Figurative Language

Comprehension

Social and Cultural Influences

This page intentionally left blank
 Figurative Language
Comprehension
Social and Cultural Influences

Edited by

Herbert L. Colston
University of Wisconsin–Parkside
Albert N. Katz
University of Western Ontario

LEA
2005
LAWRENCE ERLBAUM ASSOCIATES, PUBLISHERS
Mahwah, New Jersey London
Copyright © 2005 by Lawrence Erlbaum Associates, Inc.
All rights reserved. No part of this book may be reproduced in any
form, by photostat, microform, retrieval system, or any other
means, without prior written permission of the publisher.

Lawrence Erlbaum Associates, Inc., Publishers

10 Industrial Avenue
Mahwah, New Jersey 07430

Cover design by Katherine Houghtaling Lacey

Cover artwork by Linda K. Williamson

Table Figures, 1994; charcoal on paper, 20 x 25 cm;
Private collection.

Library of Congress Cataloging-in-Publication Data

Figurative language comprehension : social and cultural influences /

edited by Herbert L. Colston, Albert N. Katz.
p. cm.
Includes bibliographical references and index.
ISBN 0-8058-4506-2 (cloth : alk. paper)
1. Sociolinguistics. 2. Figures of speech—Social aspects. 3. Psycho­
linguistics. I. Colston, Herbert L. II. Katz, Albert N.
P40.5.F54F54 2003
306.44—dc22 2003049526
CIP

Books published by Lawrence Erlbaum Associates are printed on acid-

free paper, and their bindings are chosen for strength and durability.

Printed in the United States of America

10 9 8 7 6 5 4 3 2 1
 To Brigitte, for her love, patience, support,
and willingness to entertain reading aloud,
and to Raymond W. Gibbs, Jr.,
whose mentoring and friendship
enabled my work
—Herbert L. Colston

To Al Paiuio, who, by demonstration,

showed me what it was to be a scholar
and to Bill and Mollie Katz who, by demonstration,
tried to show me how to be a mensch
—Albert N. Katz
This page intentionally left blank
 Contents

Preface ix
Albert N. Katz

1 On Sociocultural and Nonliteral: A Synopsis and a Prophesy 1

Herbert L. Colston

Part I: Sociocultural Knowledge Influences

2 Making Sense of How We Make Sense: The Paradox 21
of Egocentrism in Language Use
Dale ]. Barr and Boaz Keysar

3 Contextual Expressions and Common Ground 43

Richard J. Gerrig and William S. Horton

Part II: Sociocultural Phenomenological Influences

4 Context and the Comprehension of Nonliteral Meanings 73
Thomas Holtgraves

5 Social and Cultural Influences on Figurative and Indirect 99

Language
Herbert L. Colston

6 Irony as Persuasive Communication 131

Raymond W. Gibbs, Jr. and Christin D. Izett

vii
viii CONTENTS

7 Do Men and Women Differ in Their Use of Nonliteral 153

Language When They Talk About Emotions?
Kristen E. Link and Roger J. Kreuz

Part III: Sociocultural Processing Influences

8 Discourse and Sociocultural Factors in Understanding 183
Nonliteral Language
Albert N. Katz

9 Social Factors in the Interpretation of Verbal Irony: 2O9

The Roles of Speaker and Listener Characteristics
Penny M. Pexman

10 Explicit Negation as Positivity in Disguise 233

Rachel Giora, Noga Balaban, Ofer Fein, and Inbar Alkabets

Part IV: New Sociocultural Influences

1 1 Stereotype Processing and Nonliteral Language 261
Trade L. Blumentritt and Roberto R. Heredia

12 On Mosquitoes and Camels: Some Notes on the 283

Interpretation of Metaphorically Transparent Popular Sayings
Carmen Cureó

13 Metaphors in Sign Language and Sign Language Users: 3O9

A Window Into Relations of Language and Thought
Marc Marschark

Author Index 335

Subject Index 345

 Preface

How is it that when I open my mouth and I say something such as, "well, you
all know that I am the Barry Bonds of my department's hockey team" that I ex­
pect I will be understood? And why is it that I am not surprised that my com­
ment may be understood differently by people who do not know me (and
take my comment as a self-serving metaphoric bragging) from those who
know me well (who will, alas, understand the comment as self-deprecating
irony). And why is that when we utter a comment that we sometimes do so
with two audiences in mind—one in the "know" and one "in the dark"—so
that two different messages are being sent intentionally (see, e.g., Katz &
Lee, 1993) ? And why is it that when we read or listen to the satire of Lenny
Bruce we recognize the "truth" in his comment that the wisest of philoso­
phers talking in a deep Southern accent will be ignored, whereas a fool talk­
ing in a cultured new England accent will be honored; this is a "truth" based
on how our prejudices about people, social class, and culture influence our
reactions to what is being said. These examples, and many others, all point
to the centrality of social and cultural factors in the production and compre­
hension of language.
Despite their seeming centrality, the history of linguistics and psycho­
linguistics has evolved to marginalize, and even ignore, social and cultural
factors. For instance, in about the mid- 19th century August Schleicher (see
the nice review by Yngve, 1996) conceptualized language "as an object of
nature" (p. 26) and, thus, as an entity to be studied on it's own, independent
of human interaction. Later theorists took a similar line of argument.
Ferdinand de Saussure (1959), for instance, made his well-known distinc­
tion between langue (language) and parole (speech) and asserted that under­
standing the langue is the true role of the linguist. Speech and the human
element were relegated to a secondary role, best studied by other disciplines,
ix
X PREFACE

a point also made by Leonard Bloomfield (1926/1957) who explicitly ex­

pelled phenomena related to people from the study of linguistics.
It was, of course, with Noam Chomsky (1957) that the modern era in lin­
guistics began. But like the linguistics just cited, Chomsky was interested in
language as an abstract entity, autonomous from secondary human-related
performance factors. According to this view, language should be considered
an autonomous function, separate from other cognitive functions. And these
other cognitive functions would and should include the belief structures that
govern what we hold about people, social class, and culture. Thus, what I have
called central to our understanding of the production and comprehension of
language is, in the Chomskian perspective, secondary and not a part of main­
line language at all! In fact, the Chomskian perspective dismissed much more
than sociocultural factors and relegated any expression that did not meet cer­
tain criteria for grammaticality as worthy of mainline study. Thus, expressions
such as, "Been there, done that" or "You're no Jack Kennedy," are considered
by those in this tradition to be nongrammatical despite the fact that they are
often used and easily understood. (see Turner, 1998, for examples of how cog­
nitive linguists treat these expressions as indeed grammatical).
The distinction between natural and nonnormal language has gone be­
yond the study of what is and is not grammatical to what is basic and derived
in other language-related domains, especially to forms of language consid­
ered literal from that considered nonliteral. The nonliteral language here
includes metaphor, irony, indirect requests, proverbs, and all forms of lan­
guage in which what is being expressed differs from what the speaker intends
to convey. In many ways linguists, psycholinguists, and various specialists in
experimental psychology have made a concerted effort over the past 10 to
15 years to break the false dichotomies just described that have treated lan­
guage as an entity separate from human beliefs and knowledge systems and
have separated so-called literal from nonliteral forms of language.
An examination of the work of the past 10 to 15 years has shown that, for
instance, one cannot examine an autonomous syntactic structure (the
backbone of the Chomskian revolution) without examining the immediate
effects of pragmatics and semantic knowledge (see review by Katz, 1998).
One cannot relegate nonliteral language to a secondary and derived form of
communication (see Gibbs, 1994; Gluckberg, 2001). One cannot examine
language as a cognitive object of nature independent of the social interac­
tions of people in communication (Clark, 1996). In these ways the under­
standing of language that is emerging is of a complex interactive system that,
even at the earliest moments of production and comprehension, simulta­
neously evaluates and integrates knowledge of linguistic structure—that is,
how language is used and the sociocultural factors of relevance.
Despite the work of the past 10 to 15 years, there is still a lack of system­
atic examination of social and cultural factors in language processing, espe­
PREFACE xi

cially with so-called nonliteral language. In fact, one can argue that
nonliteral language, because it is so contextual and reflects speaker's inten­
tion above and beyond what is being expressed, is an ideal crucible for exam­
ining the effects of social and cultural knowledge. This book is intended as a
conduit, bringing together many of the strings that are promoting and sup­
porting the importance of the human agent in our understanding of lan­
guage processing. In this volume you will find the most recent research by a
range of scholars from linguistics, psycholinguists, and experimental psy­
chology, who, in their own ways, attempt to reintroduce the human element
into language production, comprehension, and use.

—Albert N. Katz

REFERENCES

Bloomfield, L. (1957). A set of postulates for the study of language. In M. Joos (Ed.),
Readings in linguistics (pp. 26–31). Washington, DC: American Council of
Learned Societies. (Original work published 1926)
Chomsky, N. (1957). Syntactic structures. The Hague, The Netherlands: Mouton.
Clark, H. (1996). Using language. Cambridge, England: Cambridge University
Press.
Gibbs, R. W., Jr. (1994). The poetics of mind. New York: Cambridge University Press.
Glucksberg, S. (2001). Understanding figurative language. New York: Oxford Univer­
sity Press.
Katz, A. (1998). Figurative language and figurative thought: A review. In A. Katz,
C. Cacciari, R. W. Gibbs, Jr., & M. Turner (Eds.), Figurative language and thought
(pp. 3–43). New York: Oxford University Press.
Katz, A., & Lee, C. (1993). The role of authorial intent in determining verbal irony
and metaphor. Metaphor and Symbolic Activity, 8, 257-279.
Saussure, F., de. (1959). Course in general linguistics. Translated by W. Baskin. New
York: The Philosophical Library (reprinted, New York: McGraw-Hill, 1966).
Turner, M. (1998). Figure. In A. Katz, C. Cacciari, R. W. Gibbs, Jr., & M. Turner
(Eds.), Figurative language and thought (pp. 44–87). New York: Oxford University
Press.
Yngve, V (1996). From grammar to science: New foundations for general linguistics.
Amsterdam: Benjamins.
This page intentionally left blank
 1

On Sociocultural
and Nonliteral: A Synopsis
and a Prophesy
Herbert L. Colston
University of Wisconsin—Parkside

The past several decades of psychological and linguistic empirical research

and theorizing on nonliteral language has branched into a few key direc­
tions. Much of this work has focused and continues to focus on the compre­
hension of forms of figurative, indirect, and other kinds of nonliteral
language. Some of this work concerns specific, step-by-step, online process­
ing; other work is directed at the products of the comprehension and inter­
pretation processes. A related line of work has attended to the sorts of
pragmatic accomplishments that these kinds of language perform for inter­
locutors. This work seeks explanations of what is done by forms of nonliteral
language and how these achievements are brought about. A growing inter­
est in the production of nonliteral language is also emerging, as is a concern
for authenticity. And the theoretical efforts to account for the accumulating
empirical findings continue in light of these developments as well as related
phenomena like situational irony.
One implicit idea emerging in much of this work is that of a growing ap­
preciation of the importance of social and cultural influences to the compre­
hension, processing, use, pragmatic accomplishments, and so on of
nonliteral language forms. Scholars have increasingly focused on variables
such as gender, status, cultural background, age, and others in their observa­
tions, experiments, examples, and so on used to gain an understanding of
nonliteral language cognition.
1
2 COLSTON

This book is offered as a first attempt to summarize and synthesize this ap­
preciation of sociocultural influences. The chapters were prepared by a wide
variety of talented scholars of nonliteral language, with a diversity of back­
grounds in cognitive psychology, psycholinguistics, linguistics, social psy­
chology, and other areas. Thus, there is no universal theoretical or
methodological underpinning or agenda to this work. Rather, the chapters
reflect the diversity of approaches and mechanisms in the study of nonliteral
language. But the fact that these different perspectives have all unearthed
the need to pay much greater attention to social and cultural influences un­
derscores the importance of these influences.
A number of themes can be distilled from the chapters presented here.
These range from (a) the kinds of mechanisms discussed for nonliteral lan­
guage comprehension to (b) the variety of types of nonliteral language
treated, (c) the twin ideas of what is a social and what is a cultural variable,
and (d) what the actual influences in interlocutors might be. A number of
theoretical backgrounds are also used in the chapters. These emerging
themes and theories served to help organize the structure of the book. In this
chapter I discuss each briefly and follow with a short discussion of the likely
direction research on nonliteral language cognition should take given this
heightened awareness of sociocultural influences. First, however, a very
brief comment on the organization of the book is in order.
Categorizing the chapters into coherent, separable sections according to
their content proved to be a very difficult task. Organization schemes based
on the kind of theoretical approach, type of nonliteral language, particular
form of sociocultural influence, and so on were not viable because many of
the chapters discuss multiple approaches, several nonliteral forms, various
influences, and so forth. Instead, divisions based on the primary topic or
strength of the chapters were created. The resulting four parts are Socio­
cultural Knowledge Influences, Sociocultural Phenomenological Influ­
ences, Sociocultural Processing Influences, and New Sociocultural
Influences. Please note, however, that the content of a given chapter may
supercede the section in which it was placed.

MECHANISMS IN NONLITERAL LANGUAGE

COMPREHENSION: COMMON GROUND, SUPPRESSION,
NEGATION, SOCIAL PSYCHOLOGICAL PRINCIPLES,
CONTRAST, SCHEMAS, AND PRIMING

The role of common ground in the processing and comprehension of the

meaning of instances of nonliteral language was taken up in two of the chap-
ters—those by Gerrig and Horton and by Barr and Keysar. In particular,
Gerrig and Horton argue that people necessarily interpret contextual ex­
pressions against both communal and personal sources of common ground.
1. SYNOPSIS AND PROPHESY 3

An alternative view that egocentrism is more prevalent in everyday lan­

guage than mutual knowledge would allow is offered by Barr and Keysar.
This chapter makes a case that, contrary to the traditional view in
pragmatics, conventions in language use can emerge without appeal to mu­
tual knowledge between interlocutors. Rather, conventions can emerge, "as
by-products of dyadic-level mechanisms of coordination" (Barr & Keysar,
chap. 2, this book).
This is a fascinating development in that such contrary claims with evi­
dence can be provided. My suspicion is that both claims are in a sense true,
in that common ground is used, on occasion necessarily so, for important as­
pects of comprehension of many language forms. But there may also exist
parallel lower level interaction mechanisms, as described by Barr and
Keysar, as well as some others (e.g., mimicry, response in kind, script adher­
ence, priming, chaining, attitude display, acting, and mere continuance)
that are also influential. Future work might attempt to substantiate the sep­
aration of these mechanisms and address related questions such as how the
mechanisms work individually and interactively, their possible interdepen­
dence, when are each used, and so on.
Another emerging mechanism involved in nonliteral language compre­
hension is that of negation. Giora, Balaban, Fein, and Alkabets' chapter bol­
sters an ongoing attack on the notion that negation markers (e.g., not)
suppress their target concepts. Although applicable much more broadly to
other language forms, this attack on suppression, or more specifically on
obligatory suppression, plays a key role in some forms of indirect language
(e.g., litotes). This chapter presents experimental evidence that meanings
suppressed via negation markers are retained and influence later processing.
Gibbs and Izett's chapter highlights a variety of cognitive and, in particu­
lar, social psychological mechanisms of persuasion, including contrast, reci­
procity, cognitive dissonance, analogical reasoning, and others, in its review
of the persuasive ability of verbal irony. This chapter poignantly demon­
strates the great need for more cross talk between social psychologists who
study the mechanisms for how people influence one another, through lan­
guage and other means, and psycholinguists and linguists who study the
pragmatic accomplishments of forms of language.
Last, Blumentritt and Heredia's chapter discusses the mechanisms of
schema processing in the form of stereotypes, as well as priming, and how
they can influence the comprehension of forms of metonymy and metaphor.
This chapter is of particular value in its coupling of standard mechanisms of
cognitive functioning-language processing with a timely, important, and
heretofore generally ignored sociocultural variable—stereotypes.
A few other mechanisms are also discussed, including conversational
implicature and constraint satisfaction, but these are treated in the follow­
ing sections, along with the theoretical approaches.
4 COLSTON

TYPES OF NONLITERAL LANGUAGE

The second theme of this book is that it brings together work that addresses
the commonly studied forms of nonliteral language with work on relatively
understudied nonliteral forms. The chapters report studies on metaphor
and verbal irony, as well as on metonymy, proverbs, asyndeton, contextual
expressions, idioms, scalar statements, "What is X doing Y" constructions
(WXDYs), conventional indirect requests, analogies, litotes, and meta­
phorical signs.
One goal of the book was to assess whether any points of discussion might
emerge in the treatments of these diverse nonliteral forms, in terms of the
theoretical approach, mechanistic requirements, and so on. At least three of
these points are considered.
The first is the degree to which language forms depend on contexts of dif­
ferent sorts to enable appropriate comprehension. The chapters by Curco,
Gerrig and Horton, Holtgraves, and Barr and Keysar deal with this issue ex­
plicitly, and indeed all of the chapters deal with it implicitly in their recogni­
tion of the importance of the social and cultural influences, which are
certainly part of the contextual component of comprehension.
What might be emergent from these discussions, however, is the varying
degree of dependence on different kinds of contexts in the comprehension
of the different types of nonliteral language, particularly those having to do
with these sociocultural variables. Beyond the related issues of idiomaticity,
conventionalization, fixed expressions, and the like, which have received
extended treatment in many other venues, the extent of nonliteral forms
treated here and their relative dependence on social and cultural variables
for comprehension might shed new light on the role of context. Consider
that, clearly, all forms of language—nonliteral and others—require some
degree of context for appropriate comprehension. But there is an arguably
greater role played by context in many instances of nonliteral language com­
prehension, because of the less direct correspondence between the utter­
ance meaning and the intended meaning. Comprehension thus must rely on
something else.
So, comprehension of nonliteral forms relies more on context. But, and
this point is less often made, such comprehension also relies more on socio­
cultural pragmatic reasoning. Moreover, this shift toward contextual and
sociocultural sources of information may vary depending on the kind of
nonliteral form being comprehended. Arguably, verbal irony, metaphors,
and metonymies based on stereotypes, and possibly other forms, might rely
more on sociocultural information than other aspects of context. Asynde­
ton, contextual expressions, idioms, analogies, litotes, and scalar state­
ments, conversely, might rely more on other aspects of context than on
sociocultural information. Granted, this observation is speculative at this
1. SYNOPSIS AND PROPHESY

point, but it certainly behooves us to look more at different sources of con­

text, including sociocultural information, and how different nonliteral
forms would rely on these sources to varying degrees, and then perhaps at­
tempt to determine why this is the case. This observation has been made by
other scholars concerning the relative dependence of verbal irony versus
metaphor (Colston & Gibbs, 2002), but it might ought be expanded to in­
clude the entire continuum of nonliteral forms.
The second emergent point in consideration of the many nonliteral forms
treated here concerns an argument I made elsewhere about whether
nonliteral language comprehension should be approached with an inclusive
or a piecemeal theoretical goal (Colston, 2002). The former approach is
demonstrated powerfully in Holtgraves' chapter, wherein a classification of
a number of nonliteral language forms according to different kinds of
implicatures is presented. Several of the other chapters might be seen as es­
pousing the more fragmented approach, whereby the mechanisms they
cover might apply only to one nonliteral form or family (e.g., Gibbs and
Izett's contrast and verbal irony, Giora's suppression and litotes, and
Blumentritt and Heredia's schema activation and metonymy). Whether
Holtgraves' approach can account for all kinds of nonliteral language and
whether it can provide a detailed enough explanation to account for all the
rich pragmatic accomplishments of the different forms (whether it should be
required to explain the latter, however, is an open question) remains to be
seen. Nevertheless, this classification is just the sort of work that is needed
to evaluate whether we can build an explanation of language comprehen­
sion in general, based on such mechanistic or theoretical tenets, or whether
different families of nonliteral (and literal) forms will each require their own
set of explanations.
The last point of discussion to emerge in the chapters' treatments of many
nonliteral forms is that of possible mixtures of mechanisms and processes,
some of them contradictory and others complementary or parallel (see
Gibbs & Colston, 2002, for a discussion of these possible mixtures for ironic
communication), that might underlie nonliteral language use and compre­
hension. One example is the reliance on common ground versus more
dyadic-level mechanisms of coordination already discussed for the compre­
hension of contextual expressions and referential precedents in the chapters
by Gerrig and Horton and by Barr and Keysar, respectively. A second exam­
ple might be the varying degree to which suppression of meaning occurs in
nonliteral language comprehension. Giora notes that suppression, although
not obligatory, does nevertheless occasionally occur. There may thus be dif­
ferent mechanisms at play, one in which suppression occurs and one in
which it does to a lesser degree. A third example might be the conflict be­
tween contrast effect mechanisms and tinge mechanisms in verbal irony
comprehension (see Colston, 1997, for a review of this debate). The former
6 COLSTON

is discussed in Gibbs and Izett's chapter, the latter is treated in Giora's chap­
ter (although not for verbal irony). Another example might be the mix of
processes that underlie the intergroup conflict phenomenon discussed in
Colston's chapter. Here, a combination of language comprehension pro­
cesses, social tension and cognitive load, and social psychological principles
were shown to combine to produce and maintain communication difficul­
ties between different sociocultural groups.
The bottom line here is that nonliteral language use and comprehension
is a very complicated matter. The rich nuance of clever meaning exchange
and pragmatic accomplishment achieved by nonliteral language motivates
its existence, but it is also made possible only by the interplay and, on occa­
sion, conflict between an array of complex cognitive, linguistic, social, de­
velopmental, and other processes. It may not, therefore, be the case that a
particular processing mechanism (e.g., activation of a conceptual meta­
phorical mapping, creation of a blended space, contrast between context
and utterance meaning, inflated perception of an expectation-reality dis­
crepancy, and a host of others) plays a role in each and every usage and com­
prehension of a particular nonliteral form. Granted, there may be general
tendencies, but the true complexity of nonliteral forms and, in particular,
the often found clever blends of nonliteral forms (e.g., an echoic, ironic, hy­
perbolic metaphor delivered laterally) might make straightforward deter­
ministic explanations of nonliteral language incapable of accounting for the
entire phenomenon.

WHAT IS "SOCIAL" AND "CULTURAL"?

The third theme that arose from the chapters concerns different ideas about
what it means for something to be a social or a cultural influence or variable.
Perhaps this discussion is best begun with a listing and description of the va­
riety of sociocultural variables that the chapters cover. Gender is the most
prevalent sociocultural variable discussed in the chapters. Colston, Katz,
Link and Kreuz, and Pexman all offer extensive treatments of the role that
gender plays in nonliteral language use, comprehension, and processing in
terms of the gender of both speakers and listeners. The social relationship
between interlocutors (e.g., close friends vs. strangers) is taken up in the
chapters by Gerrig and Horton, Gibbs and Izett, Katz, and Pexman. Social
role (e.g., high vs. low status or power) is treated in the chapters by Katz and
Holtsgraves. Occupation (e.g., high irony [comedian] vs. low irony [nun]) is
discussed by Katz, Pexman, and Holtgraves. Geographic origin is included in
the discussion of proverb comprehension by Curco, as was religious back-
ground—a very clever insight and possibly a sociocultural variable that has
not been considered previously in discussions of nonliteral language use and
comprehension. Political background is treated by both Giora et al. and
1. SYNOPSIS AND PROPHESY 7

Blumentrit and Heredia. These same chapters also discuss speakers' and lis­
teners' ethnicity. Ethnicity and other kinds of cultural background—includ-
ing the idea that the mere degree of cooperativeness in communication or
degree of indirectness prevalence may be a cultural difference—are re­
viewed by Holtgraves. Personality traits are discussed in Pexman's chapter,
and the medium of language (e.g., spoken vs. signed)—also a very important
sociocultural variable that has received relatively little past attention—is
discussed extensively by Marschark.
As already alluded, these variables have also been treated in a variety of
ways. Occasionally, a sociocultural variable is treated as a predictor of the
likelihood of using a particular kind of nonliteral language (e.g., men use sar­
casm more than women). Other times it is considered a predictor of how a
nonliteral utterance will be comprehended (e.g., a person from outside a
given culture will be less likely to interpret correctly the nonliteral meaning
of a proverb). Still other times they are treated as more global predictors of
use or comprehension extent, quality, and so on (e.g., cultures with more in­
directness will use and comprehend nonliteral language more often than
cultures with less indirectness). In some cases a sociocultural variable is also
considered along with another expressive phenomenon (e.g., men use more
nonliteral language for negative vs. positive emotional expressions).
Other chapters discuss the role that sociocultural variables play at differ­
ent levels of nonliteral language cognition. Katz's chapter in particular pro­
vides a detailed discussion of both on- and offline effects of sociocultural
variables. Gibbs and Izett's chapter talks about the variety of ways that social
influence principles are at play in the use or comprehension of verbal irony,
both in terms of online language processing and in broad effects of irony as
an influencer (e.g., as an effective tool in advertising campaigns). Curco's
chapter discusses the influence of degree of experience with a given lan­
guage as well as degree of acculturation in comprehension of native lan­
guage proverbs—and concomitant with this, the variety of levels at which
proverbs can be interpreted (e.g., as a nonliteral language form as well as a
form of extoliation). Finally, Marschark's chapter's treatment of metaphori­
cal sign language can easily be seen as an example of nonliteral language be­
ing processed at a variety of different levels (visual processing, linguistic
processing, interpretation, etc.).
Still other chapters treat the sociocultural variables as important compo­
nents of the contextual backdrop in which nonliteral forms are always inter­
preted. These discussions focus less on sociocultural variables' influence per
se, but rather they use sociocultural variables as examples of how important
consideration of all aspects of context are when considering language, par­
ticularly nonliteral language, comprehension.
Despite how sociocultural variables were treated, the extent and vari­
ety of these variables attest to their prevalence and demonstrate their
8 COLSTON

rising importance in scholars' thinking and research about nonliteral lan­

guage cognition.

WHAT ARE SOCIOCULTURAL INFLUENCES?

What does it mean for a variable to have an influence on nonliteral language

cognition? This question constitutes the final theme derived from the chap­
ters in this book. In general, we can divide this theme into two issues—the
direction of an influence and the type of influence—and discuss several ver­
sions of each. The direction aspect is considered first.
If we consider an individual person as the object of analysis, then influ­
ence can flow in a variety of directions. A person can hold a number of roles
in a language exchange, and sociocultural influences can impact on or ema­
nate from that person in several ways. If a person is the addressee, for in­
stance, then sociocultural variables in the speakers can affect the addressee
(e.g., comprehension is different if the speaker is wealthy vs. poor). Such
variables can also exist in overhearers and affect the addressee (e.g., com­
prehension is different if the conversation is being overheard by a man vs.
woman). Sociocultural variables also exist in audiences, coaddressees,
strangers, cospeakers, and a variety of other roles, and many of these could
have an influence on the addressee's comprehension (and this is all just to
consider comprehension, other influences are taken up later). Indeed, even
to the extent that sociocultural variables can vary within the same person
(e.g., I am an authority figure to my students in the lab but a regular guy to
them if we go out to lunch), such variation might affect one's own compre­
hension of the same utterance from the same speaker when heard in these
different roles.
If instead the person in question is the speaker, then sociocultural vari­
ables in other people (e.g., addressees, coaddressees, audiences, overhear­
ers, and cospeakers) can affect production. It can also affect the
monitoring of comprehension, which in turn can affect subsequent pro­
duction, subsequent monitoring, and other aspects of the negotiation of
meaning exchange.
If the person is in one of the other roles just discussed, then sociocultural
variables in the speaker, the addressee, or in other people holding other roles
can affect the cognition of the person. Moreover, these mixtures of
sociocultural variables might interact in very complex ways. For example, if I
am obviously overhearing a conversation in which the addressee is known to
be a religious person, I might comprehend a given utterance differently than
if the addressee were known to not be religious. But this switch in my com­
prehension might only hold if the speaker were, for instance, of high social
status. If the speaker were of low social status, then my comprehension
might be more stable across the sociocultural variable in the addressee. Of
1. SYNOPSIS AND PROPHESY 9

course this particular example may not necessarily be true, but interactions
such as this might hold.
Next, the types of influence on a person encountering nonliteral language
as an addressee, overhearer, audience, and so on are considered. Individuals
or groups of people who encounter nonliteral language can be influenced by
sociocultural variables both in terms of processing or comprehension and in
broader pragmatic effects. For instance, a person hearing a nonliterai utter­
ance might process it faster or more slowly; comprehend it more or less exten­
sively; resolve ambiguity to a lesser or greater extent; compute additional,
fewer, or different types of meanings (literal or nonliteral); make different in­
ferences; or do a variety of other kinds of processing differently as a result of a
sociocultural variable in some role in the interaction. People who encounter a
nonliteral utterance may also be broadly affected in terms of their emotional
response; their attitude toward the speaker, other person, or some issue; the
degree to which they are persuaded or insulted; their level of surprise; their
subsequent thinking; their subsequent behavior; and so on all as a result of
such sociocultural variables in the speaker(s) or other people.
Sociocultural variables in other people can also have an effect on individu­
als' or groups' production of nonliteral language. Speakers might use more or
less, or perhaps different kinds of, nonliteral language depending on socio­
cultural variables in the addressees, overhearers, audiences, witnesses, and so
on. Speakers might also blend literal, nonliteral, and types of nonliteral lan­
guage in different ways according to such variables. Indeed, an often discussed
phenomenon of nonliteral language is that it affords incredible flexibility in
communication. An adroit user of nonliteral language can simultaneously
communicate several distinctly different messages individually to several dif­
ferent addressees with the same nonliteral (or even literal) utterance.
So, in summary, influence can flow in a number of different directions,
and it is comprised of changes to processing, comprehension, production,
use, emotional response, and broader pragmatic functioning. This level of
complexity once again attests to how important it is that future work on
nonliteral language continues to explore social and cultural aspects to this
domain of human cognition.

THEORETICAL APPROACHES

In addition to the aforementioned themes that emerged, a number of

theoretical approaches were also used in various chapters. These bear
brief discussion as they link with the point discussed earlier concerning
whether nonliteral language cognition is best attempted by a unified the­
ory or a family of explanations. A few of these approaches might be candi­
dates for the former approach, whereas mechanisms discussed in other
chapters illustrate the latter strategy.
10 COLSTON

Holtgraves' chapter uses Gricean and neo-Gricean distinctions of con­

versational implicatures to attempt to build a classification scheme of a vari­
ety of kinds of nonliteral language and their comprehension. Katz's chapter
introduces the possibility of using constraint satisfaction as a global theoreti­
cal mechanism for explaining nonliteral use and comprehension, an ap­
proach that holds much promise. Giora et al. discusses suppression and
salient meaning, which have been put forth previously as components of an
account for nonliteral language cognition. Lastly, the role of common
ground in comprehension is discussed, albeit with differing conclusions, in
the chapters by Gerrig and Horton and by Barr and Keysar. Each of these ap­
proaches could be an underpinning for a global account of nonliteral lan­
guage processing, comprehension, and possibly use, although they would
clearly need much work to enable explanation of all the nuance to these as­
pects of nonliteral language cognition.
Colston's chapter illustrates a number of potential cognitive and social
psychological theoretical phenomenon that could underlie gender differ­
ences in nonliteral language use, Gibbs and Izett's chapter also outlines a
number of important social psychological theoretical principles that could
underlie nonliteral language comprehension, pragmatic effects, and use.
Link and Kreuz's chapter offers a variety of potential theoretical sources for
gender differences in nonliteral language use. Pexman's chapter discusses
some theoretical components of individual differences. Blumentritt and
Heredia's chapter discusses some psycholinguistic and cognitive theoretical
underpinnings, and both of these chapters apply these components to ac­
count for sociocultural differences in nonliteral language cognition. Lastly,
the chapters by Curco and Marschark provide some key potential theoreti­
cal underpinnings to explain nonliteral language cognition in two novel
ways—the kind of nonliteral language used (proverbs and metaphorical
signs) and the sociocultural variable (s) that might be of interest (religious
background, among others, and language media). These chapters all offer
components that might be incorporated into a global theoretical attempt at
explaining nonliteral language cognition, or, and perhaps more likely, they
might supply components for more disparate explanations.
Given the incompleteness of both the potential global and disparate the­
oretical strategies, it is premature to argue which approach will likely win
out. Nevertheless, if forced to make a prediction at this point, I would have
to opt for the latter strategy. Given the wide variety of basic cognitive, lin­
guistic, social, and even perceptual processes that have been shown to play a
role in the processing, comprehension, and use of different nonliteral forms
(e.g., contrast effects, assimilation effects, conceptual metaphor, distinctive­
ness, memory processes, analogical processes, relevance, foot-in-the-door,
social alignment, and many others), and particularly, the lack of universality
of many of these processes across forms of nonliteral language, then a global
1. SYNOPSIS AND PROPHESY 11

account appears untenable. One exception to this speculation, however,

might be a global approach that is flexible with respect to what kinds of
mechanisms might be at play in a given form of nonliteral language process­
ing, comprehension, or use. Katz suggested a constraint satisfaction ap­
proach that allows the particular constraints to vary according to nonliteral
language type and contextual situation but maintains some fixed theoretical
construct for what pragmatic constraint satisfaction would be might make
for a successful compromise.

FUTURE DIRECTIONS FOR NONLITERAL

LANGUAGE RESEARCH

The topics, empirical findings, explanations, and the emergent themes and
theoretical approaches from the chapters in this book help outline some key
issues that nonliteral language research is facing. 1 briefly discuss some of
these and offer some suggestions for the future of this field. This is not a con­
clusive list, and items are not offered in any particular order of importance.

Authenticity

A number of the chapters alluded to a growing concern with authenticity in

the study of nonliteral language. Indeed, the recognized importance of socio­
cultural variables that motivated this book supports this concern. Most of the
methodologies discussed in the chapters as well as those used in empirical re­
search on nonliteral language involve experimentation. For purposes of con­
trol and manipulation, well known to most psychologists and many linguists
as a means of identifying cause–effect chains, studies typically involve the pre­
sentation of carefully designed, experimenter-crafted situations and utter­
ances to participants who are ideally representative of populations of interest.
The sets of situation-utterance items are manipulated somehow to introduce
variables that are potential causal agents. Participants are then asked to per­
form some language-related task(s) involving the items, and a wide variety of
measurements are taken to tap into potential effect agents. Although the spe­
cific details of the application of this methodology to psycholinguistic studies
may not be widely known to all scholars of nonliteral language, the general
technique is pervasive in psychological research and is very familiar to a psy­
chological and possibly linguistic audience.
Not widely known among psychologists or even among all psycholin­
guists, however, is a potentially damning problem with this methodology dis­
cussed among a number of fields also interested in nonliteral cognition. The
essential criticism is that the rich, sociocognitive nuance of actual nonliteral
talk concerning an infinity of contexts and topics, among varieties of kinds
of interlocutors, with their rich subjective experiences and mutual knowl­
12 COLSTON

edge of these, is severely compromised with the introduction of strict labora­

tory methodologies, items, tasks, and measures. Although there may be
some very low-level auditory, phonological, morphological, or perhaps
higher phenomena that can be studied with laboratory techniques, so the
criticism goes, many phenomena at those levels, and any phenomenon per­
haps involving syntax or semantics but certainly concerning pragmatics, are
made insurmountably artificial if not studied in situ. Proponents of this criti­
cism, therefore, use observational methodologies designed to capture, as
best as possible, actual language used in a variety of real-world settings and
then study recordings and rich transcripts of that language to determine its
nature and to draw inferences about speaker-listener cognitive processes,
intentions, language functioning, and so on.
Countercriticisms offered against this conversational or discourse anal­
ysis methodology are typical of arguments against observational methods:
Although they might allow cursory descriptive accounts of language phe­
nomena, they afford no means of reliable generalizeability across partici­
pant or language type populations, the precise identity of causal agents is
indeterminate, the cognitive state of the interlocutors is relatively inac­
cessible, and so on.
Whatever position one takes in this debate, there is nonetheless some­
thing to be said for the importance of authenticity in studying nonliteral lan­
guage cognition. Although it may be true that the use of strict control,
which actually may not necessitate the use of inauthentic items, impacts, for
example, the processing or comprehension of nonliteral language so mini­
mally that its disadvantages may be overlooked, the delicacy and complexity
of nonliteral language cognition lean heavily against this view. Indeed, the
recognition by the scholars in this book, as well as others of the importance
of attending to sociocultural variables and their impact on nonliteral lan­
guage cognition demonstrates the appreciation of the richness of nonliteral
language. My view is that attempts should be made to find compromise
methodologies that make use of recorded authentic language and then
adopt more controlled mechanisms to enable causal determinations.
For example, I (Colston, 2004) recently used a methodology wherein peo­
ple were placed in contrived, controlled social situations that varied some key
factor (e.g., how much the participant respected his or her addressee) and re­
quired the participant speakers to produce a particular form of language in
that situation—specifically, gratitude acknowledgments (e.g., "no problem"
and "anytime"). These verbatim utterances were then given in written form
to a separate group of participants who rated the degree of nonliteralness of
the utterances.
The evaluation of the pragmatic functions accomplished by gratitude ac­
knowledgments was made in this study with a methodology that thus provides
an appealing compromise between purely experimental and purely field- or
1. SYNOPSIS AND PROPHESY 13

corpus-based studies. By collecting real utterances offered by normal, every­

day speakers, instead of using experimenter-crafted comments, a much-
needed improvement in ecological validity is gained. Indeed, the authentic
gratitude acknowledgments collected in this study were much richer than
what would likely have been created by an experimenter. They are often
grammatically questionable (e.g., "I'm sure good" and "I'm real sorry your bike
got stole"). They do not always use complete sentences (e.g., "Refresh on old
stuff" and "You could have"). They refer to subtle details of the contexts (e.g.,
"The jacket looks good on you" and "How was the party"). They make use of
colloquialisms (e.g., "No sweat" and "it's all good"), and so on.
Conversely, by manipulating people into producing these utterances in
reasonably controlled situations rather than simply capturing usages in fully
natural settings, some of the looseness in fieldwork in terms of causal agents
or links, generalizeability, replicability, and so on, is reduced. Future work
might continue to make use of such compromise methodologies as an alter­
native to more extreme procedures.
Nonliteral Blends
Both the concern for authenticity and the work that is needed on
prevalences of kinds of nonliteral language (discussed next) call for more at­
tention to be placed on blends of nonliteral forms of language, not just on
blends of metaphorical language (see Lee & Barnden, 2001) but on blends
of different kinds of nonliteral language (Colston, 1998). Although some of
this work has begun to appear in the recent literature (e.g., Colston & Gibbs,
2002), my strong suspicion is that corpus and fieldwork will highlight the
prevalence of this kind of talk. Moreover, as already discussed, given the po­
tential offsetting nature of some mechanisms of nonliteral language cogni­
tion (suppression vs. no suppression, tinge vs. contrast effects, etc.),
investigation of such blended forms will become very important. Production
of such blends is also an important understudied area. This issue is taken up
in the Production section.
Prevalence of Nonliteral Language
Although there have been a number on studies of the prevalence of forms of
nonliteral language amid certain targeted populations (e.g., Gibbs, 2000), as
well as some very recent corpus work just emerging on this issue for figura­
tive language (Lee, 2003; also see an upcoming special issue of Metaphor and
Symbol devoted to corpus work on metaphor), there has yet to be any large-
scale measurements of actual talk in a wide array of settings and cultures for
a large number of forms and mixtures of nonliteral language.
Such work is crucial to help evaluate our emerging and refining explana­
tions. For instance, if 70% of hyperbolic utterances are of a certain type (e.g.,
14 COLSTON

specific quantified magnitude overstatements) and a smaller percentage are

of another type (e.g., extreme case formulations) and different comprehen­
sion or use explanations are supported by these different types, then the prev­
alence data can help weigh the comprehension explanations accordingly.
Benchmarked longitudinal studies of prevalences of nonliteral language
would also allow a better evaluation of mechanisms like grammaticalization,
and concomitantly the different comprehension accounts that rely on or
dismiss such processes, and the impact that sociocultural variables play in
these and similar processes.

Continued Empirical and Theoretical Work

The diversity of theoretical approaches and mechanisms and the relatively

underexplored nature of many kinds of nonliteral language also speak to
the huge need for continued work that addresses new nonliteral forms and
new nonliteral processing, comprehension, or use mechanisms—similar to
what has already been conducted. 1 do not believe that certain summary o
evaluation attempts such as this book should cease, but there is much
more territory to be explored before the process of map making will be­
come entirely fruitful.

Production

Until very recently, there has also been a dearth of work on the production of
forms of nonliteral language (see Barr, in preparation; Gibbs, 2000; Horton
& Keysar, 1996; and Link & Kreuz, this book). One standard reason for pro­
cessing and comprehension work outpacing that of production lies with the
greater difficulty in conducting work on production. Electronic communi­
cation technologies that enable relatively easy recording of nonliteral pro­
ductions are changing this situation, but these suffer from the constraints
inherent in the medium (Hancock &. Dunham, 2OOla, 2OOlb). Although
such computer mediated work is useful and should continue, parallel work
that investigates spoken nonliteral language is also crucial and should re­
ceive more attention.
More work on production of nonliteral language will help evaluate some
of the theoretical perspectives emerging in the literature. Katz's constraint
satisfaction approach in particular might benefit by more production re­
search; one could test whether speakers are attempting to satisfy a host of
complex cognitive, linguistic, and social goals and constraints in how they
talk nonliterally.
More production work could also help delineate sources for differences in
the quality of nonliteral language use, factors that lead to development of
such use, limits on such skills, and so on. In particular, production work
1. SYNOPSIS AND PROPHESY 15

could help evaluate how people use clever blends of nonliteral language
forms to accomplish complex mixtures of often conflicting social and com­
municative goals (e.g., to defend one's position from attack but to also main­
tain appropriate politeness).
Last, production work could also help address the empirical issues dis­
cussed thus far concerning the impact of sociocultural variables. Recall that,
for instance, some findings have shown more production of particular kinds
of nonliteral language across some sociocultural variables (e.g., men use sar­
casm more often than women). Production work could help continue the
exploration of these differences and their sources.

APPLICATIONS OF FINDINGS

Aside from the important directions for the basic research on processing,
comprehension, and use of nonliteral language, a number of important ap­
plications of nonliteral language might warrant further investigation.
As discussed in Colston's chapter, nonliteral language comprehension
and its breakdown might underlie some aspects of group conflict. Future
work to refine this potential mechanism and, ideally, to circumvent it (e.g.,
does knowledge of the mechanism on the part of interlocutors act to derail
or slow the communication breakdown?) could be important to alleviate
this source of social problems.
Indeed, the field of conflict resolution could potentially be benefited by
increased work on the role that nonliteral (and literal) language processing,
comprehension, and use play in misunderstandings, misinterpretations, and
other communication difficulties. Although there is already an enormous
cottage industry surrounding this kind of endeavor, it is unclear to what ex­
tent the practitioners make use of the most recent findings from research on
nonliteral language.
Given the work discussed in some chapters about the pragmatic effects of
various forms of nonliteral language (e.g., persuasion in Gibbs and Izett's
chapter and extoliation in Curco's chapter), some applied work might also
address how nonliteral language is useful for educational purposes (see
Barnard & Colston, 2000, for one example).
There may also be room for improvement to legal systems that in many ways
are built on the tenet that language is literal and people state what they mean.
Courtroom procedures, interrogation policies, beliefs surrounding testimony,
and a host of other aspects of how many legal systems operate are based on some
archaic notions about language and its use. Research on nonliteral language
comprehension applied to these issues might also help to improve such systems.
Last, and related to the aforementioned argument, there are enormous
benefits to be gained simply by educating populations much more broadly on
what language, nonliteral and literal, is and how its comprehension and use
16 COLSTON

work. I very recently read an article from a local newspaper (Barton, 2002)
that illustrates this point poignantly:

Joke empties plane: Man ticketed and sent home as passengers scurry
to resume travels.
As he boarded a commuter plane at the Dane County Regional Air­
port Tuesday morning, 33-year-old Steven M. Wiese of Cottage Grove
peeked around the cockpit door and made a joke: "I hope you haven't
been drinking."
The pilot didn't think it was funny. Neither did the other passengers.
"(The pilot) could see the other passengers' heads popping up like go­
phers coming out of their holes," said Lt. Michael Krembs of the Dane
County Sheriff's Department. "The whole plane could hear. He said it
really loud."
In an instant, Wiese's holiday travel plans were ruined. The flight was
delayed. The other passengers had to re-book on other flights on
Christmas Eve. And the pilots were checked to make sure they indeed
had not been drinking.
Wiese and his wife have not flown since their honeymoon nine years ago.
He had planned a trip to New York as a surprise Christmas gift for her.
"I shouldn't have said it. I regret saying it," Wiese said Tuesday after­
noon. "I meant it as a joke. I had a smile on my face. But they took it the
wrong way." Atlantic Coast Airlines flight 6302, with 26 passengers
aboard, was delayed. The crew reported to a medical facility to be
tested for alcohol and drugs, in compliance with Federal Aviation Ad­
ministration guidelines.
"Since 9–11, you can't joke about anything in an airport," Krembs said.
Wiese apologized to his fellow travelers as they got off the plane at the
Madison airport and lined up at the ticket counter, hoping to be
re-booked. Then he and his wife got in line, too.
"The deputies pulled them out of line and said, 'Oh, no. You're not go­
ing anywhere today,'" Krembs said.
The deputies then ticketed Wiese for disorderly conduct. He must pay
a $225 fine.
That's not all.
The FBI and the U.S. attorney's office will discuss the incident after
the holidays and decide whether the man should face the more serious
charge of interfering with a flight crew, a federal felony that carries a
1. SYNOPSIS AND PROPHESY 17

maximum 20-year prison term, said Monica Shipley, FBI spokes­

woman.
All of the other passengers were re-booked to their final destinations,
said Rick Delisi, spokesman for Atlantic Coast Airlines. The members of
the flight crew all tested negative for alcohol and other controlled sub­
stances. Their plane departed more than four hours late, at 2:05 p.m.
Delisi said the pilot followed the proper procedures. "The crew is never
meant to ask whether (people are) joking or what their motivations
are," he said. "... People just don't realize the seriousness of what
they're saying."
For Wiese, that realization came at a high price—and not just the cost of
the nonrefundable plane tickets. Instead of taking the trip, he and his
wife now plan a quiet holiday at home trying to recover from the ordeal.
Wiese has weighed in his mind exactly what he should have done dif­
ferently.
"I should have said, 'Merry Christmas.' "

If such well-intended endeavors as ensuring the safety of air travel were not
victim to these sorts of ill-placed policies stemming from a widespread sim­
plistic notion of language and its use, perhaps these and other important ef­
forts could be channeled in more effective directions.

REFERENCES

Barnard, K., & Colston, H. L. (2000, November). How to teach people about the web:
Image'Schematic metaphor and the construction of the information highway. Paper
presented at the meeting of the Society for Computers in Psychology, New Or­
leans, LA.
Barr, D. J. (in preparation). Paralinguistic correlates of discourse structure.
Barton, G. (2002, December 24). Joke empties plane: Man ticketed and sent home
as passengers scurry to resume travels. Milwaukee Journal Sentinel.
Colston, H. L. (1997). Salting a wound or sugaringa pill: The pragmatic functions of
ironic criticism. Discourse Processes, 23, 25-45.
Colston, H. L. (1998, July). On the viability of a multi-trope account of figurative lan­
guage processing. Poster session presented at the meeting of the International
Pragmatics Association, Reims, France.
Colston, H. L. (2002). Pragmatic justifications for nonliteral gratitude acknowledg­
ments: "Oh sure, anytime." Metaphor and Symbol, 17, 205-226.
Colston, H. L. (2004). On why people don't say what they mean: An analysis of authen­
tic formulaic language. Manuscript submitted for publication.
Colston, H. L., &Gibbs, R. W., Jr. (2002). Are irony and metaphor understood dif­
ferently? Metaphor and Symbol, 17, 57-80.
Gibbs, R. W., Jr. (2000). Irony in talk among friends. Metaphor and Symbol, 15,5-27.
18 COLSTON

Gibbs, R. W., Jr., & Colston, H. L. (2002). The risks and rewards of ironic communi­
cation. In L. Anolli, R. Ciceri, &G. Riva (Eds.), Say not to say: New perspectives on
miscommunication (pp. 181-194). Amsterdam: IOS.
Hancock, J. T., & Dunham, P. J. (200la). Impression formation in computer-medi-
ated communication revisited: An analysis of the breadth and intensity of im­
pressions. Communication Research, 28(3), 325-347.
Hancock, J. T., & Dunham, R J. (2001b). Language use in computer-mediated com­
munication: The role of coordination devices. Discourse Processes, 31, 91–110.
Horton, W. S., & Keysar, B. (1996). When do speakers take into account common
ground? Cognition, 59(1), 91-117.
Lee, M. G. (2003, January 10). Corpus-Based Approaches to Figurative Language
Workshop. Message posted to the Cognitive Linguistics electronic mailing list.
Lee, M. G., & Barnden, J. A. (2001). Reasoning about mixed metaphors within an
implemented artificial intelligence system. Metaphor and Symbol, 16, 29–42.
 PART I

Sociocultural

Knowledge Influences

This page intentionally left blank

 2

Making Sense of How We

Make Sense: The Paradox of
Egocentrism in Language Use
Dale J. Barr
University of California, Riverside

Boaz Keysar
University of Chicago

Language users routinely face the problem of making sense out of language:
Speakers must design utterances that listeners can understand, and listeners
must interpret utterances the way they were intended. Because ambiguity is
pervasive in language use, pragmatic theories assume that speakers and lis­
teners should strive to speak and understand against the background of a
mutual perspective. However, our findings indicate that speakers and listen­
ers are egocentric to a surprising degree. With respect to many current
frameworks in psycholinguistics, these findings are anomalous: They sug­
gest that language users are not properly designed for the task of making
sense. Our goal in this chapter is to review these findings and to try to sketch
out a new framework against which such egocentric behavior makes more
(theoretical) sense. We propose that language users can rely on simpler
mechanisms than current theories require because the work they are as­
sumed to do to achieve a mutual perspective is actually distributed among
processes in the language use environment.
The research presented in this book exemplifies how the meanings of fig­
urative and other so-called "indirect" language can vary with social and cul­
tural context. In this regard, it echoes a refrain that one encounters time and
21
22 BARR AND KEYSAR

again in the literature on language use—that meaning is underdetermined

in the sense that the same string of words can convey anything from a benign
comment to vicious sarcasm. One of the reasons why people continue to
marvel at this idea is that we are still lacking a clear picture of how, in the
face of vast ambiguity, people are able to make sense with and out of lan­
guage. What we require is a better understanding of the mechanisms that
underlie how people coordinate with each other.
One point on which all can agree is that speakers and listeners will only
properly understand one another if they process utterances against a similar
background set of assumptions and beliefs. In an influential study, Clark and
Marshall (1981) characterized this background as mutual knowledge, the set
of knowledge that is mutually held between interlocutors. Mutual knowl­
edge is different from knowledge that is merely shared, because it is not only
shared but is known to be shared. According to Clark and Marshall, mutual
knowledge is essential because it is the only true guarantee of successful
communication. They demonstrated this by showing that there are circum­
stances under which utterances that are based on information that two indi­
viduals have in common—but do not know they share—will fail. Because
Clark and Marshall developed the theory extensively for the case of definite
reference, it makes very clear predictions about how people should behave
when producing and understanding referential expressions.
In definite reference, a speaker uses language to establish some object as
the intersubjective focus of attention. Based on the cooperative principle,
when speakers describe referents to listeners they should strive to be opti­
mally informative (Grice, 1975). This means that they should provide no
more and no less information than is necessary for the purpose of securing
the listener's attention. Yet, what counts as "optimally informative" is de­
fined by their mutual knowledge (Clark & Marshall, 1981). Imagine that
Henry and Mabel are sitting at a table and there is a candle and a box placed
between them. Henry wishes to get Mabel to put the candle in the box.
What should he say? Clearly, he can simply tell her to "put the candle in the
box." He should not say "the small candle" even if he is thinking of other
larger candles that were on the shelf where he purchased the candle—these
other candles are not part of their mutual knowledge. Similarly, Mabel will
know that Henry is talking about the candle on the table and not the one
that Ted bought for her yesterday, of which the small candle reminds her, be­
cause she knows that Henry does not know about Ted's candle. She will not
ask "Which candle?" because she will know, by virtue of their mutual knowl­
edge, that Henry is referring to the candle on the table.
As cognitive psychologists, we became interested in the mutual knowl­
edge theory because of its strong implications for language processing.
Even though much of the literature on mutual knowledge shuns explicit
discussion of psychological mechanism, the theory's proponents have on
2. MAKING SENSE 23

various occasions made clear assumptions that mutual knowledge should

serve as a guiding principle in the access of information during language
processing (Clark & Carlson, 1981; Gerrig, 1986) as well as in the organi­
zation of memory (Clark & Marshall, 1981) and the lexicon (Clark, 1998).
Thus, speakers should design utterances with their specific addressee in
mind (Clark & Murphy, 1982), and comprehenders should restrict the in­
formation they consider to mutual knowledge (Clark & Carlson, 1981).
These proposals assume that language users can directly compute mutual
knowledge on the fly and, therefore, call for the existence of powerful and
efficient cognitive mechanisms that can guide the formulator and parser in
their processing decisions. Building mutual knowledge directly into the
language processing system seems advantageous not only because it would
guarantee successful communication, but it would also reduce uncertainty
and thus the complexity of the problem that language users face (Clark &
Carlson, 1981).
From this vantage point, we were surprised to find from our own experi­
ments on language use that speakers and listeners commonly violate their
mutual knowledge when they produce and understand language. Thus,
Henry will often refer to the candle as "the small candle" (Barr, 1999; Hor­
ton & Keysar, 1996), and Mabel will often consider the candle that Ted
bought her as the referent or ask "Which candle?" even though there is only
one candle that is uniquely defined by their mutual knowledge (Keysar, Barr,
Balin, & Brauner, 2000; Keysar, Barr, Balin, & Paek, 1998). This behavior is
egocentric because it is rooted in the speakers' or listeners' own knowledge
instead of in mutual knowledge.
We have also observed effects in other domains that confirmed the gener­
ality of this egocentrism. Overhearers violate mutual knowledge in assessing
whether an addressee will perceive an utterance as sarcastic (Keysar, 1994).
Addressees interpret referential expressions according to naming prece­
dents established by a previous speaker even though the current speaker was
absent when the precedents were established (Barr, 1999; Barr & Keysar,
2002). Moreover, people turn out to be quite poor estimators of what others
know. Speakers systematically underestimate the ambiguity and overesti­
mate the effectiveness of their utterances (Keysar &. Henly, 2002). When
taught the meaning of an opaque idiom such as the goose hangs high, people
overestimate the likelihood that others who are unfamiliar with the idiom
will perceive its meaning (Keysar & Bly, 1995).
These findings might appear to conflict with other studies that purport to
show effects of mutual knowledge on comprehension (e.g., Clark,
Schreuder, & Buttrick, 1983; Greene, Gerrig, McKoon, &Ratcliff, 1994).
However, Keysar (1997) noted that these studies suffer from a design flaw
which confounds information that is mutual (i.e., known to be shared) with
information that is simply known to the self (see also Lea, Mason, Albrecht,
24 BARR AND KEYSAR

Birch, & Myers, 1998). Once this flaw has been corrected, these methodol­
ogies yield the same systematic egocentrism as the studies just reviewed.
The evidence that language users are more egocentric than the mutual
knowledge theory predicts is too abundant and too robust to be ignored. Yet,
in light of the assumption that mutual knowledge is the only true guarantee
of successful communication, it raises something of a paradox. How can lan­
guage users be egocentric and still communicate effectively? In other words,
how can people get away with being so alarmingly unsophisticated in dealing
with the sophisticated problem of making sense?
We suggest that the only way to really make sense of this seemingly erratic
behavior is to take a closer look at the circumstances under which it was de­
signed to operate: the language use environment. Although certain aspects
of the language use environment have been closely studied by researchers in
disparate fields, there has been little effort to fit these pieces back together so
that they can yield their full impact on theories of language use. In what fol­
lows, we examine the structure of the language use environment with the
hope of showing how, counterintuitively, an egocentric way of speaking and
understanding can make sense. Language users' egocentric behavior does
not necessarily reflect a badly designed processing system; instead, we argue
that it reflects the operation of simple heuristics that are adaptive given the
normal circumstances of spoken language use. The burden of computing
mutual knowledge is not one that individual language users carry alone;
rather, it is one that they share by distributing it over processes in the envi­
ronment. The richness of the environment enables language users to com­
pensate for their limitations. In other words, language users have a bag of
simple tricks that makes them look more sophisticated than they actually
are. These tricks work because they are specifically designed to exploit the
structure of the environment.

DOMAINS OF THE LANGUAGE USE ENVIRONMENT

What do we mean by the language use environment? We propose a typology

that divides the environment into three subdomains: cognitive, inter­
actional, and cultural (see Table 2.1). These domains are defined by the so­
cial and language units that constitute the primary units of analysis. By
analogy, the set of domains can be construed as a set of lenses of varying
power of magnification through which the theorist observes language use
Looking through the strongest lens, we can observe language use at its finest
level of resolution, the cognitive domain. In this domain the basic social unit
is the individual language user and the basic language unit, the single clause
or utterance. This domain corresponds to the traditional level of inquiry in
psycholinguistics, whose task is to understand the moment-by-moment pro­
cesses underlying the production and comprehension of single utterances.
2. MAKING SENSE 25

TABLE 2.1
Basic Domains of the Language Use Environment

Social Language Characteristic Theoretical

Domain Unit Unit Processes Import
Cognitive Individual Clauses or Judgment and Places
utterances decision making constraints on
Attention computation
Memory
Categorization
Interactional Dyad or Conversational Epistemic exchanges Promotes shared
group turns Grounding understanding
(Clark &. among the
Brennan, 1991) dyad or group
Multimodal
communication
Cultural Community Languages Establishment or Promotes
diffusion of commonality
conventions of semantic
representation
among
community
members

The majority of our own research has primarily focused on understanding

processes within this domain. What this research has shown is the existence
of certain limitations on computation that prevent speakers and listeners
from effectively deploying mutual knowledge when they process language.
Any theory of how people establish shared understanding needs to take
these limitations into account.
The broader domain in which the individual language user is embedded is
the interactional domain, wherein the basic social unit of analysis is the dyad
or larger group of conversational interactants and the basic language unit is
the conversational turn. The focus at this level is on the interactive pro­
cesses by which individuals manage the interaction (Sacks, Schelgloff, &
Jefferson, 1974) and establish shared understanding (Clark &. Brennan,
1991; Garrod & Anderson, 1987). Research on the interactional environ­
ment provides two extremely important insights. First, in conversation a
shared perspective can be negotiated through an interactive process be­
tween interlocutors (Clark & Brennan, 1991; Clark & Wilkes-Gibbs,
1986). Second, it shows that language as traditionally construed is really
part of a multichannel system of communication (Clark, 1996) that in­
cludes paralinguistic information conveyed through the spoken channel
26 BARR AND KEYSAR

(Barr, 2003; Clark &Fox Tree, 2002) as well as body movements such as ges­
tures that are conveyed visually {McNeill, 1992).
Much of what we currently know about the interactional domain is due
to research by Clark and by other proponents of the mutual knowledge the­
ory. In their view, interaction works at a metarepresentational level; that is,
it is used as a vehicle for building models about what others know. Instead,
we suggest that interaction works not indirectly at the metarepresentational
but directly at the representational level in that it serves to coordinate indi­
viduals' conceptions of the discourse. This hypothesis is supported by some
of our findings which indicate that people egocentrically apply what they
have learned with one interlocutor when talking to the next one despite the
fact that this violates mutual knowledge (Barr, 1999; Barr &Keysar, 2002).
What is important about interactional processes, we argue, is that they
greatly reduce the amount of work that must be done by processes in the
cognitive domain.
Last, it is useful to remind oneself that communication is only made possi­
ble by the existence of shared semantic representations in the language
community. The degree to which individual perspectives diverge or con­
verge depends on the degree of overlap in how people in the community rep­
resent linguistic and world knowledge. With little overlap in semantic
representations, communication will seem difficult. With large amounts of
overlap, communication will seem effortless. To understand what is neces­
sary for successful communication, it is important to evaluate the degree of
overlap on which language users can typically rely. This requires an under­
standing the mechanisms that generate similarities or differences in seman­
tic representation among the members of a language community.
Having completed this overview of the domains of the language use envi­
ronment, we now put on our first set of lenses and take a closer look.

The Cognitive Domain

The reader might question our classification of the mind of the individual
language user as a domain within the larger environment because this term
has traditionally been used to refer to the set of information that is outside
the mind of the individual. However, following Herbert Simon (1996) we
view the structure of the human mind as a kind of an environment for
thought. We propose that the cognitive domain is relevant to pragmatic
inquiry because it imposes limits on the kinds of computations that lan­
guage users can make during real-time conversation. Furthermore, the
so-called "external" environment is itself made possible through the exis­
tence of cognitive structures that encode, store, and retrieve information.
Therefore, accessing information while planning and interpreting utter­
2. MAKING SENSE 27

ances will partake of such basic cognitive processes as judgment, categori­

zation, attention, and memory, and it will thus reflect inherent features of
how they operate.
Mechanisms of Decision Making in Language Use. Because ut­
terances are inherently ambiguous, speakers and listeners face a high degree
of uncertainty when they attempt to convey their own or decode another's
intention on the basis of linguistic evidence. They may have insufficient in­
formation about what their interlocutor knows, or they may lack good evi­
dence on which to base their assumptions. Our research shows that
language use is no different from other domains in which people must make
judgments under conditions of uncertainty. Specifically, we find that speak­
ers and listeners employ the same anchoring and adjustment heuristic in un­
derstanding language as they do in other forms of problem solving.
When people make judgments to solve a problem, they tend to anchor
their judgments in available information regardless of whether this informa­
tion is actually useful for solving the problem. They then adjust away from
this initial anchor, although the adjustment is typically insufficient; their ul­
timate response is skewed toward the initial anchor. For example, Tversky
and Kahneman (1974) found that a group of high school students who were
asked to estimate the multiplicative product of a sequence of numbers listed
in descending order came up with higher estimates than a group who was
asked to estimate the product of the same sequence with the numbers listed
in ascending order. The explanation for this difference was that the students
anchored their responses in the initial numbers of the sequence. Similarly,
Epley and Gilovich (2001) showed that people often anchor on related,
self-generated anchors and then make adjustments. When asked when
Washington became president, they anchored on 1776, the lowest possible
value, and then adjusted upward.
Likewise, research by Keysar and colleagues suggests that the same anchor­
ing and adjustment principles apply to language processing, including how
people perceive the meanings of figurative language (for a more extensive re­
view of the anchoring and adjustment approach to language processing, see
Keysar & Barr, 2002). Keysar (1994) showed that people's assessments of how
others will perceive sarcasm are anchored in their own knowledge. In his
study, participants read a passage involving two protagonists. For example, in
one passage June recommends a certain restaurant to Mark. Mark goes there
for dinner and has either a positive experience (i.e., enjoys the restaurant) or a
negative experience (i.e., hates the restaurant). The next day he leaves a note
for June that says, "The restaurant was marvelous, just marvelous." Partici­
pants were asked whether June would construe Mark's statement as sarcastic
or sincere. Keysar found that respondents tended to think that June would
perceive Mark's attitude even though she was missing the crucial information
28 BARR AND KEYSAR

about the valence of his attitude. The interpretation of this finding is that re­
spondents anchored their estimate in their own understanding, and they
failed to sufficiently adjust to the perspective of the uninformed addressee.
In addition, Keysar and Bly (1995) reported evidence that people anchor
their assessments of how others will perceive the meanings of idioms in their
own knowledge. In their study, they selected archaic English idioms such as
the goose hangs high whose meanings were unfamiliar to modern day college
students. The idioms were presented in the context of one of two passages by
which the reader could infer either the original meaning or its opposite. For
instance, in one passage, the goose hangs high was used to express optimism in
the future. In a second passage, the same idiom was used to express forebod­
ing about the future. Then, each participant read a passage in which a per­
son used the idiom in conversation with a stranger but in a context that did
not reveal its meaning. They were asked what they thought the stranger
would take the idiom to mean. Sixty-two percent of respondents tended to
think that the stranger would take the meaning to be the same thing they
learned in the first place, whereas only 32% believed the stranger would un­
derstand the opposite meaning. In short, the respondents' anchoring of their
judgments in their own knowledge of the meanings of the idioms led them to
believe that the idioms were more transparent than they actually were.
In sum, these findings underscore the relevance of processes of cognition
for theories of language use. Language users egocentrically anchor their
judgments in available information and fail to fully adjust to the perspectives
of others, just as they do in standard decision-making tasks. This suggests
that the mechanism by which people assess shared perspective in speaking
and understanding and the one they use in nonlinguistic problem solving
are one and the same and, therefore, will be subject to the same sort of limi­
tations.

The Control of Attention and Capacity Limitations. Attentional

processes determine what information becomes accessible to cognitive sys­
tems in the normal course of their operation. We propose that to operate at a
speed that is fast enough to cope with conversation, language processing sys­
tems are designed to operate on information that is made available by atten­
tion regardless of whether this information is part of mutual knowledge.
Many problems of pragmatics, such as formulating and disambiguating
referential expressions, are problems that concern the control of attention.
Speakers produce referential expressions to guide listeners' attention to ref­
erents. The expression that a speaker chooses in referring to some object—
from an elaborate, full noun phrase to a simple pronoun—will depend on
the degree to which the referent is in the focus of attention in the discourse
(Ariel, 1998; Chafe, 1976; Gundel, Hedberg, &Zacharski, 1993). Likewise,
listeners use the speaker's level of specificity as a cue to guide them in their
2. MAKING SENSE 29

search for referents. For example, when listeners hear pronouns, they will as­
sume that the referent is the current focus of attention.
However, under certain circumstances, a speaker's and a listener's focus
of attention may not coincide. Suppose that your sister is telling you about
her husband's trip to China, which her husband already told you about
when you saw him yesterday. Feeling somewhat bored, your attention strays
off, and you remember that you have not caught up with your friend Ben
since he returned from Mexico. Just as you are thinking about Ben, your sis­
ter asks, "Have you talked to him yet?" The question is, will you consider
Ben as the referent of the pronoun him, albeit temporarily, even though your
brother-in-law has been established as the topic of discourse?
Because the attentional foci of speakers and listeners may not coincide, to
avoid miscommunication it would seem optimal for comprehension systems
to restrict the search for referents to mutually known information (Clark &
Carlson, 1981). We put this restricted search hypothesis to the test by creat­
ing such a situation in our laboratory (Keysar et al., 1998). In this experi­
ment, listeners wore a set of headphones while they helped a confederate
director fill in missing details on a target picture (e.g., a picture of an air­
plane). At a critical moment, the listener's attention was distracted by a
voice in the headphones that instructed him or her to look at a competitor
picture. Just at the moment that the listener's eyes were focused on the com­
petitor, the confederate asked a question about the target picture; for in­
stance, "What color are its wings?" When the competitor picture that was at
the focus of the listener's attention was a picture of a woman, listeners were
faster to move their eyes back to the target picture than when it was a picture
of a bird. This delay indicates that listeners mistook their own private
thoughts as the referents of speakers' utterances. This is strong evidence
against the restricted search hypothesis, and it supports our contention that
language processing systems are designed to rapidly exploit the information
made available by attentional processes regardless of its mutuality.
The finding that language processing systems do not initially restrict
themselves to mutual knowledge suggests that misunderstanding will be sys­
tematic and pervasive. Yet, we know that people can compute the shared
perspective when asked. Perhaps the reason why they fail to do this during
routine language processing is because of capacity limitations; that is, the
system must operate at such a time scale that it cannot accommodate infer­
ences about mutual knowledge.
Horton and Keysar's (1996) study provides evidence in support of this
view. In this study, participants described a target shape to a listener. The
target shape was paired with a context shape that the speaker could see, but
the listener could not. For instance, a target circle appeared next to a larger
context circle that only the speaker could see. The question was whether
the speaker would describe the target circle as "the small circle" or just "the
3O BARR AND KEYSAR

circle," with the former case representing a failure to consider the listener's
lack of knowledge of the context circle. Horton and Keysar found that
speakers were more likely to produce such egocentric utterances under the
pressure of a response deadline than they were when they were allowed to
respond at leisure.
We believe that this series of findings strongly implicates that language
processing systems are designed to quickly settle matters of referential ambi­
guity by making rapid use of available information, regardless of its mutual­
ity. Speakers and listeners routinely ignore even the most blatant cues to a
referent's mutuality, such as whether it is occluded from the other's view,
when they process utterances. In other words, language processing is an­
chored in the assumption that what is salient or accessible to oneself will also
be accessible to one's interlocutor. Against the theoretical background in
which mutual knowledge is taken as the only true guarantee of successful
communication, the idea that the design of the language processor would
embody such an assumption in its design is perplexing. Yet, we contend that
although there are no guarantees of mutual understanding, there are power­
ful mechanisms in the interactional and cultural environments that pro­
mote shared perspectives among interlocutors, which can compensate for
these limitations.

The Interactional Domain

One response that we sometimes encounter when we discuss our findings is
that perhaps in real interaction, as opposed to in the psychological labora­
tory, people will be less egocentric. The assumption seems to be that to make
interaction work, people will have to be more careful about adhering to mu­
tual knowledge. Yet, we suggest that the opposite of this assumption may
hold true: People are much more busy during real conversation than they are
in the slow-moving and informationally rarified environment of the psycho­
logical laboratory; therefore, they will have fewer resources with which to
constantly monitor what others know. However, because conversation is in­
teractive and multimodal, it affords speakers and listeners copious feedback
on their performance which affords them the opportunity to be more ego­
centric.
Even though conversational interaction is the fundamental setting of
language use, the study of interaction has had little influence on the study of
the psychology of language use. In many ways studies of conversational in­
teraction developed in parallel with the laying of the foundations of sen-
tence-level pragmatics in the Gricean and speech act traditions. The field of
psycholinguistics, with its focus on single utterances and the isolated
speaker, listener, or reader, has as its primary object the cognitive domain,
and it has yet to gain the full impact of insights from studies of interaction.
2. MAKING SENSE 31

One reason why research on interaction should cause people to rethink

pragmatic theories is that it shows that pragmatic expectations are surpris­
ingly flexible, such that speakers and listeners adapt them to suit their con­
versational experience. Consider the generalized pragmatic expectation
that speakers should be optimally informative in designing referring expres­
sions: They should provide their listeners with no more and no less informa­
tion than is necessary (Grice, 1975). When choosing the word by which to
refer to some object, speakers typically have a wide variety of choices avail­
able, each of which reflects different ways of categorizing the object (Brown,
1958). For instance, a shoe can be referred to as the thing, the shoe, or the
loafer. Cruse (1977) suggested that speakers must determine the proper level
of lexical specificity or their utterances can generate unintended conversa­
tional effects. For the purpose of definite reference, he argued that speakers
should choose a specific term only when it is required, or they should default
to a conventionally neutral level. For instance, if there are two shoes of dif­
ferent styles, the speaker might call the intended shoe the loafer. However, if
there is only one shoe, the specific term loafer will sound especially "marked"
and the speaker should default to the conventionally neutral, basic-level
term shoe.
Brennan and Clark (1996) reported an experiment that challenges the
generality of this pragmatic expectation. In their experiment, speakers
played a game with listeners that required them to make reference to the
same pictures over and over in contexts requiring different levels of lexical
specificity. For instance, they referred to a shoe as the loafer multiple times
because it appeared in the constraining context of another shoe. But when
they were later presented with a test trial in which this same shoe appeared
without the constraining context of the other shoe, they continued to use
the specific term loafer even though it was now overly specific. Brennan and
Clark took this as evidence that the dyad had established a mutually ac­
cepted "conceptual pact" to refer to this particular shoe as the loafer. More­
over, they found that speakers attempted to continue to use the overly
specific term even when the test trial took place with a new listener who was
absent when the pact was established.
We conducted an experiment that was similar to Brennan and Clark's
(Barr & Keysar, 2002, Experiment 3) except that we examined listeners' ex­
pectations about the informativeness of speakers. Our experiment had a
similar design, in which listeners learned to expect that a speaker would re­
fer to a particular car as the sports car because it appeared recurringly within
the constraining context of a station wagon. Likewise, the speaker estab­
lished the precedent of referring to a particular flower as the carnation be­
cause it appeared in the context of a daisy. In a later posttest they saw a
picture containing only the car and the flower and listened to a speaker refer
to one of the objects. The question was, would listeners expect the terms
32 BARR AND KEYSAR

sports car and carnation which were overinformative in this context, or

would they expect car and flower which were sufficient? What they actually
heard was the word car. Note that if they expected to hear the overly specific
terms, they would tend to mistake the word car as the initial phonemic seg­
ment of the word carnation. By tracking listeners' eye movements, we found
that listeners expected the overly specific precedents, as revealed by more
fixations to the carnation than in a baseline condition. More important, we
found this expectation of the precedent to be equally strong among listeners
who performed the posttest with a new speaker, one who could not have
known of the precedent's existence.
The research on referring precedents is striking because it shows that in­
terlocutors consider their conversational experience more important than
prevailing norms of informativeness. Through interaction, speakers and lis­
teners are able to adapt their language use to suit their own purposes. By re­
lying on precedents, interactants reduce the complexity of the problem they
face. Speakers will face fewer options during lexical selection; hence, listen­
ers will benefit from less uncertainty in lexical identification. One reason
why speakers might use naming precedents and listeners might expect
speakers to use them is because they are part of their mutual knowledge. An­
other reason is because over the course of the interaction repeated use has
simply made the lexical item strongly available and the underlying concep­
tualization strongly entrenched. These explanations differ in that the for­
mer assumes that interaction creates changes at the metarepresentational
level, in terms of what is mutually accepted and mutually known, whereas
the latter assumes that the changes are occurring directly at the representa­
tional level. The fact that speakers and listeners both use precedents in ways
that violate mutual knowledge offer support for the latter interpretation.
Furthermore, this interpretation is consistent with other research that
shows how conversational interaction leads to conceptual convergence in
how people represent the content of the discourse (Garrod & Anderson,
1987; Markman & Makin, 1998). To be clear, we do not dispute the idea
that interacting speakers and listeners can and do keep track of what others
know. But because interaction naturally causes speakers and listeners to
similarly represent discourse information, we suggest that they need not al­
ways consult this metaknowledge to successfully communicate.
A second way in which interaction makes coordinating understanding
easier is that it enables people to engage in what we call the epistemic ex­
change, an interactive exchange that serves as a proxy for the direct compu­
tation of mutual knowledge. We propose that certain interactive episodes in
conversation represent instances of what Kirsh and Maglio (1994) termed
epistemic action—cases in which people gain information about the world
through direct action instead of through computing that information. In an
epistemic exchange, speakers or listeners proceed on the basis of an egocen­
2. MAKING SENSE 33

trie assumption and gather information about what their interlocutors know
through feedback, even though they could have directly computed the part-
ner's perspective. In the case of conversation, interaction provides interloc­
utors with ample opportunities to learn about what other people know or do
not know without having to expend effort computing it themselves.
For example, imagine you are sitting across the table from Henry, and be­
tween you there are two candles that you both can see. In addition, there is a
smaller candle that is obscured from Henry's view and is even smaller than
the two mutually visible candles. When Henry tells you to "pick up the small
candle," you could potentially compute the intended referent as the smaller
of the two visible candles, because you know that he does not know about
the even smaller hidden one. In Clark and Marshall's (1981) terms, only the
visible candles are physically co-present and part of your mutual knowledge.
However, you might simply pick up the smaller one without really thinking
or feel confused and ask, "Which candle?" It might be easier (and more ac­
curate) to get information about Henry's perspective from Henry himself,
even though you could derive the identity of the referent on the basis of your
mutual knowledge.
We revisited a set of data reported in Keysar et al. (2000) and looked for
evidence that addressees engage in such epistemic exchanges. We found
that under circumstances such as the one in this example, 27% of the time
listeners performed the following actions: (a) asking for clarification (10%),
(b) moving the small candle and then being corrected by the speaker (14%),
or (c) both (3%). In contrast, in our control condition wherein the object
corresponding to the hidden small candle was replaced with a nonreferent
(i.e., a glass), in the vast majority of cases addressees were able to go right for
the intended referent and required an interactive exchange only 6% of the
time. In summary, even when addressees are presented with clear cues to
what is mutually known, they often opt to resolve ambiguity by engaging in
an epistemic exchange rather than computing the referent themselves. Es­
pecially given our finding that people are poor estimators of what others
know, it makes a lot of sense for addressees to exploit the dynamics of inter­
action to distribute the burden of reference resolution rather than try to
compute it themselves.
Another dimension of conversation that gives interaction its dynamism
is multimodal communication. Despite the conventional terminology, a
speaker does more than just speak and a listener does more than just listen.
Speakers look and gesture as they speak. Listeners watch, nod, and make fa­
cial expressions as they listen. This background of multimodal activity pro­
vides interactants with a channel by which they can continually monitor
their level of mutual understanding, and one that is backgrounded so as not
to obtrude upon the official business of the conversation (Clark, 1996; Clark
& Brennan, 1991). When speakers witness an uncomprehending look from
34 BARR AND KEYSAR

a listener midsentence, they can choose to elaborate on or repair their utter­

ances. Listeners can nod and provide other back-channel information to
show their continuing attention and that they are following the thread of
the discourse. In other words, speakers and listeners not only communicate
in the traditional sense but use the backgrounded multimodal channel to
give constant evidence of their level of understanding. As with the interac­
tive dimension of conversation, the muitimodal character of conversation
makes things much easier on language users.
Kelly, Barr, Church, and Lynch (1999) showed that listeners can read a
speaker's pragmatic intention from their gestural behavior. In their study,
viewers were presented with a video clip in which two actors acted out an
everyday scenario that ended with a pragmatically ambiguous target utter­
ance such as it's hot in here. The target utterance could potentially be con­
strued as either a literal statement or an indirect request (i.e., to open the
window). In one condition, speakers pointed to an object while delivering
the target utterance, such as a closed window, that pertained to the intended
meaning of the indirect request. In another condition, speakers kept their
arms at their sides as they delivered it and maintained eye contact with the
addressee. In the former condition, listeners were far more likely to interpret
the utterance as an indirect request, showing that gesture can indexically
ground the meaning of utterances. Yet, not only did the gesture disambigu­
ate the speech, but the speech served to disambiguate the gesture—people
were better at identifying what the speaker was pointing to when they heard
the accompanying speech (which did not mention the object) than when
they simply saw the pointing gesture. These findings suggest that speech and
gesture work together to convey pragmatic meanings.
Another set of studies by Barr emphasizes how even a single vocal chan­
nel of communication can carry multiple dimensions of signals that enhance
conceptual and linguistic coordination. In one study, listeners learned a set
of novel color categories from a pretrained expert by viewing instances of
each category and hearing prerecorded labels from the speaker (Barr, 2003).
Listeners were able to detect the speaker's level of certainty in the classifica­
tion from paralinguistic cues such as filled pauses (e.g., um and uh), hesita­
tions, and rising or falling intonation. Given that people are more certain
about the classification of typical than atypical instances, the speaker's para-
linguistically conveyed certainty enabled listeners to differentiate good from
bad examples and thereby facilitated learning of the categories.
Finally, these same kinds of paralinguistic cues can guide listeners in the
identification of referents. Barr (2001) found that speakers produce differ­
ent "hesitation signatures" when they formulate descriptions of new refer­
ents as compared to when they retrieve established precedents to refer to old
referents. New referent signatures contained longer hesitations and were
more likely to contain a filled pause. A follow-up comprehension experi­
2. MAKING SENSE 35

ment found listeners to be sensitive to these signatures. When listeners

heard a description of a new referent that was preceded by a hesitation sig­
nature containing a filled pause, they were over 300-ms faster at compre­
hending the description than when the filled pause was replaced by
incidental noise. It is surprising that something as humble as an um can yield
such a large advantage to comprehension, and such a finding suggests that
listeners determine the identity of referents not only by what speakers say
but by the apparent effort that they put into saying it.
In this discussion of the interactional domain of language use, we at­
tempted to show how the interactive and multimodal processes of conversa­
tion simplify the coordination of mutual understanding by reducing the
corresponding burden of processes in the cognitive domain. Although many
of the same practitioners of the mutual knowledge theory have been the
prime champions of research on interaction and multimodal communica­
tion, what differs is our interpretation of the implications of this domain for
cognitive processing. Because these approaches have tended to eschew ex­
plicit discussion of mechanisms, they fail to fully appreciate the degree to
which interaction serves as a proxy for the explicit computation of a shared
perspective. Interaction serves as a vehicle not only for coordinating the
metarepresentations of language users but for the representations them­
selves. It affords speakers and listeners the opportunity to engage in
epistemic exchanges that help them learn about others' perspectives with a
minimum of effort. Finally, the multimodal nature of conversation provides
language users with immediate feedback and extra channels for communi­
cation. In short, the interactional domain enables language users to be more
egocentric because it distributes the work that must be done to achieve mu­
tual understanding over other processes in the environment.

The Cultural Domain

Whereas the interactional domain focuses on patterns of language use

within a dyad or group of interactants, the cultural domain looks at language
use through a wide-angle lens that encompasses the broader language com­
munity. We construe the culture as the repository of conventional practices
that the members of a community have in common. The cultural domain is
relevant to theories of pragmatics because it informs us about how much
language users can take for granted when they interact with other members
of their communities.
When interactants establish temporary patterns of language use to com­
municate effectively, they build on preestablished cultural patterns of lan­
guage use—specifically, linguistic conventions—that the broader language
community has shaped over time. The amount of work that language users
will need to do to understand one another, whether through direct compu­
36 BARR AND KEYSAR

tation of mutual knowledge or through interactive exchanges, will depend

on the degree to which they proceed from the same starting assumptions
about the meanings of conventions. If everyone had exactly the same
knowledge and experience, there would be little need for mutual knowledge
because people could count on others knowing what they know. Conversely,
if everyone had wildly different representations, communication would be
extremely difficult. The need for communities to develop common repre­
sentations among members presents a massive problem of social coordina­
tion. How do communities solve this problem? On what level of
commonality of representation can two average members count? And how
do these commonalities come about?
Traditionally, the field of pragmatics has sought to explain the establish­
ment and use of linguistic conventions as a product of the accumulation of
mutual knowledge among the members of a language community (Lewis,
1969). According to this view, individuals adhere to community-wide con­
ventions because they have a preference to conform to the practices of other
members of their community. Over the course of their experience, they
build up a representation of how the modal member of their community will
behave. Lewis stated, "If one has often encountered cases in which coordi­
nation was achieved in a certain problem by conforming to a certain regular­
ity, and rarely or never encountered cases in which it was not, he is entitled
to expect his neighbors to have had much the same experience" (p. 40). Mu­
tual knowledge gives individuals a justification for conforming to the con­
ventions in that it gives them reason to expect others to do the same. In
essence, what this view assumes is that there can be no conventions without
mutual knowledge.
Barr (in press) reported a series of multiagent computer simulations
that shows that mutual knowledge is not necessary for the establishment
and maintenance of semantic conventions in language communities. He
argued that such representations of communal knowledge are unneces­
sary because conventions emerge as by-products of dyadic-level mecha­
nisms of coordination. In the simulations, individual agents played a
simple signaling game with other agents in the community. Each agent
had a lexicon that mapped four symbolic forms onto four meanings. Ini­
tially, the form-to-meaning mappings were randomized for each agent.
During each round of the simulation, each agent played the signaling
game with a randomly selected partner from the community. The agent
would attempt to communicate a random sequence of four meanings to
its partner, and the partner would attempt to match the speaker's mean­
ings. Agents received feedback as to whether they were correct or incor­
rect, but they did not receive any information about the other agent's
mapping. Agents updated their mappings and then went on to play the
next round with new partners.
2. MAKING SENSE 37

Although these agents had no representation of what was going on in the

community, but just adjusted their lexicons based on their experiences dur­
ing a sequence of isolated interactions, the communities converged quite ro­
bustly to a single set of conventions. Even when they did not establish a
single system, they typically converged on several spatially organized signal­
ing systems or dialects, which are hallmarks of human language.
These simulations suggest that coordination can be achieved in language
communities as a by-product of the work that language users do in the dyad
rather than as the result of high-level calculations about what others know.
In addition, it demonstrates that the work that individual language users do
to coordinate with their language partners ultimately subserves the purpose
of making their representations ever more similar to other members of their
community. The fact that they can count on other members of their com­
munity having similar experiences with language greatly reduces the work
they must do when they speak to others who are like themselves.

CONCLUSION

We began this chapter with a discussion of the paradox of egocentrism in

language use. Although the underlying assumption of cooperation requires
language users to speak and understand against the background of their mu­
tual knowledge, language users appear to routinely disregard this knowl­
edge. They expect an addressee to perceive sarcasm even when the
addressee lacks crucial evidence about the speaker's attitude. They rou­
tinely consider hidden objects and private thoughts as the intended refer­
ents of speaker's expressions. They expect speakers to follow linguistic
precedents that were established by another speaker. All of these findings
show that language users are not designed to the exacting standards of prag­
matic theories. According to these theories, language users' egocentric be­
havior does not make sense.
We argued that to make it make sense, we need to look not at the nature of
the problem that language users must solve but at the structure of the envi­
ronment in which language use is embedded. The environment can be di­
vided into three subdomains, which is like looking at a single picture through
lenses of different powers of magnification. Our research has focused on the
cognitive domain, and it has uncovered limitations on language users' ability
to effectively deploy mutual knowledge when processing single utterances. If
we are to take these limitations seriously, then we must look elsewhere to find
mechanisms that can compensate for these limitations.
The finding of egocentrism in language use need not imply that language
users are somehow not adequately designed for the purpose of coordinating
understanding. In fact, the processing system's rapid use of available infor­
mation is exquisitely tuned to provide maximally efficient processing given
38 BARR AND KEYSAR

the interactional and cultural domains in which it operates. In essence,

what we are claiming is that language users can get away with a large degree
of egocentrism because the work of achieving shared understanding is dis­
tributed over interactive, multimodal, and cultural processes in the envi­
ronment. Interactive processes such as the epistemic exchange allow
language users to discover what other people know through negotiation
without having to compute it themselves. Multimodal channels of commu­
nication enable language users to simultaneously convey and perceive mul­
tiple dimensions of meaning, including online feedback about their level of
understanding. Cultural processes serve to indirectly coordinate the repre­
sentations of the individuals in a community, greatly reducing the work they
must do in the dyad. In essence, language users can be simpler than prag­
matic theories require because part of the burden of coordination is distrib­
uted over the domains of dyadic and communal interaction.
To be clear, we do not intend to cast doubt on the possibility of mutual
knowledge either as a theoretical construct or as a factor that is operative in
conversation. Yet, we wish to emphasize that whether or not a speaker or lis­
tener uses mutual knowledge on a particular occasion is an empirical ques­
tion. The coincidence of perspectives is not a fortuitous event but rather a
direct consequence of the operation of background processes in the lan­
guage use environment. Thus, the mere observation that a speaker produces
an utterance that is in alignment with mutual knowledge does not warrant
the inference that she or he directly computed that knowledge as mutual at
any time. The speaker may have or may have simply used information that
was simultaneously available and salient to him or her and the interlocutor.
Moreover, our purpose is not to question people's ability to compute
metarepresentations about what others know. Yet, our results imply that the
cognitive system is designed to operate efficiently on representations, not
metarepresentations. However, this does not preclude the use of such higher
order representations to monitor and correct problems that arise during
conversation. Thus, speakers' and listeners' ability to compute mutual
knowledge might be operative primarily when communication fails. Ac­
cording to our findings, it appears that mutual knowledge is most likely to be
implemented as a mechanism for detecting and correcting errors instead of
an intrinsic, routine process of the language processor.
Faced with the findings reviewed so far, we think it is important to rethink
exactly what it means to be cooperative, a concept that is at the heart of
most theories of language use. For one, the supposition that speakers strive
to be maximally informative in lexical selection does not seem to fit what
they actually do. Perhaps a better description of what they do is simply rely
on their past and current discourse experience and select the term that is
most strongly available to them. As the simulations by Barr (in press) show,
one's own conversational experience can often be a reliable guide to what is
2. MAKING SENSE 39

conventional in the community. That said, there are probably social polite­
ness norms that will cause speakers to be more careful about their lexical se­
lection (e.g., whether to refer to a woman as my mother, mom, or Mrs. Smith).
Just when and how speakers will heed social circumstances during lexical se­
lection are questions that warrant further investigation.
The presence of epistemic exchanges, hitherto unnoted in the literature,
also calls for us to rethink the notion of cooperation. If speakers and listeners
really had as strong expectations of cooperation as the theory says they
should, then they would routinely be confronted with behavior that seemed
uncooperative. Perhaps language users tolerate some slack over the short
run because it is the most effective way to share the burden of coordinating
understanding over the long run. Perhaps it is not through the individual
sentence by which language users demonstrate they are cooperative, but
rather it is how they behave over the course of the conversation.
In closing, the inherent ambiguity of language creates a complex problem
for language users and seems to call for correspondingly complex mecha­
nisms that enable language users to successfully communicate in the face of
such vast uncertainty. Our message is that this complexity should not be
sought within the head of the individual language user but rather in the cog­
nitive, interactional, and cultural domains that comprise the environment
of language use. Language users can be simpler than theories require be­
cause the environment is more complex than these theories envision. It is
only by considering the behavior of language users against this background
that we can truly make sense of how we make sense.

