
functions (i.e. a “many forms for many functions” relationship), but there is
also a good chance that individual utterances (or groups of utterances) within
a discourse themselves perform acts. This suggests that we could conceivably
end up finding hierarchies of speech acts, in which smaller speech acts would
nest within more global speech acts, e.g. we might say that the clauses in
Irene’s story are all assertions (representatives) that build her account.
The most general lesson to be drawn from all these specific problems is that
it is difficult to provide criteria allowing us to decide what counts (or doesn t
88 Approaches to Discourse Analysis
count) as an instance of a speech act in such a way that other investigators
would identify the act in the same way (see Kreckel 1981). This is often
known, in social science research, as problems of validity and reliability: do
our analytic categories correspond to similarities, and differences, among
entities in the real world? Would others agree with our analytic categories and
be able to discover them independently of our own efforts? Such problems are
troubling to all research and can potentially occur at many stages of an
analysis. Yet because the identification of utterances as actions is so important
to later stages of speech act approaches to discourse - to describing relation¬
ships among units and combinations of units into larger patterns - trying to
achieve validity and reliability at this first stage is especially important.
At the same time that we need to stress social scientific notions of reliability
and validity, it is also important to note the inherent futility (and to some
scholars, the foolhardiness) of trying to assign an understanding or function
to an utterance with which all would agree (cf. chapter 4). Thus, despite our
attempt to base our analysis of Y’want a piece of candy? on both the specific
details of what was said and the general knowledge responsible for interpret¬
ing what was said as action, we cannot really be sure that we have “correctly”
identified the speech act(s) performed by the utterance. Furthermore, we
cannot really be sure of what we mean (or want to mean) by “correct.” Many
analysts shy away from grounding correctness in what speakers themselves
would say they meant, i.e. the answer we would get if we were to ask Henry
what he intended to do with his words. One reason for this is that different
people tend to give very different answers to such questions. This discovery
has led some to doubt the possibility of speech act taxonomies at anything
other than the level of individual knowledge (Kreckel 1981; Taylor and Came¬
ron 1987). Another reason is that the aim of linguistic inquiry (even one
dealing openly with what sounds like psychological constructs of “wanting”
and so on) is generally not seen as accounting for what someone “really
means” (cf. Labov and Fanshel 1977). Rather, most analysts proceed in one
of two directions. They may work backwards from what is said to infer what
the possible meanings of those words could be and in which contexts. Or they
may attribute a hypothetical intention to a speaker: if S wants to do/mean X,
what are the possible ways that S might do so? (Bilmes 1985; Martinich 1984;
Recanati 1987). Given such analytical routes, we might then say that a “cor¬
rect result is one that allows Y want a piece of candyf to be used as question,
request, and offer, provides a description of the conditions under which it may
be so used, and explains why these are the conditions allowing these func¬
tions.

3.3.2 Sequences of speech acts

In section 3.2 we showed how we could analyze sequences of speech acts: we

viewed Irene’s No as a rejection of Henry’s offer, Zelda’s She’s on a diet as
 Speech Act Theory 89

an account for Irene’s rejection, and Irene’s story (prefaced by I’m on a diet)
as her own account for her rejection of Henry’s offer. This analysis not only
showed how our identification of an initial speech act defined our inventory
of sequential possibilities, but it highlighted problems stemming from unit size
and decisions about sequential appropriateness.
Let us focus first on the observation that, depending on how we identify an
initiating speech act, the length of a sequence can differ. We saw that sequen¬
ces initiated with a question can be closed with an answer, followed by
optional expansions. Sequences initiated with a request can be closed with a
compliance, although a non-compliance typically leads to an account (Davison
1975; Green 1975). Sequences initiated with an offer are similar to those
opened with a request. Although they can be completed with a second part if
that second part is an acceptance, refusals often le^i to third parts, i.e.
accounts. Thus, depending on how we identify an initial action, we end up
with a different view of the expected length of the sequence initiated by that
action.
It is important to note that more than mere sequence length is at issue here.
What holds a question-answer sequence together - the basis for its coherence
- is quite different from what holds an offer-rejection sequence together.
Whereas question-answer coherence is based at least partially on proposi¬
tional information, sequences initiated by offers are based on personal com¬
mitment toward action. If H accepts S’s offer, S is committed to do A (and H
is committed to allowing A to proceed); if H rejects S s offer, S has to alter a
prior course of proposed action. The role played by personal commitment
suggests that the reason why rejections to offers demand accounts has less to
do with the constitutive rules of those speech acts themselves, than with
what those rules imply about social relationships. Because speech act theory itself
offers little to say about social relationships, I reserve discussion of the social
coherence underlying speech act sequences for chapter 4. Here we can note,
however, that the coherence of the offer-rejection-account sequence is grounded
in the social meanings of these acts and their relationships to one another.
This suggests that the basis for sequential coherence between speech acts can
lie as much in the social and interactive world as in the cognitiveAvorld of
speech act categories and rules.

4 Speech act theory as an approach to discourse

In this chapter, we discussed some of the central ideas of speech act theory as
formulated by the philosophers Austin and Searle, and then applied these ideas
to a particular set of speech acts in a discourse. In this concluding section, I
briefly summarize how speech act theory provides an approach to discourse
analysis.
90 Approaches to Discourse Analysis

The essential insight of speech act theory is that language performs com¬
municative acts. In Searle’s (1969: 21) words:

The hypothesis that the speech act is the basic unit of communication,
taken together with the principle of expressibility [whatever can be
meant can be said], suggests that there are a series of analytic connections
between the notion of speech acts, what the speaker means, what the
sentence (or other linguistic element) uttered means, what the speaker
intends, what the hearer understands, and what the rules governing the
linguistic elements are.

Speech act theory, then, is basically concerned with what people “do” with
language - with the functions of language. Typically, the functions focused
upon are those akin to communicative intentions (the illocutionary force of
an utterance) that can be labelled (cf. that have a performative verb) and
realized in a single sentence. Even indirect speech acts (those that are per¬
formed “by way” of another act: Searle 1975) fall into this group: they are
drawn from the same labelled taxonomy as direct speech acts.
Language can be used for speech acts because people share rules that create
the acts: utterances “count as” successful and non-defective performances of
speech acts when they fulfill certain conditions. The rules and conditions draw
upon linguistic knowledge (e.g. the relationship between tense and the refer¬
ence time of an event) and knowledge about the world (e.g. that people may
be obliged to behave in certain ways) that allows certain linguistic devices to
indicate illocutionary force. These two bodies of knowledge, and how they
interact with one another, are assumed to be part of competence.
The conditions underlying and defining speech acts are central to speech act
theory: they are the basis for the way we recognize and classify speech acts
(and thus identify an utterance as a particular type of “unit”) and for the way
a single utterance can have more than one function (i.e. be more than one
“unit”). Note that the knowledge that participants use in linguistic exchanges
is thus relatively static knowledge: knowledge of what constitutes an act,
what type of act it is, and whether more than one act is involved in its
realization is brought “ready made” to each linguistic exchange. Yet such
knowledge is also critical to the ongoing processes of communication: it is by
identifying the units (acts) created by constitutive rules that communication
proceeds. Although what happens during such an exchange can help to fulfill
the conditions underlying a specific speech act, the circumstances of the actual
exchange do not fundamentally define or alter those conditions.
In sum, by focusing upon the meanings of utterances as acts, speech act
theory offers an approach to discourse analysis in which what is said is
chunked (or segmented) into units that have communicative functions that can
be identified and labelled. Although we can describe such acts in different ways
(e.g. as realizations of constitutive rules, as the product of form-function
relations, as the outcome of different textual and contextual conditions), the
 Speech Act Theory 91
import of such acts for discourse is that they both initiate and respond to other
acts. Acts specify (to a certain degree) what kind of response is expected: they
create options for a next utterance each time they are performed, and thus
provide a local, sequentially emergent basis for discourse. Since an utterance
can also perform more than one act at a time, a single utterance creates
different response options for a next utterance. Above I noted that what allows
us to identify what others are doing is our relatively static speech act knowl¬
edge. This is not to say, however, that what we know about speech acts prior
to any one particular linguistic exchange cannot alter the direction of an
exchange. Recall that it is our speech act knowledge that allows us to infer
not only that an interlocutor is doing something with words, but also that an
interlocutor is doing more than one thing at once with words. Mappings
between one form and multiple functions thus gives o^r exchanges a certain
degree of flexibility: if we do not respond to one possible speech act interpre¬
tation of what someone has said to us, we may respond to another. This
flexibility has an important analytical consequence: it means that a single
sequence of utterances may actually be the outcome of a fairly wide range of
different underlying functional relationships.

Exercises

1 People sometimes provide explicit speech act labels for their utterances:
we might think of this as a metaspeech act, as the speech act of "defining," or
as metapragmatics (cf. Lucy 1992). Below are examples of speakers using
several different speech act labels.

(1) During a speech in which he was discussing the United States' reaction to
the Iraqi take-over of Kuwait (October 1990), President George Bush said
the following:
Iraq will not be permitted to annex Kuwait.
That's not a threat,
it's not a boast...
That's just the way it's gonna be.
(2) Parents of a three-year-old are discussing the new shoes they have just
bought their son. Both parents had previously noted that because the
shoes have a lot of laces, they will be hard to pull off the son's feet.
FATHER: (a) You'll have to untie them.
MOTHER: (b) Oh 1 know, it's okay.
FATHER: (c) That's not a criticism, it's a reminder.
MOTHER: (d) Oh no, 1 like the shoes!
FATHER: (e) No, 1 mean you'll have to untie them.

For both of the examples, discuss the different conditions that would have to
hold for each of the acts that are explicitly named. Are the conditions related
92 Approaches to Discourse Analysis

to one another? Are there other labels that might have been assigned to these
acts? (Consider, for example, what act is conveyed in Bush's statement "That's
just the way it's gonna be.") What does the availability of multiple labels
suggest about our organization of speech act knowledge and the way we use
this knowledge during our communicative exchanges?
2 American children celebrate the holiday of Halloween by dressing in cos¬
tumes and going around to their neighbors' houses to collect candy. After
knocking on doors, and having the door opened, children say "Trick or
treat."(Berko-Gleason and Weintraub (1976), discusses how children are socialized
into this routine.) This formula is differently interpreted in different parts of
American society and these different interpretations require different respon¬
ses. The different interpretations are as follows:

(1) Give me a treat.

(2) Give me a treat, or I'll do a trick.
(3) If you give me a treat, I'll do a trick.
(4) Do you want a trick or a treat?

How would speech act theory explain these different interpretations? In what
way does the syntactic structure of "Trick or treat" and its intonation (usually
falling) convey these interpretations? (It may help to compare other "X or Y"
utterances, e.g. "Coffee or tea.")
3 Presented below are several different examples of interactions. Each con¬
tains an utterance that can be interpreted as doing more than one speech act.
(The target utterance in each case is marked with *.)

(1) *CUSTOMER: (a) Coffee to go?

SERVER: (b) Cream and sugar?
CUSTOMER: (c) Just cream.
SERVER: (d) /provides coffee/
CUSTOMER: (e) /pays/
(2) CUSTOMER: (a) Can 1 have an espresso?
SERVER: (b) You sure can!
(c) /prepares coffee/
(d) /provides coffee/
(3) A commuter is at the Metro station with a $20 bill, and no fare card. She
approaches a Metro employee.
*COMMUTER: (a) Do you know where 1 can get change for $20?
EMPLOYEE: (b) You'll have to go into a store or something
(c) There are plenty outside of the station.
COMMUTER: (d) Well there's really nothing near by.
(e) What should 1 do?
EMPLOYEE: <f) What you should do is check your money before you
leave home, and make sure you have the right change.
COMMUTER: (g) Well 1 was really in a hurry and just didn't have a chance.
(h) Anyway 1 thought 1 had it.
(i) Couldn't you just give me change if you have it?
^EMPLOYEE: (j) Why don't you rely on the goodwill of your fellow
riders?
 Speech Act Theory 93

(4) *librarian: (a) Can I help you?

patron: (b) Yeh maybe.
(c) I'm uh- I've been working up in New York City in the
theater doing acting and stage managing and one thing
and another.
(d) And I'm looking to try and get over to public relations
which is why I'm down here.
(5) Professors A and B are colleagues at the same university. A is on leave in
another city. B is in her university office, preparing for class. A calls Professor
B on the phone: B tells A that she has a few minutes to talk. After spending
a few minutes discussing an upcoming meeting, the following occurs:
*profa: (a) Would you like to call me back?
prof B: (b) No that's okay, I don't have that much more to say.
prof a: (c) No, I meant use university money instead of mine!
(d) I have some things I have to'ask yofl.
(6) Father of a six-year-old is on the phone with "Grandmom."
♦father: (a) Would you like to say hello to Grandmom now?
child: (b) No thank you.
(7) A and B are commuters on a daily train. A notices a newspaper on an
empty seat next to where B is sitting.
*a: (a) Is that your paper?
b: (b) Yeh, but you can have it.

Identify the speech acts that are being performed in the target utterances. How
does the context (including other utterances) influence your analyses? Are there
other (hypothetical) sequences in which the target utterances could perform
other speech acts? How would these sequences differ from the actual sequen¬
ces? Finally, is multifunctionality of utterances ever an issue for interlocutors?
Is multifunctionality solved (or exacerbated) by the fact that an utterance occurs
in a sequence of other utterances?

Notes

1 Many scholars have already found speech act theory to be an important source of
insight into discourse (e.g. Labov and Fanshel 1977; Sinclair and Coulthard 1975).
Others have made observations similar to the essential insight of speech act theory,
e.g. Halliday’s (1978) thesis that language is the realization of meanings. But just as
some have embraced the application of speech act theory to discourse, others have
been more reluctant to transfer insights. Taylor and Cameron (1987: chapter 3), for
example, have questioned the wisdom of relying so heavily on rules; Levinson (1983.
chapter 5) has doubted the possibility of specifying mapping relationships between
utterances and actions; Searle (1989) himself has suggested that discourse is more
readily viewed in terms of speaker goals than felicity conditions and rules. Despite
the interest and relevance of such a debate, it is important to note that I am neither
endorsing nor criticizing the use of speech act theory for discourse analysis: rather
my goal is a neutral discussion of that application that can then be compared with
the approaches created by other perspectives (part III).
94 Approaches to Discourse Analysis

2 Note that I am intentionally speaking of coherence rather than well-formedness. The

question of well-formedness is a complicated one in discourse analysis (cf. chapter
2). As we will see in succeeding chapters, not all approaches (even all structurally
derived approaches) hold the distinction between ill-formed and well-formed to be
a valid one to apply to discourse. Replacing the notion of well-formedness with
coherence not only allows for a more gradient view of sequential regularities, but it
also gives interpretive processes a greater analytical role in the differentiation of
sequences that “make sense” from those that do not.
3 Using intonation as a formal clue to the functional identification of Y’want a piece
of candy? as a question, however, brings up still other dilemmas. First, intonation is
not really linguistic form: Austin (1962: 74), for example, goes so far as to include
it in a list of items illustrating the context of utterance production. Thus, we might
be better off talking not of form per se, but in more general terms of linguistic
qualities of an utterance. Second, it is extremely difficult to say that intonations have
meanings that are independent of their contexts (e.g. Ladd 1978).
4 See chapter 5 for discussion of these questions as information-checking, rather
than information-seeking, questions. The main differences are the scope of what is
being questioned and the type of response sought: the information being sought is
not the completion of a proposition, but reception of a referent or proposition; the
response is not completion of a proposition but acknowledgment of information
status.
5 Just as different forms of a single speech act may vary depending on social relation¬
ship, so different speech acts are differently associated with what some researchers
have called the “social distance” variable. Wolfson (1988), for example, suggests
that both compliments and invitations are less likely among intimates and strangers
than among casual friends and acquaintances. Boxer (1993) suggests that indirect
complaints (unlike compliments and invitations), on the other hand, are not equally
likely among strangers and intimates.
6 Note, however, that depending on what is needed, by whom the need is felt, and to
whom the need is expressed, the more specific directive force of a “personal need”
statement can vary tremendously: I need these letters typed by 3 p.m. said by a boss
to a secretary is more likely to be interpreted as an order (a relatively non-negotiable
directive), whereas I need a vacation said by one spouse to another would receive
quite a different interpretation.
7 Note that we need not interpret Y’want a piece of candy? as an offer, e.g. if Henry
is not assumed to have candy available to him:

henry: Y’want a piece of candy?

irene: Yes.
henry: Well, it’s in the living room.
[or] Okay. I’ll put candy on the shopping list.

An example we consider later has the same possibilities:

cliff: Do you need help?

diane: Yes.
cliff: Why don’t you call the plumber then?

8 There are other relationships between directive and commissives that deserve con¬
sideration.
 Speech Act Theory 95
First, whether a single utterance serves as directive or commissive may depend
on the social identities of speaker and hearer. “Let me come with you,” for example,
is an offer if said by a parent to a three-year-old child. But said by a three-year-old
child to a parent who is about to leave the child with a babysitter, “Let me come
with you” is clearly a directive. Although these different interpretations rest upon
social identity, they also depend upon whom the act is deemed to benefit. In line
with this point, some “let” statements have multiple functions whose beneficial
value is not quite so clearcut: when a nurse tells a patient “Let’s get dressed,” for
example, one might argue that both directive and commissive are being realized -
and that both speaker and hearer will benefit.
Second, offers can be seen as pre-emptive versions of requests (Schegloff 1979b):
instead of waiting for H to ask S to do A for H, S can present H with the option
of actually doing A through an offer. Alternatively, H can try to elicit an offer of
A from S, rather than request that S do A, e.g. through a “pre-request.” Note that
.the view that S’s offer allows H to avoid a reqd'est coincides with our cultural
interpretation of offers as polite speech acts. By removing the need for H to make
a request by making an offer, S makes it possible for H to avoid an act that is face
threatening, i.e. H’s request would restrict S’s ability to decide upon his or her own
next action (thus threatening one’s negative face: Brown and Levinson 1979), as
well as do an act that benefits H rather than S (Goffman 1971b). Put another way,
by anticipating what another person wants, then one has allowed the other to avoid
having to impose upon one’s own course of action. Note that this reflects a general
civic belief that “good citizens” are supposed to avoid self-interest (e.g. avoid
making requests) and highlight altruism (e.g. make offers).
9 Note, also, that the request Can you tell me the time? can be easily analyzed as a request
for a verbal action (the provision of the time) or a request for physical action (displaying
one’s watch to provide the time) - as indicated by the substitutability of give for tell.
10 The relationship between questions and offers can be posed in terms similar to the
relationship between questions and requests, cf. discussions as to whether “can you
pass the salt” is both question and request.
11 Two points. First, one issue that we will not address in this chapter is the fact that
Zelda’s She’s on a diet is an account being offered for someone else (we consider
this in chapter 4). To do so would require us to consider other actions even more
abstract than question, request, and offer, e.g. giving solidarity, competing for the
floor. I am avoiding such analyses here in conformity with the guidelines of fairly
orthodox speech act theory, which focuses just on those speech acts whose intended
communicative (illocutionary) function can be directly performed (e.g. through a
performative verb) in a single utterance.
Second, we have already seen that Y’want a piece of candy? performs three acts,
all of which can initiate a sequence. Note, also, that this utterance opens a side
sequence (Jefferson 1972) that shifts the topic of the prior discussion (from a recent
funeral to candy; see example (21) in chapter 5). This suggests that Y’want a piece
of candy? initiates a new speech act sequence, rather than adds to or completes a
prior speech act sequence.
12 Semantic dependency is a very general property of questions and answers: as
Halliday and Hasan (1976) point out, ellipsis (i.e. recoverable deletions of text) is
a common way to show cohesion (semantic ties) across clauses in discourse.
13 Ethnic groups are typically defined as subgroups within a culture or society that are
differentiated by various complex criteria, including religious, ancestral, physical,
96 Approaches to Discourse Analysis

and/or linguistic characteristics. The importance assigned to ethnic distinctions,

however, is often economically and socially determined.
14 The way stories are fit into conversations - in terms of turn-taking adjustments,
topic, and so on - is an issue considered more by conversation analysts than speech
act analysts. See e.g. Jefferson (1978).
15 Evidence that compliance with a parent’s wishes is desired behavior (even for adult
daughters and sons) is found elsewhere in my interviews with Irene, Zelda, and
Henry. Henry, for example, boasts that his children “did what we wanted!” when
they married within their own religion; Irene complains that children do not always
“listen” nowadays. Zelda remarks that one of her daughters-in-law and her son
followed her advice to call their in-laws “Mom and Dad ; she labels the situation
with her other daughter-in-law (who does not call her Mom ) as a sore spot
(see example (13) in chapter 8).
16 Note, also, that by reporting a deception, Irene alters the relationship she has with
her audience: she creates a collusion and pulls the audience into a secret. Else¬
where (Schiffrin 1984b), I have discussed how secrets create bonds of loyalty and
solidarity: thus, the fact that Irene reveals a secret to Henry can be seen as a way
to provide compensatory strength to their relationship — strength possibly threat¬
ened by her having turned down his offer.
17 We will meet these problems again in chapter 11 when we discuss communication
and intersubjectivity, although we will discuss them there in relation to whether
participants in an exchange agree on the meaning of what is said, rather than
whether analysts of an exchange agree on its meaning.
4 Interactional Sociolinguistics

A a

1 Introduction

The approach to discourse that I am calling “interactional sociolinguistics”

has the most diverse disciplinary origins of those discussed in this book: it is
based in anthropology, sociology, and linguistics, and shares the concerns of
all three fields with culture, society, and language. The contribution to inter¬
actional sociolinguistics made by the linguistic anthropologist John Gumperz
provides an understanding of how people may share grammatical knowledge
of a language, but differently contextualize what is said - such that very
different messages are produced and understood. The contribution made by
the sociologist Erving Goffman provides a description of how language is
situated in particular circumstances of social life, and how it reflects, and adds,
meaning and structure in those circumstances. The ideas of both these scholars
have been applied extensively within linguistics, e.g. by Brown and Levinson
(1987), Schiffrin (1987a), and Tannen (1989a).
I begin in section 2 by presenting some of the concepts and ideas basic to
interactional sociolinguistics. Despite the different disciplinary starting points
of this approach, there are several basic beliefs about language, context, and
the interaction of self and other that provide unity. My choice of a particular
phenomenon upon which to base the sample analysis in section 3 stems from
these central concerns: although I begin by focusing on the very same ques¬
tion-answer exchange discussed in chapter 3, I use that exchange as a way to
consider how one particular self/other alignment (“speaking for another”)
serves as a discourse strategy and I show that different ways of contextualizing
this strategy provide it with different interactional meanings. Section 4 sum¬
marizes the critical features of the interactional sociolinguistic approach to
discourse.
98 Approaches to Discourse Analysis

2 Defining interactional sociolinguistics

In this section, I describe the basic ideas of interactional sociolinguistics. I

begin with the work of Gumperz (2.1) and go on to Goffman (2.2); in section
2.3.1 briefly summarize how the ideas of these two scholars can be combined.

2.1 The contribution of anthropology: Gumperz

In the introduction to his most recent collection of essays (Discourse

Strategies), Gumperz (1982a: vii) states that he “seeks to develop interpretive
sociolinguistic approaches to the analysis of real time processes in face to face
encounters.” After briefly describing some of Gumperz’s work prior to the
1982a collection, I describe the concepts and methods that Gumperz has
developed for the achievement of his goal.
Dil (1971) is a collection of Gumperz’s essays through 1971: the dual focus
of this volume - language and dialect diversity, language usage and social
interaction - reflects the themes that continue (and become even more unified)
in the 1982a collection. The research reported in Dil (1971) is all grounded
in an assumption basic to social and cultural anthropology: the meaning,
structure, and use of language is socially and culturally relative. The import¬
ance of this assumption is illustrated through studies focusing on a variety of
different issues. For example, Gumperz’s work in India - on regional and
social language differences, on Hindi-Punjabi code-switching, and on linguistic
convergence - focuses not just on linguistic structure, but also on how those
structures become part of the verbal repertoires of interacting social groups.
The three-part focus on language structure, use, and social groups also under¬
lies Gumperz’s (Dil 1971: 114) classic definition of the speech community as
“any human aggregate characterized by regular and frequent interaction by
means of a shared body of verbal signs and set off from similar aggregates by
significant differences in language usage.”
Despite the social and cultural emphasis of Gumperz’s early work, individual
expression (what seems later to be seen as “strategy”) also finds a place in
this research. In his studies of code switching (Dil 1971), for example, Gum¬
perz defines two types of switching from one language variety to another. First
is situational code switching: people may switch in accord with “clear changes
in . . . participants’ definition of each other’s rights and obligations” (Dil 1971:
294) . Second is metaphorical code switching: people may switch varieties
within a single situation just to convey a different view of that situation and
their relationship, such that “the language switch here relates to particular
kinds of topics or subject matters” and is used “in the enactment of two or
more different relationships among the same set of individuals” (Dil 1971:
295) .
 Interactional Sociolinguistics 99

The connections among culture, society, individual, and code are developed
in Discourse Strategies, the essays which (as noted above) seek to develop
“interpretive sociolinguistic approaches to the analysis of real time processes
in face to face encounters.” In the first article of this collection, Gumperz
(1982a: 12) points out the impact of structural linguistics (as formulated by
Saussure (1959) and applied by anthropological linguists working with speakers
of languages other than their own) on our understanding of culture and
cognition: “structural analysis furnished empirical evidence for the contention
that human cognition is significantly affected by historical forces. . . . What
we perceive and retain in our mind is a function of our culturally determined
predisposition to perceive and assimilate.” Cognition and language, then, are
affected by social and cultural forces: the way we behave and express ourselves
in relation to a linguistic code and the underlying categories of the code itself
are open to external influence. In order to understand these effects, what is
thus needed is a “general theory of verbal communication which integrates
what we know about grammar, culture and interactive conventions into a
single overall framework of concepts and analytical procedures” (1982a: 4).
The theory of verbal communication proposed by Gumperz requires the
addition of concepts and analytical procedures that build upon his earlier ideas
about culture, society, language, and the self. One new construct is contex-
tualization cue. As I explain in a moment, this is related to two other concepts:
contextual presupposition and situated inference.
Recall Gumperz’s (1982a: 12) observation that “what we perceive and retain
in our mind is a function of our culturally determined predisposition to
perceive and assimilate.” One feature of modern, urban societies is their social
and cultural heterogeneity, and one effect of such heterogeneity is that people
from very different cultural and linguistic backgrounds come into contact with
one another. Such contacts can lead to communicative difficulties precisely
because of the fact that people’s perceptions of similarities and differences in
the world are culturally bound. As Gumperz notes, however, it is not just the
core grammar of a language that exhibits such differences - despite the fact
that grammatical (syntactic, phonological, semantic) differences in one’s way
of speaking are often the easiest to note. Just as pervasive a source of com¬
municative difficulty are differences in what Saussure called the marginal
features of language: “signalling mechanisms such as intonation, speech
rhythm, and choice among lexical, phonetic, and syntactic options . . . said to
affect the expressive quality of a message but not its basic meaning” (Gumperz
1982a: 16). Gumperz’s studies of both interracial (blacks and whites in the
United States) and interethnic (Indians and British in England) settings show
that it is precisely differences in the marginal features of language that can
cause misunderstandings, lead to the formation of racial and ethnic stereo¬
types, and contribute to inequalities in power and status.
The signalling mechanisms just described are what Gumperz calls contextu-
alization cues: aspects of language and behavior (verbal and nonverbal signs)
that relate what is said to the contextual knowledge (including knowledge of
100 Approaches to Discourse Analysis

particular activity types: cf. frames; Goffman 1974) that contributes to the
presuppositions necessary to the accurate inferencing of what is meant (includ¬
ing, but not limited to, the illocutionary force). Note how the two other
constructs mentioned above (contextual presuppositions, situated inferences)
are part of this definition. Contextual presuppositions are a type of assumed
background knowledge that allows the inferencing (during the course of an
interaction) of two levels of meanings that are themselves related. One level
is the communicative activity type (cf. Hymes’s speech event, chapter 5):
whether one is teasing, lecturing, chatting, and so on. The second level is the
particular illocutionary act that the speaker intends. Crucially, the interpreta¬
tion of the illocutionary act (with the help of contextualization cues) is de¬
pendent on the use of the “frame” whose inference is also allowed by
contextualization cues.
Let us take an example (from Gumperz 1982a: 147). The example illustrates
the use of rising intonation as a contextualization cue.

teacher: James, what does this word say?

james: I don’t know.
teacher: Well, if you don’t want to try someone else will. Freddy?
freddy: Is that a “p” or a “b”?
teacher: (encouragingly) It’s a “p.”
freddy: Pen.

The teacher’s response (Well, if you don’t want to try someone else will)
indicates her interpretation of Freddy’s I don’t know not only in terms of its
literal meaning, but also as an indication that Freddy did not wish to try to
answer the question. Gumperz notes, however, that I don’t know had final
rising intonation, understood in the African American community of which
Freddy was a member as conveying a desire for encouragement (cf. “I need
some encouragement”). Thus we might say that the teacher did not retrieve
the contextual presuppositions needed to accurately interpret Freddy’s mess¬
age from his use of rising intonation.
As illustrated in the above example, Gumperz’s studies show that contextu¬
alization cues can affect the basic meaning of a message. To take another brief
example: conversational code-switching (i.e. metaphorical switching) can be a
contextualization cue because it “generates the presuppositions in terms of
which the content of what is said is decoded” (Gumperz 1982a: 98) and in
terms of which a hearer can infer a speaker’s communicative intention. Al¬
though such cues are used habitually and automatically by members of a
particular social group, they are almost never consciously noted or assigned
conventional meanings: rather, they signal the speaker’s implicit definition of
the situation, and more importantly, how the propositional content of talk is
to be understood.
As I noted above, when listeners share speakers’ contextualization cues,
subsequent interactions proceed smoothly. The methodological consequence
 Interactional Sociolinguistics 101

of this is that one can discover shared meaning by investigating the process of
interaction itself, i.e. by using the reaction that an utterance evokes as evidence
of whether interpretive conventions were shared (Gumperz 1982a: 5). It is
because contextualization cues are learned through long periods of close,
face-to-face contact, however, that many people in modern, culturally diverse,
socially heterogeneous societies are likely to interact without benefit of shared
cues. And it is the analysis of misunderstandings between people from different
groups - people who do not share contextualization cues - that can provide
the most telling evidence that such cues are at work. Furthermore, such
misunderstandings can have devastating social consequences for members of
minority groups who are denied access to valued resources, based partially
(but not totally) on the inability of those in control of crucial gatekeeping
transactions to accurately use others’ contextualization cues as a basis from
which to infer their intended meanings. A
We saw above that the individual entered Gumperz’s earlier work primarily
in social and expressive capacities - as a member of a social group, as the
purveyor and creator of social meaning. In Gumperz’s sociolinguistics of
interpersonal communication, people have an added role: they have a cognitive
capacity to make inferences. Contextualization cues are critical to this process,
for they allow conversationalists to “rely on indirect inferences which build
on background assumptions about context, interactive goals and interpersonal
relations to derive frames in terms of which they can interpret what is going
on” (Gumperz 1982a: 2). Also necessary to this process of situated inference
is the maintenance of conversational involvement - since “we know that
understanding presupposes the ability to attract and sustain others’ attention”
(Gumperz 1982a: 4). Thus, a general theory of discourse strategies must begin
by: “specifying the linguistic and socio-cultural knowledge that needs to be
shared if conversational involvement is to be maintained, and then go on to
deal with what it is about the nature of conversational inference that makes
for cultural, subcultural and situational specificity of interpretation” (Gum¬
perz 1982a: 3).
Before summarizing, it is important to note that although some of Gumperz’s
concepts seem rooted in the individual (inference, involvement), these notions
are actually grounded in a view of the self and what it does (e.g. make
inferences, become involved) as a member of a social and cultural group and
as a participant in the social construction of meaning. For example, Gumperz
(1982a: 209) reformulates Flymes’s (1974) concept of communicative com¬
petence in interactional terms, to include “the knowledge of linguistic and
related communicative conventions that speakers must have to create and
sustain conversational cooperation” (see also Gumperz 1985). And even in the
complex question of speakers’ internal differentiation of two linguistic sys¬
tems, Gumperz (1982a: 99) argues that “effective speaking presupposes socio-
linguistically based inferences about where systemic boundaries lie” and that
“members have their own socially defined notions of code or grammatical
system” (emphases added).
102 Approaches to Discourse Analysis

In sum, the key to Gumperz’s sociolinguistics of interpersonal communica¬

tion is a view of language as a socially and culturally constructed symbol
system that is used in ways that reflect macro-level social meanings (e.g. group
identity, status differences) and create micro-level social meanings (i.e. what
one is saying and doing at a moment in time). Speakers are members of social
and cultural groups: the way we use language not only reflects our group-
based identity but also provides continual indices as to who we are, what we
want to communicate, and how we know how to do so. The ability to produce
and understand these indexical processes as they occur in, and are influenced
by, local contexts is part of our communicative competence. As we see in the
next section, the work of Erving Goffman also focuses upon situated knowl¬
edge, the self, and social context - but in different ways and with different
emphasis.

2.2 The contribution of sociology: Goffman

Also contributing to the development of interactional sociolinguistics is the

work of Erving Goffman. Although Goffman does not analyze language per
se, his focus on social interaction complements Gumperz’s focus on situated
inference. Goffman places language (and other sign systems) in the very same
social and interpersonal contexts that provide the presuppositions Gumperz
finds to be an important background for the decoding of meaning. What
Goffman adds is an understanding of the forms and meanings of those contexts
that allows us to more fully identify and appreciate the contextual presupposi¬
tions that figure in hearers’ inferences of speakers’ meaning. Putting the work
of the two scholars together, we can come up with a richly textured view of
the contexts in which inferences about speakers’ meaning are situated.
Goffman’s sociology develops the ideas of several classic sociological theor¬
ists, and applies them to a domain of social life whose structural complexities
had (before Goffman’s work) gone largely unnoticed: face-to-face social inter¬
action. Building upon the work of Emile Durkheim on social facts (1895), and
primitive religion (1893), and the social psychologist George Herbert Mead
(1934) on the formation of the self, Goffman argues that the self is a social
construction or, more specifically, an interactive construction. One way of
viewing the self as a public construction is through the notion of “face”, i.e.
“the positive social value a person effectively claims for himself by the line
others assume he has taken during a particular contact.” As Goffman 1967a:
5) notes, face is “something that is diffusely located in the flow of events in
the encounter and becomes manifest only when these events are read and
interpreted for the appraisals expressed in them.” The maintenance of both
self and face is built into the fabric of social interaction (e.g. “the maintenance
of face is a condition of interaction, not its objective”; Goffman 1967a: 12;
also pp. 7, 11, 39-40) and the complementary needs of self and other
(Goffman 1963: 16; 1967b: 85). One contribution to the maintenance of face
 Interactional Sociolinguistics 103

is interpersonal ritual, both avoidance rituals (“those forms of deference which

lead the actor to keep at a distance from the recipient”; Goffman 1967b: 62)
and presentational rituals (“acts through which the individual makes specific
attestations to recipients concerning how he regards them”; Goffman 1967b:
71). Another contribution to the maintenance of face (and the presentation of
self more generally) is the material resources made available through the social
establishments and institutions in which people find themselves - resources
that not only can be used to symbolize certain favored aspects of self (Goffman
1959) or to show distance from an institutionally allocated role (Goffman
1963), but can physically facilitate the division of self into a public character
and a more private performer (Goffman 1959: chapter 3 on front and back
regions).
Goffman’s analyses of the relationship between interpersonal meanings and
social structure are balanced by careful attention to both the symbolic value
of what is said and done and the more abstract forms of social life (a duality
that seems to be inherited from Georg SimmePs (1911) distinction between
social form and meaning; see e.g. Goffman 1971a). It is in both form and
meaning that we can find the sort of detailed attention to context that I noted
above - detail that allows us to more fully understand the contextual presup¬
positions that figure in hearers’ inferences of speakers’ meaning.
Let me give two examples. We noted above Gumperz’s point that situated
inferences require interpersonal involvement, and that the maintenance of
involvement requires that certain linguistic and sociocultural knowledge needs
to be shared. Goffman’s (1963: particularly chapters 3, 4, and 10) study of
behavior in public places is relevant to Gumperz’s concern with both the
creation and the effect of involvement. What Goffman focuses on is the social
organization of involvement; he describes the way different social occasions
(and different phases of occasions) can create a wide array of expectations for
the display of involvement (e.g. access rituals such as greetings require height¬
ened involvement: Goffman 1971a; Schiffrin 1977), such that the very pro¬
cesses of both “being” involved and “showing” involvement are themselves
socially situated. This is relevant for Gumperz’s concern with involvement:
since interactions impose their own rules of involvement, we must see those
inferences that are based on involvement as also governed by broader rules of
social engagement.
The second example concerns the contextual presuppositions that underlie
hearers’ inferences of speakers’ meanings; here, let me focus on the notion of
interpretive frame. One way of describing Gumperz contextualization cues is
as a framing device, i.e. they indicate the frame (e.g. serious, joking, business,
chat) in which an utterance should be interpreted (cf. Tannen 1984, on meta¬
message). Goffman’s (1974) work on frame analysis — the frames through
which people structure experience — shows how the organization of framing
activity is itself socially situated. Thus, again, we can see Goffman s work as
providing an elaboration of the contextual presuppositions that people both
use and construct during the inferencing process, and as offering a view of the
104 Approaches to Discourse Analysis

means by which those presuppositions are externally constructed and impose

external constraints on the ways in which we understand messages.
Before I close, it is important to note that Gumperz and Goffman have
publically drawn upon each other’s work — in ways that support the sorts of
connections I am establishing here. Goffman’s more recent work on the self
(1974; 1979) builds upon his earlier (1959) division (between character and
performer) to locate the self within a participation framework - a set of
positions that individuals within perceptual range of an utterance may take in
relation to that utterance. Goffman differentiates four positions, or participa¬
tion statuses: animator, author, figure, and principal. Although these positions
can be filled by different people, a single individual can also fill different
participation slots: to simplify a bit, an animator produces talk, an author
creates talk, a figure is portrayed through talk, and a principal is responsible
for talk. Each position within a participation framework is associated with
codified and normatively specified conduct (Goffman 1981a: 3), such that our
recognition of shifts among animators, figures, authors, and principals is
facilitated by our normative expectations about the conduct appropriate for
each position. As I note in Schiffrin (1990c), however, the concepts of frame
(Goffman 1974) and footing (Goffman 1981c) provide two additional links
between participation frameworks and social interaction. Frames are the or¬
ganizational and interactional principles by which situations are defined and
sustained as experiences (Goffman 1974); footing concerns “the alignments
we take up to ourselves and the others present as expressed in the way we
manage the production or reception of an utterance” (Goffman 1981c: 128;
my emphasis). As Goffman (1981c: 126-7) further notes, what indicates shifts
in footing and alignment are not just the way we manage the production of
an utterance, but also the kinds of devices identified by Gumperz as contex-
tualization cues. This means that sociolinguists “can be looked to for help in
the study of footing” (p. 128). But there is also help that sociolinguists can
get from sociological analyses of footing: “if [sociolinguists] are to compete
in this heretofore literary and psychological area, then presumably they must
find a structural means of doing so . . . the structural underpinnings of changes
in footing” (p. 128). Thus, what Gumperz adds to Goffman’s dissection of the
self is analysis of some of the devices that convey changes in footing, and a
view of how these aspects of the production format of an utterance allow the
situated inference of a new participant alignment. And, again, what Goffman’s
work adds to Gumperz’s sociolinguistics of interpersonal communication is a
more elaborated view (whether that elaboration is of meaning or form) of
what “in” a context can provide a situated presupposition, and of the complex
organizational and experiential frames that can be signalled through contextu-
alization cues.
Two final points about Goffman’s contribution to interactional sociolinguist¬
ics. First, although it may already be obvious from our discussion, Goffman’s
view of the self (like Gumperz’s) is not grounded in individual psychology.
Nor do psychological constructs (e.g. goals, emotions) enter Goffman’s view
 Interactional Sociolinguistics 105

of interaction. The study of interaction, for example, is a study not of motives,

but of rules:

to study face-saving is to study the traffic rules of social interaction; one

learns about the code the person adheres to in his movement across the
paths and designs of others, but not where he is going, or why he wants
to get there. One does not even learn why he is ready to follow the code,
for a large number of different motives can equally lead him to do so.

Second, Goffman makes a distinction between information that is given inten¬

tionally through language and information that is given off unintentionally
through expression (Goffman 1959; 1963: 13-16). Yet, just as the distinction
between core and marginal features evaporates once their significance as
indicators of contextual presuppositions is recognized (Gumperz), so too does
the analytical distinction between communication and expression evaporate in
practice. Not only is it the case that “every linguistic message carries some
expressive information, namely that the sender is sending messages,” but “most
concrete messages combine linguistic and expressive components, the propor¬
tion of each differing widely from message to message” (Goffman 1963: 16).
In sum, Goffman’s focus on social interaction complements Gumperz’s focus
on situated inference: Goffman describes the form and meaning of the social
and interpersonal contexts that provide presuppositions for the decoding of
meaning. The understanding of those contexts can allow us to more fully
identify the contextual presuppositions that figure in hearers’ inferences of
speakers’ meaning.

2.3 Language, culture, and society as “situated”

Despite the different sets of interests reviewed above - one stemming from
concerns about language and culture, the other from concerns about the self
and society - there are two central issues underlying the work of Gumperz
and Goffman that provide a unity to interactional sociolinguistics: the inter¬
action between self and other, and context.
Gumperz’s work focuses on how interpretations of context are critical to the
communication of information and to another’s understanding of a speaker s
intention and/or discourse strategy; Goffman’s work focuses on how the or¬
ganization of social life (in institutions, interactions, and so on) provides
contexts in which both the conduct of self and communication with another
can be “made sense of” (both by those co-present in an interaction and by
outside analysts). The work of both scholars also provides a view of language
as indexical to a social world: for Gumperz, language is an index to the
background cultural understandings that provide hidden - but nevertheless
critical — knowledge about how to make inferences about what is meant
through an utterance; for Goffman, language is one of a number of symbolic
106 Approaches to Discourse Analysis

resources that provide an index to the social identities and relationships being
continually constructed during interaction. Finally, both scholars allow lan¬
guage to have more active a role in creating a world than is perhaps suggested
by the term “index”: contextualization cues can alter not only the meaning of
a message, but the participation framework of talk - such that both different
intentions and different selves and others can be displayed through subtle
changes in the way utterances are presented. Thus, the role of language in
context, and the way it provides a path for self-other communication, is
basically similar for both scholars.
As I illustrate in the sample analysis in the next section, the work of Gumperz
and Goffman can be jointly applied to the meaning of utterances; the result is
a richly textured view of the contexts in which inferences about speakers’
meaning are situated.

3 Sample analysis: “speaking for another”

Interactional sociolinguists always draw upon naturally occurring interactions

for data. Many scholars follow the tradition initiated by Gumperz and focus
upon language used by speakers from different cultural backgrounds, often in
institutional settings in which those backgrounds (and the difference in com¬
municative style for which they may be responsible) have important long-term
consequences (see, for example, the papers collected in Gumperz 1982b).
Interactional sociolinguists also draw upon naturally occurring conversations
among friends (Tannen 1984). Finally, interactional sociolinguists pay a great
deal of attention to transcription of features of talk likely to serve as contextu¬
alization cues (see appendix 2).
The data used here are somewhat different: they are drawn from the
same sociolinguistic interview that was the base for our speech act analysis.
These data are well suited for the way that I illustrate an interactional ap¬
proach, however: I recorded a good amount of speech from the participants,
spent a number of years analyzing their discourse, and was myself a partici¬
pant in the interactions (Schiffrin 1987a). Thus, I was able to incorporate
contextual knowledge - necessary to interactional sociolinguistics - into my
analysis.
Before I begin, I want to anticipate the direction the analysis will take. My
initial focus is a single utterance from one speaker. At one point during an
interview, Zelda said She’s on a diet about Irene to Henry and me. Recall
(chapter 3) that this utterance expands Irene’s own answer (No) to Henry’s
question (Y’want a piece of candy?) and provides an account for Irene’s
non-compliance with Henry’s offer. In contrast to the speech act analysis in
chapter 3, I focus here on the participation framework created when Zelda
makes a contribution relevant to Irene’s exchange with Henry: in section 3.1,
 Interactional Sociolinguistics 107

I suggest that this utterance is an act in which one person “speaks for”
another. Not only is this not the sort of act that would be discussed by speech
act theorists, but the issues raised by an analysis of participation framework,
and the methodology required to substantiate the interactional meanings of
such acts relative to their participation framework, are quite different. In
section 3.2, I contextualize the act of “speaking for another” as a way of
uncovering its interactional meaning(s). We see that speaking for another can
display at least two different participant alignments (3.2.1) within a particular
sequence of interactional moves (3.2.2), and that these alignments are associ¬
ated with gender identities (3.2.3). These analyses thus view speaking for
another as a discourse strategy that is differently used by males and females
to create either solidarity or distance. Speaking for another can also be said
to serve as a contextualization cue through which participants signal identity
and alignment. Finally, I briefly consider how this particular act is one of a
“type” of acts whereby one person takes the role of another (3.3).
Note that I am intentionally not starting my analysis by focusing on either
a particular linguistic form or problem. Rather, as noted above, I start by
noticing a particular utterance (Zelda’s She’s on a diet) in an interaction. I
continue by trying to uncover the situated meaning of this utterance through
comparisons with other utterances and interactions.
At one level of analysis, the utterance to be focused upon has situated
meaning as an account for a rejected offer issued through a question (chapter
3). But the characterization of She’s on a diet in this chapter provides an
analysis of the action sequence in which it is situated (i.e. question-answer-
expansion, offer-rejection-account) from a perspective very different from
that of speech act theory. Rather than focus on how individual acts are
governed by autonomously functioning, static rules, we focus upon the social
meaning of conduct (doing an act of a particular kind) and of the interaction
that hinges upon such conduct. We see that who participates in an action
sequence - and how they do so - is a matter of interactional significance at
several different levels of analysis.

3.1 Speaking for another: “She’s on a diet”

In chapter 3 we spent a great deal of time analyzing the speech acts in a

particular interchange. I begin with a segment of that same interchange -
Zelda’s remark She’s on a diet - reproduced as line (c) in (1) below:

(1) HENRY: (a) Y’want a piece

IRENE: (b) No.
zelda: (c) She’s on a
DEBBY: (d) ' Who’s not on [a diet.
IRENE: (e) [I’m on-
I’m on a diet
108 Approaches to Discourse Analysis

(f) and my mother [buys-=

zelda: (g) [You’re not!
IRENE: (h) = my [mother buys these mints.=
DEBBY: (i) [Oh yes I amhhhh! 7
ZELDA: (j) Oh yeh.

In chapter 3, we said that She’s on a diet is both an expansion (a sequen¬

tially dependent unit which adds information supplementary to a prior unit)
and an account (an explanation for Irene’s rejection of Henry’s offer). Both
expansions and accounts are relational acts: expansions add information to
a prior act, and accounts provide reasons and/or motivations for a prior
act that has been marked (e.g. inappropriate, impolite, insulting) in some
way.
As noted above, I focus here on a very different aspect of Zelda’s remark:
its alteration of the participation framework of talk. Recall (2.2) that a
participation framework is the set of positions which individuals within per¬
ceptual range of an utterance may take in relation to that utterance; in brief,
the way that speaker and hearer are related to their utterances and to one
another. Although the interactional meanings of speaking for another are by
no means stable or determinate, there are some stable (and identificatory)
features of this act. One key feature is that it involves three participant roles.
We might label these roles using the relatively standard terms associated with
the code model of communication (see chapter 11): a sender (who intends to
send a message), a recipient (the intended target of the sender’s message) and
a spokesperson (who mediates in the transmission of the message from sender
to receiver). But Goffman’s analysis of participation framework allows us to
be more specific about these roles. Since one person (the spokesperson) pro¬
duces a message whose content is the responsibility of another (the sender),
we might say that one person is acting as animator for another person who is
in the principal role. Although we may still speak of B as the spokesperson,
and A as the spoken-for, then, the ways in which the two parties can be said
to be “speaking” involve very different notions of selfhood. The spokesperson
draws upon the mechanical aspects of an animator; the spoken-for, the more
moral aspects of a principal. Thus, when Zelda says She’s on a diet, she is an
animator for Irene’s principal. Zelda is reporting something about Irene that
Irene could have said herself (being on a diet is personal information) and to
whose sentiments (or content) Irene is committed. Furthermore, Zelda’s utter¬
ance is in a sequential position that Irene could have occupied herself (people
often provide their own accounts when they reject an offer) and whose main
relevance bears on the exchange between Irene and Henry (Irene has turned
down Henry’s offer).
In the next section, we see that the way She’s on a diet alters the participation
framework of ongoing talk is tied to the social relationship among participants
(3.2.1), the social acts being performed (3.2.2), and the gender identities of
participants (3.2.3).
 Interactional Sociolinguistics 109

3.2 “Speaking for another” as situated meaning

One of the key insights of interactional sociolinguistics is that meaning is

situated (section 2). As shown by Bennett (1978) and Tannen (1989b), for
example, the turn-taking structure during which one person’s utterance (or
part thereof) is simultaneous with another’s can be labelled in two quite
different Ways: it can be labelled as “interruption” (with negative connota¬
tions) or “overlap” (with neutral or even positive connotations). Which label
it receives (i.e. which meta-message is conveyed) depends on how that turn¬
taking structure is contextualized by speech activity, participant, and so on.
“Speaking for another” is an act whose meaning is also interactionally situ¬
ated. This section examines three different levels $.t which the meaning of
speaking for another can be located. I begin with the temporary interactional
alignments negotiated during talk in order to view speaking for another as a
discourse strategy that is used to create either solidarity or distance (3.2.1).

3.2.1 Alignments: “chipping in” or “butting in”?

Speaking for another has little inherent meaning in and of itself: it can be
interpreted in either a positive sense as “chipping in” or a negative sense as
“butting in.” The former gloss is consistent with an interpretation of this act
as a display of positive politeness (Brown and Levinson 1987) and as a
presentational ritual (Goffman 1967b): one shares so much with another that
she is able to take her position in conversation (cf. “you know me so well you
can read my mind!”). The latter gloss, on the other hand, is consistent with
an interpretation of the act as a violation of negative face (Brown and Levinson
1987) and as a violation of an avoidance ritual: one is so invasive of another
that she is unable to allow her to maintain her own position in conversation
(cf. “you’re always putting words in my mouth,” “you don’t let me get a word
in edgewise”). Thus, speaking for another can be seen as either deferential or
demeaning to the one being spoken for.
Here I will build upon interactional assumptions about social meaning to
argue that Zelda’s She’s on a diet is both intended and interpreted in a positive
way. Before we discuss the evidence for my belief, however, it is helpful to
note some of the more general contextual factors that figure in such decisions
— factors having to do with social relationship, social identities, speech act,
and speech event (cf. chapter 5). (In section 3.2.3, we see some acts of
“speaking for” among the same four participants that seem to be more intru¬
sively interpreted as “butting in.”)
Although space prevents us from a full examination of the interactive
meanings of speaking for others, this act seems to be interpreted depending
upon how the transfer of responsibility for speaking is achieved: does the
spokesperson “take” responsibility or is it “given” to her? Sometimes the right
110 Approaches to Discourse Analysis

to make such a transfer is institutionally allocated, such that its meanings can
be partially derived from institutionally sanctioned roles that may either free
one from the need to speak for oneself or force another to accede her own
speaking rights. To take some simple examples, I recently received a phone
call in which a receptionist, referring to her employer in the third person, said
“Dr Robinson is calling you. He’ll be with you in a moment.” Part of the
employee’s institutionalized responsibility was thus to speak for another (her
employer) who was freed from the need to speak for himself. In different
circumstances, however, using a third party to initiate (or even sustain) a
phone call for oneself can reflect quite differently upon the status of the
spoken-for, as when, for example, parents arrange play dates for their children
who do not yet have the communicative competence to do so themselves.
Even during institutionally based occasions of talk, however, the meanings
of speaking for another can vary. Let us take a hypothetical example. Suppose
that a person who has just immigrated to another country is filling out papers
and answering questions during a gatekeeping encounter with authorities, and
cannot understand enough of the indigenous language to answer a particular
question from an interviewer. Another person (e.g. an official translator, an
accompanying relative, or a passing stranger) may become a spokesperson.
Participants’ interpretation of this act can vary depending on how respons¬
ibility to speak for another was transferred, e.g. did the spokesperson offer to
help or intervene, did the spokesperson agree to a request from another (either
the interviewee or the interviewer), and so on. Participant point of view may
also vary, e.g. the spokesperson may feel a sense of solidarity with the spoken-
for person that is not at all shared. Thus, even a single act of speaking for
another, performed by those in relatively institutionally allocated roles, can
have a range of different meanings.
The range of interpretations that can hold for an act of speaking for another
increases during conversation - when acts are bound not only (or not even)
to institutional status and role, but to interactional positioning and participant
footing. Because participant roles shift during conversation, the right to either
take or abdicate responsibility for one’s words also shifts. What this means is
that the interactional meaning of speaking for another can be altered depend¬
ing on current perceptions of alignments, such that speaking for another
during a conversation can just as easily be positively or negatively glossed (as
noted above). Furthermore, since social relationships are also reinforced (if
not even created) during conversation, speaking for another during conversa¬
tion can have not only local interactive meaning, but also broader implications
about one’s own (and the other’s) rights, privileges, and responsibilities.
We are now ready to consider the interactional meanings of She’s on a diet.
I start with some information about the relationships among Henry, Zelda,
and Irene. Full information of this sort (i.e. about social identities and rela¬
tionships, and about the social, cultural, and individual meanings of actions)
is available only through long-term participant observation in a community
(the sort of methodology typical of the ethnography of communication; see
 Interactional Sociolinguistics 111

chapter 5). Instead of that, I will offer here only knowledge that is based on
my sociolinguistic fieldwork with Henry, Zelda, and Irene (a method of
collecting data more typical of variationist approaches; see chapter 8) and
knowledge of the social and cultural meanings of actions based on my own
membership in a community roughly similar to the one in which Henry, Zelda,
and Irene reside.
Zelda and Henry are married, and they display the closeness and loyalties
of that relationship in their interactions, e.g. they frequently joke with and
tease each other, they argue sociably (Schiffrin 1984a), they defend each other
when challenged by a third person. Irene has lived next door to Zelda and
Henry for more than fifteen years. The typical mode of interaction between
Irene and Henry during my interviews was sociably combative: they playfully
argued and insulted each other, but again rushed tp one another’s defense if
report of a wrongdoing threatened the other’s self-presentation or image
(Schiffrin 1984b, c).
Most relevant to our current concerns is the relationship between Zelda and
Irene. Zelda and Irene are close friends. Although Zelda is about twenty years
older than Irene, both are mothers with three children each. Furthermore,
although two of Zelda’s children are married with their own children, Zelda’s
youngest daughter (a teenager) is a classmate of Irene’s eldest son. Zelda and
Irene have multiple opportunities for interaction, and multiple ways to relate
to one another. One source of evidence for their relationship is what they say
about one another. In (2), for example, Irene answered a question about her
neighborhood by describing the advantages of having a neighbor like Zelda.

(2) Cause I - I like to know that in case I need something, like with Zelda,
or: y’know she’s been a big help t’me like since I’m workin’ the kids -
kids always here or in an emergency she’s here t’get them or whatever
until I can get there. So: I know I have a girlfriend that lives in Horhsam,
and she doesn’t know her neighbor on either side. Y’know? And I’m not
that kind of person. Like in the wintertime, when you’re snowed in and
this and that, at least you can run next door, talk t’somebody if y’feel
like you’re gonna get closed in with the walls.

These comments reveal not only a closeness with, but also a dependence
on, Zelda, e.g. she’s been a big help t’me. Some of Zelda’s comments about
Irene (although not included here) also reveal the duality of their bond. For
example, Irene helps Zelda shop for clothes for her teenage daughter: since
Irene is younger, Zelda believes that she knows more about adolescent taste
in clothing.
Let us now apply what we know about Zelda and Irene’s relationship to the
interactional meaning of Zelda’s speaking for Irene. We saw above that Zelda
and Irene are close friends based on their similarities (they are neighbors,
mothers of adolescents) and on dependencies that stem partially from their
differences (e.g. in age, experience, taste). In other words, Zelda and Irene are
112 Approaches to Discourse Analysis

in a relationship of solidarity - but a solidarity that is based on both their

similarities and their differences.
The participation framework created when Zelda speaks for Irene exhibits
exactly the kind of shared responsibility and mutual dependence that is typical
of Zelda and Irene’s friendship. Recall that Irene has just turned down an offer
of candy from Henry. By providing an account for Irene’s rejection of Henry’s
offer (and by doing so before Irene’s own account which actually repeats
Zelda’s), Zelda is creating a participation framework that mirrors (on a small
scale) the more general bond that she shares with Irene. Just as Zelda “is
there” to help Irene with her children or to provide company during snow¬
storms (see 2), so Zelda “is there” to help Irene soften the potentially threaten¬
ing meanings created by her rejection of Henry’s offer. Another way of saying
this is that when Zelda says she’s on a diet, her utterance serves the same
general function as when she provides a big help . . . in an emergency. The
difference is that the former displays solidarity on a micro-interactional level
of talk relevant to an expressive domain of Irene’s life (her relationship with
Henry), and the latter displays solidarity on a macro-interactional level of
rights and responsibilities relevant to an instrumental domain of Irene’s life
(her ability to manage both her career and her family).
It is important to note that Zelda’s account is built upon the provision of
information about Irene that may be essentially private information: being on
a diet is something that one either has to observe about another or be told by
that other (unless the diet has already resulted in a visible loss of weight).
Thus, Zelda’s account works by revealing information about Irene that had
to be gleaned from a prior interaction and prior relationship (cf. Schiffrin
1984a). Although the revelation of information about another may actually
threaten, rather than build, solidarity, I do not think this is the case here (cf.
Besnier 1989). Indeed, because we know that Henry, Irene, and Zelda often
use openly competitive forms of talk for positive ends, even an overtly negative
gloss to she’s on a diet might be reframed yet again: the ability to playfully
violate an avoidance ritual (to bypass negative politeness) can attest to the
strength of a relationship. Consider, also, that the information revealed by
Zelda is essentially positive information about Irene: as typified by sayings
such as “you can’t be too rich or too thin,” American culture values being
thin. Furthermore, being overweight is commonly interpreted as a lack of
self-control. Thus, Zelda’s account for Irene’s refusal of the candy (and of
course Irene’s refusal itself) also portrays Irene in a positive way as someone
who is “in control.” Interestingly, these flattering portraits of Irene provide
an implicit negative contrast with Henry - who (despite being overweight) is
eating the candy.
Let us summarize how we have thus far analyzed the shift in participation
framework created by Zelda’s She’s on a diet. After we identified She’s on a
diet as an instance of “speaking for,” we observed some of the different
possible interactional meanings of this way of speaking and the role of spokes¬
person. We viewed these meanings as not necessarily limited to Zelda’s
 Interactional Sociolinguistics 113

particular act, but more general meanings that were possibly applicable to
Zelda’s act. We proposed that Zelda’s remark displays closeness and mutual
dependency, and that it does so at two levels: the social relationship external
to current talk, and the participation framework in current talk. Note that we
said nothing about Zelda’s intentions or goals: rather, we explained Zelda’s
remark by considering it as a way of speaking (“speaking for”) that may be
used with a person with whom one has a particular relationship (i.e. friend¬
ship). Thus, we moved from a particular utterance to the frameworks - or
contexts - in which that utterance is embedded. Implicit in this procedure is
a belief that it is the social contextualization of an utterance that motivates
and explains its use. What are provided by context, then, are situated inferences
about the meaning of an interactional move. Put another way, a particular
utterance can act as a contextualization cue to the contextual presuppositions
that inform and provide for its meaning and use.
But does such a procedure explain why Zelda says She’s on a diet at this
particular point in the conversation? And how do we know if our general
explanation fits the particular relationship among Henry, Irene, and Zelda?
In order to address this level of contextualized motivation, we need to consider
the social meanings of rejected offers and to assess the meaning of a rejected
offer for Henry and Irene in particular. In other words, we need to add another
context to our consideration of the contexts in which She’s on a diet is situated
and interpreted.

3.2.2 Social actions: providing accounts for self and other

In chapter 3, we discussed how Zelda’s She’s on a diet and Irene’s story about
her diet both provide an account for Irene’s rejection of Henry’s offer of candy.
But we had to defer our discussion of why accounts can supply a third part
to offer-rejection sequences simply because the reasons have less to do with
the constitutive rules of speech acts themselves than with what those rules
imply about social relationships and social interaction. In this section, we
illustrate how an interactional approach to discourse can help to explain the
social coherence underlying this particular speech act sequence - thus suggest¬
ing that the basis for sequential coherence between speech acts lies in the social
and interactive world in which speech acts are created. At the same time, we
see that the social meanings of acts and their relationships to one another
provide another context in which utterance meaning is situated.
In chapter 3, we said that offers are speech acts through which S makes A
available to H. We also noted that S does not know for sure whether H wants
A. Here I suggest that offers involve two kinds of “wants” with interactional
and interpersonal consequences that can be remedied by accounts.
The first want is that H wants something: since offers are presumed to be
for H’s benefit, we can infer that S assumes that H might want the good or
service being made available. As noted above, however, S has no guarantee
114 Approaches to Discourse Analysis
that A is what H wants, and this means that S must guess about what H wants
(chapter 3). The problem with this guesswork, however, is that guessing about
what another wants means making assumptions about another’s internal
states. This can be a threat to H’s negative face (Brown and Levinson 1987),
to the desire that one’s own needs and wants be unimpeded and unintruded
upon. The second want is that S wants to do something for H (akin to a Searle
sincerity condition). What this means is that when H rejects an offer, H is (in
effect) saying that he does not want what S wants. This can be a threat to
positive face (Brown and Levinson 1987), to the desire that others want the
same thing that self wants.
Although guessing about another’s wants never has a guaranteed outcome,
that outcome can be improved with familiarity. But just as right guesses about
what H wants (leading H to accept an offer) can thus reflect familiarity, wrong
guesses about what H wants (leading H to reject an offer) can display a failed
presumption of familiarity - that S does not know H well enough to guess
about what H wants. And this has two consequences. First, as already noted,
S’s guess about what H wants intrudes upon H’s negative face. Second, by
forcing H to reveal that what H wants is not what S wants, H is forced to
threaten S’s positive face.
I have suggested so far that rejected offers are threatening because of the
vulnerability of opening one’s wants to the other: rejected offers convey that
H doesn’t want what S wants, and that S was not able to guess what H wants.
Thus, rejected offers are threats both to the self (“you don’t want/like what I
want/like”) and to the self-other relationship (“I don’t know you well enough
to guess what you want/like”). These threats to the self and to the self-other
relationship are mitigated by accounts: accounts can provide a reason inde¬
pendent of the offer itself as to why H does not want what S thinks he wants
and why H does not want what S wants.
To return to Zelda’s She’s on a diet and Irene’s “diet” story, Zelda and Irene
both minimize the potential threat of Irene’s rejection of Henry’s offer. Inter¬
estingly, this remedial work ends up serving both Irene’s and Henry’s interac¬
tional needs and protecting their relationship with one another. By showing
that Irene’s rejection of Henry’s offer was not due to a lack of respect for his
values, the accounts both preserve Henry’s values and allow Irene to reject
them. Other analyses of the interactions among Henry, Zelda, and Irene show
that this kind of mutually protective work is not at all unusual.
I have suggested thus far that rejections are potentially face-threatening: they
threaten positive face by indicating that something found valuable by S is not
similarly valued by H. But what about the meaning of rejected offers for Henry
and Irene? Henry and Irene often argue with and insult one another playfully:
they engage in interactions whose forms and meanings might threaten a rela¬
tionship that was not firmly established through other bonds and threaten the
selves of those not part of such a relationship (Schiffrin 1984c). This means
that acts conventionally seen as face-threatening (e.g. rejected offers) might
not threaten Henry and Irene and not need remedial work.
 Interactional Sociolinguistics 115

In this section, we have seen that the social meanings of acts can be formu¬
lated in terms of their interpersonal and interactional consequences, thus
providing another context in which utterance meaning is situated. We see in
the next section that speaker identity provides another facet of context in
which interactional approaches situate utterance meaning: thus, speaking for
another can be not only a discourse strategy by which to maintain soli¬
darity, but also a contextualization cue that signals speaker identity and rela¬
tionship.

3.2.3 Micro- and macro-identity displays: alignments and

gender
0 *
In this section, we recontextualize She’s on a diet yet again: we consider it
within the context of who is speaking (male or female). In so doing, we will
not only consider how speaking for another can display one’s identity, but
also how speaker identity provides another facet of context in which to
discover utterance meaning. Our discussion will be helped by a brief compari¬
son with other participant realignments among the same speakers.
Here is the relevant segment again.

(1) HENRY: (a) Y’want a piece of candy?

IRENE: (b) No. [I don’t
zelda: (c) She’s on a [diet.
DEBBY: (d) [Who’s not on [a diet.
IRENE: (e) =I’m on a diet
(f) and my mother [buys-=
ZELDA: (g) [You’re not!
IRENE: (h) =my [mother buys these mints.=
DEBBY: (i) [Oh yes I amhhhh!

Note, first, that Zelda’s She’s on a diet is a remark said by one woman about
another woman’s diet. Although this utterance is directed to Henry, and not
necessarily to me (another woman in the interaction), it becomes part of an
ongoing interaction in which the women participants (me, Zelda, Irene) are
engaged in a cooperative byplay, i.e. an interactively managed interlude,
subordinate to talk about another topic, rather than an interactive focus itself
(cf. Goodwin, forthcoming; Schiffrin, forthcoming). After Zelda’s She’s on a
diet, I generalize the applicability of “diet” with a rhetorical question (Who’s
not on a diet (d)). Zelda’s response (You’re not! (g)) remedies the negative
self-assessment that being on a diet implies (i.e. being overweight), but I persist
in generalizing “diet” at least to myself (Oh yes I amhhhh! (i)). Overlapping
this exchange between me and Zelda, Irene shifts from her own utterance (I
don’t- in (b)) to repeat the account offered by Zelda (I’m on a diet (e)). Irene’s
repetition leads into the story analyzed in chapter 3 - the story showing that
\\6 Approaches to Discourse Analysis

she is so serious about dieting that she throws candy offered by her mother
out the window.
What is important to note here is that Zelda has created a topic shift that
is picked up by the two other women in the conversation - but not by Henry
(this type of topic pursuit might also be characteristic of the ethnic style of
the speakers; see Schiffrin 1984a; Tannen 1984). It might seem that this is a
topic of more interest to the women, i.e. women are traditionally assumed to
be more concerned than men with personal topics such as physical appearance
(Aries and Johnson 1983), including their weight (but see below). And some
studies have shown that women talk a great deal about food and diets: Deakins
(1989), for example, found that the third most frequent topic of women
executives at lunch was diets (see also Kipers 1987). Note, however, that
Henry is overweight (more so than Irene), and elsewhere he talks about how
thin he was when he was young, and how he plans to lose weight when he
eats more fish over the summer (since fish is lower in calories than meat).
Thus, we might expect Henry to avoid the topic of diet precisely because he
is overweight and he is the one who is eating candy.
There might still be another explanation for the women’s pursuit of the
“diet” topic - one that centers on the way they display (and create) gender
identities through the structuring of participation frameworks. That the dif¬
ferential construction of participation frameworks is a realization and reflec¬
tion of gender is suggested by a variety of studies. Kalcik (1975), for example,
suggests that women pursue topics of talk more interactively than men - that
what one person proposes as a topic is progressively built upon by another.
Cooperative topic building can easily be seen in terms of participation frame¬
work: it requires a joint alignment toward a focus of talk in which both
addressor and addressee share the roles of animator and principal. Consistent
with these findings is my own observation that Henry builds topics in ways
less dependent upon others’ active contributions (i.e. the listener places a
relatively passive, supportive role) than either Irene or Zelda. Many of my
questions to Henry, for example, were answered with lectures (Tannen 1990:
123—48): not only were Henry’s turns at talk long with very few interruptions,
but he imparted information, using traditional rhetorical devices, such as
syntactic parallelism, to make a point. Studies of gender and dispute manage¬
ment can also be seen in terms of participation framework. Goodwin (1990),
for example, shows that young African-American girls have remarkably
different ways of managing group conflict than do boys: such differences
revolve around the use of stories as a means by which to create and re-create
participation rights both inside and outside of the group.
Returning to our example in (1), the findings noted above suggest that the
way the diet topic is pursued by Zelda, Irene, and me - but not by Henry - is
due not just to topic relevance or personal importance per se, but to women’s
conceptions of what interactional needs should be pursued at that moment in
the conversation, and to the way that it is participation in talk itself that can meet
those needs. Consider, for example, that my generalization of “diet” (d, i) and
 Interactional Sociolinguistics 117

Zelda’s remark that I’m not on a diet (g) both work to minimize any one
particular person’s need to be on a diet. Thus, they function to minimize
potential insults to weight (and physical appearance) and to build solidarity
through positive politeness. Similarly, it is often noted that women attend more
to maintaining solidarity during interaction than men (e.g. by avoiding conflict
and minimizing threats) and that they focus more, in general, on the expressive
implications of what is said than do men (e.g. Tannen 1990). Such observa¬
tions are consistent with Gilligan’s (1982) findings that instead of focusing on
abstract principles and the impartial application of rules of right and wrong
(the moral path followed by men), women emphasize human connections,
caring, and the needs and situations of those affected by a problem.
(3) is a striking example of the different interactional paths - paths that seem
to be gender related - that can be pursued by Henry and Zelda in my data.

(3) DEBBY: (a) Well, Irene, is there anybody around here that you would
call a best friend?
IRENE: (b) Now?
DEBBY: (c) Yeh. hhhh
HENRY: (d) In front of us?
zelda: (e) No, we’re not her [best friends, we’re her=
HENRY: (f) [No! She’s got a best friend.
ZELDA: =neighbors!
IRENE: (g) No, I [don’t really think any one person I could say.
HENRY: (h) [=But she’s [more of a friend to a person,=
zelda: (i) [She’s not-
HENRY: (j) =than a person is to her. [Let’s put it that way.=
IRENE: (k) ZYeh. [T’be honest.
HENRY: (1) =And it- [it [Am I right or wrong?
IRENE: (m) [I think [I’ve- I’ve
been getting [hurt, too much lately.
ZELDA: (n) [Are you Irene?! 7
DEBBY: (o) Yeh.
HENRY: (P) No but am I right or wrong Irene?
(q) I could see [some damage! Yeh.=
IRENE: (r) [Yeh. I feel I’m hurt.=
HENRY: (s) =[See?]
IRENE: (t) =[I ] don’t know how the other people feel.

Note, first, that my question in (a) follows the question-answer format of our
interview (see chapter 5, section 3.3). The interchange between Henry and
Zelda (d and e), however, alters this format of the interview since they both
comment on the question, rather than allow Irene to answer the question
(chapter 5, sections 3.3.1, 3.3.5). Henry then answers for Irene in what sounds
like a supportive manner: She’s got a best friend (f). He immediately uses his
own assertion, however, to project an inequity between the way Irene treats
118 Approaches to Discourse Analysis

others (positively) and the way she is treated by others (negatively): But she s
more of a friend to a person, than a person is to her (h, j). Although Henry’s
comparison reveals more about Irene than she might otherwise prefer, Irene
does agree with Henry (Yeh. T’be honest (k), I think I’ve- I’ve been getting
hurt, too much lately, (m)). Note that once Irene herself reveals “trouble”
(Jefferson 1980), Zelda immediately offers sympathy on a personal level with
Are you Irene?! (n). But Henry uses Irene’s revelation of getting hurt as a basis
for the correctness of his own assessment: he uses it to pursue the validity of
his own point (No but am I right or wrong Irene? (p)), to reaffirm what he
himself has seen (I could see some damage! (q)), and to gain consensus about
his point (See? (s)). Thus, whereas Zelda offers support based on the personal
implications of what Irene has said, i.e. how the inequity affects Irene herself,
Henry uses Irene’s situation as a basis from which to buttress a more general
assessment of her dilemma.
Thus far, I have suggested that Zelda’s She’s on a diet proposes an interac¬
tional alignment that is more typical of the stances taken by women than those
taken by men. Above I noted that one way that gender is realized and reflected
in talk is through the differential construction of participation frameworks.
This suggests that we can further check this proposal about gender-related
interactional differences by comparing participant realignments during same-
gender talk to those created during cross-gender talk. In (4), for example, I
have been asking Zelda, Irene, and Henry about their personal contacts in the
neighborhood.

(4) debby: (a) How ’bout you Irene?

IRENE: (b) What?
DEBBY: (c) Who would you-=
ZELDA: (d) Who would you [discuss it with?
DEBBY: (e) = =[if you had a hard day
who would you complain to?
ZELDA: (f) Jayhhhhh

Irene (What? (b)) asks for clarification of the question I had initiated in (a).
In (d), Zelda provides the clarification that Irene had requested from me.
Zelda’s contribution overlaps with my own expansion of the question (in c,
e); note, also, how we both repeat the Wh portion of the question. In addition
to sharing my questioning role, Zelda also shares Irene’s answering role in (f)
by providing Jay as a potential answer. Thus, Zelda speaks first for me and
then for Irene - her adoption of the two roles reproducing the structure of the
question-answer sequence.
(4) continues with Irene’s answer building upon Zelda’s answer:

IRENE: (g) Uh:: I don’t know.

(h) Depending on what- what the problem really
was, I might [talk] to a friend easier.
debby: (i) [Yeh.] ZYeh.
 Interactional Sociolinguistics 119

In (g) and (h), Irene states that she might talk to a friend easier, note that the
implicit comparison is between a friend and Jay (Irene’s husband, the person
mentioned by Zelda). As Irene continues to compare herself to Zelda (and to
Henry) in (j), however, Henry joins the talk:

IRENE: (i) =than to a- what- I’m not that family orientated.=

DEBBY: (k) Uhhuh.
IRENE: (1) =[Like they are.
HENRY: (m) [She should be. She’s got a nice family.

We saw a moment ago that Zelda’s remarks reproduced the structure of the
ongoing discourse and allowed others to maintain their participant roles.
Henry’s assessment of Irene’s claim (from (j) and (1) that she is not family
oriented) in (m), however, shifts the participant alignment and forces others
to fit their utterances into that framework. Rather than create a duality in the
speaker role (a split between animator (the spokesperson) and principal (the
spoken for)), Henry’s assessment creates a split in the addressee role: by
assessing what Irene has said in her presence, and addressing that assessment
to me, Henry divides his audience into an addressed recipient (me) and an
unaddressed recipient (Irene).
Henry’s realignment of the audience has an outcome quite different from the
inclusionary effect of Zelda’s utterances. In one sense, Henry’s remarks ex¬
clude Irene from the interaction; at the very least, from active participation in
the dialogue that Zelda and I had initiated and maintained. Furthermore, since
Henry’s remarks offer a moral assessment of the content of Irene’s answer,
they open Irene to scrutiny from me - a relative outsider to their group (cf.
3). Thus, in one sense, Henry’s remarks create a participation framework that
can potentially separate co-participants from one another. In another sense,
however, Henry’s assessment draws Irene into interaction with him: although
she is not formally addressed by his remarks (note the third person pronoun
reference to Irene as she), she is certainly expected to hear what has been said
about her. We can consider the effects of this in terms of Goffman’s (1963)
notion of information preserve and Brown and Levinson’s ideas about polite¬
ness. Offering unsolicited advice about what is considered another’s domain
(e.g. closeness to family), particularly when what is said concerns what that
other should do, is a violation of negative face - the sort of violation com¬
monly called “not minding your own business” or “butting in.” Thus, Henry
does draw Irene into the interaction, but he does so by intruding upon her
affairs and by forcing her to defend the management of those affairs.3
Consider what happens next:

ZELDA: (m) Well [she really- wait a minute she doesn’t=

HENRY: (n) [She’s got lovely children.
ZELDA: =have any [sisters,
IRENE: (o) [I’m not talkin’ about my children.
ZELDA: Z Right.
(P)
120 Approaches to Discourse Analysis

HENRY: (q) Well you should make it so, it’s good.

ZELDA: (r) Oh! C’mon!
IRENE: (S) Hen, I’m not you, Henry!
(t) We’ve- we’ve had family problems where my husband and
his brother were in business.
(u) And at one time we were very close.
(v) And now it’s just hello, goodbye, how are you, and that’s
it.

Although Zelda’s defense of Irene (she doesn’t have any sisters (m)) continues
the participation framework opened by Henry (note her third person reference
to Irene), Zelda and Irene continue collaborating. Irene responds to Henry
(I’m not talkin’ about my children (o)) along the same lines as Zelda: both
narrow Henry’s reference to family. Zelda agrees with Irene’s own defense
(Right in p) even though Irene has bypassed the content of Zelda’s defense.
Finally, although Henry redirects his criticism to Irene herself (in q), Zelda
continues to defend Irene by discouraging Henry’s continued participation
(Oh! C’mon! (r)). Despite Irene’s and Zelda’s collaboration, Henry’s shift of
the participation framework becomes dominant: Irene defends herself to
Henry (Hen, I’m not you, Henry! (s)) but also orients that defense to me (in
t to v). Thus, although Irene rejects the relatively passive role of the unad¬
dressed recipient status, her defense still conforms to the participation frame¬
work created by Henry - simply because she defends herself relative to both
her challenger (Henry) and the recipient of the words to whom the challenge
was ostensibly addressed (me).
(3) and (4) have illustrated a difference in participant realignments predict¬
able from previous scholarship on gender and discourse. Although both Zelda
and Henry alter the participation framework, they do so in different ways:
Zelda’s realignment was more supportive and integrative; Henry’s stance was
more judgemental and divisive; in (4), Henry drew Irene into interaction by
challenging her (an act requiring a defensive response) and by treating her as
an unaddressed recipient (a position of relatively low status).
If the alignment differences just described for (1), (3), and (4) really are
related to gender we would expect them to reappear during other exchanges
- even those that do not involve speaking for another. Put another way, we
might expect the same kind of interactional alignments to reappear with
interactional moves other than the one with which we began our discussion.
Suppose, for example, that instead of providing Irene’s account by saying She’s
on a diet, Zelda had created a slot in which Irene would have to provide her
own account, e.g. Tell him why you don’t want it or Henry doesn’t know why
you refused. Fortunately, there were two interchanges in the same interview
among these three participants that contain this alternative means of reorient¬
ing the content of another’s talk: one person (either Zelda or Henry) prompts
another (Irene) to produce (animate) what the prompter (as principal) had “in
mind.” As we see, there are some striking differences in the division of
 Interactional Sociolinguistics 121

responsibility for those realignments that seem to be predicted by gender -

differences similar to those already seen.
First is an example of a realignment in same-gender talk:

(5) debby: (a) Yeh, who d’y’go bowling with?

IRENE: (b) My next door neighbors.
(c) We have a team in the summer we bowl.
(d) Every summer [they- husband and wife-
ZELDA: (e) [Tell her who you bowl with.
IRENE: (f) Who we bowl with.
ZELDA: (g) D- y’know the teams.
IRENE: (h) Oh. Wha’d’ y’mean [the kids?
ZELDA: (i) [The kids.
IRENE: (j) [Oh yeh, this year the kidsiiave= *
henry: (k) [Kids. They have a very good thing.
IRENE: =their own team,
(l) [and they’re . . . [they’re eh . . . they’re giving=
DEBBY: (m) [Oh great!
HENRY: (n) [They got a good thing going.
IRENE: (o) =us a run for our money cause we- we’re in first place and
they’re in second, and they’re- they cheer everytime we
lose a game.

The section to focus upon begins with Zelda’s overlapping speech in (e): Tell
her who you bowl with. This remark is (in the terms of ethnomethodologists;
chapter 7) an other-initiated repair: by prompting Irene to provide a specific
piece of information, Zelda is redefining something Irene has already said as
“incomplete,” i.e. as a repairable. By so doing, Zelda inserts herself into what
has thus far been a question-answer exchange between me and Irene. Her
role is somewhat like that of a monitor of the exchange: she is assessing the
adequacy of the answer and prompting Irene to animate what she herself has
“in mind” as a suitable answer to the question. In terms of participant
alignments, then, Zelda is prompting Irene to animate the words for which
she is principal.
Note, however, that Zelda repeats the content of my question (who d’y’ go
bowling with? (a)). This may account for Irene’s repetition (in f) of what Zelda
has asked for and her contrastive stress on bowl, i.e. if I had already asked
about “bowling” why should Zelda also be asking about “bowl”? Following
Irene’s repetition of the slot that requires completion (the identity of “who
you bowl with”), Zelda continues to prompt Irene: in addition to appealing
to shared knowledge (with y’know, Schiffrin 1987a: chapter 9), she broadens
the reference to the teams. In (h), Irene shows recognition with oh (Schiffrin
1987a: chapter 4) and becomes more explicit (she refers to the kids)-, however,
she maintains Zelda’s role as promptor of information - as one who has the
“right” answer “in mind” - by asking for Zelda’s confirmation (Wha’d’y’mean
122 Approaches to Discourse Analysis

the kids?). Zelda has already begun to provide her own explicit meaning: The
kids in (i) “latches onto” Irene’s Wha’d’y’mean. In (j), Irene repeats the
information given by Zelda: Oh yeh shows that she is inserting it into the slot
created by Zelda’s prompting. Note that Irene not only incorporates that
information into her own contribution (this year the kids have their own
team), but expands it into a brief narrative (j, 1, o).
We have discussed (5) because it illustrates an alternative to speaking for
another: rather than Zelda saying Irene bowls with her kids, Zelda prompts
Irene to say this herself. Although it is Zelda’s “idea” for Irene to include “the
kids” in her answer, Zelda allows Irene to share responsibility for that informa¬
tion. In addition, Zelda brings Irene into the prompting process itself, such
that they gradually build the mention of “the kids” together, each giving the
other just enough to allow that other to maintain an active stance in the
conversation. Thus, although Zelda prompts Irene to animate that for which
she is principal, she also allows her to share responsibility for the words.
An effort to prompt another to speak is also illustrated in (6); again, Irene
is being prompted to tell me something, but this time it is Henry who is doing
the prompting. Prior to (6), I had been asking Irene whom she is friendly with;
Irene is telling me about how infrequently she sees her childhood friends.

(6) IRENE: (a) I keep in touch with them, y’know I go to their affairs, but
I don’t see them that often. One lives in Jersey and one
lives up in the Northeast.
debby: (b) Umhmm.
HENRY: (c) Tell her about the [girl you were real close:,=
IRENE: (d) [But-
HENRY: (e) =you were raised, and they got money, and they don’t
know y’.
IRENE: (f) Who? Barbara? Oh. Well . . .
HENRY: (g) Well she wants t’know!
ZELDA: (H) Yeh but she’s friends with her.
IRENE: (i) I talk to her occasionally, but eh: she: she:=
HENRY: (j) [Yeh but tell the way it is.=
IRENE: (k) =[has moved t’Brookside and they have a whole new=
HENRY: (1) =[She wants t’know. She wants t’know.
IRENE: =circle of eh: friends. Y’know.

Although it is imp
interactively and sequentially, it is helpful to begin by comparing how Henry
prompts Irene in (6) to how Zelda prompted Irene in (5).

Henry’s prompts to Irene (6) Zelda’s prompts to Irene (5)

Imperative Imperative
Tell her about “X” Tell her “X”
 Interactional Sociolinguistics 123
Expands “X” (c) (e) Expands “X”
Lists critical events Creates shared knowledge
(4 narrative clauses) D- y’know the teams.
Externalizes need to tell about “X” States “X”
Well she wants t’knowl (g) The kids.
Imperative
Yeh but tell the way it is. (j)
Externalizes need to tell about “X”
She wants t’know. She wants t’know. (1)

Henry and Zelda both initiate their prompts with the imperative tell her, both
make explicit what Irene is to say and to whom she is to say it. However, the
way they pursue those prompts differs radically. In part, this difference has to
do with what they are prompting Irene to say* Henrf wants Irene to reveal
what is potentially embarrassing information, i.e. an incident in her own
history that ended “badly.” (Recall that this was essentially the outcome of
(4), in which Henry spoke for Irene concerning the importance of being close
to family; recall also that that forced Irene to reveal her failed venture into a
family business.) Following their imperatives, Henry and Zelda both expand
what it is that Irene is to tell me. These follow-up strategies display differences
that can be seen as gender related: by listing four events recapitulating the
experience that he wants Irene to report, Henry is “telling” Irene the answer,
i.e. transferring information to her; by prefacing the information with a mar¬
ker of shared knowledge, Zelda is “helping” Irene with the answer, i.e.
collaborating in the production of information. In effect, then, Henry allows
Irene little more than a role as animator; Zelda, however, allows Irene to share
in the principal role.
The prompting strategies diverge even further following the failure of the
expansions to elicit the desired information. As we saw above, Zelda gives
Irene the information - The kids. But Henry brings in an external justification
for his persistent prompting: he couples a renewed imperative (Yeh but tell the
way it is. (j)) with repeated assertions that I want the information that he is
after: She wants t’know. Note, then, that Henry begins to speak for me: he
uses me as an external validation for his own actions. Henry thus brings me
back into the conversation, but with an altered role - as an addressed recipient
- and the way he re-evokes this role alters the agenda from the one I had
established to one that he is directing.
Let us compare Irene’s responses to both Henry and Irene:

Irene’s responses to Henry Irene’s responses to Zelda

Who? Barbara? Oh. Well. . . (f) Who we bowl with.

I talk to her occasionally, Oh. Wha’d’y’mean the kids?
but eh: she: she: has moved t’ Oh yeh, this year the kids
Brookside and they have a whole have their own team (continues]
new circle of friends, y’know (i)(k)
124 Approaches to Discourse Analysis

Irene’s responses to Henry’s initial prompt show reluctance to provide what

he is after. Note that her first response (in f) is formally similar to her response
to Zelda: Irene asks Who and then offers a candidate identity (to Henry it is
Barbara?, to Zelda, it was we bowl with). But following her recognition (Oh)
that this is indeed the “story” that Henry wants, she displays what seems like
reluctance to go on with Well. . .: it is here that Henry brings me into the
prompting process with Well she wants t’know! Note, however, what happens
when Irene presents her eventual story (filled with hesitation about how to
formulate what happened):

IRENE: (i) I talk to her occasionally, but eh: she: she:=

henry: (j) [Yeh but tell the way it is.=
IRENE: (k) =[has moved t’Brookside and they have a whole new=
HENRY: (1) =[She wants t’know. She wants t’know.
IRENE: =circle of eh: [friends. Y’know.
DEBby: (m) [Yeh. Got snobbish.
HENRY: (n) Money makes a difference. Money’s important=
debby: (o) Yeh.
henry: (P) =to a lot of people. It’s a status.

The information that Irene presents (in i and k) seems to be what Henry has
been after - since Henry follows Irene’s story with a statement of a more
general theme (money changes people for the worse (n, p)) that he (a lower
middle-class blue collar worker living near an affluent white collar suburb)
has reiterated numerous times during our conversations. Yet Henry continues
to prompt Irene during Irene’s story: it is only after I present the “point” of
the story (that Irene’s friends became snobs after moving to a wealthier suburb
(m) ) that Henry states his own theme. Thus, Henry’s prompting continues
until he is sure that what he had “in mind” has not only been said (animated
by Irene) but, also, that it has been “heard” in a way consistent with the
general theme that he is putting forth.
Observe, then, that Henry has done more than alter his footing in relation
to Irene: he has shifted the structure of the discourse from a question-answer
dialogue to a discussion of a general moral issue of which he is “in control”
and to which he can orient his interlocutors. The continuation of (6) suggests
that Henry is after still further restructuring:

henry: (P) [It’s a status.

DEBBY: (q) [D’you-
(r) Yeh. Its: works in funny ways.
ZELDA: (s) Umhmm. That’s exactly it. It [works in funny ways.
HENRY: (t) [D’you take a person for face
value, or: or if they live in a big fancy house, does that make
any difference to you?
debby: (u) No. None at all.
 Interactional Sociolinguistics 125

henry: (v) Well that’s the way it should be.

(w) You’re a nice girl!

During Henry’s restatement of his point (in p), I begin to ask Irene a question,
i.e. to recreate the question-answer structure. I quickly adapt my remarks to
Henry’s theme, however, even building upon that theme in (r). Zelda supports
my statement of the theme (in s) with agreement tokens and repetition. Al¬
though this is certainly the moral theme that Henry has initiated, Henry’s
interruption of Zelda (to ask me whether my own experience conforms to his
view of what is right in t) rejects the collaborative nature of what Zelda and
I have said to shift the participant structure still further from its earlier form.
Recall that the interchange began with me asking Irene questions; Henry then
evaluated Irene’s answers, prompted Irene to shift her answers, and, finally,
asked me questions and evaluated my answers. Thuft, what the question-
answer-evaluation sequence (in t through w) does is complete the restructuring
of the participation framework to one in which Henry is “in charge.”
We have seen in (6) that Henry prompts Irene to reveal potentially embar¬
rassing information - information that is not exactly what I had requested,
but information that allows Henry to state a general theme. We commented
earlier that Henry allows Irene little more than a role as animator, and that
his persistent prompting uses me as an external validator for his own actions.
These strategies shift the discourse structure from one that I had established
to one that Henry is directing: Henry’s prompting continues until he is sure
that what he had “in mind” has been said and “heard” in a way consistent
with the general theme that he is putting forth. He also shifts the structure
from one in which I ask questions to one in which I answer questions (and
have my answers evaluated). Interestingly, the point that Henry is making is
one that unifies us - Henry, Zelda, Irene, me - in a joint moral stance
“against” another group. But the solidarity that is built is not only defined in
opposition to another group (i.e. “we” are different from those who have more
money because they reject us), it is also interactively constructed in a way that
differentiates participant roles from one another and allows one speaker
(Henry) to manage the way that others present and interpret information.
We began this section by proposing that Zelda’s She’s on a diet created a
topic shift picked up by the two other women in the conversation and elabor¬
ated upon as a way of building solidarity through positive politeness. After
noting that this interpretation was consistent with a great deal of work on
gender and discourse, we compared some other participant realignments cre¬
ated during same-gender talk to those created during cross-gender talk. Our
comparison showed that Zelda’s realignments were inclusive and reinforcing,
i.e. they seemed to allow participants to continue their prior (relatively active)
roles; Henry’s realignments were not only more divisive, but he also pursued
them more completely, and they created more radical shifts in participant
structure. Note that our interpretations have said nothing about participants’
personalities (e.g. Zelda is being “kind”) or intentions (e.g. Henry is “trying”
126 Approaches to Discourse Analysis

to show Irene that he knows “what’s best for her”); rather, we have tried to
focus upon the interactional effects of what is said. Nor have we said, despite
the seemingly cooperative nature of Zelda’s actions (i.e. Zelda and Irene seem
to share interactive goals) and the competitive nature of Henry’s actions (i.e.
Henry and Irene seem to have different interactive goals) that one alignment
strategy reveals friendship and the other antagonism; rather, both sets of
actions seem to reveal and create interpersonal involvement - the difference
being in the means by which involvement is created (Tannen 1984).
Differences such as those just noted are consistent with prior work on gender
and discourse (although they also show some of the more subtle ways that
footing and alignment can be created, maintained, and shifted). However, it
is important to note that not all realignments differ so dramatically by gender.
Recall the example in which Zelda prompted Irene to tell me who she bowls
with (5). Although I did not focus upon Henry’s contribution in this exchange,
let us examine it here:

(5) IRENE: (j) [Oh yeh, this year the kids have] their own team,=
HENRY: (k) [Kids. They have a very good thing.]
IRENE: (1) [and they’re . . .] [they’re eh . . . They’re giving us a=]
DEBBY: (m) [Oh great! ]
HENRY: (n) [They got a good thing going.]
IRENE: (o) =run for our money cause we- we’re in first place and they’re
in second, and they’re- they cheer everytime we lose a game.

As Irene incorporates Zelda’s prompted information into her answer to me,

Henry begins to evaluate Irene’s response - much as he did in (3) and (6).
Here, however, Henry’s evaluations are positive, and perhaps because of this,
they do not shift the participant structure at all. Rather, they are heard more
as background support not only for Irene’s bowling teams and for her family,
but also for her stance in our conversation.
Consider, also, what happened prior to (5) - an exchange that I present
below as (7).

(7) DEBBY: (a) Does anybody here bowl?

(b) [D’you go bowling?
ZELDA: (c) [Yeh [she does.
HENRY: (e) [We used to.
DEBBY: (f) [Do you?
HENRY: (g) [She does.
IRENE: (h) [Yeh.

My initial question (Does anybody here bowl? (a)) is addressed to the group
in general, but I quickly make it more specific and ask Irene if she bowls
(.D'you go bowling? (b)). While I am addressing my question to Irene, Zelda
answers my first question about Irene: Yeh she does (c). I use Zelda’s answer
 Interactional Sociolinguistics 127

to continue to address Irene: rather than saying Oh she does}, for example, I
bring Irene back into the question-answer format by asking for her confirmation
with Do you} (f). Note that Henry also answers for Irene - She does (g) - but
that he does so after reporting on his own situation (We used to (e)). Thus, we
might interpret Henry’s She does as a repetition of Zelda, and as an effort to
join a framework that has shifted before his own utterance. Note, finally, that
Irene also answers for herself: her Yeh (in h) follows my question to her and
overlaps with Henry’s answer about her (from g). In brief, Zelda, Henry, and
Irene all provide the information that Irene goes bowling. Zelda and Henry
both seem to be speaking for Irene: although there is a slight difference in the
realignment among participants that their utterances create (simply because
Zelda speaks for Irene before Henry does), there is nowhere near as dramatic
a difference in realignment by gender as we saw in our other examples.
We began this section by noting that the exchange among Irene, Zelda, and
myself might be (partially) due to gender-related notions of what is important
to pursue in a conversation, and even more compelling, that the kinds of
alignments taken up at various points in the conversation help display (if not
create) gender identities. I supported this idea by discussing some ways that
participant alignments seem to differ when shifts are created during cross¬
gender as opposed to same-gender talk, also noting some interchanges that
illustrate less dramatic gender differences.
We need to be careful, however, not to generalize beyond the few examples
that I have discussed here. The avoidance of any kind of generalization that
is not based on case by case analysis is the methodological counterpart of the
interactional sociolinguistic focus on situated meaning. Thus, if we wanted to
be on firmer footing about gender and participant realignments, we would have
to examine each realignment in detail: because each utterance (or exchange, or
interaction) is situated in its own particular way, we would need to pay more
careful attention to the ways that utterances (exchanges or interactions) differ
from one another, as well as the ways that they are similar. Similarly, we might
want to consider further what it means to say that someone is behaving in a
particular “gender role”: we might say, perhaps, that sounding authoritative
and traditional is part of what it means to talk like a middle-aged man in lower
middle-class Jewish American culture - not just to “act like a man.” Note,
finally, how our concerns have shifted. We began (3.2.1) by discussing par¬
ticipant roles (sender/principal, spokesperson/animator, recipient). We then
discussed how an understanding of self-other relationships adds to our ability
to account for speech act sequences (3.2.2). Now we have discussed the way
interactional alignments reveal more stable participant identities, such as gender.

3.3 “Speaking for another" and taking the role of the other

In previous sections, we situated the meaning of a single utterance, She’s on

a diet, in several different ways: as a participation framework within the
128 Approaches to Discourse Analysis

micro-structure of an interaction, as a socially motivated account within a

sequence of acts, and as a gender-based involvement strategy. Our means of
analysis included comparing the particular utterance She’s on a diet with other
utterances whose text and context were similar in some ways, but different in
others.
Trying to locate a behavior in relation to other behaviors - to that with
which it is similar in some ways, but different in others - is important, for
one might very well argue that we cannot really understand what a phenom¬
enon is without also understanding what it is not. (Of course comparing two
entities that are totally different from each other would not give us a very
good idea about what gives either one an identity: we need some shared quality
to make such a comparison meaningful.) We can also compare behaviors at a
more abstract level of analysis. What can help us here is a higher-level con¬
struct that allows us to more systematically discover both the similarities and
the differences between the two entities in relation to each other and that
construct. This sort of classificatory procedure is common in interactional
sociolinguistics: for example, Goffman (1974) helps us understand the shared
(and unshared) features of requests and apologies by considering them both
as remedial acts (acts that circumvent violations of the other). Similarly,
Tannen (1984) proposes that telling personal stories and asking rapid, over¬
lapping questions are both high involvement strategies; Brown and Levinson
(1987) suggest that “y’know” and compliments are both positive face
strategies.
In this section, we consider speaking for another as part of two higher-level
analytical constructs that are themselves related to each other: speaking for
another is an act in which one person takes the role of another, and taking
the role of another is itself a way of showing sequential coherence.
Let us begin by recalling one of Goffman’s observations critical to the
analysis of discourse: when two (or more) people come together, they not only
respond to the institutional and interactional orders in which they find them¬
selves, but they are also co-constructors of those orders. The entry of language
into this relationship is especially pervasive, simply because a current-utter¬
ance creates a range of potential contexts to which a next-utterance can
respond. It is easy to see how the context-reflecting and context-creating role
of language can locate utterances as part of a particular institutional order.
Consider, for example, the utterance Drinkf When said by a bartender to
someone sitting at a bar, we understand Drink? as an offer to sell something
(or as a directive to place an order) in a particular kind of service encounter.
We also know that an appropriate response might be Michelob, but not No
thanks, I’m not thirsty. Because Drink? would not be understood as an offer
to sell a drink in other contexts (e.g. during a dinner party) or among different
participants in the same setting (e.g. if said by one customer to another),
Drink? reflects its institutional setting. It also helps to create an interpretation
of the setting as one in which a service encounter is taking place in a way that
other utterances might not.
 Interactional Sociolinguistics 129
Utterances can also reflect and create the particulars of an interactional
order. If we take Irene’s refusal (No) of Henry’s offer of candy (Want a piece
of candy?) as a current-utterance, we can imagine a number of different
next-utterances from Henry.

(8) HENRY: Want a piece of candy?

IRENE: No. current-utterance]
HENRY: (a) Oh c’mon. next-utterances'
'(b) There’s nothing wrong with it!
(c) Suit yourself.
(d) Just testing you! I know you’re on a diet.
(e) What?
(f) I didn’t hear you.
(g) What time did the teachers leave?
(h) Y’know I bought this candy at that new place in the mall,
and when I was there /continues/

The utterances in (8) select different aspects of Irene’s No as a basis for

response, and in so doing, they provide an interpretive context for the utter¬
ance pair. Some utterances in (8) are clearly linked to the prior offer-refusal
sequence itself (a to d), whereas others are tied to more mechanical and general
aspects of language production (e and f); still others provide the prior offer-
refusal pair with an additional sequential identity, e.g. as a side sequence (g)
or as a lead into a new topic (h). As I noted above, what is important about
this range of responses (and certainly more could be added; see Goffman
1981b: 68-70, on which this example is modelled) is that each next-utterance
selects a slightly different aspect of the current-utterance as a basis from which
to respond. In so doing, it provides an interpretive context in which not only
the next-utterance, but the entire utterance pair can be understood. Put an¬
other way, a next-utterance is a slot in which a speaker can both respond to,
and create, a prior context. Although this mutually constituting relationship
- just the sort of codependency between self and context that I noted above
- might seem to constrain interactants, it also gives them a tremendous creative
potential to lead interactions in a number of different directions (accounting,
perhaps, for some of the topic fluidity that is often associated with conversa¬
tion; e.g. Dorval 1990).
One reason why the self/context codependency is important for discourse
analysis lies in the notion of sequential coherence. The utterances in (8) reflect
not just the variety of ways that a current-utterance can be understood, but a
variety of different frameworks for understanding talk - as actions (a to d),
as language production (e, f), as topic structure (g, h). What makes all of the
utterances in (8) interpretable as sequentially coherent is the availability of
different interpretive frames for talk - different ways of contextualizing what
is said. Put another way, each framework provides a different basis for under¬
standing how one utterance follows another - a different resource for sequential
130 Approaches to Discourse Analysis

coherence. (Note that we may easily express this in either Gumperz’s or

Tannen’s terms: the different utterances are contextualization cues pointing
to, or meta-messages about, different interpretive frames.)
Another resource in the repertoire being described is the participants in talk
themselves: self and other. Let us take the self first; some different ways that
the self can become a resource, again, for Henry’s next-utterance, are illus¬
trated in (9):

(9) (a) Yeh, I shouldn’t have any either.

(b) Right, I forgot you’re on a diet.
(c) Oh- I meant fruit.
(d) It really annoys me when you don’t take candy.
(e) I admire your will power.

As (9) illustrates, a next response can be a comment about one’s own action
(a, b), a self-repair (c), or a personal sentiment (d, e). Just as the self can be
a resource for response, so too, can the other; (10) illustrates some ways -
roughly parallel to those in (9) - that the other can be a basis for Henry’s
next-utterance:

(10) (a) You never accept my offers.

(b) You’re on a diet again.
(c) Did you say no?
(d) You’re stubborn about that diet.

Just as a response can focus upon the self in relation to an action, so too can
it focus upon the other (a, b); similarly, a response can be either a self repair
or an other repair (c), or offer an assessment of the self or other (d).
Although it might seem that self and other provide interchangeable alternat¬
ives as a focus for a next-utterance, the interactional repercussions of using
the other as a resource for response are quite different. One way of expressing
this difference is in terms of Brown and Levinson’s notion of face: using
another’s words can theaten the other’s negative face wants (i.e. not to be
imposed upon). Thus, as we saw in our earlier examples, the way an other-
based response is treated depends, in part, on what is said about the other:
although negative assessments of the other (as in some of Henry’s comments
about Irene) are treated as violations, positive assessments (as in some of
Zelda’s comments about Irene) may actually reinforce positive face wants.
Thus far in this section, we have proposed two analytical constructs that are
themselves related to each other. We began by considering resources for
sequential coherence, then considering the use of self or other as a particular
resource. Speaking for another is an act that fits into both these constructs: it
is one way of providing sequential coherence by drawing upon the other as a
resource for a next utterance. What differentiates it from other such acts is,
first, that it is an act that emerges during a three-party interaction - in which
 Interactional Sociolinguistics 131

a spokesperson animates the words of another (a principal) to an addressed

recipient. More critically, speaking for another displays an interpersonal
stance in which a spokesperson not only uses the other as a basis for a
next-utterance, but enters into the other’s perspective to issue a next-utterance
from that other’s point of view. Thus, speaking for another is a way of “taking
the role of the other.”
Although not often discussed as an explicit part of interactional sociolin¬
guistics, the notion of taking the role of the other is critical to the foundations
of interactional sociolinguistics because it is critical to the process of inter¬
action itself. Learning the specific cluster of behaviors normatively expected
of one who occupies a particular social position (Goffman 1961) is critical to
the process of socialization into a specific community. As argued by George
Herbert Mead (1934), this process is made possible by the symbolic resources
provided through language and evidence of shated meanings provided during
communication. To oversimplify a great deal, we learn the standards of nor¬
mative behavior by observing how others respond to us, anticipating another’s
response, and incorporating a kind of generalized response into our own
repertoire of actions and meanings. Although this process is critical to the
emergence of communicative competence (as well as the development of a
self), it hardly ends once we are socialized. Not only is much of what we say
explicitly or implicitly oriented toward reception by a hearer, but we often
continue to display an awareness of norms and standards in the way we use
language and the way we interact with others. And speaking a language is
itself a process that requires symbolically putting oneself in the other’s place
in order to know how to tailor one’s information (syntactically, semantically,
and pragmatically) so that it will be comprehendable to that other. What
speaking for another thus represents is the ritualization - the formal display
- of a process that is at the very crux of social interaction: speaking for another
can be seen as the linguistic submersion of the self in the interactive process
itself.

3.4 Summary of sample analysis

We began this section by focusing on a single utterance said by Zelda about

Irene to Henry and me: She’s on a diet. At one level of analysis, this is an
utterance that accounts for a rejected offer issued through a question (chapter
3). Here we characterized this utterance as a quite different act: one person
“speaks for” another. We situated the meaning of this act in several ways: as
a display of participant alignment, gender identity, interactional frames, and
as a means of building sequential coherence through taking the role of the
other. Each effort to uncover situated meaning depended upon a comparison
with other utterances and interactions with which it is both similar and
different, and upon a movement from a particular utterance to more general
frameworks - or contexts - in which that utterance was interactively and
132 Approaches to Discourse Analysis

socially embedded. Implicit in this procedure was a belief that it is the con-
textualization of an utterance that motivates its use: the contexts in which an
utterance occurs explain why it occurs there.
Note, also, that context allowed us to sidestep the attribution of motivation
to participants. Our sample analysis was, in some ways, an attempt to explain
why a speaker produced a certain utterance at a certain point in time. Asking
why someone does something can lead us into a search for explanations of
human conduct - a domain of inquiry that linguists are probably not prepared
(or eager) to enter. Interactional sociolinguistics tries to avoid imputing inten¬
tions (or any internal motivations or goals) to speakers. Rather than say, for
example, that Henry wanted to “dominate” the others in conversation, we can
say that Henry reoriented the participation framework in ways that main¬
tained his active role as animator and principal. What this approach thus uses
to account for why someone says or does something is not the construct of
“motivation,” but “discourse strategy”: a wide range of expressions, features,
and so on - e.g. politeness (Brown and Levinson 1987) and repetition (Tannen
1990) - are said to serve as techniques (i.e. strategies) suited to the fulfillment
of very broad interpersonal goals (e.g. face wants, involvement). In Goffman’s
(1967a: 13) terms, we may say that these strategies serve as “traffic rules of
social interaction . . . the code the person adheres to in his movement across
the paths and designs of others.” Thus, just as Henry’s way of speaking
reoriented the participation framework, Zelda’s way of speaking (“speaking
for” Irene) served as a strategy for interpersonal solidarity and involvement.
Note, then, that interactional sociolinguistics does not necessarily explain how
a way of speaking conveys a specific intention - although it can seek to explain
how a way of speaking serves a broad interactional goal. Furthermore, because
the nature of that goal is interpreted according to the contexts in which
something is said, what interactional sociolinguistics does is ground motiva¬
tion in context - such that both the meaning and the motivation of an
utterance are contextualized.
We also noted that we need to be careful not to generalize beyond the few
examples that we have discussed here: since each utterance (behavior, inter¬
action, encounter, and so on) is itself situated in a particular way, a generali¬
zation can be developed only through detailed attention to the contexts of
each particular case. The particularities of each case, however, did not prevent
us from trying to view that case in terms of higher level analytical constructs:
we viewed speaking for another as a way of taking the role of another, which
itself, is a resource for sequential coherence (cf. Goffman (1971) and Schiffrin
(1977), who viewed greetings and farewells as a type of access ritual, and
access rituals, in turn, as a type of supportive interchange). (Contrast the
ethnomethodological avoidance of such constructs; chapter 7.)
In sum, each context of an utterance has a crucial analytical role in interac¬
tional sociolinguistics. Considered along with the relatively observable con¬
texts of what was said before and what is said next are the more abstract
contexts of the place of an utterance in a repertoire of ritual interchanges,
 Interactional Sociolinguistics 133

face-saving strategies, or resources for sequential coherence. All are sources of

meaning for what is said and evidence for what is done in talk.