REFERENCES

Ariel, M. (1988). Referring and accessibility. Journal of Linguistics, 24, 65-87.

Barr, D. J. (1999). A theory of dynamic coordination for conversational interaction. Un­
published doctoral dissertation.
Barr, D. J. (2001). Paralinguistic correlates of discourse structure. Paper presented at the
43rd annual meeting of the Psychonomic Society, Orlando, FL.
Barr, D. J. (in press). The emergence of conventions in language communities. In J.
C. Trueswell &M. K. Tanenhaus (Eds.), Approaches to study ing world-situated lan­
guage use:Bridging the language-as-product and language-as-action traditions. Cam­
bridge, MA: MIT Press.
Barr, D. J. (2003). Paralinguistic correlates of conceptual structure. Psychonomic Bul­
letin & Review, 10, 462-467.
Barr, D. J., & Keysar, B. (2002). Anchoring comprehension in linguistic precedents.
Journal of Memory and Language, 46, 391-418.
Brennan, S. E., & Clark, H. H. (1996). Conceptual pacts and lexical choice in con­
versation. Journal of Experimental Psychology: Learning, Memory, and Cognition,
22, 1482-1493.
Brown, R. (1958). How shall a thing be called? Psychological Review, 65, 14-21.
4O BARR AND KEYSAR

Chafe, W. L. (1976). Givenness, contrastiveness, definiteness, subjects, topics, and

point of view. In C. N. Li (Ed.), Subject and topic (pp. 26-55). New York: Aca­
demic Press.
Clark, H. H. (1996). Using language. Cambridge, England: Cambridge University Press.
Clark, H. H. (1998). Communal lexicons. In K. Malmkjaer & J. Williams (Eds.),
Context in language learning and language understanding (pp. 63-87). Cambridge,
England: Cambridge University Press.
Clark, H. H., & Brennan, S. E. (1991). Grounding in communication. In L. B.
Resnick, J. M. Levine, &. S. D. Teasley (Eds.), Perspectives on socially shared cogni­
tion (pp. 127-49). Washington, DC: American Psychological Association.
Clark, H. H., & Carlson, T. B. (1981). Context for comprehension. InJ. Long & A.
Baddeley (Eds.),Attention and performance IX (pp. 313-330). Hillsdale, NJ: Law­
rence Erlbaum Associates.
Clark, H. H., & Fox Tree, J. E. (2002). Using uh and um in spontaneous speaking.
Cognition, 84, 73-111.
Clark, H. H., & Marshall, C. R. (1981). Definite reference and mutual knowledge.
In A. K. Joshe, B. L. Webber, & I. A. Sag (Eds.), Elements of discourse understand­
ing (pp. 10–63). Cambridge, England: Cambridge University Press.
Clark, H. H., & Murphy, G. L. (1982). Audience design in meaning and reference.
In J. F. Le Ny & W. Kintsch (Eds.), Language and comprehension (pp. 287-299).
Amsterdam: North-Holland.
Clark, H. H., Schreuder, R., & Buttrick, S. (1983). Common ground and the under­
standing of demonstrative reference. Journal of Verbal Learningand Verbal Behav­
ior, 22, 245-258.
Clark, H. H., & Wilkes-Gibbs, D. (1986). Referring as a collaborative process. Cog­
nition, 22, 1-39.
Cruse, D. A. (1977). The pragmatics of lexical specificity. Journalof Linguistics, 13,
153–164.
Epley, N., & Gilovich, T. (2001). Putting adjustment back in the anchoring and ad­
justment heuristic: Divergent processing of self-generated and experimenter-
provided anchors. Psychological Science, 12, 391-396.
Garrod, S., & Anderson, A. (1987). Saying what you mean in dialogue: A study in
conceptual and semantic co-ordination. Cognition, 27, 181-218.
Gerrig, R. J. (1986). Process models and pragmatics. In N. E. Sharkey (Ed.), Ad­
vances in cognitive science (Vol. 1, pp. 23–42). New York: Wiley.
Greene, S. B., Gerrig, R. J., McKoon, G., & Ratcliff, R. (1994). Unheralded pro­
nouns and management by common ground. journal of Memory and Language,
33,511-526.
Grice, H. P ( 1975). Logic and conversation. In P Cole & J. Morgan (Eds.), Syntax
and semantics: Vol. 3. Speech acts (pp. 41–58). New York: Academic Press.
Gundel, J. K., Hedberg, N., & Zacharski, R. (1993). Cognitive status and the form of
referring expressions in discourse. Language, 69, 274-307.
Horton, W. S., & Keysar, B. (1996). When do speakers take into account common
ground? Cognition, 59(1), 91–117.
Kelly, S. D., Barr, D. J., Church, R. B., & Lynch, K. (1999). Offering a hand to prag­
matic understanding: The role of speech and gesture in comprehension and
memory. Journal of Memory and Language, 40(4), 577-592.
2. MAKING SENSE 41

Keysar, B. (1994). The illusory transparency of intention: Linguistic perspective

taking in text. Cognitive Psychology, 26(2), 165-208.
Keysar, B. (1997). Unconfounding common ground. Discourse Processes, 24,
253-270.
Keysar, B., & Barr, D. J. (2002). Self-anchoring in conversation: Why language users
don't do what they "should." In T. Gilovich, D. W Griffin, & D. Kahneman
(Eds.), Heuristics and biases: The psychology of intuitive judgment (pp. 150-166)
Cambridge, England: Cambridge University Press.
Keysar, B., Barr, D. J., Balin, J. A., & Brauner, J. S. (2000). Taking perspective in con­
versation: The role of mutual knowledge in comprehension. Psychological Sci­
ence, 11, 32-38.
Keysar, B., Barr, D. J., Balin, J. A., & Paek, T. S. (1998). Definite reference and mu­
tual knowledge: Process models of common ground in comprehension. Journal of
Memory and Language, 39, 1-20.
Keysar, B., & Bly, B. (1995). Intuitions of the transparency of idioms: Can one keep a
secret by spilling the beans? Journal of Memory and Language, 34, 89-109.
Keysar, B., & Henly, A. S. (2002). Speakers' overestimation of their effectiveness.
Psychological Science, 13, 207-212.
Kirsh, D., & Maglio, P. (1994). On distinguishing epistemic from pragmatic action.
Cognitive Science, 18, 513-549.
Lea, R. B., Mason, R. A., Albrecht, J. E., Birch, S. L, & Myers, J. L. (1998). Who
knows what about whom: What role does common ground play in accessing dis­
tant information? Journal of Memory and Language, 39, 70-84.
Lewis, D. (1969). Convention: A philosophical study. Cambridge, MA: Harvard Uni­
versity Press.
Markman, A. B., & Makin, V. S. (1998). Referential communication and category
acquisition. Journal of Experimental Psychology: General, 127, 331—354-
McNeill, D. (1992). Hand and mind: What gestures reveal about thought. Chicago:
University of Chicago Press.
Sacks, H., Schelgloff, E. A., & Jefferson, G. (1974). A simplest systematics for the
organization of turn-taking for conversation. Language, 50, 696-735.
Simon, H. A. ( 1996). The sciences of the artificial (3rd ed.). Cambridge, MA: MIT
Press.
Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and
biases. Science, 185, 1124-1131.
This page intentionally left blank
 3

Contextual Expressions

and Common Ground

Richard J. Gerrig
William S. Horton
State University of New York at Stony Brook

In his autobiographical account of his rise to the top of corporate Amer­

ica, Jack Welch revealed that one of his favorite expressions during busi­
ness discussions was, "Don't Walter Cronkite me!" (Welch, 2001, p. 43).
Fortunately, Welch immediately provided an interpretation of his some­
what idiosyncratic meaning: "That [the utterance] was understood by
everyone to mean: 'You report the bad news, but you don't tell me how
you're going to fix it.' " It was not, presumably, Walter Cronkite's respon­
sibility as a newscaster to do more than report the news. Thus, the special
meaning Welch wished his underlings to take from the verb phrase "to
Walter Cronkite" needed to have been acquired through interactions
with Welch himself.
In this chapter, we discuss contextual expressions—the class of linguistic
phenomena to which "to Walter Cronkite" belongs. We begin by delimit­
ing this class of expressions and describing how, as in the case of "to Walter
Cronkite," speakers produce them against the background of appropriate
assessments of the common ground shared with their addressees. We then
discuss the ways in which addressees understand these types of expres­
sions. We structure this latter discussion around the questions that have
dominated research on other types of pragmatic phenomena. The conclu­
sion we draw is that addressees are well-prepared to understand the types
of contextual expressions that arise frequently in social interactions.

43
44 GERRIG AND HORTON

COMMON GROUND AND THE PRODUCTION

OF CONTEXTUAL EXPRESSIONS

The verb phrase "to Walter Cronkite" fails to appear in a standard dictio­
nary. Rather than being a conventional part of the language, the phrase
takes on meaning by virtue of the discourse context in which it appears.
This is the hallmark of contextual expressions: "Their senses depend en­
tirely on the time, place, and circumstances in which they are uttered" (H.
H. Clark, 1983, p. 300). As such, each contextual expression has an infi­
nite number of potential meanings. Jack Welch's use of "to Walter
Cronkite" provides a good example of this infinite flexibility. Most epony­
mous verb phrases, the category of language to which "to Walter Cronkite"
belongs, are based on salient acts of the eponym (H. H. Clark & Gerrig,
1983; Rapp & Gerrig, 1999). This is true of Welch's use of "to Walter
Cronkite" (i.e., Cronkite did deliver the news), but his meaning tran­
scends that starting point. We see immediately how speakers, by using
proper contexts to focus attention, can weave infinite variations on the sa­
lient acts associated with a famous individual. For "to Walter Cronkite,"
we also might think of "to show great empathy in the face of a national ca­
lamity" or "to host the Kennedy Centers Honors."
Eponymous verb phrases are one of a wide variety of contextual expres­
sions (H. H. Clark, 1983). Because we provide tokens throughout the chap­
ter, we only give a small number of examples here. We wish to give a brief
sense of the range of these expressions, from those that blend in to those that
call attention to themselves:

• Recently—thinking a mountain vacation would cheer me—I sublet my

apartment to a handsome but somber newly wed couple, who turned
out to be every bit as responsible as I'd hoped (noun-noun combina­
tion, italics added, Jen, 1999, p. 40).
• In between those times, Pino's was home to a few regular drunks, and
the kitchen was taken over by cockroaches as big as barn cats. I ate at
Pino's in spite of the roach rumor because Anthony Pino made the best
pizza in Trenton (noun-noun combination, italics added, Evanovich,
1999, p. 45).
• Tm going door-to-door with Fred's photo,' I told her [Lula]. 'Want
to help?'
'Sure, you just give me one of those posters, and I'll door-to-door the
hell out of you' (denominal verb, italics added, Evanivich, 1999, p. 132).
• All he knew was that, assuming, to begin with, that she was willing, he
couldn't sleep with a woman like Cindy and then leave her flat. She
could you folk him, he could never us folk her (denominal verb, Jen,
1999, p. 32).
3. CONTEXTUAL EXPRESSIONS AND COMMON GROUND 45

Each of these examples provides circumstances in which a contextual ex­

pression fills a gap in the mental lexicon. For that reason, we also use the
term lexical innovation to describe many tokens of contextual expressions.
We left one of these examples mysterious on purpose: What precisely do
the verb phrases "to you folk" and "to us folk" mean? It is likely to be the case
that a reader could get a general sense of what these expressions mean, but a
more precise understanding requires an accumulation of experiences over
the course of a short story. In the story, an Asian-American businessman,
Art, has accidentally reserved a room at a welfare hotel. An African-Ameri-
can woman, Cindy, who lives in the hotel comforts him after he has been
knocked unconscious (Jen, 1999, p. 30):

'Sure musta been a shock,' she said, 'End up in a place like this....'
'What about you? It's no place for you, either, you and your kids.'
'Maybe so,' she said. 'But that's how the Almighty planned it, right?
You folk rise up while we set and watch.' She said this with so little ran­
cor, with something so like intimacy, that it almost seemed an invita­
tion of sorts.

Somewhat later, Art reflects on Cindy's utterance (p. 31):

You folk. What folk did Cindy mean?

This reflection eventuates in Art's (mental) assertion, "She could you folk
him, he could never us folk her."
This example illustrates the way in which contextual expressions rely on
the common ground shared between the speaker and addressee. That is, the
meanings of such expressions can only be determined when addressees take
into account the appropriate knowledge mutually known to both interlocu­
tors. In this literary example, the reader is not exactly an addressee, but the
author still wrote with an expectation that the expressions have shared expe­
riences as their foundation (Gerrig, 1993). More generally, common ground is
established between individuals either as the result of common membership
in particular sociocultural groups or on the basis of joint personal experiences.
H. H. Clark (1996) described this distinction as one between communal com­
mon ground and personal common ground; both types of common ground are
important for understanding particular contextual expressions.

Communal Common Ground and Contextual Expressions

Communal common ground, which we also call community membership, is
the shared body of knowledge possessed by individuals who belong to partic­
46 GERRIG AND MORTON

ular cultural communities. These communities can be reasonably well-de-

fined, such as "members of the psychology department" and "Canadians," or
they may exist on a more ad-hoc basis, such as "people who were standing on
a subway platform while a performer sang 'The Way You Look Tonight.' " H.
H. Clark (1996) listed a number of types of communities based on every­
thing from nationality to occupation to political stance to hobbies. Every in­
dividual is simultaneously a member of multiple social and cultural groups,
and as a result, every individual shares partially overlapping sets of common
ground with every other individual who belongs to the same groups.
Community membership is important for providing the background for
many types of language use. Within particular communities, for example, it
is often assumed that all of the members of that group will have knowledge of
a certain vocabulary or jargon (e.g., psychology jargon such as dependent
variable or ANOVA). In these cases, the specialized vocabulary in use within
a community has a set of meanings and functions like any other set of lexical
items. Contextual expressions, by contrast, put the knowledge mutually
known to community members to use in innovative, unconventional ways.
The example with which we began, "to Walter Cronkite," provides an excel­
lent example of a contextual expression that relies on community membership.
Understanding of this phrase requires, to start, membership in the community
of individuals who remember that Cronkite was a news anchor. Speakers should
only use this phrase when they have good reasons to believe a priori that their
addresses are members of that community. Consider another eponymous
phrase uttered by a character on the situation comedy "Undeclared": "You
better shut up man, or I'm gonna Van Damme your head off." Although this
statement can be generally understood as a threat, its particular meaning will be
clear only to individuals belonging to the community who are aware that
Jean-Claude Van Damme is the star of several violent martial-arts movies. The
sitcom's writers presumably believed that the members of their target audience
were overwhelmingly members of the appropriate community.
In fact, because contextual expressions have shifting senses, the mean­
ings of particular eponymous verb phrases may differ from community to
community and may change within a community over time. For example,
there was probably a moment in time when "to do an O.J." meant to dash
athletically through an airport. As of 1995, however, "to do an O.J." took on
divergent new meanings (cited in Rapp &Gerrig, 1999, p. 613):

• From an essay by Martin Amis: A coinage has forged itself within the
media community of the West Coast: "O.J.," as a verb. Thus, "to O.J."
Or, passively (and much, much more commonly), "to be O.J.-ed" or "to
get O.J.-ed." "O.J.-ing," generally, has nothing to do with sports or mov­
ies, or sexual jealousy. It has to do with media reschedulings caused by
extra coverage of the O.J. Simpson trial.
3. CONTEXTUAL EXPRESSIONS AND COMMON GROUND 47

"People are always getting O.J.-ed off of things," explains Kathi

Goldmark, my media escort in San Francisco.
"So for example you'd say ... ?"
" 'Norman Mailer was going to do national TV, but he got O.J.-ed
off of it.'"
"I glance at my schedule and say, "Look! I'm meant to do a radio inter­
view at eleven-thirty. Live. But it says here they'll tape it if I get
O.J.-ed."
• From a letter in response to the original essay: Martin Amis ... notes that
in Los Angeles "O.J." has become a verb .... In New Jersey, this verb
has a different meaning. Recently, I was standing in line at a local deli
behind a customer who asked the counterman to "O.J." his sandwich.
Without a word of inquiry, the employee lifted his knife and cut the
sandwich in half.

We suspect that, given the divergence of opinion with respect to O.J.

Simpson's guilt, the phrase "to do an O.J." was used quite variously. Such is,
of course, the nature of contextual expressions.
Eponymous phrases are a class of contextual expression that clearly re­
quire consideration of information held in common between speakers and
addressees. Moreover, understanding such phrases engages a hierarchy of
beliefs on the part of addressees about the ways in which speakers have as­
sessed background information (H. H. Clark &.Gerrig, 1983). Each of these
beliefs makes reference to information that is in communal common
ground. First, there is the identity of the eponym. If an addressee has no idea
who Jean-Claude Van Damme is, then the expression "van Damme your
head off" will not have its full intended effect. Next, particular acts of the
eponym must be in common ground, and there must be one or more acts
that are relevant to the situation. Jean-Claude Van Damme has a number of
acts that are potentially associated with him and known to his fans, includ­
ing "bad acting" and "showing his bare chest a lot," but presumably only
something like "bloodying opponents through kickboxing" is relevant in a
threatening context. Finally, the speaker must assume that the addressee
will be able to identify the relevant act upon hearing the eponymous verb
phrase. Thus, when speakers produce a phrase such as "Van Damme your
head off" they are signaling their own belief that they have made an appro­
priate judgment about the community knowledge—with an appropriate
level of detail—that they and their addressees have in common.
In this section, we focused on phrases involving eponyms because they
appear to be circumstances in which contextual expressions engage com­
munity membership. Note, in addition, that speakers also make assessments
of community membership as a part of the process through which words or
phrases that began as contextual expressions become conventionalized (E.
48 GERRIG AND MORTON

V Clark & H.H.Clark, 1979; Gerrig &Gibbs, 1988). For example, within
his community Jack Welch's utterance "Don't Walter Cronkite me!" pre­
sumably had a meaning that had achieved the status of a conventional id­
iom. In general, speakers must be aware that expressions that function as
conventional in some communities may still be contextual in others. Con­
sider this example (Michaels, 2001, p. 1ll):

[Norbert] is always broke, but he doesn't think about getting a job.

He schemes day and night. And he dollars me. You know the expres­
sion? 'Nachman, lend me a dollar.' He never pays me back.

By uttering "You know the expression?" this speaker evokes a community

for which "to dollar" can be conventionally understood to mean "ask for a dol­
lar" but suggests that his addressee may not be part of it. Thus, speakers must
make judgments of community membership to determine whether they must
continue to provide context for what began as contextual expressions.

Personal Common Ground and Contextual Expressions

Personal common ground is the set of information shared between two (or
more) individuals arising from their joint perceptual or physical experiences,
like conversations (H. H. Clark, 1996). In the parlance of H. H. Clark and
Marshall (1981), personal common ground is based on physical or linguistic
copresence, and it is built up over the course of interactions between individu­
als. Suppose an introductory psychology lecture is interrupted by a loud bang
outside the classroom. When the professor utters, "What was that?" he is
making the assumption that, by virtue of physical copresence, the reference of
"that" will be clear. Suppose the same professor has caught two students
cheating on an extra-credit assignment. When the professor informs his col­
league, "It's happened again," he is making the assumption, by virtue of lin­
guistic copresence, that his addressee will recall their conversation about the
ubiquity of cheaters. Both types of personal common ground provide ample
opportunities for speakers to produce contextual expressions.
With respect to physical copresence, contextual expressions often serve
the function of identifying some object or activity in the environment. For
example, Downing (1977) provided an anecdote about a friend who was
"instructed to sit in the apple-juice seat" (p. 818). The intended meaning of
the phrase—"the seat in front of which a glass of apple juice had been
placed"—was clear by virtue of physical copresence. Downing observed that
the constraints on these types of deictic noun-noun combinations "appear
to be less severe than those which govern compounds suitable as long-range
category labels, usable and interpretable from situation to situation" (p.
819). That is, when speakers create novel noun-noun combinations of this
3. CONTEXTUAL EXPRESSIONS AND COMMON GROUND 49

sort, they need only call attention to salient physical features of the immedi­
ate context of use. Still, it is possible to find contextual expressions based on
physical copresence that function more conceptually. Consider these exam­
ples from an Olympic gymnastics sportscast:

• He cowboyed it.
• He has to cowboy it a little bit.

In the original contexts, the meaning of the two uses of "to cowboy" were
made clear by virtue of physical copresence: In each case, the gymnast had
bowed his legs a bit to facilitate rotation.
Speakers also routinely avail themselves of linguistic copresence to pro­
duce contextual expressions that name both ephemeral and more perma­
nent concepts. Consider this line from a short story: "And there she is, the
condiment gal" (Labute, 2001, p. 121). The author presumably expected
readers to be able to understand "condiment gal" because of the story's
opening episode: "Now there's a nice lay, he thinks to himself, as he politely
sidesteps and lets her out of line so she can carry her food over to the condi­
ment island" (p. 120). "Condiment gal" succeeds as a referring phrase be­
cause of linguistic copresence, but it does not name any intrinsic property of
the gal
Other contextual expressions appear to function via linguistic
copresence to create a new conceptual structure. This was true of examples
we cited earlier (e.g., roach rumor and "you folk"). Consider an additional ex­
ample (Evanovich, 1997, pp. 6-7).

'Well, of course you can do it,' I said. 'It's just that this situation is
sort of.. . delicate.' ...
'Hell,' she said, stuffing herself into her jacket. 'I can delicate your
ass off.'

In this case "to delicate" seems to be a good candidate for a deadjectival

verb that could potentially endure over time as a standard of behavior that
the speaker was or was not able to meet.

Conversation and the Emergence of Contextual Expressions

Although we largely cited literary examples, conversational tokens of con­

textual expressions assume the same functions. In the most interesting
cases, they emerge in conversations to serve as labels for concepts that are
not otherwise lexicalized. So, for example, as a young girl watched someone
remove a bicycle helmet before entering a grocery store, she volunteered,
5O GERRIG AND MORTON

"It's okay to wear it. It's a store helmet." In this construction store helmet pro­
jects a conceptualization of the world in which helmets can be discriminated
as to their store worthiness.
Because these contextual expressions name unlexicalized concepts, they
often represent a speaker's attempt to negotiate a perspective with an ad'
dressee. In some of our experimental work, we observed how contextual ex­
pressions unfold in time, as speakers and addressees collaborate jointly to
develop expressions that aptly encode joint perspectives. Our corpus of con­
versational data comes from a referential communication task in which par­
ticipants playing the roles of directors and matchers refer repeatedly to a set
of picture cards as part of a card-matching task (Gerrig & Horton, 2001). In
the opening round we often see overt negotiation with respect to particular
cards—here, a fish:

Director (D): A multicolored fish.

Matcher (M): Does it have like red lines?
D: Blue, red- it looks like a flag or whatever.

The second round still involves collaboration:

D: The multicolored fish.

M: Like the flag?
D: The flag yeah.

In subsequent rounds, the expression settles in and becomes truncated. The

third time the director describes this fish the utterance is "The multicolored
American flag fish." The final time, the phrase became "The American flag
one." Other directors and matchers negotiated other expressions for the
same fish, such as "the rainbow fish." We see in these examples the process
by which speakers propose contextual expressions to label novel concepts
and addressees participate in accepting or modifying those proposals. In that
way, contextual expressions emerge from linguistic copresence with the po­
tential of functioning beyond their original contexts.
We now completed a brief survey of the origins and functions of contex­
tual expressions. In each case, we showed that contextual expressions are
deeply social in their origins. Speakers produce contextual expressions
against the background of appropriate common ground. That is not to say
that speakers are always correct in their judgments. Because assessments of
common ground rely on speakers' and addressees' memory processes, the
quality of those assessments will depend on the quality of those memory pro­
cesses (Gerrig & Horton, 2001). We could, for example, gloss our earlier ex­
ample "And he dollars me. You know the expression?" as an instance in
3. CONTEXTUAL EXPRESSIONS AND COMMON GROUND 51

which the speaker decided, through monitoring of his language production,

that "dollars me" might not be sufficiently comprehensible without some
help. Still, the query "You know the expression?" reasserts the social func­
tion of language. We now turn our attention from the social circumstances
in which speakers produce contextual expressions to the mental processes
that allow addressees to understand them.

PROCESSES OF UNDERSTANDING CONTEXTUAL

EXPRESSIONS

The hallmark of pragmatic phenomena is that speakers' meanings depart

from the meanings of the words they utter. Consider this exchange (Dennis,
1958, pp. 13-14):

"Chilly out, dear," I said, kissing her. "Anything in the mail? I mean
like especially terrible Christmas cards."
Pegeen knew perfectly well what I meant and went on to say so. "I
know perfectly well what you mean. You mean is there some word from
our child or from that madwoman who carried him off. And the answer
is No. Just as it's been every day for the last four months. No! No! No!"

In this excerpt, common ground provides the link between what the narra­
tor says and what his wife, Pegeen, correctly understands him to mean. Re­
search on pragmatics has largely been directed toward uncovering the
processes and representations that allow addressees to understand what
particular speakers mean on particular occasions. With respect to our analy­
sis of the production of contextual expressions, we also pose the questions of
process and representation: With what time course are addressees able to
use common ground to narrow down contextual expressions' meaning from
infinite possibilities to (in most cases) a single possibility? And to what ex­
tent do representations of out-of-context meanings affect this time course?

How Powerfully Does Common Ground Constrain Meaning?

As we just noted, an important topic of research on pragmatic phenomena

has been the relationship between speakers' words (i.e., literal meanings)
and the meanings those words accrue when addressees bring common
ground to bear (i.e., speakers' meanings). Consider an indirect request, such
as "Could you please pass the salt?" Under appropriate circumstances, an
addressee might respond "Sure" and then pass the salt. Such a response sug­
gests that the addressee is aware both of the literal, direct force of the utter­
ance (i.e., a question of ability) and the indirect force (i.e., a request for
action; H. H. Clark, 1979). However, what is the relationship between these
52 GERRIG AND HORTON

two types of meaning? We consider, in turn, two aspects of this question that
have been addressed by researchers on pragmatics. (Because it is not en­
tirely clear how the concept of literal meaning applies to words, for contex­
tual expressions we speak of out-of-context meanings rather than literal
meanings.)

Do Out-of-Context Meanings Have Temporal Primacy? Because

most theories of comprehension have been theories of literal meaning, they
have most often made an implicit or explicit assumption that addressees re­
cover literal meaning as a necessary step along the way to recovering
speaker's meaning. On these accounts, if one examines the time course of
comprehension, the literal meaning also comes first. What has come to be
called the standard pragmatic model suggests just such temporal primacy
(Grice, 1975; Searle, 1979). According to this model, when a speaker per­
forms an indirect request (or uses irony, metaphor, and so on), the addressee
must construct the literal meaning and determine whether it is the entirety
of the speaker's intended meaning before doing anything else. As such, ad­
dressees should take less time to understand utterances conveying just lit­
eral meanings.
The temporal predictions of the standard pragmatic model have been
tested and disconfirmed with great regularity (for a review, see Gibbs, 1994).
For example, Gibbs (1979) contrasted reading times for utterances used ei­
ther as literal or indirect requests. Gibbs recorded readers' times to read the
target sentences ("Must you open the window?") and to verify a paraphrase
of the intended literal or indirect meaning. The sum of these times was reli­
ably shorter for the indirect requests than for the literal requests—a clear
contradiction of the prediction that literal meanings have temporal primacy.
These types of experiments have allowed researchers to rule out the stan­
dard pragmatic model as a general account of language processing. The re­
sults suggest that addressees are swiftly able to put common ground—in this
case, story context—to use to arrive at appropriate understandings of speak­
ers' utterances.
Contextual expressions provide an opportunity to assess the time course
with which addressees engage common ground at the level of the lexicon.
Consider this sentence: "She had never done so much wadding as she did
now, packing up to move from the fun house" (Jen, 1999, p. 193). The
phrase "fun house" has a conventional meaning as an attraction at an
amusement park. However, in this particular discourse setting "fun house"
referred to a nonamusement park house and "fun" was intended ironically
Gen, 1999, p. 182):

It was at least with a feeling of mutuality that they started to call the
house the fun house ....
3. CONTEXTUAL EXPRESSIONS AND COMMON GROUND 53

—'It mocks us,' Sven said, 'with its air of being at our service, when if
fact we live in service to it.'

"Fun house" provides an example of apreempting innovation—contextual ex­

pressions that are identical to conventional expressions (Gerrig, 1989).
Such preempting innovations are relatively common particularly because
most instances of metonymy qualify. For example, in its original conversa­
tional contest the utterance "Let me give you your orange juice" was used to
mean "Let me give you a coupon for free orange juice."
How do addressees understand preempting innovations? There are at
least two possibilities. The first possibility—the error recovery model—
gives temporal primacy to the conventional meaning. In parallel to the
standard pragmatic model, the error recovery model suggests that address­
ees first process the conventional meaning fully. Thus, the model would
anticipate that readers would access the lexicalized meaning of "fun
house" and try to create a representation of the utterance in which the
conventional meaning could function appropriately. Only when the con­
ventional meaning proved to be in error would the reader engage other
processes to recover. An alternative possibility—the concurrent processing
model—suggests that processes of developing conventional and innova­
tive meanings function in parallel.
To contrast these models, Gerrig (1989) wrote a series of stories that were
based on pairs of noun-noun combinations such as "door man" and "cave
man." For each pair, one story ended with an utterance in which the
noun-noun combination was used in its conventional sense:

A movie company is making a new multimillion dollar film. This after­

noon they are supposed to film a strange dream sequence. At the be­
ginning, a Neanderthal man is seen standing by a cave. Suddenly he
becomes the man guarding the door of a posh hotel. The director real­
ized that no one had been hired to play this part. He said, "We really
ought to get the door man."
(or:) He said, "We really ought to get the cave man."

A second story created innovative meanings for each combination:

The Fine Arts Department was hiring a new faculty member. The de­
partment needed people in two areas. They no longer had faculty who
taught about caves or doors. Unfortunately they could afford only one
new professor. Professor Rogers had an opinion about the best choice.
He said, "We really ought to get the door man."
(or:) He said, "We really ought to get the cave man."
54 GERRIG AND HORTON

For the conventional uses, it should be the case that the one or the other
of the two noun-noun combinations will be easier to understand (because,
e.g., one of the two combinations has a higher frequency of use). As such,
the error recovery model makes two predictions. First, innovative uses of
the combinations should inevitably take longer to understand than do
conventional uses (because error recovery presumably takes time). Sec­
ond, the time to understand a preempting meaning for a relatively slow
conventional use should also be relatively slow. Suppose, for example,
readers take longer to understand the utterance using the conventional
sense of door man than the parallel utterance using cave man. Assume that,
all other things being equal, processes beyond conventional meaning (e.g.,
error detection) add a standard increment to comprehension time. In that
case, readers should also take longer to understand door man in its innova­
tive use. Thus, the ordering of reading time (e.g., for the uses of door man
and cave man) should remain stable across the conventional and innova­
tive contexts. By comparison, the concurrent processing model predicts
no particular relationship between the conventional and innovative uses
of the combinations.
Participants in the experiment read 20 stories each of which included a
compound noun with a conventional or innovative meaning (i.e., partici­
pants read only one of the four versions for each pair) as well as 20 filler sto­
ries. The reading times for the sentences with the noun-noun combinations
contradicted the predictions of the error recovery model. Gerrig (1989)
sorted the pairs of combinations to gather together those that were rela­
tively fast (mean reading time = 2.10s) and relatively slow (mean = 2.53 s)
for conventional uses within each pair. The error recovery model suggests
that the same difference should be observed for innovative uses. In fact, the
pattern was reversed. Innovative uses based on "fast" conventional uses
took more time to read (mean = 2.97 s) than innovative uses based on
"slow" conventional uses (mean = 2.51 s). These means suggest, in addi­
tion, that the innovative uses of the noun-noun combinations did not inevi­
tably take longer to understand than the conventional uses. Although
overall the reading times were longer for the innovative uses, those times for
the "slow" member of each pair converged (i.e., the mean reading times
were 2.53 and 2.51 s). These data are consistent with the concurrent pro­
cessing model's suggestion that the processes that give rise to lexical and in­
novative meanings operate in parallel. In addition, the reversal in the
pattern of "slow" and "fast" hints that the processes may be competing for
the same resources. That is, when the conventional meaning is accessed
quickly (i.e., the "fast" cases) it was relatively difficult for readers to create
the innovative meaning.
The rejection of the error recovery model strongly echoes the rejection of
the standard pragmatic model. These experiments extend the important
3. CONTEXTUAL EXPRESSIONS AND COMMON GROUND 55

conclusion that common ground has a powerful effect—early and often. For
these stories, linguistic copresence, as encoded in the accumulating con­
text, enables readers to overwrite ready-made meanings. These results fore­
shadow the suggestion we make somewhat later that the processes that
allow addressees to understand contextual expressions have wider applica­
tion than just for those expressions. Any word or phrase could be a preempt­
ing innovation. As such, the processes that allow them to be understood
must be in perpetual readiness to generate appropriate readings.

Do Out-of-Context Meanings Have Conceptual Primacy? Recall

circumstances in which the addressee of the utterance "Could you please pass
the salt?" replies "Sure" and then passes the salt. In this case, it seems that the
addressee has recovered both the literal and indirect meanings. However, is it
necessarily the case that addressees inevitably recover literal meanings—that
is, even if literal meanings do not always come first, do they always come? Re­
searchers have considered this question in various ways (for a review, see
Gibbs, 1994). Consider an experiment in which participants were asked to
read stories that provided contexts that generated either literal or indirect
meaning for the same sentence (Gibbs, 1983). Gibbs demonstrated that literal
meanings were not particularly accessible in indirect contexts. This result sug­
gests that readers did not inevitably construct literal meanings for these utter­
ances. Rather, addressees used common ground to short circuit or eliminate
construction of literal meanings.
For many types of contextual expressions, it would be quite hard to define
an out-of-context meaning. It is not clear, for example, how addressees
might represent the phrases "to delicate" or "to Van Damme" without im­
mediately assessing them with respect to common ground. Still, one type of
contextual expression has generated a good deal of research as an out-of-
context phenomenon. Specifically, researchers have used noun-noun com­
binations as a way to study conceptual combination—the processes through
which the independent meanings of different concepts (e.g., roach and ru­
mor) interact to yield innovative combinations (e.g., roach rumor; e.g.,
Coolen, Van Jaarsveld, & Schreuder, 1991, 1993; Gagne &Shoben, 1997;
Murphy, 1990; Wisniewski, 1996; Wisniewski & Love, 1998; Wisniewski &
Middleton, 2002). With few exceptions, researchers have studied concep­
tual combination by presenting experimental participants with noun-noun
combinations with no supporting context. Researchers then build theories
about processes of conceptual combination based on the comparative diffi­
culty of interpreting phrases in isolation.
The difficulty with this approach is that the results for out-of-context in­
terpretation (and the corresponding theories) may prove to be of little or no
value with respect to a theory of how addressees understand noun-noun
combinations (and carry out conceptual combination) in discourse contexts
56 GERRIG AND HORTON

that are defined by richer common ground. Not surprisingly, researchers

who routinely engage participants in out-of-context interpretation have de­
fended this practice. For example, Wisniewski (1996, pp. 450–451) argued,
"It makes sense to first identify how the meanings of the constituents (i.e.,
prior knowledge) affect interpretation. Then the role of discourse setting
may be more meaningfully understood in light of these prior knowledge ef­
fects." This argument is strikingly similar to the claims some theorists have
made for the conceptual primacy of literal meaning. However, as we ob­
served, research results suggest that literal meanings are not inevitably pres­
ent in understanding. To assess the validity of the output of the conceptual
combination enterprise, it is important to determine the extent to which
these results generalize to contextual expressions.
The evidence on the temporal primacy of out-of-context meanings is rel­
evant to this issue. If it does not inevitably take longer to understand a literal
use of, for example, cave man then we have reason to wonder whether out-
of-context meanings always contribute to understanding in discourse con­
texts. In fact, Gerrig's (1989) experiments provided suggestive evidence
that the recovery of the two types of meaning interfere with each other. This
hint of interference provides the basis for a broader test of the relationship
between out-of-context and discourse contexts interpretation of noun-
noun combination.
In particular, Gerrig and Bortfeld (1999) contrasted two views of this re­
lationship. In one view, the two types of meaning recovery are interdepen­
dent. This view, represented by Wisniewski's (1996) quotation, suggests that
readers represent out-of-context meanings as a first step toward under­
standing the expressions in context. In this view, trouble should arise when
there is a discrepancy between the two interpretations. Suppose, for exam­
ple, that in the absence of discourse context most people believe doll smile
means "the smile on a doll's face," but in a particular story context it is given
the meaning "the smile a child gets when she is given a doll as a present."
The interdependence view suggests that the readers should find it relatively
difficult to understand doll smile in the story use—relatively difficult, that is,
with respect to a comparable expression (e.g., baseball smile) that has no im­
mediately available out-of-context meaning. A contrasting view—the inde­
pendence view—suggests that the two types of meaning recovery do not
necessarily depend on one another. As such, the independence view does
not predict interference when noun-noun combinations that have strong
out-of-context meanings are placed in discourse contexts.
To begin their contrast of the interdependence and independence
views, Gerrig and Bortfeld (1999) identified pairs of noun-noun combina­
tions that differed reliably in the accessibility of meanings out of discourse
contexts. They used three different methods to construct these pairs. In an
initial experiment, participants saw one member from each of 30 pairs of
3. CONTEXTUAL EXPRESSIONS AND COMMON GROUND 57

noun-noun combinations (e.g., doll smile vs. baseball smile,family lunch vs.
tax lunch, and zebra horse vs. ostrich horse). Their task was to read each
noun phrase and rate how difficult it was to come up with a meaning on a
scale ranging from 1 (very easy) to 1 (very hard). The data generally con­
firmed the experimenters' intuitions with respect to the probable accessi­
bility of meanings: The Imean difficulty rating for those noun-noun
combinations they anticipated would have relatively more accessible
meanings was 2.62; for those noun-noun combinations they anticipated
would be relatively less accessible the mean rating was 3.58. The 20 pairs
that yielded the largest differences were passed along to later experiments
(with mean ratings of 2.24 vs. 3.70).
The next pair of studies measured the accessibility of meanings in speeded
tasks. The two accounts of processing—the interdependence and independ­
ence views—make different predictions about the time course with which
contextual expressions will be processed. The two speeded tasks provided evi­
dence that the participants' intuitions about meaning accessibility out of con­
text also had an impact on processing time. In the first study, Gerrig and
Bortfeld (1999) asked participants to push a response key to indicate when
they had thought of a meaning for each noun-noun combination. Partici­
pants responded considerably more quickly for the (rated) highly accessible
tokens from each pair (mean response time = 5.61 s) versus the less accessible
tokens (mean = 8.54 s). In the second study, Gerrig and Bortfeld used a task
more closely patterned on previous processing time analyses of noun-noun
combinations: They asked participants to indicate as quickly as possible
whether they found a phrase meaningful (Murphy, 1990). Once again, partic­
ipants made responses more swiftly for highly accessible tokens (mean = 1.92
s) than for less accessible tokens (mean — 2.44 s). This response time pattern
was further corroborated by the rates at which participants judged the phrases
to be meaningful. Participants responded "yes" for 81.2% of the highly acces­
sible tokens versus only 43.4% of the less accessible tokens.
This series of results provide an appropriate context for the contrast be­
tween the interdependence and independence views. The two populations
of noun-noun combinations are reliably different out of discourse context.
As such, the interdependence view anticipates a greater degree of interfer­
ence when the relatively more accessible noun-noun combinations are
placed in discourse contexts that endow them with innovative meanings.
Consider the pair family lunch versus tax lunch. Out of discourse context,
family lunch has a meaning that is considerably more accessible. However, an
appropriate context creates meanings for bothfamily lunch and tax lunch that
are conceptually quite similar:

Each day Maggie ate lunch with her friends. They all sat together and
talked about concerns in their lives. Generally, they agreed on the
58 GERRIG AND HORTON

topic ahead of time. One day, they discussed how to avoid paying extra
income tax. Another day, they discussed how to deal with family prob­
lems. Maggie found the family lunch to be very helpful.
(or:) Maggie found the tax lunch to be very helpful.

The interdependence view predicts that readers should experience interfer­

ence when they experience family lunch in this context that is at odds with
their out of context expectations. To provide a final index of the relative ac­
cessibility of the meanings of the noun-noun combinations without particu­
lar contextual support, Gerrig and Bortfeld (1999) also used stories that
provided neutral contexts:

Each day Maggie ate lunch with her friends. They all sat together and
talked about concerns in their lives. Generally, they agreed on the
topic ahead of time. Sometimes Maggie wasn't always sure the conver­
sations were worthwhile. She thought back over the topics they'd cov­
ered over the last few days. Maggie found the family lunch to be very
helpful.
(or:) Maggie found the tax lunch to be very helpful.

Given the sparse contextual support, readers should find it easier to eke
some sort of meaning out of those combinations that have an out-of-context
advantage.
In fact, participants' reading times for the final sentences in the neutral
contexts did replicate the pattern of accessibility from the out-of-context
studies: Participants read the sentences with the noun-noun combinations
that had been otherwise highly accessible in reliably less time than they read
the sentences with the other members of the pairs (3.74 s vs. 4.34 s). How­
ever, contrary to the interdependence view, there were no hints of interfer­
ence when the combinations appeared in the innovative contexts. The
(nonreliable) trend was for readers still to take less time to understand the
highly accessible combinations (3.61 s vs. 3.71 s).
This pattern of reading times is consistent with the independence
view—that meaning recovery in and out of discourse contexts are inde­
pendent of one another. However, the reading times alone cannot confirm
that participants understood the noun-noun combinations in the way that
was intended. The rejection of the interdependence view relies on the as­
sumption that, in innovative contexts, readers represented, for example,
family lunch and tax lunch with equal specificity. To address this concern,
Gerrig and Bortfeld (1999) conducted a final experiment that added para­
phrase judgments to the ends of each of the innovative versions of the sto­
ries. (Because it was not entirely clear what paraphrases to give for the
3. CONTEXTUAL EXPRESSIONS AND COMMON GROUND 59

noun-noun combinations in the neutral contexts, the experiment omitted

those stories.)
In this final study, the reading times for the highly accessible tokens were
reliably faster than those for the less accessible tokens (2.80 s vs. 3.04 s).
This result is not consistent with a view that predicts interference. However,
the result also raises the disturbing possibility that participants may not have
arrived at the appropriate innovative meanings for the noun-noun combi­
nations: Readers might have been indicating that they had understood the
highly accessible combinations particularly quickly because they were de­
faulting to the out-of-context meanings. Fortunately, the paraphrase judg­
ments allow this possibility to be dismissed. Readers agreed with the correct
paraphrases in roughly the same amount of time for highly accessible (mean
= 3.88) and less accessible combinations (mean = 4.05). They also made
roughly the same number of errors (7.5% vs. 11.0%).
These converging measures lead to the strong conclusion that the rela­
tively high accessibility of out-of-context meanings for noun-noun combi­
nations did not lead to interference when these combinations were
endowed with innovative meanings. Thus, the results support the conten­
tion that meaning recovery in and out of discourse contexts is relatively in­
dependent. As such, Gerrig and Bortfeld's (1999) research questions the
overall value of research that focuses on out-of-context meanings. The
norm, presumably, is for people to encounter noun-noun combinations (as
well as other types of contextual expressions) in rich contexts. It is curious,
in that light, that most research on so-called conceptual combination has
used paradigms in which participants interpret expressions with no contex­
tual support. Of course, Gerrig and Bortfeld only used one type of noun-
noun combination with some unknown range of accessibility. The possibil­
ity remains that there would be some subset of noun-noun combinations for
which the out-of-context meaning would be so highly accessible that inter­
ference would arise. As we discussed earlier, readers experienced interfer­
ence for some preempting innovations (e.g., door man vs. cave man).
However, the burden clearly lies with researchers who use out-of-context
tasks to demonstrate that their results have relevance to any type of ordinary
language understanding.
In this section we showed that out-of-context meanings for contextual
expressions have neither temporal nor conceptual primacy. Rather, address­
ees appear able to use common ground to arrive at speakers' intended mean­
ings without considering how words or phrases might have been used, had
they been used out of discourse context. This is an important conclusion be­
cause it brings common ground's penetration into the process of compre­
hension down to the level of the word and the phrase. We turn now from the
relative primacy of meanings in and out of context to the processes that yield
those meanings.
6O GERRIG AND HORTON

Does Common Ground Constrain Meaning Via Special

Processes?

In the previous section, we reviewed evidence that contextual expressions

share properties with other pragmatic phenomena with respect to the diver­
gence between what a speaker said and what the speaker meant. We con­
cluded that out-of-context meanings do not have temporal or conceptual
primacy with respect to those meanings recovered in discourse contexts.
Still, that leaves open the question of whether the actual processes that
bring about the two types of meaning recovery are identical (Gibbs & Gerrig,
1989). It could be the case that two different streams of mental processes
yield their own types of meaning (i.e., out of context vs. common ground
sensitive) with the same temporal endpoint. However, researchers on prag­
matic phenomena have argued in various ways that the processes are, in
fact, identical. Keysar (1989), for example, examined the functional equiva­
lence of the processes that recover literal and metaphorical readings of the
same utterances. Consider the utterance, "Bob Jones is a magician." Keysar
demonstrated that readers' comprehension was affected by the extent to
which Bob Jones was consistently a magician, both literally (i.e., he sawed
women in half) and metaphorically (i.e., he was good with money). Keysar
suggested that consistency mattered because the same processes yielded
both literal and metaphorical readings. Results of this type support the con­
tention that pragmatic phenomena do not require special mental processes.
Rather, the same processes give rise to a variety of representations (Gibbs &
Gerrig, 1989).
By contrast, contextual expressions appear to demand special processes.
Consider a sentence without any (apparent) contextual expressions: The
cat is on the mat. Accounts of language processing universally include as a
critical step sense selection: For each word (e.g., cat and mat) addressees select
a contextually appropriate reading from the mental lexicon. This process is
most complicated when words are ambiguous, in which case addressees
must combine sources of information to select the appropriate sense. As we
note somewhat later, the details of this selection process have been the topic
of prolonged psycholinguistic debate. Still, it is clear that sense selection will
not work for contextual expressions: They are infinitely ambiguous, and (for
the vast majority of cases) none of those infinite number of meanings are
listed in the mental lexicon. For these reasons, H. H. Clark and Gerrig
(1983) proposed that the understanding of contextual expressions requires
addressees to engage in processes of sense creation. These processes of sense
creation use common ground to constrain speakers' meanings for roach ru­
mor and its peers.
Support for sense creation as a process apart from sense selection comes
from research with older adults. Zelinski and Hyde (1996) examined the
3. CONTEXTUAL EXPRESSIONS AND COMMON GROUND 61

hypothesis that older adults come to have a semantic processing deficit

that prompts them to form general rather than specific interpretations.
This deficit has implications for the understanding of contextual expres­
sions because contextual expressions require addressees to recover spe­
cific interpretations of expressions that are, on the surface, quite general.
In their experiments, Zelinski and Hyde asked younger adults (mean age =
22) and older adults (mean age = 68) to interpret eponymous verb phrases
such as "do an Elvis Edmunds" in story contexts. In the stories, various ac­
tivities were attributed to the characters (e.g., Elvis Edmunds). In some
stories, the sentence using the eponymous verb phrase included restricting
information (e.g., "Later your friend says that he has thought about doing
an Elvis Edmunds to the apples he bought."); in other stories, the use of the
phrase did not include such information (e.g., "Later your friend says that
he has thought about doing an Elvis Edmunds."). The question is whether
readers would be able to use the added information about doing an Elvis
Edmunds "to the apples" to constrain their understanding of the novel
eponymous phrase. The younger adults were more likely to use the cue to
restrict their interpretation than were the older adults. These data suggest
that the processes that give rise to appropriate interpretations of contex­
tual expressions can be selectively impaired as a cognitive concomitant of
aging.
We just briefly made the case that addressees require special processes to
recover the meanings of contextual expressions. The error here, however,
would be to conclude that because special processes are needed for contex­
tual expressions, addressees only engage those processes when they encoun­
ter contextual expressions (H. H. Clark, 1983; Gerrig, 1986). That is, the
processes that addressees use to understand contextual expressions only
seem special because psycholinguistic research has been so narrowly focused
on sense selection. We argue instead that the processes of sense cre-
ation—the processes through which addressees use common ground to fix
the precise meanings of words—function quite broadly in ordinary compre­
hension. In this light, sense creation looks special only because people have
studied it almost exclusively with respect to contextual expressions. We now
discuss three types of evidence that the processes of sense creation function
quite ubiquitously.

Construal and Negotiation. One of the central assumptions of theo­

ries of sense selection is that meanings reside ready-made in a mental lexi­
con. Thus, addressees need only access those ready-made meanings for cat
and mat, combine them in the way indicated by English syntax, and the in­
terpretation of "The cat is on the mat" becomes clear. However, most words
have the potential to take on different shades of meaning, depending on the
context. Consider these uses of tomato (Johnson-Laird, 1975):
62 GERRIG AND HORTON

• She sat on a tomato.

• She likes tomato.
• Her face was like a tomato.

Suppose the same information for tomato is accessed via sense selection in
each of these contexts. If that is the case, the addressees are clearly not done:
They need to show the same ability to adjust the representations to context
that is required for contextual expressions. Most research on meaning has
focused on how addressees settle on one or another reading of an ambiguous
word (e.g., bank or scale). We suggest, by contrast, that processes of sense
creation function within discourse contexts to narrow down the variability
within senses.
To make this observation more concrete, we return to the notion that
speakers and addressees often negotiate to establish a perspective on objects
and activities in the environment. Recall our referential communication ex­
periment that gave rise to contextual expression American flag fish. We sug­
gested that the directors offered perspectives that were adjusted through
interaction with the addressees. Now consider the phrases green fish and or­
ange fin. There is nothing about these expressions that seems, on the surface,
contextual. Still, they emerged through a process that was much the same as
American flag fish. In the first round of our task, one director performed this
utterance:
"Number twelve is like a dark green fish, it's got like an orangy fin."
In this utterance, the director hedges the description in two ways. She says
that the fish is "like a dark green" and that the fin is "orangy." In both cases,
the director signals to the matcher that she is taking a perspective that is
open to adjustment. The matcher replies, "Dark green fish, orangy fish, got
it." In the next round, the hedges have fallen away:
"Number fourteen is the green fish with the orange fin."
Thus, although green fish and orange fin do not look like contextual expres­
sions, they are. It is quite likely that other addressees, who were not partici­
pants in the negotiations, would find it relatively more difficult to identify the
appropriate picture (among the other pictures offish). We base this suggestion
on research by Schober and H. H. Clark (1989), who demonstrated that ex­
perimental participants who listened to full conversations in a referential
communication task—but did not themselves participate in negotia-
tions—were at a comprehension disadvantage with respect to direct partici­
pants. In general, we see a commonality in the construal of tomato, green, and
orange. In each case, we can evoke the processes of sense creation to explain
how addressees recover the correct specific meaning.
3. CONTEXTUAL EXPRESSIONS AND COMMON GROUND 63

Memory Processes and Meaning. We noted earlier that, within the

study of sense selection, the topic of ambiguity resolution has been one of
nearly constant controversy. Consider the word bank, which has at least two
meanings as "a financial institution" and "the side of a river." Debate has cen­
tered on the question of under what discourse circumstances are one or both
of these meanings accessible. As research methods have evolved, each of sev­
eral models has held sway for months or sometimes years (for a review, see
Simpson, 1994). In some periods, psycholinguists have believed that address­
ees automatically access all meanings equally at all times, or that all meanings
are accessed but some more so than others. In other periods, psycholinguists
have believed that only contextually appropriate meanings are accessed. Be­
cause contextual expressions have infinite possible meanings, they provide an
opportunity for researchers to test the generality of their conclusions about
ambiguity resolution. Specifically, researchers can use contextual expressions
to determine whether conclusions narrowly drawn about sense selection are
really true of more general cognitive processes.
For example, the two meanings of bank are unequal (Onifer & Swinney,
1981): "A financial institution" is the more frequent dominant meaning;
"the side of a ride" is the less frequent subordinate meaning. Research on
ambiguity resolution has often suggested that dominant meanings are privi­
leged with respect to subordinate meanings. For example, Tabossi and
Zardon (1993) wrote sentences that created biases toward the dominant
and subordinate readings of ambiguous words. Using a cross-modal lexical
decision task, they demonstrated that dominant meanings were primed irre­
spective of context, but subordinate meanings were primed only when the
context was specifically biased toward those readings. Experiments of this
sort presuppose that the privilege of dominance reflects the representation
of ready-made meanings in the mental lexicon.
By contrast, Rapp and Gerrig (1999) sought to demonstrate that the privi­
lege of dominance extends to uses of eponymous verb phrases. In the initial
phase of their research, they normed eponymous verb phrases to construct a
set with well-defined dominant and subordinate meanings. Rapp and Gerrig
asked participants in the norming study to provide interpretations of phrases
such as "do a Neil Armstrong" and "do an Albert Einstein." The instructions
were patterned after production tasks for ordinary lexical ambiguities: "Your
task is to write down the three most likely meanings of each verb phrase. Pre­
tend that you have just heard the phrase during a conversation. What would
be some meanings that seem most natural to you?" After transcribing partici­
pants' responses, Rapp and Gerrig created categories that drew together re­
sponses similar in meaning (e.g., they considered "to set foot on the moon,"
"jump up and down on the moon," and "be the first to walk on some planet"
all variations on the same interpretation of "do a Neil Armstrong"). From an
initial pool of 45 eponymous verb phrases, Rapp and Gerrig found 28 phrases
64 GERRIG AND HORTON

for which the average rate at which the dominant meaning was produced was
roughly 42% higher than the rate at which the subordinate meaning was pro­
duced (72.4% vs. 29.8% respectively). This difference between productio
frequencies is comparable to that used in experiments using ordinary ambigu­
ities (e.g., Onifer & Swinney, 1981). Note that although the subordinate
meanings were given less often, they were not particularly obscure (e.g., for
"do an Albert Einstein," the dominant meaning was "accomplished a scientific
milestone" and the subordinate meaning was "have unkempt hair").
The norming study provided the set of eponymous verb phrases Rapp and
Gerrig (1999) needed to test the relative accessibility of dominant and sub­
ordinate meanings. For the next phase of their research, they wrote stories
that supported one of these two meanings. Consider this story that engaged
the dominant meaning of "do a John Travolta:"

Rick and his sister had found some old 70s records in the attic. They
put one on the player and laughed while they listened. Then Rick be­
gan to move in time to the music. He announced, "Watch me do a
John Travolta."

An alternative story engaged the subordinate meaning:

Rick was once the most valuable player on his soccer team. A drug
habit had affected his skills, and he was removed from the team. Now
he was in rehab, hoping to kick his habit and return to the sport. He
announced, "Watch me do a John Travolta."

The purpose of the experiment was to test the hypothesis that readers
would consider the dominant meaning of the eponymous verb phrase even
in the subordinate contexts. To assess this prediction, Rapp and Gerrig
(1999) asked participants to accept or reject paraphrases of the sentences
including the following eponymous verb phrases:

Rick is going to try to disco dance [dominant paraphrase], and

Rick is going to try to make a comeback [subordinate paraphrase].

For each of 20 eponymous verb phrases, participants read a story that supported
one of the two senses and ended with one of the two paraphrases. Participants
took nearly the same amount of time to agree with both of the appropriate para-
phrases—dominant paraphrases to dominant senses (mean — 3.10s) and sub­
ordinate paraphrases to subordinate senses (mean = 2.97 s). As expected,
however, response times for the inappropriate paraphrases were reliably longer
for the dominant paraphrases in subordinate contexts (mean = 3.03 s) than
subordinate paraphrases in dominant contexts (mean = 2.66 s).
3. CONTEXTUAL EXPRESSIONS AND COMMON GROUND 65

These data support the contention that dominant senses are relatively
available even when the context does not support their relevance. As such,
they suggest that claims for the privilege of dominant meanings for ordinary
ambiguities should not unreflectively be attributed to properties of lexical
memory. Rather, common ground appears to play a broader role in shaping
experiences of ambiguity. The extension of dominance effects to at least one
type of contextual expression weakens faith in one assumption for sense se-
lection—the existence of ready-made meanings in the mental lexicon.
Contextual Expressions and Parsing. We have been focusing on
the implications of contextual expressions with respect to theories of lexical
access and meaning. However, contextual expressions also have implica­
tions for other aspects of language processing. In particular, H. H. Clark
(1983) argued that contextual expressions pose enormous problems to stan­
dard accounts of parsing. Consider his famous example from one of Erma
Bombeck's columns (H. H. Clark, 1983, p. 298):
We thought we were onto a steam iron yesterday, but we were too late.
Steam irons never have nay trouble finding roommates. She could
pick her own pad and not even have to share a bathroom. Stereos are a
dime a dozen. Everyone's got their own systems. We've just had a
streak of bad luck. First, our Mr. Coffee flunked out of school and went
back home. When we replaced her, our electric typewriter got married
and split, and we got stuck with a girl who said she was getting a leather
coat, but she just said that to get the room.
Once readers get the hang of this paragraph—that steam iron and so on are the
possessions of once and prospective roommates—there does not seem to be
anything particularly challenging about the recovery of grammatical struc­
ture. Still, Clark suggested that standard accounts of parsing will lead to a pair
of errors: nonparsing and misparsing. Non-parsing occurs when the parser
would be stumped by a construction. For example, because "married" requires
human participants, "our electric typewriter got married" would cause most
parsers to fail. Misparsing occurs when the parser yields an incorrect represen­
tation. For example, "stereos are a dime a dozen" yields the interpretation
"phonographs are very common." This is not what Bombeck intended.
To accommodate these parsing problems, Clark (1983) suggested that a
shift in emphasis was necessary from standard parsers to what he called in­
tentional parsers. He suggested that parsing should be "viewed not simply as
dividing a sentence into its parts—the traditional view—but as identifying
the goals and subgoals the speaker had in uttering each part of the sentence"
(p. 324). In Clark's view, the goals and subgoals refer to the speaker's inten­
tions. The recovery of those intentions require the parser to use information
about common ground (p. 325):
66 GERRIG AND MORTON

With contextual expressions, reference to speaker's and addressee's

common ground is mandatory. When Bombeck wrote Our electric
typewriter got married, she intended us readers to make use of the fact
that she had just written about roommates and their possessions. She
intended us to use this common ground in conjunction with the fact
that she was uttering the phrase our electric typewriter and was predi­
cating of its referent, that it got married. She intended us to use both
sources of information in inferring her hierarchy of goals.

Despite Clark's strong arguments, theorists of parsing have not moved in the
direction of intentional parsers and, for that matter, have not provided alter­
native accounts of how utterances including contextual expressions might
be parsed.
The omission of contextual expressions from theories of parsing has at
least two consequences. First, as Clark (1983) argued, if parsing theories
cannot accommodate contextual expressions, they will be incomplete or in­
correct. Second, as in other instances, contextual expressions provide an in­
teresting control for the generality of conclusions with respect to certain
aspects of parsing. For example, research attention in parsing has often fo­
cused on the extent to which different verbs have associated with thempars­
ing preferences. Consider a series of studies by Shapiro, Nagel, and Levine
(1993) that explored (among other things) the consequences of verbs' pref­
erences for taking or not taking a direct object. In a norming study, partici­
pants were asked to indicate which of two sentence forms they preferred for
common verbs such as taught:

The aging pianist taught his solo with great dignity.

The aging pianist taught with his entire family.

This norming study allowed Shapiro et al. to construct minimal pairs with
opposite preferences. In these sentences, read has a transitive preference
and sang has an intransitive preference:

1. The baby-sitter read the # story to the sick child.

2. The baby-sitter read to # the sick child.
3. The baby-sitter sang the # story to the sick child.
4. The baby-sitter sang to # the sick child.

If verbs' preferences play a role in parsing, readers should find it easier to re­
cover the structures for Sentences 1 and 4 (which respect the verbs' prefer­
ences) than Sentences 2 and 3. To test this prediction, Shapiro et al. asked
participants, while listening to the sentences, to perform a lexical decision
task (to unrelated words) at the positions marked by the #. The premise was
3. CONTEXTUAL EXPRESSIONS AND COMMON GROUND
67

that performance on a secondary task would be slowed when parsing pro­

cesses were constructing nonpreferred representations. The data bore out
this prediction: Participants took reliably more time to make lexical decision
judgments when the sentence provided a nonpreferred complement (mean
= 0.80 s) than what it provided a preferred complement (mean = 0.77 s).
In the types of experiments exemplified by Shapiro et al.'s (1993) research,
the strong assumption is made that the parsing preferences are stored along
with each verb's "entry" in the mental lexicon. Although this may be the case,
we have data that suggest that readers also have rather strong preferences
with respect to the complements that should occur with innovative
denominal verbs. In a pilot experiment, 40 participants provided the same
type of norming information Shapiro et al. had gathered. Specifically, they
were asked to choose which of two sentences sounded better to them:

1. The young mother potato-chipped the casserole to make it extra

crispy.
2. The young mother potato-chipped while she watched a game show.
3. The yuppie BMWed his girlfriend to her dance class.
4. The yuppie BMWed to the grocery store to buy some caviar.

In the norming data, participants showed a strong preference for "to po-
tato-chip" to serve as a transitive verb (Sentence 1) and for "to BMW" to
serve as an intransitive verb (Sentence 4). Because these verbs do not ap­
pear in the mental lexicon, these data present the immediate problem of ex­
plaining how the preferences arise. Suppose, furthermore, that response
time studies produced costs and benefits for mismatches and matches simi­
lar to those Shapiro et al. disclosed. With lexical innovations, it would be
impossible to attribute such results to information precompiled in the men­
tal lexicon. Of course, parallel results (should the speculation be correct)
would not necessitate identical mental processes. Still, these pilot data illus­
trate another way in which contextual expressions provide an important
control for claims about the structure and content of the mental lexicon. Ul­
timately, experiments with contextual expressions could provide strong evi­
dence for H. H. Clark's (1983) notion of intentional parsers, with common
ground at their core.

CONCLUSIONS

We conclude this chapter with a final example of a contextual expression.

Consider this brief literary excerpt (Costello, 2002, p. 213):

Gretchen said, 'What is this, Vi, a college road trip? Thirteen your
butt to Portsmouth and stay there, got it?'
68 GERRIG AND HORTON

In this excerpt, the ordinary number "thirteen" is being used as a verb. In this
chapter, we suggested that contextual expressions arise through speakers'
(and, in this case, authors') assessments of common ground. We suggested
that speakers produce contextual expressions—some for the nonce, some to
endure—with appropriate sensitivity to communal and personal co-pres-
ence. In fact, the meaning of thirteen is established much earlier in the novel,
in a memorable fashion (Costello, 2002, p. 81):

Gretchen said, Felker, that's a negative. Thirteen your ass right back here.
Thirteen' was borrowed cop code. It meant do it now.

In this original instance, the reader learns that Gretchen (who works for the
Secret Service) has borrowed the verb use of thirteen from ordinary police.
The unmarked return to thirteen late in the novel suggests that the author
believed he had successfully extended the verb to a new community—read-
ers of his book.
In the second part of the chapter, we examined the processes that make it
possible for contextual expressions like the latter use of thirteen to succeed.
We saw that processes of sense creation are highly integrated into the ordi­
nary time course of language understanding. That is why, for example, read­
ers are not baffled when thirteen—which, for most readers will previously
only named the quantity between 12 and 14—can become a verb with mini­
mal (or zero) disruption. In that light, our major conclusion is that the pro­
cesses of sense creation function quite broadly, allowing addressees to
recover the specific meanings that speakers intend.

ACKNOWLEDGMENTS

Material for this chapter was based on work supported by the National Sci­
ence Foundation under Grants No. IRI9980013 and No. ITR0082602. Cor­
respondence concerning this chapter should be sent to Richard J. Gerrig,
Department of Psychology, State University of New York at Stony Brook,
Stony Brook, NY 11794-2500. E-mail: [email protected]

REFERENCES

Clark, E. V, & Clark, H. H. (1979). When nouns surface as verbs. Language, 55,
767-811.
Clark, H. H. (1979). Responding to indirect speech acts. Cognitive Psychology, 11,
430–477.
Clark, H. H. (1983). Making sense of nonce sense. In G. B. Flores d'Arcais &. R.
Jarvella (Eds.), The process of understanding language (pp. 297-331). New York:
Wiley.
3. CONTEXTUAL EXPRESSIONS AND COMMON GROUND 69

Clark, H. H. (1996). Using language. Cambridge, England: Cambridge University

Press.
Clark, H. H., & Gerrig, R. J. (1983). Understanding old words with new meanings.
Journal of Verbal Learning and Verbal Behavior, 22, 591-608.
Clark, H. H., & Marshall, C. R. (1981). Definite reference and mutual knowledge.
In A. K. Joshi, I. Sag, & B. Webber (Eds.),Elements of discourse understanding (pp.
10-63). Cambridge: Cambridge University Press.
Coolen, R., Van Jaarsveld, H. J., &.Schreuder, R. (1991). The interpretation of iso­
lated novel nominal compounds. Memory & Cognition, 19, 341-352.
Coolen, R., Van Jaarsveld, H. J., & Schreuder, R. (1993). Processing novel com­
pounds: Evidence for interactive meaning activation of ambiguous nouns. Mem­
ory & Cognition, 21, 235-246.
Costello, M. (2002). Big if. New York: Norton.
Dennis, P (1958). Around theworld with Auntie Mame. New York: Harcourt Brace.
Downing, P A. (1977). On the creation and use of English compound nouns. Lan­
guage, 53, 810-842.
Evanovich, J. (1997). Three to get deadly. New York: St. Martin's Paperbacks.
Evanovich, J. (1999). High five. New York: St. Martin's Press.
Gagne, C. L., & Shoben, E. J. (1997). Influence of thematic relations on the com­
prehension of modifier-noun combinations. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 23, 71-87.
Gerrig, R. J. (1986). Process and products of lexical access. Language and Cognitive
Processes, 1, 187-195.
Gerrig, R. J. (1989). The time course of sense creation. Memory & Cognition, 17,
194-207.
Gerrig, R. J. (1993). Experiencing narrative worlds. New Haven, CT: Yale University
Press.
Gerrig, R. J., & Bortfeld, H. (1999). Sense creation in and out of discourse contexts.
Journal of Memory and Language, 41, 457–468.
Gerrig, R. J., & Gibbs, R. W., Jr. (1988). Beyond the lexicon: Creativity in language
production. Metaphor and Symbolic Activity, 3, 1-19.
Gerrig, R. J., & Horton, W. S. (2001, November).Judgments about common ground in
language production. Paper presented at the meeting of the Psychonomic Society,
Orlando, FL.
Gibbs, R. W., Jr. (1979). Contextual effect in understanding indirect requests. Dis­
course Processes, 2, 1-10.
Gibbs, R. W., Jr. (1983). Do people always process the literal meanings of indirect re­
quests? Journal of Experimental Psychology: Learning, Memory, and Cognition, 9,
524-533.
Gibbs, R. W., Jr. (1994). The poetics of mind. Cambridge, England: Cambridge Uni­
versity Press.
Gibbs, R. W., Jr., & Gerrig, R. J. (1989). How context makes metaphor comprehen­
sion seem "special." Metaphor and Symbolic Activity, 4, 145-158.
Grice, H. P (1975). Logic and conversation. In P Cole & J. Morgan (Eds.), Syntax
and semantics: Vol. 3. Speech acts (pp. 41-58). New York: Academic Press.
Jen, G. (1999). Who's Irish? New York: Knopf.
7O GERRIG AND HORTON

Johnson-Laird, P N. (1975). Meaning and the mental lexicon. In A. Kennedy & A.

Wilkes (Eds.), Studies in long term memory (pp. 123-142). London: Wiley.
Keysar, B. (1989). On the functional equivalence of literal and metaphorical inter­
pretations in discourse. Journal of Memory and Language, 28, 375-385.
Labute, N. (2001, May 28). Layover. New Yorker, 77, 120-121.
Michaels, L. (2001, November 12). Nachman from Los Angeles. New Yorker, 77,
110-121.
Murphy, G. L. (1990). Noun phrase interpretation and conceptual combination.
Journal of Memory and Language, 29, 259-288.
Onifer, W., & Swinney, D. (1981). Accessing lexical ambiguities during sentence
comprehension: Effects of frequency of meaning and contextual bias. Memory &
Cognition, 9, 225-236.
Rapp, D. N., & Gerrig, R. J. (1999). Eponymous verb phrases and ambiguity resolu­
tion. Memory & Cognition, 27, 612-618.
Schober, M. E, & Clark, H. H. (1989). Understanding by addressees and overhear­
ers. Cognitive Psychology, 21, 211-232.
Searle.J. (1979). Metaphor. In A. Ortony (Ed.),Metaphor and thought (pp. 92-123).
Cambridge, England: Cambridge University Press.
Shapiro, L. R, Nagel, H. N., & Levine, B. A. (1993). Preferences for a verb's comple­
ments and their use in sentence processing. Journal of Memory and Language, 32,
96-114.
Simpson, G. B. (1994). Context and the processing of ambiguous words. In M. A.
Gernsbacher (Ed.), Handbook of psycholinguistics (pp. 359-374). San Diego: Aca­
demic Press.
Tabossi, P, & Zardon, E (1993). Processing ambiguous words in context. Journal of
Memory and Language, 32, 359-372.
Welch, J. (2001). Jack: Straight from the gut. New York: Warner Business Books.
Wisniewski, E. J. (1996). Construal and similarity in conceptual combination. Jour­
nal of Memory and Language, 35, 434-453.
Wisniewski, E. J., & Love, B. C. (1998). Relations versus properties in conceptual
combination. Journal of Memory and Language, 38, 177-202.
Wisniewski, E. J., & Middleton, E. L. (2002). Of bucket bowls and coffee cup bowls:
Spatial alignment in conceptual combination. Journal of Memory and Language,
46, 1-23.
Zelinski, E. M., & Hyde, J. C. (1996). Old words, new meanings: Aging and sense
creation. Journal of Memory and Language, 25, 689-707.
 PART II

Sociocultural

Phenomenological

Influences

This page intentionally left blank

 4

Context
and the Comprehension
of Nonliteral Meanings
Thomas Holtgraves
Ball State University

Conversations abound with nonliteral meanings; sarcasm, irony, meta­

phors, hyperbole, indirect requests, hints, backhanded compliments, and so
on all occur with great frequency (Gibbs, 1994; Holtgraves, 2001a).
Nonliteral language, however, is not a monolithic phenomenon. Instead,
there are a multitude of ways in which nonliteral meanings can be conveyed
and comprehended, and one of the most important ways in which these
meanings differ is in terms of their sensitivity to the context. Context always
matters, of course, but sometimes it matters more than others. In this chap­
ter I focus on the dimension of contextual sensitivity and argue that many
types of nonliteral meaning—especially figurative expressions—are rela­
tively immune to the context; their recognized meanings transcend their oc­
casions of use. In contrast, other types of nonliteral meaning—especially
violations of the relation maxim—are completely context dependent.
These differences in contextual sensitivity have a corresponding effect on
how these different forms are comprehended; contextual variables that ef­
fect the comprehension of the latter will not have a corresponding impact
on the comprehension of the former.
In this chapter I begin with a brief review of what has become known as
the standard pragmatic model for comprehending nonliteral meanings,
Grice's (1975) theory of conversational implicature, and research designed
to test various aspects of this model. This model has not held up well, partly
73
74 HOLTGRAVES

because most research has focused on nonliteral meanings that are contex­
tually independent. Next, I review Grice's distinction between generalized
(contextually independent) and particularized (context dependent)
implicatures and attempt to classify various types of nonliteral meanings
with this scheme. Then, I describe my research on the comprehension and
interpretation of utterances yielding particularized implicatures, as well as
the role of context (face management concerns, status, and perspective) in
these processes. Finally, I discuss cultural differences in the interpretation of
nonliteral meanings.

THE GRICEAN MODEL

Grice's (1975) theory of conversational implicature is the starting point for

most attempts to explain the comprehension of indirect meanings; in the
psycholinguistic literature this has become known as the standard model
(e.g., Gibbs, 1994; Glucksberg, 1991). Grice argued that interlocutors abide
by what he termed the cooperative principle (CP), a principle stating that one
should: "Make your conversational contribution such as is required, at the
stage at which it occurs, by the accepted purpose or direction of the talk ex­
change in which you are engaged" (p. 45). This general requirement is fur­
ther specified in terms of the following four conversational maxims:

1. Quantity—Make your contribution as informative as required (i.e., do

not be either overinformative or underinformative).
2. Quality—Try to make your contribution true, one for which you have
evidence.
3. Manner—Be clear. That is, avoid ambiguity, obscurity, and so on.
4. Relation—Make your contribution relevant for the exchange.

It is usually the case that people will mutually assume adherence to the
CP and corresponding maxims, and this assumption serves as a frame for the
interpretation of a speaker's utterances. In other words, speakers' utterances
will be interpreted as if they were clear, relevant, truthful, and informative.
For example, saying "Can you pass the salt?" at the dinner table is a violation
of the relation maxim (i.e., in this context, inquiring into another's ability to
pass the salt would not be a relevant contribution). But because the hearer
assumes the speaker is being relevant, she or he will search for an ulterior
meaning and interpret the utterance (i.e., generate a conversational
implicature) in such a way so as to maintain adherence to the conversational
maxims.
The "Pass the salt" example is an instance of an implicature that arises to
preserve adherence to a specific conversational maxim. In addition, speak­
ers will sometimes intentionally flout or violate a maxim, in which case it is
4. CONTEXT AND COMPREHENSION 75

simply not possible for the hearer to assume the speaker is adhering to the
maxims. For example, abrupt topic changes ("I really think the Cubs will win
the pennant") in response to personal questions ("How did you do on that
history exam?") function as relevance violations and convey much more
than their strictly literal meaning (possible gloss: I didn't do well on the
exam). In this case, it is obvious the speaker is not complying with the rela­
tion maxim. Still, the hearer will usually assume overall cooperativeness on
the part of the speaker and generate a conversational implicature that
makes sense of the violation.

PROCESSING NONLITERAL MEANINGS

Is the Gricean model an accurate description of how people interpret in­

stances of nonliteral meaning? Research on politeness (Clark & Schunk,
1980; Holtgraves, 1997a; Holtgraves & Yang, 1990, 1992) and responses to
polite requests (Clark, 1979) provide some indirect support for this model,
primarily for the claim that both the literal and nonliteral meanings of indirect
requests are activated. For example, Clark (1979) demonstrated that when
people respond to indirect requests for information (e.g., "Can you tell me
what time you close?"), they frequently address both the literal meaning (i.e.,
do you have the ability to tell me when you close) and the indirect meaning
(i.e., tell me what time you close) of the request (e.g., "Sure, 8:00 p.m.").
Moreover, research on politeness also suggests that people recognize the
literal meaning of many indirect utterances. This is because the politeness of
an utterance is based on the remark's literal rather than indirect meaning.
For example, "Could you shut the door?" is more polite than "I want you to
shut the door," even though both are indirect requests to shut the door.
There is fairly extensive research demonstrating that the perceived polite­
ness of requests is influenced by variations in literal wording (Clark &
Schunk, 1980; Holtgraves & Yang, 1990,1992). There must be some aware­
ness of the literal meaning if politeness judgments vary in this way. Along
similar lines there is some evidence that people encode the specific wording
of an utterance when that wording varies in politeness (Holtgraves, 1997a).
In these studies, people spontaneously remembered the politeness wording
of utterances at better-than-chance levels. And even when they did not re­
member the exact wording, they did seem to retain some gist of the polite­
ness of the utterance. For example, if people had heard "I'd like you to read
the list," they were more likely to recall an equally polite form (e.g., "Could
you read the list?") rather than an impolite form (e.g., "Read the list"). These
results also suggest some activation of literal meaning.
The fact that people are attentive to variations in politeness suggests that
there must be some awareness of both the literal and the nonliteral (or con­
veyed) meaning of utterances. The problem with politeness research, of
76 HOLTGRAVES

course, is that it is not direct evidence; there is no evidence regarding the ac­
tual cognitive operations involved in the comprehension of requests. For ex­
ample, it is possible that the politeness of a request is recognized
simultaneously with (rather than prior to) the recognition of the conveyed
(indirect) meaning. Furthermore, it is conceivable that some wordings have
conventionalized politeness values that determine perceived politeness
without any activation of the literal meaning of the remark. For example,
"Can you X?" might be conventionally more polite than "I want you to X." It
is possible that this difference in wording could affect perceived politeness
without any activation of the literal meaning of the utterance.
More direct tests of this model have been provided by psycholinguists
who have examined in detail the processing of figures of speech. In general,
this research has not supported Grice's (1975) model. For example, consider
the claim that a nonliteral meaning is the result of an inference process. An
inference process is time consuming; thus people should take longer to com­
prehend figures of speech than their direct equivalents. However, numerous
studies have demonstrated that an inference process is not required for
these forms. People simply do not take more time to understand the meaning
of figurative expressions (e.g., "He spilled the beans") than they do equiva­
lent literal expressions (Gibbs, 1980; Ortony, Schallert, Reynolds, & Antos,
1978).
Moreover, in the standard pragmatic view, activation of the literal mean­
ing of a remark is obligatory and must occur prior to the (optional) recogni­
tion of the nonliteral (figurative) meaning. But research indicates that for
many figures of speech, the literal and figurative meanings are assessed si­
multaneously; in some cases, they are assessed in a reversed order.
For example, several studies have demonstrated that the nonliteral
meaning of a figure of speech is activated even when the literal meaning is
acceptable in context (Gildea & Glucksberg, 1983; Glucksberg, Gildea, &.
Bookin, 1982; Keysar, 1989), and sometimes people recognize the literal
meaning only after first considering and then rejecting the nonliteral mean­
ing, which is the reverse of the predicted ordering (Keysar, 1994).
Finally, most interpretations of Grice's (1975) model claim that once a
literal meaning has been rejected, it will no longer play a role in determining
the meaning of an utterance. But this too appears not to be the case. Rather,
the literal meaning of the words in a metaphor can continue to influence the
manner in which the metaphor is interpreted (Cacciari & Glucksberg, 1994;
Titone & Connine, 1994).

GENERALIZED VERSUS PARTICULARIZED IMPLICATURES

For the most part, psycholinguistic research on nonliteral meaning has fo­
cused on a narrow range of utterance types, primarily figures of speech
4. CONTEXT AND COMPREHENSION 77

(which are highly idiomatic) and conventional indirect requests (which are
also highly idiomatic). But clearly there are many different ways to convey
indirect meanings, with corresponding differences in how those meanings
are processed. Of particular importance for processing models is Grice's
(1975) distinction between generalized and particularized implicatures.
The basic difference between the two is that generalized implicatures are
context independent; they have preferred interpretations that occur with­
out reference to the context. Particularized implicatures, however, are con­
text dependent; their recognition requires a consideration of the utterance
in terms of a context, most notably the prior discourse context. Much of the
figurative language examined in prior psycholinguistic research appears to
produce generalized implicatures. For example, most metaphors and idioms
seem to be interpretable independent of any discourse context. Regardless
of the context, people will usually interpret "He spilled the beans" as mean­
ing "He revealed a secret." Support for this comes from the fact that the
nonliteral meaning of many metaphors is not optional; even when the con­
text supports a literal reading, the nonliteral meaning is still activated
(Gildea & Glucksberg, 1983;Glucksbergetal., 1982;Keysar, 1989). (Butsee
Levinson, 2000, for an alternative view.)
It is not only metaphors and idioms that produce generalized
implicatures. In fact, there is a very large class of utterance forms that pos­
sess preferred interpretations regardless of the context in which they are
used. Scalar implicatures (e.g., some and few) are common and generally re­
sult in what Levinson (2000) referred to as Q-implicatures. For example,
"Some of the students attended class" implies that not all of the students at­
tended class. Note that this is an implicature because the sentence would be
true even if all of the students attended class. Other examples of generalized
implicatures include sentences such as "Mark has three children" and
"Harry tried to play the stock market." The former yields the inference that
Mark has no more than three children (even though the sentence would be
true if Mark did have more than three children), and the latter yields the in­
ference that Harry failed at playing the stock marker (again, even though
the sentence would be true had he been successful). Empirical evidence re­
garding the specific manner in which these forms are comprehended is lack­
ing, although it is clear that contextual features will play little role in their
comprehension.
One particularly interesting utterance type that yields a generalized con­
versational implicature is referred to as a "what is X doing Y" (WXDY) con­
struction (Kay& Fillmore, 1999). Examples include "What you doing in my
chair?," "What is this coffee doing on the carpet?," and so on. These con­
structions typically are interpreted as indicating a state of affairs that is in­
congruent with what is normally expected in these situations (that you
should not be in my chair, that coffee should not be on the carpet, etc.). Al­
78 HOLTGRAVES

though empirical research on the processing of these forms is lacking, intu­

itively it seems that these are the default interpretations for these forms, and
there is little processing cost associated with their use (i.e., a Gricean infer­
ence process is not required).
Even illocutionary force—the speech act (e.g., promise, request, apolo­
gize, and warn) performed with an utterance—can be considered a general­
ized implicature. Note that with the exception of performative utterances
(e.g., I order you to shut the door), the illocutionary act is usually implicit
and not literally present in the utterance that performs the act (e.g., "I'll def­
initely do it tomorrow" does not contain the speech act verb promise). Yet,
people do appear to routinely and automatically recognize this level of
meaning during comprehension. For example, we have found evidence for
the online activation of illocutionary force (Holtgraves & Ashley, 2001) us­
ing both a recognition probe reaction time procedure (Experiments 1 and 2)
and a lexical decision task (Experiments 3 and 4). Specifically, when partici­
pants read utterances that performed a specific speech act (e.g., warn and re­
mind) , their responses to probes representing the performed speech act were
slowed in Experiments 1 and 2 and facilitated in Experiments 3 and 4. Al­
though illocutionary force appears to be activated during comprehension,
the process by which it is activated has not been investigated.
Finally, conventional indirect requests have received a fair amount of
empirical attention, and these too appear to be generalized implicatures.
Conventional indirect requests (e.g., Could you shut the door?) usually con­
tain the imperative form of the request (shut the door), allow for the inser­
tion of please (Could you please shut the door), and are related to the felicity
conditions underlying the performance of the request (e.g., hearer's ability).
Research has demonstrated that conventional indirect requests are pro­
cessed directly and quickly and without an inference process (Gibbs, 1983)
and that their recognition is not affected by contextual variables such as the
status of the speaker (Holtgraves, 1994).
Overall, then, generalized implicatures represent instances when
speaker meaning—what a speaker intends to accomplish with an utter-
ance—is recognized in a relatively direct fashion, without need of a
Gricean inference process. Now it is obvious that some type of inferential
processing is required to recognize a generalized implicature. But the pro­
cess outlined by Grice (1975)—recognize and then reject a literal mean-
ing—does not appear to be involved. And, of course, contextual
independence is far from absolute. It is relatively easy to envision contexts
in which a generally accepted generalized implicature is not appropriate.
For example, the literal meaning of "He spilled the beans" might be taken
seriously (and exclusively) in the context of one's youngest son spilling a
pot of beans. Generalized implicatures may be default implicatures, but
they (at least some of them) can be overridden by a context that strongly
4. CONTEXT AND COMPREHENSION 79

biases how the utterance is to be interpreted. However, this will probably

involve a time-consuming, effortful process.

PARTICULARIZED IMPLICATURES, RELEVANCE

VIOLATIONS, AND FACE MANAGEMENT

In contrast to generalized implicatures, particularized implicatures are

heavily dependent on the context and, in fact cannot be interpreted apart
from a context. A prototypical particularized implicature would be a viola­
tion of the relation maxim (be relevant). Consider the following example
(from Holtgraves, 1998b, p. 1): Bob and Al are students in the same history
class and Bob has just given a presentation in this class. The following ex­
change then takes place:

Bob: What did you think of my presentation?

Al: It's hard to give a good presentation.

How is Al's reply to be interpreted? Obviously he is asserting the belief that

the act of giving a class presentation is difficult. But most people would prob­
ably conclude that he really means something more; the most likely inter­
pretation would be that he did not like the presentation. Unlike a
generalized implicature this interpretation is not conventionally associated
with this utterance. It is only in the context of the preceding question that
this interpretation can be derived (and hence, it represents a particularized
rather than generalized implicature).
Our research on relevance violations has focused on several issues. One
major question has concerned the specific interpretation that recipients of a
relevance violation will tend to generate. This is not trivial; in theory there are
an infinite number of implicatures a recipient might produce. Grice (1975)
was not particularly clear in this regard. He argued that the implicature would
be one that makes the utterance a cooperative response, one that would fit in
the conversation. This makes perfect sense, but in many instances there are
numerous possible interpretations that would fit in the conversation; the the­
ory narrows down these possibilities somewhat but does not specify exactly
how an utterance will be interpreted. And, of course, it is impossible to specify
precisely, in advance, how an utterance will be interpreted. Still, it seems
likely that additional interpretive constraints could be developed. Research
has focused on how face management (Goffman, 1967) may play a role in the
interpretation of relevance violations.
I assume that the recipient of a relevance violation will construct an in­
ference that represents an attempt to explain why the violation occurred.
This assumption is consistent with attribution research demonstrating that
people attempt to explain why unexpected or unscripted actions occurred
80 HOLTGRAVES

(e.g., Hastie, 1984). It is also consistent with text processing research (e.g.,
Singer, Halldorson, Lear, & Andrusiak, 1992) demonstrating that readers
generate causal inferences as a means of achieving coherence in their repre­
sentation of a text (a property of text that is similar to the conversational re­
quirement of relevance) and with Graesser, Singer, and Trabasso's (1994)
model of text comprehension that is based on a search after meaning (i.e.,
readers' representations are influenced by their attempts to understand why
something is mentioned in the text).
Although there are no doubt many reasons why people might violate the
relation maxim and speak indirectly (e.g., see Roberts & Kreuz, 1994), a
fundamental motivation for indirectness is face management. According to
R Brown and Levinson (1987), indirectness is an important mechanism for
conveying politeness, and all politeness is motivated by concerns for manag­
ing face. The essence of politeness is the performance of a face'threatening
act in a manner that simultaneously attends to the face needs of the
interactants. For example, people frequently perform requests indirectly
(e.g., Could you open the door?) rather than directly (e.g., Open the door);
in this way they attend to the face of the recipient by symbolically lessening
the implied imposition of the act. Research has documented that indirect­
ness (as a form of politeness) is motivated by face management concerns (R.
Brown & Gilman, 1989; Clark & Schunk, 1980; Holtgraves, 2001b;
Holtgraves & Yang, 1990,1992). Note that this relationship has been found
for replies to personal information questions; people will be indirect and vio­
late the relation maxim as a means of managing face (Bavelas, Black,
Chovil, & Mullet, 1990), and replies that do this are perceived as more po­
lite (Holtgraves, 1986).
Almost all research on face management has focused on how speakers'
produce polite remarks; there has been virtually no research on how face
management might impact the manner in which a recipient interprets a
speaker's remark. But there is a straightforward extension here. Because
face management is a major reason for violating Gricean conversational
maxims, it is reasonable to assume that when faced with such a violation
hearers will consider the possibility that the speaker is trying to engage in
face management. This recognition then can serve as the basis for generat­
ing an interpretation of what the speaker actually means with an utterance.
This reasoning seems very likely to occur for replies to personal questions.
So, when Al replies "It's hard to give a good presentation," in response to
Bob's request for feedback, Bob is likely to infer that Al is engaging in face
management. Now, because it is a negative opinion about Bob's presenta­
tion that would be face threatening in this situation, the most likely infer­
ence is that Al does not have a positive opinion of the presentation. If the
information was positive, there would usually be no need to violate the rela­
tion maxim; a positive opinion would not be face threatening. The claim,
4. CONTEXT AND COMPREHENSION 81

then, is that in these situations relevance violations will be interpreted as

conveying nonliteral meanings; because it is negative information that
would be most face threatening in these situations, the most likely interpre­
tation is that the speaker is conveying negative information. Note that not
all relevance violations will be interpreted as conveying negative informa­
tion. Rather, the claim is that they will be interpreted as conveying negative
information if it is negative information that is face threatening.
I found support for this reasoning in several experiments (Holtgraves,
1998b). In these experiments participants read brief descriptions of situa­
tions, each followed by a question-reply exchange (similar to the example
provided earlier). Two types of situations were created, an opinion situation
in which the requested information pertained to the person asking the ques­
tion (e.g., "What did you think of my presentation?") and a self-disclosure
situation in which the requested information pertained to the recipient of
the question (e.g., "How did you do on the chemistry test?").1 The reply al­
ways violated the relation maxim either by completely changing the topic or
by providing an excuse for why the requested information might be nega­
tive. The information requested by the questions was described in the sce­
nario as being either positive, negative, or no information was given. Sample
scenarios are presented in Table 4.1.
In the first experiment participants were asked to interpret each reply and
indicate their degree of confidence in their interpretation. Their interpreta­
tions were coded in terms of whether they were literal or indirect and, if the
latter, whether the interpretation was positive, negative, or neutral. Obvi­
ously, participants should interpret the replies as conveying negative infor­
mation when they are told that the requested information is negative (the
negative information condition in Table 4.1). When no information is pro­
vided (the no information condition in Table 4.1), however, participants
could infer a number of different things (e.g., that the information is nega­
tive, that it is positive, that the speaker is not really giving an opinion at all).
But if participants attempt to uncover a reason for the relevance violation, if
a major reason for violating the relation maxim is to manage face, and if (as
in the present situations) it is negative information that is face threatening,
then participants should be most likely to infer that the information is nega­
tive. That is exactly what happened. Participants were far more likely to in­
terpret the replies as conveying negative information rather than positive or
neutral information. And, more important, they were just as likely to inter­
pret the replies as conveying negative information when no information was
provided (62.2%) as when they were told explicitly that the requested infor­
mation was negative (66.4%).

'in general, I found no meaningful differences between these two types of situations. Ac­
cordingly, in this chapter differences between opinions and disclosures are not discussed.
82 HOLTGRAVES

TABLE 4.1
Sample Experimental Materials

Opinion
Nick and Paul are taking the same history class. Students in this class have to give a
20-min presentation to the class on some topic.
No information: Nick gave his presentation and then decided to ask Paul what he
thought of it.
Negative information: Nick gave his presentation, and it was truly terrible. He
decides to ask Paul what he thought of it.
Positive information: Nick gave his presentation, and it was excellent. He decides to
ask Paul what he thought of it.
Nick: What did you think of my presentation?
Paul: It's hard to give a good presentation. (excuse)
Paul: I hope I win the lottery tonight. (topic change)
Indirect interpretation target: I didn't like your presentation.a, b
Neutral target: I gave her roses for Valentine's Day.b
Self-Disclosure
Jim is in seventh grade. Report cards were due today, and Jim's mother is curious about
how well he did.
No information: Jim just got home from school, and his mother met him at the door.
Negative information: Jim is having a bad semester and flunking several classes. Jim
just got home from school, and his mother met him at the door.
Positive information: Jim is having a great semester and getting As in all but one
class. Jim just got home from school, and his mother met him at the door.
Mom: How were your grades this semester?
Jim: I don't think the teacher grades fairly. (excuse)
Jim: It snowed very hard last night. (topic change)
Indirect interpretation target: My grades are not very good.a, b
Neutral target: His wallet was stolen.b
a
Experiment 2, Hokgraves, (1998b).
b
Experiments 1 through 3, Hokgraves, (1999).

A second experiment tested this logic further. The same materials were
presented to participants on a computer screen, and the time taken for
them to comprehend the replies was recorded. After reading the replies,
participants also indicated whether an indirect interpretation was a rea­
sonable interpretation of the preceding reply (see Table 4.1). When the vi­
gnette makes clear that the requested information is positive (the positive
information condition), then the comprehension of relevance violations
should become quite difficult and time consuming because there is no ap­
4. CONTEXT AND COMPREHENSION 83

parent reason for the violation. In contrast, in the no information and neg­
ative information conditions, the hearer can reasonably assume the
existence of face management as a reason for the violation. Consistent
with this logic, participants took far longer to comprehend replies in the
positive information condition (2766 ms) than in the no information
(2281 ms) and negative information (2263 ms) conditions. Note that the
difference between the latter two conditions was not significant. Judgment
speeds for the interpretations paralleled these results. Participants were far
slower at judging the indirect interpretation in the positive information
condition (2791 ms) than in the negative (2147 ms) and no information
(2090 ms) conditions.
These results provide support for the idea that the specific interpretation
given to a relevance violation is guided by beliefs about the reason for the vi­
olation. Again, I do not claim that relevance violations will always be inter­
preted as conveying negative information; I do claim that if face
management is recognized as a motivation for the violation, then the utter­
ance will tend to be interpreted as conveying face-threatening information.
Occasionally, it might be positive information that is face threatening. For
example, imagine a conversation between two siblings, Mark and John, in
which Mark always outperforms John in school, much to John's chagrin.
Mark is aware of John's feelings and generally tries to manage John's face.
Now, when John asks Mark how he did on his chemistry test, and Mark fails
to answer directly (e.g., "Let's go get a pizza"), John will probably interpret
the reply as conveying positive information (i.e., he did well on his exam)
rather than negative information. In this context it is positive information
that may be face threatening, and so the reply will tend to be interpreted as
conveying positive information.
Do people automatically generate indirect interpretations of relevance
violations when they comprehend the utterance, or are these interpreta­
tions constructed in a post hoc manner when people are asked to interpret
them? In the two experiments just described, participants were explicitly
asked to interpret the replies (Experiment 1) or to judge the adequacy of in­
terpretations of the replies (Experiment 2). It is possible, then, that our re­
sults reflect a post hoc judgment process rather than online comprehension.
I conducted additional experiments to examine this issue (Experiments 1
through 3, Holtgraves, 1999).
In these experiments, participants were presented materials similar to
those used in earlier research (Holtgraves, 1998b). Participants first read a
(no information) scenario, question, and reply on a computer screen, and
then they performed a sentence verification task. For this task a string of
words appeared on the screen and participants were asked to indicate, as
quickly as possible, whether or not the word string formed a sentence.
Sometimes the target string was an indirect interpretation of the reply;
84 HOLTGRAVES

other times it was a neutral string that was not related to the reply (see Ta­
ble 4.1). A pretest was conducted to select indirect and neutral strings that
were equal in comprehension difficulty. If there is any activation of the in­
direct meaning when a reply is comprehended, then sentence verification
judgments for the indirect interpretation targets should be faster than for
the control sentence (a priming effect). This is exactly what happened.
Participants were significantly faster at verifying sentences that were indi­
rect interpretations of the reply than verifying matched control sentences.
This effect occurred for excuses (Experiment 1) and topic changes (Exper­
iment 2). Note that it did not occur when the scenarios were altered so
that the reply did not violate the relation maxim. Finally, this priming ef­
fect is not the result of the indirect targets being more related to the pre­
ceding scenario than to the control targets. This possibility was eliminated
in a third experiment in which participants read only the scenario and
question (and not the reply). Under these conditions the priming effect
did not occur, suggesting that it is the comprehension of the reply in the
context of the preceding question that activates an indirect reading; facili­
tation of the indirect target is not a result of the indirect targets being more
related to the context than the neutral targets. Take together, these results
demonstrate that the indirect meaning of a relevance violation is acti­
vated when an utterance is comprehended; it is not simply the result of a
postcomprehension judgment process.
Finally, I also examined whether recognition of the conveyed meaning
of these replies involves a Gricean inference process (Experiments 4
through 6, Holtgraves, 1999). In two of these experiments, participants
read versions of the question-reply materials described earlier. After indi­
cating comprehension of the reply they performed a timed, sentence verifi­
cation task. On critical trials the target string of words was either a literal
interpretation of the reply (rather than an indirect interpretation as in the
experiments already described) or a neutral sentence matched with the lit­
eral interpretation in terms of comprehension difficulty. For most of the re­
ply types, participants were significantly faster at verifying the literal
interpretations than matched controls. (One particular reply type did not
demonstrate this pattern.) Unlike comprehension of metaphors and other
generalized implicatures, these results are consistent with the standard
pragmatic view that the literal meaning of an indirect utterance will be ac­
tivated during the comprehension process. In another experiment I timed
how long it took participants to read the reply as a function of whether the
preceding context activated or did not activate an indirect (face-threaten-
ing) reading of the reply. Participants took significantly longer to read the
replies when the indirect meaning was activated than when it was not.
This too demonstrates that these types of particularized implicatures re­
quire a Gricean inference process for comprehension.
4. CONTEXT AND COMPREHENSION 85

Preference Organization and Comprehending

Particularized Implicatures
My research (Holtgraves, 1998b; 1999) on reply interpretation demon­
strates that relevance violations are particularized implicatures and that a
nonliteral meaning can only be derived in the context of the preceding ques­
tion. Because the generated implicature is the product of a reasoning pro­
cess, recipients should be sensitive to any features of the context that are
informative as to the speaker's likely intended meaning. Thus, if face man­
agement is recognized as the motive behind a relevance violation, then any
contextual information suggesting that face management processes are op­
erative in this context should facilitate recognition of an indirect
(face-threatening) meaning. I investigated this issue by examining the im­
pact of dispreferred markers on the comprehension of indirect replies.
A question-reply sequence is an adjacency pair (Schegloff &L Sacks,
1973), and as such a reply (the second pair part) is expected given the occur­
rence of the question (the first pair part). As with all adjacency pairs, how­
ever, there is a range of possible second pair parts that could occur, and these
alternatives are not equal; some are preferred, and others are dispreferred.
Dispreferred turns are turns that are marked (in the linguistic sense) in some
way; preferred turns are unmarked. A common means of marking
dispreferred turns is with a delay, a delay that will often include prefaces
such as "well," "yes, but" (with disagreements), and apologies and/or ac­
counts (with request refusals). In contrast, preferred turns are quick, simple,
and direct. By definition, then, dispreferred turns are communicative; they
signal that the current turn is not preferred. As such, dispreferred markers
should help recipients comprehend replies with potential indirect mean­
ings, particularly replies that convey face-threatening information.
My research has focused on well as a marker of dispreferred turns
(Holtgraves, 2000). In general, the occurrence of well at the beginning of an
utterance can be interpreted as indicating that the speaker is engaging in
face management (Jucker, 1993). To refuse a request, decline an offer, and so
on are threatening to the other person's face, and the dispreferred marker
well helps soften this threat (Holtgraves, 1992). Moreover, as discourse ana­
lysts have argued, discourse markers such as well should signal to the recipi­
ent that the remark underway is indirect and needs to be interpreted within
a context that is not immediately apparent (Jucker, 1993; Lakoff, 1973). Be­
cause of this, the occurrence of the dispreferred marker well in a reply should
facilitate recognition of a face-threatening interpretation of that reply.
To examine this issue I modified the materials used in my earlier studies of
indirect replies (Holtgraves, 1998b; 1999). The major change was to manip­
ulate whether the indirect reply contained a well marker. One half of the
time it did (e.g., Well, I think it's hard to give a good presentation), and one
86 HOLTGRAVES

half of the time it did not (e.g., I think it's hard to give a good presentation).
In Experiment 1, participants read the scenario, question, and reply on a
computer screen, and then judged the adequacy of a potential interpreta­
tion of the reply. For the critical trials the interpretation was always a
face-threatening paraphrase of the reply (e.g., I didn't like your presenta­
tion). As expected, face-threatening interpretations of the replies were
more quickly verified when the reply contained a well preface (1534 ms)
than when it did not (1632 ms).
It is possible that the facilitation observed in Experiment 1 occurred be­
cause participants were explicitly asked to judge the meaning of the reply.
Would this facilitation occur automatically? A second experiment was
conducted, using a more online measure, to examine this issue. Partici­
pants in this experiment viewed the same materials as before; one half of
the time the reply contained a well preface, and one half of the time it did
not. However, rather than providing a judgment regarding the meaning of
the reply, participants performed a timed sentence verification task after
reading the reply. On the critical trials, the to-be-judged target string was
either a face-threatening interpretation of the reply (the paraphrases used
in the first experiment) or an unrelated sentence. As in my previous re­
search (Holtgraves, 1999), there was a substantial priming effect; verifica­
tion speeds were much faster for the indirect (face-threatening) targets
(1425 ms) than for the matched neutral targets (1690 ms). More impor­
tant, however, this priming effect was significantly greater when the reply
contained the well marker (a 322 ms difference) than when it did not con­
tain the marker (a 206-ms difference). These results suggest that the im­
pact of a well marker on comprehension is fairly immediate; its' presence
facilitates the comprehension of indirect meanings when the reply is com­
prehended.
It seems likely that other markers of dispreferred turns will play a role in
comprehension similar to that played by a well preface. And there is some
preliminary evidence that they do. In an earlier study (Experiment 3,
Holtgraves, 1998a) participants listened to (rather than read) question-
reply exchanges. In some conditions the reply was briefly delayed (2 s).
Brief delays are a common means of marking a turn as dispreferred, and in
these conditions participants were faster at comprehending the reply, rela­
tive to replies that were not preceded by a brief delay. A delay, similar to a
well marker, serves to mark the turn as dispreferred and, hence, facilitate
recognition of a likely indirect reading. Taken together, these studies dem­
onstrate clearly the role played by paralinguistic and noncontent features
of talk in the comprehension of speaker meaning. This is an important ave­
nue for future investigation as relatively little research has examined the
role played by nonverbal and paralinguistic behaviors in the generation of
conversational implicatures.
4. CONTEXT AND COMPREHENSION 87

Perspective and Particularized Implicatures

One particularly important feature of the context is the perspective of the

interlocutors. Perspective taking, of course, is crucial for language use (e.g.,
R. Brown, 1965; Clark, 1985; Krauss & Fussell, 1991; Mead, 1934;
Rommetviet, 1974), especially for the recognition of indirect expressions.
To successfully use an indirect expression requires the speaker to have some
sense of how the recipient will interpret the utterance, and it requires the
hearer to take the speaker's perspective in producing this particular utter­
ance. Note, in this regard, how Grice's (1975) theory of conversational
implicature requires perspective taking. Consider, for example, the quantity
maxim, or stipulation that one's conversational contribution should be as
informative as required. Obviously, adherence to this principle requires an
assessment of what one's interlocutor knows; one cannot be appropriately
informative without an assessment of the recipient's knowledge. And peo­
ple are sensitive to this; the explanations they give others vary as a function
of what they believe their audience knows (Slugoski, Lalljee, Lamb, &
Ginsburg, 1993).
Perspective taking, however, is not perfect. For example, research sug­
gests that interactants often exhibit an egocentric bias, producing and inter­
preting (at least initially) utterances from their unique perspective, without
taking into account the perspective of their interlocutors (Horton & Keysar,
1996; Keysar, 1998; Keysar, Barr, Balm, & Paek, 1998; Keysar & Bly, 1995).
That is, a people's conversational goals or agendas may blind them to the
possibility that their interlocutors have different goals or agendas (Russell &
Schober, 1999). Hence, speakers may assume they mean one thing without
considering that the hearer might interpret it differently, and hearers may
assume the speaker means one thing without considering that the speaker
may have meant something else. So, even though theories of communica­
tion suggest that perspective taking underlies successful communication, it
is possible that hearers and speakers will systematically diverge in their in­
terpretations.
This raises the issue of whether the comprehension of indirect meaning
might vary as a function of perspective. In my earlier studies (Holtgraves,
1998b, 1999, 2000) participants were asked to read and interpret replies
from the perspective of the recipient of the reply, and there was a strong ten­
dency for them to interpret relevance violations as conveying face-threat-
ening information (i.e., a negative opinion or disclosure). This makes sense
from the recipient's perspective. This person has asked for the opinion, and
failing to obtain a direct and relevant reply should assume that the requested
information is negative. If the information was positive, the speaker would
have said so. However, these participants may have been engaged in
means-end reasoning (Levinson, 2000). That is, they interpreted a rele­
88 HOLTGRAVES

vance violation (the means) as reflecting face concerns (the end); hence,
they interpreted the replies as indirectly conveying negative (i.e., face-
threatening) information. Logically, of course, this is fallacious (affirming
the consequent in deductive logic). That is, the means (a relevance viola­
tion) may occur for reasons other than face management. Consider a rele­
vance violation in response to a request for an opinion. Perhaps the
speaker's opinion was not well-formed or was ambiguous, or perhaps it did
not even exist or was one that the speaker would prefer not to convey. Ail of
these possibilities could be handled with a relevance violation, an utterance
that functions to avoid providing a specific opinion rather than conveying a
negative one. As a result, a speaker's intention in producing a relevance vio­
lation could sometimes be at odds with the recipient's interpretation of the
same utterance.
To test this possibility I asked participants to interpret replies from either
the perspective of the recipient of the reply or the perspective of the speaker,
the person who produces the reply (Holtgraves, 2003). The materials for
these experiments were adapted from those used in my earlier work on the
role of face management in reply comprehension (Holtgraves, 1998b;
1999). If meaning simply resides in an utterance, then speakers and hearers
should agree quite readily in what is being communicated. My results sug­
gest otherwise. In three experiments, participants were far more likely to in­
terpret relevance violations as conveying a negative opinion or disclosure
when they assumed the perspective of the hearer than when they assumed
the perspective of the speaker. This effect occurred for both excuses and
topic changes, as well as for requests for opinions and requests for disclo­
sures. Also, it made no difference whether participants were informed that
the requested information was negative. In addition, it made no difference
that participants alternated assuming the perspective of the speaker and
hearer, a design that should have sensitized them to the possibility that they
were interpreting the replies one way when they took the speaker's view and
another way when they took the hearer's perspective. Despite these various
features of the design, in the end replies were more likely to be interpreted as
conveying negative information when participants took the hearer's per­
spective than when they took the speaker's perspective.
This research, then, suggests that there may be a systematic, interpretive
bias built into the hearer and speaker roles. The tendency for hearers to in­
terpret relevance violations as conveying negative information is consistent
with my earlier research (Holtgraves, 1998b, 1999, 2000). Because indirect
replies are frequently used as a means of managing face, and because inter­
actants are generally aware that this is the case, recipients should interpret a
relevance violation as conveying face-threatening information (a negative
opinion or disclosure). But indirect utterances are inherently ambiguous
and open to multiple interpretations. The means–ends reasoning used by a
4. CONTEXT AND COMPREHENSION 89

recipient to recover a face-threatening interpretation can yield errors. From

the speaker's perspective, a relevance violation may be a means of signaling
an unwillingness to provide the requested information. Hence, speakers
may not believe that they are communicating negative information, but in­
stead believe they are communicating an unwillingness to convey an opin­
ion or disclosure.

Speaker Status and Nonliteral Meaning

Various speaker variables—such as his or her occupation (Katz & Pexman,

1997)—provide contextual cues regarding how a speaker's utterance is to
be interpreted. One very important variable in this regard is the speaker's
relative status (or power). Previous research has focused on language pro­
duction and demonstrated that the greater the status of the recipient, the
greater the likelihood that the speaker will use polite (or indirect) request
forms (P. Brown & Levinson, 1987; Holtgraves, 2001b; Holtgraves &Yang,
1990, 1992). Because of its role in language production, it seems likely that
speaker status will play an important role in comprehension. My research
suggests that it does. For example, in my research on the processing of con­
ventional and nonconventional indirect requests (Holtgraves, 1994), the
status of the speaker moderated the effects of request conventionality. For
conventional indirect requests (e.g., Can you shut the door?), the status of
the speaker did not affect comprehension; these forms were recognized
quickly and directly, without any type of inference process (they appear to be
generalized implicatures). In contrast, the comprehension of nonconven­
tional forms was affected by the status of the speaker. Participants were faster
at comprehending these forms when the speaker was higher rather than
equal in status to his or her conversation partner. In addition, priming exper­
iments demonstrated that both the literal and the indirect meanings were
activated for these forms when the speaker was equal status; this did not oc­
cur when the speaker was high status. These results thus demonstrate that a
Gricean inference process appears to be required for the comprehension of
nonconventional forms performed by an equal-status speaker, a process that
appears not to be necessary when the speaker is high status.
Similar to face management, speaker status is a feature of the social con­
text that impacts both utterance production and utterance comprehension.
But exactly how does status play a role in the comprehension of requests?
There are at least two possibilities. The first possibility is that knowledge
that a speaker is high status may circumvent the need for an inference pro­
cess; a directive interpretation may be activated prior to any activation of
the literal meaning of the utterance. Thus, when a speaker is high status, a
recipient may be inclined to interpret the speaker's utterances a priori as di­
rectives (cf. Ervin-Tripp, Strage, Lampert, & Bell, 1987). The finding that
9O HOLTGRAVES

an inference process is not required for nonconventional indirect requests

when the speaker is high status supports this possibility (Holtgraves, 1994).
Additional empirical support for this idea is provided by a priming study I
conducted demonstrating that directive readings can be activated based
solely on the status of the speaker (Experiment 4, Holtgraves, 1994). Partici­
pants in this study first read descriptions of situations in which speaker sta­
tus was varied, and then they made sentence verification judgments of
targets that either were or were not possible directives in each situation. The
requests themselves were not presented. When the speaker was high status,
sentence verification times were significantly faster for directives than for
related control targets (i.e., a priming effect), and this effect did not occur
when the speaker was equal status. In short, speaker status alone (without
the utterance) was sufficient for priming a directive interpretation.
A second possibility is that knowledge that a speaker is high status can fa­
cilitate an inference process if one is required. Rejection of the literal mean­
ing of an utterance depends in part on the possibility that there are
alternative interpretations of the utterance (e.g., Sperber & Wilson, 1986).
People who are high status have the right to issue directives; hence, there
exist possible directive interpretations of their utterances. The existence of
these interpretations, then, increases the likelihood that the literal meaning
will be rejected. Moreover, given that the literal meaning is rejected, knowl­
edge that the speaker is high status can increase the likelihood that a direc­
tive interpretation will be adopted.
Regardless of how speaker status impacts the comprehension of speaker
meaning, other research suggests that speaker status can influence
long-term memory for a speaker's remarks. In several experiments we had
participants read or listen to dialogues in which we manipulated the ostensi­
ble status of one of the speakers (Holtgraves, Srull, & Socall, 1989). In these
experiments there was a tendency for the utterances of high-status speakers
to be recalled as being more assertive (i.e., direct) than identical remarks ut­
tered by an equal-status speaker. These results are consistent with our re­
search demonstrating the impact of speaker status on comprehension (high
speaker status prompts a directive reading, a reading that is then retained in
long-term memory). However, I conducted additional studies demonstrat­
ing that, at times, if the wording of an utterance is at odds with normative
expectations, these wordings will tend to be remembered well. In one exper­
iment (Experiment 3, Holtgraves, 1997a) I gave participants a surprise
memory test for the experimenter's instructions. The ostensible status of the
experimenter was manipulated (high-status professor vs. equal-status un­
dergraduate research assistant) as was the politeness (or directness) of his
utterances. On subsequent memory tests, participants were more likely to
remember the polite remarks of the high-status speaker and the impolite re­
marks of the equal-status speaker. In this situation, it is utterances violating
4. CONTEXT AND COMPREHENSION 91

normative expectancies that receive additional processing and, hence, that

are remembered relatively well.

CULTURAL DIFFERENCES IN COMPREHENDING

NONLITERAL MEANINGS

Nonliteral meanings, by definition, are ambiguous. Hence, there is the pos­

sibility that people will differ in their interpretation of utterances with po­
tential indirect meanings, as my research on speaker-hearer differences
suggest (Holtgraves, 2003). And there may be systematic cultural differ­
ences in this regard. In this case the cultural background of the interlocutors
will play an important role in the comprehension of nonliteral meanings.
This issue is related to the criticism of Grice's (1975) model as not being
cross-culturally valid. For example, in a well-known article Keenan (1976)
argued that people in Malagasy routinely withhold information from one
another (information is a culturally prized commodity), an action that is in
clear violation of the quantity maxim. Because of this, violations of the
quantity maxim in Malagasy do not usually result in conversational impli­
catures (as presumably would be the case in most other cultures). There is a
more general criticism here. It has been suggested that Grice's view, with its
emphasis on individual autonomy, has relevance only in Western cultures
(Fitch & Sanders, 1994). For example, Rosaldo (1982), in her analysis of
Llongot speech acts, argued that directives in that culture are not particu­
larly face threatening, referencing as they do group membership and respon­
sibility rather than individual wants and desires. Hence, directives in that
culture will usually not be performed indirectly.
It is clearly possible that the maxims people follow in communicating co­
operatively may vary over cultures, perhaps reflecting differences in what is
regarded as rational interaction. Note, however, that even if there are cul­
tural differences in this regard, these differences will not alter the basic
Gricean insight that if a person believes a maxim has been violated, then
some interpretive work will be undertaken. In addition, Grice's (1975) fun­
damental point that conversational maxims (whatever they may look like)
serve as a basis for rational interaction and, hence, as a framework for inter­
pretation will not be altered.
Still, it is clearly important to determine the specific content of conversa­
tion maxims and how they might vary over cultures. This issue has been ex­
plored in various ways. One research line has been to examine overall
cultural differences in the tendency to interpret utterances indirectly. This
research has been guided by the cultural dimensions of individualism and
collectivism (Triandis, 1995). In general, collectivism entails a relatively
greater concern for the needs, feelings, and wants of one's in-groups; it rep­
resents a relatively greater concern for the face of in-group members
92 HOLTGRAVES

(Ting-Toomey, 1988). Because one of the major means for attending to face
is indirectness (P. Brown & Levinson, 1987), it seems reasonable that indi­
rectness will be more common in collectivist cultures than individualistic
cultures. I examined cultural variability in indirectness using the Conversa­
tional Indirectness Scale (CIS; Holtgraves, 1997b). The CIS is a 19-item,
self-report measure assessing both an interpretation dimension (the extent
to which a person looks for indirect meanings in the remarks of others) and a
production dimension (the extent to which a person phrases his or her re­
marks indirectly). My initial research demonstrated that this a reliable and
valid measure of these two dimensions. For example, people scoring high on
the interpretation dimension are more likely to recognize indirect meanings
and to be significantly faster at doing so than people scoring low on this di­
mension (Experiment 6, Holtgraves, 1997b). Note that the latter difference
occurred only for relatively unconventional forms (i.e., particularized
implicatures), forms for which an inference process is most likely required.
To examine cultural variability I (Holtgraves, 1997b) translated the CIS
into Korean and administered it to a sample of students at Korean universi­
ties. Factor analyses of the scores from this sample were highly similar to the
factor analytic results based on a sample of American students. Accordingly,
I compared the two samples in terms of their scores on the CIS. Overall, Ko­
rean students scored significantly higher on both the interpretation dimen­
sion (53.26 vs. 43.62) and production dimension (42.37 vs. 33.61). These
results provide some support for the notion that people in collectivist cul­
tures tend to produce and look for indirect meanings to a greater extent than
people in individualistic cultures. Recently, Hara and Kim (2001) demon­
strated that these broad cultural differences may be partly explained in
terms of differing self-cons truals. Specifically, they found that individuals
with interdependent self-construals were more likely to speak and interpret
indirectly than were people with independent self-construals. Of course,
not all collectivist and individualistic cultures are alike (e.g., Kashima et al.,
1995). And cultural differences in indirectness reflect only average differ­
ences. People in all cultures will vary the politeness (and thus the indirect­
ness) of their remarks as a function of the context. And, in fact, it appears
that people in collectivist cultures are more sensitive and responsive to the
context than are people in individualistic cultures (Gudykunst, Yoon, &
Nishida, 1987; Holtgraves & Yang, 1992; Triandis, Bontempo, Villareal,
Asai, & Lucca, 1988).
Another way in which cultural variability has been explored is in terms of
cultural differences regarding which specific maxims are likely to result in
conversational implicatures (Holtgraves & Drozd, 1998). Participants in
this study were students at Ball State University who were from either an in­
dividualistic culture (the United States) or collectivist cultures (East and
Southeast Asia). We created scenarios and corresponding dialogs with one
4. CONTEXT AND COMPREHENSION 93

utterance designed to convey an indirect meaning. Across these scenarios

we systematically manipulated which specific maxim was violated. For vio­
lations of the relevance maxim we also included different utterance types
(request refusals, opinions, and self-disclosures). Participants read a sce­
nario and the target utterance and then chose which of three possible inter­
pretations (one literal interpretation and two indirect interpretations) best
captured the speaker's meaning in this situation. There was a reliable Cul­
ture X Maxim Type interaction. For all but one utterance type, American
participants were less likely to interpret utterances indirectly than the East
and Southeast Asians, a finding that is consistent with research demonstrat­
ing that collectivists tend to be more indirect than individualists. However,
this pattern was reversed for violations of the quality maxim: East and
Southeast Asians were less likely to interpret the utterances indirectly than
were the American participants. Although these findings are preliminary,
they do suggest that it is too simplistic to view collectivists as more indirect
than individualists. Instead, there may be cultural variability in interpreta­
tion as a function of how the indirect meaning is conveyed (and, no doubt,
as a function of many other differences as well). This is clearly an important
avenue for future research.

CONCLUSION

Nonliteral meanings can be conveyed in a variety of ways, with corre­

sponding differences in how those meanings are processed. One major dis­
tinction useful for classifying nonliteral meanings involves the extent to
which recognition of their intended meaning is dependent on the immedi­
ate context. Some nonliteral meanings—generalized implicatures in
Grice's (1975) terms—tend to be recognized regardless of the context in
which they are used. In other words, they are highly conventionalized and
the default, nonliteral, meaning is recognized in a relatively direct fashion,
without need of a relatively time-consuming inference process. The em­
pirical evidence is really quite clear in this regard. Most figurative expres­
sions are recognized quickly and without any activation of their literal
meanings. Comprehension of conventional indirect requests is also rela­
tively quick and not influenced by features of the context such as speaker
status. Although not examined empirically, it seems likely that
Q-implicatures and WXDY constructions will also be recognized quickly
and directly. But how exactly are these nonliteral meanings recognized?
Although we know a Gricean inference process (recognize and then reject
the literal meaning) is not involved, we do not know what processes are in­
volved in their recognition. For example, which features of the utterance
or the context activate the nonliteral meaning? Is the entire meaning acti­
vated at once, or does recognition occur in stages?
94 HOLTGRAVES

Although generalized implicatures have default meanings that are acti­

vated regardless of the context, it is possible that it is the literal meaning that
is intended by the speaker and that will be recognized by the recipient. Thus,
"He spilled the beans" might be intended literally and taken as such when
reporting on the recent activities of one's preschooler. But in this case the in­
tended meaning will be the result of an inference process of some sort as the
recipient recognizes and then rejects the default nonliteral meaning. In
short, even though generalized implicatures have preferred meanings that
are independent of the context, those preferred meanings can be overridden
by the context. And when this happens, it appears to be the result of a
Gricean process in reverse—first recognize and then reject the nonliteral
meaning—with the literal meaning chosen because of its fit with the con­
versation and the context.
In contrast to generalized implicatures, the interpretation of particular­
ized implicatures is completely dependent on the context; the nonliteral
meaning is not attached a priori to the utterance that carries it. As a result,
particularized implicatures (at least those that have been investigated) re­
quire a time-consuming inference process for their recognition. And be­
cause it is a reasoning process that is required, various aspects of the
context—status, face management, and so on—will play a critical role in
the process of interpretation.
Particularized implicatures have not received as much empirical atten­
tion as generalized implicatures. As a result, not very much is known regard­
ing how they operate. For example, what mechanisms, other than relevance
violations, can yield particularized implicatures? And what features of the
context will play a role in its comprehension? And to what extent are there
cultural differences in the interpretation of particularized implicatures? Pur­
suing these and other questions will increase our understanding of the com­
prehension of nonliteral meaning.

ACKNOWLEDGMENTS

Much of the research described in this chapter was supported by grants from
the National Science Foundation (SBR–9601311 and 0131877), National
Institute of Mental Health (MH45747), and Ball State University Office of
Research and Sponsored Programs

REFERENCES

Bavelas, J. B., Black, A., Chovil, N., & Mullet, J. (1990). Equivocal communication.
Newbury Park, CA: Sage.
Brown, P, & Levinson, S. (1987). Politeness: Some universals in language usage. Cam­
bridge, England: Cambridge University Press.
4. CONTEXT AND COMPREHENSION 95

Brown, R. (1965). Social psychology. New York: The Free Press.

Brown, R., & Gilman, A. (1989). Politeness theory and Shakespeare's four major
tragedies. Language in Society, 18, 159-212.
Cacciari, C., & Glucksberg, S. (1994). Understanding figurative language. In M. A.
Gernsbacher (Ed.), Handbook of psycholinguistic research (pp. 447-478). San
Diego: Academic Press.
Clark, H. H. (1979). Responding to indirect speech acts. Cognitive Psychology, 11,
430-477.
Clark, H. H. (1985). Language use and language users. In G. Lindzey & E. Aronson
(Eds.), The handbook of social psychology (3rd ed., Vol. 2, pp. 179-232). Reading,
MA: Addison-Wesley
Clark, H. H., & Schunk, D. (1980). Polite responses to polite requests. Cognition, 8,
111-143.
Ervin-Tripp, S., Strage, M., Lampert, M., &. Bell, N. (1987). Understanding re­
quests. Linguistics, 25, 107-143.
Fitch, K. L., & Sanders, R. E. (1994). Culture, communication, and preferences for
directness in expression of directives. Communication Theory, 4, 219-245.
Gibbs, R. W., Jr. (1980). Spilling the beans on understanding and memory for idi­
oms. Memory & Cognition, 8, 449-456.
Gibbs, R. W., Jr. (1983). Do people always process the literal meaning of indirect re­
quests? Journal of Experimental Psychology: Learning, Memory, and Cognition, 9,
524-533.
Gibbs, R. W., Jr. (1994). Figurative thought and figurative language. In M. A.
Gernsbacher (Ed.), Handbook of psycholinguistic research (pp. 411–446). San
Diego: Academic Press.
Gildea, P.,& Glucksberg, S. (1983). On understanding metaphor: The role of con­
text, journal of Verbal Learning and Verbal Behavior, 22, 577-590.
Glucksberg, S. (1991). Beyond literal meanings: The psychology of allusion. Psycho­
logical Science, 2, 146–152.
Glucksberg, S., Gildea, P., & Bookin, H. (1982). On understanding nonliteral
speech: Can people ignore metaphors? Journal of Verbal Learning and Verbal Be­
havior, 21, 85-98.
Goffman, E. (1967). Interaction ritual: Essays on face to face behavior. Garden City,
NY: Anchor.
Graesser, A. C., Singer, M., &Trabasso, T. (1994). Constructing inferences during
narrative text comprehension. Psychological Review, 101, 371-395.
Grice, H. P (1975). Logic and conversation. In P. Cole & J. Morgan (Eds.), Syntax
and semantics: Vol. 3. Speech acts (pp. 41-58). New York: Academic Press.
Gudykunst, W., Yoon, Y. C., & Nishida, T. (1987). The influence of individualism-
collectivism on perceptions of communication in in-group and out-group rela­
tions. Communication Monographs, 54, 295-306.
Hara, K., & Kim, M. S. (2001, May). The effect of self-construals on conversational in­
directness. Paper presented at International Communication Association Con­
vention, Washington, DC.
Hastie, R. (1984). Causes and consequences of causal attributions. Journal of Per­
sonality and Social Psychology, 46, 44-56.
96 HOLTGRAVES

Holtgraves, T. M. (1986). Language structure in social interaction: Perceptions of

direct and indirect speech act and interactants who use them. journal of Personal­
ity and Social Psychology, 51, 305–314.
Holtgraves, T. M. (1992). The linguistic realization efface management: Implica­
tions for language production and comprehension, person perception, and
cross-cultural communication. Social Psychology Quarterly, 55, 141-159.
Holtgraves, T. M. (1994). Communication in context: Effects of speaker status on
the comprehension of indirect requests, journal of Experimental Psychology:
Learning, Memory, and Cognition, 20, 1205-1218.
Holtgraves, T M. (1997a). Politeness and memory for the wording of remarks. Mem­
ory & Cognition, 25, 106–116.
Holtgraves, T. M. (1997b). Styles of language use: Individual and cultural variability
in conversational indirectness, journal of Personality and Social Psychology, 73,
624-637.
Holtgraves, T. M. (1998a, July). Interpreting indirect replies. Paper presented at Inter­
national Communication Association Convention, Jerusalem, Israel.
Holtgraves T M. (1998b). Interpreting indirect replies. Cognitive Psychology, 37,1-27.
Holtgraves, T. M. (1999). Comprehending indirect replies: When, and how are their
conveyed meanings activated. Journal of Memory and Language, 41, 519-540.
Holtgraves, T. M. (2000). Preference organization and reply comprehension. Dis­
course Processes, 30, 87-106.
Holtgraves, T. M. (200la). Language as social action: Social psychology and language
use. Mahwah, NJ: Lawrence Erlbaum Associates.
Holtgraves, T. M. (2OOlb). Politeness. In W. P. Robinson & H. Giles (Eds.), The new
handbook of language and social psychology (pp. 341-355). Chichester, England:
Wiley.
Holtgraves, T. M. (2003). Conversational perspective taking: Diverging assumptions of
speakers and hearers.
Holtgraves, T. M., & Ashley, A. (2001). Comprehending illocutionary force. Mem­
ory & Cognition, 29, 83-90.
Holtgraves, T. M., & Drozd, B. (1998, May). Cross-cultural differences in conversa­
tional interpretation. Paper presented at the annual meeting of the Midwestern
Psychological Association, Chicago, IL.
Holtgraves, T. M., Srull, T. K., & Socall, D. (1989). Conversation memory: Effects of
speaker status on memory for the assertiveness of conversation remarks, journal
of Personality and Social Psychology, 56, 149-160.
Holtgraves, T. M., & Yang, J. N. (1990). Politeness as an universal: Cross-cultural
perceptions of request strategies and inferences based on their use. journal of Per­
sonality and Social Psychology, 59, 719-729.
Holtgraves, T. M., &Yang, J. N. (1992). The interpersonal underpinnings of request
strategies: General principles and differences due to culture and gender. Journal
of Personality and Social Psychology, 62, 246-256.
Horton, W. S., & Keysar, B. (1996). When do speakers take into account common
ground? Cognition, 59, 91-117.
Jucker, A. H. (1993). The discourse marker well: A relevance-theoretical account.
journal of Pragmatics, 19,435-452.
4. CONTEXT AND COMPREHENSION 97

Kashima, Y., Yamaguchi, S., Kim, U, Choi, S., Gelfand, M. J., & Yuki, M. (1995).
Culture, gender, and the self: A perspective from individualism-collectivism re­
search. Journal of Personality and Social Psychology, 69, 925—937.
Katz, A., &. Pexman, E (1997). Interpreting figurative statements: Speaker occupa­
tion can change metaphor to irony. Metaphor and Symbol, 12, 19–41.
Kay, P., & Fillmore, C. J. (1999). Grammatical constructions and linguistic general­
izations: The what's X doing Y? construction. Language, 75, 1-33.
Keenan, E. O. (1976). The universality of conversational implicature. Language in
Society, 5, 67-80.
Keysar, B. (1989). On the functional equivalence of literal and metaphorical inter­
pretation in discourse. Journal of Memory and Language, 28, 375-385.
Keysar, B. (1994). Discourse context effects: Metaphorical and literal interpreta­
tions. Discourse Processes, 18, 247-269.
Keysar, B. (1998). Language users as problem solvers: Just what ambiguity problem do
they solve? In S. Fussell & R. Kreuz (Eds.), Social and cognitive approaches to interper­
sonal communication (pp. 175-200). Mahwah, NJ: Lawrence Erlbaum Associates.
Keysar, B., Barr, D. J., Balin, J. A., & Paek, T. (1998). Definite reference and mutual
knowledge: A processing model of common ground in comprehension. Journal of
Memory and Language, 39, 1-20.
Keysar, B., & Bly, B. (1995). Intuitions of the transparency of idioms: Can one keep a
secret by spilling the beans? Journal of Memory and Language, 34, 89-109.
Krauss, R. M., & Fussell, S. R. (1991). Perspective taking in communication: Repre­
sentations of others' knowledge in reference. Social Cognition, 9, 2-24.
Lakoff, R. (1973). Questionable answers and answerable questions. In B. B. Kachru
(Ed.), Issues in linguistics: Papers in honor of Henry and Renee Kahane (pp. 453-467).
Chicago: University of Illinois Press.
Levinson, S. C. (2000). Presumptive meanings. Cambridge, MA: MIT Press.
Mead, G. H. (1934). Mind, self, and society. Chicago: University of Chicago Press.
Ortony, A., Schallert, D., Reynolds, R., & Antos, S. (1978). Interpreting metaphors
and idioms: Some effects of context on comprehension. Journal of Verbal Learning
and Verbal Behavior, 16, 465–477.
Roberts, R. M., & Kreuz, R. J. (1994). Why do people use figurative language? Psy­
chological Science, 5, 159-163.
Rommetviet, R. (1974). On message structure:A framework for the study of language
and communication. New York: Wiley.
Rosaldo, M. Z. (1982). The things we do with words: Llongot speech acts and
speech act theory in philosophy. Language in Society, 11, 203-237.
Russell, A. W, & Schober, M. F. (1999). How beliefs about a partner's goals affect re­
ferring in goal-discrepant conversations. Discourse Processes, 27, 1-33.
Schegloff, E., & Sacks, H. (1973). Opening up closings. Semiotica, 8, 289-327.
Singer, M., Halldorson, M., Lear, J. C., & Andrusiak, P. (1992). Validation of causal
bridging inferences in discourse understanding. Journal of Memory and Language,
31,507-524.
Slugoski, B., Lalljee, M., Lamb, R., &Ginsburg, G. P (1993). Attributions in conver­
sational context: Effects of mutual knowledge on explanation giving. European
Journal of Social Psychology, 23, 219-238.
98 HOLTGRAVES

Sperber, D., & Wilson, D. (1986). Relevance. Cambridge, MA: Harvard University
Press.
Ting-Toomey, S. (1988). Intercultural conflict styles. In Y. Y. Kim & W. B.
Gudykunst (Eds.), Theories in interculturalcommunication (pp. 213-238). Beverly
Hills, CA: Sage.
Titone, D. A., & Connine, C. M. (1994). Comprehension of idiomatic expressions:
Effects of familiarity and literality. Journal of Experimental Psychology: Learning,
Memory and Cognition, 20, 1126-1138.
Triandis, H. (1995). Individualism and collectivism. Boulder, CO: Westview.
Triandis, H., Bontempo, R., Villareal, M., Asai, M., & Lucca, N. (1988). Individual­
ism and collectivism: Cross-cultural perspectives on self-ingroup relationships.
Journal of Personality and Social Psychology, 54, 323-333.
 5

Social and Cultural Influences

on Figurative and Indirect

Language

Herbert L. Colston
University of Wisconsin-Parkside

Accounts of language comprehension have long recognized that language is

inherently underdetermined—the meaning of a statement or other instance
of spoken or written language is necessarily less than or potentially different
from the reasonably predictable intended comprehended meaning of a use
of that language by a speaker or a writer, for a hearer (s) or a reader (s), in a
context. Processes such as referential assignment, ambiguity resolution, in­
ference generation, and others nearly always take place in language compre­
hension and result in intended comprehended meanings that go far beyond
language meanings. Indeed, even relatively straightforward, veridical lan­
guage (e.g., the oft used, "The cat is on the mat") is at best a skeleton of the
meaning comprehended from that language by a hearer or reader in a
real-world context.
Often considered, and in my view erroneously so, a separate and, indeed,
by some scholars special category of language are those forms where under-
determinedness is just the beginning to how the language meaning and its
intended comprehended meaning are separated. This category of language
is comprised of indirect forms, figurative language, idiomatic, proverbial,
colloquial, and so on—expressions wherein the decontextualized meaning
of the language is more than just an underdetermined scaffold of the in­
tended comprehended meaning. The mismatch between what is said and
what is meant is instead (or additionally) one of seeming opposition (verbal
99
1QO COLSTON

irony), seemingly disparate topicality (metaphor), seemingly unequivalent

magnitude (hyperbole), seeming unrelatedness (idioms, proverbs, colloqui­
alisms, etc.), seeming laterality (emintended directives), and so on. Com­
prehension of these forms is thus an arguably more complicated matter of
gleaning the intended comprehended meaning from an underdetermined
and nonveridical language meaning.
Generally speaking, the former veridical language category wherein
underdeterminedness is the only form of separation between language
meaning and comprehended meaning has been referred to as literal language,
and the latter category where underdeterminedness is coupled with other
forms of meaning separation has been called nonliteral language.
Whether one considers literal and nonliteral types of language to be sepa­
rate categories or, as would be my preference, areas of language that, al­
though potentially overlapping, typically have the tendency to be toward
different ends of a continuum of language that varies from relatively little
complexity, density, or richness to relatively more, one must still acknowl­
edge that all language comprehension necessarily involves computation of
the speaker's intended meaning. Note that this requires consideration of
something other than the language meaning itself—something rather
ephemeral that has typically been given the umbrella term of context.
Traditional comprehension explanations have handled this role of con­
text typically by splitting comprehension into separate processes, one that
composes a semantic meaning of what is said from lexical and syntactic, as
well as semantic, information in the spoken or written language and a sec­
ondary one that then uses context to assign reference, solve ambiguities,
and compute inferences. Last comes a final process, whose extent is op­
tional, of deriving pragmatic ramifications of the use of that spoken or writ­
ten language in that context (for the given interlocutors, their social
relationship, etc.) to arrive at the speakers particular full intended meaning.
Recent revisions to this general separation strategy have begun to recog­
nize the interdependency or even the indistinguishability of these purport­
edly separate comprehension parts, and some have even done away with
some of the lexical-semantic initial processes altogether. Such accounts put
much more emphasis on top-down, context-, and expectancy-driven pro­
cesses in language comprehension.
A review of the history and current status of language comprehension
accounts is beyond the intent and scope of this chapter. Suffice it for now
to say that a movement toward incorporating contextual contributions to
language comprehension both overall and as a component of lower and
lower level bottom-up processing has taken place, and most accounts now
grapple with the fact that an intended comprehended meaning is built
from both linguistic input and context in a very complex self-dependent
relationship.
5. INFLUENCES ON LANGUAGE 1O1

However, this heightened appreciation of contextual information in lan­

guage comprehension has been limited. Perhaps unfortunately, consider­
ation of context has all too often omitted social and cultural variables of the
interlocutors and instead focused on information in the present physical en­
vironment or in the memories of the parties involved. Perhaps this is due to
the relatively heightened status given to the referential assignment over in­
terlocutor pragmatic ramification stages in traditional language processing
accounts. Nonetheless, until very recently, social and cultural variables
have not played much of a role in explanations of literal or even nonliteral
language comprehension—particularly in processing.
In this chapter I first argue for the necessity of inclusion of social/cultural
variables in consideration of all forms of language processing, comprehen­
sion, and use, particularly so for nonliteral forms. Then, I briefly review some
past and current research that concurs with this argument. Next, I demon­
strate the necessity of attending to the social/cultural components of con­
text by delineating how an understanding of the use of two forms of
nonliteral language requires this attention. Finally, I attempt to drive home
the general argument with a brief discussion of a very recent study of mine
that has begun to delineate a powerful overall mechanism that strongly ar­
gues for the crucial importance of social and cultural variables in nonliteral
language comprehension.

ARGUMENTS FOR CONSIDERATION OF SOCIAL/

CULTURAL INFLUENCES IN NONLITERAL
LANGUAGE COMPREHENSION

The relative omission to date of social and cultural influences on language

comprehension in our developing accounts is unfortunate. Two compelling
lines of reasoning strongly indicate that leaving social and cultural influ­
ences out of our explanations of all forms of language comprehension is lim­
iting our understanding of this important area of human cognition. The first
of these is a logical argument; the second is an argument by analogy. These
arguments are then followed by a discussion concerning the relative de­
pendence of literal and nonliteral language on social/cultural variables.

Argument 1: Contextual Structure Versus Contextual Content

To justify that heretofore acknowledged contextual influences on language

comprehension—such as situation models built up from preceding text or
talk or situational schematic information—are worthy of consideration,
whereas other contextual influences—such as social or cultural variables of
the interlocutors—are not, these sources must be shown to be fundamen­
tally separate or different. There is very little support for such a split. Al­
102 COLSTON

though it is possible to have different kinds of contextual support for the

interpretation of some target utterance, and these different kinds of support
will undoubtedly alter how that utterance is comprehended, I argue that
these differences are primarily structural rather than content based; there­
fore, social and cultural variables should be considered to be as important as
situational ones.
Consider first a concrete example of literal language usage. A variety of
comprehension accounts would now acknowledge that the intended com­
prehended meaning of fish, in (1):

(1) He likes fish.

would be very different when comprehended in context (2) versus (3):

(2) Your young daughter belongs to a soccer league that is expanding.

The league is planning on forming a new team, and several players
have been asked to help think of names for the team. Your daughter
is one of the players. She is at a loss for ideas, though, and asks for
your help. You don't have a ready suggestion either, so you ask her
what name her coach might like. Your daughter says,
"He likes fish."

(3) Your widowed mother has remained single for some time now, but
has finally decided to start dating. She recently introduced you to a
new guy she's been seeing who you thought would be great for her.
You invite them for dinner, and want him to have a good time, but
you don't know what to serve. You ask your mother for a suggestion
and she says,
"He likes fish."

The intended comprehended meaning of fish in (2) would likely involve

instances of fish such as shark, barracuda, or killer whale, whereas in (3) it
would instead involve instances such as salmon, cod, or whitefish. And, im­
portant technical differences in these processing accounts aside, most ac­
counts would also generally agree that this different intended
comprehended meaning of fish arises very early in processing and is influ­
enced by the preceding semantic information in the context passages.
How do these different contexts produce different intended compre­
hended meanings of the same word? Again, skirting technical difference in
the accounts, in general they agree that the first passage builds up a semantic
representation of a category of something like "names for sports teams,"
whereas the second passage builds up a semantic representation for the cate­
5. INFLUENCES ON LANGUAGE 1O3

gory of "foods for a dinner entree." Thus, before encountering the word fish,
a very different situation model is in place in the working memory of the
reader or listener. These different situational models then constrain which
senses of fish are activated, to use the processing jargon, when the word is
read or heard, and thus a different intended comprehended meaning is de­
rived but with a relatively equivalent time course in the different contexts.
However, such different processing environments, to use a broader term
than situation models, are not purely dependent on a long series of sen­
tences as in the just-noted scenarios. They can also be invoked by a variety
of other sources, including social or cultural variables that can be changed
with the replacement of a very few, or indeed just one, word(s). Consider the
comprehension of (4) in the contexts (5) versus (6):

(4) "Try the French place up the street, they serve an exquisite duck."

(5) You arrive in a new city you've never visited before, and are hoping
to get a good recommendation for a restaurant from a local citizen.
You are walking to your hotel and you meet a small group of people
waiting for a bus on the street. You strike up a conversation with a
few of them and say you are new in town and ask if someone could
recommend a good restaurant for dinner. A wealthy looking busi­
ness man in the group says,

"Try the French place up the street, they serve an exquisite duck."

(6) You arrive in a new city you've never visited before, and are hoping
to get a good recommendation for a restaurant from a local citizen.
You are walking to your hotel and you meet a small group of people
waiting for a bus on the street. You strike up a conversation with a
few of them and say you are new in town and ask if someone could
recommend a good restaurant for dinner. A poor looking homeless
man in the group says,
"Try the French place up the street, they serve an exquisite duck."

The different scenarios conjured up by contexts (5) and (6) would very
likely alter the comprehension of (4). At the very least, even if the recom­
mendation were taken seriously in both instances, the comprehension in (6)
would likely involve a degree of curiosity, skepticism, or hint of unusualness
greater than that in (5).
Note that, the source for the different interpretations in (5) and (6) is
fundamentally the same as the source for different interpretations of (2) and
(3)—the processing environment difference in which the target statement
is comprehended. For the latter scenarios, the different processing environ­
104 COLSTON

merits are invoked by altering a very few words in a passage that change the
kind of person making the statement. For the former scenarios, the process­
ing environments are altered by broader situation model descriptions that
are spelled out over several lines, which changes one's thinking regarding
the kind of fish.
One could make an argument that the more concentrated method of in­
voking processing environment differences in the latter scenarios is a stron­
ger manipulation than the more diluted method in the former ones. The
more concentrated presentation in (5) and (6) allows for greater juxtaposi­
tion of the processing environment with the target remark. This stronger
juxtaposition can enable greater contrast that can, in turn, greatly alter the
comprehension of the target utterance (Colston, 2002a). Alternatively, one
could also make a counterargument that the slow buildup allowed by the
former scenarios can create more interpretive momentum and thus send the
different interpretations of the target utterance off in broadly disparate di­
rections.
Essentially, both of these arguments have validity. Different means of pre­
senting distinct processing environments can afford different ways in which
interpretations of a target utterance can divulge. But, aside from the means
by which these processing environments are established, there is no logical
reason to suppose that the processing environment differences in the latter
scenarios are somehow more or less effective than those in the former. The
point still holds that the social/cultural differences in the former scenario
versus the situational differences in the latter scenarios are really just a mat­
ter of technique—they are not something based upon content.
Indeed, one could easily reverse the pairings of the concentrated or di­
luted mechanism just discussed and the kind of processing environment
scenarios (e.g., social/cultural vs. situational) and alter interpretation dif­
ferences accordingly. All one would need to do is invoke a social/cultural
processing environment difference with several lines worth of description
and invoke a situational processing environment difference with alterna­
tion of just one word; the structural processing environment mechanisms
would still have their effects, but these effects would be independent of the
content.
So, in essence, processing environments can differ; if arranged appropri­
ately, these differences can make social/cultural processing environments
less effective than situational ones. But if this is the case, the processing en­
vironment differences are only structural-, not content-based. One could
equally arrange the processing environments such that social/cultural envi­
ronments are more effective than situational ones.
Thus, it appears that the social/cultural versus situational difference in
contextual information bears no impact on processing or comprehen-
sion—only structural factors seem important. But there may be reason to
5. INFLUENCES ON LANGUAGE 1O5

suppose that the difference does matter when one is interpreting nonliteral
language, or at least some forms of nonliteral language (Colston & Gibbs,
2002; Winner & Gardner, 1993). For these forms (e.g., verbal irony), social/
cultural variables may in fact carry more influence than situational ones. To
demonstrate, recall that nonliteral language presents a relatively greater
separation between language meaning and intended comprehended mean­
ing. Thus, comprehension of nonliteral language may be more dependent on
deriving speakers' intention compared to literal language. According to
Colston and Gibbs (2002, p. 59), "irony says something about the speaker
(i.e., his or her attitudes or opinions about the topic).... Interpreting irony
requires that listeners draw particular inferences about the speaker's state of
mind. More specifically, understanding irony demands that listeners draw a
second-order inference about the speaker's beliefs and intentions."
Knowing some key information about the speaker may then be particularly
useful for comprehension of some forms of nonliteral language, and thus
providing or removing this information (e.g., manipulating a social/cultural
variable) would strongly influence processing or comprehension. Indeed,
this claim has empirical support; Pexman and Olineck (2002) found that
speaker occupation strongly influenced comprehension of verbal irony and
most notably so when other contextual clues were absent (see also Katz &
Pexman, 1997).

Argument 2: Analogous Omissions of Social/Cultural

Influences—Conservation and Word Learning
Another way to approach this issue of whether to consider social/cultural
variables as important components of context is to look at analogous do­
mains of research that have moved from little or no consideration of social
or cultural influences on some level of processing to a more contemporary
understanding of how these influences matter. Such a view correctly points
out how the omission of social and cultural influences in some ways is acci­
dental and understandable, given the focus of the then-current theoretical
approaches or methodologies, but at the same time is crucially inhibitive of
our attempts to understand these domains of cognition.
A number of domains of developmental psychological research, for in­
stance, were founded on principles that originally failed to consider impor­
tant sociopragmatic influences. Each of these domains has since seen an
important shift in that social/cultural influences are now at least entertained
by many scholars as being crucial components of human cognition in those
domains. Consider the domains of conservation and word learning.
Conservation used to be considered by many scholars to be a milestone
exclusively of conceptual development that was essentially immune to so­
cial influences. Children in a Piagetian preoperational stage of develop­
106 COLSTON

ment, for instance, were considered unable to comprehend that the

magnitude of substances (e.g., a fixed amount of a liquid) was constant
across various external changes to those substances (e.g., when being
poured from one container to another of a different shape). Only when chil­
dren reached a particular age and level of conceptual sophistication (in
Piagetian terms, the concrete operational stage) did they achieve the ability
to conserve. Moreover, this ability was seen as being independent of any so­
cial influences on children both in terms of the actual conservation ability as
well as how such abilities were measured.
This view has experienced a significant change. Although still consid­
ered a general conceptual milestone much like I described, conservation is
now also considered to be at least partially determined by sociopragmatic in­
fluences. This is best observed in how the means by which conservation was
measured have come under scrutiny. Children in a classical conservation
task would typically be shown a colored liquid in a transparent container of a
certain size and shape (e.g., a tall and narrow cylinder). The children would
also be asked if they can see how much of the liquid there is in the container
by an experimenter. Children would typically respond affirmatively. Next,
the experimenter would pour the liquid into a differently shaped container
(e.g., a short and wide cylinder), in direct view of the child, being careful not
to spill any liquid or to leave any liquid in the original container. Then, the
child would be asked if he or she can see how much liquid there is in the new
container and whether the amount is the same or different from what it had
been in the original container. The general pattern of results was that pre-
conserving children would say there was less liquid in the second container
but conserving children would say the amount had remained constant. The
typical reasoning given by the preconserving child was that the liquid did
not come up as high in the container and, therefore, had to be of a lesser
amount.
Consideration of the pragmatic principles involved in measuring a child
this way, however, has lead to a change in our understanding of what the
child who failed to conserve was thinking. In particular, one must appreciate
the influence of authority, questioning, and pragmatic rules for speaking to
fully understand the child's behavior. In essence, a child may have had some
semblance of conservation present in their ability, but he or she still failed to
conserve in a conservation task because of these additional influences.
When the adult finished pouring the liquid, for instance, the child may have
held an implicit understanding that the amount had remained constant. But
when the adult asked the question about the amounts, the child began to
question his or her current beliefs. The child may have wondered why the
adult would ask him or her about something so seemingly obvious or implicit
and then conclude, following a given-new assumption and the general social
knowledge that adults are more knowledgeable than children, that the adult
5. INFLUENCES ON LANGUAGE 1O7

knew something about the liquid that the child did not know. Only then
would the child invoke the height of the liquid as a clue and report that the
amount was less.
Thus, the child was using social/cultural information about the interact-
ant—adults often know more about things than do children, along with complex
pragmatic rules for interaction—(given or new contracts, relevance, etc.) to help
him or her complete what appeared to be just a relatively straightforward
problem-solving task involving physical properties of the world.
Next consider an example from a linguistic developmental task—word
learning. Although still a debated issue, the role of social/cultural and prag­
matic principles and information in vocabulary acquisition has also recently
received much greater appreciation than was the case in earlier approaches.
Initial attempts to explain a variety of problems associated with word learn­
ing produced a host of constraints or assumptions that were presumed to be
necessary for a child's word learning to take place as it does. For instance,
consider the problem of reference. How does a child know to what a new
word refers? Many things can have more than one name. For instance, a
neighbor's household pet could be a dog, doggie, pet, beagle, Musket, canine,
brat, pain, joy, boy, and so on. Apply this duplicity to all the things around a
child that can be named or referred to, and the child is faced with an enor­
mous mapping problem. Ways around this problem include simplifying the
mapping so that a child learns just one word at a time for a given thing or
having the child assume that each and every object has just a single name.
This assumption has been proposed as being necessary for children to over­
come this problem of reference and has been referred to as the mutual exclu­
sivity assumption.
To test the validity of the assumption, a variety of tasks can be employed.
A standard version involves a word that has never been encountered previ­
ously by a child (flombat) being used by a caregiver to refer to an object that is
also novel to the child (a stapler). Then, the child is shown this object along
with a second novel one (a flashlight), and is asked to pick out which object
is the spondut; another word that the child has never heard before. Typically,
the child will pick the second novel object instead of the original one. This
selection appears to demonstrate that the child is assuming that the novel
name must apply to the unnamed object because the other object already
bears a name.
Consider another problem with reference. How does a child know to
what a word refers when it is being applied to a single object? How would a
child know, for instance, to what a newly encountered word (finpot) is refer­
ring when used in the context of a novel object (a rocking horse) ? Finpot
could logically apply to a number of things including the springs that support
the horse, the head of the horse, the color of the paint, the material used to
construct the horse, the handles on either side of the head, or the rocking
1OS COLSTON

motion of the horse. But children typically take the word to refer to the en­
tirety of the object. A whole object assumption was invoked to explain the
child's focus in such situations. Proposing that the child holds such an as­
sumption, either innately or emergently as a necessary temporary corral to
channel initial word learning, also enables this problem of reference to
whole objects to be explained.
Testing for the whole object assumption is similar to that for the mutual
exclusivity assumption. A child is first shown a novel object (a rocking
horse). Then a reference is made that assigns the novel name ("this is a
finpot") by an experimenter. Next, a child is shown a variety of objects, in­
cluding component parts of the target object in isolation (the springs, the
head, etc.), as well as the whole object. The child is then asked to pick out
which object is the finpot. Typically, the child will pick the whole object in­
stead of one of its component parts. This selection appears to demonstrate
that the child is assuming that the name must apply to the entire object.
However, as was the case with conservation, such an approach to ex­
plaining word learning fails to appreciate the contribution of social/cultural
variable and pragmatic influences. Here also one must appreciate the influ­
ence of authority, questioning, and pragmatic rules for speaking to fully un­
derstand children's behavior. In the tasks that tested the mutual exclusivity
assumption, for instance, the child may not have necessarily just assumed
that the new word, spondut must refer to the novel object (the flashlight). In­
stead, because the question is being posed by a knowledgeable, cooperative,
adult interlocutor who had just negotiated the label for the other object
(flombat for a stapler), it would be very unlikely and a violation of pragmatic
rules governing common ground for the adult to now refer to that just-
named object with another name.
For the whole object assumption, a similar argument can be made. A
child in this task may not have necessarily just assumed that the label finpot
refers to the entire object. But when given the choice between a variety of
things that could be called by that name, the child again relies on the social
knowledge about their interlocutor—a knowledgeable, cooperative, adult
—as well as pragmatic rules governing relevance and common ground, to
choose the most likely object that the adult would have meant—the one
that most closely resembled the object that had been recently negotiated as
being a finpot (Bloom, 2000; Tomasello & Akhtar, 1995).
These examples demonstrate that some important domains of human
cognition have been incompletely understood due to failure to consider so-
cial/cultural influences on the cognitive activity in question. My contention
is that this same form of omission has been occurring in our understanding of
nonliteral language comprehension.
Next, I demonstrate how such an omission has lead to a limitation in
our understanding of the use of some common forms of nonliteral and indi­
5. INFLUENCES ON LANGUAGE 1O9

rect language. First, though, I briefly discuss how this omission has also in­
hibited our understanding of the processing and comprehension of
nonliteral language.

SOCIAL AND CULTURAL INFLUENCES ON NONLITERAL

LANGUAGE PROCESSING AND COMPREHENSION

It is a generally accepted point that, with adequate contextual support, the

processing of nonliteral language can occur as rapidly as the processing of
comparable literal language. The body of work that has supported this finding
has thus demonstrated that contextual information in the processing of
nonliteral language is playing its role at some point during the processing of
utterances. The question for present purposes is whether social/cultural vari­
ables in this contextual information also have an influence early in processing.
Before addressing this question, a brief discussion of some ways in which
at least one kind of nonliteral language differs from comparable literal lan­
guage is in order. Verbal irony can differ from literal language in such a way
that the influence of social and cultural variables on early processing of ver­
bal irony would almost have to be a necessity.
Consider first that the processing duration of verbal irony has revealed
conflicting findings. Some studies have shown that verbal irony processing
takes longer than processing of the same language used literally (Dews &
Winner, 1999; Giora 1995, 1997; Giora, Fein, & Schwartz, 1998;
Schwoebel, Dews, Winner, & Srinivas, 2000). Other studies have shown no
difference in the time course of verbal irony and literal language processing
(Gibbs, 1986; Gibbs, O'Brien, & Doolittle, 1995). To evaluate whether
there might be some inherent differences in verbal irony versus literal lan­
guage that could explain this discrepancy in processing duration, I con­
ducted an evaluation of the impact of referential distance and inference
complexity on the processing duration of verbal irony and comparable literal
language (Colston, 2004b). The results revealed, not surprisingly, that in­
creasing the referential distance and inferential complexity of utterances
slowed processing time. This delay in processing was also shown to be equiv­
alent for literal and ironic utterances. The key point of the study, however,
was an additional demonstration that verbal irony, by its mere nature, often
requires more complex inferences and more distal references relative to
comparable literal language. Thus, some of the variability in processing du­
ration of verbal irony compared to literal language is due to these additional
processing factors:

Ironic utterances often require complex inferences or distal referents

... that are not required for the same utterances to be interpreted liter­
ally. The mere nature of ironic utterances often introduces a require­
HO COLSTON

ment of complex inferences or distal referents for successful

comprehension, while other aspects of the ironic utterances remain
generally equivalent to when the utterances are used in literal con­
texts. Complex inferences and distal referents might then account for
a great deal of the occasional lag found in verbal irony processing
rather than the serial argument that ironic utterances must first be in­
terpreted literally. (Colston, 2004b)

Note that the oft found greater referential distances and inferential
complexities for verbal irony makes this form particularly susceptible to so­
cial and cultural influences. To give a concrete example, the Colston
(2004-b) study demonstrated that the frequently observed echoic re­
minder mechanism in verbal irony requires a more distal referent than a
comparable instance of literal language. A given ironic utterance that uses
echoic reminder is referring to an established social norm that has been vi­
olated in a particular situation but that may not have been completely sa­
lient in that situation (e.g., a speaker says, "You're so generous" to a friend
who committed a disservice to the speaker). Such a reference requires pro­
cessing effort on the part of the comprehender to establish the referent so­
cial norm. Clearly, social and cultural variables would have a great deal of
influence here. A speaker who has a particular social or cultural character­
istic that would enable references to relatively distal or nonsalient social
norms to occur more readily might be comprehended more quickly than a
speaker who does not have such a characteristic. For example, a speaker
who is stereotypically known for pointing to social norms that have been
violated (e.g., a newspaper editorialist), who makes such a reference in an
ironic comment (e.g., "people are so generous"), might be processed more
readily than a speaker stereotypically known for not referencing such dis­
crepancies (e.g., a nun).
Indeed, such a prediction has been validated. Katz and Pexman (1997)
demonstrated that the occupation of a speaker had a strong effect on how
nonliteral utterances were comprehended. Metaphorical utterances that
could be interpreted ironically or nonironically (e.g., "children are precious
gems") were more likely to be interpreted ironically when the speaker was a
person who was stereotypically apt to use irony, as indicated by norming data
(e.g., comedians, police officers, and factory workers), than a person stereo­
typically less apt to use irony (e.g., clergymen, doctors, and teachers).
We thus know that social/cultural variables do have an influence on com­
prehension. The question now is if they influence early processing. A num­
ber of other recent studies have provided an answer to this question.
Pexman, Ferretti, and Katz (2000) tested the processing duration of ironic
and metaphorical statements made by either stereotypically high-irony us­
ers or high-metaphor users based on the users' occupation. The results very
5. INFLUENCES ON LANGUAGE 11 1

clearly demonstrated that the occupation had a significant effect on pro­

cessing and that this effect was showing itself very early in the processing
stream of the utterances (see Katz, chap. 8, this book, and Pexman, chap. 9,
this book, for a synopsis of this study).
Pexman and Olineck (2002) further showed that the particular qualities
of high-irony occupations—being humorous, critical, insincere, and having
low social status—were what enabled those occupations to cue ironic in­
tent. This finding fits nicely with the previous discussion that ironic lan­
guage would almost necessitate an early influence of social/cultural
variables. The social characteristics established by Pexman and Olineck
mesh well with the process of pointing out social norms that have been vio­
lated. Being critical and humorous are particularly consistent with scrutiny
of violated expectations. Having low social status makes one likely to have
frequently experienced violated social norms. The insincerity factor also re­
flects the mechanism of using verbal irony—a statement of expectations in
the midst of a violation of those expectations.
Katz, Piasecka, and Toplak (2001) also found very early processing effects
of speaker gender on the processing of sarcastic utterances. Most notably,
they found that gender effects emerged during the act of reading the key tar­
get sarcastic utterances, with reading times for male speakers being less than
female speakers. This slowdown in the processing of sarcastic utterances by
female speakers was most pronounced when the listener was also a woman.
Thus, there is some very clear evidence emerging recently in the litera­
ture that social and cultural variables, gender and occupation among them,
have an influence on nonliteral language processing and comprehension.
Now I discuss how these variables might have an impact on the use of
nonliteral language.

SOCIAL AND CULTURAL INFLUENCES

ON NONLITERAL LANGUAGE USE

A variety of studies have been conducted on social and cultural variables

and their impact on the use of several indirect or figurative forms. Among
the kinds of variables that have been investigated are the degree of familiar­
ity between interlocutors (Colston, 1999a; Jorgensen, 1996; Kreuz, 1996),
the amount of shared knowledge of interlocutors (Katz & Lee, 1993; Kreuz,
Kassler, Coppenrath, & McLain, 1999; Kreuz &Link, 2002), the social sta­
tus of interlocutors (Holtgraves, 1997a, 1997b; Kemper &.Thissen, 1981;
Okamoto, 2002), gender (Gibbs, 2000; Holtgraves, 1991, 1992; Katz et al.,
2001), and occupation (Holtgraves, 1997a, 1997b; Katz & Pexman, 1997;
Pexman et al., 2000; Pexman & Olineck, 2002). The kinds of indirect or fig­
urative language investigated have included verbal irony, metaphor, ne­
gated sentence structure, indirect requests, and proverbs—among others.
112 COLSTON

Most of these studies have either directly or as a means of addressing

some other issue surreptitiously manipulated one of these variables across
speakers of the indirect or figurative remarks or across the relationship be­
tween the interlocutors. Comprehension or pragmatic effects are then mea­
sured, and any differences across the variables are noted. For instance, in the
Colston (1999a) study participants read short vignettes that described two
interlocutors conversing over some topic. One of the interlocutors then de­
scribed some aspect of the topic with either a direct (e.g., "It was stale" or "It
was fresh") or negated (e.g., "It wasn't fresh" or "It wasn't stale") comment.
For some of the comparisons the interlocutors' relationship was described as
being one of high common ground (e.g., they were close friends) or low com­
mon ground (e.g., they were strangers). Participants then rated the intended
meanings of the speakers making the remarks (e.g., for this example, how
fresh or stale the speaker is saying the bread was). For this particular study,
no difference was obtained for the common ground variable. But most of the
findings of this literature have shown differences across the manipulated so-
cial/cultural variables.
One shortcoming of much of this work, however, is that it does not always
afford a means by which to evaluate why the particular social or cultural dif­
ferences are found. Although the mere establishment of such differences is
very important and, as just indicated, frequently social/cultural variables are
not the primary topic of the study; thus, their causal delineation is not the
highest priority, but nevertheless their presence now begs for an explanation.
The following two studies were conducted with this goal in mind—to be­
gin an investigation of why some particular social/cultural differences are
found in nonliteral language use. The studies selected gender as the social
variable because gender has emerged as a fairly robust predictor of some ef­
fects in nonliteral language use. Gender differences in verbal irony and asyn­
deton were investigated.

GENDER DIFFERENCES IN VERBAL IRONY

The studies conducted by myself and Sabrina Lee (Colston & Lee, 2000,
2004) sought to evaluate whether and why gender differences would be
found in peoples' use of verbal irony. Prior to this study, a number of areas of
research had established, although not without controversy, some gender
differences in language use, including differences in males' and females' ver­
bal skills and communicative styles. Other lines of research had also begun
to show that a wide variety of pragmatic functions are accomplished by dif­
ferent forms of indirect or figurative language. This latter research area had
also begun to delineate which forms of indirect or figurative language were
most adept at particular pragmatic functions and why. Thus, a straightfor­
ward extension and combination of these lines of research could seek to es­
5. INFLUENCES ON LANGUAGE 113

tablish whether gender differences would be obtained in speakers' use of

indirect or figurative language. Thus, we chose to investigate gender differ­
ences in verbal irony because that form had been the most widely studied to
date for its pragmatic functions.
After a series of unsuccessful pilot studies that directly asked participants
to rate how sarcastic male and female speakers were, when depicted making
sarcastic or literal comments in short written scenarios with stereotypical
male and female names, a study that indirectly probed this phenomenon did
demonstrate a gender difference. Participants read a series of short scenarios
depicting themselves and other characters in various generally negative sit­
uations (e.g., a person was not accepted into graduate school). No cues were
given to assign gender to the characters (e.g., abstract symbols were used in­
stead of character names, and the scenarios were designed to be gender neu­
tral) . One of the characters would then make a comment that was either
literal or sarcastic, as if spoken to the participant in the experiment, at the
end of the scenario, for instance: "This is just terrible" or "This is just ter­
rific." was said after the scenario:

You and your cousin *

were on your way to another friend's house.
That friend owes * two hundred dollars.
You ring the doorbell and after two tries,
no one comes to the door.
* turns to you and says,

Participants read the scenarios and the characters' comments and rated
whether speakers using the comments were most likely male or most likely fe­
male on a 7-point rating scales. The results of both male and female partici­
pants showed that verbal irony was considered more male-like than literal
remarks. Four subsequent experiments then evaluated two possible causes
of this gender difference—discourse goal match and risk attraction.
The discourse goal match explanation stemmed from the dual possibili­
ties that verbal irony accomplishes a variety of pragmatic functions that are
different from those of other figurative and literal forms and that men might
have a different set of discourse goals on average than women. Therefore,
men might be more apt to use verbal irony because their discourse goals
match verbal irony's pragmatic accomplishments better than do women's
discourse goals.
The risk attraction explanation arose from previous research that has
documented that men are generally more risky than women in a variety of
domains. Men engage in more risky behavior, take more health risks, and
114 COLSTON

prefer more injury-risking sports and activities than do women. In addition,

men find less risky behaviors less interesting than women (Byrnes, Miller, &
Schafer, 1999; Carlson & Cooper, 1974; Howland, Hingson, Mangione, &
Bell, 1996; Wiederman, 1997). Our conjecture was that verbal irony, being
an indirect or figurative form of language, might pose a greater risk of misin­
terpretation and thus appeal more to male speakers than female speakers in
part because of this risk.
Experiments 2 and 3 (Colston & Lee, 2000, 2004) tested the possibility
that the particular pragmatic functions accomplished by verbal irony fit
better with males' than females' typical discourse goals. To make this evalua­
tion, we first needed to determine what pragmatic functions would be per­
formed by verbal irony. Verbal irony had previously been shown to enable
speakers to perform a variety of pragmatic functions, including to be humor­
ous, express surprise, express mastery over some topic or issue, diminish crit­
icism, enhance criticism, point out a deviance from expectations, show
negative emotion, and others. These pragmatic functions were tested in
studies that sought to determine purely what people thought verbal irony
would do for speakers (Roberts & Kreuz, 1994), as well as in studies that
sought to determine how and to what extent different kinds of verbal irony
would perform the functions they do (Colston, 1997a, 1997b, 1999a, 1999b,
2000a, 2000b, 2002a; Colston & Gibbs, 1998; Colston & Keller, 1998;
Colston & O'Brien, 2000a, 2000b; Dews, Kaplan, & Winner, 1995; Dews &
Winner, 1995, 1999; Gibbs, 1986, 1987, 2000; Kreuz, 2000; Kreuz, Long &
Church, 1991; Schwoebel et al., 2000; Toplack & Katz, 2000; Whaley &
Holloway, 1996).
For our purposes, we selected pragmatic functions from two general
sources. One source came from Roberts and Kreuz's study (1994) that
tested the extent to which a large number (approx. 20) of pragmatic func­
tions would be performed by verbal irony, among many other nonliteral
forms. We chose five of the functions they measured to provide us with a
range that varied from functions that were reported as not being performed
at all by verbal irony (to deemphasize—none of their participants said that
verbal irony did this), to functions that verbal irony was almost universally
reported as accomplishing (to show negative emotion—94% of their partici­
pants said that verbal irony did this). Other functions we selected from this
study were to be humorous (65%), to clarify (35%), and to manage the dis­
course (18%). Our second set of functions were derived from other studies
that investigated the pragmatic functions of verbal irony including some of
my previous work that showed that verbal irony is used to enhance
negativity; (Colston, 1997b), from work by Lee and Katz (1998) that pin­
pointed the important role of a victim in the use of verbal irony, from a claim
by Whaley and Holloway (1996) that was empirically validated by Colston
(1999b), that irony can serve to socially attack people, and from the general
5. INFLUENCES ON LANGUAGE 115

theoretical characteristic that verbal irony is indirect. From these studies we

derived the following additional pragmatic functions: to be rude, to insult,
and to imply.
We then took the scenarios from Experiment 1 and presented them to a
new set of participants along with rating scales for all of these functions. Par­
ticipants in this experiment (Experiment 2) were asked to rate how much
the remarks in the scenarios would accomplish each of the functions. The
results revealed that verbal irony is thought to express negative emotion, to
be humorous, to be rude, to de-emphasize, and to insult to a greater extent
than literal commentary, by both male and female participants.
In Experiment 3, participants were given the same situations as were pre­
sented in Experiment 2 with slight wording alterations to place the partici­
pants in the role of the speakers of the comments, but no comments were
presented. Participants were instead asked to rate the likelihood that they
would wish to accomplish each of the pragmatic functions we had selected
in each of the particular scenarios that was presented. The results revealed
that three of the five functions accomplished by verbal irony were more
sought by female participants than by male participants (to be rude, to
de-emphasize, and to insult). The other two functions were equally sought
by male and female participants. Thus, there is little evidence to support the
idea that men use verbal irony more often than women because the prag­
matic functions of verbal irony particularly suit men's discourse goals.
Experiments 4 and 5 evaluated the possibility that the heightened risk of
misunderstanding presented by verbal irony relative to literal language
would appeal to men. Experiment 4 presented a new set of participants with
the scenarios from Experiment 1; this time the wording was altered so that
the participants were not one of the characters in the scenarios, and they
were asked to rate the degree of risk of misinterpretation of the comments.
The results revealed that both men and women report that verbal irony
poses more of a risk for misinterpretation than do literal remarks. Thus, we
know that verbal irony use does involve some level of risk for a speaker that
is noticeable to people.
Results of Experiment 5 indicated a preference for use of verbal irony by
male participants, as the risk attraction account would predict. Here, partic­
ipants were given the scenarios from Experiment 3 (the ones with the partic­
ipants as the speakers), along with the comments, and the participants were
asked to imagine themselves to actually be present in these situations and to
then rate the likelihood that they would actually use the remark that was
presented.
Taken together, these findings demonstrate that verbal irony is consid­
ered a more male-like than female-like form of communication not because
the particular pragmatic functions it accomplishes fit better with males1
than females' discourse goals. The results instead support the explanation
116 COLSTON

that the risk of misinterpretation inherent in verbal irony appeals to males'

greater degree of riskiness.
Note that the explanations we explored in this study were certainly not
exhaustive. For reasons of brevity we were unable to test several other viable
explanations. Among these are the emotional expressiveness of verbal irony,
gender differences in attribution, aggressiveness gender differences, and
mastery demonstration.
The emotional expressiveness explanation stems from work that has
shown a difference in the kinds of emotions that are expressed by verbal
irony versus other kinds of remarks, including literal commentary. In addi­
tion to the negative emotion expression by verbal irony demonstrated by
Roberts and Kreuz (1994) and Colston and Lee (2000, 2004), Leggitt and
Gibbs (2000) found that verbal irony enables expression of emotions that
differ from those expressed by literal remarks. Other studies have shown
that verbal irony expresses more surprise than literal comments (Colston &
Keller, 1998). These findings, coupled with the varying cultural norms that
influence which emotions men and women are allowed to express, suggest
that men may be more likely to use verbal irony because it enables expres­
sion of the emotions that they are apt to seek to express in conversations.
The gender differences in the attribution explanation comes from work
in social psychology that has shown reliable differences in how men and
women attribute blame for contribution to failure. In general, when having
participated in some collective task that has resulted in a failure, men are
more apt to attribute blame for that failure to situational causes (e.g., there
was not enough time provided, the task asked too much, and the tools or
material were of poor quality). Women, however, are more likely to make
personal attributions for the failure (e.g., I do not have the skill and I am not
fast enough). This pattern of attribution—coupled with the notion that
nonliteral language arguably presents a greater likelihood of miscompre­
hension, which may be considered a form of failure—might make women
slightly less likely to use verbal irony. The idea is that, because women, all
else being equal, are more likely to blame themselves for a conversational
exchange that was misunderstood (whereas men are more apt to blame the
situation or their addressee) they tend to avoid ways of speaking that present
higher risks of miscomprehension and, hence, be slightly less likely to use
verbal irony.
Still another explanation might come from differences in aggressiveness
between men and women. Verbal irony may be considered a more aggressive
form of talking than literal commentary, given work that has shown verbal
irony to be more critical, condemning, and so on, than literally negative re­
marks (Colston, 1997b). Given other work that has shown a degree of
greater aggressive in certain measures of men over women, verbal irony
might be more apt to be used by men for this reason. However, note that not
5. INFLUENCES ON LANGUAGE 117

all studies have shown that verbal irony is more negative than literal com­
mentary, and not all forms of aggressiveness show a gender difference.
One other possible explanation involves how nonliteral language, in­
cluding verbal irony, enables a speaker to demonstrate mastery over the ref­
erent topic. In being able to craft and use a particularly clever and meaning-
laden nonliteral utterance in a given, frequently negative situation, speakers
can often demonstrate a degree of understanding of, unflappability toward,
and creativity regarding the situation at hand (Colston & O'Brien, 2OOOa).
This expression of control or mastery over a situation by how one talks might
also tap into some gender differences in dominance-seeking or con-
trol-seeking behaviors or something similar that could also help explain
gender differences in verbal irony use.
There is clearly much more work to be done to hone our explanations for
the gender difference obtained on speakers' use of verbal irony. Future work
should test these and other possible explanations, as well as possible combi­
nations of the explanations (e.g., demonstrating mastery and willingness to
risk misinterpretation constitute a form of aggressiveness, and demonstrat­
ing that mastery allows expression of socially acceptable emotions).
Another issue raised by the observed gender difference in verbal irony—
an issue that is gaining importance in the overall investigation of nonliteral
language—is the extent to which phenomena regarding nonliteral language
(in this case, gender differences in reported usage) are universal to all forms of
nonliteral language. Such effects may be relatively universal across all kinds of
nonliteral language, or they may be germane to just one family of nonliteral
language forms or just a single type. In terms of gender differences in use, the
question is whether a gender difference holds for all nonliteral language use,
as might be predicted by the attribution explanation just discussed (assuming
one believes all nonliteral forms pose a higher risk of misinterpretation, some­
thing I and others have challenged; Colston, 2002b), or just for verbal irony,
as would be predicted by some of the other explanations.
To initiate an investigation of this important question, I conducted a
study that is analogous to my work with Sabrina Lee on gender differences in
verbal irony. Now, however, the goal was to investigate gender differences in
speakers' use of asyndeton, a relatively unexplored form of nonliteral lan­
guage. If gender differences are also obtained here, this may support the at­
tribution explanation and, in turn, the universality view.

GENDER DIFFERENCES IN ASYNDETON

Asyndeton is a kind of utterance in which all but the skeletal syntactic cate­
gories are dropped and a reduplicated phrase structure is used for the re­
maining words (e.g., "Been there, done that" and "I go, I work, I leave").
Asyndeton can be considered a form of indirect language because a speaker
118 COLSTON

does not just say literally, albeit minimally, what he or she means. Instead,
the speaker uses the minimal form to imply a great deal more than what is
said. Moreover, the implied meaning in asyndeton goes beyond the meaning
that would have been expressed by the utterance had it contained the omit­
ted words. Speakers appear to use asyndeton to accomplish several prag­
matic functions, including to express a negative attitude toward the referent
topic (Colston, 2004a).
In six experiments conducted by myself and Virginia Lusch (Colston &
Lusch, 2004), potential gender differences in asyndeton and two of their po­
tential causes were studied. Experiment 1 presented participants with a se­
ries of short scenarios similar to those used in the Colston and Lee (2000,
2004) studies. The scenarios depicted characters in various generally neu­
tral situations (e.g., a person is describing his or her summer job as a life­
guard). Neutral statements were used to enable comparison of asyndeton
with direct positive and direct negative comments. No cues were given to
assign gender to the characters (e.g., abstract symbols were used instead of
character names, and the scenarios were designed to be gender neutral).
One of the characters would then make a comment at the end of the sce­
nario that was either literally positive, literally negative, or an asyndeton
(e.g., "I love it," "I hate it," or "I go, I sweat, I leave"). Participants read the
scenarios and the characters' comments and rated whether speakers using
the comments were most likely male or most likely female on 7-point rating
scales (as in the Colston and Lee study). The results of both male and female
participants showed that verbal irony was considered more male-like than
both kinds of literal remarks. In addition, both male and female participants
indicated that negative remarks were more male-like than positive remarks.
Thus, we have evidence for a gender difference in asyndeton use. Five subse­
quent experiments then evaluated the discourse goal match and risk attrac­
tion explanations already discussed.
Experiments 2 and 3 (Colston & Lusch, 2004) evaluated the possibility
that the heightened risk of misunderstanding presented by asyndeton rela­
tive to literal comments would appeal to men. Experiment 2 presented a
new set of participants with the scenarios from Experiment 1 and asked peo­
ple to rate the degree of risk of misinterpretation of the comments. The re­
sults revealed that both men and women reported that asyndeton poses
more of a risk for misinterpretation than do literal remarks. Both male and
female participants indicated that negative remarks were more likely to be
misinterpreted than positive remarks. Thus, we know that asyndeton use
also involves some level of risk for a speaker that is noticeable to people.
Experiment 3, however, found no reported preference for use of asynde­
ton by male participants, as the risk attraction account would predict. In­
stead, both men and women reported they would be less likely to use
asyndeton than literally positive remarks and equally likely to use asyndeton
5. INFLUENCES ON LANGUAGE 119

as literally negative remarks. Here, participants were given the scenarios

from Experiment 1 with slight rewordings to make the participants be the
speakers in the scenarios, and the participants were asked to imagine actu­
ally being present in these situations and to then rate the likelihood that
they would actually use the remark that was presented.
This result was puzzling given the results of the first experiment. Contrary
to results showing no gender differences in the likelihood of using asynde­
ton, why would people think that a speaker using asyndeton would likely be
a man? To investigate, we (Colston &. Lusch, 2004) conducted Experiment
4 which sought to test people's memory for having used asyndeton in the
past. It could be possible that people have difficulty imagining using asynde­
ton in the situations we supplied, but they do recall having used asyndeton
themselves in past situations.
In Experiment 4 we presented the same scenarios and comments as Ex­
periment 3 to a new group of participants, who were asked to rate how of­
ten in the past they had used comments such as the ones they were reading,
on 7-point rating scales ranging from not at all often and somewhat often.
The results revealed an interesting pattern. Both men and women re­
ported having used positive comments relatively frequently and having
used asyndeton relatively infrequently. But for negative comments,
women reported having used them more often than men. Thus, a gender
difference of sorts is at work in reported use of asyndeton. If men wished to
express some negative opinion, they reported having used asyndeton and
directly negative comments roughly equally. If women wished to express a
negative opinion, however, they reported using asyndeton less often than
direct negative commentary. Thus, there is some support for the risk at­
traction account. Men are, in a way, reported to have used asyndeton more
often than women—they used it as often as direct negative commentary,
whereas women used it less often than direct negative remarks and asyn­
deton can risk misinterpretation.
We (Colston & Lusch, 2004) then conducted two final experiments to
evaluate the discourse goal match explanation. Experiment 5 presented a
new group of participants with the scenarios from Experiment 1 who were
asked to rate how much the remarks in the scenarios would accomplish six
of the functions that were evaluated in Colston and Lee (2000, 2004) on
gender differences in verbal irony (i.e., to show negative emotion, to manage
discourse, to deemphasize, to be rude, to insult, to imply, and to be humor­
ous) . We exchanged to be unconventionalfor to clarify for the Colston and
Lusch (2004) study due to the relatively unconventional nature of asynde­
ton and because to clarify was the one function found to be performed to a
greater degree by literal language than verbal irony in that previous study;
thus, it was likely not something particularly suited to nonliteral language.
Both male and female participants revealed that asyndeton is thought to be
120 COLSTON

unconventional, to be humorous, and to deemphasize to a greater extent

than both kinds of literal commentary.
In Experiment 6 participants were given the same situations as were
presented in Experiment 3 (the ones in which the participants are the
speakers of the comments), but no comments were presented. Participants
were instead asked to rate the likelihood that they would wish to accom­
plish each of the pragmatic functions we had selected, in each of the par­
ticular scenarios that was presented. The results revealed that two of the
three functions accomplished by asyndeton (to be unconventional and to
be humorous) were more sought after by male participants than by female
participants. The other function was equally sought by male and female
participants. Thus, evidence supports the idea that men use asyndeton
more often than women because the pragmatic functions of asyndeton
particularly suit men's discourse goals.
Taken together, the results of this second study demonstrate that asynde­
ton is considered a more male-like than female-like form of communication
because the particular pragmatic functions it accomplishes fit better with
men's discourse goals than women's, and because the risk of misinterpreta­
tion inherent in asyndeton appeals to men's greater riskiness.
The results of these two studies have begun to tell us something about
why one social/cultural variable, gender, has an effect on nonliteral lan­
guage use. In both studies, gender differences in use of nonliteral language
were obtained that cannot be attributed to gender differences in address­
ees, situational variables, or other factors. Clearly, something about differ­
ences in men and women underlies differences in their use of nonliteral
language. Our studies raised at least six possible inherent gender differ­
ences that might underlie this nonliteral language usage difference—dif-
ferences in aggression, mastery demonstration, attribution, discourse
goals, riskiness, and emotional expression. Although these results are not
conclusive, they do begin to suggest which one(s) of these explanations
will hold up to future scrutiny.
Given that men have been measured to be more aggressive in some do­
mains than women, it makes sense that men are more likely to use verbal
irony because it is a relatively aggressive form of talking. But differences in
aggression have a tougher time explaining men's greater likelihood of using
asyndeton. This form is not typically an aggressive way of talking; thus, the
aggressiveness explanation does not bode well in explaining asyndeton's
greater usage by men.
Because many forms of nonliteral language, verbal irony and asyndeton
included, can express a degree of mastery over a topic (something not re­
ported here but for which some empirical support is emerging; see Colston &
Izett, 2004), and given a general tendency for men to wish to demonstrate
such a mastery, this explanation might be a contender. However, more work
5. INFLUENCES ON LANGUAGE 121

is still needed to find the direct link between a need or desire to demonstrate
mastery and use of nonliteral language forms.
The attribution explanation also appears to hold up to the foregoing stud­
ies. Although not all forms of nonliteral language would necessarily pose a
heightened risk for misinterpretation, the two forms here have been shown
to pose such a risk. Thus, although attribution would likely not explain gen­
der differences found on easily understood nonliteral forms, it does help to
explain the gender differences obtained in Colston and Lee (2000, 2004)
and Colston and Lusch (2004). Again, much more work is required to estab­
lish direct causal links.
The discourse goal match explanation did not hold up well under these
studies. Although it helped account to a degree for men's greater likelihood
of using asyndeton (recall that two of three of the functions that asyndeton
performed were sought by men more than women), it did not explain men's
greater usage of verbal irony (recall that three of five of the functions per­
formed by verbal irony were more sought by women than men).
The risk explanation was supported by both Colston and Lee (2000,
2004) and Colston and Lusch (2004) - Previous work has shown that men
have a greater riskiness than women (Byrnes, Miller, & Schaper, 1999;
Carlson & Cooper, 1974; Howland, Hingson, Mangione, & Bell, 1996).
Verbal irony and asyndeton were shown to pose a greater risk of miscompre­
hension. And, men did show a greater likelihood of use of these nonliteral
forms. Once again, the direct causal relationship between gender, risk and
nonliteral language use remains to be investigated.
Finally, the emotion expression explanation did not receive much sup­
port in Colston and Lee (2000, 2004) and Colston and Lusch. Although
the studies did not explicitly look at emotion expression of nonliteral lan­
guage other than the to express negative emotion pragmatic function, no
gender difference was obtained in these studies on prevalence of desiring
to express negative emotion (but gender differences in the use of nonliteral
language to express emotions were obtained by Link & Kreuz, see chap. 7,
this book).
To summarize, it appears that the mastery, attribution and risk explana­
tions were supported by the results. As already noted, it is also readily possi­
ble that these explanations can be combined in interesting ways.
Researchers should consider this issue, determine direct causal links, and
more strongly test these and other explanations. We consider our studies a
first step in the process of explaining why gender differences are found in
nonliteral language use, and welcome further explorations.
In terms of the universality issue—that is, whether phenomena or expla­
nations for a particular form of nonliteral language generalizes to all forms of
nonliteralism—the picture is also somewhat clearer but still incomplete.
First, consider which explanations are consistent with the universality
122 COLSTON

claim. The aggression, emotion, and discourse goal explanations would

likely not be consistent with universality. Not all forms of nonliteral lan­
guage are aggressive, express the same emotions, or perform the same prag­
matic functions. Second, the risk and attribution explanations are also not
likely consistent with universality if one agrees that not all forms of
nonliteral language pose a heightened risk of miscomprehension. These ex­
planations would be consistent, however, with a revised universality claim
that gender differences would be obtained in all forms of nonliteral language
that do pose a reasonably greater risk of miscomprehension. Finally, the
mastery explanation does seem consistent with universality.I suspect that
many if not most forms of nonliteral language would seem to perform this
function, although this claim needs to be evaluated for a wide variety of
nonliteral forms.
Thus, the explanations that Colston and Lee (2000, 2004) and Colston
and Lusch (2004) supported seem to be the same ones that could be consis­
tent with at least a revised universality claim that is restricted to nonliteral
forms that do pose risks of miscomprehension. Once again, our analysis is
not conclusive and many aspects of the phenomenon require confirmation,
but our studies nevertheless suggest that limited universality might hold up
to further examination. Our work also provides a blueprint for future testing
of this issue.
The studies reported here, in addition to other works that have also
looked into social/cultural variables on nonliteral language use, process­
ing, and comprehension (some of which are reviewed by other chapters in
this book), are beginning to establish the very important influence that so-
cial/cultural variables have on nonliteral language. I emphasize this impor­
tant influence by describing a very recent study of mine (Colston, 2001,
May) that not only shows the influence of a social/cultural variable on, in
this instance, nonliteral language comprehension, but also illustrates the
potential ramifications of such an influence. Essentially, social/cultural in­
fluences on nonliteral language comprehension may be a key component
to intergroup conflict.