4 Interactional sociolinguistics as an approach to

discourse

We have seen in this chapter that interactional sociolinguistics provides an

approach to discourse that focuses upon situated meaning. Scholars taking this
approach combine the ideas of the anthropologist John Gumperz and the
sociologist Erving Goffman. What Gumperz contributes to this approach is a
set of concepts and tools that provide a framework within which to analyze
the use of language during interpersonal communication; Gumperz views
language as a socially and culturally constructed symbol system that both
reflects and creates macro-level social meaning and micro-level interpersonal
meanings. Speakers use language to provide continual indices of who they are
and what they want to communicate. The work of Erving Goffman also
focuses upon situated knowledge, the self, and social context in a way that
complements Gumperz’s focus on situated inference: Goffman provides a
sociological framework for describing and understanding the form and
meaning of the social and interpersonal contexts that provide presuppositions
for the interpretation of meaning.
We applied the ideas of Goffman and Gumperz to a particular utterance,
trying to uncover its situated meaning and to see how that meaning con¬
tributed to the processes and outcome of interaction. We analyzed She’s on a
diet in several different ways: we began with its location in an action structure
(similar to those we discussed in chapter 3) and then went on to consider its
role in a participation framework within the micro-structure of an interaction
and as a gender-based involvement strategy. These analyses viewed speaking
for another as a discourse strategy used by males and females to realize
different interactional goals - as a contextualization cue through which par¬
ticipants signal identity and alignment. (Note that in viewing the social func¬
tion of contextualization cues at a micro-level of status and power, we are
extending Gumperz’s view that these cues signal status and power at a macro¬
level of social relations.) Finally, we also considered how “speaking for an¬
other” is one way of taking the role of the other, and how this, in turn, is a
resource for sequential coherence.
There were still other contexts in which we could have situated the meaning
of “speaking for another.” As Goffman’s work shows, for example, all inter¬
active activity is socially organized at multiple levels: all utterances are situated
within contexts such as “occasions,” “situations,” or “encounters” that not
only provide structure and meaning to what is said, but may themselves be
134 Approaches to Discourse Analysis

organized by what is said (e.g. Goffman 1963). What Gumperz stresses is the
interpretive importance of contexts, including, of course, the occasion in
which an utterance is produced. Contextualization cues provide information
allowing participants to interpret the meaning of what is said; one way that
they do so is to locate an utterance within an interpretive frame identifying
an encounter as a particular kind of occasion or situation. Thus, another
source of contextual meaning - and interpretations based on that meaning -
is the overall structure of an occasion (see chapter 5).
Our means of analysis included comparing the particular utterance She’s on
a diet with other utterances whose text and context were similar in some ways,
but different in others. Although much of our discussion focused on inter¬
actions other than that in which She’s on a diet occurred, we always selected
(and analyzed) those interactions in terms of what they could contribute to
our understanding of She’s on a diet.
Our overall effort also depended upon a movement from a particular utter¬
ance to more general frameworks - or contexts - in which that utterance was
interactively and socially embedded. Implicit in this procedure was a belief
that it is the contextualization of an utterance that motivates its use: the
contexts in which an utterance occurs explain why it occurs there. Consistent
with this belief, we tried to identify different aspects of contexts, the organiz¬
ational and interpretive role of contexts, and the contextualizing role of
different ways of speaking (i.e. as contextualization cues or meta-messages).
In sum, interactional sociolinguistics views discourse as a social interaction
in which the emergent construction and negotiation of meaning is facilitated
by the use of language. Although the interactional approach is basically a
functional approach to language, its focus on function is balanced in import¬
ant ways. The work of Goffman forces structural attention to the contexts in
which language is used: situations, occasions, encounters, participation frame¬
works, and so on, have forms and meanings that are partially created and/or
sustained by language. Similarly, language is patterned in ways that reflect
those contexts of use. Put another way, language and context co-constitute
one another: language contextualizes and is contextualized, such that language
does not just function “in” context, language also forms and provides context.
One particular context is social interaction. Language, culture, and society are
grounded in interaction: they stand in a reflexive relationship with the self,
the other, and the self-other relationship, and it is out of these mutually
constitutive relationships that discourse is created.

Exercises

1 Analyze the excerpt in appendix 3 (or any other of your own choosing) in
terms of participant alignments, footing, and contextualization cues.
 Interactional Sociolinguistics 135

• What kind of interaction is going on? What makes it sound like a hostile
(or sociable: Schiffrin 1984a) argument? A lecture? A friendly discus¬
sion? An interview? Anything else?
• What aspects of the utterances (e.g. contextualization cues) signal
shifts in participation structure? For example, when and how are the
interlocutors cooperating? Competing? Why do alignments seem to
shift, and how do they do so?
• Are there contextual presuppositions that are necessary to make sense
out of what is being said? If so, how is the importance of these pre¬
suppositions conveyed?
• What social and interactional identities are relevant to the interaction?
How are these identities being managed through the discussion?
2 Encounters are often bracketed by relatively straightforward interactional
rituals (i.e. greetings) which first open a period of interpersonal access (see
Goffman 1967; Schegloff 1972a; Schiffrin 1977) and then close that period of
access (Schegloff and Sacks 1973). Yet the exchange below seems to combine
parts of both an opening and a closing.

Professor A is in her office preparing for class (the door is open) when Professor
B walks by.
A: (looking into the hallway) Hi.
B: (walking by without stopping) Hi.
A few minutes later, B is in his office and A walks by.
A: (walking by without stopping) See you.
B: Hi.
On your way to class?
A: Cool, calm, and collected!

How would you explain this interaction? Under what circumstances can such
an exchange occur?
3 Everyday stories provide useful data for an interactional perspective: one
can analyze not only how the story is situated in an ongoing social interaction
(e.g. Jefferson 1978; Ochs et al. 1991; Schiffrin 1984b) but also how social
interactions within the experience being recreated create a story world (e.g.
Labov 1981; chapter 3 of this book). One linguistic device used to report social
interaction in narratives is quoted speech, or what Tannen (1989: chapter 4)
calls constructed dialogue: the storyteller animates the voices of characters in
the story. Apply an interactional analysis to the reported interactions (including
constructed dialogue) in the "I was robbed" story in chapter 6 of this book or
to a story that you have collected. For example, what social actions and inter¬
actions are reported? What participation frameworks and alignments are cre¬
ated within the story world? How does the story world justify the speaker's
telling of the story in the ongoing interaction?
4 Linguists often speak of some of the same words and expressions as
discourse markers and interjections, even though these groups of items are
defined differently. Discourse markers are sequentially dependent expressions
that optionally initiate utterances (Schiffrin 1987a). Interjections are forms that
occur "predominantly as minor sentences, entering into few or no construc¬
tions other than parataxis" (Bloomfield 1933: 176ff.). But some terms (e.g. oh)
136 Approaches to Discourse Analysis

are said to be both discourse markers and interjections, and some analysts view
both sets of items through a single functional lens (e.g. Goffman's 1981e ana¬
lysis of response cries). Use an interactional sociolinguistic approach to ident¬
ify and analyze the discourse markers and interjections in a sample of data.
Compare the interactional functions of these expressions to the function of
mmhmm, ummm, and other vocalizations that fill conversational space without
contributing referential meaning.

Notes

1 Note, also, the constant stream of new diet books and the prevalence in our culture
of food-related neuroses. Interestingly, however, there are also opposing values, e.g.
the association between weight and prosperity (I have heard portly men described as
prosperous looking) or the idea that we have “to fatten up” to be healthy.
2 Such familiarity can be based on either personal or positional knowledge about the
other: for example, personal knowledge might allow me to make a decent guess about
what my father wants for dessert (thus leading me to offer him something that he is
not likely to reject); positional knowledge might allow an employee to make a good
guess about what kind of accommodations an employer wants on a business trip
(thus leading an employee to offer a choice of hotels likely to contain a place that is
appropriate).
3 Note that the role of should (a modal of obligation) is critical here, for it invokes
Henry as an authority about Irene’s own affairs. We can see the importance of should
quite easily by imagining that, instead of saying She’s on a diet, Zelda had said either
She should be on a diet or You shouldn’t take it. Regardless of whether reference to
Irene is third person (as an unaddressed recipient) or second person (as an addressed
recipient), we would interpret Zelda as invoking authority - and as being as intrusive
as Henry. Similarly, if instead of saying She should be, Henry had just said She used
to be and then gone on with his description of Irene’s family, we would be more
likely to interpret Henry as building solidarity with Irene.
4 Efforts by one speaker to prompt another to provide information can be seen as
ritualized displays of more instrumentally motivated clarification requests. Instead
of B asking A for clarification about what A has said, B tries to get A to provide
self-clarification in a way consistent with B’s notion of what is “correct.” Both the
ritualized and instrumental versions of this move reveal the division of interactional
responsibility and control that Ochs (1985) calls an “expressed guess” strategy.
5 In Tannen’s work, the contextualizing role of occasions and situations can be con¬
veyed in terms of frames (i.e. structures of expectations) or meta-messages (Tannen
1979; Tannen and Wallat 1986): utterances are not only framed by occasions and
encounters, they may themselves convey meta-messages about the initiation, continu¬
ation, alteration or closing of a frame. Similarly, contextualization cues may be said
to provide meta-messages about how to contextualize or frame an utterance (Tannen
1984).
5 The Ethnography of Communication

1 Introduction
A «
The ethnography of communication is an approach to discourse that is based
in anthropology and linguistics. As we see in this chapter, this approach is the
most encompassing of all those considered. Not only does it focus upon a
wider range of communicative behaviors than the other approaches, but built
into its theory and methodology is an intentional openness to discovery of the
variety of forms and functions available for communication, and to the way
such forms and functions are part of different ways of life. In addition, the
ethnography of communication is not an approach that can “simply take
separate results from linguistics, psychology, sociology, ethnology, as given,
and seek to correlate them” (Hymes 1974a: 20, fn.6). Rather, it is an approach
that seeks to open new analytical possibilities (by finding new kinds of data
and asking new questions) and to propose new theories. It seeks to do so by
analyzing patterns of communication as part of cultural knowledge and beha¬
vior: this entails a recognition of both the diversity of communicative possi¬
bilities and practices (i.e. cultural relativity) and the fact that such practices
are an integrated part of what we know and do as members of a particular
culture (i.e. a holistic view of human beliefs and actions).
The key figure responsible for the development of the ethnography of com¬
munication is Dell Hymes. After describing the basic assumptions and con¬
cepts of this approach (section 2; see also Saville-Troike 1982), I apply it to
discourse by focusing on one particular speech act - questions - in two
different varieties of a speech event - interviews (section 3). Section 4 sum¬
marizes the ethnographic approach.

2 Defining the ethnography of communication

Although the ethnography of communication was developed by Hymes in a
series of papers written in the 1960s and 1970s (many of which are collected
138 Approaches to Discourse Analysis
in his 1974 Foundations in Sociolinguistics: An Ethnographic Approach), the
roots of this approach reach back to Edward Sapir’s (1933) movement away
from the study of sociocultural form and content as “product” toward their
study as “process” (Hymes 1974a: 20, fn.6). Also important was the emphasis
of the Prague School of linguistics (e.g. Mathesius 1924) on the penetration
of language structure by function. A more contemporary impetus for the
ethnography of communication as a particular mode of inquiry stemmed from
Hymes’s observation of theoretical and methodological difficulties in two
different fields: anthropology and linguistics. After explaining how some of
the central assumptions and constructs of the ethnography of communication
stem from both of these fields (2.1), I outline the methodology of this approach
(2.2) and note its relation to other approaches to discourse (2.3).

2.1 “Communication” in anthropology and linguistics

The ethnography of communication builds a single integrated framework in

which communication has a central role in both anthropological and linguistic
studies. Although I explain this role by discussing each field separately, we
will see that the key concepts and methods intentionally bring together the
two separate starting points, building an interdependence between them.
Linguistics and anthropology are disciplines whose data, problems, methods,
and theories are often seen as clearly distinct from one another. However, one
area in which both fields share an interest is “communication.” Since language
is the central means by which people communicate with one another in
everyday life (see chapter 11), understanding communication is an important
goal for linguists. The understanding of communication is also important for
anthropologists: the way we communicate is part of our cultural repertoire for
making sense of - and interacting with - the world. As Hymes observed,
however, anthropologists often ignore language as cultural behavior and/or
knowledge, neglecting the ways that language is a system of use whose rules
and norms are as integral a part of culture as any other system of knowledge
and behavior (e.g. kinship, or political, systems). Thus, the status of linguistic
communication as a grammatical system that is used for communication and
that is part of culture - and a framework for analyzing it as such - was
surprisingly neglected prior to Hymes’s work.
We noted above that anthropologists often pay little attention to language
as cultural behavior and/or knowledge. Note that I have assumed here that
“behavior” and “knowledge” are both “part of” culture. It is important to
point out, however, that it is not always easy to find agreement among
anthropologists as to what comprises “culture” or the locus of “culture” itself
(Kroeber and Kluckhohn 1952, for example, review over fifty definitions of
culture). The view that Hymes seems to adopt is that culture is a system of
ideas that underlies and gives meaning to behavior in society (cf. Keesing
1974). (As we see in a moment, however, this does not preclude the possibility
 The Ethnography of Communication 139
that behavior and knowledge can be analytically separated.) Culture thus
comprises a general “world view”: a set of assumptions and beliefs that orient
and organize the way people think, feel, and act.
However, not every aspect of culture - not every part of our cognitive
“blueprint” - needs to be shared (i.e. known) by all members. “To restrict the
concept of the cultural to something shared to the limits of a community is
an arbitrary limitation on understanding, of both human beings and the
cultural” (Hymes 1974a: 20, fn. 6). The possibility of differential knowledge
is also implied in Hymes’s (1973) point that members of a culture may have
available to them different forms, and be differentially comptetent in, the way
they draw upon a communicative repertoire (or the parts of the repertoire
from which they choose). Thus, one can differentiate the ability to engage in
a particular meaningful behavior from the fact of that behavior itself: “what
is distinctively cultural, as an aspect of behavior or $f things, is a question of
capabilities acquired or elicited in social life, rather than a question of the
extent to which the behavior or things themselves are shared” (Hymes 1974a:
20, fn. 6). But, as noted above, neither knowledge nor behavior has to be
available to, or realized by, every member. Freeing both knowledge and
behavior from a “sharedness” requirement allows almost any piece of knowl¬
edge (or any behavior) to be part of culture:

The frequency and spread of a trait is important, but secondary, so far

as concerns the criterion for its being a product of cultural behavior, as
having a cultural aspect. A sonnet, for example, is such a product,
whether or not it goes beyond a desk drawer, or even survives the
moment of completion. (Hymes 1974a: 21, fn. 6)

Thus, neither knowledge nor behavior need be completely replicated to be part

of culture. (Compare Hymes’s (1972: 54) definition of a speech community as
“a community sharing rules for the conduct and interpretation of speech, and
rules for the interpretation of at least one linguistic variety.” See also his
(1984) discussion of linguistic problems in defining the concept of “tribe.”)
We have suggested thus far that language is a system of use whose rules and
norms are an integral part of culture. We also noted that the locus of culture
is knowledge. Thus, we might say that language use in speech situations,
events, and acts (2.2) helps realize the cultural norms that underlie the way
we act toward one another: culture is continually created, negotiated, and
redefined in concrete acts between persons who are participating in some kind
of interactive situation (cf. Geertz 1973; Malinowski 1978; Ochs 1988). Thus,
the way we communicate with each other is constrained by culture (simply
because it is a part of culture), but it also reveals and sustains culture. From
an analytical standpoint, an analysis of the patterns that are formed when we
communicate thus contributes to our understanding of culture. Language use
is also a type (and a part) of social behavior in many different institutional
realms (e.g. political, economic, religious, family) that are themselves bound
140 Approaches to Discourse Analysis

to culture. Thus the norms that guide communication also reflect, and help
constitute, social institutions.
Hymes argues that ethnographers can analyze communicative patterns using
the traditional method of anthropological research: participant observation.
By participating in a wide range of activities endemic to the life of a particular
group of people, one attempts to replace one’s own way of thinking, believing,
and acting with a framework in which what is done by the members of another
group starts to seem “expected” and “natural.” The challenge faced by an
anthropologist is thus, in some ways, similar to that faced by any neophyte:
an anthropologist has to learn what native members already know about how
to “make sense” out of experience. Although ways of speaking (and more
generally, communicating) are clearly part of this knowledge, recognition and
analysis of those patterns (as noted above) were not often goals of anthropo¬
logical fieldwork. It is this that Hymes sought to rectify.
Linguists ignored the study of communicative patterns and systems of lan¬
guage use for reasons quite different from those of anthropologists. Chomsky’s
(1957, 1965) reformulation of the goals of linguistic theory excluded the
analysis of performance (cf. de Saussure’s “parole”), focusing theoretical in¬
terest instead on competence, i.e. tacit knowledge of the abstract rules of
language (cf. de Saussure’s “langue”). Rather than concentrate linguistic the¬
ory on competence, Hymes proposed that scholarship focus on communicative
competence: the knowledge governing appropriate use of grammar. Knowl¬
edge of abstract linguistic rules is included in communicative competence. But
also included is the ability to use language in concrete situations of everyday
life: the ability to engage in conversation, to shop in a store, to interview (and
be interviewed) for a job, to pray, joke, argue, tease, warn, and even to know
when to be silent. Furthermore, the study of language in use - the study of
how we are communicatively competent - contributes “in an empirical and
comparative way [to] many notions that underlie linguistic theory proper”
(Hymes 1974a: 20; also Hymes 1981), simply because it is not easy (either
theoretically or methodologically) to separate areas of language that are insu¬
lated from cultural and social processes, from those that are vulnerable to such
processes (Ochs 1988: 3).
We noted above that the ethnography of communication builds a single
integrated framework in which communication has a central role in both
anthropological and linguistic studies, such that key concepts and methods
bridge the two separate starting points. One such concept is communicative
competence: knowledge of grammar and knowledge governing appropriate use
of grammar. The methods for studying communicative competence are also
integrative (2.2). The ethnography of communication shares with much tradi¬
tional anthropology a concern for holistic explanations of meaning and beha¬
vior, i.e. explanations that locate particular behaviors (including, but not
limited to, utterances) in a wider framework of beliefs, actions, and norms.
Also shared with anthropology is an emphasis on how meanings and behaviors
need to be understood in an analytical framework in which comparisons
 The Ethnography of Communication 141
establish not only what is different in different cultures (i.e. the “diversity”
noted on p. 143), but also what is potentially the same. The discovery of
universals, however, is based only upon the ethnographic discovery of particu¬
lars: “one cannot only generalize the particularities, but also particularize the
generalities” (Hymes 1974a: 9).
The particularities that ethnographers discover are particularities of lan¬
guage use. However, in keeping with the partial linguistic heritage of the
ethnography of communication, these particularities also reside in linguistic
form and structure itself: the form of a message (and the rules governing
that form) is as critical to interpretation of particular functions as its con¬
tent (Hymes 1972b: 59). One way that Hymes makes this point is by referring
to Ernst Cassirer’s (1961: 99) analysis of the necessity for art of “two fun¬
damental factors, which constitute the whole of tl](e work only by means of
their union and interpenetration.” Hymes (1*981: 9) suggests that “interpreta¬
tion that excludes speech falls short, as would treatment of painting that
excluded paint.”1 And speech is to be analyzed as linguistic structure within
a relativistic (i.e. “emic” (2.2)) and holistic (i.e. “ethnographic”) mode of
inquiry:

The essential method ... is simply persistence in seeking systematic co¬

variation of form and meaning. The spirit of the method is “structural”
in the sense of Sapir’s linguistics, “emic” and “ethnographic” in the
sense of concern for valid description of the individual case. (Hymes
1981: 10)

2.2 Methodology: an etic grid for ethnography

We noted above that ethnographers of communication analyze communicative

patterns using the method of participant observation: a key goal is to learn
what members of a culture know about how to “make sense” out of experience
and how they communicate those interpretations. To this end, Hymes pro¬
posed a methodology by which to discover “what counts” as communicative
events. The methodology is based on the distinction between “emic” and
“etic.” Linguists studying the sound system of an unfamiliar language try to
discover phonemic patterns (i.e. what sounds are “meaningful in a particular
language) with the help of a phonetic classification (i.e. what sounds are
physically possible). So, too, one can discover communicative units with the
help of a classification system that dissects communication into the compon¬
ents of which it may be potentially comprised. One can then also discover
communicative patterns (e.g. a systematically differentiated inventory of events)
formed by interrelationships among components.
The classificatory grid that Hymes (1972b) proposed is known as the SPEAK¬
ING grid: each letter is an abbreviation for a different possible component of
communication.
142 Approaches to Discourse Analysis

S setting physical circumstances

scene subjective definition of an occasion
P participants speaker/sender/addressor
hearer/receiver/audience/addressee
E ends purposes and goals
outcomes
A act sequence message form and content
K key tone, manner
I instrumentalities
channel (verbal, nonverbal, physical)
forms of speech drawn from community repertoire
N norms of interaction and interpretation
specific proprieties attached to speaking
interpretation of norms within cultural belief system
G genre textual categories

The SPEAKING grid can be used to discover a local (i.e. culturally relative)
taxonomy of communicative “units” that are “in some recognizable way
bounded or integral” (Hymes 1972b: 56). The largest such unit is the speech
situation: the social occasion in which speech may occur (e.g. meals). Although
speech situations may provide the setting or scene in which speech occurs, they
are not themselves governed by a single set of rules. The next unit is the speech
event: “activities, or aspects of activities, that are directly governed by rules
or norms of the use of speech” (Hymes 1972b: 56). The smallest unit is the
speech act: although Hymes (1972b) does not explicitly define this, his ex¬
amples include acts that can be defined through their illocutionary force (chap¬
ter 3; e.g. commands, greetings), as well as those that cannot be so defined
(e.g. jokes). Although discourse is important to all the units, it seems to be the
speech act level that is most fundamental to the local, turn by turn management
of discourse: “Discourse may be viewed in terms of acts both syntagmatically
and paradigmatically: i.e., both as a sequence of speech acts and in terms of
classes of speech acts among which choice has been made at given points”
(Hymes 1972b: 57). Finally, the larger units in the set embed the smaller: a
party is a speech situation; a conversation during the party is a speech event;
a joke within the conversation is a speech act (Hymes 1972b: 56).

2.3 Summary: the integration of diversity

In chapter 2, we used Hymes’s (1974b) comparison between structural and

functional approaches in linguistics as a way of differentiating two definitions
of discourse analysis. The ethnography of communication falls squarely within
the functionalist paradigm: in fact, we might take Hymes’s presentation of this
paradigm as the presentation of core premises of the ethnographic approach.
I present these features again here:
 The Ethnography of Communication 143
1 Structure of speech (act, event) as ways of speaking.
2 Analysis of use prior to analysis of code; organization of use discloses
additional features and relations; shows code and use in integral
(dialectical) relation.
3 Gamut of stylistic or social functions.
4 Elements and structures as ethnographically appropriate.
5 Functional (adaptive) differentiation of languages, varieties, styles;
these being existentially (actually) not necessarily equivalent.
6 Speech community as matrix of code-repertoires, or speech styles
(“organization of diversity”).
7 Fundamental concepts taken as problematic and to be investigated.

As indicated in these features, the ethnography of communication is reluctant

to assume a closed set of language functionsrfhat aj%>ly equally to all languages
and all speech communities (3, 5): rather, diversity is assumed, and the limits
of diversity are explored. Similarly, it is categories of language use,e.g. acts,
events, not language structure, that have theoretical priority: language use
itself is patterned and structured (1) and these regularities are central to one’s
discovery of properties of the linguistic code (2). As noted above, what can
be assumed is diversity: structures and functions are relative adaptations to
different cultural systems (4, 5); a single speech community is itself an “or¬
ganization of diversity” (6). (The nature of such diversity, of course, needs to
be empirically discovered before any generalizations can be proposed.) Note,
also, that even the concepts that seem most fundamental - text, context,
structure, function, communication - are “problematic” and in need of investi¬
gation (7). (It is because of this that the ethnographic approach can encom¬
pass (or even redefine) the other approaches to discourse discussed in this
book; see below). Hymes (1974a: 9) summarizes functionalist assumptions in
this way:

primacy of speech to code, function to structure, context to message, the

appropriate to the arbitrary or simply possible; but the interrelations
always essential, so that one cannot only generalize the particularities,
but also particularize the generalities.

Consider, also, that the ethnography of communication is the most integrat¬

ive approach of those considered in this book: this reflects the fact that culture
itself encompasses or embraces a totality of knowledge and practices. Thus,
an ethnographic approach to discourse in general (and to questions in particu¬
lar; section 3) can combine speech act and interactional approaches within a
larger framework of inquiry. (I consider its relationship to the other ap¬
proaches yet to be discussed - pragmatics, conversation analysis, variation
analysis - in chapter 12.) Constitutive rules - knowledge of the conditions by
which we recognize a speech act — are part of our cultural knowledge, as are
the principles by which we organize our interactions and display our identities.
144 Approaches to Discourse Analysis

Since our knowledge of what words and meanings are appropriate for a given
time, place, purpose, and so on is cultural knowledge, the use of contextual-
ization cues to convey the contextual presuppositions of an utterance displays
our communicative competence as a member of a certain culture and situates
us in a particular web of beliefs and actions specific to that culture.
It would be misleading, however, to think that an ethnographic approach to
discourse just adds a new component (“culture”) to the basic material of
speech acts and interactional sociolinguistics. Rather, to paraphrase Durkheim
on the sui generis nature of society, an ethnographic approach creates a whole
that is greater than the sum of its parts: it seeks to define the basic notions of
the other approaches to discourse simply because it views all phases and
aspects of communication (from the cognitive to the political) as relative to
cultural meanings (e.g. Sherzer 1983). We see in the next section some of the
ways that this approach applies to the analysis of questions.

3 Sample analysis: questions as speech acts in

speech events

We saw in section 2 that the ethnography of communication is an approach

to discourse that studies communicative competence. It does so by discovering
and analyzing the patterns (structures) and functions of communicating that
organize the use of language (in speech situations, events, and acts) in the
conduct of social life. This section applies the ethnographic approach to the
analysis of one particular speech act (questions) in two varieties of a speech
event (interview).
We considered questions in earlier chapters from the point of view of speech
act theory (chapter 3) and interactional sociolinguistics (chapter 4). In chapter
3, we discussed the felicity conditions underlying questions, beginning with
the issue of how we might recognize a particular utterance {Y’want a piece of
candy?) as a question. We also considered how a single utterance might enact
more than one speech act (question, request, and offer) and how more than
one utterance (e.g. a story) could realize a single speech act. Finally, we
considered the way speech acts could be sequentially located and sequentially
defined. Our approach in chapter 4 was quite different. Although we discussed
the same utterance (Y’want a piece of candy?), our initial focus was on the
way that utterance was responded to and how those responses were situated
in a participation framework. We then went on to consider how a particular
way of speaking displayed social identity and the more general implications
of a particular interactional move (“speaking for another”). The broader
issues raised by speech act and interactional approaches also differed. The
speech act approach led us to consider issues such as the following: the role
of speech act knowledge in communication, the multifunctionality of utter-
 The Ethnography of Communication 145
ances, the functional basis of discourse structure. The interactional approach
raised other issues: how utterances both reflect and create context, how
“much” context (and what “kind of”) context is used in interpreting the
social meaning of an utterance, how utterances are potential indicators of social
meaning.
Although we will not begin with the same utterance (Y’want a piece of
candy?) here, our sample analysis in this chapter will provide yet another
framework within which this utterance (and others that are functionally sim¬
ilar) could be considered: as a speech act within a speech event; more specif¬
ically, as a question within an interview. After briefly using the SPEAKING
grid to describe two variants of the “interview” speech event (3.1), I illustrate
some of the ways that questions are normatively situated within these speech
events and related to their communicative properties: questions during inter¬
views at a library reference desk (3.2); questions**during sociolinguistic re¬
search interviews (3.3). The analysis goes well beyond discussion of the form,
meaning, and use of questions themselves. Although I did not do ethnographic
fieldwork, I try to reflect aspects of the ethnographic approach by drawing
upon my own knowledge as a “native” informant who has participated in and
observed both kinds of interviews. My descriptions preliminary to the analysis
of the questions themselves thus try to reveal the circumstances surrounding,
and perhaps accounting for, the normative meanings of those questions as
particular types of verbal behavior. A result of this method is that we will see
how ethnographers might address some of the same issues considered earlier:
the relationship between form and function in questions (cf. chapter 3), and
the contextual meanings of questions and responses (cf. chapter 4). Thus, we
will see how the form-function relationship realized in single utterances is
related to specific aspects of context. The systematic analysis of context as a
framework within which form meets function is an important feature of the
ethnographic approach, since as Hymes (1972a: xxviii) observes, form is not
a reliable indicator of illocutionary force: “one and the same sentence, the
same set of words in the same syntactic relationship, may now be a request,
now a command, now a compliment, now an insult, depending upon tacit
understandings within a community.” Section 3.4 summarizes the sample
analysis.

3.1 Interviews as speech events

Interviews are a speech event with which many people in American society
have become familiar: “Interviewing has become a powerful force in modern
society. Starting almost from birth, we are confronted by questions posed by
educators, psychologists, pollsters, medical practitioners, and employers, and
we listen to flamboyant interviewers on radio and television” (Briggs 1986:
1). Although all interviews may very well share some common core - the core
that allows them all to be understood as different variants of the same speech
146 Approaches to Discourse Analysis

event - they also differ among themselves. Interviews that serve a gatekeeping
function (Erickson and Schultz 1982), for example, are asymmetric speech
situations during which a person who represents a social institution seeks to
gain information about the lives, beliefs, and practices of people outside of
that institution in order to warrant the granting of an institutional privilege.
In other interviews (e.g. survey research interview, opinion polls), an institu¬
tional representative still seeks information from and about outsiders, but that
information will be taken back to members of the institution, with no direct
or lasting effect on the lives of those who have provided the information.
Interviews broadcast to a radio or television audience have additional (or
alternative) functions (e.g. entertainment) that create quite different con¬
straints on what is said, e.g. to have the guest talk to the audience about things
the audience wants to know (SF Focus, July 1992). Despite differences such
as these, questions are central to the information-gaining function of all
interviews.
The central role of questions for interviews makes interviews a convenient
speech event in which to locate an analysis of questions. The centrality of
questions for interviews, however, does not mean that one can analyze ques¬
tions from interviews by extracting them, as a group unto themselves, from
the interviews in which they were asked. The multifaceted relationship be¬
tween utterances and their contexts in general means that a great deal of
information about questions and answers would be lost if one considered them
as a dialogic pair isolated from surrounding actions and beliefs. As Briggs
(1986: 104) suggests, “the interview must be analyzed as a whole before any
of its component utterances are interpreted.” The questions asked in inter¬
views, then, cannot be interpreted apart from analysis of the interview itself.
Earlier we noted that systematic attention to, and analysis of, context is
one of the most critical parts of an ethnographic approach to discourse. We
also noted that a convenient way to “segment” contexts from one another (as
units of analysis) is in terms of speech acts, speech events, and speech situ¬
ations. These are all units that ethnographers analyze, and they can all be
classified according to Hymes’s SPEAKING grid (2.3). In the rest of this section,
I illustrate the use of the SPEAKING grid as a heuristic by which to describe
two types of interviews (the speech event) in which questions (the speech act)
occur.
Table 5.1 uses Hymes’ SPEAKING grid as a way of discovering the com¬
municative features and qualities that underlie our knowledge of how to
participate in two types of interviews and our identification of two events as
certain “kinds” of occurrences. Although I do not comment in detail on each
component from table 5.1 in the analyses, I will draw upon this grid when
discussing questions in each interview in later sections. I focus particular
attention on the ACT SEQUENCES and their relation to PARTICIPANTS and
ENDS. (In all discussions to follow, I use capital letters to highlight the status
of components from the SPEAKING grid.) Since questions are a key part of
interviews, we can also examine questions themselves as a way of under-
 The Ethnography of Communication 147
standing the structure of the speech events in which they are used. At the same
time, we will learn a great deal about questions because we will see how they
are used by participants whose identities are relative to a particular speech
event, to the goals defined by that speech event, and to the other acts accom¬
plished during that speech event.

Table S.l Communicative components of different interviews

Reference IV Research IV
SETTING IVee’s workplace (desk) IVee’s home (kitchen, l.r.,
d.r.)
PARTICIPANTS IVer: patron IVer: researcher
IVee: librarian IVee: researched
ENDS overt ^overt/covert
complementary some complementary
IVer/ee lacks information IVer lacks info
IVee/er gives info IVee gives info
ACT SEQUENCE opened by either opened by IVer
IVer: makes query IVer: requests info
IVee: requests clarification IVee: provides info
IVer: provides clar IVee/IVer: requests clar
IVee: resolves query IVer/IVee: provides clar
recycles recycles
rearranges rearranges
range of acts: narrow range of acts: varies;
partially controlled by IVer
closure by both depends on closure by either depends
resolution of query; on compliance with
closure is final negotiated time constraints;
closure is temporary
KEY narrow range medium range
INSTRUMENTALITIES verbal/nonverbal; physical verbal, maybe physical
materials materials
NORMS interaction is based on interaction is based on
need for information need for information
GENRE narrow range medium range

3.2 Reference interviews

American public libraries typically have departments that specialize in “refer¬

ence material”: relatively technical information, or information of interest to
a specialized audience, that is stored in formats (e.g. encyclopedias, manuals,
specialized journals, archives, computer disk) not always available outside of
the library setting. Public access to such materials is often controlled by a
specially trained librarian (a “reference librarian”) who is posted at a specific
desk (“reference desk”).
148 Approaches to Discourse Analysis

Interactions during which the reference librarian (L) increases the access of
a library user (often called a “patron,” P) to specific reference materials are
called “reference interviews.” These interviews arise because libraries regulate
access to reference material in several different ways. The most restrictive
regulation on materials is physical: the reference material can be kept at the
reference desk itself. In such cases, P must ask L for use of the specific material.
Even when such materials are physically accessible - when they are in the
generally available, circulating sections of the library known as the “stacks,”
for example - P may need help locating material. The asymmetric distribution
of knowledge between L and P also regulates P’s access to reference material.
When P is not familiar enough with either the reference materials or the format
of a particular set of materials to locate information pertaining to a particular
topic, P has to ask L for help not just in locating a source of material, but also
in identifying the appropriate material in which to search. Finally, L may
also offer P help in actually formulating a specific need - in making a query
- in a way that is appropriate to the resources available in the library.
The reference interview is the shorter and more focused of the two types of
interviews to be considered here. As noted in table 5.1, the range of ACTS,
the KEY, and the GENRES are all relatively narrow. Both L and P direct their
talk toward resolution of the problem(s) brought to the desk by P: the goal
(ENDS) of each participant is mutually known. Although more specific goals
(based upon different amounts of information about either the informational
need or the library resources) temporarily diverge from the main goal during
the course of the interview, these specific goals are all subordinate to the main
goal.
As suggested above, it is P who has a particular need and L who can fulfill
that need. This complementary distribution of needs and abilities suggests that
several different kinds of speech acts might occur during a reference interview
- speech acts through which P makes his need known to L and through which
L draws upon her abilities to fulfill that need. Even when both participants
use what we might think of as the “same” speech act, the use of that speech
act varies according to participant. As we see in a moment, the questions in
reference interviews are explicitly geared to the event-based roles of partici¬
pants and their complementary goals: P seeks to gain information and L seeks
to provide that information. Thus, although both participants ask questions,
the form and function of their questions reflect their roles and the goals
defined by those roles.
In the next sections, I turn from the description of reference interviews as
speech events to the analysis of questions within these speech events. I draw
upon fifteen audio-taped interviews from the reference desk of a public library
to show that there are three kinds of questions in reference interviews: ques¬
tions that make offers, issue queries, and request clarification about queries.
I begin by discussing the questions used to initiate reference interviews (3.2.1):
L’s question makes an offer; P’s question opens a query. I then discuss the
main query driving the exchange and the supporting (and subordinate) re-
 The Ethnography of Communication 149
quests for information (3.2.2); both of these acts can be realized through
questions. Section 3.2.3 summarizes.

3.2.1 Opening the interview

The opening portions of reference interviews are similar to the openings in

service encounters (Merritt 1976).3 After P (cf. customer) makes L aware of
his presence (e.g. by establishing eye contact) at the reference desk (cf. service
post), either party can verbally open the encounter: L can ask a question that
offers help or P can ask a question that initiates a query. (Note that I use the
term “query” rather than request for information so that I can reserve the term
“request” for more specifically defined speech acts.) We will see that the
function of initiating questions clearly establishes th® institutionalized roles of
L and P.
(1) illustrates L’s initiating offers and P’s responses to those offers.

(1) Librarian asks initial question

(A) L: (a) May I help you?
P: (b) Sure.
(c) Um if I don’t find an indication in the card catalog of which
branch the book is to be found,=
l: (d) ZUh huh.
P: (e) =what do I do next.
(B) L: (a) May I help you?
P: (b) Yes.
(c) I’m looking for . . . Fanny Mae, and Fanny Mac?
(d) And Freddie Mac, regulations. (Federal) banks.
L: (e) Okay.
(C) L: (a) Can I help you?
p: (b) Yeh maybe.
(c) I’m uh- I’ve been working up in New York City in the theater
doing acting and stage managing and one thing and another.
(d) And I’m looking to try and get over to public relations which is
why I’m down here.
l: (d) Okay.

Each reference interview in (1) is verbally opened by an utterance {May/can

I help you?) that is conventionally understood as an offer. But notice that these
opening utterances also seem like questions, both syntactically (they are inter-
rogatives) and intonationally (they have final rising intonation). This multi¬
functionality raises some of the same issues that we discussed in chapter 3:
there we used speech act theory to propose a functional relationship between
questions/requests and offers. We also pointed out, however, that these acts
could be independent. Some analyses of indirect speech acts, for example,
150 Approaches to Discourse Analysis

argue that interrogatives may be requests for action without also being lit¬
eral” questions: Can you pass the salt?, for example, seeks action rather than
information about ability. Other interrogatives that seem like questions fall
short on other underlying conditions: rhetorical questions (Athanasiadou
1991), for example, seek no verbal response or action response from others
at all; nor do they address situations about which speakers lack information.
Similarly, not all offers are questions; not all questions are offers.
Because not all interrogatives actually seek information from a respondent,
it is important to ask whether May/can I help youf actually functions as a
question. In chapter 3, we analyzed multifunctionality by finding relationships
among the conditions for each individual act - arguing that a single utterance
might perform several acts if the rules for constituting those acts were similar
in some way. An ethnographic way of addressing multifunctionality supple¬
ments the speech act approach by relating knowledge about speech acts to
knowledge about speech events, and by finding evidence in participants’ be¬
havior for the communicative norms and understandings being proposed
through the more abstract and general analysis.
In order to see whether May/can I help you? is both offer and question, let
us begin by reviewing what we learned in chapter 3 about offers. In our
discussion of Y’want a piece of candy?, we noted that a speaker (S) who makes
an offer does not necessarily undertake an obligation to perform an act (A),
simply because S cannot be certain that the hearer (H) finds A to be in his
best interest. We proposed that some offers issued as interrogatives question
the preparatory condition of promises (that H would prefer A to be done) and
allow S to check on whether A is something that H does indeed want or need.
Such an assessment is critical to the likelihood that what is being offered will
actually end up being provided.
Although offers like Y’want a piece of candy? focus on H and explicitly ask
about hearer-based conditions (H’s needs or wants), other offers focus on S
and explicitly state or ask about speaker-based conditions. Saying I want to
help you, for example, would be an offer that states a speaker-based prepara¬
tory condition, i.e. S wants to do A for H. Offers can also be issued by
questioning a speaker-based preparatory condition - questioning S’s ability to
do what H needs or wants. (Note the similarity with requests that question a
hearer-based preparatory condition, i.e. requests can question H’s ability to
do what S wants.) Thus, offers may be made when it is uncertain that H wants
A, or when it is uncertain that S is able to do A.
To paraphrase what I have suggested thus far, offers may be made when one
(or both) of two conditions are uncertain: H may want (or not want) A, S may
be sure (or unsure) that she can do A. Since both of these circumstances affect
the likelihood that S will undertake an obligation to do A (and thus that A
will actually be provided for H), it is reasonable for S to try to reduce
uncertainty before actually committing to do A. The strategy that we discussed
in chapter 3 was for S to question the hearer-based condition in order to learn
whether H wants A (“Do you want A?”). As noted above, an alternative
 The Ethnography of Communication 151
strategy is for S to question the speaker-based preparatory condition that S is
able to do A (“Can/may I do A”). This question elicits information from H
that will allow S to know whether or not she can do A: as we see later, H’s
response to this question can provide S with a description of what would
constitute “doing A.” Since it is just this information that will allow S to assess
her ability to do A, H’s answer can affirm the speaker-based preparatory
condition and allow S to go on to undertake the obligation to do A.
The discussion above suggests that Can/may I help you? does have a specific
question function as well as act as an offer. This question function rests
partially on the root and epistemic meanings of the modals can and may. The
root meanings of can and may have to do with ability (e.g. “I can/may [be
able to] lift that chair - it’s not that heavy”); the epistemic meanings have to
do with possibility and permission (e.g. “You can/may have that dessert only
if you eat your vegetables”). These meanings/work together when S asks H
whether S can (or may) do A for H. Consider two hypothetical responses to
the offer being considered:

(1') L: Can/may I help you?

pi: Maybe, but you look like you’re pretty busy already.
P2: Maybe, but I’m looking for something that’s marked “lost” in the
card files.

The hypothetical responses from PI and P2 both convey uncertainty about

whether L can really provide help, i.e. they define this help as a “possibility”
but not a certainty. The contingent nature of the “help,” however, arises for
different reasons. Consistent with the root meaning of can or may, the re¬
sponse from PI raises doubts based on L’s ability to carry through with the
offer. Although these doubts reduce the likelihood that A will be possible (the
epistemic meaning), the source of that assessment is L’s ability (the root
meaning, and a speaker-based preparatory condition of offers). Although the
response from P2 also raises doubts about the likelihood of A, these doubts
rest upon P’s own assessment of his needs in relation to available (or un¬
available) resources, and are thus less bound to the root meaning of can
or may.
I have suggested thus far that the question form of Can/may I help you? is
aimed toward the reduction of uncertainty: by questioning a speaker-based
preparatory condition that S is able to do A, Can/may 1 help you? allows H
to provide S with information that bears either directly or indirectly on S’s
ability. We might wonder, however, why S would offer A without being sure
that she is able to do A. It is here that an understanding of the particular
speech event can help: the likelihood that a particular act can actually be
provided is relative to the circumstances of the speech event in which it is either
offered or requested. We can make some simple comparisons to illustrate that
the provision of acts known as “help” is more guaranteed for some acts, and
in some kinds of public service encounters, than others. If one asks a museum
152 Approaches to Discourse Analysis

guard for help in finding the exit, for example, it is virtually certain that that
help can be provided. But other acts fall outside of one’s expectations about
what typically occurs in a specific type of encounter, e.g. asking a museum
guard for help in finding a particular postcard in the museum shop. The status
of still others is somewhere in between, e.g. asking a museum guard for help
in finding a particular painting in the museum.
The reference librarian is similar to the “in between” example just noted.
Recall that the reference librarian is posted at a particular location from which
she is supposed to help P. Note, however, that L really has no information
about the specific needs of any patrons who approach her desk - even though
it is this information that she needs in order to try to provide help. By asking
Can/may I help you?, then, L is requesting that P provide her with information
about what she needs so that she will be able to try to provide what her job
has trained her to do, i.e. help locate information. Thus, the ability of L to
actually fulfill the offer depends upon knowing the kind of help that P is
seeking.
The analysis thus far suggests that the question form and intonation of L’s
initial offer contributes to its speech act function in the interview. This sug¬
gestion was based not only on our understanding of speech act conditions, but
on the linguistic meaning of the utterance (e.g. the modals) and our contextual
knowledge of a particular speech event. This multifaceted approach is consist¬
ent with the ethnographic belief that the function of an utterance resides in
its “text” (i.e. its form and meaning) and its “context” (i.e. knowledge of the
circumstances of a speech event), a relationship that is itself mutually constitu¬
tive (chapter 10).
An ethnographic perspective also seeks evidence for a particular analysis
within participants’ conduct: how do participants themselves display knowl¬
edge of communicative norms and understandings? Two aspects of the form
and content of P’s responses suggest that P treats Can/may I help you? as both
question and offer.
First, P’s response to L may address the literal meaning of Can/may I help
you?, i.e. the actual ability of L to “help” P. Although some patrons just
respond with Yes (as in b in IB) or just describe the sort of help they need (as
I note in a moment), other responses are more informative. Compare (1A) and
(1C):

(1A) L: (a) May I help you?

p: (b) Sure.
(1C) l: (a) Can I help you?
p: (b) Yeh maybe.

Sure and Yeh maybe differ along a scale of certainty - the former conveying
more certainty that L is able to help P than the latter. That P’s responses to
L’s Can/may I help you? can convey such differences suggests that P may bring
a general notion of what kind of help can be reasonably expected to a reference
 The Ethnography of Communication 153

interview and compare his own particular needs to what he thinks L can
provide. Focus for a moment on (1C). P’s Yeh maybe anticipates a problem
with the upcoming query, e.g. it is atypical, inappropriate, difficult to convey.
These doubts are also conveyed in the way P goes on to formulate his query:
Yeh maybe is followed by an unusally detailed description of P’s previous job
and his move from New York, neither of which is directly relevant to his query
itself. (Note that P later states his uncertainty very directly: I don't quite know
how to phrase my question.) Thus, by conveying relative uncertainty that L is
able to help P, P is responding to the “question” meaning of Can/may I help
youf, i.e. P interprets this utterance as a question about L’s ability (the root
meaning) to provide “help.”
Second, P may respond to L’s offer by issuing a query. In all the examples
in (1), P’s affirmation (line b in 1A, IB, and 1C) is followed immediately by
4 li
his query (lines c),

(1) (A) L: (a) May I help you?

p: (b) Sure.
(c) Um If I don’t find an indication in the card catalog of which
branch the book is to be found /conts /
(B) L■: (a) May I help you?
p: (b) Yes.
(c) I’m looking for . . . Fanny Mae, and Fanny Mac? /conts /
(C) L•: (a) Can I help you?
P (b) Yeh maybe.
(c) I’m uh- I’ve been working up in New York City /conts /

It is important to note that P begins a query without L having to elicit a query

more explicitly. That is, the following sequence does not occur:

l: May I help you?

P: Yes.
L: Tell me what I can do.
p: I’m looking for . . .

Notice, however, that L’s ability to provide help depends upon knowing what
kind of help - what specific information - is needed by P. The queries from
P in (1) either state information that P needs (B) or lead into a request for
information (A, C). This information has to be provided by L for P if L’s offer
is to be actualized, i.e. if S is to commit herself to doing A for H. Thus, by
describing the kind of “help” that P needs, P is responding to the “offer”
meaning of Can/may I help you?, i.e. P interprets this utterance as an offer
from L to do A for P.
Not all reference interviews open with an initial question from L. The
examples in (2) show that sometimes it is P who can ask the first question in
a reference interview.
154 Approaches to Discourse Analysis

(2) Patron isks initial question

(A) p: (a) Hi I’m in the business section looking for something on mutual
funds ratings?
(b) Now I don’t even know where to start.. .
(c) could you help me?
(B) p: (a) Yes. Can you tell me where I might find a numismatic catalog
and if it can be taken from the library for a couple of days?

After recognition tokens (Hi (A), Yes (B)) acknowledging L’s implicit avail¬
ability, P uses questions as part of his query about specific information in the
library, e.g. the interrogative forms (Could you help me?, Can you tell me
where). P’s queries also include other forms that fulfill the felicity conditions
of both questions and requests. For example, first person statements about
what P seeks (e.g. I’m looking for (A)) fulfill a preparatory condition of
questions (S lacks knowledge) and the sincerity condition of requests (S wants
A). P’s queries in (2) also provide information about what is being sought:
something on mutual funds ratings (A), a numismatic catalog (B). P may also
provide information about his own knowledge (I don’t even know where to
start (A)) that states a preparatory condition of questions. (See section 3.1.2
in chapter 3 for discussion of the relationship between questions and requests.)
Finally, as noted above, P also asks questions that seek L’s help (Could you
help me? (A), Can you tell me (B)): these questions not only realize the
essential condition of requests, but they compensate interactionally for L’s lack
of an explicit offer and open the way for L to actually begin to offer help.
In this section, we have seen two ways that participants in reference inter¬
views use questions to initiate the body of the interview: L may ask a question
that offers help, P may ask a question that makes a query. Because the
questions used by L and P have different functions in the interview, the options
for realizing these functions differ. Rather than ask Can I help you?, for
example, L may use a nonverbal signal of openness to interaction (e.g. raised
eyebrows; Ekman 1979). And rather than use a question to make a query, P
may use a number of other request forms that also open a phase of joint
formulation and clarification of the reference material being sought (as dis¬
cussed in the next section). Note, finally, that the functions of the initiating
questions examined here clearly establish the asymmetric roles of the partici¬
pants: P solicits a service (P wants information) and L provides a service (L
gives information). Despite this global asymmetry in participant role, the
devices for formulating and eventually satisfying a query are more symmetric¬
ally distributed on a turn-by-turn basis. In the next section, we examine the
way P and L work together to formulate and satisfy a query.

3.2.2 The query: how it is issued, clarified, and satisfied

The formulation and satisfaction of a query are the main ENDS of the refer¬
ence interview: P’s query seeks information and it is up to L to help P find
 The Ethnography of Communication 155

that information.5 The query thus has a central role in organizing the NORMS
by which the ACT SEQUENCE is constructed. We saw in section 3.2.1 that
although the query does not initiate the encounter, it is the query to which
the initiation of an interview leads. We see in this section that questions from
L and P work towards local clarification of the query (on a turn-by-turn basis)
and global resolution of the query (at the level of the entire speech event): L
and P use questions to issue, clarify, and try to resolve the query. Our discus¬
sion will also note that questions are not necessary to the formulation of a
query: other forms can be used for some of the same ENDS.
I begin with an example in which P uses a statement to request a specific
item that is then provided by L.

(3) p: (a) I would like the um Mansfield Chart s^pck chart.

L: (b) [gives it to P]

The statement in (a) is a query even though it does not use a question form
to ask for information. Since P knows specifically what he wants and where
he can find it (at L’s desk), he can use a standard request form (i.e. state a
sincerity condition of requests; see chapter 3, p. 71) to seek that item from L.
Because L can give P the requested item, there are no requests for clarification
of the query and the query is satisfied.
In (4), P also knows what he wants but is less familiar with either its location
or rules about its use. P phrases his query (in a) as a question (can you tell
me) that asks about L’s ability to satisfy the query (i.e. questions a preparatory
condition of requests; see chapter 3, p. 71).

(4) p: (a) Yes. Can you tell me where I might find a numismatic catalog
and if it can be taken from the library for a couple of days?
L: (b) Is numismatics coins?
P: (c) Yeah.
L: (d) Yes there- we do have some circulating books.
(e) It would be on the second floor
(f) and the call number for the coin books is 737.* . *
(g) There’s a stairway to the left of the front door and there’s
someone up there who can help you find . . .
p: (h) Thank you.
L: (i) You’re welcome.

Instead of immediately providing P with the requested information, L checks

the meaning of numismatics (b), proposing a definition of numismatics imme¬
diately following the query. After P affirms the meaning of numismatics with
Yeah (b), L provides both pieces of information requested by P: the location
(where I might find. . . (a)) and how to reach it (d to g), and the fact that
some are circulating books (d). Although this information does not address
the particular item being sought (a numismatic catalog (a)), it tells P how he
156 Approaches to Discourse Analysis

might find (a) that item himself. Thus, L satisfies P’s query (d to g), but
requests (b) and receives (c) clarification of the query before doing so.
P can also check the formulation of the query or request clarification of his
own query; the outcome of these acts may involve work from both P and L.
In (5), P both offers and requests from L clarification of his own query. (5
continues an earlier example (1C).)

(5) l: (a) Can I help you?

p: (b) Yeh maybe.
(c) I’m uh- I’ve been working up in New York City in the theater
doing acting and stage managing and one thing and another.
(d) And I’m looking to try and get over to public relations which is
why I’m down here.
l: (e) Okay.
p: (f) Uh what I’m trying to find out is a source of PR firms, and
related sorts of things that would be interested in people that
deal in the media.
(g) In other words in the- I don’t quite know how to phrase my
question . . .
(h) in the sort of technical end of the image making end of it. . .
(i) the people who deal in commercials in television, radio, that end
of it.
l: (j) Straight public relations?
p: (k) Yeh that’s a great- that’s a good place to start.

After giving reasons for his query (c, d), P presents his query (f) as a request
that presupposes a need for help (through the pseudo cleft what I’m trying to
find out) and states what he needs in very general terms (a source of PR firms,
and related sorts of things) that are relevant to himself and his own previously
stated qualities {that would be interested in people that deal in the media). In
(g), P begins to rephrase his own query (in other words) and conveys uncertainty
about how to make the query (I don’t quite know how to phrase my question
. ..). He then offers a reformulation of his query that modifies the item for
which he is searching: in the media is rephrased as in the sort of technical end
of the image making end of it. . . the people who deal in commercials in
television, radio, that end of it (h, i). L also reformulates the item being requested
(Straight public relations? (j)). When this reformulation is acknowledged by P
(Yeh that’s a great- that’s a good place to start (k)), the query is resolved.
(6) is another example in which P opens the query to reformulation from L.
The reformulation here is facilitated by a question-like feature: rising intona¬
tion. (We already discussed the relationship between rising intonation and
questions in section 3.1.1 of chapter 3.)

(6) P: (c) I’m looking for . . . Fanny Mae, and Fanny Mac?
(d) And Freddie Mac, regulations.
 The Ethnography of Communication 157
(e) (Federal) banks.
L: (f) Okay.

In (c), P uses a “question-like” intonation on one part of his query (Fanny

Mac?): this intonation opens a possible turn transition space in which an item,
whose information status is uncertain, is opened to interactional scrutiny (see
chapter 3, section 3.1.1, and chapter 7, section 3.1.2). Such a device can be
used if P himself is uncertain about how to identify a referent: note that what
follows Fanny Mac? (i.e. Freddie Mac (d)) can be interpreted as a self-repair.
Rising intonation can also be used if P is uncertain about L’s recognition of
the referent or prior use of the referent as a next-mention (chapters 6-8). We
might thus call the utterance produced with this intonation contour an informa¬
tion check, as in section 3.3.2 of this chapter. The important point here,
however, is the interactional function of this intonatfbn within the reference
interview itself. Fanny Mac? opens the way for L to begin her own turn in a
number of fairly constrained ways: L can ask for clarification, show recogni¬
tion of, or provide, the information being sought in the query. When L does
not begin her own turn, however, P continues his turn: P continues to formu¬
late the query (d, e) until L recognizes what P is seeking and takes responsi¬
bility for fulfilling the query (Okay (f)).
Both P and L use rising intonation in (7). Here the initial part of P’s query
is followed by P’s direct request for help and a question: L later questions P’s
formulation of the query.

(7) p: (a) Hi I’m in the business section looking for something on mutual
funds ratings?
(b) Now I don’t even know where to start. . .
(c) could you help me?
l: (d) That exact title is what you’re looking for?

Although P’s query in (7) is issued through a request form (i.e. I’m looking
for (a) states the sincerity condition), P also presents the item for which he is
looking (mutual funds ratings?) with rising intonation. When L does not offer
any verbal acknowledgment or recognition, P continues with another request
format that states his own need more directly (Now I don’t even know where
to start. . . (b)). When P then asks for help, could you help me? (c) realizes
the essential condition of requests. It is after P directly asks L for help that L
checks P’s formulation of the query: That exact title is what you’re looking
for? (d) facilitates the resolution of the query by putting the terms of P’s query
into a category (i.e. “title”) that can be searched within the library resources.
L’s use of rising intonation with this question transfers responsibility back to
P, thus showing again that P and L jointly share in the overall ENDS of the
speech event, i.e. solving the query.
Our examples thus far have suggested that P can use a variety of forms —
questions, statements, phrases with rising intonation - to elicit L s help in
158 Approaches to Discourse Analysis

formulating, and thus resolving, a query. (8) is another example of joint work
from both P and L (note that 8 follows 3).

(8) p: (c) Uh okay do you have the latest Forbes Magazine?

(d) In September they put out a
L: (e) Mutual fund.
P: (f) [Yeh.
l: (g) [Yeh.

In (8), P asks a direct question about a particular item (Do you have the latest
Forbes Magazine? (c)). When L does not respond, P adds information about
the item (In September they put out a (d)) that begins to clarify and explain
the item. It is then that L shows her recognition of what is being sought:
Mutual fund (e) is latched onto P’s supplemental information. The accuracy
of this description is noted by P (Yeh (f)) just as L affirms (Yeh (g)) that this
magazine is available. Thus, P and L work together to reformulate the query:
even though P does not create a turn transition space (e.g. through rising
intonation) in which L can respond to the query, P’s own clarification is
completed by L.
The examples have illustrated several aspects of the use of questions to issue
queries. First, queries can be issued without questions: they can be issued as
requests made through declarative statements (3). Second, questions can have
different roles within queries: questions can occupy an entire query (4) or they
can be part of a query (8). Third, question-like features can also serve func¬
tions within queries: rising intonation allows P to check on whether the
information being provided can allow L to begin satisfying the query (6, 7);
it also allows L to check on a possible resolution of the query (line j in 5).
Fourth, L tries to satisfy the query even while it is being issued: L proposes
refinements of the query (with or without rising intonation: compare 5 and 7
to 8), asks for clarification (through a question (4)), and marks recognition of
the information being sought (5). Finally, the examples above show that
although L may ask for clarification from P (e.g. 4), and P may begin to offer
his own reformulation of a query (e.g. 5), P and L also work together to
reformulate (and thus resolve) a query (5 to 8).
We have seen in this section that two levels of questioning activity facilitate
the formulation and satisfaction of the query in reference interviews. First,
questions and question-like features figure in the formulation of the query
itself. P presents a query using either question forms or request forms that are
clear realizations of Searle’s conditions for felicitous questions: the speaker
lacks knowledge of a particular state of affairs (preparatory rule) and wants
to gain that knowledge (sincerity rule) by eliciting information from the hearer
(essential rule). Second are the clarification sequences. In order to provide P
with information, L may have to seek information about the query or redefine
P’s query in terms compatible with the organizational schema used by the
library to organize reference materials.6 P may also reformulate his own query
 The Ethnography of Communication 159
under several different circumstances: when P himself anticipates L’s need for
further specificity or recategorization; when L shows a need for more informa¬
tion (perhaps by withholding a response at a turn-transition space). Thus,
questioning facilitates the reformulation and resolution of the query on a turn
by turn basis, and on the level of the speech event itself.

3.2.3 Summary: questions in reference interviews

Patrons and librarians use questions for three different specific ENDS in
reference interviews: questions from L make offers, questions from P issue
queries, and questions from both L and P seek clarification (cf. check the
formulation) of the query. The chart below summarizes the question-answer
sequences in reference interviews. Note that thiS'chart ft not meant to illustrate
all possible arrangements of questions or their functions; nor is it an exhaust¬
ive description of all speech acts and functions in reference interviews. Ques¬
tion-answer sequences that are subordinate to others have indices indicating
their subordination (e.g. question-2a) and the fact that they are also next-ques¬
tions (e.g. question-3).

l: question-1 [offer]
p: (answer-1 [accepts] )
question-2 [begins query: seeks information]
{L/P}: question-3/2a [seeks clarification of q-2] or [checks information in q-2]
[p/l]: answer-3/2a [provides clarification of q-3/2a] or [confirms
information in q-3/2a]
l: answer-2 [resolves query: provides information]

This chart shows that questions can be used to offer help and thus prompt the
query, to issue the query, and to reformulate the query (as either L or P seeks
clarification of, or checks the information in, P’s query) as a way of facilitating
the resolution of the query. Requests for clarification and information checks
open a sequence subordinate to the main query. Furthermore, regardless of
who requests clarification or checks information from whom, the floor belongs
to L following the provision of clarification or confirmation of information:
L uses the reformulated query as a basis from which to resolve P’s query. Thus,
questions play a critical role in reference interviews: P enters a reference
interview with the explicit goal of seeking information and both P and L use
questions as a way to find that information.
We began our discussion of reference interviews by using Hymes’s SPEAK¬
ING grid to describe the communicative components comprising this speech
event. By way of conclusion, we can relate the questions in reference interviews
to these communicative components. Questions in reference interviews are
distributed among participants in ways that reveal both asymmetric and sym¬
metric PARTICIPANT roles. We noted a global asymmetry underlying the
160 Approaches to Discourse Analysis

entire interview: P solicits a service and L provides a service. We also noted,

however, that the clarification sequences that help to formulate and eventually
satisfy a query are symmetrically distributed. This allocation of acts to actors
suggests a more collaborative effort than might be suggested by the term
“interview” (see section 3.4). This symmetry could also reflect the fact that L
and P share ENDS (and both know that they do so). And the sharing of goals
could help to account for the ACT SEQUENCE: a series of questions and
answers, interchanged between participants, that work cumulatively toward
fulfilling the main task of the interview.

3.3 Questions in sociolinguistic interviews

This section focuses upon questions in another variety of interview: socio¬

linguistic research interviews. Sociolinguistic interviews have some features
typical of an interview. One important feature is their ENDS: since one person
(a sociolinguist, S) seeks to gain specific information from another (a respond¬
ent, R), the structure of activities and the expectations for behavior revolve,
at least partially, around a desire to resolve an asymmetrical distribution of
information. These ENDS are similar to those of other interviews. This sim¬
ilarity is important, for as Cameron et al. (1992: 13) point out: “interaction
between researcher and researched does not produce some anomalous form of
communication peculiar to the research situation and misleading as to the
nature of ‘reality’. Rather such interaction instantiates normal communication
in one of its forms.” Thus, to rephrase the point from above in the terms of
Cameron et al., we can say that “interviews” are one of the forms of ‘normal’
communication to which sociolinguistic interviews are similar (see also Wolf-
son 1976).
As noted earlier, one of the goals of the ethnography of communication is
to discover the communicative features or qualities that underlie our identifica¬
tion of particular events, encounters, or situations as certain “kinds” of oc¬
currences. Since questions are a key part of interviews, we can examine the
questions in sociolinguistic interviews as a way of understanding at least part
of the structure of this particular speech event - perhaps the part that is most
“interview” like. And at the same time, we will learn something about ques¬
tions by seeing how they are located in a particular configuration of actions
and beliefs that differs from other varieties of the same speech event. We will
see that what influences the form and function of questions in sociolinguistic
interviews are the following communicative components: PARTICIPANTS,
ENDS, ACT SEQUENCE, and GENRE. Just as we described the general
circumstances of reference interviews before focusing on questions within
those interviews, so too will we briefly consider these components before
discussing NORMS for questions in sociolinguistic interviews.
Let us start by considering (in fairly general terms) the PARTICIPANTS.
Goffman (1961: 323) divides occupations into two broad categories: “Spe-
 The Ethnography of Communication 161

cialized occupational tasks can be divided into two categories, one where the
practitioner ‘meets the public’ through his work, a second where he does not,
performing it only for the established members of his work organization.”
Goffman’s categories provide a base from which to discuss different types of
interviewers. The division noted above minimizes the importance of the activ¬
ity per se and maximizes the importance of the audience to whom the activity
is directed. Although all interviewers “meet the public,” the results of such
meetings are not always directed toward the needs and wants of that particular
public in either an immediate or long-term sense. In one sense, of course,
sociolinguistic researchers are often very involved in interpreting the meanings
and intentions of their respondents: such involvement is a critical part of trying
to obtain certain kinds of responses from informants. But the primary goal
(END) of a researcher’s interview is nevertheless the production of knowledge
for other researchers - for the established members a work organization (in
■Goffman’s terms).
The information that researchers seek to gain through a research interview
depends largely on the theories, hypotheses, issues, and problems current in
their field, as well as a varied collection of (sometimes implicit) methodolo¬
gical assumptions about the nature of data and its means of analysis. The
information sought in a sociolinguistic interview, for example, is information
that is interesting to sociolinguistic theory, practice, and application, e.g. in¬
formation about phonological, grammatical or discourse patterns, language
and social attitudes, and ways of life in a community. Researchers’ knowledge
can also serve the public (both the segment of the public interviewed and a
larger public): researchers can become advocates for public causes or empower
the public to become advocates for their own causes. But, as noted above, the
intended and actual products of a research interview are not geared toward
the public with whom they are meeting, but toward other researchers. This
contrasts sharply with the ENDS of reference interviews: librarians’ goals are
limited to the interview itself, i.e. to help a patron find information.
Another difference between sociolinguistic interviews and reference inter¬
views concerns the degree to which the interviewer’s goals are shared with the
interviewee. Not all of S’s goals are explicitly shared with R. This is because
of what Labov (1972e) calls the “observer’s paradox”: linguists want to
observe how people use language when those people are not being observed
(or are not aware of being observed). Although S often tells R of an interest
in language, this interest is often put in terms of broader social and cultural
issues (issues that are consistent with sociolinguistic interests in general).
Sociolinguists do not share many of their more specific goals with their
informants: they do not tell informants that they want them to speak in
different styles, that they want to hear the pronunciation of a particular sound
in different phonological environments, that they want them to tell narratives
of personal experience. This asymmetric knowledge of the ENDS of a socio¬
linguistic interview differentiates them from reference interviews and also has
an effect on the questions asked. As seen earlier, the fact that librarian and
162 Approaches to Discourse Analysis

patron share a goal (and both know that they share a goal) helps to account
for the ACT SEQUENCE: a series of collaborative questions and answers that
work cumulatively toward fulfilling the main task of the interview (resolving
a query). The asymmetric knowledge of ENDS also bears on the overall
structure of the sociolinguistic interview as a speech event, and on the ways
that questions are used during the interview.
A brief consideration of some of the ways in which sociolinguists achieve
their goals suggests still other differences between the two speech events being
considered. In keeping with the sociolinguists’ goal of obtaining a variety of
different speech styles, for example, both S and R are encouraged to introduce
topics and to shift topics away from an already loosely structured agenda
(Labov 1984). One effect of this is a sometimes explicit avoidance of the
question-answer format and fixed topic structure typical of many institutional
interviews. Rather than follow what can be called a “serial” format (asking a
set of questions whose order is predetermined by an interview agenda), for
example, S often uses what can be called a “stepwise” format (cf. chapter 7)
in which S’s next-question is based on a topic initiated by R’s answer to a
prior-question. Another effect of the intentional loosening of constraints on
the maintenance of topic is that sociolinguistic interviews allow a variety of
exchange types besides question-answer sequences, and permit a number of
different genres, e.g. stories, descriptions, and arguments. As I noted at the
outset of this book, sociolinguistic interviews provide most of the data used
for the sample analyses in this book, e.g. the exchanges in chapters 3, 4, and
7, the story in chapter 6, and the lists in chapter 8. This fluidity in ACT
SEQUENCE helps to produce an important difference in terms of how ques¬
tions relate to the overall structure of the interviews in which they are asked:
whereas reference interviews are centered on the joint resolution of a single
query (or series of queries), sociolinguistic interviews seek to integrate the
asking and answering of questions within other kinds of talk.
Above I noted that S and R are both encouraged to avoid the question-
answer format and fixed topic structure typical of many interviews. It is
important to note, however, that this “encouragement” comes from different
sources and in different guises. Sociolinguists are taught how to conduct
loosely structured interviews (e.g. in courses in field methods, through practice
with interview modules) and they use their knowledge to guide the way they
themselves speak and the way R speaks. Sociolinguists, for example, often talk
about how to “get” people to tell stories or how to “elicit” stories; similarly,
it is believed that casual speech can be increased by “allowing” people to
introduce topics that may seem “tangential” to the topic proposed by S (Labov
1984). During an interview, S transforms these goals into interactional prac¬
tices that often have the desired results. By intentionally altering (e.g. supple¬
menting or even avoiding) formats that are explicitly constrained only by S,
however, S is often forced into another kind of position of control: S monitors
the speech of another for reasons that are not at all known to interlocutors,
e.g. to elicit one form of speech rather than another, to allow (rather than try
 The Ethnography of Communication 163
to prevent) the saying of something. And this means, again, that the ENDS of
the interview - even the local, utterance-by-utterance goals of participants -
are quite different for S than for R.
Different orientations towards speech and towards participation in a speech
event may differentiate the social meaning of participants’ utterances more
radically than we might expect. One of the goals of my own interviews, for
example, was to record people’s pronunciation of the vowel [as] in the word
“planet.” To this end, I found that asking people if they thought there was
life on other planets did not always produce the desired token, i.e. people
could just say “yes” or “no.” I built upon my observations that my respond¬
ents often completed one another’s sentences, however, to ask the question in
the following way: “Do you think there’s life on other ...” My respondents
would then use the word “planet” to fill in my question, usually going on to
discuss their beliefs as well. This question,//then, ^Certainly had multiple
meanings and functions for me - not all of which were shared by my inter¬
locutors. Thus, formally equivalent productions of what seems to be the same
utterance may have very different interactional meanings depending upon the
speaker.
Finally, it is important to note that sociolinguistic interviews can be seen as
mixed GENRES. In my own work, for example, I sometimes refer to my
interviews as “interviews” per se. The direction of my analyses, however, has
always assumed that they reveal interactional patterns shared with conversa¬
tions as well (e.g. Schiffrin 1984a, b, 1987a). Because the term “genre”
concerns the formal characteristics of texts more than the situational under¬
standings supplementing text types (cf. chapter 8), however, I will refer to
sociolinguistic interviews as “mixed” or “hybrid” speech events, rather than
genres (but see Bauman and Briggs 1990; Briggs and Bauman 1992). Classi¬
fying sociolinguistic interviews in this way (part institutional talk, part con¬
versation) can suggest an underlying duality in sociolinguists’ sense of what is
going on during such an occasion, i.e. in their definition of the situation as
one based on power and/or solidarity.
As further support for this duality, consider the way sociolinguists talk about
what goes on during an interview. People whose language is being researched
are often referred to as informants: people who “give” information about
language to one with less information about that language. The one to whom
information is given is one who is assumed to have the power to gain such in¬
formation. But sociolinguists’ use of the term “consultant’ rather than
“informant” (Guy 1992) suggests a different status and role: here the re¬
searched party is either an expert from whom one seeks advice or an equal
with whom one jointly pursues information. Descriptions of how different
speech activities are located within the interview can also rely upon an as¬
sumption of an asymmetric power distribution, e.g. the goals noted above of
“getting” people to tell stories or “allowing” people to introduce their own
topics. But other talk about interviews suggests a sense of cooperation that
emerges from solidarity and affiliation — definitions of the situation more allied
164 Approaches to Discourse Analysis
with conversation. Sometimes such a sense of cooperation is intentionally
fostered by the researcher. In asking questions about another’s educational
level, sociolinguists are often told to phrase the question as How many years
of school did you manage to finish} Since the factive verb “manage” implicates
that the subject tried to finish school (but may have been prevented from doing
so), this question seeks to reduce a probable status discrepancy between the
educational levels of S and R; one might also argue that this question assumes
that both parties value education. The hybrid status of sociolinguistic inter¬
views is also reflected in the variety of relatively monologic genres (stories,
descriptions, arguments) that arise during interviews. The forms and meanings
of questions from different participants during different parts of the interview
also help to provide for the sense that sociolinguistic interviews are comprised
of more than one type of speech event.
I have suggested thus far that there are several properties of sociolinguistic
interviews that influence the types of questions asked and their functions in
the interview: PARTICIPANTS, ENDS, ACT SEQUENCE, GENRE.8 The ana¬
lysis to follow is organized around who asks what kinds of questions. We
noted earlier that the interview replaces one’s everyday roles with the institu¬
tionally structured roles of interviewer and interviewee (cf. Briggs 1986: 2).
These roles influence the mode of questioning available to participants (Goody
1978), such that not all modes are equally available to interviewer and inter¬
viewee. However, it is not only role relationships themselves that influence
questions: the degree to which role relationships are externally defined influ¬
ences the ability of questions to define a relationship (Goody 1978: 39). When
roles are clearly defined (e.g. customer and server in a service encounter,
patron and reference librarian in a reference interview), role expectations bias
the interpretation of any questions asked. But when roles are weakly defined
(e.g. two strangers sitting next to one another in an airplane), the mode of
questioning itself can help to establish a role relationship.
In what follows, I will have space for only a relatively brief discussion of
who asks what questions, and how this distribution of questions is related to
the speech event. Space also prohibits an analysis of all the questions in a
single sociolinguistic interview (comparable to our analysis for reference in¬
terviews). Not only are sociolinguistic interviews often fairly long (typically
more than an hour), but their form and content vary depending on a number
of factors: social characteristics of S and R, number of participants (e.g. more
than one R) and their relationship, first or later interview with informants,
early or later section of the interview. The discussion here is based on the
questions in ten sociolinguistic interviews carried out by the same researcher
(Anne Bower): the informants were working-class, Irish American or Italian
American men and women, ranging in age from early twenties to mid seven¬
ties. I also draw upon observations from seven interviews in which I was the
researcher and the informants (described in my earlier work, e.g. Schiffrin
1987a) were lower middle-class, middle-aged (mid thirties to early sixties),
Jewish American men and women.
 The Ethnography of Communication 165

I begin with a mode of questions used primarily by sociolinguists (3.3.1) and

go on to questions used by both sociolinguists and informants (3.3.2, 3.3.3).
I then consider the relationship between questions and participants in different
phases of the interview (3.3.4). After a brief summary of the use of questions
in sociolinguistic interviews (3.3.5), I summarize the entire sample analysis
(3.4).

3.3.1 Information-seeking questions

The mode of questioning conventionally associated with an IVer’s role is

information-seeking questions. These questions are straightforward realizations
of Searle’s felicity conditions for questions: the speaker lacks knowledge of a
particular state of affairs (preparatory rule) and wante to gain that knowledge
(sincerity rule) by eliciting information from the hearer (essential rule). The
overwhelming majority of information-seeking questions are asked by S. S
typically uses questions to seek either information about general topics (“What
kinds of games did you play as a kid?”) or specific information (e.g. “How
did you play jacks?”). The information elicited by an information-seeking
question may be related to prior discourse. Suppose S asks a general question
about games, R responds by listing a number of games, including “jacks,” and
then S asks how R played jacks. Both of S’s questions would be considered
information-seeking questions — even though the second question pursued a
topic opened by R’s answer to the first question.
I suggested earlier that sociolinguistic interviews define roles for participants
that influence the mode of questioning available to them. But the way that
participants sometimes adopt the mode of questioning conventionally associ¬
ated with the other’s position suggests that they try to redefine those roles,
and thus that the participant roles in sociolinguistic interviews are less rigid
than those in reference interviews.
When R does ask S information-seeking questions, the questions often have
unexpected effects (i.e. the “outcome” aspect of ENDS (2.2)). First, R’s infor¬
mation-seeking questions can lead to interchanges that seem less like inter¬
views than conversations. (As noted in appendix 1, we can think of con¬
versations as sociable (rather than task-driven) speech events with relatively
fluid boundaries between topics, flexibility in allocation of turns and partici¬
pant roles.) Second, R can adopt the role of S and ask questions to another
R. I illustrate each of these possibilities below.
Consider, first, how R’s questions help alter our interpretation of an activity.
They can be part of interchanges that seem less like interviews than conversa¬
tions.10 Let us compare (9) and (10). (9) is a segment from one of my interviews
in which I ask information-seeking questions.

(9) s: (a) Is the- are there ever block parties on this block?
ri: (b) No.
166 Approaches to Discourse Analysis

s: (c) No?
R2: (d) Ump um.
S: (e) Did you ever go to a block party any[where?
R3: (f) [No.
Rl: (g) Years ago, yeh.
S: (h) Did you ever go to one?
R2: (i) No.
S: (j) They’re supposed to be fun. I’ve never been to one,

(9) seems very “interview” like, perhaps to an extreme since the only responses
elicited by my questions are monosyllabic affirmations or denials. Note, also,
that I introduce and continue the topic, and that my questions select the
next-speaker, such that I am exercising control over who speaks when.
(10) is from a later interview with one of the same speakers (R3, identified
here just as R). In addition to noting R’s questions, I note certain features that
seem more typical of a conversation than an interview. Prior to (10), R had
been telling S about her twin grandsons. Included were two short stories about
children’s language, a topic that R had initiated. One story focused on what
the twin grandsons call each other; the other on what another grandchild calls
her. (10) begins as R ties the two topics together by anticipating what the
twins (when they can talk more) will call her.

(10) r: (a) So probably my other ones will call me that.

(b) I think Grandmom’s nice.
S: (c) I think it’s nice too.
R: (d) I like it.
S: (e) Yeh. I [call my-
R: (f) [Does your baby talk?
S: (g) Uh: he doesn’t t- [say my name yet.
R: (h) [Oh.
s: (i) Um . . . he: makes a lot of noises!
R: (j) Yeh.
s: (k) And he has a very deep voice [for his age=
R: (1) [Does he?!
s: (m) =hh yeh it’s really funny t’hear him!
(n) Because he’s so little, y’know, and his voice is deep,
(o) it sounds like it’s coming from an older person.
R: (P) Right.
s: (q) And uh:
R: (r) He’ll probably have a deep voice.
s: (s) Yeh I bet he will.
(t) Well my brother has a deep voice.
R: (u) Oh.
S: (V) And so does my father.
(w) He took- he knows mama, y’know and [cat.=
 The Ethnography of Communication 167
r: (x) [Yeh.=
S: (y) =[They have a cat.
r: (z) =[Oh that’s goo:d!
s: (aa) =And um: I guess there are [more words than=
R: (bb) [That’s very good!
s: =1 can recognize y’[know in all those noises.
R: (cc) [Yeh: right. Right.
(dd) The mother knows what- what d’you [call=
S: (ee) [Yeh.
R: =your grandmother?
s: (ff) Well one I called um Nanny, and the other I call Grandma.

(10) is a section that seems less like an interview and more like a conversation.
We can see this not only by observing R’s questions (f?l, dd; to be commented
on below) but in the way R and S manage the topic of talk and negotiate
speaking rights.
After R’s coda to her stories about her grandchildren (So probably my other
ones will call me that (a)), S and R occupy three turns (b to d) mutually
evaluating the name Grandmom. This brief sequence neither adds nor expli¬
citly seeks new information; rather its use of repetition creates interpersonal
involvement (Tannen 1989). S and R then both begin to follow up on the
general topic of language in the family, although in different ways. After still
another acknowledgement (Yeh (e)) of the evaluation, S begins to position
herself as “grandchild” to report what she calls her grandmother (I call my-
(e)). R seeks to position S as an adult (rather than child) by asking Does your
baby talk? (f). (“Your baby” here refers to S’s nephew, whom S had previously
mentioned as “my brother’s child.”) The first part of S’s response (Uh: he
doesn’t t- (g)) to R’s information-seeking question about S’s relative provides
the information sought by R. The second part (the repair to say my name yet)
also ties that information to the more local topic of what children call adults
- the topic just pursued by R. In lines (h) to (p), S elaborates on her response
while R provides continuers that neither claim a turn for R, nor motivate S to
offer R a turn (Schegloff 1981): R’s information checking question (Does he?!
(1)) works here as a continuer. This same distribution of speech continues in
(r) to (cc), even though the topics shift slightly. R’s next question (What d’you
call your grandmother? (dd)) continues the pattern in which S provides informa¬
tion sought by R, although it switches the topic back to what S had begun to
initiate with I call my- (e).U Thus, R’s questions open and maintain a phase
of the interview in which R and S add to a jointly pursued topic. It is because
of this joint allocation of participant rights that (10) seems less like an inter¬
view and more like a conversation
Recall Goody’s observation that a mode of questioning can help define a
role relationship when roles are not strongly defined by institutional norms.
Roles clearly help define normative expectations and obligations for partici¬
pants in a sociolinguistic interview: who expects what kind of behavior from
168 Approaches to Discourse Analysis

whom is clearly a key part of participants’ definition of the interview. But who
asks whom questions can define not only the capacity in which one person is
acting towards another, but also the situation of which those actions are a
constituent part. In (9), S’s questions conformed with normative expectations
about the role of an interviewer. But in (10), R’s questions in (f), (1), and (dd)
allowed S and R to jointly pursue and collaborate on a topic of talk. Thus,
R’s questions help to alter participants’ interpretation of the ongoing activity.
The second way that R uses information-seeking questions arises in multi¬
party interviews: one R adopts the role of S and asks questions to another R.
One of my own informants, for example, sometimes paraphrased my questions
for me (Schiffrin forthcoming, and 1992a; also chapter 4). In (11), I am asking
R1 (Irene) with whom she discusses her problems. Another informant, R2
(Zelda), is also present, but she is not explicitly addressed in this particular
interchange. (This is example (4) in chapter 4.)

S: (a) How ’bout you Irene?

Rl: (b) What?
S: (c) Who would you-=
R2: (d) Who would you [discuss it with?
S: (e) = [if you had a hard day who would you
complain to?
R2: (f) Jayhhhhh
Rl: (g) Uh:: I don’t know.
(b) Depending on what- what the problem really was, I might talk
to a friend easier.

S’s question (How ’bout you Irene (a)) is clearly addressed to Rl. It is R1 who
asks for clarification (What? (b)) and S who begins to provide clarification
(Who would you- (c)). But as S continues in her own questioning role (if you
had a hard day who would you complain to? (e)), R2 also takes S’s role as
questioner (Who would you discuss it with? (d)). As Rl continues in her own
role to provide an answer (g, h), R2 also takes Rl’s role as respondent
(Jayhhhhh (f)).
Although R can temporarily adopt S’s questioning role, too overt and pro¬
longed adoption of this role can be rejected as behavior inappropriate for an
interview. In one of my own interviews, for example, a wife commented on
her husband’s repeated questioning to me by reminding him that she’s inter¬
viewing you! In the example in (12), R asks S a question that S answers, but
S then reworks the participant structure so that she is the one asking the
question. Prior to (12), S and two respondents had been wondering what
happened to someone who had been present.

(12) s: (a) Maybe he’s a ghost.

R2: (b) Yeh, did you have any experiences in ghosts?
S: (c) Oh, yeh.
 The Ethnography of Communication 169
(d) That was one thing I did want to ask you.
R3: (e) As a ghost?
Ri: (f) [I bet you were never a ghost.
s: (g) [As a ghost hh.
(h) Like, for instance, in some families, there’s somebody that
has dreams that come true.
R2: (i) Yeh.
s: (j) Or, y’know, a family that, y’know, there’s somebody that
dreams something and it means that something’s gonna
happen.
(k) Like the Irish are big for that.
(1) I wonder, like, did you ever dream that you felt (something)
[for anybody in your family?
Ri: (m) [Uh: well, uh, before my mother dieSii. I. .. uh /continues/

After S jokingly comments that Maybe he’s a ghost (a), R2 proposed this as
a new topic of talk by asking S if she had any experiences in ghosts (b). S’s
Oh, yeh does not answer R’s question in the sense that it affirms an experience
with ghosts. Instead, Oh, yeh acknowledges the topic of the question, such
that S explicitly formulates the topic in terms of a question that she wanted
to ask (That was one thing I did want to ask you (d)). Although this topic is
temporarily treated in a joking key (e to g), S then resumes with a straightfor¬
ward request for information about ghosts (h to 1) and dreams to which R1
responds. S later comments to R2 (the one who asked the question in (b)) that
she would be a good coworker.
We have observed thus far that information-seeking questions are used
primarily by S: when they are used by R, they help to create a change in
activity (or genre) or they are adaptations of S’s questions that are directed to
another R. These restrictions on the use of information-seeking questions by
R, however, do not mean that R does not ask questions. As we see in the next
sections, there are two types of questions that both S and R ask.

3.3.2 Information-checking questions

One type of question asked by both S and R is the information-checking

question. Information-checking questions may have a variety of forms: inter¬
rogative sentences, tag questions in which the tag is a subject-aux inversion
with reversed polarity (e.g. “You can do it, can t you? ), particle like tags (e.g.
“see?,” “y’know?”) at the end of a declarative statement, particles with rising
intonation (e.g. “really?,” “oh?,” “right?”), rising intonation on a declarative
statement or part of a statement (see chapter 3, section 3.1.1). Largely because
of the meaning associated with final rising intonation (again, see chapter 3,
section 3.1.1), information-checking questions fulfill the sincerity, prepara¬
tory, and essential conditions of information-seeking questions. The main
170 Approaches to Discourse Analysis

quality differentiating information-checking from information-seeking ques¬

tions is the scope of what is being questioned and the type of response sought:
the information being sought is not the completion of a proposition, but
reception of a referent or proposition. These features also create different
turn-taking consequences: whereas information-seeking questions always se¬
lect a next-speaker, information-checking questions either serve in a back
channel capacity (Duncan 1972) or allow a current speaker to continue pend¬
ing receipt of acknowledgment or recognition.
Although both S and R use information-checking questions, the way they do
so is quite different. The chart below suggests how S and R use these questions.
(As previously, in section 3.2.3, subordinate question-answer sequences have
indices indicating that they are next-questions and that they are subordinate
to prior questions. The table does not illustrate all possible arrangements of
the act being considered.)

Information-checking questions from S

s: question-1 [seeks information]
R: answer-1 [provides information]
S: question-2/1 a [information-check]
R: answer-2/1 a [affirms information]
answer-1

Information-checking questions from R

S: question-1 [seeks information]

r: answer-1 [provides information
question-2/la [information-check]
s: answer-2/1 a [affirms information]
r: answer-1 [provides information]

Although information-checks and their responses can always be seen as sub¬

ordinate to a main sequence, the participation status occupied by S and R
differs when they use these questions. S typically uses information-checks to
address some aspect of what R has said; R uses them to address some aspect
of his own talk. In other words, S uses these questions to check her own
reception of another’s talk, and R to address another’s reception of his own
talk. Note, however, that what information-checks from both S and R are
directed to is S’s reception of talk; also, regardless of who asks an informa¬
tion-checking question, what follows is R’s answer.
(13) illustrates how one particular form of S’s information-checking ques¬
tions (an affirmative particle with rising intonation) can be situated in a
sociolinguistic interview as a turn-continuer (Schegloff 1981).

(13) s: (a) Well, like did you have a team when you were goin’to school
on a little league, or-
R: (b) Oh yeh. [We used to-]
 The Ethnography of Communication 171
s: (c) [Really?]
R: (d) = yeh, I play- I was on an organized I was five years old when
I was on my first organized baseball team.

During R’s answer (Oh yeh (b)) to S’s question (did you have a team (a)), S
asks Really? (c). R’s response treats Really? as a yes-no question about his
own prior response (Oh yeh (b)) that had affirmed the existence of a team.
But in addition to answering yeh, R begins a long list of the different teams
that he was on (example 25 in chapter 7). Thus, S’s question (Really? (c))
prompts R to expand his own answer to S’s prior information-seeking question
(did you have a team (a)).
The next examples illustrate how R’s information-checking questions check
on R’s own talk. We also see that they may be issued through a greater variety
of forms than S’s information-checking questions. In«(14), R’s question is a
full interrogative that checks S’s familiarity with a specific referent.

(14) s: (a) Who was your favorite star? [Or-

R: (b) [Oh, uh, Wallace
Reed, I guess.
(c) Do you remember readin’ about him?
S: (d) Uh huh. I- I like old films, I’ve watched a lot.

After providing an answer (Oh, uh, Wallace Reed, I guess, (b)) to S’s question
(Who was your favorite star? (a)), R checks on S’s reception of the answer,
i.e. S’s familiarity with the referent, in (c), Do you remember readin’ about
him?
In (15), R is describing a particular ball game (Error) and the type of ball
used for the game.

(15) r: (a) And what we used to do we used to, we used to play it with a
golf ball. . . Right?
(b) You ever know what a golf ball, how fast it come back to you
off of the brick wall?
s: (c) I can ima[gine.
R: (d) [Anyway, it could shoot back.
(e) We used to throw ’em all at the brick wall, and you hadda
catch it.

R first uses the tag Right? (a) to check on S’s reception of a golf ball. After
receiving no verbal response, R switches to a you know question (Schiffrin
1987a) to check on S’s knowledge of the properties of that referent (You ever
know what a golf ball, how fast it come back to you off of the brick wall?
(b)) before using it in his description. S’s acknowledgment of the properties
(I can imagine (c)) overlaps with R’s restatement of the properties (Anyway,
it could shoot back (d)). R then returns to his description of the game (e), this
172 Approaches to Discourse Analysis

time referring to “a golf ball” pronominally (see discussion of this referring

pattern in chapters 6 and 8).
Information-checking questions allow R to check the reception not only of
referents, but also of propositions that contribute to the point being developed.
In (16), R is contrasting the attitudes of children today with those of her own
youth.

(16) R: (a) They don’t respect their parents as much as we did.

(b) Now, if my mother got sick, we all got sick with her . . .
Y’know?
s: (c) Right. Yeh.
r: (d) Today, the children can’t say the same thing.

After arguing in general terms about children’s current attitudes (They don’t
respect their parents as much as we did (a)), R gives a specific example (Now,
if my mother got sick, we all got sick with her . . . (b)) illustrating how she
and her siblings shared her mother’s problems. R checks this example with
Y’know? After S’s acknowledgment (Right. Yeh (c)), R completes her com¬
parison (d).
So far, we have seen how various forms allow R to check S’s receipt of
different amounts of information, i.e. of referents (14), properties of referents
(15), and propositions (16). When information is being packaged and presented
for a particular interactive purpose during an interview, the same kinds of
forms can check on how that information suits the purpose defined by that
particular interactional slot. In (17), two pieces of information (in e and i) fill
a clarification slot.

(17) r: (a) I never drank so much- so many sarsaparillas in all my life, hh

S: (b) What’s a sarsaparilla? I’ve heard about ’em, [but uh-
R: (c) [What’s that?]
S: (d) Wh- what is a sarsaparilla?
r: (e) Oh:, it’s something on the order of uh cream soda.
S: (f) Oh, oh.
r: (g) Y’know?
S: (h) Yeh.
R: (i) And it’s- uh pretty near the same: uh taste, I think.

In (17), S initiates a clarification sequence (What’s a sarsaparilla? (b)) that

occupies several turns. After R’s what’s that? (c) requests a repeat that S
provides (Wh- what is a sarsaparilla? (d)), R provides clarification (something
on the order of uh cream soda (e)). Although S marks her receipt of the new
information (Oh, oh (f); Schiffrin 1987a), R checks that reception with
Y’know? (g). Following S’s affirmation (Yeh (h)), R then continues his clari¬
fication by describing a slightly different feature of the drink (i) in a new
intonation and information unit. Thus, the information-checking Y’know? in
 The Ethnography of Communication 173
(17) marks a transition between two pieces of information that occupy a single
interactional slot, i.e. the provision of clarification.
The information-checking question in (18) also has a transitional role, but
to a new interactional slot within a sequence. (18) is from a section of a
sociolinguistic interview in which informants are asked about their familiarity
with regional lexical variants. Sections such as these are typically comprised
of either two-part question-answer exchanges or three-part question-answer-
evaluation sequences. In (18), S asks whether R ever heard of a baby cab (a).

(18) s: (a) Did you ever hear of a baby cab?

r: (b) I’ve heard of it.
(c) I don’t know,
(d) I- I- I think uh England uses that?
ft
(e) Some [country anyway.
s: (f) [Yeh.

After offering one answer {I’ve HEARD of it (b)) and conveying uncertainty
(I don’t know (c)), R offers a possible location where baby cab is used: I- I- I
think uh England uses that? (d). S’s acknowledgment of the information as a
proposed answer overlaps with R’s continuation of her answer (Some country
anyway (e)). Note that the answer is accepted despite continued uncertainty
about its veracity: not only does R become more vague (some country) but
the use of anyway implicates that England may not be right. S then goes on
to a new question.
We have seen in this section that S and R use information-checking questions
to check some aspect of ongoing talk: S typically checks her reception of R’s
talk, and R checks S’s reception of his own talk. R regains the floor after S:
R either acknowledges that S’s reception is sufficient and then continues (as
in 13), or S acknowledges adequate reception of what R has said and then R
continues (as in 17). At first glance, this allocation of speaking rights seems
to contrast with that achieved by information-seeking questions: whereas S’s
information-seeking questions select a next-speaker, R’s information-checking
questions allow a current speaker to continue. Note, however, that both
question types create a slot in which it is R who continues talking: just as R
is the next speaker after S’s information-seeking questions, both S’s and R’s
information-checking questions return the floor to R. As we see in the next
section, the turn-taking consequence of clarification-questions is similar to
that of information-seeking and information-checking questions.

3.3.3 Clarification-questions

Although clarification-questions are used by both S and R, their locations and

functions differ according to participant status. The chart below shows the
participant distribution of clarification-questions. Three features of this chart
174 Approaches to Discourse Analysis

enable comparison with the chart on page 159 (showing clarification se¬
quences in reference interviews): the clarification-questions are listed in terms
of question-answer sequences; subordinate question-answer sequences have
indices indicating that they are next-questions that are subordinate to prior
questions. Again, the chart does not illustrate all possible arrangements of the
act being considered.

Clarification-questions from R
s: question-1 [seeks information]
R: question-2/la [seeks clarification of question-1]
s: answer-2/la [provides clarification of question-1
R: answer-1 [provides information sought by question-1]

Clarification-questions from S
s: question-1 [seeks information]
R: answer-1 [provides information sought by question-1]
s: question-2/la [seeks clarification of answer-1]
R: answer-2/1 b [provides clarification of answer-1]
answer-1 [provides information sought by question-1]

As shown, R typically seeks clarification of a question from S; S typically seeks

clarification of an answer from R. When R seeks clarification of S’s ques¬
tion, the floor is nevertheless returned to R after S’s clarification, so that R
can provide an answer to S’s question. When S seeks clarification of R’s
answer, then R’s clarification continues the answer initiated earlier. Thus, the
completion of either sequence returns the floor to R. This further suggests that
although the clarification requests from R clearly serve his own needs - since
they facilitate the provision of the answer - they also provide a space for
exactly what S wants, i.e. continued speech from R.
The next examples illustrate the distribution just noted - how clarification-
questions are situated in sociolinguistic interviews and how their functions
depend upon who does the asking. In (18), R requests clarification of S’s
question; the distribution is reversed in (19).

(18) S: (a) And your wife’s passed on, right?

R: (b) Pardon me?
S: (c) Your wife has passed on?
R: (d) Is my wife what?
s: (e) Is your wife passed on, she’s passed away, right?
R: (f) Oh, my wife’s dead?
s: (g) Yeh.
R: (h) Yeh, that’s why I- I have a room here /continues/

(18) contains multiple clarification-questions from R (b, d, f). Each request

makes more explicit guesses (Ochs 1985) about what is being asked by incor¬
porating more information from S’s clarifications (c, e) of her initial question
 The Ethnography of Communication 175
(a). The entire clarification sequence is embedded in the question-answer
exchange (a, h) and facilitates the provision of an answer.
Although it is S who requests clarification in (19), the clarification sequence
is similarly embedded in the question-answer exchange, and again it is geared
toward getting a clear formulation of an answer from R.

R: (a) There was- there was one person up the street,

(b) he was a number writer,
(c) and he owned a car.
s: (d) A number writer?
R: (e) A number writer.
s: (f) What’s that?
R: (g) Well, a number writer, um, man that uh- /continues/
# ft
Following R’s description of one person up the street (a to c), S requests
clarification of one part of that description (A number writer? (d)) with an
echo question (Christian 1973). When the echo question format of S’s request
is treated only as a “non-hearing” (i.e. it receives only repetition of a number
writer (e) from R), S asks more directly for clarification (What’s that? (f)) and
R provides clarification (beginning in g). Thus, although it was S who re¬
quested clarification in (19), that clarification results in an answer from R (g)
that also supplements the informational content of R’s prior answer (a to c).
(20) is a slightly more complex example with several clarification sequences.

(20) s: (a) What did kids do on dates?

(b) When you were [growin’ up.
r: (c) [When I was younger
s: (d) Yeh.
r: (e) [Clears throat] Excuse me.
(f) It would either be like maybe going b<
And then take in- like you stop, u like for a soda or
(g)
something.
s: (h) Uh huh.
R: (i) Or a movie.
(j)
Or the biggest thing was the dance.
S: (k) Like where’d you go for dances?
r: (1) Schools.
(m) Like [your schools used to-
s: (n) [High schools?
r: (o) Eh, eh, no, grade schools.
/r continues/

The first clarification sequence in (20) is subordinate to S s question What did

kids do on dates? (a). As S begins to modify this question with clarifying
information (When you were growin’ up (b)), R also requests clarification of
176 Approaches to Discourse Analysis

S’s question (When I was younger? (c)). After S provides clarification (Yeh
(d)), R answers the question from (a). In answering the question, R mentions
the dance (j). S then asks a question whose focus is much more local than that
of her earlier question about what kids did on dates (a): Like where’d you go
for dances? (k) asks specifically about the prior referent the dance (j) and its
answer clarifies that referent (schools (1)). The next question from S is also
local and could be said to seek further clarification of the referent: S asks High
schools? (n) and R answers Eh, eh, no grade schools (o).
To summarize: we saw earlier that information-seeking questions from S,
and information-checking questions from both S and R, allow R to continue
talking. Clarification sequences seem to have a similar turn-taking con¬
sequence: regardless of who requests clarification from whom, the floor be¬
longs to R after completion of a clarification sequence.

3.3.4 Questions, participants, ends, and act sequence

We have seen thus far that some questions during sociolinguistic inter¬
views are asked primarily by S (information-seeking questions) and some are
asked by both R and S (information-checks, clarification-questions). All of
these questions can create participation structures that replicate the more
global structure and participatory norms of an interview: one person seeks
information, one person provides information. As noted earlier, however,
sociolinguists and their informants can also temporarily trade or alter the roles
assigned by the interview structure: one way that informants do so during a
multiparty interview is by adopting the questioning strategies of the inter¬
viewer.
Interviewing more than one person at a time has a general methodological
advantage for sociolinguists: it increases the likelihood that informants will
shift styles, thus providing important data on phonological variation (Labov
1972c). Multiparty interviews also mean that a greater number of interactive
roles - and genres - become available for each participant. Although members
of a family or friends who are being interviewed together may talk just to S
(i.e. they may do no more than separately answer S’s questions, as in 9), they
often talk to one another about other matters, or they may construct joint
responses to S’s questions that lead to other speech activities. Thus, multiparty
interviews may reveal a greater variety of genres (conversation, argument,
discussion) than individual interviews.
(21) illustrates the use of questions in a multiparty interview: it highlights
not only the way different PARTICIPANTS use questions to create an ACT
SEQUENCE relative to different ENDS, but also the way a mode of question¬
ing can help differentiate phases of the speech event. (21) also contains the
question discussed in chapters 3 and 4, Y’want a piece of candy? To maintain
consistency with our previous use of this example, I use personal names
instead of S (Debby) and R (Henry, Irene, Zelda).
 The Ethnography of Communication 177

Before I present (21), it is important to note that ongoing with the physical
activities of setting up our interview was a discussion about the unexpected
death of a teacher from the neighborhood school; in fact, this was the main
topic of our talk prior to the beginning of the interview itself. (Because the
conversation about the teacher had been in progress before I turned the tape
recorder on, I do not have a record of it and cannot include it in (21).) I was
planning to start this interview with a word list: one of the most formal
sections of an interview during which informants read a list of words (targeted
for study because of the phonological environment they provide for a particu¬
lar sound).

(21) henry: (a) Y’want a piece of candy?

IRENE: (b)
N°z , «
ZELDA: (c) She’s on a diet.
DEBBY: (d) Who’s not on [a diet.
IRENE: (e) [I’m on-
I’m on a diet
(f) and my mother [buys-=
ZELDA: (g) [You’re not!
IRENE: (H) =my [mother buys these mints.=
DEBBY: (i) [Oh yes I amhhhh! 7
ZELDA: (i) — Oh yeh.
IRENE: (k) The Russell Stouffer mints.
(1) I said, “I don’t want any Mom.”
(m) “Well, I don’t wanna eat the whole thing.”
(n) She gives me a little tiny piece.
(o) I eat it.
(P) Then she gives me an[other,=
henry: (q) [Was =
IRENE: (r) =so I threw it out the window=
henry: =there a lot of people?=
IRENE: (s) =1 didn’t [tell her.=
HENRY: (t) [Was there=
IRENE: (u) =[She’d kill me.
henry: =[a lot of people at the house?
ZELDA: (v) All: the teach[ers.
IRENE: (w) [A lot of teachers will- probably will all be
there till late.
HENRY: (x) Je:sus Christ.
ZELDA: (y) All: the teachers.
HENRY: (z) What a heartache.
ZELDA: (aa) And that Miss DiPablo? She was d- she must be her very
best friend. She was there. She said she was with her all:
week. She must be stayin’ at her house!
IRENE: (bb) Probably.
178 Approaches to Discourse Analysis

ZELDA: (cc) [With-

DEBBY: (dd) [Did it happen today?
IRENE: (ee) [Saturday.
ZELDA: (ff) [No:
DEBBY: (gg) Oh, oh.
HENRY: (Hh) [Damn shame.
IRENE: (ii) [All right, let’s go.
HENRY: (ii) All right.
DEBBY: (kk) Okay, who wants to go first?
ZELDA: (ID Oh: I’ll go first.
DEBBY: (mm) Okay.
ZELDA: (nn) I better get my glasses.
DEBBY: (oo) Okay.
IRENE: (PP) Well I’ll go first, then.
DEBBY: (qq) Okay.
IRENE: (rr) What am I su- just read¬
DEBBY: (ss) just read them. Umhmmm.

The interchange in (21) can be clearly divided into two phases: a pre-inter-
view phase (a to ii); a lead-in phase (ii to ss) that makes a transition to the
word list reading in the interview. These two phases are characterized by
radically different topic structures and participant orientations. (21) also con¬
tains eight questions, repeated below:

(a) HENRY: Y’want a piece of candy?

(d) DEBBY: Who’s not on a diet.
(q) henry: Was there a lot of people?
(t) HENRY: Was there a lot of people at the house?
(aa) ZELDA: And that Miss DiPablo?
(dd) DEBBY: Did it happen today?
(kk) DEBBY: Okay, who wants to go first?
(rr) IRENE: What am I su- just read-

The functions of these questions are related to the phases of the speech event
in which they occur.
Let us take the pre-interview phase first. It is here that Henry offers Irene
candy (Y’want a piece of candy? (a)), Irene refuses (No (b)), Zelda provides an
account for the refusal (She’s on a diet (c)), and Irene tells a story further
accounting for her refusal (c to u). My rhetorical question (Who’s not on a
diet (d)) establishes solidarity through positive politeness (chapter 4, p. 117).
Henry’s questions (Was there a lot of people? (q), Was there a lot of people
at the house? (t)) overlap with Irene’s story to initiate a return to the main
topic (the teacher’s death). Irene and Zelda jointly answer Henry’s question
(v, w, y). Henry evaluates the situation (x, z). Zelda’s And that Miss DiPablo?
(aa) is an information-checking question concerning a new referent.
 The Ethnography of Communication 179
The pre-interview talk has a relatively fluid topic organization: three topics
(candy, diets, the teacher) cross cut one another with none overtly reinforced
by participants. In addition, turn selection (who speaks when) is, for the most
part, a matter of self-selection (chapter 7), such that the participant roles of
speaker and hearer are relatively available to any party. The five questions in
the pre-interview phase have a range of different functions. Henry’s first ques¬
tion (Y’want a piece of candy? (a)) is simultaneously an offer and request that
opens a three-part exchange (see chapter 3). My question (Who’s not on a diet
(d)) is part of a gender-specific byplay about “diet.” Henry’s next two questions
(Was there a lot of people? (q), Was there a lot of people at the house? (t))
return the discussion to the topic prior to “candy” and “diet.” Zelda’s question
(And that Miss DiPablo? (aa)) is an information check that introduces a new
referent into talk about “the teacher” (see chapters 6, 7, and 8). Thus, questions
have a range of functions in this phase of the interview: this lack of functional
rigidity matches the fluid topic and turn organization noted earlier.
The structure of the lead-in phase is in dramatic contrast to the pre-interview
phase just discussed. Critical to the shift in topic and participant structure is
Irene’s All right, let’s go (in ii). Note, however, that prior to this I have asked
a question (Did it happen today? (dd)) that overlaps Zelda’s elaboration of a
referent (Miss DiPablo) pertinent to the ongoing discussion (aa, cc). My
question reflects an asymmetry in information: I am seeking information that
I do not have. Thus, it reveals my status as an outsider to the topic, and
displays a participation framework more typical of an interview. Irene’s
answer (Saturday (ee)) fits the interview structure simply because it does no
more than provide the information that I had requested. (Zelda’s response No:
is more evaluative.) It is after Irene’s answer to my question that Irene says
All right let’s go, noted above as critical to the lead into the interview itself.
Irene’s topic pre-closing (all right in (ii); Schegloff and Sacks 1973) overlaps
Henry’s evaluation (Damn shame (hh)). Irene follows her pre-closing with a
directive (let’s go) that makes explicit not only that there is an agenda (we are
gathered together for a purpose) but that it is time to start attending to that
agenda. Thus, Irene’s all right, let’s go does lead into the interview. However,
this relatively global transition in interactional structure has been eased by a
prior question-answer exchange that conforms to the norms and participant
structure of an interview: S seeks information from R.
In the remainder of the lead-in phase (ii to ss), topic and participation
framework are explicitly directed toward starting our agenda. My question
(Okay, who wants to go first? (kk)) opens a series of short turns through which
participants negotiate the specifics of who is to do what when (i.e. begin
reading the words in the word list). The topic of each turn is narrowly
constrained toward the goal of getting us started with the task: Henry agrees
with the proposed transition (All right (jj)), I agree and offer others the
opportunity to self-select the order in which they will do the task (Okay, who
wants to go first? (kk)). I continue to allow the others latitude in the way they
plan to fulfill the task: my repeated okay’s (mm, oo, qq) transfer responsibility
180 Approaches to Discourse Analysis

to the others to take the next action (Merritt 1976), whether that action is
Zelda offering to go first (Oh: I’ll go first (11)), Zelda saying she needs her
glasses (I better get my glasses (nn)), or Irene offering to go in Zelda s place
(Well I’ll go first, then (pp)). Once it is decided that Irene goes first, respon¬
sibility for instructing the others in performance of the task is clearly trans¬
ferred to me: Irene’s question (What am I su- just read- (rr)) about what she
is supposed to do places me in a directive role. Thus, what gives this segment
its status as a lead-in phase is that both topic and participation are oriented
toward making a transition to the main agenda: the topic is task-driven and
participation rights are allocated and relatively circumscribed by the nature of
the task.
We have seen in this section that questions have different functions in
different phases of the interview. Just as the mode of questioning may alter
our sense of role relationships and our identity of the speech event as inter¬
view” or “conversation,” so too can it help to create different phases of an
interview and organize the flow from one interview activity to another.