SOCIAL/CULTURAL VARIABLES, NONLITERAL

LANGUAGE, AND INTERGROUP CONFLICT

Consider the following three, seemingly unrelated research domains in

psycholinguistics, cognitive and social psychology: language comprehen­
sion, social tension, and cognitive load. Language comprehension in­
volves a complex interaction of ascertaining the meaning of the language
that speakers use and the intentions the particular speakers have in using
that particular language in the given contexts. Research has demonstrated
the necessity of this process most clearly for nonliteral forms including fig­
5. INFLUENCES ON LANGUAGE 123

urative language such as metaphor (e.g., "This weather is garbage" to refer

to inclement weather) or verbal irony (e.g., "This weather is lovely").
Studies have also revealed this process in indirect language when, for in­
stance, an utterance is said to one person but an intended meaning is di­
rected at a third party (e.g., as in, "Some people think this is nice weather")
or when an intended meaning is largely dependent on a listener's likely in­
ference or response (e.g., as in, "How about this weather?"). Indeed, the
process can be observed even in direct, literal language, which often is de­
ceptively underdetermined (e.g., as in saying, "The weather is bad," but
with no direct reference to the weather here). Although this process is re­
flected in many, and indeed arguably, all, kinds of language, it is arguably
more obvious in the figurative forms.
This process of language and intention interdependence is also widely
recognized by language scientists and theorists, as reflected in a wide variety
of language comprehension theories. Notable among these are Gricean
pragmatics, with its conversational maxims and conversational implicature
mechanisms; relevance theory, with its system of computing elaborative
contextual effects within reasonable processing constraints; the large body
of inference literature, which shows that readers and listeners make a wide
variety of coherence and elaborative inferences either online or shortly after
encountering language; and, arguably, the direct access model, which claims
that broad nonlinguistic contextual information is brought to bear during
initial language comprehension.
A very different, but nonetheless large, cross-disciplinary literature has
also reliably demonstrated an increase in tension (measured subjectively and
objectively) as a product of factors like being in unfamiliar social- and cul-
tural-group environments. This tension can be found in low-level measures
normally outside of conscious control (e.g., Galvonic Skin Responses
(GSRs)), as well as in subjective ratings. It also arises from a variety of sources
such as being among a group of individuals of different gender, ethnicity, age,
social status, and others. Phenomena along these lines have been studied in
areas as diverse as adolescents' gender interaction development, personality
trait work on shyness, and anxiety disorders in clinical psychology.
Last, still another large and cross-disciplinary literature has demon­
strated a reliable breakdown in a variety of cognitive processes as a function
of the presence of a diversity of stressors. Work on this phenomenon can be
found in much of the research on divided and other forms of attention, from
human factors research on topics like vigilance, and in much of the cogni­
tive aging literature.
A recent new project of mine (Colston, 2001, May) sought to wed these
research lines. The complexity of figurative language comprehension pro­
cesses might make them vulnerable to the usual breakdowns from which in­
tricate cognitive functions suffer (e.g., overload in short-term memory
124 COLSTON

capacity hinders language inference processes). Social tension, as often oc­

curs in diverse group interactions, may be one source of such processing
breakdowns. Thus, I sought to determine if social tension is detrimental to
complex, higher order language processing (e.g., comprehension of speaker
intention in using figurative language) such that communicative misunder­
standings ensue or increase. Moreover, I also sought to determine whether
this mechanism is a component in group relations.
In two experiments, an affirmative answer to this research question was
found. Participants rated subtle aspects of the intended meanings of tape-re-
corded speaker-actors who made comments in a variety of situations. Group
tension in Experiment 1 was operationally defined as being in different-gen-
der versus same-gender pairs when publicly making such evaluations, with
different-gender pairs being the higher tension situation (norming data con­
firmed this definition).
Results showed a small but significant decrease, as a function of social
tension, in the ability to comprehend sarcastic intent from a speaker's into­
nation. Essentially, people who were in same-gender groups were able to de­
tect sarcastic intent based on the words and the intonation that the
recorded speaker-actors used. This result was similar to earlier comparison
studies that measured perceived sarcasm, humor, and pretense using the
same recordings. But, people in different-gender groups were not able to use
the speaker-actor's intonation to detect sarcasm.
Experiment 2 enhanced social tension by having people publicly make
ratings in larger groups of one to three peers. Here, high tension was defined
as having all other-gender members in the group, whereas low tension was
defined as having at least one same-gender member in the group. This ex­
periment replicated the first with a pointedly stronger effect size. The take-
home message of these experiments is that when people are placed in unfa­
miliar social environments, based on gender, they lose the ability to use a
speaker's intonation to detect sarcasm.
I have subsequently conducted an analogous study using ethnicity as the
social/cultural variable instead of gender (Colston & Jones, 2004). To en­
able manipulation of ethnicity, I used videotaped scenarios. Each scenario
first had a narrarator describe a negative situation involving the participant
and another person (e.g., "A classmate of yours is told to go up and speak in
front of the class, something she (or he) hates doing. Before she (or he) gets
up, she (or he) says to you"), which was followed by a comment by that other
person. Half of the comments were literal (negative words, e.g., "I'm totally
scared about this"), and the other half were sarcastic (positive words, e.g.,
"I'm totally thrilled about this"). Actor-confederates were hired to portray
the speakers making these comments. Half of the confederates were African
American, and the remaining confederates were European American. One
half of each of the members of these ethnic groups was men. A videotape was
5. INFLUENCES ON LANGUAGE 125

then constructed that presented a series of these scenarios. A given scenario

was preceded by 10 s of dead air, followed by the voice of the narrator de­
scribing the situation, and finally the audio and video recording of the con­
federate making his or her comment. The tape had an equal number of each
kind of comment spoken by each type of confederate, which were presented
in random order.
A set of 7-point rating scales ranging from not at all sarcastic (1) to ex­
tremely sarcastic (7) was used to measure participants' assessment of the de­
gree of sarcasm in the speakers' comments. One group of the participants
was randomly assigned to a low-stress condition in which they simply made
their ratings with no additional instructions. The other group was told that
the experimenters were additionally interested in the accuracy of their rat­
ings. These participants were tested with the experimenter in the room,
standing behind them, looking over their shoulders, and purportedly mak­
ing accuracy ratings on each of their sarcasm ratings. This group was labeled
the high-stress condition. Half of the participants were African Americans
and half were European American, appropriately assigned to conditions for
counterbalancing.
Analyses on the sarcasm ratings were conducted separately for the two
ethnic groups. For the European-American speakers, the key finding was a
three-way interaction between confederate ethnicity, language type, and
stress. The pattern of the means was as follows: When rating their own eth­
nic group, no difference was observed in the magnitude of difference of rat­
ings given for sarcastic and literal comments in the low-stress condition
(literal = 1.86, sarcastic = 6.31) versus the high-stress condition (literal =
1.73, sarcastic = 6.23). But for ratings of African-American speakers, the
magnitude of difference between sarcastic comments and literal comments
dropped when moving from the low-stress (literal = 2.34, sarcastic = 6.18)
to the high-stress condition (literal = 2.44, sarcastic = 5.35).
The pattern of ratings for the African-American participants interest­
ingly did not reflect this pattern. Although the other general findings (e.g.,
that, all else being equal, sarcastic comments are rated as more sarcastic
than literal comments) were mostly the same for African-American and Eu-
ropean-American participants, the cross-ethnic comprehension degrada­
tion as a function of stress was not observed. Future research is required to
determine the reason for the different pattern in African-American partici­
pants. Perhaps it has to do with the relative familiarity of African Americans
to European Americans versus European-Americans to African Americans.
Essentially, what these results show is that, when placed in a high-stress
situation, some people's ability to comprehend a dissimilar (based on ethnic­
ity) speaker's nonliteral remark breaks down. These results support the hy­
pothesis concerning the mechanism of intergroup conflict, at least for the
majority (in the United States, Europe an-American) participants. More re­
126 COLSTON

search is required to explicitly link the remaining components of this mech­

anism. Researchers should consider (a) that unfamiliar social/cultural
encounters bring about the kind of stress shown here to affect comprehen­
sion, (b) that degraded comprehension leads to misunderstandings, and (c)
that misunderstandings can create or enhance intergroup conflicts, and so
forth, but the straightforward nature of such links makes them highly plausi­
ble, and they are consistent with the present data.
Thus, there is a very strong indication from this last line of research that
social/cultural variables not only have an effect on nonliteral language com­
prehension but that this effect is part of a larger mechanism that plays a sig­
nificant role in intergroup relationships. Thus, it might behoove researchers
to further investigate this phenomenon if for no other reason than that it
might enable us to derive better solutions to intergroup conflicts.

CONCLUSION

1 made arguments that social/cultural variables are on par with other kinds
of contextual information in providing an influence on language use, com­
prehension, and even processing—most notably for nonliteral language. I
reviewed my and other scholars' work that has shown clear empirical evi­
dence of such an influence at all of these levels of nonliteral language cogni­
tion. Finally, I showed that such influences may underlie other broader
phenomenon such as intergroup conflicts. I conclude by noting that, al­
though the work discussed here has made important strides in recognizing
the importance of social/cultural variables in language cognition, we have
much further to go both in the continued work on nonliteral language and
in work on more traditional literal language.
Mood, emotional state, physiological status, and a vast array of other
heretofore psychological influences have been shown to have tremendous
influences on many allegedly impenetrable cognitive, perceptual, and even
sensory processes. Moreover, juxtapositions, sequences, and reinstantia­
tions of these influences have their own set of ramifications that compound
the influence on such cognitive processes. To consider just one, simple ex­
ample, people's judgments of the magnitudes of quantities, a presumably
straightforward sensory-perceptual-cognitive process, are influenced by all
of the aforementioned factors. The amount of assigned work can seem
greater if one is overwhelmed. The weight of something to be carried can
seem lighter if one is happy. Even the sweetness of some substance can seem
greater if one is hungry. Also, the shade of a color will seem darker when
backdropped with bright light. The temperature of a liquid will seem colder
after having felt something hot. Even a memory for some past experience is
more likely to occur if the recaller is in the same emotional or other state as
when the original experience first happened.
5. INFLUENCES ON LANGUAGE 127

Many such straightforward cognitive, perceptual, or even sensory pro­

cesses are thus influenced by very broad contextual and structural factors.
And I considered just those influences on a given person's cognition, per­
ception, and sensation of their world—for instance, I as an individual am in­
fluenced in my cognition, about the world by factors such as my emotional
state, what has just come before this emotional state, and so on. When one
looks at language cognition, the picture is even more complex—when I act
as an interlocutor, empathic processes can make influences on the other per­
son have an effect on me. The other person can manipulate these contex­
tual and structural influences in the ways he or she acts or talks to have an
influence on me. Indeed, moving to a social interlocutorial setting opens up
its own enormous set of social influences that have been greatly explored in
social psychology and other fields. Just my mere tendency to categorize peo­
ple into in-groups and out-groups can have a tremendous influence on how I
will comprehend, process them, and so on (e.g., an unusual form of dress will
appear innovative on an ally but ridiculous on an enemy without my having
to explicitly recognize that he or she is wearing a shirt).
Thus, in our continued exploration of how and what people comprehend
and how and why they use nonliteral and indeed all kinds of language, we
clearly have to pay much greater attention to these structural, social, and
other kinds of broad influences in our consideration of heretofore very
low-level, bottom-up language processes. For nonliteral language the need
is perhaps more pronounced because of the somewhat greater importance of
contextual information, but the influences are present for all language types.
Nonliteral language researchers who have begun this consideration are thus
to be applauded, but we need to do more, and our colleagues who study
other arenas of language need to do more as well.

ACKNOWLEDGMENT

I thank Albert Katz for his comments on this chapter.

REFERENCES

Bloom, P. (2000). How children learn the meaning of words. Cambridge, MA: MIT
Press.
Byrnes, J. P., Miller, D. C., & Schafer, W. D. (1999). Gender differences in risk taking:
A meta analysis. Psychological Bulletin, 125(3), 367-383.
Carlson, K., & Cooper, R. E. (1974). A preliminary investigation of risk behavior in
the real world. Personality and Social Psychology Bulletin, I (1), 7-9.
Colston, H. L. (1997a). "I've never seen anything like it": Overstatement, under­
statement and irony. Metaphor and Symbol, 12, 43-58.
Colston, H. L. (1997b). Salting a wound or sugaring a pill: The pragmatic functions
of ironic criticism. Discourse Processes, 23, 25-45.
128 COLSTON

Colston, H. L. (1999a). "Not good" is "bad," but "not bad" is not "good": An analy­
sis of three accounts of negation asymmetry. Discourse Processes, 28, 237-256.
Colston, H. L. (1999b). The pragmatic functions of rebuttal analogy. Metaphor and
Symbol, 14, 259-280.
Colston, H. L. (2000a). Comprehending speaker intent in rebuttal analogy use: The
role of irony mapping, absurdity comparison and argumentative convention.
Language and Speech, 43(4), 337-354.
Colston, H. L. (2000b). "Dewey defeats Truman": Interpreting ironic restatement.
Journal of Language and Social Psychology, 19(1), 44-63.
Colston, H. L. (2001, May). Figurative language comprehension processes and group re­
lations. Paper presented at the meeting of the Midwestern Psychological Associa­
tion, Chicago, IL.
Colston, H. L. (2002a). Contrast and assimilation in verbal irony. Journal of
Pragmatics, 34, 111-142.
Colston, H. L. (2002b). Pragmatic justifications for nonliteral gratitude acknowl­
edgments: "Oh sure, anytime." Metaphor and Symbol, 17, 205-226.
Colston, H. L. (2004a). The pragmatics of asyndeton. Manuscript in preparation.
Colston, H. L. (2004b). Seriality versus complexity in verbal irony comprehension.
Manuscript submitted for publication.
Colston, H. L., & Gibbs, R. W., Jr. (1998). Analogy and irony: Rebuttal to "Rebuttal
Analogy." Metaphor and Symbol, 13, 69-75.
Colston, H. L., & Gibbs, R. W., Jr. (2002). Are irony and metaphor understood dif­
ferently? Metaphor and Symbol, 17, 57-80.
Colston, H. L, & Izett, C. (2004). Figurative language in social status management.
Manuscript submitted for publication.
Colston, H. L., & Jones, C. Q. (2004). Indirect language comprehension and social ten­
sion: Implications for group relations. Manuscript in preparation.
Colston, H. L., & Keller, S. B. (1998). You'll never believe this: Irony and hyperbole
in expressing surprise. Journal of Psycholinguistic Research, 27(4), 499–513.
Colston, H. L., & Lee, S. Y. (2000, May). On gender differences in verbal irony. Paper
presented at the meeting of the International Gender and Language Associa­
tion, Stanford University, Palo Alto, CA.
Colston, H. L., & Lee, S. Y. (2004). Gender differences in verbal irony use. Manuscript
submitted for publication.
Colston, H. L., & Lusch, V. (2004) .A gender difference in indirect language: The case of
asyndeton. Manuscript submitted for publication.
Colston, H. L., & O'Brien, J. (2000a). Contrast and pragmatics in figurative lan­
guage: Anything understatement can do, irony can do better, journal of
Pragmatics, 32, 1557-1583.
Colston, H. L., & O'Brien, J. (2000b). Contrast of kind versus contrast of magni­
tude: The pragmatic accomplishments of irony and hyperbole. Discourse Pro­
cesses, 30, 179-199.
Dews, S., Kaplan, J., & Winner, E. (1995). Why not say it directly? The social func­
tions of irony. Discourse Processes, 19, 347-367.
Dews, S., & Winner, E. (1995). Muting the meaning: A social function of irony.
Metaphor and Symbolic Activity, 10, 3-19.
5. INFLUENCES ON LANGUAGE 129

Dews, S., & Winner, E. (1999). Obligatory processing of literal and nonliteral mean­
ings in verbal irony. Journal of Pragmatics, 31 (12), 1579-1599.
Gibbs, R. W., Jr. (1986). What makes some indirect speech acts conventional? Jour­
nal of Memory and Language, 25(2), 181-196.
Gibbs, R. W., Jr. (1987). Memory for requests in conversation revisited. American
Journal of Psychology, 100 (2), 179-191.
Gibbs, R. W, Jr. (2000). Irony in talk among friends. Metaphor and Symbol, 15,5-27.
Gibbs, R. W, Jr., O'Brien, J. E., &. Doolittle, S. (1995). Inferring meanings that are
not intended: Speakers' intentions and irony comprehension. DiscoursePro­
cesses, 20, 187-203.
Giora, R. (1995). On irony and negation. Discourse Processes, 19, 239-264
Giora, R. (1997). Understanding figurative and literal language: The graded sa­
lience hypothesis. Cognitive Linguistics, 7, 183-206.
Giora, R., Fein, O., & Schwartz, T. (1998). Irony: Graded salience and indirect ne­
gation. Metaphor and Symbol, 13, 83-101.
Holtgraves, T. (1992). Interpersonal underpinnings of request strategies: General
principles and differences due to culture and gender. Journal of Personality and So­
cial Psychology, 62, 246-256.
Holtgraves, T. (1997a). Politeness and memory for wording of remarks. Memory and
Cognition, 25, 106-116.
Holtgraves, T. (1997b). Styles of language use: Individual and cultural variability in
conversational indirectness. Journalof Personality and Social Psychology, 73,624-637.
Holtgraves, T. (1991). Interpreting questions and replies: Effects of face-threat,
question from, and gender. Social Psychology Quarterly, 54, 15-24-
Howland, J., Hingson, R., Mangione, T. W, & Bell, N. (1996). Why are most drown­
ing victims men? Sex differences in aquatic skills and behaviors. American Jour­
nal of Public Health, 86(1), 93-96.
Jorgensen, J. (1996). The functions of sarcastic irony in speech. Journal of
Pragmatics, 26, 613–634.
Katz, A., & Lee, C. (1993). The role of authorial intent in determining verbal irony
and metaphor. Metaphor and Symbolic Activity, 8, 257-279.
Katz, A., & Pexman, P (1997). Interpreting figurative statements: Speaker occupa­
tion can change metaphor to irony. Metaphor and Symbol, 12, 19–41.
Katz, A., Piasecka, I., & Toplak, M. (2001, November). Comprehending the sarcastic
comments of males and females. Poster session presented at the 42nd annual meet­
ing of the Psychonomic Society, Orlando, FL.
Kemper, S., & Thissen, D. (1981). Memory for the dimensions of requests. Journal of
Verbal Learning and Verbal Behavior, 20(5), 552-563.
Kreuz, R. J. (1996). The use of verbal irony: Cues and constraints. In J. S. Mio & A.
N. Katz (Eds.), Metaphor: Implications and applications (pp. 23-38). Mahwah, NJ:
Lawrence Erlbaum Associates.
Kreuz, R. J. (2000). The production and processing of verbal irony. Metaphor and
Symbol, 15, 99-107.
Kreuz, R. J., Kassler, M., Coppenrath, L., & McLain, A. (1999). Tag questions and
common ground effects in the perception of verbal irony. Journal of Pragmatics,
31(12), 1685-1700.
130 COLSTON

Kreuz, R. J., & Link, K. E. (2002). Asymmetries in the use of verbal irony,journal of
Language and Social Psychology, 21(2), 127-143.
Kreuz, R. J., Long, D. L., & Church, M. B. (1991). On being ironic: Pragmatic and
mnemonic implications. Metaphor and Symbolic Activity, 6, 149-162.
Lee, C., & Katz, A. (1998). The differential role of ridicule in sarcasm and irony.
Metaphor and Symbol, 13, 1-5.
Leggitt, J., & Gibbs, R. W., Jr. (2000). Emotional reactions to verbal irony. Discourse
Processes, 29, 1-24.
Okamoto, S. (2002). Politeness and the perception of irony: Honorifics in Japanese.
Metaphor and Symbol, 17, 119-139.
Pexman, P. M., Ferretti, T., & Katz, A. (2000). Discourse factors that influence
on-line reading of metaphor and irony. Discourse Processes, 29, 201-222.
Pexman, P. M., & Olineck, K. M. (2002). Understanding irony: How do stereotypes
cue speaker intent? Journal of Language and Social Psychology, 21, 245-274.
Roberts, R. M., & Kreuz, R. J. (1994). Why do people use figurative language? Psy­
chological Science, 5(4), 159-163.
Schwoebel, J., Dews, S., Winner, E., & Srinivas, K. (2000). Obligatory processing of
the literal meaning of ironic utterances: Further evidence. Metaphor and Symbol,
15,47-61.
Tomasello, M., & Akhtar, N. (1995). Two-year-olds use pragmatic cues to differen­
tiate reference to objects and actions. Cognitive Development, 10, 201-224.
Toplak, M. & Katz, A. N. (2000). On the uses of sarcastic irony. Journal of
Pragmatics, 32, 1476-1488.
Whaley, B. B., & Holloway, R. L. (1996). "Rebuttal" analogy: A theoretical note.
Metaphor and Symbolic Activity, 11, 161-167.
Wiederman, M. W. (1997). The truth must be in here somewhere: Examining the
gender discrepancy in self-reported lifetime number of sex partners. journal of Sex
Research, 34(4), 375-386.
Winner, E., & Gardner, H. (1993). Metaphor and irony: Two levels of understand­
ing. In A. Ortony (Ed.), Metaphor and thought (2nd ed., pp. 425-443). New York:
Cambridge University Press.
 6

Irony as Persuasive
Communication
Raymond W. Gibbs Jr.
Christin D. Izett
University of California, Santa Cruz

Persuading others to think or act in certain ways can be very difficult to ac­
complish. For example, how would you persuade teenagers not to smoke cig­
arettes? Many people suggest that the best way to convince others not to
engage in this risky behavior is to have a serious, somber conversation in
which the facts about the effects of smoking on health and life expectancy
are described and discussed. This kind of plain talk may be impressive to
some adults, but do such arguments convince teenagers?
A way to persuade teenagers is through the use of special rhetorical de­
vices that may catch their attention and possibly seduce them to adopt some
belief. One widely employed figure of speech in advertising that may nicely
serve this function is irony. Consider the following radio advertisement,
sponsored by the California Department of Health Services, played in May
1998 on California radio stations. The radio spot is spoken in the voice of a
60-year-old man in a very sincere tone (California Department of Health
Services):

We the Tobacco Industry, would like to take this opportunity to thank

you, the young people of America, who continue to smoke our ciga­
rettes despite Surgeon General warnings that smoking causes lung
cancer, emphysema, and heart disease. Your ignorance is astounding,
and should be applauded. Our tobacco products kill 420,000 of your
parents and grandparents every year. And yet, you've stuck by us. That
13 1
132 GIBBS AND IZE1T

kind of blind allegiance is hard to find. In fact, 3,000 of you start smok­
ing everyday because we tobacco folks tell you it's cool. [Starts to get
carried away.] Remember, you're rebels! Individuals! And besides, you
impressionable little kids are makin' us tobacco guys rich!! Heck, we're
billionaires!! [Clears throat/composes himself.]. In conclusion, we the
tobacco conglomerates of America, owe a debt of gratitude to all teens
for their continued support of our tobacco products despite the unfor­
tunate disease and death they cause. Thank you for your understand­
ing. Thank you for smoking. Yours truly, The Tobacco Industry.

Listeners likely have different reactions to this radio message. Many adults
who decry the way the tobacco industry seduces children to smoke see great
humor in the speaker's irony (e.g., "That kind of blind allegiance is hard to
find"). Others may not get the point of this ad despite its heavy-handed use of
irony throughout the speaker's heartfelt expression of gratitude. But most
people certainly pay attention to the ad, precisely because it adopts a perspec­
tive that seems so incongruous (e.g., thanking people for engaging in a behav­
ior that may likely end up killing them). We do not know whether this specific
ad is persuasive to teenagers and gets them to not smoke. Yet, it is clear that
advertisers see irony as a special weapon in their attempt to capture people's
attention and persuade them to act in certain ways.
This chapter describes the role that irony plays in persuasive communica­
tion. We argue that irony has several special features that make it a wonder­
ful tool for persuasion, especially the ability to highlight the contrast
between expectation and reality. Irony works to create ad hoc intellectual
communities that lead audiences to view themselves, even temporarily, as
"conspirators" in accepting the values to which irony indirectly alludes. We
describe several instances of this thesis in various applied contexts (e.g., ad­
vertisements, personal arguments, political debates, intellectual discus­
sions, and literature) and suggest how contemporary research in cognitive
and social psychology provides good evidence for why irony is employed so
frequently in persuasive communication.

WHY USE IRONY?

Irony can be fraught with perils, enough so that one may reasonably ask why
any person should ever use irony. Irony divides its audience in two different
ways (Kaufer, 1977). First, there is the distinction between those who recog­
nize the irony and those who do not. Those who recognize the irony under­
stand what the speaker-author intends to say; because of their wisdom they
may be called wolves. Those who fail to recognize the irony mistake what the
speaker-author appears to say for what he or she intends to say. For their
gullibility, these people may be called sheep.
6. PERSUASIVE COMMUNICATION 133

Another difference in audiences is between those who agree with the

speaker-author's intended meaning and those that do not. Supporters of
irony are confederates; those that disagree are victims. These two groups are
not the same as wolves and sheep, because understanding what the
speaker-author intends to say and agreeing with it are distinct aspects of
communication (Kaufer, 1977).
Employing irony, therefore, divides the audience into four groups
(Kaufer, 1977). First, there are those who recognize the irony and agree with
the author's intended message (i.e., wolf-confederates). For example, a
speaker says "President Bush is such a humanitarian" and the addressee un­
derstands this statement as a sarcastic comment about President Bush and
agrees with the speaker's opinion. A second group includes people who rec­
ognize the irony but disagree with the author's intended message (i.e.,
wolf-victims). Thus, an addressee may hear the speaker's sarcastic com­
ment about President Bush, understands the speaker ironic meaning, but
disagrees with the speaker's assessment. A third group includes those who
do not recognize the irony but would agree with the author's message if they
had correctly understood it (i.e., sheep-confederates). For example, a lis­
tener may not recognize the speaker's intended irony in saying "President
Bush is a humanitarian" but would agree with the speaker's assessment if he
or she had understood it correctly. Finally, there is a group that does not rec­
ognize the irony and would not accept the author's communicative message
(i.e., sheep-victims).
One proposal suggests that the main job of an ironic speaker-author is to
create as many wolf-confederates as possible and minimize the number of
sheep-confederates who wrongly believe themselves opposed to the cre-
ator's position (Kaufer, 1977). For example, in the antismoking ad just pre­
sented, there are sufficient cues to the ad's ironic message (e.g., obvious
incongruity, statements like "Your ignorance is outstanding," and dramatic
tones of voice) for listeners to readily understand the ad and agree with its
underlying message (i.e., wolf-confederate). Of course, there may be some,
like the Tobacco Industry, who recognize the irony but are not persuaded by
its urgent appeal for young people not to smoke (i.e., wolf-victim). Yet, the
California Department of Health Services may specifically intend to mock
these wolf-victims as part of their message. In fact, teenagers may be more
likely persuaded to not smoke by realizing that they are part of a special
group of individuals who are smart enough to make fun of the Tobacco In­
dustry for its continued efforts at promoting cigarette smoking.
In addition to these rhetorical complexities, irony it is often regarded as
special because it represents a mode of intellectual deattachment. "Irony
engages the intellect rather than the emotions" (Walker, p. 24) and sits on
"the cutting edge of not caring" (Austin-Smith, p. 51). Commentators since
Aristotle have seen irony in speaking or writing as a sign of sophistication, at
134 GIBBS AND IZETT

the very least in the use and understanding of language. Irony is inherently
elitist in setting apart an elite (one who understands and employs irony)
from the masses (those who neither use nor understand irony). This elitism
accounts in part for irony's reputation since the time of Cicero as a gentle­
manly form of discourse. Academic scholars and cultural critics have fre­
quently characterized the late 20th-century as the Age of Irony. Many
Americans enjoyed the pleasures of ironic detachment—skepticism, satiri­
cal humor, and, at times, a downright lack of caring—when it came to both
private and public affairs. Younger people, particularly those of Generation
X (i.e., born in the 1970s and later), employed irony as their dominant dis­
course style. Popular culture, for better or worse, has been dripping with
irony, as books, films, television, magazines, and music all indulged in vari­
ous forms of irony to highlight the great divide between expectations and re­
ality. The antismoking ad is a perfect illustration of this.
But psychological research suggests that irony does not necessarily reflect
or promote detachment or lead to apathy. People use irony to achieve a com­
plex set of social and communicative goals (Colston, 1997; Dews & Winner,
1995; Kumon-Nakamura, Glucksberg, & Brown, 1995; Lee & Kutz, 1998;
Roberts & Kreuz, 1994), including being humorous, acting aggressively,
achieving emotional control, elevating one's social status, expressing atti­
tudes, provoking reactions, mocking others, and muting the force of one's
meaning. Some scholars have maintained that one form of irony, ironic criti­
cism, automatically reduces the amount of condemnation that listener's ex­
perience (Dews & Winner, 1995). By stating literally positive words in an
ironic criticism (e.g., "A fine friend you are!"), the speaker ensures that lis­
teners will interpret what is said in a more positive manner. The results of
several studies show that critical ironic statements are rated as less severe
than literal ones (Dews & Winner, 1995). Studies like Dew and Winters'
support the idea that irony can work to reduce the personal impact of a
speaker's message.
However, other studies demonstrate that ironic criticism actually en­
hances, rather than dilutes, condemnation and criticism (Colston, 1997).
Thus, the positive direct meaning of an ironic utterance, such as "How
pleasing!" increases the degree of criticism listeners perceive, compared to
when they hear direct literal remarks, such as "How disgusting!" The con­
trast between the actual situation and the speaker's positive remark, again
relative to the literal comment, gives rise to an enhanced sense of criticism
(Colston, 1997).
Irony's ability to mock, attack, and ridicule, provoking embarrassment,
humiliation, even anger, suggests that it may, in many circumstances, work
to evoke strong emotional reactions in an audience. Part of the appeal of the
antismoking ad is that it clearly makes listeners feel something, regardless of
whether that emotion is interpreted positively (e.g., "I am sure glad that I'm
6. PERSUASIVE COMMUNICATION 135

not one of those ignorant persons who smoke") or negatively (e.g., "I feel
guilty because I smoke"). Research shows that different forms of irony evoke
different kinds of emotional reactions, especially when compared to plain,
literal talk (Leggitt & Gibbs, 2000).
Imagine yourself in the following situation: One day while parking your
car at work Mary is splashed with mud. Mary walks over to your car while
you are getting out. You look at Mary and ask why her clothes are such a
mess. Mary looks back at the mud puddle in the road and answers: "You
splashed mud on me with your car." Mary's utterance might very well make
you feel guilty for doing what you, perhaps unintentionally, did to dirty her
clothes. However, Mary's comment, although quite factual, does not convey
much about her own attitude or emotion in regard to your act. She might
simply be calling attention to what you did while forgiving your actions, or
she might actually be displeased or quite angry with you.
Suppose, instead, that Mary actually uttered the sarcastic remark,
"Thanks a lot for giving me a bath." Once again, you might feel guilty upon
being alerted to your misdeed. Yet, by virtue of informing you about what
you did in a sarcastic manner, you are likely to recognize that Mary appears
rather angry. Sarcasm is considered especially appropriate for conveying a
person's hostile attitude toward or ridicule of some other individual, usually
the addressee (Gibbs, 1994; Lee & Katz, 1998). Consequently, being the ob­
ject of another person's sarcasm might provoke intense emotional reactions.
People may, for example, feel angry when they are the victims of sarcastic
comments, because they resent the mocking style of the speaker's implied
criticism. Moreover, speaking sarcastically may signal quite a different con­
ceptualization of some event than if someone made literal comments, such
as if Mary, in the aforementioned scenario, said "You have splashed mud on
me" or "What you did makes me angry."
Leggitt and Gibbs (2000) demonstrated that people experience different
emotional reactions to various ironic statements than to literal ones and feel
more intense emotions having heard ironic remarks than literal ones. For in­
stance, sarcasm, rhetorical questions, and overstatement all evoked similar
and quite negative reactions, whereas understatement and satire evoked
relatively neutral reactions. In general, a person's emotional reaction to an
ironic statement depends on the degree to which the speaker directly chal­
lenges an addressee or makes a big deal of an issue. Listeners' evaluations of
what ironic speakers must be feeling also differ depending on the type of
ironic language employed. Speakers of sarcasm and rhetorical questions ap­
peared to feel a wide range of negative emotions, but speakers of overstate­
ments appeared to feel relatively neutral emotions. Moreover, speakers of
satire were seen to feel more positive emotions and a lower degree of hostile
emotions, and speakers of understatements were seen to feel relatively neu­
tral emotions. These complexities in people's emotional reactions to differ­
136 GIBBS AND IZETT

ent forms of irony highlight the pleasures and pitfalls associated with ironic
communication. Speaking ironically may evoke strong emotions, yet these
may not always be experienced as positive feelings.
People often use irony not to disparage others but to remind each other of
the bonds that tie them together. Consider the following exchange between
two college students (from the data described in Gibbs, 2000, p. 7). Melissa
and a friend are in the kitchen of Melissa's apartment talking about the pre­
vious night. As they were talking, Melissa's housemate Jeanette, who was in
her own bedroom, heard her name being mentioned:

Jeannette: (yelling from her room) Are you talking about me again?
Melissa: I have no life Jeannette. All I do is talk about you. All the
time.
Jeannette: (laughing) Get a life!

The housemates' interaction here is jocular, rather than sarcastic, in that

each speaker was not entirely serious about the impact of what they were
saying. Both speakers clearly enjoyed the exchange and found some humor
in what each other said. But this minor exchange reflects one way that irony
helps persuade others. When Jeannette, perhaps glibly, accuses Melissa of
talking or gossiping about her, Melissa's jocular retort works to reassure
Jeannette that she is not the constant topic of conversation or gossip. This
slight instance of reassurance is persuasive in the sense that it reminds the
two women of the good-natured friendship they have, precisely because
they can speak jocularly without significant risk of having their intentions
be misunderstood.
Irony is, in this way, a particularly compelling means of reaffirming pre­
suppositions common to both the speaker-author and the audience. Thus,
irony involves foregrounding the normative standards that make possible
the correct interpretation of ironic statements. Summoning common norms
and reinforcing them by apparently violating them is another way by which
an ironic speaker-author can establish an implicit solidarity with the audi­
ence and persuade to adopt, or be reminded of, a certain belief.
Ironic language also enables speakers to establish bonds by marking who
is part of the "in-group" and who is not. Consider the following conversation
in which a group of college students disparage another person. This ex­
change took place outside a campus coffee shop (Gibbs, 2000, p. 8):

Kayla: How are you doing?

Cherie: Um ... good, we're going to study Latin but the coffee
shop is just packed.
David: It's rockin'
6. PERSUASIVE COMMUNICATION 137

Sarah: I ... study Latin ... Latin language?

Kayla: It's wet out here.
Sarah: You guys are taking Latin? (laughs)
Cherie: Yeah ... (laughs)
Kayla: (whiny tone of voice) But that's a dead language (every­
one laughs) I'm just kidding is that not what everyone
tells you?
Cherie: It's true and we don't really know how to pronounce ev­
erything.
David: It's really hard.
Cherie: Yeah, but it's only a year long program.
David: So, you're fluent in Latin after a year. (everyone laughs)
Kayla: Right ... right.
David: It's true, (everyone laughs)
Sarah: You read all those ancient texts, that's cool, (laughs)
Cherie: Why you guys dissin' on Latin?
David: (mocking tone) What, wo-ah, you're dissin' my Latin.
Kayla: Actually, Latin helps because, doesn't it, it helps with ety­
mology it help with words, breaking words down.
David: totally ... yeah, yeah, she got it ... yeah.
Cherie: structure, parts of speech, yeah.
David: I'm a changed person since the last couple of weeks of
Latin.

Teasing and jocularity, like that seen in this dialogue, offer gentle, indirect
ways of pointing out people's deviation from social standards and is central
to socialization practices between parents and children, friends, and roman­
tic partners. By ironically noting the discrepancy between the reality of
Cherie and David studying Latin and the expectation or belief that Latin is
nothing more than a dead language, Kayla and Sarah aim to persuade the
others of their beliefs about what is "hip" and what is not. The content of an
ironic message can solidify the bond between ironist and audience, which
also enhances the persuasive impact of an ironic argument. Yet, ironic argu­
ments also communicate to an audience that the ironist shares certain be­
liefs and values (i.e., conspirators) that others, to their detriment, do not
(i.e., victims).
A more blatant example of how noting the discrepancy between expecta­
tion and reality works to persuade is found in Swift's (1729/1996) A Modest
138 GIBBS AND IZETT

Proposal Readers are forced to recognize the desperate poverty of the Irish
people by the ironic solution of infanticide and cannibalism. Swift's purpose
was to enrage his readers against the policies that had allowed the situation
in Ireland to deteriorate to the point where a scheme to sell children as food
was no more than a "modest proposal." Our understanding of Swift's ironic
message creates an ad hoc community of those who share Swift's beliefs and
values. The pleasure derived from the recognition of irony is compounded
by our awareness of our own competence as its audience.
Various empirical research, in fact, demonstrates that irony is especially
useful for highlighting the contrast between what was expected and what
ensued. Experimental findings suggest that irony creates more contrast be­
tween expectations and resulting events than does understatement,
whereas both forms of irony create more contrast than do literal statements
(Colston & O'Brien, 2OOOb). Furthermore, the degree of contrast between
expectation and reality also affects listeners' judgments of how humorous,
how condescending, how expressive of surprise, and how protective of a
speaker any ironic remark appears to be (Colston & Keller, 1998; Colston &
O'Brien, 2000a). Other studies show that when a speaker restates another
person's inaccurate remark (e.g., "Sure, Ronald Reagan was the president
during the 1970s"), the ironic message works better than either refutation or
correction to communicate the idea that the first speaker should have
known better (Colston, 2OOOc). This body of work indicates that people em­
ploy irony to specifically and succinctly comment on the disparity between
expectations or beliefs and what is actually happening.

WHY IS IRONY PERSUASIVE?

SOCIAL PSYCHOLOGICAL EVIDENCE

Social psychology research supports our contention that irony may be espe­
cially effective in persuasive communication because it highlights the con­
trast between expectation and reality. For instance, the contrast principle of
human perception describes how people view the difference between two
options that are presented in succession. The basic premise of the contrast
principle is that if the second item presented is distinctly different than the
first, we will perceive it as being more different than it is in actuality. Con­
sider the following example: If you lift a light item first, and then a heavy
item, you will estimate that the second item is heavier than it actually is.
Conversely, if a heavy item is lifted first, lifting a lighter item in turn will ap­
pear to be almost effortless in contrast to the heavy item.
The contrast principle is well established in the field of psychophysics and
has been applied in a variety of social influence situations. Sales organiza­
tions and marketing executives worldwide make good use of the contrast
principle in influencing consumers to purchase goods. Automobile dealers
6. PERSUASIVE COMMUNICATION 139

use the contrast principle to increase accessory sales. Dealers will wait until
the price for the automobile is negotiated before they suggest one accessory
option after another to the buyer, who in wake of a multithousand dollar
purchase perceives a few extra hundred dollars as trivial in comparison.
Home real estate agents utilize the contrast principle in increasing the dollar
amount of properties that they sell. When showing a potential buyer a set of
properties, an agent will almost always start by showing a couple of undesir­
able properties. Real estate companies call these homes set-up properties. Es­
sentially, the companies maintain a few properties at inflated prices on its
lists. These homes are not intended to be sold to buyers, but are used as a
comparison to the genuine properties to increase the likelihood that they
are sold. After looking at a few run-down and high-priced properties, the
genuine properties, although more expensive, appear to be the deal of a life­
time (Cialdini, 1975). The main point of the contrast principle is that any el­
ement of human perception can be made to appear very different than it
really is, depending on the nature of the event that precedes it.
A wonderful example of the contrast principle at work is seen in the fol­
lowing letter that a college student wrote to her parents (Cialdini, 1984, pp.
15-16):

Dear Mother and Dad:

Since I left for college, I have been remiss in writing and am sorry for
my thoughtlessness in not having written before. I will bring you up to
date now, but before you read on, please sit down. You are not to read
any further unless you are sitting down, okay? Well, then, I am getting
along pretty well now. The skull fracture and the concussion I got
when I jumped out the window of my dormitory when it caught on fire
shortly after my arrival here is pretty well healed now.
I only spent two weeks in the hospital and now I can almost see nor­
mally and only get those sick headaches once a day. Fortunately, the
fire in the dormitory, and my jump, were witnessed by an attendant at
the gas station near the dorm, and he was the one who called the fire
department and the ambulance. He also visited me in the hospital and
since I had nowhere to live because of the burnout dormitory, he was
kind enough to invite me to share his apartment with him. It's really as
basement room, but it's kind of cute. He's a very fine boy and we have
fallen deeply in love and are planning to get married. We haven't got
the exact date yet, but it will be before my pregnancy begins to show.
Yes, Mother and Dad, I am pregnant. I know how much you are look­
ing forward to being grandparents and I know you will welcome the
baby and give it the same love, devotion and tender care you gave to
me when I was a child. The reason for the delay in our marriage is that
14O GIBBS AND IZETT

my boyfriend has a minor infection, which prevents us from passing

our premarital blood tests and I carelessly caught it from him.
Now that I have brought you up to date, I want to tell you that there is
no dormitory fire, I did not have a concussion or a skull fracture, I was
not in the hospital, I am not pregnant, I am not engaged, I am not in­
fected and there is no boyfriend. I am getting a "D" in American His­
tory and an "F" in chemistry and I want you to see those marks in their
proper perspective.
Your loving daughter, Sharon

By getting her parents to believe one set of events, the reality of Sharon's
poor grades seems almost insignificant. In this way, Sharon has used the con­
trast principle to persuade others of her point of view. Irony often works to
persuade others through the contrast principle. A good place to find exam­
ples of this is in advertisements. Consider the headline for the Range Rover:
"The British have always driven on the wrong side of the road," accompa­
nied by a picture of the automobile driven on a steep slope off to one side of
the road. Understanding this headlines requires that observers be aware
that the British drive on the left side of the road, and that the left side is the
correct side in Britain, although it seems wrong to those accustomed to the
alternative. An observer may thus reflect that for a 4-wheel drive vehicle,
the wrong side of the road (i.e., off the road altogether) is the right side. Yet,
people may readily infer that it may be wrong for an auto to leave the road
but right (pleasurable and advantageous) not to be bound by the road. Not
all observers will necessarily draw these complex inferences, but the adver­
tisers' choice of an ironic message destabilizes one's understanding of what is
right or wrong which liberates the observer's thoughts; perhaps he or she
sees the beauty or even the necessity of driving or owning a Range Rover.
A different form of contrast is seen in a Kodak ad stating "This picture
was taken by someone who didn't bring a camera." This message is self-con-
tradictory and ironically destabilizes the idea that taking pictures requires a
camera, and it leads observers to the possibility of buying a disposable cam­
era (made by Kodak) on the spot. Thus, one's conventional assumption of
bringing along a camera to take photographs has been ironically upended.
Consider one final ad to demonstrate irony's potential effectiveness in
persuading by extreme contrast. In an effort to persuade people to devote
their valuable time to a volunteer effort (to increase literacy by helping the
illiterate learn how to read), an ad agency based in San Diego launched a
creative campaign that employs the heavy use of sarcasm, guilt, and humor
as persuasive devices. They describe this approach as a form of reverse psy­
chology. Previous campaigns utilized a more traditional approach to persua­
sion whereby potential volunteer recruits where told that increasing literacy
6. PERSUASIVE COMMUNICATION 141

among fellow citizens is simply "the right thing to do." The sarcasm cam­
paign employs a very different approach.
Here are some examples of their campaign ads: "Keep America Stupid!"
and "Fight Literacy!" and "This country has enough smart people, what we
need are more stupid people!" The ads then urge message recipients not to
call the associated toll-free number to enlist as a volunteer. Within the first 2
days of the campaign launch, 60 people contacted the San Diego Council on
Literacy to express an interest in volunteering. This number is staggering in
contrast to the negligible number of volunteers recruited by ads emphasizing
that volunteering simply is the right thing to do. The sarcasm campaign is
being expanded nationwide due to its initial effectiveness (Stewart, 2000).
The norm of reciprocity is another well-known social psychological con­
cept that utilizes the contrast principle to increase compliance and influ­
ence. A norm is a specific guide to conduct. Breaking a norm will most likely
lead to some form of social sanction, and being put in a position of trans­
gressing a norm causes even small children tremendous anxiety. To avoid
feeling anxious or being disapproved of, most people comply with social
norms without question. The norm of reciprocity is a good of example of a
social norm in American culture: If I do something for you, you are then ob­
ligated to return the favor and do something for me (Cialdini, 1978).
One example of the norm of reciprocity working with the contrast princi­
ple to enhance social influence is the door-in-the-face technique. This tech­
nique increases compliance by operating as a type of perceptual hyperbole
(creating a large contrast). First, a person is asked for a very extreme favor.
The favor must be something to which almost no one would comply. Sec­
ond, the requester makes a concession, thus invoking the norm of reciproc­
ity, and asks for a much smaller favor. The implicit message by the requester
is that "because I made a concession and am compromising now it is your
turn to reciprocate my concession and grant me this favor."
For instance, Cialdini and Ascani (1976) asked university passersby to
donate either a unit of blood sometime tomorrow or donate a unit of blood
once every 2 months for a period of 2 years. When this extreme favor was in­
evitably rejected, the passersby were asked to simply donate a unit of blood
sometime tomorrow. Results of this study showed that people not only
agreed to give more blood but actually followed through on their agreement
when they received the more extreme request first.
The contrast principle and the norm of reciprocity together make certain
forms of irony especially persuasive. Consider a situation in which one friend
asks another for a loan by saying "I need to borrow ten thousand dollars to
pay my mortgage, which is way overdue, and to send in my son's college tui­
tion, and to keep my car from being impounded." Listeners, especially
friends, may react sympathetically to this appeal, despite its extreme nature,
because of their desire to do the right thing and be helpful to a friend in need.
142 GIBBS AND IZETT

Of course, $10,000 is a great deal of money to most people, and this request
would be almost impossible to fulfill. But if the speaker then said, "Okay, 1 re­
ally just need fifty bucks to pay my phone bill," then the listeners will experi­
ence a great sense of relief given the contrast to the original request.
Consequently, the listener will be more likely lend the $50 than would be the
case if the speaker had simply come out with this request in the first place.
The norm of reciprocity, along with the contrast principle, makes hyperbole
(one form of irony) especially useful in persuading others to comply with
one's wishes.
Because ironic statements or arguments by nature use contrast to high­
light the discrepancy between expectation and reality, irony appears to be an
excellent example of utilizing the contrast principle to increase persuasion.
Also, ironic statements act to destabilize the listener by presenting messages
that appear contradictory to their previous beliefs or the reality of the situa­
tion. This destabilization forces listeners to acknowledge their beliefs, see if
they are consistent with the present situation, and then reevaluate their po­
sition in light of the ironic statement or ad. In this way the destabilizing force
of irony opens up listeners or observers to potentially adopting new beliefs,
or at least seriously questioning their old ones. Once in this position, listen­
ers or observers are more vulnerable to being persuaded or influenced by the
speaker or ad.
By highlighting the discrepancy between expectation and reality, irony
often produces a state of cognitive dissonance whereby an individual simul­
taneously holds to cognitions (ideas, attitudes, beliefs, and opinions) that
are discrepant or psychologically inconsistent. This discrepancy produces a
state of tension that can be very uncomfortable, and people become highly
motivated to reduce it. Reducing cognitive dissonance involves changing
one or both cognitions in such a way as to render them more compatible
(more consonant) with each other or adding more cognitions to act as a
bridge to the original cognitions (Festinger, 1957).
Consider the following example: A man who smokes cigarettes reads a re­
port citing medical evidence linking cigarette smoking to cancer, lung dis­
ease, and premature death. The cognition "I am a smoker" is dissonant with
the cognition "smoking will kill me." In an effort to reduce dissonance, a
smoker may try to quit smoking. When he finds this task too challenging he
may try to change the cognition "smoking will kill me." He may convince
himself that the evidence that smoking kills is inconclusive. He may switch
to a filtered brand, convinced that the filter will block the cancer-causing
materials from entering his lungs. He may convince himself that he needs
cigarettes to relax and ultimately to be happy and, therefore, conclude he
would rather live a shorter, happier life. Reducing dissonance in any of these
ways helps stabilize listeners by decreasing the discrepancy they experience
between expectation and reality.
6. PERSUASIVE COMMUNICATION 143

The antismoking ad discussed at the beginning of the chapter offers an

excellent example of a commercial that produces cognitive dissonance. This
ad highlights the two psychologically inconsistent cognitions: "I am a
smoker" and "smoking cigarettes will kill me." The ad then uses irony to
eliminate the possibility of changing the cognition "smoking will kill me" by
making sarcastic remarks about possible dissonance reducing cognitions,
such as "you are a rebel!" or "you are an individual!" The listener's only op­
tion is to change the cognition "I am a smoker."
A famous example of how ironic images provoke cognitive dissonance is
seen in Margaret Bourke-White's photograph of a breadline during the
American Depression of the 1930s. The photograph shows a group of im­
poverished African Americans waiting in line for a handout. Above them is
a billboard with a picture of a White family of four driving happily in a car.
The caption in the billboard reads "World's Highest Standard of Living.
There's No Way Like the American Way!" Bourke-White invites us to expe­
rience a heightened sense of dissonance from the recognition of the contrast
between expectation and reality and by threatening ones self-image as a car­
ing, generous person. Part of this dissonance arises from observers' apprecia­
tion of the hypocrisy in the American ideal of us being a kind and prosperous
nation that ignores the plight of the poor. In general, when presented with a
communication or image that produces dissonance, the listener or viewer is
very vulnerable to influence. By setting the decision-making criteria to in­
clude options that are desirable outcomes for the speaker or ad and will re­
duce dissonance for the listener or viewer, the ultimate resolution will
benefit both parties (i.e., the listener or viewer is persuaded to comply to the
request of the speaker or ad and listener or viewer reduces dissonance and
reestablishes a positive self-image).
Another example of how irony can lead to dissonance and persuasion is
seen in a study on decreasing HIV transmission (Aronson, Fried, & Stone,
1991; Stone, Aronson, Grain, Winslow, & Fried, 1994). One group of col­
lege students was asked to deliver a compelling speech to their peers about
the benefits of using condoms and the dangers of HIV and disease transmis­
sion. The speech was videotaped, and the students were told that the tape
would be shown to younger (i.e., more impressionable) high school students
as part of a sexual education course. The students were then asked to recol­
lect instances from their daily lives when they found it difficult, if not impos­
sible, to use condoms during intercourse.
The experimenters set up a situation whereby the college students, who
were more than likely not practicing safe sex, were made to be cognizant of
their own actions and the action that they were advocating in their speech.
Through the use of irony, hypocrites were confronted with their own hy­
pocrisy. This irony produced a state of cognitive dissonance; to reduce dis­
sonance and reestablish integrity, college students resolved to change
144 GIBBS AND IZETT

their actual actions to be consonant with the actions they endorsed in

their speech. Indeed, compared to college students who simply produced
videotapes promoting condom use but were not asked to recall instances of
not practicing safe sex (i.e., not made mindful of their own hypocrisy), stu­
dents who were made to feel like hypocrites purchased far more condoms
at the end of the study. Moreover, in a phone interview conducted 3
months later, 2% of these students reported using condoms regularly, more
than twice as many as reported by hypocritical students who were not con­
fronted with their own hypocrisy. Making people aware of their own ironic
or hypocritical attitudes seems to be a very effective way of influencing hu­
man behavior. Advertisers may be well aware of this fact and use it to their
profitable advantage.
Finally, social psychological research suggests another reason why ironic
messages may be especially persuasive. A popular theory of social influence
claims that attitude change arises from two sources (Petty & Cacioppo,
1986). The central route to persuasive communication relies on solid argu­
ments, logic, and essential facts and figures. Listeners carefully scrutinize all
the information presented and think carefully about the issue at hand. Con­
versely, the peripheral route to persuasion does not rely on engaging a per-
son's thinking, but rather it relies on providing distracting cues that
stimulate the acceptance of an argument with little or no thought. Irony
may be especially persuasive because it works on both the central and pe­
ripheral routes for attitude change. For example, in the California anti­
smoking campaign, people are persuaded by central arguments such as the
fact that smoking kills 420,000 people per year and by peripheral cues such
as humor and the speaker's mocking tone of voice. Thus, irony probes peo­
ple to think about some matter by highlighting the contrast between expec­
tation and reality, simultaneously evoking more extreme emotional
reactions than typically felt when hearing literal statements.

IRONY IN INTELLECTUAL DEBATES

Our claim that irony is persuasive because of its special ability to highlight
contrast and evoke cognitive dissonance is supported by other work on irony
in intellectual debates. Whaley and Holloway (1996) argued, for example,
that certain kinds of rebuttal analogies are especially effective in argumenta­
tion. For instance, one economist may use the analogy "Giving tax breaks to
the rich is like putting sprinklers in the rainforest" when arguing with econo­
mists favoring tax cuts for wealthy Americans to stimulate the economy.
Colston and Gibbs (1998) claimed that irony is a central element in rebuttal
analogies. Quite specifically, in rebuttal analogy, a seemingly nonironic, un-
seen-as-ironic, or simply less ironic target domain is paired with a clearly
ironic base domain, which highlights via irony the widely contrasting rela­
6. PERSUASIVE COMMUNICATION 145

tions between the base and target. Consider the following two examples of
ironic rebuttal analogies (Colston & Gibbs, 1998, p. 72):

The White House trotted out Dick Darman, the budget director, to
blast Governor Clinton's deficit reduction. This is a man who's run $1
trillion 238 million of deficits. Isn't that a little like the Boston Stran­
gler criticizing street crime? (Cable News Network, 1992)

Base: The Boston Strangler criticizing street crime.

Target: A person who ran huge deficits criticizing overspending.
Irony: It is ironic for a criminal to condemn crime because crime is
the behavior in which a criminal engages. It is analogously
ironic for a person guilty of overspending to criticize that very
behavior.

An inexperienced Secretary of Defense getting "on--the-job'training" is

like taking your first drink of water from a fire hose. (National Public Ra­
dio, 1989)

Base: Taking your first drink of water.

Target: An inexperienced Secretary of Defense getting "on-the-job-
training."
Irony: Taking one's first drink of water from a first hose is ironic be­
cause one cannot expect to get nourishment, especially for the
first time, at the delicate task of drinking water will be forth­
coming with enough pressure to knock a person down. It is
equally ironic for a person to expect to gain job experience by
taking on the enormously important task of managing the
world's most powerful armed forces.

These examples illustrate how irony is a fundamental element of rebuttal

analogies and may be especially effective as persuasive devices by pointing
out the incompatibility of what is expected and what is presented. A recent
set of studies confirmed this claim (Colston, 2000a). Analogies with ironic
bases (e.g., "Doubling the defense budget to intimidate North Korea is like
using a chainsaw to file your nails") were perceived by participants to be
more argumentative, and more attacking than were analogies with absurd
and nonironic bases. These findings support the idea that listeners' recogni­
tion of the ironic structure between base and target is more central to their
understanding of rebuttal analogies than is absurdity comparison or argu­
mentative conventions. Another study showed, however, that speakers who
use rebuttal analogies were unfavorably perceived by listeners (Whaley &
146 GIBBS AND IZETT

Wagner, 2000). Thus, ironic rebuttal analogies may be persuasive in argu­

ments, even if speakers may not always be positively perceived for using
these devices.
Academic writing, although generally seen as containing few instances
of irony and humor, actually contains many examples in which writers ex­
press certain ideas by ironically highlighting the contrast between other
people's beliefs or expectations and some reality. One way this is accom­
plished is through a writer simply quoting someone else's previous work
without comment, with the irony resulting from juxtaposition of the
quoted passage with some other passage in the text. For instance, in one
debate between some linguistics and computer scientists, Dresher and
Hornstein (1976) in a footnote quoted without comment text from
Winograd (1974), one of the people whose work they were attacking, in
their discussion of practical issues in building computational models of
natural language understanding. Dresher and Hornstein wrote,
(1976:330) Thus, one could start with fairly simple components (a small
number of syntactic components, a small lexicon, etc.) which could be im­
proved indefinitely (by adding more syntactic patterns, more lexical items)
according to practical considerations such as time, money and computer
space. The quotation they cited from Winograd (1974, p. 93) was as fol­
lows: If someone is trying to build the best robot which can be completed
by the next year, he will avoid any really hard problems that come up,
rather than accepting them as a challenge to look at a new area. There will
be pressure from the organization of the projects and funding agencies to
get results at the expense of avoiding hard problems.
By quoting him directly, Dresher and Hornstein (1977) used
Winograd's (1974) own words to support their contention about the im­
plausibility of artificial intelligence models given the practical limitations
faced by computer scientists. Again, simply quoting someone else's words
allows writers to convey ironic intentions about the quoted work and au­
thor (s). How people come to process these quoted statement as ironic
must be complex, as readers do not simply establish ironic intentions by
recognizing certain textual features that conventionally mark irony. In­
stead, readers enter texts uncertain and use several possible relations
among the reader, the writer, and another writer to establish several inten­
tions, some of which might be ironic. Readers look for intentions consis­
tent with the new incongruent relations proposed among the reader, the
ironic writer, and the victimized writer. We assume that the quoted writers
believe their original assertions as they are quoted. Irony arises, however,
because we assume that the statement was made with a different intention
in the original text. Scholars see these ironic mentions of other people's
words as an effective device to persuade others of their own beliefs (see
Colston, 2000b).
6. PERSUASIVE COMMUNICATION
147

IRONY IN LITERATURE

People do not usually read literature to be persuaded by writers in the way

they may when reading expository essays. But literature may indeed work to
persuade readers of particular ideas and beliefs through the use of ironic
contrast. An excellent example of this is seen in the work of the 20th-cen-
tury British author Evelyn Waugh (Beaty, 1992). Waugh was a master at us­
ing irony as a way of perceiving life, and as an aesthetic device for reporting
its unresolved paradoxes. For Waugh, irony was the most congenial aes­
thetic means of displaying the existential dilemmas of 20th-century man.
Many of Waugh's earlier novels—Decline and Fall, Vile Bodies, Black Mischief,
A Handful of Dust, and Scoop—illustrate his distinctive style to comment on
the often chaotic environments that his characters inhabit yet allow him as
author to remain somewhat disengaged from other people and events. The
early novels convey the strong sense of Waugh's understanding of the absur­
dities of life that he clearly took great pleasure in writing about. Whether he
was dealing with life in public schools and universities, religion, the shenani­
gans and hypocrisy of the upper class, or British history and politics, Waugh's
purpose was not only to amuse readers but also to persuade them of his own
critical assessment. In fact, some of Waugh's nonfictional essays expressed
his strong belief in the importance of style in the art of persuasion. Irony was
surely one of his greatest tools.
Irony also provided Waugh with the semblance of open mindedness, a
feature that likely attracted readers, and offered him the tools for describing
his very definite perceptions of reality. This detachment is a hallmark of
most ironic literature, with Waugh's light style still coming across as relevant
to very serious matters. We consider just two brief excerpts of Waugh's writ­
ing to illustrate how irony informed his imaginative creations.
Waugh's (1932) Black Mischief was perhaps his finest example of using
irony to invite readers to discover with the narrator the complexities of life's
incongruities. By using language inappropriate in context that highlights in­
congruities between appearance and reality, Waugh indirectly disclosed his
own thoughts but also aimed to elicit similar reactions for a morally respon­
sible audience. An example of this is seen in the one character's tribute to a
slain comrade at his funeral. In this eulogy, the speaker attributes many qual­
ities that were clearly not the case in real life: "Thousands fell by his right
hand. The words of his mouth were like thunder in the hills. Weep, woman
of Azania, for your royal lover is torn from your arms. His virility was inex­
haustible, his progeny numerous beyond human computation .... When he
led you to battle there was no retreating. In council the more guileful, in jus­
tice the most terrible, Seth the magnificent is dead" (p. 299).
Waugh's (1932) aim here was to blatantly mock the dead man's preten­
sions about his effectiveness as a loyal British subject serving the colonial
148 GIBBS AND IZETT

empire in Africa. The speaker cannot be accused of speaking ill of the dead
given the words he uses. But the incongruity of the words with people's rec­
ognition of the dead man's real-life attributes and accomplishments gives
rise to a wonderful ironic feel that is both pleasurable and persuasive.
In The Loved Ones, Waugh (1948) also mocked the pretensions of a char­
acter, Aime, for hyperextensive preparations for a romantic evening:

With a steady hand Aimee fulfilled the prescribed rites of an American

girl preparing to meet her lover—dabbed herself under the arms with a
preparation designed to seal the sweat glands, gargled another to
sweeten the breath, and brushed into her hair some odorous drops
from a bottle labeled "Jungle Venom"—(advertised as coming) "From
the depth of a fever-ridden swamp, with the remorseless stealth of the
hunting cannibal." Thus fully equipped for a domestic evening ... she
was all set to accept her manifest destiny, (p. 1ll)

Waugh's (1948) exaggerated diction highlighted how Aimee's prepara­

tions were ridiculous and that she took herself way too seriously with
thoughts of her romantic fate as "manifest destiny." The exaggeration here
also marks the incongruity with Aimee's self-concept as a highly ethical per­
son yet one that readily uses an arsenal of drugstore products to enable her
own seduction, one that never happens despite her expansive efforts and su­
preme confidence.
Literature contains many examples like these in which authors try to in­
still a set of beliefs or attitudes through the characters' words and deeds.
This kind of persuasion can be quite subtle yet surprisingly effective in se­
ducing readers to think in particular ways while experiencing the pleasures
of narrative worlds.

CONCLUSION

There are many ways to persuade people to think or act in particular ways.
Speakers and writers employ many rhetorical devices in their attempt to get
others to adopt their respective points of view. Our claim is that irony has
special features that make it a highly useful tool or weapon in persuasive
communication. Irony succinctly acts to emphasize the incongruity be­
tween some expectations or beliefs and the reality of a situation. Of course,
speakers and writers can point out incongruous, inconsistent, even hypo­
critical thinking using literal language. But irony often seduces listeners to
think hard about a speaker's message because of the highlighted incongruity
and the emotions that arise from recognizing this disparity between expecta­
tion and reality.
There are clear risks to using irony for persuasion. History offers us many
lessons of the negative consequences of using irony and failing to get one's
6. PERSUASIVE COMMUNICATION 149

message across. The songwriter Randy Newman experienced some of these

perils in his 1982 song "Short People" which mocked short people for all
their deficiencies, with the indirect aim of undercutting people's silly preju­
dices against others. Radio stations refused to play the song, and one state
legislature even passed a resolution against the song. Like Jonathan Swift ex­
perienced in the 17th-century with his classic, A Modest Proposal and the
contemporary singer Eminen with some of his ironic messages on homosex­
uals and others, any person trying to persuade others of some belief using
irony risks failing to be understood. Yet, these perils are often counterbal­
anced by the positive persuasive effects that speakers achieve when trying to
maximize the number of wolf-confederates and minimize the number of
sheep-confederates (Kaufer, 1977). The prominence of irony in many social
influence situations, ranging from personal conversations to political de­
bates, is testimony to the widely held, but not universal, belief that irony is
well worth the risks it sometimes entails.
Part of the reason why irony is a popular persuasive device is that listen­
ers often feel differently about themselves and the speaker as a result of
successfully interpreting ironic messages. First, recognizing incongruity
can lead to a pleasurable feeling of arousal (Berlyne, 1971). This aesthetic
reaction to irony makes it a joy to use in commenting on life's events and in
understanding other's recognition of ironic situations. Simply detecting
the irony in some situation can lead to a positive feeling about one's own
abilities to recognize the complexities in life. But people experience addi­
tional aesthetic and emotional pleasures in knowing that they share ironic
perceptions with others. As is often stated about metaphor (Gibbs, 1994),
irony works to create a sense of intimacy between those who use and those
who understand it. The feeling that one is part of a conspiracy in seeing
irony is a strong reason why it is so frequently observed in persuasive com­
munication. This heightened degree of arousal may also suggest one rea­
son why ironic language is often more memorable than corresponding
literal language (either literal uses of the same words or nonironic para­
phrases; Gibbs, 1986). Not surprisingly, empirical research demonstrates
that college students find ironic ads (reversals plus destabilization) more
artful and cleaver than ads that express simple poetic figures (i.e., repeti­
tion plus substitution; McQuarrie & Mick, 1996).
Our arguments on irony's persuasive effects must be softened by the rec­
ognition of the complexities of ironic language and artwork. Similar to met­
aphor, irony has many linguistic and pragmatic features, each of which need
not leads to enhanced persuasion.
For instance, irony is really a global term for a variety of figurative forms,
such as sarcasm, jocularity, rhetorical questions, understatements, and hyper­
bole. Leggitt and Gibbs (2000) showed these varying types of irony many
evoke different emotional reactions, as well as different beliefs about a
15O GIBBS AND IZETT

speaker's emotional state. Moreover, different forms of irony may work differ­
ently in emphasizing the contrast between expectations and reality (Colston
& O'Brien, 2000a, 2000b). Thus, we resist making any blanket statement
about whether all forms of irony are equally effective in persuasive communi­
cation. We also note that related cousins to irony, such as parody and satire,
make frequent appearances in persuasive discourse and advertisements. Our
suspicion is that these related figurative schemes share similarities to irony in
highlighting discrepancies between expectation and reality. Yet, there may be
important differences too in how this cluster of tropes function to seduce and
persuade. We are enthusiastic about blending ideas from social psychology
with research methods from psycholinguistics to empirically explore more of
the social, interpersonal dimensions of ironic talk.

REFERENCES

Aronson, E., Fried, C., & Stone, J. (1991). Overcoming denial and increasing the
intention to use condoms through the induction of hypocrisy. American]oumal of
Public Health, 81, 1636-1638.
Austin-Smith, B. (1990). Into the beast of irony. Dimension, 27, 51-52.
Beaty, F. (1992). The ironic world of Evelyn Waugh. DeKalb, IL: Northern Illinois Press.
Berlyne, D. (1971). Aesthetics andpsychobiology. New York: Appleton.
Cialdini, R. B. (1975). Reciprocal concessions procedure for inducing compliance:
The door-in-the-face technique. Journal of personality and Social Psychology, 31,
206-215.
Cialdini, R. B. (1978). The lowball procedure for producing compliance: Commit­
ment, then cost. Journal of Personality and Social Psychology, 36, 463–476.
Cialdini, R. B. (1984). Influence: The psychology of persuasion. New York: Morrow.
Cialdini, R. B., & Ascani, K. (1975). The test of concession procedure for inducing
verbal, behavioral and further compliance with a request to give blood. Journal of
Applied Psychology, 61, 296-300.
Colston, H. L. (1997). Salting a wound or sugaringa pill: The pragmatic functions of
ironic criticism. Discourse Processes, 23, 25-45.
Colston, H. L. (2000a). Comprehending speaker intent in rebuttal analogy use: The
role of irony mapping, absurdity comparison and argumentative convention.
Language and Speech, 43, 337-354.
Colston, H. L. (2000b). "Dewey defeats Truman": Interpreting ironic restatement.
Journal of Language and Social Psychology, 19, 44–63.
Colston, H. L. (2000c). On necessary conditions for verbal irony comprehension.
Pragmatics & Cognition, 8, 277-324.
Colston, H., & Gibbs, R. W., Jr. (1998). Analogy and irony: Rebuttal to "Rebuttal
analogy." Metaphor and Symbol, 13, 69-76.
Colston, H., & Keller, S. (1998). You'll never believe this: Irony and hyperbole in
surprise. Journal of Psycholinguistic Research, 27, 49–513.
Colston, H. L., & O'Brien, J. (2000a). Contrast and pragmatics in figurative language:
Anything understatement can do, irony can do better. Journal of Pragmatics, 32,
1557-1583.
6. PERSUASIVE COMMUNICATION 151

Colston, H. L., & O'Brien, J. (2000b). Contrast of kind vs. contrast of magnitude:
The pragmatic accomplishments of irony and hyperbole. Discourse Processes, 30,
179-199.
Dews, S., & Winner, E. (1995). Mutiny to meaning: A social function of irony. Dis­
course Processes, 19, 3-19.
Dresher, E., & Hornstein, N. (1976). On some supposed contributions of artificial
intelligence to the scientific study of language. Cognition, 4, 321-398.
Festinger, L. (1957). A theory of cognitive dissonance. Stanford, CA: Stanford Univer­
sity Press.
Gibbs, R. W., Jr. (1986). On the psycholinguistics of sarcasm. Journal of Experimental
Psychology: General, 115, 3-15.
Gibbs, R. W., Jr. (1994). The poetics of mind: Figurative thought, language, and under­
standing. New York: Cambridge University Press.
Gibbs, R. W., Jr. (2000). Irony in talk among friends. Metaphor and Symbol, 15,5-27.
Kaufer, D. (1977). Irony and rhetorical strategy. Philosophy and Rhetoric, 10, 94-98.
Kumon-Nakamura, S., Glucksberg, S., &. Brown, M. (1995). How about another
piece of pie: The allusional pretense theory of discourse irony. Journal of Experi­
mental Psychology: General, 124, 3-121.
Lee, C., & Katz, A. (1998). The differential role of ridicule in sarcasm and irony.
Metaphor and Symbol, 13, 1-15.
Leggitt, J., & Gibbs, R. W., Jr. (2000). Emotional reactions to verbal irony. Discourse
Processes, 29, 1-24.
McQuarrie, E., & Mick, D. (1996). On resonances: A critical pluralistic inquiry
into advertising rhetoric. Journal of Consumer Research, 19, 180-197.
Petty, R. E., & Cacciopo, J. T. (1986). Communication and persuasion: Central and pe­
ripheral routes to attitude change. New York: Springer-Verlag.
Roberts, R., &Kreuz, R. (1994). Why people use figurative language? Psychological
Science, 5, 159-163.
Stewart, A. (2000, February). Mathews/Mark's anti-crusade. Adweek, 50, 7.
Stone, J., Aronson, E., Grain, A. L., Winslow, M. R, & Fried, C. B. (1994). Inducing
hypocrisy as a means of encouraging young adults to use condoms. Personality' and
Social Psychology Bulletin, 20, 116-126.
Switt, J. (1996). A modest proposal and other satirical works. New York: Dover.
Walker, N. (1990). Feminist alternatives: Irony and fantasy in contemporary novels by
women. Jackson: University of Mississippi Press.
Waugh, E. (1932). Black mischief. Boston: Little.
Waugh, E. (1948). The loved one: An Anglo-American tragedy. Boston: Little.
Waugh, E. (1930). Vile bodies. Boston: Little.
Waugh, E. (1938). Scoop. Boston: Little.
Whaley, B., & Holloway, R. (1996). Rebuttal analogy: A theoretical note. Metaphor
and Symbol 11, 161-167.
Whaley, B., & Wagner, L. (2000). Rebuttal analogy in persuasive messages: Com­
municator likeability and cognitive responses. Journal of Language and Social Psy­
chology, 19,66-84.
Winograd, T. (1974)- Five lectures on artificial intelligence (Memo AIM No. 246).
Stanford, CA: Stanford University, Artificial Intelligence Laboratory.
This page intentionally left blank
 7

Do Men and Women Differ

in Their Use of Nonliteral
Language When They Talk
About Emotions?
Kristen E. Link
State University of New York College at Oswego

Roger J. Kreuz
The University of Memphis

The goal of our research was to determine whether there are gender differ­
ences in nonliteral language use when people communicate about emo­
tions. In Experiments 1 and 2, participants watched characters in film clips
experience emotional events. In Experiment 3, participants read narra­
tives that were analogous to the film clips. Later, they wrote a description
of the character's emotion or a description of how they would have felt in
the same situation. There was no correlation between the perceived inten­
sity of the emotion and the amount of nonliteral language in the partici­
pants' descriptions, which is problematic for Ortony's (1975) vividness
hypothesis. Men used more nonliteral language in descriptions of negative
emotions than positive emotions, whereas no difference was found for
women. Finally, men tended to use more nonliteral language in descrip­
tions of others' emotions, whereas women tended to use more in descrip­
tions of their own emotions. These findings suggest that men and women
do use nonliteral language differently, at least in the context of emotional
communication.
153
154 LINK AND KREUZ

The communication of emotion can be accomplished in a variety of ways.

For example, emotion can be expressed through nonverbal means (e.g.,
Ekman, Friesan, & Ellsworth, 1972; Izard, 1971) and through language
(Rime, Mesquita, Philippot, & Boca, 1991; Shimanoff, 1985a). This re­
search has documented the characteristics of emotional language, such as
which emotions are expressed and to whom they are disclosed. For example,
Rime et al. illustrated that shame is less likely to be discussed than other
emotions, and they noted that emotions are typically expressed to a close
friend, spouse or partner, or family member.
A great deal of this research has investigated the claim that men and
women talk about emotions in different ways. For the most part, this work
has not focused on the role of nonliteral, or figurative, language in emo­
tional communication. The purpose of the research presented in this chap­
ter is to explore whether men and women use nonliteral language differently
when they are asked to describe emotions. We review the relevant research
on gender, emotion, and nonliteral language, and report the results from a
series of experiments that were designed to explore whether gender differ­
ences exist.

GENDER AND EMOTIONAL LANGUAGE

The research on gender and emotional expression through language has not
yet produced a consistent picture. A number of studies have documented
gender effects, but they are difficult to summarize simply due to differences
in methodology and subject populations.
First, results have varied depending on whether direct or indirect
self-report measures were used (LaFrance & Banaji, 1992). With direct
self-report measures, participants rate their subjective emotional states or
expression. For example, Snell, Miller, and Belk (1988) developed the
Emotional Self-Disclosure Scale that measures individual's willingness to
disclose eight different emotions to specific individuals (e.g., female
friend, spouse or partner, etc.). An investigation of gender differences uti­
lizing this scale have illustrated overall gender differences, with women re­
porting a greater willingness to disclose their emotions than men (Snell,
Miller, Belk, Garcia-Falconi, & Hernandez-Sanchez, 1989). However,
Stein and Brodsky (1995) found this difference only when the participants
experienced an unpleasant emotion before they filled out the question­
naire. Rime et al. (1991) also used a direct measure, in which participants
recalled emotional experiences and described how the emotions were ex­
pressed. These authors found no gender differences regarding how often
participants chose to talk about their emotional experiences with others.
These results might differ from those reported by Snell et al. (1989) be­
cause these participants recalled specific experiences, whereas those in
7. USE OF NONLITERAL LANGUAGE 155

Snell et al. (1989) had to make a judgment based on their memories of sev­
eral experiences with a particular emotion. However, both direct measures
discussed thus far required that participants rely on their memories to indi­
cate their emotional expressiveness. Feldman Barrett, Robin,
Peitromonaco, and Eyssell (1998) used a diary technique in which partici­
pants indicated how much they expressed their emotions immediately af­
ter every interaction. In this case, women reported that they expressed
their emotions more often than men. LaFrance and Banaji (also see
Shimanoff, 1985b) suggested that when individuals must make a general
rating of their emotionality based on their memories, they are likely to rely
on the own gender-based stereotypes of emotional expressiveness.
With indirect self-report measures, individuals' verbalizations are coded
for emotion content and scores are derived from this analysis (LaFrance &
Banaji, 1992). For example, in Shimanoff (1983, 1985a), participants
tape-recorded their conversations outside of the laboratory, and segments of
these conversations were analyzed for their emotional content. No overall
gender differences were found in terms of the number of affect words (e.g.,
angry, happy, and sad) used. A similar result was reported by Banaji and
LaFrance (1989; cited in LaFrance & Banaji, 1992). Leaper, Carson, Baker,
Holliday, and Myers (1995) tape-recorded 5-min conversations between
friends in the laboratory. In this case, however, men were found to disclose
significantly more private facts and personal thoughts and feelings than
women. However, the authors did not distinguish emotional disclosures
from other types of disclosures in their analyses. In addition, they did not
consider whether men talked more in the conversations in general, allowing
for a greater number of disclosures. There is the possibility that a higher pro­
portion of the women's conversation included references to emotion.
Fussell and Moss (1998) used a somewhat different indirect procedure.
Film clips were shown to participants who were then asked to describe the
emotions in the clips. Results suggest that men and women used the same
number of words in their descriptions of emotion. The results of the studies
reviewed so far suggest that when participants are asked to estimate how fre­
quently they talk about their emotions in general, greater gender differences
are observed than when more objective measures are used (see LaFrance &
Banaji, 1992).
In addition, investigations of gender differences in talk about emotions
have varied depending on the relationship between the research partici­
pants. Researchers have studied the emotional communication of different
subject populations, and some have focused on how husbands and wives
communicate with one another. Shimanoff (1985b) found that wives re­
ported disclosing significantly more emotions than their husbands. In addi­
tion, wives had a more positive attitude toward disclosing vulnerable,
hostile, and regret emotions than their husbands. However, when she ana­
156 LINK AND KREUZ

lyzed their tape-recorded conversations, Shimanoff found no differences in

the number of affect words husbands and wives used. However, Notarius
and Johnson (1982) found that wives expressed more neutral and negative
emotions than their husbands. Consistent findings were reported by Snell et
al. (1988) using a direct self-report measure, in which men reported less will­
ingness to disclose negative emotions such as depression, anxiety, anger, and
fear to their spouses or partners than women. Other studies have found simi­
lar gender differences when couples interact in conflict situations (e.g.,
Christensen & Heavey, 1990; Kelley et al., 1978; Noller, 1993).
However, studies that did not employ married couples are typically less
consistent in terms of the valence of the emotion. For example, Feldman
Barrett et al. (1998) reported that women expressed their emotions more
than men regardless of the emotion. Similar results were reported by Snell et
al. (1989); women reported more willingness to disclose feelings of depres­
sion, happiness, anger, calmness, and fear. Men and women were equally
willing to disclose jealously, anxiety, and apathy. Somewhat different results
were indicated by Brody and Hall (1993), whose review of several studies re­
vealed that women were more likely to express their emotions than men ex­
cept for anger.
Studies investigating dyads composed of friends or strangers reveal differ­
ent patterns. Specifically, in a meta-analysis, Dindia and Allen (1992) found
greater gender differences in self-disclosure (with women self-disclosing
more than men) when the dyads were in relationships than when they were
strangers. This suggests that several factors are important when considering
the effect of gender on emotional expression. Some of the inconsistencies in
the literature may be clarified if the particular type of language that is used in
self-disclosures is investigated. That is, research on gender and emotional
communication has not specifically focused on the use of nonliteral lan­
guage. Nonliteral language is used to describe emotion, and this issue is ad­
dressed in the next section.

NONLITERAL LANGUAGE AND EMOTION

A consistent finding in the literature is that people find it difficult to express

their emotions via literal language (e.g., Bowers, Metts, & Duncanson,
1985) and that nonliteral language may be used for this purpose (e.g.,
Fainsilber & Ortony, 1987; Gibbs & Nascimento, 1996; Leggitt & Gibbs,
2000; McMullen & Conway, 1996). Examples of nonliteral language in­
clude metaphor (e.g., "His anger was a wild animal"), simile (e.g., "She was
like a kid in a candy store"), idiom (e.g., "He blew his top"), and hyperbole
(e.g., "She was as high as the clouds"). Some of these forms of nonliteral lan­
guage (e.g., metaphor and simile) are often used to express abstract ideas,
like emotions, that are difficult to convey using literal language. In addition,
7. USE OF NONLITERAL LANGUAGE 157

nonliteral language might be more useful to distinguish different intensities

of the same emotion. For example, the simile "She was like a kid in a candy
store" better captures the degree of happiness than the literal expression
"She was really happy." These and other types of nonliteral language have
been explored in the psychological literature (for a review, see Gibbs, 1994).
Ortony (1975, p. 45) suggested that metaphor and simile may be used to
communicate ideas that are difficult to verbalize using literal language (i.e.,
the inexpressibility hypothesis), and they may help communicate one's subjec­
tive experience more vividly (i.e., the vividness hypothesis). To test these
ideas, Fainsilber and Ortony (1987) had participants describe their actions
and feelings for mild and intense experiences of positive and negative emo­
tions. The authors found that more metaphors were used in descriptions of
feelings than in descriptions of actions, supporting the inexpressibility hy­
pothesis. In addition, more metaphors were used in descriptions of intense
emotional experiences than mild ones, supporting the vividness hypothesis.
Anderson and Leaper (1998) provided additional evidence that could sup­
port the inexpressibility hypothesis. In their study, direct and indirect refer­
ences to emotion were coded in tape-recorded conversations between pairs
of friends. Direct references were defined as specific emotion terms (e.g., an­
gry, scared, and happy), whereas indirect references included words or
phrases that "substituted for, or related to, an emotion term," including an­
noyed, bummed, and "other phrases and metaphors that referred to emotion
terms" (p. 428). Indirect references, although not exclusively nonliteral,
were more common than direct references, particularly for negative emo­
tions. It is clear from these studies that nonliteral language is very commonly
employed to describe affective states.

MOTIVATION AND DESIGN

The purpose of our research was to further explore how nonliteral language
is used in the context of emotional expression. Experiment 1 was designed to
test Ortony's (1975) vividness hypothesis, using a different methodology
and procedure to assess the generalizability of this effect. In addition, mate­
rials for later experiments were selected in this experiment. Experiments 2
and 3 were designed to examine whether men and women use nonliteral
language in emotional communication in different ways. We showed that
there are many inconsistencies in the gender and emotional expression liter­
ature, many of which are explained by factors such as the relationship be­
tween the conversation partners and the gender makeup of the dyad.
However, perhaps these inconsistencies also result from not considering the
use of nonliteral language in emotional disclosures. In this study we exam­
ined whether nonliteral language plays a differential role in how men and
women express their emotion through language; the particular emotional
158 LINK AND KREUZ

experience being described and the individual to whom it is described were

held constant.
Experiments on gender differences in emotional expression via nonliteral
language are needed due to limitations in previous research. For example,
Shimanoff (1983, 1985a, 1985b) explicitly removed nonliteral language
from her analysis of emotional communication in conversations. In Ander­
son and Leaper's (1998) analysis of direct and indirect references, nonliteral
language was included, but it was collapsed with other indirect, but literal
forms. Other researchers who have examined the role of nonliteral language
in emotional expression have looked only at metaphor (Fainsilber &
Ortony, 1987; Gibbs & Nascimento, 1996; Williams-Whitney, Mio, &
Whitney, 1992). In addition, researchers who have examined nonliteral
language more generally have not included analyses of different types of
nonliteral expressions (Fussell & Moss, 1998). The possibility that certain
forms of nonliteral language are used more often than others to express emo­
tion has not been examined.
Because gender effects may be manifested in different ways, we chose to
manipulate both perspective and valence in Experiments 2 and 3. Perspec­
tive refers to whether a person is reporting his or her emotional experience or
the experience of someone else. Perspective was manipulated by Williams ­
Whitney et al. (1992) and was shown to affect the amount of nonliteral lan­
guage that participants used. Specifically, nonexpert writers used more
nonliteral language when describing their own emotions than others' emo­
tions. Shimanoff (1983) also examined the role of perspective and found
that it interacted with gender: Men talked more about their own emotion
than others' emotions, whereas there was no difference for women.
Valence refers to whether an emotion is positive or negative. It, too, has
been shown to affect emotional communication. Fussell and Moss (1998)
found that participants used more nonliteral language in their descriptions
of negative emotions, and Anderson and Leaper (1998) found that negative
emotions were particularly likely to be referenced indirectly. Other re­
searchers (Notarius & Johnson, 1982; Shimanoff, 1983, 1985b) have found
that women expressed more negative emotions than men, but only when
the participants were married couples.
Finally, participants' verbal ability was measured in Experiments 2 and 3.
This individual difference has been used as a control in some previous work
on nonliteral language (e.g., Trick & Katz, 1986). We controlled for verbal
ability for two reasons. First, individuals with high verbal ability may be more
likely to use nonliteral language than individuals with low verbal ability. In
addition, there is a possible confound with gender. For example, if women
used more nonliteral language in their descriptions, then it might be a true
gender effect or simply a reflection of higher verbal ability, an effect that has
been repeatedly demonstrated (e.g., Hyde & Linn, 1988).
7. USE OF NONLITERAL LANGUAGE 159

We chose to use film clips to elicit descriptions of emotions in Experi­

merits 1 and 2. This is similar to the approach used by Fussell and Moss
(1998) who argued that these more objective stimuli avoid the variability in­
herent in autobiographical experiences of emotions. We also chose to elicit
descriptions of emotions using narrative descriptions of the film clips in Ex­
periment 3. Text-based stimuli have been used previously to elicit nonliteral
language. Gibbs and Nascimento (1996, Experiment 4) used love poetry to
elicit conceptual metaphors, and Williams-Whitney et al. (1992) employed
short scenarios to elicit descriptions of emotions.
It may be that individuals process films and texts in different ways. On the
one hand, films can provide a very rich sensory experience. On the other
hand, texts can provide an equally rich experience if the reader creates a
fleshed-out situation model (Kintsch, 1998). These differences between
films and texts may affect how people process them. Evidence for such dif­
ferences has been reported by Pezdek and her colleagues (Pezdek, Lehrer, &
Simon, 1984; Pezdek, Simon, Stoeckert, &Kiely, 1987). They reported that
the comprehension processes involved in watching television differ from
those involved in reading narrative texts. We hypothesize that these pro­
cessing differences may influence the language people use to describe these
experiences. Therefore, both film clips and narratives were utilized in our re­
search to assess the existence of any gender effects.
It is difficult to derive predictions about men's and women's use of
nonliteral language in their expression of emotion because of the lack of em­
pirical research on the topic. However, two opposing hypotheses can be gen­
erated. First, because women experience emotions more intensely (e.g.,
Brody & Hall, 1993) and descriptions of intense emotions are more likely to
contain nonliteral language (Fainsilber & Ortony, 1987), one might con­
clude that women's linguistic expressions of emotion will contain more non-
literal language. Second, because there is a sex-role expectation for men to
control their emotional expression (except for anger; see Brody & Hall,
1993, for a review), they may be more likely to use indirect methods of ex­
pression, including nonliteral language. To the degree that the discomfort
involved with men's expression of emotion is face threatening, the use of an
off-record speech act might be warranted. Brown and Levinson (1987) de­
scribed several off-record strategies that employ nonliteral language (e.g.,
metaphor and hyperbole). Men may resort to using nonliteral language in
such cases, particularly in descriptions of their own emotions.

EXPERIMENT 1

This experiment was designed to test Ortony's (1975) vividness hypothesis

and to select materials for later experiments. Thirty-two undergraduate in­
troductory psychology students from the University of Memphis partici­
16O LINK AND KREUZ

pated for course credit. Seventeen excerpts from 17 different films were
used. The films were contemporary English-language productions (e.g.,
Steel Magnolias, Stark & Ross, 1998, and The Shawshank Redemption,
Glotzer, Lester, Marvin, & Dourabont, 1994) and were chosen from a vari­
ety of genres. The films were chosen based on suggestions made by col­
leagues of the authors and the authors themselves, who recalled particularly
emotional scenes in movies. The length of the films clips ranged from 35 s to
4 min (M = 2 min 16 s, SD = 55 s). The clips contained characters experi­
encing a variety of emotions. The length of each clip and the emotion repre­
sented by each clip appears in Appendix A.
Participants watched the film clips in small groups. After each clip was pre­
sented, they provided a written description of the emotion that was being ex­
perienced by the specified character in the clip. Participants then wrote down
the name of the emotion that they perceived the character was experiencing.
Next, they rated, on a 6-point Likert scale, the intensity of the emotion and
the genuineness of the emotional display. Finally, participants indicated
whether they had previously seen the film from which the clip was drawn.

Test of the Vividness Hypothesis

For the purposes of determining reliability, the participants' descriptions

were divided into sentences. In most cases, sentences were defined by termi­
nal punctuation (i.e., periods, question marks, and exclamation points).
However, when participants did not use proper punctuation, their contribu­
tions were divided into sentence units by two judges based on the following
criteria: (a) a topic shift and/or (b) a capitalized first word in the absence of
punctuation. In these cases, divisions were made conservatively rather than
marking all possible sentences.
We identified the metaphors and similes in participants' descriptions to
evaluate Ortony's (1975) vividness hypothesis, which predicts that meta­
phors and similes should be used more frequently to describe emotionally in­
tense experiences. Two independent judges identified the metaphors and
similes in each sentence. Nonliteral expressions that were repetitions of
characters' statements were not coded. The judges agreed 86% of the time
on the number of nonliteral expressions in each sentence. The correlation
between the coders was 0.53, indicating a modest level of agreement
(Bakeman & Gottman, 1986). However, the judges discussed all disagree­
ments until 100% agreement was reached, and only those expressions that
were agreed upon as being nonliteral were included in the analyses. A total
of 234 metaphors and 58 similes were identified. The mean number of meta­
phors and similes per 100 words was 1.75 (SD = 2.96).
The number of metaphors and similes per 100 words used by each partici­
pant for each clip was calculated. This metric of nonliteral language use was
7. USE OF NONLITERAL LANGUAGE 161

chosen to be used in this and further analyses because it controlled for the
number of words in participants' descriptions, and it was deemed meaning­
ful and easily interpretable. The vividness hypothesis predicts a positive cor­
relation between the proportion of metaphors and similes produced and the
perceived intensity of each clip, irrespective of the valence of the emotion.
Such a relationship was not supported by these data, r(544) = -0.06, ns.
The film clips used in this experiment contained emotional experiences
that varied in perceived intensity. However, essentially no relationship was
found between the intensity of the emotion and the amount of nonliteral
language used to describe it. This lack of correlation is probably not due to a
restriction of range because participants used the entire intensity scale (M
= 4.62, SD = 1.35) and varied greatly in the number of metaphors and simi­
les they employed. Therefore, a strong form of the vividness hypothesis,
which predicts more nonliteral language with increasing emotional inten­
sity, was not supported.

Selection of Materials for Experiment 2

Another goal of this experiment was to select 8 film clips (4 positive and 4
negative) from this set of 17 to be used in Experiment 2 by measuring the
nonliteral language produced by the participants and examining the ratings
of the clips.
The nonliteral expressions in participants' descriptions were identified by
one judge using a taxonomy developed by Kreuz, Roberts, Johnson, and
Bertus (1996). Eight forms of nonliteral language were identified: hyper­
bole, idiom, indirect request, irony, metaphor, rhetorical question, simile,
and understatement. Examples and definitions of these eight forms appear
in Table 7.1. The mean number of nonliteral expressions per 100 words was
1.72 (SD = 2.89). Mean genuineness (M = 4.84, SD = 1.20) and intensity
(M = 4.62, SD = 1.35) ratings were also calculated across all film clips and
participants.
To determine whether participants interpreted the emotional experiences
of the characters in the clips uniformly, the most common emotion label (e.g.,
happiness) for each clip was determined. The percentage of participants who
provided this label was averaged across the clips (M = 63.53, SD = 19.90).
Because two thirds of the participants agreed on each label, it seems likely that
the emotions in these clips were highly interpretable.
Previous exposure to the films may be a confound in the data provided by
the participants. It may be that individuals who have seen the films can draw
on a richer context for their descriptions than subjects who saw only clips
from these films. For example, genuineness ratings might be inflated for par­
ticipants who had seen the entire film from which a clip was drawn. There­
fore, we conducted analyses to determine whether this problem exists.
162 LINK AND KREUZ

TABLE 7.1
Examples and Definitions of Nonliteral Language Types

Figure Definition Example

Hyperbole Deliberate overemphasis "It takes all of his strength to
write his letter."
Idiom Conventionalized expression "He doesn't think sorry will cut
it"
Indirect request Requests stated as questions Do you have the time?.
about ability
Irony Opposite meaning expressed Thanks for all your help!
Metaphor Implicit comparison "Through all of it I would be
empty."
Rhetorical question Assertions framed as questions "What did I do wrong?"
Simile Explicit comparison "I would feel like my heart will
just jump out of my chest ..."
Understatement Deliberate underemphasis "He clearly has issues." [Written
about a suicidal character]

Note. Examples within quotations were taken from Experiments 2 and 3. Nonliteral expressions are
italicized.

A 17 (film) X 2 (seen previously vs. not seen) analysis of variance

(ANOVA) was performed for each dependent measure, with the seen ver­
sus not seen variable nested within each film. Therefore, a subject might
have been in the seen condition for one film but in the not seen condition for
another film. Participants who had previously seen a film rated the clip from
that film as being more intense (M = 4.86, SD = 1.31) than participants
who had not seen the film (M = 4.56, SD = 1.36), and the difference was
significant, F(l, 507) = 5.97,£> < .05.1 Participants who had previously seen
a film also rated the clip from that film as being more genuine (M = 5.20, SD
= 1.05) than participants who had not (M = 4.65, SD = 1.27); once again,
the difference was significant, F(l, 507) = 18.99, p < .001. A lack of an in­
teraction in both analyses (Fs < 1) indicates that films clips rated as more in­
tense and genuine by participants who had seen the films were also rated as
more intense and genuine by participants who had not seen the films.
Separate mean nonliteral language scores were calculated for partici­
pants who had seen the films (M = 1.72, SD = 2.91) and for participants
who had not seen the films (M = 1.73, SD = 2.89). A 1 7 x 2 ANOVA re­
vealed no main effect of having previously seen the films, F(l, 507) = 1.28,

Large degrees of freedom resulted from subjects being nested within each film.
7. USE OFNONLITERAL LANGUAGE 163

ns, and no interaction, F(16, 507) = 1.16, ns. These results indicate that
whether participants had previously seen the films did not affect the amount
of nonliteral language that they produced.
Mean genuineness and intensity ratings, as well as amount of nonliteral
language, were calculated for each clip. Because the two ratings were rela'
tively high and correlated, r(l7) =0.77, the selection of clips for later exper­
iments was based on two other criteria: the amount of nonliteral language
the clips elicited and the amount of nonliteral language within the clips
themselves. Clips that contained a significant amount of nonliteral lan­
guage were eliminated. We then selected the four positive emotion clips that
had elicited the most nonliteral language from the participants so that they
would be likely to elicit nonliteral expressions in Experiment 2. The emo­
tions depicted in these clips were happiness and love. In a similar way, four
negative emotion clips, depicting sadness and anxiety, were chosen.

EXPERIMENT 2

We addressed the question of whether men and women differ in their use of
nonliteral language when describing positive and negative emotions from
different perspectives in Experiment 2. Participants were presented with de­
pictions of emotional experiences in film clips selected from Experiment 1.
They were asked to describe the emotions as experienced by the character or
as the participants would have experienced them. These descriptions were
analyzed to determine the effects of gender, valence, and perspective on the
use of nonliteral language in descriptions of emotion.
Participants were 20 male and 20 female undergraduate students from The
University of Memphis taking an introductory psychology course and partici­
pating for course credit. The eight film clips selected from Experiment 1 were
used. The clips ranged in length from 1 min 21 s to 4 min (M = 2 min 52s, SD
= 53 s). (See Appendix A for additional information.) Four different random
orders of the film clips were recorded onto videotape. A practice clip (i.e., The
Shining, Fryer et al, 1980) was placed at the beginning of each tape.
Participants were tested in small groups. They first received the Author
Recognition Test (ART; Stanovich &West, 1989), which is a measure of ex­
posure to print. The ART has been shown to correlate significantly with
measures of verbal ability (Stanovich & Cunningham, 1992). For example,
the correlations between the ART and two measures of verbal ability—the
Nelson-Denny vocabulary subtest and a verbal fluency task in which partic­
ipants wrote down as many words as they could in a given category—were
0.60 and 0.40, respectively. We used the ART to control for individual and
possible gender differences in verbal ability.
Participants viewed the practice clip and wrote a description of how the
specified character was feeling. The experimenter checked each partici­
164 LINK AND KREUZ

pant's description to ensure that they were following directions, and not
simply describing the depicted events. The participants then watched the
eight film clips. After each clip, participants were given 5 min to (a) write a
description of the emotional experience of the character in the film clip (the
other condition) or (b) imagine themselves in the same situation, and write
a description of how they would feel (the self condition). Across the four
tapes, each clip appeared equally often in each condition. Each participant
responded with a self description for two positive and two negative clips and
with an other description for two positive and two negative clips.
The design was a 2 X 2 X 2 mixed design with gender (male vs. female) as
the between-subject factor and valence of the emotion (positive vs. nega­
tive) and perspective (self vs. other) as the two within-subjects factors.
The mean ART score was 8.95 (SD = 6.52).Men's (M = 9.65, SD = 7.41)
and women's (M = 8.25, SD = 5.58) scores on the ART did not differ signifi­
cantly, t(38) = 0.68, ns. However, ART scores were included as a covariate in
later analyses to control for individual differences in verbal ability.
As in Experiment 1, eight forms of nonliteral language were coded by two
independent judges. The judges agreed with each other 72% of the time on
the number of nonliteral expressions in each sentence. The correlation be­
tween the judges was modest (r = 0.57). However, the judges discussed all
disagreements until 99.7% agreement was reached. Only two instances of
disagreement were left unresolved, and these were not included in the anal­
yses. A total of 899 nonliteral expressions were identified.
Participants indicated whether they had previously seen the films so that
we could assess whether this factor affected their written descriptions. The
amount of nonliteral language produced in each description was calculated
as in Experiment 1 (i.e., mean number of nonliteral expressions per 100
words). Separate mean nonliteral language scores were calculated for par­
ticipants who had seen the films (M = 3.69, SD = 4.40) and for participants
who had not seen the films (M = 3.82, SD = 3.63). An 8 (film) X 2 (seen
previously vs. not seen) AN OVA revealed no significant main effect of prior
viewing, F( 1,303) < 1, and no significant interaction, F (7,303) = 1.49,ns.
A similar analysis was performed to determine whether having previously
seen the films influenced the number of words participants used in their de­
scriptions. The analysis revealed a significant main effect of having previ­
ously seen the films, F(l, 302) = 15.32, p < .001. That is, participants who
had seen the films used significantly more words (M = 82.25, SD = 45.86)
in their descriptions than participants who had not seen the films (M =
67.83, SD = 29.57). However, upon inspection of the data, an outlier was
identified who had seen seven of the eight films and whose shortest descrip­
tion (232 words) was 101 words longer than the longest description of any
other participant. In addition, this participant used a mean of 264.13 words
(SD = 20.99) per description, compared to the remaining participants'
7. USE OF NONLITERAL LANGUAGE 165

mean of 69.88 (SD = 24.27). When the outlier was removed from the analy­
sis, the mean number of words used by participants who saw the films was
72.40 (SD = 22.51), and the mean number of words used by participants
who did not see the films was 67.56 (SD = 25.62). This difference was not
significant, F(l, 294) — 2.57, ns. Therefore, whether participants previously
saw the film was not included in the remaining analyses in which repeated
measures analyses could be performed (and each participant could act as his
or her own control).
The amount of nonliteral language produced was analyzed with a 2 (gen­
der) X 2 (valence) X 2 (perspective) mixed-design analysis of covariance
(ANCOVA), with ART scores as the covariate. The analysis was performed
with both subjects (F1) and items (i.e., the film clips; F2) as random factors
(Clark, 1973). There was a significant gender X valence interaction, F 1 (l,
37) = 7.66, p < .01 (see Fig. 7.1), which was marginally significant in the
item analysis, F 2 (l, 6) = 5.39, p = .059. Follow-up tests performed by sub­
jects indicated that men used significantly more nonliteral language when
describing negative emotions (M = 4.81, SD = 2.40) than positive emo­
tions (M = 2.60, SD = 2.19),F^l,38) = 25.09,p < .001. Women, however,
did not differ in the amount of nonliteral language used in descriptions of
negative (M = 4.02, SD = 2.59) and positive (M = 3.61, SD = 3.20) emo­
tions, F 1 (l, 38) < 1. Men and women did not differ significantly in the
amount of nonliteral language used to describe either positive, F(l, 37) =
1.03, ns or negative emotions, F(l, 37) < 1.
The number of words used in each participant's description was analyzed
with a 2 (gender) X 2 (valence) X 2 (perspective) mixed-design ANCOVA,
with ART scores as the covariate. There were no significant effects in the sub­
ject analysis. Note that men and women did not differ in the number of words
they used to describe emotions. This is consistent with many of the studies re­
viewed by James and Drakich (1993). However, there was a main effect of
gender in the item analysis, F 2 (l, 6) = 148.83, p < .001, with women using
more words (M = 82.89, SD = 47.00) in their descriptions than men (M =
66.59, SD = 17.59). This may be due to the fact that the ART scores could
not be used as a covariate in the item analysis because the items (i.e., the film
clips) were the cases rather than the participants. Therefore, individual differ­
ences in verbal ability were not controlled. For example, the outlier discussed
earlier may be driving this effect; she wrote a large amount for each clip, which
inflated the mean for the women, but we could not control for her verbal abil­
ity in the item analysis, which also may have been higher than average. What­
ever the reason for the effect, this finding was only significant in the item
analysis and, therefore, may not generalize to other participant populations.
Experiment 2 revealed that men are more likely to use nonliteral lan­
guage when describing negative than positive emotional experiences. This
is consistent with the idea that men are less comfortable when they describe
166 LINK AND KREUZ

FIG. 7.1 Mean number of nonliteral expressions per 100 words for men and women
when describing positive and negative emotions. Error bars are standard errors.

negative emotions (e.g., Shimanoff, 1985b). As a result, men may resort to

using nonliteral language in such cases. Our findings support the notion that
nonliteral language is used in descriptions of emotion as a means of being in­
direct.

EXPERIMENT 3

Given that films differ from other types of aesthetic experience, it is worth
exploring the nonliteral language elicited by narrative descriptions of emo­
tional experiences. As mentioned earlier, the processing of films and texts
may lead to different representations, and it may cause people to talk about
them in different ways. Experiment 3 was designed to explore whether text
versions of the film clips would elicit descriptions in the same way or
whether differences between text and film would lead to a different pattern
of results.
Participants were 20 male and 20 female undergraduates from the Uni­
versity of Memphis taking an introductory psychology course and partici­
pating for course credit. A narrative was created for each film clip in
Experiment 2. The narratives, containing verbatim dialogue, actions, and
descriptions of the scene, served as stimuli. The narratives ranged from 335
to 845 words (M = 582.63, SD = 200.59). A pilot study was conducted to
ensure that the narratives closely corresponded to the film clips.
Twenty-four individuals from the same population that participated in the
experiments rated how well the narratives represented the film clips. On a
7. USE OF NONLITERAL LANGUAGE 167

six-point scale ranging from 1 (low match) to 6 (perfect match), the mean cor­
respondence rating was 5.48 (SD = 0.77). A sample narrative appears in
Appendix B. Four different random orders of the narratives were prepared,
and a practice narrative began each set.
Participants were tested in small groups. They first received the ART. Par­
ticipants then read the practice narrative and wrote a description of how the
specified character was feeling. The experimenter checked each partici-
pant's description to ensure that they were following directions. The partici­
pants then read the eight narratives. They were given 1 min, 30 sec to read
each page of the narratives. After each narrative, participants were given 5
min to write a description in either the self or other condition. Across the
four different random orders, each narrative appeared equally often in each
condition such that each participant responded with a self description for
two positive and two negative clips and with an other description for two
positive and two negative clips.
The mean ART score was 9.28 (SD = 5.19). As in Experiment 2, men's
(M = 9.55, SD = 5.72) and women's (M = 9.00, SD = 4.75) scores on the
ART did not differ significantly, t(38) = 0.331, ns.
The eight types of nonliteral language were coded by the same judges as
in Experiments 1 and 2. The judges agreed with each other 82% of the time
on the number of nonliteral expressions in each sentence. The correlation
between the judges was 0.68. As before, the judges discussed all disagree­
ments until 100% agreement was reached. A total of 595 nonliteral expres­
sions were identified.
The same procedure as in Experiment 2 was used to analyze the effect of
having previously seen the films on which the narratives were based. The
mean nonliteral language score and mean number of words used for partici­
pants who had seen the films (M = 2.60, SD = 2.53; M = 67.71, SD = 23.9;
respectively) and for participants who had not seen the films (M = 3.00,
SD = 3.01; M = 68.94, SD = 27.80; respectively) were compared. Separate
8 (film) X 2 (previously seen vs. not seen) ANOVAs revealed no main ef­
fects, Fs < 1, and no interactions, Fs < 1. Therefore, whether participants
previously saw the film was not included in the analyses.
The amount of nonliteral language was calculated and analyzed as in Ex­
periment 2. There was a marginally significant gender X perspective inter­
action, Fj(l, 37) = 4.00, p = .053 (see Fig. 7.2). This interaction was not
significant in the item analysis, F 2 (l, 6) =2.77, ns. Follow-up tests produced
no significant simple effects. However, it appears that, although men tend
to use more nonliteral language to describe others' emotions (M = 2.95,
SD = 1.78) than their own emotions (M = 2.66, SD = 1.65), women tend
to use more nonliteral language to describe their own emotions (M — 2.87,
SD = 1.37) than others' emotions (M = 2.39, SD = 1.68). In addition, an
item analysis revealed a significant main effect of valence, F 2 (l, 6) = 9.18,
168 LINK AND KREUZ

FIG. 7.2 Mean number of nonliteral expressions per 100 words for men and women
when describing their own and others' emotions. Error bars are standard errors.

p < .05 in which more nonliteral language was used in descriptions of neg­
ative emotions (M = 3.21, SD = 2.02) than positive emotions (M= 2.22,
SD = 1.54). This finding is consistent with that of Fussell and Moss (1998).
The number of words used in each participant's description was calculated
and analyzed with a 2 (gender) X 2 (valence) X 2 (perspective) mixed- design
ANCOVA, with ART scores as the covariate. There were no significant
effects in the subject analysis. However, the ART proved to be a significant
predictor of the number of words participants used in their descriptions,
F(l, 37) = 5.02, p < .05. The number of words used in descriptions of emo­
tions increased as ART scores increased. There was a marginally signifi­
cant gender X valence interaction in the item analysis, F 2 (l, 6) = 5.67, p
= .055; women used more words in their descriptions of negative
emotions (M = 70.94, SD = 28.07) than men (M = 65.34, SD = 21.35),
whereas the number of words used for describing positive emotions did not
differ (Ms = 68.48 and 68.65, SDs = 27.78 and 26.25, respectively).
In participants' descriptions from both Experiments 2 and 3, six of the
eight types of nonliteral expressions shown in Table 7.1 were identified:
metaphor, hyperbole, idiom, simile, rhetorical question and understate­
ment. The mean number of nonliteral expressions per 100 words was 3.76
(SD = 4.01) in Experiment 2 and 2.72 (SD = 2.69) in Experiment 3, and this
difference is significant, F(l, 75) = 5.31, p < .05. This effect supports the idea
that the film clips and narratives were processed differently, resulting in dif­
ferential amounts of nonliteral language in the participants' descriptions.
7. USE OF NONLITERAL LANGUAGE 169

The data from these two studies were collapsed to allow the relative
frequencies of the smaller nonliteral language categories to be assessed.
These frequencies appear in Fig. 7.3. A 5 (type of nonliteral expression)
X 2 (gender) repeated measures ANCOVA with ART scores as the
covariate was performed (the two instances of understatement were not
included in the analysis). The analysis revealed significant differences
among the nonliteral expressions in their rate of use, F(2.1, 163.9) =
12.43, p < .001.2 Neither the main effect of gender nor the interaction
was significant, Fs < 1.
Post hoc pairwise comparisons revealed that metaphors were used sig­
nificantly more often than any other type of expression. The frequency of
use for hyperbole, similes, and idioms did not differ, but each was used sig­
nificantly more often than rhetorical questions. The relative frequencies
of each type of nonliteral expression are similar to those found in another
area of study. Specifically, Kreuz et al. (1996) found that, in a literary cor­
pus, metaphor was used most frequently, followed by hyperbole, idiom,
rhetorical questions, simile, irony, understatement, and indirect requests.
There are, therefore, similarities in the use of nonliteral language in two
very different domains.

Type of Nonliterai Expression

FIG. 7.3 Mean number of each type of nonliteral expressions per 100 words in
Experiments 2 and 3. Error bars are standard errors.

"Degrees of freedom were adjusted using the Greenhouse-Geisser epsilon because the as­
sumption of sphericity was not met.
1 70 LINK AND KREUZ

DISCUSSION

The results from these experiments provide some evidence about how in­
dividuals use nonliteral language to describe emotions. One result that is
surprising is the test of Ortony's (1975) vividness hypothesis in Experi­
ment 1. We found essentially no relationship between the intensity of the
emotion and the amount of nonliteral language used. However,
Fainsilber and Ortony (1987) found support for the vividness hypothesis
using autobiographical emotional experiences in which participants
identified mild and intense experiences of a specific emotion and de­
scribed them.
The results of the current study may differ from Fainsilber and Ortony's
(1987) work because of differences in methodologies. Perhaps the effect of
intensity on nonliteral language use only occurs in autobiographical emo­
tional experiences. That is, in our experiment, participants described how
another person was feeling, whereas in Fainsilber and Ortony's study, partic­
ipants described their own feelings. One may argue, however, that more ob­
jective stimuli were used in Experiment 1 because the descriptions did not
rely on idiosyncratic autobiographical memories (Fussell & Moss, 1998),
Therefore, variations in intensity were more tightly controlled in our experi­
ment, in that we did not rely on participant's intuitions of what constitutes a
mild or an intense emotional experience.
Note that the amount of nonliteral language used by the participants was
relatively high. In Experiments 2 and 3, the participants produced a figure of
speech once every 31 words. This compares to a rate of one figure of speech
per 37 words in a corpus of contemporary short stories (Kreuz et al, 1996).
In other words, the college students used more nonliteral language than
published authors. This counterintuitive finding can be explained by the
task: The participants were specifically describing emotions, whereas au­
thors must tell a complete story.
More important, our results shed light on several questions concerning
nonliteral language, gender, and emotion. First, they illustrate that men
and women, in general, do not use different amounts of nonliteral lan­
guage, and men and women do not differ in their preference for a certain
type of nonliteral expression. In addition, men and women did not differ in
the number of words they used in their descriptions of emotion. This could
be viewed as evidence against gender differences in emotional communi­
cation, at least when descriptions of emotions are obligatory. This result is
consistent with Anderson and Leaper (1998), who suggested that their
lack of gender differences in emotional expression might have been due to
task demands. Specifically, "when people are placed in a situation where
self-disclosure is expected, women and men self-disclose similarly" (p.
423). However, according to other researchers, if men tend to shun discus­
7. USE OF NONLITERAL LANGUAGE 171

sions of emotion (e.g., Gallois, 1993), they should say as little as possible
even when obliged to describe emotions.
This lack of gender differences, however, must be qualified by the two
interactions involving gender. Specifically, the results from Experiment 2
revealed that men produced more nonliteral language when they de­
scribed negative emotions than when they described positive emotions.
Women, however, used the same amount of nonliteral language when they
described positive and negative emotions (see Fig. 7.1). This finding was
obtained even when controlling for differences in verbal ability. Some of
the previous studies that examined only literal language found that men
and women (specifically, unmarried couples) in a natural setting talk
about positive and negative emotions with equal frequency (e.g.,
Shimanoff, 1983), whereas others indicate that women are more likely to
disclose negative emotions than men (e.g., Shimanoff, 1985b; Snell et al.,
1989). Our results support those of the first group of researchers: Men and
women did not differ in the number of words they used to describe positive
and negative emotions. However, the analysis of nonliteral language re­
vealed a different pattern; men used more figurative language in their de­
scriptions of negative emotion, whereas this difference was not significant
for women. It is possible that men's negative attitude toward disclosing
negative emotions (Shimanoff, 1985b) makes them prefer to be less direct
and, hence, use more nonliteral language in these cases. Alternatively, this
difference in nonliteral language use might explain why some researchers
find that women are more likely to disclose negative emotion. Because this
research has typically disregarded nonliteral expressions, much of what a
man says about negative emotions is disregarded. It may be, then, that men
and women are more similar in the amount they talk about positive and
negative emotions (as was found in this study with respect to the number
of words used to describe emotions), but the difference lies in how they
talk about these emotions.
Experiment 3 also revealed a marginally significant gender X perspective
interaction. Men tended to use more nonliteral language when they de­
scribed others' emotions while women tended to use more nonliteral lan­
guage when they described their own emotions. With a similar population,
Williams-Whitney et al. (1992) found that participants used more
nonliteral language in descriptions of their own emotions than another per-
son's emotions. This pattern was found only for the women in Experiment 3.
This difference for men may be due to the specific emotions used in the two
studies. Specifically, Williams-Whitney et al. used the emotions pride and
shame, whereas Experiment 3 used emotions (e.g., happiness and sadness)
that may have been perceived differently. This is particularly important
given gender differences reported by Brody and Hall (1993) in which men
were more likely than women to express outwardly directed emotions, such
1 72 LINK AND KRELJZ

as pride, whereas women were more likely to express more inwardly directed
emotions, such as shame. The effect of the particular emotion needs to be
investigated more thoroughly in future research. However, a possible expla­
nation for our finding is that women's tendency to use more nonliteral lan­
guage in descriptions of their own emotions reflects their tendency to
experience emotions more intensely.
Shimanoff (1983) found that men talked more about their own emotions
in naturally occurring conversations. As previously noted, she did not ana­
lyze nonliteral language. Therefore, it could be that more of what men said
about others' emotions was eliminated from her analysis than what women
said about others' emotions. Therefore, a similar conclusion might be
drawn: Men and women do not differ in the amount of talk about their own
and others' emotions but in how they talk about their own and others' emo­
tions. Specifically, men use more indirect, nonliteral forms of speech when
they talk about others' emotions. Perhaps because of societal expectations,
they do not feel like experts in the area of emotion, resulting in the tendency
to talk indirectly about how others feel.
Therefore, both interactions, along with data from other experiments
that show gender differences, seem to indicate that it is not the amount of
emotional communication that differs, but the way in which men and
women describe emotion. Perhaps the differences between men and
women in the amount of emotional communication appear because indi­
rect references to emotion have been disregarded in analyses of emotional
communication, and these differences would disappear if both literal and
nonliteral expressions were included. Of course, this conjecture needs to
be tested empirically.
Finally, although the gender X valence and gender X perspective inter­
actions were not found in both experiments, it is important to remember
that the stimuli used in Experiments 2 and 3 were not the same. Specifically,
participants in Experiment 2 saw and heard the actors experiencing emo­
tions, whereas those in Experiment 3 simply read about these characters.
Previous research has documented differences in how spoken and written
language are processed (Ferreira & Anes, 1994; Pezdek et al., 1984; Pezdek
et al., 1987). Perhaps such differences affected the participants' task of de­
scribing emotions. In addition, there are four cinematic devices that are
available to film makers, which are not available to writers, that aid in con­
veying the events in a story (Magliano, Dijkstra, & Zwaan, 1996): mise en
scene (i.e., setting design, costume and make-up, lighting, and direction of
actors), montage (i.e., shots edited together), cinematography (i.e., aspects
of filming including film stock, camera speed, and lens) and framing (i.e.,
placement and focus of camera); and sound (i.e., dialogue, external noises,
and music). These devices may function to more readily portray emotion
than a transcription of a film. Specifically, these devices include the use of
7. USE OF NONLITERAL LANGUAGE 173

music, facial expressions, and prosodic and paralinguistic cues that are not
easily translated into the written word. Bourg (1996) stated that "in order to
interpret the emotional states of others, one must first engage in the cogni­
tive processes of perceiving and discriminating cues" (p. 243). Therefore,
because these additional cues are present in the film clips, the task of infer­
ring the characters' emotions and how intensely the character is experienc­
ing the emotion becomes easier. For example, the film clips may have
provided a richer sensory experience, and the intensity of the emotions may
have come across more clearly, resulting in a greater reliance on nonliteral
language use to describe these emotions. Evidence for this suggestion can be
found in the differential amounts of nonliteral language produced in the two
experiments. Participants who saw the film clips generated 28% more
nonliteral expressions than participants who read the narratives.
These experiments provide some support for the existence of gender dif­
ferences in emotional communication and the role that nonliteral language
plays. Researchers need to consider linguistic emotional expression not only
in terms of how frequently words describing affect are used, but also in terms
of the nonliteral language that is used to describe emotions. It has been illus­
trated, both in our research and elsewhere (Fainsilber & Ortony, 1987;
Gibbs & Nascimento, 1996; Ortony, 1975; Williams-Whitney etal, 1992),
that nonliteral language is an important part of expressing one's emotions.
The larger question of whether there are gender differences in the use of
nonliteral language across different discourse contexts remains to be inves­
tigated. In the specific realm of describing emotional experiences, however,
evidence has been provided that men and women do use nonliteral lan­
guage differently.
Roberts and Kreuz (1994) suggested that different types of nonliteral ex­
pressions can be used to fulfill different discourse goals. One might expect
that gender differences exist depending on both the discourse context and
the discourse goals of the speaker. Men and women may have different dis­
course goals in the same discourse context. Specifically, men's use of
nonliteral language may have depended on whether they wished to be more
or less expressive about the emotion that they were describing. In other dis­
course contexts, gender differences may be present or absent depending on
whether men's and women's discourse goals are similar or dissimilar. For ex­
ample, Tannen's (1990) gender-as-culture hypothesis has implications for
the way men and women approach conversations in general. According to
this hypothesis, men are more likely than women "to manage the discourse,"
a discourse goal discussed by Roberts and Kreuz (p. 161). Because this dis­
course goal can be achieved through the use of rhetorical questions (Roberts
& Kreuz), men might use rhetorical questions more often than women in
conversations. Therefore, the question of gender differences in nonliteral
language use might be addressed with future studies that integrate research
174 LINK AND KREUZ

on differences in men's and women's discourse goals and show how these
goals are achieved using nonliteral language.

ACKNOWLEDGMENTS

We thank Craig Stewart for his assistance in nonliteral language coding and
Katherine Kitzmann and Brent Olde for helpful comments on earlier ver­
sions of this chapter. This research was supported by a Center of Excellence
grant to the Department of Psychology at the University of Memphis by the
state of Tennessee.
Correspondence concerning this chapter should be addressed to Kristen
E. Link, Department of Psychology, State University of New York College at
Oswego, Oswego, NY 13126. E-mail: [email protected]

REFERENCES

Anderson, K.J., & Leaper, C. (1998). Emotion talk between same- and mixed-gen-
der friends: Form and function. Journal of Language and Social Psychology, 17,
419-448.
Bakeman, R., & Gottman, J. M. (1986). Observinginteraction: An introduction to se­
quential analysis. Cambridge, England: Cambridge University Press.
Banaji, M., & LaFrance, M. (1989, May). Gender differences and emotionality: Differ­
ences in verbal expression and similarities in rated intensity. Paper presented at the
meeting of the Eastern Psychological Association, Boston.
Bourg, T. (1996). The role of emotion, empathy, and text structure in children's and
adult's narrative text comprehension. In R. J. Kreuz & M. S. MacNealy (Eds.),
Empirical approaches to literature and aesthetics (pp. 241-260). Norwood, NJ:
Ablex.
Bowers, J.W., Metts, S. M., &.Duncanson, W. T. (1985). Emotion and interpersonal
communication. In M. L. Knapp & G. R. Miller (Eds.),Handbook of interpersonal
communication (pp. 500—550). Beverly Hills, CA: Sage.
Brody, L. R., & Hall, J. A. (1993). Gender and emotion. In M. Lewis & J. M.
Haviland (Eds.), Handbook of emotion (pp. 447–460). New York: Guilford.
Brown, P., & Levinson, S. (1987). Politeness: Some universals in language usage. Cam­
bridge, England: Cambridge University Press.
Christensen, A., & Heavey, C. L. (1990). Gender and social structure in the de-
mand/withdrawal pattern of marital conflict. Journal of Personality and Social Psy­
chology, 59, 73-81.
Clark, H. H. (1973). The language-as-fixed-effect fallacy: A critique of language
statistics in psychological research. Journal of Verbal Learning and Verbal Behavior,
12, 335-359.
Dindia, K., & Allen, M. (1992). Sex differences in self-disclosure: A meta-analysis.
Psychological Bulletin, 112, 106-124.
Ekman, P, Friesan, W. V, & Ellsworth, P. (1972). Emotion in the human face. New
York: Pergamon.
7. USE OF NONLITERAL LANGUAGE 175

Fainsilber, L., & Ortony, A. (1987). Metaphorical uses of language in the expression
of emotions. Metaphor and Symbolic Activity, 2, 239-250.
Feldman Barrett, L., Robin, L., Pietromonaco, P. R., & Eyssell, K. M. (1998). Are
women the "more emotional" sex? Evidence from emotional experiences in so­
cial context. Cognition and Emotion, 12, 555-578.
Ferreira, F, & Anes, M. (1994). Why study spoken language? In M. A. Gernsbacher
(Ed.), Handbook of psycholinguistics (pp. 33-56). New York: Academic Press.
Fryer, R., Harlan, J., Johnson, M. L., Kubrick, S., Richards, M. (Producers), Kubric,
S. (Director). (1980). The Shining [Motion picture]. United States: Warner Bros.
Fussell, S. R., & Moss, M. M. (1998). Figurative language in emotional communica­
tion. In S. R. Fussell & R. J. Kreuz (Eds.), Social and cognitive approaches to interper­
sonal communication (pp. 113-141).Mahwah,.NJ: Lawrence Erlbaum Associates.
Gallois, C. (1993). The language and communication of emotion: Universal, inter­
personal, or intergroup? American Behaviorist, 36, 309-338.
Gibbs, R. W, Jr. (1994). The poetics of mind. New York: Cambridge University Press.
Gibbs, R. W., Jr., & Nascimento, S. B. (1996). How we talk when we talk about love:
Metaphorical concepts and understanding love poetry. In R. J. Kreuz & M. S.
MacNealy (Eds.), Empirical approaches to literature and aesthetics (pp. 291–307).
Norwood, NJ: Ablex.
Glotzer, L, Lester, D., Marvin, N. (Producers), & Dourabont, F. (Director). (1994).
The shawshank redemption [Motion picture]. United States: Columbia Pictures.
Hyde, J. S., & Linn, M. C. (1988). Gender differences in verbal ability: A meta-anal-
ysis. Psychological Bulletin, 104, 53-69.
Izard, C. E. (1971). The face of emotion. New York: Appleton-Century-Crofts.
James, D., & Drakich, J. (1993). Understanding gender differences in amount of
talk: A critical review of research. In D. Tannen (Ed.), Gender and conversational
interaction (pp. 281–312). Oxford, England: Oxford University Press.
Karsch, A. S. (Producer), & Streisand, B. (Producer & Director). (1991). The prince
of tides [Motion picture]. United States: Columbia Pictures.
Kelley, H. H., Cunningham, J. D., Grisham, J. A., Lefebvre, L. M., Sink, C. R., &
Yablon, G. (1978). Sex differences in comments made during conflict within
close heterosexual pairs. Sex Roles, 4, 473-489.
Kintsch, W. (1998). Comprehension: A paradigm for cognition. Cambridge, England:
Cambridge University Press.
Kreuz, R. J., Roberts, R. M., Johnson, B. K., & Bertus, E. L. (1996). Figurative lan­
guage occurrence and co-occurrence in contemporary literature. In R. J. Kreuz
& M. S. MacNealy (Eds.), Empirical approaches to literature and aesthetics (pp.
83-97). Norwood, NJ: Ablex.
LaFrance, M., & Banaji, M. (1992). Toward a reconsideration of the gender-emo-
tion relationship. In M. S. Clark (Ed.), Emotion and social behavior: Review of per­
sonality and social psychology (Vol. 14, pp. 178-201). Newbury Park, CA: Sage.
Leaper, C., Carson, M.., Baker, C., Holliday, H., & Myers, S. (1995). Self-disclosure
and listener verbal support in same-gender and cross-gender friends' conversa­
tions. Sex Roles, 33, 387-404.
Leggitt, J. S., & Gibbs, R. W, Jr. (2000). Emotional reactions to verbal irony. Dis­
course Processes, 29, 1-24.
176 LINK AND KRELJZ

Magliano, J. E, Dijkstra, K., & Zwaan, R. A. (1996). Generating predictive infer­

ences while viewing a movie. Discourse Processes, 22, 199—224.
McMullen, L., &.Conway, J. (1996). Conceptualizing the figurative expressions of
psychotherapy clients. In J. S. Mio & A. N. Katz (Eds.), Metaphor: Implications
and applications (pp. 59–71). Mahwah, NJ: Lawrence Erlbaum Associates.
Noller, E (1993). Gender and emotional communication in marriage: Different cultures
or differential social power? Journal of Language and Social Psyclwlogy, 12,132-152.
Notarius, C. I., & Johnson, J. S. (1982). Emotional expression in husbands and
wives. Journal of Marriage and the Family, 44, 483-489.
Ortony, A. (1975). Why metaphors are necessary and not just nice. Educational
Theory, 25, 45-53.
Fezdek, K., Lehrer, A., & Simon, S. (1984). The relationship between reading and
cognitive processing of television and radio. Child Development, 55, 2072-2082.
Pezdek, K., Simon, S., Stoeckert, J., & Kiely, J. (1987). Individual differences in tele­
vision comprehension. Memory & Cognition, 15, 428-435.
Rime, B., Mesquita, B., Ehilippot, E, & Boca, S. (1991). Beyond the emotional event:
Six studies on the social sharing of emotion. Cognition & Emotion, 5, 435–465.
Roberts, R. M., & Kreuz, R. J. (1994). Why do people use figurative language? Psy­
chological Science, 5, 159-163.
Shimanoff, S. B. (1983). The role of gender in linguistic references to emotive states.
Communication Quarterly, 31, 174-179.
Shimanoff, S. B. (1985a). Expressing emotions in words: Verbal patterns of interac­
tion. Journal of Communication, 35, 16—31.
Shimanoff, S. B. (1985b). Rules governing the verbal expression of emotions be­
tween married couples. The Western journal of Speech Communication, 49,
147-165.
Snell, W.E., Jr., Miller, R. S., & Belk, S. S. (1988). Development of the Emotional
Self-Disclosure Scale. Sex Roles, 18, 59-73.
Snell, W. E., Jr., Miller, R. S., Belk, S. S., Garcia-Falconi, R., & Hernadez-Sanchez, J.
E. (1989). Men's and women's emotional disclosures: The impact of disclosure
recipient, culture, and the masculine role. Sex Roles, 21, 467-486.
Stanovich, K. E., & Cunningham, A. E. (1992). Studying the consequences of liter­
acy within a literate society: The cognitive correlates of print exposure. Memory
& Cognition, 20, 51-68.
Stanovich, K. E., & West, R. F. (1989). Exposure to print and orthographic process­
ing. Reading Research Quarterly, 24, 402-433.
Stark, R. (Producer), & Ross, H. (Director). (1989). Steel Magnolias [Motion pic­
ture] . United States: TriStar Pictures.
Stein, L. B., & Brodsky, S. L. (1995). When infants wail: Frustration and gender as
variables in distress disclosure. The journal of General Psychology, 122, 19-27.
Tannen, D. (1990). You just don't understand: Women and men in conversation. New
York: Ballantine.
Trick, L., & Katz, A. N. (1986). The domain interaction approach to metaphor pro­
cessing: Relating individual differences and metaphor characteristics. Metaphor
and Symbolic Activity, I, 185-213.
Williams-Whitney, D., Mio, J. S., & Whitney, E (1992). Metaphor production in
creative writing, journal of Psycholinguistic Research, 21, 497-509.
 Appendix A

INFORMATION ABOUT FILM CLIPS USED

IN EXPERIMENTS 1 AND 2

Film Emotion Length

The Bridges of Madison County Sadness 0.09375
The Prince of Tides* Sadness 0.166667
A Christmas Carol Happiness 0.02431
The Princess Bride* Love 0.154167
A Cry in the Dark Anxiety 0.08125
Roxanne Love 0.116667
Falling Down* Anxiety 0.08194
The Shawshank Redemption* Happiness 0.1125
A Few Good Men Anger 0.127083
The Shining Anxiety 0.07292
The Fisher King* Sadness 0.08958
Steel Magnolias* Sadness /anger 0.05625
Indiana Jones and the Last Crusade* Happiness 0.104167
The Untouchables Anxiety 0.03542
It's a Wonderful Life* Happiness 0.126389
Vertigo Anxiety 0.0625
An Officer and a Gentleman Sadness 0.104167

Note. All films listed were used in Experiment 1. Clips used in Experiment 2 are marked with an
asterisk.

177
This page intentionally left blank
 Appendix B

SAMPLE NARRATIVE USED IN EXPERIMENT 3

From The Prince of Tides (adapted from Karsch &. Streisand, 1991)

Sally is lying on her bed in her robe, holding the phone to her ear, with her
other arm wrapped around herself. She is squirming uncomfortably as she
tells the person at the other end, "Uh, Tom ... I don't think it's a good idea
for you to come home this weekend."
Tom, the man she is talking to, is standing by the refrigerator in his
kitchen shuffling papers as he responds to Sally: "Why not?"
"Well, I'm not sure ... I'm just not sure that I want to see you right now. I
... uh ... I have a lot to figure out."
As he walks out of the kitchen, Tom says doubtfully, "Uh-huh." He con­
tinues walking into a dim living room as Sally explains further.
"What's the point, Tom? I mean, let's face it. We don't make each other
feel good anymore."
"Right," says Tom as he sits down on a couch and runs his hand through
his hair.
Sally's voice begins to quiver as she continues. "Look. Um ... I didn't
want to tell you this over the telephone. I wanted to tell you before you left,
but the way you left...." Sally pauses as she begins to cry. "There wasn't any
time."
Tom, leaning forward intently, asks, "What'd you want to tell me, Sally?"
There is a long pause while Tom waits and Sally cries. Tom becomes impa­
tient and asks, "What's his name?... What's his name, Sally? The man must
have a name!"
"Jack Cleveland."
179
180 APPENDIX B

Tom puts his head in his free hand and leans back against the couch say­
ing, "Oh no, oh no!" He rubs his forehead as he continues, "Jesus, Sally."
Tom now begins to get angry, stands up, and continues in a stronger voice,
"Geez, Jack Cleveland?! You mean that aging pompous hippie from the hos­
pital who still rides a motorcycle?!"
As Tom is screaming this at her, Sally rests the phone on her shoulder and
shuts her eyes. He continues, asking, "Oh, for Christ's sake, Sally, why him?!
Why him?!"
Sally is crying harder, now, with dark streaks of make-up underneath her
eyes. She wipes her nose with a tissue, breathes deeply, and says, "Because he
knows how he feels about me."
At this, Tom is more calm. He collapses wearily into a chair, clasps his
hand to his forehead, and whispers, "Oh, Sally." Tom sighs and leans for­
ward, with his head still being held up by his hand. "This is too difficult to
talk about over the phone .... Um, just think about it carefully, 'kay?" His
hand is nervously playing about his mouth.
"I hardly think about anything else .... Goodnight, Tom."
As she begins to hang up the phone, Tom stops her when he calls out,
"Sally?" She brings the phone back up to her ear, shaking her head, and asks
with annoyance, "What?"
"Are you really in love?" Tom asks.
Sally answers with a quivering voice, "I'm not sure. I might even be doing
this to hurt you. I gotta go," and hangs up the phone.
Tom sits there, holding the phone for a few seconds more before hanging
up, listening to the dial tone. He sighs and covers his face in his hands. He
sits there wringing his hands for a moment, and glancing to the side, gets up
and walks to a desk in the same room. He sits at the chair in front of it and
turns on the desk lamp. He pulls out a sheet of stationery from a drawer in
the desk, picks up a pen, and stares at it as he turns it in his hands. He begins
to compose a letter to Sally in his head, and then starts to write: "Dear Sally, I
wish the words 'I love you' weren't so difficult for me. I've missed you. I miss
touching you. I don't know what keeps me at such a distance." Tom pauses,
and looks up, with a pained expression on his face. He continues writing,
"I'm sorry I disappoint you, Sally. But you're right to feel that way. How else
could you react to half a man?"Tom sighs deeply. "How could you not be dis­
appointed? Hell, I seem to disappoint anyone who tries to find the best in
me." When he finishes writing, Tom slumps with exhaustion into the chair
and leans his head back, clenching his eyes shut.
 PART III

Sociocultural

Processing Influences

This page intentionally left blank

 8

Discourse and Sociocultural

Factors in Understanding
Nonliteral Language
Albert N. Katz
University of Western Ontario

There is a well-established tradition in linguistics and in certain fields of

psycholinguistics that the study of language comprehension should em­
ploy sentences presented without discourse or social context (e.g., see
Glucksberg, 2001). That is, participants presented with a target sentence
such as, "The night sky was filled with drops of molten silver" (1), might be
asked to rate whether the sentence is syntactically correct, semantically
meaningful, easy to comprehend, or literally true. What is surprising about
this tradition, at least from my perspective (see Katz 1996b), is that as nat­
ural language processors, humans never encounter sentences out of either
discourse or cultural context. Discourse context refers to the characteris­
tics about the language in which a sentence, such as (1), is embedded, cul­
tural context refers to factors such as the shared beliefs between speaker
and listener (s), conventions shared between writer and reader, or, in a
more general sense, the beliefs that surround who the speaker is, who the
listener is, and the environment in which Speaker X makes an utterance to
Listener Y (see Clark, 1996).
In this chapter I discuss both discourse and cultural contexts usually in
tasks in which a target sentence is read within a passage. Discourse factors
that can be manipulated include variables such as whether the reader is
warned in the passage that a given target is forthcoming by use of a marker,
such as, "in a manner of speaking," "really," or "truly"; the number of plausi­
183
184 KATZ

ble continuations invited by the discourse in which the statement is embed­

ded, and so forth. Note that in these cases any influence on the
understanding of the target is based on factors limited to our knowledge of
language and language use.
Cultural or social factors are also presented within the confines of a pas­
sage presented for processing. Now, however, the information being ac­
cessed is knowledge (or beliefs) about the sociocultural world that goes
beyond our knowledge of language use but reflects instead broad beliefs
about our culture and various subgroups within that culture. Within an ex­
periment, variables of interest might be whether the target sentence is ut­
tered by a man or woman (and the gender of the audience), whether it is
uttered by a priest versus a car salesman, or whether it is uttered by a boss
speaking to a subordinate. Note that in these cases there is no logical neces­
sity for the meaning of the utterance to be influenced by these factors; how­
ever, on pragmatic grounds the speaker and listener both might be tapping
into accepted beliefs about the world; if this were the case, then such factors
should be evident in how the target is processed. The interest here is
whether sociocultural context (e.g., our beliefs about female speakers or car
salesmen) is taken into account at the very earliest moments of comprehen­
sion and, if so, whether discourse factors moderate these effects. I discuss
this matter both in terms of utterances in which the expressed meaning is
the same or very similar to the intended meaning and in terms of utterances
in which the expressed meaning differs considerably from the intended
meaning, such as metaphor, irony, indirect requests, and proverbs.

ON CONTEXT

I begin with a simple mind experiment. Consider the sentence, "He cer­
tainly is Abelard to her Heloise" (2), and write down what that sentence
means to you. When I have conducted this exercise informally, it is clear
that people have intuitions about this sentence. They recognize that it is
syntactically correct and usually recognize that Abelard and Heloise refer to
people involved in some social interaction; but, if ignorant of the story of
these two people, they can give us little more. Lacking knowledge about
Abelard or Heloise, my respondents appear to base their inferences on the
presented grammatical information alone. It is as if my respondents assume
that "X (a person = He) certainly is Y (a person = Abelard) to her Z (a per­
son = Heloise)" acts as a construction in construction grammar terms (see
Kay, 1997), in which the construction is used to describe a limited number of
types of interpersonal relationships with the exemplar (Y is to Z) being a pro­
totype of each type (i.e., acts metaphorically).
Contrast sentence (2) with a similar construction in which the requisite
context tends to be accessible to many in our culture: "He certainly is Romeo
8. UNDERSTANDING NONLITERAL LANGUAGE 185

to her Juliet" (3). Now my respondents have little difficulty in coming up with
an elaborated interpretation. Indeed, one can easily see how one could de­
scribe the interpretation that the phrase "Romeo and Juliet" serves as a proto­
type for a certain type of romantic relationship and that the interpretation of
the statement would arise by classifying the topic (he) to the category sug­
gested by the nominal vehicle represented by "Romeo and Juliet" (see, e.g.,
Glucksberg, 1991). That is, the interpretation is one in which the person be­
ing described is depicted in a passionate but tragically doomed relationship.
The point is that even when presumably out of context, the interpreta­
tion of a given statement is inextricably linked to the manner in which it is
presented, and when an explicit context is not available, one is constructed
from stored knowledge during the act of comprehension. Language use is
never context free, despite the logic of many experiments that may suggest
otherwise. In this chapter I do not make the pretense that language is con­
text free; instead, I emphasize examining the comprehension of sentences
such as those just presented when they are presented in a discourse or
sociocultural context. For instance, consider if (3) was as follows: The priest
said, "He certainly is Romeo to her Juliet" (4) versus the comedian said, "He
certainly is Romeo to her Juliet" (5).
Would social stereotypes about priests and comedians come to the fore
and influence how the target is understood, for instance as a metaphoric de­
scription on one hand and as an ironic or sarcastic commentary on the
other? I present some data that indicates the answer is "yes."
Consider another aspect of a sociocultural context, one not based on
stored declarative knowledge (as in the aforementioned examples) but on
the ongoing sociocultural ecology that is active as the comment is being
made. For instance, consider the scenario in which three people are in a con­
versation, but only two of these people know that a friend (John) has sepa­
rated from another friend, Mary. If one of the people "in the know" states
"John and Mary are acting like love birds these days" (6), the two recipients
of this statement would inevitably draw different interpretations, because of
what they know. Indeed the speaker might have framed the message in the
way that he did on purpose: namely, to send one message to the other person
in the know (who should recognize the irony intended by the speaker) and
another message to the unaware communicant—a message intended to
keep that person ignorant of the facts (Kierkegaard, 1996, argued that this is
in fact one of the functions served by ironic usage).
Naturally, the sociocultural ecology described in (6) can also be made sa­
lient in ongoing discourse; moreover, discourse-based and declarative
knowledge-based context can interact. Consider the following:

Speaker A: John can't stand Mary. You can't even have them in the
same room anymore.
186 KATZ

Speaker B: Yeah. He certainly is Romeo to her Juliet.

It is obvious that Speaker B is using the metaphor to be ironic or even sar­

castic. That is, understanding the comment requires knowledge both of the
metaphor and of the ironic spin made about it. Given the familiarity of the
metaphoric base, the ironic spin should be quite transparent. Consider next
the analogous conversation:

Speaker A: John can't stand Mary. You can't even have them in the
same room anymore.
Speaker B: Yeah. He certainly is Abelard to her Heloise.

The situation is somewhat different because the ironic spin is not appar­
ent given the relative unfamiliarity of the contrast. No doubt one could un­
derstand Speaker B: The syntax is not ambiguous and, assuming that
Gricean principles are being followed, one can assume that Speaker B was
following conversational maxims. But what should Speaker A make of the
comment given this speaker's ignorance of the Abelard and Heloise story?
Assuming a metaphoric default, it is quite possible that the ironic intent is
lost completely, with Speaker A making the assumption that the example
provided is of two people who typify an estranged couple.
The informal examples just presented are illustrative of the points I make
in the rest of the chapter. I describe some research from my laboratory and
from others that demonstrates the importance of the social environmental
context in disambiguating speakers' intent that a nonliteral interpretation
of an utterance is in order and that, in many instances, this contextual infor­
mation occurs at the onset of processing.

The Nature of Nonliteral Language

The psycholinguistic literature on nonliteral language often ignores the dif­
ference between literal and figurative language versus familiar and unfamil­
iar expressions. Although literal use and the familiar expressions are
doubtlessly correlated, it is important to make the distinction between
them. Familiar or canonical language refers to language as it is most com­
monly used. Thus, familiarity and literality are orthogonal to one another.
That is, there are instances of nonliteral language in which the dominant
use of certain phrases are intended to express and are understood as express­
ing something different than what is being described by individual words.
Examples abound in many types of nonliteral language: When one states
that his or her car is a lemon or that the grass is greener on the other side of
the fence or that Joe is acting as my best friend, it is generally understood
that the car is deficient (and not a fruit), that envy is being expressed (and
8. UNDERSTANDING NONLITERAL LANGUAGE 187

not a statement about fertilizers), and that Joe has acted as a jerk. In other
cases the use is novel, such as when someone might compare their car to a
banana or hear a proverb with which they are not familiar. This distinction
between fixed or familiar use versus unfamiliar or novel use is expressed in
many ways in the literature as, for instance, the distinction between frozen
and novel metaphors, fixed expressions versus creative usage (see Moon,
1998), and familiar and unfamiliar tropes and is the basis for a general theory
of language processing—namely, the graded salience hypothesis of (Giora,
in press).
This distinction between familiar and literal use has certain implications.
It does not permit one to generalize findings or their theoretical implications
from studies that use conventional metaphor (or salient ironies) to the less
canonical counterparts. Of more direct interest to the issue of context is the
role played by familiar and less familiar tropes and by different types of
tropes, such as irony versus metaphor. Assuming that the salient meaning
comes to mind more readily than a nonsalient or noncanonical meaning,
one can predict that familiar, fixed expressions, used in their usual way,
should be understood more readily with minimal support because the ex­
pression has a well-established sense that transcends specific contexts. Un­
familiar tropes, by contrast, might require context to help disambiguate
speakers' intent and, in general, would be more difficult to integrate into the
ongoing discourse. That is, a syntactically acceptable string of words might
be processed and integrated into ongoing discourse quite readily if this string
was a fixed expression whose dominant meaning was nonliteral. Consider,
however, what would happen if the canonical use is violated—that is, if a fa­
miliar proverb or metaphor is used in a manner that is inconsistent with the
dominant sense, such as when one states that their car is a lemon, referring
to how it is decorated as a float in a parade. Now, based on the same logic
(i.e., that one has ready access to the dominant sense), one can predict that
interpretative problems occur because the dominant sense has to be ignored
for the contextually driven (and, in this case, the literal) sense to be derived.
There is another way in which our understanding of the role played by
context has direct implications for understanding nonliteral language. The
degree of contextual support required for comprehension differs for differ­
ent forms of nonliteral language. This is in contrast with the usual approach
(e.g., see the reviews presented in Gibbs, 1994; and Glucksberg, 2001; but
also see Colston & Gibbs, 2002) which has implicitly argued for commonal­
ity of processing for all forms of nonliteral language but is consistent with the
approach of, for instance, Winner (1988) who argued that comprehension
of irony involves knowledge of social norms and expectations, which is not
the case for the comprehension of metaphor. Thus, knowledge of who said
what might be much more important in the processing of irony than, for ex­
ample, the processing of metaphor. Ellen Winner (1988), in fact, noted that
188 KATZ

a child learns to recognize metaphor at an earlier age than he or she recog­

nizes irony because understanding social conventions (on which irony is
based) develops later than the cognitive skills required to understand meta­
phor. Colston and Gibbs (2002) made a complementary argument with the
processing of metaphor and irony by adults based on the more complex
metarepresentational reasoning required to infer ironic messages compared
to metaphoric messages.
I argue that the distinctions made by Winner (1988) and by Colston and
Gibbs (2002) represent the natural and basic reality of language processing
and should not be shunted as peripheral or minor: What is recovered from
the intent of a speaker is intimately tied to our knowledge and beliefs about
the discourse and social environments in which language is used. Over the
last few years, in my laboratory my colleagues and I have addressed the ques­
tion of the roles played by discourse and social context in the processing of
various forms of nonliteral language, employing both off-line and on-line
methodologies. We showed that the effects of social factors can be quite sub­
tle. For instance, in Lee and Katz (1998) we demonstrated that sarcasm, but
not irony in general, involves the ridicule of a specific person or group of
people. That is, social factors are more important in producing a sense of sar­
casm than they are in producing a sense of irony.
It is a sad commentary on much of the experimental literature that social
factors are not even considered in the construction of the experimental ma-
terials—let alone systematically manipulated. The experimental stimuli
should reflect real-life contextual constraints. Gibbs (2000) reinforced the
need for such care. He analyzed the conversation of university students and
their friends for instances of ironic speech and noted that five distinct types
of irony were frequently employed: teasing-jocularity, sarcasm, rhetorical
questions, hyperbole, and understatement. Moreover, social factors were
found to be important: Some types were more frequently used by men, oth­
ers by women; and, most important, use of irony was often a collaborative
social event, with an ironic comment being followed by ironic responses by
others in the conversation. Presumably, to create experimental situations
that simulate real-life conditions (and, hence, the contextual constraints
that would be active in real-life language production and comprehension)
experimenters should at least construct discourse contexts that reflect the
social conditions in which various forms of language are employed. To date,
this degree of care has not generally been observed, and, in fact, the even
more modest attempt to ensure that the experimental stimuli are realistic is
not the norm. Consider, for example, a study Maggie Toplak and I (Toplak &
Katz, 2000) conducted to examine the reasons a person might use an indi­
rect sarcastic statement rather than the direct criticism. We employed items
taken from a seminal article on irony processing (Kreuz & Glucksberg,
1989) but normed these items to see if our sample could imagine themselves
8. UNDERSTANDING NONLITERAL LANGUAGE 189

in the scenario being described and whether, in that scenario, they would
have used the statement attributed to the speaker. To our surprise, only
about one third of the items from that article met our relatively modest at­
tempt to ensure ecologically realistic discourse contexts. Clearly much more
work is required in the experimental preparation of the stimuli used to ex­
amine selected theoretical issues in the processing of language, especially
nonliteral language.

Off-line Examination of Social Factors in the Use

and Comprehension of Nonliteral Language

Much of the literature on social factors is nonexperimental and involves, for

instance, trying to identify a unique form of woman's speech (e.g.,
Weatherall, 2002) or as a form of speech that changes with age (Ekert,
1997). These approaches are off-line in the sense that there is no monitoring
of the moment-to-moment processing of the nonliteral use; some text is an­
alyzed at some temporal distance from when it might have been originally
produced or encountered. This tradition is also well-represented in experi­
mental psycholinguistic work. Typically the psycholinguistic studies use
textoids, a set of short passages in which some nonliteral use is presented and
participants are asked to make some decision about the use, such as rating
the degree to which sarcasm is expressed or the subsequent memory of the
textoid. Experimental manipulations, when they occur, are produced by
changing the context in which the target sentence is presented. 1 present
some instances next.

Relationships Among the People in Communication. The

amount of literature is growing, which points to the importance of systemati­
cally examining the role played by different participants in a conversation.
For instance, the use of sarcasm has been shown to vary as a function of the
degree of relatedness of the speaker and the victim of the speaker's barb
(Jorgensen, 1996; Kreuz, 1996). Toplak and Katz (2000) showed that differ­
ent members in communication give somewhat different reasons why some­
one would make an indirect sarcastic statement. And, Kreuz, Kassler,
Coppenrath, and McLain (1999) showed that the appropriateness of an
ironic expression is associated with the amount of shared information be­
tween speaker and listener.
Katz and Lee (1993) manipulated contextual information based on
whether or not members of a conversation shared background information
(by virtue of being mutually present when some information was pre­
sented) or not (when the requisite information is presented when one
member of the conversational group was not present). The manipulation
occurred by varying the nature of various texts, which our respondents
190 KATZ

were asked to rate: Participants were presented a set of scenarios in which

one person echoed an earlier statement made by someone else, with the
echoic mention now being metaphoric is use. Two manipulations were im­
portant: (a) whether the context would lead a listener to believe that the
speaker was endorsing the statement or was being pragmatically insincere
and (b) whether the listener was present or not when the original mention
of the statement had been made. This last manipulation would lead to two
groups of listeners: a privileged audience that would share with the speaker
the knowledge of the context in which the statement had been made origi­
nally and a nonprivileged audience that would not have access to this
background contextual knowledge.
The findings, based on ratings given to authorial intent and degree of
metaphoricity or ironicity, were very clear. If a listener thought the speaker
was endorsing the truth value of the statement, then the comment was
taken as a metaphoric description; but if the listener thought the speaker
was actually rejecting the truth value of the statement, then it was taken as
an ironic comment. Moreover, perception of the statement as being ironic
varied as a function of audience privilege: An interpretation of irony was es­
pecially salient when the listener not only perceived a speaker to be making
an incongruent (i.e., a rejecting) statement but also recognized that some­
one else in the conversation (the nonprivileged listener) was unaware that
the speaker was not in fact endorsing the position he or she just espoused.

The Impact of Social Roles. The variables just noted could be

thought of as contextual constraints engendered by the social dynamics of
communicating with people of varying degrees of relatedness and shared
knowledge. In addition, there are contextual constraints engendered by so­
cial convention. That is, one could argue that there are social conventions
associated with gender, socioeconomic class, and power or status inequali­
ties that could be used to disambiguate speaker's intent. An example of so­
cioeconomic factors is evident in the use of indirect requests, such as when
one asks: "Can you open the window?"; taken literally this is a question
about ability but pragmatically is understood as a request to perform the act.
Holtgraves (1997a) and Kemper and Thissen (1981) showed that state­
ments differing in politeness effect our memory: Participants were more
likely to remember the wording of statements that are incongruous with so­
cial status, such as when a low-status person talks in an impolite manner to a
high-status person.

Effect of Gender. Analogous effects can be found with gender:

Women are more likely than men to interpret statements as being indirect
(e.g., Holtgraves, 1991); in addition, as a function of power relationships,
closeness of relationship with the person to whom she or he is talking, and
8. UNDERSTANDING NONLITERAL LANGUAGE 191

the size of the request being made, women differ from men in their use of
polite language (Holtgraves, 1992). Gender differences have been noted
especially with the use of irony and sarcasm. For instance, Gibbs (2000)
found that men make sarcastic remarks almost twice as often as women,
and in general agreement, Colston and Lee (2000) found that irony is con­
sidered a "male-like" trait by both men and women. Katz, Piasecka, and
Toplak (2001) conducted an experimental study in which participants
read a set of short textoids in which several constraints that might influ­
ence a statement as being sarcastic were manipulated. Of interest here was
whether the actor in the textoid who made the comment was a man or a
woman. After reading each scenario, participants were asked a set of ques­
tions about the intent of the speaker, such as whether the speaker was be­
ing sarcastic, ironic, humorous, aggressive, and so on. Gender effects
emerged: When the comment was not obviously intended, men were per­
ceived as more sarcastic, whereas when the comment was intended,
women were rated as more sarcastic. There were no gender-related effects
on the ratings of irony, indicating (consistent with Lee & Katz's 1998
study), that sarcasm involves a specified victim.

Effect of Occupational Stereotypes. Taken together the data can be

viewed as strong evidence for socially driven interpersonal factors in inter­
preting nonliteral and other forms of somewhat ambiguous statements. Penny
Pexman and I (Katz & Pexman, 1997) asked whether we could demonstrate
another socially driven effect on the comprehension of nonliteral language. In
this case the social factor was beliefs (or stereotypes) regarding the occupation
of a speaker. We wished to find out whether these stereotypes could change
the interpretation of a statement from being treated as a metaphor to being
treated as a sarcastic commentary on the metaphoric contrast. Moreover, we
were interested to see whether any such effects would be moderated by the
discourse context or the familiarity of the statement.
Our study (Katz & Pexman, 1997) was quite simple. To conduct this
study we first had to demonstrate a difference in the perceptions of language
usage for members of different occupations. We obtained norms that clearly
showed some differences: For example, clergymen, doctors, and teachers
were perceived as likely to use metaphor (and less likely to use irony) in ev­
eryday speech, whereas comedians, police and factory workers were per­
ceived as more likely to use irony (but not metaphor; for a related matter, see
Holtgraves, 1997b, who provided evidence of cross-cultural and individual
differences in using indirect language).
Once we obtained these occupational differences, we constructed
textoids. In one scenario a statement in the form of a nominal metaphor
(e.g., "children are precious gems") was placed in a discourse that either
brought out the metaphoric sense or an ironic usage or was neutral with re­
192 KATZ

spect to the metaphor versus ironic difference. The nominal metaphors

were either highly familiar instances of basic conceptual metaphors or were
novel instances. For each of the three contexts, the nominal metaphor
would be uttered by a person who was perceived as a high metaphor user
(e.g., clergyman) or high irony user (e.g., comedian). A fourth (control)
condition was also constructed: This condition was the same as the neutral
condition just described except that the speaker was not associated with any
specific occupation. Naturally, passages were counterbalanced such that
any one person only saw the target sentence in one context, which was ut­
tered by a person from one occupation; but, across the study, all key passages
were uttered by someone from each occupation (or, in the final control con­
dition, by a person from no specified occupation).
Two types of dependent measures were taken. First, after each scenario,
participants answered a set of questions regarding the intent of the speaker,
two of which were especially relevant—namely, whether the speaker was
sarcastic and whether the speaker was mocking someone else. Across stud­
ies we manipulated, in the rating questions themselves, whether the occu­
pation of the speaker was made salient. Second, after all the scenarios were
rated and after completing a distraction task, participants were asked to re­
call as many of the target sentences as possible.
The results were very clear. With respect to the rating data, the discourse
contexts acted as one would expect: The key statement was perceived as
more sarcastic when placed in a counterfactual context meant to bring out
the ironic commentary. There was no difference between the neutral con­
text and the context meant to bring out the metaphoric sense, suggesting
that the default was to treat the comparison as a metaphor. Of greater inter­
est here is the influence of speaker occupation. When the neutral condi­
tions were compared (the two in which speaker occupation was mentioned
and the one in which it was not), speakers from high-irony occupations were
perceived as more sarcastic, at least for the highly familiar statements.
When the occupation was made salient in the rating scale, the effects were
even more general: Statements made by speakers of high-irony occupations
were seen as being more sarcastic than the exact same statements made by
high-metaphor speakers. Overall, familiar statements were perceived as es­
pecially sarcastic when uttered in a irony-inducing context or when uttered
by a member of a high-irony occupation. The memory data confirmed the
influence of speaker occupation: Statements uttered by members of high-
irony occupations were better recalled than the same statements made by
members of high-metaphor occupations. Because the statements and con­
texts were identical in every other way, these data suggest that occupation
was coded somewhere in the process of comprehension.
Taken together, these findings indicate that the social role of the person
who makes a statement plays an important role in our perception of a given
8. UNDERSTANDING NONLITERAL LANGUAGE 193

statement as being a metaphor or a sarcastic commentary on that metaphor.

Moreover, these effects interact with the familiarity of the statement itself.
Our interpretation of these data was that the metaphoric content of a famil­
iar statement would be available for use regardless of context. Thus, when
the context is consistent with the metaphoric interpretation, one need not
look for an alternative (ironic) usage. However, when the context is nonen­
dorsing and suggests a sarcastic usage, the incongruity between the familiar
metaphoric use and the ironic context is made especially salient, leading to a
heightened sense of sarcasm. Incongruity is less marked with unfamiliar
statements because those statements are not associated with metaphoric
use. An analogous argument is made with respect to speaker occupation: In
the neutral condition participants have little explanation of why the speaker
made the comment. Participants provided with a highly familiar statement
know that the statement is likely to be used as a metaphor, and it is made by a
person associated with either metaphoric or ironic speech. Only in the latter
case a sense of incongruity was obtained, and it is reflected in the sarcasm
ratings. With unfamiliar items, however, the metaphoric content of the
statement is much less salient, and the sense of incongruity does not arise.
Taken together, these data provide evidence that people are sensitive to
the social roles of a speaker and that the influence of these roles interacts
with constraints engendered by the discourse context and the salience of the
statement itself. The interactive effects of speaker occupation, discourse
context, and statement familiarity suggest that multiple sources of informa­
tion are been weighed when one attempts to understand the intent behind
the use of a given statement.

On-Line Processing and the Role of Contextual Constraints:

A Constraint Satisfaction Approach
I already presented data from off-line studies that support my contention
that social factors play an important role in helping us understand meta­
phor, irony, and indirect requests. The question of interest now is whether
these contextual factors act on-line—that is, during the act of compre­
hending the text. In fact, there is a vigorous ongoing debate whether se­
mantic and pragmatic factors come into play at the earliest stages of
comprehension or whether they only come into play later, after some oblig­
atory processing has occurred. The contrast between whether contextual
constraints are primary and limit what interpretations occur to some ver­
bal information during the act of reading that information and whether
contextual constrains come into play after the fact is relevant not only to
our understanding of nonliteral language comprehension per se, but to a
larger set of issues in language processing. This controversy can be found
from low-level activities such as at the level of word access (Giora, in
194 KATZ

press) and resolution of syntactic ambiguities (McRae, Spivey-Knowlton

& Tanenhaus, 1998) to high-level activities such as the processing of met­
aphor (Glucksberg, 1991), sarcasm (Giora, Fein, & Schwartz, 1998), and
proverbial language (Turner & Katz, 1997).
With respect to the debate on the processing of nonliteral language, the
classic position (e.g., Searle, 1979) is that nonliteral meaning is not pro­
cessed at the earliest moments. This position holds that one is obligated to
initially process for the literal sense of a trope and only when that fails to
provide a contextually adequate interpretation does one then go on and
try to find a plausible (and nonliteral) interpretation. In contrast, there is
by now a 25-year-old experimentally based position and the currently ac­
cepted received wisdom which claims that, with appropriate contextual
support, the nonliteral meaning is made available as rapidly as the literal
sense and that the access of the nonliteral interpretation is not dependent
on a failure to achieve a plausible literal interpretation (see Katz, 1996a,
for a review). In short, the alternative holds that at the earliest moments
nonliteral language can be accessed directly but only if there is an appro­
priately rich supportive context.
My understanding of the literature is that the received wisdom in favor of
the direct-access approach has been best demonstrated with familiar tropes
and, in line with the classic notion supporting obligatory processing of literal
meaning, that less familiar nonliteral meaning might still involve the initial
processing of either literal meaning (see Turner & Katz, 1997) or salient
meaning—that is, coded meaning such as occurs when the nonliteral sense is
the most familiar usage of the item (see Giora, in press; Giora et al., 1998). In
Giora's graded salience hypothesis, there is an obligatory activation of the
dominant sense of an item regardless of context; context can boost the acti­
vation level of nonsalient meaning but will never do so at a cost to the acti­
vation level of the dominant or salient meaning. Thus, in general, the salient
sense is context independent (i.e., it is always activated regardless of con­
text) , whereas the initial activation of nonsalient meaning is context de­
pendent. One implication of this position is that the critical variable is
saliency, not the literalness of the trope: If the most familiar usage is
nonliteral, then that would be activated at the earliest moments of compre­
hension; in fact, the literal sense of such items would only occur later (if at
all) in processing.
Naturally, the opposing position is that context determines what be­
comes aroused at the earliest moments of processing and, consequently,
does not give a processing priority to either a salient or a literal sense of an
item. Rather, with an appropriate context one should be able to find early ac­
tivation of either literal or nonliteral meaning. From my perspective, how­
ever, this approach is underspecified and it is not clear what constitutes
contextual support that would encourage either a literal or nonliteral un­
8. UNDERSTANDING NONLITERAL LANGUAGE 195

derstanding of a message. The argument that I will make presently is that a

constraint satisfaction perspective permits the necessary specification.
Examination of these issues and testing of the contrasting theoretical po­
sitions requires measures that are sensitive to the moment'to-moment pro­
cessing that occurs as one reads or hears the critical information. For
instance, consider again statements such as:. "John and Mary really seem to
hate each other. You can't even leave them alone in the same room anymore.
You can say that he is certainly Romeo to her Juliet." Compare these state­
ments to the following: "John and Mary really seem to love each other. You
can't even leave them alone in the same room anymore. You can say that he
is certainly Romeo to her Juliet." Assuming familiarity with the contrast, the
former usage is ironic whereas the latter one is metaphoric. We already dis­
cussed how familiarity, the nature of the discourse context, and social fac­
tors (e.g., the person who would make such a comment) all play a role in
helping us realize the ironic intent. The issues of interest are (a) whether the
ironic sense would emerge as rapidly as the metaphoric sense if the context
was such that discourse and social factors were both consistent with either
the metaphoric or ironic usage and (b) whether, during the processing of tar­
get items such as in the Romeo and Juliet example, there is evidence that the
metaphoric or ironic meaning arises very early or more slowly, well after the
target is no longer being read or heard.
There are a limited number of relevant on-line studies. Using a cross-
modal priming procedure, Blasko and Connine (1993) demonstrated that
the nonliteral meaning of familiar and apt metaphors is accessed early in the
processing sequence. Using measures of brain activity, Pynte, Besson,
Robichon, and Poli (1996) indicated that, when contextually relevant, the
metaphorical meaning of a sentence is the only one accessed. Both of these
studies are consistent with the received wisdom that, when contextually
supported, the metaphoric meaning is processed early in the sequence and,
hence, unlikely to be dependent on a prior (and obligatory) processing of lit­
eral meaning.
The evidence with sarcastic statements is less clear: Gibbs (1986) showed
that the sarcastic sense is processed as rapidly as the literal sense (and thus
occurs early in the processing sequence and is constrained by context),
whereas Giora et al. (1998) provided evidence (supporting her graded sa­
liency position) that the ironic sense of a statement takes longer to develop
than the literal sense (and thus occurs later in processing; for nonsalient
items, context does not constrain which meaning is accessed; see Giora et
al., 1998 for details).
Although these studies were well executed, neither Giora et al. (1998)
nor Gibbs (1986) study took into account (a) the richness of contextual
constraints that are available to aid comprehension; (b) the multiplicity of
contextual constraints that are, in principle, available for use; or (c) the
196 KATZ

notion that social factors might be especially important in inviting an ironic

but not a metaphoric interpretation. As such, the differences between Giora
et al. and Gibbs' findings may merely represent differences in the contextual
constraints available and differences in when a constraint might come into
play: Perhaps some constraints are more likely to work early in processing,
whereas others develop more slowly and thus only work later in the process­
ing sequence.
The emphasis on constraints has led me to consider an alternative theo­
retical account of nonliteral language processing—multiple sources of infor­
mation are evaluated and integrated continuously. Constraint satisfaction
models, the name typically assigned to such a theoretical approach, have
been quite successful in explaining much of the variability in linguistic ambi­
guity resolution, such as that found in syntax (see McRae et al., 1998). We
argue that, in principle, these models should be applicable to resolving the
ambiguity inherent in the processing of nonliteral language.
The logic of constraint satisfaction is that the understanding of text in­
volves constructing an interpretation that fits the available information
better than plausible alternative interpretations. Such models assume that
different sources of information (i.e., constraints) provide immediate proba­
bilistic support for competing interpretations in parallel over time (e.g., lit­
eral vs. nonliteral or metaphoric vs. ironic interpretation) and, as such, do
not posit modular obligatory processing priorities, such as suggestedby mod­
els that assume one is obligated to process for literal or salient meaning. In
general, competition duration (and thus reading time) is itself a function of
the strength of the various alternatives: If the constraints all point to the
same interpretation, then competition is resolved rapidly, whereas settling
on an interpretation is delayed as support for different alternatives become
more equal.
I argued (e.g., Katz &. Ferretti, 2001) that constraint satisfaction might
prove to be a useful way to resolve some of the issues in the time taken to
comprehend a statement used either literally or in its nonliteral, figurative
sense. Among the constraints of importance would be the context itself (the
extent to which it supports a literal or a nonliteral reading) and the nature of
fixed expressions (the reading of the initial words of a familiar fixed expres­
sions, e.g., as a proverb, idiom, or frozen metaphor, might trigger the rest of
the expression). I argue that additional discourse contexts, such as explicit
markers to usage and, by extension, sociocultural context, should also act as
constraints favoring one or another interpretation.
We (Katz & Ferretti, 2001) began to test this approach in a series of read­
ing studies involving familiar and unfamiliar proverbs placed in contexts
constructed to support either the nonliteral (i.e., proverbial) or literal (i.e.,
nonproverbial) sense of the item. We also employed proverbs that were
rated as highly familiar or unfamiliar by our population because, by manipu­
8. UNDERSTANDING NONLITERAL LANGUAGE 197

lating proverb familiarity and whether the context supports the literal or
nonliteral sense of the item, we could disentangle effects of literality from sa­
liency. In Katz and Ferretti (2001), we took just those two constraints into
account—that is, the extent to which the context was biased toward a literal
or nonliteral use—and whether the trope was familiar or not.
Our task employed a self-paced moving window procedure in which
readers advance a text word-by-word across a computer screen at their own
pace. This procedure mimics the eye movements found in normal reading;
removes spurious strategies adopted by readers forced to read at an uncom­
fortable and, for them, unusual pace; and permits the computation of read­
ing latencies (or reading times, RTs) for each word in a text. Thus, we got a
measure of how long a person lingered over each word, and the amount of
time taken to go from one sentence to the next.
We found that familiar proverbs were read more rapidly that nonfamiliar
proverbs and that this familiarity effect emerged very early—by the second
word of the proverb. We also observed that, for familiar proverbs, there was
no difference in RT by the end of the proverb regardless of context. That is,
the proverbial meaning was as easily integrated into the discourse on-line
whether used in its familiar (proverbial) or less familiar (literal) sense. With
unfamiliar proverbs, relative to the literal biasing context, there was a slow­
ing of reading when the item was placed in a context that brought out its
nonliteral (proverbial) sense—an effect that began to emerge about midway
through the reading of the proverb. This difference was very noticeable at
the last word of the proverb, and it extended into the reading of the sentence
following the proverb. Thus, the resolution or integration of the nonliteral
sense into the discourse was not completed until some time after the proverb
itself was no longer being read.
Without going into detail, these data do not support models that do not
emphasize the context in which nonliteral language is embedded, and they
do not clearly support either those models that posit an obligatory initial
processing of literal or salient meaning or those models that argue that, with
an elaborated context, literal and nonliteral meaning are equally easy to in­
tegrate into discourse. But they are more compatible with a constraint satis­
faction approach. Thus, we argued that the more rapid reading of the
familiar (vs. unfamiliar) proverbs and for unfamiliar proverbs placed in a lit­
eral biasing (compared to a nonliteral biasing) context was due to the fact
that there is less competition between the various sources of information be­
ing evaluated in those conditions. Although the two constraints that we ex­
amined could explain much of the data, another prediction that would be
made by a constraint satisfaction perspective was not observed—namely,
the prediction that a familiar proverb used literally should be more readily
integrated than the same proverb used nonliterally because, in literal usage,
there are two sources of information (i.e., the expressed literal and the
198 KATZ

nonliteral or salient sense) providing information consistent with the literal

interpretation. We did not find this pattern possibly because the constraints
pointing toward a literal interpretation were not very strong. Moreover, by
an analogous argument, the constraint satisfaction approach would predict
that differences in processing unfamiliar proverbs used literally and
nonliterally should be resolved much earlier in processing if there was a con­
straint that signaled that an item was to be read as a proverb.
To test this possibility, we directly manipulated what some linguists would
label as introductory formulae, explicit markers that are the standard ways of
signaling that the receiver will be presented with a proverb (Katz & Ferretti,
in press). Toward this end, we chose to examine the markers "Proverbially
speaking," "In a manner of speaking," and "Literally speaking" because they
disambiguate the intended meanings of proverbs and are consistent with the
types of markers typically used in introducing proverbial (see Mieder, 1982,
1990) and noniiteral usages (Moon, 1998). Thus, we replicated Katz and
Ferretti's (2001) study but added one or another explicit discourse marker.
The findings were clear and striking. Warning the reader about the pro­
verbial nature of the upcoming statement appears to allow for the compre­
hension and integration of unfamiliar proverbs used nonliterally by the end
of the statements. That is, with the appropriate constraint—the marker,
"proverbially speaking"—we eliminated, for unfamiliar proverbs used pro­
verbially, the difference in RT between a literal and nonliteral usage on-line,
as predicted by the constraint satisfaction approach. Moreover, for the first
time, we found evidence that familiar proverbs were integrated more easily
when presented in a literal than nonliteral (proverbial) biasing context—
namely, the first word of the sentence following the familiar proverb was
read more rapidly in the literal inviting context. Thus, as predicted again
from the constraint satisfaction perspective, the dominant tendency toward
a specific interpretation can be overridden if there is sufficient support to­
ward an alternative (in this case, literal) interpretation.
Thus, overall we found encouraging support for a constraint satisfaction
approach for understanding the processing of nonliteral language. To date,
the explicit tests of this theoretical position have been based on manipula­
tions of discourse context. But, as already reviewed, we have off-line evi­
dence that sociocultural context also plays a role in how we interpret and
understand nonliteral language; in principle, these factors can also act in a
constraint satisfaction manner. In fact, there is a growing body of evidence
that, as predicted by a constraint satisfaction perspective, sociocultural con­
text comes into play early in processing.

Occupational Stereotypes. As a first step at examining the role of

sociocultural contextual constraints, Penny Pexman, Todd Ferretti, and I
(Pexman, Ferretti, & Katz, 2000) examined the role of the three constraints
8. UNDERSTANDING NONLITERAL LANGUAGE 199

that were examined in Katz and Pexman (1997). Recall that in the earlier
study we presented target items in the form of a nominal metaphor—
namely, "An A is (does) B." These items were either familiar or unfamiliar
instantiations of basic conceptual metaphors, and they were always pre­
sented in short passages, being uttered by one of the characters in the story.
Thus, the first constraint we examined was the role that expressed familiar­
ity played on disambiguating the meaning intended by the character who ut­
tered the statement. The second constraint was the nature of the discourse
that preceded the target item: The discourse either suggested the target
statement was an endorsement and hence being used as a metaphor (the
metaphor-inviting condition) or suggested that the speaker was rejecting
the truth value of the statement and hence using the metaphor to make a
sarcastic commentary (the sarcasm-inviting condition) or was neutral with
respect to endorsement rejection. The third constraint was a social factor—
the occupation of the person who made the statement (either high-irony
speaker, high-metaphor speaker, or a control condition in which the
speaker's occupation was not presented)—which was manipulated. The
findings of this earlier study were clear using off-line dependent measures
(either recall or ratings of sarcasm and mocking): Each of these constraints
helped to disambiguate the metaphoric or sarcastic intent of the speaker,
which often acted interactively.
We wished to see whether any of these three constraints acted on-line
during the reading of the target item or whether effects only occurred later
in the processing sequence. The latter point is especially important because
it allows us to determine when each of the contextual constraints exerts its
influence on the target sentence. Moreover, we eliminated the possibility
that our sample adopted special metaphor or sarcasm reading strategies (see
Steen, 1993) by intermixing the critical experimental passages among a
large number of filler passages. The critical experimental passages were very
similar to those employed in Katz and Pexman (1997) with the major alter­
ation being the addition of text after the target statement, allowing us to in­
vestigate contextual effects that arise after the target has been read.
The RT data were very clear. Averaged over all conditions, there was lit­
tle variability in reading the target sentence until the last word, which was
read 170 ms. more slowly than the preceding words. This increase in RT at
sentence end is a standard finding and is known as sentence wrap-up in the
language processing literature, which indicates "A search for referents that
have not been assigned, the construction of inter-clause relations ... and an
attempt to handle any inconsistencies that could not be resolved within the
sentence" (Just & Carpenter, 1980, p. 345). Given that most theories of
metaphor processing put the processing load for comprehension with the
vehicle (in our case, the last word), variability in RT for the last word in the
sentence should depend on how readily the last word is integrated into the
200 KATZ

representation of text. If the processing of a metaphor in context occurs di­

rectly, it should be expected that the resolution of any ambiguity would be
completed by sentence end and not spill over to the RTs taken downstream
(as evidenced by longer RTs taken to start reading the sentence following
the target or even the first few words of the following sentence). Thus, we
wished to see if spill over occurred and if such an occurrence varied with the
contextual constraints.
There was clear evidence that the constraints came into play quite early
in the processing sequence. First, on comparing the two neutral conditions
in which speaker's occupation (high-metaphor or high-irony) was men­
tioned with that for which there was no mention of occupation, we found re­
liably longer RTs for the last word of the ambiguous statement in the two
conditions involving occupational mention; this effect was most evident for
statements that were unfamiliar. Taken together, these data indicate that
the effect of a social constraint and the indicator of speaker's intent is avail­
able no later than the end of the utterance—that is, at an early stage of state­
ment processing. The familiarity effect suggests that at these early stages of
processing the cue is mainly useful in the comprehension of unfamiliar state­
ments, presumably those without a well-established canonical (or salient)
meaning.
Second, there were clear effects based on the nature of the discourse that
preceded the target statement. Targets that were embedded in the meta-
phor-inviting context produced much less spillover to locations down­
stream than the same items placed in a sarcasm-inviting condition. In fact,
the RTs in the two conditions were the same at the last word in the target lo­
cation, suggesting that, regardless of the nature of the context, the meta­
phoric interpretation of the statement was largely completed by statement's
end. However, the time taken to proceed to the first word of the sentence
following the target was slower (by 54 ms) in the sarcasm-inviting condition.
These data suggest that although the metaphoric content of the statement
might be available early, the recognition of the sarcastic commentary takes
longer to develop.
Third, there was an interactive effect of discourse context and occupa­
tional mention: When the target statement was uttered by a person from a
high-metaphor occupation, RTs at the last word in the statement were sub­
stantially slower in the neutral condition (by over 100 ms) than either the
irony-inviting or metaphor-inviting contexts, which did not differ from one
another. When the target is uttered by a person from a high-irony occupa­
tion, once again we find slower RTs in the neutral condition relative to the
other two conditions. However, now there are additional effects, with differ­
ences found in RTs beyond the statement itself: The time taken to proceed
to the next sentence is faster in the metaphor-inviting condition but slower
in the irony-inviting condition, a difference of over 100 ms.
8. UNDERSTANDING NONLITERAL LANGUAGE 2O1

Taken together, the effects show that providing enriched context leads to
some resolution of meaning of an ambiguous statement by the last word of
that statement. Given that in our case the statements are in the canonical
form of a nominal metaphor, these data are consistent with the claim that,
first, with contextual support, metaphoric meaning is available very early in
the processing sequence. Second, when the statement is uttered by a person
typically understood as a high-metaphor user, further analysis of the state­
ment does not appear to occur (as evidenced by a failure to find marked
spillover effects in RT beyond the sentence itself). Third, when the speaker
is typically understood as a high-irony user, additional processing occurs
downstream, almost immediately when the discourse context invites an
ironic interpretation and more slowly when the context just supports the de­
fault metaphoric interpretation. Thus, it appears that at least one social fac­
tor (speaker's occupation) plays an important role in the on-line reading of
nonliteral language. When the discourse context is sarcastic, interpretation
of the speaker is assessed early and aids in the integration of the target state-
ment's meaning very shortly after the statement has been read; when there
is an incongruity between the sarcastic nature of the speaker and a context
that invited a nonsarcastic interpretation, this incongruity appears to be ini­
tially ignored but is noted by the beginning of the next sentence. Although
we did not discuss it in these terms in Pexman et al. (2000), these data are
completely consistent with a constraint satisfaction perspective.