3.3.5 Summary: questions in sociolinguistic interviews

We have focused here upon three types of questions used during sociolinguistic
interviews: questions primarily from S seek information, questions from both
S and R check the information being provided and request clarification.
Although the distribution of these modes of questioning generally supports the
norms of the interview structure, participants also exploit these convention¬
alized expectations as ways to alter their role relationships and their definition
of the situation.
Like the reference interviews discussed in section 3.2, the norms underlying
the use of questions during sociolinguistic interviews are most influenced by
the ENDS and PARTICIPANTS of the speech event. Since the specific ends
and participants of sociolinguistic interviews differ from those of reference
interviews, however, the distribution and social meaning of questions also
differ between the two speech events. Despite differences in the specific re¬
search goals of sociolinguists, one relatively consistent goal is to record a great
deal of speech from an informant. S’s information-seeking questions clearly
support this goal. Information-checking questions and clarification-questions
also support this goal: they all allow R to return to, or maintain, a turn at
talk. The goals and participant identities specific to reference interviews pro¬
duced not only different forms and functions of questions (offers and collab-
oratively reformulated queries), but also an important difference in the rela¬
tionship between questions and the overall ACT SEQUENCE of the interview.
Reference interviews are centered on the joint resolution of a single query (or
series of queries). Sociolinguistic interviews integrate the asking and answering
of questions within other kinds of talk, including talk that can seem less like
interview than conversation.
 The Ethnography of Communication 181

3.4 Summary of sample analysis

We noted earlier that the ethnography of communication is the most integrat¬

ive approach of those considered thus far: it can combine other approaches
within a larger framework of inquiry into cultural knowledge and the social
and linguistic practices for which it provides. Our analysis of the ways in
which a speech act (question) can be used during a speech event (interview)
thus illustrates cultural knowledge of language structure and function, the
latter broadly enough construed to include how language functions as a
resource through which to organize social interactions and enact social roles.
But as also noted earlier, it is misleading to view an ethnographic approach
to discourse as just the addition of a new component (“culture”) to the basic
material of other approaches. Rather, an ethnographic approach seeks to
define the basic constructs of other approaches (along with its own constructs)
as aspects of communication whose identities and interrelationships all reflect
particular cultural meanings. In keeping with the idea that an ethnographic
approach seeks culturally relative definitions of the basic “stuff” of communica¬
tion, let us summarize our sample analysis by considering how this approach
can add to our knowledge of questions as speech acts.
Most scholars who study questions find them difficult to define, and they
often note a number of criteria (syntactic, intonational, lexical, semantic,
functional) by which to do so. Applying the criteria to actual utterances is
complicated not just because of the multiple criteria themselves, but also
because the criteria themselves seem to fluctuate in importance: what seems
important to recognition of a question in some instances is not important in
others.
Interactional and contextual criteria for questions are not operationally
failsafe either. Goody (1978: 26), for example, suggests that “the direct illo¬
cutionary force of all questions is the elicitation of a response - but not
necessarily a verbal response” (cf. Lund 1984). Broadening the notion of
response to include a cognitive change in information state without verbal
recognition of that change would allow many information-checking devices to
pass as questions (as in the examples in section 3.3.3). But the identification
of responses that are not manifested verbally (or nonverbally, e.g. through a
head nod) is difficult for discourse analyses that rely for data upon naturally
(rather than experimentally) occurring communicative events.
Context further complicates the identification of an utterance as a question.
Although participant identity and setting clearly influence the interpretation
of utterances as questions, for example, the degree to which these contextual
parameters have an influence can vary. Even something so basic as the responsi¬
bility for deciding upon meaning can be unevenly distributed among partici¬
pants. McHoul (1987) suggests, for example, that the way a hearer interprets
an utterance is more critical to its function (e.g. the interpretation of an
interrogative as question or warning) than the speaker s intention. But even
182 Approaches to Discourse Analysis

this distribution is culturally relative, reflecting deeply held cultural assump¬

tions about responsibility for knowledge and the ability of one person to
interpret another’s thoughts and feelings (DuBois 1987; Duranti 1988; Ochs
1985).
We have identified in our analysis a greater variety of utterances as questions
than might be included in either a straightforward speech act analysis or by
reliance upon formal clues alone. We incorporated many aspects of Searle s
analysis of questions in doing so: we spoke repeatedly of how utterances
fulfilled preparatory, sincerity and/or essential conditions of questions. But we
also added different aspects of context to our analysis of the utterances that we
called “questions.” Perhaps more importantly, we did so in a way that built
upon an underlying assumption that communicative events themselves are
organized and that their properties have systematic effects on constituent acts.
Let me be more specific about the methodology. We first described the
communicative properties of events. What is it like for people to participate
in a reference interview? What understandings underlie participation in a
sociolinguistic interview? We then used some relatively traditional means of
defining questions to choose some utterances to focus upon. We tried to
discover how different communicative properties of speech events would
define the function of those utterances and how they did so. By considering
the utterances within the overall communicative organization of the speech
event, we found that all the utterances that we called “questions” shared a
general function: they all sought a response that facilitated the transmission
of information in line with the overall goal of the speech event. Note that this
function was not something we presumed initially: we did not begin with a
definition of questions that assumed their goals to be so neatly attuned to the
overall goal of the speech event.
The communicative properties best differentiating types of questions both
within and across speech events were PARTICIPANT, ENDS, and (to a lesser
degree) ACT SEQUENCE. Thus, who says what, why, and when, explains a
great deal about questions in interviews. This analysis of context allowed us
not only to identify a group of utterances that share a function, but to reveal
the more subtle differences among those utterances. We might pursue this still
further to propose a contextually motivated taxonomy of speech act functions
- one that seeks to show the similarities and differences among questions
themselves, as well as between questions and other utterances. Although the
questions discussed all seek a response facilitating the transmission of informa¬
tion, there are some important differences among them: how much partici¬
pants can presume about each other’s knowledge, the kind of response being
sought, the way a response facilitates information transmission. Differences
such as these can serve as heuristics by which to define different types of
questions.
Some approaches to discourse would seek to support the kinds of trends
described here with quantitative analyses that compare the number of question
types asked by different participants in different locations (both within and
 The Ethnography of Communication 183
across speech events). Our discussion has proceeded as if such analysis is a
possible next step in an ethnographic approach: I presented examples of each
question type as if they were always mutually exclusive categories and as if
each utterance filled just one question function. An ethnographic approach to
discourse, however, does not assume either the mutual exclusivity of categories
or the monofunctionality of an utterance. These, too, are subject to variation
depending upon context. The following example suggests that although informa¬
tion-seeking questions, information-checking questions, and clarification-
questions can be separately realized by different utterances, their functions can
also overlap in a single utterance or be difficult to discriminate. In (22), S asks
three questions.

(22) s: (a) So are most of the guys that hang in your crowd uh what,
Italian, Polish, Irish? ^ ft
R: (b) Indians . . .
s: (c) Indians?
(d) You had everybody, huh?
r: (e) Yeh.

S’s question in (a) is the easiest to categorize. So are most of the guys that
hang in your crowd uh what, Italian, Polish, Irish? seeks new information: R
had been describing his crowd, but said nothing about their nationalities. The
questions in (c) and (d) are more difficult to categorize. S’s question (Indians?
(c) ) can be interpreted either as a request for clarification or an information
check (since Indians (b) is a surprising addition to this list of ethnic identities).
These two question types can be especially difficult to tease apart because the
categories “information-check” and “clarification-question” are not neces¬
sarily mutually exclusive: one reason why people might check their reception
of another’s utterance (information-check) is that it is unclear to them (clarifica¬
tion-question). This is not to say, however, that these two categories cannot
be differently defined in principle: information-checks can be addressed either
to one’s own talk (“Do you understand what I mean?”) or to another’s talk
(“Do I understand what you mean?”), whereas clarification-questions can be
addressed only to another’s talk simply because they presume a lack of under¬
standing in what another has said.
The categorization of S’s question You had everybody, huh? (d) in (22) is
also difficult, but for slightly different reasons. Like S’s Indians? (c), the
question in (d) is semantically dependent on prior talk. But in addition to being
identifiable as an information check, You had everybody, huh? can be identi¬
fied as an information-seeking question that builds a stepwise topic transition
(chapter 7) in which S’s question (d) builds on a topic initiated by R’s answer
(b) to a prior-question (a). Thus, here the data themselves do not allow a clear
formulation of criteria by which to differentiate question types.
The discrimination of question types gets even more complicated if we try
to identify the same types of questions across speech events. In several sections
184 Approaches to Discourse Analysis

(3.2, 3.3.2, 3.3.3), we discussed examples in which questions were issued

through rising

(6) p: (c) I’m looking for . .. Fanny Mae, and Fanny Mac?
(13) s: (a) Well, like did you have a team when you were goin’ to school
on a little league, or-
r: (b) Oh yeh. [We used to-]
s: (c) [Really?]
(15) R: (a) And what we used to do we used to, we used to play it with a
golf ball . . . Right?
(19) R: (a) There was- there was one person up the street,
(b) he was a number writer,
(c) and he owned a car.
s: (d) A number writer?

Despite the formal similarity of these questions, we identified each one as

slightly different: Fanny Mac? (6) opened a turn-transition space for joint
reformulation of the query. Really? (13) and Right? (15) were information-
checks, A number writer? (19) was a request for clarification. Although all
these functions may very well be related to one another at one level of analysis,
the differences captured through their divergent descriptions reflect important
differences in the way they are situated in the speech event. For example, in
our discussion of (15) in section 3.3.2, we said that R uses this type of
utterance to check on S’s reception of a proposition or referent and to provide
a slot in which S can request clarification if that reception is deemed problem¬
atic. The function of Fanny Mac? (6, in section 3.2) is similar, but it is tailored
more specifically to the overall goal of the query: if L does find P’s formulation
of the query adequate (i.e. L does not need to request clarification), L can
resolve the query by providing the information being requested. Thus, al¬
though we might very well call Fanny Mac? in (6) an information-check, this
would ignore its more situated meaning within the reference interview. Ignor¬
ing differences such as these would ignore the way speech act identity is
embedded in speech events - a source of contextual variability that is crucial
to the ethnographic approach to discourse. Note, however, that the level of
specificity with which we discussed the function of specific utterances that
we called “questions” (and utterances that we labelled as different types of
questions) makes it difficult to draw conclusions about questions in general.
There are several points to draw from these examples before going on to the
next issue. First, categorizing questions may be difficult because the categories
themselves are not necessarily mutually exclusive: this is a major obstacle to
quantitative comparisons. Second, the discrimination of question types may
depend on semantic relationships across clauses. Information-seeking ques¬
tions can be semantically independent of prior talk, but information-checks
and clarification requests cannot. Thus, the degree to which the information
in a question is semantically tied to a prior proposition helps to determine its
 The Ethnography of Communication 185
question type. What an ethnographic perspective might point out is that
participants’ expectations for semantic continuity between propositions might
vary in different speech events - such that what seems like an information¬
seeking question in one speech event would not seem like one in another.
Third, a form that serves as one mode of questioning in one context may not
do so in another. The most general point, then, is that both types and tokens
of questions (and other speech acts) can be contextually relative and can
benefit from systematic analysis of communicative contexts. (I return to this
point in chapter 9.)
Consider, finally, that our sample analysis suggested that an ethnographic
approach to questions in interviews required consideration of how questions
fit into the interview itself, and thus an analysis of the interview as a speech event.
We described two different types of interviews and the questions asked within
them: we saw that the form and function of tjaese questions were intricately
tied to the communicative structure of the different interviews. This interrela¬
tionship between speech act and speech event suggests that speech acts stand
in a reciprocal relationship with speech events - both analytically and in our
own ways of thinking and knowing about communication. Thus, an ethno¬
graphic approach to discourse might also use findings about a speech act
(questions) to learn about a speech event (interviews) and posit knowledge of
this relationship as part of our communicative competence. To this end, one
would also need to examine more general cultural beliefs about transmission
of information, norms for interactions between strangers, and behavior in
public and private social establishments (e.g. library versus home), for these
(and other) beliefs also underlie the way we use language to question and be
questioned in interviews.

4 An ethnographic approach to discourse

We have seen in this chapter that an ethnographic approach to discourse seeks

to discover and analyze the structures and functions of communicating that
organize the use of language in speech situations, events, and acts. Knowledge
of these structures and functions is part of our communicative competence:
what we say and do has meaning only within a framework of cultural knowl¬
edge. The ways that we organize and conduct our lives through language are
thus ways of being and doing that are not only relative to other possibilities
for communicating, but also deeply embedded within the particular frame¬
work by which we - as members of our own specific communities - make
sense out of experience.
Although an ethnographic approach provides an analysis of language, it
views language as but one part of a complex pattern of actions and beliefs
that give meaning to people’s lives. Consistent with this assumption, our
 186 Approaches to Discourse Analysis

sample analysis did not just focus on questions within different types of
interviews, but on the goals, settings, participants, and other acts that con¬
stituted (and were constituted by) the speech events. Similarly, we also noted
that the questions themselves contributed to the definition of the speech event
“interview.” Thus, “talk is always constitutive of some portion of reality: it
either makes something already existing present to (or for) the participants or
creates something new” (Duranti 1988: 225).

Exercises

1 Analyze the following example of a library reference interview through an

ethnographic perspective.

L: (a) Yes. Can I help you?

P: (b) Is there a way that I can find out where in periodicals this
periodical is.. ..
(c) Archives in [Pathology and Laboratory Medicine.
L: (d) [Pathology. Yeh.
[6 seconds!
L: (e) Okay.
(f) Um, there's a listing under Archives of Pathology.
(g) It said it was called the Archives of
Pathology and Laboratory Medicine=
P: (h) periodical?
L: (i) from nineteen twenty six to twenty eight?
1
( ) [But now it's
P: (k) [This is just their March issue-
L: (I) Now it's just called the Archives of
Pathology, according to this.
(m) It's .. . I'd- when- that could be wrong or this could be wrong
(n) I don't know which.
(o) Uh it's at the Georgetown Medical School Library and at the
George Washington . .. University, Medical Library, and at
Howard's Medical Library.
P: (P) Okay,
(q) I'm going to be going to G. W. tomorrow.
(r) Would that be- is that the main library, or is it a separate [li-
L: (S)
[There's
a separate library.
(t) Are you going on the subway? [Be-
P: (U) [I'm going to a class at-at-at uh
George Washington so . ..
L: (V) The- the medical school is when you
come up the escalator,
(w) it's immediately to your right.
 The Ethnography of Communication 187
(x) You have to turn right,
(y) and then, turn left to go into the building,
(z) but, when you come up the escalator it's the building.
(aa) It says Medical School.
P: (bb) Okay. So it's in the medical school and it's=
L: (cc) z And=
P: =[not in the library.
L: (dd) =[it's ju- z Yes. And I'm just checking to-it does say open
to the public .. . the Medical School Library.
p: (ee) So I can just go in and use it.
l: (ff) Uhhuh.
p: (gg) Thank you very much.

2 Suppose you were setting out to do an ethnography of communication in

ypur own culture. What speech situations and-'events^would you try to tape-
record? Why?
3 Tape-record a session either in a classroom where English is being taught
as a second language, or in a classroom where a foreign language is being
taught to native English speakers. How are the assumptions of an ethnographic
approach to discourse reflected in the teaching strategies used in the classroom
(or in the textbooks)? If they are not reflected, how might they be? If the
students are having particular difficulties, how might they be explained in
ethnographic terms?
4 Although the sample analysis in this chapter focused upon question-
answer sequences, the ethnography of communication can also offer a rich
contextual framework for the analysis of pronouns (e.g. Hanks 1990) and other
referring terms. For example, what cultural assumptions are revealed by the
use of the vague pronoun they, or the expression a colored guy, in the "I was
robbed" story presented in chapter 6? In what other speech situations are the
same referring terms used? Or consider the referring expressions in the sample
data in appendix 3: e.g. who are the different they'sl How are they woven not
only into the communicative structure of the argument (e.g. participants, goals)
but also into the cultural themes underlying the argument?
5 Collect fifteen to twenty examples of service encounters. Use the etic grid
to compare the communicative structure of these speech events (including, but
not limited to, the use of questions) with the interviews discussed in this
chapter. Alternatively, tape-record and transcribe examples of other interviews
(e.g. on television talk shows, news interviews) and compare their communica¬
tive structure (and their question-answer sequences) to reference interviews
and research interviews.

Notes

1 Although Hymes is using Cassier’s point in relation to the speech used during Native
American verbal art, I believe that the same point applies to all communicative events.
2 Since P is the participant in need of information, and L is the one who supplies
information, P can be seen as the IVer and L the IVee. This role designation clearly
188 Approaches to Discourse Analysis
predicts the distribution of acts among participant: P presents a query (often through
a question) just as an IVer in a research interview (3.3) requests information (often
through questions). However, an alternative to the division of interactional work
noted in table 5.1 is to suggest that P and L both function as IVer and IVee. Since
L is in an institutionalized role that provides access to, and control over, information
that P needs, we might compare L’s role to that of a doctor (or a job counselor) who
conducts an intake interview with a patient (or a client) so as to make a diagnosis
(or assess the other’s needs or abilities). It is in this sense that we can say that L is
interviewing P so as to more precisely assess what it is that P (as IVee) needs. This
designation of participant role interacts with the classification not just of events as
interview or not (e.g. we might prefer to think of parts of this speech event as more
like a “consultation” between expert and novice (Jacoby and Gonzales 1991) than
like an “interview”), but also with our underlying ideas of occupational relations and
obligations. Thus, an analysis of questions might help us build a more exact taxonomy
of speech events whose main goal is the seeking of information. Similarly, we might
find that although reference interviews are not prototypical interviews, they are more
“interview” like than sociolinguistic interviews (3.3) since their central goal (resolution
of a query) is a macro-level realization of an information-seeking question.
3 Reference interviews share some features with service encounters, e.g. the server (cf.
the librarian) may ask a series of questions about the product being considered for
purchase. Responses to such questions, however, do not guarantee that the customer
in a service encounter will actually make a purchase; the two speech events also differ
in that the reference interview is an exchange of information, whereas the service
encounter is a potential economic transaction that can benefit the server.
4 Note that yes from P need not be interpreted as an affirmative response to a question.
Yes is used not only when responding to requests, but also when accepting responsi¬
bility for the next move. For example, a graduate student taking her oral PhD exams
once responded yes to a WH question (What is the difference between . . . ) and then
proceeded to give the substantive answer - thus accepting the responsibility for
answering the question prior to actually providing the information.
5 In some ways, a query is very much like a question: P’s purpose in making a query
is to elicit information from L which he does not have, but which he presumes L to
have. I discuss this in section 3.4.1. Note also that although space prevents discussion
of other evidence of the centrality of the query, it is helpful to note the following: if
an interview is interrupted (e.g. by a phone call to L), participants return to con¬
sideration of the query; if the query cannot be satisfied, alternative help is offered
(cf. the provision of “accounts” when requests cannot be met or offers are rejected;
chapter 3, pp. 78-84); new queries are structurally separated from one another and
independently managed.
6 A common complaint among reference librarians concerns the patron’s formulation
of a query in terms that are too general (e.g. “do you have books on foreign
languages?”) to allow resolution within the organizational schema of library mater¬
ials. Compare the need of physicians to “translate” patients’ descriptions of their
own physical conditions into terms that can be interpreted as “symptoms” and
understood in relation to a physiological model of the body and the schemas through
which diseases and syndromes are defined.
7 See Cameron et al. (1991) for arguments to this effect. Language and Communica¬
tion volume 13, no. 2 (April 1993) is devoted to discussion of the social and political
role of research interviews.
 The Ethnography of Communication 189

8 These properties are not only related to the form and meaning of utterances within
an interview: they are also interrelated with one another. The sociolinguist’s role
as “interviewer,” for example, emerges from the goals of the interview and allows
greater control of its act sequence.
9 I do not include informants’ questions to each other. My impression is either that
these questions duplicate the interview framework (e.g. when one informant takes
the role of the interviewer to replay the interviewer’s question to another informant)
or that they are questions that are outside of the interview frame itself (Schiffrin
1992a).
10 Compare the use of politeness forms that are conventionally used in one social
setting, or for one social relationship, as a strategy through which to alter the
meanings of another setting or relationship (Ervin-Tripp 1976).
11 Although we cannot be sure that the informant had even attended to the inter¬
viewer’s pre-empted description - let alone completed it in the way S intended - it
is important to note that this kind of effort to create bot£ local and global cohesion
by drawing from others’ contributions is what 1 often find myself doing when trying
to be conversational during an interview.
6 Pragmatics

1 Introduction

Pragmatics is another broad approach to discourse: it deals with three con¬

cepts (meaning, context, communication) that are themselves extremely vast
and unwieldy. Given such breadth, it is not surprising that the scope of
pragmatics is so wide, or that pragmatics faces definitional dilemmas similar
to those faced by discourse analysis. Levinson (1983), for example, devotes
an entire chapter of his pragmatics textbook to the problem of definition, and
he finds none that successfully limit the scope of the field as well as represent
it as a unified scholarly endeavor. A consequence of this breadth is the need
to devote more space to discussion of how pragmatics can provide an ap¬
proach to the specific discourse problem considered by our sample analysis
(referring sequences) than is necessary for the other approaches to discourse
being discussed.
This chapter focuses on one particular type of pragmatics - Gricean prag¬
matics - not only because some other definitions of pragmatics cover much of
the same ground as discourse analysis, but because this theory has become
“the hub of pragmatics research” (Fasold 1990: 128). In section 2.0,1 describe
the philosopher H. P. Grice’s ideas about speaker meaning (2.1) and the
cooperative principle (2.2). After some general points about the application
of Gricean pragmatics to referring terms (section 3), I use one particular
discourse to suggest how Gricean pragmatics provides an approach to dis¬
course analysis (section 4). Note that the sample analysis in this chapter begins
the second focus of the analyses in this book: referring terms. I concentrate
on the way the Gricean maxims of quantity and relevance explain the sequen¬
tial alternation between referring terms, particularly the use of indefinite/defi¬
nite, and explicit/inexplicit referring terms. These sequences reveal patterns,
and raise issues, to which we return in chapters 7 and 8. Section 5 summarizes
the pragmatic approach.
 Pragmatics 191

2 Defining pragmatics

Pragmatics was defined by Morris (1938) as a branch of semiotics, the study

of signs (but see Givon 1989: 9-25, for discussion of its earlier roots). Morris
(p. 81) viewed semiosis (the process in which something functions as a sign)
as having four parts. A sign vehicle is that which acts as a sign; a designatum
is that to which the sign refers; an interpretant is the effect in virtue of which
the sign vehicle is a sign; an interpreter is the organism upon whom the sign
has an effect. Put another way, something is a sign of a designatum for an
interpreter to the degree that the interpreter takes account of the designatum
in virtue of the presence of the sign (p. 82). In Morris’^ own terms: “Semiosis
is •. . . a mediated-taking-account-of. The mediators are sign vehicles; the tak-
ings-account-of are interpretants; the agents of the process are interpreters;
what is taken account of are designata” (p. 82; emphasis in original).
In addition to defining different aspects of the semiosis process, Morris
identified three ways of studying signs: syntax is the study of formal relations
of signs to one another, semantics is the study of how signs are related to the
objects to which they are applicable (their designata), pragmatics is the study
of the relation of signs to interpreters. Thus, pragmatics is the study of how
interpreters engage in the “taking-account-of” designata (the construction of
interpretants) of sign-vehicles.
Contemporary discussions of pragmatics (although not viewed within the
behaviorist framework of Morris) all take the relationship of signs to their
users to be central to pragmatics. In the rest of this section, I describe Gricean
pragmatics: a contemporary version of pragmatics which focuses on meaning
in context, but expands both the “sign” and the “user” ends of the sign-user
relationship. I discuss two central concepts of Gricean pragmatics: speaker
meaning (2.1) and the cooperative principle (2.2).

2.1 Speaker meaning

The first concept important to Gricean pragmatics is speaker meaning. As we

will see, speaker meaning not only allows a distinction between two kinds of
meaning (and hence, a division between semantic and pragmatic meaning), it
also suggests a particular view of human communication that focuses on
intentions (chapter 11).
Grice (1957) separates non-natural meaning (meaning-nn) from natural
meaning. Natural meaning is devoid of human intentionality, as in, for
example, Those spots mean measles. Non-natural meaning is roughly equival¬
ent to intentional communication (but see below): Grice’s example is Those
three rings on the bell (of the bus) mean that the bus is full (p. 53). A critical
feature of meaning-nn is that it is intended to be recognized in a particular
192 Approaches to Discourse Analysis

way by a recipient. As Grice (p. 58) states: “ ‘A meant-nn something by x’ is

(roughly) equivalent to ‘A intended the utterance of x to produce some effect
in an audience by means of the recognition of this intention’.” Implicit in this
definition of meaning-nn is a second intention — the intention that a recipient
recognize the speaker’s communicative intention. Pursuing Grice’s example of
three rings on the bell, we might say that for the three rings to have meaning-
nn, it is sufficient for the bus riders to recognize that the conductor (or
whoever caused the three bells to ring) intended the effect of the bells (i.e.
“the bus is full”) to be produced by their recognition of his intention. (See
Recanati 1987 for discussion of whether such recognition is necessary and
sufficient.)
Strawson (1964: 155) separates not two, but three, intentions in Grice’s
formulation. In Strawson’s terms, to mean something by x, S must intend the
following:

(a) S’s utterance of x to produce a certain response r in a certain

audience A;
(b) A to recognize S’s intention (a);
(c) A’s recognition of S’s intention (a) to function as at least part of A’s
reason for A’s response r.

We can view these intentions as cycling back upon one another: intention (c)
is that intention (b) function as at least part of the reason for the fulfillment
of intention (a). It is only when the three intentions are activated that an
individual is acting as a communicator; it is only when the three intentions
are realized that communication has occurred. Thus, to convey a speaker
meaning, a communicator makes manifest (displays) the sort of reflexive
intention described above.
Where does semantic meaning enter into Grice’s framework? That is, how
are the stable senses of words (their literal, conventional meanings) related to
speaker meaning? (See Grice 1968.) Certainly a person may intend to com¬
municate by the words “Sam is a boy” precisely the conventional under¬
standings associated with those words; in Grice’s terms, a speaker may intend
to create in a hearer a recognition of the intention to convey that proposition
about the world, i.e. about Sam’s identity. But what if someone states Boys
will be boys, or Sam is Sam? If we were to focus just on the meanings that
these utterances have by virtue of their conventional meanings, then we would
be able to do little more than identify them as tautologies. Yet speakers
routinely use such utterances to mean considerably more than what is con¬
veyed by their semantic meanings (but see Fraser 1988; Ward and Hirschberg
1991; Wierzbicka 1987).
What Grice’s framework does is allow speaker meaning to be relatively free
of conventional meaning; in fact, the critical insight of Grice’s meaning-nn for
our purposes is that what the speaker intends to communicate need not be
related to conventional meanings at all, and not conventionally “attached” to
 Pragmatics 193

the words being used. Thus, as we discuss in section 2.2, speaker-meaning

need not be code-related, i.e. it may be inferred through processes quite
different from the encoding and decoding processes assumed by the code
model of communication (chapter 11). But how do speaker-meanings arise?
We will find an answer to this question when we discuss the next concept
critical to Gricean pragmatics: the cooperative principle.

2.2 The cooperative principle

In order to understand the cooperative principle (CP), it is helpful to first

describe Grice’s view of logical meaning in relation to natural language. Our
discussion will also focus on the concept of implicature: an inference about
speaker intention that arises from a recipient’s usei*of both semantic (i.e.
logical) meanings and conversational principles. Because implicatures are re¬
lated to semantic meaning, the “sign” remains important in meaning-nn. But
because implicatures are also dependent on conversational principles, “con¬
text” mediates the sign-user relationship.
Grice (1975) is concerned with the relationship between logic and conversa¬
tion (in fact, “Logic and conversation” is the title of the paper). As Grice
explains, natural language utterances do not seem to convey the same
meanings that corresponding logical propositions would:

It is a commonplace of philosophical logic that there are, or appear to

be, divergences in meaning between, on the one hand, at least some of
what I shall call the formal devices —, A, V, z>, (x), 3(x), J(x). . . and, on
the other, what are taken to be their analogs or counterparts in natural
language - such expressions as not, and, or, if all, some, or (at least one),
the. (Grice 1975: 41)

For example, the utterance I went to the store and I put gas in the car might
be represented as “P & Q” (in which “P” represents the proposition in the
first clause, and “Q” the second). But the interpretation of the utterance is
broader than that which would rest upon the logical meaning of the connective
The logical meaning of can tell us, for example, that “P & Q” is
true only if both P and Q are true. We typically understand such an utterance,
however, as conveying temporal sequence (I went to the store before I put gas
in the car). Grice differentiates such understandings from logical meanings by
saying that logical meanings are part of what someone “says”: to “say” is
“closely related to the conventional meaning of the words (the sentence) he
has uttered” (p. 44). The broader interpretations are what someone “impli¬
cates” (although he reserves discussion of the meaning of “implicate” until
after his discussion of conversation; see below). This distinction allows Grice
to propose that natural language expressions really do not diverge from the
formal devices of logicians. Rather, the added meanings that seem to appear
194 Approaches to Discourse Analysis

in utterances are implicatures, and they are due to rules and principles of
conversation; more precisely, to the cooperative principle underlying com¬
munication.
In order to support his proposal, Grice (p. 45) makes several initial observa¬
tions about conversation:

Our talk exchanges do not normally consist of a succession of discon¬

nected remarks, and would not be rational if they did. They are charac¬
teristically, to some degree at least, cooperative efforts; and each
participant recognizes in them, to some extent, a common purpose or set
of purposes, or at least a mutually accepted direction ... at each stage,
some possible conversational moves would be excluded as conversation¬
ally unsuitable, (emphasis in original)

Going on from these initial and very general observations, Grice (p. 45)
proposes a general principle which participants will be expected to observe:
“Make your conversational contribution such as is required, at the stage at
which it occurs, by the accepted purpose or direction of the talk exchange in
which you are engaged.” This principle is labelled “the cooperative principle”
(CP) and it consists of four more specific maxims (p. 46):

Quantity:
1 Make your contribution as informative as is required (for the
current purposes of the exchange).
2 Do not make your contribution more informative than is
required.
Quality
Try to make your contribution one that is true.
1 Do not say what you believe to be false.
2 Do not say that for which you lack adequate evidence.
Relation: Be relevant.
Manner:
Be perspicuous.
1 Avoid obscurity of expression.
2 Avoid ambiguity.
3 Be brief (avoid unnecessary prolixity).
4 Be orderly.

It is important to note that these maxims (indeed, the CP in general) derive

not from the nature of conversation per se, but from the fact that talking is
“a special case or variety of purposive, indeed rational, behavior” (p. 47). We
will see in a moment that it is these principles that provide a basis for the
inference of implicatures: the CP and its attendant maxims allow speakers to
lead their hearers to interpretations of their communicative intent (speaker
meaning) that go beyond the logical meanings of what they “say.” Note that
 Pragmatics 195

since such principles are not principles of conversation - indeed, not even
principles of language use - they do little to return us to the “sign” side of
the sign-user relationship so central to Morris’s pragmatics. But when we take
a closer look at the way hearers arrive at implicatures, we will see that this
inferencing process includes a hearers’ interpretation of what a speaker “says,”
and thus builds upon semantic meaning.
A crucial feature of implicatures is that they must be capable of being
calculated by a hearer. Grice (p. 50) describes the process as follows:

To work out that a particular conversational implicature is present, the

hearer will rely on the following data:
(1) the conventional meanings of the words used, together with
the identity of any references that may be involved
(2) the CP and its maxims
(3) the context, linguistic or otherwise, of the utterance
(4) other items of background knowledge
(5) the fact (or supposed fact) that all relevant items falling under
the previous headings are available to both participants and
both participants know or assume this to be the case.

Given this basic process, implicatures can be created in one of three ways: a
maxim can be followed in a straightforward way, a maxim can be violated
because of a clash with another maxim, or a maxim can be flouted. I illustrate
each below.
Let us begin with our earlier example I went to the store and I put gas in
the car to show how implicatures can be calculated in a straightforward way.
We have already said (consistent with (1) above) that when I present this
utterance, I “say” something, and that what I “say” has conventional meaning,
including the meaning of the logical connective We also suggested that
an implicature of temporal order is created, i.e. that I went to the store before
I put gas in the car. How does this happen? We see from Grice’s description
that we need to make a general assumption that both I and my hearer know
the CP and its maxims (2 above) as well as the context (linguistic or otherwise)
of the utterance (3 above) and other items of background knowledge (4
above). Given these assumptions, we might say that an implicature of temporal
order arises from the Manner maxim “Be orderly,” i.e. it is orderly to present
events in the order in which they occur, from the Manner maxim “Be brief,”
or from the Quantity maxim, i.e. it is not necessary to present temporal informa¬
tion about events if that information can be inferred iconically (from the order
of clauses in the text). Whichever maxim is responsible for the implicature, it
is important to note that the implicature arises without my hearer assuming
that I have violated a maxim: the implicature can be inferred under the
assumption that I am following the maxims in a straightforward way.
Although conventional meaning is said to be part of what must be known
for the calculation of implicatures, it does not seem to play that central a role
196 Approaches to Discourse Analysis

in implicatures that arise when the maxims are followed in a straightforward

way. Nor is it that important when an implicature arises from a clash between
maxims (Grice’s second possibility). Let me provide Grice’s example (p. 51).

A: Where does C live?

b: Somewhere in the South of France. (Gloss: There is no reason to suppose
that B is opting out; his answer is, as he well knows, less informative than
is required to meet A’s needs. This infringement of the first maxim of
Quantity can be explained only by the supposition that B is aware that to
be more informative would be to say something that infringed the maxim
of Quantity, “Don’t say what you lack adequate evidence for,” so B
implicates that he does not know in which town C lives.)

It is the third possible route toward implicatures that shows most clearly the
role of conventional meaning: a flouting of the maxims. Grice (pp. 52-6)
explains that people can say things that seem to violate one or more of the
maxims, e.g. one can violate the Quantity maxim by presenting less informa¬
tion than seems to be required, the Quality maxim by saying things which
seem to be false.2 When this occurs, however, hearers still assume that a
speaker is following the general CP; thus, they find an interpretation of the
violation that allows them to preserve this assumption of cooperation. It turns
out that some maxim violations are apparent only because of semantic prob¬
lems - and it is here that we can see the role of conventional meaning most
clearly in the calculation of implicatures. Grice’s example of metaphor, which
flouts the maxim of quality, provides the clearest illustration:

Examples like You are the cream in my coffee characteristically involve

categorial falsity, so the contradictory of what the speaker has made as
if to say will, strictly speaking, be a truism; so it cannot be THAT that
such a speaker is trying to get across. The most likely supposition is that
the speaker is attributing to his audience some feature or features in
respect of which the audience resembles (more or less fancifully) the
mentioned substance, (p. 53)

Thus, hearers would not know that a maxim had been violated - and not
search for an implicature - if they did not first know that the semantic
properties of “human” (evoked by the pronoun you) exclude the possibility
of also being “liquid” (cream).
In this section, I discussed implicatures in relation to the CP and its maxims.
Before closing, let us recall (from section 2.1) that the independence of
meaning-nn from the linguistic sign per se freed the study of speaker meaning
from the sign-user relationship (the focus of Morris’s pragmatics). We have
just seen that implicatures rest primarily on a principle of communication -
the cooperative principle - rather than principles of language per se: implica¬
tures allow us to account for how people convey messages not provided
 Pragmatics 197
through the stable semantic meaning of their words. But we have also seen
that the process by which speaker-meaning is inferred includes within it the
interpretation of conventional meaning. And because conventional meaning
plays a role in hearers’ inference of speaker meaning, the linguistic sign (the
code) also has a role in Gricean pragmatics (see chapter 11).

3 Referring terms: pragmatic processes in

discourse

Gricean pragmatics provides a way to analyze th« inference of speaker

meaning: how hearers infer the intentions underlying a speaker’s utterance. It
is not intended as an approach to the analysis of discourse, i.e. to sequences
of utterances. In the remainder of this chapter, however, I show how the CP
can help define the way information from one utterance contributes to the
meaning of another utterance, thereby contributing to our knowledge of rela¬
tionships between utterances.
Our sample analysis in section 4 focuses on the organization of referring
terms in a narrative. Before beginning this analysis, I introduce several terms
to be used in section 4 and in the other chapters (7 and 8) dealing with similar
issues (3.1). I also present brief examples suggesting the importance of the
maxims of quantity and relation (3.2), motivate a discourse analysis of refer¬
ring sequences (3.3), and comment on methodology (3.4).

3.1 Referring as a pragmatic process in discourse

Although reference and referring terms have been analyzed through many
different perspectives in philosophy and linguistics, scholars often view the
process of referring to entities in the universe of discourse as pragmatic -
simply because it is a process involving speakers, their intentions, actions, and
knowledge. Givon (1989: 175), for example, states that:

Reference in a Universe of Discourse is already a crypto pragmatic affair.

This is because every universe of discourse is opened (“established”) -
for whatever purpose - by a speaker. And that speaker then intends
entities in that universe of discourse to either refer or not refer. And it
seems that in human language it is that referential intent of the speaker
that controls the grammar of reference, (emphasis in original)

Similarly, Searle (1969: chapter 4) views referring as a speech act, governed by

conditions similar to those governing the performance of actions such as promises
198 Approaches to Discourse Analysis

or requests (see chapter 3). Clark and Wilkes-Gibbs (1986) go still further to
speak of referring as a “collaborative process”: they suggest that although a
speaker can propose a referent, the identification of the referent needs to be
seen as an outcome of speaker-hearer interaction (see also Martinich 1984:
161-2). Some particular types of reference are especially dependent on mutual
knowledge: Clark and Murphy (1982), for example, argue that definite refer¬
ence (reference to an entity with the expectation that the hearer will be able
to make a similar identification; see below) depends critically upon mutual
knowledge, beliefs, and suppositions (see also Clark and Marshall 1981). Thus,
as seen from both the speaker’s and the hearer’s point of view, the process by
which expressions refer to an entity can be seen as pragmatic: “the mechanism
by which referring expressions enable an interpreter to infer an intended
referent is not strictly semantic or truth-conditional, but involves the cooper¬
ative exploitation of supposed mutual knowledge” (Green 1989: 47).
Note, however, that Grice himself does not consider reference in detail. Grice
includes “the identity of any reference that may be involved” (1975: 50), along
with the conventional meanings of words, as part of the information relied
upon in the calculation of a conversational implicature. This exclusion of
reference from the domain of pragmatics suggests that the identity of a referent
is not inferred through use of the CP (cf. Clark 1975). It is also faithful to
Morris’s (1938) view of semantics as the study of how signs are related to the
objects to which they are applicable, and pragmatics as the study of the
relation of signs to interpreters. Yet, Grice’s discussion also suggests that
specific maxims (e.g. quantity and relevance) constrain the speaker’s choice of
referring terms - an effect that can be more easily reconciled with the view
that reference is a relationship between language and an entity in one’s (men¬
tal) discourse model (Green 1989; Webber 1979; see also Kronfeld 1990). I
return to this point in sections 3.2 and 4.
Viewing referential processes as pragmatic is important not just for the
general reasons sketched above, but also because it may help us account for
distributional differences between definite and indefinite forms, and explicit
and inexplicit forms. Referring terms that are definite are noun phrases with
the definite article the, possessives, pronouns, names, titles; those that are
indefinite include noun phrases with the indefinite article a, with quantifiers,
and with numerals. Indefinites can also be the predicates in existential there
sentences (but see Bolinger 1977; Ziv 1978; also chapters 7 and 8). This
distributional restriction is important since some forms (e.g. this: Prince 1978)
can be used for both definite and indefinite reference.
Although it is relatively easy to formally differentiate referring terms that
are definite from those that are indefinite, it is not as easy to define their
functional differences, i.e. to reveal the conditions under which different terms
can be appropriately used. Referents are often introduced into discourse with
terms that are indefinite and explicit (e.g. a woman I work with) and conti¬
nued with terms that are definite and inexplicit (e.g. she). The reasons for this
distribution can be conveyed in pragmatic terms. Definite forms are often said
 Pragmatics 199

to indicate S’s intention to refer to a single entity that can be specifically

identified by S, and that S expects H to be able to identify from whatever clues
(textual, contextual) are available (Ariel 1990; Green 1989: 46).4 Indefinite
forms, on the other hand, are said to indicate “the fact that the term is being
introduced into the discourse for the first time - and without any speaker’s
presumption of bearer’s familiarity” (Givon 1989: 179, emphasis in original).
The explicitness of referring terms can also be viewed in pragmatic terms.
Whereas definiteness has to do with S’s assumption that H will be able to
identify a single, specific entity to which S intends to refer, explicitness has to
do with the presentation of information that actually enables H to correctly
identify a referent, i.e. the lexical clues that allow H to single out whom (or
what) S intends to differentiate from other potential referents. (In this sense,
explicitness is similar to informativity.) Definiteness and explicitness can easily
crosscut one another. For example, I can use a-H of thfe following (and more)
definite descriptions to refer to the same person: my husband, Louis, Dr Scavo,
the man I live with. These are all more explicit than the indefinites an adult
I live with, someone I met in college. The definite he, however, is less explicit
than the indefinites just given. Note, finally, that the explicit-inexplicit dis¬
tinction is more continuous than discrete: someone I met in college is more
explicit than someone l knew when I was younger, but less explicit than
someone I met in my third year of college; he is less explicit than any of the
above expressions, but more explicit than zero anaphora.
We have seen so far that the process of referring may be seen as pragmatic,
simply because it is a process involving speakers, their intentions, actions, and
knowledge. Similarly, both definiteness and explicitness can be defined in
pragmatic terms. Definiteness is concerned with S’s intentions and assump¬
tions about what H can be expected to know. Explicitness is partially motiv¬
ated by S’s cooperative intentions, i.e. information is presented to enable H
to identify an intended referent.

3.2 A preliminary Gricean analysis: quantity and relevance

Differences in both definiteness and explicitness among referring terms can be

associated not just with pragmatics in general, but with the Gricean maxims
of quantity and relevance. As I noted above, for example, I may refer to the
very same person in numerous ways: definite descriptions (my husband, Louis,
he. Dr Scavo, the man I live with) and indefinite descriptions (an adult I live
with, someone I met in college). Although I may intend each expression to
refer to the same person, they are not all equally suited to the task of helping
you identify that person. For example, use of the referring expression someone
I met in college may very well be accurate, but it is not explicit enough. Not
only do I know a great deal more about the person to whom I intend to refer
than where we met, but because I met many people in college, this description
would not allow you to uniquely identify who it is that I am talking about.
200 Approaches to Discourse Analysis

This problem can easily be cast in Gricean terms. Although someone I met in
college adheres to the maxim of quality, it violates the maxim of quantity: I
have not provided you with information sufficient for your task of selecting
from the set of referents “people I met in college” the one particular entity to
whom I intend to refer. Thus, the various ways that we may refer to someone
are ordered in terms of how much information they supply about that person
and about our relationship with that person.
Equally important as the quantity of information conveyed in a particular
referring term, however, is the relevance of that information (cf. Kronfeld’s
(1990) distinction between functional and conversational relevance). Here I
will consider an example that will allow us to begin to address not only
paradigmatic, but also syntagmatic, choices among referring terms. During a
sociolinguistic interview, a middle-aged woman (Sue) was talking about her
high school friends. She stated the following:

sue: (a) I always wanted to marry an Italian guy

IVer: (b) How come?
SUE: (c) I just wanted to.
(d) And I said it.
(e) And I did.

Although Sue did marry an Italian guy (And I did (e)), her statement in (a) of
her wish to do so is referentially opaque: we do not know whether she intends
to refer to a specific “Italian guy” who she wants to marry, or to anyone who
meets the description “Italian guy.” Because Sue has been talking about a
period of time (high school days) prior to her acquaintance with any Italian
guys, however, we can use this background knowledge to infer that Sue does
not have one specific “Italian guy” in mind. When the evoked time period of
Sue’s discourse switches in (e) to after her marriage - and I did (e) - we
understand that there is one specific “Italian guy” whom she has married.
When Sue goes on to describe how her husband’s family viewed his marriage
to her (since her family is part Italian and part Irish) she explicitly brings her
husband into the discourse. In (2), Sue is describing how her sister-in-law
married someone who was not completely Italian:

(2) sue (a) So she was the oldest daughter.

(b) And she broke the ice for me like she said.
(c) Y’know, for Tony, my husband.
IVer: (d) How did you meet Tony?

In (c), Sue mentions her husband in two ways: his name (Tony) and his kin
relationship (my husband). Thus, (1) and (2) have supplied three descriptive
resources for a next-mention: ethnicity, name, relationship.
Although the referring term (Tony, a proper noun) used by the IVer in (d)
(How did you meet Tony?) identifies a specific person, let us see how Sue
 Pragmatics 201

refers to Tony in her answer to the IVer’s question. (3) presents the section of
a narrative describing the first time Sue met Tony:

(3) (a) So anyway, this real dark looking, gangster looking guy, comes out
the door,
(b) and he looked like an old man to me.
(c) And he looked like a gangster.

Sue’s description of her first encounter with Tony uses a referring expression
(this real dark looking, gangster looking guy) that is explicit, but much less
restrictive than the expressions (Tony, my husband) that she used earlier.
(There are many more people who can fit into the category of “real dark
looking, gangster looking guys” than into the category of “my husband.”)
Note, however, that the shift in time frame and in information state alters the
descriptive categories that are available for reference. Jme’s description in (3a)
matches the quantity of information that she then had about Tony. It is also
relevant to her first meeting with Tony (or at least to the impression she wants
to convey about that meeting). Thus, the shift in time frame creates a scenario
to which a referring term can (or even, must) be both informationally appro¬
priate and relevant.
Our discussion of the way Sue mentions her husband in (2) and (3) illustrates
that relevance can be as important a pragmatic constraint on referring terms
as quantity of information. In (3), relevance shifted along with information
quantity: this real dark looking, gangster looking guy is consistent with S s
knowledge - her quantity of information - about Tony at that time. Relevance
can also figure in the choice of a referring term even without a shift in quantity.
In Schiffrin (1988b), for example, I discuss a narrative told in response to a
question I had asked about local availability of medical help. Despite her
knowledge of explicit information about a particular doctor (e.g. his name),
my respondent refers initially to a neighborhood doctor who we use - simply
because it was that categorization that was relevant to my prior question.
So far, we have seen that referring terms convey different quantities of
information and can be more or less relevant to ongoing discourse. I have also
noted that quantity and relevance may work together (or separately) to con¬
strain referring terms. In the sample analysis in section 4, we see in more detail
how the maxims of quantity and relation work relative to information pro¬
vided in other utterances, thereby influencing both paradigmatic and syntag-
matic choices within referring sequences. Before we do this, however, it will
be helpful to view referring not just as a pragmatic process, but also as a
sequential process.

3.3 Referring as a discourse process

Referring may be seen as a sequential problem. Rather than analyze referring
terms per se, we may analyze referring sequences-, how is reference initiated?
202 Approaches to Discourse Analysis

How is reference continued? This shift in perspective is important primarily

because it allows us to view reference in terms familiar to discourse analysis.
In chapter 2, we saw that discourse analysts often focus on sequential patterns,
e.g. of clauses, moves, actions. This is because discourse creates syntagmatic
options: what occurs at one point in time (in one slot in a sequence) constrains
the options (and selection of an option) at a later point in time (in a later slot
in the sequence). Discourse is frequently noted to be constrained in this way.
The occurrence of what is heard as a question in [slot 1], for example,
constrains the way a remark in [slot 2] will be heard, i.e. as an answer (see
chapters 3, 4, 5, and 7). Research on referring terms in discourse reveals the
existence of similar kinds of expectations about sequences of referring terms:
a particular referring term in [slot 1] constrains what terms will occur in [slot
2] and, even further, the way a sequentially next term, e.g. a pronoun, will be
interpreted. Such research has focused a great deal of attention not only on
characterizing the slots in which particular terms appear, e.g. in terms of
paragraph, rhetorical, or story structure (Fox 1987; Tomlin 1987), or in terms
of recency and accessibility (Ariel 1990; Givon 1989), but also on accounting
for why particular expressions are suited for use in different slots. Thus,
referring terms exhibit a dependence across clauses that can be seen as similar
to other sequential dependences in discourse.
Focusing on referring sequences also allows us to clearly differentiate first-
mentions of a referent (cf. [slot 1]) from next-mentions (cf. [slot 2]). As noted
earlier, first-mentions are often indefinite noun phrases (e.g. a + NP) and
explicit (e.g. nominals rather than pronominals). Next-mentions may be defi¬
nite and less explicit (e.g. pronouns rather than nominals). The main func¬
tional factor underlying this distribution is information status (cf. Chafe 1974,
1976, 1980, 1987) and accessibility (Ariel 1990). A first-mention evokes an
entity that may be new to H or that H cannot be assumed to currently have
in consciousness. A next-mention evokes an entity that need not be assumed
to be new to H: H might either have that entity in consciousness or be able
to retrieve it from memory.
In sum, some very different questions can be asked depending on the dis¬
course location of a referring term. For first-mentions, for example, we may
ask how it is that H identifies new entities. Does S provide assistance by tying
a new entity to information that is already either situationally or textually
evoked (Prince 1981)? How is the specific formulation of such references
sensitive to different aspects of physical or social context (e.g. Schegloff
1972b; Sacks and Schegloff 1979)? Our questions are quite different for
next-mentions. How does S provide sufficient information to guide H toward
identification of a referent even if a referring term is very inexplicit (e.g. a
pronoun or even zero anaphora)? Alternatively, why would S use a referring
term that provides more than enough information for identification of a
referent?
Now that we have motivated the analysis of referring terms in relation not
only to pragmatics, but to discourse analysis in general, we should be ready
 Pragmatics 203

to present a more detailed Gricean analysis of referring sequences. However,

there is just one more issue to mention: a methodological issue.

3.4 The importance of method and data

Before going on to apply Gricean pragmatics to some of the issues raised

above, it is important to make some observations about method and data.
Although I am applying Gricean pragmatics to discourse, the application that
I propose does not always adhere to the typical methodology of Gricean
analyses. As we discussed above, the disciplinary origins of Gricean pragmatics
are in philosophy. Contemporary pragmatics is certainly a part of linguistics
(Levinson 1983), but it is part of a linguistics that takes as data a hypothetical
sentence, adds to it a hypothetical context, and,sails it%n utterance (e.g. Cole
1981). However, if we want to use Gricean pragmatics as an approach to a
discourse analysis that is committed to analyzing how people actually use
utterances to communicate with one another, then we need to make the sorts
of adjustments that I will be making in this section (cf. Levinson 1983: chapter
6; Schiffrin 1987b). Most importantly, rather than rely on constructed sen¬
tences that are treated as if they are contextualized utterances, we must use
actual utterances that have been produced by speakers in contexts. Such a shift
in data can have far-reaching effects on Gricean pragmatics itself.
Although I do not discuss the effects of using actual utterances on Gricean
pragmatics in general (but see Schiffrin 1987b), some effects will become
apparent in my analysis. The decision to base my illustration of Gricean
pragmatics on the referring terms in one story means that the analysis is guided
in large part by the referring sequences that are in that story. Thus, rather
than propose relatively abstract relationships between maxims and try to find
examples that substantiate (or refute) those relationships (e.g. Horn 1985b),
we will try to use the maxims to help explicate particular uses of referring
terms in particular utterances in one particular story. This is not to say that
such an analysis might not end up influencing our understanding of abstract
relationships between different maxims. But this is not what guides the ana¬
lysis. Rather, what guides the analysis is an effort to resolve problems that
stem from concrete observations about language use.

4 Sample analysis: referring sequences in

narrative

The sample analysis in this section is based upon the referring expressions in
one particular discourse - a narrative. After presenting the narrative (4.1), I show
204 Approaches to Discourse Analysis

how the maxims of quantity and relevance help to account for the referring
sequences in this particular discourse (4.2). I summarize in section 4.3.

4.1 The data

Stories are useful texts in which to analyze referring sequences. In telling a

story, a speaker constructs a story world in which a limited number of entities
act and interact with one another in a defined location and for a limited period
of time. In addition, although stories are situated within conversations, a story
world can be somewhat independent of that world and can involve distinct
time (and information state) shifts away from that world. This means that
referents not yet textually evoked may have to be introduced and made
relevant to domains that are not currently relevant to the conversation per se.
Thus, stories offer the opportunity to find referring sequences in which new
referents are introduced and continually used in a particular framework to
7
which they are relevant.
The story on which I focus is quite long, and I will be drawing from different
sections of it to make various points. For these reasons, it will help to see it
as a whole before we begin its analysis; I present it as (4). Lines in the narrative
itself (the story world) are alphabetized; others are numbered.

IVer: Have you ever been robbed?

GARY: I was robbed. At night.
IVer: When was this?
GARY: I guess about six or eight weeks ago.
IVer: How’d it happen?
Gary: (a) I picked up a fare [1] at Board and Elm.
(b) He [1] was a colored guy.
(c) He [1] flagged me down.
(d) So I wasn’t- it was late at night.
(e) It was around one thirty in the morning.
(f) And I was gonna turn it [1] down.
(g) So I figured, y’know, I’d pick him [1] up anyway.
(h) But then two friends [2] came out.
(i) So they [?3] said, “We’re only going right around the
corner more or less,” y’know. “Could you take us up?”
(j) They [?3] gave me an address.
(k) They [?3] said, “We’ll get- show you how to get there.”
(1) So it was maybe only three blocks away,
(m) so they [?3] said- I think they [?3] gave me an address
that’s at the center of the block,
(n) and there was a small light on at the house. /Mhm./
(o) So I pulled up,
(P) I turned on the light,
 Pragmatics 205

(q) and the guy he [? 1] grabbed me from the back of my head

(r) and [?1] put a gun to it.
Andhh uh,
IVer: Jesus.
GARY: (s) I had about fifty dollars in my sock
(t) and I had about nine dollars on me.
(u) So I gave ’em [?3] the nine dollars
(v) and they [?3] wanted more.
(w) So they [?3] took my wallet.
(x) I didn’t want to give them [?3] the money.
(y) So then the guy up front he [?4] stuck a gun on my temple,
asking for more money.
(z) They [?3] were gonna, y’know, shoot me.
(aa) So I gave them [?3] the resf'of theftmoney.
IVer: (1) In your sock, [right
GARY: (2) [Yeh.
IVer: (3) If you hadn’t pulled out the money in your sock, do you
think they [?3] would’ve looked?
(4) Do they [?3] know you keep it here?
GARY: (5) No.
(6) But, from what other cab drivers [5] told me,
(7) [5] said I probably would’ve been shot.
IVer: (8) Oh, Gary
GARY: (9) That’s what they [5] tell me.
(10) Maybe they’re [5] doin’ it to scare me.
(bb) But, I gave ’em [?3] the money.
(cc) Like I figured-
(dd) I didn’t really get upset.
(ee) I wasn’t scared.
(ff) I wasn’t scared till after it was all over with.
(gg) At the time I don’t think- it never even-
(hh) all I thought of was Diane and the baby.
IVer: Su:re.
GARY: (ii) And uh, I gave the guy [?4] the money.
IVer: Of course.
GARY: (jj) And they [?3] said- well, they [?3] ripped out the radio
cabinet,
(kk) they [?3] ripped the microphone out of it.
(11) So I went- I backed down the street,
(mm) and I found a cop [6] right around the corner.
IVer: Jesus.
GARY: (nn) And uh, then when-
(oo) y’know, I told the cop [6] what happened.
(PP) He [6] said, “All right, get in your car and go back onto
Broad Street. I’ll meet you up there.”
206 Approaches to Discourse Analysis

(qq) Cause they [6a] were trying to pick people up for

identification.
(rr) That’s when it hit me.
(ss) My leg started jumpin’ up and down,
(tt) I just couldn’t control my nerves.=
IVer: [Jesus.]
GARY: (uu) =[But] then the whole time I was-
I- I didn’t even get scared or upset.
(vv) And I- I- maybe I thought subconsciously if I would’ve
been scared or shown that I’d have been scared, then I
probably would’ve been hurt.
(ww) Maybe they [?3] would’ve gotten jumpy.
(xx) Y’know, and . . .

The referring terms to be discussed have bracketed numbers to index the

referents and to help trace their mentions through the text. Some expressions
(e.g. they) are ambiguous; these have question marks preceding their indices.
Although it would be useful to be able to use Gricean pragmatics to resolve
referential ambiguity, this sample analysis has a more limited goal: show how
Gricean principles provide an interpretive basis for the various referential
possibilities that can be inferred and for the referring sequences that they
would create. This limitation follows Grice’s (1975: 50) view that “the identity
of any reference” is part of the information relied upon to calculate implicatures;
“identity” (and more generally, how signs are related to the objects to which they
are applicable, Morris’s view of semantics) is not itself a product of implicature.
(But the view that reference is a relationship between language and an entity in
one’s “mental/discourse model” more readily allows the use of Gricean prag¬
matics for reference itself.) Although I proceed for now as if each expression has
a clear referent, I discuss the ambiguity of several terms in the appropriate sections
and the general problem of ambiguity in section 4.4.
My analysis focuses on each human referent with a continued existence in
the story. These are listed below with the location (first-or next-mention) of
each referring term, whether the terms are indefinite (Indf) or definite (Df),
explicit (Ex) or inexplicit (Inex), and the referring terms used for each referent.
To avoid confusion, I use single quotation marks to distinguish my identifica-
tory uses of expressions as referents from the specific realizations of those
referents, e.g. the referent ‘cop’ may be realized as a cop, the cop, or he.

First mention Next mention

Referent 1 ‘passenger’
Indf/Ex Df/Inex Df/Ex
a fare (a) he (b, c), it (f), him (g), ‘zero’ (r) the guy he (q)

Referent 2 ‘two friends’

Indf/Ex none
two friends (h)
 Pragmatics 207

Referent 3 ‘they’
Df/Inex Df/Inex
they (i) they (j, k, m, u, v, w, z, jj, kk, ww) them (u, x,
aa, bb)

Referent 4 ‘guy up front’

Df/Ex Df/Ex
the guy up front he (y) the guy (ii)

Referent 5 ‘other cab drivers’

Indf/Ex Df/Inex
other cab drivers (6) ‘zero’ (6) they (9) they (10)

Referent 6 ‘cop’
Indf/Ex Df/Ex Df/Inex
a cop (nn) the cop (oo) he (pp)"

Referent 6a ‘cops’
Df/Inex none
they (qq)

This chart shows that some of the referring sequences in (4) fit a common
pattern in which first-mentions presuppose less about the ability of hearers to
identify a referent than do next-mentions. First-mentions of ‘other cab drivers’
and ‘cop’ are indefinite and explicit; next-mentions of these same referents are
definite and inexplicit. The chart also shows other referring sequences that are
less typical. Next-mentions of ‘cop’ alternate between explicit (the cop) and
inexplicit terms (he). And whereas referent 3 (‘they’) is mentioned only pro-
nominally (they) regardless of location, referent 4 (‘guy up front’) is mentioned
only nominally (the guy up front he, the guy). Still other referents (‘two
friends’, ‘cops’) are evoked only once, the former explicitly (two friends), the
latter inexplicitly (they). Finally, it is also important to note the complex referent
‘passenger’ (that can be metonymically evoked as it) and that referents 6 (‘cop’)
and 6a (‘cops’) are linked by a relationship of inclusion. Since full discussion
of each referring sequence would occupy too much space for our current
purposes, I devote most of what follows to those sequences raising problems
that can be best addressed by a Gricean approach to discourse analysis.

4.2 The analysis

We see in this section that the maxims of quantity and relation work together
in a very general way to guide H’s interpretation of S’s referential intention.
We will use the story in (4) as a source of information (the quantity maxim)
to which successive referents in the story are relevant (the relation maxim).
Thus, we will see how referring terms are fit to the information presented in
a text, and made relevant to that information. Although my discussion is
208 Approaches to Discourse Analysis

centered on first- and next-mentions in (4), I try to point out the more general
applicability of Gricean pragmatics for analyses of referring sequences.

4.2.1 Referent 1: ‘passenger’

The referring expressions used for referent 1 raise several important issues. The
first issue concerns the labelling of the referent as ‘passenger’ and the use of
the word fare to evoke that referent. A ‘passenger’ is a person who acts in a
particular capacity within a particular situation: a ‘passenger’ pays a fee to
travel. Cabdrivers’ passengers are sometimes called “riders” and what they
pay is a “fare.” Thus, the referent ‘passenger’ combines two conceptually
related entities: a ‘person’ and a ‘fare’. It is this relationship that allows the
word fare to be used to evoke the referent ‘passenger’. More specifically,
saying fare to evoke ‘passenger’ is allowed by the conceptual relationship
known as metonymy: we use one entity (a fee paid) to refer to another (a
person who pays the fee) to which it is related (Lakoff and Johnson 1980:
chapter 8). The general relationship between ‘fare’ and ‘passenger’ is one in
which an object used (‘fare’) stands for its user (‘passenger’). An example of
this from Lakoff and Johnson (1980: 38) is “The sax has the flu today”. The
particular relationship between ‘fare’ and ‘passenger’ is one in which the user
is a source of the object (the fare) and transmits that object to a recipient (the
cabdriver). A similar relationship would hold if a salesperson said “The cash
and carries always need help to their cars,” or a newspaper delivery person
said “The weeklies are behind again.” In these hypothetical examples, the cash
and carries would evoke the people paying cash and carrying their own
merchandise (rather than waiting for its delivery); the weeklies would evoke
the people paying each week (rather than each month) for their newspapers.
As we noted in the chart above, referent 1 is introduced as an indefinite
explicit noun (a fare) and its next-mentions (he, it) are definite. This distribu¬
tion raises three other issues. First, why is the first-mention of “passenger” an
indefinite explicit noun (a fare) and why are its next-mentions definite inex¬
plicit pronouns (he, it)? Second, why do the next-mentions switch between it
and he, i.e. why isn’t the metonymic it always used to evoke the person who
pays a fare? Third, why do the referring expressions used for “passenger”
(he/him (b, c, g), the guy he (q), and ‘zero’ (r)) vary in explicitness? I consider
each issue in turn.
Consider, first, how Gricean pragmatics might explain the definiteness or
indefiniteness of first-mentions. Although first-mentions are typically indefin¬
ite, it is well known that background information can allow a first-mention
to be definite (Prince 1981; Brown and Yule 1983: 186). In Gricean terms,
information (quantity maxim) can allow a hearer to infer the existence of a
referent by providing a framework within which the existence of such a referent
is expected (relation maxim). As (5) illustrates, background information can
be textually presented.
 Pragmatics 209

(5) I went back to school.

The truant officer came after me.

In (5), the textual mention of ‘school’ is informative enough to allow us to

assume that a particular entity (‘truant officer’) is in (i.e. “exists” in) a
particular setting (‘school’). Background information can also be situationally
provided: if one or more interlocutors are near a school, for example, one
might use a definite description for a first-mention of a number of entities
typically relevant to that setting.
Let us return to (4): the first-mention of ‘passenger’ is indefinite, not definite.
A Gricean analysis might explain this in terms similar to those just suggested:
first-mention of ‘passenger’ is not definite because Gary did not present informa¬
tion (nor was such information contextually available) that would allow his
hearer to expect (i.e. find relevant) the existence'of suA an entity. To be more
concrete, ‘passenger’ was neither textually anchored (e.g. Gary did not men¬
tion a phone call from a dispatcher) nor situationally anchored (e.g. Gary was
not receiving instructions from a dispatcher). We can suggest, then, that
first-mentions are indefinite if neither prior text nor context provides informa¬
tion sufficient for S to assume that H can expect an entity to be relevant
enough to have an existence. (Compare discussion of ‘guy up front’ (4.3.4)
and initial definites in chapter 8, section 3.2.2.)
Although ‘passenger’ is not anchored in prior text or context, its first-men¬
tion is not without any relevance at all: ‘passenger’ is anchored to the speaking
situation, to a prior topic of talk, and to the story topic (cf. DuBois 1980;
Prince 1981). In saying I picked up a fare, Gary makes a fare (the metonymic
reference to ‘passenger’) relevant to himself and his own reported action. Since
the speaker is assumed present during an act of speaking, the link between
‘passenger’ and T anchors the referent to the speaking situation. Other informa¬
tion in the clause relates the referent to both prior and upcoming topics.
Earlier in the interview, Gary had been talking about ‘tough neighborhoods’.
Since Broad and Elm is an area of town known to be relatively dangerous,
Gary’s location of ‘fare’ at this intersection is relevant to that prior topic, as
well as the threat that he will face. Finally, notice that Gary begins his story
with an event (I picked up a fare) that is critical to what happens later in his
story. Had Gary not picked up a fare (an action about which he had doubts:
I was gonna turn it down (f)), he would not have been robbed. Thus, although
‘passenger’ is not anchored in prior text or context, its multiple relevancies
(to the speaker himself, to a prior topic of talk, and to later story events) are
conveyed through information within the same clause in which it is intro¬
duced.
Gricean pragmatics can also help explain the use of the lexical item fare
(rather than, for example, a robber, someone, or a guy) for the first-mention.
The general topic being discussed prior to Gary’s story was cabdriving. First-
mention of a fare situates the story as a type of experience relevant to this
topic, and as an encounter with someone who occupies an occupationally
210 Approaches to Discourse Analysis

defined position. Fare is also matched in terms of quantity^ and relevance to

what Gary knew when the experience itself began to unfold. Stories routinely
move speakers to a prior information state at which they did not know what
they came to know at later points in the story world. Story tellers often present
what they know later (time 2) as unknown to them when they are situated
earlier in the story world (time 1). This shift in temporal perspective also
creates a different world in which entities are inferrable as relevant (cf. our
discussion of this real dark looking, gangster looking guy in 3).
Mention of ‘passenger’ continues in the story as it (f). This use of the
pronoun it is not surprising: having introduced a fare, Gary has provided
information sufficient to allow his hearer to identify a referent without repeti¬
tion of the explicit noun. Furthermore, we can easily find an explicit, textually
accessible antecedent for it (see e.g. Ariel 1990; Chafe 1974, 1980; Clancy
1980; Givon 1989; Wong 1990). But as I noted above, both a fare and it are
metonymic references to ‘passenger’. Rather than continue to use metonymy
to evoke this referent, Gary more frequently evokes ‘passenger’ as he (b,c) him
(g), the guy he (q) and ‘zero’ (r). As noted earlier, this variation raises two
additional issues: the switch between it and he, and the difference in explicit¬
ness among he, the guy he, and ‘zero’.
Consider the he/it alternation first:

(a) I picked up a fare at Broad and Elm.

(b) He was a colored guy.
(c) He flagged me down.
(d) So I wasn’t- it was late at night.
(e) It was around one thirty in the morning.
(f) And I was gonna turn it down.
(g) So I figured, y’know, I’d pick him up anyway.

One possible explanation for why Gary uses it in (f) rather than he lies in the
notion of information state noted above and the relevance of different descrip¬
tions for different information states. Immediately prior to (f), Gary had pro¬
vided descriptive background information (late at night (d), around one thirty
in the morning (e)) relevant to his first encounter with ‘passenger’ as a potential
job. This embedded orientation (Labov and Waletsky 1967) suggests a nar¬
rative shift in information state during which Gary defines the relevance of
‘passenger’ not as a person per se, but as a job, i.e. a source of money. Thus,
using it as a next-mention for ‘passenger’ during an embedded orientation
coincides with the textual relevance of this referent as a job-related category.
Although we are illustrating the application of Gricean pragmatics to dis¬
course by focusing on the referring sequences in one particular story, a Gricean
approach would also try to find evidence in sequences from other texts. Note,
then, that Gary uses a metonymic it (‘fare’ stands for ‘passenger’) whenever
the most relevant identifying category is job rather than person related. (6) is
from another conversation:
 Pragmatics 211

(6) (a) Like, this one colored lady I picked up yesterday.

(b) She’s a schoolteacher.
(c) She lived in- well, a nicer part of Germantown.
(d) Over towards Mount Airy.
(e) And we were talkin’.
(f) It was a long fare
(g) so we were talkin’ for a while.
(h) And she told me . . . [continues]

In (6), Gary is telling a story about this one colored lady (a). His next-mentions
alternate between she (b, c, h) and it (f). She evokes the referent in personal
terms: ‘she’ does human actions (she lived (c), she told (h)), ‘she’ has an
identity outside the world of cabdriving (she’s a schoolteacher (b)). Notice that
although it would be infelicitous to use it in any of th%se utterances (e.g. It’s
a schoolteacher (b)), either it or she can be used as the subject of was a long
fare since this defines the referent in relation to a job (as in our earlier
comments on metonymy.) It in Gary’s story in (4) is similar: it is situated in
an earlier point in the story world when the most relevant identifying category
for ‘fare’ is not person, but job.
Another next-mention of ‘passenger’ in Gary’s story is a definite nominal
the guy he (q). What seems to motivate the explicitness of this particular
term is the displacement of this referent from the text (Ariel 1990: chapter 2;
Brown and Yule 1983: 183). Textual displacement can be described in Gricean
terms. After talking about referent 1 (through both explicit and inexplicit
forms), Gary introduces new referents (‘two friends’ (h) and ‘they’ (i to m))
that became the focus of the narrative action (4.3.2, 4.3.3). It is only when
Gary arrives at the destination that ‘passenger’ is again given a more central
role:

(o) So I pulled up,

(p) I turned on the light,
(q) and the guy he grabbed me from the back of my head
(r) and put a gun to it.

As noted above, the entity most relevant to the story prior to the collective
referents (‘two friends’ ‘they’) was ‘passenger’. But once the collective referents
enter the story, the relevance of ‘passenger’ is temporarily suspended. Because
the relevance of this referent can no longer be textually or contextually
assumed, Gary uses a referring term (the guy he (q)) that presents enough
information to bring it back into the story. As noted above, we can thus
explain the effect of displacement on referring terms in Gricean terms: when
an individual referent loses its relevance (its identity) in a collectivity, that
relevance can be reinstated with an explicit next-mention. More generally,
what can compensate for a temporary loss of textual relevance is a highly
informative lexical item.
212 Approaches to Discourse Analysis

Evidence for this explanation is a sequence from another text. In (7), Gary
is describing his family and uses explicit nominals to individuate them.

(7) IVer: (a) Did you have older brothers and sisters and stuff?
GARY: (b) I have an older brother and a younger sister.
(c) Like we’re all two years apart.
(d) So like my brother he’s gonna be twenty six,
(e) and I’ll be twenty four in August.
(f) He’ll be twenty six three days after me
(g) and my sister, she just turned twenty one.

Gary introduces his siblings in (b) with indefinites; more specifically, in exist¬
ential predicates in which the siblings are related to him (i.e. I have; see
chapters 7 and 8). A collective referent we follows in (c). Then, when Gary
returns to his description of the individual members of the group, he initially
differentiates them using my brother he (d) and my sister, she (g) rather than
just pronouns. Thus, again, when an individual referent loses its relevance in
a collectivity, that relevance can be reinstated with an explicit next-mention.
(Another example like this is the guy up front he (4.3.3).)
Finally, it is important to note that Gary also uses a zero pronoun for
‘passenger’ in (r):

(q) and the guy he grabbed me from the back of my head

(r) and put a gun to it.

Such uses seem to be syntactically and textually constrained, e.g. they occur
with verbs that are conjoined or in clauses with close textual antecedents (e.g.
the use of ‘zero’ for ‘other cab drivers’ in line 7 in this story (4.3.5)). Zero
pronouns, however, can also be explained in Gricean terms consistent with
those proposed for the guy he. Above, I suggested that a temporary suspension
of relevance in the text can lead to a greater amount of information in the
referring term. Here I am suggesting the opposite: maximal relevance in prior
text can lead to minimal information in a next-mention slot (cf. Ariel 1990:
chapter 6).
We have concentrated in this section on the referent ‘passenger’. We saw
that a fare is a referring expression that is metonymically related to the entire
referent through a relationship of “object used” to “user.” We also saw that
referent 1 is made relevant to the speaker and his topics of talk within the
same clause in which it is introduced. We explained the alternation between
mentions of ‘passenger’ as it and he in terms of relevance and quantity. We
suggested that explicit next-mentions (the guy he) can compensate for sus¬
pended relevance and that inexplicit next-mentions (‘zero’) are allowed under
conditions of maximum relevance. On a procedural note, we applied a set of
general principles to some specific utterances: we used those principles to try
to explain the selection of different options within and across those utterances.
 Pragmatics 213

The principles not only explained what occurred in a particular sequential slot
in a text (i.e. first-mention, next-mention), they also drew upon prior text as
a source of information (the quantity maxim) to which referents in the story
are relevant (the relation maxim). We also provided some other examples that
supported the explanations being proposed. Although slightly different prob¬
lems arise with the next referents to be discussed, we can use the same general
procedure to address these referents.

4.2.2 Referent 2: ‘two friends’

The first (and only) mention of referent 2 is as two friends. The main import¬
ance of this referent in the story is to be part of the later ‘they’, a collective
referent comprised of ‘passenger’ and ‘two friends’ (t*it see section 4.2.3).
Like the initial reference to ‘passenger’, ‘two friends’ is introduced in relation
to an entity assumed to be more familiar and already accessible. This is
because the lexical item friend is inherently tied to another entity (i.e. friends
are friends “of” somebody), in this case to the textually evoked ‘passenger’.
Thus, again, the speaker presents in his text a certain amount of information
that allows a new referent to be heard as relevant to already familiar informa¬
tion, in this case to a prior character in the story.

4.2.3 Referent 3: ‘they’

Referent 3 is the most problematic in the story, and it raises several difficult
issues that concern not just reference, but pragmatics in general, e.g. whether
hearers are able to recognize speakers’ referential intentions, the potential
vagueness and shifting meaning of pronouns, and the importance of discourse
structure for pronoun interpretation. The basic problem with referent 3 is that
it is so inexplicit that we cannot be sure to whom they is intended to refer.
This inexplicitness also makes it difficult to know whether they is a first-men¬
tion of a new referent or a next-mention of a prior referent - both possibilities
to be considered below.
One possibility is to consider ‘they’ broadly as ‘passenger’ and ‘two friends’.
If we do so, then they in (i):

(i) So they said, “We’re only going right around the corner more or less,”
y’know. “Could you take us up?”

is a first-mention, just as they in (j), (k), (m), (u), (v), (w), (z), (jj), (kk), and
(ww), and them in (u), (x), (aa), and (bb), are next-mentions. Although so
inexplicit a first-mention as they might seem somewhat unusual, we could
argue that more explicit information about who comprises the group has
actually already been given and that they is a containing inferrable (Prince
214 Approaches to Discourse Analysis
1981). Indeed, as a previous example showed, referents may be introduced
individually and then collectively: after mentioning ‘an older brother’ and ‘a
younger sister’ in (7), Gary added yet a third (implicit) referent (himself) to
the entire sibling group (we). Thus, given explicit introduction of its separate
members, a first-mention of a collectivity can be an inexplicit pronoun. In this
sense, we may see they in Gary’s story as evoking a referent that is anchored
in the prior text. Indeed, we could even use the maxims of relation and
quantity to explain the tendency for they to be used to include as many
relevant individual referents as are informationally available.
An obvious alternative to this broad interpretation of ‘they is that they is
co-referential with just two friends. Not only would this follow a common
pattern (first-mention is indefinite and explicit; next-mention is definite and
inexplicit) but it also adheres to the well-established tendency for pronouns
to be co-referential with the most recent semantically compatible nominal.
Note that two friends immediately precedes the first occurrence of they:

(h) But then two friends came out.

(i) So they said /continues/

A problem with the narrow interpretation of they as just ‘two friends’, however,
is that it conflicts with our knowledge about Gary’s experience, i.e. we know
that there were three people in Gary’s cab. Why would Gary talk about only
two of these people - especially given the role that all three play in the robbery?
Gricean pragmatics provides still another direction from which to approach
the referent of they. Rather than puzzle over which interpretation of they is
what the speaker intended, we may ask about our inability as hearers to settle
upon the exact referents evoked by they. Although we might take this inability
as incompatible with the Gricean view of communication in which the hearer’s
goal is to recognize speaker intentions, we might also view the speaker’s use
of vagueness as intentional and as strategic to his purposes. Furthermore, we
can express this in terms completely consistent with the Gricean view of com¬
munication and implicatures: if the use of they violates the maxim of quantity
because it is informationally insufficient for hearer recognition of a referent, we
need to nevertheless assume the constant underlying operation of the cooperative
principle and try to discover why the quantity maxim has been violated.
One reason for the violation of the quantity maxim might be to allow they
to have a membership - to evoke referents - that fluctuates according to its
relevance to the story. This would allow us to see the vagueness of they not
as a communicative failure, but as a communicative resource (cf. the CA
treatment of seeming violations: Schegloff 1972a, and chapter 7 in this book).
With this in mind, let us examine the different uses of they in the story.
Observe, first, that none of the actions predicated of they as a collectivity can
possibly be actions that were performed by the specific group as a collectivity
with no internal differentiation among members. All the verbal actions - the
constructed dialogue (Tannen 1989: chapter 4) - are presented as group actions:
 Pragmatics 215

(i) So they said, “We’re only going right around the corner more or less,”
y’know. “Could you take us up?”
(j) They gave me an address.
(k) They said, “We’ll get- show you how to get there.”
(l) So it was maybe only three blocks away,
(m) so they said- I think they gave me an address that’s at the center of the
block

Use of they with constructed dialogue suggests not a “choral” presentation of

speech (Tannen 1989: chapter 4) but, rather, that individual responsibility for
what was said is less important than the presentation of the actual dialogue.
Similarly presented are physical actions in which the group is agent of the
action:
A ft

(V) and they wanted more.

(w) So they took my wallet.
(z) They were gonna, y’know, shoot me.
(jj) And they said- well, they ripped out the radio cabinet,
(kk) they ripped the microphone out of it.
(ww) Maybe they would’ve gotten jumpy.

or the group is recipient of Gary’s actions:

(u) So I gave ’em the nine dollars

(X) I didn’t want to give them the money.
(aa) So I gave them the rest of the money.
(bb) But, I gave ’em the money.

Again, the use of they here suggests that individual responsibility for what was
said and done is less important than the conduct itself.
Despite the numerous collective referring terms, the group is broken apart
at the most pivotal points in the story: the actions that physically threaten
Gary. It is the event reported in (q) that actually initiates the robbery and leads
Gary to realize that he is being robbed:

(q) and the guy he grabbed me from the back of my head

(r) and put a gun to it.

(I have already discussed the guy he (q) as a next mention of passenger

(4.3.1).) It is after Gary gives into the demand for money (u, v) that the
robbery is physically intensified:

(y) So then the guy up front he stuck a gun on my temple, asking for more
money.
216 Approaches to Discourse Analysis

Note that two different people are involved in the events reported in (q) and
(y). They are explicitly differentiated from one another not in terms of their
earlier introduction as referents (e.g. one of the two friends, the fare I origin¬
ally saw), but in physical terms relevant to the pivotal events in the story: the
location of the three men in the cab is relevant to the physical threats felt by
Gary, i.e. whether a gun is put either at the back or the side of his head. (I
discuss the guy up front he in the next section.) Finally, although Gary does
use a pronoun in reporting his response to the demands for money, he also
does so with a more explicit the guy:

(aa) So I gave them the rest of the money.

(bb) But, I gave ’em the money
(ii) And uh, I gave the guy the money

I discuss these alternative next-mentions in the next section.

Thus far I have suggested that explicit next-mentions separate individuals in
the group when those individuals are performing actions particularly relevant
to - even pivotal to - the reported experience. This, in turn, supports the idea
that they is an intentionally inexplicit reference to the passengers in Gary’s
cab. We can phrase this in Gricean terms: the inexplicitness of they is an
intentional violation of the maxim of quantity motivated by the communicat¬
ive goal of tailoring the relevance of actors and actions to the details of a
reported experience. In Gary’s story, they was thus used to portray a group
as an undifferentiated collectivity rather than a set of individuated members.
Before we turn to the next referent, it is important to note that Gary uses
the pronoun they for several referents: it seems to be co-referential with ‘other
cab drivers’ (4.2.5) and it establishes a collective identity for ‘cop’ (4.2.6). I
discuss the way these other uses of they textually intervene with Gary’s
continual reference to the three people in the cab in sections 4.2.5 and 4.2.6.
We have seen in this section that they can be argued to have a narrow
interpretation (‘two friends’), a broad interpretation (‘passenger’ and ‘two
friends’), or a vague interpretation that violates the quantity maxim, but
allows referents to shift according to their relevance for the story. The value
of Gricean pragmatics for this analysis was not necessarily to choose the right
interpretation from among these options, but to explain the principles provid¬
ing an interpretive basis for the various options. Note, again, that our applica¬
tion of the principles depended upon viewing the text as a source of information
to which referents in the story are relevant.

4.2.4 Referent 4: ‘guy up front’

I mentioned ‘guy up front’ in the previous section as a referent who was

individuated from the collective ‘they’. Both first- and next-mentions of this
referent are definite and explicit: the guy up front he and the guy. Before
 Pragmatics 217

discussing this sequence it is important to note why I have inferred that these
are mentions of a single referent. Prior text provides us with information
allowing us to infer that the guy (ii) is a next-mention of ‘the guy up front’.
The action ‘ask for money’ (y) is linked to ‘give money’ (ii) as a request-
compliance pair (chapter 3, p. 3.1.2). Since people usually comply with a
request by directing an action to the person who has issued the request, we
interpret the guy up front he (y) and the guy (ii) as the same referent.
Let us consider the first-mention:

(y) So then the guy up front he stuck a gun on my temple, asking for more
money.

Since the guy up front is understood as part of the previously mentioned

collectivity ‘they’, the initial definiteness is not surprising. In Prince’s (1981)
terms, this first-mention is a contained inferrable, i.e. a member of a previously
mentioned set. We motivated the explicitness of this expression in terms of its
heightened relevance to the experience in the previous section.
The next-mention is part of a reported action that responds to the threat
from ‘the guy up front’:

(ii) And uh, I gave the guy the money.

Since this event is central to the story - it is Gary’s unwilling concession to

the robbery and the resolution to the complicating action - it might not be
surprising to find a definite explicit noun here rather than a pronoun. Further¬
more, this next-mention is also textually distant from the first-mention.
‘Giving money’, however, is an action that has already been reported twice
in the story with them. Although I discussed these mentions of they in section
4.2.3, it is important to note them again during our discussion of ‘guy up
front’ — simply because they co-occur with what seems to be the same event.

(aa) So I gave them the rest of the money.

(bb) But, I gave ’em the money
(ii) And uh, I gave the guy the money

The questions that we need to address are: (1) Why does Gary switch between
them (and its variant ’em) and the guy in different reports of the same event?
(2) How can Gricean pragmatics explain those alternations?
Earlier we saw that quantity violations may serve the goal of relevance
(4.3.1). By checking the sections of text in which the event ‘give money’ is
reported with different referents, we may be able to discover whether the guy
(ii) and/or them (aa, bb) are quantity violations motivated by relevance.
Procedurally, then, we will again be relying upon text as a source of informa¬
tion and relevance.
The event ‘give money’ is first reported in (aa):
218 Approaches to Discourse Analysis

(y) So then the guy up front he stuck a gun on my temple, asking for more
money.
(z) They were gonna, y’know, shoot me.
(aa) So I gave them the rest of the money.

Gary gives the money to them (aa), not to the guy. These two referents convey
links with two different antecedent events: whereas them links ‘give money’
with the just mentioned group threat (They were gonna, y’know, shoot me
(z)), the guy would have linked ‘give money’ with the prior event (the guy up
front he stuck a gun on my temple, asking for more money (y)). Thus, they in
(aa) helps portray the event ‘give money’ as an outcome of Gary’s evaluative
inference of what would have happened to him (he would be shot (z)), not as
an outcome of what was done to him (y). In this sense, we might say that them
is an informationally adequate expression that serves a goal of relevance
because it establishes a relationship with a prior event.
The next report of ‘give money’ follows the interchange between Gary and
the IVer; it returns to the story frame itself:

(bb) But, I gave ’em the money

It is surprising that ’em is used here. ’Em seems to violate what scholars (e.g.
Givon 1992) think of as distance and clarity constraints: not only was the last
mention of this referent eight clauses earlier, but since those other clauses used
they for a different referent (‘other cab drivers’ (4.3.5)), the use of ’em in (bb)
is potentially ambiguous. Thus, ’em seems to be a quantity violation because
it gives too little information. Although we suggested that quantity violations
giving too much information were a way of lexically compensating for a
suspension of textual relevance, it is unclear whether too little information
can also serve relevance. Note, however, that we might interpret But, I gave
’em the money (bb) as a way of returning the discourse to a non-adjacent
focus. But, I gave ’em the money (bb) paraphrases the event So I gave them
the rest of the money (aa), which was the last story event to be mentioned -
the most recent story event before the evaluative digression with the IVer.
Furthermore, this paraphrase is initiated with But, which establishes a global
link in the discourse (Schiffrin 1987a). Thus, the repetition of the same refer¬
ring term — despite its informational shortcomings — might also figure in the
return to the story frame.
It is in the third report of ‘give money’ that Gary uses the guy. What precedes
this is an evaluation of the experience:

GARY: (cc) Like I figured-

(dd) I didn’t really get upset.
(ee) I wasn’t scared.
(ff) I wasn’t scared till after it was all over with,
(gg) At the time I don’t think-it never even-
 Pragmatics 219

(hh) all I thought of was Diane and the baby.

IVer: Su:re.
Gary: (ii) And uh, I gave the guy the money

Note that (ii) continues the narrative action initiated in (y):

(y) So then the guy up front he stuck a gun on my temple, asking for more
money.
[evaluative digression from narrative sequence]
(ii) And uh, I gave the guy the money

As I indicate in the excerpt, the two events in (y) and (ii) are interrupted by
an exchange with the IVer (which functions as external evaluation: Labov
1972b; see chapter 8) and by Gary’s own report of his feelings (also evaluat¬
ive). Despite this interruption, we understand these two actions as sequent¬
ially tied together as request and compliance: (y) So then the guy up front he
stuck a gun on my temple, asking for more money reports a demand from a
particular person for money; (ii) And uh, I gave the guy the money reports
compliance with that particular demand. Thus, despite the interruption of this
sequence, we understand these two actions as sequentially tied together in an
act sequence (cf. chapter 3). Note, also, that Gary continues his story after
this event: I gave the guy the money is the first event to return to the
temporally sequential events making up the complicating action of the story
(ii to pp). Thus, we may explain the explicit reference to the guy in (ii) as a
way of making apparent a non-adjacent sequential tie (cf. Fox 1987) that
reinstates a prior textual focus. Note that this is similar to our suggestion that
repetition of the pronoun (from aa to bb) established sequential relevance:
quantity is linked to the need to show relevance. But because we said that the
guy was overinformative (and ’em was underinformative) it is only the guy
that illustrates the trade-off between textual relevance and lexical information
noted earlier: an informative referring term can compensate for a temporary
loss of textual relevance.

4.2.5 Referent 5: ‘other cab drivers’

‘Other cab drivers’ follows a typical pattern of indefinite/definite and ex¬

plicit/inexplicit mentions. Like other indefinite first-mentions discussed (4.2.1,
4.2.2), an introduction is made in relation to the speaker himself:

(6) But, from what other cab drivers told me

Other cah drivers includes the speaker as a member of the group cab drivers
(i.e. he could not have felicitously said other truck drivers). Nor would the use
of generic cab drivers be related to the speaker in the same way. In addition.
220 Approaches to Discourse Analysis

‘other cab drivers’ is introduced in terms of their actions toward the speaker,
i.e. they told me.
During our discussion of ‘they’ (and the use of they as next-mentions) in
section 4.2.3,1 mentioned that Gary also uses they as a next-mention for ‘other
cab drivers’. Here I will return to a question that was implicit in that section:
can a speaker use they to evoke different referents without causing confusion
about which entity is being mentioned? This question is important because the
third person pronouns {they, ’em) in the adjacent clauses (10, bb) could evoke
two different referents, i.e. ‘cab drivers’ and ‘they’ (the robbers):

IVer: (3) If you hadn’t pulled out the money in your sock, do you think
they would’ve looked?
(4) Do they know you keep it here?
GARY: (5) No.
(6) But, from what other cab drivers told me,
(7) said I probably would’ve been shot.
IVer: (8) Oh, Gary.
GARY: (9) That’s what they tell me.
(10) Maybe they’re doin’ it to scare me.
(bb) But, I gave ’em the money.

Although it is clear that ’em (bb) is a next-mention of the robbers, they in

Maybe they’re doin’ it to scare me (line 10) is less easily interpreted. By
labelling this line with a number rather than a letter (p. 204), I am obviously
implying that they is a next-mention of a referent not involved in the story
events, i.e. ‘other cab drivers’. But they in line 10 is problematic because it
has another possible antecedent: referent 3, the ‘they’ who robbed Gary. This
interpretation would require not only a more distant antecedent for they (prior
to the introduction of ‘other cab drivers’), but also a different interpretation
for doin’ and it, i.e. as the threatening actions in the story.
In the Gricean analysis illustrated thus far in this section, we have viewed
text as a source of information in relation to which referring expressions
establish (or convey) relevance. To solve the problem of the seemingly am¬
biguous ‘they’ in line 10, then, we can examine the section of text in which
they is used to see whether the quantity and relevance of information presented
in that text leads to an interpretation of they as ‘other cab drivers’.
Let us take a closer look at the text in which Maybe they’re doin’ it to scare
me is embedded. The IVer’s question (Do they know you keep it here? (line
4)) asks about what the robbers (‘they’) knew about where Gary kept his
money. It is during Gary’s answer to this question that Gary introduces ‘other
cab drivers’ and then reports what they tell him: But, from what other cab
drivers told me, said I probably would’ve been shot (lines 6, 7). Notice that
Gary’s initial mention of ‘other cab drivers’ (6) is clearly demarcated from the
story per se: he opens the segment in which they are introduced with but
followed by a pause. Despite the temporary nature of this referent, its relev-
 Pragmatics 221

ance to Gary’s interchange with the IVer is high enough to allow a zero subject
in line 7 (cf. section 4.2.1).
Although they in Maybe they’re doin’ it to scare me (line 10) seems ambigu¬
ous, the text itself suggests that they is a next-mention of ‘other cab drivers’.
When people contrast what someone says with what he “really means,” they
may do so by stressing the verb say, e.g. “He said he was sorry but he didn’t
really mean it.” This is exactly what we see in lines 9 and 10:

gary: (9) That’s what they tell me.

(10) Maybe they’re doin’ it to scare me.

The contrastive stress on tell in line 9 suggests a contradiction between what

the cab drivers say and what they “mean” (their intentions). The continuation
of the present tense from line 9 to line 10 also suggest^ that these actions are
contrastively paired.
We noted earlier that the digression about ‘other cab drivers’ was opened
by But,:

(6) But, from what other cab drivers told me,

Gary returns to the story frame by again using But, this time with a repetition
of the last story event from (aa):

(aa) So I gave them the rest of the money.

(bb) But, I gave ’em the money.

The fact that Gary pauses after But in both line 6 and (bb) is important. But,
marks a global (non-adjacent) contrast, rather than a local contrast with a
proposition in a just-prior clause (Schiffrin 1987a). More specifically, But, in
line 6 opens a digression from the story frame, and But, in (bb) closes that
digression from the story frame. But, also brackets referents inside and outside
of the story world: they in Maybe they’re doin’ it to scare me is outside of the
story world (i.e. ‘other cab drivers’) and ’em in But, I gave ’em the money is
clearly a next-mention within the story world.
I have argued thus far that a referring term that is potentially ambiguous -
they in line 10 - is a next-mention of ‘other cab drivers’. But I also suggested
earlier that the very next use of they is a next-mention of a different referent
(referent 3, the robbers). The reason they can be next-mentions of two differ¬
ent referents is that the speaker provides clear discourse boundaries that
separate the story being told (the story world) from the interchange with the
IVer (the conversational world). These same boundaries separate the two uses
of they, such that one use of they (relevant to the conversational world) can
be a next-mention of ‘other cab drivers’ (referent 5), and the other use of they
(relevant to the story world) can be a next-mention of ‘they’ (referent 3, the
robbers). Thus, the explicit demarcation of the conversational from the story
222 Approaches to Discourse Analysis

world allows Gary to evoke referents in both worlds without using explicit
referring terms to do so. In Gricean terms, what happens is a kind of trade-off
between information provided in a text and information provided in a refer¬
ring expression itself: when discourse provides information sufficient to the
identification of a specific world in which an entity can be found relevant,
then discourse is a source of informativeness that can offset the inexplicitness
(and potential ambiguity) of a referring expression.

4.2.6 Referents 6 and 6a: ‘a cop’ and ‘cops'

The final referent is ‘a cop’, introduced (again) in relation to the speaker:

(mm) and I found a cop right around the corner.

It is the speaker who found a cop, and furthermore did so in a location defined
in relation to his own physical location (right around the corner rather than,
e.g. at Broad and Lee). Next-mentions of ‘a cop’ are definite: first, a nominal
(the cop) and then a less explicit pronoun (he).
‘Cops’ are also evoked collectively as they (qq). Note that this use of they
precedes the use of they as a next-mention of referent 3, the robbers (ww).
But we see here, again, that discourse boundaries can help to separate these
potentially ambiguous referring terms from one another. In this case, however,
the basis for the discourse boundary between the two successive uses of the
referring term they (in qq and ww below) is temporal.

GARY: (nn) And uh, then when-

(oo) y’know, I told the cop what happened.
(PP) He said, “All right, get in your car and go back onto Broad
Street. I’ll meet you up there.”
(qq) Cause they were trying to pick people up for identification.
(rr) That’s when it hit me.
(ss) My leg started jumpin’ up and down.
(tt) I just couldn’t control my nerves.
IVER: [Jesus.
GARY: (uu) [But then the whole time I was-
I- I didn’t even get scared or upset.
(w) And I- I- maybe I thought subconsciously if I would’ve been
scared or shown that I’d have been scared, then I probably
would’ve been hurt.
(ww) Maybe they would’ve gotten jumpy.

Gary concludes the events of his narrative by reporting his encounter with
the cop (oo, pp) and the motivation underlying their action (qq). Gary then
evaluates his experience by tying it to the time of that encounter (that’s when
 Pragmatics 223

it hit me (rr)) and comparing his reaction after the actual robbery (rr to tt) to
the way he felt during the robbery itself (uu to xx). Note how Gary clearly
differentiates the two time periods: the period during which Gary reports the
crime begins with that’s when it hit me (rr); the return to the period of the
robbery itself is noted with but then the whole time (uu). So, again, we see
that by clearly separating different worlds - here, different temporal worlds -
Gary can use discourse information to separate the different referents of two
inexplicit referring terms: they evokes ‘cops’ in one time period, they evokes
‘robbers’ in another time period. Put another way, discourse provides enough
information to allow Gary to evoke two different referents with a single
inexplicit referring term.

4.3 Summary: referring sequences, relevanc#, and

quantity

I have been suggesting in this section that one way that Gricean pragmatics
applies to discourse analysis is by providing a description of the pragmatic
conditions during which different referring terms are interpreted. The maxim
of quantity helps to guide H toward information that can provide clues about
the identity of a referent. The maxim of relation leads H to search for the
relevance of a particular referent. Although the maxims do similar work for
both first-mentions and next-mentions, the sequential location of a referent
(first-mention, next-mention) influences the source of information in relation
to which it is interpreted. A first-mention is interpreted relative to textually
and contextually provided background assumptions about shared knowledge:
first-mentions are relevant to information based in H’s knowledge of context
(including general background knowledge about other entities in the world).
Next-mentions draw upon an additional source of information, i.e. informa¬
tion that has been textually grounded (including, but not limited to, information
available through a first-mention). Next-mentions are thus relevant to informa¬
tion based in H’s knowledge of both text and context.
We can gain more insight into the contribution of Gricean pragmatics to
discourse analysis by going over how the referring sequences in Gary’s story
in (4) are sensitive to the maxims of quantity and relation. Although I will
state some of our findings in relatively general terms, it should be kept in mind
that these suggestions are based on a very small amount of data, and that they
are meant more to illustrate the way Gricean pragmatics would approach the
analysis of referring sequences than as a definitive Gricean analysis of refer¬
ence.
Let us consider first-mentions first. Referents were introduced in ways rele¬
vant to surrounding information. Initial definites (e.g. they, the guy up front)
were relevant to previously presented (or assumable) information, i.e. they were
used when the speaker could assume a hearer s ability to use text (or context)
to infer the existence and relevance of a specific referent. Initial indefinites
224 Approaches to Discourse Analysis

were a more varied group: although most of these (a fare, other cab drivers,
a cop) were relevant to information presented in the same clause, one initial
indefinite was relevant to prior textual information.
Next-mentions were likely to be definite and less explicit than first-mentions.
This suggests - in line with virtually all other work on referring terms - that
the pragmatic conditions underlying next-mentions differ from those under¬
lying first-mentions. But Gricean pragmatics also provides a description of
next-mentions similar to that for first-mentions. Thus, we can say that sub¬
sequent definites occur under pragmatic conditions very similar to those for
initial definites, i.e. when a referent can be interpreted as relevant to already
accessible information. The only difference, of course, is that the information
to which a referent is relevant has already been textually encoded, i.e. it is the
prior mention of the referent in the text itself. Prior textual information also
allows next-mentions to be inexplicit. In brief, textual encoding of a referent
creates background relevant to a subsequent definite referent and provides the
information allowing subsequent reference to be inexplicit.
The next-mentions in Gary’s story also suggest that quantity and relevance
can act together to define the pragmatic conditions underlying the interpreta¬
tion of different referring terms.10 We found one example of quantity and
relevance working together in a relatively straightforward way: the alternation
between he and it as next-mentions for ‘passenger’ (4.2.1). He and it convey
different information: it relies on a metonymy between ‘passenger’ and ‘fare’,
he and it provide different information about gender and animacy. However,
both he and it follow the quantity maxim. Their difference lies in the way they
follow that maxim: they select information according to which identificatory
category (person versus job) is relevant to a particular event in the story. In
brief, we might say that these inexplicit forms alternated according to different
conceptions of what was relevant.
Other referring sequences in Gary’s story suggested that quantity viol¬
ations can be explained in terms of the maxim of relation. Recall our discus¬
sion of explicit next-mentions. We explained the use of the guy he (q) for
‘passenger’ and the guy (ii) for ‘guy up front’ in terms of a trade-off between
lexical informativeness and suspended relevance. Explicit next-mentions can
thus be viewed as a violation of the quantity maxim designed to show relev¬
ance.
Another example suggesting that quantity violations are related to relevance
was the use of they for three different referents (‘passenger’ and ‘two friends’,
‘other cab drivers’, ‘cops’) without reinstatement of the referent through a
more explicit term. These adjacent uses of they with different referential intent
also violate the maxim of quantity although in a different direction: not
enough explicit information is presented in the referring term itself to allow
the differentiation of referents. Despite this violation, we said that they was
interpretable as three different referents, simply because the adjacent uses were
clearly demarcated from one another by structural markers (discourse mar¬
kers, temporal markers) that defined different subunits of the text. Thus,
 Pragmatics 225

what separated these inexplicit and potentially ambiguous forms from one
another was the clear presence of discourse boundaries that defined different
frames of relevance for adjacent occurrences of the inexplicit form they. Put
another way, what compensated for the lack of lexical informativeness was
an increase in explicitness at the discourse level, specifically the marking of
different discourse worlds to which each referent of they could be found
relevant.
A final example of how quantity violations were motivated by the need for
relevance is the way they served as first- and next-mentions of a collective
referent (‘passenger’ and ‘two friends’ (4.2.3)). We saw that it was not always
clear to whom the inexplicit pronoun they was intended to refer. One way to
interpret this vagueness is as an intentional violation of the maxim of quantity
- a violation allowing the speaker to allow different members of a collectivity
to be relevant to different portions of the nar^tive action.
' Although they as a collective referent raised the issue of ambiguity most
clearly, there were also other expressions about whose referents we were
somewhat uncertain. Recall that I identified both the guy he (q) and ‘zero’ as
next-mentions of ‘passenger’ (4.2.1):

(q) and the guy he [?1] grabbed me from the back of my head
(r) and [? 1] put a gun to it.

However, it is also possible that the guy he and ‘zero’ are not meant to evoke
‘passenger’ at all — but one of the two friends (4.2.2) who had just been
mentioned. Although it would be useful to be able to use Gricean pragmatics
to resolve referential ambiguity, the goal of this sample analysis was more
limited: I have tried to show how Gricean principles provide an interpretive
basis for the various referential possibilities that can be inferred.
Consider, however, another way of viewing referential ambiguity in a Gri¬
cean framework. Recall that the quantity maxim stated the following:

1 Make your contribution as informative as is required (for the current

purposes of the exchange).
2 Do not make your contribution more informative than is required.

Rather than view unclear cases like the guy he, and ambiguous expressions
like they, as violations of the quantity maxim, we might view them as informa¬
tionally appropriate to the purposes of the exchange. Consider, for example,
if Gary had been telling his story to the cops he finds. The purposes of an
exchange with the police would be quite different: rather than describing a
dangerous experience, Gary would be reporting a crime (with possible legal
ramifications) and would thus need to be specific about who did what. Informa¬
tion about exactly who put a gun to his head (r) and who stuck a gun to his
temple (y) would be critical to that purpose. What I am suggesting, then, is
that the referential ambiguity in Gary’s story need not be seen as a violation
226 Approaches to Discourse Analysis

of a maxim at all: the vagueness might be perfectly well suited to the purposes
of the story.
This raises a broader question about the application of Gricean pragma¬
tics to language use. We noted earlier (3.4) that linguists using a Gricean
perspective often analyze utterances that are hypothetical sentences in hypo¬
thetical contexts. The use of actual utterances, however, forces us to attend
to the actual situations in which language is used - situations that include
speakers and hearers, whose needs, goals, and wants are tailored to a particu¬
lar socially and culturally defined communicative situation (see chapters 4 and
5). Just as cross-cultural analyses of Gricean pragmatics suggest that the
Gricean maxim of quantity may be sensitive to culturally specific (and vari¬
able) ideas of how much information is appropriate (Keenan 1979), so ana¬
lyses of different communicative situations within one culture might suggest
that how much information we provide is tailored to our socially based
conceptions of what our interlocutors need for the purposes of our social
exchange. To this end, it is important to note that all the uncertain cases
that we discussed were clustered within the complicating action of Gary’s
story: it is in this section that establishing the precise identity of who did what
is less critical to the point of the story than establishing the danger of the
experience.
To conclude: the maxims are general principles of communication; their
application to specific conversational or linguistic phenomena is an outcome
of the way they are used by a hearer to infer speaker meaning. When the
maxims are applied to a referential problem (i.e. what inferences can a hearer
draw about the speaker’s referential intention upon hearing a particular refer¬
ring term?), the work they do is very general. The quantity maxim leads
hearers to search for information, i.e. the amount of information in a text, the
amount of information in a description itself. The maxim of relation leads
hearers to use information in a certain way, i.e. to find its relevance to the
rest of the text and to the context in which it is situated. The maxims of
relation and quantity do not themselves provide a hearer with the identity of
a speaker’s intended referent; rather, they lead a hearer toward information
and provide a means of using that information to help hearers infer a speaker’s
referential intention. It is these pragmatically based choices that help create
sequential patterns of referring terms and are thus partially responsible for
different discourse structures.

5 Gricean pragmatics as an approach to discourse

In this chapter, we described Grice’s ideas about speaker meaning and the
cooperative principle (section 2) and then applied these ideas to a particular
 Pragmatics 227

problem: we used the CP (specifically, the maxims of quantity and relevance)

to describe the conditions under which people use different expressions to
communicate referential intentions in discourse (section 4). We concluded by
saying that referring sequences are the outcome of pragmatically based choices
concerning the provision of appropriate quantities of information in relevant
ways, and thus that discourse structures are created (in part) by the cooperat¬
ive principle.
We saw that Gricean ideas about information quantity and relevance (cf.
Horn 1985b; Levinson 1983, 1987; Sperber and Wilson 1986) can help solve
problems of utterance interpretation that hinge on assessing the contribu¬
tion of various kinds of contexts to such interpretations. The application of
the CP to discourse leads to a particular view of discourse and its analysis:
discourse as a text whose contexts (including cognitive, social, and linguistic
contexts) allow the interpretation of speaker ■'fneaniiflg in utterances. What
Gricean pragmatics thus provides is a set of principles that constrains speakers’
sequential choices in a text and allows hearers to recognize speakers’ inten¬
tions by helping to relate what speakers “say” (in an utterance) to its text and
contexts.
Although I more fully discuss the assumptions of a Gricean approach in
relation to several key issues in discourse analysis in chapters 9, 10, and 11,
it is important to briefly foreshadow those points here. The approach that
Gricean pragmatics offers to discourse analysis is based in a set of general
principles about rationally based communicative conduct (the CP) that tells
speakers and hearers how to organize and use information offered in a text,
along with background knowledge of the world (including knowledge of the
immediate social context), to convey (and understand) more than what is said
- in brief, to communicate. The operation of these principles leads to a
particular view of discourse structure in which sequential dependencies -
constraints imposed by one part of a discourse on what occurs next - arise
because of the impact of general communicative principles on the linguistic
realization of speaker meaning at different points in time. Our analysis of
referring sequences suggested, for example, that the textual and contextual
information provided in an initial position [slot 1] in a discourse serves as a
background by which to judge how much information is appropriate in a next
position [slot 2], as well as the relevance of that information for [slot 2]. The
operation of the cooperative principle in one part of a discourse thus helps to
define options in another: it is this functionally based interdependency that
helps to create the sequential regularities characteristic of discourse and that
allows people to use both text and context as a resource by which to com¬
municate with each other. Thus, what Gricean pragmatics offers to discourse
analysis is a view of how participant assumptions about what comprises a
cooperative context for communication (a context that includes knowledge,
text, and situation) contribute to meaning, and how those assumptions help
to create sequential patterns in talk.
228 Approaches to Discourse Analysis

Exercises

1 Although Gricean pragmatics highlights the cooperative assumptions that

we bring to conversations, not all verbal interchanges seem cooperative (e.g.
Grimshaw 1990). The sample data in appendix 3, for example, contain some
exchanges that seem overtly hostile, and others that seem more competitive
than cooperative. After identifying what seem like the non-cooperative parts of
the interchange, describe how Gricean pragmatics could account for these
sections.
2 Many scholars have supplemented or altered Grice's cooperative prin¬
ciple. Examples of neo-Gricean work can be found in Horn (1985a, b), Lakoff
(1973), Leech (1983), Martinich (1984), Recanati (1987), and Sperber and Wilson
(1986); see also the special issue of Brain and Behavior 1989. After reading some
of the works just noted, apply both Grice's framework and those suggested
above to the following data (or other data of your own choice).

(1) Professors A and B are colleagues at the same university. A is on leave in

another city. B is in her university office, preparing for class. A calls Pro¬
fessor B on the phone: B tells A that she has a few minutes to talk. After
spending a few minutes discussing an upcoming meeting, the following
occurs:
prof a: (a) Would you like to call me back?
prof B: (b) No that's okay, I don't have that much more to say.
prof A: (c) No, I meant use university money instead of mine!
(d) I have some things 1 have to ask you.
(2) A and B are commuters on a daily train. A notices a newspaper on an
empty seat next to where B is sitting.
a: (a) Is that your paper?
B: (b) Yeh, but you can have it.
[or] No, you can have it.

3 We often infer that two propositions are related to one another even when
that relationship is not explicitly stated (e.g. Halliday and Hasan 1976; Thomp¬
son and Mann 1987). In the following pairs of sentences, for example, we are
likely to infer some relationship between the two propositions even though
none is explicitly stated.

(a) I stopped at the bank on the way home from work.

I withdrew $200.
(b) It rained last night.
The rabbit got wet.
(c) I went into the kitchen.
The cookies burned.
(d) Jay was in the hospital last week.
He needed emergency surgery.
(e) My father fell on the ice.
He broke his leg.
 Pragmatics 229

(f) Our neighbors must be on vacation.

Their newspapers are piling up.

After identifying the possible relationships between the propositions in the

above sentence pairs, use the Gricean maxims to account for the inferred
relationships. How did you decide between the relevance of one maxim rather
than another? Suppose you were to add the connectives and, then, so, because
to each of the above examples. How could you integrate the Gricean analysis
of inferred relationships into descriptions of the meanings of these different
connectives? How do the inferences supplied through the use of the connect¬
ives differ from those available without connectives?
4 A shampoo that is intended to remove head lice has the following short
text in large, bold letters on the bottle:

A Single Application a
• Kills Lice & Their Eggs
• Prevents Reinfestation

Given this amount of information, what interpretation(s) could a consumer

draw about the amount of shampoo to apply and the frequency of use?
The box containing the shampoo repeats the same text, again in large bold
letters, on the front; the text is centered on the box. The box front also notes,
in smaller letters at the bottom, that it is a Family Pack with two bottles. What
inference about amount and frequency could this information evoke?
Now consider information presented on each bottle, and on the back of the
box.

A sufficient amount should be applied to saturate hair and scalp.... A single

application is sufficient.

Does this text evoke the same interpretation(s) about the amount of shampoo
to apply and the frequency of use? How can Gricean pragmatics explain the
alternative interpretations? How could this product labelling be clearer?

Notes
1 What Levinson terms the “continental” view of pragmatics is more similar to what
Americans call “sociolinguistics” and “discourse analysis.” Research in this vein is
represented in the journals Pragmatics and Journal of Pragmatics.
2 The central theoretical role played by “violations” stems from the level of generality
at which the principles are defined. Compare the discussions of interactional
strategies in chapter 4 and local rules in chapter 7.
3 Levinson (1983: 147-62) discusses alternative theories of metaphor. See also Lakoff
and Johnson (1980) and Lakoff (1987), whose dependence on the framework known
as “cognitive linguistics” defines the relationship between semantics and pragmatics
quite differently than does the formal semantics perspective assumed by Grice.
4 Exactly how identifiable an entity must be, however, is not clear. DuBois (1980:
233), for example, counts a reference as identifiable “if it identifies an object close
230 Approaches to Discourse Analysis

enough to satisfy the curiosity of the hearer” (see also Grimes 1975; Kronfeld 1990).
Donnellan (1978) and Klein (1983) also argue that one can refer to entities with a
definite pronoun without knowing their identity.
5 Although our analysis refers to both definites and indefinites, I am using these terms
to refer to categories that are actually much narrower: definites and indefinites that
are both referential and specific. Donnellan (1966: 198) describes the differences
between using expressions attributively and referentially:

A speaker who uses a definite description attributively in an assertion states

something about whoever or whatever is the so-and-so. A speaker who uses a
definite description referentially in an assertion . . . uses the definite description
to enable his audience to pick out whom or what he is talking about and states
something about that person or thing.

(la) and (lb) illustrate the attributive-referential distinction for definites; (2a) and
(2b) for indefinites.

(la) Definite, Attributive

The vandals who burned the books should go to jail. Whoever would do
such a thing should be punished, and I hope that’s what happens when
they eventually find out who did it.
(lb) Definite, Referential
The vandals who burned the books should go to jail. I hope that’s what
the judge decides when he hears their case today.
(2a) Indefinite, Attributive
I never saw a two-headed person.
(2b) Indefinite, Referential
There was a two-headed person at the circus who was the hit of the
whole show.