Gender Effects. As noted earlier, sociolinguistic and off-line psycho­

linguistic work implicates gender as an important factor in the processing of
sarcasm. In my laboratory, we (Katz et al., 2001) started to examine the
on-line role of gender using the moving windows methodology. We asked
participants to read a set of passages as in the studies already described. The
critical passages were pretested to ensure that the speaker was being sarcas­
tic. Of most interest was our manipulation—namely, the gender of speaker
and listener—which we orthogonally varied in a 2 (male or female speaker)
X 2 (male or female listener) design. The passages were exactly the same ex­
cept for the gender assigned to speaker and listener (as operationalized by
gender-specific names). The results were intriguing: First, overall passages
were read more quickly by a male rather than a female speaker; this effect
was most noted for the last two words of the utterance made by the speaker.
That is, gender effects emerged during the act of reading the critical sarcas­
tic statement, once again supporting the argument that social constraints
come into play very early in the process of comprehension—an interpreta­
tion that is supported by other lines of emerging research (Garnham,
Oakhill, & Reynolds, 2002). One can speculate that the slowdown in read­
ing for the female-speaker condition suggests that it is more difficult to settle
on an interpretation of a possible sarcastic comment when the statement is
2O2 KATZ

uttered by a women, perhaps because sarcasm is less likely to be associated

with women. The effect just noted was modified by the gender of the lis­
tener. The effect emerged, once again, for the last two words of the possibly
sarcastic comment. That is, the slower reading of comments made by female
speakers is found most notably when the comment is directed at another
woman, once again supporting the notion that, because sarcasm is more
likely to be associated with men, noncanonical usage is delayed as people at­
tempt to integrate what they are reading with both the text and their stored
knowledge of the world. And this occurs on-line during the act of reading.

SUMMARY AND CONCLUSIONS

Based on the data presented, I draw the following conclusions and take the
liberty to speculate freely. First, I argue that social constraints are a ubiqui­
tous force in the disambiguating of speakers intent. In the aforementioned
review, I limited myself to certain linguistic forms in which what a speaker
expresses differs from what he or she intended to convey: metaphor, sar­
casm, and, to a lesser extent, indirect requests. And I limited most of the dis­
cussion here to two such social constraints: occupational status and gender.
In principle, however, any of a number of linguistic forms could be consid­
ered and a host of additional sociocultural factors examined, such as age or
ethnicity or whether the people interacting in a passage are related in some
way. Second, the effects of these social constraints are found not only in
off-line tasks, but also (from my perspective, even more provocatively), in
on-line tasks. Social constraints come into play almost immediately during
the act of comprehension. Third, the effects of social constraints appear to
interact with a host of other variables, such as the familiarity of the expres­
sion being used by the speaker; although the evidence for this is most clearly
seen in the off-line studies, it also is observed in some of the on-line data.
These data have direct implications for the type of processing models that
are suitable for explaining language comprehension in general and
nonliteral language comprehension more specifically. The standard theo­
retical contrast, in both cases, has been between two classes of models. In
one case, the extraction of some information is obligatory and in which the
effects of context (however it is defined) limits comprehension to one (or, at
most, a very few) context-appropriate meanings. The second and contrast­
ing theoretical approach does not put an emphasis on the processing of
obligatory information but instead considers that context plays a central
and immediate role in the extraction of information.
Therefore, I favor the second set of models in which discourse is consid­
ered to play an immediate role in the ongoing extraction of meaning from
language. But I especially favor a specific version of such models based on
the finding that the interpretation that we make is based on a set of factors
8. UNDERSTANDING NONLITERAL LANGUAGE 2O3

that interact with one another—namely, constraint satisfaction. Thus, a

sociocultural factor, such as knowledge of a persons occupation, might in­
vite a metaphoric or an ironic understanding of the same phrase, but this ef­
fect is moderated by the familiarity of the phrase itself. Generally, the effect
of occupation, gender of the speaker, or whether one is speaking to a close
friend by itself will not be sufficient enough to provide an unambiguous in­
terpretation; in fact, these factors might only come into play when other fac­
tors are not present. For instance, occupational knowledge might be a strong
constraint when other information in the context is not available for
disambiguation to occur. As such, and in line with the evidence that I al­
ready reviewed, I favor an approach like that instantiated in constraint satis­
faction models.
Note that in many ways the most recent version of the graded salience
model is also compatible with much of the data discussed as supportive of
constraint satisfaction (see Giora, in press). In this version the original the­
ory has been modified such that Giora's original modular component (sensi­
tive to salience) is now complemented by an additional context-sensitive
mechanism. Giora posited that context can boost the activation of non-
salient material but not at the cost of the activation of salient material. Al­
though the addition of a second component often leads to similar
predictions to those offered by constraint satisfaction models, the theoreti­
cal differences between the two positions should not be minimized. Con­
straint satisfaction models do not give processing priority to salient
information; rather, these models view language processing as a function of
the ongoing interaction of multiple sources of information, each of which
will constrain the interpretation that will emerge to various degrees. Con­
straints that are very strong (e.g., item saliency) may appear to be obligatory
because the other constraints present in the context are not strong enough
to overcome the dominance of a very strong constraint. Moreover, and most
important, constraint-based models hold that increasing the strength of op­
posing constraints can, in principle, lead to greater activation of less salient
meanings than salient meanings and, in opposition to Giora's position,
might even lead to the salient meaning not be activated at all (see Vu, Kellas,
& Paul, 1998, for a demonstration of word access). Thus, in principle, con-
straint-based and the graded salience models can be empirically disentan­
gled, although, given the dominance of salience, one might need a
methodology more sensitive than the moving windows technique that I
used. One possibility is to use event-related brain potentials that, because of
the multidimensional nature of the data sets produced, have proven sensi­
tive to subtle differences in the processing of language and other cognitive
processes (see Osman, 1998, for a review).
In conclusion, there are two consequences that follow from adopting a
constraint satisfaction approach. First, the standard approach taken in the
2O4 KATZ

nonliteral language processing literature wherein a researcher looks to see if

in a given experiment some obligatory processing occurs or not (e.g., the ex­
traction of literal meaning) is, from the constraint satisfaction approach, a
fruitless enterprise because one would expect that literal meaning might in
fact be extracted before nonliteral meaning if the constraints all point to­
ward a literal interpretation but would not be extracted initially if the con­
straints point toward a figurative interpretation. The emphasis is thus
neither on the characteristics of the target phrase per se nor, generally
speaking, just on the discourse characteristics, but rather on the specific fac­
tors that appear in the presentational context and the meaning of those fac­
tors to the interpreter. The second point follows directly from the first.
Given the importance of understanding the constraints available in the dis­
course, a major job for experimenters will be to systematically identify con­
straints and to quantify the strength of these constraints in different
contexts. I argue further that central to this mission would be a systematic
examination of social factors, such as gender, occupation, and social status.
Research that identifies pragmatic sociocultural constraints include, but is
not limited to, identifying (a) when a specific person might employ humor,
sarcasm, or hyperbole, (b) socioeconomic knowledge about language use;
and (c) gender-related effects. In essence, a constraint-based approach re­
quires an in-depth examination of the reasons why a given expressions
might be used (e.g., see Toplak & Katz, 2000, for sarcastic irony) and when it
might be employed (e.g., see Drew & Holt, 1995, for a well-developed exam­
ple with idioms).
ACKNOWLEDGMENT

This research was supported by a grant from the Natural Sciences and Engi­
neering Research Council of Canada (Operating Grant No. 06P007040).
Correspondence concerning this chapter should be addressed to Albert
Katz, Department of Psychology, The University of Western Ontario, Lon­
don, Ontario, Canada, N6A 5C2. E-mail: [email protected]

REFERENCES

Blasko, D., & Connine, C. (1993). Effects of familiarity and aptness on metaphor
processing. journal of Experimental Psychology: Learning, Memory, and Cognition,
19, 295-308.
Clark, H. (1996). Using language. Cambridge, England: Cambridge University Press.
Colston, H., & Gibbs, R. W., Jr. (2002). Are irony and metaphor understood differ­
ently? Metaphor and Symbol, 17, 57-80.
Colston, H., & Lee, S. (2000, May). On gender differences in verbal irony. Paper pre­
sented at the meeting of the International Gender and language Association,
Stanford University, Palo Alto, CA.
8. UNDERSTANDING NONLITERAL LANGUAGE 2O5

Drew, P, & Holt, E. (1995). Idiomatic expressions and their role in the organization
of topic transition in conversation. In M. Everaert, E. J. van der Linden, A.
Schenk, & R. Schreuder (Eds.), Idioms: Structural and psychological perspectives
(pp. 117-132). Hillsdale, NJ: Lawrence Erlbaum Associates.
Ekert, P (1997). Age as a sociolinguistic variable. In. F. Coulmas (Ed.), The handbook
of sociolinguistics (pp. 151-167). Oxford, England: Blackwell.
Garnham, A., Oakhill, J., & Reynolds, D. (2002). Are inferences from stereotyped
role names to characters' gender made elaboratively? Memory and Cognition, 30,
439-446.
Gibbs, R. W., Jr. (1986). On the psycholinguistics of sarcasm. Journal of Experimental
Psychology: General, 115, 3-15
Gibbs, R. W., Jr. (1994). The poetics of mind. Cambridge, England: Cambridge Uni­
versity Press.
Gibbs, R. W., Jr. (2000). Irony in talk among friends. Metaphor and Symbol, 15,5-27.
Giora, R. (2003). On our mind: Salience, context and figurative language. New York:
Oxford University Press.
Giora, R., Fein, O., & Schwartz, T. (1998). Irony: Graded salience and indirect ne­
gation. Metaphor and Symbol, 13, 83-101.
Glucksberg, S. (1991). Beyond literal meanings: The psychology of allusion. Psycho­
logical Science, 2, 146-152.
Glucksberg, S. (2001). Understanding figurative language: From metaphors to idioms.
New York: Oxford University Press.
Holtgraves, T. (1991). Interpreting questions and replies: Effects of face-threat,
question form, and gender. Social Psychology Quarterly, 54, 15-24.
Holtgraves, T. (1992). Interpersonal underpinnings of request strategies: General
principles and differences due to culture and gender. Journal of Personality and So­
cial Psychology, 62, 246-256.
Holtgraves, T. (1997a). Politeness and memory for wording of remarks. Memory &
Cognition, 25, 106-116.
Holtgraves, T. (1997b). Styles of language use: Individual and cultural variability in
conversational indirectness. Journal of Personality and Social Psychology, 73,
624-637.
Jorgensen, J. (1996). The functions of sarcastic irony in speech. journal of
Pragmatics, 26, 613-634.
Just, M., & Carpenter, P (1980). A theory of reading: From eye fixation to compre­
hension. Psychological review, 87, 329-354.
Katz, A. (1996a). Experimental psycholinguistics and figurative language: Circa
1995. Metaphor and Symbolic Activity, 11, 17-37.
Katz, A. (1996b). On interpreting statements as metaphor or irony: Contextual
heuristics and cognitive consequences. In J. S. Mio & A. N. Katz (Eds.), Meta­
phor: Implications and applications (pp. 1—22). Mahwah, NJ: Lawrence Erlbaum
Associates.
Katz, A., & Ferretti, T. (2001). Moment-by-moment comprehension of proverbs in
discourse. Metaphor and Symbol, 16, 193-221.
Katz, A., & Ferretti, T. (2003). Reading proverbs in context: The role of explicit
markers Discourse Processes, 36, 19—46.
2O6 KATZ

Katz, A., & Lee, C. (1993). The role of authorial intent in determining verbal irony
and metaphor. Metaphor and Symbolic Activity, 8, 257-279.
Katz, A., & Pexman, P (1997). Interpreting figurative statements: Speaker occupa­
tion can change metaphor to irony. Metaphor and Symbol, 12, 19-41.
Katz, A., Piasecka, I., &Toplak, M. (2001). Comprehending the sarcastic comments of
males and females. Paper presented at the 42nd annual meeting of the Psycho-
nomic Society, Orlando, FL.
Kay, P (1997). Words and the grammar of context. Stanford, CA: CSLI Publications.
Kemper, S., & Thissen, D. (1981). Memory for dimensions of requests. Journal of
Verbal Learning and Verbal Behavior, 20, 552-563.
Kierkegaard, S. (1996).The concept of irony, with constantreference to Socrates (L. M.
Capri, Trans.). London: Collins.
Kreuz, R. J. (1996). The use of verbal irony: Cues and constraints. In J. S. Mio & A.
N. Katz (Eds.), Metaphor: Implications and applications (pp. 23-38). Mahwah, NJ:
Lawrence Erlbaum Associates.
Kreuz, R. J. & Glucksberg, S. (1989). How to be sarcastic: The echoic reminder the­
ory of verbal irony. Journal of Experimental Psychology: General, 118, 374-386.
Kreuz, R. J., Kassler, M., Coppenrath, L., & McLain, A. B. (1999). Tag questions and
common ground effects in the perception of verbal irony.Journal of Pragmatics,
31, 1685-1700.
Lee, C., & Katz, A. (1998). The differential role of ridicule in sarcasm and irony.
Metaphor and Symbol, 13, 1-5.
McRae, K., Spivey-Knowlton, M., & Tanenhaus, M. (1998). Modeling the influ­
ence of thematic fit (and other constraints) in on-line sentence comprehension.
Journal of Memory and Language, 38, 283-312.
Mieder, W. (1982). International proverb scholarship: An annotated bibliography (Vol.
1). New York: Garland.
Mieder, W. (1990). International proverb scholarship: An annotated bibliography (Vol.
2). New York: Garland.
Moon, R. (1998). Fixed expressions and idioms in English; A corpus-based approach.
Oxford, England: Clarendon.
Osman, A. (1998). Brainwaves and mental processes: Electrical evidence of atten­
tion, perception and intention. In D. Scarborough & S. Sternberg (Eds.), An invi­
tation to cognitive science: Vol. 4. Methods, models and conceptual issues (pp.
865-915). Cambridge, MA: MIT Press.
Pexman, P, Ferretti, T, & Katz, A. (2000). Discourse factors that influence on-line
reading of metaphor and irony. Discourse Processes, 29, 201-222.
Pynte, J., Besson, M., Robichon, F. H., & Poli, J. (1996). The time course of meta­
phor comprehension: An event-related potential study. Brain and Language, 55,
293-316.
Searle, J. (1979). Metaphor. In A. Ortony (Ed.), Metaphor and thought (pp. 92-123).
Cambridge, England: Cambridge University Press.
Steen, G. (1993). Metaphor and literary reception. London: Longman.
Toplak, M., & Katz, A. (2000). On the uses of sarcastic irony. Journal of Pragmatics,
32, 1476-1488.
Turner, N., & Katz, A. (1997). The availability of conventional and of literal mean­
ing during the comprehension of proverbs. Pragmatics and Cognition, 5, 199-133.
8. UNDERSTANDING NONLITERAL LANGUAGE 2O7

Vu, H., Kellas, G., & Paul, S. (1998). Sources of constraints on lexical ambiguity res­
olution. Memory & cognition, 26, 979-1001.
Weatherall, A. (2002). Gender, language and discourse. New York: Routledge.
Winner, E. (1988). The point of words: Children's understanding of metaphor and irony.
Cambridge, MA: Harvard University Press.
This page intentionally left blank
 9

Social Factors in the

Interpretation of Verbal Irony:

The Roles of Speaker

and Listener Characteristics

Penny M. Pexman
University of Calgary

The topic of this chapter is the role of social factors in interpretation of

ironic intent. To illustrate the type of factors that are discussed, imagine
you are the addressee of the following remark: "You are so punctual." How
will you interpret this remark? How will you determine what meaning the
speaker intends? Your metalinguistic knowledge may indicate that al­
though most remarks of this type are intended literally, some are intended
ironically. In determining whether this remark is intended literally or iron­
ically you may attend to clues in the situation in which the remark oc­
curred: If you just arrived 20 min late for an important meeting, then the
speaker likely has a negative attitude about your punctuality. Incongruity
between the negative tone of the situation and the positive tone of the
statement is a reliable cue to ironic intent (e.g., Colston & O'Brien, 2000a;
Gerrig & Goldvarg, 2000; Katz & Lee, 1993; Katz & Pexman, 1997). Per­
haps earlier in the day you made a positive claim to the speaker about your
punctuality (e.g., "Don't worry, I'm really punctual. I'll make the meeting
on time"). The fact that the speaker's remark echoed your claim could also
be a cue to ironic intent (e.g., Colston, 2000; Gibbs, 1986; Jorgensen,
Miller, & Sperber, 1984; Kreuz & Glucksberg, 1989; Kumon-Nakamura,
Glucksberg, & Brown, 1995). Even if this claim had not been made, how­
2O9
21O PEXMAN

ever, the fact that the remark echoes an implicit norm or social convention
(that people should be punctual) could cue ironic intent. Another cue
could be the speaker's use of the word so in the remark. The exaggeration
implied by this word may be another hallmark of verbal irony (Colston &
O'Brien, 2000a; Kreuz, 1996). Certainly, tone of voice can also signal
irony, but ironic intent can be detected without tone of voice information
(Kreuz & Roberts, 1995). In addition to these verbal cues, nonverbal cues
like gesture and facial expression indicate irony. Much of the psychological
research on verbal irony has focused on identifying these verbal and non­
verbal cues to ironic intent. Less research has addressed the social cues
that may be relevant. It seems likely that characteristics of the speaker
would be considered: Is this someone who often speaks ironically? Does
the speaker like the listener? It also seems possible that characteristics of
listeners could influence their interpretation: Are some people (perhaps
because they frequently speak ironically themselves) more prone to an
ironic interpretation? I first summarize research on these issues, then con­
sider how current theories of irony would (or would not) predict a role for
social factors, and finally suggest ways in which these social factors might
be relevant for interpreting other forms of figurative language.

SPEAKER CHARACTERISTICS

Past research has shown that certain characteristics are attributed to

speakers making ironic statements. Compared to speakers who make
nonironic remarks, speakers who make ironic remarks are perceived to be
more angry, disgusted, and scornful (Leggitt & Gibbs, 2000), as well as ver­
bally aggressive and offensive (Toplak & Katz, 2000). Colston (1997) re­
ported that speakers who used ironic criticisms were perceived to be more
condemning than speakers who used literal criticisms. Roberts and Kreuz
(1994) found that irony was used to express negative emotion. It has also
been argued, however, that irony does not serve only negative communi­
cation goals. Kreuz, Long, and Church (1991; see also Colston & O'Brien
2000a, 2000b) reported that ironic statements were intended to mock
someone but also to be funny. Dews and Winner (1995; Dews, Kaplan, &.
Winner, 1995) reported that speakers who made ironic criticisms were per­
ceived to be less annoyed than speakers who made literal criticisms. Dews
and Winner proposed the tinge hypothesis to explain these findings. Ac­
cording to the tinge hypothesis, ironic criticisms are less critical than lit­
eral criticisms because the positive surface meaning of an ironic statement
tinges the listener's perception of the negative, ironic meaning. Thus,
irony lessens the criticism conveyed.
Although Dews and Winner (1995) reported findings that supported
the tinge hypothesis, Colston's (1997) results suggested a very different
9. INTERPRETATION OF VERBAL IRONY 211

function for verbal irony: to enhance conveyed criticism. Colston noted

that his procedure was somewhat different than Dews and Winner's. As is
typical, Dews and Winner presented their stimuli to participants in written
form (a target statement preceded by a context paragraph). In addition,
and less typically, Dews and Winner gave participants tone of voice infor­
mation for the statements in the form of a tape recording. Ironic criticisms
were read in a nasal, mocking tone of voice, and literal criticisms were read
in an angry tone of voice. As Colston pointed out, tone of voice informa­
tion alone could have been used to perceive that speakers of ironic criti­
cisms were less annoyed than speakers of literal criticisms. Colston did not
present tone of voice information.
In a recent study, Kara Olineck and I (Pexman & Olineck, 2002a) pre­
sented some resolution to the debate over the tinge hypothesis. We argued
that the social function of irony might depend on the perspective taken in
interpretation. If one considers dimensions that tap perceived speaker in­
tent by asking, for instance, whether the speaker was mocking the addressee,
then ironic criticisms may be rated higher (more mocking) than literal criti­
cisms. In contrast, if one considers dimensions that tap the general social im­
pression created by the statement, such as whether the speaker is being
impolite, then ironic criticisms may be rated lower (less impolite) than lit­
eral criticisms. This is precisely what we found. We concluded that the tinge
hypothesis is relevant to the general social impression created by the ironic
criticism (i.e., it is considered more polite than a literal criticism) but not
necessarily to perceived speaker intent (i.e., it is perceived to be more mock­
ing than a literal criticism).
Although there are clearly many different communication goals at­
tributed to ironic speakers, the research reviewed thus far does not
demonstrate that these attributes would necessarily function as cues to
ironic intent. To demonstrate this, the attributes would need to be pre­
sented as speaker characteristics and be shown to increase the likeli­
hood of an ironic interpretation. The attributes could be presented
directly as traits (e.g., the speaker is labeled as a funny person) or could
potentially be inferred from other types of social information (e.g., so­
cial categories). If these speaker characteristics were shown to influ­
ence attributions of speaker intent, then the controversial implication
would be that communicative intentions are a perceived to be a func­
tion of both the situation and the person. That is, listeners consider fac­
tors like speaker personality traits to be relevant to communicative
choices. In the following sections I describe research in which speaker
attributes have been manipulated to examine their effects on interpre­
tation of verbal irony. These attributes have included the nature of the
speaker-addressee relationship, speaker occupation, speaker personal­
ity traits, and gender of the speaker.
212 PEXMAN

Speaker-Addressee Relationship
Slugoski and Turnbull (1988) examined the influence of relationship vari­
ables on interpretation of potentially ironic remarks. That is, speakers and
addressees were identified as either close or distant and as either liking or
disliking each other. Results showed that relationship distance had little ef­
fect, but that liking had a strong effect. When the speaker and listener
clearly liked each other, literal insults were more likely to be interpreted as
ironic compliments (as compared to situations in which speaker and listener
disliked each other). In contrast, literal compliments were more likely to be
interpreted as ironic insults when the speaker and listener clearly disliked
each other. These findings suggested that listeners consider relationship
variables in the process of inferring speaker intent. Relationship affect, in
particular, seems to be a useful cue to ironic intent. This may be because it
reveals the speaker's attitude, which tends to be more positive for ironic
compliments than for literal insults, and more negative for ironic insults
than for literal compliments.
Speaker Occupation
Occupations are social categories that can be associated with particular
traits. Slusher and Anderson (1987) examined stereotypical traits of law­
yers, artists, and clergymen. They found that participants characterized
these professions in the following ways: wealthy and aggressive lawyers, tem­
peramental and creative artists, and kind and friendly clergymen. Albert
Katz and I (Katz &. Pexman, 1997) investigated whether occupation stereo­
types extended to beliefs about language use. For 50 different occupations,
we asked participants to rate members of each occupation for (a) likelihood
of using irony and (b) likelihood of using metaphor. Results showed that par­
ticipants had quite consistent perceptions about tendencies to use irony and
metaphor, such that members of certain occupations (e.g., comedian) were
reliably perceived as much more likely to use irony than were members of
other occupations (e.g., nurse). Similarly, there were large differences across
occupations in perceived likelihood of using metaphor, with members of cer­
tain occupations (e.g., English professor) rated as much more likely to use
metaphor than members of other occupations (e.g., factory worker).
We further speculated that people's shared perceptions about language
use by members of different occupational categories might function as cues
to speaker intent. That is, if a speaker was described as a member of a
high-irony occupational category, then his or her remarks might be more
likely to be interpreted as ironic. To examine this issue, we devised stimuli
that were metaphors (e.g., "children are precious gems"). These stimuli
could be interpreted as positive metaphoric remarks but could also be inter­
preted as ironic remarks.
9. INTERPRETATION OF VERBAL IRONY 213

Our goal was to investigate whether speaker occupation cued ironic in­
tent as a function of the incongruity of the statement with the discourse con­
text and also the familiarity of the statement. As such, we presented
participants with short context paragraphs followed by target metaphors.
Speakers were identified as members of high-irony occupations, high-meta-
phor occupations, or no occupation was specified. Contexts were devised
such that some were irony-inducing (a negative situation was described),
some were metaphor-inducing (a positive situation was described), and
some were neutral (outcome was not described as positive or negative). The
target metaphors were all novel instantiations of root metaphors (Lakoff &
Johnson, 1980) but were selected based on pilot study ratings such that half
were relatively more familiar and half were relatively less familiar. Partici­
pants were asked to consider each context and target statement and rate
speaker intent. Afterwards, they were given an unexpected free recall task
for the target statements.
Results showed that speaker occupation did cue perceptions of speaker
intent, because metaphoric statements made by members of high-irony oc­
cupations were considered to be more sarcastic than the same statements
made by members of high-metaphor occupations. Also, in making the meta­
phoric statements, high-irony speakers were perceived to be more mocking
than high-metaphor speakers, particularly when the statements were rela­
tively less familiar. We argued that speaker occupation cued ironic intent
through heightened incongruity. That is, the incongruity created when a
highly ironic speaker made a positive metaphoric remark facilitated detec­
tion of ironic intent.
In addition, results of the free recall task indicated that statements made
by members of high-irony occupations were remembered more accurately
than statements made by members of high-metaphor occupations. We inter­
preted these memory data to mean that occupation was coded in the com­
prehension process. Left unresolved, however, was the issue of when in the
comprehension process occupation information was activated and inte­
grated. Was this an early or a late effect? Relatedly, in the Katz and Pexman
(1997) study participants were required to make overt ratings of speaker in­
tent, and it was possible that the occupation effects observed might be arti­
facts of this methodology.
In a subsequent study we investigated whether speaker occupation is ac­
tivated and integrated in the comprehension process when participants are
not required to make explicit decisions about speaker intent (Pexman,
Ferretti, & Katz, 2000). We used the same stimuli as in Katz and Pexman
(1997, Study 1), but we adopted a self-paced, moving windows reading task
to assess processing. In the moving windows reading task stimuli were pre­
sented on a computer screen. Because participants advanced the moving
window themselves, it was possible to measure reading time (RT) for each
214 PEXMAN

word in each context paragraph and each target statement. In moving win­
dows experiments, processing for the target statement sometimes continues
even after the last word in the statement has been read. To capture any
spillover effects of this type, we added short wrap-up sentences to follow
each target metaphor. Although participants answered a simple, factual
comprehension question after each passage, they were never asked about
speaker intent. As such, it was assumed that the task captured typical inte­
grative processing.
Using this on-line procedure, results showed that for the last word in
the target metaphor, RTs were significantly longer when speaker occupa­
tion had been mentioned in the context paragraph. In particular, RTs were
almost 200 ms longer at the last word in the target when speaker occupa­
tion (either high-irony or high-metaphor) was mentioned for unfamiliar
target metaphors, as compared to RTs for that location when no occupa­
tion was provided in the context paragraph. There were also interactive ef­
fects of speaker occupation and context incongruity slightly further
downstream, at reading locations immediately after the target statement.
For instance, when the speaker was a member of a high-irony occupation,
and when the context suggested a nonironic interpretation, there was in­
creased processing time at locations past the end of the target statement.
These results suggested that participants coded speaker occupation infor­
mation and integrated this information with other cues at the end of the
target statement. Thus, speaker occupation information was activated
and integrated fairly early in the processing of potentially ironic remarks.
Note that participants used this cue even when they were not required to
make overt ratings of speaker intent. We argued that these results sup­
ported a context-driven, interactive model of language processing. Rele­
vant cues to ironic intent were activated quite rapidly and integrated to
derive the best possible interpretation.
The results of the Katz and Pexman (1997) and Pexman et al. (2000)
studies suggested that speaker occupation was a salient cue to ironic in­
tent, but several issues were left unresolved. The investigations had, for in­
stance, been limited to metaphoric stimuli, and the speaker occupations
were either congruent with metaphoric speech (the high-metaphor occu­
pations) or were incongruent with that type of speech (the high-irony oc­
cupations). The speaker occupation effects were attributed to this
incongruity. It is also possible, however, that speaker occupation effects
could be created not just by the incongruity of a high-irony speaker using a
metaphor but also by the occupation cueing traits associated with ironic
speakers. Those traits might cue an ironic interpretation even when the
nonironic sense of the target statement itself was not obviously out of
keeping with the speaker's conversational tendencies. This possibility was
tested in a recent study by Kara Olineck and I (Pexman & Olineck,
9. INTERPRETATION OF VERBAL IRONY 215

2002b), in which the target statements were all nonmetaphors like "you
are a wonderful friend."
In the Pexman and Olineck (2002b) study, we also addressed other unre­
solved issues with regard to speaker occupation effects. We assessed the im­
pact of speaker occupation on interpretation of ironic criticisms but, unlike
the previous Katz and Pexman (1997) and Pexman et al. (2000) studies, also
included ironic compliments (e.g., "you are a terrible friend" said when your
friend has just done something gracious). In everyday speech, ironic compli­
ments are less common than ironic criticisms (Gibbs, 2000), and they may
also be more difficult to interpret (Kreuz & Glucksberg, 1989). Because of
this additional ambiguity for ironic compliments, we hypothesized that
speaker occupation might be a particularly salient cue to ironic intent for
these remarks.
The last unresolved issue that we addressed in the Pexman and Olineck
(2002b) study was why speaker occupation cued ironic intent. Our expecta­
tion was that the occupational categories were associated with particular
traits, but which traits were relevant in the context of potentially ironic
speech?
In the Pexman and Olineck (2002b) study we conducted three experi­
ments. The first two were ratings studies. In Experiment 1, participants were
presented with short context paragraphs followed by positive or negative tar­
get statements like "you are a wonderful friend" or "you are a terrible friend."
The context paragraphs described either a positive event or a negative event.
In addition, these contexts were either strong (very positive or very negative)
or weak (mildly positive or mildly negative). Finally, speaker occupation was
manipulated. A pilot study was conducted to collect new occupation ratings.
In the previous studies (Katz & Pexman, 1997; Pexman et al., 2000) my col­
leagues and I had used high-irony (and low-metaphor) or high-metaphor
(and low-irony) occupations. In the Pexman and Olineck pilot study we asked
participants to rate a set of occupations for likelihood that members of each
occupation would use sarcasm. We suspected that participants would have
clear ideas about sarcasm use but if asked to rate irony might confuse verbal
irony with situational irony. Results showed that certain occupations were
perceived to be very likely to use sarcasm (e.g., comedian, talk show host,
movie critic, and cab driver) whereas other occupations were perceived as
very unlikely to use sarcasm (e.g., accountant, doctor, clergyman, and scien­
tist) . Certainly, several of these occupations overlapped with the high-irony
and high-metaphor occupations used in the previous studies.
For each target statement, participants were asked to make the following
ratings: (a) Is the speaker being sarcastic? (b) Is the speaker saying some­
thing polite? (c) Is the speaker mocking someone? (d) How certain are you
that you correctly interpreted the speaker's intent? They also completed an
unexpected recall task.
216 PEXMAN

The results of Experiment 1 showed that speaker occupation influenced

ratings of the extent to which the speaker was mocking someone: State­
ments made by sarcastic speakers were perceived to be more mocking than
statements made by nonsarcastic speakers. This effect interacted with con­
text strength, such that the speaker occupation effect was stronger for state­
ments presented in weak (as compared to strong) contexts. Although
speaker occupation effects were observed in the mocking ratings, there were
no significant effects in ratings of politeness, certainty, or, most important,
sarcasm ratings. There were, however, effects of speaker occupation in the
recall data. Recall tended to be highest for potentially sarcastic remarks,
such as for positive statements made by sarcastic speakers, particularly when
the remark was relatively more ambiguous, as in the weak context condi­
tions or in situations in which the remark could be an ironic compliment (as
compared to situations in which the remark could be an ironic criticism).
Based on the results of Experiment 1, we concluded that speaker occupation
information was coded in the comprehension process (hence, the recall ef­
fects) and influenced ratings of speaker intent (mocking ratings) but did not
add significantly to perceived level of sarcasm. We hypothesized that the
null occupation effects for sarcasm ratings could be explained by the pres­
ence of other cues: The contexts provided information about the positive or
negative outcome of events, and the congruity or incongruity of this infor­
mation with the obvious positive or negative tone of the target statements
could have been a powerful (and sufficient) cue to ironic intent. Thus,
speaker occupation, although coded in the comprehension process, would
have been a redundant cue for level of perceived sarcasm.
In Experiment 2, we tested this hypothesis by modifying the context in­
formation. We removed all information in the context passages about
whether the outcome of events had been positive or negative and thus cre­
ated a neutral context condition. All target statements (both positive and
negative) were presented in these neutral contexts. Speaker occupation in­
formation was, however, still provided. Under these conditions, when con­
text incongruity was not signaled, there was a significant speaker occupation
effect: Both positive and negative target statements were perceived to be
more sarcastic when the speaker was a member of a sarcastic occupation.
Speaker occupation effects were also observed for mocking ratings, polite­
ness ratings, and certainty ratings. That is, participants were more certain
that they had correctly interpreted speaker intent when the speaker was
identified as a member of a sarcastic occupation.
These findings suggested that speaker occupation cues speaker intent,
even for nonmetaphoric statements, particularly when other cues to speaker
intent (e.g., context incongruity) are absent. The results of the previous
studies (e.g., Katz & Pexman, 1997) had been taken to mean that speaker
occupation effects are generated by incongruity between a speaker's per­
9. INTERPRETATION OF VEREBAL IRONY 2 17

ceived conversational tendencies (i.e., tendency not to use metaphor) and

the type of utterance made (i.e., a metaphor). The more recent findings
(Pexman & Olineck, 2002b) suggested that speaker occupation effects may
also be generated by activating certain traits for the speaker, based on the oc­
cupational stereotype, that lead the listener to expect an ironic remark (or
not). This type of effect would influence interpretation of many different
types of utterances, even familiar, nonmetaphoric remarks, as was the case
in the present study.
What are the traits that are perceived to predict ironic intent? To address
this question, we asked participants to rate a set of occupations for particular
traits (Pexman & Olineck, 2002b, Experiment 3). As noted, previous re­
search had established that several traits are attributed to speakers who
make ironic remarks, for instance, tendency to be humorous (e.g., Kreuz et
al., 1991), tendency to criticize (e.g., Colston, 1997; Kreuz et al., 1991), ten­
dency to be aggressive (e.g., Toplak & Katz, 2000), and social status or edu­
cation level (Katz & Pexman, 1997). In addition, certain dimensions seem
to be negatively associated with ironic speech, such as tendency to be sin­
cere (e.g., Clark & Gerrig, 1984; Haverkate, 1990; Kumon-Nakamura et al.,
1995) and tendency to have close relationships (Slugoski & Turnbull,
1988). We asked participants in Experiment 3 to rate 45 occupations on
each of these dimensions. Recall that in the pilot study, we had asked a dif­
ferent set of participants to rate the same 45 occupations for likelihood of us­
ing sarcasm. In a regression analysis, we considered whether any of the trait
ratings (collected in Experiment 3) predicted the sarcasm ratings (collected
in the pilot study). Results showed that four of the trait dimensions had sig­
nificant, unique relationships with sarcasm ratings: These were tendency to
be humorous, tendency to criticize, tendency to be sincere, and education
level. Of these four predictors, tendency to be humorous accounted for the
largest proportion of variance in sarcasm ratings (28%). With all four of
these predictors, the regression equation had an R2 value of .74. The impli­
cation is that 74% of the variance in sarcasm ratings was explained by the
combination of humor, criticism, sincerity, and education level. As such, we
argued that these variables capture many of the critical components of oc­
cupational stereotypes in the context of potentially ironic speech. Certainly,
there are other important factors such as stereotype strength and consis­
tency that have not been captured in this analysis. Nonetheless, it seems
reasonable to conclude that members of occupations that are associated
with sarcastic speech are perceived to be funnier, more critical, less sincere,
and to have lower education levels. This information, when integrated with
the statement and other aspects of context, shapes perceptions of speaker
intent. The on-line results reported in Pexman et al. (2000) suggested that
this integration process is not merely a function of being asked to make ex­
plicit decisions about speaker intent. In the on-line study, in which partici­
2 18 PEXMAN

pants were not required to make an explicit decision about speaker intent,
there was evidence that participants integrated speaker occupation infor­
mation relatively early in processing—as they read the last word in the tar­
get statement. It seems likely that the same type of processing occurred in
this study.

Speaker Personality Traits

In the literature on speaker occupation effects (already reviewed), it is as­

sumed that speaker traits are inferred from occupation stereotypes. It seems
highly plausible that direct knowledge of a speaker's personality traits could
also cue speaker intent. To my knowledge, this has only been tested in one
study. This study was conducted in my laboratory and the participants were
school-age children.
The developmental literature suggests that children begin to understand
verbal irony around age 6 (e.g., Dews et al., 1996; Hancock, Dunham, &
Purdy, 2000). The developmental literature also suggests that children be­
gin to understand personality traits and to use those trait concepts to make
nonobvious inferences even earlier, as young as age 4 (Heyman & Gelman,
1999). These ages mark early understanding of irony and personality traits,
with children's concepts of verbal irony and traits becoming more complex
as they mature.
In a recent study, my coauthors and I (Harris, Ivanko, Jungen, Hala, &
Pexman, 2001) explored children's abilities to use speaker personality traits
as cues to ironic intent. We investigated whether consistent personality trait
information (a mean speaker making an ironic criticism) would facilitate de­
tection of ironic intent and how children would interpret inconsistent per­
sonality trait information (a nice speaker making an ironic criticism). Literal
compliments and ironic criticisms were presented to children in the context
of short puppet shows with prerecorded narrative. Intonation was provided
such that ironic criticisms were delivered with a slightly scornful tone of
voice, whereas literal compliments were delivered with a more positive, ear­
nest tone.
Before combining personality trait information with potentially ironic re­
marks, we first conducted a control experiment (Experiment 1) in which we
tested 5- and 6-year-old children's understanding of personality traits sepa­
rately from their understanding of potentially ironic remarks. In Experiment
1, participants were first asked to predict helping and sharing behavior of the
mean and nice characters based on personality information alone. With a
high degree of accuracy, the children predicted that the nice characters
would be likely to help and share and that the mean characters would not.
Separately, the children were also presented with the contexts and poten­
tially ironic statements and were asked three questions devised to tap the
9. INTERPRETATION OF VERBAL IRONY 219

children's understanding of speaker belief ("When Betty said 'you are so

careful,' did she think Alex was careful or not careful?") and speaker intent
("When Betty said 'you are so careful,' was what she said nice or mean?" and
"Now look at the faces and show me how nice or mean Betty was being").
The face scale depicted five faces, with expressions ranging from nice
(broadly smiling face) to mean (scowling face).
Results of Experiment 1 showed that for statements presented in the posi­
tive contexts (literal compliments) the children responded correctly to the
belief question in every case (100% correct) and were only slightly less accu­
rate in their responses to the intent question (97% correct). For responses to
statements presented in the negative contexts (ironic criticisms), the chil­
dren were much less accurate on the belief question (55% correct), and they
tended to be even less accurate in their responses to the intent question
(45% correct). This difference in accuracy for speaker belief (attitude) and
intent has been observed in other developmental studies (e.g., Ackerman,
1981, 1983; Hancock et al, 2000) and has been taken as evidence that de­
tection of speaker attitude is a prerequisite for detection of speaker intent
(Winner & Leekam, 1991).
We then combined the personality trait information with the contexts
and statements in Experiments 2 and 3. The participants in Experiment 2
were again 5- and 6-year-old children. Speaker trait information did not
influence performance on the speaker belief question, but it did affect ac­
curacy on the speaker intent question. That is, in the negative context sit­
uations, in which the remark was intended to be an ironic criticism, the
children tended to assess speaker intent more accurately when the speaker
was a mean person (consistent trait information, 69% correct) than when
the speaker was a nice person (inconsistent trait information, 50% cor­
rect) . The same type of difference was observed in the positive context sit­
uations, in which the remark was intended to be a literal compliment: The
children tended to assess speaker intent more accurately when the speaker
was a nice person (consistent trait information, 88% correct) than when
the speaker was a mean person (inconsistent trait information, 75%).
These results suggested that young children used information about
speaker personality traits to modify their interpretations of speaker intent
for potentially ironic remarks.
The same procedure was used with older children (7- and 8-year-olds) in
Experiment 3. These older children were generally more accurate in their
interpretations of speaker belief and speaker intent. Their responses to the
first intent question did not seem to be modified by speaker trait information
(for literal compliments, 100% correct with nice speaker, 94% correct with
mean speaker; for ironic criticisms, 87% correct with mean speaker, 81%
correct with nice speaker). There was evidence, however, that the older
children integrated the speaker trait information with the statements in
22O PEXMAN

their responses to the second intent question (face scale). Here, they rated
ironic criticisms as somewhat less mean when the speaker was labeled as nice
(as compared to their ratings for ironic criticisms made by mean speakers).
In contrast, they considered all literal compliments to be very nice regard­
less of speaker traits.
There is an interesting developmental story in these findings. Younger
children relied more heavily on trait information, and it biased their inter­
pretations of both ironic and literal utterances. Older children had a stron­
ger grasp on ironic intent and a more sophisticated understanding of
personality traits. The older children seemed to realize that personality
traits did not predict the essential intent behind ironic utterances (a person
could be nice and still make a mean remark) but that the traits could be
taken into consideration in determining how mean a remark is intended to
be. Perhaps most surprising is the fact that both groups of children were able
to deal with inconsistent or conflicting information and make decisions
about speaker intent in very ambiguous contexts.
We think that this developmental work has some interesting implications
for the issue of teasing and bullying. Sarcasm is often part of teasing behavior
(Keltner, Capps, Kring, Young, & Heerey, 2001) because it allows a bully to
simultaneously entertain bystanders and mock the target. Harris et al.
(2001) suggested that children's detection of sarcasm is enhanced when
they are told that the speaker is a mean person. This enhancement probably
occurs because the trait provides the child with valuable information about
the speaker's attitude and motive. The implication, albeit indirect, is that
children's understanding of bullying behavior might be helped by explicit in­
formation about the motive and attitudes underlying that behavior.

Gender of the Speaker

The teasing literature has also provided some conflicting claims about gen­
der differences and teasing. In research in which irony was considered to be
one form of teasing, Lampert (1996) reported that men were more likely to
tease than women. In contrast, Keltner et al. (2001) concluded that there
was only weak support for gender differences in extent of teasing and found
no evidence of gender differences in the style or manner of teasing. Gibbs
(2000) reported that men were more likely to use sarcastic irony in conver­
sation with friends.
Katz, Piasecka, and Toplak (2001) investigated whether gender, as a so­
cial category, could cue a speaker's tendency to use ironic language (or not).
In a ratings task, Katz et al. found that men were perceived to be more sar­
castic than women. Also, in an on-line reading task, there was evidence that
gender of the speaker and gender of the addressee were processed as the sar­
castic comment was read. As in the Pexman et al. (2000) study, the on-line
9. INTERPRETATION OF VERBAL IRONY 221

data suggested that this social cue was integrated quite early in process-
ing—as participants read the last word in the sarcastic statement.
The research just described suggests that speaker characteristics are
coded in the comprehension process for verbal irony by adults and by chil­
dren, and those characteristics can act as cues to speaker intent. These find­
ings suggest that listeners perceive speaker intent to be a function of
interacting situational and personality factors. As such, the participants in
these experiments seem to share the view of some researchers in the field of
communication—that personality traits are a significant predictor of com­
munication behavior (e.g., McCroskey, Daly, Martin, & Beatty, 1998). Ac­
cording to this view, there should be individual differences in production of
verbal irony: Some individuals should tend to use verbal irony more fre­
quently than others. I consider this possibility next.

INDIVIDUAL DIFFERENCES

In a recent study in my laboratory (Ivanko, Pexman, & Olineck, 2003), we

investigated individual differences and verbal irony. In particular, we tested
the view that some individuals tend to use verbal irony more frequently than
others. We also investigated the possibility that individual differences in
irony production might be related to irony comprehension; that is, if a per­
son tends to use irony quite often in their own speech, then they might be
more likely to detect it in the speech of others. The issue of individual differ­
ences has received very little research attention in the irony literature.
In one of the few previous studies related to this issue, Jorgensen (1996)
investigated gender differences in emotional reactions to verbal irony. She
reported that men were more likely than women to perceive humor in sar­
castic irony (Experiment 2). Also, Jorgensen found that women were more
likely than men to be offended or angered by sarcastic remarks (Experi­
ment 3). Given these findings, it seemed possible that gender might be one
of the individual differences related to production and interpretation of
verbal irony.
Another relevant characteristic might be captured by conversational in­
directness. Holtgraves (1997) investigated individual differences in the pro­
duction and comprehension of indirect requests (e.g., saying "it's cold in
here" as a request for someone to close the door). He devised and validated
the Conversational Indirectness Scale (CIS) to measure the extent to which
individuals differ in their tendencies to express themselves indirectly and
understand indirect meanings. The scale involves items that were designed
to tap both interpretation (10 items) and production (9 items) of conversa­
tional indirectness. To the extent that verbal irony is an indirect form of
speech, it seemed possible that the CIS might capture individual differences
in production and comprehension of verbal irony.
222 PEXMAN

To assess individual differences in use of verbal irony we devised a self-re-

port measure in which participants were asked to rate their own tendencies
to speak sarcastically in general (e.g., "How sarcastic are you?" and "How
sarcastic would your friends say you are?") and in specific contexts (e.g.,
"How likely are you to make a sarcastic remark in an argument with your
roommate over household chores?"). A large group of participants (n =
251) completed this Sarcasm Self-Report Scale (SSS), and a principal com­
ponents analysis of their responses produced four components. These com­
ponents implicated several different functions for sarcasm: to gently criticize
a close friend (general sarcasm component), to save face when delivering a
compliment (face-saving component), to diffuse potential embarrassment
in a very positive situation (embarrassment diffusion component, e.g., "just
got engaged and telling friends" and "score winning point in final basketball
game of the season"), or to diffuse frustration in a very negative situation
(frustration diffusion component, e.g., "mile-long line at the grocery store"
and "friend just locked your keys in car"). We also found significant gender
differences for responses on the SSS, with male participants generally giving
higher self-reports of sarcasm use than female participants. That is, male
participants had higher scores than female participants for three of the SSS
components (general sarcasm, face saving, and embarrassment diffusion);
no gender difference was found for the fourth component (frustration diffu­
sion) .
It has been argued that men and women have different motives for using
self-directed humor (Lampert, 1996): Men tend to use self-direct humor "in a
self-protective manner to reduce social vulnerability, whereas for women, it
worked more to increase social vulnerability and promote intimacy" (p. 585).
Our findings seem consistent with these claims, because male participants re­
ported that they were more likely to use sarcasm when it would decrease vul­
nerability (through face saving or embarrassment diffusion) but not when
sarcasm would not decrease vulnerability (frustration diffusion). Explana­
tions for these gender differences likely involve complex sociocultural factors
(e.g., Tannen, 1993) beyond the scope of the present work.
Participants in the Ivanko et al. (2002) study also completed a produc­
tion task (role playing) in which several situations were described. Half of
the situations involved the speaker and a new acquaintance, and half of the
situations involved the speaker and a best friend. Participants were asked to
choose (from four options, including both literal and ironic utterances) the
statement they would most likely make in each situation. We found that
statement choices were predicted, to some degree, by SSS scores. In particu­
lar, individual scores for the face-saving component of the SSS were a signif­
icant predictor of statement choices for a situation involving a new
acquaintance, and individual scores for the general sarcasm component
were a significant predictor of statement choices for a situation involving a
9. INTERPRETATION OF VERBAL IRONY 223

best friend. These results provided some validation for the SSS, because
self-rated conversational tendencies were related to predicted communica­
tion behavior. These results also suggest that there are, indeed, individual
differences in tendency to use sarcastic irony. We next tested whether those
differences were related to differences in detection and interpretation of
verbal irony.
Participants completed an interpretation task in which potentially ironic
statements were presented. Participants were asked to rate various aspects
of speaker intent: sarcasm, mocking, humor, and politeness. We examined
whether any of the measured individual difference variables (SSS scores,
CIS scores, and gender) predicted ratings on the interpretation task. Note
that none of these individual difference variables predicted sarcasm ratings
for ironic insults (e.g., saying "That sounds pretty exciting" in a conversation
about a boring date). For ironic compliments (e.g., saying "That sounds
pretty dull" in a conversation about an exciting date), however, gender was a
significant predictor of perceived level of sarcasm, with women tending to
perceive these statements as more sarcastic than men. We speculated that
the gender difference observed in sarcasm ratings for ironic compliments
was produced because the term sarcasm has a negative connotation and
women's higher ratings to this dimension may reflect sensitivity to the nega­
tive tinge inherent in ironic compliments (e.g., Dews & Winner, 1995;
Pexman & Olineck, 2002a). This conclusion is supported by the finding
that women also rated ironic compliments as less polite than did men.
In addition, although the SSS and CIS did not predict sarcasm ratings in
the interpretation task, they did predict ratings of the social impact of these
statements. For instance, the perceived politeness of ironic criticisms was
predicted, to some degree, by individuals' CIS-production scores (higher
scores were associated with higher politeness ratings) and by their
SSS-face-saving scores (a tendency to use sarcasm to save face was associ­
ated with lower politeness ratings). Thus, participants who rated themselves
as likely to use indirect speech were less likely to perceive offense in ironic
criticisms. Perhaps these participants were relatively more sensitive to the
tinge function of ironic criticisms. In contrast, participants who rated them­
selves as likely to use sarcasm with new acquaintances or when compliment­
ing others (face-saving component of the SSS) may have been more
sensitive to the critical intent underlying these statements.
Ivanko et al.'s (2003) study also involved a processing task. This was a
self-paced moving windows reading task in which literal and ironic state­
ments were presented following short context paragraphs. We found that
relative RTs for literal and ironic statements could be predicted by several
individual difference variables. Participants who rated themselves as being
more likely to use sarcasm in frustrating situations (SSS-frustration diffu­
sion) or with new acquaintances (SSS-face saving) or to look for or use indi­
224 PEXMAN

rect speech (CIS-production and CIS-interpretation) tended to read the

ironic statements faster than or in equivalent time to the literal statements.
In contrast, older participants tended to read the ironic statements more
slowly than the literal statements.
These results suggest that there are individual differences in verbal irony.
These include differences in production, with some individuals tending to
use sarcastic irony more than others. These differences in production de­
pend, to some degree, on the situation. For instance, some individuals re­
ported that they tend to use sarcasm with best friends but not with new
acquaintances. Some use sarcasm in very positive situations, and others use
sarcasm in very negative situations. These individual differences in produc­
tion do have consequences for interpretation: influencing the perceived so­
cial impact of verbal irony and influencing the speed with which irony can
be processed relative to literal language.

SUMMARY OF PREVIOUS FINDINGS

The research summarized in this chapter points to several conclusions.

There is clear evidence that readers use social cues in the process of inferring
speaker intent. They do this in laboratory tasks in which they are asked to
make judgments of speaker intent, and they also do this in tasks in which no
explicit judgments of speaker intent are required. They access these cues
rapidly and integrate them with other available information in order to ar­
rive at a coherent representation of the situation described. Although the
evidence suggests that these cues are always coded in the comprehension
process, they seem to influence judgments of ironic intent primarily when
other verbal cues are minimal. One might wonder to what extent these so­
cial cues would be accessed in everyday interpersonal communication, in
which the context is assumed to be richer, involving both verbal and non­
verbal cues. On that issue, I first argue that there are many everyday situa­
tions in which verbal and nonverbal cues to speaker intent are few, as in
e-mail and telephone conversations. Further, even in these everyday con­
texts in which nonverbal cues are absent, people still choose to speak ironi­
cally. In fact, Hancock and Dunham (2001) compared irony use in
face-to-face conversations to irony use in computer-mediated communica­
tion and found that irony was actually produced more often in the com-
puter-mediated context. In addition, despite the missing nonverbal cues,
comprehension was just as good in the computer-mediated context as it was
in the face-to-face context. The fact that participants in the studies re­
ported here so readily used social cues to comprehend speaker intent even
when not required to do so suggests that this is the normal procedure: I ar­
gue that these social cues are noted and considered in the context of every­
day conversation.
9. INTERPRETATION OF VERBAL IRONY 225

This research also suggests that young school-age children are building
theories of how personality traits are related to communicative intent. The
data suggested that older children (7- and 8-year-olds) are beginning to un­
derstand that traits are not perfect predictors of behavior but that they are
still relevant to deciding how mean a speaker intended to be in making an
ironic criticism. Adult participants seem to have a similar perception:
Speaker characteristics are relevant to decisions about, for instance, how
mocking a speaker intends to be, but they are not the most reliable cues to
irony. Speaker characteristics are, however, sufficiently reliable cues to irony
to be relevant when the situation is relatively ambiguous, either because the
statement is very unfamiliar or because the context is neutral with respect to
outcome.

FITTING SOCIAL CUES AND INDIVIDUAL DIFFERENCES

WITH THEORIES OF VERBAL IRONY

In the research described here it was demonstrated that social factors influ­
ence interpretation of ironic intent. These include characteristics of the
speaker and characteristics of the listener. An important issue is the extent
to which theories of verbal irony interpretation can account for such effects.
Theories of irony do not make explicit predictions about the role of social
cues, such as gender and speaker occupation, or about individual differences
but for present purposes I attempted to derive predictions based on the gen­
eral assumptions of each theory.
The echoic mention theory (Sperber & Wilson, 1981, 1986) holds that
irony is made possible by an utterance that implicitly or explicitly echoes a
previous event, expectation, or social norm. According to the echoic re­
minder theory (e.g., Kreuz & Glucksberg, 1989) the echoic nature of an
ironic utterance reminds the listener of a failed expectation or violated so­
cial norm. As such, the ironic utterance expresses the speaker's negative at­
titude about the situation. According to these echoic theories, detection of
irony should depend on the presence of an implicit or explicit antecedent. It
is not obvious why speaker characteristics should influence the echoic as­
pect of ironic statements. One possibility, however, is that speaker charac­
teristics cue particular expectations on the part of the speaker, and those
expectations could be echoed in the statement. If that were the case, one
would expect that ratings of irony would be higher when speakers were per­
ceived to have strong expectations. These speakers would likely be the
members of groups who are highly aware of social convention. Perhaps occu­
pations like clergyman and doctor would fit this description. Yet, research
shows these were actually the occupations that tended to generate lower
ratings of irony (Katz & Pexman, 1997; Pexman et al, 2000; Pexman &
Olineck, 2002b). The occupations that produced higher irony ratings (e.g.,
226 PEXMAN

comedian and cab driver) do not seem candidates for highly conventional
social expectations. The echoic theories would seem to have a difficult time
explaining effects of social cues.
Echoic theories do, however, seem capable of explaining individual dif­
ferences in processing. One could speculate that some individuals attend
more than others to failed expectations. This type of difference could be re­
lated to a pessimistic or cynical outlook. This difference in outlook could
mean those individuals have stronger expectations for an ironic statement;
consequently, they process ironic statements more readily.
According to the pretense models (e.g., Clark & Gerrig, 1984), an ironic
speaker pretends to be a very optimistic person, addressing an imaginary lis­
tener and assuming that the listener would interpret the utterance literally.
Consequently, the ironic speaker expresses a negative attitude toward the
situation, the imaginary listener, and the pretended optimist. Perhaps this
theory would predict effects of social cues in terms of the consistency of the
cues with this type of behavior. For instance, if a speaker's characteristics
were consistent with jocularity or with insincerity, then a listener might be
more likely to detect the speaker's pretense. This possibility is supported by
our finding that humor and insincerity were two traits associated with the
sarcastic occupation stereotype (Pexman & Olineck, 2002b). Similarly, if
the listener was a funny and/or insincere person, then he or she might be
more apt to detect the speaker's pretense.
The allusional pretense theory (Kumon-Nakamura et al., 1995) involves
two main arguments: (a) ironic utterances allude to a failed expectation ei­
ther by implicit or by explicit echo and (b) ironic utterances involve prag­
matic insincerity (e.g., overpoliteness). As with pretense theory, it seems
possible that social cues and individual differences affect perception of
ironic intent by enhancing (or attenuating) the perception of pragmatic in­
sincerity. Thus, there may be a mechanism by which this model predicts an
influence for social cues in the perception of ironic intent.
Recently, Utsumi (2000) argued that none of these theories provides a
complete account of verbal irony. Although echoic mention and pretense
maybe important factors in some instances of verbal irony, they are not nec­
essary or sufficient conditions for irony (see also Colston, 2000). As an alter­
native, Utsumi proposed the implicit display theory of verbal irony, which
involves three main assumptions. First, an ironic utterance assumes an
"ironic environment" defined as "a situational setting which motivates ver­
bal irony" (p. 1778). Second, an ironic utterance implicitly displays the
ironic environment. Third, irony is characterized as a prototype-based cate­
gory. As such, utterances with relatively more characteristics of implicit dis­
plays will be perceived as more ironic.
In elaborating on the notion of ironic environment, Utsumi (2000) speci­
fied that this environment involves a speaker having an expectation that is
9. INTERPRETATION OF VERBAL IRONY 227

not met and having a negative emotional attitude toward the incongruity
between expectation and outcome. Utsumi also noted that the speaker's
negative attitude could be implicitly displayed by a host of cues, including
verbal cues such as intonation and exaggeration and nonverbal cues such as
gesture and facial expression. It seems possible that these cues might also in'
elude beliefs about the speaker. If the speaker's characteristics help to con­
vey his or her negative attitude about the situation, then those
characteristics might enhance detection of irony. This implicit display ac­
count holds that irony is understood more readily when the utterance is
more similar to prototypical verbal irony. In other words, irony is subject to
typicality effects. To the extent that speaker characteristics cue features in­
volved in prototypical irony, they may make the utterance seem more similar
to the prototype and thus enhance detection.
Therefore, it seems possible for implicit display theory to predict that social
cues will influence interpretation of irony. It also seems that the model would
have specific predictions about what types of characteristics should act as cues
to irony. That is one would predict that effective cues would include charac­
teristics that suggest a negative attitude, such as cynicism or a critical outlook.
Further, the theory claims that the negative attitude must be conveyed indi­
rectly, and so it would predict that characteristics such as insincerity and, per­
haps, humor would also be effective cues. These predictions seem consistent
with the results reported by Pexman and Olineck (2002b): Sarcastic speakers
were perceived to be humorous, critical, insincere, and to have lower educa­
tion or social status. The implicit display theory may even be able to explain
this last, somewhat surprising association between sarcastic speech and lower
education levels. Lower education levels (strongly correlated with lower so­
cial status) may signal a need to express criticism indirectly to lessen the risk of
offending higher status others. Because the implicit display theory emphasizes
indirect expression of negative attitude, it seems capable of explaining this re­
sult. As such, implicit display theory seems able to explain effects of social cues
on interpretation of verbal irony.
The implicit display theory also gains a great deal of flexibility from the as­
sumption that irony is a mental category that is subject to typicality effects.
Through this assumption, the theory seems to be able to explain individual
differences of the type we observed in Ivanko et al. (2003). The category of
verbal irony could show effects of expertise, as other categories do (e.g.,
Tanaka & Taylor, 1991). Some individuals may have more experience with
ironic utterances by virtue of using those utterances quite frequently. Con­
sequently, they could have more knowledge of the types of situations in
which those statements can be used and of the types of reactions with which
those statements are met. This enhanced mental representation for verbal
irony could facilitate processing of ironic statements relative to literal state­
ments, as we observed.
228 PEXMAN

Utsumi's (2000) theory is a good fit for the data reported in this chapter
because it allows multiple sources of information to be considered in the
process of interpretation and because it construes verbal irony as a mental
category. Similar characteristics are described by Katz and Ferretti (2001)
for the constraint satisfaction view of language processing. With this view,
comprehension is achieved by a mechanism that taps all of the sources of in­
formation available to the comprehender. Different sources of information
(lexical, conceptual, etc.) compete for activation. The interaction of these
constraints provides probabilistic evidence for different interpretations un­
til one alternative wins. The findings reviewed here suggest that these con­
straints also include characteristics of the speaker and characteristics of the
listener. The constraint satisfaction view allows constraints to have different
weights as a function of their reliability. That is, incongruity between the
tone of the context situation and the tone of the statement is a strong and re­
liable cue to ironic intent. Speaker occupation is a less reliable cue (scien­
tists can be sarcastic, but their perceived tendency is to be nonsarcastic).
Thus, incongruity between context and statement can constrain interpreta­
tion so rapidly that less reliable or less strongly weighted constraints like
speaker occupation or speaker gender, although coded by the compre­
hender, show little influence on interpretation. Similarly, my hunch is that
individual differences constrain interpretation only when other constraints
(like context and statement incongruity or lexicalized statement meaning)
are weak. This hunch has yet to be tested.

Social Cues and Other Forms of Figurative Language

I argued that social cues and individual differences constrain the process of
comprehending ironic speech. I suggest that this reflects the seamless inte­
gration of social, linguistic, and conceptual information that normally oc­
curs in the course of language comprehension. Because the notion of
constraint satisfaction is relevant to language comprehension in general,
there seems little reason to believe that the influence of social cues would be
unique to irony. Other forms of figurative language likely have distinct social
functions, although those functions have received little research attention
thus far (cf. previous research on indirect requests, e.g., Holtgraves, 1994,
1997, 1998). For instance, proverb (e.g., "strike while the iron is hot") and
metaphor (e.g., "children are precious gems") can be used to convey wis­
dom, although they can certainly be used ironically, perhaps to mock those
who would use them to convey wisdom. Idioms (e.g., "kick the bucket") al­
low us to communicate with others in our culture, but they seem to provide a
means of excluding those with less shared knowledge. Metonymy (e.g., "the
ham sandwich in the corner wants his bill") may also function as a type of
cultural shorthand, allowing speakers to present themselves as witty and
9. INTERPRETATION OF VERBAL IRONY 229

quick thinking. If these types of figurative language serve specific social

functions, then inferences of speaker intent may be influenced by the extent
to which social cues are consistent (or inconsistent) with this function.
There may also be widely held beliefs about the characteristics of people
who use these forms of figurative language; however, as noted, there has
been little research on this issue. For instance, it has been claimed that prov­
erbs are particularly appropriate for conversation with older adults (Jack­
son, 1994). The implication is that there may be generational (cohort)
effects in use and comprehension of proverbial statements. If so, speaker age
could act as a constraint on proverb interpretation. Similarly, there may be a
perception that speakers who offer metonymic statements tend to be cynical
and humorous. Knowledge of these speaker characteristics could constrain
interpretation of metonymy. As such, I predict that social cues constrain fig­
urative language comprehension in general, but that different types of figu­
rative language are cued by different social variables.
Gibbs (1999) suggested that in understanding language (and other hu­
man actions) we have a strong tendency to make inferences about
intentionality. To the extent that social cues are tapped in the cognitive pro­
cess of inferring intention, they should be relevant in almost every linguistic
context. Research suggests that listeners consider social cues, such as char­
acteristics of the speaker, to be relevant to inferences about intent. Thus, lis­
teners have an implicit belief that individual traits and membership in
particular social categories influence communicative behavior. The re­
search summarized in this chapter provides evidence to support this implicit
belief: Individual characteristics influence both interpretation and produc­
tion of verbal irony.
REFERENCES

Ackerman, B. P. (1981). Young children's understanding of a speaker's intentional

use of a false utterance. Developmental Psychology, 17, 472–480.
Ackerman, B. P. (1983). Form and function in children's understanding of ironic ut­
terances. journal of Experimental Child Psychology, 35, 487-508.
Clark, H. H., & Gerrig, R. J. (1984). On the pretense theory of irony.Journal of Ex­
perimental Psychology: General, 113, 121-126.
Colston, H. L. (1997). Salting a wound or sugaringa pill: The pragmatic functions of
ironic criticisms. Discourse Processes, 23, 24-45.
Colston, H. L. (2000). On necessary conditions for verbal irony comprehension.
Pragmatics and Cognition, 8, 277–324.
Colston, H. L., & O'Brien, J. (2000a). Contrast of kind versus contrast of magni­
tude: The pragmatic accomplishments of irony and hyperbole. Discourse Pro­
cesses, 30, 179–199.
Colston, H. L., & O'Brien, J. (2000b). Contrast and pragmatics in figurative language:
Anything understatement can do, irony can do better. Journal of Pragmatics, 32,
1557-1583.
23O PEXMAN

Dews, S., Kaplan, J., & Winner, E. (1995). Why not say it directly? The social func­
tions of irony. Discourse Processes, 19, 347–367.
Dews, S., & Winner, E. (1995). Muting the meaning: A social function of irony.
Metaphor and Symbolic Activity, 10, 3–19.
Dews, S., Winner, E., Kaplan, J., Rosenblatt, E., Hunt, M., Lim, K., McGovern, A.,
Quaker, A., & Smarsh, B. (1996). Children's understanding of the meaning and
functions of verbal irony. Child Development, 67, 3071–3085.
Gerrig, R. J., & Goldvarg, Y. (2000). Additive effects in the perception of sarcasm:
Situational disparity and echoic mention. Metaphor and Symbolic Activity, 15,
197-208.
Gibbs, R. W, Jr. (1986). On the psycholinguistics of sarcasm.Journal of Experimental
Psychology: General, 115, 3–15.
Gibbs, R. W, Jr. (1999). Intentions in the experience of meaning. Cambridge, England:
Cambridge University Press.
Gibbs, R. W, Jr. (2000). Irony in talk among friends. Metaphor and Symbolic Activity,
15, 5–27.
Hancock, J. T., & Dunham, P J. (2001, November). Irony use inface-to-face and com-
puter-mediated interactions. Poster presented at the 42nd annual meeting of the
Psychonomic Society, Orlando, FL.
Hancock, J. T., Dunham, PJ., & Purdy, K. (2000). Children's comprehension of crit­
ical and complimentary forms of verbal irony. journal of Cognition and Develop­
ment, I, 227–248.
Harris, M., Ivanko, S., Jungen, S., Hala, S., & Pexman, P (2001, October). You're re­
ally nice: Children's understanding of sarcasm and personality traits. Poster presented
at the second biennial meeting of the Cognitive Development Society, Virginia
Beach, VA.
Haverkate, H. (1990). A speech act analysis of irony. Journal of Pragmatics, 14,
77-109.
Heyman, G. D., & Gelman, S. A. (1999). The use of trait labels in making psycho­
logical inferences. Child Development, 70, 604-619.
Holtgraves, T. (1994). Communication in context: Effects of speaker status on the
comprehension of indirect requests.Journal of Experimental Psychology: Learning,
Memory, and Cognition, 20, 1205-1218.
Holtgraves, T. (1997). Styles of language use: Individual and cultural variability in
conversational indirectness. Journal of Personality and Social Psychology, 73,
624-637.
Holtgraves, T. (1998). Interpersonal Foundations of Conversational Indirectness.
In S. R. Fessell & R. J. Kreuz (Eds.),Social and cognitive approaches to interpersonal
communication (pp. 71–89). Mahwah, NJ: Lawrence Erlbaum Associates.
Ivanko, S. L., Pexman, P M., & Olineck, K. M. (2003, July). How sarcastic are you?:
Individual differences in the interpretation, production, and processing of verbal irony.
Poster presented at the 8th International Pragmatics Conference, Toronto, ON.
Jackson, V. R. (1994). Proverbs: A tool for work with older persons. Activities, Adap­
tation, and Aging, 19, 5–13.
Jorgensen, J. (1996). The functions of sarcastic irony in speech. Journal of
Pragmatics, 26, 613-634.
9. INTERPRETATION OF VERBAL IRONY 231

Jorgensen, J., Miller, G. A., & Sperber, D. (1984). Test of the mention theory of irony.
Journal of Experimental Psychology: General, 113, 112–120.
Katz, A. N., & Ferretti, T. R. (2001). Moment-by-moment reading of proverbs in lit­
eral and nonliteral contexts. Metaphor and Symbol, 16, 193–221.
Katz, A. N., & Lee, C. J. (1993). The role of authorial intent in determining verbal
irony and metaphor. Metaphor and Symbolic Activity, 8, 257-279.
Katz, A. N., & Pexman, P. M. (1997). Interpreting figurative statements: Speaker
occupation can change metaphor to irony. Metaphor and Symbol, 12, 19–41.
Katz, A. N., Piasecka, I., & Toplak, M. (2001, November). Comprehending the sarcas­
tic commentsof males and females. Poster presented at the 42nd annual meeting of
the Psychonomic Society, Orlando, FL.
Keltner, D., Capps, L, Kring, A. M., Young, R. C., & Heerey, E. A. (2001). Just teasing:
A conceptual analysis and empirical review. Psychological Bulletin, 127, 229–248.
Kreuz, R. J. (1996). The use of verbal irony: Cues and constraints. In J. S. Mio 6k A.
N. Katz (Eds.),Metaphor: Implications and applications (pp. 23–38). Mahwah, NJ:
Lawrence Erlbaum Associates.
Kreuz, R. J., & Glucksberg, S. (1989). How to be sarcastic: The echoic reminder the­
ory of verbal irony. journal of Experimental Psychology: General, 118, 374–386.
Kreuz, R. J., Long, D. J., & Church, M. B. (1991). On being ironic: Pragmatic and
mnemonic implications. Metaphor and Symbolic Activity, 6, 149-162.
Kreuz, R. J., & Roberts, R. M. (1995). Two cues for verbal irony: Hyperbole and the
ironic tone of voice. Metaphor and Symbolic Activity, 10, 21–31.
Kumon-Nakamura, S., Glucksberg, S., & Brown, M. (1995). How about another
piece of pie: The allusional pretense theory of discourse irony. Journal of Experi­
mental Psychology: General, 124, 3–21.
Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago: University of Chi­
cago Press.
Lampert, M. (1996). Gender differences in conversational humor. In D. I. Slobin, J.
Gerhardt, A. Kyratzis, & J. Guo (Eds.), Social interaction, social context, and lan­
guage (pp. 579-596). Mahwah, NJ: Lawrence Erlbaum Associates.
Leggitt, J. S., & Gibbs, R. W., Jr. (2000). Emotional reactions to verbal irony. Dis­
course Processes, 29, 1-24.
McCroskey, J.C.,Daly,J. A., Martin, M. A., & Beatty, M.J. (Eds.). (1998). Commu­
nication and personality: Trait perspectives, Creskill, NJ: Hampton.
Pexman, P M., Ferretti, T. R., & Katz, A. N. (2000). Discourse factors that influence
on-line reading of metaphor and irony. Discourse Processes, 29, 201-222.
Pexman, P M., & Olineck, K. M. (2002a). Does sarcasm always sting? Investigating
the impact of ironic insults and ironic compliments. Discourse Processes, 33,
199-217.
Pexman, P M., & Olineck, K. M. (2002b). Understanding irony: How do stereo­
types cue speaker intent?Journal of Language and Social Psychology, 21, 245–274.
Roberts, R. M., & Kreuz, R. J. (1994). Why do people use figurative language? Psy­
chological Science, 5, 159–163.
Slugoski, B. R., &Turnbull, W. (1988). Cruel to be kind and kind to be cruel: Sar­
casm, banter, and social relations. Journal of Language and Social Psychology, 7,
101-121.
232 PEXMAN

Slusher, M. E, & Anderson, C. A. (1987). When reality monitoring fails: The role of
imagination in stereotype maintenance. journal of Personality and Social Psychol­
ogy, 52, 653–662.
Sperber, D., & Wilson, D. (1981). Irony and the use–mention distinction. In P. Cole
(Ed.), Radical pragmatics (pp. 296–318). New York: Academic Press.
Sperber, D., & Wilson, D. (1986). Relevance: Communication and cognition. Cam­
bridge, MA: Harvard University Press.
Tanaka, J. W, &.Taylor, M. (1991). Object categories and expertise: Is the basic level
in the eye of the beholder? Cognitive Psychology, 23, 457–482.
Tannen, D. (Ed.). (1993). Gender and conversational interaction. New York: Oxford
University Press.
Toplak, M. T, & Katz, A. N. (2000). On the uses of sarcastic irony. Journal of
Pragmatics, 32, 1467–1488.
Utsumi, A. (2000). Verbal irony as implicit display of ironic environment: Distin­
guishing ironic utterances from nonirony. Journal of Pragmatics, 32, 1777–1806.
Winner, E., & Leekam, S. (1991). Distinguishing irony from deception: Under­
standing the speaker's second–order intention. British Journal of Developmental
Psychology, 9, 257–270.
 1O

Negation

as Positivity in Disguise

Rachel Giora
Tel Aviv University

Noga Balaban
Tel Aviv University

Ofer Fein
The Academic College of Tel Aviv

Inbar Alkabets
Tel Aviv University

Once there was a poor woman who had three or four little children,
and she used to lock them up in her room when she went out to work,
to keep them safe. One day when she was going away she said, "Now,
my dears, don't let baby fall out of window, don't play with the matches,
and don't put beans up your noses." Now the children had never
dreamed of doing that last thing, but she put it into their heads, and the
minute she was gone, they ran and stuffed their naughty little noses full
of beans, just to see how it felt, and she found them all crying when she
came home.
"Did it hurt?" asked Rob, with such intense interest that his mother
hastily added a warning sequel, lest a new edition of the bean story
should appear in her own family.

233
234 GIORA ET AL.

"Very much, as I know, for when my mother told me this story, I was so
silly that I went and tried it myself ..."
—Alcott, (1962), 115–1161

INTRODUCTION: ON THE ROLE OF NEGATION

How explicit negation affects the representation in memory of negated con­

cepts has been considered in recent empirical research into the discourse
and cognitive functions of negation markers (not and no). The consensus
among psycholinguists is that a negation marker is an instruction from a
speaker to a hearer to suppress the negated information. Accordingly, a ne­
gation marker reduces the levels of activation of the negated concepts to the
extent that eventually they are no more accessible than unrelated controls
and significantly less accessible than equivalent positive concepts (Hasson,
2000; Kaup, 1997, 2001; Lea & Mulligan, 2002; MacDonald & Just, 1989;
Mayo, Schul, & Burnstein, in press, Experiment 2). Folk wisdom, however,
would have it otherwise (see the Alcott quote just cited). The belief here is
that what is negated prevails, as the following exchange, taken from the
Santa Barbara Spoken American Corpus, exemplifies:

(1) P: ... it was very clear. You know.

... She kept saying,
... prefacing everything with,
... you know, this is not a personal attack.
... This is not a personal vendetta,
B: Yeah, yeah yeah yeah.
Right Right.
P: Which tells you, that it is.
B: Yeah.
P: That's immediately what it said.
And that's what everybody perceived it.
B: Yeah.
(Du Bois, 2000, italics added).

Adopting the same attitude, the following journalistic text suggests that
it is how people conceive of negation that accounts for why, despite explicit
denials, the person discussed, Kochi Mordechay, then wife of Itsik
Mordechay (former Israeli Defense Minister convicted of sexual abuses), is
taken to affirm what she denies (that she was a battered woman):

We thank Dana Zimmerman for this citation.

10. NEGATION 235

(2) I think Kochi [Mordechay] was wrong in disputing the rumors in de­
tail, focusing on their specifics .... Sometimes people are stupid, and
occasionally they would hear Kochi talking about "Itsik"
[Mordechay] and "beatings" but they wouldn't note the connectors
"there were no beatings, the rumors are vile." Now after she had been
specific, the story has become official kind of.... (Linor, 1999).

Admittedly, even a limited and random scan of how negation is used ren­
ders the suppression hypothesis suspect. In the following example (cited and
discussed in Jefferson, 2002), negation is used to provide for a supportive
and affiliative response, following a negative turn (line 6):

(3) 1 Maggie: .hh because I(c) (.) you know I told Mother what'd
ha:ppened yesterday
2 there at the party,
3 Sorrell: [°Yeah.°]
4 Maggie: [a : : ] nduh, .hhhhh (0.2) uh you know she asked me if
it was
5 (–) because I'd had too much to dri:nk and I said no=
6 Sorrell: (–) =[N o n : : : .]
7 Maggie: — [because at the t] i:me I'd only ha:d,h you
know that drink 'n a ha:If
8 When we were going through the receiving line.
9 Sorrell: Ri:ght.

It is also quite apparent that the speaker of the following testimony does
not want us to deactivate the negated concepts:

(4) We suffer from shortage in medicines, milk for children, diapers.

There are no vegetables, no fruits, no meat and milk products. We
basically eat rice and what we grew in our yard. (Badra El–sha'ar, a
resident of Tko'a, a Palestinian village in the occupied territories, as
cited in ad by Betselem—the Israeli Center for Information of Hu­
man Rights, 2002).

Clearly, by publishing it, Betselem intended us to attend to rather than

dispense with the negated information about the elementary supplies the
Palestinians are in desperate need of as a result of the devastating destruc­
tion inflicted on them by the Israeli army and the siege and curfews that were
not lifted.
Similarly, when Naomi Klein (2002) entitled her recent book, No Space,
No Choice, No Jobs, No Logo, she by no means expected us to attenuate the
236 GIORA ET AL.

meanings of space, choice, jobs, or logo. Rather, she used these elements as ob­
jectives to be reclaimed by the people, by those entitled to them. The book
laments the expulsion of the people from these public domains and deplores
their exclusive control by corporations and governments (not, however,
without indicating subversive ways to repossess them).
The following example (about a dozen bullets fired by an Israel Defense
Forces soldier that pierced the windshield of a taxi in which the journalist,
Gideon Levy, was traveling) further confirms that explicitly negated con­
cepts are retainable (at least in the mind of the speaker):

(5) NOTHING HAPPENED

Nothing happened. Soldiers opened fire, no one was hurt. Not a
thing happened. The soldiers evacuated the bullet–riddled taxi and
its passengers from the zone of fire and no officer appeared: not to
investigate, not to take testimony, not to explain, not to apologize,
and above all not to show the soldiers that, after all, something did
happen. (Levy, 2002).

The passage (as, in fact, the article throughout) is imbued with negated
concepts. "Nothing happened" (literally "nothing didn't happen" in He­
brew) is actually an echoic (though negative) irony,2 intending us to per­
ceive that something did happen (which necessitates the retention of the
meaning of happen). Similarly, the negations that follow ("no officer ap­
peared: not to investigate, not to take testimony, not to explain, not to
apologize, and above all not to show the soldiers that, after all, something
did happen"; negations markers are in bold for convenience) do not dis­
miss the negated concepts but instead construct a set of expectations of
what should have happened. Not only do the negation markers not oblit­
erate these negated entities; in a way, they serve to bring them out and spell
out the irresponsibility and indifference of the military (on the evaluative
function of negation, see Labov, 1972). They all belong in the same class or
ad hoc category of events that should have followed this shooting
event—also in support of the retention–of–negated–concepts hypothesis
(for a similar view on how a negated concept cannot be entirely eradi­
cated, see Horn, 1989, pp. 50–51).
The following example is also illustrative of the retainability of negated
concepts:

(6) THE FACTS WERE IN ON ISRAEL'S ARABS

Two statements by former Prime Minister Ehud Barak in his testi­
mony before the Or Commission, which is investigating the events
2
On the echoic mention view of irony see Sperber and Wilson (1986/1995).
1C). NEGATION 237

of October 2000,3 should raise an eyebrow. Or rather, raise hackles.

One was, "There was no concrete intelligence assessment" of the
possibility that disturbances of these dimensions would break out,
and the other was that the reason no discussion had been held on
the issue of the Arabs of Israel was "because in any case long–term
problems would have come up in any such discussion."
[.–.]
A discussion of the question of the Arabs of Israel? Why wear our­
selves out with it when all that could come up would be only
long–term problems? Now, of all times, do we need to start dealing
with the question of discrimination against Arabs? Or the fact that
no Arab city has been established to date? Or perhaps of the NIS 4
billion that the government had allocated on paper to close the
gaps? After all, what we have before us is an acute problem—Arabs
are throwing stones at policemen and blocking roads. This is the
only problem, and there is nothing more to it. No history of failings
and no future of civil revolt. It began and ended in the month of
October. Why should the prime minister deal with a problem like
that? It's a problem for a squad commander, maximum a regional
commander. (Bar'el, 2002).

The journalist is of course ironic. When writing "This is the only problem,
and there is nothing more to it. No history of failings and no future of civil re­
volt. It began and ended in the month of October. Why should the prime
minister deal with a problem like that?," he definitely intends us to entertain
the possibility that "there is ... more to it," which necessitates its retention
(see also Giora &Fein, 1999a, 1999b; Giora, Fein, & Schwartz, 1998). In­
deed, assuming its accessibility allows the writer to immediately elaborate on
it and go into details (of "history" and "future of civil revolt"). Only when
the negated statement is retained we can make sense of the elaborations
that follow (viewing them as members of "there is ... more to it" category).
In the same way, the journalist does not intend us to reduce the possibility of
"a history of failings" and of "a future of civil revolt." Rather, the availability
of an ironic interpretation relies on the retention of these negated items.
Or take the following example from Kate Chopin's (1894/1976) The Story
of an Hour, in which negated information allows us to draw a contrastive
comparison between the heroine's reactions to the news about her hus-
band's death and other women's reaction to such news:

Following the breakout of the second Intifada (uprising) in the Palestinian Occupied
Territories in September 2000, the Israeli Palestinians demonstrated against the Israeli occu­
pation and were treated violently by the Israeli police who shot and killed 14 Arab demon­
strators. The killing of Israeli citizens is called "the events of October 2000."
238 GIORA ET AL.

(7) She did not hear the story as many women have heard the same,
with a paralyzed inability to accept its significance. She wept at
once, with sudden, wild abandonment.... (p. 198)

The use of a negated sentence ("she did not hear the story ...") highlights
the event in the foreground; it brings out and evaluates the heroine's unusual
reaction (cf. Labov, 1972). It also allows the specification of how women go or
should go about hearing such news. In contrast to expectations made explicit
by the negated phrase, the heroine is engulfed by a sense of relief:

(8) There would be no one to live for during those coming years; she
would live for herself. There would be no powerful will bending hers
in that blind persistence with which men and women believe they
have a right to impose a private will upon a fellow creature, (p. 199)

Here too the positive statements are evaluated by the negative state­
ments that clarify and emphasize them, acquainting us with the heroine's
previous life experiences and that, therefore, cannot be assumed or ex­
pected to be suppressed.
But even evidence accumulated in the lab does not entirely support the
suppression hypothesis. For instance, findings show that jurors are influ­
enced by information they have been instructed to disregard (Thompson,
Fong, & Rosenhan, 1981). Further, media audiences are influenced by news
they are told is untrue (Wegner, Wenzlaff, Kerker, & Beattie, 1981). When
asked not to think of a concept (e.g., pink elephants, white bear, or house),
subjects cannot suppress that concept;4 at times, the to–be–suppressed con­
cepts even gain in accessibility (Wegner, 1994; Wegner & Erber, 1992;
Wegner, Schneider, Carter, & White, 1987). Such findings contest the sup­
pression hypothesis.
In this chapter we further question the suppression hypothesis and pro­
pose instead that suppression of negated items is not obligatory but op­
tional. The following pair of sentences (taken from Tottie, 1994, p. 414)
might illustrate the claim. They have identical initial clauses ("Fred didn't
see a cyclist") but different continuations, each necessitating the retention
of a different constituent from the previous identical clause, attesting that
suppression (following negation) cannot be obligatory and automatic but a
matter of deliberation, taking into consideration the scope of negation.
Thus, if (9) is acceptable, then cyclist should not be suppressed by the pre­
ceding negation marker but rather (if at all) see. Similarly, if (10) is accept­
able then see should not be suppressed by the preceding negation marker
but rather cyclist:
4
These concepts are not suppressed even when they are entirely irrelevant to an accessi­
ble context—to their stream, of consciousness at the moment.
10. NEGATION 239

(9) Fred didn't see a cyclist who was coming down the hill and hit him.
(10) Fred didn't see a cyclist but a man on a horseback.

The evidence we present here supports the retention hypothesis. Ac­

cording to the retention hypothesis (Giora, 2003; Giora & Fein, 1999a,
1999b), suppression is not obligatory: An activated meaning need not be
suppressed if it does not interfere with comprehension and might instead be
instrumental in constructing the intended meaning. Accordingly, speakers'
choice of a negated positive instead of an antonym ("the book is not inter­
esting" vs. "the book is boring" and "the book is not boring" vs. "the book is
interesting") can be viewed (among other things) as aimed at introducing
information to the discourse (e.g., about expectations)5 rather than elimi­
nating it from the mental representation. Given the retention hypothesis,
we propose that information introduced via negation would be retained and
tinge the interpretation of the negated item so that the outcome is a miti­
gated product involving both the negativity of the negation marker and also
the expressed meaning of the negated item (see also Givó0243n, 1993; Jespersen,
1924/1976).
If indeed negation markers hedge information rather than discard it, they
could convey social and pragmatic intentions and be used when, for in­
stance, people wish to downplay information such as when they want to
break bad news somewhat indirectly (see Experiment 4 below) or introduce
new information in a nonconfrontational manner. Indeed, negation was
found to be used when addressing controversial issues (Giora, 1994). Giora
(1994) analyzed the use of negation in public addresses (e.g., the late Egyp­
tian president Anwar Sadat's address to the Israeli Parliament). She showed
that negation allows the speaker to make a claim without asserting it—to in­
troduce new information nonassertively.6

ON THE ROLE OF NEGATION: EMPIRICAL FINDINGS

To reject the suppression hypothesis and support the retention hypothesis,

we first have to show that negated meanings are indeed accessed, regardless
of a prior negating context. Experiment 1 was designed to test this hypothe–

5
Moxey and Sanford (2000) expressed a somewhat similar view: "It appears that 'not
many' makes participants think that the speaker herself expected more, and that the speaker
believed the listener expected more. In contrast, only the first of these holds for 'few' and
'very few' " (p. 245).
We focus here on the hedging effect of explicit negation when negating is processed in a
compositional manner. Familiar, fixed expressions and idioms involving a negation marker
are excluded from the analysis because their pragmatic meanings are lexicalized and are not
constructed on the fly.
24O GIORA ET AL.

sis. We aimed to tap initial processes and test the assumption that a negation
marker (not as in not X) will not inhibit the access of salient meanings (of X)
(as might also be deduced from Clark & Chase, 19727). In Experiments 2, 3,
and 4, we examined the effect of negation on later integrative processes. We
wished to show that, contra the received view (e.g., Hasson, 2000, Experi­
ment 3; Kaup, 2001; MacDonald & Just, 1989, Experiment 1), a negation
marker will not suppress salient meanings activated initially but only modify
them (for a similar view, see Horn, 1989, pp. 236–240). In Experiments 1
and 3 we used scalar adjectives (sharp and rotten). In Experiments 2 and 4
the negated elements were not necessarily adjectives, and they were not
necessarily gradable ('fail'/'succeed'). In all the experiments, the partici­
pants, native speakers of Hebrew, were presented Hebrew items (translated
here for convenience).

EXPERIMENT 1

According to the modular view (Fodor, 1983) and the graded salience hy­
pothesis (Giora, 1997, 2003; Peleg, Giora, & Fein, 2001, in press), lexical ac­
cess is invariant across contexts. Consequently, contexts containing both
negated (not X) and nonnegated (X) constituents should initially facilitate
salient (coded and prominent) responses related to X, whereas a context
containing Y, which is the antonym of X, should not, because it does not in­
volve an explicit mention of X or any of its salient features. Thus, if piercing is
a salient feature of sharp, both (11) and (12) would prime it; (13) however,
would not:

(11) This instrument is sharp.

(12) This instrument is not sharp.
(13) This instrument is blunt.

By measuring response times (RTs) to two types of probes (related

and unrelated) in a nonnegated–positive condition, we first aimed to
establish that the related probe (piercing) is indeed a salient meaning of
the target (sharp) and would be facilitated compared to the unrelated
probe. We predicted that the priming to be exhibited in a non-
negated–positive (X) condition would be replicated in a negated–posi–
tive (not X) condition, but would not be replicated in an antonym (Y)
condition. Specifically, RTs to salient (related) probes (piercing) follow–

' According to Clark and Chase (1972), the negation operator is dissociated from the mes-
sage's core concepts and would, therefore, involve processing the core supposition and then
negate it. Because the core supposition is processed as a cognitive unit, which is then marked
with a negation tag, Mayo et al. (in press) terms this "the schema–plus–tag model."
10. NEGATION 241

ing a negated (positive) adjective ("This instrument is not sharp11}

would replicate those following a nonnegated (positive) adjective
("This instrument issharp"). In contrast, the priming effect expected in
the negated condition ("This instrument is not sharp") would not be
replicated in an antonym condition ("This instrument is blunt"). Al­
though, on the face of it, there seems to be a greater semantic affinity
between a negated adjective (not X) and its antonym (blunt = not
sharp) than between a negated and nonnegated adjective (not sharp =
sharp), the graded salience hypothesis would predict that, initially, the
latter are much more alike. Both the negation (not X) and its positive
opposite (X) share the same stimulus (X), whereas the antonym (Y) and
its equivalent negation (not X) do not. Their processing, therefore,
should involve different accessing routes.

Method

Design. A 3 X 2 factorial design was used with context type (positive,

negative, and antonym) and probe type (related and unrelated) as within–
subjects factors.

Participants. Participants were 36 graduate and undergraduate stu­

dents of Tel Aviv University, between the ages of 19 and 33 (27 women and 9
men). All were native speakers of Hebrew.

Stimuli. Stimuli were 72 triplets, 36 of which were target triplets, each

including a nonnegated–positive context (11), a negated–positive context
(12), and an antonym context (13) (repeated in (14–16) for convenience),
followed by two (related and unrelated) probes controlled for number of syl­
lables (with antonym related probes not being conventionally associated
with any item of the pairs as do direct opposites such as black and white, see
Clark, 1970):

(14) This instrument is sharp.

(15) This instrument is not sharp.
(16) This instrument is blunt.
Probes: piercing (related); leaving (unrelated)

Of the 72 triplets, 36 were filler items (12 of which included a negation

particle), followed by a nonword probe. In addition, 5 practice trials and 10
buffer trials were included.
242 GIORA ET AL.

Procedure. Participants were seated in front of a computer screen and

were tested individually. They were first given oral and written instructions.8
Reading of experimental sentences was self–paced: Participants pressed a key
when they have read the sentence. Each participant saw one sentence of each
of the triplets (e.g., one sentence of (14–16)). The interstimulus interval (ISI)
between offset of each target sentence and onset of the probe was 100 ms. The
probe was centrally displayed for 300 ms and the subjects had to make a lexical
decision as to whether the probe was a word or a nonword. The participants
responded by pressing one of two (yes or no) keys. The latency between the
offset of the probe and the pressing of the key was measured by the computer
and served as an RT To guarantee an attentive reading of the experimental
items, each lexical decision was followed by a yes or no comprehension ques­
tion (that was, however, irrelevant to the target word). Interitem interval was
600 ms. Two subjects who did not respond correctly to a minimum of 80% of
the comprehension questions were replaced.
Results
Means and standard deviations for the two conditions are presented in Ta­
ble 10.1. In the nonnegated–positive (X) condition (e.g., (14)), there was a
significant difference between the RTs to the related and unrelated probes
(in the subject analysis), ts (35) = 2.16, p < .01; ti(35) = 1.29, p = .1. Simi­
larly, as predicted, in the negated (not X) condition (e.g., (15)), there was
also a significant difference between the RTs to the related and unrelated
probes (in the subject analysis), ts(35) = 2.13, p < .01; ti(35) = 1.5, p = .07.
In the antonym (Y) condition, the difference between the related and un­
related probes did not reach significance ts(35) = 0.24, p = .4; ti(35) =
1.45, p = .07. Responses to the yes/no questions that followed the different
types of sentences did not vary significantly.
Discussion
Results of the subject analysis, which was backed by the same trend in the
item analysis, support the graded salience and modular hypotheses accord–

The first screen stated, "Thank you for participating in this experiment." The second
screen stated, "A short sentence will be presented on the screen. Read it carefully and press
the space bar when you are done. When you press the space bar, the sentence will disappear
and a letter string will be displayed.
If the letter string makes up a word (e.g., house) press the 'L' key; if it does not make up a
word (e.g., hois) press the 'S' key. Right after you have made the lexical decision, you will have
to respond to a Yes/No question related to the sentence you have read. Press the 'L' key for
'Yes' and the 'S' key for 'No.' For practice press the space bar."
Then five practice trials followed.
The third screen stated, "Make your decision as fast as you can, without compromising
precision. To start, press the space bar."
10. NEGATION 243

TABLE 10.1
RT (in ms) for Related and Unrelated Probes Following Positive,
Negative, and Antonym Contexts

Nonnegated–Positive Negated–Positive Antonym

Probe Type Related Unrelated Related Unrelated Related Unrelated
M 393 430 393 430 411 417
SD 195 199 178 193 208 167

ing to which lexical access is insensitive to contextual information. As pre­

dicted, a negating context did not inhibit access of salient meanings of target
words. (For converging results, see Fischler, Bloom, Childres, Rocus, &
Perry, 1983; Hasson, 2000, Experiments 1 and 2, which show that the affir­
mative meaning of negated metaphors are sustainable immediately and
even after a 500–ms delay; MacDonald & Just's, 1989, reading times in all
three experiments; and Mayo et al.'s (2002, pp. 10–16) Experiment 1). In
contrast, an antonym did not facilitate the related probe, because its mean­
ing was not made explicit in that context. Although a negation of X (not
sharp) and its antonym Y (blunt) may be viewed as akin, initially they are less
alike than a positive and a negative articulation of the same item (X/—X),
which are commonly viewed as opposites (sharp–not sharp).

EXPERIMENT 2

Experiment 1 demonstrates that initial processing of negated and non–

negated articulations of the same linguistic item (sharp and not sharp) prime
the same concept (s). In spite of their semantic difference, both conditions
facilitate the same salient meaning. Such findings demonstrate that a nega­
tion marker has no inhibitory effects on lexical access. They support the
view that processing is initially insensitive to contextual information, as as­
sumed by the graded salience hypothesis and the modular view.9 The ques­
tion, however, is whether a negation marker has postlexical suppressive
effects. Would the meaning activated initially due to its salience be sup­
pressed by the negation marker so as to allow for the contextually appropri­
ate interpretation to be retrieved and exclusively integrate with prior
context? Recall that according to the retention hypothesis (Giora, 2003;
Giora & Fein, 1999a, 1999b), suppression is not obligatory. An activated
meaning need not be suppressed if it may be functional in constructing the
intended meaning. We, therefore, predict that the meaning activated ini–

9
It also supports the ordered and reordered access views, see Rayner, Pacht, and Duffy
(1994).
244 GIORA ET AL.

tially would be retained and tinge the interpretation of the negated element
so that the result is a mitigated product (for information on the tinge hy­
pothesis, see Dews & Winner, 1995).
To test this hypothesis, we ran three off–line experiments. Off–line tasks
involve no time constraints and thus allow for suppression, which requires
extra processing time, to take place. However, if, following negation, sup­
pression is not obligatory, we should find traces of the negated concepts. In
Experiment 2 we aimed to show that speakers retain at least some aspects of
the negated element (for a similar view, see Hegel, cited in Horn, 1989, p.
64) .10 Specifically, we wished to show that a negated item constrains the ac­
ceptability of the item that follows it. The assumption that, following nega­
tion, suppression is obligatory predicts that any element can follow a
negated item, because an entity that is not represented cannot constrain the
next discourse element. In contrast, assuming that negation does not eradi­
cate all the features of the negated element (activated initially because of
their salience), the negated element should constrain the acceptability and
classifiability of the next constituent in the string (as would nonnegated ele­
ments). The retention hypothesis thus predicts that only elements classifi­
able as members of the category in which the negated item is a member can
follow that negated item (as would be the case with lists of nonnegated
items). (For information regarding how nonnegated items are classified into
ad hoc categories, see Barsalou, 1983.) We thus expected sentences such as
(17a and 18a) to be evaluated as more acceptable and appropriate than sen­
tences such as (17b and 18b). Although in (17a and 18a), the two ele-
ments—the negated and the nonnegated items—can be grouped on the
basis of class membership, this does not hold for (17b and 18b):

(17) a. What I bought yesterday was not a bottle but a jug.

b. What I bought yesterday was not a bottle but a closet.11
(18) a. I don't want coffee; I want tea.
b. I don't want coffee; I want shoes.

Method
Participants. Participants were 40 students of Tel Aviv University,
who were between the ages of 20 and 40 (20 women and 20 men). They were
all native speakers of Hebrew.

10
"For Hegel, a 'pure negative judgment' like the rose is not red suggests that a different
predicate from the same semantic class applies to the subject: 'To say that the rose is not red
implies that it is still coloured.' " (Hegel, 1892:306 as cited in Horn, 1989, p. 64)
11
Only some of the sentences included but clauses. On the constraints of the but operator,
see Winter and Rimon (1994).
10. NEGATION 245

Stimuli Stimuli were 30 pairs of sentences such as (17) and (18) and
10 filler pairs.

Procedure. Participants were asked to indicate which of the alterna­

tives in each pair is a preferable, a more acceptable, or a more natural sen­
tence in their language, given that they do not want to be funny or
entertaining.
Results and Discussion
Results are straightforward. Ninety–six percent of the participants preferred
the sentences in which the negated and nonnegated items are classifiable as
members of the same category (17a and 18a), and a clear–cut dispreference
(4%) for the sentences in which these items are less amenable to such cate­
gorization (17b and 18b). These results demonstrate that a negation marker
does not eliminate the mental representation of the negated item altogether.
Even when negated, the negated concept constrains the next discourse ele­
ment; it determines which concepts can follow it and which ones cannot.
This finding demonstrates that at least some of the aspects of the negated
concept must be retained.
Taken together, the results of this experiment and the examples from nat­
urally occurring discourses (1–8) testify to speakers' sensitivity to the re­
tained aspects of items explicitly negated.
EXPERIMENT 3

To further test the retention hypothesis, we conducted another experi­

ment in which we wanted to be more accurate about speakers' sensitivity
to the tingeing or hedging effect of explicit negation. Therefore, here we
asked participants to rate pairs of statements on a 7–point polarity scale
(1 = X, 7 = Y). The target pairs included an item (X) and its negated ant­
onym (not Y): "This instrument is sharp" and "This instrument is not
blunt" The retention hypothesis predicts that the negated item (not Y)
would be rated as distinct from (X) because of the difference between X
and Y, despite the entailment relation obtaining between X and not Y,
whereupon sharp entails not blunt (see, e.g., Bartsch & Vennemann, 1972;
Brewer & Lichtenstein, 1975; Lyons, 1977).
The retention hypothesis further predicts that, when explicitly negated,
both items with negative (rotten) and positive (fresh) associations would be
similarly retained and tinged. This stands in contrast to the asymmetry as­
sumed by Ducrot (1973) and Horn (1989). Ducrot (1973, as cited in Horn,
1989, p. 334) contended that negative and positive moral adjectives exhibit
an asymmetry in that negating the positive (not right) implies the opposite
(wrong); however, negating the negative (notwrong) does not imply the op–
246 GIORA ET AL.

posite (right), but a weaker version of it. This view was empirically supported
by Colston (1999). Similarly, outside the moral sphere, Cornulier (1974,
cited in Horn, 1989, p. 335) maintained that not rich is taken to implicate
poor but not poor does not implicate rich (see also Brewer & Lichtenstein,
1975). (For a different view, see Clark & Chase, 1972; Fiedler, Walther,
Armbruster, Fay, & Naumann, 1996; Johnson–Laird &.Tridgell, 1972; Just
& Carpenter, 1976; Wason, 1963.) We tested this assumption vis a vis the
retention hypothesis.
Method
Participants. Participants were 30 graduate and undergraduate stu­
dents of Tel Aviv University, who were between the ages of 18 and 30 (19
women and 11 men). All were native speakers of Hebrew.
Stimuli. Stimuli were comprised of 40 experimental pairs selected from
Experiment 1 and 20 filler sentences. Of the experimental items, 10 con­
tained nonnegated positive items coupled with their negated antonyms
(e.g., "The vegetables looked fresh" and "The vegetables looked not rotten").
Another 10 repeated the positive item and its hedged antonym (e.g., "The
vegetables looked fresh" and "The vegetables looked fairly rotten"). Another
set of 10 items contained nonnegated negative items coupled with their ne­
gated (positive) antonym (e.g., "Sarit's dress was ugly" and "Sarit's dress was
not pretty"). The matched set repeated the nonnegated negative items cou­
pled with a hedged (positive) antonym (e.g., "Sarit's dress was ugly" and
"Sarit's dress was fairly pretty"). The 20 fillers involved various hedges and
intensifiers such as really, fairly, entirely, little, and so on.
Procedure. Subjects were presented booklets containing pairs of sen­
tences and were asked to rate each sentence of the pair on a 7–point polarity
scale (e.g., sharp–blunt and fresh–rotten).12

12
The instructions read as follows:
Dear Participant,
You will be presented with pairs of statements that express the impression two different
people have of a certain object. Your task is to compare the expressions and to grade each
of them on 1–7 scale. For example:
a. This painting is terrible
b. This painting is pretty good

| _ a _ | |__|__|__|_b_|__|

1 7
terrible good
There are no right or wrong responses. We are interested in studying the way you under­
stand the different expressions. Thank you for your cooperation!
10. NEGATION 247

Results

Results are presented in Tables 10.2 and 10.3 and Fig. 10.1. Ratings for the
nonnegated positive items (fresh) differed significantly from their negated
antonyms (not rotten), t(14) = 17.07, p < .0001, and from their hedged ant­
onyms {fairly rotten), t(14) = 28.55, p < .0001. Likewise, ratings for the non–
negated negative items (ugly) differed significantly from their negated
antonyms (notpretty), t( 14) = 21.29, p < .0001, and from their hedged ant­
onyms (fairly pretty), t(14) = 25.20,p < .0001. As illustrated by Fig. 10.1, in­
verting the scale findings for the positive or negative adjectives shows that
they act as a mirror image of each other.

Discussion

Results support the retention hypothesis, according to which salient mean­

ings that might be instrumental in constructing the intended meaning
should not be suppressed. Indeed, although a negation marker did not block
salient meanings (piercing) of negated constituents (not sharp; cf. Experi­
ment 1), it did not suppress them. The bluntness of not blunt or the "pretti–
ness" of not pretty was not eradicated even when an off–line task allowed

TABLE 10.2
Mean Ratings of the Negative Adjectives
Relative to Negated and Hedged Positive Adjectives

Adjective Negative Negated Positive Hedged Positive Positive

Type (ugly) (not pretty) (fairly pretty) (pretty)
1.23 (0.26) 3.61 (0.43)
1.33 (0.32) 5.51 (0.39)

Note. Items were rated on a 7–point scale (1 = ugly and 7 = pretty). Standard deviations are shown
in parentheses.

TABLE 10.3
Mean Ratings of the Positive Adjectives Relative to Negated and
Hedged Negative Adjectives

Adjective Negative Hedged Negative Negated Negative Positive Adjective

Type (rotten) (fairly rotten) (not rotten) (fresh)
2.29 (0.34) 6.59 (0.32)
4.23 (0.44) 6.64 (0.28)

Note. Items were rated on a 7–point scale (1 = rotten and 7 = fresh). Standard deviations are shown
in parentheses.
248 GIORA ET AL.

FIG. 10.1 Ratings of positive adjectives versus negated and hedged negative ad­
jectives compared to ratings of negative adjectives versus negated and hedged posi­
tive adjectives (top panel). A mirror image of negative X appears in bottom panel.

subjects extra processing time during which suppression could have become
effective. Instead, the "bluntness" of blunt or the "prettiness" of pretty tinged
the meaning of not blunt and not pretty and made it distinctly different from
sharp and ugly. Negation directed the modification of the negated concept
towards the mid, neutral position on the scale.
Our results further show that negative (ugly) and positive (fresh) adjec­
tives do not differ when compared with their negated opposites. The differ­
ence between negatives (ugly) and their negated positives (not pretty) is
equivalent to that obtaining between positives (fresh) and their negated
10. NEGATION 249

negatives (not rotten). Contra the asymmetry assumption (Colston, 1999;

Horn, 1989), our findings demonstrate that the salient meanings of both
negatives and positives affect interpretation similarly (psycholinguistically
and partly semantically).
In sum, our findings concerning the tingeing effect allow us to classify ne­
gation markers within the category of modifiers rather than within the cate­
gory of suppressors. Indeed, explicit negation is comparable to other
modifiers such as fairly. Al though fairly is a moderate modifier preserving the
salient meaning of the modified adjective rather actively (fairly rotten is
closer to rotten than not rotten and fairly pretty is closer to pretty than not
pretty), not is a slightly (although distinctly) stronger modifier than fairly.
Like fairly, however, it does not eradicate the salient meaning of the modi­
fied constituent (for more information on other modifiers sensitive to
gradability, see Paradis, 2001).

EXPERIMENT 4

Are speakers aware of the tingeing effect involved in negation? Would they
select a negated element when wishing to downplay unpleasant informa­
tion? Experiment 4 was designed to test whether speakers are sensitive to
the pragmatic functions of the negation marker not. Having shown that ne­
gation acts as a modifier rather than as a suppressor, allowing features of the
negated item to infiltrate and affect the interpretation of the negative con­
stituent (not Y), we aimed to show that speakers employ negated items
when being polite.

Method
Participants. Participants were 30 graduate and undergraduate stu­
dents of Tel Aviv University, between the ages of 21 and 36 (22 women and 8
men). All were native speakers of Hebrew.

Stimuli. Stimuli included 16 experimental pairs and 8 filler pairs. The

experimental pairs comprised a sentence that contained a nonnegated neg­
ative item ("What you said was a lie") and a sentence that contained its ant­
onym, which was marked for negation and conveyed the same semantic
interpretation ("What you said was not true"). The fillers included two sen­
tences with different semantic meanings ("The dinner you cooked for us was
delicious" and "The dinner you cooked for us was filling"). The order of pre­
sentation of each of the items of the experimental pairs was randomized.

Procedure. Subjects were presented booklets containing these pairs

and were asked to indicate which of the two sentences they would select if
they wished to be polite or less offensive.
25O GIORA ET AL.

Results

To quantify the responses, a selection of a not X (nottrue) structure scored 1;

a selection of an X (a lie) structure scored 0. The maximum score for a sub­
ject then would be 16. Results showed that indeed most of the subjects se­
lected the sentences that contained a negated constituent. The mean score
of all the responses was 14.27 (SD = 1.95).

Discussion

When speakers were aiming at politeness and were faced with two alterna­
tives, they preferred a negated, semantically positive constituent (not true)
to a nonnegated semantically negative constituent (a lie). This choice indi­
cates that speakers are sensitive to the pragmatic function of negation,
which often provides one with a mitigated, hedged version of the negated
constituent. Such findings are consistent with the view that a negation
marker (not) is not a strong suppressor, although it might reduce initial levels
of activation (as shown by Hasson, 2000, Experiment 3 and Mayo et al.,
2002, Experiment 2). Rather, it is a strong modifier.

GENERAL DISCUSSION

In this study, we tested the retention–of–negated–concepts hypothesis. Con­

trary to the received view, which assumes that suppression of negated con­
cepts is obligatory, we adduced evidence showing that it is not. We showed
that information introduced via negation is often retained rather than sup­
pressed and tinges the negativity of the explicit negation. The interpretation
of a negated item (not pretty) is therefore, more mitigated than its alternative
opposite (ugly), which allows negation to be used when politeness or hedg­
ing is required. Findings in our lab as well as an analysis of naturally occur­
ring discourses demonstrate that comprehenders neither block nor
eliminate from mental representations salient meanings of negated con­
cepts. As a result, speakers select negated positives as opposed to non–
negated negatives when they are after a polite or a low–key mode of
expression and when they wish to hedge a positive expression.
In Experiment 1, we showed that negation does not have inhibitory ef­
fects. A negation marker (not in not–X) did not block access of the salient
meanings of a target (X in not-X). Thus, piercing was primed following both
sharp and not sharp. These results corroborate those of MacDonald and Just
(1989) and Hasson (2000, Experiments 1 and 2). Although MacDonald
and Just's findings may demonstrate some suppression effects, their reading
time phase (in all their experiments), which is the only measure in their
studies that could tap initial processes, shows no inhibition of salient mean–
1O. NEGATION 251

ings of concepts following negation. Similarly, in Hasson's study, tapping re­

lated concepts immediately (150 ms and even 500 ms) after offset of the
negated targets showed no inhibition effects.
How would comprehenders manage salient meanings of negated con­
cepts, which are initially insensitive to negation effects? Would they sup­
press them postlexically to allow the derivation of the appropriate
meaning? Experiment 2 shows that salient meanings of negated concepts
are not wiped out even when comprehenders are allowed extra processing
time. Rather, they are retained and affect the ongoing discourse process­
ing. Thus, lists including negated items were shown to behave like lists of
nonnegated items; that is, they were viewed as acceptable only when
categorizable in the same set. Specifically, subjects found "What I bought
yesterday was not a bottle but a jug" acceptable. In contrast, "What I
bought yesterday was not a bottle but a closet" was not acceptable. Such
findings cannot be accounted for in terms of suppression. If a negation
marker discards an entity altogether, the acceptability of the next item in
the list should not be constrained by the prior occurrence of that entity.
Experiment 2 thus suggests that at least some of the features of negated
items are preserved and affect the classifiability of the next item in line.
Note that this is also true of negated items preceded by nonnegated el­
ements. The following string (19), which describes the Israeli Chief of
Staff, Ya'alon, is appropriate because the negated element can be classi­
fied within the same category as the nonnegated item that precedes it,
showing that the negated element must have preserved at least some of
its properties:

(19) The word is that he is intelligent but not brilliant. Tough but not
cruel, demanding but not macho. (Shavit, 2002)

This seems to be also true of the neither—nor construction. Although (20)

is appropriate, (21) is not:

(20) The word is that he is neither tough nor cruel.

(21) The word is that he is neither tough nor thirsty.

Indeed (22) shows that such negation requires that the negated items
make up a category (of predicates applicable to Sami Michael's—the new
president of the Association for Civil Rights in Israel—identity):

(22) Michael says that ever since he came to this country [Israel] at the
age of 22, he has not known what peace of mind is, neither as a Jew
nor as an Israeli, nor as a person who believes in democracy and a
just society. (Eldar, 2002)
252 GIORA ET AL.

Besides, it is quite obvious that double negation necessitates the retain–

ability of the negated items. For instance, "I don't think this is implausible,"
approximately meaning "I think this is plausible," necessitates the retain–
ability of both think and plausible.
Similarly, constructions such as X if not Y do not in fact suppress Y but
rather mean "X and possibly Y," thus relying on the retainability of the affir­
mative meaning of Y:

(23) Most if not all of the remaining Gricean rules respond to the Speaker's
Economy, either directly ... or indirectly .... (Horn, 1984, p. 12).
(24) Well yeah but you see that the trouble is they've been now two if not
three pilot phases. (British National Corpus, 1993: H5E 929,
http://sara.natcopr.ox.ac.uk/lookup.html).
(25) Unfortunately for such critics it has been found that acupuncture
works equally well, if not more effectively, on animals. (British Na­
tional Corpus, 1990: CB9 1459, http://sara.natcopr.ox.ac.uk/
lookup.html).
If negation neither inhibits nor necessarily suppresses negated items,
how does it eventually affect interpretation? Experiment 3 shows that a
negation marker induces mitigation13: Not pretty is distinguishable from
ugly (see also Clark & Clark, 1977: 426): It is perceived to be halfway be­
tween pretty and ugly; that is, it is conceived of as "less than pretty." How­
ever, not pretty is also quite distinguishable from fairly pretty, which is
rated as closer to pretty, which suggests that not is a stronger modifier
than a more positive hedge (e.g., fairly) and would make not pretty less
pretty than fairly pretty but also more pretty than ugly. No wonder the
American administration, favoring the Israelis over the Palestinians, de­
scribed the Israeli siege of Arafat as "unhelpful" (Sky News, 25 Septem­
ber 2002). "Unhelpful" certainly sounds less disastrous than disastrous or
even harmful.
The following example illustrates this further. It shows that "not liqui­
date" does not mean "let be." Rather (the question of who will "not liqui­
date" Arafat set aside), "not liquidate" allows for a "less than liquidate"
interpretation such as neutralize or not be:

(26) The Americans made it clear that they are not going to liquidate
him, but that if the Palestinians want to see light at the end of the
tunnel, they themselves should neutralize him. That is an unequiv­
ocal statement: Arafat will not be the decision–maker. He will not
be. (Shavit, 2002)

On other forms of mitigation see Caffi (1990, 2001).

10. NEGATION 253

Experiment 4 shows that speakers are sensitive to the modifying effect

of explicit negation. Participants showed a clear–cut preference for ne­
gated items (not succeeding) that contained a negated positive (succeed­
ing) over their antonym (failing) when they were asked to be polite. The
negation marker, which failed to suppress the positivity of the negated
concept, allowed this positivity to dilute the negativity of the negation
marker and offer a more positive articulation of an undesirable state of af­
fairs (failing).
In fact, a close look at some of the studies that assumed suppression re­
veals that they do not quite support a suppression hypothesis. For In­
stance, MacDonald and Just's (1989) third experiment, which uses related
associates as probes, shows no suppression effects. It is only when partici­
pants had to make a decision as to whether a probe appeared or did not ap­
pear in the text that findings clearly supported the suppression hypothesis
(MacDonald & Just, 1989, Experiment 1). However, as MacDonald and
Just observed, it is quite possible that having to be positive and say "yes"
about a target that appeared in a negated context (i.e., the target was pre­
ceded by no) involved some Stroop (1935) interference that slowed down
responses to probes featuring negated targets. Using the same suspect
methodology, Kaup (2001) replicated these results, but then it is possible
that they, in fact, also testified to the accessibility of the negated concepts
as negated concepts rather than to their inaccessibility as a result of a nega­
tive context.
Specifically, Kaup (2001) tried to support the view that accessibility or
inaccessibility of concepts is a function of their representation in the sit­
uation model (Johnson–Laird, 1983; van Dijk & Kintsch, 1983; Zwaan,
1999). Thus, "Mary bakes bread but no cookies" should lead to a reduced
accessibility of the negated concept (cookies), because it should not be
kept in mind. (After all, no cookies came into being). In contrast, the ne­
gated concept (photograph in "Elizabeth burns the letters but not the
photographs") should be included in the mental representation and
should not be suppressed (because there must have been some photo­
graphs) . Such an account, however, will not explain the speaker's expec­
tation that the addressees (restaurant diners) retain the concept of "12%
service" appearing on their bill (29) when paying it:

(27) The price does not include 12% service. Thanks!!!14

Given that the price does not include 12% service, no mental representa­
tion of it is expected. According to Kaup, it should have been suppressed.

4
We thank Mira Ariel (23–25 and 27).
254 GIORA ET AL.

However, it is quite obvious that the addressees were supposed to keep pre­
cisely this concept in mind and act upon it when tipping (for which they are
even thanked in advance; also see the bean story in the epigraph).
Another example that stresses what is absent ("the real reasons why
the United States has found itself under attack") is shown in (28). Al­
though, according to Kaup (2001), what is absent should be suppressed
(because it should not be represented in our mental model), here, what is
absent is apparently what the journalist is after and is what he wants us to
keep in mind:

(28) But the real lie in the President's speech—that which has dominated
American political discourse since the crimes against humanity on 11
September last year—was the virtual absence of any attempt to ex­
plain the real reasons why the United States has found itself under
attack.
In his mendacious article in this newspaper last week, President
Bush's Defense Secretary, Donald Rumsfeld, also attempted to mask
this reality. The 11 September assault, he announced, was an attack
on people "who believe in freedom, who practice tolerance and who
defend the inalienable rights of man." He made, as usual, absolutely
no reference to the Middle East, to America's woeful, biased poli–
cies in that region, to its ruthless support for Arab dictators who
do its bidding—for Saddam Hussein, for example, at a time when
the head of Iraqi nuclear research was undergoing his Cal-
vary—nor to America's military presence in the holiest of Muslim
lands, nor to its unconditional support for Israel's occupation of
Palestinian land in the West Bank and Gaza. (Fisk, 2002)

Apparently, negation will not affect suppression in a question such as

(29), no matter what the answer to that question is:

(29) Is this emphatically not true for our "postmodern" time? (Zizek,
2002).

A negation marker, then, might have different functions: It might indi­

cate the absence of the negated entities (as when negating nouns), which ei­
ther introduce the negated entities to the discourse (see Giora, 1994, and (4,
5, and 30)), or lead to similar alternatives (Experiment 4). It might hedge
properties (as in the case of predicates, see Experiments 3 and 4). It is not en­
tirely implausible that it might also suppress entities. Further research
should investigate the conditions under which negation involves suppres­
sion. Our findings, however, argue against an obligatory view of suppression.
They show that when exploitable, so–called irrelevant meanings partake in
10. NEGATION 255

the interpretation process and affect speakers' linguistic choices vis a vis
their social environments (for a similar view regarding irony and metaphor,
see Giora, 2003; Giora & Fein, 1999a, 1999b).

ACKNOWLEDGMENTS

This research was supported by a grant from the Tel Aviv Basic Research
Fund to the first author.
We thank Mira Ariel, Herb Colston, Jack Du Bois, Uri Hasson, Brooke
Lea, Carita Paradis, Noa Shuval, and Aldo Savi for very helpful comments
on an earlier draft.

REFERENCES

Alcott, L. M. (1962). Little Man; Life at Plumfield With Jo's Boys. New York:
MacMillan Publishing Co. (Original work published 1871)
http://www.digital.library.upenn.edu/women/alcott/men/men.html
Bar'el, Z. (2002). The facts were in on Israel's Arabs. Ha'aretz in English. Retrieved
25 August 2002 from http://www.haaretz.vco.il
Barsalou, L. W. (1983). Ad hoc categories. Memory & Cognition, 11, 211–227.
Bartsch, R., & Vennemann, T. (1972). Semantic structures. Frankfurt: Anthenaum
Verlag.
Brewer, W. F., & Lichtenstein, E. H. (1975). Recall of logical and pragmatic implica­
tions in sentences with dichotomous and continuous antonyms. Memory & Cog­
nition, 3, 315–318.
British National Corpus. (1990). http://sara.natcorp.ox.ac.uk/lookup.html
British National Corpus. (1993). http://sara.natcorp.ox.ac.uk/lookup.html
Caffi, C. (1990). Modulazione, mitigazione e litote. In M.-E. Conte, P. Ramat, & A.
Giaalone Ramat (Eds.), Diminsioni della linguistica (pp. 169–199). Milano:
Angeli.
Chopin, K. (1976). The story of an hour. In B. H. Solomon (Ed.) The awakening and
selected stories and Kate Chopin (pp. 198–200). New York: Signet. (Original work
published 1894).
Clark, H. H. (1970). Word association and linguistic theory. In J. Lyons (Ed.), New
horizons in linguistics (pp. 271–286). Baltimore: Penguin Books.
Clark, H. H., & Chase, W. G. (1972). On the process of comparing sentences
against pictures. Cognitive Psychology, 3, 472–517.
Clark, H. H., & Clark, V E. (1977). Psychology and language: An introduction to
psycholinguistics. San Diego: Harcourt Brace Javanovich.
Colston, H. L. (1999). "Not good" is "bad," but "not bad" is not "good": An analysis
of three accounts of negation asymmetry. Discourse Processes, 28, 237–256.
Cornulier, B., de. (1974). Remarques à propos de la negation (Notes on expected
negation) anticipée. Le français moderne, 41, 43–57.
Dews, S., & Winner, E. (1995). Muting the meaning: A social function of irony.
Metaphor and Symbolic Activity, 10, 3–19.
256 GIORA ET AL.

Du Bois, W. J. (200). Santa Barbara corpus of spoken American English. CD–ROM.

Philadelphia: Linguistic Dat a Consortium [www.ldc.upenn.edu/Publications/
SBC/].
Ducrot, O. (1973). La preuve et le dire (The proof and the said). Paris: Herman.
Eldar, A. (2002, 16 September). Trumpeting a blast of despair. Ha'aretz in English.
Fiedler, K., Walther, E., Armbruster, T., Fay, D., & Naumann, U. (1996). Do you re­
ally know what you have seen? Intrusion errors and presuppositions effects on
constructive memory. Journal of Experimental Social Psychology, 32, 484–511.
Fischler, I., Bloom, P. A, Childres, D. G., Rocus, S. E., & Perry, N. W. (1983). Brain po­
tentials related to stages of sentence verification. Psychophysiology, 20, 400–409.
Fisk, R. (2002, 15 September). America's case for war is built on blindness, hypoc­
risy and lies. The Independent.
Fodor, J. (1983). The modularity of mind. Cambridge, MA: MIT Press.
Giora, R. (1994). On the political message: Pretending to communicate. In H.
Parret (Ed.), Pretending to communicate (pp. 104–123). Berlin/New York: Walter
de Gruyter Verlag (Communication and Cognition Series).
Giora, R. (1997). Understanding figurative and literal language: The graded sa­
lience hypothesis. Cognitive Linguistics, 7, 183–206.
Giora, R. (2003). On our mind: Salience, context, and figurative language. New York:
Oxford University Press.
Giora, R., & Fein, O. (1999a). Irony: Context and salience. Metaphor and Symbol,
14, 241–257.
Giora, R., & Fein, O. (1999b). On understanding familiar and less–familiar figura­
tive language. Journal of Pragmatics, 31, 1601–1618.
Giora, R., Fein O., & Schwartz, T. (1998). Irony: Graded salience and indirect nega­
tion. Metaphor and Symbol, 13, 83–101.
Givon, T. (1993) English grammar: A function–based introduction, Vol. I. Amster-
dam/Philadelphia: John Benjamins.
Hasson, U. (2000). Comprehension of negation in negated metaphors. Unpublished
master's thesis, Princeton University, Priceton, NJ.
Hegel, G. W. F. (1892). The logic of Hegel: Translated from the encyclopedia of the philo­
sophical sciences. W. Wallace, trans. 2nd ed. London: Oxford University Press.
Horn, L. R. (1984). A new taxonomy for pragmatic inference: Q–based and R–based
implicatures. In D. Schiffrin (Ed.), Meaning, form and use in context (pp. 11–42).
Washington DC: Georgetown University Press.
Horn, L. R. (1989). A natural history of negation. Chicago: University of Chicago
Press.
Jefferson, G. (2002). Is "no" an acknowledgment token? Comparing American and
British uses of (+)/(–) tokens. Journal of Pragmatics, 34, 1345–1383.
Jespersen, O. (1976). The philosophy of grammar. New York: Norton. (Original work
published 1924).
Johnson–Laird, P N. (1983). Mental models. Cambridge, MA: Harvard University
Press.
Johnson–Laird, P. N., & Tridgell, J. M. (1972). When negation is easier than affirma­
tion. Quarterly Journal of Experimental Psychology, 24, 87–91.
10. NEGATION 257

]ust, M. A., & Carpenter, P. A. (1976). Eye fixations and cognitive processes. Cogni­
tive Psychology, 8, 441–480.
Kaup, B. (1997). The processing of negatives during discourse comprehension.
In M. G. Shafto & P. Langley (Eds.), Proceedings of the nineteenth conference of
the cognitive science society (pp. 370–375). Mahwah, NJ: Lawrence Erlbaum
Associates.
Kaup, B. (2001). Negation and its impact on the accessibility of text information.
Memory & Cognition, 29, 960–967.
Klein, N. (2002). No Logo. New York: Picador.
Labov, W. (1972). The transformation of experience in narrative syntax. In Lan­
guage in the inner city (pp. 354–395). Philadelphia: University of Pennsylvania
Press.
Lea, B. R., & Mulligan, E. J. (2002). The effect of negation on deductive inferences.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 28, 303—317.
Levy, G. (2002). Nothing happened. Ha'aretz Supplement in English. Retrieved 16
August 2002 from http://www.haaretz.co.il
Linor, E. (1999). Shiv'a Leilot, Yedi'ot Achronot (A Hebrew daily, September 17,
1999).
Lyons, J. (1977). Semantics. Cambridge, England: Cambridge University Press.
MacDonald, M. C., & Just, M. A. (1989). Changes in activation levels with nega­
tion. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15,
633–642.
Mayo, R., Schul, Y., & Burnstein, E. (in press). "I am not guilty" versus "I am inno­
cent": The associative structure activated in processing negations. Journal of Ex­
perimental Social Psychology.
Moxey, L. M., & Sanford, A. J. (2000). Communicating quantities: A review of
psycholinguistic evidence of how expressions determine perspective. Applied
Cognitive Psychology, 14, 237–255.
Paradis, C. (2001). Adjectives and boundedness. Cognitive Linguistics, 12, 47–65.
Peleg, O., Giora, R., & Fein, O. (2001). Salience and context effects: Two are better
than one. Metaphor and Symbol, 16, 173–192.
Peleg, O., Giora, R., & Fein, O. (in press). Contextual strength: The whens and
hows of context effects. In I. Noveck & D. Sperber (Eds.), Experimental
pragmatics. Basingstoke, England: Palgrave.
Rayner, K., Pacht, J. M., & Duffy, S. A. (1994). Effects of prior encounter and global
discourse bias on the processing of lexically ambiguous words: Evidence from eye
fixations. Journal of Memory and Language, 33, 527–544.
Shavit, A. (2002). The enemy within. Ha'aretz Supplement in English. Retrieved 30
August 2002 from http://www/haaretz/co/il
Sperber, D., & Wilson, D. (1986/1995). Relevance: Communication and cognition.
Oxford, England: Blackwell.
Stroop, J. R. (1935). Studies of inference in serial verbal reactions. Journal of Experi­
mental Psychology, 18, 643–662.
Thompson, W. C., Fong, G. T., & Rosenhan, D. L. (1981). Inadmissible evidence
and juror verdicts. Journal of Personality and Social Psychology, 40, 453–463.
258 GIORA ET AL.

Tottie, G. (1994). Any as an indefinite determiner in non–assertive clauses: Evi­

dence from present–day and early modern English. In D. Kastovsky (Ed.), Studies
in early modern English (pp. 413–427). Berlin, Germany: Mouton de Gruyter.
van Dijk, T. A., & Kintsch, W. (1983). Strategies of discourse comprehension. New
York: Academic Press.
Wason, P. C. (1963). The contexts of plausible denial. Journal of Verbal Learning and
Verbal Behavior, 4, 7–11.
Wegner, D. (1994). The ironic processes of mental control. Psychological Review,
101, 34–52.
Wegner, D., & Erber, R. (1992). The hyperaccessibility of suppressed thoughts. Jour­
nal of Personality and Social Psychology, 63, 903–912.
Wegner, D., Schneider, D., Carter, S., & White, T. (1987). Paradoxical effects of
thought suppression. Journal of Personality and Social Psychology, 53, 5–13.
Wegner, D., Wenzlaff, R., Kerker, R. M., & Beattie, A. E. (1981). Incrimination
through innuendo: Can media questions become public answers? Journal of Per­
sonality and Social Psychology, 40, 822–832.
Winter, Y., & Rimon, M. (1994). Contrast and implication in natural language. Jour­
nal of Semantics, 11, 365–406.
Zizek, S. (2002). Welcome to the desert of the real. Retrieved October 17, 2002 from
http://www.lacan.com/desertsymf.htm
Zwaan, R. A. (1999). Situation models: The mental leap into imagined worlds. Cur­
rent Directions in Psychological Science, 8, 15–18.
 PART IV

New Sociocultural

Influences

This page intentionally left blank

 11

Stereotype Processing

and Nonliteral Language

Tracie L. Blumentritt
University of Wisconsin–La Crosse

Roberto R. Heredia
Texas A&M International University

Following the April 19, 1995, Oklahoma City bombing, news reports began
circulating that the incident may have been the work of terrorists. The next
day a Jordanian American was wrongfully detained and returned from Lon­
don to the United States in connection with the terrorist attack. A few
months later, the American Civil Liberties Union (ACLU) filed a claim
against the United States on behalf of Abraham Ahmad. According to the
ACLU's claim, "it [was] wrong to target someone in a criminal investigation
based on cultural stereotypes ... [and] Arab Americans have been sub­
ject [ed] to a great deal of prejudice and discrimination, both before and after
the bombing" (ACLU, Freedom Network, 1995, p. 1). Unfortunately, the at­
tacks on September 11, 2001 further reinforced and extended the stereotype
that not only are all Arabs terrorists but also that all Muslims or people of the
Islamic faith are terrorists. However, such generalizations are without founda­
tion. No particular ethnic or cultural group holds an exclusive trademark for
terrorism because other groups have been and are involved in such destruc­
tive acts (e.g., Anglo Americans, Europeans, and Latin Americans). In this
chapter, the term stereotype describes the seemingly pervasive human ten­
dency to perceive others not as individuals but as members of particular social
groups with shared characteristics (Lippman, 1922). In other words, stereo–

261
262 BLUMENTRITT AND HEREDIA

types are cognitive frameworks that consist of beliefs and generalizations

about perceived typical characteristics for certain social groups. Thus, for
some people, the term terrorist can be understood in terms of a dominant
meaning, which typically includes acts of terrorism, such as bombing and gun­
fire, and a stereotype meaning that characterizes Arabs or Muslims as people
who commit acts of terrorism. A vast amount of research has been conducted
on stereotypes, the most recent of which involves examining stereotypes as a
specific type of cognitive structure and discerning the processes by which
these structures exert influence on behavior, attitudes, memory, social judg­
ment, inferences, and various other perceptual processes (e.g., Banaji &
Greenwald, 1995; Banaji, Hardin, & Rothman, 1993; Bodenhausen, 1988;
Cohen, 1981; Darley & Gross, 1983; Kunda & Sherman–Williams, 1992;
Synder, Tanke, & Berscheid, 1977). In short, when a stereotype is activated,
practically all aspects of social information processing are affected. Clearly,
the possession of a stereotype and its activation have extremely important im­
plications for social life.
Within this chapter, we explore stereotype processing and the connection
between stereotypes and language processing, and we propose that stereotypes
can be construed as a special case of figurative language. For example, in the
sentence (1) "It is important to assure that TERRORISTS do not access air­
ports," it is possible to interpret the term terrorist in two ways. First, this term
may trigger general information about things that denote acts of terrorism such
as bombs, weapons, and highjackings. Second, when people listen to this term,
they may retrieve information about a specific ethnic group that is associated
with such terrorists attacks. This stereotypic meaning can be seen as an exten­
sion of a nonliteral interpretation (i.e., a metonymic relationship) in which a
perceived dominant characteristic of a particular ethnic group is used to refer to
the entire group (e.g., Frisson & Pickering, 1999; Gibbs, 1990, 1993, 1994). In
fact, Heredia and Blumentritt (2002) showed that social stereotypes can be ex­
amined from a psycholinguistic (as well as sociocognitive) perspective and that
by doing so several interesting and novel processing effects emerge. We begin by
reviewing the relevant sociocognitive literature on stereotype processing and
then introduce important concepts from the psycholinguistic literature that
bear directly on stereotype processing. Finally, we review a recent study that il­
lustrates how concepts and methods from the field of psycholinguistics can in­
form and guide the empirical study of stereotypes.

STEREOTYPE PROCESSING

Stereotype Activation and Structure

Broadly speaking, current conceptualizations construe stereotypes as a spe­

cific type of cognitive structure that contains information relevant to the
1 1. STEREOTYPES AND NONLITERAL LANGUAGE 263

categorization of groups of people (Dijksterhuis & Van Knippenberg, 1996;

Stephan, 1989). These cognitive structures enable the perceiver to effi­
ciently, simplistically, and often inaccurately make judgments about others.
The dominant view holds that information contained within the structure is
highly interconnected, and forms large networks of abstract information of­
ten referred to as schemas. Social schemas are networks representing knowl­
edge about members of a given social category and can include such things
as perceived physical characteristics, (stereo) typical traits and attributes,
and expected behaviors of members of the particular social group. Specific
pieces of information (often referred to as nodes) within the schemas are
thought to be organized semantically and according to frequency and re­
cency of activation. Activation occurs when the node is made accessible to
working memory and can occur either through consciously controlled pro­
cesses or through automatic, largely unconscious processes (e.g., Bargh,
1994; Blair & Banaji, 1996). Following initial node activation, a process of
spreading activation occurs in which related information about the group is
automatically activated until lexical decay occurs (Anderson, 1983;
Stephan, 1989).
In recent models of stereotyping, a conceptual distinction is drawn be­
tween stereotype activation and stereotype application (Blair & Banaji,
1996; Fiske & Neuberg, 1990; Gilbert & Hixon, 1991). Stereotype activa­
tion occurs when the stereotype is brought to mind through various means
(e.g., being asked to think about a group of people) or modalities (e.g., view­
ing a human face or listening to stereotypic remarks). Stereotype application
involves making a stereotypic judgment or response—for example, associat­
ing the word competence with the word male. However, for the perceiver to be
able to apply the stereotype (i.e., to make the judgment), the stereotype
must first be activated. According to this view activation is a necessary but
insufficient condition in stereotyping.

Methodological Techniques for the Study of Stereotyping

The primary methodology used in the experimental study of stereotype activa­
tion has been semantic priming. In the typical single-word priming experiment,
participants are visually presented racially salient words, such as black or white,
and then asked to make a judgment or decision about stereotype-congruent
target words, such as lazy or smart. Reaction times of participants are faster
when the prime-target relation is congruent with the negative or positive ste­
reotypes associated with the relevant social category (e.g., Dovidio, Evans, &
Tyler, 1986). The majority of studies have used visually presented priming stim­
uli of different types, including words (primarily trait words, but also nouns re­
lated to gender (Banaji & Hardin, 1996; Dijksterhuis & Van Knippenberg,
1996; Dovidio, et al., 1986), scrambled sentences (Bargh, Chen, & Burrows,
264 BLUMENTRITT AND HEREDIA

1996), names of historical figures (Stapel, Koomen, & Van der Plight, 1997), so­
cial category labels (Rothbart, Sriram, & Davis–Stitt, 1996), and human faces
(Bargh et al., 1996; Chen & Bargh, 1997).
The particular priming method varies as well, with many studies using the
well–known priming technique in which the participant is consciously ex­
posed to the critical prime (bread), followed by the presentation of a related
(butter) or unrelated target (floor). However, some studies have used non–
conscious (i.e.., subliminal) methods in which the priming stimulus is pre­
sented below the participants' level of reliable detection. Devine (1989) used
a form of this method by having participants identify the location of the stim­
uli (stereotype-related words) briefly presented in the participants' parafoveal
visual field, quickly followed by a pattern mask. Other studies have relied on
masking techniques to prevent conscious recognition of the content of the
stereotype-related priming stimuli. For example, masking stimuli have been
used to subliminally present photographs of faces of different ethnicities
(Bargh et al., 1996; Chen & Bargh, 1997) and trait or nontrait words (Bargh
& Pietromonaco, 1982; Locke, MacLeod, & Walker, 1994).
In a related vein, Banaji, Hardin, and Rothman (1993) examined implicit
stereotyping using a priming technique in which individuals are exposed to
positive or negative trait words that are either semantically relevant or irrel­
evant to a brief story participants subsequently read (see Higgins, Rholes, &
Jones, 1977). The story involved a character named Donald who performed
various empirically established, ambiguous behaviors with respect to the
trait categories. Participants who were previously exposed to positive traits
rated Donald's behavior more positively than participants exposed to nega­
tive traits, and vice versa. Exposure to semantically irrelevant positive or
negative traits showed no influence on ratings. Using a version of this tech­
nique, Banaji et al. demonstrated that implicitly primed, stereotyped infor­
mation influenced judgments of targets consistent with the stereotype. For
example, participants exposed to primes connoting dependence (a
stereotypically female trait) subsequently rated a female target more de­
pendent than a male target who performed identical behaviors.
With very few exceptions (see Dijksterhuis & Van Knippenberg, 1996;
Heredia & Blumentritt, 2002; for examples of lexical decisions), the de­
pendent measures for most priming–based studies of stereotyping have been
reaction times to various kinds of judgments about the target stimuli. The
presumption (generally borne out by research) is that reaction times will be
faster for stereotype consistent prime-target relations than for stereotype in­
consistent, prime–target relations. Examples of the types of tasks used in­
clude judgments of pronouns (e.g., is it a male or female pronoun? Banaji &
Hardin, 1996), judgments of proper nouns (e.g., is it a male or female name?
Blair & Banaji, 1996), and judgments of whether a target stimuli could ever
be true of the prime category or was always false (Dovidio et al., 1986).
1 1. STEREOTYPES AND NONLITERAL LANGUAGE 265

Automaticity and Controlled Processes in Stereotype

Activation

A great deal of research has examined the extent to which stereotypes are au­
tomatically activated and the conditions under which their activation may be
consciously controlled (e.g., Banaji & Greenwald, 1995; Banaji & Hardin,
1996; Banaji et at., 1993; Blair & Banaji, 1996; Devine, 1989; Dovidio et al.,
1986). Automatic processing of social information has been extensively elab­
orated by Bargh and colleagues (e.g., Bargh, 1984, 1989, 1994; Bargh,
Chaiken, Govender, & Pratto, 1992) who proposed that automatic cognitive
processes are those that are largely involuntary, effortless, and unconscious.
They run on autopilot, so to speak, and are viewed as relatively static pro­
cesses that, once initiated, are not constrained by capacity limitations and are
not under the conscious control of the perceiver. Specific automatic processes
arise from repeated experience with a particular stimulus or stimulus domain
and are automatically triggered by exposure to the stimulus (Bargh, 1984).
Controlled processes are aspects of cognition believed to be under con­
scious, voluntary control and are effortful and limited by the availability of
attentional energy (e.g., Bargh, 1989; Shiffrin & Dumais, 1981). Conscious
control over activation can occur only if the perceiver is given enough pro­
cessing time to allow suppression (or inhibition) of activation or if the
perceiver has a conscious intention to suppress activation (Bargh, 1989;
Blair & Banaji, 1996). An early influential study by Neely (1977) examined
specific experimental conditions under which automatic processing of sin­
gle words can occur. For one condition, participants were presented with
words that primed their own semantic category (e.g., BODY–arm); for the
second condition, participants were told to expect a target from a semanti–
cally unrelated category when given a certain prime word. For example, they
were told to expect the name of a type of bird when presented with the
BODY prime (e.g., BODY–sparrow). For both conditions, the experimental
task was a lexical decision, with prime-target intervals set at 250 ms and
2000 ms. Neely found that for the short prime–target interval (250 ms), the
prime facilitated lexical decisions to semantically related target words irre­
spective of experimenter instructions. At the longer prime–target interval
(2000 ms), the conscious expectancy conditions produced facilitation for
the expected target and inhibition for unexpected targets regardless of the
targets' semantic relation to the prime. Neely (and others) have taken these
results to indicate that automatic activation occurs at very brief time inter­
vals (i.e., under very high cognitive constraints); at longer time intervals,
conscious attention is allowed to develop and to potentially inhibit the auto­
matic processes (Devine, 1989; Logan, 1980).
Using designs that assess on-line automaticity effects, Blair and Banaji
(1996) reported a series of experiments that examined automatic activation
266 BLUMENTRITT AND HEREDIA

of gender stereotypes under baseline conditions of automaticity (i.e.,

prime-target interval less than 500 ms) and under conditions designed to
counteract automatic stereotyping. They used a semantic priming proce­
dure in which the primes were either positively or negatively valenced
stereotypically masculine traits (e.g., bold and crude, respectively), feminine
traits (e.g., gentle and fickle, respectively), or gender neutral non–traits (e.g.,
birthday and horror, respectively). Targets consisted of male or female proper
nouns. Results indicated automatic activation of the stereotype when the
prime–target interval was very brief (350 ms), and no explicit intention to
counter the stereotype was created. When given an explicit strategy to
counter the stereotypical association, participants were able to significantly
reduce (but not eliminate) stereotype priming, even under high cognitive
constraint. Further, when given more processing time, in combination with
the counterstereotype strategy, stereotype priming was reversed. A second
on–line study by Banaji and Blair (1996) revealed that gender stereotypes
are automatically activated even when the judgment task is unrelated to
gender and even if the participants are unaware of the gender relation of the
prime-target pairings.
In summary, the study of the social stereotype—its cognitive structure,
processes, and influence on behavior—has been an extremely active area in
the extant sociocognitive literature. Drawing on models and methods from
cognitive psychology, social psychologists have delineated many of the
mechanisms underlying stereotyping, and they have highlighted important
areas that need further investigation. Our work, too, is based in the socio­
cognitive literature. However, we also offer a new and different perspective
on stereotypes. It is to this perspective to which we now turn.

FIGURATIVE LANGUAGE AND STEREOTYPES

What is the relationship between figurative language and stereotypes? To il­

lustrate consider the sentence, (2) "Be careful James, those SHARKS will tell
you one thing today and tomorrow say something completely different." The
term sharks in this case is clearly not describing a marine animal. It describes
the idea that lawyers are vicious and cunning. This type of figurative language
is known as metaphoric reference (Gibbs, 1990, 1994; Onishi & Murphy,
1993; Stewart & Heredia, 2002), and it contrasts with predicate metaphor of
the form "A(n) is a B" (e.g., lawyers are sharks). In predicate form, metaphor
consists of a topic (e.g., lawyers) and a vehicle (e.g., sharks). When these
domains are combined (i.e., "lawyers are sharks"), what is common about the
two domains comprises the metaphorical meaning of the statement which is
referred to as the "ground." Finally, in referential metaphor, the topic and ve­
hicle occur apart from one another. The vehicle is usually made explicit, and
the listener must abstract the topic from the previous context. For example, to
1 1. STEREOTYPES AND NONLITERAL LANGUAGE 267

understand (2), information relevant to the topic (e.g., lawyer) must come be­
fore the metaphoric reference; otherwise, it would be extremely difficult to
make sense of such an utterance. Notice also that the metaphoric reference
can be generalized to all lawyers. In this case, it would be possible to say that all
lawyers are vicious and cunning and that one would have to be careful with
them, especially when it comes to monetary issues. Such logic could also be
applied to other instances in which, for example, all surgeons are viewed as
butchers, all women are described as female dogs, and all men are viewed as
pigs. In fact, it is not surprising to hear women lament that all men are pigs. We
contend that it is possible to view and study social stereotypes as extensions of
metaphoric referential descriptions.
In addition, stereotypes can also be viewed and classified as parallel cases
of metonymic reference. In metonymy, a salient aspect of an entity is used to
refer to the entity as a whole or to some other part of the entity (Frisson &
Pickering, 1999, pp. 1366–1367). That is, the name of one thing is used for
that of another of which it is an attribute or with which it is associated. For
example, in (3) "We need a new glove to play third base" (Gibbs, 1993), the
metonymic reference glove maps a salient characteristic of one domain. The
glove in this case is part of the baseball player that in turn represents the en­
tire domain player. The connection between the two domains form a rela­
tionship between part (glove) and whole (baseball player). In metonymic
comparisons, the two concepts being compared must belong to the same
conceptual domain (Gibbs, 1990, 1993, 1994; also see Frisson & Pickering,
1999 for a comprehensive list of different types of metonymy).
How are stereotypes related to metonymic references? Let us reconsider
our earlier example of terrorists from (1), "It is important to assure that
TERRORISTS do not access airports." The term terrorist has become a ste­
reotype attributed to the Arab community. It can be argued that this stereo­
type is being used as metonymic reference due to the perceived notion that
because some Arabs have been involved in terrorist attacks (a salient char­
acteristic of this perceived generalization), it represents the entire domain
(all Arabs). Following this reasoning, other ethnic stereotypes are possible
such as (4) "Jane stopped and said hello to the BEANER who was reading a
novel." In (4), the stereotype may describe a Mexican or a Hispanic individ­
ual. The idea here is that this ethnic group becomes associated with the as­
sumed food that they consume. That is, if Mexicans eat beans, they must be
described as beaners, and this follows the somewhat risible general principle
that you are what you eat.
Therefore, stereotypes can be seen as composed of a nonliteral and a lit­
eral interpretation. In the case of the stereotype beaner, the nonliteral mean­
ing describes the Mexican ethnic group, and the literal meaning describes
the direct related association to food. Now, the question of interest is, how
are stereotypes processed? Are people able to access the nonliteral sense of
268 BLUMENTRITT AND HEREDIA

the stereotype directly (e.g., Banaji & Greenwald, 1995; Banaji & Hardin,
1996; Banaji et al., 1993; Blair & Banaji, 1996; Devine, 1989)? Or, during
stereotype processing, must the literal interpretation of the stereotype be
triggered first and only after the nonliteral interpretation can be triggered
(cf.Glucksberg& Keysar, 1993;Grice, 1975;Searle, 1979)? Or, can both the
literal and nonliteral interpretations be triggered simultaneously? (e.g., Fris­
son & Pickering, 1999; Swinney, 1979; and also see Simpson, 1994). Before
reviewing our work on stereotype processing using this perspective, we
briefly review relevant studies in metaphoric and metonymic reference.

EVIDENCE ON FIGURATIVE REFERENCE

Metaphoric Reference

In one of the first studies to investigate this issue, Gibbs (1990) had partici­
pants read paragraphs such as (1a) in Table 11.1. The purpose of this manip­
ulation was to determine comprehension differences between metaphoric
and literal reinstatements in reading. In the literal reinstatement condition,
the final sentence referred to its antecedent with a word that was a synonym
of the person or thing being described by the context (e.g., fighter). For the
metaphoric reinstatement condition, the final sentence referred to its ante­
cedent with a metaphoric term (e.g., creampuff) that was semantically felici­
tous to the context of the passage. The participants' objective was to simply
read and comprehend each sentence one at a time. Reading times were re­
corded at the final sentence of the passage (metaphoric vs. literal). The re­
sults from the reading task showed that final sentence containing
metaphoric reinstatements took longer to comprehend than the sentence
containing literal reinstatements. This result was interesting because, as
suggested by the accompanying probe recognition task, participants were
indeed able to make the connection between the metaphoric reference
creampuff and the antecedent boxer described by the preceding context.
In a follow up study, Onishi and Murphy (1993) suggested that using the
vehicle creampuff to reference the topic boxer may have required readers to
infer semantic information beyond that provided by the story context. To
address this possibility, Onishi and Murphy enriched Gibbs' (1990) stimuli
so that there would be as little confusion as possible as to whom or what in
each story might be serving as the topic of the figurative description. Re­
gardless, in two different reading time experiments, Onishi and Murphy
were unable to show that comprehension of nonliteral reference could be as
effortless as understanding comparable literal reference. Thus, Gibbs and
Onishi and Murphy's results supported the idea that during reading com­
prehension it is more difficult to understand metaphoric than literal referen­
tial descriptions. With respect to whether literal or nonliteral language can
1 1. STEREOTYPES AND NONLITERAL LANGUAGE 269

be accessed faster, the results from these two experiments suggest that literal
processing has precedence during language processing (also see Swinney &
Osterhout, 1990; and cf. Gerrig & Healy, 1983; Gildea & Glucksberg, 1983;
Keysar, 1989; McGlone, Glucksberg, & Cacciari, 1994).
However, in a recent study, Stewart and Heredia (2002) examined the
activation of the antecedent during real–time comprehension of meta–
phoric reference. Unlike the original studies by Gibbs (1990) and Onishi
and Murphy (1993), Stewart and Heredia used the cross–modal lexical
priming technique (CMLP; Swinney, 1979), a psycholinguistic task
known to be sensitive to semantic and associative relations as well as con­
textual effects (e.g., Swinney & Osterhout, 1990; Tabossi, 1988, 1996).
The CMLP was used to study the activation of the literal and metaphoric
referential descriptions throughout the sentence at Positions 1 (at prime
offset, depicted by the subscript [**1] in Table 11.1, paragraph 1b), and 2
(1000 ms after prime offset). As participants listened to sentences of the
type in paragraph 1b, at the offset of the critical prime creampuff (Position
1), for example, a literal–related (e.g., pastry) and its controlled–unrelated
probe were presented in the middle of a computer screen. This probe re–

TABLE11.1
Sample Stimuli From Gibbs (199O) and Stewart and Heredia (2OO2)

Paragraph la: Metaphoric Reference From Gibbs (1990, p. 59)

Stu went to see the Saturday night fights. There was one boxer that Stu hated. This guy
always lost. Just as the match was supposed to start, Stu went to get some snacks. He
stood in line for ten minutes. When he returned, the bout had been cancelled. "What
happened?" Stu asked a friend. The friend replied,
"The creampuff didn't even show up." (metaphoric reinstatement)
"The fighter didn't even show up." (literal reinstatement)
Paragraph 1b: Metaphoric Reference From Stewart and Heredia (2002, p. 36)
Stu and his buddy went to see the Saturday night fights. There was one fighter they both
hated because the guy always lost. Just as the match was about to begin, Stu went to get
some snacks. When he returned he saw that the bout had been cancelled. "What
happened?" Stu asked. His friend replied,
"Aw, the creampuff[**1] didn't even show[**2] up, I can't believe it!"
(metaphoric probe: boxer) vs. (literal probe: pastry)
Paragraph 1c: Metonymic Reference From Gibbs (1990, p. 60)
Mr. Bloom was manager of a high–school baseball team. He was concerned about the
poor condition of the field. He also was worried about one athlete. His third baseman
wasn't a very good fielder. This concerned the manager a good deal. The team needed all
the help it could get. At one point, Mr. Bloom said to his assistant coach,
"The glove at third base has to be replaced" (metonymic reinstatement)
"The player at third base has to be replaced" (literal reinstatement)
Note. Probes presented at prime offset are noted as **.
27O BLUMENTRITT AND HEREDIA

fleeted the literal meaning of creampuff associated to bread. The related

nonliteral form, contrarily, reflected the meaning of the prime word im­
plied by the context (e.g., boxer). The participants' objective was to simply
listen to the sentences presented over headphones and pronounce the
probe appearing on the computer screen. To summarize, the results re­
vealed significant activation for the nonliteral interpretation and no acti­
vation for the literal interpretation of the metaphoric referential prime.
That is, it was possible to access the figurative meaning but only when
measurement took place immediately after the metaphoric reference term
(i.e., the critical prime).These results contrasted with Gibbs' (1990) and
Onishi and Murphy's (1993) findings in which readers experienced great
difficulty comprehending sentences containing the same nonliteral de­
scription. Thus, according to Stewart and Heredia (2002), access to the
nonliteral interpretation of a metaphoric reference is possible provided
that appropriate and sensitive methods to early stages of language process­
ing are used (cf. Dascal, 1989; Frisson & Pickering, 1999; Heredia, 2002).
Next we review two important studies in metonymic reference.

Metonymic Reference

Again, in one of the few metonymy experiments, Gibbs (1990) had partici­
pants read a passage such as the one in Table 11.1, paragraph 1c. Reading
times were recorded for the final sentence (metonymic vs. literal reinstate­
ment) . Notice that in the metonymic referential description, the glove is
meant to refer to the baseball player being described by prior context. Player,
contrarily, is the literal and direct meaning of the contextual information
preceding the target sentence. Similar to metaphoric referential descrip­
tions, the results showed that participants were faster to comprehend literal
than metonymic reinstatements. Although participants had more difficulty
in comprehending metonymic reinstatements, evidence by the accompany­
ing recognition task showed that participants were actually able to make the
connection between the metonymic reference and the antecedent informa­
tion related to the player. Although both metonymic and metaphoric refer­
ential processing appeared to have behaved in the same manner by showing
similar patterns, a subsequent experiment comparing both tropes showed
that, indeed, metaphoric referential statements were easier to comprehend
than metonymic references. In both cases, the literal reinstatement was eas­
ier to comprehend than both tropes. However, in a more recent study using
an eye–tracking paradigm (e.g., Tanenhaus & Spivey–Knowlton, 1996),
Frison and Pickering (1999) demonstrated that, like other research on am­
biguity (e.g., Seidenberg, Tanenhaus, Leiman, & Bienkowski, 1982;
Swinney, 1979), both the metonymic and literal referential descriptions can
be accessed simultaneously. Clearly, more research is needed in both areas to
11. STEREOTYPES AND NONLITERAL LANGUAGE 271

investigate metaphoric and metonymic reference to rule out the possibility

that differences among Gibbs (1990), Stewart and Heredia (2002), and
Frison and Pickering (1999) are merely methodological ones. In any case,
the studies reviewed in this section provide both a conceptual and method­
ological framework to study stereotype processing under a psycholinguistic
and, more specifically, a figurative language perspective. We now return to
our discussion on stereotype processing.

STEREOTYPE PROCESSING

Our exploration of stereotypes differs from that of Stewart and Heredia

(2002) and Heredia and Stewart (2002). Again, our general theoretical
framework is that social stereotypes can be viewed as an extension of figura­
tive language and, more specifically, to include metaphoric and metonymic
relations. In this view, the social stereotype terrorist is seen as containing a
literal meaning (e.g., bomb), which is directly associated with acts of destruc­
tion, and a nonliteral meaning (i.e., Arabs). The nonliteral meaning, in this
case, represents the generalization that when people think about the term
terrorist, they immediately activate information related to Arabs. In our re­
search, we asked the following questions: How do people process social ste­
reotypes? Are stereotypes processed in such a way that during
comprehension only the nonliteral sense is activated? To what extent is
prior context necessary to active the nonliteral sense of the stereotype? In
the following section we attempt to answer these very important questions.
In perhaps the only study of its kind, we (Heredia & Blumentritt, 2002)
employed the CMLP task to examine the automatic on–line activation of
stereotypes during spoken language. Participants listened to complete sen­
tences presented aurally, and then they made yes or no lexical decisions to
visually presented target words. As previously discussed, this type of task has
been widely used in psycholinguistic studies, and its sensitivity to semantic
and associative relations, as well as contextual effects, is well documented.
In addition, we were interested in using a method that increased the general
ecological validity of the study of stereotypes. Specifically, we maintained
that exposure to stereotypes normally occurs during the course of spoken
conversation (see also van Dijk, 1984, 1987). For example, a common sce
nario is that a conversational partner makes an offhand racist or sexist re­
mark, or he or she uses a stereotypic term to reference a particular social
group or, perhaps, tells a sexist or racist joke that relies on stereotype infor­
mation for its humor value. The point is that we typically hear stereotypical
remarks rather than, say, read them in print form. Although many people
subscribe to joke listservs and receive all kinds of jokes describing certain
ethnic groups stereotypically, in this case such jokes have to be read. But
Stewart and Heredia (2002, p. 35) pointed out, "speech contains a number
272 BLUMENTRITT AND HEREDIA

of prosodic cues that we as listeners have come to rely on to resolve ambigu­

ity in an expeditious manner." Our goal then was to examine activation of
both the stereotype meaning and the dominant meaning during spoken lan­
guage and as language unfolds (cf. Frisson & Pickering, 1999). The CMLP
task is an appropriate methodology to do just this. We used as our dependent
measure reaction times to a lexical decision. We wanted to examine
real–time processing of stereotypes, outside of postlexical episodic memory
influences. A lexical decision task gives us a purer measure of activation, es­
pecially when compared to other kinds of judgment tasks that have been
used (e.g., Blair & Banaji, 1996; Devine, 1989). For example, in some tasks
used in the stereotype literature, participants are given a particular stereo­
type such as secretary and then they are asked to judge whether the stereo­
type is more relevant to a male or female employee. Selections are made by
pressing a computer button labeled m for male or f for female as fast and as ac­
curately as possible (e.g., Banaji & Hardin, 1996). To say the least, this tech­
nique is hardly an example of an implicit task. Moreover, by using the CMLP
we were able to take advantage of the priming effect to assess the extent to
which a particular meaning was activated or inhibited. According to this
logic, significant priming facilitation signified that the meaning of the par­
ticular concept of interest was indeed activated. Third, we systematically
manipulated the effects of contextual information on the comprehension of
social stereotypes. A well–known finding from the psycholinguistic litera­
ture is that lexical activation is faster when the word appears within a con–
textually appropriate sentence than when it appears in a contextually
inappropriate or neutral sentence (Simpson, 1994; Tabossi, 1988; Tabossi,
Colombo, & Job, 1987). Here, we report here a series of four experiments
that addressed these issues. Table 11.2 shows sample sentences and the ex­
perimental manipulations. Target words are presented immediately at the
offset (0 ms) of the critical prime (probe [*1] ) or 300 ms after the offset of the
critical prime (probe [*2] ).
As shown in Table 11.2, for the contextually unbiased conditions (Exper­
iments 1 and 2), the portion of the sentence preceding the critical prime
(e.g., terrorist) provided no specific information about its nonliteral (e.g.,
Arabs) or its literal meaning (e.g., bombs). For experiment 1, we wanted to
see if activation of the nonliteral word would occur immediately at prime
offset (0 ms). Based on the sociocognitive literature (e.g., Blair & Banaji,
1996; Banaji & Hardin, 1996), we expected that it would. To assure that the
nonliteral stereotype meaning was not less familiar than the literal term,
thus taking longer time to achieve full activation, in Experiment 2 we pre­
sented the critical targets 300 ms after prime offset. We predicted that with
greater processing time we would see increased activation of the nonliteral
meaning. Finally, for both prime offset conditions, we expected facilitation
priming for the literal targets (Swinney & Osterhout, 1990).
1 1. STEREOTYPES AND NONLITERAL LANGUAGE 273

In Experiments 3 and 4 (see Table 11.2) we evaluated the effects of con­

textual information on the processing of social stereotypes. To what extent is
the nonliteral word dependent on an appropriate context to trigger its acti­
vation? We drew here from the literature on lexical ambiguity to inform our
hypotheses and expected that, with the addition of a biasing context, we
would see even greater activation of the nonliteral word (e.g., Simpson,
1994). As shown in Table 11.2, in the contextually biased conditions, the
term Middle East is included in the sentence prior to the critical prime. Our
logic was that Middle East would bias the sentence toward the stereotype
that all people from the Middle East (i.e., Arabs) are terrorists. Moreover, we
evaluated time course effects in a contextually biased condition by present­
ing the targets at 0 ms and 300 ms prime offset. We predicted that with in­
creasing processing time context would become even more powerful in its
effect on activation and reaction times would become increasingly faster to
the nonliteral words (e.g., Tabossi, 1988; Tabossi et al., 1987).
Across all experiments the literal forms of the critical prime showed
facilitatory priming. In fact, the priming effect was very similar across experi­
ments. However, contrary to our predictions, the nonliteral forms generally
demonstrated inhibitory priming. That is, we found slower reaction times to
the nonliteral meanings of the critical prime relative to unrelated control
words. And, as shown in Tables 11.3 and 11.4, reaction time differences
clearly indicate a pattern of increasing inhibition of the nonliteral forms. To
summarize, in the absence of any contextual information (Experiment 1),
processing of nonliteral targets was no different than the processing of unre–

TABLE 11.2
Sample of Sentences Used in the Heredia and Blumentritt (2OO2) Study

Experiment 1 and 2: Contextually Unbiased Toward the Stereotype

Pete realizes the importance of assuring that TERRORISTS [Probe*1] do not have access
[Probe*2] to airports or any other type of security related system.

Experiment 3 and 4: Contextually Biased Toward the Stereotype

Pete has lived in the Middle East and realizes the importance of assuring that
TERRORISTS [Probe*1] do not have access [Probe*2] to airport systems or any other related
system.

Sample Targets Related Unrelated Control

Literal: BOMB YARD
Nonliteral: ARAB CAIN

Note. From "On–line Processing of Social Stereotypes During Spoken Language Comprehension," by
R. R. Heredin and T. L. Blumentritt, 2002, Experimental Psychology, 49, p. 211. Copyright 2002 by
Hogrefa and Huber. Reprinted with permission.
274 BLUMENTRITT AND HEREDIA

TABLE 11.3
Priming Effects for Experiment l Prime Offset (O ms)
and Experiment 2 (3OO ms)

Prime Offset
Visual Probes
Probes (Targets) 0 ms 300 ms

Literal (bomb) 58* 45*

Nonliteral (Arab) 5 –26*

Note. Lexical decision results are given as (control – experimental) priming reaction times in milli­
seconds; *p < .05.

TABLE 11.4
Priming Effects for Experiment 3 Prime Offset (0 ms)
and Experiment 4 (3OO ms)

Prime Offset
Visual Probes
Probes (Targets) 0 ms 300 ms

Literal (bomb) 55* 50*

Nonliteral (Arab) –31* –39*

Note. Lexical decision results are given as (control — experimental) priming reaction times in ms;
*p < .05.

lated control targets. However, by increasing the processing time window of

the critical stereotype (prime) to 300 ms (Experiment 2), reaction times to the
nonliteral targets were significantly reduced, producing inhibition. In addi­
tion, when we added a biasing context we saw inhibition immediately (Experi­
ment 3) and even greater inhibition after a 300–ms delay (Experiment 4).
What should we make of this interesting pattern of results? First, we won­
dered if our participants were simply inadequately familiar with the stereo­
types used in the experiments. Blasko and Connine (1993) examined lexical
decision reaction times to targets that were either figuratively (secure) or lit­
erally related (hard) to a metaphor ("the family is a rock"). They found that
reaction times were slower for figuratively related words, but only when the
metaphor was relatively unfamiliar. They also found that when they in­
creased prime offset time from 0 ms to 300 ms they produced decreased inhi­
bition of the figurative form. This contrasted with our results, which
1 1. STEREOTYPES AND NONLITERAL LANGUAGE 275

indicated increasing inhibition with increased processing time. Note that

we also pretested our word pairs (e.g., terrorist–Arab) for familiarity. Mean
familiarity rating for our stimuli was 4.3 (SD = .906), well beyond the 3.9
criterion proposed by Blasko and Connine (1993).
Second, if low familiarity was not a plausible explanation, could it be that
the stereotype was simply less dominant than the standard meaning and
that during processing it was not only poorly activated, but actually inhib­
ited? The sociocognitive literature on selective attention and person per­
ception (e.g., Dijksterhuis & Van Knippenberg, 1996; Macrae,
Bodenhausen, & Milne, 1995; Macrae, Milne, & Bodenhausen, 1994;
Neumann & DeSchepper, 1992) suggests this explanation. In general, this
literature shows that not only are dominant stimuli automatically activated
(i.e., show a pattern of facilitation), but subordinate, less dominant stimuli
are actively inhibited. Apparently, as Macrae et al. (1995) noted, the subor­
dinate meaning "is not simply neglected; instead, it is actively dampened
through a spreading inhibition process, even to levels below its resting state
... one advantage of active inhibition ... lies in the suppression of poten­
tially distracting (hence disruptive) mental representations" (p. 399). Per­
haps a similar process occurred in our study in which the literal and more
dominant meaning was automatically activated, whereas the nonliteral
meaning was dampened. With more processing time and a biasing context,
this dampening effect was allowed to develop and becomes even more po­
tent in its effect and, consequently, produced even greater inhibition.
Third, we considered the possibility that some sort of controlled process­
ing of the stereotype targets occurred. Controlled processing of stereotype-
related information has been reported in the literature (e.g., Blair & Banaji,
1996; Devine, 1989), although it has not been previously demonstrated in
on–line tasks under conditions of high cognitive load (i.e., less than 500 ms
prime offset) and without a specific strategy to counter the stereotype.
Nonetheless, it is conceivable that the stereotype-related primes and associ­
ated targets evoked feelings of uneasiness or even discomfort in our partici­
pants (who were primarily Mexican American) which then produced an
inhibitory response. This is speculative at this point because, to our knowl­
edge, no other study in the sociocognitive literature has examined stereo­
type processing in an ethnic minority sample. Moreover, it is clear from
previous work that (at least in off–line tasks) individuals who are motivated
to do so are able to inhibit activation of stereotypes (e.g., Devine, 1989;
Kunda, 1999).
What are the theoretical implications of our results? Certainly one is that
they point to some sort of inhibitory mechanism involved in stereotyping
that remains poorly understood. We propose that further study needs to be
conducted to more fully understand the specific mechanisms underlying
this inhibitory process. Relatedly, it is reasonable to conclude that automatic
276 BLUMENTRITT AND HEREDIA

activation of stereotypes may not occur as previously theorized (e.g.,

Banaji & Greenwald, 1995; Banaji & Hardin, 1996; Banaji et al., 1993;
Blair & Banaji, 1996; Devine, 1989; Dovidio et al., 1986), and further
work needs to be done to understand under which specific conditions
automaticity of stereotype activation occurs. It may be that with certain
populations who are particularly sensitive to the effects of stereotyping the
accepted boundaries of automaticity (i.e., 500 ms) will need to be revised.
What are the real–life implications of stereotype inhibition? Are some peo­
ple more likely to inhibit stereotypes than others? For example, are there
certain person variables, such as prejudice level, that systematically influ­
ence stereotype processing? And, perhaps most important, how does inhi­
bition (or activation) at the moment of processing translate into attitudes,
impressions, expectations, and behaviors? No doubt these questions have
been addressed at the postaccess level of analysis in the sociocognitive lit­
erature (e.g., Banaji & Greenwald, 1995; Banaji et al., 1993; Chen &
Bargh, 1997; Devine, 1989; Dijksterhuis& Corneille, 2000; Levy, 1996).
However, very few (if any) studies have addressed these theoretically and
practically important questions utilizing methods that enable investiga­
tion of on–line processing of stereotype-related information. It is our con­
tention that psycholinguistic concepts and associated methods provide an
interesting way to approach these issues. Before closing our discussion, we
address the issue as to whether stereotypes are triggered automatically
during the communicative process. In a subsequent study, we attempted to
replicate our results of Experiment 2 with the same methods described in
Table 11.2, but this time we utilized a naming task instead of the lexical de­
cision task. We used a naming task because naming is thought to provide a
deeper and less strategic measure of lexical access (e.g., Hernandez, Bates,
& Ávila, 1996). Needless to say, we were unable to replicate the results of
Experiment 2. These results and the finding that no activation for the
nonliteral meaning was evident in Experiment 1 at prime offset, but acti­
vation in both Experiment 2 and 3 has lead us to conclude that stereotypes
may not be as automatic as once thought. Thus, to comprehend the
nonliteral sense of a stereotype, prior biasing context should be provided.
Alternatively, about 300 ms are needed to fully process the social stereo­
type and grasp the implied meaning of it.

FINAL REMARKS

The purpose of our chapter was to provide a novel perspective on how stereo­
types can be studied. We believe that stereotypes are closely related to both
metaphoric and metonymic referential descriptions. Our overall argument is
that theoretical and methodological advances in figurative language and am­
biguity resolution could be used as a tool to understand how humans process
1 1. STEREOTYPES AND NONLITERAL LANGUAGE 277

stereotypes during the communication process. Although we clearly favor re­

search techniques such as the ones described in Heredia and Blumentritt
(2002) and Stewart and Heredia (2002), future research would benefit from
replicating our results using other reading techniques that rely on priming. It is
important to underscore that Heredia and Blumentritt's (2002) results are
based largely on Mexican Americans. The next step should be to replicate our
results with Anglo Americans. That is, we would like to see our results ex­
tended and replicated both cross-regionally (within the United States) and
cross-nationally.
We argued the possibility that stereotypes can be seen as extensions of
metaphoric and metonymic referential descriptions. Certainly, this view
may or may not apply to all stereotypes; that is, some stereotypes may not
meet the exact criteria for metaphoric or metonymic referential descrip–
tions. Ultimately, the researcher should decide if the stereotype can be con­
sidered a metonymy or a metaphoric reference. Finally, we note that in
Heredia and Blumentritt (2002) no distinction was made between possible
metaphoric– and metonymic-like stereotypes. And, it is quite probable that
our results would not have changed given the similar patterns observed by
Gibbs (1990) for both kinds of tropes. However, one direction for future
work would be to analyze social stereotypes from a metaphoric and meto­
nymic language perspective. It would be interesting to observe if differential
processing effects emerge for different kinds of language-based interpreta­
tions of social stereotypes. In summary, we echo the comments of Maas and
Arcuri (1996) who observed that "stereotypes are closely—if not insepara-
bly—linked to language. Indeed, it is difficult to imagine language-free ste­
reotyping ... our knowledge of stereotypes will remain incomplete without
an analysis of the language that defines a given stereotype" (p. 220). We
would simply add to this by reiterating that a very useful and, we believe, po­
tentially fruitful way of studying stereotypes is by viewing them as a special
case of figurative language. This will open new research avenues, and new
information will emerge.

REFERENCES

American Civil Liberties Union: Freedom Network. (1995, November 9). ACLU files
claim against U.S. government on behalf of Jordanian—American wrongly detained for
Oklahoma City bombing. Retrieved July, 8, 2002, from http://www.aclu.org/
news/nl 109951.html
Anderson, J. R. (1983). The architecture of cognition. Cambridge, MA: Harvard Uni­
versity Press.
Banaji, M. R., & Greenwald, A. G. (1995). Implicit gender stereotyping in judg­
ments of fame. Journal of Personality and Social Psychology, 68, 181–189.
Banaji, M. R., & Hardin, C. D. (1996). Automatic stereotyping. Psychological Sci­
ence, 7, 136–141.
278 BLUMENTRITT AND HEREDIA

Banaji, M. R., Hardin, C., & Rothman, A. J. (1993). Implicit stereotyping in person
judgment. Journal of Personality and Social Psychology, 65, 272–281.
Bargh, J. A. (1984). Automatic and conscious processing of social information. In R.
S. Wyer, Jr., & T. K. Srull (Eds.), Handbook of social cognition (Vol. 3, pp. 1–43).
Hillsdale, NJ: Lawrence Erlbaum Associates.
Bargh, J. A. (1989). Conditional automaticity: Varieties of automatic influence in
social perception and cognition. In J. S. Uleman & J. A. Bargh (Eds.), Unintended
thought (pp. 3–51). New York: Guilford.
Bargh, J. A. (1994). The four horsemen of automaticity: Awareness, intention, effi­
ciency, and control in social cognition. In R. S. Wyer & T. K. Srull (Eds.), Hand­
book of social cognition (2nd ed., pp. 1–40). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Bargh, J. A., Chaiken, S., Govender, R., & Pratto, F. (1992). The generality of the au­
tomatic activation effect, Journal of Personality and Social Psychology, 62, 893—912.
Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social behavior: Di­
rect effects of trait construct and stereotype activation on action. Journal of Per­
sonality and Social Psychology, 71, 230–244.
Bargh, J. A., & Pietromonaco, P. (1982). Automatic information processing and social
perception: The influence of trait information presented outside awareness on im­
pression formation. Journal of Personality and Social Psychology, 55, 599–605.
Blair, I. V, & Banaji, M . R. (1996). Automatic and controlled processes in stereo­
type priming. Journal of Personality and Social Psychology, 70, 1142–1163.
Blasko, D. G., & Connine,C. M. (1993). Effects of familiarity and aptness on meta­
phor processing.Journal of Experimental Psychology: Learning, Memory, and Cogni­
tion, 19, 295–308.
Bodenhausen, G. V. (1988). Stereotypic biases in social decision making and mem­
ory: Testing process models of stereotype use. Journal of Personality and Social Psy–
chology, 55, 726–737.
Chen, M., & Bargh, J. A. (1997). Nonconscious behavioral confirmation processes:
The self–fulfilling consequences of automatic stereotype activation. Journal of
Experimental Social Psychology, 33, 541–560.
Cohen, C. E. (1981). Person categories and social perception: Testing some bound­
aries of the processing effects of prior knowledge. Journal of Personality and Social
Psychology, 40, 441–452.
Darley, J. M., & Gross, P. H. (1983). A hypothesis–confirming bias in labeling effects.
Journal of Personality and Social Psychology, 44, 20–33.
Dascal, M. (1989). On the roles of context and literal meaning in understanding.
Cognitive Science, 13, 253–257.
Devine, P. (1989). Stereotypes and prejudice: Their automatic and controlled com­
ponents. Journal of Personality and Social Psychology, 56, 5–18.
Dijksterhuis, A. P.,& Corneille, O. (2000). On the relation between stereotype activa­
tion and intellectual underperformance. Unpublished manuscript, University of
Amsterdam, Amsterdam, The Netherlands.
Dijksterhuis, A. P., & Van Knippenberg, A. D. (1996). The knife that cuts both
ways: Facilitated and inhibited access to traits as a result of stereotype activation.
Journal of Experimental Social Psychology, 32, 271–288.
1 1. STEREOTYPES AND NONLITERAL LANGUAGE 279

Dovidio, J. F., Evans, N., & Tyler, R. B. (1986). Racial stereotypes: The contents of
their cognitive representations. Journal of Experimental Social Psychology, 22,
22–37.
Fiske, S., & Neuberg, S. (1990). A continuum of impression formation, from cate-
gory–based to individuating processes: Influences of information and motivation
on attention and interpretation. Advances in Experimental Social Psychology, 23,
1–74.
Frisson, S., & Pickering, M. J. (1999). The processing of metonymy: Evidence from
eye movements. Journal of Experimental Psychology: Learning, Memory, and Cogni­
tion, 25, 1366–1383.
Gerrig, R. J., & Healy, A. F. (1983). Dual process in metaphor understanding: Com­
prehension and appreciation. Journal of Experimental Psychology: Learning, Mem­
ory, and Cognition, 9, 667–675.
Gibbs, R. W., Jr. (1990). Comprehending figurative referential descriptions. Journal
of Experimental Psychology: Learning, Memory, and Cognition, 16, 56–66.
Gibbs, R. W., Jr. (1993). Process and products in making sense of tropes. In A.
Ortony (Ed.), Metaphor and thought (2nd ed., pp. 252–276). New York: Cam­
bridge University Press.
Gibbs, R. W., Jr. (1994). The poetics of mind. New York: Cambridge University Press.
Gilbert, D., & Hixon, J. (1991). The trouble of thinking: Activation and application
of stereotypic beliefs. Journal of Personality and Social Psychology, 60, 509—517.
Gildea, P.,& Glucksberg, S. (1983). On understanding metaphor: The role of con­
text. Journal of Verbal Learning and Verbal Behavior, 22, 577–590.
Glucksberg, S., & Keysar, B. (1993). How metaphors work. In A. Ortony (Ed.), Meta­
phor and thought (2nd ed., pp. 401–424). New York: Cambridge University Press.
Grice, H. P. (1975). Logic and conversation. In P. Cole & J. Morgan (Eds.), Syntax
and semantics: Vol. 3: Speech acts (pp. 41–58). New York: Academic Press.
Heredia, R. R. (2002). Bilingual sentence processing: An examination of context effects
and cross–language priming. Unpublished manuscript, Texas A&M International
University, Laredo, TX.
Heredia, R. R., & Blumentritt, T. L. (2002). On–line processing of social stereotypes
during spoken language comprehension. Experimental Psychology, 49, 1–14.
Heredia, R. R., & Stewart, M. T. (2002). On–line methods in bilingual spoken lan­
guage research. In R. R. Heredia & J. Altarriba (Eds.), Bilingualsentence process­
ing (pp. 7–28). Amsterdam: Elsevier.
Hernandez, A. E., Bates, E. A., & Àvila, L. X. (1996). Processing across the lan­
guage boudary. A cross–modal priming study of Spanish–English bilinguals. Jour­
nal of Experimental Psychology: Learning, Memory, and Cognition, 22, 846–864.
Higgins, E. T., Rholes, W. S., & Jones, C. R. (1977). Category accessibility and im­
pression formation. Journal of Experimental Social Psychology, 13, 141–154.
Keysar, B. (1989). On the functional equivalence of literal and metaphorical inter­
pretations in discourse. Journal of Memory and Language, 28, 375–385.
Kunda, Z. (1999). Social cognition: Making sense of people. Cambridge, MA: MIT
Press.
Kunda, Z., & Sherman–Williams, B. (1992). Stereotypes and the construal of indi­
viduating information. Personality and Social Psychology Bulletin, 19, 90–99.
280 BLUMENTRITT AND HEREDIA

Levy, B. (1996). Improving memory in old age through implicit self–stereotyping.