Both definite and indefinite expressions can be used either specifically or non-
specifically. (3) illustrates the specific-non-specific distinction for definite forms; (4),
for indefinite forms.

(3a) Specific definite

Last night the cat knocked our trash can over.
(3b) Non-specific definite
The cat is an animal that often acts as a scavenger.
(4a) Specific indefinite
Every night a cat knocks our trash can over. It must be hungry.
(4b) Non-specific indefinite
Every night a cat knocks our trash can over. They’ve become real
scavengers lately.

In both (3a) and (4a), the speaker seems to have in mind a specific cat that can be
identified; this assumption does not seem to hold for either (3b) or (4b).
As I stated earlier, I will refer in what follows to definites and indefinites - although
what I mean by this is referential, specific indefinites and definites.
 Pragmatics 231

6 Although space prevents me from a comprehensive review of other analyses of referring

terms (or reference in general), I try to build on the insights of other approaches and
to note some compatible concepts and tools from other perspectives - most notably,
from the information flow perspective (Chafe 1976, 1987; DuBois 1987; Givon 1979,
1983, 1984; Prince 1981). However, fuller discussion of how Gricean pragmatics
compares to other approaches to discourse is reserved for the conclusion of the book.
7 Much of the literature on referring in discourse makes use of narrative as a type of
discourse in which referring must be accomplished (see e.g. Givon 1992). Argu¬
ments for the theory of emergent grammar (Hopper and Thompson 1980, 1984)
also seem to assume that narrative is the basic form in which lexical categories, like
nouns and verbs, and clause relationships are to be found.
8 Not all relevant textual or contextual information, however, allows the use of a
definite noun phrase for initial reference. Gary had been talking about “tough
neighborhoods” in the conversation prior to his story, and he mentioned that much
of his knowledge about the city came from driving a cabko different locations and
picking up people. Rather than pursuing either of these issues, however, he conti¬
nued to talk about different city neighborhoods, and it was this topic that led to
the IVer’s question Have you ever been robbed? Because driving a cab and picking
up people had been mentioned, we might have expected these activities to provide
enough information to allow an initial definite reference (the fare) for “passenger”.
But perhaps because these activities had a subordinate role in relation to Gary’s
main topic (they were mentioned only to explain how he acquired his knowledge
of different neighborhoods), they were not sufficient to warrant the assumptions
needed for an initial reference to the fare. Compare the use of initial definites for
referents 3 (4.2.3) and 4 (4.2.4).
9 Note also that even though Gary knew when he was telling the story (and indeed,
knew when he was grabbed and had a gun put to his head (q)) that his fare was
“a robber,” it is not surprising that he did not introduce his story by saying I picked
up a robber at Broad and Elm. If he had, his hearer might very well have wondered
about why Gary would do such a thing - a question that does not at all come to
mind if he picks up a fare.
10 The relationship between quantity and relevance has been the focus of recent work
in pragmatics. Key to much of the discussion is the second part of Grice’s quantity
maxim - which essentially places an upper limit on quantity. I repeat both parts:
“Make your contribution as informative as is required (for the current purposes of
the exchange). Do not make your contribution more informative than is required.”
Horn (1985b: 12) notes that the second quantity principle is “essentially akin to
Relation (what would make a contribution more informative than required, except
the inclusion of material not strictly relevant to and needed for the matter at
hand?).” Levinson (1988), however, argues that relevance does not really concern
information at all (for him, it has more to do with “timely helpfulness in relation
to interactional goals”). Levinson (1988) suggests instead that the second quantity
principle is a principle of informativeness: from the speaker’s end, this is a maxim of
minimization (“say as little as necessary”); from the recipient’s end, this is a maxim
of maximization (“amplify the informational content of the speaker’s utterance, by
finding the most specific interpretation, up to what you judge to be the speaker’s
m-intended point”). In short, for Horn, the upper limit on information quantity is
akin to relevance (i.e. too much information would not be relevant); for Levinson,
the upper limit pushes the speaker to say less and the hearer to interpret more.
7 Conversation Analysis

1 Introduction

Conversation analysis (CA) offers an approach to discourse that has been

extensively articulated by sociologists, beginning with Harold Garfinkel who
developed the approach known as ethnomethodology (influenced by the phe¬
nomenology of Alfred Schutz), and then applied specifically to conversation,
most notably by Harvey Sacks, Emanuel Schegloff, and Gail Jefferson. Con¬
versation analysis differs from other branches of sociology because rather than
analyzing social order per se, it seeks to discover the methods by which
members of a society produce a sense of social order. Conversation is a source
of much of our sense of social order, e.g. it produces many of the typifications
underlying our notions of social role (Ciccourel 1972). Conversation also
exhibits its own order and manifests its own sense of structure.
CA is like interactional sociolinguistics in its concern with the problem of
social order, and how language both creates and is created by social context.
It is also similar to the ethnography of communication in its concern with
human knowledge (cf. our discussion of communicative competence in chapter
5) and its belief that no detail of conversation (or interaction) can be neglected
a priori as unimportant. All three approaches also focus on detailed analysis
of particular sequences of utterances that have actually occurred. But CA is
also quite different from any of the approaches discussed thus far: CA provides
its own assumptions, its own methodology (including its own terminology),
and its own way of theorizing.
I begin in section 2 with an overview of CA assumptions and methods, as
well as some of its key findings and concepts (see also Goodwin and Heritage
1990; Heritage 1989; Wootton 1989; Zimmerman 1988). Section 3 is a sample
analysis that focuses upon the way a particular construction (that I call “there
+ BE + ITEM”) helps to solve some recurrent problems of talk. Since one of
the problems addressed by this construction has to do with how people
 Conversation Analysis 233

mention and recognize referents (i.e. the ITEM noted above), this sample
analysis also continues and amplifies our concern with referring sequences.
Section 4 summarizes.

2 Defining conversation analysis

Since CA is derived from ethnomethodology, it is important to very briefly

discuss this sociological perspective before discussing CA. Although I cannot
possibly hope to do justice to this field (see Button 1991; Garfinkel 1967;
Heritage 1984a; Turner 1974), I will try to present some key ideas that can
b.e connected to the assumptions, methods, and findings of CA.
Garfinkel’s term “ethnomethodology” was modelled after terms used in
cross-cultural analyses of ways of “doing” and “knowing.” Ethnobotany, for
example, is concerned with culturally specific systems by which people “know
about” (classify, label, etc.) plants. Garfinkel (1974: 16) states that he used
the term “ethno” for the following reason: “ ‘Ethno’ seemed to refer, somehow
or other, to the availability to a member of common-sense knowledge of his
society as common-sense knowledge of the ‘whatever’.” The “whatever” en¬
compassed by ethnomethodology is not a specific body of knowledge about
one domain (e.g. plants). Rather it is the “ordinary arrangement of a set of
located practices” (p. 17). What ethnomethodology is thus concerned with is:
“a member’s knowledge of his ordinary affairs, of his own organized enter¬
prises, where that knowledge is treated by us as part of the same setting that
it also makes orderable.”
The above quotes suggest that uncovering what we know is a central concern
for ethnomethodology. They also suggest that knowledge is neither autonom¬
ous nor decontextualized; rather, knowledge and action are deeply linked and
mutually constitutive. (I suggest below that this has an important bearing on
the study of language.) Much of Garfinkel’s research reveals that participants’
understandings of their circumstances provide for the stable organization of
their social activities. Such understandings, however, are not ready and wait¬
ing apart from human activity: participants continually engage in interpretive
activity - and thus reach understandings - as a way of seeking order and
normalcy during the course of their everyday conduct. Furthermore, the sense
of order that emerges is displayed through ongoing activity that provides a
practical basis, and a sense of intersubjectivity, through which further activity
can be sustained. Social action thus not only displays knowledge, it is also
critical to the creation of knowledge: one’s own actions produce and repro¬
duce the knowledge through which individual conduct and social circumstance
are intelligible.
The link between knowledge and action has an important bearing on the
study of language. Ethnomethodological research avoids idealizations, arguing
234 Approaches to Discourse Analysis

instead that what members produce are “typifications”: categories that are
continuously adjusted according to whether the anticipation of an actor (an
anticipation on which the actor’s own conduct was based) is confirmed by
another’s actions. The anticipations to which actors orient, and to which
actions are held accountable, can be formulated in terms of interactional rules
(or systems), i.e. “a detailed texture of institutionalized methods of talking to
orient to” (Heritage 1984a: 292). These methods provide actors with “cognit¬
ive bearings” that are essential to their ability to “make continuous sense of
their environments of action” (Heritage 1984a: 292).
Language (and action through language) is no less a situated product of rules
and systems than other typifications. Although language is the medium
through which common-sense categories are constituted, the meaning and use
of a particular term (and thus the boundaries of a category) are nevertheless
indeterminate and negotiable. The relationship between words and objects is
as much a matter of the world of social relations and activities in which words
are used, as of the world of objects that is being “named.” Put another way,
the meaning of a particular utterance (including the sense of particular de¬
scriptive terms) is indexical to a specific context and purpose. The contextual-
ization of language in these ways allows its entry into the mutually constitutive
relationship between action and knowledge: speakers produce utterances as¬
suming that hearers can make sense out of them by the same kind of practical
reasoning and methodic contextualizing operations that they apply to social
conduct in general. It is these methods and operations that allow actors to
“proceed on the basis that their experiences are ‘identical for all practical
purposes’ ” (Heritage 1984a: 60). Furthermore, it is because actors succeed in
using the sequential progression of interaction to display their understandings
of its events and rules that the shared world that has been jointly achieved is
publicly available for analysis (Taylor and Cameron 1987: 104).
Although specific CA analyses do not always acknowledge their ethnomethodo-
logical heritage, some of the key ideas in CA can be related to the ideas
briefly noted above. The focus of CA on conversation, for example, arises out
of the ethnomethodological distrust of idealizations as a basis for either social
science or ordinary human action. Sacks (1984) argues that many idealizations
in social science produce general concepts that have only a vague and indeter¬
minate relationship with a specific set of events. Since it can be difficult to
decide whether a specific event supports a generalization, this distance be¬
tween “type” and “token” hinders the development of a cumulative body of
knowledge. (By the same line of argument, CA distrusts linguistic categoriza¬
tions of the functions of particular words or expressions: such categorizations
may be generalizations that do not at all reflect the specific uses of an item;
see section 4 and chapter 10.) Sacks chose to work on conversation in an effort
to remedy the idealizations of sociologists. He wanted to: “handle the details
of actual events, handle them formally, and in the first instance be informative
about them in the direct ways in which primitive societies tend to be informat¬
ive, that is, that anyone else can go and see whether what was said is so”
 Conversation Analysis 235

(Sacks 1984: 26). Tape-recorded conversation thus provided data that could
be available for many analysts and subjected to many analyses. Again, in
Sacks’ words: “I could get my hands on it and I could study it again and again
. . . others could look at what I had studied’ (Sacks 1984: 26).
The goals and beliefs that I have just described continue to influence CA.
CA focuses upon the details of actual events: analysts record conversations
that occur without researcher prompting (cf. our discussion of sociolinguistic
interviews in chapters 5 and 8). Analysts also produce transcriptions of events
that attempt to reproduce what is said (both linguistic details such as pronun¬
ciation and nonlinguistic details such as inbreaths) in ways that avoid presup¬
positions about what might be important for either participants or analysts
themselves (e.g. Jefferson 1989; see appendix 2).1 Similarly, analysts avoid
positing generalizations about what participants “know”: the focus, instead,
is on specific events that occur during the conversation. And in keeping with
the relationship noted above between action and knowledge, the events that
are focused upon are said to reflect and realize practical knowledge.
The CA treatment of context is also ethnomethodologically based. We noted
above that utterances are indexical: this indexicality locates utterances not just
in a world of social relations (more on this in a moment), but in a world of
other utterances. Furthermore, each utterance in a sequence is shaped by a
prior context (at the very least, and most typically, the immediately prior
utterance) and provides a context for a next utterance (again, for a very next
utterance). In Heritage’s (1984a: 242) terms, “the significance of any speaker’s
communicative action is doubly contextual in being both context-shaped and
context-renewing.” This notion of context as being both retrospective and
prospective can be seen as yet another way that meanings (and knowledge)
are continually adjusted and sequentially emergent.
Although CA assumes that utterances always have contextual relevance for
one another, not all aspects of context are assumed to have so constant a
relevance. We noted above that the indexicality of utterances locates them in
a world of social relations. However, CA transcripts of talk pay little attention
to social relations and to what other approaches call “social context,” e.g.
social identities of participants, setting, personal attributes, and so on. By
intentionally ignoring what are often assumed to be static features of a social
world (e.g. the occupation of a participant), CA reflects yet again the ethno-
methodological avoidance of premature generalizations and idealizations. So¬
cial identity (setting, and so on) is viewed instead as a category of social life
and conduct that is subject to locally situated interpretive activity: the relev¬
ance of a social identity can be no more presumed to hold across different
times and places than can the relevance of a one second pause (on the latter,
see Jefferson 1989). As Schegloff (1987a: 219) points out:

the fact that they [social interactants] are “in fact” respectively a doctor and
a patient does not make those characterizations ipso facto relevant . . .
their respective ages, sex, religions, and so on, or altogether idiosyncratic
236 Approaches to Discourse Analysis

and ephemeral attributes (for example, “the one who just tipped over
the glass of water on the table”) may be what is relevant at any point in
the talk.. .. That is, there should be some tie between the context-
as-characterised and its bearing on “the doing of the talk or doing the
interaction”.

As Coulter (1989: 103) says, “it is only within organizations of sayings and
doings that assignably ‘personal’ attributes are made manifest” (see also
Schenkein 1978). Thus, although CA is an approach to discourse that empha¬
sizes context, the relevance of context is grounded in text.
Now that we have discussed some of the basic ideas of ethnomethodology
and their reflection in CA, let us turn to some of the specific methods and
findings of CA. Heritage (1984a: 241) lists three assumptions of CA: (a) inter¬
action is structurally organized; (b) contributions to interaction are contex¬
tually oriented; (c) these two properties inhere in the details of interaction so
that no order of detail can be dismissed, a priori, as disorderly, accidental, or
irrelevant. The latter two assumptions have already been discussed above; the
assumption of structural organization will be discussed in a moment in relation
to adjacency pair organization. Consistent with the CA avoidance of prema¬
ture generalization and the ethnomethodological focus on action as the locus
of knowledge, CA views the empirical conduct of speakers as the central
resource out of which analysis must develop (Heritage 1984: 243). Further¬
more, what is said provides not only the data underlying analysis, but also the
evidence for hypotheses and conclusions: it is participants’ conduct itself that
must provide evidence for the presence of units, existence of patterns, and
formulation of rules. To this end, CA searches for recurrent patterns, distribu¬
tions, and forms of organization in large corpora of talk.
We noted above that the CA view of interaction is a structural view. One
such structure is the adjacency pair: a sequence of two utterances, which are
adjacent, produced by different speakers, ordered as a first part and second
part, and typed, so that a first part requires a particular second part or range
of second parts (Schegloff and Sacks 1973: 295-6; also Schegloff 1972a).
We saw an example of an adjacency pair in chapter 1: a summons-answer
sequence. There we noted that a telephone ring is a summons; a summons
opens a conditional relevance for a second part of a sequence, an answer.
When an answer is not forthcoming, it is heard as officially absent, such that
a summons can be reissued and/or an account offered for its absence. The
summons-answer adjacency pair not only has structure in and of itself, it
also provides for a coordinated entry into, and an orderly exchange of turns
within, the conversation. Conversational closings also depend upon two-part
sequences that provide for coordinated exit from turn exchange (Schegloff and
Sacks 1973). Thus, adjacency pairs are organized patterns of stable, recurrent
actions that provide for, and reflect, order within conversation. Their socio¬
logical importance is that they provide a normative framework for actions that
is accountably implemented (Heritage 1984a: 249). Their linguistic import-
 Conversation Analysis 237

ance is that they provide an environment in which inferences about relevance

can be assigned across utterances (cf. Gricean pragmatics, chapter 6) and a
sequence in which expectations about form and meaning can be specified
across utterances.
Adjacency pairs also reflect the local nature of conversational structure. So
pervasive is the pairwise organization of talk that a next-utterance need not
be explicitly tied to a prior utterance: rather, adjacency itself can provide a
basis for assumed relevance (cf. Gricean pragmatics, chapter 6). Sacks (1971:
lecture 4, 11-12) points out the importance of next-position:

There is one generic place where you need not include information as to
which utterance you’re intending to relate an utterance to .. . and this is
if you are in Next Position to an utterance. Which is to say that for
. adjacently placed utterances, where a next intends^o relate to a last, no
other means than positioning are necessary in order to locate which
utterance you’re intending to deal with.

From a speaker’s point of view, next-position thus offers a location in which

to find the recipient’s analysis of the utterance - to see whether an anticipated
response is confirmed. From a recipient’s point of view, next-position offers an
opportunity to reveal aspects of the understanding of prior talk to which own
talk will be addressed. (Recall the context-shaping and context-renewing fea¬
tures of utterances.) Thus, next-position is a crucial location for the building
of intersubjectivity: each next turn provides an environment in which recipients
can display many understandings, including problematic understandings (i.e.
misunderstandings) that lead to “third position repair” (Schegloff 1992).
Although conversation typically proceeds in a pairwise fashion, there are
also ways that a pair can be expanded, before its initiation (see Schegloff 1980;
Levinson 1983: 345-64, on pre-sequences), after its completion (see Fox 1987:
23-8, on post-elaborations), or even during its creation (see Jefferson 1972,
on side sequences). That such modifications are specifically constructed as
modifications shows the robustness of the adjacency pair organization for
participants themselves (recall the need to find evidence for patterns within
participants’ own talk; but see Tsui 1989). Finally, although the first part of
an adjacency pair defines a slot for the second, more specific preferences
within that slot can be structurally indicated. A preference for agreement
(rather than disagreement) with an assessment, for example, can be indicated
by a lack of structural marking (Pomerantz 1984), i.e. no qualifiers, prefaces,
pauses, or accounts. Consistent with the focus on structure (and the reluctance
to posit psychological states that are not empirically grounded in conduct;
Pomerantz 1990), preferences are seen as structural preferences defined on the
basis of the skewed distribution of forms and behaviors. (But see discussion
in Bilmes 1988; Taylor and Cameron 1987: 109-15).
The pervasiveness of local organization in talk is also revealed in the CA
analysis of turn-taking (Sacks et al. 1974). A central problem underlying talk
238 Approaches to Discourse Analysis
is what might be called a “distribution” problem: how do speakers allocate
turns at talk? How do they know during what period of time one will be
expected to speak and the other obliged to be silent? How does one person
know when to end speaking, and another when to begin speaking, with
minimal gap and minimal overlap between turns? Note that some of these
problems arise from empirical observation of talk, e.g. the minimal gap/over¬
lap problem.
CA offers a solution to these (and other) problems centered on turn exchange
- a solution whose operation is demonstrable in actual talk. The solution is
“a basic set of rules governing turn construction, providing for the allocation
of a next turn to one party, and coordinating transfer so as to minimize gap
and overlap” (Sacks et al. 1974: 12). The rules are locally managed in that
they apply to all possible points of turn exchange:

1 At initial turn-constructional unit’s initial transition-relevance place:

(a) If the turn-so-far is so constructed as to involve the use of a
“current speaker selects next” technique, then the party so se¬
lected has rights, and is obliged, to take next turn to speak, and
no others have such rights or obligations, transfer occurring at
that place.
(b) If the turn-so-far is so constructed as not to involve the use of a
“current speaker selects next” technique, self-selection for next
speakership may, but need not, be instituted, with first starter
acquiring rights to a turn, transfer occurring at that place.
(c) If the turn-so-far is so constructed as not to involve the use of a
“current speaker selects next” technique, then current speaker
may, but need not, continue, unless another self-selects.
2 If, at initial turn-constructional unit’s initial transition-relevance place,
neither 1(a) nor 1(b) has operated, and, following the provision of
1(c), current speaker has continued, then the rule-set (a)-(c) reapplies
at next transition-relevance place, and recursively at each next transi¬
tion-relevance place, until transfer is effected (Sacks et al. 1978: 12).

Central to the rules is “transition-relevance place” - a location at which the

rules may apply. Such places reflect the existence of various “unit types”
(sentential, clausal, phrasal, and lexical constructions) through which a
speaker may construct a turn. Included within that construction is a projection
of unit type and, thus, its possible completion point. The rules above operate
recursively upon the completion of successive unit types, thus providing one
important motivation for recipients of talk to continue attending to that talk.2
Note, finally, that the exchange of turns is as critical to the emergent archi¬
tecture of intersubjectivity and accountability built through talk as are ad¬
jacency pairs (Heritage 1984a: 254-60; Atkinson and Heritage 1984: 7). It is
the exchange of turns that provides for “next-position,” and thus for the
emergence of sequentially ordered displays of understandings to which actors
 Conversation Analysis 239
are mutually accountable - not only in conversation, but presumably in other
interactions as well (Greatbatch 1988).
In sum, CA approaches to discourse consider the way participants in talk
construct systematic solutions to recurrent organizational problems of conver¬
sation. The existence of those problems - and the need to find such solutions
- arises out of the ethnomethodological search for members’ knowledge of
their own ordinary affairs, knowledge that reveals and produces a sense of
order and normalcy in everyday conduct. Since the sense of order that emerges
is publicly displayed through ongoing activity, one can examine the details of
that activity for evidence of its underlying order and structure - searching not
just for evidence that some aspect of conversation “can” be viewed in a certain
way, but that it is viewed that way by participants themselves (Levinson 1983:
318-19).

3 Sample analysis: “there + BE + ITEM”

We noted above that CA approaches to discourse consider how participants

in talk construct systematic solutions to recurrent organizational problems.
Among the many problems that are solved are opening and closing talk, turn
taking, repair, topic management, information receipt, and showing agreement
and disagreement. Solutions to such problems are discovered through the close
analysis of how participants themselves talk and to what aspects of talk they
themselves attend: CA avoids positing any categories (whether social or lin¬
guistic) whose relevance for participants themselves is not displayed in what
is actually said. Analysts have demonstrated that aspects of talk as varied as
particles (Heritage 1984; Pomerantz 1984), error correction (Jefferson 1974),
laughter (Jefferson 1979), silence (Jefferson 1989), and syntactic structure
(Ford and Thompson 1986; Goodwin 1979) are relevant to the ongoing
management of conversation.
The sample analysis in this section considers how a single device addresses
several problems of conversational management. In order to do this, we
need to locate all occurrences of the device in a corpus of talk, and search
for recurrent patterns of use (sequentially based distributions) in the data,
showing that what are hypothesized to be sequential expectations are ac¬
tually oriented to by participants. In so doing, we will be showing that sequen¬
tial expectations and structures solve certain organizational problems: these
solutions may also raise (or be applied to) other problems (Levinson 1983:
326).
The device considered is a linguistic construction - the existential there
construction. Analyses of this construction have often noted that it presumes
very little in the way of new, semantically complex information at the begin¬
ning of the sentence, but focuses a great deal of information at the end of a
240 Approaches to Discourse Analysis

sentence. The information focused upon need not be assumed to be shared

with the hearer: indeed, this “newness” is what motivates its location at the
end of the sentence and its introduction by material that imposes few proces¬
sing demands upon the hearer.
(1) is an example of an existential there construction used by a library patron
(P) to open a query during a reference interview with a librarian (L) (see
chapter 5). The there construction focuses the item for which P is searching.

(1)
P: (j) There used to be a monthly report that comes from S-Securities
Exchange Commission ... on insiders’ transactions=
l: (k) Uh huh
P: (1) =and many years ago you used to carry it
(m) and I haven’t seen it in a long time.

The query is opened with there used to be: this predicates nothing but the
existence (prior to speaking time) of what we will call an ITEM. P’s description
of the ITEM being queried includes a great deal of information within the
noun phrase: when the publication comes out (monthly), its source (Securities
Exchange Commission), and its topic (insiders’ transactions). When P con¬
tinues to add information about this ITEM (when it was carried (1), her own
knowledge of its availability (m)), however, the ITEM is evoked through the
pronoun it. Thus, the existential there construction in (1) initiates a referring
sequence (chapter 6) in which a first-mention is indefinite and explicit (a
monthly report. . .) and the next-mention is definite and inexplicit (it).
I hope to have suggested thus far that focusing on the existential there for
our CA sample analysis will allow us to consider some of the same questions
raised in our earlier discussion of referring terms and referring sequences: how
are referents introduced? How do first-mentions differ from next-mentions?
What the CA approach adds to this discussion is an analysis of the way a
particular construction that seems to do referential work is integrated into the
systems through which participants organize talk. In still broader terms, the
CA approach can reveal how referring sequences are part of the mechanics of
conversation.
We noted earlier that CA research (especially when initiated and pursued by
sociologists) typically pays little attention to linguistic structure per se.4 In
keeping with this practice, I will not refer to the focus of our analysis as
existential there constructions, but as “there + BE + ITEM” constructions. This
descriptive label allows us to ignore what have often been seen as structural
and semantic differences among different surface forms of this construction.
Consider the examples in (2).

(2a) There’s the old ladies from the home.

(2b) And there was him- there was him- my brother and the dog, Prince.
(2c) There’s a little boy down here I’d say he’s eight, not no more.
(2d) And then there’s the Jewish section, too.
 Conversation Analysis 241

Although I will use the label “there + BE + ITEM” construction to describe

the uses of “there” illustrated in (2), examples like these have been identified
by linguists as different constituents of grammatical structure, and the four
sentence types have been said to derive from different underlying syntactic
structures (e.g. Erdmann 1976). These distinctions would not typically be
discussed by CA, and we, too, will disregard them for now. (In section 4,
however, I consider some of the advantages and disadvantages of this termino¬
logy.) Thus, in all the examples in (2), whatever information is relevant to the
noun following the copula will be considered an ITEM: the old ladies from
the home (2a), him, my brother, the dog. Prince (2b), a little boy down here
I’d say he’s eight, not no more (2c), and the Jewish section (2d).
I noted above that the CA approach can reveal how referring sequences -
first-mentions and next-mentions of an ITEM - are pa^*t of the mechanics of
conversation. We see in this analysis that “therd* + BE + ITEM” constructions
address two problems of conversational management: the sequential mention
of individual ITEMS and the organization of ITEMS into larger topics. The
solutions offered by “there + BE + ITEM” for the former problem intersect
with two other mechanisms of conversational organization: adjacency pairs
(3.1) and turn-taking (3.2). The way “there + BE + ITEM” addresses problems
of topic management is an outgrowth of its use with mention-pairs, but the
work of the construction at the topic level is less dependent on other systems
of conversational management (3.3). The means of presenting the analysis that
I have chosen highlight the CA view of conversational structure and the ways
that a device can be reflexive of that structure. This presentational style also
allows us to continue our focus on referring sequences. However, we will also
note separately how “there + BE + ITEM” can be used for first-mentions,
next-mentions, and only-mentions of a referent outside of the adjacency pair
organization (3.4).
Before we begin, I want to briefly compare the upcoming analysis to the
Gricean pragmatic analysis and highlight the main features of the CA ap¬
proach. In chapter 6, we used the maxims of quantity and relevance to explain
the use of definite/indefinite and explicit/inexplicit referring terms in a story.
At first, a CA approach to first-mentions and next-mentions might seem
similar to the pragmatic approach: both locate referring as something a
speaker does within a communicative setting. However, CA turns the prag¬
matics approach around: rather than posit generally applicable abstract prin¬
ciples that are used to explain specific utterances, CA seeks to discover general
principles from specific patterns of speaking. The CA search for patterns is
also a distributional method of analysis. Although this is rarely explicitly
acknowledged as part of a CA procedure, this kind of item and arrangement
(Hockett 1958) procedure is actually fairly typical (Bilmes 1990; Levinson
1983: 295). Above I noted that I will present the analysis of “there + BE
+ ITEM” in relation to adjacency pairs and turn-taking. However, I did not
assume prior to the analysis that these aspects of conversational organization
were related to the construction. Rather, the importance of adjacency pairs
242 Approaches to Discourse Analysis

and turn-taking emerged after I had located all instances of “the item ( there
+ BE + ITEM”) and tried to identify the environments in which it was used.
In what follows, then, what I characterize is the sequential “arrangement of
the item in relation to (and independent of) other sequentially definable units,
both within and across turns.
Discovering the sequential patterns in which a construction is integrated
leads CA to ask more general questions about both the construction and the
conversation. What does the distribution of a construction reveal about how
participants manage tasks of conversation? Are the tasks being managed part
of a broader problem of conversational management that is being systematic¬
ally addressed by a single device? By asking these questions, CA tries to
identify the very same problems that participants themselves - rather than
linguists or other outside observers - address when they enter into a conversa¬
tion.

3.1 “Mentions “There + BE + ITEM,” and adjacency

pairs

Since conversation typically proceeds in a pairwise fashion, a basic means of

conversational organization is the two-part adjacency pair sequence (section
2). The pairwise system can also undergo modification, however: a pair can
be expanded before its initiation, after its completion, or even during its
creation. These extensions of the adjacency pair organization support the idea
that adjacency pairs are central to sequential implicativeness.
This section considers how first- and next-mentions of an ITEM presented in
“there + BE + ITEM” are related to adjacency pair organization. “There + BE
+ ITEM” is found in two types of question/answer (Q/A) pairs: Q/A pairs that
are structurally independent of surrounding talk (3.1.1); Q/A pairs embedded
in surrounding talk, as either pre-sequences or insertion sequences (3.1.2).

3.1.1 “There + BE + ITEM” in independent Q/A pairs

This section begins with several examples of “there + BE + ITEM” in the first
parts (questions) of Q/A pairs. These examples suggest that “there + BE
+ ITEM” in questions projects the form and content of answers, such that the
second part of the Q/A adjacency pair displays its conditional relevance to the
first. I begin with examples (3 to 6) that show that “there + BE + ITEM” in
a question can project on to the first utterance in an answer, thereby imme¬
diately displaying the conditional relevance of the answer for the question.
Other examples (7 and 8) illustrate how “there + BE + ITEM” in a question
can project an organizational basis for the entire answer. We also note briefly
that “there + BE + ITEM” can be used in the second parts of Q/A pairs (but
see section 3.3).
 Conversation Analysis 243

(3) and (4) illustrate the simplest way that “there + BE + ITEM” in a question
can project onto an answer. The format projected by the question onto the
answer is just the question form itself, i.e. the unknown polarity conveyed by
a yes/no question form.

(3) IVee: (a) Fact, I never hung on the corner in my life,

(b) I never set in a bar in my life.
—> IVer: (c) Yeh? Was- was there a corner bar or a tap room that your
dad went to?
IVee: (d) Oh, yeh hhhh.

In (;a) and (b),

(through a stepwise topic transition; see section 3.2) then asks was there a
corner bar or a tap room that your dad went to*(c). Tfle IVee’s answer builds
upon the question function itself with its yeh response: the existence of “a
corner bar or a tap room” is unknown and yeh establishes that existence.
(4) is similar to (3), but it illustrates that an answer can be constructed with a
negative (i.e. X does not exist) as well as an affirmative (i.e. X does exist) token.

(4) —» IVer: (a) Were there any social functions that the church had that
you kids went to?
IVee: (b) Not me, no.

(5) illustrates how “there + BE + ITEM” can project slightly different ma¬
terial onto the format of the answer. In (5), it is “there + BE” that constitutes
the answer.

(5) -> IVer: (a) So there were a lot of mills around here in the old days.
IVee: (b) There was.
IVer: (c) Wow.
IVee: (d) But not no more. [Yeh.]
IVer: (e) [What] happened to ’em? Did they close
down, or-
IVee: (f) Well, yeh, they w- they moved away or they moved to
Can- some moved to Canada, some moved down south.

In (5), it is “there + BE” from the question in (a) that is repeated (There was)
in (b). Note that the ITEM itself has not been mentioned in any of the answers
in the examples considered thus far: in (5b), for example, the IVee does not
say there were mills. However, ellipsis of the ITEM is a source of internal
evidence that participants “know” what they are talking about. Furthermore,
the ITEM in (5) is subsequently mentioned pronominally (they) in both the
IVer’s questions (What happened to ’em? Did they close down? (e)) and the
IVee’s own further descriptions (e.g. they moved away (f)).
(6) shows that the use of “there + BE + ITEM” as a first-mention in a
question can project multiple formats onto an answer. Prior to (6), the IVer
244 Approaches to Discourse Analysis

and IVee have been discussing Irish parties; the topic of “drinking” has come
up (see 3.2). (Note that IVee-1 is hard of hearing, and her niece (IVee-2) is
acting as a spokesperson in this section of the interview; see chapter 4, and
chapter 5, pp. 168-9)

IVee-1 (a) Oh, they used to get feeling good and a lot of them did get
(6)
drunkhhhh I won’t deny thathh
IVer: -» (b) Was there ever any trouble?
IVee-2: (c) Was there ever any trouble?
IVee-1 (d) Hm, yes, there was, sometimes:,
(e) but uh they used to squash it down pretty good y’know.
There’d always be somebody there that would watch for
(f)
trouble,
you’d always have somebody to take care of anybody that
(g)
would get into trouble.
IVer (h) Uh huh.
IVee-1: (i) But she always had very nice parties and they always came
back to them, like if- she had another one herself, and Irish
dances and stuff like that.

Following the IVer’s question Was there ever any trouble? (b) and its repeti¬
tion by IVee-2 (d), IVee-l’s answer builds upon all the components of the
“there + BE + ITEM” format of the question in Hm, yes, there was, sometimes
(d). Yes affirms the existence of “trouble” (fixing the polarity of the question
per se), there was restates the “there + BE” component, and sometimes
modifies the existence of trouble per se. Following this initial part of her
answer, the IVee modifies “trouble” with a disclaimer ((e) but uh they used
to squash it down pretty good) that introduces a new ITEM they. “There +
BE + ITEM” in (f) provides a next-mention of “they” that specifies the
capacity in which “they” is relevant to “trouble”: somebody there that would
watch for trouble. Further specification is given in (g): somebody to take care
of anybody that would get into trouble. IVee-1 then returns to the topic of
“parties” (i).
Thus far, we have concentrated on how “there + BE + ITEM” in a question
can project onto the first utterance in an answer - thereby immediately dis¬
playing the conditional relevance of the answer for the question. The next two
examples illustrate how “there + BE + ITEM” in a question can project an
organizational basis for the entire answer. Since space prevents lengthy con¬
sideration of both examples, I will be brief with (7) and provide more detail
for (8).

(7) IVer: (a) -> Do you think there’s a difference between kids gettin’
married now and when you got married?
IVee: (b) Well, to me, when we got married, we got married for love,
(c) Today, they don’t even know when they’re in love.
 Conversation Analysis 245

In (a), the IVer asks whether there’s a difference between X (kids gettin’
married now) and Y (when you got married), i.e. she asks for a comparison.
Although the ITEM in this question is relatively complex, all three parts of it
(a difference, kids getting married now, and when you got married) provide
an organizational format for the answer. In (b), the IVee begins with a
repetition of the time frame when we got married and a description of what
happened at that time (a reason for marriage). The contrast in (c) follows the
same format: first the time frame (today), then a description of a different
reason for marriage. Although I do not include it here, the rest of the answer
expands this format, as the IVee alternates between time frames (today versus
in my days) and their described features.
(8) is an example of an answer that builds extensively upon the “there + BE
+ ITEM” format in the question. In (8), an IVer is asking three informants
about the population on the block. The initial questioft in (c) Are there any
old folks livin’ on this block? is prefaced by but gettin’ back to, showing that
the IVer is returning to a prior topic, i.e. the people on this block. The collect¬
ivity (“any old folks”) in the “there + BE + ITEM” question in (c) is dissected
in the answer in ways that show that the people (mentioned in the answer)
are included in the larger collectivity (mentioned in the question). (Compare
our discussion of group and individuating referents in chapter 6, pp. 213-14.)

IVer: (a) But gettin’ back to the people on this block,

(b) so it’s mostly kids, right?
(c) Are there any old folks livin’ on this block?
IVee-1: (d) Yes,=
IVer: (e) Like older people.
IVee-1: (f) =all down this street [( )
IVee-2: (g) [There’s a couple.
(h) There’s um . . .
IVee-1: (i) There’s [three.
IVee-3: (j) [That kook hangin’ out the window.
IVee-2: (k) Yeh. [That’s-
IVee-1: (1) [( ) you don’t like him ( )?
IVee-2: (m) [Kids call him Uncle Ed? Hh.
IVer: (n) [What’s that address?
IVee-2: (o) That his name?
IVee-3: (P) Uncle Ed, ( ) Uncle Ed.
(q) I’ll break his jaw.
IVee-2: (0 Then there’s one woman here, crippled or
whatever,=
IVer: (s) ZMhm.
IVee-2: (t) =you want to call her. She’s-
IVee-3: (u) Delirious.
(V) She just walks around the street with a walker?=
IVer: (w) Mhm.
246 Approaches to Discourse Analysis

IVee-3 (x) =She’ll stop any car cornin’ down the street, too.
IVee-2 (y) (No,) [hh she’s not. Hh.
IVee-3 (z) [Tell you her whole life history.
IVee-2 (aa) She gets you to come in her house and-=
IVee-3 (bb) Fix her ( ).
IVee-2 (cc) =read her mail for her and-
IVer: (dd) Hh, oh, bless her heart.
(ee) She really sounds lonely.
IVee-2: (ff) And then there’s a- another couple up the street who
never come out.=
IVer: (gg) Uh huh.
IVee-2 (hh) =And, God forbid you parked your car in their pavement
and they’re- they’re out there ready to throw bricks at
you and, they call the cops . . .
IVer: (ii) What’s- what’s that address?
IVee-2: (jj) I don’t know the address.
[TAPE ENDS]

The IVer’s question (Are there any old folks livin’ on this block? (c)) uses a
“there + BE + ITEM” format: she asks about the existence of a particular type
of person (“old folks”) in a particular location (“on this block”). Following
an initial affirmation token (yes in (d)), IVee-1 alters the locational description
of old folks livin’ on this block (c) to all down this street (f). This description
is overlapped by an affirmation from IVee-2 (There’s a couple (g)): IVee-2
repeats “there + BE” and describes the referent numerically as a “couple”.
Thus, both IVee-1 and IVee-2 answer the question affirmatively, but in differ¬
ent ways: IVee-1 with yes, IVee-2 with repetition. In addition, both IVee-1
and IVee-2 build a description of the referent, but again they do so differently:
IVee-1 provides locational information, IVee-2 provides numeric information.
In (h), IVee-2 initiates another “there + BE + ITEM” construction that is not
completed. IVee-1 then does a post-position other-repair (Schegloff 1992) of
IVee-2’s numeric description, switching from there’s a couple (g) to there’s
three (i). Recall that the IVer’s question about whether there are any old folks
contained a first-mention that focused not just on existence, but on quantity
(through any). Here we have seen that this question is followed by two
next-mentions that similarly focus on existence (through “there + BE”) and
on quantity (couple, three).
Thus far, I have noted four features of this example: the IVer’s question
incorporates a “there + BE + ITEM” format, both IVee-1 and IVee-2 answer
the question affirmatively, both IVee-1 and IVee-2 build a description of the
referent, both IVee-1 and IVee-2 initiate “there + BE + ITEM” constructions
as extensions of their initial answers.
“There + BE + ITEM” is also used for mentioning ITEMS in the answer.
Recall that the ITEM first-mentioned in the question is any old folks (c). The
first member of the group of three old folks (i) is introduced in (j): That kook
 Conversation Analysis 247

hangin’ out the window. Although IVee-3 does not explicitly say that that
kook is one of the “three old folks,” that membership is sequentially infer¬
rable. Because that kook follows a numeric specification (there’s three (i)), it
is interpreted as an enumerated item, i.e. as “counting as” one of the “three
old folks.” Compare, for example, picking cakes out at a bakery. If I say “I’ll
take three. That chocolate cake . . .”, the salesperson would understand “choc¬
olate cake” as the first item on my three part list. Further evidence that that
kook is intended as a specific member of the group of “three old folks” is the
way it is followed by a next member of the group. In Then there’s one woman
here (r), “one woman” is introduced with then (see chapter 8; also Schiffrin
1992b). By establishing “one woman” as a second member of the group of
“three old folks” - as “next” in the list - then retroactively establishes the
prior ITEM (that kook) as first in the list. Again, consider the bakery example.
If I continued the order for cakes by saying “ihen tlfe coconut,” the way I
introduced “coconut” would retroactively fix the identity of the previously
mentioned “chocolate cake” as a member of the list. In addition, the indefinite
article one (in one woman here) is a numeric mention of the ITEM: this
displays a sequential tie with three (i).
The third member of the group of “old folks” is introduced in (ff): And then
there’s a- another couple up the street who never come out (ff). This ITEM is
also established as part of the group of old folks livin’ on this block (c) in
several ways. In addition to using then and “there + BE + ITEM,” IVee 2 uses
a numeric descriptor (another) that establishes another couple as subsequent
in a list, and a locative (up the street) that establishes their presence on this
block (c). Thus, the mention of a third ITEM in the answer also reflects a
contingency upon the “there + BE + ITEM” format in the question.
In sum, (8) illustrates that the use of “there + BE + ITEM” in a question can
project the form and content of an answer well beyond its first utterance. Any
old folks livin’ on this block is the ITEM in the question introduced through
“there + BE + ITEM.” The first specification of this ITEM (as three) was also
through “there + BE + ITEM.” Following the initial member (that kook) of
the list of “three old folks” were two other entries also in the “there + BE +
ITEM” format: Then there’s one woman here, crippled or whatever, And then
there’s a- another couple up the street who never come out. All the ITEMS
mentioned in the answer were understood as members of the general ITEM
introduced through “there + BE” in the question. Thus, “there + BE + ITEM”
in a question can project an organizational basis for an entire answer, thereby
allowing a speaker to provide a continual display of the conditional relevance
of the answer to that question.
We have seen in this section that first-mentions through “there + BE +
ITEM” can occupy the first part of a question/answer pair. We have also seen
some ways that “there + BE + ITEM” can project the second part of a
question/answer pair. Discussion of the dependency between first-mention
through “there + BE + ITEM” and next-mention of that ITEM is reserved until
section 3.3.
248 Approaches to Discourse Analysis

3.1.2 “There + BE + ITEM” in embedded Q/A pairs

“There + BE + ITEM” creates a first-mention in another kind of Q/A adjacency

pair: Q/A pairs that are structurally and topically dependent on surrounding
talk as either a presequence (Levinson 1983: 345-64; Schegloff 1980) or
insertion sequence (Jefferson 1972; cf. chapter 5, pp. 159, 170). This depend¬
ency is reflected in the location of the next-mention of the ITEM in relation
to the first-mention. Instead of projecting a next-mention within the Q/A pair
itself (i.e. in the answer), the first-mention created by the “there + BE + ITEM”
format in the first part of the pair projects a next-mention in post-pair talk.
(9) illustrates “there + BE + ITEM” in a variant of the then there’s X list
format (Schiffrin 1992b; also chapter 8), acting here as a pre-sequence. Since
this presequence attends to problems in prior talk, we will also need to discuss
the talk leading up to the use of “there + BE + ITEM.”

IVer: (a) Well, you said you’d go dancin’ and then go to oyster saloon.
(b) Where did you all dance?
IVee: (c) On Second Avenue.
IVer: (d) Well- at a ballroom or a dance hall?
IVee: (e) Well, what they called a dance hall.
IVer: (f) Yeh.
IVee: (g) McKenzie.
IVer: (h) McKenzie?
IVee: (i) McKenzie.
IVer: (j) Ah.
IVee: (k) Then there was one on Main Street right down uh Main and
Parker?
IVer: (1) Uh huh.
IVee: (m) That was Rap’s.
(n) And they had them, all around.
(o) Wherever there was a dance goin’ on and they used to sell
tickets.

Let us start with the IVer’s question: Where did you all dance? (b). This
question can be interpreted in two ways: asking for location or asking about
a type of place (Schegloff 1972b). The IVee’s answer displays the former
understanding: On Second Avenue (c). The IVer then asks a question dis¬
playing the latter understanding: ballroom or a dance hall? (d). Note that this
question can be sequentially interpreted in one of two ways. Ballroom or a
dance hall? (d) can be a next-turn other-initiated repair to (b), implying that
the IVee misheard where in (a) as a question about a street name rather than
a place type. Alternatively, ballroom or a dance hall? (d) can be a post¬
elaboration, i.e. a question initiating a second adjacency pair that seeks infor¬
mation about the information given in the first (Fox 1987; 23). Regardless of
 Conversation Analysis 249

which interpretation reflects participants’ actions, the IVee selects a member

of the category “place type” from the disjunction ballroom or a dance hall?
(d), marking its descriptor (the place type, dance hall (e)) as a term used by
others but not by herself (what they call (e)). She then self-confirms it (yeh (f))
and states the name of the dance hall (McKenzie (g)).
Note, however, that McKenzie, the name of the dance hall (g), is also
problematic. After the IVee provides this name in (g), the IVer issues a next-
turn response that repeats the answer with rising intonation (McKenzie? (h)).
(We could also use the terms of chapters 4 and 5 to call this a request for
clarification or information-check.) The IVer’s questioning repetition is fol¬
lowed by the IVee’s own repetition of McKenzie (i). Put another way, the
IVer’s McKenzie? (h) offers the IVee an opportunity for an other-initiated
self-repair. The IVee’s repetition of McKenzie (i) shows that the opportunity
for repair can be bypassed. ^ ft
The “there + BE + ITEM” construction in (k) (Then there was one on Main
Street right down uh Main and Parker?) opens a presequence to mention of
the next dance hall (Rap’s (m)). This pre-sequential introduction is, in part, a
consequence of the just noted uncertainty about McKenzie. Note that Then
there was one on Main Street right down uh Main and Parker? is offered prior
to a specification of the next dance hall by name (That was Rap’s (m)). When
issued with a terminal rise (as in k), then there’s X opens a pre-sequence in
which what is being questioned is the other’s knowledge about the existence
of X. (Compare discussion of information-checking questions in chapter 5,
(3.3.3).) This use of “there + BE + ITEM” can be reissued until it is followed
by a recognition token (yes, right, um). Thus, uh huh (1) in the second part of
the embedded Q/A pair closes the pre-sequence prior to the use of the ITEM
in a next-mention (that was Rap’s (m)).
Then there was one on Main Street right down uh Main and Parker? (k) also
attends to an earlier problem from prior talk: the dual understanding of
“where” as calling for either a place category (ballroom or dance hall) or a
specification of physical location. By saying Main Street right down uh Main
and Parker, the IVee re-establishes physical location as the proper under¬
standing of “where.”
We have just seen “there + BE + ITEM” used in a pre-sequence completed
prior to the specification of the ITEM in an identificatory frame. More can be
said about an ITEM following its first-mention in a “there + BE + ITEM”
pre-sequence. In (10), for example, two people being interviewed are about to
do what is called a minimal pair test. The IVer is interested in whether
Philadelphians can perceive differences between two vowels that are very close
in pronunciation (i.e. near-mergers, Labov et al. 1991). To this end, one
person will read a word (e.g. “merry”) and another person will guess whether
the reader is trying to say the word “merry” or the name Murray. (10)
occurs when an IVee (IVee-1) who is familiar with the test is telling another
IVee (IVee-2) who is new to the test how she should announce which word
she thinks is being read by IVee-3 (referred to in 10a, b as Barb).
250 Approaches to Discourse Analysis

(10) IVer: (a) Barb, you read them.

IVee-1: (b) Yeh, you read ’em, Barb.
(c) Now you ha-
(d) now there’s a guy’s named Murray, right?
IVee-2: (e) Yeh.
IVee-1:—> (f) And there’s merry Christmas, right?
IVee-2: (g) Yeh.
IVee-1: (H) ZNow you gotta tell her which [mAri] she’s sayin’
(i) The guy’s name or Christmas.
(1) Go ahead.

In (d), IVee-1 uses the “there + BE + ITEM” format with a final rise (on the
tag right) to first-mention a guy’s named. Murray as a way for IVee-3 to
announce her identification of IVee-2’s intended meaning. After IVee-2’s in¬
formation receipt (yeh (e)), IVee-1 uses the same format to mention merry
Christmas as the announcement procedure for the alternative interpreta¬
tion of IVee-2’s intended meaning. Both of these first-mentions are prelimin¬
ary not just to the description of the minimal pair test, but to the actual use
of the terms “guy” or “Christmas” as ways to identify IVee-2’s intended
meaning during the enactment of the test. Following the completion of both
Q/A pairs, IVee-1 summarizes the use of the terms (Now you gotta tell her
which [mAri] she’s sayin (h), The guy’s name or Christmas (i)) and opens
the test itself (Go ahead (j)). Thus, the “there + BE + ITEM” format in (10)
creates first-mentions in a pre-sequence to an activity that relies on next-
mentions.
In addition to providing a first-mention in a pre-sequence, “there + BE +
ITEM” can provide a first-mention in another kind of embedded Q/A pair: an
insertion sequence. In (11), “a school” (subsequently referred to during a story
illustrating the problems of forcing schools to bus students for racial balance)
is introduced with “there + BE + ITEM.”

(11) IVee: (a) And this is what the thing is.

(b) They get in,
(c) and i- what’s it- what’s it bein’ done for?
IVer: (d) Yeh.
(e) That’s right.
IVee: (f) There- I’ll tell you.
—» (g) There’s a school down in my mother’s town, right?
IVer: (h) Right.
IVee (i) Strictly colored.
(j) It’s the newest school in North Beach.
(k) It was all colored.
(1) And because of this law of integrating,
(m) these colored children had to go to this white school,
(n) which they didn’t want.
 Conversation Analysis 251

In (a), (b), and (c), the IVee presents generalizations that both follow from her
prior points (about educational inequalities) and project further development
of those points. After beginning what sounds like a “there + BE + ITEM”
format in (f), she self-interrupts with I’ll tell you (see section 3.2) before
completing the construction as There’s a school down in my mother’s town,
right? (g). The tag right? makes explicit that this first-mention is seeking
recognition and that it is not an end in and of itself. This dependency is also
shown by the IVer’s response (right (h)) and by the fact that the IVee continues
her description of the school (i to k) after the IVer’s recognition. All of the
next-mentions of the school are inexplicit and definite: ellipsis in (i), it in (j)
and (k). In addition, the description of the school in (g) and (i) through (k)
allows the IVee to use these colored children in (m) as a referring term for
children who attended the school, without mentioning the school another
sj to
time. A
We have seen in this section that “there + BE + ITEM” opens a mention in
the first part of an embedded Q/A pair (either an insertion sequence or
pre-sequence). The next-mention does not occur within the adjacency pair
itself (i.e. not in the answer, the second part of the pair), but in post-pair talk
that follows what does occupy the second part of the pair (i.e. recipients’
acknowledgment). This is a distributional difference that is defined in relation
to units of conversational structure.

3.1.3 Summary: “there + BE + ITEM" in adjacency pairs

In this section, we discussed the use of “there + BE + ITEM” for first-mentions

in questions (the first part of an adjacency pair) and the way this construction
projects the form and content of answers (the second part of an adjacency
pair) and post-pair talk. We also saw that “there + BE + ITEM” can provide
for ITEMS in answers.
The occurrence of first- and next-mentions through “there + BE + ITEM”
in adjacency pairs is not surprising. Since the first part of an adjacency pair
creates a slot for the second part, first- and next-mentions have ready-made
conversational slots into which they can be fit (see also Fox 1987). Nor is it
surprising that “there + BE + ITEM” can project a construction format onto
the second part of the Q/A pair. The adjacency pair structure provides a slot
in which a second part is sequentially implicative on the first: what is said in
the first part will be pursued in the second part, and will be assigned an
identity based on the identity of the first part. The projectability of “there +
BE + ITEM” from the first part to the second part is thus an extension of this
conditional relevance between the two parts.
When “there + BE + ITEM” opens a mention in the first part of an embedded
Q/A pair, however, the next-mention does not occur within the adjacency pair
itself, but in post-pair talk. The relationship between the location of a next-
mention of an ITEM (in an answer or in post-pair talk) and the autonomy of
252 Approaches to Discourse Analysis

the Q/A pair is an important distributional difference. In independent Q/A

pairs, Q opens a slot for A and the provision of A completes the Q/A pair.
Thus, the projectability of “there + BE + ITEM” in Q extends only to A. In
dependent Q/A pairs, however, Q itself is sequentially implicative upon prior
talk. Thus, the contingency that Q opens is not confined to an upcoming A,
but extends into surrounding talk. The use of “there + BE + ITEM” in
pre-sequences and insertion sequences also highlights a more general CA point
about talk. Any aspect of talk can be sequentially implicative in two directions:
utterances respond to prior talk and create contingencies for further talk; that
is, they are both context shaped and context renewing (section 2).

3.2 “There + BE + ITEM” and the organization of a turn

at talk

Our discussion of the use of “there + BE + ITEM” in adjacency pairs allowed

us to focus on several ideas critical to CA: the importance of sequential
structure and implicativeness in talk, how the distribution of a specific con¬
struction is related to the pairwise organization of talk, how a specific
construction is distributed across turns in talk. This section focuses on another
major source of conversational structure and organization: the exchange struc¬
ture created by the need to take turns. The use of “there + BE + ITEM” is
sensitive to the management of individual turns and how these turns are
designed for a recipient: we find a preference for “there + BE + ITEM” to
occupy one turn-constructional unit, and to be located in a turn-internal
position. Although this section illustrates the CA method in a way similar to
that just considered (again, we are locating a linguistic device in conversational
structure), the constraint imposed by the turn exchange system is very different
from that imposed by the adjacency pair system.
Let us start by observing that almost all of the examples in the corpus of
sixty “there + BE + ITEM” constructions considered here were produced under
a single continuous intonation contour. Note that the cases that are intona-
tionally continuous include not only those with a few words and no syntactic¬
ally defined turn-transition spaces (as in 12 below) but also those that are
longer and that span syntactically defined turn-transition spaces (as in 13
below):

(12) There’s three.

(13) There’s always something to do in a home where there is a family.

That speakers produce - under one continuous intonation - even those tokens
of “there + BE + ITEM” cases that extend past syntactically definable turn-
transition spaces suggests a preference for “there + BE + ITEM” to be a single
turn-constructional unit. In other words, an example like (14) can be produced
as it actually was, without an intonational break:
 Conversation Analysis 253

(14) because there was homes over through there that I knew like from a
kid.

Or it can be produced with intonational breaks marking the syntactically

definable turn-transition spaces:

(14') because there was homes, over through there, that I knew, like from a
kid.

The fact that such utterances are routinely produced under one continuous
intonation (as in 14) suggests a preference for “there + BE + ITEM” to be
issued as a single turn-constructional unit.
CA typically finds further evidence for a preference system by comparing
preferred structures to those thought to be diapreferftd: dispreferred struc¬
tures are usually structurally marked in some way (Pomerantz 1984). Let us
examine some examples of “there + BE + ITEM” that are not intonationally
continuous to see if they are structurally differentiated from the others in
a way that suggests that they are, in fact, dispreferred. The following are
examples, from the data considered thus far, that are intonationally discon¬
tinuous.

(1) (j) There used to be a monthly report that comes from S- Securities
Exchange Commission ... on insiders’ transactions /Uhhuh/ and
many years ago you used to carry it and I haven’t seen it in a long
time.
(8) (r) Then there’s one woman here, crippled or whatever, /Mhm./ you
want to call her. She’s- /Delirious./

These intonationally discontinuous “there + BE + ITEM” constructions are

multiple turn-constructional units in particular sequential contexts that make
interactive use of the boundaries between those units. I discuss these examples
below, highlighting the way the possible turn-transition points are interactive¬
ly fit and recipient designed.
The intonational discontinuity in the example from the library reference
interview stems from the way the utterance is designed for reception by a
particular participant, i.e. a librarian. (The fact that the utterance is designed
for a librarian is of course part of our knowledge about the particular speech
event, although the term “speech event” is not used in CA.) We noted in
chapter 5 that patrons open an informational query for librarians’ response in
multiple locations. Likewise, librarians often begin to reformulate patrons’
queries at the first possible location at which they achieve some recognition
of what is being sought. In (1), the patron pauses after having provided some
key pieces of information about the ITEM being sought: its frequency (a
monthly report) and source (that comes from S- Securities Exchange Com¬
mission). This pause is thus an opportunity for the librarian to display her
254 Approaches to Discourse Analysis

recognition — if possible - and to satisfy the query. When that recognition is

not provided, the patron continues her description of the ITEM. Thus, the
“there + BE + ITEM” in (1) is designed for a particular recipient (a librarian):
the intonational discontinuity at the turn-transition spaces facilitates the pro¬
vision of a next appropriate action.
The next example, (8) from this chapter, is also recipient designed, but in
quite a different way. In (8), each part of the “there + BE + ITEM’ construc¬
tion is a separate turn-constructional unit that figures in the negotiation of
participant rights in the conversation. Put another way, the interactive work
accomplished by the intonational breaks in (8) is geared to the turn-taking
mechanism itself and the way that different participants have already estab¬
lished their rights to speak. We can best see this by reviewing earlier portions
of (8) from a turn-taking perspective:

IVer (c) Are there any old folks livin’ on this block?
IVee-1: (d) Yes,=
IVer: (e) Like older people.
IVee-1: (f) =all down this street [( )
IVee-2: (g) [There’s a couple.
(h) There’s um . . .
IVee-1: (i) There’s [three.
IVee-3: (j) [That kook hangin’ out the window.
IVee-2: (k) Yeh. [That’s-
IVee-1: (1) [( ) you don’t like him ( )?
IVee-2: (m) [Kids call him Uncle Ed? Hh.
IVer: (n) [What’s that address?
IVee-2: (o) That his name?
IVee-3: (P) Uncle Ed, ( ) Uncle Ed.
(q) I’ll break his jaw.

Three people are participating in the provision of an answer: they latch onto
one another’s talk (e), repair (i), and overlap with one another. The intona-
tionally discontinuous “there + BE + ITEM” construction not only continues
this interactive pattern, but reflects the specific way of participating in which
IVee-2 has already been engaged. Note that IVee-2’s contributions - at least
up until this point - are not attended. IVee-2’s there’s a couple (g) is corrected
by IVee-l’s there’s three (i) when IVee-2 (in what sounds like the beginning
of a list) closes herself off (There’s um . . . (h)). IVee-2 (Yeh. That’s- (k)) is
interrupted by IVee-1 (you don’t like him (1)). IVee-2 (Kids call him Uncle Ed?
(m), That his name? (o)) twice questions the name of someone who has been
mentioned. IVee-3 uses IVee-2’s information about Uncle Ed (in (p)), but
ignores the way IVee-2 had sought recognition of that information through
questioning intonation (m, o; see chapter 5, pp. 156-7). Thus, IVee-2 is
accorded less status as an authority about the block and is less likely to be
“heard” as someone who can provide an answer.
 Conversation Analysis 255

It is within this interactive context that the intonationally discontinuous

“there + BE + ITEM” occurs:

(8) -> IVee-2: (r)

whatever,=
IVer: (s) Mhm.
IVee-2: (t)
IVee-3: (u) Delirious.

The pause after here

its continued description in a way that allows the allocation of participation
rights through the turn-taking distribution system to continue in the pattern
already established, i.e. another participant can take over either to amplify or
repair what has been said. When no such transfer of Speaking rights occurs
after here, IVee-2 continues. IVee-2 pauses again after or whatever, as the IVer
issues a continuer {Mhm). Note that when IVee-2 does continue (you want to
call her (t)), she is interrupted again by IVee-3, who completes the description
of one woman as Delirious (u). Thus, the multiple turn-constructional units
created by intonationally discontinuous “there + BE + ITEM” in (8) is a way
by which participants create or replicate participant speaking rights.
Turn-constructional units in “there + BE + ITEM” also allow recognitional
work in descriptions. Before illustrating this, however, it will help to look
briefly at (15) to see how turn-constructional units can accumulate toward a
recognizable description of a location - but only pending receipt of each unit
cumulative to that

(15) IVee: (a) Did you ever ride in Grays Ferry Road?
(b) It runs like past Shunk Avenue?
(c) On one side is the uh Naval Home.
IVer: (d) Right.
(e) There’s a wall there.
IVee: (f) Big d- high wall?
IVer: (g) Uh huh.
IVee: (h) (It) would come up around the-
IVer: (i) I think I’ve even seen that.
IVee: (j) Yeh, that’s the Naval Home.
IVer: (k) Yeh.
IVee: (1) Well, they- I think they’re talkin’ about=
IVer: (m) Yeh.
IVee: =doin’ away with that.

In (15), an IVee is
(the Naval Home (c)): I think they’re talkin’ about doin’ away with that (1). Prior
to that description, recognition of the location of the place is achieved through
a series of descriptors (each a separate turn-constructional unit offered by the
IVee) that is coupled with a series of receipt markers (from the IVer).
256 Approaches to Discourse Analysis

(16) is similar, but here it is the turn-constructional units within an intona-

tionally discontinuous “there + BE + ITEM” construction that build a cumu¬
lative description of a location pending a receipt marker.

(16) IVee: (a) There’s a few pockets in West Philadelphia that are- is white.
(b) I think there’s still J- there’s uh s- a small Jewish section,
(c) of around uh S- Springfield Avenue,
(d) say,
(e) like [Fiftieth and Springfield or=
IVer: (f) [You’re right.
IVee: =something.
IVer: (g) Z That’s right.

The IVee uses “there + BE + ITEM” to present both a general category (a) and
a specific category member (b): after introducing a few pockets in West
Philadelphia that are- is white (a), he specifies a member of the set of “white
neighborhoods” {a small Jewish section (b)). Although the general category
is presented in one turn-constructional unit, presentation of the specific ca¬
tegory member is comprised of multiple turn-constructional units. In (b), the
two separate qualities (“Jewish,” initiated as /-, and “small,” initiated as s-)
are linked as a small Jewish section prior to a turn-transition point. Note,
however, that this turn-transition space is marked with “continuing” intona¬
tion. Locational information is provided about a small Jewish section in two
further turn-constructional units. The first unit offers a description of around
uh S- Springfield Avenue (c) that also has “continuing” intonation, receives
no continuer from the IVer, and then is made tentative with another turn-
constructional unit say (d). The next description is overlapped by the IVer’s
recognition markers. Thus, (16) illustrates that “there + BE + ITEM” can be
an intonationally discontinuous description when it is “added on to” pending
the receipt of recognitional tokens.
Finally, (17) illustrates a multiple turn-constructional use of “there + BE +
ITEM” in which a temporal when clause is intonationally separated from the
“there + BE + ITEM” mention.

IVee: (a) This was an Irish neighborhood at one time.

(b) Years and years ago.
IVer: (c) Right around here, huh?
IVee: (d) Yeh. [This area-
IVer: (e) [( )
IVee: (f) Oh, yeh.
—> (g) There was Jewish and Irish and Polish up this street, when
we moved in.
IVer: (h) Mhm.
IVee: (i) And then as uh, well, now I’m talking about forty years ago,
Ada.
 Conversation Analysis 257
IVer: (j) Mhm.
IVee: (k) And then I’d say in about two or three years, it started
getting strictly Italian.
(1) The neighborhood had changed.

In (17), the IVee has been describing the ethnic composition of her neighbor¬
hood (this was an Irish neighborhood (a)) in the past (years and years ago (b)).
She continues that description but shifts to another time period when we
moved in (g). The “there + BE + ITEM” format describes the ethnic compo¬
sition of the neighborhood (There was Jewish and Irish and Polish up this
street) at a point in time that differs from the prior time period mentioned.
Thus, the intonationally separated “when” clause is part of a larger descrip¬
tion of “who lived here when” (Ford 1993); it need not be seen as part of the
ITEM per se. (Note that the IVee continues ttf move^through time with her
description, shifting next to in about two or three years [past the time they
moved in] it started getting strictly Italian (k).
Thus far, I have shown that “there + BE + ITEM” shows a preference for
intonational packaging as one turn-constructional unit. This preference is
indicated not just by the frequency of “there + BE + ITEM” constructions
occurring under one continuous intonational contour, but by observations
about the particular locations of multiple turn-constructional versions of
“there + BE + ITEM.”
Another way that “there + BE + ITEM” is fit to the turn-taking system of
conversation is through its positioning in relation to turns themselves: “there
+ BE + ITEM” shows a preference for turn-internal position. I show this
preference in two kinds of examples. First are examples in which turn-initial
“there + BE + ITEM” is a repairable (18, 19, 20). Next are examples in which
turn-internal “there + BE + ITEM” is repaired-to from some other first-
mention construction (21, 22, 23).
Beginning with (18), we see a turn-initial use of “there + BE + ITEM” that
is self-interrupted, such that its completed version ends up in turn-internal
position. (18 was discussed earlier as 11.)

(18) IVee: (a) And this is what the thing is.

(b) They get in,
(c) and i- what’s it- what’s it bein’ done for?
IVer: (d) Yeh.
(e) That’s right.
IVee: (f) There- I’ll tell you.
—> (g) There’s a school down in my mother’s town, right?
IVer: (h) Right.

In (18), the speaker self-interrupts (There- (f)) to insert a pre-announcement

(I’ll tell you (f)). She then completes the “there + BE + ITEM” construction in
(g) as a recognitional (with right?), and goes on to embed an announcement
258 Approaches to Discourse Analysis

(cf. a point) in a story about students at the school just first-mentioned by

“there + BE + ITEM.” Thus, (18) suggests that a turn-initial “there + BE +
ITEM” construction may be self-interrupted to check on the recognitional
status of an ITEM and/or to allow a suspension of turn-taking rules so as to
secure a turn space long enough for an activity in which next-mentions are
embedded.
Several other examples in which turn-initial “there + BE + ITEM” is self-
interrupted suggest similar reasons for moving this format away from turn-initial
to turn-internal locations. In (19), for example, the speaker’s self-interruptions
and intonational breaks (a to c) are geared toward mentioning an ITEM (or
ITEMS) that his interlocutor could recognize (and show recognition of) before
he relies upon those mentions for a next-description:

(19) IVee: (a) Well, there’s uh certain sections, where-

(b) now like in-
(c) oh, I guess you, y’know,
->(d) in the newspapers there’s a Catholic high school,
(e) Saint Thomas Moore out there,=
IVer: (f) Yeh.
IVee: (g) =in West Philly.
(h) And they were sayin’ it’s the ideal situation.

Note that the ITEM in the “there + BE + ITEM” construction in (a) is certain
sections. The ITEM in the next “there + BE + ITEM” construction (d) is a
Catholic high school that is named and physically located in a way that is
progressively specified (from out there (e) to in West Philly (g)). Note, also,
that the location (West Philly (g)) of the second-mentioned ITEM (Saint
Thomas Moore) is a specification of the first-mentioned ITEM (certain sec¬
tions (a)): in the IVee’s continuation of his turn (not included above), this
location becomes the focus of a long description. Thus, again, it looks as if
turn-initial “there + BE + ITEM” does not secure a turn space sufficient for
an extended description.
(20) is similar: what is interesting here is that the movement of “there + BE
+ ITEM” away from turn-initial position is also a movement to a floor that
is free from competing talk.

(20) IVer: (a) Like what, for instance?

(b) I bet you could tell some really good stories.
IVee: (c) Hh I’ve had a lot happen to me, but-
IVer: (d) Well, [roll ’em out, cause this is- [this=
IVee: (e) [Well, there’s- [Y’know,=
IVer: =[is your chance.
IVee: (f) =[there’s-
-» (g) a couple weeks ago there was just a cab driver was
shot.
 Conversation Analysis 259

Observe, first, that when reintroducing there’s from (e) in (f), the IVee replaces
Well with Y know. This is a switch to a device that is more geared toward
getting the floor than responding to a prior utterance (Schiffrin 1987a). More
critically, the IVee does not complete - or even restart - the “there + BE + ITEM”
mention for the third time (a couple weeks ago there was just a cab driver was
shot (g)) until competing talk has ended. When the IVee finally does have the
floor, the temporal adjunct a couple weeks ago precedes the “there + BE +
ITEM” construction and the first-mention of a cab driver. The preposing of
the temporal adjunct is evidence that the IVee has secured a turn (Ford 1993).
In the excerpts thus far, we have said that speakers’ repairs from turn-initial
“there + BE + ITEM” suggest a preference for this construction to be turn-
internal. These excerpts also suggest that a reason for this preference is to
guarantee a turn long enough for extended descriptive use of the ITEM
first-mentioned by “there + BE + ITEM.” This guarantee does not seem to be
provided by “there + BE + ITEM” itself.
The same preference is shown by a somewhat different set of examples. (21),
(22), and (23) show that turn-internal “there + BE + ITEM” can be repaired
to form some other turn-initial (but interrupted) construction. These examples
also show that even ITEMS that are background to a different topic of talk
are turn-internal.
In (21), an IVer has asked an IVee about a childhood rhyme (a). The name
of the rhyme becomes a source of extensive discussion.

(21) IVer: (a) Do you remember um a rhyme that went “doggie diamond?”
IVee: (b) Yeh. [Yeh.
IVer: (c) [Was it a game, or was it a rhyme . . .?
IVee: (d) No, that was just a rhyme.
(e) See, remember- uh we used to call-
—> (f) in them days uh there were men,
(g) they were kinda half witted people.

The IVee begins with no more than a recognition token (Yeh (b)) to the IVer’s
initial question (Do you remember um a rhyme that went “doggie, doggie,
diamond?” (a)). Upon repetition of this token, the IVer asks another question
(Was it a game, or was it a rhyme . . .? (c)) that pursues the topic of the first.
The IVee’s answer (No, that was just a rhyme (d)) does not initially expand
the topic. In (e), however, the IVee does begin to expand the topic by invoking
recognitional work on the part of the IVer with See, remember, and we used
to call. Both of these beginnings are self-interrupted to a “there + BE + ITEM”
construction prefaced by a time adjunct (in them days uh there were men (f)).
What the IVee introduces with the “there + BE + ITEM” construction (men,
they were kinda halfwitted people (f, g)) seems to have little to do with the
topic opened by the IVer. Although space prevents me from presenting the
entire exchange relevant to the rhyme, the men introduced in (f) actually are
important because their job was to pick up dog droppings (the dog “diamonds”
260 Approaches to Discourse Analysis

of the rhyme). Thus, without knowing about these men, the IVee’s explanation
for the name of the rhyme will make no sense.
The analysis of (21) offered thus far suggests that See, remember- uh we used
to call- (e) is what Polanyi (1978) calls a true start: it is a repairable designed
specifically as something to be repaired from. (I explore the relationship
between repairs and “there + BE + ITEM” further during discussion of next-
mentions in section 3.4.) Rather than information designed to “stay on the
floor” in and of itself, the function of such a start is to gain the floor in order
to create a turn space long enough for presentation of another piece of
information (i.e. to suspend the relevance of turn-transition spaces for turn
exchange). Thus, the true start used by the speaker in (21) can create the same
result (turn-internal “there + BE + ITEM”) as the repairs from turn-initial
“there + BE + ITEM” in (18), (19), and (20).
I present (22) and (23) very briefly just to support the generality of the repair
format suggested in (21). In (22), the “there + BE + ITEM” construction is.
again, repaired-to.

(22) IVee: (a) We used to play cowboys and Indians.

IVer: (b) Oh, [really?
IVee: (c) [I remember that when I was about-1 guess I was about
four or five.
IVer: (d) Geez, that’s ( )
IVee: (e) [And we used to-
(f) there’s a lamppost in the one alley over on Hope Street.
IVer: (g) Uh huh.
IVee: (h) And we used to pretend we’d make like a little fire there.

In (22), the IVee switched from And we used to (e) to there’s a lamppost in
the one alley over on Hope Street (f). The ITEM introduced by the “there + BE
+ ITEM” construction in turn-internal position serves as the location of dif¬
ferent games being described. After receiving recognition of that item (the
IVer’s Uh huh (g)), the IVee returns to her true-start from (e) to expand the
description of the game in a way that incorporates the ITEM just introduced:
And we used to pretend we’d make like a little fire there (h).
In (23), an IVee is comparing the skills required for different kinds of sports.

(23) IVee: (a) Where hockey you can have-

—> (b) there- there’s- there is a guy he’s about five two, and maybe
not even that tall.
IVer: (c) Yeh.
IVee: (d) And he- he was one of the best players for the Montreal
Canadians for years.
IVer: (e) Huh.
IVee: (f) See, now that- that’s a sport where you have to be good to
play.
 Conversation Analysis 261
After saying that basketball is not a fair sport (because people who are tall
have an unfair advantage), the IVee brings up hockey (a). The “there + BE +
ITEM” construction is, again, repaired-to turn-internally: there- there’s- there
is a guy he’s about five two, and maybe not even that tall (b). The ITEM
first-mentioned in that construction is important for the “hockey” part of the
comparison between basketball and hockey (d): it provides a concrete example
supporting the general value of hockey (f).
To summarize this section: I have shown two ways that “there + BE + ITEM”
is fit to the turn-taking system of conversation. First, “there + BE + ITEM”
shows a preference for intonational packaging as one turn-constructional unit.
We indicated this preference not just by frequency (“there + BE + ITEM”
typically occurs under one continuous intonational contour), but also by
observations about the particular sequential locations of multiple turn-
constructional versions of “there + BE + ITEMr” Second, “there + BE + ITEM”
shows a preference for turn-internal position. We indicated this preference by
showing that turn-initial “there + BE + ITEM” is repaired-from (it is a repair¬
able), and that turn-internal “there + BE + ITEM” is a repaired-to. Both repair
formats suggest the same reason for the turn-internal preference: to locate
“there + BE + ITEM” in a turn in which current-speaker has already secured
the time needed for extended descriptive use of the ITEM first-mentioned by
“there + BE + ITEM.”

3.3 “There + BE + ITEM” and topic organization

In previous sections, we considered “there + BE + ITEM” in relation to first-

and next-mentions of an ITEM within adjacency pairs and turns. Conversa¬
tions, however, may cover more than just one ITEM at a time. Furthermore,
even when there are extended discussions of a single ITEM, such discussions
might not exhaust the boundaries of the conversation itself, i.e. people talk
about more than one ITEM during a single conversation. These observations
suggest two additional problems that must be resolved by conversationalists:
the organization of talk about multiple ITEMS and the management of tran¬
sitions between multiple ITEMS.
In this section, we discuss how “there + BE + ITEM” can help to manage
these problems. I focus on the role of “there + BE + ITEM” in what Sacks
(1972: 15-16) calls stepwise topic transitions:

A general feature for topical organization is movement from topic to

topic, not by a topic-close followed by a topic-beginning, but by a
stepwise move, which involves linking up whatever is being introduced
to what has just been talked about, such that, as far as anybody knows,
a new topic has not been started, though we’re far from wherever we
began.
262 Approaches to Discourse Analysis

Stepwise topic transitions may be built in several ways. Although Sacks’s

description of this movement from topic to topic does not require lexically
explicit transitions, it is possible to linguistically mark a stepwise transition,
e.g. through repetition (or anaphora) and a metalingual link (as in “Speaking
of that reminds me of [topic 2]”). Less explicit transitions can lack clear
markers or rely upon inferred category relationships between topics. A discus¬
sion of houses [topic 1], for example, can become a discussion of split-level
houses [topic la]; a discussion of split levels [topic la] can become a discussion
of bungalows [topic lb]. Despite the different kinds of topic relationships
possible - with or without explicit markers of linkage - all these examples
accomplish a transition by mentioning topics without explicitly opening or
closing those topics.
Before we consider “there + BE + ITEM” in relation to topic, it is important
to make a procedural point. CA assumes a methodological dependency be¬
tween how analysts pose problems and how they discover solutions. The past
sections have been considering “managerial” problems of talk to which people
must attend and the way the sequential location of a specific construction in
conversational structure offers solutions to those problems. Although a par¬
ticular conversation analysis can begin with consideration of a single problem,
analysis of what a device seems to offer a “solution for” can also reveal other
problems of talk to which people attend - simply because what serves as a
solution for one problem of talk may be applicable to another problem as well.
Although we are introducing in this section a new problem (talking about
different topics) with what seems to be a new solution (stepwise topic transi¬
tions), our analysis of “there + BE + ITEM” in relation to this problem and
solution can build upon what we have already learned. We saw earlier that
“there + BE + ITEM” could be sequentially implicative in two directions: it
could respond to prior talk and create contingencies for further talk. Stepwise
topic transitions are themselves a means of sequential organization that works
both retrospectively and prospectively: they link up whatever is being intro¬
duced to what has just been talked about (Sacks 1972: 15-16). Thus, we might
expect “there + BE + ITEM” to allow stepwise topic transitions - simply
because both build upon the same type of sequential implicativeness.
We start with (24), an example in which topic 1 (“relief”) is explicitly
provided prior to topic la (“soup houses”), and topic la-x (“a place up on
Francis Avenue”). Topics 1 and la-x both enter talk through “there + BE +
ITEM.”

(24) -» (a) I mean, in them days there was no such thing as rel- as relief.
(b) You had to make a livin’, y’know.
(c) And they had free soup houses.
-Md) There’s a place up on Francis Avenue here, oh about three miles
up.
(e) That’s still in existence yet.
(f) They se- they go in there and they make- give you soup, for free.
 Conversation Analysis 263

The speaker begins with a “there + BE + ITEM” construction that notes the
non-existence of “relief” (topic 1). After stating an alternative to “relief” ((b)
You had to make a livin’, y’know), he states that one type of assistance was
“free soup houses”: And they had free soup houses (c). Although “free soup
houses” (topic la) is a set member of the larger category “relief” (topic 1), it
is also a category that can itself be further specified. This is precisely what
happens in (d) through (f): the speaker first-mentions a specific soup house
(topic la-x) with the “there + BE + ITEM” format (d), and then provides two
properties of that soup house (e, f). Thus, “there + BE + ITEM” introduces
topic 1 (“relief”) and helps shift from topic la (“soup houses”) to the more
specific topic la-x (“a place up on Francis Avenue here”).
Although space prevents us from discussing the next example in full, (25)
illustrates the role of “there + BE + ITEM” in handling complex topic connec¬
tions among more than one ITEM at once. *

(25) IVer: (a) Yeh. Well, like did you have a team when you were goin’
to school or a little league, or-
IVee: (b) Oh yeh. [We used to- yeh, I play- I was on an=
IVer: (c) [Really?
IVee: (d) =organized- I was five years old when I was on my first
organized baseball team.
(e) I was six years old when I was on my first organized
football team.
(f) And I was seven years old when I first got in my soccer league.

After beginning to answer the IVer’s question (Well, like did you have a team
when you were goin’ to school or a little league, or-) with several repairables
(b, d), the IVee instead expands his answer with a list organized around two
topics. Topic 1 is “age”: the list is organized chronologically (I was five (d)/
six (e)/ seven (f) years old when . . .). Topic 2 is “sport”: the list is comprised
of the teams on which the IVee played different sports (“baseball” (d), foot¬
ball” (e), “soccer” (f)). Note that these two topics are related: the general
category of “age” provides a time frame within which each sport/organized
team is introduced (i.e. “I was YEARS OLD when SPORT ).
In (g), the speaker typifies himself (I was the type of kid that. . .) by bringing
together topics 1 and 2 (“age” and “sport”) and linking them with a new one
that exemplifies the type of kid he was. It is here that we find “there + BE +
ITEM.”

(g) I was- I was the type of kid that when I turned seven years old, there
was not a day in the week I wasn’t playing some kinda sport on some
kinda team.

In (g), topic 1, “age” (when I turned seven years old), and topic 2, “sport”
(playin’ some kinda sport on some kinda team), continue. What is added by
264 Approaches to Discourse Analysis
“there + BE + ITEM” is topic 3, “day”: there was not a day in the week. . ..
Thus, the “there + BE + ITEM” format in (g) provides a first-mention of a
new topic (“day in the week”) that is linked to the prior topics of “age” and
“sport.” It is in this sense that it makes explicit a stepwise topic transition:
from topic 1, “age,” and topic 2, “sport” (d, e, f), to topic 3, “day” (g). The
IVee then continues the list exactly as projected by the “there + BE + ITEM”
format: he describes further subtopics of both topic 2 (“sport”) and topic 3
(“day”) within a subtopic of topic 1 (“age”, seven years old):

(h) I used to play football on Sundays,

(i) Saturdays was my soccer night,
(j) I used to play Mondays and Wednesdays was my basketball,
(k) eh: it was Tuesdays and- what is it.
(l) Mondays and Thursdays was basketball

Thus, schematically, the list is continued exactly as projected by the “there

+ BE + ITEM” format:

Topic 1 “age”
Topic la “seven years old”
Topic 2 “sport” Topic 3 “day”
Topic 2a football Topic 3a Sundays
Topic 2b soccer Topic 3b Saturdays
Topic 2c basketball Topic 3c Mondays, Thursdays

Following an insertion sequence (not discussed here), the IVee closes his list
as follows:

(t) I used to always, I- I was playin’ sports every day.

The list thus concludes with the topics that had previously been jointly
presented in the “there + BE + ITEM” construction (g) and expanded in the
list.
We have discussed (25) in some detail to show that “there + BE + ITEM”
can contribute to the management of several ITEMS at once. In (25),
the speaker linked the prior topics of “age” (turned seven) and “sports”
(some kinda sport on some kinda team) with a new topic “day in the week.”
Thus, there + BE + ITEM linked ITEMS from prior talk (i.e. treating them
as next mentions) with an ITEM important to upcoming talk (i.e. a first-
mention).
In sum, our discussion of “there + BE + ITEM” in the management of topic
has shown that this construction can help organize larger sequential transi¬
tions between different ITEMS that serve as topics in talk. Our discussion has
also illustrated the CA dependency between the posing of problems and the
discovery of solutions. Although we introduced a new problem (talking about
 Conversation Analysis 265

different topics) and what seemed to be a new solution (stepwise topic transi¬
tions), we also noted that the use of “there + BE + ITEM” with topics might
be able to build upon the fact that “there + BE + ITEM” could be sequentially
implicative both retrospectively and prospectively. Since stepwise topic transi¬
tions build upon the same type of sequential implicativeness, it is not surpris¬
ing that “there + BE + ITEM” allows stepwise topic transitions. Thus, although
a particular conversation analysis can begin with consideration of a single
problem, analysis of the way a device offers a solution can also tell us some¬
thing about other problems of talk to which people attend.

3.4 “There + BE + ITEM” and the organization of

mentions ft

We have seen thus far that “there + BE + ITEM” constructions address two
problems of conversational management: the organization of mentions (first-
mention, next-mention) of an item and the organization of topic. First- and
next-mentions are related to two other mechanisms of conversational organ¬
ization, adjacency pairs and turn-taking. Although our focus on adjacency
pairs and turn-taking highlighted the CA view of conversational structure and
the integration of a device into specific structures, it is also important to note
that “there + BE + ITEM” can organize first and next-mentions independently
of the adjacency pair organization. On a procedural note, we will again see
how the solutions provided by “there + BE + ITEM” can be applied to slightly
different problems of conversational management.
I begin with examples showing that “there + BE + ITEM” can provide
first-mentions (outside of the adjacency pair structure) that project a variety
of next-mentions and do some of the same recognitional work as was accom¬
plished in embedded question/answer pairs. These examples also illustrate the
referring sequences familiar to us from chapter 6: first-mention is indefinite
and explicit, next-mention is definite and inexplicit.
Let us begin with (26). The IVer has asked about gangs, and the IVee begins
her answer by mentioning her girlfriends (a).

(26) IVee: (a) We did have uh three or four or five girl friends, but uh like
that. . .
(b) Like I say, we had, a radio.
IVer: (c) Yeh.
IVee: (d) So we danced at home, and different girls’ houses if we
didn’t have: any place to go.
(e) You’d uh:, go maybe one girlfriend’s house an’ you’d dance:
in the parlor, or so like that.
—» (f) Or else if uh if there’s a couple boys in the neighborhood
that you did talk to,
(g) you bring them in.
266 Approaches to Discourse Analysis

After describing activities with her girlfriends, the IVee offers an alternative
activity in (f). This alternative (you bring them [boys] in) hinges on the
existence of a couple boys in the neighborhood that you did talk to (f) - the
ITEM first mentioned in the “there + BE + ITEM” format. Note the lengthy
description in the first-mention of the ITEM: a syntactically complex descrip¬
tion that crosses a syntactically, but not intonationally marked, turn-transition
space (after neighborhood in (f), section 3.2). The term used in the next-
mention is truncated: them (g).
(27) is similar to (26) in that the ITEM first mentioned in the “there + BE
+ ITEM” format receives an immediate next-mention. It differs, however,
because it is a short ITEM (a ferry run), and because it is issued as part of a
self-repair.

(27) IVee: (a) So uh they liked to go away for them- that two days,
y’know,
(b) and that-=
IVer: (c) Yeh.
—» IVee: (d) =and then there used to be a ferry run and that-
(e) just for uh twenty cents it would take you all the way
down to what they call Billingsport.
IVer: (f) Twenty cents, (yeh).

In (a), the IVee is describing how people used to go away for summer week¬
ends. And that in (b) is self-interrupted and replaced by then there + BE and
the ITEM a ferry run (d). Although the IVee reissues and that after “there +
BE + ITEM,” and that is self-interrupted yet again and followed by a descrip¬
tion of where it (a truncated mention of “the ferry run”) will take you (e).
Note that it is difficult to say whether that (b) is a first-mention of “ferry
run” that is repaired to a more typical indefinite, explicit first-mention (a ferry
run). If this is the case, then it looks as if “there + BE + ITEM” can do remedial
work in the organization of referring sequences by providing a slot for a
first-mention that will allow a recipient to achieve recognition of the ITEM.
Next-mention it (c) can then be definite and inexplicit.
(28) illustrates that a first-mention in a “there + BE + ITEM” format need
not be lexically explicit to project a truncated next-mention. It also suggests
that “there + BE + ITEM” can be used to accomplish recognitional work
similar to that achieved in a presequence or insertion sequence (3.1.2). In (28),
the speaker is describing the time he almost fell off a roof.

(28) (a) And the ladder gave way.

(b) And it was a slanty roof.
(c) And I’m rollin’ and rollin’
(d) and somehow or other I- I managed to grab the edge of the roof
(e) and there happened to be-
(f) it was a good pla- it was-
 Conversation Analysis 267

—» (g) there- there was a what’ cha call them in there, y’know?
(h) And I could hold onto it.

Like the “there + BE + ITEM” mention in (27), the first-mention in (28) also
follows self-interruptions - here, interruptions that display difficulty in finding
a description of the item on the edge of the roof (d) that the speaker grabbed.
The first-mention (in g) ends up being formulated as a non-specific item (a
what’ cha call them in there) that seeks recognition with y’know? and thus
accomplishes some of the same recognitional work that “there + BE + ITEM”
did with pre-sequences. Use of the “there + BE + ITEM” format with a what’
cha call them in there and y’know resolves the descriptive difficulty, such that
the speaker then returns to his narration of how he saved himself from falling
off the roof with a truncated reference to what saved him: I could hold onto
it (h). *
(29) also suggests that “there + BE + ITEM” first-mentions can do recogni¬
tional work, and that such work can be tailored to a recipient’s informational
need.

(29) IVee-1: (a) Yeh, I think I’m goin’ down- downtown tonight and see
the-
-♦(b) y’know, the- the- there is such a team in Philadelphia as
the Flyers.
IVee-2: (c) How’re you gonna get in, [Butch ?=
IVer: (d) [I’ve heard of the Flyers.
IVee-2: (e) =That’s gonna cost you ninety dollars a seat.
IVee-1: (f) No, I’m gonna- they- they have the cable downtown,
down in South Philly.

In (a), IVee-1 announces his plans for tonight, but self-interrupts with a
recognitional (y’know) and a “there + BE + ITEM” first-mention of such a
team in Philadelphia as the Flyers (b). This first-mention is not intended to
help IVee-2 identify who is being spoken about. Not only are the two IVee’s
close friends (who know one another to be familiar with the Flyers), but
IVee-2’s response (he questions the feasibility of the plans (c, e)) makes
clear that he knows who the Flyers are. Rather, the “there + BE + ITEM”
first-mention is intended for the IVer. The IVer’s receipt of the information
(I’ve heard of the Flyers (d)) acknowledges the ITEM not as new knowl¬
edge per se, but as knowledge that can now be used in the course of under¬
standing IVee-1’s announcement about his plans. Thus, “there + BE + ITEM”
is a way not only of identifying the ITEM as a first-mention, but also of
making the subsequent use of an ITEM (in a later description, not included
here) recognizable to someone who cannot be presumed to be part of the
category of person who would typically know the ITEM. The “there + BE +
ITEM” format does both these tasks by providing a spot for first-mention of
an ITEM.
268 Approaches to Discourse Analysis

So far we have seen that “there + BE + ITEM” can produce first-mentions

outside of the adjacency pair structure. Such mentions vary in length, speci¬
ficity, and detail; in the terms used in chapter 6, they are indefinite and
explicit. The next-mentions projected are relatively truncated; in the terms of
chapter 6, they are definite and inexplicit. We have also been suggesting that
the first-mentions presented through “there + BE + ITEM” can help a specific
recipient achieve recognition of an ITEM.
“There + BE + ITEM” can also provide for next-mentions outside of the
adjacency pair structure. Like the first-mentions just discussed, these mentions
seem to do repair work for an ITEM whose recognition is potentially trouble¬
some to a recipient. In (30), for example, the IVer has asked if the IVee knows
what a Father, Son, and Holy Ghost house is. The IVee’s answer begins in (a).

(30) IVee: (a) That’s three- has three rooms on top of one
another. Right?
IVer: (b) Zhhhh Z You’re right! That’s the sign of an old
Philadelphian, somebody [who=
IVee: (c) [Yeh.
IVer: (d) =knows what one of [those are.
IVee: (e) [I like- I was in one.
IVer: (f) You were?
IVee: (g) Yeh.
IVer: (h) Where?
IVee: (i) A girlfriend of mine, downtown somewhere.
(i) She m- uh got married and she moved into one.
(k) One room here and then it’s one room on the second floor
and one room on the top.
IVer: (1) Yeh, that’s right.
IVee: (m) So she used her downstairs as her parlor.
IVer: (n) Uh huh.
IVee: (o) Her second-
—» (P) there’s only three rooms [like] y’know.
IVer: (q) [Yeh.
IVee: (r) The second story of the room was her bedroom and the
third room was her bedroom.
IVer: (s) Where did she cook? Where did she cook?

The initial definition of a Father, Son, and Holy Ghost house is in (a): That’s
three- has three rooms on top of one another. The IVee then identifies the
three rooms (k): each mention of a room is a member of the set “three rooms.”
She then begins to describe what each room is used for (So she used her
downstairs as her parlor (m)) but self-interrupts the description of the second
room (in o) to repeat the general category there’s only three rooms (p) in a
recognition seeking format (like y’know) prior to the description of the second
and third stories/rooms (r).8
 Conversation Analysis 269

Before going on to the next use of “there + BE + ITEM” (for only-mentions),

it is important to note that some of the “there + BE + ITEM” examples already
discussed suggest that “there + BE + ITEM” can do remedial work in the
organization of referring sequences. I repeat the relevant examples. In one
group of examples, a speaker initiates a first-mention, which is then repaired-
to as an ITEM in “there + BE + ITEM.” (As noted earlier, we cannot be sure
of the referential status of that in line (b) in (27) - such that a ferry run may
be a first or next-mention.)

(6) (e) but uh they used to squash it down pretty good y’know.
(f) There’d always be somebody there that would watch for trouble
(27) (b) and that-
(d) and then there used to be a ferry run and that- just for uh twenty
cents it would take you all the xftay down to what they call
Billingsport.
(29) (a) Yeh, I think I’m goin’ down- downtown tonight and see the-
(b) y’know, the- the- there is such a team in Philadelphia as the Flyers.
(30) (o) Her second-
(P) there’s only three rooms like y’know.
(q) The second story of the room was her bedroom and the third room
was her bedroom.

(6), (29), and (30) are examples in which “there + BE + ITEM” provides a
next-mention (a repaired-to) for a prior mention (a repaired-from). (Recall our
uncertainty about 27.) The first-mention in each case is definite [the- (29a),
her second (30o)) and possibly inexplicit as well (they (6), that (27)). The
ITEMS in “there + BE + ITEM” that are repaired to are all indefinite and
explicit.
In the second type of example, the speaker self-interrupts before first men¬
tion of an item in a “there + BE + ITEM” format:

(22) (e) And we used to¬

ff) there’s a lamppost in the one alley over on Hope Street.
(h) And we used to pretend we’d make like a little fire there.

In (22), the speaker repeats and completes the self-interrupted And we used
to from (e) in (h), after using “there + BE + ITEM” to first-mention a lamppost
in the one alley over on Hope Street (f). This continuation after “there + BE
+ ITEM” suggests a need to have provided information that would allow
hearer recognition of the location of there (i.e. the location of the game in
(h)). The use of “there + BE + ITEM,” then, seems to provide a slot in which
to mention an ITEM in a way that is recipient designed for recognition of that
ITEM.
The remedial use of “there + BE + ITEM” just suggested helps make sense
out of its use for only-mentions, i.e. when a speaker uses “there + BE + ITEM”
270 Approaches to Discourse Analysis

to mention an ITEM just once. As we see, an only-mention can be inserted

into a sequence in a backgrounded capacity: “backgrounding” might be seen
as a way that a speaker anticipates a hearer’s lack of recognition or familiarity
with an ITEM, i.e. a type of pre-emptive repair.
In (31), for example, “there + BE” introduces an ITEM during a list. The
ITEM, however, is not in and of itself a category member in the list (see
chapter 8). A speaker is describing how her mother was a very good neighbor.
After listing five activities engaged in by her mother, the speaker lists a sixth.

(31) (a) And if there was sickness in any of the homes

(b) she was always there ready to help.
(c) She was a very good neighbor,
(d) everybody loved her, really did, I mean that.

The activity she was always there ready to help (b) continues the list of the
mother’s typical activities. The “there + BE + ITEM” presented prior to that
activity provides a background condition motivating that particular activity:
it is the presence of sickness in any of the homes (a) that motivates the
mother’s help. Because familiarity with or recognition of this background
condition cannot be presumed, we might say that introducing it prior to
continuation of the list pre-empts a need for later repair. Once the mention
of “sickness” has served this function, “sickness” is not mentioned again;
rather, the speaker closes the list with a repetition of what each listed activity
established, i.e. she was a very good neighbor (c).
(32) shows that “there + BE + ITEM” can be used to insert background
ITEMS in lists of events that are interpreted as having occurred successively
in time. In (32), the speaker is describing her mother’s activities during a wake
for a deceased relative and uses “there + BE + ITEM” to mention an item
relevant to those events.

(32) (a) When he died, my mother had all the clay pipes: and the tobacco,
(b) and she bought all different brands of cigarettes,
(c) and you laid them around.
(d) So as: your friends came in,
-♦(e) there was someone to take them upstairs and give them a shot.
(f) And then they’d go downstairs
(g) and then they’d s-
(h) y’know, each person come in would get a shot of whiskey.

The events in (32) are a list of activities comprising “a wake”: they are
presented as “one-after-another” events in “a story.” Interspersed with the
story events are several background conditions (when he died (a), as: your
friends came in (d)) that provide a temporal framework for the events. The
“there + BE + ITEM” format also provides a background condition for an
event - not a temporal condition, of course, but an existential condition: a
 Conversation Analysis 271

person (someone to take them upstairs) is present to facilitate a next activity

(“have a drink”) in the wake by giving participants a shot (of whiskey). The
speaker continues to another event (which itself follows a return to downstairs
(f)), but self-interrupts to restate the general circumstance from (e) as an event
per se without explicit mention of the “someone” who would facilitate it:
y’know, each person come in would get a shot of whiskey (h).
(31) and (32) suggested that “there + BE + ITEM” can insert only-mentions
into lists of events in which the only-mentioned ITEM serves in a background
capacity. (33) is an example in which an only-mentioned ITEM provides back¬
ground for an important event. The speaker is telling about a childhood fight.

(33) (a) So the next thing I know I’m standing out front
(b) and she comes up
(c) and she happened to say something very fcul about my mother
(d) and that just drew the line.
(e) With that I went up the playground
(f) and all the neighbors-
(g) I had a viewer-
-Mh) uh like y’know there was so many people up there like you
might’ve thought Cassius Clay was fighting or somebody, y’know.
(i) And uh, my Aunt Betty came up and got me.

The story events are in (a) through (e). Following these events, the speaker
offers three possible descriptions of an audience to the fight. First is all the
neighbors (f). This is self-repaired to 7 had a viewer (g), which, in turn, is
reformulated not only in a recognition seeking format (“like y’know X,
y’know”), but in a “there + BE + ITEM” format to like y’know there was so
many people up there like you might’ve thought Cassius Clay was fighting or
somebody, y’know (h). Despite the length and complexity of this ITEM, it is
not used again in the story: rather, the speaker returns to the events themselves
to continue the story.
In sum, “there + BE + ITEM” can be used in a variety of ways that are
independent of the adjacency pair structure (3.1.1). This construction can
introduce first-mentions that project a truncated next-mention. It can also be
used for next-mentions and only-mentions of an ITEM. When used for next-
mentions, “there + BE + ITEM” may serve as a recipient designed repair. When
used for only-mentions that are backgrounded to other talk, there + BE +
ITEM” may pre-empt the need for repair.

3.5 Summary: “there + BE + ITEM”

Our sample analysis has examined how “there + BE + ITEM” fits into two
conversational systems, adjacency pairs (3.1) and turn-taking (3.2). We have
also examined the role of this construction in topic management and in the
272 Approaches to Discourse Analysis

organization of mention pairs (i.e. referring sequences) outside of adjacency

pairs. Thus, we have ended up considering several different facets of conversa¬
tion together: referring sequences, question/answer pairs, turn-taking, and
topics. That the analysis of “there + BE + ITEM” led us to consider these
different tasks suggests not just something about the construction itself (i.e.
one construction can serve different tasks), but more generally about conver¬
sational management: conversationalists need to manage multiple problems
simultaneously.
That one construction can fit the resolution of simultaneous tasks also
suggests that these tasks themselves might be related to one another, having
more in common than is sometimes thought. Take, for example, the recogni-
tional work achieved by the turn construction of “there + BE + ITEM.” We
saw that “there + BE + ITEM” is packaged as a single turn-constructional unit
as a way of leading a recipient toward recognition of an ITEM. We also saw,
however, that “there + BE + ITEM” is a multiple turn-constructional unit
when used in repeated attempts to accomplish recognition of an ITEM. The
location of “there + BE + ITEM” in an insertion pair is also geared toward
hearer recognition. This suggests that the task of helping a hearer identify a
referent has a systematic bearing on how turns are constituted and exchanged
and on how (and when) adjacency pairs are embedded in ongoing talk. Thus,
by explicating the distribution of a construction in conversation, we have
gained insight not only into how that construction contributes to the sequen¬
tial progression of talk, but also into how different tasks, and different sys¬
tems, are related to one another.
We can also consider what a CA analysis of “there + BE + ITEM” contributes
to our understanding of referring sequences. Although referring sequences in
general are not pairs that are as tightly organized as adjacency pairs, we have
seen that “there + BE + ITEM” does project a next “slot” in several ways: the
form and content of an answer can reflect the prior “there + BE + ITEM”
construction in the question; a next-mention can be truncated (definite and
inexplicit) in the second parts of adjacency pairs or in post-pair talk to an
embedded adjacency pair. We have also seen that a next-mention through
“there + BE + ITEM” may serve a remedial purpose, or that a first-mention
may pre-empt the need for remedy.
As noted above, however, referring sequences are not pairs that are as tightly
organized as adjacency pairs. Referring sequences are often produced by one
person, thus violating the two party requirement. Furthermore, a speaker may
very well refer just once to an entity, with no “noticeable gap” produced upon
the absence of a second mention. Also distinguishing referring sequences from
adjacency pairs is that the constraint to continue a referring sequence can vary
depending upon the format of the first-mention. Y’know X?, for example,
allows a first-mention of X that seems to unconditionally project a second
mention of X (cf. discussion of information checks in chapter 5, pp. 169-73).
Thus, in (34), it would be surprising if the item brought into discourse by
y’know was not subsequently used in the discourse:
 Conversation Analysis 273

(34) zelda: (a) Y’know, that eh orthopedic doctor?

(b) Y’know that took care of Henry when he had his back?
(c) Problems?

Although not included in (34), Zelda goes on to use the referent “that ortho¬
pedic doctor that took care of Henry when he had his back problems” in a
short narrative. Thus, as I noted above, the y’know X? format carries with it
an expectation that X will be mentioned again and will be used in some way
in the discourse. (See Schiffrin (1987a: 285) for a description of this exchange
structure. Schiffrin (1987a: 267-89) also discusses how the y’know X? format
involves the hearer in the interactive recognition of an entity.) In contrast to
the way y’know X? projects a second mention, other ways of mentioning an
entity may very well be heard as an only mention. (35) illustrates the X as an
only mention. *

(35) And I happened to look and there the blood was cornin’ out of the
horse’s leg, y’know, and then, the foot.

In (35) - which is a climactic moment in a story about the cavalry - there are
three items that are mentioned once and only once: “the blood,” “the horse’s
leg,” and “the [horse’s] foot.”
In sum, what a CA approach contributes to our knowledge of referring
sequences is the following. Referring is an interactive process that is inter-
meshed with other structures of talk: first- and next-mentions are located
relative to adjacency pair structure and turn-taking constraints. Linguistic
constructions are used in ways that allow a recipient to recognize an item (i.e.
a referent) at the same time that other conversational tasks are being pursued
and accomplished. Finally, referring terms are sequentially implicative, but in
a looser sense than adjacency pairs.

4 Conversation analysis as an approach to

discourse

We have seen in this chapter that a CA approach to discourse considers the

way participants in a conversation construct systematic solutions to recurrent
organizational problems of talk. It discovers such solutions through the close
analysis of the sequential progression of talk. By focusing on the way “there
+ BE + ITEM” allows first-mentions, next-mentions, and only-mentions - in
relation to adjacency pairs, turn-taking, and topic management - we have seen
not only how attention to the sequential progression of talk can reveal the
distribution and interactional functions of a conversational phenomenon (both
274 Approaches to Discourse Analysis

within and across turns), but also how such attention can suggest interdepend¬
ent relationships among tasks (functions) and among solutions (systems).
Thus, although I began with a single construction in relation to a single
problem (i.e. referring), the way the problem seemed to be solved by the
construction led to discussion of other problems and other systems of talk.
It is in this sense that the order in which I presented examples of “there +
BE + ITEM” was actually more important - in a methodological sense - than
it might have seemed. I presented each example (and set of examples) as both
a response to and a prompt for another example (and set of examples):
different ways of presenting examples might have suggested a different mode
of organization in the data. Far from seeking to impose an external “out¬
sider’s” organization on the data, however, I would argue that this attention
to order in the data was actually quite the opposite: an attempt to arrange
instances of talk in a way that would reveal the “natural” and “correct”
arrangement of items in the sequential and distributional (cf. syntagmatic and
paradigmatic) patterns with which speakers themselves work. Put another
way, my sequential presentation of data in this chapter tried to mirror the way
conversational phenomena (items, constructions) are themselves patterned for
speakers. This is yet another reflection of the use of talk as a resource for the
discovery and definition of particular problems and their solutions (“rules,”
“systems,” “connections among systems,” “functions”) from the participants’
point of view.
Consider, also, that traditional terms of linguistic analysis played almost no
role in my sample analysis. As noted initially, “there + BE + ITEM” has been
considered quite widely by linguists as existential there, i.e. the use of there
with a copula predicates existence of a referent (often an indefinite NP). (See
Erdmann (1976) for an extensive review and Hannay (1985) for an analysis
in functional grammar.) Despite the lack of attention to grammar, CA analyses
of discourse can have an important bearing on our understanding of grammar
(e.g. Ford 1993; Ford and Thompson 1992; Fox 1987; Goodwin 1979; Sche-
gloff 1979a). Existential there sentences are right branching constructions that
can focus a great deal of information (recall the length and syntactic complex¬
ity of some of the ITEMS) at the end of the sentence. This might be said to
serve particular communicative problems very similar to those addressed in
our account. Put another way, we might say that the syntactic qualities of
existential there sentences arise from the problem (discussed in our analysis)
of trying to get another to achieve recognition of a first-mention. The inter¬
actional packaging of existential there sentences as multiple turn-construc¬
tional units (with each unit adding another syntactic constituent) might be
motivated by the need to achieve recognition of an ITEM in a way that
provides its own opportunity for the confirmation of intersubjectivity at each
syntactic (also, each informational and possibly intonational) juncture (cf.
Schegloff 1992).
I also noted initially that, consistent with the CA approach, I referred to
“there + BE + ITEM” in a way that presumed very little about linguistic
 Conversation Analysis 275
structure. Before we close this chapter, it will be helpful to compare three
different types of “there” that CA would not differentiate: demonstrative,
adverbial, and pronominal (see also Freeze 1992). By so doing, we will be able
to assess the strengths and weaknesses of the CA practice of ignoring language
structure.
(36) illustrates demonstrative there. The IVee had been talking about his
time in the cavalry (a, b), and had interrupted himself to greet some acquaint¬
ances who passed by (c). In (d), he identifies the acquaintances who he has
just greeted with a there construction. (36 was 2a.)

(36) IVee: (a) And uh, but, you couldn’t do nothin’.

(b) See, what happened when that cavalry was there-
(c) Hello there.
(d) There’s the old ladies from the hom£.=
IVer: (e) [I figured this.
IVee: (f) [=A11 these women who we’ve talked to?
IVer: (g) Yeh.
IVee: (h) They all know me because I- I do errands for them.
(i) Those that can’t get out, y’know?

There in (36) is a demonstrative that has a deictic function: the referent to

which it points (“the old ladies from the home”) is situationally present and
there locates it at a relative distance from the egocentric locus of the speech
event (i.e. the speaker himself). More generally, there is a spatial deictic that
occupies the distal end of a proximal-distal axis: there contrasts with here to
indicate distance from the speaker’s current location (e.g. “I worked there”
would indicate that “there” is not the speaker’s current location). Notice that
demonstrative there in (36) introduces a first-mention as a definite explicit
noun phrase (the old ladies from the home (d)); next-mentions are pronominal
(they (h), those (i)). Notice also that the IVee “points” to a situationally
present referent after he has already greeted them (Hello there (c)) - after they
have become salient in the interaction. As we also noted, the IVee continues
to identify these referents for the IVer. This sequential location is similar to
that already seen for existential there sentences (e.g. in 3.4).
Adverbial there in (37) also has a locational meaning. (Because there does
not physically situate the referent in the immediate here and now, however,
uses such as this are not typically seen as deictic.) Here is the fuller context:
(37) is from a story in which the speaker (an adult) is telling about how his
brother (as a child) had discovered some treats that his mother had hidden
from the children. (37 was example 2b.)

(37) (a) So, my mother went out,

(b) and my brother says, “I gotta search around here.”
(c) So he went upstairs in the bedroom
(d) and there he found them hid.
276 Approaches to Discourse Analysis

(e) You know, she had hid them.

—> (f) And there was him- there was him- my brother and the dog, Prince,
(g) They got on the bed and he was goin’ to the dog, One for me and
one for you.”

The brother’s search for the treats begins in (c) as he travels upstairs in the
bedroom; this location is marked by an adverbial there in (d). There in (f) -
the case of interest - is also a marker of location, indicating where (in the
bedroom) the brother discovered his other brother and the dog. Thus, both
cases of there in (37) are locational adverbs: they convey the location of a
situation, i.e. an action (“he found them” in (d)) and a state (“my brother and
the dog were at X” in (f)). Notice, again, that although the first-mentions in
(f) are definite noun phrases (him, my brother, the dog, Prince), the next
mentions are less explicit (they). And again, there is a repair from a less explicit
him to a more explicit my brother within the there sentence itself.
(38) illustrates the case of “there + BE” that has been said to differ most
dramatically from those with the locational meaning. In (38), the speaker is
telling a story that reports on the behavior of some neighborhood boys. The
general point of the story is to complain about the parents of those boys -
since the parents are nowhere to be found during the disturbing incident being
reported. (38 was example 2c.)

(38) (a) They wouldn’t take-

(b) They got one little boy over there-
(c) now these are boys about fourteen, thirteen, fourteen.
->(d) There’s a little boy down here
(e) I’d say he’s eight, not no more.
(f) I seen ’em cornering him over there about four of them.
(g) On the lady’s: step in the corner.
(h) He was cryin’
(i) and they were spittin’ on him.

The basic idea underlying the type of “there” in (37d) is that it is not a
locative: the construction “there + BE + [indef] NP” asserts the existence,
rather than the location, of an entity. (Important - but not absolutely critical
- to the identification of this type of “there” is that the NP is an indefinite
noun phrase (see discussion of definiteness and indefiniteness in chapter 6,
pp. 198-9; cf. Abbott 1993; Rando and Napoli 1978; Ward and Birner 1993;
Ziv 1982). It is this case of “there” that has been called not just existential,
but expletive (Fillmore 1968), pronominal (Erdmann 1976), and presentative
(Hannay 1985).
Since “there” can be used in a sentence that asserts existence of an entity
(in particular, as an indefinite NP), it is often suggested that “there + BE +
[indef] NP” is a very useful way to introduce a referent into discourse. That
is, because “there + BE + [indef] NP” predicates very little about an entity
 Conversation Analysis T71

other than its existence, it is an efficient way to provide information about

something about which hearers cannot be assumed to have prior knowledge
(cf. the definition of indefinites) in a separate chunk of talk (idea or tone unit:
Chafe 1992) without also forcing hearers to immediately make use of that
entity in a more elaborated way (e.g. as someone doing something). Further¬
more, by placing a weak predicate (“be”) prior to what is often a long and
highly informative noun phrase, “there + BE + [indef] NP” creates a word
order in which last position is reserved for what Firbas (1964) calls the rheme,
i.e. new information with the highest degree of communicative dynamism.
Returning to our example in (37d), we can see that this is exactly what
happens here: a new entity (“a little boy down here”) is introduced, to which
subsequent reference is then brief, inexplicit, and definite (a pronoun in e, f,
h, and i). Note, also, that the referent is introduced with a lengthy predicate
(a little boy down here) followed by a brief independent clause (I’d say he’s
eight, not no more) that further modifies the description (he is not just little
but eight, not no more); it is also a repair of an earlier introduction in (b)
They got one little boy over there-.
I have discussed here uses of “there” that are demonstrative (deictic, 36),
adverbial (37), and pronominal (38). The demonstrative and adverbial uses
are typically said to occur in locative sentences, the pronominal there in
existential sentences. Although it seems relatively easy to differentiate the uses
just noted, there are some cases that are less clearcut. (39) illustrates.

(39) IVer: (a) What- what would it be mostly around here?

IVee: (b) Oh, combination Irish:,
(c) Irish and there-
-►(d) and then there’s the Jewish section, too.
(e) There’s no Jewish people right in this area, either.
(f) I guess it’s just Irish, Polish,
all the leftovers, from Italian and Jewish.

After having been asked what it would be around here (i.e. what ethnic groups
would live here) in (a), the IVee lists Irish (b), Jewish (d), and Polish (f). In
(38d), “there” in and then there’s the Jewish section, too seems to be existen¬
tial despite the definite NP, simply because the existence of a new entity is
being predicated. In other ways, however, “there” seems to have a locational
meaning and thus to be adverbial: although the IVer asked about around here,
the speaker is listing sections in the physical space designated by “around
here,” and could be seen as travelling (on an imaginary tour) through her
neighborhood. (Note, also, that the “there” construction in (e) denies the
existence of a Jewish section right in this area: this might be seen as a “weak
locative.”) In still another way, the “there” construction in (d) can be said to
be locative in a kind of metaphorical way in a discourse world, i.e. to locate
an item as an item being enumerated as part of a list comprised of a set of
such items (cf. Bolinger 1977; Lakoff 1987: 462-582). This is also suggested
278 Approaches to Discourse Analysis

by the addition of then as a list marker (chapter 8), i.e. the self-repair from
there (c) to and then there (d).
Although CA would not compare different uses of there, I have done so to
suggest some of the strengths and weaknesses of the CA practice of ignoring
language structure. One weakness is that syntactic and semantic differences
such as those suggested here are ignored by calling all cases of ‘there + BE
the same construction. Recognizing the existence of different cases of there,
however, actually gives us more of an appreciation for how the different uses
seem to crosscut one another (i.e. 39). This might then be said to challenge
syntactic and semantic classifications that assume mutually exclusive structures
and functions. My discussion has hinted at an underlying shared function of
these constructions in talk with a similar consequence for next-mentions. Not
only did each use of a there construction allow a truncated next-mention,
but each use located an entity (in some way) and did so in a sequence that
required recognitional (and some remedial) work on the part of the hearer. A
CA analysis that examines all there constructions, but also classifies them
according to conventional syntactic and semantic categories, might be able to
best explicate the functional basis of the structural categories.
Just as traditional linguistic analyses not explicitly mentioned by CA can
actually be related quite nicely to the results of CA analyses, so, too, we find
that terms and concepts not explicitly mentioned can be said to underline such
analyses and to be part of the implicit assumptions and procedures of CA.
Although I discuss these terms and concepts in chapters 9, 10, and 11, it is
worth giving an example here. As I have repeatedly said, CA is an approach
to discourse that works from “what is said”: more specifically, it takes what
is said as the source of observations, of evidence, and of explanations. This
dependence on what is said makes CA an approach to discourse that is
strongly bound to text. This is not to say, however, that people are assumed
to communicate only on the basis of what is said at a particular moment in
time. As ethnomethodologists make clear, what is said is not only indexical
of a larger social world (as it simultaneously creates and re-creates that world),
it also cannot be understood without the taken-for-granted assumptions that
we bring to (and reconfirm with) our utterances. Nevertheless, it is text that
remains central to CA procedures when applied to talk: text is the way our
taken-for-granted worlds are discovered. Put another way, text is the route to
our discovery (as analysts, as participants) of the contexts that text creates
and by which it is created.
Consider a final point about the CA approach to discourse that emerges from
our analysis of “there + BE + ITEM.” We have analyzed the distribution of
“there + BE + ITEM” in several different environments: adjacency pairs (in¬
dependent, dependent), turns, topics, and mention-pairs. Although we dis¬
cussed the use of “there + BE + ITEM” in each of these environments
separately, it should be clear that we actually used our observations about the
construction in one environment to inform our analyses of another. We noted,
for example, that “there + BE + ITEM” can do recognitional work in em-
 Conversation Analysis 279

bedded question/answer pairs: this was an inference about the function of the
construction based on observations about its sequential distribution in relation
to particular units of conversational structure. When we found that the same
construction occurred in a quite different environment - for first, next, and
only-mentions outside of adjacency pairs - we nevertheless tried to explain
this set of occurrences in terms of the functions already identified from our
previous analyses. CA is reluctant to assign either meaning or function to a
form apart from a thorough investigation of how that form is sequentially
embedded in actual talk. Once a function is found in one environment, how¬
ever, it is reasonable to see whether that same function can be applied to other
tasks. We thus related our knowledge that “there + BE + ITEM” could do
recognitional work in embedded Q/A pairs to other sequences - finding the
same kind of function. We made a similar point through our discussion of the
relationship between “there + BE + ITEM” and stepwise topic transitions:
since “there + BE + ITEM” could be retrospectively and prospectively implic¬
ative in embedded Q/A pairs, it was not surprising that this structural ability
could be applied to stepwise topic transitions. Thus, the CA reluctance to
assign meaning or function to a construction might better be seen as a reluct¬
ance to assign a meaning or function that has been taken out of context, i.e.
decontextualized.

Exercises

1 Analyze the turn-transitions in a five-page sample of spoken data. What

turn-taking rules apply and where do they apply? (For example, do certain
speakers select others more frequently? Do some speakers self-select more
than others? Where is self-selection likely?) Which instances of overlapping
speech are interruptions, and by what criteria did you identify them as inter¬
ruptions? Upon what linguistic qualities (e.g. syntax, semantics, pragmatics,
intonation) of an utterance do current-speaker and next-speaker seem to rely
as a guide to the identification of turn-transition spaces?
2 Our CA sample analysis differentiated independent from embedded ques¬
tion/answer pairs. What are the sequential differences between these two loca¬
tions? Can these differences be related to the linguistic (syntactic, semantic,
pragmatic) structure of questions?
3 Pre-sequences are utterances that prefigure a more focused upon speech
action, e.g. a pre-announcement such as / have to tell you something prefigures
an announcement (Terisaki 1976); a pre-question such as Can I ask you some¬
thing? prefigures a question (Schegloff 1980). Using a corpus of conversational
data, try to identify the interactional work that is accomplished by different
pre-sequences.
4 Telephone openings are summons-answer sequences that help to set up
the exchange structure of a phone conversation, as well as provide an environ¬
ment for speaker-hearer recognition (Godard 1977; Schegloff 1972a, 1979a).
280 Approaches to Discourse Analysis

Tape-record twenty to thirty phone openings in either a single setting (e.g. a

home) or several settings (e.g. home, business). What is the typical sequence
of steps? How are relationships between the caller and the called party (e.g.
their social identities, level of familiarity, intimacy, power differences) revealed
in the openings? (Helpful readings include Godard 1977; Robert 1992; Sifianou
1989.)
5 Conversation analysts rely upon both phone conversations and face-to-
face conversations as data, without explicitly assessing the possibility that
these two different channels of communication may influence (or be displayed
in) what is said. Tape-record several exchanges on the phone and several
during face-to-face interactions. If possible, try to record roughly comparable
speech events from both sources, e.g. a friendly chat, requesting information
from a public service, complaining about an insurance policy or the quality of
a service. How are the different channels of conversation reflected in the turn¬
taking and adjacency pair organization of the exchanges? Are there other dif¬
ferences, e.g. in pause length, in use of discourse markers, in openings and
closings, that may be related to the channel difference? (This exercise may also
be applied to comparisons across spoken and written modalities, or to compari¬
sons across speech situations.)
6 Our sample analysis in this chapter focused upon there is constructions.
Examine the sequential location of this construction in a different corpus. After
comparing the results of the two analyses, assess the relationship between
sentence structure and discourse function in conversation analytic terms. (Help¬
ful work to read might be Ford 1993; Fox 1987; Goodwin 1979; Schegloff 1979b.)

Notes

1 CA transcripts do not always use traditional orthography or spelling, depending

instead on what has been called “eye dialect” to replicate the way people speak.
Jefferson (1984: 348), for example, provides the following transcribed line:

I thought that was pretty outta sight didju hear me say are you a junkie.

A key feature of this system is that phonetic details (e.g. associated with rapid speech
rules, voicing (surprise might be transcribed as “surprize”), and so on) are repres¬
ented orthographically. The use of “eye dialect” has been criticized by linguists (e.g.
Macauley 1991; Preston 1982).
2 A key issue in the organization of turns centers on the identification of turn-
constructional units and the parameters (e.g. intonational, syntactic, semantic) that
underline their identification (see Ford and Thompson 1992). This issue intersects
with the identification of the basic linguistic units used by speakers in spontaneous
conversation.
3 Although CA research has not focused extensively on referring terms (or on problems
of reference), much has been said that is relevant to these issues. CA research relevant
to referring sequences includes the following. Schegloff (1972b) discusses how ways
of formulating location reflect participant decisions about membership: for example,
how one selects from among alternate place descriptions such as “65-year-old house,”
 Conversation Analysis 281

“Silver Spring,” “Washington,” “the East Coast,” “the US” reflects speakers’ cat¬
egorization of hearer. Sacks’s (1972b) discussion of membership categories touches
on the attribution of relationships between referents: when we hear a sequence like
“The baby cried. The mommy picked it up,” for example, we understand “the
mommy” to be the mother of the baby. Schegloff and Sacks (1973) point out, in
their discussion of minimization in reference, that our referring terms provide the
least amount of information needed for recognition by another. General ethnomethodo-
logical views on the indexicality of meaning (Garfinkel 1967) are also relevant to
reference: e.g. the inherent contextual basis of all meaning is exemplified by the
referential vagueness of an utterance like “it is nice.” Most recently, Fox (1987) has
shown that alternation between nominal and pronominal reference not only follows
sequential constraints, but also helps to create such sequential structures. Conversa¬
tion analysts (e.g. Sacks 1972) have also discussed topic in talk, i.e. how participants
accomplish “talking about something.”
4 Linguists who are interested in how syntactic s*ructurd» emerges from discourse,
however, sometimes combine CA techniques with syntactic analysis (e.g. Ford 1993;
Fox and Thompson 1990).
5 Prince (1981) and Halliday and Hasan (1976) offer different perspectives on the
relationship between ellipsis and recoverability of information.
6 Note also that the questions in embedded Q/A pairs are yes/no questions about
existence with no subject-auxiliary inversion, i.e. they are there’s X? but not is there
X? In this sense, they are more similar to try markers or information-checking
questions (chapter 5, section 3.3.3); see also our discussion of rising intonation with
declarative sentences in chapter 3 (pp. 65-8).
7 Definitions and analyses of “topic” within linguistics are not consistent, partially
because of the conflation of the notion of topic with information status: topic is often
seen not only as “what we are talking about,” but also as “what we are familiar
with,” i.e. given information (Schiffrin 1992c). By the former definition, the ITEM
introduced by “there + BE + ITEM” can certainly become a topic - although it is
unlikely to be a topic at the time of mention. Furthermore, if the ITEM is a
first-mention, the latter criterion (its familiarity to hearer) cannot be assumed.
8 Note that the IVee’s mention of two bedrooms (in line r) is apparently a mistake and
is treated by the IVer as an opportunity for other-initiated self-repair. The interaction
continues, but is interrupted (when another person enters the room) before the
self-repair can be completed:

IVER: Where did she cook? Where did she cook?

IVEE: On the second floor. That was her-z
IVER: Her bedroom and kitchen, huh?
IVEE: No, no. See you have three- like- [there’s=
IYee: [No, come on in, it’s OK.
IVEE: =three rooms.
Oh, did you get your new shoes, hon?
Oh, they’re cute.
IVER: Hey, those look pretty good.
8 Variation Analysis

1 Introduction

The variationist approach to discourse is the only approach discussed in this

book whose origins are solely within linguistics. This approach stems largely
from studies of variation and change in language: fundamental assumptions
of such studies are that linguistic variation (i.e. heterogeneity) is patterned
both socially and linguistically, and that such patterns can be discovered only
through systematic investigation of a speech community. Thus, variationists
try to discover patterns in the distribution of alternative ways of saying the
same thing, i.e. the social and linguistic factors that are responsible for varia¬
tion in ways of speaking.
Both the initial methodology and the theory underlying such studies are
those of William Labov (who has also developed a speech act approach to
discourse; chapter 3). Although traditional variationist studies have been
limited to semantically equivalent variants (what Labov calls “alternative ways
of saying the same thing”), such studies have also been extended to texts. It
is in the search for text structure, the analysis of text-level variants and of how
text constrains other forms, that a variationist approach to discourse has
developed.
After an overview of variation analysis in relation to discourse in section 2,
I present a sample analysis that focuses upon a particular discourse unit, the
list, in section 3. My sample analysis takes two directions. First, I compare
the structure of lists to the structure of narratives (3.1). Second, I analyze
linguistic variation within lists, focusing here upon referring terms and se¬
quences (3.2). These two directions show that variation analysis provides a
two-pronged approach to discourse, one that focuses upon texts per se, and
one that focuses upon how texts constrain other variants (3.3). Section 4
summarizes.
 Variation Analysis 283

2 Defining variation analysis

As I noted above, the variationist approach to discourse is the only approach

discussed in this book whose origins are solely within linguistics. Although
the linguistics in which it is grounded is “socially realistic” (a term used by
Hymes 1974d: 196), Labov resists the term sociolinguistics “since it im¬
plies that there can be a successful linguistic theory or practice which is not
social” (1972f: xiii). Furthermore, although social factors (e.g. social stratifi¬
cation) are considered in actual analyses, and in more general formula¬
tions of patterns and explanations of distributions, the influence of sociology
(its assumptions, concepts, and theory) was not heavily incorporated into
early studies of language variation and change. Thus, the influence of lin¬
guistics pervades the variationist approach to discourse. (But see sections 2.2
and 2.3.)
Many of the problems addressed by variationists are problems familiar
to linguists in general, e.g. the problem of linguistic change. Other problems
that prompted the development of variation analysis required solutions
whose analysis exceeded the theoretical and/or methodological tools of de¬
scriptive linguistics (e.g. free variation) or dialectology (e.g. urban dialects).
Discourse also presents problems for linguistic analysis; as we see below,
solutions to these problems are both inside (and outside) traditional lin¬
guistics.
We see in this section that variationists search for units in (and of) discourse
that bear systematic and patterned relationships to one another. After describ¬
ing the variationist approach to “units” in discourse (2.1) and the ways that
units have “relationships” with one another, including methods for analyzing
those relationships (2.2), I turn to a brief consideration of data collection (2.3).
We will see that despite its linguistic heritage, many aspects of the variationist
approach to language differ quite dramatically from traditional (and formalist)
linguistics. Section 2.4 summarizes.

2.1 Units in discourse

Labov (1972b; Labov and Waletsky 1967) provided a systematic framework

for the analysis of oral narrative - a framework that illustrates quite well the
variationist approach to discourse units. As I discuss in a moment, this frame¬
work defines a narrative as a particular bounded unit in discourse, and it
defines parts of narrative as smaller units whose identities are based on their
linguistic (syntactic, semantic) properties and on their role in the narrative.
Labov’s interest in narrative, however, arose not just out of a search for
discourse units per se, but also out of a set of social and political concerns
stemming from the notion of verbal deprivation:
284 Approaches to Discourse Analysis

The notion of verbal deprivation . . . leads its sponsors inevitably to the

hypothesis of the genetic inferiority of black children that it was origin¬
ally designed to avoid. The most useful service which linguists can per¬
form today is to clear away the illusion of verbal deprivation and to
provide a more adequate notion of the relations between standard and
nonstandard dialects. (Labov 1972a: 201-2)

Labov argues that the sort of verbal behavior used as evidence for verbal
deprivation is hardly due to a genetic deficit, but to regular sociolinguistic
factors operating upon a white adult and black child in an asymmetrical
interview situation (p. 207). The verbal behavior found in narratives of per¬
sonal experience, however, is quite different: far from being impoverished,
narratives reveal that “the black English vernacular is the vehicle of communica¬
tion used by some of the most talented and effective speakers of the English
language” (Labov 1972b: 396).
I have introduced narratives in this way not only to show one way that
variation analysis is socially realistic (recall our comment on p. 283), but also
to lead to an understanding of why narratives have provided so fruitful a
source of data for analysis: narratives are a discourse unit with a fairly regular
structure that is largely independent of how they are embedded in surrounding
talk. (Their methodological value is further noted in section 2.3.) Although I
describe part of this structure during our sample analysis (to provide a com¬
parison with lists (3.1)), I also do so here to highlight the variationist view of
discourse units.
Narratives have a linear structure in which different sections present differ¬
ent kinds of information. Each section has a different function within the
story. In addition, each section is comprised of different types of clauses whose
syntactic and semantic properties contribute to their identity as units within
the story, and to their function. Narratives are opened by an abstract, a clause
that summarizes the experience and presents a general proposition that the
narrative will expand. Orientation clauses (typically with stative predicates)
follow the abstract: they describe background information such as time, place,
and identity of characters. The main part of the narrative is comprised of
complicating action clauses. Each complicating action clause describes an
event - a bounded occurrence in time - that is understood to shift reference
time, i.e. it follows the event immediately preceding it, and precedes the event
immediately following it. Evaluation pervades the narrative: speakers can
comment on events from outside of the story world, suspend the action
through embedded orientation clauses, and report events that themselves in¬
dicate the significance of the experience. Speakers can also modify clause
syntax as a way of revealing the point (cf. general proposition) of the story.
Finally, the story is closed by a coda - a clause that shifts out of the past time
frame of the story to the time frame of current talk.
In addition to the narratives providing data in chapters 3 and 6, we saw an
example from a narrative in chapter 1, repeated here:
 Variation Analysis 285

One of the most dramatic danger-of-death stories was told by a retired

postman on the Lower East Side: his brother had stabbed him in the head
with a knife. He concludes:
And the doctor just says, “Just about this much more,” he says, “and
you’d a been dead.” (Labov 1972b: 387)

The utterance presented above is an example of an evaluation. I noted in

chapter 1 how this example is derived from the tools of linguistic analysis: not
only is it classified linguistically (as a “comparator” that is realized by differ¬
ent linguistic forms), but it is a labelled segment of a text that has a function
in relation to other sections. Despite this functional interdependence, vari-
ationist approaches allow a textual segment more autonomy relative to its
sequential location in a specific story than would be granted in explicitly
contextualizing approaches like interactional sociolinguistics (chapter 4) or the
ethnography of communication (chapter 5), or in more sequentially based
approaches like conversation analysis (chapter 7). As we noted in chapter 1,
an indication of this in the example above is Labov’s willingness to discuss
the evaluating clause as separate from the rest of the story - to treat it as an
example of a structural unit and functional type that can be extracted from
its story for comparisons with other evaluative devices.
Although narrative clauses can be extracted from their text and compared
for analytical purposes, units (clauses) within a particular narrative are also
defined in relation to one another. As noted above, for example, an abstract
is a generalization that prefigures the point of the story; a complicating action
clause is defined according to its temporal relation to an adjacent clause. A
particularly striking example of this text-internal dependency is that the func¬
tion of a clause can change depending upon its location in the story. A clause
such as “she was only eight years old” presented prior to the complicating
action would be a descriptive background clause (part of the orientation). The
same clause embedded in the complicating action would be evaluative -
serving to suspend the action prior to a point of resolution.
I have been suggesting that variationists view the structure of narrative as
largely independent of surrounding talk: narratives are autonomous textual
units whose internal parts stand in systematic relationships with one another.
Furthermore, as we note in section 2.2, narrative structure can create an
environment within which different linguistic forms are variable ways of
saying (or doing) the same thing. Although narratives are thus units that can
be studied apart from the immediate textual or interactional contexts in which
they are told, they are not independent of personal meanings and significance.
In fact, it is the personal value of an experience for the story teller that is
responsible for evaluative structure in narrative, and hence for the way that
narratives “command the total attention of an audience in a remarkable way,
creating a deep and attentive silence that is never found in academic or
political discussion” (Labov 1972b: 396). The telling of stories is also sensitive
to the social context (e.g. interactional setting, speaker identity) in which they
286 Approaches to Discourse Analysis

are told. Recall Labov’s argument that the notion of verbal deprivation is due
to regular sociolinguistic factors operating upon a white adult and black child
in an asymmetrical interview situation. Social context operates upon displays
of verbal competence in general - such that “an average middle class speaker
[may be] enmeshed in verbiage, the victim of sociolinguistic factors beyond
his control” (Labov 1972c: 214). Thus, although narrative structure is inde¬
pendent of surrounding talk, narratives are nevertheless personally meaningful
and their means of production is socially anchored.
Consider, also, that although narrative structure might seem to be found at
a relatively surface level of analysis, other texts have structure beyond what
is revealed in the utterances themselves. Analyses of apartment descriptions
(Linde and Labov 1975), for example, reveal that text structure can be mod¬
elled on spatial perceptions and physical structure. Although not often con¬
sidered part of variation analysis per se, Labov’s analyses of ritual insults in
Black English Vernacular (Labov 1972d), and of therapeutic discourse (Labov
and Fanshel 1977), are built on some of the same abstract premises as vari¬
ation analysis. The analyses just noted seek to be explicit about the speech act
relationships between spoken utterances. The premise underlying such ana¬
lyses is that coherence in interaction cannot be understood from the surface
linguistic form of utterances, but has to be found at an underlying level of
actions and reactions (see also Labov 1981). To this end, Labov and Fanshel
(1977) propose an extensive speech act analysis of discourse termed “compre¬
hensive discourse analysis.” The meaning of what is said is expanded with the
help of background knowledge, including social information about partici¬
pants’ rights and obligations: the result of such expansions is a set of proposi¬
tions (somewhat like underlying thematic statements). Propositions are, in
turn, related to speech actions: “how the speaker meant to affect the listener,
to move him, to cause him to respond, and so forth” (Labov and Fanshel 1977:
59). Mapping rules relate expanded meanings to utterances (“what is meant”
and “what is done” to “what is said”); sequencing rules connect abstract
speech actions (at the level of “what is done”).
The abstract speech actions focused upon by comprehensive discourse ana¬
lysis and the linguistic variables focused upon by Labovian variation analysis
(see below) seem very different at first. Labov proposes, however, that co¬
herence depends upon a complex hierarchical organization that “is plainly
derived from the linguistic analysis of phonology and grammar” (Labov and
Fanshel 1977: 350). Whereas linguistic variables form a set of closed options
(again, see below), however, speech actions do not - simply because of their
contextual dependence (Labov and Fanshel 1977: 358).
I have suggested in this section that variationists extend beliefs about lan¬
guage structure to the analysis of texts: the variationist approach to narratives
treats them as relatively autonomous textual units within which smaller parts
are linguistically defined and systematically related to one another. One way
that personal identity enters into the study of discourse units is in relation to
the evaluative structures used to convey the personal significance of an experi-
 Variation Analysis 287

ence. Social context is also part of the study of discourse units, since the setting
in which a story is told allows (or inhibits) the display of linguistic com¬
petence. Finally, social context enters into the construction of speech actions
and the formulation of speech act rules.

2.2 Variable relationships in discourse

Although the narrative is an example of a textual unit, it is important to note

that variationists typically focus upon other units of analysis that are not text
sized. As Labov (1972a: 188) observes:

It is common for a language to have many alternate ways of saying

. “the same” thing. Some words like car and tlutom<%ile seem to have the
same referents; others have two pronunciations, like working and wor-
kin’. There are syntactic options such as Who is he talking to? vs. To
whom is he talking? or It’s easy for him to talk vs. For him to talk is
easy.

One of the main tasks in variation analysis is to discover constraints on

alternative realizations of an underlying form: such constraints (that can be
linguistic and/or social) help determine which realization of a single underlying
representation appears in the surface form of the utterance. Consider, briefly,
the alternation noted above between working and workinmore precisely,
between the velar and apico-velar nasal. This alternation is constrained both
linguistically and socially. The use of in’ is more frequent in progressives and
participles than in gerunds - reflecting different patterns of historical develop¬
ment (Houston 1989). In’ is also more frequently used by working class
than middle class speakers and by men than women; all speakers use in’ more
than ing when they are speaking casually instead of carefully (Labov 1972c;
Trudgill 1974).
Since variationists try to discover patterns in the distribution of alternative
ways of saying the same thing, i.e. the social and linguistic constraints on
linguistic variation, an initial step in variationist studies is to establish which
forms alternate with one another and in which environments they can do so.
Constraints are then proposed in line with different hypotheses that may
propose a variety of different explanations for a given phenomenon, ranging
from the physiology of articulation to processing considerations to social or
biological universals. Variationist approaches compare different explanations
by searching for data that confirm (or cast doubt upon) the co-occurrences
predicted by each explanation. Although this is not, of course, a goal unique
to variationists, variationist approaches add the strengths (and limitations) of
quantitative analysis to such efforts. Quantitative analyses require definitions
of the variants (all the different possible realizations of an underlying type,
i.e. a closed set), a classification of factors in the environment with which
288 Approaches to Discourse Analysis

those variants may be associated, and a comparison of the frequencies and

probabilities with which different variants co-occur with different factors.
As noted above, the discovery of patterns in the distribution of linguistic
variants requires an obvious first step: establish which forms alternate with
one another (are “the same”) and in which environments they can do so.
Labov (1972a: 188) intends “sameness” to be defined narrowly “by some
criterion such as ‘having the same truth value’.” This restriction has made it
challenging (and, for some, theoretically problematic) to extend the notion of
variable to levels of analysis beyond the phonological. Morphological and
syntactic forms, for example, convey meanings that may or may not be neu¬
tralized in different discourse environments (Sankoff 1988). Forms that are
semantically distinctive in one environment may not seem so in another -
although it is difficult to say whether residues of those meanings remain with
each occasion of use and whether they are exploited for textual and/or prag¬
matic purposes (Lavendera 1984). Finally, wedding the notion of “semantic”
equivalence to truth value assumes a relatively narrow view of semantics that
not all variationists (e.g. Lavendera 1978) and not all linguists (e.g. Bolinger
1977) are eager to accept. The narrow view of semantics concerns a level of
referentiality associated with abstract sentence-types; the broader view of
semantics concerns a level of actual utterances in specific contexts (Silverstein
1976; cf. our discussion of formalist and functionalist definitions of discourse
in chapter 2). For reasons such as these, some variationists have suggested
replacing the requirement of semantic equivalence with functional equivalence
(cf. Myhill 1988). By the latter criterion, generalizing tags with obvious se¬
mantic differences - tags like and whatever, or something, etcetera - would
be considered variants (Dines 1980). As noted above, discussion of these issues
considers whether discourse neutralizes (Sankoff 1988) or builds upon (Laven¬
dera 1978) meaning-related differences in syntactic form. (See also Labov
1978; Lefebvre 1989; Romaine 1981.)
The search for variants in discourse inherits many of the problems of the
search for syntactic variants (e.g. Schiffrin 1985a; Vincent 1983). There are
additional problems. One stems from the status of discourse as a linguistic
unit per se. Texts do not seem to have the same kind of internally constrained,
externally bounded structure as sentences (Schiffrin 1988a). This makes it
difficult to define the theoretical status of discourse variants. They are not
alternative realizations of a single underlying form or representation (as one
could argue for phonological, morphological, or syntactic variants); nor do
they occupy a “slot” (a phonological segment, a syntactic constituent) in a
grammar or set of grammatical rules. Analyzing variation within a discourse
unit (such as the narrative, section 2.1) whose structure is relatively easy to
delineate partially obviates the need to make such a decision about grammat¬
ical status, i.e. one can analyze alternative forms that appear within specific
slots in a narrative structure.
Note, finally, that the way one analyzes distributions across social groups
in a speech community may also be altered by a reliance on functional equi-
 Variation Analysis 289
valents: the form that one segment of a community uses to realize a specific
function may not be identical to the form used by another segment of the
community. Thus, formally distinctive, but functionally equivalent, forms can
be differently distributed among speakers in a community (Sankoff and Thi-
bault 1981).
We have seen in this section that an important unit of analysis for vacation¬
ists is the linguistic variable. Incorporated into the definition of the variable
as a unit is an empirically grounded relationship between surface forms of
utterances: different ways of saying the same thing are grouped together as
alternative realizations of an underlying form only because they stand in an
“equivalence” relationship to one another (either a referential or functional
equivalence).

2.3 Locating vernacular speech in a community

Variationists assume that the social context of speech influences the use of
speech - so much so that very different judgements about verbal competence
(and community members’ judgements about social status) may be made
depending upon the particular sample of speech to which one has access (2.1).
Social context has a still deeper importance for variationist studies. Because
linguistic analysis is based upon what people say, rather than upon linguists’
intuitions of grammaticality and appropriateness (cf. the comparison between
formalist and functionalist paradigms in chapter 2), it is critical to have access
to a sample of speech that provides data adequate for analysis. When people
know that their language is being observed and/or judged, however, they may
alter their ways of speaking toward socially prestigious forms and/or toward
forms more like those of their interlocutors. What variationists thus seek is a
mode of speech called the vernacular. The vernacular is the variety used when
speakers pay minimum attention to speech (Labov 1984: 29; what Labov
1972e calls “casual style”); it is also that mode of speech acquired in pre-ado¬
lescent years (Labov 1984: 29). The highly regular character of the vernacular
is a result of the lack of attention to speech, and the fact that the linguistic
rules acquired early in life appear to be “more regular than those operating
in more formal ‘super-posed’ styles that are acquired later in life” (Labov
1984: 29).
In addition to needing speech representing the vernacular, variationists also
need speech representing a range of styles (because people often shift their
pronunciation of sounds that are undergoing change when they shift styles) as
well as a large volume of high-quality recorded speech from different members
of a speech community (to facilitate analysis of the social distribution of
linguistic forms). The technique traditionally used to collect data with these
qualities is the sociolinguistic interview (Labov 1984: 32-42). Sociolinguistic
interviews are a mixed genre of talk (chapter 5). One way in which they
differ from other interviews is that they encourage topic shifting and group
290 Approaches to Discourse Analysis

interactions among people present. Another difference is that respondents are

encouraged to tell narratives of personal experience. This is not only because
narratives reveal community norms and styles of personal interaction, but also
because speakers regularly shift toward the vernacular when telling a story
(Labov 1984: 32). Thus, narratives have another value for variationists apart
from their status as discourse units: the search for the vernacular within a
sociolinguistic interview is believed to be greatly facilitated when a respondent
tells a story.
In sum, variationists require data that allow the discovery of the highly
regular rules of language and the social distribution of variants governed by
those rules. This type of data - a variety of language termed the vernacular -
emerges only during certain social situations with certain interactional condi¬
tions. One such condition is when a speaker tells a narrative of personal
experience. Thus, the same discourse unit that is useful for variationists be¬
cause of its regular textual structure (2.1), and because it enables the definition
of environments in which to locate specific linguistic variants (2.2), is also
useful as a source of vernacular speech in which patterns of linguistic variation
and change may be discovered.

2.4 Summary

The variationist approach to discourse is based within a socially realistic

linguistics. In some ways, linguistics clearly pervades the variationist approach
to discourse. Although other approaches to discourse (e.g. conversation ana¬
lysis, chapter 7) also search for units and posit regular relationships (e.g.
sequential relationships) between units, for example, the variationist approach
defines those units in ways that incorporate their linguistic properties. Fur¬
thermore, a discourse unit such as the narrative is seen as more autonomous
from its immediate interactional context than would be the case for interac¬
tional, ethnographic, or conversation analytic approaches - approaches sharing
their disciplinary origins with sociology and anthropology. However, the vari¬
ationist approach also differs from traditional linguistics in many ways.
Variationists collect data to obtain samples of the vernacular; they never rely
just upon their intuitive judgements about grammaticality or use constructed
examples to test a hypothesis (but see Rickford 1990). Variationists use quan¬
titative methods of analysis to test hypotheses about constraints on the dis¬
tribution of forms within connected speech; these methods differ markedly
from those of formal linguists.
A variationist approach to discourse is thus a linguistic approach that con¬
siders social context under certain methodological and analytical circumstan¬
ces. Although a discourse unit such as the narrative is sensitive to the social
context in which it is told, and its point is grounded in the speakers’ values
and subjective experience, the structure of a story can be analyzed apart from
the way it is locally situated (cf. Schiffrin 1984b). Variationist analyses of
 Variation Analysis 291

discourse other than narratives show a similar separation between language

and social life - a reluctance to mesh them as two mutually constitutive
systems (cf. interactional sociolinguistics, ethnography of communication,
conversation analysis). For example, variationist analyses of two-party dis¬
course (in which speakers exchange turns) rely more upon considerations of
social meaning and context than do analyses of narratives. Although Labov
(1972e: 252) proposes that the “fundamental problem of discourse ana¬
lysis is to show how one utterance follows another in a rational, rule-
governed manner,” he also notes that “there is no formal basis in sentence-
grammar to explicate our reaction” to many everyday exchanges as coherent
and well formed (Labov 1972d: 299). In analyses of exchanges such as the
following:

A; Are you going to work tomorrow? A

B: I’m on jury duty.

Labov locates coherence at a level of actions and reactions, but he does so in

a way that adds social context to methods and concepts borrowed from
linguistic analysis. Labov (1972d) and Labov and Fanshel (1977: chapter 3),
for example, propose rules that connect actions and meanings to words. This
is a major task of discourse analysis and should be pursued in formal terms:
“Linguists should be able to contribute their skill and practice in formalization
to this study . . . formalization is a fruitful procedure even when it is wrong:
it sharpens our questions and promotes the search for answers” (Labov 1972d:
298). Variationist formalizations of discourse rules include social information
with linguistic primitives because discourse is an area “of linguistic analysis
in which even the first steps towards . . . rules cannot be taken unless the social
context of the speech event is considered” (Labov 1972e: 252). Thus, a
variationist approach to discourse is a linguistically based approach that adds
social context to analyses of the use of language.

3 Sample analysis: the “list” as text

The sample analysis in this section focuses upon a particular discourse unit,
the list. In section 2 we saw the importance of discourse units for variation
analysis: they reveal discourse structure, provide an environment in which to
define constraints on linguistic variants, and can be a site for the study of the
vernacular. Most variationist attention to discourse units has focused on
narratives (cf. Horvath 1987; Linde and Labov 1975). But it is also important
to find another type of discourse with regular forms and functions. My dis¬
cussion of lists suggests that lists are descriptive structures that center on
categories and category members (3.1). Although we have already discussed
292 Approaches to Discourse Analysis

the structure of narratives, I use narrative structure as a backdrop for the

discussion of lists. This is meant not only to elucidate the structure of lists,
but also to suggest that comparing textual structures (variation across text
types) is as much an application of variation analysis as is comparing forms
and structures within texts (variation within a text type). The former level of
comparison can contribute not only to what we know about text level vari¬
ation (e.g. the distribution of predicate types across clauses (3.1.3b)), but also,
even more fundamentally, to our understanding of how it is that we (as
members of a speech community) attribute to a particular string of utterances
an identity as an instance of one genre rather than another.
I also illustrate variation analysis by analyzing variation within a text type:
referring terms within lists (3.2). This is meant to illustrate the latter focus
noted above: the analysis of variants within a text type. Specific problems
pertaining to referring terms have already been posed in the sample pragmatic
(chapter 6) and conversation analyses (chapter 7). We see in this chapter how
variationists might address some of the same questions concerning the dis¬
tribution of definite/indefinite, and explicit/inexplicit referring terms (chapter
6), as well as more general questions concerning first-mentions and next-men¬
tions (chapter 7). To maintain continuity with the sample analyses in the
previous chapters, it would help to be able to begin our analysis with referring
terms themselves. However, the analysis of variation within text depends on
a prior understanding of that text itself: indeed, the formulation of text-
specific constraints (which is a key part of a variationist approach to referring
terms) requires this. Our discussion of lists themselves will also help us see
how the list is a discourse unit that is especially relevant to our understanding
of reference and referring expressions.
Before I begin, it is important to comment on the data to be used and the
methodology. Variationist studies of discourse face some of the same metho¬
dological requirements as variationist studies of phonology and syntax. As
noted earlier, such studies proceed by identifying variants and different fea¬
tures of their linguistic environment that are hypothesized to constrain (either
favor or disfavor) particular variants. These hypotheses are then supported (or
refuted) with the help of quantitative analysis. Each step of this process
requires a corpus whose size is sufficient to its goals. Not only does one need
to find “enough” of each variant, but the features used to categorize a lin¬
guistic environment cannot be so narrow that the environment is subdivided
into groups too small for meaningful comparison. The same holds for social
categories: studying the distribution of a phonological variant in a speech
community requires a selection of informants who adequately represent the
differences in social status and role (e.g. social class, gender, ethnicity, and so
on) that are meaningful in that community. One also needs a number of cases
sufficient for the application of statistical tests of significance. Unfortunately,
it is never simple to say how many cases are “enough.” Rather, the answer to
this question depends on the number of variants being analyzed and the
complexity of the constraints, i.e. the number of “factors” into which social
 Variation Analysis 293
and linguistic environments are divided. Regardless of exact numbers, the
point is that the method of analysis requires a relatively large corpus that
encompasses both linguistic and social heterogeneity. Many variationists have
found that such a corpus is best provided through sociolinguistic interviews
in a community (2.3; also chapter 5, section 3.3).
Finally, let me briefly look ahead to some general features of the variationist
approach that I will illustrate. As I noted above, I begin by comparing two
text types (lists and narratives): I then focus on variation in referring terms in
lists. These analyses reveal several assumptions that variationists make about
discourse. First, text is built from linguistic constituents that can have formal
and functional relations to one another. Second, although text is comprised
of smaller units, the coherence underlying the text is based on the way those
units are related to one another at other levels of analysis. Third, analyzing
the distribution of variants within texts is a way to di&over differences among
texts, as well as features of texts themselves. Finally, limiting the study of
linguistic variants to a particular text type whose structure is familiar allows
the relatively precise identification of textual constraints. It also permits a view
of sentence and discourse as interdependent systems.

3.1 Linguistic variation across text types

One of the most basic differences between narratives and lists is that narratives
tell what happened and lists describe a category. This difference bears on both
the content of what is said and its function in discourse. Before I discuss
several differences between narratives and lists that stem (at least partially)
from the fact that telling about an experience differs from describing a ca¬
tegory, however, it will be helpful to present an example of a list to show one
way that the form of a list can be explicitly related to the organization of a
category (3.1.1). I then discuss some specific linguistic differences between lists
and narratives stemming from the difference noted above: the unit of a list is
an entity (a “list-item”) rather than an event (3.1.2), lists build descriptive
rather than temporal and evaluative structures (3.1.3). Section 3.1.4 sum¬
marizes what is revealed here about the variationist approach to text structure:
text is built from linguistic constituents that have formal relations to one
another.

3.1.1 An example of category organization in a list

I suggested above that the focus of lists is on a descriptive category and its
members. Lists present a collection of items that can be classified as the
same” in some ways, but “different” in other ways. The sameness of the
items justifies their inclusion in a single class, and the differences between them
separate them into subclasses. Just as “sameness” is criterial to inclusion in
294 Approaches to Discourse Analysis

the category that unites a list, so too, are “differences” criterial to segmenta¬
tion into the subcategories that divide a list.
I begin with an example of a list that reveals, in an unusually explicit way,
the organization of the category being described. (1) is also a “good” example
of a list. Although variationist approaches to discourse also analyze texts that
are not “good” representatives of a particular text type, they often begin with
examples that display the full text structure in as clear a way as possible.
Specific differences that appear in other examples (e.g. different clause order¬
ing, elaboration or reduction of specific parts) are then seen not as aberrations,
but as motivated variations of the “basic” structure.
(1) is a list of “race tracks around here.” What is shared is the general
identity of the places named (each is a “race track around here”) and what is
different is the particular location (around here) and identity of each race
track.

(1) IVer: (a) Racing’s big around here, isn’t it?

IVee: (b) Yeh.
IVer: (c) Yeh.
IVee: (d) Well, you got uh, Jersey.
(e) You got. . . Monmouth
(f) and you got Garden State.
(g) Y’got Atlantic City:
IVer: (h) Mhm.
IVee: (i) And then uh here you got Liberty Bell.
(j) And they’re building a new one up in Neshaminy.
IVer: (k) That’s right. [I’ve never seen that, though.
IVee: (1) [And uh . . . you got Delaware.
(m) And of course, if you want to re- be- really go at it you can
go up to New York.=
IVer: (n) Mhm.
IVee: (o) =You got Aquaduct
(P) and you got Saratoga
(q) and you have that Belmont, y’know.
(r) I mean like uh . . .

The list in (1) is prompted when an IVer makes a general assertion (Racing’s
big around here) to which a tag isn’t it (requesting confirmation, or at least
hearer uptake) is appended. The IVee’s list of specific “race tracks around
here” can thus be seen as specifying (or supporting) the IVer’s generalization
through cumulative instances. In sequential terms, the list provides a response
(confirmation) to the prior request for confirmation.
Let us focus here on what (1) tells us about lists as linguistic reflections of
category members. In a moment, I will present a diagram in which the category
and subcategories are labelled in ways showing their hierarchical relationship
to one another. Although this step is not absolutely critical to variationist
 Variation Analysis 295
approaches to discourse, it reveals several variationist beliefs about text. First,
it suggests that just as narratives can reveal the speaker’s organization of a
world apart from language - a world of experience - so too can lists. Second,
it suggests that the order in which people present information (e.g. narrative
events, list-items) is an important source of information about text itself. This
order is assumed to be motivated in some way, and discovering that motiva¬
tion can help to reveal the source of underlying coherence. A related belief is
that texts have a structure beyond what is revealed in the utterances them¬
selves. Labov and Waletsky (1967), for example, diagrammed displacement
sets that were based on temporal relations between events, but also suggested
the functional identity of clauses. Similarly, Labov (1972d) and Labov and
Fanshel (1977) sought to be explicit about the speech act relationships be¬
tween spoken utterances: they argued that coherence in interaction cannot be
understood from the surface linguistic form of utterandbs, but has to be found
at an underlying level of actions and reactions.
Let us thus examine (T). This diagram labels the subcategories (list-items)
in the order in which they were presented in (1). It also shows their categorical
relationships to one another. On the far right are the line numbers from (1)
that correspond to each item.

(1') The structure of (1)

X Racing around here (a)
XI Racing in New Jersey (d-g)
Xla . .. Race tracks in New Jersey (e-g)
X2 Racing here [in Pennsylvania] (i, j)
X2a . .. Race tracks in Pennsylvania (i, j)
X3/X3A Racing in Delaware/race track in Delaware (1)
X4 Racing in New York (m-q)
X4a . .. Race tracks in New York (o-q)

Three levels of structure appear in (1'). The highest (most general, superordin¬
ate) is the category itself: “racing around here.” The next level of structure is
the states (New Jersey, Pennsylvania, Delaware, New York). The lowest level
of structure is the particular race tracks in the states. (Despite their difference
in level of subordination, both the states and the race tracks are list-items.)
(1) reveals an iconicity between the order of clauses in the text and category
structure. Note, first, that the order of items in (1) is iconic. As shown in (1'),
the order in which each race track is mentioned matches the hierarchical
relationships between items being listed: the category XI (“race tracks in New
Jersey”) is presented prior to the subcategories Xla (“Monmouth”), Xlb
(“Garden State”), and Xlc (“Atlantic City”). X2 (“race tracks in Pennsyl¬
vania”) and X4 (“race tracks in New York”) are similarly presented. The one
apparent exception is X3/X3a. From the way this category is presented (And
uh . . . you got Delaware) it is impossible to tell whether the speaker intends
to mention the state “Delaware” as X3 (cf. you got uh, Jersey as XI) or the
296 Approaches to Discourse Analysis

“race track in Delaware” as X3a (cf. you got Garden State as XIb). This
referential ambiguity, however, may in itself be iconic simply because there is
a race track in Delaware that is called Delaware race track. (In addition,
the “race track in Delaware” (X3a) is the only race track in the state of the
type (horse, not harness, racing) being listed here, such that X3 cannot be
instantiated beyond X3a.) Thus, just as you got is used to introduce two levels
of subcategory (XI or XIb), so too can the lexical item Delaware evoke a
state or a race track.
The way that items are introduced in the list is also iconic. Many of the
list-items are introduced in the same way (you got + ITEM in d, e, f, g, i, o,
p) regardless of their category level. Such repetition is iconic simply because
the introduction of different items through a single predicate structure is a
linguistic reflection of their coexistence in a common conceptual realm. Simi¬
larly, the means of introducing a list-item diverges from the you got pattern
when an atypical item is mentioned: a race track that has not yet been built
(And they’re building a new one up in Neshaminy) and race tracks in New
York further from the speaker’s home than the others (And of course, if you
want to re- be- really go at it you can go up to New York). The deviations
from the you got pattern in (1) may be seen as reflections of the peripheral
(versus core) status of these racetracks as examplars of the category “race
track around here”: the one not yet built is not yet a “race track”; the one in
New York is not quite “around here.” Note, also, that Belmont is introduced
in a way (you have and as that Belmont) that marks its special status (Belmont
is a nationally known race track, a place where major races, comparable to
the Kentucky Derby, are run). Thus, it is not only repetition itself that is iconic,
but the interplay between repetition and innovation: the list-items introduced
through one structure share a categorical status (as core members) and the
items introduced through different structures have different categorical statuses
(as peripheral members).
Although lists are iconic in the ways noted above, it is important to note
that not all aspects of lists are linguistic reflections of categorical organization.
For example, we might expect each separate item in a list to be syntactically
and/or intonationally packaged as a separate unit. We do find an example of
this in (1) when a higher-level list-item is mentioned separately from its
subcategories: XI (“New Jersey”) is introduced in its own independent clause
and its own intonation unit as Well, you got uh, Jersey (d). But there are also
counterexamples: X2 (“Pennsylvania”) is introduced pronominally as here in
the same clause and the same intonation unit as X2a (“Liberty Bell”), And
then uh here you got Liberty Bell (i).
(1) has illustrated one way that a list is a linguistic reflection of categories
and categorical organization. I turn now to a discussion of more general
characteristics of lists (contrasting them briefly with narratives) in order to
suggest that lists are a discourse level realization of a descriptive category in
which the cumulative presentation of category members (list-items) instan¬
tiates the general category.
 Variation Analysis 297

3.1.2 Units of lists are entities

In this section, I compare the basic units of lists (entities) to the basic units of
narratives (events). This comparison reflects yet another variationist assump¬
tion about text: text is comprised of smaller units that form a configuration
which is based on the way those units are related to one another.
Whereas the basic syntactic unit of a narrative is a clause with temporal
juncture, i.e. an event that moves reference time forward (2.3), the basic unit
of a list is an entity, i.e. anything of which something may be predicated
(individuals, sets, propositions; Webber 1979). (One type of entity may very
well be an event, and people may list events rather than tell about them in
stories (see 3.1.4c)). Evidence that lists are segmented into entities is that items
are introduced into lists in ways that focus on entitie^per se, rather than on
what may be predicated about those entities. One common way to introduce
an entity into a list is to use a right-branching construction in which seman¬
tically weak (and given) information precedes semantically new information
(the list-item itself). We already saw an example of this in (1) in which the
core members of the “race tracks around here” category were all introduced
with you got + ITEM; we also discussed the use of “there + BE + ITEM” for
the introduction of both first-mentions and second-mentions in lists in chapter
7. Constructions of this type include not only you got and there is, but other
possessive verbs (have, got) with either first person (I/we) or third person
(theyl(s)he) pronouns: such constructions focus upon the entity itself, rather
than a predication about the entity.
Another way to focus upon entities in a list is through ellipsis: the presenta¬
tion of an item with no surrounding information at all. Since it is the item
alone that is presented, ellipsis is a striking illustration of how lists reduce the
need to predicate something about each individual item: rather, any properties
of the item that need to be known to merit its inclusion in the list are assumed
(or presupposed) from shared knowledge about what category is being de¬
scribed.
The type of ellipsis allowed in lists differs sharply from that allowed in
narratives. Whereas narratives may very well have numerous intonation units
that lack NP subjects (regardless of whether we see these as verb chains
having undergone subject deletion or as conjoined verbs), lists may have
numerous intonation units that lack predicates. (2) illustrates two variants of
list ellipsis.

(2) IVer: (a) Well, was this a big Irish neighborhood when you were a
kid, [growin’ up? Really?
IVee: (b) No. [Just um
Irish, German, and Jewish.
IVer: (c) This neighborhood.
IVee: (d) Yeh, mhm.
298 Approaches to Discourse Analysis

(e) Irish, German, and Jewish.

IVer: (f) Mhm, yeh.
IVee: (g) In fact, there’s a Jewish woman the first house here.
IVer: (h) Been there a long time?
IVee: (i) She’s still there, yeh.
(j) Since I was a kid.
IVer: (k) Yeh?
IVee: (1) And then there was a Jewish family next door.
-Mm) Then us.
->(n) Then Jewish,
-Mo) then Jewish.
IVer: (p) What did they do?

(2') presents the structure of (2). Although I exclude then us (m) from ('), note
how an external physical structure (here, the physical structure of houses on
a block) can impact upon list structure (cf. Linde and Labov 1974).

(2') Structure of (2)

X Ethnic groups in the neighborhood
X1-[T1] A big Irish neighborhood [then] (a)
X1-[T1] Irish (b)
X2-[T1] German (b)
X3-[T1] Jewish (b)
X3a-[T2][T1] A specific Jewish woman [now] and [then] (g)
X3b-[T1] Jewish family (1)
X3c-[T1] Jewish (n)
X3d-[T1] Jewish (o)

[Tl] and [T2] are time indices

Two types of ellipsis are illustrated in (2). The ellipsis in Irish, German, and
Jewish (b, e) is very typical of short three-part) lists in which entities are
themselves short (NPs) and not elaborated upon. What is important for our
purposes here is that nothing is predicated of the entities themselves: rather,
their mention is sufficient to their inclusion as members of a category. The
second type of ellipsis illustrated in (2) is less typical. Then + ITEM (then
Jewish (n, o) is potentially extendable into lists that include more than three
items, and that allow a great deal to be said about each item after its introduc¬
tion into the list. Here the entity being listed is prefaced by a sequential marker
that marks the progression of the items through discourse time (see (3.1.3a)),
and perhaps also the progression of houses down the block (see Linde and
Labov 1975, on tours). Although these forms of ellipsis differ, the point here
is that they both allow mention of the entity itself, with no supporting proposi¬
tion, and nothing overtly predicated about the entity. Furthermore, we under¬
stand the entities to be category members even without overt indications of
their status in relation to the category.
 Variation Analysis 299

In sum, the basic unit of a list is an entity. Evidence for this is the way entities
are introduced into lists, either by constructions that syntactically and seman¬
tically focus upon an entity or by elliptical mentions that allow a focus upon
the entity as a “next item” without any overt predication about that entity.

3.1.3 Information structures

This section continues the variationist approach to text, but rather than focus
upon the basic building blocks of a text type (events or entities: 3.1.2), I
focus upon the information structures that are formed through clauses in the
text. We noted earlier that the focus of lists is on a descriptive category and
its members: lists present a collection of items that can be classified as “the
same” in some ways, but “different” in other ways. AltHbugh there are numer¬
ous analyses of spoken narrative from both linguistic and literary perspectives
(e.g. Chatham 1978; Fleischman 1990; Sheldon 1988), virtually all scholars
agree that narrative is concerned with experience. To oversimplify, narratives
tell about something that happened to someone. The experiences upon which
narratives focus can include a circumstance that the narrator himself or herself
had an active role in bringing about or a problem for which the narrator
sought a resolution. Thus, narratives frequently “tell about” what happened
to a human being - a human who can take an agentive role in bringing about
a particular change in circumstance.
This general difference between lists and narratives extends to the informa¬
tion structures that prevail in these text types. We see in this section that
narratives and lists both build two types of information structures - temporal
(3.1.3a), and descriptive (3.1.3b) - although the text types do differ in terms
of which information structure is dominant. The speaker’s subjective involve¬
ment in recounting an experience, however, means that narratives also build
evaluative structures (3.1.3c).
One way to see how lists and narratives rely upon different information
structures is to compare the function of different linguistic items in both struc¬
tures. Notice that this reveals another variationist assumption about discourse
analysis (one to be pursued at length in section 3.2): syntactic and semantic
differences among linguistic items are sensitive to text structure. In addition
to analyzing such differences within one text type (as we do in section 3.2),
one may analyze such differences across text types (as we do here). The
direction pursued here is also consistent with the view of text that we noted
earlier: text is a structure by virtue of its particular configurations of “lower
level” linguistic forms. Thus, we can discover differences among text types by
examining how linguistic forms themselves fit certain distributional patterns.

a Temporal structure
The Labovian view of narrative defines temporal structure as the criterion
central to the definition of narrative (2.3): temporal information (what
300 Approaches to Discourse Analysis

occurred when) is inferred largely from the sequential presentation of two

event clauses in discourse. Put another way, the interpretation of narrative
rests largely upon the assignment of reference time (the time period in relation
to which something can be assumed to have happened; Reichenbach 1947) to
events: it is the linear presentation of event clauses in narrative that typically
allows one to assume that reference time shifts from one event to the next (but
see Dry 1983; Ehrlich 1990; Fleischman 1990). The temporal structures of
lists, however, are not typically structures that are concerned with reference
time. Rather, the time relations in lists are textual relations between items
presented successively in the discourse itself, i.e. a matter of textual rather
than propositional structure.
We noted above that one way to see the difference between the two text
types is to compare the function of different linguistic items in both structures.
Then in narrative is linked primarily to reference time (even when it marks
global changes in episode; Segal et al. 1991). (3) illustrates. (I have chosen a
stative clause from narrative, simply because stative predicates are so common
in lists.)

(3) (a) And then, and uh, you rode twenty miles one way
—> (b) and then uh there was an old deserted ranch house there
(c) and that’s where I- we used to meet the other fellow cornin’ from
the other direction.

Rode twenty miles in (a) is an achievement predicate (Vendler 1967); was in

(b) is a state (part of an existential predicate, used frequently in lists; see pp.
302-4). Although “an old deserted ranch house” is assumed to exist prior to
the speaker’s mention of it, we interpret its existence in relation to the speaker’s
prior action, i.e. we infer that he observed the house only after he rode twenty
miles (the distance noted in (a)). In this sense there was an old deserted ranch
house is almost like an achievement, as if the speaker had noticed the house,
or realized that it was there, and is conveying this observation to the hearer
(cf. Dowty 1986: 50). Then plays a central role in evoking this interpretation:
then provides both an ending point for the achievement and a starting point
for the state. Note, also, that the clause position of then in narrative clauses
has a dramatic effect on reference time, such that clause-final then in (3b)
would continue rather than shift reference time (Schiffrin 1990b, 1992b).
The use of then in lists is quite different. Then in lists is always clause-initial,
a restriction consistent with its function. That is, the clause-initial position of
then reflects its use as a discourse level marker of temporal succession that is
based on the S-occurrence of then (in initial position) rather than the VP-
occurrence (in final position; Ernst 1983; Schiffrin 1992b). Then in lists has
temporal meaning only as a discourse marker (Schiffrin 1987a: chapter 8): it
shows progression not through reference time (events, states), but through
discourse time. Then in (4) illustrates. A sociolinguistic interviewer has asked
whether any policemen live in the respondent’s neighborhood.
 Variation Analysis 301

(4) There’s a cop up the next block.

There used- there was a cop over here.
He’s retired.

The speaker begins by listing two members of the set “policemen around
here.” Following the two list-items, the speaker describes policework in
general - a section not included above - and then adds another entity to the
list:

Then uh, there are a couple uniform cops around here somewhere.

Then establishes a return to the list. Despite a fairly extensive description

pertinent to the more general topic under discussion (that is not part of the
list per se), a couple uniform cops is clearly the next (1st item. Note that the
use of the stative predicate “be” with each list-item establishes a descriptive
structure (3.1.3b). In sharp contrast to then in the narrative (then uh there
was an old deserted ranch house there (3b)), however, then with stative
predicates in lists does not shift reference time. Then in lists has meaning only
as a discourse marker that marks the sequential progression of items through
discourse time.
In sum, the temporal structure of a narrative is central to its identity as a
discourse unit and to its semantic interpretation; temporal structure in the
propositional sense, however, has little relevance to the structure of lists.
An indication of this difference is that then in narrative is a marker of refer¬
ence time whose clause position can alter our interpretation of what hap¬
pened when; then in lists is a marker of text sequence, i.e. the “next item in
the list.”

b Descriptive structure
Another informational difference between narratives and lists concerns their
descriptive structures. Description in narratives is typically assigned to a back¬
ground “orientation” function (Labov 1972b). The identification of who was
present, where something happened, and when it happened often prefaces the
narrative action itself. Such information is typically assumed to endure
throughout the time of the story world, such that changes in background states
need to be explicitly noted.
Narrative orientation may also be embedded within the complicating action,
and here it may have a descriptive and/or evaluative function. In (5), for
example, the speaker interrupts his narration to describe physical features of
a location. (5 presents the clauses prior to the story excerpt in 3).

(5) —> (a) See, every five miles they had a little stone.
(b) Say on one side was Mexico and the other side was the United
States. Y’know?
(c) And then, and uh, you rode twenty miles one way
302 Approaches to Discourse Analysis

Clauses (a) and (b) are restricted background clauses (Labov 1972b): static
descriptive situations relevant for a limited period of time in the reported
experience. The states in (5) also serve as landmarks (Psathas 1979): they are
locations that the speaker uses as a base from which to orient a next action
(in c). Thus, the embedded orientation clauses in (5) have a limited relevance
in the story: they are pivotal locations for a next-action that returns to the
temporal structure of the story. Although not illustrated here, I noted that
some embedded orientation clauses have an evaluative function in narrative
(3.1.3c). Thus, although the information itself may be a static description (i.e.
a stative predicate that does not move forward in reference time), it can also
have a secondary function of conveying the significance of certain events for
the point of the story.
Description in lists plays a very different role. Since lists are themselves text
level category descriptions, it is not surprising that stative predicates (such as
“have” (5a) and “be” (5b)) are the main verb types used to build the descrip¬
tive structure of a list. What such predicates provide are just what lists need:
descriptions of situations that are assumed to hold for an unspecified period
of time (with no beginning or ending points), or items that are assumed to
have stable conceptual identities. Put another way, part of what makes a text
into a list is the use of predicates representing situations that can serve as
continuing instantiations of a given category.
Consistent with the different functions of description in the two text types
is a different pattern of active and stative predicates. As I noted above, stative
predicates in narrative either precede active predicates (we see a specific
example below, as 8) or are inserted within a sequence of active predicates (as
we saw in 3). The distribution of active and stative predicates in lists, however,
is quite different.
(6) is a section of an example (seen in full as 8 in chapter 7). An IVer asks
some teenage friends (two females, one male) whether there (are) any old folks
livin’ on this block. The section in (6) reverses the distribution of active and
stative predicates typical in narratives: rather than a stative or two inserted
into a sequence of active predicates which comprises the complicating action,
the statives in (6) provide the skeletal organization of the list, with active
predicates adding supplementary information about each item.

IVee-2: (r) Then there’s one woman here, crippled or

whatever,=
IVer: (s) ZMhm.
IVee-2: (t) =you want to call her. She’s-
IVee-3: (u) Delirious.
(v) She just walks around the street with a walker?=
IVer: (w) Mhm.
IVee-3: (X) =She’ll stop any car cornin’ down the street, too.
IVee-2: (y) (No,) [hh she’s not. Hh.
IVee-3: (z) [Tell you her whole life history.
 Variation Analysis 303

IVee-2: (aa) She gets you to come in her house and-=

IVee-3: (bb) Fix her ( ).
IVee-2: (cc) =read her mail for her and-
IVer: (dd) Fih, oh, bless her heart.
(ee) She really sounds lonely.
—> IVee-2: (ff) And then there’s a- another couple up the street
(gg) who never come out.=
IVer: (hh) Uh huh.
IVee-2: (ii) =And, God forbid you parked your car in their pavement
(jj) and they’re- they’re out there ready to throw bricks at
you
(kk) and, they call the cops . . .

Two list-items are introduced here with stative" predates: Then there’s one
woman here, crippled or whatever (r) And then there’s a- another couple up
the street who never come out (ff). Following its introduction into the list,
each referent is described with active predicates (v, x, z, aa, cc; ii-kk). Note,
also, that the two active predicates (parked your car (ii), call the cops (kk)),
and the stative predicate (they’re out there (jj)) form a short temporal sequence
embedded within the list structure.
(7) is a slightly more complicated example: the speaker presents two inter¬
secting lists that are differentiated by predicate type. An IVee (an elderly
woman) is answering a question from an IVer about where a young man would
have taken a young lady when they were dating long ago. The IVee begins her
response by saying that there were not many places to go (because the De¬
pression reduced everyone’s income). What she presents, instead, is a list that
is partially responsive to the focus of the IVer’s question (a list of “places”),
but also an intersecting list of “what dates used to do” (a list of “activities”).
We see that the former list uses stative predicates, and the latter list active
predicates.

(7) IVee: (a) We did a lot of walking. /Hh./

IVer: You could go a long way walkin’, no kidding.
IVee: Yeh. Mhm.
IVer: Walkin’ and talkin’ I bet.
IVee: Yeh. That’s all you could do.
I mean uh he, uh he wasn’t workin’ at the time.
And to get a job I mean, oh, it was- it was awful.
IVer: Was it very difficult?
IVer: Yeh, terrible.
(b) And uh, every once in a while we’d stop and get a soda,
y’know, or something like that.
-> (c) And then- they- they always had these little urn ice cream
parlors around, /Uh huh/ which they don’t have today.
(d) You go in, y’know,
304 Approaches to Discourse Analysis

(e) and you sit at a table

(f) and- and uh you can get yourself a soda or a dish of ice cream
or something like that, y’know.
—> (g) In fact we even had a bakery.
(h) You can go in there
—> (i) and they had tables.
-> (j) And they had the homemade ice cream.
(k) You could get a big dish of ice cream for a dime.

After a short description of activities (“walking” (a), “stop and get a soda”
(b)), the IVee introduces a place (“ice cream parlors” (c)) that has a pivotal
role in the list. Note that immediately prior to ice cream parlors, the speaker
had self-corrected from then. This self-correction suggests that the speaker was
about to continue the list of activities from (a) and (b): in fact, she does return
to the activities after specifying a location (“ice cream parlors”) for one of
those activities. But the specification of “place” (they always had these little
um ice cream parlors around (c)) has a dual role in the list. As already noted,
mention of “ice cream parlors” provides the locational frame in which the IVee
can elaborate a previous item (“stop and get a soda” (b)) from the “activity”
list initiated in (a). This elaboration is facilitated by repetition: the repetition
in (f) of get a soda ... or something like that from (b) ties the two mentions
of the “activities” together; a dish of ice cream (f) also ties the “activity” to
the specific location ice cream parlor (c). But mention of “ice cream parlors”
also provides a first-item in a list of “places” (that continues with “a bakery”
(g)). Thus, by initiating a list of “places”, “ice cream parlors” also shifts focus
back to the IVer’s question about where people would go on dates.
Note that the distribution of predicate types in (7) helps to separate the
different intersecting lists from one another. Not surprisingly, the items in the
“activities” list are all event clauses with active predicates (“walk” (a), “stop
and get” (b), “go in” (d, h), “sit” (e), “get” (k)). The items in the “place” list
(and in a subordinate “things in the place” list) are all statives (all are “have”
(c, g, i, j,). Thus (7) shows, again, how the distribution of predicate types in
lists differs from that in narratives: whereas only active predicates can com¬
prise a narrative, either active or stative predicates can comprise a list.
Still another way to see the difference between description in narratives and
lists is to contrast the use of the same syntactic construction in both types:
existential there constructions (chapter 7). Example (8) (already seen as 11 in
chapter 7) shows the use of an existential predicate in the orientation to a
narrative.

(8) IVee: (a) And this is what the thing is.

(b) They get in,
(c) and i- what’s it- what’s it bein’ done for?
IVer: (d) Yeh.
(e) That’s right.
 Variation Analysis 305

Wee: (f) There- I’ll tell you.

—> (g) There’s a school down in my mother’s town, right?
IVer: (h) Right.
Wee: (i) Strictly colored.
(i) It’s the newest school in North Beach.
(k) It was all colored.
(1) And because of this law of integrating,
(m) these colored children had to go to this white school,
(n) which they didn’t want.

“There’s + ITEM” in (b) is used in a turn-position in which the referent

focused upon is guaranteed a slot in which to be described: prior to the
introduction of a school in (b), the speaker self-repaired from there to a
pre-announcement I’ll tell you that secures the* floor for her story. The tag
right? also makes explicit that the speaker is seeking recognition of this item
as preliminary to its further use in the story. Although the school is described
in more detail (i-k) prior to the story, the school does not figure in the
narrative action itself: rather it remains background information in the story
that allows the speaker to talk about what happened (i.e. the narrative action)
to the children who attended the school.
The background function of existential there in narratives can also be seen
when this construction serves as an embedded orientation. (9) is a section of
a long narrative in which a cabdriver tells about being robbed (the narrative
discussed in chapter 6). In the section of the story reproduced here, the speaker
interrupts the complicating action to describe the destination to which he
is sent.

(9) (1) So it was maybe only three blocks away,

(m) so they said- I think they gave me an address that’s at the center of
the block,
_> (n) and there was a small light on at the house.
(o) So I pulled up,
(p) I turned on the light,
(q) and the guy he grabbed me from the back of my head
(r) and put a gun to it.

Lines (1) through (n) present material that is backgrounded to the complicating
action. In (n), there is introduces one particular feature of the destination being
described, a small light on at the house. The light introduced in (n), however,
has no further role in the story. In fact, when the speaker returns to the
complicating action of his story, he turns on a different light (I turned on the
light (p)) - the light in his cab - after reaching the destination (o). It is
surprising that the same lexical item “light” appears in two different clauses.
Not only is it used with different referential intentions, but there is often
introduces an item that is then treated as an accessible referent (e.g. conveyed
306 Approaches to Discourse Analysis

as a definite) in subsequent clauses (see chapter 7). But it is precisely because

this sequence is unexpected that it is so instructive. What is introduced by the
existential predicate is in a backgrounded descriptive section of the story - a
section structurally and functionally segmented from the complicating action
that does not play an important role in the story.
There is has a much different role in lists than it does in narratives. In (10),
an IVer asks about children’s relationship with an absent father.

(10) IVer: (a) Have any of the kids ever expressed a desire to see him?
IVee: (b) No.
(c) No, they never did.
(d) I mean like uh, they never asked about daddy or where did
daddy go or-
-Me) I mean like, there was me
->(f) and there was my father
-♦(g) and there was my cousins.
(h) And the kids were pretty well occupied.

The list of items in (10) begins in (e) with there was me, continues in (f) with
there was my father and ends in (g) with there was my cousins. All the items
in the list are definite descriptions (see 3.2.2c). The same kind of structure -
the same string of existentials - occurs with indefinites.

(11) (a) There’s uh, there’s string bands,

(b) there’s brigades,
(c) there’s comic divisions,
(d) there’s- y’know, there’s quite a few divisions.

The existential predicates in (10) and (11) contrast sharply with the existent¬
ials in the narratives in (8) and (9). Rather than being used to introduce
background (a character, a place, etc.) prior to a text, the entities introduced
by there is are central to the text.
We have seen thus far that narratives build temporal structures and lists
build descriptive structures; use of the temporal marker then, and the distribu¬
tion of active versus stative predicates, reflect these different structures. We
see in 3.1.3c that there is still another difference between narratives and lists:
narratives do not just report, they also evaluate, experience.

c Evaluative structure
Narratives and lists differ in the degree to which human subjectivity guides
the productive processes underlying the speech activity itself. Stories are often
said to be constructions of an experience, rather than representations of a
reality (e.g. Bruner 1990). A story teller’s active involvement in the crea¬
tion of an experience pervades the story telling process: speakers may impose
their own subjectivity on what happens at a variety of levels, e.g. what to
 Variation Analysis 307
report (or leave out), the relevance (or irrelevance) of specific occurrences,
temporal and causal relations between events, and so on (see e.g. Bamberg
1991; Bamberg and Marchman 1991; Slobin 1991, for cross-linguistic views
at the syntactic level). The telling of a story is also intersubjectively guided. A
story is a reconstruction of an experience, told at a specific time, in a specific
place, to a specific audience for whom the storyteller seeks to demonstrate the
validity of a general claim (e.g. about oneself, one’s experience, or the world).
Put another way, a story seeks to establish an intersubjectively agreed upon
point (a point which may involve the audience as co-author: Duranti 1988;
Ochs et al. 1991).
Members of a category seem slightly less immune to individual subjectivity:
some categories may even have members whose qualities justify their inclusion
in the category independent of an “observer’s” subjective view of those
qualities. Rosch (1978), for example, emphasizes that categorization is struc¬
tured according to perceived structure and function in the world: naturally
occurring groups of attributes may be non-arbitrarily related to particular
functions. Rosch also points out that categorization is structured by a principle
of cognitive economy that defines a “basic level” of categorization whose
concepts are among the first to be learned, the most commonly used, and are
labelled with short, structurally simple words (see also Berlin et al. 1974;
Berlin 1976; Downing 1980; Rosch 1973; Rosch et al. 1976). This is not to
say, however, that categorization is totally “independent of who is doing
the categorizing and on what basis” (Lakoff 1987: 50). The specific details of
categorization can vary across individuals, subgroups, and cultures, and can
change over time (Kay 1975; MacLaury 1991); some studies also find that
different strategies are used to place objects into categories (e.g. Reed 1972).
Despite such variation, however, categorization seems to be a less subjectively
guided process than narration: the involvement of the self seems less likely to
pervade the “logical” processes underlying category membership than the
“narrative” processes underlying memory of experience (cf. Bruner’s 1986 use
of the terms paradigmatic versus imaginative).
Speakers’ involvement in the reconstruction of what happened has an im¬
portant consequence for the language used in story telling; so, too, the relative
lack of involvement in category reconstruction has a consequence for the
language used in list making. I suggested above that subjective involvement is
geared partially toward making a point through a story. The point of a story
is indicated through what Labov calls evaluation (2.3). Evaluative structures
in stories are often deviations from textual norms (Polanyi 1979), e.g. from
simple clausal syntax (Labov 1972b) or from unmarked tense forms (Fleisch-
man 1990; Schiffrin 1981a; Silva-Corvalan 1983; Wolfson 1979), but a variety
of other means are also used for expressive purposes, e.g. prosody, repetition,
and direct quotes (see also Bauman 1986). Thus, evaluation pervades not only
the processes whereby “what happens” becomes part of one s experience, but
also the processes whereby an experience is situated in the here and now,
the “why,” and the “to whom” of its telling.
308 Approaches to Discourse Analysis

Although lists are also situated and can also be used to make a point, the
evaluation of items in lists is fundamentally different from the evaluation of
events in narratives. Evaluation is central to the task of reconstructing an
experience and to the task of embedding a story in an interaction. Although
evaluation in lists may very well figure in the interactive function of the list,
it is less crucial to the internal construction of a list per se. More generally,
evaluation is required and pervasive in a story (e.g. Polanyi 1979, 1985), but
optional and minimal in lists. These points are important because when the
items in a list are proposition-sized (e.g. event clauses), a list seems more like
a narrative than we might expect.
The next examples have two purposes: to show how lists themselves (not
their specific items) can have an evaluative function in discourse (cf. Johnstone
1983; Rains 1992); and to differentiate lists of events from narratives.
(12) is a relatively simple example showing that list-items may be enumer¬
ated to make a point. The speaker (a 70-year-old woman) has been comparing
the relatively easy housework of nowadays to the time when she was a
housewife and found housework to be demanding.

(12) (a) But to me, eh: I always found something to do.

(b) There’s always something to do in a home where there is a family.
/Urn hm./
(c) There is such things as closets to clean out
(d) and drawers to clean out
(e) and curtains to sew
(f) and windows to do
(g) and, and dishes to do
(h) and cooking to do
(i) and you t- name it
(j) and scrubbin’ to do
(k) there’s always is constantly something to be done if you want a
nice, clean home.

The accumulation of items (of “things to do”) in (12) provides evidence for
the general claim that there’s always something to do in a home (b): each
list-item provides a specific instantiation of the indefinite NP something to do
(b). None of the items are highlighted as more central to the general experience
of housekeeping than any others: rather, it is the fact that such tasks can be
enumerated without differentiation that provides the constancy underlying the
claim that there is always something to do.
(13) is another list in which the speaker supports the validity of a general
claim by enumerating specific instances. It is slightly more complicated than
(12), however, because each specific instance is itself evaluated as a way of
legitimizing it as support for the claim. Prior to (13), the speaker had been
complaining that her younger daughter-in-law does not call her “Mom.” The
speaker states that the only solution to this problem is to start in the beginning
 Variation Analysis 309

(of the in-law relationship (a)), and she presents evidence that this is the
appropriate solution.

(13) (a) And you have to start in the beginning.

(b) Now my daughter in law did.
(c) My older daughter in law from the very beginning she said Mom,
(d) so she’s used to it.
(e) Mom and Dad.
(f) See it does m- it’s only a name!
(g) And Ron- we told Ronny too,
(h) in the beginning, you call- if you can’t say Mom and Dad, call
them by their first name.
(i) But call them something.
. (i) Not “uh:::::” a *
(k) And she’s an intelligent girl
l
( ) and she’s a nice girl.
(m) She gives us the biggest respect.
(o) And she’s really nice.
(P) I like her very much.
(q) It’s just that she can’t say it.
(r) Now I remember when I first got married,
(s) and I was in that situation.
(t) And eh the first- like the first... few times, I wouldn’t say
anything.
(u) And my husband said to me “now look it isn t hard, just say
‘Mom’.”
(v) He says, “And I want you to do it.”
(w) And I did it.
(X) And I got used to it.

The list-items in (13) are the specific people who adhered to the general rule
you have to start in the beginning (a): “older daughter in law” (b to f),
“Ronny” (the speaker’s son (g to j)), and the speaker herself (r to x). We can
summarize the members of the list as follows:

(13') The categories in (13)

X “people who did the right thing”
XI “my older daughter in law” (b-f)
X2 “my son Ronny” (g-j)
X3 “myself” (r”x)

Although the items above constitute the categories in the list itself, there is
clearly more in the discourse than just the three items in (13'). In addition to
defending the person who has done the wrong thing (the daughter in law who
does not call the speaker “Mom” (k to q)), the speaker establishes for each
310 Approaches to Discourse Analysis

list-item the validity of its membership in the list. One way that she does so
is to tie each list-item to the general claim through repetition or paraphrase
of a critical part of the claim. In (a), the speaker states that the behavior has
to occur in the beginning of the in-law relationship. The behavior of each
person mentioned is indexed to the appropriate time: the older daughter-in-
law to the very beginning (c), the son to the beginning (h), the speaker herself
to when she first got married (r). People in the list are also presented in ways
that highlight their role either in the comparison with the daughter-in-law who
behaved inappropriately, or in the collection of people who behaved appropri¬
ately. The older daughter-in-law’s behavior (b to f) contrasts with that of the
younger one. This is shown through contrastive stress (did (b)) and ellipsis
(“did [call me ‘mom’]”). Ronny (g to j) and the speaker herself (r to x) are
additions to the descriptive set being built in support for the claim about
appropriate behavior. This is shown through contrastive stress on too (g) when
Ronny is first mentioned, and on / (s) when the speaker mentions herself. Still
another way that the speaker in (13) shows that the list-items support her
general claim is the use of devices that are similar to narrative devices in both
form and function. The speaker uses, for example, a meta-comment that
functions like an external evaluation (See it does m-it’s only a name! (f)) of
the older daughter-in-law, reported speech (to the son (h to j) and from the
speaker’s husband (u, v)) that functions like an internal evaluation, and re¬
ported actions (w, x) that are temporally sequenced in relation to the con¬
structed dialogue.
Before going on to more general points about the role of evaluation in lists
and narratives, it is important to summarize our discussion of (13). (13) is a
list in which the accumulation of items substantiates a claim about the world.
The list-items mentioned (“older daughter-in-law”, “Ronny”, “I”) would be
part of the list with or without the devices just discussed, i.e. repetition,
contrastive stress, meta-comments, and reported actions. It is in this sense that
evaluation in the list in (13) is not internal to the construction of the list itself.
Rather, what evaluative devices do in (13) is legitimate the use of each specific
instance as support for the general proposition being put forth in the inter¬
action: they are part of the ongoing construction of the list as evidence for the
speaker’s claim.
Thus far we have treated the list-items in (12) and (13) as entities that are
no different from the entities considered in earlier examples: (12) is a list of
“things to do,” (13) is a list of “people.” There is another way, however, to
consider the entities in these lists. Since the basis for the inclusion of “people”
in the list in (13) is their behavior, for example, we might categorize (13) in
terms of propositions. Each list-item could be said to be an event clause in
which the predicate “called mother-in-law ‘Mom’ ” has a succession of differ¬
ent arguments (“older daughter-in-law”, “Ronny”, “I”). Rather than be a list
of people as in (13'), then, (13) would be a list of propositions about people.
Similarly, the list-items in (12) can also be recast as propositions, e.g. as
instantiations of stative predicates “there is X to do.”
 Variation Analysis 311

When list-items are syntactically and semantically defined as clause-sized

propositions, rather than as NP-sized referents, temporal order can play a
greater role in the construction of lists than previously indicated. This is
because events that do follow a particular temporal trajectory in the speaker’s
experience may very well be presented in terms of that trajectory. When this
happens, a list of events might seem very much like a story. But as I suggest
in the next examples, it is only when those events are evaluated that we begin
to interpret the discourse as story-like.
In (14), an IVer asks an IVee about his first job. This descriptor becomes a
subcategory in a list of ‘jobs’ in which the jobs are temporally ordered. I
include time indices in the left margin.

(14) IVer: Well, hey, do you remember your first job?

The first job you had? a
IVee: (a) The first job.
[T(la)] (b) I think my first job,
(c) one of my first,
(d) I must’ve been about ten years old.
XI (e) I used to stock the shelves in a grocery store.
(f) I used to work eight hours on a Saturday for two dollars.
Hh.
IVer: Oh, what slave labor.
IVee: Yeh.
[T( lb)] (g) Well, I think as I got older,
X2 (h) I was a janitor in a cleaners.
(i) Not any- well, AJh huh./ I cleaned up
(!) and made hangers up for the-
(k) put the plastic on the hangers
(l)
and the cardboard on the hangers
(m) and the paper around the hangers you put in your suit
jackets.
[T( 1)] IVer: Then, that’s when- you were a kid, [right?
IVee: [Yeh.
[T(2)] IVer: And, so, did you work when you were in high school?
X3 IVee: (n) Well, I was a short order cook in a hoagie shop.
IVer: Oh, really?
Tesus, you’ve had a checkered career, haven’t you?
IVee: Hh. /Hh./
X4 (o) I was- I was a busboy at a country club while I was in
school.
I was a dishwasher, /Uh huh/ for a pancake house.
X5 (p)
(q) They had dishwasher
(r) then you were a busboy. /Yeh./
(s) Half the night you wash dishes,
(t) the other half you cleaned the tables up.
312 Approaches to Discourse Analysis

IVer: (u) Did you like bein’ a short order cook at a hoagie-
factory?

Most of the jobs are formulated as stative predicates, i.e. I was + “job title”:
I was a janitor in a cleaners (X2 in h), I was a short order cook in a hoagie
shop (X3 in n), I was-I was a busboy at a country club while I was in school
(X4 in o), I was a dishwasher, for a pancake house (X5 in p). The one job
not formulated as a stative is XI: I used to stock the shelves in a grocery store
(e). This deviation from the pattern reflects the non-prototypical nature of the
job: not only is there no conventionally used nominal (e.g. shelf Stocker) for
this job, but the speaker stocked shelves as a child (a period in life during
which people do not have what we think of as a “real” job). Underlying the
presentation of jobs is a collaboratively built temporal organization - an
organization that is all the more notable because it involves shifts in reference
time between stative predicates (see Dry 1983; Schiffrin 1992b) that are not
typically temporally ordered in lists (see 3.1.3c). It is the IVee who initiates
the temporal structure: after his description of his first job (b to f), he moves
forward in time (Well, I think as I got older . . . (g)) to describe his next job
(h to m). This temporal framework is picked up by the IVer’s proposal that
these jobs be grouped together as part of a certain stage of life: Then, that’s
when- you were a kid, right?. The IVer then pursues this more global, episodic
structure of jobs by asking about a second stage (And, so, did you work when
you were in high school?) during which the next jobs are understood to have
occurred.
Notice that there is more information presented in this list than just names
of jobs. The IVee and IVer, however, diverge in terms of what kinds of
information they add. The IVee adds information relevant to his descriptions
of the jobs per se: he describes the tasks that comprised the jobs in terms that
continue the list format. The IVer (in addition to the temporal information
noted above), adds evaluative information: after evaluating the job herself
(Oh, what slave labor), she proposes an evaluation for the IVee to consider
(Jesus, you’ve had a checkered career, haven’t you?) and finally directly asks
the IVee to evaluate a job (Did you like bein’ a short order cook at a hoagie-
factory?). That the IVer proposes shifts in stance - shifts that are not pursued
by the IVee - is important for my point about the difference between lists of
events and stories. The IVee is presenting a list in which the main focus is not
so much his participation in the job, but the type of job per se. Thus, the
subcategories in the list are not the speaker’s relationship to the jobs, but the
jobs themselves:

XI stock the shelves in a grocery store

X2 a janitor in a cleaners
X3 a short order cook in a hoagie shop
X4 a busboy at a country club
X5 a dishwasher for a pancake house
 Variation Analysis 313
The IVer’s evaluative comments, on the other hand, seek to open a different
kind of speech activity - more akin to story telling - in which what is central
is not the categories of an experience, but a person reacting to an experience.
(This effort is consistent with the goals of sociolinguistic interviewing; see
section 2.3, also chapter 5.) Thus, (14) illustrates that a sequence of temporally
ordered events provides material for two very different speech activities: lists
(which categorize experience) and stories (which evaluate experience). Put
another way, although both stories and lists can have constituents that are
temporally ordered, it is evaluation that differentiates the two: evaluation is
required and pervasive in a story, but it is optional and minimal in lists.
The same point is illustrated by (15) and (16). Both are reports (from a single
speaker) about where she lived, and in both, events are marked (sometimes
with then) as temporally successive.
a *
(15) (a) Well, I was born at 22nd and em ... tsk ... oh I forgo-well I think
it’s 22nd and Lanham.
(b) And urn .. . and uh I grew up really, in the section called Larkspur.
(c) And then, I went into the service, for the two years,
(d) and then, when I came back, I married . . . ITT got married.
(e) And then I lived at uh 51st and Lyle. Which is West Ph- it’s right
off of 51st and Granite Avenue. That’s where I lived.
(f) And then we moved here.

Except for some embedded descriptive material concerning specific locations

(a, e) and an event (“getting married”) that creates a natural juncture (i.e.
people usually move from their parents’ home when they are first married),
the speaker lists the places in which she lived in the order in which she lived
there. Nothing is said about how the speaker experienced life in any of the
places: there is no evaluation of any of the locations. In (16), the same speaker
(in a later interview) provides more details about life at 51st and Lyle. (This
is example (44) in Schiffrin 1987a: 249.)

(16) (a) We lived there for two weeks without water, or gas.
(b) We had electricity.
(c) And it was wonderful that we could wake up in the morning, and
play the radio, and do what we want.
(d) Because this landlord- landlady was terrible.
(e) And then we lived there for five years,
(f) and we bought a triplex across the street.

Like (15), (16) contains a sequence of temporally ordered events: We lived

there for two weeks without water, or gas (a) precedes we lived there for five
years (e), which precedes and WE bought a triplex across the street (f). These
events however, are interwoven with descriptions and evaluations. What was
good about living at 51st and Lyle is presented as a short list: it was wonderful
314 Approaches to Discourse Analysis
that we could wake up in the morning, and play the radio, and do what we
want (c)). These good qualities compensated for what was bad (without water,
or gas (a)) to make life at 51st and Lyle a better place to live than their former
home (where the landlady was terrible (d)). Not only is this location thus
evaluated from the speaker’s point of view, but even the demarcation of time
is evaluative. As noted above, reference time moves forward in (e), we lived
there for five years. What shifts from (a) to (e), however, is not location:
rather, it is the qualities of that location (presumably the home had water and
gas after the two-week period ended) and thus the speaker’s experience of that
location. Thus, even the temporal demarcation of events - the temporal struc¬
ture of the sequence - is evaluatively motivated.
I have suggested in this section that the relative necessity and pervasiveness
of evaluation is important to the distinction between narratives and lists, and
even more critical to the distinction between narratives and lists of events. We
noted earlier that the basic unit of a list is an entity: anything of which
something may be predicated (individuals, sets, propositions). Now we are
seeing that lists whose items are propositions do the same kind of categorical
work as other lists: they predicate of each proposition a group membership,
such that a next-proposition is understood as an instantiation of the same
general category as prior propositions. When the reference time of one situ¬
ation (whether event or state) is understood to end prior to the time that
another begins (as in 14 and 15), a list of events may begin to seem very much
like a narrative - but a narrative without evaluation (i.e. a report). The
difference, however, is that events in a narrative work together to make a point
about oneself or one’s world: the events in a narrative are linked together
through an evaluative structure. Lists of events, on the other hand, have a
cumulative effect: events are added together as separate instances and it is their
accumulation itself that substantiates a generalization.

3.1.4 Summary: variation across text types

In this section, we have compared the text structure of narratives and lists:
narratives tell what happened and lists describe a category. This basic dif¬
ference is related to differences in the basic unit of the two text-types (event
versus entity) and in the relative importance of different information structures
(temporal and evaluative versus descriptive). The comparison of text-types is
a key part of a variationist approach to discourse: it contributes not only to
our knowledge of text level variation (e.g. in the distribution of predicate types
across clauses) but also to our understanding of how it is that we attribute
functional identities (and labels) to strings of utterances. And this, in turn, can
allow a more systematic analysis (again, from a variationist perspective) of
how text-types are socially situated, e.g. how linguistic differences in the
stories told (or lists constructed) reflect the different situations in which texts
are used.
 Variation Analysis 315
The discussion in this section has also helped to make more general points
about variationist approaches to discourse. Texts are built from linguistic
constituents that have formal relations to one another: these relationships
help to reveal the basis for coherence within the text. The comparison of
what seems like the “same” linguistic item (or the “same” information
structure) is key to the discovery of these relationships. Notice that this
comparison reveals another variationist assumption about discourse analysis:
syntactic and semantic differences among linguistic items are sensitive to text
structure both within and across text types. We have also seen that what
seem like the very same linguistic units (e.g. then, existential there, stative
predicates, event clauses) have radically different functions depending on
the text-type in which they occur. Although this dependency illustrates the
point made earlier (text-types are built from linguistic units), it also helps to
make a point fundamental to variationist approaches t<$ discourse: one cannot
assume that syntactic difference (or equivalence) is always associated with
functional difference (or equivalence) in text. I expand upon this point in
section 4.

3.2 Variation within text

Now that we have illustrated the variationist approach to discourse by com¬

paring text-types, we turn to another direction taken by variationists: variation
within a single text-type. Although our focus is still linguistic variation, the
locus of that variation shifts from text-type per se to lower level forms (e.g.
syntactic, morphological) whose distribution can be discovered either gener¬
ally in discourse, or more specifically in particular text-types. For example,
one might analyze coordinate and subordinate markers either in general, or
in specific text-types, e.g. arguments, stories. Limiting one’s study of linguistic
variants to a particular text-type has both methodological and theoretical
advantages. A methodological advantage is that we can use our knowledge of
the structure of a text-type to define the linguistic environments in which
formal alternations are possible (what Labov calls the “envelope of variation”:
2.3) and to define different features of those environments as variable con¬
straints on the occurrence of the formal alternants (2.3). A theoretical advant¬
age is further understanding of the functions of formal variants in discourse,
how variation both reflects and realizes particular textual structures. Our
discussion of predicate types in section 3.1.2, for example, showed their
different distribution in narratives and lists, suggesting that statives function
as background clauses only in narrative. Form-function correlations such as
these allow us to approach textual structure from two complementary direc¬
tions: the way a particular text-type contrasts with another text-type (as
we saw in section 3.1), and the way a particular text-type is internally con¬
stituted by combinations of structures and meanings both within and across
its sentences.
316 Approaches to Discourse Analysis

The focus in this section is variation in referring terms in lists. Specific

problems pertaining to reference and referring terms have already been posed
in the sample pragmatic (chapter 6) and conversation analyses (chapter 7). In
this section, we see how variationists might address some of the same ques¬
tions. Before I begin this portion of our sample analysis, it is important to
note that the list is a discourse unit that is especially relevant to our under¬
standing of reference and referring expressions. We have already discussed
referring terms in another text-type (a story, chapter 6) and mention-pairs in
other discourse sequences (e.g. question/answer pairs, chapter 7). Like other
texts, lists also allow the analysis of sequential relationships between “men¬
tions” of an item. Lists present a clear set of items whose relevance begins and
ends not only in the list itself, but also at specific junctures within the list.
These features help us to identify sequential constraints on referring. Lists are
also a closed discourse unit in which a group of conceptually related entities
is brought into focus for a hearer: a new referent is brought into discourse,
used for a while, and then dropped. Rather than becoming agents involved in
the actions of a reported experience (as in narrative), it is the entities them¬
selves that are the focus of talk - such that we might expect to find constraints
on informativeness that are quite different from those in narrative. In addition,
it is well known that reference to entities is sensitive to the structural junctures
and transitions within texts (e.g. Fox 1989): in lists, such junctures are cat¬
egory based.
Lists allow another view of referring expressions - a view that stems from
the way lists display categorical organization, or, more generally, from the
relationship between discourse and conceptual categories. Conceptual cat¬
egories are intimately tied to choices between different referring terms. To
take a brief example, in the story analyzed in chapter 6 we noted that the
speaker alternated between he and it when referring to “a passenger” (i.e. a
person paying a fee to travel). We suggested that this alternation reflected two
different ways of situating “a passenger”: the former in a world of people, the
latter in a world of occupational roles and identities. It is precisely such
conceptual worlds (or, more accurately, pieces thereof) that are presented in
lists.
Getting a view of conceptual worlds through discourse has an important
bearing on our understanding of referring sequences. Virtually all analyses of
pronouns in discourse assume that pronouns can be used when a more precise
understanding of a referent (or in pragmatic terms, recognition of a speaker’s
referential intention) is available (or inferrable) from context (see Kronfeld’s
(1990) concept of functional relevance). We saw in chapter 6, for example,
that speakers may initially mention an entity with a definite noun phrase if
they can assume that a hearer has information sufficient to allow recognition
of that specific entity. Although one source of such information is prior text,
a critical part of accessing a referent is drawing inferences based not only upon
what is said, but also upon what is assumed to be known about the world
(e.g. Ariel 1990). The knowledge upon which such inferences depend can very
 Variation Analysis 317

well be knowledge about category organization. Prince’s (1981) “containing

inferrables” (a type of given information), for example, are defined in relation
to set membership. Clark and Haviland’s (1977) view of “bridging” is similar:
an entity such as “beer” is considered given information if it can be considered
a member of a group of “picnic supplies.” Since it is precisely the organization
of category members that lists can help to reveal, lists may help us learn how
categories allow the inferences underlying the use of particular referring ex¬
pressions (e.g. definite noun phrases). Put another way, lists are a means by
which a speaker uses discourse to convey an organization of a category;
category membership is, in turn, an extremely important notion in our under¬
standing of how referents are accessed.
In the next two sections, I consider constraints on variation in referring terms
within lists. I begin with a very general factor, not limited to lists per se, that
constrains definiteness and explicitness of list-items: ftrder of mention (3.2.1).
I then turn to how text-specific factors that arise from the organization of lists
themselves constrain referring terms in lists (3.2.2).

3.2.1 A general discourse constraint: order of mention

Since a key interest of discourse analysts is sequential patterns, it is not

surprising that many analyses of referring terms in discourse focus upon
referring sequences (see discussion in chapter 6). Two general patterns emerge
from such research: one concerning definiteness; the other, explicitness. First-
mentions are often indefinite noun phrases; next-mentions are often definite.
The second sequential pattern concerns explicitness: first-mentions are typi¬
cally more explicit (i.e. full nouns) than next-mentions (pronouns).
Variationist approaches to discourse seek to add precision to generalizations
such as those noted above (about what “often” or “typically” occurs) through
quantitative analyses: variationists count the number of times a particular
variant appears and compare that with the number of times that the variant
could have appeared in a particular linguistic environment (2.2). This proced¬
ure is repeated for each linguistic environment in which the alternation can
be found. Variationists actually avoid statements about typical patterns unless
they are based on frequency counts and evaluated by statistical tests of signi¬
ficance.
Table 8.1 counts the number of definite referring expressions (compared to
indefinites) in both first- and next-mentions of items in lists. I include here
just those list-items that were mentioned more than once. Consistent with the
notation used in our diagrams of the hierarchical organization of list-items, I
will use (X) as the general category underlying the list, (XI . .. n) as the
subcategories (the list-items), and so on, in subsequent discussion. Tables 8.1
and 8.2 do not differentiate mentions of list-items by level (e.g. XI versus
XIa); nor do they include mentions of X itself.
318 Approaches to Discourse Analysis

Table 8.1 Referring sequences and definiteness in lists

First-mention Next-mentions Total

Indefinite 23 0 23
Definite 9 74 83
Total 32 74 106

As table 8.1 shows, referring sequences in lists follow a general definiteness

pattern similar to that noted for referring sequences in other discourse: 72
percent (23/32) of the first-mentions, but none of the next-mentions, are
indefinite.
Referring sequences in lists also show the same general explicitness pattern
as sequences in other discourse. Table 8.2 shows the relationship between
order of mention (first-mentions versus next-mentions) and explicitness, going
from most explicit (full noun phrases) to less explicit (pronouns, zero pro¬
nouns).

Table 8.2 Referring sequences and explicitness in lists

First-mention Next-mentions Total

Full NP 32 10 42
Pronouns 0 61 61
Zero 0 3 3
Total 32 74 106

Table 8.2 shows that 100 percent (32/32) of the first-mentions are full noun
phrases, but only 13 percent (10/74) of the next-mentions are full noun phrases.
Note, also, that none of the next-mentions of (XI . . . n) are more explicit or
more informative than their first-mentions. This is not to say, of course,
that more information cannot be added about a list-item after its first-mention.
But rather than being added by more explicit referring terms, such information
is always added in a construction (e.g. a predicate nominal) that links the
information to a pronoun.
The list of “old folks living on this block” (6) was an example of both
patterns seen above. Rather than present the entire example again, I just
present the referring terms here as (17).

(17) Referring expressions from example (6)

First-mention Next-mentions
that kook hangin’ out the door him/his (4 times)
one woman here she/her (12 times) “zero”
another couple up the street they/their (3 times)

The first-mentions are both indefinite and more lexically informative (i.e.
explicit) than next-mentions: “kook” is evaluative, “woman” is gender-specific,
 Variation Analysis 319

and “couple” provides information about number. Reference to each entity is

also textually bound, i.e. (XI) is not mentioned again after (X2 . . . n) is
introduced (see 3.2.2b).
As we discussed in section 2.2, variationist approaches do not always seek
to explain individual cases that might be seen by others as violations of a
categorical constraint. This is because variationists assume that constraints
operate in a probabilistic fashion: one expects to find less then perfect com¬
plementary distributions and heterogeneity at both social and linguistic levels
of analysis. Nevertheless, it is still important to know the boundaries of
variation, i.e. to know in what environments (if any) realizations of language
structure are categorical. The more precise one can be about constraints on
variation, the more one can use variation studies to compare and evaluate
competing hypotheses and explanations for empirical observations. In addi¬
tion, heterogeneity and variability play an important role in sociolinguistic
theory: language variation is related to social identity, language variation both
reflects and creates language change, and so on. Finally, the possibility that
one constraint interacts with (e.g. supplements, cancels) another raises import¬
ant questions about whether formalist assumptions of autonomy and modu¬
larity of language are justified (chapter 1).
Reasons such as these lead variationists to try to explain as much variation
within a particular realm as possible. Thus, we might try to explain why the
constraints in tables 8.1 and 8.2 are not categorical: there were 9/32 definite
first-mentions and 10/74 explicit next-mentions. A number of different
strategies for explaining variation within a particular realm are available:
identifying new types of constraints (either internal or external to the language
system), analyzing the operation and possible intersection of multiple con¬
straints. In our discussion of referring expressions in chapter 6, we made
several observations about initial definites that may help explain the “atypi¬
cal” cases in tables 8.1 and 8.2. We noted that first-mentions could be definite
if hearers have accessible to them information that can be used as a conceptual
framework within which to locate a referent. It is this factor that accounts for
the nine initial definites in table 8.1. Note, however, that the source of
accessibility underlying these initial definites was due to different contextual
parameters, only some specific to lists. Some initial definites were referents
anchored to the speakers themselves, e.g. speakers always began lists about
their relatives by anchoring the person occupying a particular kinship role to
themselves with a first person possessive pronoun (e.g. my grandmother, my
older brother). Other initial definites were proper names whose referent is
assumable as part of shared knowledge, e.g. the name of a country or state.
Recall from (1), for example, that the speaker named the states with race
tracks without first introducing the states as indefinites (e.g. “There’s a state
around here called New York”). As noted earlier, other initial definites in
lists were inferrable (either textually or as containing inferrables: Prince
1981) from the general category underlying the list: I discuss these in section
3.2.2.
320 Approaches to Discourse Analysis

The approach to atypical cases that I pursue here is to find constraints arising
from the particular text-type serving as the locus of variation. I show how
variation in referring sequences in lists is due to the structure of lists them¬
selves by drawing upon observations from our earlier discussion (3.1) and
trying to identify features of lists with enough precision to make quantitative
comparisons of referring expressions in different textual environments - com¬
parisons that can then be used to evaluate different hypotheses about the
organization of referring expressions in talk.

3.2.2 Constraints specific to lists

Although mentions of list-items form referring sequences much like those in

other discourse, mentions of list-items also vary in ways that arise from the
list organization itself. We saw in section 3.1 that lists focus on a general
category (X) and its members (XI ...«). A speaker presents items that share
a common attribute and are inferrable as part of a collection: the cumulative
presentation of such items in talk reflects their similarity and their shared
category membership. Although category members are partially the same, they
are also partially different from one another.
Constraints on variation in referring terms stem from the categorical struc¬
ture of lists. Evidence that the language used in lists is sensitive to categorical
structure is provided in table 8.3. Table 8.3 shows that and is used with
coordinate relationships between category members (regardless of their level,
e.g. XI and X2, Xla and Xlb), but not with subordinate relationships (going
either up a level (Xla to XI) or down a level (XI to Xla)).

Table 8.3 Connections between list-items

Subordinate Coordinate Total

And 1 35 36
“Zero” 15 5 20
Total 16 40 56

Table 8.3 shows that although and links coordinate level relationships be¬
tween categories (e.g. XI and X2), it rarely links subordinate relationships
between a category and subcategory (e.g. XI and Xla): 87 percent (35/40) of
the coordinate relationships are linked by and, compared to only 6 percent
(1/16) of the subordinate relationships. Thus, the categorical relationships
between list-items establish structurally differentiated boundaries and transi¬
tions within lists that are linguistically reflected. We see in the next subsections
that variation in referring terms is constrained by these structural boundaries.
I begin by suggesting that lists provide environments for referring expressions
that are at least partially segmented from surrounding discourse (3.2.2a); this
suggestion further justifies the idea that the referring sequences in lists are
 Variation Analysis 321

sensitive to constraints stemming from the text-type itself. I then focus on how
the general category (the “sameness”) in lists constrains referring expressions
(3.2.2b). I conclude the analysis by defining thematic and recency constraints
within lists and comparing their effects on referring terms (3.2.2c).

a Inside and outside of lists

Lists provide environments for referring expressions that are at least partially
segmented from surrounding discourse. One way to see this is to find an
example in which an item is mentioned both before and during a list. If lists
are relatively closed discourse units, we might find that mentions of an item
prior to the inception of the list have little sequential effect on mentions of
that item within the list.
Let us consider (18). (18) begins as an IVer asks a question about the IVee’s
in-laws: "

(18) IVer: (a) Did you ever live with your in-laws after you got married?
IVee: (b) Eh: no, I didn’t.
(c) We had- well, they lived on the first floor:
(d) they owned this eh triplex.
(e) And eh my mother in law and father in law lived on the
first floor,
(f) and my sister in law and her husband lived on the second,
(g) and we lived on the third.
(h) And it was a very good arrangement because we- we had
our own: um apartments.

The item to focus upon in (18) is “in-laws.” “In-laws” is first mentioned

outside the list in the IVer’s question (Did you ever live with your in-laws after
you got married} (a)). “In-laws” is then treated as a next-mention (an inex¬
plicit definite they) as the IVee begins her answer to the question (they lived
on the first floor (c), they owned this eh triplex (d)). The answer requires an
explanation of a particular kind of house which is introduced as this eh triplex
(d). It is the living arrangement in the triplex that becomes the general category
underlying the list: what is listed is the occupants of different floors in the
triplex. Recall that “in-laws” had previously been treated as old information:
its next-mention was they in (c) and (d). Within the list, however, “in-laws”
is introduced in terms relevant to the list itself: And eh my mother in law and
father in law lived on the first floor (e) establishes “in-laws” as occupants of
the first floor of the triplex. Central to my point is that “in-laws” is also
brought into the discourse as if it were new information: it is mentioned
through explicit full nouns. Thus, an item previously used in the discourse is
treated as a first-mention once it enters the list.
Although lists can open environments for reference that differ from prior
discourse, it is not quite so clear that their environments for reference end
once the list closes. One reason for this is that lists do not end as neatly as
322 Approaches to Discourse Analysis

they begin (cf. Labov’s (1972b) point that the endings of narratives are less
clear-cut than their beginnings; also Schiffrin 1984a). Another reason is that
lists can intersect with other discourse units whose boundaries may or may
not coincide with those of the list (see (23) in 3.2.2c) What this means is that
it is not always easy to know when a person has temporarily left a list and
will return to that list, or when a person closes a list. Furthermore, speakers
themselves may not have a clear idea of when their list will end simply because
the mention of one item may remind them of another.

b Category constraints
We saw above that lists open a new environment for referents. An important
characteristic of such an environment is the presence (either stated explicitly
or assumed implicitly) of a general category (X) in relation to which other
referents (XI . . . n) are understood. In this section, we see how the underly¬
ing presence of such a category constrains referring expressions (to X and to
XI ... n) within lists.
We noted earlier that first-mentions can be definite if speakers can assume
hearers’ accessibility to a conceptual framework within which to locate a
referent. Since the category underlying a list provides just such a framework,
the referential existence of first-mentions can be inferred in relation to that
framework - hence, some definite first-mentions in lists.
(19) illustrates list-items whose referents can be inferred to be specific mem¬
bers of an assumed collectivity (cf. Prince’s containing inferrables), i.e. “family
members.” (Note that these list-items were not included in tables 8.1 and 8.2
because they were mentioned only once.) In (19), definite NPs are the focus
of the existential predicate there is. (19 was seen earlier as 10.)

IVer: (a) Flave any of the kids ever expressed a desire to see him?
IVee: (b) No.
(c) No, they never did.
(d) I mean like uh, they never asked about daddy or where did
daddy go or-
(e) I mean like, there was me
(f) and there was my father
(g) and there was my cousins.
(h) And the kids were pretty well occupied.
(i) I mean like,
IVer: (j) Yeh, it sounds like they were busy.

The IVer’s question about whether the IVee’s children ever missed their father
(the parents are separated) is responded to with a list of people who kept the
kids . . . pretty well occupied: me (e), my father (f), my cousins (g). Each of
these list-items is presented as the focus of the existential predicate there was.
Although such constructions are often said to require indefinite NPs, it has
also been noted that lists are an exception to this generalization. Flannay
 Variation Analysis 323

(1985), for example, describes enumerative uses of existential predicates in

which entities are understood as particular cases, i.e. members of a sometimes
implicit set (see also Rando and Napoli 1978). Definite NPs in there is con¬
structions thus present an entity that is thought to be of interest not so much
because it itself exists, but because it exists as some member of a larger set.
In (19), me, my father, and my cousins are presented as members of the
implicit set “family members”: their existence is relevant to the category being
described. Put another way, each definite description is presented as an example
that fills a specifying role.
The category (X) underlying a list constrains references to individual list-
items (XI ... n) in ways other than definiteness. (20) illustrates that the lexical
informativeness of first-mentions can be constrained by the implicit category
unifying the list. (20 is a summary of 6; I include only the first-mentions.)
*
(20) IVer: (c) Are there any old folks livin’ on this block?

IVee-3: (j) That kook hangin’ out the window.

IVee-2: (r) Then there’s one woman here, crippled or whatever, you
... want to call her.

IVee-2: (ff) And then there’s a-another couple up the street who never
come out.

The information provided about each list-item is relevant to the IVer’s ques¬
tion, which not only prompts the list but also provides the general category
underlying the list. Although two unifying factors (“age” and “location”) are
made salient in the question Are there any old folks livin' on this block? (c),
only “location” continues to explicitly unify the list-items, first in the general
affirmative response all down this street (f) and then in first-mentions of
list-items: the window (j) is a boundary between the privacy of home and the
public nature of the street; here (r) and up the street (ff) are relative to the
speaker’s own location on the block. Thus, each mention of (XI . , . n) is made
relevant to the category (X) of which it is a member.
The category (X) underlying a list constrains mentions not just of subca¬
tegories (XI ...«), but also mentions of the category (X) itself. The next
example illustrates the salience (cf. the topic accessibility; Givon 1983a) of the
category (X) in the list. This example will also help us begin to compare two
constraints on pronouns: recency versus thematicity (see Clancy 1980; Grimes
1978; Tomlin 1985). Finally, it will suggest the importance of subcategory
boundaries within a list.

(21) IVer: (a) That’s right, I’m-slowly getting your family straight here
[hh. It’s-
IVee: (b) [hhhOh, when my mother died, they u-
324 Approaches to Discourse Analysis

(c) this is funny,

(d) she had forty five great grand- forty-five grandchildren,
and over: twenty-five great grandchildren.
IVer: (e) Holy [Mackerel.
NEICE: (f) [Now it’s more than doubled.
IVee: (g) Oh, it’s doubled since she died, I think she’s must had
about fifteen more born since she died.
neice: (h) More than that.
IVee: (i) What, in ten years?
neice: (j) Su:re.
IVer: Wow.
IVee: (k) [Well, I know I’m losing track myself.
neice: (1) [I had two, I had two,
(m) Susie had two,
(n) Peggy had two,
(o) Cathy had one,
(P) Teresa had three,
(q) Patsy had two more since she died. Three [more.
IVee: (r) [Three. Uh:
neice: (s) And then there’s a lot of other ones in the family.

(2Y) highlights the categorical structure of (21). The reference to “my mother”
(who is a key part of the category) crosses subcategory boundaries:

(2T) Categorical structure of (21)

X “my mother’s descendants”
XI “my mother’s descendants born before she died”
Xla forty-five grandchildren,
XIb over: twenty-five great grandchildren. (d)
X2 “my mother’s descendants born after she died”
Oh, it’s doubled since she died, I think she’s must had about
fifteen more born since she died. (g)
X2a I had two,
X2b Susie had two,
X2c Peggy had two,
X2d Cathy had one,
X2e Teresa had three,
X2f Patsy had two more since she died. (q)

(X) in (21) is “mother’s descendants.” The descendants are differentiated

according to their time of birth in relation to the mother’s life: XI are those
“born before she died,” X2 are those “born after she died” (g). After noting
those born “before she died” (forty-five grandchildren, and over: twenty-five
great grandchildren (d)), the next subcategory “mother’s descendants born
after she died” (X2) is introduced: it’s doubled since she died (g). Note that
 Variation Analysis 325
in introducing X2, “mother” is evoked by she. The use of a pronoun here is
not surprising: “mother” is still a very salient (and recent) referent, and there
are no competing antecedents (Clancy 1980; Givon 1983a).
The pronoun in Patsy had two more since she died (q) - in which she evokes
“mother,” not “Patsy” - is surprising. Note that a long list of descendants
born since she died is presented: these are subcategories of X2. It is within the
first-mention of one such subcategory (X2f Patsy had two more since she died)
that “mother” is mentioned again. The pronominal reference to “mother” as
she is especially striking here, given both the distance between she and the
explicit mention of “mother” (my mother (b)) and the other intervening
referents. Thus, this use of she seems to violate what scholars (e.g. Givon
1983a; Tomlin 1987) have called a recency constraint. What she in (23) seems
to suggest, instead, is the operation of a thematic constraint: an entity that
plays an important textual role (cf. in the te'xt topic; Schiffrin 1988b) re¬
mains an accessible topic even if not mentioned recently. Put another way,
although the antecedent of she in (q) is certainly across structural (i.e. subcat¬
egory) boundaries, it is still located within an open focus space (Grosz 1981;
Sidner 1986) simply because the referent is the category underlying the list
itself.
Although the competition between recency and thematic constraints is hard¬
ly confined to lists, lists provide an especially clear source of data in which to
evaluate these two constraints. Thematicity is particularly easy to identify: the
general category (X) underlying the list is thematic. Recency can also be easily
identified, but more importantly, our understanding of recency can be sup¬
plemented: the presence of clear structural boundaries between subcategories
(e.g. XI and X2) allows us to consider recency not just in terms of the number
of clauses intervening between a pronoun and its antecedent, but also in terms
of the location of successive mentions of a list-item (e.g. XI) relative to the
mentions of other items (i.e. the subcategory boundaries) in the list.
Table 8.4 provides a quantitative view of a thematic constraint in lists: it
compares next-mentions across subcategory boundaries depending upon
whether next-mentions evoke the category itself (X) or a subcategory (XI . . . n).
In other words, table 8.4 compares next-mentions of (X) to next-mentions of
(XI) when (X2) precedes that next-mention. A thematic constraint predicts that
next-mentions of (X) can be less explicit because they are higher on a scale of
topic accessibility (Givon 1983a, b, c) than next-mentions of (XI . .. n).

Table 8.4 Category level and next-mentions across subcategory

boundaries

Next-mention is of

Category (X) Subcategory (XI ... n) Total

Pro/“zero” 11 0 11
Full NP 0 6 6
Total 11 6 17
326 Approaches to Discourse Analysis

As table 8.4 shows, 100 percent (11/11) of the next-mentions of (X) are
pronouns. This is in sharp contrast to the distribution of pronouns for next-
mentions of (XI . . . n): no pronouns at all were used for next-mentions across
subcategory boundaries. This difference supports the idea that the general
category whose specific members the list instantiates is thematic in the list.
Put another way, the focus space of a general category (X) never closes as long
as the list continues; the focus spaces of the subcategories (XI ...«), on the
other hand, open and close during the course of the list.
Before I go on to supplement the data from table 8.4 with data that reveal
more about the importance of subcategory boundaries, let me summarize thus
far. We have seen three ways in which the category being described in a list
can influence mentions of items in a list: first-mentions of (XI . . . n) can be
definite because they are all members of a general category (X); lexical informa¬
tion ties mentions of (XI . . . n) to (X); pronouns can have antecedents that
cross subcategory boundaries when they evoke the category (X) underlying
the list. In 3.2.2c we see that just as the category (X) influences referring terms
in a list, so too do differences among subcategories (e.g. the difference between
(XI) and (X2)).

c Subcategory constraints
Just as the overall unity of a list (the category) is reflected in its referring terms
(3.2.2a), so too are the differences and boundaries between subcategories.
We saw earlier that if a next-mention of a list-item (XI) is after another
list-item (X2) has been mentioned, then the next-mention of (XI) is a full
noun. Table 8.5 presents these data again, but contrasts them with next-
mentions within subcategory boundaries, i.e. next-mentions of (XI) before
(X2) is introduced.

Table 8.S Next-mentions within and across subcategory boundaries

Cross boundaries Do not cross boundaries Total

Pro/"zero" 0 64 64
Full NPS 6 4 10
Total 6 68 74

Table 8.5 shows that 92 percent (68/74) of the next-mentions are within sub¬
category boundaries; 95 percent (64/68) of these are pronominal or “zero”.
Table 8.5 also shows that 8 percent (6/74) of the next-mentions have antece¬
dents that cross subcategory boundaries, i.e. return to a list-item (XI) after
another list-item (X2) has already been discussed. All six next-mentions are
full nouns: five of these appear in descriptive structures whose sequential
organization is quite different from the lists considered thus far. After discuss¬
ing these five cases, I briefly comment on the one other case.
All the lists discussed so far in this chapter can be called depth-before¬
breadth lists. List-items and descriptive material about those items are together
 Variation Analysis 327

in the text; each list-item (XI) is completed prior to the introduction of a next
list-item (X2).

XI . . .
X2. . .
X3 . . .

In this structure, speakers provide “depth” to the list (information about (XI))
before adding “breadth” to the list by mentioning (X2). Notice that pronomi¬
nal next-mentions can be clearly linked with a single unambiguous antecedent
in this structure, simply because no new list-items are introduced until descrip¬
tion of the prior item is complete. Not only is (XI) closed before mentioning
(X2), but list-items are not typically returned to once they are closed. Thus,
in the depth-before-breadth list structure, references^ list-items typically do
not cross subcategory boundaries.
Lists can also be presented in a breadth-before-depth structure: speakers
provide “breadth” to the list (they first mention (XI), (X2), and (X3)) before
providing much information (“depth”) about each item.

XI
X2
X3
XI . . .
X2 . . .
X3 . . .

In this structure, each list-item is presented prior to its elaboration. This means
that speakers return to prior list-items (e.g. to XI) cyclically and they cross
the subcategory boundaries of other list items (e.g. X2 and X3) to do so. (22)
illustrates. The IVee is describing what is known in Philadelphia as a “father,
son, and holy ghost” house. (This was example (30) in chapter 7.)

(22) IVee: (a) A girlfriend of mine, downtown somewhere

(b) she m- uh got married and she moved into one.
XI (c) One room here
X2 (d) and then it’s one room on the second floor
X3 (e) and one room on the top.
IVer: Yeh, that’s right.
XI IVee: (f) So she used her downstairs as her parlor.
IVer: Uh huh.
X2 IVee: (g) Her second,
X (h) there’s only three rooms [like,] y’know.
IVer: [Yeh.]
X2 (h) The second story of the room was her bedroom
X3 (i) and the third room was her bedroom.
328 Approaches to Discourse Analysis

The IVee in (22) begins by mentioning all three list-items as indefinite full
nouns: (XI) is one room here (c), (X2) is one room on the second floor (d),
(X3) is one room on the top (e). She then provides more information about
them in the same order in which they were introduced. Note that although
she provides that information as predications about the list-item, she also
explicitly labels each list-item again: (XI) her downstairs (f), (X2) the second
story of the room (h), (X3) the third room (i).
As noted earlier, five of the six explicit next-mentions of list-items across
subcategory boundaries are in breadth-before-depth lists. The one remaining
example is (23). We noted earlier that lists can be environments for reference
that differ from surrounding discourse, but that it is not always easy to establish
when a list ends. (23) is an example in which a speaker embeds several nar¬
ratives into a list, and references to entities move between the list structure per
se and the narratives. In (23), the speaker lists his jobs in temporal order
(3.1.3a) and tells stories that explain his transition from job to job. Although
space prevents presentation of the whole discourse, I include a mention of a
“job” (X2 at time 2) that is within a narrative and does not cross subcategory
boundaries, as well as the next-mention of primary interest: mention of a “job”
(XI at time 4) within a narrative that crosses subcategory boundaries.

(23) Xl/Tl Mostly what happened was I was-I worked there as a metal
polisher,
X2/T2 and then I went to a loom company that was cuttin’ looms.
For about six months.
[describes illnesses]
All right, I was workin’ for them about six months.
[doctor forbids him to go back to work]
[argument with supervisor: he’s fired]
X3/T3 And I went and got a job right over here at Baxter and Kelly’s
makin’ uh plush materials.
[describes problem with second job]
X1/T4 So I went back to metal polishing.

In (23), the speaker describes how he went from metal polishing (XI at time
1) to a loom company (X2 at time 2) to making plush materials (X3 at time 3)
and then back to metal polishing (XI at time 4). When the speaker mentions
a work place currently in focus in the list, the referring term is pronominal
even if a long description intervenes between the first and next-mentions: he
mentions “the loom company” as them in All right, I was workin’ for them
about six months. Although not included above, the speaker mentions Baxter
and Kelly’s makin’ uh plush materials (X3 at time 3) after a story but within
the same subcategory description, and again uses a pronoun to do so. Note,
however, that when “metal polishing” (XI) is mentioned again (at time 4) it
is evoked through a full noun: So I went back to metal polishing. This
difference shows, again, that list-items can be treated as given information
 Variation Analysis 329

(and be evoked through pronouns) only if a new subcategory has not been
introduced.
We have seen thus far that pronominal next-mentions of list-items (XI . . .
n) typically have antecedents within their own subcategory boundaries: the
only exceptions are when the referent evoked by the pronoun is the thematic
category (X) itself. Next-mentions of list-items (e.g. XI) that cross subcategory
boundaries (e.g. after mention of X2) are full nouns.
Recall, however, that some next-mentions of (XI . . . n) are explicit nouns
even though their first-mentions are not across subcategory boundaries: 4 out
of 68 of the referring expressions within subcategory boundaries were full
nouns (table 8.5). All four of these next-mentions had evaluative functions in
their lists (3.1.3). (24) illustrates.

(24) IVee: (a) But- see, years ago, every neighborhood had its own little,
uh stores,
(b) y’know, corner stores and everything,
(c) which the supermarkets knocked out of [commission.
IVer: [No kidding.
That’s right.
IVee: (d) And there were little restaurants,
(e) that you don’t find today.
IVer: Well, like-
IVee: (0 See, H and H come in and knocked all the little restaurants
out of business.
IVer: Well, that’s really a shame.
IVee: (g) Yeh. Right. Progress is ba:d.

The list in (24) is part of a discussion of differences between the past and the
present: the speaker is listing the types of establishments (little stores (a), little
restaurants (d)) that are no longer in her neighborhood. The speaker s attitude
toward change is not only summarized at the end of the list (Progress is ba:d
(g)): it is conveyed through the description and evaluation of list-items them¬
selves. In (a) through (c), the supermarkets (c) are portrayed metaphorically
(Lakoff and Johnson 1980) as antagonists who knocked the little stores (XI)
out of commission. The first-mention of little restaurants (X2) is evaluative.
The repetition of little (from little, uh stores (a)) in (d) (little restaurants) and
the intonationally separate adjunct (that you don t find today) both prefigure
the powerless position, and fate, of the restaurants; that you don’t find today
(e) also ties (X2) to the comparison underlying the list. The next-mention of
“little restaurants” (X2) is a full noun all the little restaurants (rather than the
pronominal them) that is also evaluative. Not only does the speaker increase
the generalizability of this example with all, but the explicit next-mention of
(X2) - repeating little restaurants from the first-mention - conveys that the
fate of the restaurants is part of a repeated pattern, thus also contributing to
the generalization.
330 Approaches to Discourse Analysis

To summarize: we have focused here upon how referring terms in lists reflect
differences among subcategories. Specifically, we have seen that the presence
of subcategory boundaries constrains the use of pronouns, i.e. pronominal
mentions of list-items have antecedents within the section of the list describing
that list-item itself. Put another way, the alternation between pronouns and
full noun phrases in lists is constrained by the way different sections of the
list are segmented from one another. Thus, just as the overall descriptive unity
of a list is reflected in its referring terms (3.2.2a), so too are differences among
the parts of a list.

3.2.3 Summary: variation within text

In this section, I have illustrated how variationist approaches to discourse

analyze variation within text. I focused on variation in referring terms in lists.
We saw that referring terms in lists reflect the category structure underlying
a list. The alternation between pronouns and full noun phrases in lists is
constrained by category level and subcategory boundaries - by the way differ¬
ent units in the list are related to, and segmented from, one another. Such
findings drew upon our knowledge of the structure of lists (3.1) to define
possible constraints on referring terms and to analyze variation in referring
terms under different structural conditions.

4 The variationist approach to discourse

I noted at the outset of this chapter that the variationist approach to discourse
stems from quantitative studies of linguistic change and variation. Although
such analyses typically focus on social and linguistic constraints on semantic-
ally equivalent variants, such an approach has also been extended to texts.
In this chapter, I illustrated extensions at two different levels of analysis:
comparing text types (3.1) and analyzing variation within text (3.2). I did so
to show not only that variationist approaches to discourse can take two direc¬
tions, but that there is an important relationship between these two directions:
our comparison between text-types provided valuable information about
different types of textual structure, and thus information about discourse
constraints on grammatical variation that are specific to particular text
types.
In section 3.1, we identified and compared the textual structures of nar¬
ratives and lists. We saw that the basic unit of narrative is the event; the basic
unit of lists is the entity. The primary information structures of narratives are
temporal and evaluative; the primary information structure of lists is descript¬
ive. Such comparisons reflect the variationist tendency to speak of discourse
 Variation Analysis 331

in terms similar to those used by structurally oriented linguists: “units” are

akin to constituents; “information” is propositional in nature (despite the fact
that propositions may themselves have evaluative interpretations); “struc¬
tures” are syntagmatic and paradigmatic arrangements of units in recurrent
patterns. Texts are thus built from linguistic constituents that have formal
relations to one another: these relationships help to reveal the basis for co¬
herence within a text (or across turns in an interaction: Labov 1972d; Labov
and Fanshel 1977).
Comparing text types also contributes to our knowledge of variation in text
structure per se. Recall that we depended upon variation in the distribution
of specific linguistic forms (e.g. temporal markers, stative predicates) across
text-types as a source of information about those text-types. We began these
comparisons thinking that we were comparing the “same” linguistic form, just
in different linguistic environments. Our results shdWing how syntactic and
semantic differences among linguistic items reflect (and help realize) text
structure, however, suggest that what seemed the same at a sentence level is
really very different at a text level. Then, existential there, stative predicates,
and event clauses all have dramatically different functions depending on the
text-type in which they occur. Although such differences clearly illustrate the
interdependency between sentence and text, they also raise an important issue
about referential and functional equivalence.
We noted earlier (section 2) that variationists try to discover patterns in the
distribution of alternative ways of saying the same thing, i.e. the social and
linguistic factors that constrain linguistic variation. Thus, an initial step in
variationist studies is to establish which forms alternate with one another. We
also saw that applying variation analysis at a discourse level has sometimes
required loosening the criterion by which alternating forms are grouped
together from referential equivalence to functional equivalence. Although I am
not advocating the use of one criterion over another, what I am suggesting is
that the adoption of any kind of formal basis for identifying equivalents
(whether it is referential or syntactic) does not allow us to ignore functions:
in fact, it may end up highlighting functional variability at another level of
analysis. This result is the converse of what happens when we begin with
functional equivalents and find that a great variety of different forms can serve
similar functions. In our discussion of library reference interviews in chapter
5, for example, we discussed the different forms and devices (interrogative or
declarative statements, rising intonation) that patrons could use to accomplish
a single function (i.e. issue a query). Thus, regardless of the locus of analysis
from which variationists initiate a study of variation in and of discourse, they
may end up at a destination quite different from the one anticipated - and may
find that they reached that destination using a rather surprising path.
In section 3.2 we focused on referring terms as linguistic variants within
texts: we analyzed how such variants are constrained by both general features
of discourse and particular features of the text type. After noting a general
discourse tendency for first-mentions to be indefinite and explicit (cf. chapter
332 Approaches to Discourse Analysis

6), we concentrated on how referring sequences are sensitive to the particular

structure of lists. Put another way, we analyzed how formal variants within
lists (nouns versus pronouns) are constrained by the descriptive structure of
the list.
Limiting the study of linguistic variants to a particular text-type - as we did
- has both methodological and theoretical advantages that point out the
importance of taking a two-pronged approach to variation in discourse. By
focusing on referring terms within a discourse whose structure we had just
described, we were able to use our knowledge of that structure to identify the
linguistic environments in which formal alternations were possible, and to
define different features of those environments as constraints on variation
among formal alternants. We saw, for example, that the distribution of full
nouns and pronouns was constrained by the location of an item in the hierarch¬
ical structure of a list. We interpreted this constraint as evidence that pronouns
can be constrained by thematicity, as well as by recency. Efforts to relate the
findings of variation analysis to other issues and controversies within linguis¬
tics are not unusual: variationists often form their hypotheses and use their
findings to address the adequacy of an explanation, or try to find evidence
that will help resolve particular problems or decide among competing solu¬
tions for those problems.
The way variationist approaches to discourse compare different possible
explanations is by searching for data which confirm (or cast doubt upon) the
co-occurrences predicted by each explanation. Although this is not, of course,
a goal unique to variationists, variationist approaches add the strengths (and
limitations) of a quantitative paradigm to such efforts. Quantitative analyses
require definitions of the variants (the different possible realizations of an
underlying type), a classification of factors in the environment with which
those variants may be associated, and a comparison of the frequencies with
which different variants co-occur with different factors. In our analysis in
(3.2.2b), for example, we defined the variants as pronouns and nouns, we
identified environments in terms of categorical structure (i.e. as next-mentions
of items at different levels of category structure), and we compared the relative
frequencies of pronouns and nouns according to their location in the categor¬
ical structure of the lists in which they occurred. Quantitative procedures
depend upon certain assumptions about the relationship between text and
context that limit the range of discourse variation to which they can be
applied, and the way that such variation can be interpreted.
Although the quantitative thrust of variationist approaches to discourse has
the advantages suggested above, it also limits the amount of attention paid to
particular cases, concentrating instead upon general trends and patterns. Vari¬
ationist approaches do not always seek to explain the cases that seem to violate
a general pattern: this is consistent with the belief that constraints operate in
a probabilistic fashion (one expects to find less then perfect complementary
distributions) in response to both synchronic and diachronic pressures in social
life and language structure. (Inherent variability plays a tremendously import-
 Variation Analysis 333

ant role in both sociolinguistic and linguistic theory; to oversimplify a bit,

language variation is related to both social identity and language change.) This
does not mean, however, that it is not important to know the boundaries of
variation. The more precise one can be about constraints on variation, the
more one can use variation studies to compare and evaluate competing hypo¬
theses about social life and language structure. A number of different strategies
for explaining variation within a particular realm are available: identifying
new types of constraints (either internal or external to the language system),
analyzing the operation and possible intersection of multiple constraints. The
approach illustrated here was to find constraints on referring sequences in lists
that were due to the structure of lists themselves, and to make quantitative
comparisons of referring expressions in different list environments - compar¬
isons that could be used to evaluate different hypotheses about the organiza¬
tion of referring expressions in talk.
Finally, although variationists can also focus upon interactional and situ¬
ational variation in texts, this was not a direction pursued in our sample
analysis. Note, however, that such analyses would be built upon - and incorp¬
orate - studies such as those provided here. The comparison of text-types and
the analysis of textual constraints on variation both allow more systematic
analyses of how text-types are socially situated: they first “factor” out how
particular features of a text are linguistically, rather then socially, motivated.

Exercises

1 Collect versions of the same story (or the same list) in different modalities
(e.g. a shopping list that is spoken and written) and different situations (e.g. a
story told to a friend and to a family member). How do the text structures vary
in ways that might be due to modality and situation? Are there any syntactic
variants that seem to appear in one modality/situation but not the other?
2 Thematic and recency constraints can be compared in text types other
than lists. Using a written text as data (e.g. a short story, an expository text, a
recipe, a newspaper article), analyze how thematicity and recency constrain
pronominal/nominal variation in the text.
3 Recipes are a form of written discourse that combine features of both lists
(the ingredients necessary to make a certain dish) and narrative (steps for
making the dish are typically presented in temporal order). Collect ten to fifteen
recipes for the same dish (e.g. pie crust, meat loaf) from a range of cookbooks
intended for different audiences (e.g. novice to expert). Analyze the textual
structure of the recipes in terms of the two text-types noted, as well as the
features that do not seem to conform to either text-type. Identify text-level
variants (e.g. in the order in which ingredients are listed) and clause-level vari¬
ants (e.g. in the use of referring terms, adverbial clauses, conjunctions). Do
the textual structures and/or the variants seem to be related to the different
audiences to which the recipes are directed? If not, what might be responsible
334 Approaches to Discourse Analysis

for the differences that you found? (Note: The same exercise can be applied to
other procedural texts in which instructions are given in a comparable format.)
4 Texts can be identified by a number of different criteria, discussed in work
such as Bauman and Briggs (1990), Biber (1988), Briggs and Bauman (1992),
Enkvist (1991), Ferguson (1983), Horvath (1987), and Virtanen (1992). Compare
the criteria used in this chapter to differentiate lists from narratives to the
criteria proposed in some of the above-mentioned references. What other crite¬
ria might have been used to differentiate lists from narratives? How can we
evaluate the principles underlying a text typology? What is the value of con¬
structing text typologies, and can this goal be accomplished?
5 Research by Ellen Prince (Prince 1984, 1985) and Gregory Ward (Ward
1990; Ward et al. 1991) has examined how principles derived from Gricean
pragmatics influence the use of syntactic variants. After reading one or two
selections from each of these authors, assess the compatibility of the Gricean
and variationist approaches, either by focusing on a specific construction in
your own data sample or by reanalyzing the referring sequences in chapters 6
and 8 of this book. (You might also include referring sequences from the data
in chapters 4, 5, and 7.)
6 The constraints on referring terms discussed in this chapter have
been largely structural constraints, defined according to the location of a "men¬
tion" in a text. However, we also noted some exceptions to the structural
patterns that seemed to be evaluatively motivated. Some of the constraints
considered in other chapters (adjacency pair, cooperative principle) might also
be considered evaluative: they arise from the way a speaker weaves together
the need to make a "mention" recognizable to a hearer with a subjective point
being made by a text and by the role of a referent within the text. Analyses by
Duranti (1984), Maitland and Wilson (1987), Muhlhausler and Harre (1990), and
Varenne (1984) pursue this line of inquiry more thoroughly. Using some (or all)
of the referring sequences considered in chapters 7-9, the data in appendix 3,
or any other data of your choice, consider referring sequences from an evalua¬
tive (non-structural) perspective.

Notes

1 Variation analyses often check the statistical significance of quantitative distribu¬

tions. However, I have not done so in this chapter: the extremely skewed distributions
(i.e. the contrast between cells with a very low percentage of tokens and those with
a high percentage of occurrences) seem to be sufficient to my claims in this chapter.
Part III Conclusion
’ f

N
We began this book by observing that discourse analysis is widely recognized
as one of the most vast, but also one of the least defined, areas in linguistics.
We saw in previous chapters that one reason for this is that our understanding
of discourse is based on scholarship from a number of academic disciplines
that are actually very different from one another. The chapters in part II
have not only reviewed six different approaches to discourse, but illustrated
how these approaches address some important issues bearing on the produc¬
tion and interpretation of utterances as both linguistic and social practice.
In this final part, I compare approaches to discourse in a more systematic
and abstract way that I hope will enable us to^considier them not just as tools
for analysis, but as possible candidates for theories of discourse. My discussion
is grounded in three issues that discourse analysts face: the relationship be¬
tween structure and function, text and context, utterances and communica¬
tion. Although these issues are quite different in some ways, they all require
reconciling what is often defined as part of “language” (or at least, part of a
linguistic system) with what is not typically defined as part of language. Many
scholars are comfortable making such a distinction in principle, but find it
difficult to maintain such a distinction in practice. The three issues on which
my comparison is centered all obscure the borders between different analytic
domains, and are thus partially responsible for the vastness of discourse
analysis.
 .

.
9 Structure and Function

1 Introduction

Discourse has often been viewed in two different ways: a structure, i.e. a unit
of language that is larger than the sentence; and the realization of functions,
i.e. as the use of language for social, expressive, and referential purposes
(chapter 2). Despite the diversity of structural approaches, structural analyses
all focus on the way different units function in relation to each other: they
disregard “functional relations with the context of which discourse is a part”
(van Dijk 1985: 4). Since it is precisely the relationship between discourse and
the context of which discourse is a part that characterizes functional analyses,
it might seem that the two approaches have little in common.
All the approaches discussed in this book, however, incorporate both struc¬
ture and function into their analyses. After describing how approaches starting
from structure lead to function, and vice versa (section 2), I compare the ap¬
proaches more explicitly in terms of how they relate individual utterances to
structures holding across utterances (section 3). We will see that all the
different approaches to discourse analysis try to bridge or somehow combine
the two analytical worlds of structure and function: not only do the types of
structures analyzed in discourse lead to the identification of functions, but the
types of functions analyzed in discourse are linguistically realized in ways that
create structures (section 4).

2 Structure and function in discourse analysis

Whereas some approaches to discourse begin from structure and lead to

function (2.1), others take the opposite route (2.2). Approaches that are
340 Conclusion

theoretically oriented toward either structure or function, then, actually end

up combining structure and function in practice.

2.1 From structure to function

I discuss two approaches in this section: conversation analysis (2.1.1) and

variation analysis (2.1.2). Although both begin with structure (but at a differ¬
ent location within “text”; chapter 10), both also bring function into their
means of analysis.

2.1.1 Conversation analysis

One of the central assumptions of conversation analysis (CA) is that interac¬
tion is structurally organized. Rather than posit generalizations about struc¬
ture apart from the details of specific events, however, it is participants
conduct itself that must provide evidence for the presence of units, existence
of patterns, and formulation of rules. To this end, CA searches for recurrent
patterns, distributions, and forms of organization in talk. One structure dis¬
covered in talk is the adjacency pair: a sequence of two adjacent utterances
produced by different speakers, and ordered as a first part and second part.
Adjacency pairs are also “typed” so that a first part requires a particular
second part or range of second parts.
In addition to analyzing structure, CA also seeks to discover the solutions
that people construct for recurrent organizational problems of social life. One
such problem is the exchange of turns at talk. This “distribution” problem is
resolved by the operation of a series of ordered rules that operate at turn
transition locations. One might say, then, that turn-taking rules have the
function of allocating turns at talk.
The description of turn-taking rules (and other devices, see below) as having
a “function” might seem as though CA is an approach to discourse that is
explicitly structural and functional. Because CA derives functions from se¬
quential structures and regularities that hold across utterances, however, CA
does give structure a more central methodological importance than function.
Whatever can be established as serving a function - linguistic devices, rules,
a one second silence (Jefferson 1989), laughter (Jefferson 1984) - serves that
function only by virtue of its sequential arrangement with other parts of talk.
CA tries to uncover patterns in the way participants use particular terms or
devices through a distributional method of analysis akin to a method used in
structural linguistics whereby items are “arranged” in relation to one another.
In our sample analysis in chapter 7, for example, we characterized the sequen¬
tial “arrangement” of a linguistic construction (although the construction was
characterized without consideration of its linguistic structure as “there + BE
+ ITEM”) both within and across turns, and in relation to (and independent
 Structure and Function 341
of) other sequentially definable units (questions, answers, topics). Thus, the
distrust of generalizations that are neither empirically grounded nor appliable
to specific events extends to the analysis of functions within CA. What CA
finds to be an antidote to premature generalizations about function is the
recurrence of an item in a particular slot in a sequential structure. Put another
way, CA analyses of sequential structure provide a methodological resource
for the derivation of function.
I noted above that CA assumes a structural framework at the level of
relationships between utterances, rather than linguistic structure. Note that
this does not preclude linguistic structure from serving a function in a text -
a function derived (as just noted) from recurrent observations about its loca¬
tion in sequential structure. Just as my discussion of the “functions” of “there
+ BE + ITEM” in chapter 7 was based on its sequential location, so too I
suggested that an understanding of the functions of different types of “there”
(demonstrative, adverbial, pronominal) could be based on observations about
sequential structure. Thus, even the CA discovery of the functions of language
structure is based on sequential distribution of a form in relation to conver¬
sational structure.
In sum, CA is a structural methodology for the analysis of talk. The sequen¬
tial regularities and patterns that it discovers, however, reveal how specific
features of talk are used by participants to accomplish tasks that are central
to the organization of talk, i.e. how they function to create solutions for
recurrent problems of talk.

2.1.2 Variation analysis

Although variation analysis also searches for structure, the means by which it
does so, the structures that it discovers, and the units within those structures
all differ from CA. Discourse structure for variationists is found in texts
defined as either primarily monologic (e.g. narratives, lists) or dialogic (e.g.
question-answer or request-response sequences). In the former case, structure
is found within a discourse unit whose external boundaries and internal
relationships define an autonomous unit that is largely independent of how it
is embedded in surrounding talk (cf. Goodwin 1984; Jefferson 1978). In the
latter case, structure is found at a level of underlying actions and reactions
that are related to one another through sequencing rules, and that are related
to utterances through rules of production and interpretation (see section 2.2.1
on speech act theory).
The variationist view of linguistic variables also builds upon the notion of
structure. A linguistic variable is part of a system of interrelated parts: it is a
set of linguistic realizations defined in relation to a single underlying form or
representation that occupies a “slot” (a phonological segment, a syntactic
constituent) in a grammar. Furthermore, the linguistic environment by which
variables are constrained is a structured environment: “factors” are segments
342 Conclusion

of text that themselves can have different values and impact upon the occur-
rence/non-occurrence of the variable. Recall, for example, our analysis of
textual constraints on referring terms (chapter 8): we tried to identify features
of lists with enough precision to make quantitative comparisons of referring
expressions in different textual environments. To this end, we found that the
distinction between categories and subcategories, and the presence or absence
of category boundaries between first-and next-mentions of a list item, con¬
strained the explicitness of a referring term (full noun versus pronoun). Thus,
factors in the linguistic environment create textual constraints on which vari¬
ants are dependent.
I have been suggesting so far that the structure assumed by variationists is
both syntagmatic and paradigmatic. Descriptions and comparisons of different
text-types focus upon syntagmatic structure, and analyses of textual con¬
straints on variation focus on paradigmatic structure. Thus, variationists as¬
sume discourse structure in terms similar to those used by structurally oriented
linguists: “units” are similar to constituents, “information” is propositional in
nature, “structures” are syntagmatic and paradigmatic arrangements of items
in recurrent patterns.
When we consider the locus of variation a bit more closely, however, we see
that function also enters into variationist approaches to discourse. Variation¬
ists have to establish which forms are in alternation with one another. Doing
this at a discourse level has sometimes led analysts to shift from a requirement
of referential equivalence to one of functional equivalence. As I noted in chapter
8, however, the locus of analysis from which variationists begin a study of
text-related variation may not be the point at which they end such a study. The
adoption of a referential basis for identifying equivalents does not allow us to
ignore functions: in fact, it may end up highlighting the functional constraints
(and thus functional differences) at another level of analysis. This result is the
opposite of what happens when we begin with functional equivalents.
Let us consider this point in a bit more detail in relation to referring terms.
A variationist analysis of referring terms can define equivalence either referen-
tially or functionally. Depending upon one’s definition of equivalence, the forms
upon which one would focus would be quite different. Analyses of referential
equivalents would begin by focusing upon the different forms used to realize
the same referential intent. An example from chapter 6 (presented here as 1)
illustrates. Sue is describing how her sister-in-law married someone who was
not completely Italian:

(1) sue: (a) So she was the oldest daughter.

(b) And she broke the ice for me like she said.
(c) Y’know, for Tony, my husband.
IVer: (d) How did you meet Tony?

A bit later, Sue is describing how she met Tony, and she uses two other
expressions that have the same referential intent:
 Structure and Function 343

(2) (a) So anyway, this real dark looking, gangster looking guy, comes out
the door,
(b) and he looked like an old man to me.
(c) And he looked like a gangster.

(1) and (2) contain four different referring expressions that all realize the same
referential intent: Tony, my husband, this real dark looking, gangster looking
guy, and he are alternative ways of conveying a single entity to which Sue
intends to refer. Note, however, that the referring terms vary in terms of their
functional relevance (see discussion in chapter 6). What a variationist ap¬
proach would do is operationalize the functional differences among these
expressions as constraints on their use (along with whatever structural con¬
straints also seemed to be important).
Analyses of functional equivalents, on the other ha%d, focus on the realiza¬
tion of a single function. Although such analyses might begin by focusing upon
a single form that would realize a variety of different referential intents, they
also broaden their focus to other forms that fill the same function. The
indefinite description “this X” (used as first-mention), for example, has been
said to have a particular function: it displays a negative evaluation of a referent
(it is pejorative) that will become a topic of next talk (Prince 1978; Wright
and Givon 1987). One of the examples above, this real dark looking, gangster
looking guy (3a), used “this X” in a way that seems consistent with this
function. (Note that if we want to consider “this X” as a first-mention in (2),
we can do so only if we consider it in relation to the shift in information state
accompanying entry into a story world; chapter 6, p. 210.) Not only is this
real dark looking, gangster looking guy an unflattering description (which
continues in the next lines of the story), but, consistent with others’ predic¬
tions, this referent becomes a topic of next talk: the focus of the story initiated
with (2) is how a romantic relationship evolved, such that the referent is a
topic at both sentence and text levels (Schiffrin 1988b). (3) is another example:
a speaker uses “this X” while complaining about the demands of the church.

(3) (a) And in September, the beginning of October, they have this block
collection.
(b) And you are supposed to give twenty-five dollars.
(c) Each family is supposed to give twenty-five dollars.
(d) Well, they- I used to give whatever I could afford, before.
(e) And now, “Well, Father, if I catch you, fine. If not, sorry about
that.”

Since block collection is being viewed negatively - and is the text topic
underlying the next few utterances - this would be another example in which
“this X” realizes the suggested function. The distribution of “this X” would
be checked in all other texts in a sample to see whether it is a referring
expression whose uses are functionally equivalent - even though different
344 Conclusion

tokens of this referential type would of course be realizations of equivalent

referential intents. Pursuing this a bit further, one could then examine the
distribution of “this X” by different speakers (their social characteristics), in
different text types, different modalities, and different registers.
Although a functionally based analysis of “this X” could continue in the
way suggested above, it would also have to identify other forms that might
serve one (or more) of the same functions as “this X” (e.g. “this one X,” “one
X”). But once we do this, we are back to the study of formal variants - not
formal variants that are referentially equivalent (since they would probably be
used to evoke different referents), but formal variants that are functionally
equivalent. Thus, as I noted in chapter 8, regardless of the locus of analysis
from which variationists begin a study of text-related variation, they may end
up at a destination quite different from the one anticipated, and find that they
reached that destination using a path that they may have wanted to avoid.
Thus far, I have been suggesting that the variationist focus on structure leads
to an analysis of function. I want to also suggest that the sorts of structures
(both within and across clauses) that might be considered by variationists can
serve text-level functions. This interdependency also forces an analysis of both
structure and function.
In chapter 2, we discussed an example in which a speaker (Henry) used five
clauses to introduce the main characters of a narrative. I present this example
here as (4):

(4) (e) Because there’s this guy Louie Gelman,

(f) he went to a big specialist,
(g) and the guy . . . analyzed it wrong.
(h) In fact his doctor didn’t know,
(i) and the specialist didn’t know.

Note that the introduction of these characters is critical to Henry’s story:

before Henry can use his story to make a general point about doctors (doctors
are not God!), he has to introduce the characters whose experience and
interaction with one another provide a specific instance of this general point.
The old/new information structures in clauses (e), (f), and (g) provide this
introduction.

(e) Because there’s this guy Louie Gelman [new],

(f) he [old] went to a big specialist [new],
(g) and the guy [old] . . . analyzed it wrong.

By placing new information in the context of familiar information within each

clause, Henry facilitates the introduction of characters important to the point
of his story. Thus the information structure within the clauses serves the
text-level function of introducing the characters whose experience is central
to the story.
 Structure and Function 345

Structures across the clauses in (4) also facilitate the general point to be made
by the story. Clauses (e) to (g) above mention two referents, twice each. The
first-mentions and next-mentions form referring sequences, presented below:

First-mention Next-mention
Referent 1 this guy Louie Gelman he
Referent 2 a big specialist the guy

Referent 1 is evoked initially through an indefinite NP (this guy) and a proper

noun (Louie Gelman)-, its next-mention is a pronoun (he). Referent 2 is evoked
initially through an indefinite noun (a big specialist); its next-mention is a
definite noun (the guy). Sequences such as these recur throughout discourse:
indeed, the clauses in which we can find dependencies between referring terms
need not even be adjacent to one another. These distributional regularities
suggest that speakers construct sequences even apart from these that we
intuitively identify and label as particular chunks of discourse (e.g. answer,
story). The point here, however, is somewhat different. The referring terms in
(4) shift from explicit to inexplicit, from indefinite to definite. These shifts
reflect the incorporation of the referents into a story world. Put another way,
one reason why Henry can go on to make the point of his story is because the
characters central to that point have been structurally embedded in the story
world. Thus, a text-level function is allowed by structures that hold across
clauses.
We can make a similar point by considering temporal structures across
clauses. Formal analysts often try to model temporal interpretation in dis¬
course, taking into account how different linguistic forms and meanings within
clauses influence temporal interpretations across clauses (e.g. Dowty 1986;
Dry 1981, 1983; Fleischman 1990; Hinrichs 1986; Kamp and Rohrer 1983;
Partee 1984; Webber 1988). We noted in chapter 2 that a modification of the
basic rule of narrative ordering - in which reference time shifts with each
complicating action clause - helps Henry make the point of his story: Henry
reports his friend’s encounter with a specialist before the friend’s temporally
prior encounter with his own doctor. Henry s deviation from a temporal order
in which time moves forward can certainly be formally represented - but it is
not formally motivated. Rather, the shift back in reference time has the
discourse function of helping Henry increase the membership of the group of
professionals who was wrong, adding to the overall point of his story.
In sum, variationists begin with structure: they search for structure within
monologue and dialogue, they assume that the underlying form whose differ¬
ent realizations constitute a variable are part of the structure of language, and
they assume that variants are structurally constrained. Yet they are also led
to analyses that build upon an interdependency between structure and func¬
tion: analyses of textually constrained variants veer between structure and
function, and analyses of text-level functions reveal that these functions can
be fulfilled by structure within and across clauses.
346 Conclusion

2.2 From function to structure

This section discusses approaches to discourse that begin with function and
end up incorporating structure into their analyses: speech act theory (2.2.1),
ethnography of communication (2.2.2), interactional sociolinguistics (2.2.3),
and pragmatics (2.2.4).

2.2.1 Speech act theory

Speech act theory begins by grounding its units of analysis in a system outside
of language (in speaker intention and action, in our knowledge of constitutive
rules), but its application to discourse leads to a structural approach in which
units are sequentially arranged in certain patterns. The units focused upon by
speech act theory are a relatively small subset of language functions: communic¬
ative actions that can be realized in a single sentence, labelled by a particular
culture, and often performable with a verb naming the act, e.g. promise, thank,
offer, hint, regret, conclude. As Searle (1979: 22) notes:

If we adopt illocutionary point as the basic notion on which to classify

uses of language, then there are a rather limited number of basic things
we do with language: we tell people how things are, we try to get them
to do things, we commit ourselves to doing things, we express our feeling
and attitudes and we bring about changes through our utterances.

Speech acts are discovered by examining texts and contexts (chapter 10):
conditions underlying the realization of a particular act are both linguistically
met and contextually satisfied (see also section 2.2.3). These conditions are
the basis for the way we classify speech acts (and thus identify an utterance
as a particular type of “unit”), for the way a single utterance can have more
than one function (i.e. be more than one “unit”), and ultimately, for the
patterns formed by the use of utterances as acts during social interactions.
There are several reasons why the functional thrust of speech act theory
noted above leads to discourse structure. We saw several examples in chapter
3 of speech acts that were definable only sequentially: answers, expansions,
and accounts. This means that identifying the speech act function of an
utterance often requires looking at where it occurs in relation to other utter¬
ances, i.e. its location in what is assumed to be an orderly arrangement of
other units. To take a simple (hypothetical) example, compare the role of yes
in (5a) through (5c):

(5a) S: Do you want to go to the movies tonight?

H: Yes.
(5b) s: That’s a beautiful view.
 Structure and Function 347
H: Yes.
(5c) s: Do you promise you’ll be there?
H: Yes.

In (5a), the function of yes is answer to a question; in (5b), agreement with

an assessment; in (5c), making a promise. Since each of these functions is
obviously dependent upon the preceding act, we must conclude that the speech
act functions of yes in (5) result from its location in a discourse structure
defined through the functional identity of the first constituent in that struc¬
ture. Other approaches (e.g. conversation analysis) that have reanalyzed speech
act sequences suggest that the effect of structure on speech act function is even
more penetrating than we have suggested: there are numerous occasions where
the illocutionary force of an utterance - the very bedrock of speech act theory
- may not even be clear until it is followed by another utterance. Such
structures thus have a retrospective effect on the identity of a prior utterance,
such that some speech act categories depend on what precedes and/or what
follows the utterance in which they are realized.
The multifunctionality of utterances can be accommodated in the speech act
view of structure. Since an utterance can perform more than one act at a time
(have more than one function), a single utterance creates different response
options for a next utterance. This gives our exchanges a certain degree of
flexibility: multiple interpretations of X can lead to multiple possibilities for
responding to X. In analytical terms, this means that whatever sequence results
may be the outcome of very different underlying functional relationships. It
also means that whatever structure results is comprised of only one of a range
of (possibly) very different functional units.
In sum, speech act theory leads to the discovery of well-formed sequences
of discourse. This outcome is exactly what is aimed for by structural analyses
- despite the fact that the structures are functionally (and contextually) based.

2.2.2 Ethnography of communication

The ethnography of communication is a functional approach to discourse:

language is always constitutive of some portion of social and cultural life,
either indexing a prior reality, or creating a new reality, for participants. The
meanings and functions of language are personal, social, and cultural func¬
tions that mesh with other systems of actions and beliefs. Consistent with the
comparative and relativistic premises of this approach, however, little can be
assumed about such functions: it is up to ethnographers to discover both the
range of and ways of realizing such functions.
What allows language functions to mesh with context - and to be discovered
through contextual analyses - is the fact that context itself is not random and
disorganized: rather, context has a communicative structure built from a
variety of communicative components that co-occur and alternate with one
348 Conclusion

another. In our sample analysis of questions in chapter 5, for example, we

added different aspects of context to our analysis of particular utterances in
a way that built upon an underlying assumption that the organization of
communicative events has a systematic effect on (helping to define the function
of) constituent acts. We ended up suggesting that all the utterances that we
called “questions” share a general function: they all seek a response that
facilitates the transmission of information in the speech event. And this sug¬
gested that the smaller “units” of an ethnographic analysis (e.g. the utterances
used to realize speech acts) may be mutually constitutive with the larger units
(e.g. the speech events) in which they occur.
The procedure that I have just sketched suggests an intentional interdepend¬
ency between structure and function: the communicative structure of a speech
event reveals the function of constituent acts, and the function of the act itself
contributes to the structure of the event. Thus, an analysis of the structure of
a speech event can end up revealing how that structure accommodates the
functions realized by the event. Put more generally, “the recognition of social
function brings recognition of new structure, transcending conventional com¬
partments” (Hymes 1974c: 150).
The interdependence between the communicative structures of context and
the function of individual utterances can be put to further use in relating
structure and function: systematic analysis of communicative contexts can help
to explain the multifunctionality of utterances. We spent a great deal of time
in chapter 3 discussing how Y’want a piece of candy? could serve as three
speech acts (question, request, offer), and in chapter 5 discussing how Can/
may I help you? could serve as two speech acts (question, offer). In both
discussions, I noted that the interpretation of an utterance as an offer depends
on S’s belief that A (the act) would benefit H and that S is able to do A. What
counts as beneficial and within the realm of S’s ability, however, is contex¬
tually relative, and rests partially upon S’s assessment of H’s goals in a speech
event. One reason why we assume that a patron in a library can understand
a librarian’s May I help you? during a reference interview as an offer, for
example, is because we know that the patron has come to the librarian’s desk
for help and that the librarian expects to be able to provide help. Similarly,
during one of my sociolinguistic interviews, I interpreted a question addressed
to me by an informant when her husband entered the room, You want Henry
to talk more for you?, as an offer to get her husband to talk. This was
consistent with my knowledge that one of my goals during that interview was
to record people’s talk and with the informant’s definition of the situation and
her expectations of her own role and responsibilities.
Although the point that the fulfillment of felicity conditions is contextually
bound (and thus contextually relative) may seem obvious, it helps to explain
why other questions about H’s wants are not offers. The hypothetical ex¬
amples in (6) suggest some other functions served by such questions.

(6a) You want a new car? You have to work for it.
 Structure and Function 349
(6b) You want to stay here for vacation or go away?
(6c) You want to get the phone? I’m in the shower.

None of the questions in (6) are offers by S to provide A for H. (6a) functions
like a conditional sentence (cf. “if you want a new car . . . ”, Schiffrin 1992c),
not an offer by S to buy H a car. (6b) seems to be a straightforward informa¬
tion question. (6c) acts as a request. In brief, the interpretation of A as bene¬
ficial and possible depends on situational parameters. Thus, analyzing the
interdependence between contextual structures and utterance functions can
help us disentangle multiple functions of utterances from one another.
The analysis of the communicative structure of contexts so important to an
ethnographic approach to discourse can lead to still another way of relating
form and function: the identification of functional ^alternatives. An ethno¬
graphic approach makes no more assumptions about the mapping between
form and function within a sentence than it does within discourse. One
consequence of this is that a tremendous variety of different forms may serve
the same interactional function. The formal properties of an utterance may
very well be compatible with its functions: we can sometimes use sentence
meaning and structure as clues by which to identify function (e.g. our analysis
of offers as questions in reference interviews in chapter 5, section 3.2.1). But
the functions of other utterances seem to be independent of their forms. For
example, a librarian’s opening offer in a reference interview may be inter¬
changeable with eye contact, a patron’s query in a reference interview may be
issued as a statement, a request for clarification can be made by repetition
(with or without rising intonation). Although these alternatives can all sub¬
stitute for questions, they can do so only within very specifically defined act
sequences and settings.
Another example illustrates that functional alternatives can also depend
upon participant. Take requests for clarification from a researcher and an
informant during sociolinguistic interviews. A researcher’s requests for clari¬
fication are functionally similar to information-seeking questions that build
on an informant’s answer to a prior question as a way of progressing from
one topic to another in a stepwise fashion (chapter 7). But an informant’s
requests for clarification are more similar to informants’ own information¬
checking questions: both help to tailor the informant’s answer to the re¬
searcher’s expectations and knowledge. Again, the general point is that a
systematic analysis of the communicative structure of context helps analysts
to define functional alternatives.
It is important to note that the discovery of a cluster of different forms that
fulfill a similar function has a consequence not just for the analysis of how
specific acts are realized, but also for the analysis of discourse organization
itself. As we noted above, functional equivalents are defined only relative to
particular contexts. The discovery of contextually defined alternatives suggests
the existence of higher level functions that provide for slots in still other
sequential structures. To return to the reference interview for a moment, we
350 Conclusion

can very well identify question-answer sequences within that interview, but
we can also identify a structure of query-resolution that encompasses almost
the entire reference interview. In some ways, a query is very much like a
question: a patron’s purpose in making a query is to elicit information that he
does not have, but which he presumes the librarian to have. But a query
realizes these question functions at a discourse level, such that other utterances
(which also serve speech act functions) are pressed into service at a higher
level of analysis.
To summarize: I have been suggesting that an ethnographic approach to
discourse finds the function of a specific unit (e.g. a speech act) by locating
that unit within a context that is partially structured as another unit (i.e. speech
event). Thus, the function of an utterance is intricately tied to the communi¬
cative structure of speech events.

2.2.3 Interactional sociolinguistics

Interactional sociolinguistics is also a basically functional approach to dis¬

course that incorporates structure into its analyses. Like other functional
approaches, interactional sociolinguistics focuses upon language use: it is
concerned not just with the way people intend language to serve referential
meanings, but also with the social, cultural, and expressive meanings stemming
from how utterances are situated in contexts. One of the key tasks of the
interactional approach is to describe the contexts in which language is used,
and to relate the meanings and specific functions of an utterance to those
contexts. The range of functions allowed-by interactional sociolinguistics in¬
cludes illocutionary force, but it is also more inclusive than most of the other
approaches discussed: face strategies, the creation of involvement, and the
realignment of participants can all be seen as functions. (Note that these
functions are more interactively and interpersonally situated than need be the
case in ethnographic analyses.) The willingness to allow so wide a range of
functions (and multiple functions of a single utterance) is in keeping with the
theoretical belief in the importance of context and its methodological counter¬
part of progressive contextualization: the addition of more information about
the contexts (both general and specific) in which a single form occurs allows
us to identify more functions of that form.
Note, however, that the discovery of functions - even those that are highly
contextualized - has an effect on structure. We can make this point by
returning to a hypothetical example from chapter 2, presented here as (7):

(7) jan: (a) Are you free for lunch today?

Barbara: (b) I have to advise students all day.

Jan’s Are you free for lunch today? can have a number of functions - all
grounded in contexts - other than a questioning function. We noted (in
 Structure and Function 351
chapter 2) Jakobson’s (1960) differentiation of a conative function from an
emotive function, according to which aspect of the situation (addressee, ad-
dressor) was the focus of speech. Jan’s utterance can be said to function as a
display of solidarity with Barbara (a conative function) and to display gender
identity (an emotive function) by using a conventionalized form of indirectness
to check on the other’s availability. Although these functions seem to have
little to do with structure, they do set limits upon what follows next in the
interaction. If Are you free for lunch today? is heard as a display of solidarity,
for example, it would be considered impolite to provide a negative response
by just saying no (cf. chapter 3, pp. 77-8 and chapter 4, p. 113-15). A more
appropriate response would draw from the same functional repertoire: make
a counter offer for lunch (e.g. How about Friday?), show appreciation (by
saying thank you), or provide a reason for non-availability (e.g. I have to
advise students all day). Thus, the linguistic realization of contextualized
interpersonal functions sets up expectations about what comes next - expecta¬
tions that act as a cognitive foundation for sequential structure.
It is also important to note that some interactional sociolinguistic studies
do concern themselves explicitly with structure, e.g. with the “traffic rules
of social interaction” (Goffman 1967a), participation structures (Goffman
1981a), or patterned exchanges of moves (Goffman 1971a, b). (Since moves
are similar to speech acts that are sequentially located and sequentially relev¬
ant, I compare such structures to speech act structures in section 3.) Some
interactions may thus have structures that can be both indexed (cf. the role of
contextualization cue) and constructed by particular utterances.
Before we close this section, it is important to comment on the importance
of the term “interaction” in interactional sociolinguistics. Discourse is inher¬
ently an interactive activity in which what one person says and does is both
response to past words and actions, and input to future words and actions.
The focus on interaction gives interactional sociolinguistics a more elaborated
view of participant role, identity, and relationship than the other approaches
discussed. It also provides another way that structure and function enter into
this approach. We noted in chapter 4 that much of what we say is oriented
toward reception by a hearer: speaking a language is a process that requires
symbolically putting oneself in the other’s place in order to know how to tailor
one’s information (syntactically, semantically, and pragmatically) so that it
will be comprehendable to that other. Sequential structures in talk are based
not just upon the communicative functions of linguistic expressions per se, but
also upon the emerging set of understandings that participants gain through
the give and take of interaction - through the process of orienting toward the
other person. As critical to the construction of coherence as speaker intention
(cf. speech act theory, pragmatics), then, is the reciprocity of perspective
between self and other: each builds upon what the other has said to offer a
meaning that the other can adapt, and to which the other can respond. In
effect, then, the structure of interaction arises from the same “chain of cere¬
mony” that is responsible for the maintenance of self and other: “individuals
352 Conclusion

must hold hands in a chain of ceremony, each giving deferentially with proper
demeanor to the one on the right what will be received deferentially from the
one on the left” (Goffman 1967a: 85). That is, each utterance in an interaction
receives part of its meaning from what another person offered before, and
gives part of its meaning back to that other person to use in what comes next:
it is from this chain of self/other reciprocity that the structure of interaction
emerges. Language enters into this reciprocity: because of its contextualizing,
indexical functions, language stands in a mutually constitutive relationship
with the self, the other, the self/other relationship, and the contextualized mean¬
ings that are continually negotiated during interaction.

2.2.4 Pragmatics

Functional explanations of language build on the very basic idea that language
has functions by finding explanations for language structure in systems out¬
side of language itself. By this criterion, Gricean pragmatics is a functional
approach to language (Levinson 1983: 97): its main constructs are located
outside of language per se in speaker meaning (i.e. speaker intention) and
rational principles of human communication (i.e. the cooperative principle).
Although communication between people is based partially on shared knowl¬
edge of structural representations and stable meanings in a linguistic code (i.e.
a text; chapters 10 and 11), communication also depends upon a cognitive
ability to use context to make inferences. As Sadock (1985: 141) puts it,
“many of what were conceived of as essentially inexplicable structural facts
of language are really reflections of the fact that natural language is used by
real speakers in real contexts to accomplish real goals.” But rather than go
into detail about what is involved in human communication (as Sadock would
suggest, the real goals of “real speakers in real contexts”), Grice posits as
the underlying basis of conversation a general cooperative principle, and as the
underlying basis of the CP human rationality itself. Consistent with the func¬
tionalist idea of searching for explanations for language structure in systems
outside of language itself, Gricean analyses rely upon the CP and its attendant
maxims for explanations of language structure.
Since Gricean pragmatics is not concerned with discourse per se (the “inex¬
plicable structural facts” for which it provides functional explanations are
usually syntactic or semantic), we have to search a bit for its impact on
discourse level structures. One example we have already seen comes from our
analysis of referring sequences. In chapter 6, we used the maxims of quantity
and relation to describe referring sequences: a particular referring term in
[slot 1] constrains what terms will occur in [slot 2], and, even further, the way
a sequentially next term will be interpreted. Our analysis showed that the
maxims of quantity and relevance constrained the choice between definite and
indefinite, and explicit and inexplicit, mentions of a referent: thus, pragmatic
constraints were partially responsible for different discourse structures.
 Structure and Function 353

Another example of how the Gricean approach to discourse structure is

functionally grounded comes from the more traditional discourse analytic con¬
cern with question/answer structure. As we discussed in earlier chapters, dif¬
ferent approaches to discourse analysis view the source of coherence between
question and answer (what allows an utterance following a question to be heard
as an answer?) very differently. Speech act theory, for example, finds sequential
coherence to stem from well-formed structures of speech acts. Conversation
analysis depends most heavily on the identity of the sequential slot “question”
- arguing that whatever follows what is heard as a question will have sequential
relevance as an answer. I briefly mention these two other approaches just
because the Gricean approach stands in such stark contrast: what matters for
the Gricean approach is not the identity of the discourse slots (and the basis of
this identity can certainly differ depending upon approach), but the cooperative
principle. (8) illustrates (this is Levinson’s 1980: 102f example 18).

(8) a: Where’s Bill?

B: There’s a yellow VW outside Sue’s house.

Levinson argues that B’s utterance violates the maxims of quantity and relev¬
ance. Because we try to interpret B’s utterance as cooperative at some level,
however, we infer that Bill may be at Sue’s house, i.e. we interpret B’s
utterance as an answer to A’s question. Thus, it is because people assume
cooperation (at the level of either the general CP or specific maxims) that an
utterance can be heard to occupy a particular sequential position. Put another
way, the sequential identity of an utterance emerges from (is a product of) the
cooperative principle: an utterance is heard as an “answer,” and can thus be
a constituent in a question/answer structure, only because interlocutors share
an underlying assumption of cooperation.
In sum, the Gricean approach to discourse is basically a functionalist ap¬
proach to language: explanations for language structure are sought in a general
cooperative principle that rests ultimately upon human rationality. Both the
constituents of discourse structure and their orderly arrangement as coherent
text thus arise because of the impact of communicative principles on the
linguistic realization of speaker meaning at different points in time.

3 Comparisons among approaches: from

utterance to discourse

Although we differentiated a structural from a functional definition of dis¬

course in chapter 2, our review of how different approaches to discourse
incorporate both structure and function within their analyses suggests that this
354 Conclusion

is by no means as absolute a distinction in practice as it is in principle: many

discourse analysts end up trying to bridge or somehow combine the two
analytical worlds. Before concluding this chapter, I more directly compare the
way different approaches combine structure and function.
A convenient way to begin our comparison is in terms of how different
approaches relate what is said within individual utterances to structures hold¬
ing across utterances. We will see that the aspects of a constituent utterance
that have an impact on discourse structure vary for different approaches: some
structures are based more on the functions of individual clauses than others.
Similarly, the functions that are proposed vary: not only is there a different
range, but some are more situated in ongoing talk than others, some more
abstract than others. Finally, both structures and functions are more or less
grounded in text, or in an interplay between text and context.
We noted earlier that conversation analysis assumes a structural framework:
relationships between utterances are structurally organized. CA subordinates
the analysis of function to the analysis of structure: the function of an item
can be discovered only by analysis of its location relative to other items in
talk, i.e. its sequential distribution. CA also relates properties of utterances to
regularities across utterances. Turns are exchanged, for example, because of
a distribution system that holds across different parties’ utterances. Yet what
enables turn-taking is not just a set of context-free sequentially applicable and
recursive rules, but the fact that the rules are context-sensitive (Sacks et al.
1974): the rules apply at particular locations (turn-transition spaces) defined
within utterances. Preference structures provide another example of how CA
relates properties of utterances to structures across utterance boundaries.
Preference structures are defined across turns at talk: one second-pair part (e.g.
an agreement) is preferred over another (e.g. a disagreement). Evidence for a
preference, however, is found in the distribution of items (e.g. particles, pauses)
within an utterance (an utterance located as the second part of a two-part
pair). Thus, what occurs within utterances facilitates the emergence, and the
discovery, of structures across utterances.
We also noted earlier that CA distrusts generalizations that are not empiric¬
ally grounded, and that this distrust extends to the analysis of functions and
structures (both across and within utterances). Although this need not preclude
linguistic structure from serving a function in a text, CA itself makes abso¬
lutely no assumptions about linguistic structure and how it may be related to
structure across utterances.
In sharp contrast, variation analysis is the approach to discourse that incorp¬
orates linguistic structure and methodology most extensively. Variation ana¬
lysis considers how surface features of utterances, and actions performed by
utterances, build structures across clauses; these structures lead to coherence.
Our analyses of particular text-types (e.g. narratives, lists) in chapter 8 showed
that one way the surface features of utterances contribute to coherence is by
realizing functions associated with that text-type, at a particular location
within that text. Recall, for example, how the distribution of active and stative
 Structure and Function 355
predicates differs across narratives and lists: we said that this reflects the
differential provision of temporal and descriptive information within the two
text-types. Surface features of utterances are also sometimes analyzed as vari¬
ables: alternative realizations of referential or functional equivalents. Different
realizations of a variable may be sensitive to location in a text. Again, as we
saw in chapter 8, pronouns and nouns in lists are sensitive to the category
level of a list-item and to the location of a next-mention of that item within
the list. Whereas speakers use pronouns to evoke a general category (X)
regardless of its location in the list, they do not use pronouns to evoke a
specific list-item (XI) if a different list-item (X2) was mentioned after a prior
mention of the item (XI). The use of one variant (e.g. a full noun) rather than
another (e.g. a pronoun), and the way such variants are conditioned by textual
constraints, helps build the overall coherence of a discourse.
. We also noted in chapter 8 that the linguistic variables focused upon by
Labovian variation analysis, and the abstract speech actions focused upon
by Labov’s method of comprehensive discourse analysis, seem very different
at first. Both constructs, however, rest upon a single assumption about dis¬
course coherence: coherence depends upon the sort of complex hierarchical
organization that “is plainly derived from the linguistic analysis of phonology
and grammar” (Labov and Fanshel 1977: 350).
Of course variables and abstract speech actions do differ: they contribute to
coherence in different ways, and at different locations in the complex hierar¬
chical organization on which coherence depends. Traditionally defined lin¬
guistic variables are segments of underlying representations that are encoded
in surface forms of language: it is possible (in fact, necessary) to view a variable
as a closed set of options. Abstract speech actions are quite different. Not only
are they not encodable in language alone, but they are not a closed set of
options. Labov and Fanshel (1977: 358) point out that the existence of a
closed set of options for speech actions seems unlikely because of their con¬
textual dependence: “A very large portion of the social structure enters into
the construction of such speech actions, and, at the present time, a compre¬
hensive grammar of [higher level speech actions, such as] insults or challenges
seems to be out of the question.” However, the degree to which social context
(including prior utterances) enters into the formulation of speech act rules
varies for different actions. Labov and Fanshel (1977: 93-7) suggest, for
example, that a repeated request is a challenge when H fails to comply with
a prior request whose appropriateness had been assumed by S because of S s
beliefs about H’s social competence. Unlike the CA assumption that all talk
is sequentially relevant and sequentially implicative, then, variationists weigh
and measure the importance of sequential embedding. (We see in chapter 10
that variationists treat both text and situation as “optional contextual con¬
straints on language use.) The quantitative methodology used by variationists
segments environments into factors that act as probabilistic constraints on
variants. Even the abstract speech actions that figure in sequencing rules can
be more or less dependent on the talk whose sequences they are helping to
356 Conclusion

construct: the rule for challenges noted above (but not for requests), for
example, relies upon assumptions about role-related competence.
We have seen thus far that variationists use two different constructs (vari¬
ables, abstract speech actions) to relate what is said within individual utter¬
ances to structures holding across utterances; both of these constructs depend
on the assumption that the hierarchical organization of language contributes
to discourse coherence. Abstract speech actions, however, are not linguistic
constituents within a hierarchy: they are language functions. Speech act theory
also focuses on actions. However, the finite set of actions upon which it
focuses are more narrowly, and less socially, defined than those focused upon
by variationists (and other scholars who take a more social interactional
approach to speech acts: Ciccourel 1980; Schegloff 1987). Furthermore, speech
act theorists do not group together surface features of utterances into the
construct of “variable,” e.g. they do not say that the range of illocutionary
force indicating devices that can be used to realize a single speech act are
alternative ways of “filling” an underlying speech act “slot.” (As we note in
chapter 10, however, speech act theorists do use the constructs of “rules” and
mapping relationships among encoded (and encodable) meanings.) Thus, vari¬
ation analysis and speech act theory both assume that it is text-context
relationships encoded within an individual utterance that create structures
across utterances.
We noted earlier that the initial task (and one of the key tasks) of a structural
analysis is the discovery and identification of constitutents. But as both vari¬
ation and speech act theory make clear, units cannot be identified by linguistic
criteria alone. The shift to a level of action, however, threatens the most basic
assumption of all structural analyses: structural units are linguistic units, not
units defined by what a speaker does and intends to do through language.
Thus, once we begin to search outside language itself for criteria by which to
identify structural units, we are forced to deal with the relationship between
language and the contexts of which it is a part - precisely the text-context
relationship that structural analysts purportedly ignore (see chapter 10).
Like conversation analysis, variation analysis, and speech act theory, the
ethnography of communication also focuses on sequential structures across
utterances. Like speech act theory, the constituents of these structures for
ethnographers are often functionally labelled speech acts, whose identity is
related to linguistic features of clauses. Unlike speech act theory, however,
little is assumed about universal inventories of such acts. More crucially, such
acts are not separable from the larger speech events in which they occur: what
seems like the very same utterance may realize radically different acts depend¬
ing on the communicative structure of its speech event. Thus, both functions
and functional alternatives are grounded in communicative structures.
The ethnographic approach thus seems similar, in some ways, to the conver¬
sation analytic approach. Both situate functions (actions and meanings) within
sequential structure. Neither CA nor ethnography assumes the relevance for
one particular speech event of any a priori categories. Furthermore, both
 Structure and Function 357

approaches are built upon particularities and the discovery of generalizations

that account for each particular case.
Despite these similarities, some important differences between the two ap¬
proaches remain. Whereas CA does not venture beyond what is displayed in
a transcript itself (seeking evidence only in participants’ conduct), ethnography
depends on “thick interpretation” (Geertz 1973) of cultural meanings - inter¬
pretation that requires not only extensive fieldwork and participation in a com¬
munity, but also the use of knowledge about a wide matrix of beliefs and actions
in a particular culture. The ethnography of communication also makes more
explicit a two-way relationship between structure and function. This interdepend¬
ence is assumed to hold across “unit” size: just as the communicative structure
of a speech event reveals the function of constituent acts, so too does the function
of the act contribute to the structure of the event. This interdependence is also
assumed to hold across “levels” and conventional compartments of grammar:
just as linguistic structure can reveal social functions, social functions can reveal
new levels of structural organization (e.g. Hymes 1981). Thus, the ethno¬
graphy of communication assumes a two-way mapping between the functions
of utterances (and their linguistic properties as clauses) and discourse structure.
We have been focusing thus far on how functions of language can serve as
constituents in discourse structure. However, not all functions can be neatly
chunked and arranged. We noted in our discussion of interactional socioling¬
uistics, for example, that Jan’s Are you free for lunch today f (7) can function
as a display of solidarity. Although this function sets limits upon what follows
next in an interaction, we might not want to segment and label this function
as a speech act. This is not to say, of course, that Are you free for lunch today?
would not lead to a particular kind of response. It would be inappropriate to
follow a display of solidarity with just no, for example: a more appropriate
response would draw from the same strategic repertoire, e.g. propose another
time for lunch, convey appreciation, explain one’s non-availability.
Our analysis of “speaking for another” (chapter 4) also illustrated the
interactional socioliguistic reluctance to always segment and label utterance
functions. Although we did consider this an act that one person takes in
relation to another, it is not an act that can be defined in terms of illocutionary
force or conventional formulae. One reason is that it is an act that is always
overlaid on another act, e.g. one provides an “answer” or an “account” for
another. In addition, the expressive meaning of this act is always relative to
its context, including the sorts of institutional and interactional contexts
whose interpretive relevance can be indexed by contextualization cues. Put in
still stronger (but perhaps overly simple) terms, the act itself is dependent on
the interpretive contexts in which it is situated. But, again, this is not to say
that speaking for another would not lead one to expect a particular kind of
response — a response dependent on its interactional meaning.
I have been suggesting, then, that the acts considered by interactional socio¬
linguists are contextualized functions that set up expectations about what
comes next: they do not delimit a closed set of well-formed structures whose
358 Conclusion

constituent acts have autonomous identities. Although some of these expecta¬

tions do seem to result in structures similar to those posited by Labov’s com¬
prehensive discourse analysis (see e.g. Goffman 1971a, b; Merritt 1976, 1984)
or speech act approaches, not all sequential regularities in talk can be chunked
into discrete, functionally based actions that can be mapped onto utterances.
The interactional sociolinguistic view of structure differs in another way
from the speech act view: the role of rules. A speech act approach to discourse
relies upon three kinds of rules: constitutive rules that create the act per se,
mapping rules that relate an utterance to the constitutive rules, sequencing
rules that relate the actions that result from the mapping rules (rather than
the surface linguistic details of each utterance in the text). Instead of constitu¬
tive and mapping rules, interactional sociolinguistics speaks of norms of con¬
duct (Goffman 1959, 1963), strategies (Brown and Levinson 1987; Gumperz
1982a), and principles (Tannen 1981). When interactional sociolinguists do
speak of rules (e.g. Goffman’s 1963, “traffic rules”), the focus is on the
patterns of conduct created by norms, and the social meaning of conforming
to (or deviating from) such norms. Furthermore, instead of the fixed knowl¬
edge states (constitutive rules) assumed to underlie speakers’ meanings, the
interactional view of speakers’ meanings is that they are situated, inherently
flexible, and multifaceted. Instead of hearers mapping from text/context to
constitutive rules, the interactional view of hearers’ inferences relies upon
many different aspects of an utterance, including the situated meanings and
implications of acts (e.g. chapter 4, section 3.2.2).
Note, also, that implied through the speech act construct of rule is a more
deterministic view of human behavior than that with which most interactional
sociolinguists would be comfortable. By assuming that human behavior is
guided by norms of conduct, strategies, or principles, interactional sociolin¬
guistics allows human behavior - both unintentional conduct and intentional
communication (see chapter 11) - not only to be probabilistic, but to have
meanings and interpretations that can shift from user to user and context to
context (cf. Gricean principles, chapter 6). This view of human action also
frees interlocutors from a dependence upon a fixed inventory of speech act
types (as usually assumed by speech act theorists). Rather, a range of acts (and
act types) can emerge from the particular expectations and obligations that
arise during an ongoing interaction.
We noted above that speech act theory is wed to a relatively deterministic
identification of a limited set of functions whose identity is shared by inter¬
locutors. This assumption means that a process that is important to functional
approaches - the process of locating functions in many different aspects and
types of contexts of an utterance - creates a danger for a speech act approach
to discourse because it creates a sense of “indeterminacy.” It creates no such
danger for interactional sociolinguistics, however. Not only does this ap¬
proach assume that meaning is more protean, but the underlying model of
communication that is assumed does not require symmetry between speaker
intention and hearer interpretation (chapter 11).
 Structure and Function 359
A final point. Interactional sociolinguistics is less eager than the other ap¬
proaches to rely upon structure either within or across clauses (but see discus¬
sion of pragmatics below). As noted above, rather than discover well-formed
structures across clauses, such researchers account for sequential expectations.
And such expectations emerge not from interlocutors’ reliance upon grammat¬
ical (i.e. “core”) structures of what is said, but from both verbal (e.g. prosodic,
phonological, morphological, syntactic, rhetorical) and nonverbal (kinesic,
proxemic) aspects of what is said. These qualities serve as cues used by
participants to infer an interpretive framework in which to identify relation¬
ships between utterances. Thus, interactional sociolinguistic analyses of rela¬
tionships across utterances do not depend upon linguistic structures within
clauses. Rather than assume that speakers’ meanings depend upon knowledge
of a stable system of shared, context-free rules (cf. discussion of intersubject¬
ivity in chapter 11), interactional sociolinguistics assumes that interlocutors’
knowledge is situated: principles and strategies (not rules) underlie language
use. Such principles are inherently flexible and multifaceted, and as noted
above they do not always lead to actions with conventional labels. (This is
similar to the Gricean pragmatics view; see below.) Hearers are assumed to
make inferences by relying upon many different aspects of an utterance,
including the situated meanings and implications of acts, and crucially the
contextualization cues that may signal those meanings.
Like interactional sociolinguistics, pragmatics rejects the assumption that
what lead to regular structures across utterances are stable, shared rules
that allow interlocutors to encode meanings within utterances. The principles
of communication captured by Grice’s cooperative principle are similar to
norms. Although norms are a type of rule (cf. Taylor and Cameron 1987),
norms are more like conventional expectations for behavior than formal algo¬
rithms that map linguistically encoded forms from one level of representation
to another. Furthermore, although the Gricean approach seeks to explain
structural facts of language, these facts are often part of the syntax and
semantics of individual clauses - not structures that hold across clauses. Thus,
unlike variation and speech act approaches to discourse, Gricean pragmatics
does not map from surface form to any higher level functionally labelled unit
at all. Like interactional sociolinguistics, Gricean pragmatics views speakers’
meanings as flexible and situated (although what “situates” meanings is a very
different type of context; see chapter 10). Such meanings need not be realized
in sequentially defined slots in a text.

4 Conclusion

We observed in chapter 2 that the availability of two different perspectives -

one structural, one functional - is partially responsible for the tremendous
360 Conclusion

scope of discourse analysis. If we focus on structure, our task is to identify

and analyze units, determine procedures for assigning utterances a constituent
status, discover regularities underlying combinations of units, and make prin¬
cipled decisions about whether or not particular arrangements are well formed
(or whether they conform to, or violate, sequential expectations). If we focus
on function, on the other hand, our task is to identify and analyze actions
performed by people for certain purposes, interpret social, cultural, and per¬
sonal meanings, and justify our interpretations of those meanings for the
participants involved. Dealing with both structure and function is a delicate
and difficult task - one that requires stepping into two different analytical
worlds.
We have seen that each approach discussed in this book depends upon both
structure and function. This means that each approach inherits the strengths
and the weaknesses of both perspectives (although in sometimes different
ways) - but also that the strengths of one might complement, or at least
partially compensate for, the weakness of the other.
As just noted, structural approaches face an “identification” problem: it is
difficult to find the right criteria to use to decide upon the identity of units.
Nor can we be absolutely sure that particular units (or the criteria by which
we chose them) are the ones that participants in the interaction would actually
rely upon (or even that there is one set upon which all people consistently
rely).1 Functional analyses can sometimes provide an antidote for structural
problems such as these. Several of the approaches discussed view units in
discourse through a functional lens: what does an utterance “do” for inter¬
locutors? Yet functional solutions are far from perfect and themselves raise
other problems. Functional analyses are built on a process of progressive
contextualizing: the addition of more information about the contexts (both
general and specific) of a single form allows the induction of more functions
of that form. As necessary as this process is for the interpretation of function,
it creates a danger: the more we learn about the many different contexts in
which a single form is embedded, the less sure we are that any single interpre¬
tation (based on any single context) is correct. In fact, our entire notion of
what it means to “be correct” - and whether there is a single correct analysis
of something - becomes open to skepticism. Thus, our identification of func¬
tionally based units, and the structures that they form, may also end up being
uncertain and unstable.
Consider, also, that discourse organization and structure is emergent: dis¬
course emerges over time (it is constructed on an utterance-by-utterance basis).
This means that discourse is continually imposing its own contexts and hence
its own sources of indeterminacy. What is said is always in response (in some
way or another) to what was said before and in anticipation of what comes
next (cf. the CA idea of context-shaped and context-renewing). Thus, the
function of an utterance (and hence its identity) must be defined in relation
to (and as appropriate to) a context that is not static, but dynamic, and even
more critically, a context that is still in the process of being interactively
 Structure and Function 361
formed. This has an impact on the structural side of analyses of discourse: as
Bateson (1955) points out, we cannot always see an outline of a conversation
when we are in the middle of it as clearly as when it is finished.
The emergent structure of discourse makes it difficult to know from what
set of response options (what “system”) a response will be selected until that
choice has been made (Schiffrin 1987a: 23). Once a response is issued, however,
we know more about the structure being built, and more about the functional
repertoire from which a response has selected. In some ways, then, this is
another way that we can draw upon the strengths of one type of analysis to
compensate for the weaknesses of the other. Just as we may try to identify
units through their functions, we may also try to discover (and thus limit)
functions by seeing which function seems to be attended to and receive a
response.
.To summarize, the approaches to discourse Analysis described in this book
suggest that what is appropriate “is to examine structure in the light of func¬
tional requirement and function in the light of structural requirement” (Sadock
1984: 142). Thus, neither radical structural nor radical functional analyses are
appropriate. Rather, combining facets of both types of analyses may help to
balance the weaknesses of one mode of analysis with the strengths of another.

Note
1 Despite the difficulties of identifying structural units of discourse, there seems to be
a fundamental assumption in all analyses of language, and of social life, that people
categorize the world that surrounds and informs them. Such categorizations not only
enable people to inform others about that world: they also help people make sense
out of what would otherwise be a steady stream of stimuli from sources of informa¬
tion that are often simultaneously available. Put another way, we routinely and
automatically segment our environment and relate those segments to one another in
ways that capture both their similarities and their differences. There is no reason to
expect that we - in our capacity as speakers and hearers engaging in social interaction
- stop doing this when we produce and interpret utterances. And there is no reason
to expect discourse analysts to stop trying to categorize those utterances (parts of
utterances, or combinations of utterances) as units of some kind, to try to identify
the “stuff” of which they are made, and to analyze relationships among units. Like
other categories of experience, such units would seem to help people make sense out
of their interactive world and the utterances that are produced and interpreted within
that world. (Also like other categories, we might find that some utterances are
“better” examples of a category member than others.)
10 Text and Context

1 Introduction

The focus of this chapter - how different approaches to discourse view text
and context - provides a bridge between chapter 9 (on structure and func¬
tion) and chapter 11 (on communication). In chapter 9 we discussed how
discourse has been viewed as structure (a unit of language that is larger than
the sentence) and as the realization of functions (the use of language for social,
expressive, and referential purposes). We saw that all the approaches discussed
in this book incorporate both structure and function into their analyses: the
structures analyzed in discourse lead to the identification of functions, and
the functions analyzed in discourse are linguistically realized in ways that
create structures.
In chapter 11 we turn to linguistic assumptions about communication. Lin¬
guists often assume that messages are created through an interaction be¬
tween two different types of information. The first type of information is often
called “semantic”: a stable core of propositional meanings conveyed through
the language (i.e. the linguistic code) itself. The semantic information pres¬
ented in units larger than a clause is often thought of as textual information.
The second type of information is contextual information. Context is more
difficult to define than text. Contextual information is always information
that is identified in relation to something else that is the primary focus of
our attention. This means that it is impossible to talk about context in a
vacuum: context cannot exist unless we are thinking of “something else” (e.g.
an image, a smell, a sound, a word, an utterance, a sequence of utterances)
that is located relative to it. The identity of that “something else” (and what
kind of sense we are trying to make of it) influences our decisions about what
counts as context and about what “parts of” context we find important. Thus,
although it seems possible to find a single source of “text” (i.e. in the linguistic
system), the source of contextual information is necessarily more varied:
 Text and Context 363
context has multiple sources that can be quite different from one another
and can shift depending on our focus of attention. (See, for example, Becker
1984.)
The belief that messages are created through an interaction between text and
context implies a procedure by which to identify the communicative content
of an utterance: combine linguistic meaning with context to derive inferences
about messages. Identifying the communicative content of an utterance is
important for discourse analysis: it can help us understand coherence relations
across utterances. Actually using the procedure noted above to identify either
the communicative content of an utterance or relations across utterances,
however, is more complicated than it seems. First, it is often difficult to decide
whether it is linguistic or contextual information that is responsible for com¬
municative content. Such difficulties are illustrated by debates about the
meaning and function of the conjunction and, i.£ it is not clear how to account
for the added meanings (e.g. temporal, causal) that use of the word and seems
to convey (Posner 1980; Schiffrin 1986a). Second, it is not always clear how
text and context interact to allow inferences about communicative content.
For example, it is not clear whether the directive function of I need these
papers typed by 2 p.m. is inferred through recipients’ use of linguistic informa¬
tion (e.g. Searle 1975), contextual information about social role and setting
(e.g. Ervin-Tripp 1976), and/or contextually based inferences about communic¬
ative strategies (Brown and Levinson 1978). Third, since interpretive contexts
are continually being created through what is said, inferences about communic¬
ative context are dynamically situated; thus, whatever one infers about a
particular utterance can be modified (or even cancelled) through later utter¬
ances. Finally, it is difficult to even be sure about what should count as
communicative content. Levinson (1983: 131), for example, classifies different
types of communicative content that all contribute to meaning-nn (chapter 6),
i.e. what a speaker intends to convey by means of a hearer’s recognition of
that intention. But other scholars focus less exclusively upon intention as a
basis for communicative content (chapter 11).
Despite such difficulties, the discovery and use of some kind of procedure
for identifying the communicative content of utterances is an important goal
for discourse analysis. The current chapter focuses on the way different ap¬
proaches to discourse view the two types of information that contribute to
communicative content: text and context.
In order to compare different views of text and context, we need to define
both terms in a way general enough to apply to all the approaches. As noted
earlier, I will use the term “text” to differentiate linguistic material (e.g.
what is said, assuming a verbal channel) from the environment in which
“sayings” (or other linguistic productions) occur (context). In terms of utter¬
ances, then, “text” is the linguistic content: the stable semantic meanings
of words, expressions, and sentences, but not the inferences available to
hearers depending upon the contexts in which words, expressions, and sen¬
tences are used. Put another way, text is the propositional meanings that
364 Conclusion

are linguistically realized (e.g. that might constitute the “semantics” of

a grammar) in grammatically definable units such as clauses, and through
the relations that are conveyed between (and among) such units. Context is
harder to define in a way that covers all the approaches to discourse that
we have described: in fact, we will see that it is precisely their consider¬
ation of context that most clearly differentiates the approaches. We can pro¬
vide an initial definition of context, however, in terms of utterances: if we say
that text provides for the “what is said” part of utterances, then it is context
that combines with “what is said” to create an utterance. Context is thus a
world filled with people producing utterances: people who have social,
cultural, and personal identities, knowledge, beliefs, goals and wants, and
who interact with one another in various socially and culturally defined situ¬
ations.
All the approaches discussed in this book consider both text and context as
interdependent contributions to utterance meaning and to discourse coherence
across utterances. Yet they also make assumptions about text and context, and
about their relationship, that are not always clearly articulated and that differ
quite surprisingly from one another. At first glance, structurally motivated
approaches (conversation analysis, variation analysis; chapter 9) might seem
to be “text” based - to focus upon the propositional meanings that are lin¬
guistically realized through units such as clauses or sentences, and through the
relations conveyed or implied between such units. We saw in chapter 9,
however, that although variation analysis (chapter 8) does focus upon textual
structures based (partially) on relationships between clauses, conversation
analysis (chapter 7) does not consider the linguistic structures underlying
clauses, sentences, and texts at all. We might also assume that functionally
motivated approaches are more concerned with context: the non-linguistic
environment (be it cognitive, social, and/or cultural) in which texts are pro¬
duced and interpreted. However, there are also differences among these
approaches. Pragmatics (chapter 6) assumes a relatively clear distinction be¬
tween text and context as two autonomous sources of information.' But the
ethnography of communication (chapter 5) intentionally obviates the distinc¬
tion between what is text and what is not: recall, for example, the inclusion
of both verbal and nonverbal “instrumentalities” in the etic grid through
which the communicative components of different events can be comprised.
Thus, whether an approach is structurally or functionally motivated does
not necessarily predict either its theoretical or methodological emphasis on
text or context, or how it incorporates text and context into theory and
method.
We noted above that what most clearly differentiates approaches to dis¬
course is their treatment of context: scholarly attention to context can focus
upon knowledge, situation, and/or text itself (section 2). Approaches to dis¬
course also differ in their view of “text” in relation to “context” and in terms
of the unit of analysis (i.e. single or multiple utterances) upon which they
initially focused (section 3). Section 4 summarizes.
 Text and Context 365

2 What is context?

The approaches to discourse discussed in this book make different assump¬

tions about what aspects of context are relevant to the production and inter¬
pretation of utterances. Speech act theory (2.1.1) and pragmatics (2.1.2) view
context primarily as “knowledge”; although a key part of such knowledge is
“knowledge of situation,” “situation” is largely unanalyzed by these ap¬
proaches. Interactional sociolinguistics (2.2.1) and the ethnography of com¬
munication (2.2.2) also view context as “knowledge,” and they, too, include
“knowledge of situation”; these approaches, however, propose frameworks
and constructs through which to analyze “situation” as part of “knowledge.”
Speech act theory also depends partially upon fhe kinds of concrete situations
analyzed by interactional sociolinguistics and the ethnography of communica¬
tion (2.2.3). Variation analysis (2.3) views context as “situation,” without
explicitly incorporating it into “knowledge”; it also categorizes “text” as part
of context. Conversation analysis (2.4) focuses on how text is a means of
displaying “situation,” and on how text creates knowledge including, but not
limited to, knowledge of “situation.”

2.1 Context as knowledge

Speech act theory and pragmatics both view context in terms of knowledge:
what speakers and hearers can be assumed to know (e.g. about social institu¬
tions, about others’ wants and needs, about the nature of human rationality)
and how that knowledge guides the use of language and the interpretation of
utterances.

2.1.1 Speech act theory and context

People can use language to do things - to perform speech acts - because the
rules through which speech acts are realized are part of linguistic competence.
Although the rules specify both textual and contextual conditions underlying
language use, a central goal of speech act theory is to characterize our knowl¬
edge of those rules (rather than, for example, the range of forms and situations
in which the rules take effect). What this means is that the social circumstances
that might help to define a particular act are incorporated into the description
of “what we know” when we speak. This abstract knowledge of text and context
is what allows us to identify different types of speech acts at both a general
level (e.g. directives, commissives) and a specific level (e.g. questions, offers).
An emphasis on knowledge pervades the application of speech act theory to
discourse. Although this application requires the analysis of actual utterances,
366 Conclusion

the analysis of how those utterances perform speech acts assumes that lin¬
guistic competence includes speaker and hearer mastery of constitutive rules.
Such rules are part of mutual knowledge and have an important role in
governing language use: it is mutual knowledge of these rules that allows
hearers to discover a speaker’s intended speech act, and mutual knowledge of
abstract rules that makes possible a sense of coherence across utterances.
Note that not all approaches to discourse rely so heavily upon mutual
knowledge as a path to, or locus of, coherence (see chapter 11). The ethno¬
graphy of communication, interactional sociolinguistics, and conversation ana¬
lysis, for example, avoid the assumption of mutual knowledge by relying upon
the actual utterances produced in response to a given utterance as a way to
discover the hearer’s interpretation of a speaker’s intention - an interpretation
that may or may not coincide with the speaker’s actual intention. Thus, in
contrast to speech act theory (which, as noted above, locates coherence in the
linguistic rules that are part of speaker/hearer mutual knowledge), the other
approaches just noted search for coherence in a behavioral context: they
examine a speaker’s own utterance and the utterance issued by someone else
in response to that utterance. In brief, very different kinds of contextual as¬
sumptions can be said to provide a route to coherence.
I have suggested thus far that context for speech act theorists is a specific
kind of background knowledge called “constitutive rules,” i.e. knowledge
about what conditions need to hold if an utterance is to count as a particular
speech act. Although people draw upon constitutive rules much as they might
draw upon other contextual information that is used to make inferences about
meaning (e.g. Grice’s cooperative principle), constitutive rules also have an¬
other role. Recall that constitutive rules create speech acts (in contrast with
regulative rules, which regulate pre-existing behavior). This means that the
rules appropriate to a given utterance have to be cognitively “activated” in
order for a given utterance to count as a realization of a particular speech act.
Put another way, constitutive rules provide the organized framework of knowl¬
edge in which to discover a solution to the problem of speech act identity.
Thus, constitutive rules can be seen as a kind of cognitive context in relation
to which what is said may be interpreted and found to be meaningful.
We noted above that our knowledge of speech act rules - the cognitive
contexts that underlie how we use language to do things - includes both
linguistic and social information about what constitutes a particular speech
act. This means that social contexts, such as social institutions, others’ wants
and needs, and so on, do contribute to the definitional framework for speech
act interpretation. Our discussion of Austin (chapter 3, section 2.1), for example,
mentioned contextual conditions (“circumstances”) and textual conditions
(e.g. the availability of explicit performative formulae) that allow an utterance
to have a certain illocutionary force. Like Austin, Searle’s rules and conditions
for speech acts draw upon both context and text: they also elevate intentions
and other psychological states as conditions enabling a speech act, by assigning
to them their own type of rule. (Also like Austin, Searle classifies conditions
 Text and Context 367
and rules according to their necessity for the act.) But in contrast to Austin,
Searle classifies different kinds of conditions (and rules) according to what
aspect of text and context is focused upon in the condition or rule. The
different conditions also overlap (partially) with different components (i.e.
locutionary, illocutionary) of a total speech act.
In sum, the central contribution of context to utterance interpretation in
speech act applications to discourse is its contribution to the knowledge (i.e.
the conditions and rules) underlying the successful performance and interpreta¬
tion of speech act types. The abstract knowledge of text and context just noted
is what allows us to identify different types of speech acts. Abstract knowledge
of context, however, is only one way that context figures in speech act
applications to discourse. The actual social circumstances during which an
utterance is produced also provide a framework in which an utterance is
successfully realized as a token of a particular type &f act. Thus, context is
not just a matter of knowledge: context is also a set of social circumstances
in which utterances can be produced and interpreted as realizations of their
underlying constitutive rules. I discuss this in section 2.2.

2.1.2 Gricean pragmatics and context

Gricean pragmatics also views context as a cognitive contribution to utterance

interpretation. The context that is proposed, however, is quite different from
that assumed by speech act theory. The context proposed by Gricean prag¬
matics is a general principle that participants assume one another to believe
and observe: “Make your conversational contribution such as is required, at
the stage at which it occurs, by the accepted purpose or direction of the talk
exchange in which you are engaged.” It is this principle that is called the
cooperative principle (CP).
The CP allows the inference of speaker meaning - meaning that can go well
beyond (or even substitute for) semantic meaning. Such inferences (called
“implicatures”) are calculated with the use of the following information:

(1) the conventional meaning of the words used, together with the iden¬
tity of any references that may be involved; (2) the CP and its maxims;
(3) the context, linguistic or otherwise, of the utterance; (4) other items
of background knowledge; and (5) the fact (or supposed fact) that all
relevant items falling under the previous headings are available to both
participants and both participants know or assume this to be the case.
(Grice 1975: 50)

Item (1) is what someone “says” (the semantic meanings of propositions).

Although Grice mentions context only briefly (as item (3) “the context, lin¬
guistic or otherwise, of the utterance”), context is actually critical to implica-
ture calculation. We can extrapolate from Grice’s division of context into
368 Conclusion

“linguistic” and “otherwise” (in item 3), and his notion of “linguistic” as
what someone “says,” to define two types of context. First is a linguistic
context: the text within which the utterance is linguistically encoded (see
section 3.1). Second is the “otherwise” context: the situation during which
the utterance is produced. These contexts bear on inferencing only because
they contribute information that can be used as mutually known background
knowledge. That contexts are akin to mutual knowledge is suggested by
Grice’s item (4), noting other items of background knowledge (presumably,
information about the world not provided from text or situation), and by
Grice’s item (5), the availability of mutual knowledge.
Calculating implicatures also requires the cooperative principle (item 2).
Despite the central role played by the CP and its maxims in inferencing,
however, the CP is actually quite similar to the other contexts listed: it, too,
is background knowledge that participants assume one another to share. This
similarity allows us to suggest that what allows hearers to calculate implicatures
and infer speaker meaning is conventional meaning (item 1) and background
knowledge from a number of different sources, i.e. assumptions about human
nature (item 2), text (item 3), situation (item 3), and the world (item 4).
Despite the importance and diversity of the “otherwise” and “background
knowledge” contexts, Gricean pragmatics offers no suggestions about how to
analyze those contexts. Some of the issues raised in our sample analysis in
chapter 6 illustrate the kinds of contextual analyses that might be needed, at
least for a Gricean analysis of referring sequences. Background knowledge is
pertinent to the identification of referents (e.g. the use of metonymic refer¬
ences) and the forms used for first-mentions (e.g. initial definites) and next-
mentions. We might analyze this background knowledge as a schema (frame,
or script) that provides structured expectations about what kinds of people
and things typically appear in a given setting, and what kinds of actions
typically occur there. The analysis of text itself is also important (3.1). Text
can reinforce expectations about the relevance of schematic knowledge for
interpretation of a particular referent. Analyses of text can also help to explain
shifts from explicit first-mentions to inexplicit next-mentions and the depend¬
ence of inexplicit forms on different descriptive frameworks (e.g. story world
versus conversational world) that are textually created. Finally, the circum¬
stances of talk itself can provide a framework of expectations relative to which
cooperation is geared: the actual situations in which language is used include
speakers and hearers, whose needs, goals, and wants are tailored to a particu¬
lar socially and culturally defined communicative situation. Thus, interac¬
tional and ethnographic analyses of different communicative situations are
also relevant: they might suggest, for example, that how much information we
provide is tailored to our socially and culturally based conceptions of what
our interlocutors need for the purposes of a particular social exchange.
In sum, a Gricean pragmatic approach to discourse analyzes the way speaker
meaning is dependent upon a cognitive context of shared beliefs and assump¬
tions. Although this view of context is similar in some ways to that of speech
 Text and Context 369
act theory, there are also important differences. Speech act theory compart¬
mentalizes context into different aspects of social circumstances, and associ¬
ates different clusters of conditions (both textual and circumstantial) with
specifically labelled functional units (i.e. speech act types). Gricean prag¬
matics, on the other hand, compartmentalizes context into different sources
of background knowledge (assumptions about human nature, text (see section
3.1), situation, the world) without grouping specific types of knowledge into
labelled units (see Sperber and Wilson 1986: 243-54).

2.2 Context as situation and knowledge

Both approaches to discourse discussed thus far include knowledge of the

social circumstances framing an utterance as pftrt of &e cognitive context in
which utterances are produced and interpreted. Yet neither approach guides
the analysis of the actual circumstances in which utterance tokens (not abstract
utterance types, i.e. speech act types (2.1.1)) are embedded. This section
considers the way two other approaches to discourse - interactional sociolin¬
guistics (2.2.1) and the ethnography of communication (2.2.2) - incorporate
situational analyses into their view of context. In so doing, however, they also
alter the cognitive contexts considered by speech act theory (from linguistic
to either communicative competence (Hymes 1974a) or situated competence
(Gumperz 1985), and pragmatics (from inference to situated inference). This
section also considers the way speech act applications to discourse need to
incorporate analyses of context as situation as well as knowledge (2.2.3).

2.2.1 Interactional sociolinguistics and context

Context as “situation” is critical to interactional sociolinguistics. In fact, one

of the main features of this approach is that it provides a richly textured view
of social interaction and social situations, including the way participation
frameworks and presuppositions arise from situated interaction. Language and
context constitute one another: language contextualizes and is contextualized,
such that language does not just function “in” situated interactions, language
also forms and provides for these interactions (Duranti and Goodwin 1991).
The contexts upon which interactional sociolinguistics focus are both in¬
ternal and external to the individual. Nowhere is this better illustrated than
in a well-known formulation from one of the sociological “fathers” of inter¬
actional sociolinguistics: “if men define situations as real, they are real in their
consequences” (Cooley 1902). Although Cooley’s notion of the definition of
a situation focuses on inner subjectivity (that need not be shared by others,
however; chapter 11), it also focuses on one’s perception of social circum¬
stance and on the real social consequences of that perception. This means that
a definition of a situation can probably not be formed in complete isolation
370 Conclusion

from the social “here and now,” and, furthermore, that its material con¬
sequences in that “here and now” are very concrete. (Note that these links to
the external world allow definitions of the situation to escape the problem of
infinite regress faced by Gricean pragmatics’ dependence on mutual knowl¬
edge: Smith 1982). Thus, even the cognitive contexts that are considered by
interactional sociolinguists (e.g. frames: Goffman 1974; Tannen 1979) are
socially grounded: knowledge is knowledge of social circumstances or expec¬
tations about social conduct.
Interactional sociolinguistics also provides a way to analyze social context
and to incorporate those contexts into the procedures through which we infer
meaning. Goffman’s sociological research focused attention on the interac¬
tional order that underlies social occasions, situations, and encounters. Knowl¬
edge of this general organization leads to appreciation of the socially
constituted “moves” that help create a sense of “reality” in a particular
interaction (i.e. a definition of the situation). This sense of reality is intimately
related to Gumperz’s ideas about contextualization cues, contextual presup¬
positions, and situated inferences.
Recall that situated inferences are interpretations at two levels of inferen-
cing: what is going on in the “here and now” (i.e. are we teasing? arguing?)
and what typically happens during this kind of situation (i.e. what usually
characterizes teasing? arguing?). Recall, also, that contextualization cues are
aspects of language and behavior that relate what is said to contextual knowl¬
edge. These cues signal contextual presuppositions: assumed background
knowledge about the different situations in which we act. The contextual
presuppositions signalled by contextualization cues inform us about our cur¬
rent activity by tying it to a more general framework of expectations about
situated activity-types, of which our current activity is but one specific
example. Contextual presuppositions are strikingly similar to definitions of a
situation: what we know about, and what we expect to find, in a particular
activity (or situation) provides the framework through which we characterize
and define that activity (or situation). Thus, we draw situated inferences about
speakers’ messages through the use of contextualization cues that signal our
contextual presuppositions, i.e. our definition of the situation.
In sum, interactional sociolinguistics provides a way to identify different
kinds and levels of contexts, to conceptualize the organizational and interpret¬
ive role of contexts, and to describe how utterances and parts of utterances
have a contextualizing role that allow us to draw situated inferences.

2.2.2 Ethnography of communication and context

The ethnography of communication combines other approaches within a

larger framework of inquiry into cultural knowledge and the social and lin¬
guistic practices for which it provides. It views context as both cognitive (what
we know, embedded in our communicative competence) and social (the social
 Text and Context 371
and cultural components that combine to define communicative events). The
ethnography of communication also provides a way to discover the organiza¬
tion of context: the SPEAKING grid segments social context into different
components that not only define a particular communicative situation (event,
and act) as a closed and bounded unit (Hymes 1972b: 56), but also provide
a way to systematically differentiate from one another those situations (events
and acts) that comprise the communicative repertoire of a given community.
Ethnographically motivated analyses of context provide an important sup¬
plement to the interactional sociolinguistic focus on contextual presupposi¬
tions and definition of the situation. Above we noted that contextualization
cues are clues by which we retrieve knowledge (i.e. contextual presuppositions)
about what is expected to occur during a particular activity (or situation). We
use this knowledge to draw situated inferences about what others mean. We
noted that these inferences are “situated” in twt) ways? they are located in the
“here and now” of an ongoing interaction; they depend upon our general
knowledge of social situations. An important way to discover the knowledge
that allows situated inferences is an ethnographic analysis of communicative
situations, events, and acts. Suppose, for example, that we wanted to say that
a particular phrase or intonation served as a contextualization cue by which
to signal participants’ definition of a situation as a reference interview in a
library (chapter 5): we would need to know what typically occurs during such
an interview in order to describe the contextual presuppositions evoked by the
contextualization cue. It is in this way that the ethnography of communication
can provide us with the contextual presuppositions upon which interactional
sociolinguistics relies as a basis for situated inferences about speakers’ meanings.
In sum, the ethnography of communication makes the role of context central
to the analysis of communication - to the meaning of specific speech acts, to
the intentions underlying a specific utterance, to the relationships assumed to
hold between utterances, to the organization of acts within events, and events
within situations. What we say and do has meaning only within a framework
of cultural knowledge - not linguistic, but communicative, competence. The ways
that we organize and conduct our lives through language are thus ways of being
and doing that are deeply embedded within the particular contexts - cultural
frameworks - by which we make sense out of experience. The detailed analyses
of context that this approach provides sometimes make it difficult to propose
generalizations about language functions apart from the contexts through
which they arise. However, it is just such detailed knowledge about situations,
events, and acts that provides the contextual presuppositions needed by interac¬
tional sociolinguists, and that forms a key part of our communicative competence.

2.2.3 Speech act theory: context as situation

In section 2.1 we noted that context plays a dual role for speech act theorists:
it is the abstract knowledge underlying speech act types; it is a set of social
372 Conclusion

circumstances in which utterances can be produced and interpreted as realiza¬

tions of their underlying constitutive rules. The relationship between the
cognitive contexts underlying the general production and interpretation of
speech act types, and the actual circumstances during which a specific utter¬
ance serves as a token of a particular speech act, is complex. To make this
point more concretely, let us consider the following hypothetical examples. In
(la), a text produced by one speaker seems to state all the underlying condi¬
tions of a particular speech act.

(la) doris: I just noticed that we’re out of milk. We’ll need some tomorrow
morning. I won’t have a chance to buy any today. You can pass
the grocery store on your way home from work later. Even
though I know that you didn’t plan to buy milk there, I want
you to buy some milk there. Will you do that?
andy: Okay.

Requests like (la) rarely occur in conversational interaction simply because at

least some of the information needed to understand such a text as a request
can be assumed to be either already known or easily inferrable by a listener.
That is, Doris may very well be able to assume that Andy knows the conditions
and the constitutive rules for requests without stating them explicitly.
Contrast the hypothetical (lb):

(lb) doris: Uhoh. [looking in the refrigerator]

andy: Okay.

Although what Doris says in (la) and in (lb) is linguistically different, her
utterances are functionally equivalent in some ways: they both issue a request.
Thus, the same abstract conditions and the same constitutive rules underlie
these two different realizations of the “same” speech act type. More specific¬
ally, the same abstract constitutive rules hold for Doris’s act in both (la) and
(lb): S predicates a future act A of H (the propositional content rule); S
assumes that H is able to do A and S believes H is able to do A; it is not
obvious to both S and H that H will do A in the normal course of events of
his own accord (preparatory rules); S wants H to do A (the sincerity rule); S’s
utterances count as an attempt to get H to do A (the essential rule). However,
the way the utterances in (la) and (lb) realize this act is dramatically different.
What Doris says in (la) requires little attention to the physical context in
which it occurs, (lb) is the opposite: interpreting uhoh as a request relies
almost entirely upon how that remark is situated in its physical context.
Regardless of the way the “texts” in (la) and (lb) are bound to their actual
situational contexts, the request receives the same response: Andy agrees to
comply with the request to buy milk.
The contrast between (la) and (lb) bears in several ways on the role of
context in speech act applications to discourse. First, (la) and (lb) show that
 Text and Context 373
the relationship between text and context is basic not only to an understanding
of messages, but also to an analysis of relationships across units of discourse.
Second, (la) and (lb) suggest that the text-context relationship is not fixed:
although some utterances in a text can be analyzed as relatively autonomous
from their physical context (as in la), others must be analyzed as totally
dependent on their context (as in lb). Third, (la) and (lb) illustrate a kind of
trade off between text and context: in fact, as both Austin and Searle note, it
is possible to perform an act “without invoking an explicit illocutionary
force-indicating device where the context and the utterance make it clear that
the essential condition is satisfied” (Searle 1969: 68). These points all bear on
our main concern here: the role of context as situation in speech act applica¬
tions to discourse. Although social circumstances always have a role in the
abstract knowledge underlying speech act types, their role in the interpretation
of actual speech act tokens may be diminished*!as in*la) or enhanced (as in
lb) depending upon the actual situation in which an utterance is issued.
Similarly, uncovering the basis for sequential coherence across utterances may
require supplementing abstract speech act knowledge with understandings
about social proprieties and interactive norms (chapter 4, section 3.2.2). Thus,
although cognitive context (mutual knowledge of speech act rules) is critical
to the production and interpretation of speech acts in general, the circumstan¬
ces during which an utterance is produced (and the social meaning of those
circumstances) cannot have a theoretically fixed (or a priori) relationship with
the utterances through which speech acts are realized.
I have suggested in this section that abstract knowledge of context is only
one way that context enters speech act applications to discourse: social cir¬
cumstances also provide a framework in which utterances are successfully
realized as particular acts. Context has this extra “situated” role simply
because the linguistic form and content of an utterance rarely provide a guide
sufficient to the task of identifying its speech act. Situated context gains special
importance in the analysis of those acts for which no performative verb exists
and those acts that are often labelled “indirect.” Thus, although the linguistic
form and content of an utterance do give clues to its speech act function, such
clues must work with contextually provided information. What is said works
with information available from actual circumstances to allow utterances to
do things. More specifically, social situations and their meanings provide the
clues that allow hearers to infer the illocutionary force of specific utterances:
they help hearers find the “right” constitutive rules by which to interpret the
function of, and formulate a response to, an utterance.

2.3 Context as situation and text: variation analysis

We have seen that several approaches to discourse consider context in two

different, but interrelated, ways: context as knowledge (incorporated in lin¬
guistic or communicative competence) and context as situation (including
374 Conclusion

knowledge of both the “here and now” situation, and knowledge of general
“types” of situations).
Variation analysis also considers context as situation, but in a way somewhat
different from the approaches discussed thus far. Like the ethnography of com¬
munication, variation analysis divides situation into separate components.
Such components are a subset of those in Hymes’s SPEAKING grid: the social
situation (setting and scene), the participants (their social identities, such as
gender, age, ethnicity), and the key (formal or informal, careful or casual style
of speaking). Unlike ethnography, however, variationists view situational fac¬
tors like setting, identity, and key as discrete and mutually exclusive factors
that can be coded, counted, and compared across different circumstances (cf.
chapter 5, section 3.4).
The variationist view of context differs even more dramatically from the
interactional sociolinguistic view of context. Let us compare, for example, the
views of social identity from both perspectives. Variationists typically treat
identities as social and cultural categories that we “step into”: our identities
are not easily open to our own control or to alteration by another (cf. Eckert
and McConnell-Ginet 1992). This view of identity underlies the variationist
practice of coding identity as a categorical variable: a speaker is coded as a
white, middle class, middle-aged male from Philadelphia, for example, and he
is assumed to maintain that very same identity regardless of the activity or
interaction in which he is engaged. A very different view of social identity is
provided by interactional sociolinguistics. This perspective suggests that iden¬
tity is locally situated: rather than an identity being maintained throughout an
extended period of time (be it a speech activity, a situation, a span of life, or
even a lifetime), different facets of identity may be highlighted or submerged
during different periods of time. This more dynamic view suggests that identity
is open to intentional manipulation by self, and to interpersonal negotiation
between self and other. Thus, identity is less amenable to the “once and for
all” coding used by variationists.
A similar difference appears in the interactional and variationist views of
situation itself. In the interactional approach, participants’ sense of “what is
happening” is not supplied when an activity begins, but is molded and re¬
molded during different phases of the activity, such that definitions of the
situation emerge at the same time as the situation itself. Thus, a situation is
open to definition and redefinition. Variationists do not consider a situation
in toto to change once a definition of that situation is established. How¬
ever, variationists do conceive of different parts of a situation as open to
change. This is revealed somewhat indirectly through the methods by which
variationists code the variable “speaking style.” Labov (1972c) differentiates
styles (e.g. careful, casual) according to the amount of attention paid to
speech. Since attention to speech is difficult to measure, variationists rely upon
indirect clues of style shifts. Whereas some of these clues concern the way a
speaker produces an utterance (what Labov speaks of as “channel cues”),
other clues are situational. A shift in addressee, for example, can accompany
 Text and Context 375
a shift in style: if an interviewee speaks to someone other than the interviewer,
that utterance is coded as “casual.” Thus, although the different components
of a situation can shift, the situation itself is viewed as relatively static.
Another difference between the variationist, and ethnographic and interac¬
tional, approaches to context concerns the necessity of bringing situation into
an analysis of text. Our sample analysis in chapter 8, for example, considered
the textual structure of lists without saying anything at all about the purposes
of the lists, the situations in which they were presented, the speakers, and so
on. This is in stark contrast to the ethnographic and interactional analyses in
which the form and function of even a single utterance could not be considered
apart from its situation (e.g. as part of a speech event, a participant structure,
or in relation to contextual presuppositions). That situation can be an optional
part of a variationist analysis reflects the variationist view of discourse as an
autonomous unit whose internal parts stand in'systerrfktic relationships with
one another. When compared to most of the other approaches to discourse
being discussed in this book, then, variationists’ assumptions about text - and
language - are closer to a key assumption of formal grammarians: language
has a modular structure with interacting, but autonomous, components. (In
many other respects, of course, variation analysis is firmly within a function¬
alist paradigm; see chapter 2.) This assumption is responsible for the variation¬
ist practice of separating language structure per se (not only at textual levels,
but at syntactic and phonological levels) from social and communicative
function (e.g. Labov 1987). Function and context are relevant only when
structural explanations fail to account for the range and distribution of lin¬
guistic variants.
In addition to viewing components of “situation” as context, variationists
also view “text” as context. Let us briefly return to the analysis from chapter
8 to illustrate. Recall that we were interested in how a referring term (call it
R-3) in a particular location in a list (call it clause-3, C-3) was constrained by
its location in the structure of the list. One part of our procedure was to
characterize the structure of the list, and the location of both R-3 and C-3, in
that structure. Note, however, that what this amounted to was treating the
other clauses in the list (C-l, C-2), and the other referring terms (R-l, R-2)
as part of the environment in which R-3 and C-3 were located. Put another
way, we treated the text in which an option (R-3), and its containing clause
(C-3), were located as a context - as information to be considered only in
relation to something else that was the primary focus of our attention. Thus,
text itself became a context (cf. Brown and Yule’s 1983: 46-50, term “co¬
text”) that could be compartmentalized, coded, and counted.
Note, then, that the same kind of approach is applied by variationists to
both text and situation: both are divided into discrete and mutually exclusive
factors that can be coded, counted, and compared. The compartmentalization
of text is easily understandable given our earlier observation (chapter 9) that
variation analysis is a structurally motivated approach. Not only do variation¬
ists define discourse as a unit of language larger than a single sentence (chapter
376 Conclusion

2, section 3), but variationists analyze how that unit is comprised of smaller
units (whether those units are conceived of as clauses, sentences, turns, utter¬
ances, and so on) that are systematically related to one another. Given the
focus upon text as comprised of smaller units, and the analysis of the relation¬
ship of those units with one another, the segmentation and categorization of
text into units that have systematic effects upon each other is not surprising.
It is this basic structuralist methodology that also seems to be applied to
variationist analyses of situation.
In sum, variationists consider both situation and text to be context. They
segment and categorize both aspects of context, viewing different facets of its
form and content as discrete and mutually exclusive factors that can be ex¬
tracted from their surroundings (e.g. their broader cultural frameworks) to be
coded, counted, and compared across different circumstances. The different
components into which variationists divide and classify situation and text are
defined as “constraints” on the variant being focused upon: they are contextual
influences on linguistic realizations in a particular sequentially defined slot.

2.4 Context as knowledge, situation, and text:

conversation analysis

At first glance, conversation analysis seems to combine the way other ap¬
proaches view context: context is knowledge (speech act theory, pragmatics),
situation (interactional sociolinguistics, ethnography of communication), and
text (variation analysis). But a very particular view of knowledge - and its
relationship to action and language - differentiates the CA view of context
from the others.
As we discussed in chapter 7, CA is an approach to discourse that inherits
from ethnomethodology the goal of accounting for the common-sense knowl¬
edge that members have for constructing talk. The knowledge on which
members rely is treated by them as part of the same setting that it also makes
orderable (Garfinkel 1974: 17). Thus, knowledge is, in a sense, context (cf.
the speech act (2.1.1) and pragmatic (2.1.2) views). But knowledge cannot be
considered apart from either the situations in which, or the actions for which,
it is used. Knowledge and situation are mutually constitutive: participants’
understandings of their circumstances, and participants’ role in constructing
those circumstances through actions, are all intertwined. Activities serve to
publicly display an emergent sense of order: activities provide a practical basis,
and a sense of intersubjectivity, through which other activities (and the situ¬
ation that they help to create) can be sustained. Social action thus not only
displays knowledge (cf. the speech act view), but is critical to the creation of
knowledge: one’s own actions produce and reproduce the knowledge through
which individual conduct and social circumstance are intelligible.
Not only does this relationship differentiate the CA view of knowledge from
the speech act and Gricean pragmatics views (in which knowledge seems to
 Text and Context 377
exist even when it is not knowledge “in use”), it also has an important bearing
on the study of language and on the relationship between text and context.
As we noted in chapter 7, members produce “typifications”: categories that
are continuously adjusted according to whether the anticipation of an actor
(an anticipation on which the actor’s own conduct was based) is confirmed by
another’s actions. Language (and action through language) is a situated product
of rules and systems that provide actors an ability to “make continuous sense
of their environments of action” (Heritage 1984a: 292). Language serves as a
medium through which common-sense categories (typifications) are constituted.
Language also enters into the mutually constitutive relationship between ac¬
tion and knowledge: speakers produce utterances assuming that hearers can
make sense out of them by the same kind of practical reasoning and methodic
contextualizing operations that they apply to social conduct in general.
.1 have been suggesting thus far that the CA view of Context is built upon the
ethnomethodological view of language, action, knowledge, and situation. What
CA analyses show is that the sequential progression of interaction - the
positioning of utterances - is critical to members’ display of their under¬
standings (including their knowledge about situation) through language. Each
utterance in a sequence is shaped by a prior context and provides a context
for a next utterance: “the significance of any speaker’s communicative action
is doubly contextual in being both context-shaped and context-renewing”
(Heritage 1984a: 242). From a speaker’s point of view, next-position thus
offers a location in which to find the recipient’s analysis of the utterance - to
see whether an anticipated response is confirmed. From a recipient’s point of
view, next-position offers an opportunity to reveal aspects of the under¬
standing of prior talk to which own talk will be addressed. It is not surprising
that two constructs central to CA reflect the doubly contextual role of utter¬
ances. Adjacency pairs are organized patterns of stable, recurrent actions that
provide a normative framework in which actions are accountably implemented
(Heritage 1984: 249). The exchange of turns allowed by turn-taking rules
provides for “next-position,” and thus for the emergence of sequentially or¬
dered displays of understandings to which actors are mutually accountable.
The terminology used above - an utterance is context-shaped and context-
renewing - clearly shows that CA views “text” as “context.” This terminology
does not mean, however, that the CA view of text-as-context is the same as
the variationist view (2.3). The CA view of text-as-context as both retrospect¬
ive and prospective reflects the ethnomethodological view that meanings (and
knowledge) are continually adjusted and sequentially emergent. In sharp con¬
trast to variation analysis, then, utterances are not positioned relative to one
another because of their underlying linguistic properties (or speech actions) or
their abstract textual relations. Rather, utterances are positioned because they
provide frameworks of understanding and meaning for both past and upcom¬
ing utterances.
We noted above that interactional sociolinguistic and ethnographic ap¬
proaches to discourse do not readily separate what is said from the situation
378 Conclusion

in which it is said; the variationist view, on the other hand, resorts to situ-
ation-as-context when structural (text-based) explanations cannot fully ac¬
count for the range and distribution of linguistic variants. The CA treatment
of situation-as-context bears a surface similarity to the variationist view -
since situation-as-context is not assumed to be relevant - but again, it differs
in a critical way. Although CA assumes that utterances always have contextual
relevance for one another, not all aspects of situation-as-context are assumed
to have so constant a relevance. Although CA transcripts (and analyses) of
talk pay a great deal of attention to the productive details of utterances (e.g.
pauses, inbreaths, overlaps) they pay little attention to components of situation,
e.g. participants, setting. By intentionally ignoring what are often assumed to
be static features of a social world (e.g. the occupation of a participant), CA
reflects still again the ethnomethodological avoidance of premature generaliza¬
tions and idealizations. Categories of social life and conduct cannot be assumed
apart from the activities in which (and through which) they are realized: like
knowledge and language, the definition of the situation is subject to locally
emergent interpretive activity. The relevance of a social identity, for example,
can be no more presumed to hold across different times and places than can
the relevance of a one second pause. Thus, although CA is an approach to
discourse that emphasizes context, context-as-knowledge and context-as-situ-
ation are grounded in - and can only be discovered by - context-as-text.

2.5 Summary: different approaches to context

We have seen in this section that different approaches to discourse make

different assumptions about what aspects of context are relevant to the pro¬
duction and interpretation of utterances. Whereas speech act theory and prag¬
matics view context primarily as “knowledge,” interactional sociolinguistics
and the ethnography of communication view context as “knowledge” and as
“situation.” Variation analysis views context as “situation” and “text.” Con¬
versation analysis focuses on the reflexive relationship among “knowledge,”
“situation,” and “text.” The next section briefly considers the way different
approaches view “text,” and how their analyses of text are related not only
to their view of context, but also to the unit of analysis (i.e. single or multiple
utterances) upon which they initially focused.

3 What is text?

Earlier in this chapter we defined “text” as the linguistic content of utterances:

the stable semantic meanings of words, expressions, and sentences, but not
the inferences available to hearers depending upon the contexts in which
 Text and Context 379
words, expressions, and sentences are used. Text provides for the “what is
said” part of utterances; context combines with “what is said” to create an
utterance. Although all the approaches to discourse that we discussed are
concerned with language and with the “utterance,” not all the approaches are
explicitly concerned with text and “utterances.” Speech act theory and Gri¬
cean pragmatics, for example, are applicable to discourse, but they were not
initially concerned with relationships across utterances (3.1). Virtually all the
other approaches, on the other hand, were initially interested in utterances,
whether that interest stemmed from the contextual or textual contribution to
coherence across utterances (3.2).

3.1 Text and context for “utterance” based approaches

Both speech act theory and Gricean pragmatics focus on single utterances.
Given this focus, it is not surprising that these approaches offer no way to
analyze text either as part of context (cf. sections 2.3, 2.4) or as its own unit
of analysis: neither approach considers how to analyze multiple utterances,
the relationship between utterances, or how adjacent utterances contribute to
the communicative content of one another. I suggest here, however, that both
approaches can extend their view of how language and context are related
within one utterance to an analysis of multiple utterances.
Let us begin by recalling that both Austin and Searle differentiate the pro-
positional meaning of an utterance from its illocutionary force: a locutionary
act is the utterance of sounds and words with meaning (sense and reference).
Searle (1969: 22-33) goes on to argue that the propositional content of an
utterance (a part of the locutionary act) can remain constant despite variation
in the illocutionary act. Propositional content can have a role in speech act
interpretation: what is said and how it is said plays a key role in making
particular background knowledge relevant to the interpretation of illocution¬
ary force. Even in the most direct speech acts, for example in the performative
I promise, use of the verb “promise” is the cue that accesses the relevant
knowledge about the conditions underlying the speech act “promise.” In
addition to performative verbs, sentences have illocutionary force indicators
as varied as verbs, syntactic form, and intonation. Recall, for example, our
discussion of intonation and the semantics of the verb “want” in the identifica¬
tion of Henry’s Y’want a piece of candy? (chapter 3) as a question. Our
discussion of Y’want a piece of candy?, and a librarian’s May I help you?
(chapter 5), showed that analyses of indirect speech acts also consider “what
is done” (despite the fact that syntactic form itself may not be a direct reflection
of the multiple functions of an utterance). Thus, linguistic form and meaning
can provide clues to the rules (and the knowledge that they represent) that are
responsible for illocutionary force.
One way to extend speech act analyses from “utterance” to “utterances” is
to extend the distinction between propositional content and illocutionary force
380 Conclusion

from sentence to text. For example, we might analyze how the linguistic form
and meaning of a story contributes to its communicative function: this was an
essential part of our analysis of how Irene’s “diet” story served as an account
(chapter 3). However, we might also need to consider how each sentence
within a text has its own illocutionary force, and how (or if) those embedded
speech acts contribute to a text-level illocutionary force. We might also need
to add an analysis of propositional meanings across utterances (e.g. the rhet¬
orical predicates identified by Thompson and Mann 1987) or the macro-level
propositions that texts produce (e.g. van Dijk 1977).
Another possibility is to find that not all sentences in a text perform their
own autonomous speech acts. As we noted in chapter 3, text can enter the
constitutive rules of a speech act somewhat surreptitiously through the felicity
conditions for a specific act: an act may require a response to be felicitous
(e.g. S’s bet requires H’s uptake; Austin 1962: 37) or require prior utterances
to be felicitous (e.g. a conclusion has to follow prior evidence; Searle 1979).
Similarly, one utterance can provide information about the speech act being
performed by another utterance. Conversation analysts, for example, have
proposed that a speaker may anticipate the speech act to be performed in an
upcoming utterance through a prior utterance that engages an interlocutor in
a “pre-sequence,” e.g. a speaker may issue a pre-request, pre-announcement,
and so on (Levinson 1983: 345-64; Schegloff 1980, 1988a). For example, I
might not ask you to take me to the train station (a request) unless I have first
checked to see whether you are free at the time I need to go (a preparatory
condition, i.e. whether you have the ability to perform the act). Thus, what is
said by one utterance (e.g. “What time do you have to be at work tomorrow?”)
can prefigure the illocutionary force of a later utterance (e.g. “Can you take
me to Union Station in time for a 9:05 train?”). However, it is not just the
propositional (cf. locutionary) meaning of a prior utterance that helps perform
an upcoming act: it is how that prior utterance represents the felicity condi¬
tions of the upcoming act. Thus, multiple utterances can create the very same
context-as-knowledge necessary for interpretation of an act as a single utter¬
ance. Put another way, the representation of a knowledge base through which
to interpret the illocutionary force of a speech act may be not only a sentential
task, but a textual task.
Gricean pragmatics also differentiates propositional meaning (what speakers
“say”) from communicative meaning (i.e. meaning-nn, speaker intention).
Recall that propositional meaning is one type of information that we use when
we calculate implicatures: implicatures are allowed, in part, by the conven¬
tional meaning of words (2.1.2). But implicatures also rely upon the context,
“linguistic or otherwise” (Grice 1975: 50), of the utterance.
Our sample analysis in chapter 6, for example, showed that “linguistic”
context can provide information appropriate in quantity to the use of a refer¬
ring term in a later clause. Put another way, one could implicate the referent
evoked by a referring term by integrating the linguistic content of prior text
into assumed knowledge. Likewise, “linguistic context” can also provide in-
 Text and Context 381
formation to which a later referring term can be relevant: again, one could
infer the referent evoked by finding the assumed knowledge to which it was
relevant. In addition to showing how “linguistic context” (text) could figure
in implicatures, our sample analysis also suggested that the situation (the
“otherwise context”) could figure in implicature calculation. Consistent with
the idea that quantity of information is suited to the current purposes of an
exchange, we suggested that ambiguous expressions (e.g. they) might be per¬
fectly well suited for the telling of a story (or part of a story) to one kind of
audience, but not to another. Thus, the use of actual utterances in their
“linguistic and otherwise” contexts forces us to attend to how speakers’ and
hearers’ needs, goals, and wants are tailored not just to the conventional
meanings of prior text, but also to particular socially and culturally defined
communicative situations.
We have seen in this section that, although ^speechftact theory and Gricean
pragmatics are two approaches to discourse that do not begin from the ana¬
lysis of utterances, the insights that they offer for the analysis of a single
utterance can be extended to multiple utterances.

3.2 Text and context for “utterances” based approaches

In contrast to the two approaches to discourse that began in philosophy

(speech act theory, pragmatics), the approaches to discourse analysis that
began in anthropology, sociology, and linguistics all began with a focus on
utterances. These approaches offer more explicit concepts and methods by
which to relate what is said in a text to the context in which it is said.
Interactional sociolinguistics is an approach to discourse that views all
language as inherently contextualized and contextualizing. Aspects of lan¬
guage as varied as rhythm, repetition, and syntactic structure, and units as
small as single expressions or as large as texts, all reflect (and construct) the
participation frameworks, identities, encounters, situations, occasions, and so
on in which a single message is situated. To return to the terms used above,
language (including text) is contextualized: relationships between utterances
come to be interpreted as coherent by virtue of the contexts in which they are
situated. But language is also contextualizing: one utterance constructs an
interpretive context (in part, by providing information about background
expectations associated with a given activity or situation) in which the
meaning of a next utterance can be inferred.
The contextualizing processes noted above are achieved through links be¬
tween language and participants’ knowledge of situation. Language provides
connections to context through contextualization cues. We are able to convey
and interpret meaning because we use core (e.g. phonology, syntax) and
marginal (e.g. prosody, rhythm) features of language to signal relationships
between what we say and contextual knowledge. The knowledge signalled by
these features is crucial to the accurate inferencing of what we mean. Because
382 Conclusion

core and marginal features both provide contextualization cues, interactional

sociolinguists sometimes collapse the distinction between language as gram¬
mar (abstract rules) and language as use (performance features). Thus, con¬
textualization cues (in “text”) signal contextual presuppositions (knowledge
about “situation”) that allow the inferencing of a speaker’s meaning.
The ethnography of communication is also a contextual approach to text.
This approach assumes that language is always constitutive of some portion
of social and cultural life. Consistent with the comparative and relativistic
premises of this approach, however, few a priori assumptions are made about
the degree to which language (or which aspect of language) is constitutive of
context. Thus, the relative autonomy (context-dependence or context-inde¬
pendence) of text from context cannot be assumed: it must be discovered
through empirical investigation.
Ethnographers do believe that language is a system that is firmly located
within a system of cultural beliefs and understandings that work together to
define communication and communicative events. The fact that language is so
deeply embedded in culture means that the ethnographic study of language is
broader than the study of grammar per se. Furthermore, the aspects of lan¬
guage that do mesh with context - and the linguistic functions that can be
discovered through contextual analyses - are due to the fact that context itself
has a communicative structure comprised of a variety of components that
co-occur and alternate with one another. It is the higher level structures
provided by the communicative events defined through these components to
which language itself accommodates. However, ethnographers also believe
that some aspects of language (including aspects of text) may very well be part
of a system comprised of smaller systems that are not only partially autono¬
mous from one another, but also perhaps independent of context. The point
is that all fundamental concepts - including the relationship between text and
context, structure and use, code and performance - are to be taken as proble¬
matic and to be investigated (Hymes 1974a).
Finally, as we saw earlier, both conversation analysis and variation analysis
include text as a type of context. We also noted, however, that the CA view
of text-as-context is not the same as the variationist view. The CA view of
text-as-context reflects the ethnomethodological belief that meanings (and
knowledge) are continually adjusted and sequentially emergent. This belief is
also responsible for the idea that whatever aspects of situation-as-context are
relevant to relations between utterances are manifested in utterances them¬
selves. Variation analysis, on the other hand, attributes the sequential posi¬
tioning of utterances to their underlying linguistic properties or abstract
(functional or structural) relations. Variation analysis also separates text-as-
context from situation-as-context; the former are linguistic constraints, the
latter non-linguistic constraints, on linguistic realizations in a particular tex-
tually defined slot. Thus, variationists allow linguistic variants and textual
segments more autonomy relative to their sequential location in a text than
do explicitly contextualizing approaches like interactional sociolinguistics or
 Text and Context 383

the ethnography of communication, or sequentially based approaches like

conversation analysis. (We see in chapter 11 that this separation of text from
situation-as-context is more consistent with the code model of communication
than the interactional model of communication.)
It is important to note, however, that although variationist analyses of
discourse seem reluctant to mesh text and context as two mutually constitutive
systems (cf. interactional sociolinguistics, ethnography of communication,
conversation analysis), variationist analyses of two-party discourse (in which
speakers exchange turns) rely more upon considerations of social meaning
and context than do analyses of narratives. Although Labov (1972e: 252)
proposes that the “fundamental problem of discourse analysis is to show how
one utterance follows another in a rational, rule-governed manner,” he also
notes that “there is no formal basis in sentence-grammar to explicate our
reaction” to many everyday exchanges as coherent and well-formed (Labov
1972d: 299).

4 Conclusion

All the approaches to discourse discussed in this book consider both text and
context as crucial to the analysis of utterances. The approaches vary, however,
in their definitions of context and in their views of how text is related to con¬
text. These differences mean that although all the approaches assume an
interdependency between text and context, they differ in their understanding
of the processes by which speakers/hearers rely upon text and context to
construct/discover the meaning of an utterance and the coherence relations
across utterances.
We noted at the outset of this chapter that context is difficult to define. We
suggested that one reason for this difficulty is that contextual information is
always information that is identified in relation to something else that is the
primary focus of our attention: the identity of that “something else” influen¬
ces our decisions about what counts as context. We have seen in this chapter
that different approaches to discourse define context quite differently. One
reason for this may be that each approach began by focusing on a variety of
issues and questions (on a different “something else”) that then defined the
sort of context likely to be of interest. Those approaches that began by
focusing on the meaning of a single utterance, for example, were less likely to
offer an explicit way to focus on text per se as a source of background
information - simply because text itself was not already included as an object
of inquiry. However, if discourse analysts hope to use such perspectives to
analyze utterances - whether it is the relationship between utterances, or how
adjacent utterances contribute to the communicative content of one another
- then the inclusion of text-as-context seems unavoidable.
384 Conclusion

In conclusion, I would like to suggest — in keeping with the eclecticism

reviewed in this chapter - that text and context cannot be related in any simple
or single way. Not only can context be conceived of very differently, but the
interpretation of “what is said” cannot rest upon a fixed interdependency with
the contexts (whether context as knowledge, situation, or text) in relation to
which an utterance is issued. Put another way, contexts can impose themselves
upon texts (or aspects of texts) to different degrees and in different ways (see
Duranti and Goodwin 1991; Schiffrin 1987a: chapter 10, 1990a).
Our constructed example from an earlier section in this chapter can help
make these points. Compare (la) and (lb) again:

(la) doris: I just noticed that we’re out of milk. We’ll need some tomorrow
morning. I won’t have a chance to buy any today. You can pass
the grocery store on your way home from work later. Even
though I know that you didn’t plan to buy milk there, I want
you to buy some milk there. Will you do that?
ANDY: Okay.
(lb) doris: Uhoh. [looking in the refrigerator]
ANDY: Okay.

Analyzing the text/context relationship in Doris’s utterances is important if

we want to explain how it is that Andy’s Okay seems coherent in relation to
what Doris has just said. To be more specific: Andy can use Okay as a
coherent response to Doris’s utterances in either (la) or (lb). In either case.
Okay can convey Andy’s willingness to undertake the action that Doris has
requested (see also Merritt 1984). But it is precisely in understanding the basis
for the coherence of Okay as a response that we need to consider the text/con¬
text relationship: we cannot possibly understand how Andy’s Okay fits into
the dialogue structure, or how it functions as an agreement to comply with
Doris’s request, without first analyzing the way that text and context worked
together to create Doris’s message.
Earlier I suggested that (la) is a text that can be understood almost com¬
pletely apart from a context, i.e. it relies very little upon its context-as-situ-
ation. (lb) is a text that cannot be understood without its situation-as-context.
We might thus say that (la) and (lb) illustrate that text and context intertwine
with one another to different degrees.
However, (la) and (lb) actually illustrate more complexities in the text/con¬
text relationship than I previously suggested. We said earlier that (la) is
independent of its context in that the request (“I want you to buy milk”) can
be interpreted without the use of situational information. However, the reason
this message is so easily retrieved is that (la) encodes all the background
conditions typically assumed to underlie the formulation of felicitous requests
(chapter 3). If we allow background conditions (and background knowledge
in general) to be a part of context (a cognitive part), we then get quite a
different picture of the text/context relationship. Since (la) displays a great
 Text and Context 385

deal of background information in its text, it can actually be seen as a

reflection of context, and thus totally dependent on (rather than independent
of) context.
There is still another way to analyze text/context in (la) and (lb) - in terms
of conditions of use. Ways of speaking vary according to their situational
appropriateness, such that what sounds right in one situation would sound
very wrong in another. Although (la) and (lb) are appropriate under very
different circumstances, each way of formulating a directive is appropriate in
some situation. In this sense, both (la) and (lb) are equally bound to their
contexts: their difference lies in the type of situational context to which they
are bound.
I have used (la) and (lb) to suggest some of the difficulties of trying to
formulate a single view of the text/context relationship. It is beginning to seem
that we get different views of this relationship'depending upon what facet of
text we focus on, what aspect of context, and what kind of relationship we
are examining. If we focus on interpretability of a text, an elliptical utterance
(such as lb) can be seen as heavily dependent on context; if we focus on
background knowledge, very explicit and detailed utterances (such as la) can
be seen as heavily dependent on context; if we focus on appropriate conditions
of use, all texts can be said to depend on their contexts. Variation such as this
suggests not only that the text/context relationship may be impossible to
analyze in any single way, but that describing the text/context relationship
may require empirical analyses of different aspects of texts, different aspects
of contexts (drawing here, from the contextual analyses offered by the ap¬
proaches most explicitly interested in context as “situation”), and the ways
they are bound to one another.

Notes

1 The term “text” has been used in various ways, including a cognitive sense in which
it refers to a mental representation of what is said (e.g. Webber 1979), and a linguistic
sense in which it refers to passages of sentences that “form a unified whole” and
exhibit semantic cohesion (Halliday and Hasan 1976). Briggs and Bauman (1992)
and Bauman and Briggs (1990) discuss how our cultural conceptions of “text” are
intertwined with genre, social power, context, and intertextuality.
2 Note, however, that the availability of implicatures can constrain the structure of the
lexicon. Horn (1972), for example, suggests that a concept will not be directly
lexicalized if it is generally implicated through an existing word.
3 Rather than analyze my question about your availability as an indirect request (see
chapter 3), we might analyze it as a pre-sequence to my upcoming request (see
chapter 7).
11 Discourse and Communication

1 Introduction

Discourse is used for communication: people use utterances to convey informa¬

tion and to lead each other toward an interpretation of meanings and inten¬
tions. This role greatly increases the scope of discourse analysis, simply because
one has to address how the language of utterances is related to aspects of the
communication process (such as knowledge or intentions) that bear an indirect
(and controversial) relationship to language per se.
Some of the approaches to discourse analysis that we have discussed are
quite explicit in relating their frameworks to communication. Speech act
theory, for example, locates the speech act as the basic unit of human linguis¬
tic communication. Interactional sociolinguistics aims to integrate what we
know about grammar, culture, and interactive conventions into a general
theory of verbal communication. The ethnography of communication seeks
to discover how communication (including, but not limited to, language
use) is culturally organized. Gricean pragmatics begins with a general prin¬
ciple that underlies not just language use, but communication in general.
Although conversation analysis and variation analysis are less openly con¬
cerned with communication, they too end up addressing issues that bear on
how people produce and interpret language during interactions with one
another. It seems, then, that what discourse analysts learn about utterances
has to do with what we know about communication and that the products
of discourse analyses can be integrated with our knowledge of communica¬
tion.
As I discuss in this chapter, however, the approaches to discourse reviewed
in this book differ quite surprisingly in terms of what model of communication
they assume: variationist approaches assume a code model; pragmatic and
speech act approaches assume an inferential model; interactional, ethnographic,
and conversation analytic approaches assume an interactional model. Before
 Discourse and Communication 387

discussing these models in detail, I first describe some aspects of communica¬

tion that are differently conceived by the models, incorporating some compar¬
isons among the six approaches to discourse within this description (section 2).
Section 3 is devoted to a more focused comparison among the different models
of communication assumed by different approaches to discourse. Section 4
summarizes.

2 Aspects of communication

Although virtually all models of communication identify at least three com¬

ponents as pertinent to the exchange of information, those components are
neither labelled nor conceived of in the same way. Smith (1977: 14), for
example, refers to a communicator, a signal, and a recipient, but Martinich
(1984: 17) refers to a sender, a message (something with considerably more
“meaning” than the signal), a medium (e.g. the code in which signals are
conveyed), and a receiver. Most models also assume that participants are ca¬
pable of sharing knowledge and experience, or at least of reaching similar
interpretations, i.e. of intersubjectivity. Indeed, as Taylor and Cameron (1987:
161) note, a principle of intersubjectivity is “so central to the study of verbal
communication in modern times that it might be called its fundamental prin¬
ciple” (see also Taylor 1992). However, models differ in terms of how extens¬
ively they assume participants to rely upon intersubjectivity, and at what point
in the communication process intersubjectivity is salient.
Let us begin our general discussion of communication with participants:
how do we speak of those who communicate and what do we imagine them
doing when we say that they are communicating? Although scholars agree that
the process of communicating involves at least two participant roles, they
disagree on the nature of those roles and their contribution to the process of
communicating. This disagreement pervades even so basic an issue as the
choice of terms through which to refer to participants. For example, the terms
that are often used, “sender” and “receiver,” are borrowed from information
theory: they stress the “transmission” aspect of communication, whereby a
message is assumed to be created at one end of the process and sent to the
other end of the process along a channel (cf. the familiar conduit metaphor;
Reddy 1979). Many linguistic analyses of communication often label “sender”
and “receiver” as “speaker” and “hearer” respectively. However, this choice
of terms is also theoretically implicative and consequential: use of “speaker”
and “hearer” conforms to a tacit assumption that communication is typic¬
ally verbal and dyadic, and that communicative roles can be neatly segre¬
gated as to relatively active versus passive roles (i.e. speaking versus hearing).
Even the use of the more seemingly neutral terms “individual,” “self,” and
“person” reflect different ways of conceptualizing human beings, e.g. as a
388 Conclusion

member of the human kind, as a locus of experience, and as an agent-in-

society respectively (Harris 1989), thus conveying different assumptions about
how human beings are oriented toward participation in the communication
process.
Note that we made implicit assumptions about participant role in each
chapter in part II — each time we presented an example and each time we
had to label the identities of participants. In the speech act chapter I often
used the names of participants (since they were familiar to me), but I also
referred to S (for speaker) and H (for hearer) in the discussion of general
speech act rules. This is consistent with the notion that speech act rules apply
across various situations and situated identities, despite the fact that other
aspects of social context may enter into the choice of linguistic forms by which
to realize the acts. Pragmatics maintained a focus on S and H. In discussion
of the story in the sample analysis, however, I used the speaker’s name - to
reflect the personal basis of the experience being reported. The interactional
sociolinguistic approach is concerned with social and personal identities of
interactants: hence, the use of personal names in the examples, and the extens¬
ive discussion of social identities and personal relationships in explicating the
examples. The examples in the ethnography of communication chapter la¬
belled the identities bestowed by the speech event: L for librarian and P for
patron; S for sociolinguistic interviewer and R for respondent. The implication
here is that participant identity is couched within normatively guided events
for the use of speech. Conversation analysis might rely upon similar labels of
situated identity - if it can be shown that those identities are relevant to, and
displayed through, what is being said. Since the data being used for the CA
analysis were from sociolinguistic interviews, I labelled them that way -
although perhaps just S and H would have been more appropriate since not
all talk seemed bound to the interview per se. Variationist examples elsewhere
in the literature usually use informants’ names to capture their identity as real
people with actual identities, as opposed to hypothetical idealized speakers.
However, I also relied upon IVer and IVee for the variationist examples just
to reflect the speech situation in which the lists being focused upon were
produced.
It is not only the choice of a particular pair of terms for participants that
has theoretical consequences: the use of any two-part division that is expected
to hold for all communicative domains necessarily ignores the ways in which
participation rights may be socially and culturally allocated for particular
speech acts, events, or situations (chapter 5). The use of a two-part division
also implies that no more than two participation roles are available, thus
ignoring the many ways that participants both create and display multiple
participation statuses depending upon their (productive or receptive) orienta¬
tion to what is said (see chapter 4).
During an act of communication, one person is assumed to make something
available to another - a message - that the other did not have before. Receipt
of a message can possibly alter the other’s course of action, even if that
 Discourse and Communication 389
alteration is no more than the creation of a shared piece of information that
is stored for reference or use in the indefinite future (Smith 1977: 15). The
source of a message is differently conceived. For some analysts, for example,
messages are essentially thoughts that are externalized and made available to
others through their representation in a shared code. For others, messages are
communicative intentions, e.g. what a speaker believes or wants. Even when
the content of a message is assumed to be referential (i.e. to say something
about the world), messages can be said to accomplish this referential function
in ways that are relatively code-dependent or context-dependent. Furthermore,
the basically referential meaning of a message can be said either to exhaust its
communicative function or to work along with other types of information (e.g.
social, expressive) that may or may not be conveyed in the same manner or
through the same channel.
• The production and interpretation of messages depends upon the availability
of a medium of communication. Some models of communication are very
code-dependent: they place much of the communicative burden for the receipt
of messages on the signals conveyable in the code, such that messages are said
to be encoded and decoded. But other models depend much less on the code,
assuming that a code is an imperfect and incomplete guide to a communica¬
tor’s intentions. Such models reduce the import of the code by becoming more
context-dependent: without denying the importance of the code, they assume
that codes can convey messages only in relation to the way an utterance is
situated in time, place, and so on, and in the way it is related to what other
messages have already conveyed and what other information is potentially
available for interpretation.
Despite the usual tendency of linguists to focus primarily on language as a
medium of communication, then, many scholars focus on both verbal and
nonverbal channels as mediums of communication. Even among linguists who
focus on the verbal channel, scholars make different assumptions about how
information is made available. As noted above, information may be seen as
codified through the grammatical resources of a language, e.g. a meaning can
be mapped onto a sentence and represented through a sequence of sounds. But
participants may also make inferences (grounded in context and contextual
assumptions) that supplement the content of a verbal-based message with
meanings that could not be conveyed through the grammar. Linguists differ
tremendously on the proper way to describe such a grammar (e.g. as a priori
or emergent; Hopper 1988) and on the number and content of contextually
derived principles.
We noted earlier that a principle of inter subjectivity is central to the study
of communication (Schiffrin 1990a; Taylor 1992). Simply put, intersubjectiv¬
ity has to do with the sharing of knowledge or experience. According to
many scholars, intersubjectivity is seen as relevant to communication at the
two ends of the communicative activity itself: its inception and its completion.
That is, it is often assumed that in order for communication to proceed at
all, people must share certain basic knowledge, e.g. knowledge about the
390 Conclusion

world, about the language to be used, and so on. It is also assumed that
one of the main purposes of communication is the transmission of informa¬
tion from one person to another. This means that one product of communica¬
tion is shared knowledge: a key part of the communicative process involves
person A sending to person B a message (sometimes seen to originate as a
thought or representation of the world) that goes from his own mind to
then reside in the other’s mind. In short, intersubjectivity is often assumed
to have a dual role: it both allows communication, and is achieved by com¬
munication.
At first glance, it might seem as if discourse analysts cannot help but be¬
lieve in intersubjectivity: certainly, communication does both require and cre¬
ate shared knowledge. To take a simple example, suppose you are my
neighbor, you own a dog that is barking late into the night, and I want you
to get your dog to stop barking. There are a number of ways that I can try to
convey this intention to you: I can buy a muzzle and send it to you anony¬
mously, I can complain to you that I am tired because I have not been
sleeping well, I can ask you if you know whose dog has been barking at night,
and so on. Suppose I choose a relatively direct route, and during the course
of a conversation with you, I say the following: “I want your dog to stop
barking.” Now, if this message is to be understood, I must assume that
you share certain world knowledge with me (e.g. dogs bark, owners have
some control over dogs) and certain linguistic knowledge (e.g. when I say
“I,” I am referring to myself, when I say “stop barking,” the change of state
verb “stop” leads to the inference that barking has occurred). All this knowl¬
edge - and of course other knowledge as well - is assumed by me to be
shared with you in order for our communication to proceed at all. Once I
actually present my utterance, and assuming that it is understood, what have
I accomplished? Regardless of what you decide to do about your dog (and
regardless of what you may think of my rather direct tactics in making my
request), what I have accomplished is the transformation of my previously
private intention to communicate something into a publicly shared piece of
information.
It is because communication does have so intimate a relation with shared
knowledge (which is of course more intricate than my example suggested) that
intersubjectivity seems to play so necessary a role in communication. How¬
ever, intersubjectivity is actually an extremely problematic notion, in part
because it dwells at the complicated crux between what is private experience
and/or knowledge (and known just to an internal self) and what is public (and
available for others to witness and/or share).
Now that we have reviewed four aspects of communication (participants,
message, medium, intersubjectivity), let us go on to describe how different
approaches to discourse adopt different perspectives on the nature of com¬
munication. In so doing, we will be speaking of three different models of
communication that define and treat the aspects just discussed (both individ¬
ually and together) quite differently.
 Discourse and Communication 391

3 Discourse analysis and models of

communication

In this section, I describe three different models of communication: a code

model (3.1), an inferential model (3.2), and an interactional model (3.3). The
approaches to discourse discussed in this book are more or less aligned with
one of these models: variationist approaches assume a code model; pragmatic
and speech act approaches assume an inferential model; interactional, ethno¬
graphic and conversation analytic approaches assume an interactional model.
I begin description of each model by considering the participant role of the
person who initiates communication: differences in participant role are related
to different conceptions of the message and the medium, and to assumptions
about intersubjectivity. My discussion of each model includes the way differ¬
ent approaches to discourse are more or less wed to the assumptions of that
model.

3.1 Code model

The main participant role assumed by the code model of communication is a

sender. Examined a bit more closely, a sender has three sequentially ordered
roles. First, a sender has an internally represented proposition (perhaps we can
think of this as a “thought”) that she intends to make accessible to another
person. Second, a sender transforms a thought into a set of externally and
mutually accessible signals, here drawing upon knowledge of a code that is
shared with an intended recipient of the message. Finally, a sender transmits
that thought (a transmission allowed by conversion of the thought into code-
derived signals) to its intended recipient; the recipient then relies upon essen¬
tially the same procedures to decode the signal, retrieve the message, and thus
access another’s thought.
The code model of communication is “entrenched in Western culture” (Sper-
ber and Wilson 1986: 6): it is said to trace back to Aristotle and it underlies
many contemporary linguistic theories. For example, as stated by Katz (1966:
104-104):

The speaker’s message is encoded in the form of a phonetic repres¬

entation of an utterance by means of the system of linguistic rules with
which the speaker is equipped. This encoding then becomes a signal to
the speaker’s articulatory organs, and he vocalizes an utterance of the
proper phonetic shape. This is, in turn, picked up by the hearer’s auditory
organs. The speech sound that stimulates these organs is then converted
into a neural signal from which a phonetic representation equivalent to
392 Conclusion

the one into which the speaker encoded his message is obtained. This
representation is decoded into a representation of the same message that
the speaker originally chose to convey by the hearer’s equivalent system
of linguistic rules.

Various scholars (Sperber and Wilson 1986; R. Harris 1981) argue that Saus-
sure’s development of a theory of signs in the early twentieth century helped
to guarantee the code model of communication a prominent place in linguist¬
ics. In part, this is because the translation of internally represented proposi¬
tions about the world (thoughts) into a format accessible to another person
can be assumed to be greatly facilitated by the use of a shared and fixed code
- a code in which sounds (signifiers) are conventionally (and stably) linked
with meanings (signifieds). What this means for our current purposes is that
the view of the sender (as someone with the three sequential roles noted
above) is not only associated with a particular view of the message (as an
internal proposition), but is also deeply dependent upon a particular view of
the medium used to communicate, i.e. a view that language is a fixed code
(R. Harris 1981). That is, the originator of a message is assumed to map the
thought onto (or realize the thought through) a set of paradigmatic and
syntagmatic choices provided through the code, that are then interpretable by
the receiver because that receiver has access to the same code.
Note that this model of communication restricts messages to thoughts that
the speaker intends to convey, i.e. what sets the whole process in motion is
not a thought per se, but an intention to convey a thought. In fact, various
scholars (e.g. Ekman and Freisen 1969; MacKay 1972) have argued for the
need to make a clear-cut distinction between material that is communicative
(because of its intentional basis, or in MacKay’s terms “goal-directed”) and
material that is informative (that may be interpreted regardless of whether or
not it is intended to be so). Critically, it is only signals encoding a thought
that the speaker intends to transmit that end up being communicative - just
so long as the receiver decodes the signals (relies upon the code) in such a way
as to create in his own mind a thought matching the content of the sender’s
own thought.
Intersubjectivity plays a critical role in the code model of communication.
Not only is the goal of communication to achieve a shared message, but the
process by which that goal is achieved rests upon the prior existence, and use,
of a shared code, i.e. shared units, rules, and so on. Thus, it is no surprise that
Taylor and Cameron (1987: 161) call intersubjectivity the “fundamental prin¬
ciple” of verbal communication - for it is the code model of verbal communica¬
tion that is so often being assumed.
Variation analysis is an approach to discourse that is wed to the code model
of communication. Many of the variationist concepts discussed make sense
only if we assume that sender and recipient encode and decode a code. Take
the notion of variable, for example. Not only does a variable link together
surface forms at a deeper level of representation within the code, but what
 Discourse and Communication 393
provides the basis for the link (i.e. the relationship of referential equivalence)
resides in a shared and fixed code. Put another way, the reason we identify
different ways of speaking as variants is because we assign them the same
conventionally based link between sound and meaning. Furthermore, it is
because meaning is attached to the linguistic code that variationists can ana¬
lyze discourse units (e.g. narratives, lists) apart from the way they are locally
situated in talk.
We also noted that the source of coherence across exchanges in talk is often
located at levels other than the surface forms and meanings of utterances: this
certainly seems to de-emphasize the role of the code. Recall, however, that the
processes by which coherence is created (i.e. by which utterances are produced
and interpreted) are viewed as similar to the encoding and decoding proced¬
ures underlying linguistic rules: the only difference is Jjhat they map form to
action, instead of form to meaning. (Recall, also, the point in chapter 9 that
the abstract speech actions (focused upon by Labov’s method of comprehen¬
sive discourse analysis) rest upon an assumption that discourse coherence
depends upon a complex hierarchical organization derived from the linguistic
analysis of phonology and grammar.) Finally, although social context is clearly
important to the variationist approach, context is viewed as a constraint on
the way people use the code - not as something that pervades the definition
of categories in the code.

3.2 Inferential model

Even though the view of the communicator, the message, and the code are
quite different in the inferential model than in the code model, the inferential
model of communication also depends upon a principle of intersubjectivity.
Let me begin again with a description of the person initiating a communicative
act.
As we saw in section 3.1, the code model of communication focuses only
upon those thoughts that a sender intends to transmit. A considerably more
restricted, and differently defined, set of intentions figures in the sort of
meaning pertinent to communication in the inferential model, and critically
those intentions are not directly concerned with propositions or thoughts per
se. The description of those intentions owes much to the philosopher Grice
(1957, 1968), and it is from his description that we can create a picture of a
communicator as one whose primary role in communication is to display
intentions (see fuller discussion in chapter 6). As we noted earlier, the Gricean
view of communication involves three intentions:

(a) S’s utterance of x to produce a certain response r in a certain audience A

(b) A to recognize S’s intention (a)
(c) A’s recognition of S’s intention (a) to function as at least part of A’s
reason for A’s response r.
394 Conclusion

We viewed these intentions as cycling back upon one another: intention (c) is
that intention (b) function as at least part of the reason for the fulfillment of
intention (a). We also noted that it is only when the three intentions are
activated that an individual is acting as a communicator; it is only when the
three intentions are realized (cf. recognized by an audience, see below), that
communication has occurred. Thus, the inferential model assumes that the
initiator of communication is not one who transmits thoughts, but one who
makes manifest (displays) the sort of reflexive intentions described above.
Although I have done no more than sketch the different intentions claimed
to be crucial to communication, it should be obvious from my description that
a particular capacity of the individual as actor is being focused upon: the
capacity of the individual to make intentions obvious to others. Such a focus
has a striking consequence. Once we view “having and displaying intentions
as the central means by which individuals communicate, we also have a means
by which to differentiate behavior which is communicative from behavior
which is not communicative, and, even further, what it is we communicate
when we do communicate. In other words, our view of the person provides a
way to define what we communicate when we send a message to our audience;
that is, we communicate our intentions. And the way we do so is through
utterances: “a modification of the physical environment designed by a com¬
municator to be perceived by an audience” (Sperber and Wilson 1986: 29).
Nothing that I have said thus far implies that speakers’ intentions need be
limited to referential information: for example, one may intend an utterance
to communicate an understanding that its speaker has a feeling of gratitude
(expressive information), or one may intend an utterance to produce a feeling
of interpersonal solidarity in its audience (social information). And in another
sense, intentions are not propositional (i.e. descriptive of the world) at all:
they are more akin to attitudes or beliefs. However, analyses of communication
that focus on intentions tend to focus almost totally on how utterances pro¬
duce understanding of the speaker’s intention to convey a proposition, i.e. the
referential content of a message. Thus, messages are essentially propositions
(properties or actions predicated of entities) that the speaker intends to convey.
Let me use a brief example to illustrate this point. Suppose I invite you to a
party that I’m having on June 16 at 8 p.m. One intention in doing so is to
convey a feeling of good will toward you; another is to include you in a
celebration of a particular event; still another is to reciprocate for a dinner
that I had at your house in May. Now, although these may all be my inten¬
tions, in models of communication in which the actor’s primary role is to
display intentions, it is only the purely propositional content of the message
that is the critical intention. My speaker meaning (Grice 1957, 1968), in other
words, would be something like the following: “I want you to come to a party
at my house on June 16 at 8 p.m.”
The role of the addressee in this process is surprisingly simple in some ways,
yet complex in others. As I described above, the communicator’s intentions
are reflexive. Although these intentions are formulated from the speaker’s
 Discourse and Communication 395

point of view, the recipient is assumed to act as a mirror-image of the speaker:

the recipient recognizes intention (c), thereby recognizing intention (b), which
is that the speaker intended recognition of intention (a) which is that an
utterance produce a certain response.
Now, what would the recipient in our hypothetical example above do? My
invitation would produce a certain response (intention (a)) - a response that
amounts to no more than the recipient’s recognizing my intention and thus
understanding the referential meaning of the message that I intended to con¬
vey. The way that such recognition was achieved would have involved recog¬
nition that I had intended this understanding.
An additional feature of the inferential model of communication - one with
consequences for its view of message and recipient - is that messages may be
communicated by way of information that is not itself propositional informa¬
tion. In other words, although a communicatdr’s intention may focus on the
communication of propositional information, that information may be con¬
veyed in ways other than the referential means encoded in the semantics of
the language per se.
Although I discussed this more in chapter 6, recipient recognition of a
communicator’s intentions can be achieved somewhat independently of shared
reliance on (and decoding of) conventional signs. What supplements a shared
code is a shared set of communicative principles: added information about
communicative intentions is inferrable through joint reliance on a set of
maxims concerning the quality, quantity, relevance, and manner of informa¬
tion. The analysis of implicatures thus allows one to account for how people
convey considerably more than is provided through their shared linguistic
code, in particular, through the stable semantic content of their words. Since
messages can be inferred despite reduced attention to the linguistic code,
individuals need not be assumed to rely solely on their shared knowledge of
the sound-meaning correspondences assigned so central a role in the code
model of communication (2.1). But, as I noted above, individuals are assumed
to rely on shared knowledge of another sort - shared knowledge of communica¬
tive principles - for it is these principles that also allow the inference of
communicators’ intentions.
Our discussion thus far has led to the role of intersubjectivity: despite the
considerable shift in our view of the message (from “thoughts” to “inten¬
tions”) and the medium (from “code” to “code plus general communicative
principles”), intersubjectivity has as pervasive a role in the inferential model
of communication as it did in the code model. First, the goal of communication
is the achievement of intersubjectivity, i.e. one person’s recognition of an¬
other’s intentions. Second, intersubjectivity is achieved through a procedure
in which recipient recognition of intentions mirrors the communicator’s dis¬
play of intentions. Third, procedures for achieving intersubjectivity are based
in prior knowledge: people share the same linguistic code, as well as the same
principles of communication. Thus, intersubjectivity remains a fundamen¬
tal principle of verbal communication.
396 Conclusion

Since the inferencing of speaking meaning is central to Gricean pragmatics,

it is not surprising that this perspective assumes the inferential model of
communication. The Gricean view, however, also incorporates the code
model. We saw above that conventional meaning plays a role in the construc¬
tion of messages: what people “say” is “closely related to the conventional
meaning of the words (the sentence)” (Grice 1975: 44); in addition, what is
“said” figures in the processes by which implicatures are calculated. Of course
Grice adds to these “explicit” meanings (Sperber and Wilson 1986: 21-38)
the more “implicit” and code-independent meanings created through the co¬
operative principle. But it is the role of implicated meanings that alters Grice’s
model of communication from being merely a code-based model to an inferen¬
tial model of communication.
Even though we have not dealt with this at length in chapter 6, still another
feature of the Gricean view of communication has to be mentioned. The
implicatures created through utterances are inherently more open ended than
code-based meanings: not only can different assumed contexts lead recipients
toward different implicatures, but the lack of grounding in what is actually
said can lead to the cancellation of an implicature. A well-known example is
that an utterance like Marge has one child leads to the implicature “Marge
has only one child.” This implicature, however, can be cancelled (since one
does not mean “only one”): if we are discussing the availability of financial
assistance which requires that a person have one child, for example, I might
then quite reasonably say Marge has one child, in fact, she has several. That
implicatures can be cancelled is critical to the Gricean view of communication
- for it presents a view of communication as flexible and indeterminate
(Levinson 1983: chapter 3).
Finally, we have repeatedly stated that speakers display intentions that are
intended to be recognized by recipients, and that although one part of such a
display is code-based, another part rests on the cooperative principle and the
other contexts that figure in the calculation of implicatures. This means that
two kinds of knowledge must be assumed to be shared by communicators: one
is shared knowledge of the linguistic code; the other is shared knowledge of
the contexts that help create implicatures (Schiffrin 1990a). That the CP has
status as shared knowledge means that people communicate not just by relying
on linguistic rules (formal grammatical rules that are part of linguistic com¬
petence; Chomsky 1965), but also by relying on a very general set of rationally
based principles that apply not just to language use, but to behavior in general.
And these principles do not have the formal algorithmic nature of linguistic
rules, but are more like norms (Taylor and Cameron 1987), simply because
they are socially shared (in fact, supposedly universally shared) expectations
about behavior whose violations also have the ability to convey meaning.
The model of communication implicit in speech act theory also combines
the code and inferential models (but as I note below, there are some similarities
with the interactional model). Although Gricean principles are assumed to play
a role in inferencing (e.g. of indirect speech acts; Searle 1976), speech act
 Discourse and Communication 397
theory also allows the code a central contribution to meaning (Searle 1969:
42-50; but also Recanati 1987; Martinich 1984). Communication proceeds
when people recognize and classify others’ speech acts. The way they do so is
by matching their knowledge of certain linguistic devices (illocutionary force
indicating devices), along with their knowledge of context, intentions, and so
on, with a particular speech act label. The ability to make such “matches” -
regardless of their code-based or context-based source - is part of linguistic
competence. Thus, speech act identities (indicated partially by the code) are
critical in the communication of intention in the speech act view: we recognize
an utterance as a promise not only by knowing the underlying conditions for
promising, but by knowing the labelled category “promise.” But in the neo-
Gricean inferential model proposed by Sperber and Wilson (1986: 243-54)
communicative intentions (including intentions to perfq^rm certain actions) can
be inferred without the use of such categories.
Although speech act theory stresses speaker intentions (albeit intentions
more limited to illocutionary force than the Gricean view), it also makes a
place for effects upon recipient other than the recognition of speaker intention,
i.e. what we described earlier as perlocutionary effect. Perlocutionary effect is
not the focus of much attention in orthodox speech act theory (although
Austin (1962) and Searle (1969) both note that it cannot always be easily
separated from illocutionary force). But the fact that it is identified as one type
of meaning or act (along with locutionary and illocutionary) does suggest that
the speech act view of communication is also somewhat of an interactional
model. I do not want to overemphasize this similarity, however: like interac¬
tional models of communication, speech act theory assigns a role to unin¬
tended meanings and acts; but unlike interactional models, speech act theory
assumes that intentional meanings play a role much more central to communi¬
cation than unintentional meanings.
In sum, both Gricean pragmatics and speech act theory assume a model of
communication that is centered on the inferencing of speaker meaning, while
also allowing the code a role in such inferencing. Inferences about speaker
meaning are allowed not only by conventional meanings that are linguistically
encoded (either at a sentential or textual level), but also by the operation of
the cooperative principle - a particular kind of cognitive context - in conjunc¬
tion with background and situational knowledge.

3.3 Interactional model

The interactional model of communication shifts our view of participant roles

(the communicator and the recipient), the message, and the medium; it also
places less emphasis on the principle of intersubjectivity. Put most simply, this
model assumes that what underlies communication is behavior - regardless of
whether that behavior is intentional or not. As we will see, this belief shifts
much of the responsibility for communication from initiator (one who displays
398 Conclusion

information) to recipient (one who witnesses and interprets information); it is

this shift that diminishes the role of intersubjectivity.
Let us begin, again, with the communicator. Some scholars believe that
communication can occur regardless of a person’s intentions, and regardless
of whether an utterance is designed to be perceived by a recipient. Such
scholars view an individual as one who communicates not by transmitting a
thought or manifesting an intention, but by displaying information - informa¬
tion which may not even be designed for perception by an audience (cf.
Sperber and Wilson’s definition of utterance, quoted on p. 394). Consider, in
this regard, a key observation from Watzlawick et al. (1967: 48-9):

First of all, there is a property of behavior that could hardly be more

basic and is, therefore, often overlooked: behavior has no opposite. In
other words, there is no such thing as nonbehavior or, to put it even
more simply: one cannot not behave. Now, if it is accepted that all
behavior in an interactional situation has message value, i.e. is communica¬
tion, it follows that no matter how one may try, one cannot not com¬
municate. Activity or inactivity, words or silence all have message value:
they influence others and these in turn, cannot not respond to these
communications and are thus themselves communicating, (emphasis in
original)

Thus, as long as an individual is in an interactive situation (i.e. as long as

behavior is available to others’ observations), that individual communicates
information: even efforts not to communicate would be, by default, communi¬
cative.
This view of what is involved in communication is often characteristic of
those whose focus is the situated nature of all behavior (not just verbal
behavior). Poyatos (1983), for example, argues that both physical reactions
(e.g. sweating, blushing) and paralinguistic qualities of verbal utterances (of
which an individual may be unaware) may convey messages - even if such
behavior is neither intentional nor conscious. That such behavior is considered
potentially communicative is a radical departure from the view described
earlier, not only because it is not intentional behavior, but also because it is
behavior which is not necessarily designed for perception by an audience (see
Schiffrin 1987b).
In some analyses of the individual as one whose communicative role is to
display information, an actor is assigned two different capacities by which
to display information. For example, Goffman’s (1959) distinction between
information given and information given-off clearly differentiates the intentional
from the non-intentional role. Information given is information intentionally
emitted by a person and received by another in the manner intended by the
actor; information given-off, on the other hand, is information which is inter¬
preted for meaning by a recipient even if it has not been intended to convey
that meaning. Because either type of information may be interpreted by a
 Discourse and Communication 399

recipient, either type is communicative. Furthermore, communicative primacy

may be accorded to information either given or given-off: individuals at both
the productive and receptive ends of the communication process may become
quite adept either at manipulating the supposedly unintended meanings of
their own displays or at interpreting the unintended meanings of others’
displays (Goffman 1959; see also Goffman 1963).
How does this portrait of the communicator alter the role of other parts of
the communication process? Above I quoted Watzlawick et al.’s (1967) view
that one cannot not communicate. I will use the examples with which they
continue to illustrate how our conceptions of message and recipient also
change.

The man at a crowded lunch counter who looks straight ahead, or the
. airplane passenger who sits with his eyes closed, arelboth communicating
that they do not want to speak to anybody or be spoken to, and their
neighbors usually “get the message” and respond appropriately by leav¬
ing them alone. This, obviously, is just as much an interchange of com¬
munication as an animated discussion, (p. 49)

Note that in these examples, individuals may very well wish to be left alone,
and may even wish to convey that message to others. But it is not critical that
they intend their gaze per se - or their lack of gaze - to convey that message.
Thus, behaviors such as silence, gaze, and so on can communicate messages
even if they are not intended to do so.
Or take my earlier example of an invitation to come to a party on June 16
at 8 p.m. I suggested that a number of intentions may underlie such an
invitation, e.g. a desire to convey a feeling of good will toward you, to include
you in a celebration of a particular event, to reciprocate for a prior meal. We
noted earlier that for those who view a speaker’s role as primarily to display
intentions, the meaning would be something like “I want you to come to a
party at my house on June 16 at 8 p.m.” For those who view the speaker’s
role as centered around the display of information, however, any of those
above-mentioned intentions could be communicated. Even more strikingly,
those same meanings could be interpreted even if they were not my intentions.
One question that arises from these examples is how unintended meanings
- meanings that may, in fact, not even be verbal or representational - are
created. In both examples, what we need to assume is that an audience uses
its background knowledge to draw inferences about social and expressive
meanings of the particular information to which they have access (either
nonverbal actions such as gaze direction, or verbal actions such as invitations).
Now, such knowledge may indeed be conventional, e.g. interpretations of gaze
are based on socially and culturally created conventions. However, it is prob¬
ably not knowledge that is code-based in the same way (or, more weakly, to
the same degree) as linguistic knowledge - if only because the interpretation
of the information is so dependent on how that information is situated.
400 Conclusion

The notion of “situated” information leads us to another point critical to

understanding the interactional model of communication: it is less code de¬
pendent and more context-dependent than either the code or the inferential
models. As Smith (1977: 14-15) notes, analyses may very well begin with the
signals emitted in a code, but “many more sources of information than one
signal have a significant effect on a recipient’s behavior. Expanding the min¬
imal description to accommodate these other sources of information is vital
to understanding the process of communicating.” Suggested as other sources
of information are not only those typically found in a “context” (e.g. Hymes s
SPEAKING grid, chapter 5), but also the information that recipients them¬
selves bring to communication, e.g. their own background knowledge, psycho¬
logical states, etc.
Let me take an example of a simplified greeting sequence to illustrate how
bringing the notion of “situated information” into a model of communication
can alter our consideration of this process by complicating our analysis of the
meaning of a message. Greetings provide a good example: they rely upon a
limited set of items from the linguistic code to do a great deal of social and
expressive work in a well-circumscribed set of occasions (Schiffrin 1977).
Suppose I enter a room where you are sitting and having a conversation with
someone else. I do not interrupt you, but when you look up and see me, you
say “Hi”; I then respond “Hi.”
How do we analyze the two utterances of “Hi?” We can easily say (as would
the code model) that each utterance is a conventionalized greeting signal; or
we can say (as would the inferential model) that each “Hi” is recognized as
an intention to greet. What the interactional model offers, however, is the
possibility of quite different analyses for each “Hi” - simply because of our
notion of situated information. For example, we might view the initial “Hi”
as a response to the situation itself: you and I have entered a setting that offers
increased mutual access to one another; once you have exhausted a current
involvement and you notice the increased access, you mark that increase with
a relatively informal greeting (that not only reveals something about our social
relationship, but also displays that our prior contact was not all that long ago).
We might also say that the second “Hi” is responding to the same situation
of increased mutual access to which you responded. Alternatively, we might
say that the second “Hi” is responding to the situation that the initial “Hi”
has created - a situation in which the increase in mutual access has already
been marked (and must thus be responded to as a display of that access) and
in which certain possibilities for our upcoming contact have been created.
Regardless of which description we adopt, the point is that once we allow
“situated information” into a model of communication, we can argue that an
utterance is a response not only to information conveyed in a prior utterance,
but to the situation in which the prior utterance is located, or to the situation
that the prior utterance helps create.
Another change in our view of how messages are created arises from the
diminished role assigned to intentions, and relatedly, changes in our concep-
 Discourse and Communication 401
tion of participant role. Since we are allowing two kinds of information to be
communicative (information intentionally designed to be perceived as com¬
municative, and information not so designed) we are, in effect, allowing two
sources for messages - one residing in that aspect of a person emitting informa¬
tion, the other in the person receiving information. Thus, in contrast to both
the code and the inferential models, the recipient in the interactional model
is assigned a much more active role: the recipient finds meaning in another’s
situated behavior and tries to assign possibly multilevelled interpretations
(referential, emotive, social) to whatever information becomes available. This
means that the message that is communicated need not be rooted in either
thoughts or intentions: it is just as likely to emerge from interactions among
information intentionally emitted by an actor and unintentionally emitted by
an actor, the way that information is situated, and the interpretations that the
recipient assigns to that information. *> *
Finally, the principle of intersubjectivity in this view of communication has
a role less pervasive than that described for other models. First, communica¬
tion need not rest upon the achievement of intersubjectivity; as Watzlawick et
al. (1967: 49) state, we cannot “say that ‘communication’ only takes place
when . . . mutual understanding occurs” (cf. Smith 1977: 15). Rather, the goal
of communication is recipient achievement of an interpretation of displayed
information. Second, the procedures used to interpret others’ displays need
not mirror those used to produce displays. A related point is that what a
recipient of a message knows may lead to an interpretation that diverges from
what was intended by its producer. The one place where intersubjectivity does
play a role is in the realm of prior linguistic knowledge: shared knowledge
allows decoding of linguistic information. But because the interactional model
places more emphasis on how information is situated, it is less code-dependent
and thus less dependent on the intersubjectivity assumed to underlie a shared
code.
The interactional model of communication underlies three approaches to
discourse discussed in this book: interactional sociolinguistics, ethnography of
communication, and conversation analysis. Since it is these approaches that
pay most attention to social context (chapter 10), it is not surprising that they
rely upon the model that provides context with a role that is more than
cognitive.
Consider, first, interactional sociolinguistics. Analyses from this approach
often reveal that a single exchange can have multiple interpretations, interpreta¬
tions that may be associated with different social identities and the use of
different background knowledge. This suggests that what is communicated is
inherently situated, and often situated (in an interpretive sense) in different
ways for different people. Situated interpretations are made possible by con-
textualization cues: it is this construct that also closely links interactional
sociolinguistics to the interactional model of communication.
As we discussed in chapter 4, contextualization cues are signalling mechan¬
isms (such as intonation, speech rhythm, lexical, phonetic, syntactic, and
402 Conclusion

textual options) that affect the expressive quality of a message (Gumperz

1982a: 16). Not only can these cues be core or marginal features of language,
they can also be nonverbal, e.g. facial expression, physical posture. Thus, the
verbal/nonverbal, and core/marginal, distinction evaporates once the signifi¬
cance of contextualization cues as indicators of contextual presuppositions is
recognized. Since these cues are not always based in the linguistic code, there
is obviously no place for them in the code model of communication. Rather,
the reliance of these cues upon a great variety of behaviors to contextualize the
meaning(s) of a message fits within a model of communication in which the
code works along with other information sources, and in which all of that
information has meaning only by being interpreted in context.
The nature of messages and the contribution of contextualization cues to
those messages also belong firmly within the interactional model. Contextual¬
ization cues themselves are information given-off: they are almost never con¬
sciously noted or assigned conventional meanings. But by helping to signal
contextual presuppositions, contextualization cues influence both the ex¬
pressive quality of a message and its basic (propositional) meaning (Gumperz
1982a: 131). Thus, contextualization cues facilitate the inferencing of informa¬
tion given as well as information given-off: they contribute to the linguistic
and expressive components found in messages (Goffman 1963: 16).
Contextualization cues are a construct that reflects the assumptions of the
interactional model of communication in still other ways. Consider their
cognitive and social functions, and how these functions bear on participation
in communicative exchanges. We said above that contextualization cues are
information that is given-off. The proposal that people utilize contextualiza¬
tion cues to make inferences (as Gumperz calls them, situated inferences)
means that the ability to rely upon information given-off is part of the cognit¬
ive capacity through which interactants interpret what is going on. Contextu¬
alization cues also have the social function of displaying social identity and
definition of situation. Recall our discussion of footing: “the alignments we
take up to ourselves and the others present as expressed in the way we manage
the production or reception of an utterance” (Goffman 1981c: 128). One way
that interactants indicate shifts in footing and alignment is through contextu¬
alization cues (Goffman 1981c: 126-7). Thus, the social and expressive
functions of contextualization cues are to convey the capacity in which some¬
one who produces an utterance is acting and how that person is relating to
the recipient of the utterance. This function fits comfortably within the inter¬
actional model of communication in which the main capacity of a speaker is
not necessarily to emit information (cf. the sender in the code model) or to
display intentions (cf. the inferential model): a message can be dissected into
different parts for which different aspects of a self might be responsible. The
central role of contextualization cues in interactional sociolinguistics is also
consistent with the added contribution of the recipient in the interactional
model of communication: the recipient finds meaning in another’s situated
behavior and tries to assign possibly multi-levelled interpretations to informa-
 Discourse and Communication 403

tion both given and given-off. Likewise, contextualization cues reflect the
interactional view that communication is situated: cues help to convey the
expressive meaning and the illocutionary force of a message, in part by indic¬
ating in what “frame” (e.g. what speech activity, what key) the recipient
should locate an utterance.
Consider, finally, another way that the participant role assumed by interac¬
tional sociolinguistics is embedded in the interactional model of communica¬
tion. We noted earlier that speaker intention is the foundation of communication
for both pragmatics and speech act theory. Interactional sociolinguistics sup¬
plements the construct of speaker intention with discourse strategy. Whereas
speaker intention is essentially internal to the individual, discourse strategy is
directed from the speaker to a recipient and its presence becomes inferrable
through an observable effect on the recipient. This balance between what is
observable (a recipient response) and what is fiot (an intention, a strategy)
arises from the interactional sociolinguistic insistence upon building general¬
izations from repeated observations of behavior, and from a willingness to
posit constructs that go beyond what can only be seen or heard (cf. conversa¬
tion analysis). Interestingly, however, the willingness to propose unobservable
constructs such as “strategy” returns interactional sociolinguistics to a more
static assumption of intersubjectivity than proposed by conversation analysts
(see below), i.e. intersubjectivity would be based on shared discourse strategies.
The ethnography of communication also assumes an interactional model of
communication. However, the status of the code, the nature of a message, the
importance of context, and the role of participants are all open to empirical
and comparative investigation. In Hymes’s (1972b: 12) words, “just as what
counts as phonemic feature or religious act cannot be identified in advance,
so with what counts as a communicative event.” Nor can one assume the
constant relevance of the code or the context for any particular communicative
event, not only across cultures but also within a single culture. Also not
warrantable apart from empirical investigation is the assumption that mess¬
ages are constructed along identical principles, and that participants have
constant roles in producing and interpreting messages.
Like interactional sociolinguistics, the ethnographic approach views com¬
munication as a process that revolves around a single construct - not contextu¬
alization cue, but the considerably broader construct of culture. We saw how
culture provides the fundamental lens through which behavior is interpreted
in our very first example of the ethnographic approach in chapter 1, in which
an Ojibwa Indian reacted to the sound of thunder in the same way that he
would respond to a human being whose message had not been clear. Culture
defines not only what counts as communication (“no phenomenon can be
defined in advance as never to be counted as constituting a message”: Hymes
1974b: 13), but also how communication is chunked and labelled as speech
acts and speech events. These meaningful units provide contexts in which
messages are produced and interpreted. Thus, an utterance is always culturally
situated (and culturally relative) as a message only through its relationships
404 Conclusion

with other utterances and within an encompassing framework of actions,

beliefs, and situations.
As we discussed in earlier chapters, the linguistic code — so central to
communication in the code model — is also part of culture. Communicative
competence assumes a relationship between code and culture in which culture
encompasses linguistic knowledge; communicative competence is cultural
knowledge that includes social and psychological principles governing the
use of language, as well as abstract “grammatical” rules pertaining to the
linguistic code.
Just as culture allows us to “make sense” of experience, so too does lan¬
guage: “the communicative event is the metaphor, or perspective, basic to
rendering experience intelligible” (Hymes 1974b: 16). In a way, then, language
“allows” the interpretations, the sense-making, and the tacit assumptions that
embody culture. To a certain degree, it is the design features of the linguistic
code that provide language with its sense-making capacity (the capacity so
critical to our notion of culture): “Of codes available to human beings, lan¬
guage, as the one more than any other capable at once of being explicitly
detailed and transcendent of single contexts, is the chief beneficiary under
many circumstances of the primary centrality of communication” (Hymes
1974b: 16). Yet it is not only the code-like properties of language that allow
people to “render experience intelligible”. Hymes (1981: 9) adopts Cassirer’s
(1961: 113) point to make this clear: “individuals do not simply share what
they already possess; it is only by virtue of the sharing process that they attain
what they possess, constructing a ‘shared world’ of meaning within the me¬
dium of language.” It is not just the code per se that enables experience to be
intelligible: it is interactions between human beings that allow language to
have a sense-making capacity - the capacity so central to culture itself. Thus,
the linguistic code serves communication by allowing the potential construc¬
tion of a shared world of meaning (i.e. intersubjectivity, culture).
Finally, the fact that “the sharing process” is so central to communication,
to sense-making, and to culture has an important consequence for the view of
participant role in communication. Hymes (1974b: 13) observes that the
communicative status of approaching footsteps, or the setting of the sun, “is
entirely a question of their construal by a receiver.” The elevated status of the
recipient is completely in keeping with the interactional model of communica¬
tion. Yet, consistent with the ethnographic approach, even the status of the
recipient is particular and relative: one cannot assume that participant roles,
whether they are senders’ intentions or receivers’ interpretations, have a con¬
stant role in the process by which meanings come to be shared.
Although conversation analysis also assumes the interactional model of
communication, it does so almost by default: the code and inferential models
make too many assumptions about stable systems of representations (in the
linguistic code) and single goals underlying communication (whether those
goals are information transmission or recognition of intention) to provide a
framework of assumptions compatible with the practice of CA. Unlike inter-
 Discourse and Communication 405

actional sociolinguistics and the ethnography of communication, CA has no

central construct in relation to which different aspects of communication are
defined. This lack of a central construct is intentional: to assume any single
construct as central to communication would be to work with just the kind
of idealization that Sacks believed would misrepresent what happens in spe¬
cific sets of events. In keeping with the idea that meaning and action are both
context-shaped and context-renewing, however, CA obviously assumes that con¬
text is important during communication. Of course, what is “in” context
cannot be assumed: its relevance must be made manifest through what is said
(chapter 10). The stable and determinate meanings assumed present in a
linguistic code are similarly suspect to generalization. Similarly, intersubject¬
ivity cannot be assumed: it must be displayed in next-turn or immediately
afterward (Schegloff 1992). Finally, since recipient response is methodologic¬
ally and theoretically critical to CA, the fact -that a if interactional model of
communication provides a wider communicative role for a recipient is com¬
patible with CA.

4 Summary

In this chapter, I described three different models of communication. The code

model assumes that an individual transmits a message, through use of a shared
code, to a receiver. The inferential model assumes that an individual displays
intentions that are inferred by a recipient who relies upon both a shared code
and a shared set of communicative principles allowing the use of inferencing
strategies. The interactional model assumes that an individual displays situated
information that is interpreted by a recipient. We have seen that although all
discourse analyses (indeed, all linguistic analyses) assume some underlying
model of communication, the approaches to discourse discussed in this book
differ quite surprisingly in terms of which model they assume.
12 Conclusion: Language as Social
Interaction

1 Introduction

I noted at the outset of this book that discourse analysis is one of the most
vast and least defined areas in linguistics. What I have tried to accomplish in
this book is the description and comparison of six different approaches to the
linguistic analysis of discourse: speech act theory, interactional sociolinguistics,
ethnography of communication, pragmatics, conversation analysis, variation
analysis. As I noted initially, my aim was not to reduce the vastness of
discourse analysis, but to clarify the scope of discourse analysis in such a way
that it can continue to deal with a wide range of problems and phenomena,
but hopefully in a more systematic and theoretically coherent way.
In this closing chapter, I summarize the approaches to discourse that we have
discussed (section 2). I also suggest, despite the sometimes substantial differ¬
ences among the approaches, some general principles of discourse analysis that
emerge from all of the different approaches (section 3). Section 4 concludes
the book.

2 Starting points: differences among approaches

Each approach to discourse that we discussed began by noting a different

issue, problem, or phenomenon, sometimes from the vantage point of very
different frameworks of inquiry and traditions of research. Despite these dif¬
ferent starting points, I have tried to show that each approach can be used to
address some of the same general problems of discourse analysis (e.g. how one
utterance contributes to the communicative content of another, how utterances
 Language as Social Interaction 407
are related to one another) and some of the same specific problems (e.g.
question-answer sequences, referring sequences).
Speech act theory began with the observation that language could be used
not just to describe the world, but to perform actions: its focus was communi¬
cative acts performed through speech. What speech act theory offers to dis¬
course analysis is a set of constitutive rules by which particular labelled speech
actions can be defined. Speakers and hearers are assumed to share these rules
and to use the rules to produce and interpret a variety of actions (regardless
of whether those actions occupy one, or more than one, utterance). The
particular action defined through the rules provides expectations for a next-
action. Thus, discourse coherence emerges on a local, action-by-action basis;
the sequential relationships between actions are derived from the knowledge
that we use to relate an utterance to an action.
.A speech act approach to discourse leads to*a particular view not only of
structure (chapter 9), and of the text/context relationship (chapter 10), but
also of coherence (the overall sense of a discourse: Schiffrin 1985a, 1987a:
21-9, 326-30; see also Dorval 1990) and the processes by which coherence is
found. Coherence is the result of underlying mapping rules (relating an utter¬
ance to the rules defining it as a speech act) and sequencing rules that relate
not the surface linguistic details of what is said, but the actions that have
resulted from the mapping rules. Thus, the speech act approach to coherence
is rule-centered: not only does it require constitutive rules for the very defini¬
tion of speech acts, but it requires mapping rules to link an utterance (or
utterances) to an act, and sequencing rules to link utterance-action correla¬
tions to one another (cf. Labov and Fanshel 1977). We can thus say that
coherence is created by the application of rules. The status of the three
different rule types, however, deserves further analysis. Although mapping and
sequencing rules seem more normative (and thus regulative) than constitutive,
deviations from all three types have similar results, i.e. behavior that seems in¬
appropriate, rather than ungrammatical, and that often can be made appropri¬
ate by being relocated in a different context.
Interactional sociolinguistics began with the social and linguistic meanings
created during interaction. Compared to speech act theory, interactional socio¬
linguistics thus offers a framework in which utterances can be analyzed not
just as actions, but also as indicators of social, cultural, and personal meaning.
Utterances are situated in both local and global contexts of interpretation. Any
one utterance is assumed to be sequentially relevant to what came before (its
local context) and to a general framework of understandings about a particu¬
lar “type” of situation (its global context). Contextualization cues are aspects
of language that evoke both of these contexts - signalling the relevance of
frameworks within which messages are to be understood. Discourse can thus
be seen as a contextualized and contextualizing vehicle for the construction
of different levels of meaning. Utterances contextualize, and are contextualized
by, one another: each utterance is itself linked to a broader framework of
assumptions to which it provides ongoing indices.
408 Conclusion

The verbal and nonverbal cues that are part of cultural repertoires for
signalling meaning are important not just to interactional sociolinguistics, but
also to the ethnography of communication. The ethnographic approach began
with observations of a gap in traditional anthropological and linguistic theory:
a lack of analytical attention to “ways of speaking.” The focus of the ethno¬
graphy of communication is communication as cultural behavior: language is
part of a matrix of meanings, beliefs, and values that extend beyond knowl¬
edge of grammar.
A key concept in the ethnography of communication is communicative
competence. This concept integrates linguistic competence into cultural knowl¬
edge: communicative competence is cultural knowledge that includes social
and psychological principles governing the use of language, as well as abstract
“grammatical” rules pertaining to the linguistic code. But language is more
than an individual possession, capability, or “instrument” that represents
experience (cf. the code model of communication, chapter 11). Language
allows us to “make sense” of experience (“the communicative event is the
metaphor, or perspective, basic to rendering experience intelligible”; Hymes
1972b: 16) not only because of its design features as a code, but also because
its use creates the potential for a shared world of meaning. Thus, discourse is
constitutive of a larger social or cultural reality only because of the properties
of language and because of the ability of human beings to use language to
create, if not intersubjectivity itself (Taylor 1992), at least a belief in intersub¬
jectivity.
Like interactional sociolinguistics, the ethnography of communication also
offers a contextual approach to the analysis of utterances. As just noted, what
is said is always constitutive of a larger social and cultural reality. This means
that the analysis of any single aspect of form or function can only be part of
the overall framework of understanding and interpretation through which
actions and beliefs are created. The communicative units studied by ethno¬
graphers are also believed to constitute one another. Events and situations
stand in a dependency relationship with acts: just as the communicative struc¬
tures of speech events and situations reveal the function of constituent acts,
the function of the act itself contributes to the structure of the event and the
situation. Thus, discourse (and language in general) is a part of culture:
because culture is a framework for acting, believing, and understanding, cul¬
ture is the framework in which communication (and the use of utterances)
becomes meaningful.
Pragmatics began with a focus on a very different kind of meaning than the
contextual approaches just discussed - not social and cultural meaning, but
individual, intention-based meaning that could supplement the logical, pro-
positional, and conventional meanings representable through a linguistic code.
Despite its focus on a different level of meaning, pragmatics also offers another
broad contextual framework for the analysis of utterances. Its contextual focus
is not situational or cultural contributions to utterance meaning, however, but
the very general assumptions that speaker and hearer bring to each and every
 Language as Social Interaction 409
occasion of speaking - assumptions about one another’s cooperative nature.
These assumptions work with textual and situational information to allow
very particular inferences about speaker meaning. Because what is said in one
utterance can contribute to the inference of speaker meaning in another utter¬
ance, discourse can be seen a chain of inferential relationships whose links are
based in relationships that arise from the operation of the maxims (e.g.
quantity, relevance) as they apply across utterances.
Conversation analysis began by searching for ways to discover our ordinary,
everyday procedures for constructing a sense of social and personal reality. Its
main focus is the way language is shaped by context, and, in turn, the way
language shapes context. The contexts upon which conversation analysts focus,
however, are only those that can be empirically attested through actual speech
or behavior. Likewise, inferences about language (or textual) structure, about
speaker intention, about relationships across Utterances, and so on, must be
grounded in actual doings and sayings. Conversation analyses thus end up
offering very close and detailed analyses of the workings of specific devices or
structures in the construction of talk. The contribution that utterances make
to one another’s meanings has to be attested in what is actually said. Similarly,
relations between utterances cannot be assumed (except in next position; Sacks
1971: lecture 4, 11-12): interdependency between utterances (cf. coherence)
must be demonstrated through the way recipients orient toward talk.
Whereas conversation analysis began with a general interest in the local
construction of social order, variation analysis began with an attempt to find
order in areas of language structure and behavior whose patterns seemed to
defy traditional means of discovery. Variation analysis offers an approach to
discourse that applies some of the same tools and concepts to texts as are
applied to lower level (phonological, syntactic) variation. Text and situation
are both divided into discrete and mutually exclusive factors that can be coded,
counted, and compared. Not only do variationists define discourse as a unit
of language larger than a single sentence, they also analyze how that unit is
comprised of smaller units that are systematically related to one another.
Given the focus upon text as comprised of smaller units, and the analysis of
the relationship of those units with one another, the segmentation and cate¬
gorization of text into units that have systematic effects upon each other is
not surprising. Thus, discourse is a unit of linguistic analysis whose coherence
arises because of systematic relationships between the units (whether these are
words, meanings, clauses, or actions) that comprise a text.
My initial differentiation of approaches to discourse was according to what
I still believe is their most significant characteristic: they have very different
origins. The origin of an approach provides different theoretical and meta-
theoretical premises that continue to influence assumptions, concepts, and
methods. For example, different origins may be responsible for different assump¬
tions and beliefs about how human beings engage in meaningful discourse.
Both approaches originating in philosophy, for example, make assumptions
about the role of speaker intention that are not shared by those approaches
410 Conclusion

that begin from social scientific or humanistic directions. Interactional sociolin¬

guistics, the ethnography of communication, and conversation analysis all
downplay the role of speaker intention by constructing more sociogenic ex¬
planations for human behavior, by allowing for the empirical discovery of
culturally diverse concepts of the self (and allied notions of human responsi¬
bility and will), or by subjugating the discovery of speaker intention to em¬
pirical confirmation in hearer response. Despite their specific differences, these
approaches all view human behavior as part of the interactive construction of
meaning. Not only is each utterance actually (rather than hypothetically)
interpreted by a recipient, but each utterance is part of a larger framework of
social and cultural meaning. Thus, utterances can be assigned a level of
particularity (and potential diversity) not at all possible if one relies solely on
constructed, hypothetical utterances as data.
Different origins are also responsible for different conceptions of language
- including assumptions about the stability of linguistic meaning, the contribu¬
tion of linguistic meaning to interactive meaning, and the degree to which
language is designed for communicative purposes. Several approaches (inter¬
actional sociolinguistics, ethnography of communication, conversation analysis)
view communicative meaning as inherently contextualized and contextualiz¬
ing. Some interactional sociolinguistic analyses propose that language is not
only inherently contextual, but communicatively designed. This suggests that
linguistic structures emerge not from the static rules of abstract grammar, but
from a process whereby strategies for constructing texts get “sedimentized” into
seemingly stable forms (DuBois 1987b; Hopper 1987; Hopper and Thompson
1980, 1984). This belief is carried still further in the conversation analytic
distrust of idealizations - a distrust that means that idealization has no role
as a basis for research in social science, as the grounds for ordinary human
action, or as the structural scaffolding upon which the construction of an
utterance depends.
The contextual approaches just noted allow utterances to be less context-
independent (cf. the view that language has a modular structure with auto¬
nomous components, chapter 2) than does the approach most grounded in the
linguistic analysis of text - variation analysis. We can trace this difference to
one of the initial motivations of variation analysis: a desire to solve traditional
problems of linguistic theory using new methods of analysis. Similarly, we
can trace Searle’s belief about the value of idealization in speech act theory to
his interest in abstract grammatical theory and linguistic competence. Searle
(1969) argues that our intuitions provide “idealized models” (p. 56) of the
conditions necessary and sufficient for the utterance of a given sentence as a
successful, non-defective performance of a given act, and that we cannot hope
to be systematic if we do not engage in idealization (p. 55).
Other differences that can be at least partially traced to different origins
include beliefs about methods for collecting and analyzing data. For example,
some approaches focus intensively on interpretation of a few fragments of talk
(e.g. interactional sociolinguistics); others focus on distributions of discourse
 Language as Social Interaction 411
items across a wide range of texts (e.g. variationists). Some approaches avoid
quantitative analysis. The ethnographic approach to discourse, for example,
does not assume either the mutual exclusivity of categories or the ability to
extract one function of an utterance from the myriad functions that are
realized by an act of speaking. Both of these are prerequisites to the kind of
coding, counting, and comparing that quantitative analyses (such as those
carried out by variationists) would require.
Some approaches to discourse (e.g. ethnography of communication, interac¬
tional sociolinguistics) require a great deal of social, cultural, and personal
information about interlocutors and may use interlocutors as informants in
analysis of their own talk. Other approaches (e.g. pragmatics, speech act
theory) assume an idealized speaker/hearer whose specific social, cultural, or
personal characteristics do not enter into participant strategies for building
text at all. Conversation analysis requires that*whate$er aspects of participant
identity are relevant to the production or interpretation of a particular utter¬
ance have to be revealed in locally situated interpretive activity: “it is only
within organizations of sayings and doings that assignably ‘personal’ at¬
tributes are made manifest” (Coulter 1989: 103).
Methodological differences such as these are due, partially, to different
theoretical assumptions - assumptions that are based in the different origins
noted above. If it is assumed, for example, that linguistic meaning is less
important to interactive meaning than are sequential structures of talk (as in
conversation analysis), then an analyst would pay little attention to linguistic
form and structure per se. But if it is assumed that linguistic form and meaning
are constrained by context (i.e. that one is embedded in the other, as in the
variationist approach) then one needs to pay a great deal of attention to form
and function at many different levels of analysis. Or, if it is assumed that many
different features of language provide clues to the contextual presuppositions
by which we interpret personal, social, and cultural meanings (as in interac¬
tional sociolinguistics), then one has to devote attention to the aspects of
language (rhythm, prosody, and so on) that are often overlooked by more
structural analyses. In short, no methodological preferences are reached in a
vacuum: they are all the product of more general beliefs in what constitutes
data and what counts as evidence and “proof.”
Although each approach to discourse has a different starting point - com¬
plete with interest in a different set of problems, a different motivation, and
different assumptions - I have tried to show in this book that each approach
can be used to address some of the same problems of discourse analysis. In
chapters 9 to 11, and thus far in this chapter, I have been discussing how each
approach addresses relatively general problems concerning relationships be¬
tween utterances. I return now to some of the specific problems on which our
sample analyses in each chapter concentrated.
We considered question-answer sequences in chapters on speech act theory,
interactional sociolinguistics, and ethnography of communication. Our speech
act analysis began with the conditions underlying questions and how they
412 Conclusion

applied to one particular utterance. It then considered the conditions under¬

lying two other speech acts - requests, offers - that this utterance also seemed
to perform. Next-utterances were then considered only in relation to how they
provided sequentially appropriate acts defined in relation to an initiating
utterance. Our interactional sociolinguistic analysis of question-answer se¬
quences (chapter 4) was quite different. Although we began with the same
utterances (implicitly acknowledging their multiple functions and sequential
relationships), we considered how different people participated in the three-
part exchange initiated by the question—request—offer. We focused upon the
social meaning created when one person provided a response sequentially
available to another person (“speaking for another”). To do so, we drew upon
other data as a way of proposing that this particular interactional move could
have larger social meanings and could be related to participant identity. Thus,
whereas chapter 3 began with individual knowledge of speech acts, chapter 4
went on to the social meanings created by self/other participation in a speech
act sequence. Our ethnographic analysis (chapter 5) then located “participants
in a speech act sequence” in a still broader framework of analysis. In chapter
5, we considered questions and answers in different speech events (reference
interview, sociolinguistic research interview), showing how the form and func¬
tion of questions could be part of two different communicative events, defined
not only in terms of who participated in each, but also in terms of participant
goals, act sequences, and norms for interaction and interpretation. Thus, what
chapter 5 added to the analysis of question-answer sequences was a concern
for the social and cultural meanings that provide a shared framework for
participation in a communicative event.
Although the shift in framework in chapters 3, 4, and 5 led to a shift in
issues and data, there were some phenomena that were addressed (or could
have been addressed) in each chapter. In chapter 3, for example, we discussed
final rising intonation as an illocutionary force indicating device. By showing
that final rises (in particular types of exchanges) could satisfy some of the
conditions underlying the speech act “question,” we were able to propose that
this intonation could lead to the identification of a declarative statement as a
question. Although we did not explicitly address final rises in chapter 4, this
is an issue that could easily be considered in the interactional sociolinguistic
framework: final rising intonation is a contextualization cue that helps to
trigger a situated inference about a particular speech action. Chapter 5 did
consider final rising intonation - not in relation to abstract speech act condi¬
tions, but in relation to actualized speech acts in speech events. Our ethno-
graphically motivated discussion of final rises was less wed to the abstract
meaning of this intonation than to its concrete and particular uses within a
communicative event.
The topics of the sample analyses in chapters 6, 7, and 8 were different, yet
the transitions between the framework of inquiry in each chapter were sur¬
prisingly consistent. In our chapter on pragmatics (chapter 6), we focused on
the way one person told a story, and on how that person presented the
 Language as Social Interaction 413
characters in his story. But our focus upon an individual telling a story had
significance beyond the choice of data: it reflected the pragmatic focus upon
speaker meaning, and the kind of knowledge upon which a hearer would need
to rely in order to interpret speaker meaning. Thus, our focus in the pragmatics
chapter was somewhat similar to our focus in the speech act chapter: we were
interested in how an individual relies upon a particular kind of knowledge to
produce and interpret language.
Our conversation analysis (in chapter 7) moved the analysis of speaker
meaning into the analysis of how speaker and hearer construct a conversation.
Here we concentrated on the way a particular linguistic construction fit into
the mechanics of conversation: we focused upon “there + BE + ITEM” in relation
to adjacency pairs, turn-taking, topic transitions, and repairs. We also con¬
sidered the role of this construction in relation to figst-and next-mentions of
a referent - allowing some comparisons with the Gricean pragmatic approach
from chapter 6. But rather than address individual knowledge, we addressed
the way self and other - joint participants in talk - used a construction to
organize mentions in talk, and how that organization was related to the other
tasks facing people who are jointly engaged in the construction of conversation.
Our variation analysis (in chapter 8) located the analysis of referring se¬
quences in a textual structure (the “list”) whose organization contrasted with
that of another text (the “narrative”). We said little about the interactions in
which lists (and their referring terms) are presented, focusing instead on how
the underlying organization of the list could influence the use of one form
rather than another. In this sense, we moved from an analysis in which one
person’s interactional mission is to interpret the other’s intention (by drawing
upon shared assumptions about cooperation), to an analysis of how the struc¬
tures of conversation are jointly achieved, and finally, to an analysis of how
norms for constructing a text-type are assumed to underlie the production of
each text-token, in some sense superseding the particular instance in which
each text appears.
As I noted above, although a shift in framework sometimes leads to a shift
in issues and data, there are some phenomena that can be addressed through
different frameworks. Chapters 6, 7, and 8, for example, all addressed the use
of definite noun phrases for first-mentions of a referent, i.e. initial definites.
Our pragmatic analysis (chapter 6) viewed initial definites in relation to the
way the quantity and relation maxims help hearers to use background knowl¬
edge to draw inferences: situational or textual information can allow a hearer
to infer the existence of a referent by providing a framework within which the
existence of such a referent is expected. Our conversation analysis (chapter 7)
suggested that first-mentions - whether definite or indefinite - in there con¬
structions are integrated into the turn-taking and adjacency pair structure of
conversation. Our variation analysis analyzed the alternation between initial
indefinites and definites in terms of the textual structure of lists - showing
that initial definites were a reflection of the category organization underlying
lists.
414 Conclusion
In sum, the order in which I discussed each approach, and the type of inquiry
for each area of empirical focus (each sample analysis), was not random. They
reflected a broadly based transition from a focus upon individual actions,
knowledge, or intentions (speech act theory, pragmatics) to a focus upon how
self and other interactively construct what is said, meant, and done (interac¬
tional sociolinguistics, conversation analysis) to a focus upon the semiotic
systems shared and used by self and other during their interactions (to make
sense of utterances, the ethnography of communication draws upon culture
and the structure of communicative events; variation analysis draws upon the
linguistic structure of texts). It is difficult to allow and account for transitions
from self to self/other to shared semiotic systems in one s analysis of discourse
- even more challenging to incorporate such transitions into a theory of
discourse. But such a goal seems to be a crucial part of a discourse analysis
that seeks to integrate what so many different approaches can offer both
individually and together to the analysis of utterances. In the next section, I
offer a preliminary attempt to bring together some general principles of dis¬
course analysis from the particular approaches we have been discussing.

3 Principles of discourse analysis: language as

social interaction

The subtitle of this section matches the subtitle of this chapter: “language as
social interaction.” I have done this because I believe that it is this formulation
that best unites (at a very general level) the different approaches to dis¬
course that we have discussed. Each approach somehow incorporates this
insight into its specific methods and concepts. Speech act theory focuses upon
the linguistic actions that we perform toward another person - the actions
that initiate (or continue) interaction. The cooperative principle so crucial to
Gricean pragmatics is a principle applicable to human interaction: it is this
assumption that governs the way people interpret one another’s meaning
during interactions with one another. Interactional sociolinguistics and con¬
versation analysis are quite clear in their beliefs that social interaction is the
locus of language use: what we know and understand about interaction com¬
plements our ability to use language. The ethnography of communication
focuses upon communicative situations, events, and acts: these are “units of
analysis” that can be realized only if one assumes that human beings are
interacting with one another. Variation analysis takes as one of its central
goals the analysis of “language as it is used in everyday life by members of
the social order, that vehicle of communication in which they argue with their
wives, joke with their friends, and deceive their enemies” (Labov 1972f: xiii).
It is within everyday speech - or, as I would say, the speech used in everyday
 Language as Social Interaction 415

social interactions — that one can find the “basis of intersubjective knowledge
in linguistics” (Labov 1972f: xiii).
Saying that each approach to discourse views “language as social interac¬
tion has two important consequences. First is to focus discourse analysis on
praxis and process. To be more specific: social interaction is a process whereby
one person has an effect on another. To be involved in a social interaction is
to be involved in an interchange in which our own activities are directed to
other people and others’ activities are directed to us. Ochs (1988: 15) has
stated that “activity mediates linguistic and sociocultural knowledge and that
knowledge and activity impact one another.” (She attributes this view, more
generally, to Piaget 1952; Vygotsky 1962; Bourdieu 1977; Giddens 1979.)
What I would like to add to Ochs’s view is that the activity most pertinent to
our understanding of discourse is interactive activity^ activity that is directed
to another person and has a potential for affecting that other person. Dis¬
course analysis, then, views language as an activity embedded in social inter¬
action. It is thus interactive activity that mediates linguistic and sociocultural
knowledge.
This view of language as an interactive activity mediating linguistic and
sociocultural knowledge has a consequence for the paradigmatic school of
linguistics into which discourse analysis fits. In chapter 2, we reviewed two
different paradigms within linguistics: formalist and functionalist. We noted
that definitions of discourse can be framed within the assumptions of either
paradigm: “discourse is language above the sentence” is more formalist; “dis¬
course is language use” is more functionalist. We proposed an alternative
definition of discourse as “utterances” in an effort to avoid the problems noted
with the other two definitions. If we take “language as social interaction” as
the core shared assumption of all the approaches reviewed, however, we can
say that discourse analysis studies not just utterances, but the way utterances
(including the language used in them) are activities embedded in social inter¬
action. Although this view of the domain of discourse analysis does not
preclude formal analyses of linguistic or interactional structures, it does place
discourse analysis squarely within a functionalist paradigm as described by
Leech (1983: 46), but with some slight modifications. According to Leech,
language is regarded “primarily as a societal phenomenon.” I am proposing
here that the language of discourse is primarily a social interactional phenome¬
non. Linguistic universals tend to be derived from “the universality of the uses
to which language is put in human society.” Here we can substitute the idea
that discourse universals are derived from the universality of uses to which
language is put in human interaction. And as Leech concludes: “formalists
study language as an autonomous system, whereas functionalists study it in
relation to its social function.” Again, we can say that discourse analysts study
the language of utterances in relation to its function in social interaction.
In addition to suggesting that all the approaches to discourse view language
as social interaction, and all are more compatible with a functionalist than a
formalist paradigm, I want to also propose several general principles of
416 Conclusion

discourse analysis. These principles are not intended to supplant those aspects
of theory and methodology which differentiate the specific approaches. Nor
do I want to imply that all the assumptions of one specific approach are
necessarily shared by others; for example, assumptions about the stability of
linguistic meaning, about whether particular features can be studied in isola¬
tion from other features, and so on. Despite such differences, I propose that
all the approaches to discourse share a single set of underlying principles (see
also Schiffrin 1988b, 1990d).

1 Analysis of discourse is empirical.

Data come from a speech community: data are about people using
language, not linguists thinking about how people use language.
Analyses are accountable to the data: they have to explain the data
in both sequential and distributional terms.
Analyses are predictive: they produce hypotheses that can be falsified
or modified by other data.
2 Discourse is not just a sequence of linguistic units: its coherence
cannot be understood if attention is limited just to linguistic form and
meaning.
3 Resources for coherence jointly contribute to participant achievement
and understanding of what is said, meant, and done through everyday
talk. In other words, linguistic forms and meanings work together
with social and cultural meanings, and interpretive frameworks, to
create discourse.
4 The structures, meanings, and actions of everyday spoken discourse
are interactively achieved.
5 What is said, meant, and done is sequentially situated, i.e. utterances
are produced and interpreted in the local contexts of other utteran¬
ces.
6 How something is said, meant, and done - speakers’ selection among
different linguistic devices as alternative ways of speaking - is guided
by relationships among the following:
(a) speaker intentions;
(b) conventionalized strategies for making intentions recognizable;
(c) the meanings and functions of linguistic forms within their
emerging contexts;
(d) the sequential context of other utterances;
(e) properties of the discourse mode, e.g. narrative, description,
exposition;
(f) the social context, e.g. participant identities and relationships,
structure of the situation, the setting;
(g) a cultural framework of beliefs and actions.

It is important to note that these principles are both empirically and theo¬
retically motivated. For example, the cumulative results of analyses of actual
 Language as Social Interaction 417

talk are what motivate the principle of sequentiality. Another equally import¬
ant source is the assumptions implicit in individual approaches to discourse.
For example, the focus on “interactive achievement” is based on conversation
analysis and interactional sociolinguistics. And the focus on how speakers
select among different linguistic devices (in the sixth principle) stems from
variation analysis, specifically the assumption that heterogeneity of form can
be explained by attention to function and to (linguistic and social) context.
I also want to note that there are individual differences in the way these
principles actually guide analyses. Take the second principle: coherence can¬
not be understood if attention is limited just to linguistic form and meaning.
How discourse analysts go beyond language - what they add to language to
explain coherence - clearly differs. Gricean pragmatics, for example, posits
communicative maxims which allow speakers to express intentions beyond
what can be conveyed in the semantic meanings of their words, and allow
hearers to recognize those intentions. Speech act theorists propose multiple
mapping relationships - and rules - linking utterances into sequences, and
linking what is said to what is meant and done. Interactional sociolinguistics
proposes that the process of conveying meaning requires speaker and hearer
to share underlying interpretive schemata whose relevance for the interpreta¬
tion of a particular message is signalled through the use of specific verbal and
nonverbal devices. Despite the differences among these three approaches, each
adheres to the general principle, simply because each is an attempt to account
for discourse coherence by supplementing semantic meaning with context¬
ualizing factors.
Finally, approaches to discourse may be united not only by the shared
principles noted above, but also by the way specific analyses jointly contribute
to three overall research questions. First, how do social and linguistic resources
differentiate a discourse from a random sequence of sentences? The answers
given to this question can contribute to the goals of discourse analysis noted
initially in this book (p. 41). Sequential analyses seek to discover the principles
underlying the order in which one utterance, or one type of utterance, follows
another; semantic and pragmatic analyses seek to describe how the organiza¬
tion of discourse, and the use of particular expressions and constructions
within certain contexts, allow people to convey and interpret meaning. The
general goal of differentiating discourse from a random sequence of sentences
(or utterances) - sequentially, semantically, and pragmatically - underlies all
discourse analysis: the solution suggested by the approaches that view “lan¬
guage as social interaction” is that both social and linguistic resources are
responsible for discourse.
Second, how does our understanding of utterances in discourse contribute
to our knowledge of language? This question has traditionally been posed in
relation to syntactic structures and their semantic interpretations: how does
our knowledge of how a particular construction is used in discourse contribute
to our knowledge of the form and meaning of that construction? But it can
also be asked in relation to areas often considered as part of phonological
418 Conclusion

theory, e.g. what does the use of intonation as a contextualization cue (or as
an indicator of illocutionary force) tell us about the role of intonation in
language? Can phonological variation serve discourse purposes, and if so,
what does that tell us about the reasons for alternating between two different
ways of saying the same thing?
Third, how are social and linguistic resources distributed throughout a
speech community and across speech communities? This is basically a com¬
parative question that focuses on distribution within a speech community (but
across speakers, situations, styles), as well as distribution across speech com¬
munities. What do “unequal” distributions tell us about the way our human
ability to use language becomes valued as symbolic capital? Can access to
social and linguistic resources change over time?
Of course, not all analyses of discourse need to focus specifically on these
questions. Some focus on the linguistic resources differentiating text-types
(narratives versus descriptions) or modalities (spoken versus written). Some
focus on the social and linguistic strategies by which actions are performed
(e.g. how direct and indirect requests are performed). And there are of course
other specific goals. But the answers provided by such studies can be pulled
together in a common effort to answer these three broad research questions.
And like the principles stated above, this is still another means of synthesizing
different approaches to discourse.

4 At the borders of different disciplines

In this book, I discussed six different approaches to discourse analysis that

began in different disciplines. I also considered three issues which greatly
expand the scope of discourse analysis: structure and function, text and con¬
text, discourse and communication. Now I am suggesting, despite the internal
diversity among the approaches and despite the external expansiveness of
discourse analysis, that these approaches share some very general principles.
In closing, however, I want to suggest that discourse cannot be analyzed -
even if one considers one’s analysis linguistically motivated and linguistically
relevant - through one discipline alone. Consider the issues about which all
discourse analysts make assumptions: structure and function, text and con¬
text, discourse and communication. In each pair of concepts, the first member
is the one that fits most comfortably into the realm of linguistic inquiry. To
be specific: structures can be identified at many levels of linguistic organization
(sounds, sentences), but functions are usually seen as non-linguistic (e.g. cog¬
nitive, social); texts are linguistic, but contexts include non-linguistic situ¬
ations and people; even discourse, although rarely seen as confined to language
per se, is certainly more language-centered a concept than communication
(which involves people, intentions, and knowledge).
 Language as Social Interaction 419
In a sense, then, the need to combine the study of structure with that of
function, to understand the relationship between text and context, and to
make clear how discourse is related to communication, is actually a single need.
This need bears directly on the interdisciplinary basis of discourse analysis. I
have said that it is difficult to always know how to separate (and relate)
structure and function, text and context, discourse and communication. But
what I am really saying is that it is difficult to separate language from the rest
of the world. It is this ultimate inability to separate language from how it is
used in the world in which we live that provides the most basic reason for the
interdisciplinary basis of discourse analysis. To understand the language of
discourse, then, we need to understand the world in which it resides; and to
understand the world in which language resides, we need to go outside of
linguistics. When we then return to a linguistic analysis of discourse - to an
analysis of utterances as social interaction - I believe that we will find that
the benefits of our journey have far outweighed its costs.

Notes

1 Labov (personal communication) has contrasted this modus operandi of variation

analysis with that of ethnography of communication, which seeks to use traditional
methods of analysis (i.e. participant observation) to discover and resolve new prob¬
lems (e.g. patterns of communication).
2 It is important to note that researchers in one tradition have also drawn quite
explicitly from the insights of other traditions. One particularly rich area of interaction
is between conversation analysis and ethnography of communication (Moerman
1988); areas whose connections have also been explored are pragmatics and conversa¬
tion analysis (Bilmes 1993; Levinson 1983; chapter 6).
Appendix 1 Collecting Data

Given the vastness and variety of topics and issues that fall under the label
“discourse analysis,” it should not be surprising that a wide range of data can
be used to illustrate how to go about doing an analysis of discourse. In order
to apply the approaches discussed in this book, e.g. by replicating the sample
analyses, using the exercises suggested at the end of each chapter, or finding
other ways to illustrate the approaches, I suggest having available one or more
of the following types of data:
1 A taped conversation between two or three personal acquaintances. This
can be collected during a variety of gatherings or occasions, e.g. a dinner, a
study session, a discussion, a phone conversation. A section of the conversation
should be transcribed (see appendix 2), preferably a section with more than
one “main” speaker, more than one “text type” (e.g. a story, an explanation,
a description), more than one “exchange type” (e.g. question-answer, repairs,
clarifications), and more than one topic (or subtopics). If there is some section
that seems especially interesting for some reason (e.g. maybe it seems puzzling,
maybe there was a misunderstanding, maybe the topics are interesting), this
section might also (or alternatively) be a good choice for transcription.
2 One (or more) instances of some type of institutional talk (e.g. service
encounters in stores) - although these may be more difficult to tape. En¬
counters between people of institutionally differentiated statuses may be help¬
ful, e.g. professor and student, barber and customer, head of a division and
assistant(s). Again, these should be transcribed: although some such en¬
counters are short, it is difficult to remember enough details of the utterances
for analyses illustrating all the different perspectives.
3 Family conversations between parents and children of different ages, or
interactions among children. Again, these should be transcribed (but see Ochs
1979).
4 Sociolinguistic (or other research) interviews (see chapter 5) can provide
a varied group of text-types. They can also be supplemented by other types of
interviews (e.g. television talk shows, news interviews), publicly accessible
 Appendix 1 Collecting Data 421
discussions (e.g. radio call-in shows), or conversational data (as noted in point
1). Again, these should be transcribed.
5 Written discourse from a variety of sources: news articles, academic
articles, letters to the editor, instructional texts (e.g. recipes, home repairs),
etc.
Describe for whatever data you collect as much as possible about the par¬
ticipants and the setting (cf. chapter 5 on the ethnography of communication),
including personal background knowledge about the participants (if avail¬
able).

Suggested readings are: Labov (1984), Milroy (1987), Ochs (1988: chapter 2),
Schiffrin (1987a: 41-8), Stubbs (1983: chapter 11), Tannen (1984: chapter 3),
and Wolfson (1976).
Appendix 2 Transcription Conventions

From DuBois (1991)

Symbols for discourse transcription

Units
Intonation unit (carriage return}
Truncated intonation unit
Word (space)
Truncated word
Speakers
Speaker identity/turn start
Speech overlap []
Transitional continuity
Final
Continuing 5

Appeal ?

Terminal pitch direction

Fall \
Rise /
Level
Accent and lengthening
A
Primary accent
«
Secondary accent
Booster f

Lengthening
Tone
Fall \
Rise /
Fall-rise V
 Appendix 2 Transcription Conventions 423
Rise-fall A
Level
Pause
Long ...(N)
Medium
Short
Latching (0)
Vocal noises
Vocal noises 0
Inhalation (H)
Exhalation (Hx)
Glottal stop %
Laughter @
Quality *
Quality (Y Y)
Laugh quality <@ @>
Quotation quality <QQ>
Multiple quality features (Y(Z Z)Y)
Phonetics
Phonetic/phonemic transcription (/ /)
Transcriber’s perspective
Researcher’s comment (( ))
Uncertain hearing (XX)
Indecipherable syllable X
Specialized notations
Duration (N)
Intonation unit continued &
Intonation subunit boundary I
Embedded intonation unit <1 l>
Reset (Capital Initial)
False start 0
Codeswitching <L2 L2>
Non-transcription lines
Non-transcription line $
Interlinear gloss line $G
Reserved symbols
Phonemic/orthographic
Morphosyntactic coding + *#{)
n
User-definable

Synopsis of Transcription Design Principles

Category definition: define good categories

1 Define transcriptional categories which make the necessary distinc¬
tions among discourse phenomena.
424 Appendix 2 Transcription Conventions

2 Define sufficiently explicit categories.

3 Define sufficiently general categories.
4 Contrast data types.
Accessibility: make the system accessible
5 Use familiar notations.
6 Use motivated notations.
7 Use easily learned notations.
8 Segregate unfamiliar notations.
9 Use notations which maximize data access.
10 Maintain consistent appearance across modes of access.
Robustness: make representations robust
11 Use widely available characters.
12 Avoid invisible contrasts.
13 Avoid fragile contrasts.
Economy: make representations economical
14 Avoid verbose notations.
15 Use short notations for high frequency phenomena.
16 Use discriminable notations for word-internal phenomena.
17 Minimize word-internal notations.
18 Use space meaningfully.
Adaptability: make the system adaptable
19 Allow for seamless transition between degrees of delicacy.
20 Allow for seamless integration of user-defined transcription categories.
21 Allow for seamless integration of presentation features.
22 Allow for seamless integration of indexing information.
23 Allow for seamless integration of user-defined coding information.

From Jefferson (1979)

Transcript notation

1 Simultaneous utterances

Utterances starting simultaneously are linked together with either double or

single left-hand brackets:

rr Tom: rrl used to smoke a lot when I was young

Bob: llI used to smoke Camels

2 Overlapping utterances

When overlapping utterances do not start simultaneously, the point at which

an ongoing utterance is joined by another is marked with a single left-hand
 Appendix 2 Transcription Conventions 425

bracket, linking an ongoing with an overlapping utterance at the point where

overlap begins:

Tom: I used to smoke ^a lot

[
Bob: He thinks he’s real tough

The point where overlapping utterances stop overlapping is marked with a

single right-hand bracket:

i Tom: I used to smoke ra loti more than this

Bob: I see

3 Contiguous utterances

When there is no interval between adjacent utterances, the second being

latched immediately to the first (without overlapping it), the utterances are
linked together with equal signs:

Tom: I used to smoke a lot=

Bob: = He thinks he’s real tough

The equal signs are also used to link different parts of a single speaker’s
utterance when those parts constitute a continuous flow of speech that has
been carried over to another line, by transcript design, to accommodate an
intervening interruption:

Tom: I used to smoke ^a lot more than this=

Bob: You used to smoke
Tom: = but I never inhaled the smoke

Sometimes more than one speaker latches directly onto a just-completed ut¬
terance, and a case of this sort is marked with a combination of equal signs
and double left-hand brackets:

Tom: I used to smoke a lot=

= [[ Bob: _ rrHe thinks he’s tough
Ann: So did I

When overlapping utterances end simultaneously and are latched onto by a

subsequent utterance, the link is marked by a single right-handed bracket and
equal signs:

Tom: I used to smoke ra lot,

]= Bob: I see
Ann: = So did I
426 Appendix 2 Transcription Conventions

4 Intervals within and between utterances

When intervals in the stream of talk occur, they are timed in tenths of a second
and inserted within parentheses, either within an utterance:

(0.0) Lil: When I was (0.6) oh nine or ten

or between utterances:

Hal: Step right up

(1.3)
Hal: I said step right up
(0.8)
Joe: Are you talking to me

A short untimed pause within an utterance is indicated by a dash:

Dee: Umm - my mother will be right in

Untimed intervals heard between utterances are described within double

parentheses and inserted where they occur:

((pause)) Rex: Are you ready to order

((pause))
Pam: Yes thank you we are

5 Characteristics of speech delivery

In these transcripts, punctuation is used to mark not conventional grammatical

units but, rather, attempts to capture characteristics of speech delivery. For
example, a colon indicates an extension of the sound or syllable it follows:

codon Ron: What happened to you

and more colons prolong the stretch:

co::lons Mae: I ju::ss can’t come

Tim: I’m so::: sorry re:::ally I am

The other punctuation marks are used as follows:

A period indicates a stopping fall in tone, not necessarily the end

of a sentence.
 Appendix 2 Transcription Conventions 427

, A comma indicates a continuing intonation, not necessarily between

clauses of sentences.

? A question mark indicates a rising inflection, not necessarily a

question.

? A combined question mark/comma indicates a rising intonation

weaker than that indicated by a question mark.

! An exclamation point indicates an animated tone, not necessarily

an exclamation.

A single dash indicates a halting, abrupt cutoff, or, when multiple

dashes hyphenate the syllables of a word or connect strings of
words, the stream of talk so markecj, has a stammering quality.

Marked rising and falling shifts in intonation are indicated by upward and
downward pointing arrows immediately prior to the rise or fall:

i T Thatcher: I am however (0.2) very i fortunate

(0.4) in having (0.6) a t mar:vlous
depiuty

Emphasis is indicated by underlining:

Ann: It happens to be mine

Capital letters are used to indicate an utterance, or part thereof, that is spoken
much louder than the surrounding talk:

Announcer: an the winner: liz:s (1.4) RACHEL ROBERTS

for YTANKS

A degree sign is used to indicate a passage of talk which is quieter than the
surrounding talk:

00 M: 'hhhh (.) °Um::°'Ow is yih mother

by: th'wa:y.h

Audible aspirations (hhh) and inhalations ('hhh) are inserted in the speech
where they occur:

hhh Pam: An thi(hh)s is for you hhh

'hhh Don: 'hhh O(hh) tha(h)nk you rea(hh)lly

A ‘gh’ placed within a word indicates gutturalness:

428 Appendix 2 Transcription Conventions

gh J: Ohgh(h) h hhuh huh huh huh

A subscribed dot is used as a “hardener.” In this capacity it can indicate, for

example, an especially dentalized “t”:

dot J: Was it Tla:s’ night.

Double parentheses are used to enclose a description of some phenomenon

with which the transcriptionist does not want to wrestle. These can be vocal¬
izations that are not, for example, spelled gracefully or recognizably:

(( )) Tom: I used to ((cough)) smoke a lot

Bob: ((sniff)) He thinks he’s tough
Ann: ((snorts))

or other details of the conversational scene:

Jan: This is just delicious

((telephone rings))
Kim: I’ll get it

or various characterizations of the talk:

Ron: ((in falsetto)) I can do it now

Max: ((whispered)) He’ll never do it

When part of an utterance is delivered at a pace quicker than the surrounding

talk, it is indicated by being enclosed between “less than” signs:

> < Steel: the Guandian newspaper looked through >the

manifestoes< la:st Tweek

6 Transcriptionist doubt

In addition to the timings of intervals and inserted aspirations and inhalations,

items enclosed within single parentheses are in doubt, as in:

( ) Ted: I (’spose I’m not)

(Ben): We all (t- )

Here “spose I’m not,” the identity of the second speaker, and “t-” represent
different varieties of transcriptionist doubt.
Sometimes multiple possibilities are indicated:
 Appendix 2 Transcription Conventions 429
Ted: (spoke to Mark)
(’spose I’m not)

Ben: We all try to figure a anB^) for jt

(stuffing girl)

When single parentheses are empty, no hearing could be achieved for the string
of talk or item in question:

Todd: My ( ) catching
( ): In the highest ( )

Here the middle of Todd’s utterance, the speaker of the subsequent utterance,
and the end of the subsequent utterance could not be recovered.

7 Gaze direction

The gaze of the speaker is marked above an utterance, and that of the
addressee below it. A line indicates that the party marked is gazing toward
the other. The absence of a line indicates lack of gaze. Dots mark the transition
movement from nongaze to gaze, and the point where the gaze reaches the
other is marked with an X:

Beth: .... rX_

Terry- LJerry’s fa rscinated with elephants
Don: . X_

Here Beth moves her gaze toward Don while saying “Terry”; Don’s gaze shifts
toward and reaches hers just after she starts to say “fascinated.”
If gaze arrives within a pause each tenth of a second within the pause is
marked with a dash:

Ann: . X• • . rA_

Well (— r-) We coulda used la liddle, marijuana.=

Beth: lX __

Here Beth’s gaze reaches Ann three-tenths of a second after she has said
“Well-,” and one-tenth of a second before she continues with “We coulda
used....”
Commas are used to indicate the dropping of gaze:

Ann: __
Karen has this new hou:se. en it’s got all this
Beth:

Here Beth’s gaze starts to drop away as Ann begins to say “new.”
430 Appendix 2 Transcription Conventions

Movements like head nodding are marked at points in the talk where they
occur:

Ann: ____
Karen has this new hou:se. en it’s got all this
Beth: _,,, ((Nod))

Here Beth, who is no longer gazing at Ann, nods as the latter says “got.”
Asterisks are used in a more ad hoc fashion to indicate particular phenomena
discussed in the text. In the following fragment, for example, Goodwin uses
them to indicate the position where Beth puts food in her mouth:

Ann:
= like- (0.2) ssilvery:: g-go:ld wwa: dlpaper.
Beth: ****** . . . X_

8 Applause

Strings of X’s are used to indicate applause, with lower- and uppercase letters
marking quiet and loud applause respectively:

Audience: xxXXXXXXXXXXXXxxx

Here applause amplitude increases and then decreases.

An isolated single clap is indicated by dashes on each side of the x:

Audience: -x-

Spasmodic or hesitant clapping is indicated by a chain punctuated by dashes:

Audience: -x-x-x

A line broken by numbers in parentheses indicates the duration of applause

from the point of onset (or prior object) to the nearest tenth of a second. The
number of X’s does not indicate applause duration except where it overlaps
with talk, as in the second of the following examples:

Speaker: I beg >to suppiort the rmlotion<=

I-«-=-1
Audience: =x-xxXXXXXXXXXXXXxxxx-x

Speaker: THIS iWEEK fSO > THAT YOU CAN STILL MAKE,
Audience: 1 xx-XXXXXXXXXXXXXXXXXXXXJ
Speaker: _ f[YEAR MINDS UP<
Audience: lLXXXXXXXXXXXXXX ((edited cut))
 Appendix 2 Transcription Conventions 431

9 Other transcript symbols

The left-hand margin of the transcript is sometimes used to point to a feature

of interest to the analyst at the time the fragment is introduced in the text.
Lines in the transcript where the phenomenon of interest occurs are frequently
indicated by arrows in the left-hand margin. For example, if the analyst had
been involved in a discussion of continuations and introduced the following
fragment:

Don: I like that blue one very much

-> Sam: And I’ll bet your wife would like it
Don: If I had the money I’d get one for he/ tt
Sam: And one for your mother too I’ll bet

the arrows in the margin would call attention to Sam’s utterances as instances
of continuations.
Horizontal ellipses indicate that an utterance is being reported only in part,
with additional speech coming before, in the middle of, or after the reported
fragment, depending on the location of the ellipses. Thus, in the following
example, the parts of Don’s utterance between “said” and “y’know” are omitted:

Don: But I said . . . y’know

Vertical ellipses indicate that intervening turns at talking have been omitted
from the fragment:

Bob: Well I always say give it your all

Bob: And I always say give it everything

Codes that identify fragments being quoted designate parts of the chapter
authors’ own tape collections.

From Schiffrin (1987a)

Transcription conventions

Key to transcription conventions

falling intonation followed by noticeable pause (as at end of

declarative sentence)
432 Appendix 2 Transcription Conventions

? rising intonation followed by noticeable pause (as at end of

interrogative sentence)
5
continuing intonation: may be slight rise or fall in contour
(less than or may be followed by a pause (shorter
than or “?”)
f animated tone
noticeable pause or break in rhythm without falling
intonation (each half-second pause is marked as measured by
stop watch)
self interruption with glottal stop
lengthened syllable
italics emphatic stress
CAPS very emphatic stress

When speech from A and B overlap, the starting point of the overlap is marked
by a left-hand bracket, and the ending point of the overlap is marked by a
right-hand bracket.

a: Do you know what time the party’s supposed to start?

B: Six o’clock.

When lack of space prevents continuous speech from A from being presented
on a single line of text, then *=’ at end of A1 and *=’ at beginning of A2 shows
the continuity.

ai: Do you know what time the party’s supposed to start?=

B: Six o’clock.J
A2: = Because I have to work late tonight.

When speech from B follows speech from A without perceptible pause, then
Z links the end of A with the beginning of B.

A: Do you know the time?

b: Six o’clock. Six o’clock.

When speech from B occurs during what can be heard as a brief silence from
A, then B’s speech is under A’s silence.

a: I can’t wait to go to the party! It’ll be fun.

B: Oh yeh!
 Appendix 2 Transcription Conventions 433

From Tannen (1989a)

Transcription conventions

The following transcription conventions are used.

indicates sentence final falling intonation

indicates clause-final intonation (“more to come”)
?! indicates exclamatory intonation
three dots in transcripts indicate pause of 1/2 second or more
two dots indicate perceptible pause of |ess than 1/2 second
three dots show ellipsis, parts omitted in quotations from other
sources
accent indicates primary stress
CAPS indicate emphatic stress
t Accent on words already in CAPS shows emphatic stress
^ Brackets (with or without top flap) show overlap.
1 Two voices going at once. L Simultaneously.
Brackets with top flap reversed show
latching.
No perceptible inter-turn pause
colon following vowel indicates elongated vowel sound
extra colon indicates longer elongation
hyphen indicates glottal stop: sound abruptly cut off
quotation marks highlight dialogue
Underlining highlights key words and phrases
Left arrows highlight key lines
arrow at right of line indicates —»
speaker’s turn continues without interruption —>
so look for continuation on succeeding line
A upper case “A” indicates pronunciation of the indefinite article
(“a”) as the diphthong ley/. (Note that distinguishing between
the unstressed form of the article “a” and the hesitation marker
“uh” is always an interpretation, as they both have the same
phonetic realization (/A /).
/words/ in slashes show uncertain transcription
/?/ indicates inaudible utterance
( ) Parentheses indicate “parenthetical” intonation: lower
amplitude and pitch plus flattened intonation contour

Suggested readings are: DuBois (1991), Edwards (1991), Edwards and Lam-
pert (1992), Macaulay (1991), Ochs (1979), and Preston (1982).
Appendix 3 Sample Data

HENRY: (1) That’s what I was tellin’ Deborah.

(2) Your children have t'earn what they want.
(3) I once talked to a wise man, his name happened to be un
m-Nixon.
(4) The man was a very smart man.
(5) And I was working very hard,
(6) and I told him, I said “I must save money t’send my children
t’college.”
(7) Y’know what he told me for an answer?
(8) He says, “Henry, children find their own way t’go t’college if
they want to.”
(9) He says, “They make better children.”
IRENE: (10) I agree with [that.=
HENRY: (11) [And I never forgot what he said.=
IRENE: (12) =[I agree with that.
HENRY: (13) =[Absolutely.
DEBBY: (14) Yeh. I think it’s much better t’[raise your children-
HENRY: (15) [Yes. You have-
you don’t give- you won’t- you f-
(16) I’ll tell y’somethin’ else,
(17) them real wealthy people, in most cases their children have
t’work.
DEBBY: (18) Yeh it’s [the middle
IRENE: (19) [That’s right.
HENRY: (20) It’s the middle class . ..
IRENE: (21) It’s the middle class [that gives their kids everything.
DEBBY: (22) [Yeh.
HENRY: (23) [I got-=
DEBBY: (24) Yeh.
 Appendix 3 Sample Data 435

HENRY: (25) =1 got mo[n:ey, and I’ll give y’this, and I’ll=
IRENE: [That’s right! 7
DEBBY: (26) ^ yeh
HENRY: (27) =give y’[that. And not-
DEBBY: (28) [Yeh.
IRENE: (29) But it’s not a [matter of- even- Hen if they have=
HENRY: (30) [And there is less respect=
IRENE: (31) =money- even if they don’t have mo[ney-
HENRY: (32) [But today there is less
respect for the mother, and the father of the
(33) house. The father’s not the kingpin in his own=
DEBBY: (34) um
HENRY: (35) =house. And this is one of the bad things also.=
DEBBY: (36) 2 Hmmm. + "
HENRY: (37) =The fac- that’s basically every thing in America.
(38) Y’don’t have the father, as the kingpin of the house.=
DEBBY: (39) Hmmm.
HENRY: (40) He is the shluf today!=
IRENE: (41) It depends on [the house.
HENRY: (42) [He is not the- he is not the kingpin.
DEBBY: (43) Why d’y’think that happened?
HENRY: (44) Why?
DEBBY: (45) Yeh.
HENRY: (46) Because the- well then- this is the- this is-=
IRENE: (47) ZThe=
HENRY: (48) =[this came out- this came out-
IRENE: (49) =[mothers went t’work. It’s a=
HENRY: (50) [No! This came about-
IRENE: (51) financial situation that t’live the mothers had=
HENRY: (52) [No! This came about-
IRENE: (53) =t’go t’work.
henry: (54) Right. You’re right you’re right you’re right.
(55) This came about, before the second world was when most of
the mothers were home.
(56) When their children went t’school, they came home,
(57) and they fou- the mother was waiting for their children.
(58) But then we had the war,
(59) and the women applied.
(60) and they done a wonderful job,
(61) there’s no question to it.
(62) Eh, the women done the job,
(63) they got- they carried on with the country,
(64) and a lot of them held it,
(65) and the money was good.
(66) Because the needs were greater,
436 Appendix 3 Sample Data

(67) and the more money, the better that you lived.
(68) Y’spend more, you live better than ever.
(69) So then the man and wife worked,
(70) and the kids were running around.
(71) Well you know if you get a sixteen- seventeen year old kid
and the mothers’ workin.
(72) and she’s got a boyfriend,
(73) they’re gonna play!
IRENE: (74) [They’re gonna play anyway.
HENRY: (75) =[They’re gonna play. They’re gonnna make babies!
(76) And this is what happened, [because you got more-
IRENE: (77) [They’re gonna play=
HENRY: (78) =[babies than y’ever had. [That are not married.=
IRENE: (79) =[regardless [It’s the morality.
HENRY: (80) =And y’can’t blame ’em. Look, their nature.
(81) They’re eh people!
(82) [They’re full of- full of eh Heh?
IRENE: (83) [The standards though are different today.
(84) The standards are different today.
HENRY: (85) Standards are different.
(86) But I’m tellin’ y.
(87) If the father is respected, [an:d eh
IRENE: (88) [Henry, lemme ask you question.
(89) Lem- you made a statement that the mothers: run the house.
Right?
(90) All right how many fathers today, hold down two jobs,
because they can’t afford t’live, and are never around.
HENRY: (91) This is bad. Well [this is the society that we’re in.
IRENE: (92) [Look, my husband holds down one job.
(93) He’s not home now. [He hasn’t been home since=
HENRY: (94) He [works hard.
IRENE: =seven thirty this morning.
(95) Basically, I run that house yes.
(96) Because who’s gonna run it?
henry: (97) You can run a hou- [whatcha- now whatcha you can=
IRENE: (98) [you mean he can’t- I mean any kinda-
HENRY: (99) =ran a house- you can run a house a- and do the job,
(100) =which is important, y’ can’t y- a man can’t do it himself,
and a woman can’t do it himself w- if y’ want it to be
successful. In most cases.
IRENE: (101) Why do [they: even, the Jewish people from years ago,=
HENRY: (102) [But eh-
IRENE: =where- where- where there were intermarriages,
(103) why was the point left to being, that whatever the mother is,
that’s what the child is.=
 Appendix 3 Sample Data 437

HENRY: (104) Well this is [debatable that ( )

IRENE: (105) =[Because the mother is the one that raises: the
child.
(106) And makes the child go t’Hebrew, or go t’church or go here,
or go there.
HENRY: (107) But- well this is because she’s home t’see that the child
does it.=
IRENE: (108) Z Right!
HENRY: (109) =But the child- the mother’s not there, so the kid plays hooky.
The kid is playin’ hooky.
IRENE: (110) Right but there’s plenty of mothers that are home, and the
kids play hooky.
HENRY: (HD Heh? *
IRENE: (112) There are plenty of mothers that are home, and [the=
HENRY: (113) [But=
IRENE: =[kids play hooky.
HENRY: =[in most cases eh eh percentage wise, it will work better that
way.
(114) It wor- Irene, you gotta remember one thing.
(115) You cannot put yourself in that category, because you are in
a different circumstance.
(116) You know you went through, and you went through
something else.
(117) T’catch up.
(118) Y’know you got a lot t’catch up.
(119) An:d you’re both working hard,
(120) but it ain’t always gonna be like that.
(121) it’s gonna pay, it’s gonna be different.
IRENE: (122) But [I could never-
HENRY: (123) [But you still have a family structure.
IRENE: (124) Right. [But I could never go back to the point wh-=
HENRY: (125) [Right? Y’still have a family-
IRENE: =where I didn’t work.
(126) Give it up now that I have been exposed to it for a year and
a half,
(127) and feel [being home=
HENRY: (128) Because the- [the goodies are there.=
IRENE: (129) =and just [Money-Not so=
HENRY: (130) = Z Because the goodies are [there!
IRENE: =much the [goodies yo-
HENRY: (131) [But if Jay made a livin’, do you think it would
be right for you to work?
(132) Would your-would your [fam-=
IRENE: (133) [YES. ZNO. Y’know why?=
438 Appendix 3 Sample Data

henry: (134) Absolutely not. [I’m dea- dead against it.

IRENE: (135) Z=Because [I don’t think a woman today is
content to be the stagnant housewife of cooking, and
cleaning,
(136) and you feel like your brain, is- gets to a point, and it stops.=
HENRY: (137) Then have [more babies. That’ll keep y’busy.
IRENE: (138) =[It stops. Oh that’s a-
(139) Y’can’t go on havin’ babies for the rest [of your life.=
HENRY: (140) [Why not?
IRENE: (141) =There’s other things t’do.
HENRY: (142) These women are bored because they [( )
IRENE: (143) [If I’m not allowed
t’work, so I go around t’department [stores and spend=
HENRY: (144) [All right maybe=
IRENE: =money.
henry: (145) =that’s a foolish statement.
(146) But let’s put it this way.
(147) A woman is needed in the house,
(148) t’clean the house.
(149) and t’cook the hou- uh cook the meals,
(150) and clean the clothes,
(151) there is- there is a tremendous amount of work for a [woman.
IRENE: (152) [( )
You’re ( ) Henry.
HENRY: (153) You don’t think there’s a d-a lot of work for [yourself?
ZELDA: (154) [You can get-
you can get anybody t’come in an’ clean: [the=
HENRY: (155) [All week?
ZELDA: (156) =house. That is not the point.
IRENE: (157) That’s not r- No=
ZELDA: (158) [That’s off. No. That’s off, [Henry.
IRENE: (159) =[That’s not true.
HENRY: (160) [You say that’s wrong?
ZELDA: (161) Yep. That’s not a mother’s duty. Just t’clean and cook and
clean.
Bibliography

Abbott, B. (1993) A pragmatic account of the definiteness effect in existential sentences.

Journal of Pragmatics, 19(1): 39-56.
Aries, E. and Johnson, F. (1983) Close friendship in adulthood: Conversational content
between same-sex friends. Sex Roles, 9: 1183-96.
Ariel, M. (1990) Accessing Noun-phrase Antecedents. London: Routledge.
Athanasiadou, A. (1991) The discourse function of questions. Pragmatics, 1(1): 107-22.
Atkinson, M. and Heritage, H. (eds) (1984) Structures of Social Action. Cambridge:
Cambridge University Press.
Austin, J. (1962) How to Do Things with Words. Cambridge, MA: Harvard University
Press.
Bach, E. (1981) Tenses and aspects as functions on verb phrases. In C. Rohrer (ed.),
Time, Tense and Quantifiers. Tubingen: Niemeryer, 19-37.
Bach, K. and Harnish, R. (1982) Linguistic Communication and Speech Acts. Cam¬
bridge, MA: MIT Press.
Bamberg, M. (1991) Narrative activity as perspective taking: The role of emotionals,
negations, and voice in the construction of the story realm. Journal of Cognitive
Psychotherapy, 5(4): 275-90.
Bamberg, M. and Marchman, V. (1991) Binding and unfolding: towards the linguistic
construction of narrative discourse. Discourse Processes. 14(3): 277-306.
Basso, K. (1972) “To give up on words.” Silence in Western Apache culture. In P. Giglioli
(ed.), Language and Social Context. Harmondsworth: Penguin, 67-86.
Bates, E. and MacWhinney, B. (1982) A functionalist approach to grammar. In E. Wanner
and L. Gleitman (eds), Language Acquisition: the State of the Art. New York:
Academic Press, 167-214.
Bateson, G. (1955) A theory of play and fantasy. American Psychological Association
Psychiatric Research Reports, II. Reprinted in Steps to an Ecology of Mind. New
York: Chandler (1972), 177-93.
Bauman, R. (1974) Speaking in the light: the role of the Quaker minister. In R. Bauman
and J. Sherzer (eds), Explorations in the Ethnography of Speaking. Cambridge:
Cambridge University Press, 144-61.
Bauman, R. (1986) Story, Performance, and Event. Cambridge: Cambridge University
Press.
440 Bibliography

Bauman, R. and Briggs, C. (1990) Poetics and performance as critical perspectives on

language and social life. American Review of Anthropology, 19: 59—88.
Becker, A. (1984) Biography of a sentence: a Burmese proverb. In E. Bruner (ed.), Text,
Play and Story: Proceedings of the American Ethnological Society, 135—55.
Becker, A. L. (1988) Language in particular: a lecture. In D. Tannen (ed.), Linguistics
in Context. Norwood, NJ: Ablex Publishing, 17-35.
Bennett, A. (1978) Interruptions and the interpretation of conversation. Proceedings of
the Fourth Annual Meeting of the Berkeley Linguistics Society, 557-75.
Berko-Gleason, J. and Weintaub, S. (1976) The acquisition of routines in child language.
Language in Society, 5: 129-36.
Berlin, B. (1976) The concept of rank in ethnobiological classification: some evidence
from Aguaruna folk botany. In R. Casson (ed.), Language, Culture and Cognition:
Anthropological Perspectives. New York: Macmillan.
Berlin, B., Breedlove, D. and Raven, P. (1974) Principles ofTzeltal Plant Classification.
New York: Academic Press.
Besnier, N. (1989) Information withholding as a manipulative and collusive strategy in
Nukulaelae gossip. Language in Society, 18(3): 315-42.
Biber, D. (1988) Variation across Speaking and Writing. Cambridge: Cambridge Univer¬
sity Press.
Bilmes, J. (1985) “Why that now?” Two kinds of conversational meaning. Discourse
Processes, 8: 319-55.
Bilmes, J. (1988a) The concept of preference in conversation analysis. Language in
Society, 17(2): 161-82.
Bilmes, J. (1988b) Category and rule in conversation analysis. Papers in Pragmatics,
2 (1/2): 25-59.
Bilmes, J. (1993) Accounting practices as an ethnomethodological and linguistic prag¬
matic concern: the case of response priority. Paper presented at International Prag¬
matic Association Meeting, Kobe, Japan.
Blakemore, D. (1987) Semantic Constraints on Relevance. Oxford: Basil Blackwell.
Blakemore, D. (1988) The organization of discourse. In F. Newmeyer (ed.), Linguistics:
the Cambridge Survey, 299-350. Cambridge: Cambridge University Press.
Bloomfield, L. (1933) Language. New York: Henry Holt and Company.
Bolinger, D. (1977) Meaning and Form. London: Longman.
Bolinger, D. (1982) Intonation and gesture. In Papers from the Parassession on Non-
declaritives. Chicago: Chicago Linguistic Society, 1-22.
Bourdieu, P. (1977) Outline of a Theory of Practice. Cambridge: Cambridge University
Press.
Boxer, D. (1993) Social distance and speech behavior: the case of indirect complaints.
Journal of Pragmatics, 19(2): 103-26.
Briggs, C. (1986) Learning How to Ask. Cambridge: Cambridge University Press.
Briggs, C. and Bauman, R. (1992) Genre, intertexuality, and social power. Journal of
Linguistic Anthropology, 2(2): 131-72.
Brown, G. (1977) Listening to Spoken English. London: Longman.
Brown, G., Currie, K. and Kenworthy, J. (1980) Questions of Intonation. London:
Croom Helm.
Brown G. and Yule, G. (1983) Discourse Analysis. Cambridge: Cambridge University
Press.
Brown, P. and Levinson, S. (1987) Politeness. Cambridge: Cambridge University
Press.
 Bibliography 441

Bruner, J. (1986) Actual Minds, Possible Worlds. Cambridge, MA: Harvard University
Press.
Bruner, J. (1990) Autobiography as self. In Acts of Meaning. Cambridge, MA: Harvard
University Press, 99-138.
Button, G. (ed.) (1991) Ethnomethodology and the Human Sciences. Cambridge: Cam¬
bridge University Press.
Button, G. and Casey, N. (1984) Generating topic: the use of topic initial elicitors. In
J. Atkinson and J. Heritage (eds), Structures of Social Action: Studies in Conversation
Analysis. Cambridge: Cambridge University Press, 167-90.
Cameron, D., Frazier, E., Harvey, P., Rampton, B. and Richardson, K. (1991) Ethics,
Advocacy and Empowerment: Issues of Method in Researching Language. London:
Routledge.
Carlson, L. (1981) Aspect and quantification. In Tedeschi and Zaenen (eds.) Syntax and
Semantics 14: Tense and Aspect. New York: Academic Press, 31-64.
Carlson, L. (1984) “Well” in Dialogue Games. Amsterdam: John Benjamins Press.
Cassirer, E. (1961) The Logic of the Humanities. New Haven, CT: Yale University Press.
Chafe, W. (1974) Language and consciousness. Language, 50: 111-13.
Chafe, W. (1976) Givenness, contrastiveness, definiteness, subjects, topics, and point of
view. In C. Li and S. Thompson (eds), Subject and Topic. New York: Academic Press,
25-56.
Chafe, W. (1980) The deployment of consciousness in the production of a narrative. In
W. Chafe (ed.), The Pear Stories: Cognitive, Cultural and Linguistic Aspects of
Narrative Production. Norwood, NJ: Ablex Press, 9-50.
Chafe, W. (1987) Cognitive constraints on information flow. In R. Tomlin (ed.),
Coherence and grounding in discourse. Amsterdam: John Benjamins Press.
Chafe, W. (1992) Prosodic and functional units of language. In J. Edwards and M. Lampert
(eds), Talking Data: Transcription and Coding in Discourse Research. Hillsdale, NJ:
Lawrence Erlbaum Associates.
Chatham, S. (1978) Story and Discourse. Ithaca, NY: Cornell University Press.
Cherry, R. (1990) Politeness in written persuasion. Journal of Pragmatics,
Chomsky, N. (1957) Syntactic Structures. The Hague: Mouton.
Chomsky, N. (1965) Aspects of the Theory of Syntax. Cambridge, MA: MIT Press.
Christian, D. (1973) What do you mean by request for clarification? In R. Fasold (ed.),
Variation in the Form and Use of Language. Washington, DC: Georgetown Univer¬
sity Press, 26-74.
Ciccourel, A. (1972) Basic and normative rules in the negotiation of status and rule. In
D. Sudnow (ed.), Studies in Social Interaction. New York: Free Press, 229-58.
Ciccourel, A. (1980) Language and social interaction: Philosophical and empirical issues.
In D. Zimmerman and C. West (eds), Language and Social Interaction. Special issue
of Sociological Inquiry, 50 (3/4): 1-30.
Clancy, P. (1980) Referential choice in English and Japanese narrative discourse. In
W. Chafe (ed.), The Pear Stories. Norwood, NJ: Ablex Publishers.
Clark, H. (1979) Responding to indirect speech acts. Cognitive Psychology, 11: 430-77.
Clark, H. and Carlson, T. (1981) Context for comprehension. In J. Long and A. Baddeley
(eds), Attention and Performance IX. Hillsdale, NJ: Lawrence Erlbaum, 313-30.
Clark, H. and Carlson, T. (1982) Hearers and speech acts. Language, 58: 332-73.
Clark, H. and Haviland, J. (1977) Comprehension and the given-new contract. In
R. Freedle (ed.), Discourse Production and Comprehension. Norwood, NJ: Ablex
Publishers, 1-40.
442 Bibliography

Clark, H. and Marshall, C. (1981) Definite reference and mutual knowledge. In A. Joshi,
B. Webber and I. Sag (eds), Elements of Discourse Understanding. Cambridge:
Cambridge University Press, 10-63.
Clark, H. and Murphy, G. (1982) Audience design in meaning and reference. In J. LeNy
and W. Kintsch (eds), Language and Comprehension. Amsterdam: North Holland
Publishing Company.
Clark, H., Schreuder, R. and Buttrick, S. (1983) Common ground and the understanding
of demonstrative reference. Journal of Verbal Learning and Verbal Behavior, 22:
245-58.
Clark, H. and Wilkes-Gibbs, D. (1986) Referring as a collaborative process. Cognition,
22: 1-39.
Cole, P. (ed.) (1981) Radical Pragmatics. New York: Academic Press.
Cole, P. and Morgan, J. (1975) Speech Acts (Syntax and Semantics, Volume 3). New
York: Academic Press.
Collett, P. (1983) Mossi salutations. Semiotica, 45 (3/4): 191-248.
Comrie, B. (1985) Tense. Cambridge: Cambridge University Press.
Cooley, C. H. (1902) Human Nature and the Social Order. New York: Scribner.
Corsaro, W. (1979) “We’re friends, right?”: children’s use of access rituals in a nursery
school. Language in Society, 8: 315-36.
Coulter, J. (1989) Mind in Action. Atlantic Highlands, NJ: Humanities Press Interna¬
tional.
Coulthard, M., Montgomery, M. and Brazil, D. (1981) Developing a description of
spoken discourse. In M. Coulthard and M. Montgomery (eds), Studies in Discourse
Analysis. London: Routledge and Kegan Paul.
Coupland, N. (1983) Patterns of encounter management. Language in Society, 12(4):
459-76
Craig, R. and Tracy, K. (1983) Conversational Coherence: Form, Structure, and Strat¬
egy. Beverly Hills, CA: Sage Publications.
Cumming, S. (1984) The sentence in Chinese. Studies in Language, 8(3): 365-95.
Davison, A. (1975) Indirect speech acts and what to do with them. In P. Cole and
J. Morgan (eds), Speech Acts (Syntax and Semantics, Volume 3). New York: Aca¬
demic Press, 143-86.
Deakins, A. (1989) Talk at the top: topics at lunch. Manuscript William Patterson
College, New Jersey.
Dil, A. (ed.) (1971) Language in Social Groups. Stanford, CA: Stanford University Press.
Dines, E. (1980) Variation in discourse: “And stuff like that”. Language in Society,
9(1): 13-33.
Donnellan, K. (1978) Speaker reference, descriptions, and anaphora. In P. Cole (ed.),
Pragmatics (Syntax and Semantics 9). New York: Academic Press, 47-68.
Dorval, B. (ed.) (1990) Conversational Organization and Its Development. Norwood,
NJ: Ablex.
Downing, P. (1980) Factors influencing lexical choice in narrative. In W. Chafe (ed.),
The Pear Stories: Cognitive, Cultural and Linguistic Aspects of Narrative Production.
Norwood, NJ: Ablex Press, 89-126.
Dowty, D. (1986) The effects of aspectual class on the temporal structure of discourse:
semantics or pragmatics? Linguistics and Philosophy, 9: 37-61.
Dry, H. (1981) Sentence aspect and the movement of narrative time. Text, 1: 233-40.
Dry, H. (1983) The movement of narrative time. Journal of Literary Semantics, 12:
19-53.
 Bibliography 443

DuBois, J. (1980) Beyond definiteness: the trace of identity in discourse. In W. Chafe

(ed.), The Pear Stories. Norwood, NJ: Ablex Press, 203-74.
DuBois, J. (1987a) Meaning without intention: lessons from divination. Papers in
Pragmatics, 1(2): 80-122.
DuBois, J. (1987b) The discourse basis of ergativity. Language 63(4): 805-55.
DuBois, J. (1991) Transcription design principles for spoken discourse research. Prag¬
matics, 1(1): 71-106.
Duncan, S. (1972) Some signals and rules for taking speaking turns in conversations.
Journal of Personality and Social Psychology, 6: 341-9.
Duranti, A. (1984) The social meaning of subject pronouns in Italian conversation. Text,
4: 4.
Duranti, A. (1988) Intentions, language, and social action in a Samoan context. Journal
of Pragmatics, 12: 13-33.
Duranti, A. (1989) Ethnography of speaking: toward a linguistics of the praxis. In
. F. Newmeyer (ed.), Linguistics: The Cambridge Survey. Cambridge: Cambridge
University Press.
Duranti, A. and Brenneis, D. (1986) The audience as co-author: an introduction. In
A. Duranti and D. Brenneis (eds), The Audience as Co-author, special issue of Text,
6(3): 239-47.
Duranti, A. and Goodwin, C. (1991) Rethinking Context. Cambridge: Cambridge Univer¬
sity Press.
Durkheim, E. (1893) The Elementary Forms of the Religious Life. Glencoe, IL: Free Press.
Durkheim, E. (1895) The Rules of Sociological Method. New York: Free Press.
Eckert, P. and McConnell-Ginet, S. (1992) Communities of practice: where language,
gender, and power all live. In K. Hall, M. Bucholtz and B. Moonwomon (eds),
Locating Power, Proceedings of the Second Berkeley Women and Language Con¬
ference, University of California, Berkeley, 89-99.
Edwards, J. (1991) Transcription. In W. Bright (ed.), Oxford International Encyclo¬
pedia of Linguistics. New York: Oxford University Press.
Edwards, J. and M. Lampert (eds) (1992) Talking Data: Transcription and Coding in
Discourse Research. Hillsdale, NJ: Lawrence Erlbaum Associates.
Ehrlich, S. (1990) Referential linking and the interpretation of tense. Journal of Prag¬
matics, 14: 57-75.
Ekman, P. (1979) About brows: emotional and conversational signals. In M. von
Cranach, K. Fopa, W. Lepenies and D. Ploog (eds), Human Ecology. Cambridge:
Cambridge University Press, 169-249.
Ekman, P. and Freisen, W. (1969) The repertoire of nonverbal behavior: categories,
origins, usage and coding. Semiotica, 1: 49-98.
Erdmann, P. (1976) There Sentences in English. Munich: Tuduv.
Erickson, F. and Shultz, J. (1982) The Counselor as Gatekeeper: Social Interactions in
Interviews. New York: Academic Press.
Ernst, T. (1983) Towards an Integrated Theory of Adverb Position in English. Bloom¬
ington: Indiana University Linguistics Club.
Ervin-Tripp, S. (1976) Is Sybil there? The structure of American English directives.
Language in Society, 5: 25-66.
Fairclough, N. (1989) Language and Power. London: Longman.
Fasold, R. (1990) Sociolinguistics of Language. Oxford: Blackwell.
Feld, S. (1982) Sound and Sentiment: Birds, Weeping, Poetics, and Song in Kaluli
Expression. Philadelphia: University of Pennsylvania Press.
444 Bibliography
Ferguson, C. (1983) Sports announcer talk: syntactic aspects of register variation.
Language in Society, 12(2): 153-72.
Ferrera, A. (1985) Pragmatics. In T. van Dijk (ed.), Handbook of Discourse Analysis,
Volume 2: Dimensions of Discourse. New York: Academic Press, 137-157.
Figueroa, E. (1990) Sociolinguistic metatheory: an utterance based paradigm. PhD,
Dissertation, Georgetown University.
Fillmore, C. (1968) The case for case. In E. Bach and R. Harns (eds), Universals in
Linguistic Theory. New York: Holt, 1-88.
Firbas, J. (1964) On defining the theme in functional sentence analysis. In Travaux
Linguistiques de Prague, Volume 1. University of Alabama Press, 267-80
Fishman, P. (1983) Interaction: the work women do. In B. Thorne, C. Kramarae and
N. Henley (eds), Language, Gender and Society. Newbury, MA: Newbury House,
89-101.
Fleischman, S. (1990) Tense and Narrativity. Austin: University of Texas Press.
Ford, C. (1993) Grammar in Interaction: Adverbial Clauses in American English Con¬
versation. Cambridge: Cambridge University Press.
Ford, C. and Thompson, S. (1986) Conditionals in discourse: a text-based study from
English. In E. Traugott, C. Ferguson, J. Snitzer Reilly and A. ter Meulen (eds), On
Conditionals. Cambridge: Cambridge University Press.
Ford, C. and Thompson, S. (1992) Projectability in conversation: grammar, intonation
and semantics. Manuscript, University of Wisconsin, Madison.
Foucault, M. (1982) Power/Knowledge: Selected Interviews and Other Writings by
Michel Foucault, 1972-1977. New York: Pantheon.
Fox, B. (1987) Discourse Structure and Anaphora. Cambridge: Cambridge University
Press.
Fox, B. and Thompson, S. (1990) A discourse explanation of the grammar of relative
clauses in English conversation. Language, 66(2): 297-316.
Fraser, B. (1988) Motor oil is motor oil: an account of English nominal tautologies.
Journal of Pragmatics, 12: 215-70.
Freeze, R. (1992) Existentials and other locatives. Language, 68(3): 553-95.
Garfinkel, H. (1967) Studies in Ethnomethodology. Englewood Cliffs, NJ: Prentice Hall.
Garfinkel, H. (1974) On the origins of the term “ethnomethodology”. In R. Turner
(ed.), Ethnomethodology. Harmondsworth: Penguin.
Geertz, C. (1973) Interpretation of Cultures. New York: Basic Books.
Gees, J. and Michaels, S. (1989) Discourse styles: variations across speakers, situations,
and tasks. Discourse Processes, 12(3): 263-6.
Geluykens, R. (1987) Intonation and speech act type: an experimental approach to rising
intonation in queclaratives. Journal of Pragmatics, IT. 483-94.
Geluykens, R. (1988) On the myth of rising intonation in polar questions. Journal of
Pragmatics, 12: 467-85.
Geluykens, R. (1989) Raising questions: question intonation revisited t Journal of Prag¬
matics, 13: 567-75.
Giddens, A. (1979) Central Problems in Social Theory: Action, Structure, and Contra¬
diction in Social Analysis. Berkeley and Los Angeles: University of California Press.
Gilligan, C. (1982) In a Different Voice: Psychological Theory and Women’s Develop¬
ment. Cambridge, MA: Harvard University Press.
Givon, T. (1979) On Understanding Grammar. New York: Academic Press.
Givon, T. (ed.) (1983a) Topic Continuity in Discourse: Quantified Cross-language
Studies. Amsterdam: John Benjamins Press.
 Bibliography 44 5

Givon, T. (1983b) Introduction. In T. Givon (ed.) Topic Continuity in Discourse: a

Quantitative Cross-language Study. Amsterdam: John Benjamins Press.
Givon, T. (1983c) Topic continuity in discourse: a quantitative cross-linguistic study.
In T. Givon (ed.) Typological Studies in Language. Amsterdam: John Benjamins Press.
Givon, T. (1983d) Topic continuity in spoken English. In T. Givon (ed.), Topic con¬
tinuity in Discourse: a Quantitative Cross-language Study, Amsterdam: John Ben¬
jamins Press.
Givon, T. (1989) Mind, Code, and Context: Essays in Pragmatics. Hillsdale, NJ: Law¬
rence Erlbaum Associates.
Givon, T. (1992) The grammar of referential coherence as mental processing instruc¬
tions. Linguistics, 30(1): 5-56.
Godard, D. (1977) Same setting, different norms: phone call beginnings in France and
the United States. Language in Society, 6: 209-19.
Goffman, E. (1959) The Presentation of Self in Everyday Life. New York: Anchor
Books. ff ft
Goffman, E. (1961) The medical model and mental hospitalization. In Asylums. New
York: Anchor Books, 323-86.
Goffman, E. (1963) Behavior in Public Places. New York: Free Press.
Goffman, E. (1967a) On face work. In Interaction Ritual. New York: Anchor Books,
5-46.
Goffman, E. (1967b) The nature of deference and demeanor. In Interaction Ritual. New
York: Anchor Books, 49-95.
Goffman, E. (1971a) Supportive interchanges. In Relations in Public. New York: Basic
Books, 62-94.
Goffman, E. (1971b) Remedial interchanges. In Relations in Public. New York: Basic
Books, 95-187.
Goffman, E. (1974) Frame Analysis. New York: Harper and Row.
Goffman, E. (1981a) Introduction. In Forms of Talk. Philadelphia: University of Penn¬
sylvania Press, 1-4.
Goffman, E. (1981b) Replies and responses. In Forms of Talk. Philadelphia: University
of Pennsylvania Press, 5-77.
Goffman, E. (1981c) Footing. In Forms of Talk. Philadelphia: University of Pennsylvania
Press, 124-59.
Goffman, E. (1981d) Radio talk. In Forms of Talk. Philadelphia: University of Pennsyl¬
vania Press, 197-330.
Goffman, E. (1981e) Response cries. In Forms of Talk. Philadelphia. University of
Pennsylvania Press, 78-123.
Goodwin, C. (1979) Conversation Organization. New York: Academic Press.
Goodwin, C. (1984) Notes on story structure and the organization of participation. In
M. Atkinson and J. Heritage (eds), Structures of Social Action. Cambridge: University
Press, 225-46.
Goodwin, C. and Heritage, J. (1990) Conversation analysis. Annual Review of Anthropo¬
logy, 19: 283-307.
Goodwin, M. (1990) Tactical uses of stories: participation frameworks within girls’ and
boys’ disputes. Discourse Processes, 13: 33-71.
Goodwin, M. (forthcoming) Byplay: playful rendering of talk in stories. In J. Baugh,
G. Guy and D. Schiffrin (eds), Towards a Social Science of Language.
Goody, E. (1978) Towards a theory of questions. In E. Goody (ed.), Questions and
Politeness. Cambridge: Cambridge University Press, 17-43.
446 Bibliography
Gordon, D. and Lakoff, G. (1975) Conversational postulates. In P. Cole and J. Morgan
(eds), Speech Acts (Syntax and Semantics, Volume 3). New York: Academic Press,
83-106.
Greatbatch, D. (1988) A turn taking system for British news interviews. Language in
Society, 17(3): 401-30.
Green, G. (1975) How to get people to do things with words. In P. Cole and J. Morgan
(eds), Speech Acts (Syntax and Semantics, Volume 3). New York: Academic Press,
107-41.
Green, G. (1989) Pragmatics and Natural Language Understanding. Hillsdale, NJ:
Lawrence Erlbaum Associates.
Grice, H. P. (1957) Meaning. Philosophical Review, 67: 377-88.
Grice, H. P. (1968) Utterer’s meaning, sentence-meaning, and word-meaning. Founda¬
tions of Language, 4: 1-18.
Grice, H. P. (1975) Logic and conversation. In P. Cole and J. Morgan (eds), Speech Acts
(Syntax and Semantics, Volume 3). New York: Academic Press, 41-58.
Grice, H. P. (1981) Further notes on logic and conversation. In P. Cole (ed.), Radical
Pragmatics (Syntax and Semantics, Volume 9). New York: Academic Press, 113-28.
Grimes, J. (1975) The Thread of Discourse. The Hague: Mouton.
Grimes, J. (ed.) (1978) Papers on Discourse. Dallas, TX: Summer Institute of Lin¬
guistics.
Grimes, J. (1982) Topics within topics. In D. Tannen (ed.), Analyzing Text and Talk.
Washington, DC: Georgetown University Press.
Grimshaw, A. (1981) Language as Social Resource. Stanford, CA: Stanford University
Press.
Grimshaw, A. (ed.) (1990) Conflict Talk. Cambridge: Cambridge University Press.
Grosz, B. (1981) Focusing and description in natural language dialogues. In A. Joshi,
B. Webber and I. Sag (eds). Elements of Discourse Understanding. Cambridge:
Cambridge University Press, 84-105.
Gumperz, J. (1957) Some remarks on regional and social language differences in India.
(Reprinted in A. Dil (ed.), Language in Social Groups. Stanford, CA: Stanford
University Press, 1-11.)
Gumperz, J. (1958) Dialect differences and social stratification in a North Indian village.
(Reprinted in A. Dil (ed.), Language in Social Groups. Stanford, CA: Stanford
University Press, 25-47.)
Gumperz, J. (1964) Linguistic and social interaction in two communities. American
Anthropologist, 6: 137-53.
Gumperz, J. (1981) The linguistic bases of communicative competence. In D. Tannen
(ed.), Analyzing Discourse: Text and Talk. Washington, DC: Georgetown University
Press, 323-34.
Gumperz, J. (1982a) Discourse Strategies. Cambridge: Cambridge University Press.
Gumperz, J. (1982b) Language and Social identity. Cambridge: Cambridge University
Press.
Gumperz, J. (1985) In D. Schiffrin (ed.) Meaning, Form and Use: Linguistic Applica¬
tions. Washington, DC: Georgetown University Press.
Gunter, R. (1974) Sentences in Dialog. South Carolina: Hornbeam Press.
Guy, G. (1992) Explanation in variable phonology: an exponential model of morpho¬
logical constraints. Language Variation and Change, 3(1): 1-22.
Guy, G., Horvath, B., Vonwiller, J., Daisley, E. and Rogers, I. (1986) An intonational
change in progress in Australian English. Language in Society, 15(1): 23-52.
 Bibliography 447
Halliday, M. (1967) Notes on transitivity and theme in English. Parts 1 and 2. Journal
of Linguistics, 3: 37-81, 199-244.
Halliday, M. (1973) Explorations in the Functions of Language. London: Edward
Arnold.
Halliday, M. (1974) The place of “functional sentence perspective” in the system of
linguistic description. In F. Danes (ed.), Papers on Functional Sentence Perspective.
The Hague: Mouton.
Halliday, M. (1975) Learning How to Mean: Explorations in the Development of
Language. London: Edward Arnold.
Halliday, M. (1978) Language as a Social Semiotic. London: Edward Arnold.
Halliday, M. and Hasan, R. (1976) Cohesion in English. London: Longman.
Hallowed, A. (1964) Ojibwa ontology, behavior and world view. In S. Diamond (ed.),
Primitive Views of the World. New York: Columbia University Press, 49--S2.
Hanks, W. (1990) Referential Practices. Chicago: University of Chicago Press.
Hannay, M. (1985) English Existentials in Functional Grammar. Dordrecht: Foris
Publications.
Harris, R. (1980) The Language Makers. London: Duckworth.
Harris, R. (1981) The Language Myth. London: Duckworth.
Harris, R. and Taylor, T. (1991) Landmarks in Linguistic Thought. London: Routledge.
Harris, Z. (1951) Methods in Structural Linguistics. Chicago: University of Chicago
Press.
Harris, Z. (1952) Discourse analysis. Language, 28: 1-30.
Harris, Z. (1988) Language and Information. New York: Columbia University
Press.
Heath, S. (1982) What no bedtime story means. Language in Society, 11: 49-76.
Herbert, R. (1990) Sex based differences in compliment behavior. Language in Society,
19(2): 201-24.
Heritage, J. (1984a) Garfinkel and Ethomethodology. Oxford: Basil Blackwell.
Heritage, J. (1984b) A change-of-state token and aspects of its sequential placement. In
J. Atkinson and J. Heritage (eds), Structures of Social Action: Studies in Conversation
Analysis. Cambridge: Cambridge University Press, 299-345.
Heritage, J. (1989) Current developments in conversation analysis. In D. Roger and
P. Bull (eds), Conversation: an Interdisciplinary Perspective. Clevedon: Multilingual
Matters, 21-47.
Heritage, J. and Atkinson, J. (1984) Introduction. In J. Atkinson and J. Heritage (eds),
Structures of Social Action: Studies in Conversation Analysis. Cambridge: Cambridge
University Press, 1-16.
Hinrichs, E. (1986) Temporal anaphora in discourses of English. Linguistics and Philo¬
sophy, 9: 63-82.
Hockett, C. (1958) A Course in Modern Linguistics. New York: Macmillan.
Holker, K. (1989) Con and co: continuity and marqueurs in oral discourse. In M. Conte,
J. Petofi and E. Sozer (eds), Text and Discourse Connectedness. Amsterdam/Phil¬
adelphia: John Benjamins Press, 83-92.
Holmes, J. (1989) Sex differences and apologies: one aspect of communicative com¬
petence. Applied Linguistics, 10: 194-213.
Holmes, J. (1990) Apologies in New Zealand English. Language in Society, 19(2):
155-200.
Hopper, P. (1987) Emergent grammar. Proceedings of the Thirteenth Annual Meeting,
Berkeley Linguistics Society. Berkeley: Berkeley Linguistics Society, 139-157.
448 Bibliography
Hopper, P. (1988) Emergent grammar and the a priori grammar postulate. In D. Tannen
(ed.), Linguistics in Context: Connecting Observation and Understanding. Norwood,
NJ: Ablex Press, 117-34.
Hopper, P. and Thompson, S. (1980) Transitivity in grammar and discourse. Language,
55: 251-99.
Hopper, P. and Thompson, S. (1984) The discourse basis for lexical categories in
universal grammar. Language, 60(4): 703-52.
Horn, L. (1972) On the Semantic Properties of the Logical Operators in English.
Indiana University Linguistics Club.
Horn, L. (1985a) Metalinguistic negation and pragmatic ambiguity. Language, 61(1):
121-74.
Horn, L. (1985b) Toward a new taxonomy for pragmatic inference. In D. Schiffrin (ed.),
Meaning, Form and Use in Context: Linguistic Applications (Georgetown University
Round Table on Languages and Linguistics). Washington, DC: Georgetown Univer¬
sity Press, 11-42.
Horvath, B. (1987) Text in conversation: variability in story-telling texts. In K. Denning
et al. (eds), Variation in Language: NWAV-XV at Stanford. Stanford, CA: Stanford
University, Linguistics Department, 212-23.
Houston, A. (1989) The English gerund: syntactic change and discourse function. In R.
Fasold and D. Schiffrin (eds), Language Change and Variation. Amsterdam/Phil¬
adelphia: John Benjamins Press, 173-96.
Hurford, J. and Heasley (1983) Semantics: a Coursebook. Cambridge: Cambridge
University Press.
Hymes, D. (1961) Functions of speech: the evolutionary approach. In F. Gruber (ed.),
Anthropology and Education. Philadelphia: University of Pennsylvania Press,
55-83.
Hymes, D. (1972a) Toward ethnographies of communication: the analysis of communic¬
ative events. In P. Giglioli (ed.), Language and Social Context. Harmondsworth:
Penguin, 21-43 (excerpts from Hymes, D. (1966) Introduction: toward ethnographies
of communication. American Anthropologist, 66(6): 12-25).
Hymes, D. (1972b) Models of the interaction of language and social life. In J. Gumperz
and D. Hymes (eds), Directions in Sociolinguistics: the Ethnography of Communica¬
tion. New York: Holt, Rinehart and Winston, 35-71.
Hymes, D. (1973) Speech and language: On the origins and foundations of inequality
in speaking. Daedalus, Summer: 59-86.
Hymes, D. (1974a) Toward ethnographies of communication. In Foundations in Socio¬
linguistics: an Ethnographic Approach. Philadelphia: University of Pennsylvania Press,
3-28.
Hymes, D. (1974b) Why linguistics needs the sociologist. In Foundations in Sociolin¬
guistics: an Ethnographic Approach. Philadelphia: University of Pennsylvania Press,
69-82.
Hymes, D. (1974c) Linguistic theory and functions in speech. In Foundations in Socio¬
linguistics: an Ethnographic Approach. Philadelphia: University of Pennsylvania Press,
145-78.
Hymes, D. (1974d) Linguistics as sociolinguistics. In Foundations in Sociolinguistics:
an Ethnographic Approach. Philadelphia: University of Pennsylvania Press,
193-209.
Hymes, D. (1981) In Vain I tried to Tell You. Philadelphia: University of Pennsylvania
Press.
 Bibliography 449
Hymes, D. (1984) Linguistic problems in defining the concept of “tribe”. In J. Baugh
and J. Sherzer (eds), Language in Use: Readings in Sociolinguistics. Englewood Cliffs,
NJ: Prentice Hall.
Jacoby, S. and Gonzales, P. (1991) The constitution of expert-novice in scientific
discourse. Issues in Applied Linguistics, 2(2): 149-82.
Jakobson, R. (1960) Closing statement: linguistics and poetics. In T. Sebeok (ed.), Style
in Language. Cambridge, MA: MIT Press, 350-77.
Jefferson, G. (1972) Side sequences. In D. Sudnow (ed.), Studies in Social Interaction.
New York: Free Press, 294-338.
Jefferson, G. (1974) Error corrections an interactional resource. Language in Society,
3: 181-99.
Jefferson, G. (1978) Sequential aspects of storytelling in conversation. In J. Schenkein
(ed.), Studies in the Organization of Conversational Interaction. New York: Free
Press, 219-48.
Jefferson, G. (1979) A technique for inviting laughter and its subsequent acceptance/
declination. In G. Psathas (ed.), Everyday Language: Studies in Ethnomethodology.
New York: Irvington, 79-96.
Jefferson, G. (1980) On “trouble premontory” response to inquiry. Sociological Inquiry,
50: 153-85.
Jefferson, G. (1984) On the organization of laughter in talk about troubles. In J. Atkinson
and J. Heritage (eds), Structures of Social Action: Studies in Conversation Analysis.
Cambridge: Cambridge University Press, 346-69.
Jefferson, G. (1989) Preliminary notes on a possible metric which provides for a
“standard maximum” silence of approximately one second in conversation. In
D. Roger and P. Bull (eds), Conversation: An Interdisciplinary Perspective. Clevedon:
Multilingual Matters, 166-96.
Johnstone, B. (1983) Presentation as proof: the language of Arabic rhetoric. Anthropo¬
logical Linguistics, 25: 47-60.
Kalcik, S. (1975) “. . . like Ann’s gynecologist or the time I was almost raped”: personal
narratives in women’s rap groups. In C. Farra (ed.), Women and Folklore. Austin:
University of Texas Press, 3-11.
Kamp, H. and Rohrer, C. (1983) Tense in texts. In R. Bauerle, C. Schwarze and A. von
Stechow (eds), Meaning, Use and Interpretation of Language. Berlin: W. de Gruyter,
250-69.
Kaplan, R. and Bresnan, J. (1982) Lexical-functional grammar: a formal system for
grammatical representation. In J. Bresnan (ed.), The Mental Representation of Gram¬
matical Relations. Cambridge, MA: MIT Press, 173-281.
Katz, J. (1966) The Philosophy of Language. New York: Harper and Row.
Kay, P. (1975) Synchronic variability and diachronic change in basic color terms.
Language in Society, 4: 257-70.
Keenan, E. (1979) Universality of conversational postulates. Language in Society, 5:
67-80.
Keenan, E. and Schieffelin, B. (1976) Topic as a discourse notion. In C. Li and S. Thompson
(ed.), Subject and Topic. New York: Academic Press, 335-84.
Kipers, P. (1987) Gender and topic. Language in Society, 16(4): 543-58.
Kreckel, M. (1981) Communicative Acts and Shared Knowledge in Natural Discourse.
New York: Academic Press.
Kroeber, A. and Kluckhohn, C. (1952) Culture: a Critical Review of Concepts and
Definitions. New York: Vintage.
450 Bibliography
Kronfeld, A. (1990) Reference and Computation. Cambridge: Cambridge University Press.
Labov, W. (1966) On the grammaticality of everyday speech. Paper presented to the
Linguistic Society of America.
Labov, W. (1972a) The study of language in its social context. In Sociolinguistic
Patterns. Philadelphia: University of Pennsylvania Press, 183-259.
Labov, W. (1972b) The transformation of experience in narrative syntax. In Language
in the Inner City. Philadelphia: University of Pennsylvania Press, 354-96.
Labov, W. (1972c) The isolation of contextual styles. In Sociolinguistic Patterns. Phil¬
adelphia: University of Pennsylvania Press, 70-109.
Labov, W. (1972d) Rules for ritual insults. In Language in the Inner City. Philadelphia:
University of Pennsylvania Press, 297-353.
Labov, W. (1972e) The social stratification of (r) in New York City department stores.
In Sociolinguistic Patterns. Philadelphia: University of Pennsylvania Press, 43-69.
Labov, W. (1972f) Introduction. In Sociolinguistic Patterns. Philadelphia: University of
Pennsylvania Press, xiii-xviii.
Labov, W. (1978) Where does the linguistic variable stop? A reply to Beatriz Lavandera.
Working Papers in Sociolinguistics, 44. Austin, TX: Southwest Educational Develop¬
ment Laboratory.
Labov, W. (1981) Speech actions and reactions in personal narrative. In D. Tannen
(ed.), Analyzing Discourse: Text and Talk (Georgetown University Round Table).
Washington, DC: Georgetown University Press, 219-47.
Labov, W. (1984) Field methods of the project on linguistic change and variation. In
J. Baugh and J. Sherzer (eds), Language in Use. Englewood Cliffs, NJ: Prentice Hall,
28-53.
Labov, W. (1987) The overestimation of functionalism. In R. Dirven and V. Fried
(eds), Functionalism in Linguistics. Philadelphia/Amsterdam: John Benjamins Press,
311-32.
Labov, W. and Fanshel, D. (1977) Therapeutic Discourse. New York: Academic Press.
Labov, W., Karen, M. and Miller, C. (1991) Near-mergers and the suspension of
phonemic contrast. Language Variation and Change, 3(1): 33-74.
Labov, W. and Waletzky, J. (1967) Narrative analysis. In J. Helm (ed.), Essays on the
Verbal and Visual Arts. Seattle: University of Washington Press, 12-44.
Ladd, R. (1978) The Structure of Intonational Meaning. Bloomington: Indiana Univer¬
sity Press.
Lakoff, G. (1987) Women, Fire and Dangerous Things. Chicago: University of Chicago
Press.
Lakoff, G. and Johnson, M. (1980) Metaphors We Live by. Chicago: University of
Chicago Press.
Lakoff, R. (1973) The logic of politeness: or minding your p’s and q’s. Proceedings of
the Ninth regional meeting of the Chicago Linguistics Society, University of Chicago,
292-305.
Lakoff, R. (1975) Language and Woman’s Place. New York: Harper and Row.
Lavandera, B. (1978) Where does the sociolinguistic variable stop? Language in Society,
7: 171-83.
Leech, G. (1983) Principles of Pragmatics. London: Longman.
Lefebvre, C. (1989) Some problems in determining syntactic variables: The case of WH
questions in Montreal French. In R. Fasold and D. Schiffrin (eds), Language Change
and Variation. Philadelphia: John Benjamins Press, 351-66.
Levinson, S. (1983) Pragmatics. Cambridge: Cambridge University Press.
 Bibliography 451

Levinson, S. (1987) Minimization and conversational inference. In M. Papi and

J. Verschueren (eds), The Pragmatic Perspective. Amsterdam: John Benjamins Press,
61-129.
Levi-Strauss, C. (1967) Structural Anthropology. New York: Anchor Books.
Linde, C. and Labov, W. (1975) Spatial networks as a site for the study of language
and thought. Language, 51: 924-39.
Lucy, J. (1992) Language Diversity and Thought. Philadelphia: University of Pennsyl¬
vania Press.
Lyons, J. (1977) Semantics. Cambridge: Cambridge University Press.
Macaulay, R. (1991) “Coz it izny spelt when they say it”: Displaying dialect in writing.
American Speech, 280-91.
McHoul, A. (1987) Why there are no guarantees for interrogators. Journal of Prag¬
matics, 11: 455-71.
MacKay, D. (1972) Formal analysis of communication processes. In R. Hinde (ed.),
Nonverbal Communication. Cambridge: Cambridge University Press, 3-26.
MacLaury, R. (1991) Social and cognitive motivations of change: measuring variability
in color semantics. Language, 67(1): 34-62.
Maitland, K. and Wilson, J. (1987) Pronominal selection and ideological conflict.
Journal of Pragmatics, 11: 495-512.
Martinich, S. (1984) Communication and Reference. Berlin: Walter de Gruyter.
Mathesius, V. (1924) Some notes on the function of the subject in modern English.
CMS 10: 244-8.
Matthiessen, C. and Thompson, S. (1987) The structure of discourse and “subordina¬
tion”. ISI Reprint Series, 87-183.
Mead, G. (1934) Mind, Self and Society. Chicago: University of Chicago Press.
Merritt, M. (1976) On questions following questions (in service encounters). Language
in Society, 5: 315-57.
Merritt, M. (1984) On the use of “okay” in service encounters. In J. Baugh and
J. Sherzer (eds), Language in Use. Englewood Cliffs, NJ: Prentice Hall, 139-47.
Milroy, L. (1987) Observing and Analyzing Natural Language. Oxford: Blackwell.
Moerman, M. (1988) Ethnography and Conversation Analysis. Philadelphia: University
of Pennsylvania Press.
Morgan, J. (1975) Some interactions of syntax and pragmatics. In P. Cole and
J. Morgan (eds), Speech Acts (Syntax and Semantics, Volume 3). New York: Aca¬
demic Press.
Morris, C. (1938) Foundations of the theory of signs. In O. Neurath, R. Carnap and
C. Morris (eds), International Encyclopedia of Unified Science. Chicago: University
of Chicago Press, 77-138.
Muhlhausler, P. and Harre, R. (1990) Pronouns and People. Oxford: Blackwell.
Murray, S. (1985) Toward a model of members’ methods for recognizing interruption.
Language in Society, 13: 31—41.
Myhill, J. (1988) A quantitative study of future tense marking in Spanish. In K. Ferrara
et al. (eds), Linguistic Change and Contact. Austin: The University of Texas, Lin¬
guistics Department, 263-72.
Newmeyer, F. (1983) Grammatical Theory. Chicago: University of Chicago Press.
Newmeyer, F. (1991) Functional explanation in linguistics and the origins of language.
Language and Communication, 11 (1/2): 3-28.
Ochs (Keenan), E. (1976) The universality of conversational implicature. Language in
Society, 5: 67-80.
452 Bibliography

Ochs, E. (1979) Transcription as theory. In E. Ochs and B. Schieffelin (eds), Develop¬

mental Pragmatics. New York: Academic Press, 43-72.
Ochs, E. (1985) Clarification and culture. In D. Schiffrin (ed.), Meaning, Form and Use:
Linguistic Applications. Washington, DC: Georgetown University Press.
Ochs, E. (1988) Culture and Language Development. Cambridge: Cambridge University
Press.
Ochs, E., Taylor, C., Rudolph, D. and Smith, R. (1991) Storytelling as a theory-building
activity. Discourse Processes, 15(1): 37-72.
Partee, B. (1984) Nominal and temporal anaphora. Linguistics and Philosophy,
7: 243-86.
Perlmutter, D. (ed.) (1983) Studies in Relational Grammar. Chicago: University of
Chicago Press.
Philips, S. (1974) Warm Springs “Indian time”: how the regulation of participation
affects the progression of events. In R. Bauman and J. Sherzer (eds), Explorations in
the Ethnography of Speaking. Cambridge: Cambridge University Press, 92-109.
Piaget, J. (1952) The Origins of Intelligence in Children. New York: Norton.
Piaget, J. (1970) Structuralism. New York: Basic Books.
Pike, K. (1967) Language in Relation to a Unified Theory of the Structure of Human
Behavior. The Hague: Mouton.
Polanyi, L. (1978) False starts can be true.
Polanyi, L. (1979) So what’s the point? Semiotica, 25: 207-41.
Polanyi, L. (1982) Linguistic and social constraints on storytelling. Journal of Prag¬
matics, 6: 509-24.
Polanyi, L. (1985) Conversational storytelling. In T. van Dijk (ed.). Handbook of
Discourse Analysis. Volume 3. Discourse and Dialogue. London: Academic Press,
183-201.
Polanyi, L. (1988) A formal model of the structure of discourse. Journal of Pragmatics,
12: 601-38.
Pomerantz, A. (1984) Agreeing and disagreeing with assessments: Some features of
preferred/dispreferred turn shapes. In J. Atkinson and J. Heritage (eds), Structures of
Social Action: Studies in Conversation Analysis. Cambridge: University Press, 57-101.
Pope, E. (1976) Questions and Answers in English. The Hague: Mouton.
Posner, R. (1980) Semantics and pragmatics of sentence connectives in natural language.
In F. Kiefer and J. Searle (eds), Pragmatics and Speech Act Theory. Dordrecht:
D. Reidel and Co., 87-122.
Potter, J. and Wetherell, M. (1987) Discourse and Social Psychology: Beyond Attitudes
and Behavior. London: Sage Publications.
Poyatos, F. (1983) New Perspectives in Nonverbal Communication. London: Pergamon
Press.
Preston, D. (1982) ‘Ritin fowklower daun ‘rong: folklorists’ failures in phonology.
Journal of American Folklore, 95: 304-26.
Preston, D. (1985) The Li’l Abner syndrome: written representations of speech. Amer¬
ican Speech, 60(4): 328-36.
Prince, E. (1978) On the function of one this and this really weird one. Presented at the
LSA Summer Meeting, Urbana, IL.
Prince, E. (1981) Toward a taxonomy of given-new information. In P. Cole (ed.),
Radical Pragmatics. New York: Academic Press, 223-56.
Prince, E. (1984) Topicalization and left-dislocation: a functional analysis. Annals of
the New York Academy of Sciences.
 Bibliography 453

Prince, E. (1985) Fancy syntax and “shared knowledge”. Journal of Pragmatics,

9: 65-81.
Prince, E. (1988) Discourse analysis: a part of the study of linguistic competence. In
Linguistics: the Cambridge Survey. Cambridge: Cambridge University Press, 164-82.
Pufahl, I. (1988) How to assign work in an office: a comparison of spoken and written
directives in American English. Journal of Pragmatics, 10: 673-92.
Quirk, R., Greenbaum, S., Leech, G. and Svartvik, J. (1972) A Grammar of Contem¬
porary English. London: Longman.
Rains, C. (1992) “You die for life”: on the use of poetic devices in argumentation.
Language in Society, 21(2): 253-76.
Rando, E. and Napoli, D. J. (1978) Definites and there sentences. Language, 54(2):
300-13.
Recanati, F. (1987) Meaning and Force: the Pragmatics of Performative Utterances.
Cambridge: Cambridge University Press.
Reddy, M. (1979) The conduit metaphor. In A. Orfony (eel.), Metaphor and Thought.
Cambridge: Cambridge University Press, 284-324.
Reed, S. (1972) Pattern recognition and categorization. Cognitive Psychology, 3: 382-407.
Reichenbach, H. (1947) Elements of Symbolic Logic. London: Macmillan.
Reichman, R. (1985) How to Get Computers to Talk Like You and Me. Cambridge,
MA: MIT Press.
Rickford, J. (1987) The haves and have nots: Sociolinguistic surveys and the assessment
of speaker competence. Language in Society, 16(2): 149-78.
Robert, H. (1992) Telephone Conversations. Bloomington: University of Indiana Press.
Romaine, S. (1981) On the problem of syntactic variation: a reply to B. Lavandera and
W. Labov. Working Papers in Sociolinguistics, 82. Austin, TX: Southwest Educa¬
tional Development Laboratory.
Rosch, E. (1973) Natural categories. Cognitive Psychology, 4: 328-50.
Rosch, E. (1978) Principles of categorization. In E. Rosch and B. Lloyd (eds), Cognition
and Categorization. Hillsdale, NJ: Laurence Erlbaum, 27—48.
Rosch, E., Mervis, C., Gray, W., Johnson, D. and Boyes-Braem, P. (1976) Basic objects
in natural categories. Cognitive Psychology, 8: 382-449.
Sacks, H. (1971) Lecture notes. School of Social Science, University of California at
Irvine.
Sacks, H. (1972a) Lecture notes. School of Social Science, University of California at
Irvine.
Sacks, H. (1972b) On the analyzability of stories by children. In J. Gumperz and
D. Hymes (eds), Directions in Sociolinguistics. New York: Holt, Rinehart and
Winston, 325-45.
Sacks, H. (1984) Notes on methodology. In J. M. Atkinson and J. Heritage (eds),
Structures of Social Action. Cambridge: Cambridge University Press, 21-7.
Sacks, H. and Schegloff, E. (1979) Two preferences in the organization of reference to
persons in conversation and their interaction. In Psathas (ed.), Everyday Language:
Studies in Ethnomethodology. New York: Irvington, 15-21.
Sacks, H., Schegloff, E. and Jefferson, G. (1974) A simplest systematics for the organ¬
ization of turn-taking in conversation. Language, 50: 696-735.
Sadock, J. (1985) Whither radical pragmatics? In D. Schiffrin (ed.), Meaning, Form,
and Use in Context: Linguistic Applications (Georgetown University Round
Table on Languages and Linguistics). Washington, DC: Georgetown University
Press, 139-49.
454 Bibliography

Sankoff, D. (1988) Sociolinguistics and syntactic variation. In F. Newmeyer (ed.),

Linguistics: The Cambridge Survey. Cambridge: Cambridge University Press,
140-61.
Sankoff, D. and P. Thibault. (1981) Weak complementarity: tense and aspect in Mon¬
treal French. In B. Johns and D. Strong (eds), Syntactic Change (Natural Language
Studies, 26). Ann Arbor: University of Michigan.
Sapir, E. (1933) Communication. Encyclopedia of the Social Sciences, 4: 78-81.
Saussure, F. de (1959) A Course in General Linguistics. New York: Philosophical
Library (originally published 1916).
Saville-Troike, M. (1982) The Ethnography of Communication. Oxford: Blackwell.
Schegloff, E. (1972a) Sequencing in conversational openings. In J. Gumperz and
D. Hymes (eds), Directions in Sociolinguistics. New York: Holt, Rinehart and
Winston, 346-80.
Schegloff, E. (1972b) Notes on a conversational practice: Formulating place. In D. Sundow
(ed.), Studies in Social Interaction. New York: Free Press, 75-119.
Schegloff, E. (1979a) The relevance of repair to syntax-for conversation. In T. Givon
(ed.), Syntax and Semantics, 12: Discourse and Syntax. New York: Academic Press,
261-88.
Schegloff, E. (1979b) Identification and recognition in telephone conversation openings.
In Psathas (ed.), Everyday Language: Studies in Ethnomethodology. New York:
Irvington, 23-78.
Schegloff, E. (1980) Preliminaries to preliminaries. In D. Zimmerman and C. West (eds),
Language and Social Interaction. Special issue of Sociological Inquiry, 50 (3/4):
104-52.
Schegloff, E. (1981) Discourse as an interactional achievement: Some uses of “uh huh”
and other things that come between sentences. In D. Tannen (ed.), Analyzing Dis¬
course: Text and Talk. Washington, DC: Georgetown University Press, 71-93.
Schegloff, E. (1984) On some questions and ambiguities in conversation. In J. Atkinson
and J. Heritage (eds), Structures of Social Action: Studies in Conversation Analysis.
Cambridge: Cambridge University Press, 28-52.
Schegloff, E. (1987a) Between micro and macro: contexts and other connections. In
J. C. Alexander et al. (eds), The Micro-Macro Link. Berkeley: University of Califor¬
nia Press.
Schegloff, E. (1987b) Confirming allusions. Talk presented to Linguistics Society of
America Summer Institute, Stanford University.
Schegloff, E. (1988) Presequences and indirection: applying speech act theory to ordin¬
ary conversation. Journal of Pragmatics, 12: 55-62.
Schegloff, E. (1992) The last site for intersubjectivity. American Sociological Review.
Schegloff, E., Jefferson, G. and Sacks, H. (1977) The preference for self-correction in
the organization of repair in conversation. Language, 53: 361-82.
Schegloff, E. and Sacks, H. (1973) Opening up closings. Semiotica, 7 (3/4): 289-327.
Schenkein, J. (ed.) (1978) Studies in the Organization of Conversational Interaction.
New York: Academic Press.
Scherer, K. and Giles, H. (eds) (1979) Social Markers in Speech. Cambridge: Cambridge
University Press.
Schieffelin, B. (1990) How Kaluli Children Learn What to Say, What to Do and How
to Feel. Cambridge: Cambridge University Press.
Schiffrin, D. (1977) Opening encounters. American Sociological Review, 42(4): 671-91.
Schiffrin, D. (1981a) Tense variation in narrative. Language, 57(1): 45-62.
 Bibliography 455

Schiffrin, D. (1981b) Handwork as ceremony: the case of the handshake. In A. Kendon

(ed.), Nonverbal Communication, Interaction, and Gesture. The Hague: Mouton,
237-50.
Schiffrin, D. (1984a) Jewish argument as sociability. Language in Society, 13(3): 311-35.
Schiffrin, D. (1984b) How a story says what it means and does. Text, 4(4): 313-46.
Schiffrin, D. (1985a) Multiple constraints on discourse options: a quantitative analysis
of causal sequences. Discourse Processes, 8(3): 281-303.
Schiffrin, D. (1985b) Conversational coherence: the role of “well”. Language, 61(3):
640-67.
Schiffrin, D. (1986a) The functions of and in discourse. Journal of Pragmatics, 10(1):
41-66.
Schiffrin, D. (1986b) Turn-initial variation: structure and function in conversation. In
D. Sankoff (ed.), Diversity and Diachrony. Philadelphia: John Benjamins Press,
367-80. h
■ Schiffrin, D. (1987a) Discourse Markers. Cambridge: Cambridge University Press.
Schiffrin, D. (1987b) Toward an empirical base in pragmatics. Review article of Stephen
Levinson, Pragmatics. Language in Society, 16(3): 381-95.
Schiffrin, D. (1988a) Conversation analysis. In F. Newmeyer (ed.), Linguistics: the
Cambridge Survey. Cambridge: Cambridge University Press, 251-76.
Schiffrin, D. (1988b) Sociolinguistic approaches to discourse: Topic and reference in
narrative. In K. Ferrera et al. (eds), Linguistic Contact and Variation. Austin: Univer¬
sity of Texas Press, 1-28.
Schiffrin, D. (1990a) The management of a cooperative self in argument: the role of
opinions and stories. In A. Grimshaw (ed.), Conflict Talk. Cambridge: Cambridge
University Press, 241-59.
Schiffrin, D. (1990b) Between text and context: deixis, anaphora and the meaning of
then. Text, 10(3): 245-70.
Schiffrin, D. (1990c) The principle of intersubjectivity in conversation and com¬
munication. Review article of Talbot Taylor and Deborah Cameron Conversation
analysis (1988), and Lauri Carlson, “Well” in dialogue games (1987). Semiotica,
80: 121-51.
Schiffrin, D. (1990d) Conversational analysis. Annual Review of Applied Linguistics,
11: 3-19.
Schiffrin, D. (1992a) Gender displays among family, friends, and neighbors. In K. Hall,
M. Bucholtz, and B. Moonwomon (eds), Locating Power, Proceedings of the Second
Berkeley Women and Language Conference, University of California, Berkeley,
515-27.
Schiffrin, D. (1992b) Anaphoric then: aspectual, textual and epistemic meaning. Lin¬
guistics, 30(4): 753-92.
Schiffrin, D. (1992c) Conditionals as topics in discourse. Linguistics, 30(1): 165-97.
Schiffrin, D. (1993) Research talk on, for, and with subjects. Peer commentary on
Cameron (et al.) Ethics, advocacy and empowerment: Issues of method in researching
language. Language and Communication, 13(2): 133-6.
Schiffrin, D. (forthcoming). The transformation of experience, identity, and context. In
J. Baugh, G. Guy and D. Schiffrin (eds), Towards a Social Science of Language.
Scott, M. and Lyman, S. (1968) Accounts. American Sociological Review, 33: 46-62.
Searle, J. (1962) What is a speech act? In M. Black (ed.), Philosophy in America. Ithaca,
NY: Cornell University Press, 221-39.
Searle, J. (1969) Speech Acts. Cambridge: Cambridge University Press.
456 Bibliography

Searle, J. (1975) Indirect speech acts. In P. Cole and J. Morgan (eds), Syntax and
Semantics. Volume 3: Speech Acts. New York: Academic Press, 59-82.
Searle, J. (1979) A taxonomy of illocutionary acts. In Expression and Meaning. Cam¬
bridge: Cambridge University Press.
Searle, J. (1989). On conversation. In J. Searle (ed.), (On) Searle on Conversation.
Amsterdam: John Benjamins Press.
Segal, E., Duchan, J. and Scott, P. (1991) The role of interclausal connectives in
narrative structuring: evidence from adults’ interpretations of simple stories. Dis¬
course Processes, 12(1): 27-54.
Sells, P. (1985) Lectures on Contemporary Syntactic Theories. Stanford, CA: Center for
the Study of Language and Information.
Selting, M. (1992) Prosody in conversational questions. Journal of Pragmatics, 17(4):
315-45.
Sheldon, A. (1990) Pickle fights: gendered talk in preschool disputes. In D. Tannen
(ed.), Discourse Processes: Special Issue on Gender and Conversational Interaction,
13(1): 5-32.
Sherzer, J. (1983) Kuna Ways of Speaking. Austin: University of Texas Press.
Sidner, C. (1986) Focusing in the comprehension of definite anaphora. In B. Grosz,
K. Jones and B. Webber (eds), Readings in Natural Language Processing. Los Altos,
CA: Morgan Kaufman Publishers, 363-94.
Sifianou, M. (1989) On the telephone again! Differences in telephone behavior: England
versus Greece. Language in Society, 18: 527—44.
Silva-Corvalan, C. (1983) Tense and aspect in oral Spanish narrative. Language,
59: 760-80.
Simmel, G. (1911) The sociology of sociability. American Journal of Sociology, 55: 3.
(Reprinted in 1950 in K. Wolff (ed.) The Sociology of Georg Simmel. New York:
Free Press, 40-57.)
Sinclair, J. and Coulthard, M. (1975) Towards an Analysis of Discourse: the English
Used by Teachers and Pupils. London: Oxford University Press.
Slobin, D. (1991) Learning to think for speaking: native language, cognition, and
rhetorical style. Pragmatics, 1(1): 7-26.
Smith, N. (1982) Mutual Knowledge. London: Academic Press.
Smith, W. (1977) The Behavior of Communicating. Cambridge, MA: Harvard Univer¬
sity Press.
Sperber, D. and Wilson, D. (1986) Relevance. Cambridge, MA: Harvard University Press.
Stenstrom, A. (1984) Questions and Responses in English Conversation. Stockholm:
CWK Gleerup.
Strawson, P. (1964) Intention and convention in speech acts. Philosophical Review,
73: 439-60.
Stubbs, M. (1983) Discourse Analysis. Chicago: University of Chicago Press.
Tannen, D. (1979) What’s in a frame? Surface evidence for underlying expectations. In
R. Freedle (ed.), New Directions in Discourse Processing. Norwood, NJ: Ablex
Publishing, 137-81.
Tannen, D. (1981) Review of Labov and Fanshel, Therapeutic discourse in Language.
Tannen, D. (1984) Conversational Style. Norwood, NJ: Ablex Press.
Tannen, D. (1989a) Talking Voices: Repetition, Dialogue, and Imagery in Conversa¬
tional Discourse. Cambridge: Cambridge University Press.
Tannen, D. (1989b) Interpreting interruption in conversation. Papers from the 25th
annual regional meeting of the Chicago Linguistics Society, Chicago, 266-87.
 Bibliography 457

Tannen, D. (1990) You Just Don’t Understand! New York: William Morrow.
Tannen, D. (ed.) (forthcoming) Linguistic Framing in Conversation. Oxford: Oxford
University Press.
Tannen, D. and Saville Troike, M. (eds) (1985) Perspectives on Silence. Norwood,
NH: Ablex Publishing.
Tannen, D. and Wallat, C. (1986) Medical professionals and parents: a linguistic
analysis of communication across contexts. Language in Society, 15(3): 295-312.
Taylor, T. (1992) Mutual Misunderstanding. North Carolina: Duke University Press.
Taylor, T. and Cameron, D. (1987) Analyzing Conversation. New York: Pergamon
Press.
Terasaki, A. (1976) Pre-announcement sequences in conversation. Social Science Work¬
ing Paper 99, School of Social Science, University of California, Irvine.
Thompson, S. and Mann, W. (1987) Rhetorical structure theory: a framework for the
analysis of texts. Papers in Pragmatics 1(1): 79-J05. h
Tomlin, R. (1985) Foreground-background information and the syntax of subordina¬
tion. Text, 5 (1/2): 85-122.
Tomlin, R. (1987) Linguistic reflections of cognitive events. In R. Tomlin (ed.),
Coherence and Grounding in Discourse. Philadelphia: John Benjamins Press, 455-79.
Trudgill, P. (1974) The Social Differentiation of English in Norwich. Cambridge: Cam¬
bridge University Press.
Tsui, A. (1989) Beyond the adjacency pair. Language in Society, 18(4): 545-64.
Turner, R. (ed.) (1974) Ethnomethodology. Harmondsworth: Penguin.
van Dijk, T. (1972) Some Aspects of Text Grammars. The Hague: Mouton.
van Dijk, T. (1977) Text and Context. London: Longman.
van Dijk, T. (1985) Introduction: discourse as a new cross-discipline. In T. van Dijk
(ed.), Handbook of Discourse Analysis, Volume 1: Disciplines of Discourse. New
York: Academic Press, 1-10.
Varenne, H. (1984) The interpretation of pronominal paradigms: speech situation,
pragmatic meaning, and cultural structure. Semiotica, 50 (3/4): 221-48.
Vendler, Z. (1967) Linguistics in Philosophy. Ithaca, NY: Cornell University Press.
Vincent, D. (1983) Les ponctuants du langage. Dissertation, Universite de Montreal.
Virtanen, T. (1992) Issues of text typology: narrative - a “basic” type of text? Text,
12(2): 293-310.
Vygotsky, L. (1962) Thought and Language. Cambridge, MA: MIT Press.
Ward, G. (1990) The discourse functions of VP preposing. Language, 66: 742-63.
Ward, G. and Birner, B. (1993) There-sentences and information status. Presented at
LSA Annual Meetings, Los Angeles, CA.
Ward, G. and Hirschberg, J. (1991) A pragmatic account of tautological utterances.
Journal of Pragmatics, 15: 207-20.
Ward, G. Sproat, R. and McKoon, G. (1991) A pragmatic analysis of so-called ana¬
phoric islands. Language, 67: 3.
Watzlawick, P., Beavin, J. and Jackson, D. (1967) The Pragmatics of Human Communi¬
cation. New York: Norton.
Webber, B. (1979) A Formal Approach to Discourse Anaphora. New York and London:
Garland.
Webber, B. (1988) Tense as discourse anaphor. Computational Linguistics, 14(2):
61-73.
Wierzbicka, A. (1987) English Speech Act Verbs: a Semantic Dictionary. New York:
Academic Press.
458 Bibliography

Wierzbicka, A. (1987) “Radical semantics” vs. “radical pragmatics”. Language, 63:

95-H4.
Wolfson, N. (1976) Speech events and natural speech: some implications for sociolin-
guistic methodology. Language in Society, 5: 189-209.
Wolfson, N. (1979) The conversational historical present alternation. Language,
55: 168-82.
Wolfson, N. (1988) The Bulge: a theory of speech behavior and social distance. In J. Fine
(ed.), Second Language Discourse: a Textbook of Current Research. Norwood, NJ:
Ablex, 21-38.
Wong, M. (1990) Referential choice in spoken Cantonese discourse. PhD Dissertation,
Georgetown University.
Wootton, A. (1989) Remarks on the methodology of conversation analysis. In D. Roger
and P. Bull (eds), Conversation: an Interdisciplinary Perspective. Clevedon: Multi¬
lingual Matters, 238-58.
Wright, S. and Givon, T. (1987) The pragmatics of indefinite reference. Studies in
Language, 11(1): 15-31.
Zimmerman, D. (1988) On conversation: the conversation analytic perspective. In
Communication Yearbook 11. Beverly Hills, CA: Sage.
Ziv, Y. (1982) Another look at definites in existentials. Journal of Linguistics, 18:
73-88.
Index

a *

a priori grammar see formalist paradigm pragmatics 220-1, 222, 225-6

abstracts 284, 285 variation analysis 296
acceptance see compliance anaphora 28
accounts animators
interactional sociolinguistics 108, identity displays 119, 123, 125
113-15 speaking for another 108
speech act theory 80-4, 89 answers
act sequences 147 speech act theory 77
reference interviews 148, 155, 160 stories as 34, 35-8
sociolinguistic interviews 160, 162, see also question-answer pairs;
176-80 summons-answer pairs
actions anthropology
conversation analysis: ethnography of communication 8,
ethnomethodology 233-4; text and 137-41
context 376-7 interactional sociolinguistics 7, 97,
interactional sociolinguistics 113-15 98-102
language use 32 applause 430
see also speech act theory attributive referring terms 230n
active predicates 302-4 Austin, John 6, 49, 50-4
additions 79 indirect acts 60
adjacency pairs 16 intonation 94n
conversation analysis 236-7, 242-52, sample analysis 61
271-2: structure and function 340; taxonomies of acts 57
text and context 377 authors 104
structuralism 27 autonomy 22-3
see also question-answer pairs; avoidance rituals 103
summons-answer pairs
adverbial there 275-6, 277 background mentions
affricates 35 conversation analysis 270-1
alignments 109-13 pragmatics 208-9
micro-and macro-identity displays behavior 138-9
115-27 benefits
ambiguity directives and commissives 95n
460 Index

benefits (cont’d) ethnography of communication 8,

requests and offers 75, 83 140, 408
Bloomfield, L. 26 interactional sociolinguistics 101, 131
breadth-before-depth lists 327-8 communicative function 43n
bridging 317 comparators 11
butting in 109-13 comparisons, discourse analysis
identity displays 119 approaches 15-19
complaints 94n
categories compliance
functionalist paradigm 22 pragmatics 217, 219
variation analysis 293-6, 307: speech act theory 83, 89, 96n
constraints 320, 322-6 complicating action clauses 284, 285
children, acquisition of language 22 compliments 94
chipping in 109-13 comprehensive discourse analysis 11,
Chomsky, N. 286, 355
competence, explanation of 8, 140 conative functions 351
sentences, definition 28 language use 33
circumstances, speech act theory 50-1 constatives 50, 51-3
clarification constitutive rules
ethnography of communication ethnography of communication 144
183-4: reference interviews 154-9; speech act theory 55, 56, 57:
sociolinguistic interviews 168, 172-6 questions 67; text and context
interactional sociolinguistics 136n 366, 372
closed questions 65 context
codas 284 conversation analysis 10, 235-6
code model 391-3, 396 functionalism 34
code switching 98 interactional sociolinguistics 7-8,
cognition 99 105-6, 134, 181-2
cognitive function 43n speech act theory 6-7: intonation 94n
cognitive linguistics 229n utterances, discourse as 40
coherence 417 variation analysis 11, 285-6, 289
pragmatics 353 contextual conditions 55-6
speech act theory 63, 89, 94n, 407: constatives 52
text and context 366 performatives 50
structuralism 25 contextual presuppositions 100, 103-4
variation analysis 355 speaking for another 113
collaboration see solidarity contextualization cues
collective referents ethnography of communication 371:
conversation analysis 245 culture 144
pragmatics 211-12,213-15 interactional sociolinguistics 99-101,
commissives 57, 58 106, 370: communication, nature
offers 72-3, 76 of 401-3; footing 104;
commitment 89 interpretive frames 103; sequential
commissives 73 coherence 130; speaking for
promises and offers 74 another 113, 133—4
communication, nature of 18, 42, contiguous utterances 425
386-405, 418-19 conventional meaning 56-7
communicative activities 100 conversation analysis (CA) 232-9,
communicative competence 273-9, 409, 413
 Index 461
communication, nature of 388: digressions 221
interactional model 401, 404-5 directives 57, 58
“core” example 9-10 indirect acts 59-60
exercises 279-80 offers 72-3, 75
focus 12 requests 70
sample analysis 239-73 displacement, textual 211
stories 96n disputes 116
structure and function 340-1, 354: diversity 143
and ethnography of doubt see uncertainty
communication 356-7
text and context 376-8, 382-3 echo questions 175
cooperative principle (CP) 9, 193-7, ellipses
226-7 conversation analysis 243
inferential model 396 structiyalism &S
referring terms 198 variation analysis 297-8
structure and function 352-3 embedded evaluations 11, 305, 308
text and context 367-8 embedded question-answer pairs 248-51
correctness 88 emergent paradigm see functionalist
criticality of rules 56 paradigm
culture emic schemas
ethnography of communication 8-9, ethnography of communication 141
137-41, 181-2: diversity 143-4; functionalism 32, 33
interactional model 403-4 emotive functions 351
interactional sociolinguistics 7, 97: language use 33
anthropology 98-9, 100-2, 106 encounters 27
speech act theory 95n encouragement 100
current-utterances 128-9 ends 146
reference interviews 154-5, 157, 159,
data collection and analysis 420-1 160
discourse analysis approaches 13 sociolinguistic interviews 160-3,
sample analyses 14 176-80
deceptions 83-4, 96n entities 297-9
declarative sentences epistemic meaning 151
as questions 30 equivalence 342-4
speech act theory 57: directives 70; essential rules 56
questions 65 indirect acts 60
defensiveness 121 offers and promises 73
definite referring terms questions 64, 66, 67, 68-9
conversation analysis 240 requests 71
pragmatics 198-9, 230n: sample taxonomies of acts 58
analysis 206-7, 208-9 ethnography of communication
variation analysis 317-19 137-44, 185-6, 408, 412
demonstrative there 275, 277 communication, nature of 388:
depth-before-breadth lists 326-7 interactional model 401, 403-4
descriptive structure 301-6 “core” example 8-9
designata 191 exercises 186-7
development, discourse analysis focus 12
approaches 14 sample analysis 144-85
dialogic pairs see adjacency pairs structure and function 347-50, 356-7
462 Index

text and context 370-1, 382: and pragmatics 202: sample analysis
variation analysis 375 207-9, 213-14, 216-17, 219-20,
ethnomethodology 9-10, 232, 233-6 222-5
etic schemas variation analysis 317-20: structure
ethnography of communication 141-2 and function 344-5
functionalism 32-3 flexibility 91
evaluative structure 11,306-14 footing 104, 124
exhaustiveness 78 form
existential predicates 36 and function 85-7
existential there construction and meaning 103
conversation analysis 239—40, 274, formalist paradigm 20-3, 41-2
276-8: see also “there + BE + language across the sentence 23-31
ITEM” constructions social interaction, language as 415-17
variation analysis 304-6 formulation, queries 154-9
expansions frames 103-4, 136n
interactional sociolinguistics 108 fricatives 35
speech act theory 78-84 function
explanations and form 85-7
of competence 8 and structure 18, 42, 339-61,
speech act theory 77 418-19
explicit performatives 52-3 functionalist paradigm 20-3, 41-2
explicit referring terms ethnography of communication 138,
conversation analysis 240 142-3
pragmatics 198-9: sample analysis language use 31-9
206-7, 212, 215, 224-5 social interaction, language as 415-17
variation analysis 317-18
expressibility, principle of 54, 90 Garfinkel, Harold 9, 232, 233
expression gaze direction 429-30
individual 98 gender 115-27
interactional sociolinguistics 105 generative grammar 28
expressives 57, 58 genre 147
reference interviews 148
face notion 102-3, 130 sociolinguistic interviews 160, 163
Fairclough, N. 31 Givon, T. 197
falsity 51-2, 53 Goffman, Erving
familiarity 114 interactional sociolinguistics 7, 97,
family conversations 420 102-6, 133-4
Fasold, R. 40 systems and ritual conditions 32
felicity 64 grammar
performatives 51-2 conversation analysis 274
questions 69 ethnography of communication 140
requests 71 generative 28
text and context 380 interactional sociolinguistics 99
Figueroa, E. 40-1 greetings 400
figures 104 Grice, H. P.
first-mentions inferential model 393
conversation analysis 240-52, 258-9, pragmatics 9
265-73 Gricean pragmatics 190-203, 226-7,
language use 36, 43n 408-9, 412-13
 Index 463

communication, nature of 388: conversation analysis 240

inferential model 396-7 pragmatics 198-9, 230n: sample
and conversation analysis 241 analysis 206-7, 208-9, 212
exercises 228-9 variation analysis 317-18
sample analysis 203-26 indirect complaints 94n
structure and function 352-3, 359 indirect questions 30
text and context 367-9, 379-81 indirect speech acts
guesses 114 “core” example 6, 7
Gumperz, John 7, 97, 98-104, 105-6, speech act theory 59-60, 75, 90
133-4 individual expression 98
inexplicit referring terms
Harris, Zellig 18, 24 conversation analysis 240
historical present tense 87
pragmatics 198-9: sample analysis
accounts 81, 84
206-7, 212*213, 215
history, discourse analysis 19n variation analysis 317-18
Horn, L. 23In
infelicity 51
Hymes, Dell 8, 137-43 inferences
pragmatics 9
identifiability of entities 199
situated: ethnography of
identification
communication 371; interactional
of speech act sequences 76-84
sociolinguistics 100-1, 103, 104,
of utterances as speech acts 63-76
113, 370
identity, interactional sociolinguistics
speech act theory 59
115-27
variation analysis 316-17
ill-formedness 94n
inferential model 393-7
structuralism 26-7
information-checking questions 94n
illocutionary acts
ethnography of communication
interactional sociolinguistics 100
183-4: reference interviews 157;
speech act theory: Austin 51, 53-4;
sociolinguistic interviews 169-73
commissives 73; multifunctionality
information flow perspective 23In
and indirect acts 59; Searle 55-7;
information-seeking questions
sequences 76-7; taxonomies of
ethnography of communication
acts 57-8
183-4: sociolinguistic interviews
illocutionary force
165-9, 173
interactional sociolinguistics 100
speech act theory 68-72
speech act theory 66-7, 68, 90:
inside of lists 321-2
inferential model 397; intonation
institutional encounters 420
66-7, 68; offers 72; structure and
function 347; text and context instrumentalities 147
379-80 insults 53
intentional information 105
imperative sentences
directives 70 intentions, speaker
as questions 30 interactional sociolinguistics 132
implications 52 pragmatics 191-2
implicatures speech act theory 56-7
cooperative principle 193-6 interactional model 397-405
inferential model 396 interactional sociolinguistics 98-106,
text and context 367-8, 380-1 133-4, 407, 412
incompleteness 66 communication, nature of 388:
indefinite referring terms interactional model 401-3
464 Index
conversation analysis comparison 10 Jakobson, R. 32-3
“core” example 7-8 Jefferson, Gail 10, 232
exercises 134-6
focus 12 keys 147
sample analysis 106-33 reference interviews 148
structure and function 350-2, 357-9 “knowability” 82
text and context 369-70, 381-2: and knowledge
variation analysis 374-5 context as 365-73, 376-8
interactive paradigm see functionalist conversation analysis 233
paradigm ethnography of communication 138-9
interests of speakers and hearers 58
interpretants 191 Labov, William 10-11,282-91
interpretations 7-8 language use 31-9
interpreters 191 lead-in-phase 179-80
interpretive frame 103 learning of languages 22
interrogatives “let” statements 95n
ethnography of communication 150 Levinson, S. 231n
speech act theory 65, 69, 70 lexical-functional grammar 22
interruptions linguistics
conversation analysis 257-8, 266-7 ethnography of communication
ethnography of communication 188n 137-41
interactional sociolinguistics 109 interactional sociolinguistics 7, 97
pragmatics 219 speech act theory 54, 61
intersubjectivity 387, 389-90 variation analysis 10, 282, 283
code model 392 lists 291-317
inferential model 395 constraints 320-30
interactional model 401 locutionary acts 51, 53
intervals in speech 426 logic 193
interviews 145-7 Lyons, J. 27
reference 147-60
sociolinguistic: data from 14, 420; macro-identity displays 115-27
ethnography of communication manner maxim 194
160-80; variation analysis 289-90 many-for-one relationships 87
intimates 94n marginal features of language 99
intonation 412 material resources 103
conversation analysis 252-7 mathematical representations 22
ethnography of communication 149, meaning
157 ethnography of communication 151
interactional sociolinguistics 100 interactional sociolinguistics 7-8: and
language use 38 form 103
question identification 30 pragmatics 9, 191-3, 226-7
speech act theory 65-8, 69 situated 109-27
intonational closure 25 speech act theory 56-7, 88
invitations 73, 94n mediums of communication 389
involvement, social organization of code model 392
103 inferential model 395
issuance, queries 154-9 interactional model 401
items 240-1: see also “there + BE + messages 388-9
ITEM” constructions code model 392
 Index 465
inferential model 394-5 observer’s paradox 161
interactional model 400-1 offers
meta-messages 136n ethnography of communication
metalinguistic function 33 149-50, 152-9
metaphorical code switching 98 interactional sociolinguistics 113-14
metaphors 196 speech act theory 72-6, 86:
metonyms 208 sequences 76, 77, 82-3, 89
micro-identity displays 115-27 old information 36
misfires 51, 52 one-for-many relationships see
modularity 22-3 multifunctionality
morphemes 24 only-mentions 269-71
Morris, C. 191 open questions 65
motivation 132 orientation clauses 284
multifunctionality origins, discourse analysis approaches
ethnography of communication: 13
reference interviews 149-50; outside of lists 321-2
structure and function 348-9 over-initiated repairs 121
speech act theory 6-7, 59-60, 63, 75, overlaps
85-7: “let” statements 95n; offers interactional sociolinguistics 109
72; sequential completion 80; transcription notation 424-5, 432
structure and function 347
multiparty interviews 168, 176-7 paradigmatic structure 342
parataxis 28
narrative clauses 11 parents, compliance with 83, 96n
narrative ordering 37 participants 387-8
variation analysis 345 code model 391-2
narratives 11, 283-7 ethnography of communication 147:
and lists 291-2, 293, 297, 299-315 reference interviews 159-60;
see also stories sociolinguistics interviews 160-1,
natural meaning 191 176-80: structure and function 349
negative statements 83 inferential model 393-5
new information 36 interactional model 398-401
Newmeyer, F. 22 observation 140, 141
next-mentions participation framework 104
conversation analysis 237, 240-52, contextualization cues 106
258-9, 265-73 identity displays 116, 118-27
interactional sociolinguistics 128-31 speaking for another 108, 112-13,
language use 36, 43n 132
pragmatics 202: sample analysis patterns in texts 11
207-8, 210-25 perceptual function 43n
variation analysis 317-20: structure performatives 50-4
and function 345 directives 70
nominal anaphora 28 indirect acts 60
non-compliance see rejections perlocutionary acts
non-natural meaning (meaning-nn) 191-2 Austin 51, 54
non-specific referring terms 230n Searle 56
norms 147 personal knowledge 136n
reference interviews 155 phatic function 33
sociolinguistic interviews 160 phenomenology 9
466 Index

philosophy principals 104

pragmatics 9 identity displays 119, 121, 122, 123
speech act theory 61 speaking for another 108, 131
phonemics 32 principle of expressibility 54, 90
phonetics 32 promises 56, 73-4, 75-6
poetic function 33 felicity conditions 64
politeness prompts 120-5
ethnography of communication 189n pronomial there 276-7
interactional sociolinguistics: identity pronouns
displays 117, 119; speaking for pragmatics 212, 213-15
another 109 variation analysis 325-6, 328-9:
speech act theory 77, 78 structure and function 355
positional knowledge 136n pronunciation 35
pragmatic goals 41 proper nouns 82
pragmatics 190-203, 223-7, 408-9, propositional acts 55
412-13 propositional content rules 56
communication, nature of 388: questions 64-7, 68, 76
inferential model 396-7 requests 71-2
and conversation analysis 241 taxonomies of acts 58
“core” example 9 propositions 24
exercises 228^9 prosody 65
focus 12 psychological state 58
sample analysis 203-23 punctuation conventions 25
structure and function 352-3, 359
text and context 367-9, 379-81 quality maxim 194
Prague school of linguistics 138 quantitative analyses 287-8, 317
pre-interview phase 178-9 quantity maxim 194, 199-201
pre-requests 95n sample analysis 208, 213, 214-15,
pre-sequences 248-9 217-18, 223-6
preferences 237, 252-3, 257-9 queries 149, 153-9
structure and function 354 question-answer pairs 411-12
preparatory rules conversation analysis 242-52
ethnography of communication structuralism 27, 29, 30
150-1, 154 questions
speech act theory 56: indirect acts ethnography of communication
59-60; offers and promises 73-4, 144-85
75; questions 64-6, 67, 68-9, 86; sample analyses 15-17, 18
taxonomies of acts 58 speech act theory 64-70, 86: indirect
present tense acts 59; and offers 74, 75-6; and
historical 87: accounts 81, 84 requests 70, 71-2; sequences 76,
speech act theory: accounts 81; 77, 89
performatives 50, 53 structuralism 29-30
presentational rituals 103
speaking for another 109 recency 325
preterite tense 84, 87 recipients 108
primary acts reference interviews 147-60
offers 76 references 28
speech act theory 59-60 referential function 33
primary performatives 52-3 referents 36
 Index 467
referring terms interactional sociolinguistics 358
conversation analysis 240-52, speech act theory 54, 55-6, 358
265-73: stepwise topic transitions
261-5 Sacks, Harvey 10, 232, 234-5
pragmatics 197-203: sample analysis next-position 237
203-26; structure and function 352 stepwise transitions 261-2
sample analyses 15-17, 18-19 sample analyses 14
variation analysis 292, 316-30: comparisons 15-17
structure and function 342-5; text Sapir, Edward
and context 375 ethnography of communication 138
reformulation, queries 156-9 functionalist paradigm 22
refusals see rejections satisfaction, queries 154-9
regulative rules 55 Schegloff, Emmanuel 10, 232
rejections Schutz, Alfred $
■interactional sociolinguistics 112, Searle, John 6, 49-50, 54-7, 61, 90
113-14 multiple functions and indirect acts
speech act theory 77-8, 89 59, 60
relation maxim 194, 199-201 offers 73, 75
sample analysis 208, 213 questions 64
relational grammar 22 requests 71
relevance 199-201 taxonomies of acts 57-8
sample analysis 208, 213, 223-6 secrets 96n
reliability 88 segmentation of texts 7
repairs 258-61 self
repetition 296 anthropology 101
replacements 79 sequential coherence 130
representatives 57, 58 sociology 102-3, 104
requests speaking for another 108
ethnography of communication 150 self-interruption 257-8, 266-7
pragmatics 217, 219 self-repairs
speech act theory 70-2, 86: context ethnography of communication 157
and text 372; and offers 73-5, speech act theory 79
76, 95n; sequences 76, 77, 89 self-selection
resolution, queries 154-9 conversation analysis 238
revelation of information 112 ethnography of communication 179
rhemes 277 Sells, P. 26-7
rhetoric semantic closure 25
ethnography of communication 150 semantic goals 41
history 19n semantics
ritual conditions 32 pragmatics 191: speaker meaning
roles 192
ethnography of communication 164, structuralism 43n
165, 167-8 text and context 362
interactional sociolinguistics 127-31 variation analysis 288
root meaning 151 semiosis 191
rules senders 108
constitutive 55, 56, 57: ethnography sentences
of communication 144; questions definitions 26, 28, 40
67; text and context 366, 372 language above 23-31
468 Index

and utterances 40-1 variation analysis 289-90

sequential coherence 129-30 sociolinguistic variables 35
sequential goals 41 sociology
serial format 162 conversation analysis 9-10, 233
setting 147 interactional sociolinguistics 7, 97,
side sequences 95n 102-5
sign vehicles 191 solidarity 112
simultaneous utterances 424 identity displays 117, 120, 125
sincerity rules speaking for another 106-33
ethnography of communication 154, structure and function 357
155 SPEAKING grid 141-2, 146
speech act theory 56: indirect acts context 371
59; questions 64, 67, 68, 69; specific referring terms 230n
requests 71, 75, 86; taxonomies of speech act theory 49-61, 89-91, 407,
acts 5 8 408-9
situated inferences communication, nature of 388:
ethnography of communication 371 inferential model 396-7
interactional sociolinguistics 100, “core” example 6-7
101, 103: alignment 104; context exercises 91-3
370; speaking for another 113 focus 12
situated information 399-400 functionalism 35
situated meaning 109-27 sample analysis 61-89
situation, context as 369-78 structure and function 346-7, 356, 358
situational code switching 98 text and context 365-7, 369, 371-3,
social actions 113-15 379-80
social distance 94n speech acts 142, 144-85
social function 43n speech delivery 426-8
social identities speech events 142, 144-85
conversation analysis 235-6 speech situations 142
directives and commissives 95n spokespersons
language use 35 conversation analysis 244
text and context 374 interactional sociolinguistics 108,
social interaction 110,131
language as 406-19 statements 59
speech act theory 85 stative predicates 302-4
social order 10, 232 status of speakers and hearers 58
social relationships stepwise format
interactional sociolinguistics 110-13 conversation analysis 261-5
speech act theory 69-70, 89 sociolinguistic interviews 162, 183
socialization 131 stops 35
society stories
interactional sociolinguistics 7, 97: beginnings, language use 34, 35-8
anthropology 98-9, 100-2; conversation analysis 96n
sociology 105 functions 87
language use 31 interactional sociolinguistics 116
sociolinguistic interviews pragmatics 204
data from 14, 420 speech act theory 80—4
ethnography of communication see also narratives
160-80 strangers 94n
 Index 4 69
strengths of presentations 58 Austin 50, 52-3
structural linguistics 99 Searle 55-6
structural paradigm see formalist paradigm thematicity 325
structure “there + BE + ITEM” constructions
conversation analysis 237 240-73, 278-9
and function 18, 42, 339-61, 418-19 structure and function 340-1
interactional sociolinguistics 7 threats 112, 114
Stubbs, Michael 18 topics
subcategories 295-6 conversation analysis 261-5
constraints 320, 322, 323-30 ethnography of communication 162
subject-verb inversion 29 interactional sociolinguistics 115-16
subjectivity 306-7 structuralism 27
summons-answer pairs 10, 236 transcription
sympathy 118 conventions 14-15, 422-33
ft* . Bf
syntactic closure 25 conversation analysis 235
syntactic goals 41 interactional sociolinguistics 106
syntagmatic structure 342 language “above the sentence” 25
syntax transition-relevance place 238
ethnography of communication 149 truth 51-2, 53
pragmatics 191 turns
speech act theory 70 conversation analysis 237-8, 252-61,
structuralism 28, 43 question 272: structure and function 340,
identification 30 354; text and context 377
system conditions 32 ethnography of communication 179
system-sentences 27-8 structuralism 27
typotaxis 28
Tannen, D. 136n
uncertainty
taped conversations 420
ethnography of communication 151-3
taxonomies of acts 57-9
intonation 38: speech act theory 66
telephone calls 10, 236
speech act theory 66, 68-9: offers
temporal anaphora 28
74, 76
temporal ordering 37
transcription 428-9
variation analysis 345
unintentional information 105
temporal structure 299-301
units
tense
speech act theory 77
historical present 87: accounts 81, 84
structuralism 24-5
present: accounts 81; performatives
variation analysis 283-7: lists 297-9
50, 53 universal, linguistic
preterite 84, 87 ethnography of communication 141
speech act theory 84, 87 formal and functionalist paradigms 22
text and context, relationship between uptake 50
18, 42,362-85, 418-19 utterance acts 55
conversation analysis 278 utterances
ethnography of communication 152 as definition of discourse 39-41
pragmatics 209 pragmatics 203
speech act theory 51-3, 55-6: speech act theory 61, 63-76, 85-8
indirect acts 60
text-sentences 27-8 vagueness 214, 225-6
textual conditions validity
470 Index

interactional sociolinguistics 118 wants

speech act theory 88 interactional sociolinguistics 113-14
variation analysis 282-91, 330-3, 409, speech act theory 68
413 well-formedness 94n
communication, nature of 388, 392-3 structuralism 26-7
“core” example 10-11 word order 28
exercises 33-4 words-to-world fits 58
focus 12 world-to-words fits 58, 73
sample analysis 291-30 written language
structure and function 341-5, 354-6 data collection 421
text and context 373-6, 382-3 sentences 25
vernacular speech 289-90
violations 196, 214-15, 217-18, 224-6 zero pronouns 212
 A ft

% 4

>'

-- ■
This book is a guide to the various frameworks, concepts, and metl
2425'

available for the analysis of discourse within linguistics. It compares six

dominant approaches to discourse analysis: speech act theory,
pragmatics, ethnomethodology, interactional sociolinguistics,
ethnography of communication, and variation theory. The author not
only considers each approach from several standpoints but she also
illustrates each approach through extensive applications to a variety
of concrete social and linguistic problems facing discourse analysts.
Exercises pose problems to which each approach can be applied.

Blackwell Textbooks in Linguistics

Introduction to Government and Binding Theory

Liliane Haegeman

Morphological Theory
Andrew Spencer

Language Acquisition
Helen Goodluck

Introduction to Sociolinguistics (Second Edition)

Ronald Wardhaugh

Children's Syntax
Martin Atkinson

Understanding Utterances
Diane Blakemore

Phonology in Generative Grammar

Michael Kenstowicz

Approaches to Discourse
Deborah Schiffrin

Deborah Schiffrin

The author is Associate Professor of Linguistics at Georgetown University.

She is the author of Discourse Markers (1987).