Journal of Personality and Social Psychology, 71, 1092–1107.
Lippman, W. (1922). Public opinion. New York: Harcourt Brace Jovanovich.
Locke, V, MacLeod, C., & Walker, I. (1994). Automatic and controlled activation
of stereotypes: Individual differences associated with prejudice.British Journal of
Social Psychology, 33, 29–46.
Logan, G. D. (1980). Attention and automaticity in Stroop and priming tasks: The­
ory and data. Cognitive Psychology, 12, 523–533.
Maass, A., & Arcuri, L. (1996). Language and stereotyping. In C. N. Macrae, C.
Stangor, & M. Hewstone (Eds.), Stereotypes and stereotyping (pp. 193–226). New
York: Guilford.
Macrae, C. N., Bodenhausen, G. V, & Milne, A. B. (1995). The dissection of selec­
tion in person perception: Inhibitory processes in social stereotyping. Journal of
Personality and Social Psychology, 69, 397–407.
Macrae, C. N., Stangor, C., & Milne, A. B. (1994). Activating social stereotypes: A
functional analysis. Journal of Experimental Social Psychology, 30, 370–389.
McGlone, M. S., Glucksberg, S., & Cacciari, C. (1994). Semantic productivity and
idiom comprehension. Discourse Processes, 17, 167–190.
Neely, J. H. (1977). Semantic priming and retrieval from lexical memory: Roles of
inhibitionless spreading activation and limited-capacity attention. Journal of Ex­
perimental Psychology: General, 106, 226–254.
Neumann, E., & DeSchepper, B. G. (1992). An inhibition–based fan effect: Evi­
dence for an active suppression mechanism in selective attention. Canadian
Journal of Psychology, 46, 1–40.
Onishi, K. A., & Murphy, G. O. (1993). Metaphoric reference: When metaphors are
not understood as easily as literal expressions. Memory & Cognition, 21, 763–772.
Rothbart, M., Sriram, N., & Davis–Stitt, C. (1996). The retrieval of typical and
atypical category members. Journal of Experimental Social Psychology, 32,
309–336.
Searle, J. (1979). Expression and meaning. Cambridge, England: Cambridge Univer­
sity Press.
Seidenberg, M., Tanenhaus, M., Leiman, J., & Bienkowski, M. (1982). Automatic
access of the meanings of ambiguous words in context: Some limitations of
knowledge-based processing. Cognitive Psychology, 14, 489–537.
Shiffrin, R. M., & Dumais, S. T. (1981). The development of automatism. In J. R.
Anderson (Ed.), Cognitive skills and their acquisition (pp. 111–140). Hillsdale, NJ:
Lawrence Erlbaum Associates.
Simpson, G. B. (1994). Context and the processing of ambiguous words. In M. A.
Gernsbacher (Ed.), Handbook of psycholinguistics (pp. 359–374). San Diego, CA:
Academic Press.
Stapel, D. A., Koomen, W., &. Van der Plight, J. (1997). Categories of category ac­
cessibility: The impact of trait concept versus exemplar priming on person judg­
ments. Journal of Experimental Social Psychology, 33, 47–76.
Stephan, W. G. (1989). A cognitive approach to stereotyping. In D. Bar-Tal, C. F.
Graumann, A. W. Kruglanski, & W. Stroebe (Eds.), Stereotyping and prejudice:
Changing conceptions (pp. 37–57). New York: Springer-Verlag.
1 1. STEREOTYPES AND NONLITERAL LANGUAGE 281

Stewart, M. T., & Heredia, R. R. (2002). Comprehending spoken metaphoric refer­

ence: A real–time analysis. Journal of Experimental Psychology, 49, 34–44.
Swinney, D. A. (1979). Lexical access during sentence comprehension: (Re)consid-
eration of context effects. Journal of Verbal Learning and Verbal Behavior, 18,
645–660.
Swinney, D. A., & Osterhout, L. (1990). Inference generation during auditory lan­
guage comprehension. Psychology of Learning, 25, 17–33.
Tabossi, P. (1988). Accessing lexical ambiguity in different types of sentential con­
texts. Journal of Memory and Language, 27, 324–340.
Tabossi, P. (1996). Cross–modal semantic priming. Language and Cognitive Processes,
11, 569–516.
Tabossi, P., Colombo, L., & Job, R. (1987). Assessing lexical ambiguity: Effects of
context and dominance. Psychological Research, 49, 161–167.
Tanenhaus, M. K., & Spivey–Knowlton, M. J. (1996). Eye–tracking. Language and
Cognitive Processes, II , 583–588.
van Dijk, T. A. (1984). Prejudice and discourse: An analysis of ethnic prejudice in cogni­
tion and conversation. Amsterdam: Benjamins.
van Dijk, T. A. (1987). Communicating racism: Ethnic prejudice in thought and talk.
Newbury Park, CA: Sage.
This page intentionally left blank
 12

On Mosquitoes and Camels:

Some Notes on the
Interpretation of Metaphorically
Transparent Popular Sayings *
Carmen Curcó
National Autonomous University of Mexico

All languages have proverbs and popular sayings that appear to be meta­
phorically transparent. Consequently, their interpretation should pose no
serious problems for normative speakers and second language learners. This
work explores an empirical observation: Their metaphorical transparency
does not necessarily lead to one single interpretation across subjects. For al­
most any popular saying there are a number of possible interpretations, all
compatible with its linguistic form. Normally, native speakers are unaware
of interpretations different from that generally accepted in their commu­
nity, but these tend to emerge naturally when fluent nonnative speakers
participate in joint conversations. For instance, for a Swedish saying that
translates roughly as "picking out mosquitoes and swallowing camels," the
interpretations of subjects unfamiliar with the expression fell into two dis­
tinct groups: For some it meant to attempt something very difficult; for oth­
ers it meant to pay excessive attention to insignificant details while
overlooking serious faults (the standard interpretation in Sweden), The aim
*I am grateful to Ana Villa for her enthusiastic and valuable participation in gathering the
data, and to Marilyn Buck for her valuable comments on a previous version of this paper. This
research was supported by project IN313001-DGAPA of the National Autonomous Univer­
sity of Mexico.
283
284 CURCÓ

of this chapter is to explore the extent to which this kind of variation is sys­
tematic and, if so, whether it can be traced to specific social, cultural, and
linguistic factors. I report the initial results of an exploratory cross-linguistic
study, wherein the preferred interpretations that native and nonnative
speakers gave to a set of metaphorical proverbs were considered. Subjects
came from a diversity of linguistic, social, and cultural backgrounds. Results
suggest that complex inferences are at work in the interpretation and learn­
ing process of these expressions. Some of them are triggered by lexical mean­
ings and associations specific to the languages with which a particular
subject is familiar, while others seem to be derived from cultural practices,
social knowledge, and salient values of specific groups.

INTRODUCTION

The study reported here arose from an empirical and accidental observation
involving a mild misunderstanding. A group of colleagues were having an
informal conversation about work issues. At a certain point, one colleague,
who was Swedish in origin but had lived and worked in Mexico for over 30
years, used a translated version of a Swedish proverb and uttered, "I then re­
alized we were sieving for mosquitoes and swallowing camels." The meaning
of this statement was unfamiliar to other colleagues, who nonetheless un­
derstood her expression and enjoyed it. Or so they thought, as it later
emerged that their interpretations fell squarely into two groups: those who
had (incorrectly) interpreted it as "we were doing something very difficult
and very absurd," and those who had got the intended point "we were get­
ting obsessed about unimportant details and overlooking serious faults."
Both interpretations made sense in the context of the conversation. When
commenting about this regular difference in interpretation, the Swedish
woman remarked that she thought it was because of the influence of the
country of origin of participants in the conversation (France, Greece, Mex­
ico, and United States). She advanced the hypothesis that people coming
from Protestant countries would tend to recover the Swedish proverbial
meaning more easily, because in those countries there is tradition in the reli­
gious practice of reading the Bible individually and as a direct source. In
Catholic countries, by contrast, people are less prone to read the Bible them­
selves, and they often have access to its content through priests and public
religious activities; thus, the biblical context of interpretation for proverbs
with a scriptural origin would be less accessible for them.
This anecdote gave rise to a more general query about the effect of the ac­
cessibility of cultural values, practices, and beliefs on the favored interpreta­
tion given to unfamiliar proverbs by foreigners. An exploration into this
interpretation process bears directly on a number of issues, and it should
have theoretical implications for our understanding of linguistic and con–
12. INTERPRETATION OF POPULAR SAYINGS 285

ceptual metaphor, proverbs, and the workings of verbal interpretation, in

addition to our second language teaching practices. In particular, it could
give some indications about the role—if any—of proverbs in testing and as­
sessing linguistic competence in first and foreign languages (Chapman et al.,
1997; Cheng, 1996; Nippold, Uhden, & Schwartz, 1997; Ormsby, personal
communication, December 2, 2001; Ulatowska, Chapman, Highley, &
Prince, 1998), an increasingly important issue that is not addressed here.
The question of how cultural elements impinge on the interpretation of
metaphorical proverbs involves many others. How do speakers interpret ut­
terances? Ho do we process figurative language? What is the nature of meta­
phor? Are concepts metaphorical themselves? What are the essential
features of a proverb? What counts as cultural concepts or beliefs? Are they
homogeneous across a given population? Is it possible to study cultural rep­
resentations systematically? None of these questions is unproblematic, and
we do not seem to have any definite answers to them yet. In what follows I
sketch a few assumptions made throughout this chapter, leaving aside the
major controversies surrounding them.
VERBAL COMMUNICATION AND THE INTERPRETATION
OF UTTERANCES

Linguistic underspecification seems to be widely recognized nowadays in the

semantics and pragmatics literature. Although there are a number of differ­
ent ways in which it can be understood depending on what is meant bylin­
guistic meaning, it remains fairly uncontroversial that every utterance has a
number of possible interpretations, all compatible with its linguistic form.1
There is also wide agreement that, given linguistic underdeterminacy, prag­
matic inference is required if a listener is to successfully recover a speaker's
meaning. What is less settled is the nature of the principles involved in the
pragmatic inferential process (for leading positions, see Grice, 1975/1989;
Levinson, 2000; and Sperber & Wilson 1986/21995) and whether pragmatic
additions are strictly guided by the underlying structure of the sentence ut­
tered or are relatively free of linguistic constraints (for a discussion, see
Carston, 2002; Stanley, 2002; and Stanley & Szabo 2000).
Here I assume that the interpretation of verbal stimuli is guided by the
search for an optimal balance between the cognitive gains obtained from
processing an utterance and the effort invested by the hearer in doing so.
This is what Sperber and Wilson (1986/21995) called the search for an inter­
pretation consistent with the principle of relevance. The automatic search

The underspecification thesis can take at least these forms: (a) Linguistic meaning
underdetermines speaker meaning (what is communicated), (b) linguistic meaning under–
determines explicitly communicated content, and (c) linguistic meaning underdetermines
the proposition expressed (see Carston, 2002, for a discussion).
286 CURCÓ

for relevance is an assumed inbuilt disposition rooted in the human cogni­

tive system.
Among the factors known to affect the processing effort invested in inter­
preting a linguistic expression are frequency and recency of use, the logical
and linguistic complexity of the expression, and the size and accessibility of
the intended context of interpretation. This last factor is especially impor­
tant in the study reported here, as I will show.
I also follow Sperber and Wilson (1986/21995) in viewing the context of
interpretation of an utterance as a psychological construct, a subset of he
hearer's beliefs and assumptions about the world that are mentally repre­
sented and employed as a set of premises in the derivation of cognitive ef­
fects (Sperber & Wilson, 1986, p. 16).
A hearer interpreting an utterance will follow a path of least effort: He or
she will start from the smallest set of the most accessible contextual assump­
tions in which the linguistic expression used can be processed to derive cogni­
tive effects and will stop expanding the context and/or searching for
alternative interpretations as soon as enough cognitive effects in exchange for
the processing effort invested have been derived. The first interpretation that
meets this criterion and is compatible with what the hearer knows about the
speaker's preferences and abilities will be selected as the intended one.
Members of the same cultural group share a number of experiences,
teachings views, beliefs, and assumptions about the world. Among the set
of assumptions mentally represented by any individual, those that are
shared by the cultural group he or she belongs to are likely to be more sa­
lient and accessible than they are for nonmembers of the culture, although
such cultural representations can be entertained by outsiders too. It is the
relative accessibility of cultural assumptions mentally represented that has
a direct bearing on this study, given that the particular way in which these
assumptions interact with the incoming ostensive stimulus (the meta­
phorical proverb, in this case) will make one interpretation more readily
available than other potential candidates that are also compatible with the
linguistic form of the proverb.

THE NATURE AND INTERPRETATION OF METAPHOR

Views about the nature of figurative language in general and of metaphor in

particular abound, and a number of interesting issues about the relationship
between figurative language and thought (see Gibbs, 1994; Katz, Cacciari,
Gibbs, & Turner, 1998), and how figurative language is processed have been
explored in the psycholinguistics and pragmatics literature by a great many
authors. Gibbs (1994, 2001a) offered recent and interesting reviews of con­
temporary models of figurative language understanding. Here I assume (as
does Sperber & Wilson, 1985–1986, 1986/21995) that understanding meta–
12. INTERPRETATION OF POPULAR SAYINGS 287

phor requires no special interpretive abilities or procedures, and that meta­

phor is merely a particular case of loose talk. When hearers encounter a
metaphorical utterance, they search for an interpretation consistent with the
principle of relevance in exactly the same manner as when interpreting any
other type of utterance. The ostensive use of a metaphoric expression, how­
ever, can be viewed as an instruction for the hearer to find relevant common
properties between the concepts associated with the topic and the vehicle
terms (Gibbs 2001a, p. 326; Sperber & Wilson, 1985–1986, 1986/21995).
Concepts encoded in the linguistic form of the utterance typically give
access to encyclopedic knowledge and stereotypical contexts associated
with them. When processed in such contexts, any utterance, whether meta­
phorical or not, will yield a range of explicatures and implicatures. In the
case of a metaphor, hearers find the intended resemblances among concepts
use association through metaphor in their search for optimal relevance, in­
hibiting the recovery of those resemblances whose derivation would detract
from relevance. The relevance of a metaphor is thus established by finding a
wide array of cognitive effects that can be retained as strong or weak
implicatures.2 In highly standardized metaphors, relevance is achieved
through the derivation of a few strongly implicated assumptions. More cre­
ative cases involve putting together the encyclopedic entries of encoded
concepts to yield a weaker, less determinate range of implicatures for which
the hearer takes more responsibility.
We have known for a long time that our similarity judgments are asym­
metrical in metaphor. There is good evidence to believe that an important
role of the vehicle term in a metaphor is to make more salient a number of
features of the topic term (Winner, 1988). But not all features of the vehicle
term are equally important in interpreting a metaphor. The subset of charac­
teristics that a metaphor directs our attention to is determined by salience,
which in turn is a property linked both to the context of interpretation and
to the intention of the communicator. Hence, when people understand met­
aphors, a great deal of irrelevant information is filtered out (Glucksberg,
Newsome, & Goldvarg, 2001).
The study of the interpretation processes underlying our understanding
of metaphor poses a crucial question for cognitive science concerning the
nature of concepts. Many researchers of figurative language claim that met­
aphorical language is a direct reflection of the structure of concepts, which

'Implicatures are said to be strong if they are fully determinate premises or conclusions
that must be supplied if an interpretation is to be consistent with the communicative princi­
ple of relevance and for whose derivation the speaker seems responsible. Weak implicatures
fall on the other extreme of a continuum, being assumptions that the hearer is encouraged
but not forced to supply, so that a great part of the responsibility for their derivation falls
largely on the hearer's side. The less the responsibility on the part of the speaker, the weaker
the implicature.
288 CURCÓ

are themselves thought of as metaphorical (Gibbs 1994, 2001a; Lakoff &

Johnson, 1989), and they argue that most of our thinking and cognition is
metaphorical too (Lakoff & Núñez, 2002). Other cognitive scientists have
strongly argued that concepts cannot be metaphorical (Fodor, 1998;
Murphy, 1996). This debate is beyond the scope and purpose of this chapter,
but it is certainly one of the fundamental questions underlying the scientific
study of metaphor.
THE NATURE AND INTERPRETATION OF PROVERBS

Proverbs are short utterance types—public representations—that people

quote with the purpose of addressing some aspect of about human nature, life,
behavior, or experience. Because proverbs are public representations widely
distributed in human populations, they are cultural representations. It is rec­
ognized that human cognitive abilities act as a filter on the representations
that are capable or likely to become cultural (Sperber, 1996, p. 70). As often
noted (e.g., Gibbs, 2001b), an important number of proverbs are metaphori­
cal. In particular, their interpretation depends on our ability for abstract
thought, given that proverbs establish a link between something specific and
concrete and a generic, more abstract aspect of human experience.
Perhaps the pervasiveness of proverbs in all human societies is due to our
capacity to recognize similarities between what proverbs mean and different
life situations. Besides, if Sperber and Wilson (1995) are right that the aim of
the human cognitive system is the maximization of relevance,3 an abun­
dance of proverbs in human societies is to be expected. Proverbs encode
concrete manifestations of more abstract themes to which they bear a re­
semblance relation. They can thus be relevantly used in wide ranges of situa­
tions, condensing as they do a huge potential to achieve cognitive effects at
a rather low processing cost: Proverbs usually have simple and short linguis­
tic and logical forms. This links their social importance to our most basic
cognitive dispositions. Socially, as Gibbs (2001b) put it, proverbs "function
as devices to unite speakers, speakers to their communities, and communi­
ties to ideas of universal truth in human experience" (p. 167). Their effi­
ciency as public representations and their impact as elements of internal
cohesion in a sociocultural group are thus consequences of their extraordi­
nary efficiency as cognitive devices.
Whether proverbs are interpreted through a single generic metaphoric
pattern of thought, "generic is specific," as some have argued (e.g., see
Gibbs, 2001b; Lakoff & Turner, 1989), or through a general comprehension
procedure employed in communication, as would follow from other views
(Carston, 2002; Sperber & Wilson, 1986, 1995) is an open issue. In the first
3
Contrast this with the aim of natural language interpretation in ostensive communica­
tion, which is merely the optimization of relevance.
12. INTERPRETATION OF POPULAR SAYINGS 289

case, a hearer would map knowledge from specific, familiar, and concrete
domains to very general events that are less familiar and abstract as part of a
complex conceptual—itself metaphorical—process. In the second case,
hearers would recover interpretations guided by their search for optimal rel­
evance in the interpretation of ostensive stimuli and their general knowl­
edge of the world. Here I take the second view. This framework allows for a
good understanding of cross-cultural differences in the interpretation of
proverbial language. For one thing, it clearly predicts that salient culture
specific assumptions can, in principle, intervene in the selection of one of
several possible interpretations when subjects are confronted to unfamiliar
proverbial language. Notice, besides, that the assumption that proverbs en­
code information about specific and concrete events that should be inter­
preted as referring to more general aspects of human life and experience in
particular situations is most likely to be part of the knowledge of the world
that an individual brings to his or her interpretation process. Hence, the role
of this contextual knowledge in interpreting can be acknowledged without
committing to the view that there is a single generic metaphoric pattern of
thought "generic is specific" responsible for proverb interpretation.
Metaphorical proverb interpretation is, along with metaphor, a con-
text-dependent process, sensitive to the beliefs held by the hearer and the
mutually manifest assumptions shared by participants.4 Interpretation of
unfamiliar proverbs presented out of context, or in a very limited one, is a
rather unnatural task. However, observing how subjects choose an interpre­
tation in such circumstance offers an advantage: Their choice will rest
heavily on the hearers' most firmly established beliefs and salient assump­
tions, among which cultural representations and values should play an out­
standing role. In natural contexts, cultural assumptions may be—and often
are—overridden by other assumptions that are more specific to the partici­
pants and to the time and place of the utterance.
Note that feature salience is the result of the interpretation process, not a
preestablished fact. Notions such as salience should result from instantiations
of the optimization of relevance, rather than having to be separately specified.
Therefore, they are attached to a particular individual, interpreting a particu­
lar utterance token in a particular context of interpretation, at a particular
time. However, in comparing preferred interpretations across individuals with
different cultural backgrounds and without a biasing context we can safely
4
Sperber & Wilson (1986, 1995) define the manifestness of an assumption to an individ­
ual as the degree to which an individual is capable of mentally representing an assumption
and holding it as true or probably true at a given moment. The set of assumptions that are
manifest to an individual at a given time constitute his cognitive environment. A mutual cog­
nitive environment is one which is shared by a group of individuals and in which it is manifest
to those individuals that they share it with each other. Every manifest assumption in a mutual
cognitive environment is mutually manifest.
29O CURCO

assume that in general culture-specific beliefs are relatively more salient for
members of that culture than for outsiders.
Conversely, if Lakoff and Turner (1989) and Gibbs (2001b) are right that
unfamiliar proverbs are interpreted by using the "generic is specific" concep­
tual metaphor, the way in which salient cultural properties play any special
role in the process seems less clear. If I understand them correctly, the idea
behind the "generic is specific" metaphor is that it provides a general cogni­
tive mechanism for understanding the general in terms of the specific, in
which the mapping of the source domain knowledge onto the target do­
mains of experience preserves the cognitive structure of the source domain.
This would help us understand why certain cultural representations and
values play important roles in the coming of new (or a series of) proverbs
linked with specific conceptual domains,5 but it does not give many reasons
to explain why interpretations other than the generally accepted are possi­
ble for fluent nonnative speakers and why some of these interpretations ap­
pear to be more accessible to certain groups.
Indeed, Gibbs (2001b) acknowledged the importance of cultural assump­
tions in interpreting proverbs. In discussing some proverbs mentioned in
20th-century African literature, he mentioned, "Each of these proverbs
poses riddles whose solution is found in the specific cultural contexts in
which they are used. Their rhetorical power comes from people's tacit recog­
nition of the tension between these sayings' concrete messages and their ap­
plication to abstract themes" (pp. 174-175). In this view, cultural factors
determine the interpreter's understanding of the concrete and specific do­
main, and they would interfere negatively with interpretation whenever this
culturally specified knowledge cannot be used in linking it to the generic
and abstract.
One important assumption underlying the study reported here is that not
only is the concrete and specific culturally shaped, but that abstract themes
can be more or less salient to an individual depending on cultural factors too.
In any case, Gibbs' (2001b) view that proverbs are public representa­
tions of schemas that characterize an open-ended category of situations
and that they function as a variable template that can be filled in different
ways is right. Cultural background is one of the many factors likely to affect
the actual manner in which the template is filled and a particular interpre­
tation is selected.
CULTURAL FACTORS AFFECTING INTERPRETATION

Cross-cultural psychology has succeeded in many respects as a new aca­

demic discipline, but it nonetheless faces some general methodological diffi-

For instance, it would help understand why proverbs involving food concepts are so basic
to all cultures.
12. INTERPRETATION OF POPULAR SAYINGS 291

culties. First, cross-cultural psychologists and anthropologists do not

completely agree on what precisely should count as a cultural unit of analy­
sis. Clearly, a culture-bearing unit of analysis cannot be defined independ­
ently of the level at which the independent variable studied operates, so that
appropriate units will vary from study to study. Second, assuming homoge­
neity of cultures and social groups with respect to the factor studied is usu­
ally misleading. Third, the spreading of cultural characteristics through
contacts between groups, known in anthropology as Galton's problem, is no­
ticeably on the rise as the global era takes over (see Berry, Poortinga, Segall,
& Dasen, 1992, for discussion). Hence, in cross-cultural research there is a
risk of drawing erroneous inferences based on empirically observed differ­
ences across groups, given the difficulties in effectively isolating a cultural
homogeneous population and defining a culture-bearing unit. Further com­
plications include the difficulty of ruling out all possible alternative explana­
tions to the differences found.
The study of the variation of selected interpretations to a given proverb
reported here is not free of these limitations. Even if one were to find some
form of special association between group membership and interpretation
selected, tracking such differences to some specific cultural factor is not
straightforward. Variation in the selection of an interpretation is likely to
arise out of differences in the ease with which a subject can integrate one of
the possible interpretations of a proverb with the content of certain individ­
ual mental representations, which could or could not be cultural. Concepts
and beliefs are representations that can be internalized individually and that
are cultural as a result of a great many individuals internalizing them
(Sperber, 1996, p. 75). Cultural classifications are set apart by their uniform
distributions: Closely similar versions of the same representation are spread
throughout a human population. The results report initial explorations. In
future studies, cultural assumptions with a potential to affect interpretation
will have to be studied independently to ascertain the extent to which they
have been replicated and distributed homogeneously across the populations
studied. Also, the probes employed in empirical studies should be chosen or
designed to match specific target assumptions isolated this way.
DESCRIPTION OF THE STUDIES
In the studies reported here, the preferred interpretations of proverbs given
by native speakers of Spanish (familiar with the proverbs) and those selected
by a group of foreign students learning Spanish as a second language (less fa­
miliar or totally unfamiliar with the same proverbs) were compared. The
studies were conceived as initial tools in an exploration of the question
whether there are any preferred interpretations and, if so, whether these in­
terpretations are associated with group membership, such as geographical or
religious origin, or gender.
292 CURCO

Presented out of context, the meaning of a proverb is radically underde­

termined, especially if the proverb is unfamiliar to a subject. For proverbs, as
well as for any other utterance, it is possible to recognize a number of poten­
tial interpretations compatible with their linguistic form, and the choices
given by subjects may be influenced by a very wide array of factors, among
which cultural background is but one. Future studies will develop more fine-
grained tools to explore it.
A set of proverbs was chosen and subjects were presented with two possi­
ble paraphrases of each: One that represents the generally accepted use in
the community where they are mostly used and another that is compatible
with its linguistic form, but that is not standard in the way the first one is.
The first objective was to find out whether Mexican native speakers of
Spanish shared a canonical conventional meaning associated with each
proverb. Participants were asked to choose one of two possible interpreta­
tions: There are no intended interpretations, because probes are not embed­
ded in an ostensive act of communication. The study merely looked for more
salient and preferred ones. In general, the context in which an utterance is
produced disambiguates and helps hearers recognize which of the various
possible meanings of an expression is intended. The quest here was precisely
about a psychological, individual, culturally determined subset of context
that should remain when the discourse and situational contexts are omitted.
The second objective was to explore the extent to which certain factors,
such as country of origin or religious background, were in some special asso­
ciation with the interpretations to the same proverbs selected by foreigners
learning Spanish as a second language in Mexico.
STUDY 1: STABILITY OF PROVERB MEANING
AMONG NATIVE SPEAKERS

Stability of Meaning Among Mexican Native Speakers

of Spanish
The aim of Study 1 was to determine the extent to which there is stability
among Mexicans in their interpretation of a proverb. Selection of the gener­
ally accepted figurative use paraphrase over the possible figurative use para­
phrase was expected as a tendency6 (see Appendix).
Method
Participants. Twenty-four Mexicans—15 men and 8 women—who
were native speakers of Spanish, from middle-class homes, and between the
ages of 25 and 30 (mode = 27) volunteered for participation.
6
Generally accepted interpretations were based on the intuitions of a few native speak­
ers independently questioned and checked against three handbooks of proverbs and popu­
lar sayings.
12. INTERPRETATION (3F POPULAR SAYINGS 293

Materials. The items consisted of 14 proverbs and a set of two possible

interpretations for each (see Appendix for a full list). All proverbs but one
are independently recognized as familiar by Mexicans.

Procedure. The 14 proverbs were included in a questionnaire. Their

order of presentation was randomized, as was the order of the possible inter­
pretations. For each proverb, participants were asked whether they knew it
and whether they understood the meaning of all the words. They were also
instructed to select the interpretation of the proverb they considered closest
to its meaning.

Results and Discussion

Frequencies for each answer to all proverbs given by the participants are dis­
played in Table 12.1. Chi-square tests were conducted for each probe to ex­
plore whether the null hypothesis (that Mexicans choose the figurative use
paraphrase) can be maintained.
Results show that Mexicans choose the conventional interpretation for 9
of the 14 proverbs, although stability of meaning was shown for a few more of
them. For instance, note the consistency of answers to Proverb 4, which do
not match the generally accepted paraphrase. None of the subjects chose
the expected answer for this probe. There is a straightforward explanation
for this outcome. The proverb is not of Mexican origin, but it comes from
Spain. In Mexico parrots are culturally associated with nonstop talking,
which explains the salience of the possible paraphrase over the generally ac­
cepted one in Spain.

Interpretation and Gender. The same set of data was analyzed to test
for a possible relation between gender and choice of answer for the group of
Mexicans. The frequencies of response are displayed in Table 12.2. Again,
chi-square tests were conducted to see if the null hypothesis that there is no
special association between gender and the selected interpretation could be
maintained. Results show that, except for Probe 1, there is no significant re­
lation between choice of interpretation and gender of speaker. Thus, al­
though there is some variation in the answers selected by native speakers,
this is not linked to their gender.7

7
An interesting result emerges from the distribution of answers in Probe 1. The two possi­
ble answers differ only in the presence of a syntactic negation in one of them. This issue is, of
course, beyond the scope of this chapter.
294 CURCO

TABLE 12.1
Frequency of Selection of the Canonical Interpretation
by Mexicans—Study l

x2 Value
Proverb Number (Critical Value of
(Ordered as in the Appendix) Frequency % x2 = 3.84, p < .05)
1 15 62.5% 3.3
2 19 79% 1.04
3 11 45.8% 7.04
4 0 0% 24
5 19 79% 1.04
6 18 75% 1.28
7 6 25% 11.57
8 10 41.6% 7
9 17 70.8% 1.75
10 16 66.6% 2.66
11 15 62.5% 3.3
12 12 50% 6
13 18 75% 1.28
14 19 79% 1.04

STUDY 2: STABILITY OF MEANING AMONG

NON-NATIVE SPEAKERS

Method
Participants. Thirty-two foreign students of Spanish attending the
National Autonomous University of Mexico—16 men, 16 women—partici-
pated in the study. They were paid $5. The participants' ages ranged from 18
to 34 (mean = 25.5, mode = 25). On average, they had studied Spanish for
16.75 months (exposure to language ranged from 2 to 6 years). Half of them
lived in Mexico for 2 months, and the other half lived in Mexico between 3
and 12 months. Nine of them were studying Spanish as their second lan­
guage, 10 as their third language, and 7 as their fourth language. For the rest,
Spanish was being studied either as fifth, sixth, or seventh language. Their
countries of origin included Austria, Brazil, Bosnia, Cameroon, Canada,
China, England, France, Holland, Indonesia, Japan, Poland, Slovenia,
South Korea, Tanna Vanuatu, and the United States. For the analysis of the
data, the subjects in the sample were grouped into four main geographical
12. INTERPRETATION OF POPULAR SAYINGS 295

TABLE 12.2
Frequency of Selection of the Canonical Interpretation
by Male and Female Mexicans—Study l

Proverb Number X2 Value

(Ordered as in the (Critical Value x2
Appendix) Interpretation Women Men X2 = 3.84, p < .05
1 GAP 0 15
PFP 8 1 19.99
2 GAP 7 12
PFP 1 4 .45
3 GAP 4 7
PFP 4 9 .29
4 GAP 0 0
PFP 8 15 0
5 GAP 5 12
PFP 2 3 .17
6 GAP 7 11
PFP 1 5 .99
1 GAP 2 4
PFP 6 12 0
8 GAP 3 7
PFP 5 8 .13
9 GAP 6 11
PFP 2 5 .34
10 GAP 4 12
PFP 4 4 1.2
11 GAP 6 8
PFP 1 6 1.6
12 GAP 4 8
PFP 4 7 .04
13 GAP 7 11
PFP 1 5 .99
14 GAP 7 12
PFP 0 4 1.68

Note. GAP = generally accepted paraphrase; PFP = possible paraphrase.

areas: Asia (9 subjects), Europe (10) subjects, America8 (12 subjects, Cana­
dians and Brazilians are included here), and Africa (1 subject). The religious
breakdown was as follows: 7 Catholics, 14 Protestants, and 11 non-Chris-
8
Throughout this chapter 'America' includes all countries in the continent and does not
refer to the United States.
296 CURCO

tians. Half of them had lived in countries other than Mexico, and Mexico
was the only foreign country where the other half had stayed for longer than
2 months. According to the language course in which they were enrolled,
15% of them had a basic command of Spanish, 65% were considered inter­
mediate, and 18% were classified as advanced students.
Materials. The materials used were the same as those in Study 1.
Procedure. The same 14 proverbs were included in a questionnaire.
Their order of presentation was randomized, as was the order of the two pos­
sible paraphrases offered as what was believed to be the most appropriate in­
terpretation. For each proverb, participants were asked whether they knew
it and whether they understood the meaning of all the words. They were also
instructed to select the interpretation of the proverb they considered closest
to its meaning. Whenever participants indicated that they did not under­
stand all the words in a proverb, their answer to that probe was not taken
into account.9 The objective was to find out (a) whether foreigners chose
the generally accepted paraphrase of a proverb, (b) whether there is an asso­
ciation between being Mexican or foreigner and the interpretation selected,
(c) whether there is an association between the geographical location of the
country of origin of subjects and their choice of paraphrase, and (d) whether
there is an association between the dominant religion of the country of ori­
gin of subjects and their choice of paraphrase (in most of the cases, the main
religion in the country of origin of a subject coincided with his or her own).
Results and Discussion
As in study 1, chi-square tests were conducted for each probe to explore
whether the null hypothesis (that all foreigners will choose the generally ac­
cepted paraphrase) should be rejected. Frequencies for each answer givenby
the group of foreigners to all proverbs are displayed in Table 12.3.
Interpretations of Nonnative speakers. The null hypothesis—that
all foreign subjects select the generally accepted paraphrase, as native
speakers do—was rejected for all probes but three (Proverbs 1, 2, and 5), in
some cases, p < .05, and in others, p < .01.
Association Between Being a Native Speaker or a Foreigner and
Interpretation Chosen. We then explored whether there was a special
relation between being Mexican or being a foreigner and the type of para­

9
A11 participants answereda second questionnaire, which presented them with a number
of conversational exchanges in written form (contexts) for each proverb. Subjects were asked
to choose the context in which a particular proverb would be most adequately used. The re­
sults of this study are not reported in this chapter.
12. INTERPRETATION OF POPULAR SAYINGS
297

TABLE 12.3
Frequency of Selection of the Canonical Interpretation
by Foreigners—Study 2

x2 Value
Proverb Number (Critical Value of
(Ordered as in the x2 = 3.84, p < .05 and
Appendix) Frequency % x2 = 6.63, p < .01)
1 24 75% 2
2 27 84% 0.78
3 11 34% 13.78
4 14 43% 10.12
5 23 71% 2.53
6 13 40% 11.28
1 10 31.2% 15.12
8 12 37% 12.5
9 11 34% 13.78
10 20 62.5 4.5
11 17 53.1% 7
12 12 37% 12.5
13 13 40% 11.28
14 17 53.1% 7

phrase chosen. For each probe (excluding Proverbs 1, 2, and 5), we com­
pared the answers of both groups and conducted a chi-square test. The
results, displayed in Table 12.4, lead us to conclude that there is a significant
relation between being a native speaker of Spanish or a foreigner and the in­
terpretations selected for Proverbs 4, 6, 9, 12, and 13. For the rest of the
probes, the null hypothesis is maintained.
Effect of Geographical Area of Origin on Interpretation Chosen.
To explore whether the type of answers selected were related to the geo­
graphical location of the country of origin of a subject, we partitioned our
sample into three major geographical areas: America, Europe, and Asia (the
only African subject in the sample was excluded from this section of the
study).10 To test the null hypothesis that geographical origin is not related to
10
Murdock (1967, cited in Berry et al., 1992) distinguished six cultural areas on the basis
of geography, namely Sub-Saharan Africa, Asia, Australasia, Circurn-Mediterranean, North
America, and South America. Our categories are a simplification of his system, given the size
and composition of our sample. Note, however, that the range of variation within certain cul­
tural areas can be as large as the variation between these areas for a number of dimensions, so
we need to explore the crucial dimensions intervening in interpretation in more detail.
 TABLE 12.4
Frequency of Selection of the Canonical Interpretation
and the Possible Paraphrase by Foreign Speakers of Spanish
and by Mexican Native Speakers—Study 2

x2 Value
Proverb Number Foreign Mexican Native (Critical Value
(Ordered as in Speakers of Speakers of of x2 = 3.84,
the Appendix) Interpretation Spanish Spanish p < .05)
1 GAP Not Not Not
PFP considered considered calculated
2 GAP Not Not Not
PFP considered considered calculated
3 GAP 11 11
PFP 17 13 .26
4 GAP 14 0
PFP 17 23 16.2
5 GAP Not Not Not
PFP considered considered calculated
6 GAP 13 18
PFP 16 6 5.8
7 GAP 10 6
PFP 20 18 .24
8 GAP 12 10
PFP 15 13 .004
9 GAP 11 17
PFP 17 7 4.72
10 GAP 2 16
PFP 9 8 .15
11 GAP 17 14
PFP 12 7 .28
12 GAP 6 12
PFP 21 11 4.83
13 GAP 13 18
PFP 18 6 5.97
14 GAP 17 19
PFP 11 4 2.9

Note. GAP = generally accepted paraphrase; PFP = possible paraphrase.

298
12. INTERPRETATION OF POPULAR SAYINGS 299

choice of paraphrase, we conducted chi-square tests for each of the probes

(2 possible answers X 3 geographical areas, df = 2). The results are displayed
in Table 12.5

Effect of Dominant Religion on Interpretation Chosen

To explore whether the type of answers selected were related to the domi­
nant religion of the country of origin of a subject, we partitioned our sample
into three major religious categories: Catholic, Protestant, and non-Chris-
tian. To test the null hypothesis that the dominant religion is not related to
choice of paraphrase, we conducted chi-square tests for each of the probes
(2 possible answers X 3 major religious backgrounds, df = 2). The results are
displayed in Table 12.6.

GENERAL DISCUSSION AND CONCLUSION

For the probes used in this exploration, it was found that there is a very sig­
nificant amount of stability of interpretation for native speakers. For the four
proverbs in which more variation in interpretation was found (Proverbs 3, 7,
8, and 12; see Appendix), it was noticed that one of the alternative re­
sponses can be seen as a particular case of the other, a fact that may have led
some of the subjects to select the most general one. Further studies should
take this into account and test different interpretations derived from more
distant conceptual domains, in which the connection to cultural themes is
ideally more precise. Variation in responses for native speakers was not sig­
nificant and not linked to gender factors.
Nonnative speakers showed a pattern of responses that contrasts signifi­
cantly with native speakers. The null hypothesis that nonnative speakers se­
lect the same interpretation as native speakers was rejected in most cases.
For 78.6% of the proverbs in Study 2, participants did not select the gener­
ally accepted paraphrase, and this was statistically significant in all cases but
one, in which only a tendency was found.11
A special association between being native or nonnative speaker of Span­
ish and the type of answer selected for several items was found. For Proverb
4, Mexicans overwhelmingly preferred an interpretation based on the con­
cept of talking, whereas foreigners were divided between this and an inter­
pretation based on the concept of eating. For Proverb 6, native speakers
were inclined to choose an interpretation based on the conceptual con­
struction of "nonintrusion in other people's affairs," whereas nonnative

11
Despite the limitations of the conclusions that can be drawn, this finding deserves con­
sideration in second language testing practices, wherein the use of proverbs is sometimes
found in proficiency exams. (Ormsby, personal communication, December, 2, 2001)
 TABLE 12.5
Frequency of Selection of the Canonical Interpretation
and the Possible Paraphrase by Foreign Speakers of Spanish
Coming from the American continent, Europe, and Asia—Study 2

Foreign
Proverb Subjects Foreign
Number Coming Subjects Foreign x2 Value
(Ordered as From the Coming Subjects (Critical Value
in the American From Coming of x2 = 5.99,
Appendix) Interpretation Continent Europe From Asia p < .05, df = 2
1 GAP Not Not Not Not
PFP considered considered considered calculated
2 GAP Not Not Not Not
PFP considered considered considered calculated
3 GAP 4 8 3
PFP 8 2 6 6.14
4 GAP 8 7 3
PFP 4 3 6 3.70
5 GAP 8 5 9
PFP 4 3 0 4.35
6 GAP 4 4 4
PFP 8 3 5 1.14
7 GAP 2 4 5
PFP 10 5 4 3.64
8 GAP 7 0 5
PFP 5 8 4 7.20
9 GAP 5 3 2
PFP 7 3 7 4.12
10 GAP 6 6 8
PFP 6 2 1 3.80
11 GAP 9 3 5
PFP 3 5 4 3.26
12 GAP 1 1 4
PFP 11 5 5 5.07
13 GAP 5 5 1
PFP 7 3 8 5.33
14 GAP 6 8 4
PFP 6 2 4 2.74

Note. GAP = generally accepted paraphrase; PFP = possible paraphrase.

3OO
 TABLE 12.6
Frequency of Selection of the Canonical Interpretation and the
Possible Paraphrase by Foreign Speakers of Spanish With Catholic,
Protestant, and Non-Christian Religious Background—Study 2

Proverb Foreign Foreign Foreign x2 Value

Number Subjects Subjects Subjects With (Critical Value
(Ordered as With a With a a Non- of x2 = 5.99,
in the Catholic Protestant Christian p < .05,
Appendix) Interpretation Background Background Background df = 2)
1 GAP Not Not Not Not
PFP considered considered considered calculated
2 GAP Not Not Not Not
PFP considered considered considered calculated
3 GAP 4 6 3
PFP 3 8 8 1.06
4 GAP 3 7 4
PFP 4 7 6 0.22
5 GAP 5 7 10
PFP 2 5 0 5.41
6 GAP 2 6 5
PFP 5 6 5 0.95
7 GAP 3 2 5
PFP 4 12 5 3.4
8 GAP 1 7 6
PFP 6 6 4 3.24
9 GAP 2 4 3
PFP 4 8 7 0.031
10 GAP 1 7 6
PFP 6 6 4 6.22
11 GAP 4 8 6
PFP 2 6 4 0.18
12 GAP 1 1 4
PFP 5 11 5 5.07
13 GAP 5 4 3
PFP 1 9 8 5.86
14 GAP 7 5 7
PFP 0 8 3 7.68

Note. GAP = generally accepted paraphrase; PFP = possible paraphrase.

301
3O2 CURCO

speakers were split between this and an interpretation based on the concep­
tual scheme of "life style matching what one can afford." For Proverb 9, non­
native speakers selected interpretations based on the concepts of necessity
and surprise, but native speakers preferred the interpretation based on the
concept of surprise. Proverb 12 elicited interpretations based both on the
concept of ugliness and the concept of humbleness in native speakers; but in
nonnative speakers, most choices were linked to humbleness. For Proverb
13, native speakers' interpretations were related to the concepts of kindness
and usefulness, whereas nonnative speakers chose interpretations linked to
concepts of clarity and understanding. The variations in responses were sig­
nificant, as shown in the tables, along with other strong—though not statis­
tically significant—tendencies.
Some associations between the type of answer selected and the geograph­
ical origin of the subjects were also found. For Proverb 3, Europeans had a
significant stronger tendency to select an interpretation based on the con­
cept of price, whereas Americans and Asians preferred an interpretation
based on the concept of solution. Proverb 4 elicited answers based on the
concept of eating in most Europeans, but Americans and Asians tended to
select an interpretation based on the concept of talking. For Proverb 5,
speakers of all geographical areas showed a tendency for an interpretation
based on the concept of good times, but note that no Asian subject chose an
interpretation based on bad times, whereas almost half of the Americans
and the Europeans did. In Proverb 7, most Americans preferred an interpre­
tation linked to the notion of boasting, whereas other groups were divided
among the interpretation options. The concept of efficiency underlied the
choice of interpretation of Europeans in Proverb 8, but Americans and
Asians were divided between this interpretation and one based on the con­
cept of bravery. Results also showed that Asians tended to choose an inter­
pretation of Proverb 9 based on the concept of necessity, whereas
Americans and Europeans did not have a preferred interpretation. For Prov­
erb 10, Asians preferred an interpretation based on the notions of fitness
and ability, whereas Americans and Europeans were divided among the al­
ternative choices. Americans showed a strong tendency to select an inter­
pretation based on the notion of spending for Proverb 11, Europeans
preferred one based on the notion of understanding, and Asians were split
between the two possibilities. Finally, for Proverb 13, Asians chose an inter­
pretation based on notions of clarity and understanding, whereas Europeans
preferred one based on helpfulness, and Americans were divided between
the two.
Religious background also was significantly associated to the answers in
several cases. For instance, all subjects from mainly Catholic areas selected
the no action option in Proverb 1, all Christians selected the interpretation
based on the notion of good times in Proverb 5, Protestants preferred an inter­
12. INTERPRETATION OF POPULAR SAYINGS 3O3

pretation linked to the concept of boasting over one based on the concept of
humbleness in Proverb 7. People from Catholic regions chose an interpreta­
tion based on the notion of bravery for Proverb 8, whereas Protestants and
non-Christians selected one based on the concept of efficiency. Protestants
were significantly inclined to choose the interpretation constructed on the
notion of concern for Proverb 10, but Catholics and non-Christians preferred
one based on concepts of fitness and ability. For instance, and against the as­
sumptions mentioned in the introduction, all Catholics associated the "pick­
ing out for mosquitoes and swallowing of camels" with an obsession for details
and with an overlooking of serious faults, an interpretation also favored by
non-Christians, whereas Protestants chose an interpretation based on the no­
tions of difficulty and absurdity.
No definite associations between these answers and cultural specificity
can be made before more detailed studies consider the cultural representa­
tions of the populations studied as such and finer instruments are designed
to attest to the salience and impact of these cultural assumptions in the in­
terpretation process, but the results presented suggested some lines along
which future research can be designed.

REFERENCES

Berry, J., Poortinga, Y., Segall, M, & Dasen, P. (1992). Cross-culturalpsychology: Re­
search and Applications. Cambridge, England: Cambridge University Press.
Carston, R. (2002). Thoughts and utterances: The pragmatics of explicit communica­
tion. Oxford, England: Blackwell.
Chapman, S., Ulatowska, H., Franklin, L., Shobe, A., Thompson, J., & Mc Intire, D.
(1997). Proverb interpretation in fluent aphasia and Alzheimer disease: Implica­
tions beyond abstract thinking. Aphasiology, 11 (4-5), 337-350.
Cheng, L. (1996). Beyond bilingualism: A quest for communicative competence.
Topics in Language Disorders, 16(4), 9–21.
Davis, S. (Ed.). (1991). Pragmatics: A reader. Oxford: Oxford University Press.
Fodor, J. (1998). Concepts: Where cognitive science went wrong. Oxford, England:
Clarendon.
Gibbs, R. W., Jr. (1994). The poetics of mind. Cambridge, England: Cambridge Uni­
versity Press.
Gibbs, R. W., Jr. (2001a). Evaluating contemporary models of figurative language
understanding. Metaphor and Symbol, 12, 295-316.
Gibbs., R. W., Jr. (2001b). Proverbial themes we live by. Poetics, 29, 167-188.
Glucksberg, S., Newsome, M., & Goldvarg, Y. (2001). Inhibition of the literal: Fil­
tering metaphor-irrelevant information during metaphor comprehension. Meta­
phor and Symbol, 12, 277-293.
Grice, H. P. (1989). Studies in the way of words. Cambridge, MA: Harvard University
Press. (Original work published 1975)
Katz, A., Cacciari, C. Gibbs, R. W., Jr., & Turner, M. (1998). Figurative language and
thought. Oxford, England: Oxford University Press.
3O4 CURCO

Lakoff, G., & Nunez, R. (2002). Where mathematics come from. New York: Basic
Books.
Lakoff, G., & Turner, M. (1989). More than cool reason. Chicago: Chicago University
Press.
Levinson, S. (2000). Presumptive meanings. The theory of generalized conversational
implicature. Cambridge, MA: MIT Press.
Murphy, G. (1996). On metaphorical representation. Cognition, 60, 173–204.
Nippold, M., Uhden, L., & Schwarz, I. (1997). Proverb explanation through the
lifespan: A developmental study of adolescents and adults. Journal of Speech,
Language and Hearing Research, 40(2), 245–253.
Sperber, D. (1996). Explaining culture. A naturalistic approach. Oxford, England:
Blackwell.
Sperber, D., & Wilson, D. (1985/1986). Loose talk. In Proceedings of the Aristotelian
society (pp. 153-172). Oxford, England: Blackwell (reprinted, Davis, 1991).
Sperber, D., & Wilson, D. (1986). Relevance: Communication and cognition. Oxford,
England: Blackwell.
Sperber, D., & Wilson, D. (1995). Relevance: Communication and cognition (2nd ed.).
Oxford, England: Blackwell.
Stanley, J., 6k Szabo, Z. (2000). On quantifier domain restriction. Mind and Lan­
guage, 15, 219–261.
Stanley, J. (2002). Making it articulated. Mind and Language, 17(1–2), 149-168.
Ulatowska, H., Chapman, S., Highley, A., & Prince, J. (1998). Discourse in healthy
old-elderly adults: A longitudinal study. Aphasiology, 12(7–8), 619–633.
Winner, E. (1988). The point of words: Children's understanding of metaphor and irony.
Cambridge, MA: Harvard University Press.
 Appendix

PROVERBS AND THEIR FIGURATIVE USE PARAPHRASES

1. Ver el temblor y no hincarse (proverb in Spanish)

• Seeing the earthquake and not kneeling down (not kneeling down
at the sight of an earthquake) (approximate literal translation)
• To see danger approaching and do nothing to prevent it (generally
accepted figurative use paraphrase)
• To see danger approaching and do something to prevent it (possible
figurative use paraphrase)

2. Dar margaritas a los cerdos (proverb in Spanish)

• To give pigs daisies (approximate literal translation)
• To give something valuable to someone who cannot appreciate it
(generally accepted figurative use paraphrase)
• To encourage someone who has no real hope to succeed (possible
figurative use paraphrase)

3. Depende del sapo la pedrada (proverb in Spanish)

• The type of stone you throw depends on the toad you aim it at (ap­
proximate literal translation)
• The charge for a good or service depends on the customer (gener­
ally accepted figurative use paraphrase)
• The type of solution you give to a problem depends on how serious it
is (possible figurative use paraphrase)

3O5
3O6 APPENDIX

4. Perico, corto de manos, largo de pico (proverb in Spanish)

• Parrot, short hands, long beak (approximate literal translation)
• Said of someone who works little and eats much (generally ac­
cepted figurative use paraphrase)
• Said of someone who does little and talks much (possible figurative
use paraphrase)

5. A cada santo le llega sufuncion (proverb in Spanish)

• Each saint gets its performance day (approximate literal translation)
• Everyone gets good times (generally accepted figurative use para­
phrase)
• Everyone gets bad times (possible figurative use paraphrase)

6. A tu palo, gavildn y a tu matorral, conejo (proverb in Spanish)

• To your tree, hawk and to your bush, rabbit (approximate literal
translation)
• People must mind their own business (generally accepted figurative
use paraphrase)
• People's lifestyle must be compatible with what they can afford
(possible figurative use paraphrase)

7. Perro que ladra no muerde (proverb in Spanish)

• A dog that barks does not bite (approximate literal translation)
• People who have fits of temper are usually good natured and noble
(generally accepted figurative use paraphrase)
• People who boast are usually lacking (possible figurative use para­
phrase)

8. A ver de que cuero salen mas correas (proverb in Spanish)

• Let's see which leather yields more belts (approximate literal
translation)
• Let's see who is bravest (generally accepted figurative use para­
phrase)
• Let's see who is most efficient (possible figurative use paraphrase)

9. Hasta el mas chimuelo come davos (proverb in Spanish)

• Even the toothless eat nails (approximate literal Spanish)
• You can get a surprise from the people you less expect it (generally
accepted figurative use paraphrase)
• When in extreme need, people cannot afford to be choosy and de­
manding (possible figurative use paraphrase)
APPENDIX 3O7

10. No hay que meterse en la danza si no se tiene sonaja (proverb in Spanish)

• Don't get in the dance if you don't have a rattle (approximate literal
translation)
• Don't get involved in something for which you are unfit (generally
accepted figurative use paraphrase)
• Don't get involved in other people's affairs (possible figurative use
paraphrase)

11. Comenfrijoles y repiten polio (proverb in Spanish)

• They eat beans and belch out chicken (approximate literal transla­
tion)
• They have a low income and spend like rich (generally accepted fig­
urative use paraphrase)
• They're told one thing and they understand another (possible figu­
rative use paraphrase)

12. A cada capillita le llega su fiestecita (proverb in Spanish)

• Every little chapel gets is little celebration (approximate literal
translation)
• Even if a person is ugly, someone will eventually fall in love with her
(generally accepted figurative use paraphrase)
• Even those with a humble background get a moment of glory (possi­
ble figurative use paraphrase)

13. Candil de la calk, oscuridad de su casa (proverb in Spanish)

• Street lamp, darkness at home (approximate literal translation)
• Said of someone who is helpful and kind to others, but not to his
own family (generally accepted figurative use paraphrase)
• Said of someone who is very clear about the problems of other peo­
ple, but very confused about their own (possible figurative use para­
phrase

14. Estamos colando mosquitos y tragando camellos (proverb in Spanish,

translated from Swedish)
• We are sieving for mosquitoes and swallowing camels (approximate
literal translation)
• We are overconcerned about details and overlooking serious faults
(generally accepted figurative use paraphrase)
• We are doing something very difficult and absurd (possible figura­
tive use paraphrase)
This page intentionally left blank
 13

Metaphors in Sign Language

and Sign Language Users:

A Window Into Relations

of Language and Thought

Marc Marschark
National Technical Institute for the Deaf,
Rochester Institute of Technology, and University of Aberdeen

As its title suggests, this chapter is about the metaphoric aspects of sign lan­
guage and the use of such figurative devices by sign language users.1 Distin­
guishing and then integrating these domains requires consideration of three
primary areas of investigation, each of which is described here to only a lim­
ited extent—albeit for different reasons. After a brief history of research into
linguistic and psychological study involving metaphor, the chapter focuses
on metaphorical qualities that appear to be inherent in signed languages to
some greater degree than in spoken languages. Although it could be argued
that, in terms of the totality of artistic impression, a spoken language can be
as metaphorical as a signed language, the discussion here primarily considers
everyday language, a domain in which spoken language often seems to be a
dull cousin. Of particular interest here is the issue of metaphor versus
iconicity. These two aspects of sign language frequently have been seen in­
correctly as a single dimension, likely because iconicity does not occur in
spoken languages (its closest parallel is onomatopoeia).

Except where noted otherwise, metaphor is used metonymically to refer to figurative

expressions in general.

3O9
310 MARSCHARK

The second area of investigation concerns the ways in which various as­
pects of sign language are used to express ideas in figurative ways. This dis­
cussion focuses on common metaphors in sign language. The author's
firsthand aspects of this discussion are largely limited to American Sign Lan­
guage (ASL), although several secondhand examples are included. Beyond
these language-specific examples, however, it appears that all sign lan­
guages, from Croatian Sign Language (Hrvatski Znakovni Jezik) to Italian
Sign Language (La lingua italiana dei segni or LIS) and from Australian Sign
Language (Auslan) to Austrian Sign Language (Österreichischen
Gebädrdensprache), have similar means of expressing metaphor, even if their
differing linguistic rules make for normal cross-linguistic variation.
The third area considered, and the one most obviously psychological
rather than linguistic, concerns processes involved in the comprehension
and production of figurative language by individuals who use sign language.
Regrettably, this area is perhaps the most wanting of the three examined
here. Most notable are the lack of research on (a) the comprehension of
metaphor in sign language, (b) the use of figurative language by deaf individ­
uals who use spoken language or by hearing individuals who use sign lan­
guage, and (c) the development of metaphor in children acquiring sign
language as a first language. Although it might be assumed that the lack of
such work suggests that there is nothing to investigate (i.e., that a metaphor
by any other name smells as sweet), several considerations suggest this is not
the case. Rather, a combination of the immaturity of this research area, the
relative paucity of psycholinguists (as opposed to linguists and cognitive psy­
chologists) interested in sign language, and some sociopolitical constraints
within the Deaf 2 community have delayed what is likely to be a fascinating
and informative domain of enquiry, especially as it relates to children.
Meanwhile, the study of figurative and artistic use of sign language by adults
continues apace in both theater and linguistics.

BIRTHING THE STUDY OF METAPHORS IN SIGN

LANGUAGE AND SIGN LANGUAGE USERS

As many of the chapters in this book demonstrate, figurative language has

long been recognized as both a reflection of creativity and a window into the
language and thought of the user. Paivio (1979) once referred to this meta­
phorically by saying "For the student of language and thought, metaphor is a
solar eclipse" (p. 150). Mac Cormac (1985, p. 2) noted that metaphors can
operate as cognitive processes that offer new insights, whereas Wilcox

2
Note that literature in this field uses deaf to refer to audiological matters and Deaf to refer
to cultural matters. The metaphoric character of lower case d and uppercase D is not lost on
many familiar with the Deaf community (see Marschark, 1993).
13. METAPHORS IN SIGN LANGUAGE 311

(2000, p. 35) suggested that we use metaphors to make sense of what goes on
around us. If metaphor is such a valuable tool, how is it learned or does it de­
velop spontaneously?
These questions were addressed in an early study by Piaget (1923/1974)
which examined the ability of 8- to 11-year-olds to comprehend and explain
the meanings of proverbs. Piaget asked children either to describe how each
proverb was related to its literal meaning or simply to explain isolated prov­
erbs. His results suggested that younger children create "holistic schemas"
or "general images" of the literal contents of a proverb, based on their knowl­
edge of the topic (subject) and vehicle (object) rather than the underlying
figurative meaning (p. 151). Although the children in his study "all had the
feeling that each proverb had a symbolic sense" (p. 161), Piaget found that it
was not until the children started to apply analogical reasoning to the parts
of the proverbs and move away from their global, concrete images that they
could correctly explain figurative meanings.
Since Piaget's (1923/1974) study, comprehension of figurative language
has been seen as an indicator of cognitive flexibility in children, both deaf
and hearing. For theoretical reasons, Piaget did not expect hearing children
younger than 11 years (i.e., not having achieved formal operations) to be able
to understand the figurative meanings of proverbs, a correct prediction on
his part. Yet, research demonstrating younger deaf children's failures to un­
derstand nonliteral language frequently led to the conclusion that they are
both concrete and literal in language and thought (Blackwell, Engen,
Fischgrund, & Zarcadoolas, 1978; Boatner & Gates, 1969; Conley, 1976;
Myklebust, 1964; but also see Marschark & Nall, 1985). As discussed in this
chapter, these latter studies typically entailed stimulus materials drawn from
the vernacular and presented in print (e.g., "The students studied the poem
until they knew it by heart"). The fact that deaf children typically lag behind
hearing peers in literacy skills (Traxler, 2000) was ignored or taken as con­
vergent support for the conclusion.
The development of metaphoric understanding in deaf children is con­
sidered at length in a later section. At this point, however, it is important to
recognize that, at least in part, the negative view of their skills from early
work in the area followed naturally from the fact sign languages such as ASL
and British Sign Language (BSL) were not recognized as true, natural lan­
guages until the early 1960s (Stokoe, 1960; Stokoe, Casterline, &
Croneberg, 1965) and did not become an object of study in psychology until
the 1970s. By the late mid- to late 1970s, linguists were beginning to exam­
ine the structure and rules of signed languages (e.g., see Friedman, 1977, and
Siple, 1978), and the first forays into the psycholinguistics of sign language
use emerged at about the same time (e.g., Klima & Bellugi, 1979). More re­
cently, the combination of evidence from cognitive psychology,
psycholinguistics, cultural anthropology, and neuropsychology has offered
312 MARSCHARK

new insights into sign language and sign language users, and further studies
of sign language have spawned new conceptions about language (e.g.,
Brennan, 1990; Liddell, 1990; Wilcox, 2000).
As I clarify in the remainder of this chapter, recent studies of signed lan­
guages have been augmented by extensive and sophisticated studies of
iconicity and metaphor. Meanwhile, creativity expressed in the theatrical
and poetic sign productions of deaf native signers has long been taken as a
reflection of the flexibility of sign language rather than the language user, a
position that appears to be implicit in recent analyses of sign language poetry
(e.g., Taub, 2OOla; Wilcox, 2000, chap. 7). Thus, despite discussions about
the unique figurative potential of sign language, there has been little consid­
eration of potential interactions between sign language and the cognitive or
other psychological characteristics of sign language users.
To a considerable degree, this situation appears to have developed be­
cause of the absence of discussion about the possibility that individuals who
grow up using a signed language might have psycholinguistic or metaling­
uistic skills different than other hearing or deaf individuals (but see
Emmorey, 2002; Marschark, 1993; Marschark, Lang, & Albertini, 2002).
The history of research involving deaf individuals has been haunted by a sci­
entific culture that equated speech with language and language with intelli­
gence. Lacking a full understanding of sign language, disagreement and
discord both among deaf people (e.g., those using spoken vs. sign language)
and between deaf and hearing individuals have created significant barriers
to investigations in several potentially interesting domains (Marschark &
Spencer, 2003). The related view of deaf people as deficient or inferior also
led to an understandable resentment among many deaf people toward those
hearing people who sought to study them.
The lack of discourse concerning interactions between language and
cognition among deaf adults and children is regrettable for both theo­
retical and pragmatic reasons. With regard to recent reconsideration of
the Sapir-Whorf notions of linguistic relativity, it might be suspected
that the modality difference between signed and spoken language
might increase the apparent effects of linguistic relativity, just as
greater differences between any two spoken languages might produce
larger collateral effects than when those languages are more similar. In
the case of signed and spoken languages, however, the two co-occur in
the same culture (broadly speaking). This similarity of context suggests
the possibility that whereas reciprocal effects of language and culture
might both contribute to observed differences between two aural—oral
cultures, any such effects will be attenuated for a deaf subculture im­
mersed within a hearing culture. Deaf culture thus appears to provide a
paradigm case for examining the effects of cultural and linguistic rela­
13. METAPHORS IN SIGN LANGUAGE 313

tivity, but such an exercise still has not yet been undertaken. Particu­
larly with regard to metaphor, the kinds of figurative constructions
produced, the range of their most common meanings, and their interac­
tions with the educational and cultural backgrounds of deaf individuals
would be particularly enlightening.
Consider, for example, Glucksberg's (2001) discussion of ASL as one
of several classifier languages that "do not normally have names for
superordinate categories" (p. 39). If true, what does this say about the
cognitive processes of ASL users? Glucksberg cited the example of the
category furniture which, he suggested, is signed as CHAIR-TA-
BLE-BED, ETC., that is, as a list of category exemplars rather than with a
single, lexicalized category name. 3 In fact, there are some categories,
such asfurniture, that used to lack signed category names, and some oth­
ers, such as tools, that still do. The very existence of such situations sug­
gests that native users of ASL might have somewhat different
conceptions of those categories than others and that the organization of
their knowledge might well affect problem solving, reading, and memory.
Glucksberg noted that the relations of categories are intimately tied to
the abstraction of essential properties of objects, and one presumes that a
collective lexical or conceptual term for a set of objects might well be
more conducive to seeing that set as an abstraction than would a listing
of concrete exemplars (cf. Luria, 1976, chap. 5; Piaget, 1923/1974). Al­
though investigations of that possibility are in their early stages, some re­
lated findings are now emerging (e.g., Marschark & Everhart, 1999;
Marschark, Convertino, McEvoy, & Masteller, 2001), indicating rela­
tively small but consistently significant differences between deaf and
hearing individuals in relations among concepts and the spontaneous
use of associative connections in a variety of tasks.
Over the last 20 years or so, the use of multiple exemplar category names
in ASL appears to have become far less frequent, and now it is relatively
rare.4 In some cases, such as furniture, new category name signs have
emerged. (For example, FURNITURE entails an inclined F-hand shape (see
Fig. 13.1) moved side-to-side in a short arc, whereas PREACH entails the
same hand shape moved forward and backward.) Other category names,
such as tools, are finger spelled, whereas many others have long had common
sign names. Still, the potential cognitive impact of such linguistic and mor­
phological differences between signed and spoken languages or among
signed and among spoken languages has yet to be determined. To fully un­

3
English glosses of signs typically are written in capital letters.
The apparent rapidity of such changes suggests that sign languages might be evolving
faster than spoken languages, but the topic has not been investigated formally.
314 MARSCHARK

FIG. 13.1 Selected ASL hand shapes (F, 1, V or 2, bent V).

derstand the interactions among cognition, metaphor, and sign language, it

is clear that a more complete understanding of metaphors inherent in sign
language and metaphors expressed through sign language is needed. An ex­
cellent starting place for this discussion is the distinction between iconicity,
signs appearing to look like their referents, and metaphor, the symbolic lik­
ening of one thing to another.
ICONIC AND METAPHORIC ASPECTS OF SIGN LANGUAGE

For many years, investigators of ASL (and many nonnative users of ASL)
downplayed the role of iconicity in the language in an attempt to focus at­
tention on its linguistic status rather than its gestural qualities.5 But times
are changing. Taub (2001Ib, p. 1) described ASL as involving "an integration
of visual imagery with linguistic structure on a scale that no spoken language
can equal." She and Wilcox (2000) provided in-depth analyses of the iconic
qualities of ASL poetry, and several linguists have extensively investigated
the iconic and metaphoric properties of ASL and other sign languages, even
if they have not always distinguished among them (e.g., Boyes Braem, 1981;
Brennan, 1990; Marschark & West, 1985).
Apparently the first treatment of metaphor and iconicity in sign language
was provided by Boyes Braem (1981). In her pioneering and influential ex­

5
Gesture in this sense is not intended in the verbal sense described by McNeill (1992), but
precisely the opposite: the appearance to many laypersons, including some hearing parents of
deaf children, that sign language is not really language at all (cf Stokoe, 1960).
13. METAPHORS IN SIGN LANGUAGE 315

amination of ASL handshapes, Boyes Braem identified a level of symbolic

representation that she referred to as metaphorical. Although many of the
uses of devices she identified are now referred to as iconic rather than meta­
phoric (she described visual metaphors of various handshapes), her work set
the stage for later analyses of metaphor and iconicity across a variety of
signed languages. Brennan (1990) similarly blurred metaphor and iconicity
in her morphological analysis of BSL, whereas Marschark and West (1985)
confounded iconic gestures with iconic and metaphoric characteristics of
ASL. The latter study also may have underestimated what was seen at the
time as an amazing degree of figurative language competence by deaf chil­
dren (discussed later). Beyond their specific findings, however, these earlier
investigations pointed the way toward the need to distinguish linguistic and
psychological aspects of metaphor and iconicity in signed languages. To
date, only the linguistic distinction has been considered in any detail.

Iconicity of Classifiers
Signs are essentially composed of handshapes, locations, and movements
(see the hand shapes of Fig. 13.1, and all three components in Figs. 13.2 and
13.3). Distinguishing the iconic properties of signs and signing from their
metaphoric properties requires consideration of the use of these compo­
nents, within the sign space, to map linguistic form. Inherent in such map­
ping is the way that signers use classifiers as a kind of frozen trope (i.e.,
figurative devices that have entered into common usage) to describe space
and time. Classifiers are particular handshapes that can be used to represent
people, animals, vehicles, or objects. The referent of a classifier is first estab­
lished in the discourse via signs, and the classifier handshape then assumes
that role much in the manner of a pronoun, albeit a dynamic one. In ASL,
for example, a bent V-hand (see Fig. 13.1) is used represent an animal; or an
upside down bent V-hand could indicate a dead animal. (Metaphoric exten­
sion of such signs is considered in the next section.)
Emmorey (2001, pp. 148-151) described the three primary categories of
mapping between signs and space: (a) articulators map physical and concep­
tual elements, (b) locations in signing space map to locations in physical or
conceptual, and (c) space movements of the articulators map to the motion
of elements. Handshapes may be fixed in space, indicating a relative loca­
tion, a relative time, or a qualitative aspect. Alternatively, movement of the
handshape can indicate changes in relative location, time, or quality (e.g.,
IMPROVE, reflecting "good is up," and DETERIORATE, reflecting "bad i
down"). In these cases, the signs entail both iconic and metaphoric charac­
teristics, a double mapping (considered later).
Taub (2001b, p. 34) described classifiers as forming "a set of iconic build­
ing blocks for the description of physical objects, movements, and loca­
316 MARSCHARK

FIG. 13.2 Two ASL signs for PROMOTION.

FIG. 13.3 ASL classifier used to show a bicycle (or other vehicle) riding over a hill.

tions." Beyond iconicity in the shapes of signs is the iconic quality referred to
as virtual depiction (Mandel, 1977) or path-for-sliape iconicity (Taub, 2001Ib).
Rather than the hands or other parts of the body offering the iconic ana­
logue for some part of an object or event, virtual depiction entails the move­
ment of the articulators providing the representational quality, as in the two
alternative signs for PROMOTION in ASL shown in Fig. 13.2. When hand­
13. METAPHORS IN SIGN LANGUAGE 317

shapes take on iconic similarity to concrete referents (e.g., the PERSON or

VEHICLE classifier), they also can represent the movement of those refer­
ents in space—which is what Mandel (1977) referred to as temporal motion.
Figure 13.3 shows an example involving the VEHICLE classifier.
It should be evident by this point that the perception of iconicity in sign
language is not limited to sign language users. Nonsigners often can pick up
an occasional sign in an otherwise inscrutable language stream, just as a
speaker of English might recognize the odd word in a conversation held in
French. For example, a one-hand (see Fig. 13.1) classifier, representing a
person descending a staircase might easily be recognized by someone with­
out sign language experience (this is usually referred to as transparency).
Klima and Bellugi (1979) asked American hearing individuals with no
knowledge of ASL to identify the relations between signs and their English
meanings or simply identify the meanings of isolated signs. They found only
limited transparency of their 90 stimulus signs, ranging from 9% in a free iden­
tification task to 20% in a multiple-choice task. O'Brien (1999) examined
iconic and metaphoric characteristics of isolated signs, asking hearing indi­
viduals to identify meanings of iconic, metaphorical, and arbitrary ASL signs.
She found that although iconic signs were the most obvious of the three, peo­
ple found metaphorical signs to be more sensible, natural, and transparent
than arbitrary signs. Finally, Pizzuto and Volterra (2000) showed 40 signs from
LIS, 20 considered transparent and 20 not, to nonsigning individuals from six
countries and 12 language communities (both signed and spoken). Overall, at
least 50% the non-Italians were able to correctly identify the meanings of 13
of the 20 transparent signs, whereas none of the nontransparent signs reached
that criterion (and only two approached it). Deaf individuals from other
countries were better able to guess meanings than hearing individuals for all
but two signs, but variation across hearing status and country of origin indi­
cated cultural as well as linguistic bases for perception of iconicity (cf.
O'Brien, 1999). Taken together, such findings offer some consistency, but
there is considerable variability in the ability of nonsigners to see the regularity
or bases of signs. Such findings should not be surprising, as they mirror well
deaf people's difficulty in mastering the irregular aspects of many spoken and
written languages. It would be expected that the most irregular languages, like
English, would be particularly difficult for deaf individuals to learn, whereas
more phonologically regular languages, like Italian, would be somewhat eas­
ier. Unfortunately, differences in deaf education across countries like the
United States and Italy make such comparisons difficult, and they have not
yet been undertaken.6

6
There are, however, suggestive findings showing that dyslexia is more common among
hearing children learning irregular languages such as English or French than those learning
regular languages such as Italian (Paulesu et al., 2001).
318 MARSCHARK

Representation of Time

One of the most potentially interesting metaphoric qualities of sign lan­

guage is its representation of time. Within all signed languages, temporal in­
formation is represented via the use of space in ways that preserve
dimensions of relative time. Several different time lines can be found in
ASL, and other languages have additional ones or variations of those found
in ASL (see, e.g., Engberg-Pedersen's, 1993, description of Danish Sign Lan­
guage) . The most obvious of these time lines is referred to as the deictic time
line, extending forward and backward from the signer's dominant shoulder
to indicate time relative to the time of the utterance (Emmorey, 2001;
Engberg-Pedersen, 1993). Whereas signs like NOW and TODAY (and,
nonmetaphorically, HERE) as well as time-related signs like WEEK and
MONTH are made in the neutral sign space just in front of the signer's body,
the deictic time line is used to indicate both past and future. Thus, MON­
DAY is made relatively close to the signer's body, to the right side of the neu­
tral sign space, whereas NEXT MONDAY is usually signed NEXT-WEEK
MONDAY, with the sign WEEK made with a forward moving arc indicating
"into the future." Alternatively, the next Monday or two or three Mondays
in the future might be made, metaphorically, by signing MONDAY and
"bumping" it forward, in the future, one, two, or three times.
A second, fairly obvious time indicator is the sequence time line. Extending
from left to right across the front of the signer's body, the sequence time line
indicates relative placement in a temporal sequence, with the reference
point determined by the time of focus of the description, placed (signed) di­
rectly in front of the signer. Thus, a recounting of the various steps involved
in the preparation of this chapter centered on the "present" task of sitting at
the keyboard and composing (as this sentence first is being written) might
also include, to the left, going to the library and collecting references, writ­
ing notes, and assembling information. To the right might be revising the
document and, in a more creative moment, the sign indicating PAPER
COMING OFF PRINTER, representing the publication of the chapter in
the book.
Somewhat less obvious to naive observers of signing is the anaphoric time
line (Engberg-Pedersen, 1993). Used to compare or contrast time periods re­
lated to the topic of a discourse (Emmorey, 2001), the anaphoric time line
extends diagonally through the sign space, from the nondominant side of
the signer forward to the dominant side. Unlike the deictic or sequence time
lines, the anaphoric time line does not have a neutral here-and-now default.
Rather, the events or topics are described relative to each other, as if each
was represented by a separate positional pronoun with anaphoric reference
relative to others. Describing the evolution of metaphor research, for exam­
ple, might make better use of an anaphoric time line rather than a sequence
13. METAPHORS IN SIGN LANGUAGE 319

time line, capturing the various paradigms and theoretical perspectives that
have been seen in the field over the past half century or so, rather than a se­
quence of specific events relative to a particular time. Starting wherever the
focus of the discussion might be (e.g., research on deaf children's sign lan­
guage metaphors in the 1980s), the signer's body would be placed along the
diagonal anaphoric time line relative to other events being described, not in
order of mention (i.e., requiring shifting forward and backward in time along
the time line).
Beyond these primary time lines are two other uses of a time indicator
that occurs in a two-dimensional plane in front of the signer, parallel to the
body. The calendar plane (Engberg-Pederson, 1993) iconically captures a
calendar in some ways, for example, as the sign EVERY-FRIDAY is made by
drawing the FRIDAY F-handshape down a column of Fridays on a calendar.
CLASS EVERY-MONDAY AND EVERY-WEDNESDAY also would pre­
serve the relative positions in the (horizontal) week. In Danish Sign Lan­
guage, the calendar plane is used primarily for describing the year, whereas
in ASL, it is most often used for descriptions within a week. In addition, the
calendar plane serves as a generic indicator of successive topics in a dis­
course, not necessarily temporally related outside of the discourse, indicat­
ing the NEXT in a sequence, moving from higher to lower in the plane (e.g.,
changing topics in a classroom psychology lecture, which is not particularly
tied to time).
In short, the use of space to represent time in ASL and other sign lan­
guages offers rich metaphorical opportunities to highlight temporal quality
and relative position. Unfortunately, there does not appear to be any re­
search concerning the development of time lines in either deaf children or
second language learners of a signed language. The possibility that the de­
piction of the abstraction of time might result in some cognitive correlates in
children (as might the difference between learning to tell time with a digital
vs. an analog clock), however, suggests that it might be a fruitful area of
study (Stokoe & Marschark, 1999).

METAPHORS IN SIGN LANGUAGE

AND USING SIGN LANGUAGE METAPHORICALLY

The previous section noted that classifier handshapes are frequently used to
represent various kinds of objects, often located or moving in a designated
environment. As in spoken language, such signs can also be used metaphori­
cally to represent abstract concepts. Indeed, many of the conceptual meta­
phors found in spoken English are also found in ASL, and presumably a
considerable number of those occur in other languages as well. Conceptual
metaphors, such as "events are movers" (Lakoff & Johnson, 1980; Lakoff &
Turner, 1989), involve abstract domains described in terms of concrete do­
32O MARSCHARK

mains, with numerous and consistent transfers between them (also see
McNeill's, 1992, treatment of metaphoric gestures). Without a sensory basis
for understanding and communicating abstract ideas, language users often
resort to mapping those ideas on parallel physical entities and relations. The
target (or topic) domain is that actually being referred to, whereas the source
(or vehicle) domain is that being used in the description. How it is that chil'
dren come to understand conceptual metaphors and their psychological
utility for language users of all ages are two of the more fascinating areas of
psycholinguistic investigation (e.g., Glucksberg, 2001). Unfortunately, they
are also beyond the scope of this chapter.
In ASL, many signs for abstract concepts involve visual aspects of ob­
jects or actions that are figuratively part of or associated with the abstrac­
tions. One of the most commonly cited examples in treatments of
metaphor in ASL (as well as in English) is that involving mapping from the
abstract domain of communication (the target) to the sending and receiv­
ing of objects or ideas (the source), that is, the conceptual metaphor "com­
municating is sending." In ASL, however, "communicating is sending"
involves a double mapping of both metaphoric and iconic components
(Taub, 2001a). Thus, in one oft-cited example, an idea represented by a
one-hand (see Fig. 13.1) can move from the signer's head to another per­
son but "not get through"—communicated via the sign THINK­
BOUNCE, in which the handshape bounces off of a raised flat hand repre­
senting the impenetrable mind of another. Alternatively, if the point fi­
nally does sink in, THINK-PENETRATE shows the idea (the index
finger) slowly penetrating the fingers of the previous blockage. The
iconicity of sign language thus is often reflected in the shapes and move­
ments of the hands representing analogous aspects of the referent, whereas
the metaphoric quality overlays and binds the abstract (or concrete) rela­
tion between the abstract and the concrete referents.
A related example common to ASL, spoken English, and other languages
is the conceptual metaphor of "mind as container." In addition to mapping
cognitive processes onto containers, either in the head or elsewhere, iconic
properties of the ASL metaphor, similar to those in "communicating is send­
ing," provide a double mapping not available in spoken language. Thus, for­
getting or overlooking something can be signed by making the EMPTY or
VACANT sign across the forehead instead of its usual location on the back
of the nondominant hand, indicating MIND-EMPTY. Similarly, a sign for
NONE, a zero-hand, usually made with a brief movement away from the
body, can be moved to the forehead, indicating the same thing. The feeling
of losing a thought is often signed with the 1-hand of a thought gradually
sinking through a horizontal flat hand (the conceptual and physical oppo­
site of THINK-PENETRATE) until it disappears completely; recognition of
a complete if momentary loss of reason (also known as realizing you have
13. METAPHORS IN SIGN LANGUAGE 321

done something really stupid) might be signed by taking the sign for BRAIN
(made with two hands), removing it from the head, and, while holding it like
an ostrich egg, breaking it onto the floor or whisking it in a bowl.
Wilbur (1987) described the use of the C-hand classifier, normally used
to represent size and shape of cylindrical objects, in the "mind as con­
tainer" metaphor. Made at the forehead, the C shape can open (expand) or
close to indicate amount of knowledge. Signers put information into the
container, take information out, and a double-handed C comprises the
sign BRILLIANT (as in intelligence; there is a different sign for use with
light). Liddell (1995) described use of the same hand shape, also used in
the sign CULTURE (with the same movement as FAMILY, GROUP and
CLASS), to capture "culture as container," as individuals "carry" their at­
titudes and behaviors in their perceptions of their own culture.
Related to the "mind as container" and "communicating is sending"
metaphors is the "ideas are objects" metaphor. The signs for LEARN and
STUDENT (literally, LEARN-PERSON) both involve drawing ideas up
from the printed page into the head. INFORM and TEACH both entail
taking ideas from the head and sending them outward. As verbs, LEARN,
INFORM, and TEACH all can be inflected to denote manner, duration,
and so on. But, as a directional sign, INFORM also can be iconically in­
flected in an established sign space to indicate who informed whom, as well
as in what manner. Similarly, the directional sign GIVE can be inflected to
indicate periodic or continual passing out of papers to members of a class in
either a uniform or seemingly random order. As indicated earlier, such
combinations of metaphor and iconicity in sign language are referred to as
double mapping.
Several investigators (e.g., Emmorey, 2001; Taub, 200la, 2001b; Wilcox,
2000) have described the double mapping of qualitative information in the
use of sign space, which is comparable to the (single) mapping of quality and
space in spoken language. For example, the common observation is that
"good is up" and more good is more up (see Fig. 13.3). As indicated in Table
13.1, power, status, happiness, quantity, and value are also up in the sign
space, either made with upward movements or signed at higher levels in the
sign space to indicate relative quality, whereas their want is indicated by
downward movement or lower signing (also see Lakoff & Johnson, 1980).7
The availability and use of double mapping may or may not bestow any cog­
nitive or linguistic benefits to the native or near-native signer, but it cer­
tainly provides an added creative characteristic to the artistic use of signing,
as in poetry or storytelling, as is described in the next section.

The "good is up" relation is not just seen in signed and spoken languages. In Bliss symbols,
good is represented by an upward arrow, feeling by a heart, and happy by a heart with an upward
arrow beside it (and a caret above the heart).
 MARSCHARK

TABLE 13.1
Examples of ASL Signs

"Positive is up" "Negative is down"

HAPPY SAD
IMPORTANT LOUSY
THRILLED DEPRESSION
INSPIRED WORTHLESS
BETTER, BEST BAD
IMPROVE DETERIORATE
SUCCESSFUL OPPRESSION
RICH CHEAP
WIN LOSE
OUTSTANDING FAILURE
FAMOUS IGNORE
APPEAR DISAPPEAR

Nonmetaphor Trope in ASL

Before moving from the linguistics of figurative language to its psychology, it is
worth noting that the other categories of trope typically found in spoken lan­
guages are also found in signed languages. As alluded to in the previous sec­
tion, hyperbok (exaggerating magnitude) and litote (minimizing magnitude)
occur frequently and naturally in most signed languages by modifying sign size
or movement, often with accompanying facial expression (Marschark,
LePoutre, & Bement, 1998). Simile is fairly obvious and iconic in ASL; for ex­
ample, the sign LIKE is made with a prone Y-hand shape that moves side to
side, often between two objects, people, or concepts that have been estab­
lished in the sign space. Metonymy (part representing the whole) is also a com­
mon aspect of ASL; a number of signs are represented by an iconic depiction
of one of their parts. Signs for many animals, in particular, are metonymic,
showing the stripes for TIGER, ears for RABBIT and HORSE, whiskers for
CAT and MOUSE, and the trunk for ELEPHANT, among many others. Al­
though such observations are encouraging, apparently the only empirical re­
search to date on nonmetaphoric trope in sign language is the work by
Marschark and colleagues (e.g., Everhart &. Marschark, 1988; Marschark,
West, Nall, & Everhart, 1986), documenting that such constructions are pro­
duced by deaf children and by deaf and hearing mothers who sign.
ASL also includes a number of examples of what Goossens (1990) referred to
as metaphtonymy, or metaphorical use of metonymy. Wilcox (2000, pp. 92-96)
13. METAPHORS IN SIGN LANGUAGE 323

described one example of metaphtonymy, THINK-HEARING, which she saw

as illustrating "the powerful conflicting cultural influences" experienced by cul­
turally Deaf individuals through their interactions with hearing people and deaf
individuals who are not part of the Deaf community. First consider the sign
SPEAK, which involves a 1-hand rotating, clockwise, parallel to and just in
front of lips. Metonymically, the sign also means HEARING, as in a hearing per­
son (but not in hearing in the auditory sense). The sign THINK has the same
movement as HEARING, but it is normally made to the side of the temple in­
clined to a 45° angle. However, when the HEARING sign movement (horizon­
tal) is made in front of the forehead instead of the lips, the result is
THINK-HEARING, a sign that negatively and metaphtonymically refers to a
deaf person who thinks and acts like a hearing person. More neutrally,
HEARING also is used metaphtonymically in HEARING-SCHOOL, mean­
ing public school. As with other kinds of trope seen within sign language, such
documentation hopefully will soon lead to empirical investigation.

COMPREHENSION AND PRODUCTION OF METAPHOR

BY DEAF INDIVIDUALS

Over the past 20 years or so, metaphor research has yielded a rich body of lit­
erature concerning the psychological processes involved in the comprehen­
sion of figurative language. Research involving hearing individuals, both
adults and children, has been reviewed elsewhere and will not be reconsid­
ered here (see Gibbs, 1994; Glucksberg, 2001; Katz, Cacciari, Gibbs, &
Turner, 1998; Ortony, 1993). Research concerning the comprehension of
metaphor by deaf adults and children, however, has been more limited. For
adults, there appears to be no published research at all concerning compre­
hension of figurative language either in sign language or in the vernacular.
For figurative language in printed materials, this gap in the literature likely
reflects the well-documented finding that deaf individuals have significant
challenges in the domain of literacy, with the 50th percentile of deaf
18-year-olds reading at approximately the fourth-grade level (Traxler,
2000). Teachers of deaf students of all ages (including the author) are well
familiar with their struggle to understand English metaphors and idioms, at
least in part as a result of their relatively impoverished early language experi­
ence (see Marschark et al., 2002, chaps. 4,5, & 8, for discussion) .8 Teachers
often work with deaf students on comprehension of metaphors in text, but it
is unclear whether the general principles underlying figurative construc­
tions are understood and transfer across domains or types of trope.

8
These generalizations refer to the 95% of deaf children who have hearing parents. Deaf
children of deaf parents have rich early language experience and tend to show better aca­
demic and reading achievement (see Marschark et al., 2002).
324 MARSCHARK

The comprehension and production of metaphor by deaf children has re­

ceived some empirical attention (discussed later), but research concerning
how or when deaf adults comprehend metaphors within sign language is
clearly needed. Even if deaf college students and other deaf adults were to
show lower levels of performance relative to hearing peers, examination of
factors that influence their comprehension could be enlightening. Al­
though there has been a history of assuming that deaf and hearing individu­
als have identical cognitive processes, we now know that there are some
reliable differences with regard to imagery and verbal processes, as well as
memory. In particular, despite considerable overlap, deaf and hearing col­
lege students have recently been found to show significant variation in the
organization of conceptual knowledge, associations among conceptual do­
mains, and the symmetry or asymmetry (relative to hearing peers) in the
links between exemplars and their category names (Marschark et al., 2001;
McEvoy, Marschark, & Nelson, 1999).
Several investigators have demonstrated that the distance between the
domains of a metaphoric tenor (that being described, usually the subject)
and vehicle (that to which a comparison is being made, usually the object)
can affect comprehension and memory of metaphors (e.g., Marschark &
Hunt, 1985; Tourangeau & Sternberg, 1981; Trick & Katz, 1986). In gen­
eral, the typical finding is that domains that are farther apart, but not too far,
lead to better comprehension, memory, and judgment of a metaphor's qual­
ity, supporting interactionist theories of metaphor comprehension (e.g.,
"Juliet is a rose" versus "Juliet is a bartender" or "Juliet is a fax machine").
Such studies have not been conducted with deaf individuals or, apparently,
with individuals presented with metaphors in their second language. In view
of the findings indicating less well-bounded and strongly interconnected
conceptual domains among deaf individuals, such research might be partic­
ularly enlightening, both theoretically and pedagogically.

Metaphor Comprehension by Deaf Children

Deaf children's comprehension of metaphor has historically been examined

in deaf studies as a means of evaluating their linguistic and cognitive abili­
ties. Following the early work of Piaget (1923/1974), many investigators ex­
amined figurative language abilities in hearing children as an indicator of
both linguistic flexibility and a more global cognitive ability to consider the
world from alternative perspectives (e.g., Gardner, Winner, Bechofer, &
Wolf, 1978). Studies of deaf children occurred naturally from such work, fol­
lowing from earlier studies of nonverbal creativity (see Marschark & Clark,
1987, for a review). In particular, figurative language abilities have been
shown to depend on classification skills and, especially, the ability to see re­
lationships across domains as superseding superficial similarities, an area in
13. METAPHORS IN SIGN LANGUAGE 325

which deaf children are seen to lag behind hearing peers (Ottem, 1980).
Studies by Rittenhouse, Morreau, and Iran-Nejad (1981) and Inman and
Lian (1991) showed that metaphor comprehension in deaf children is re­
lated to performance in Piagetian conservation tasks, further indicating a
link between those skills.
If deaf children think as concretely and literally as early investigators re­
ported (e.g.,Blackwelletal., 1978;Boatner&Gates, 1969; Myklebust, 1964),
they would not be expected to either comprehend or produce nonliteral lan­
guage. Alternatively, because of its compactness, figurative language provides
an extremely efficient and vivid form of communication that could allow deaf
children to communicate ideas that might not be expressible in literal terms.
Several investigators have emphasized the importance of this aspect of figura­
tive language for young hearing children who have much to say but meager
linguistic resources at their command (e.g., Gardner et al., 1978; Petrie,
1979). Thus, metaphor should be of particular utility to deaf children, across a
wider age range, due to their having smaller vocabularies than hearing peers.
Given the deaf children's relative lack of experiential diversity and lan­
guage skills in the vernacular (Marschark et al., 2002), it would not be sur­
prising to find that they have little skill in understanding the many
nonliteral aspects of English (Boatner & Gates, 1969). Conley (1976) was
apparently the first to evaluate that assumption, comparing deaf and hear­
ing students on a test of English idiom comprehension. Despite the fact that
the vocabulary of test sentences were selected from lower reading levels, she
found that the deaf children were significantly poorer in their (English) id­
iom comprehension than hearing peers, at least at the middle range of read­
ing levels. However, she also found that deaf and hearing children at lower
reading levels did not differ significantly in their performance (see also
Easterbrooks, 1983).
Iran-Nejad, Ortony, and Rittenhouse (1981) argued that the lack of
comprehension indicated in such studies resulted from deaf children's diffi­
culties in understanding written English, rather than in anything specific to
nonliteral language (Paul, 2003). They presented deaf 9- to 17-year-olds
with short stories in which syntax, vocabulary, and world knowledge were
controlled. The students read the stories, which were accompanied by a sim­
ple picture depicting the main point, and then selected from a set of four lit­
eral sentences, similes, or metaphors the sentence that best completed each
story. Iran-Nejad et al. found that even their youngest deaf students could
select nonliteral sentences as best completions for the stories, a finding repli­
cated by Rittenhouse et al. (1981) and Fruchter, Wilbur, and Eraser (1984)
with deaf children who signed and Rittenhouse and Kenyon (1991) with
children who used spoken language.
Payne and Quigley (1987), in contrast, controlled vocabulary and syntax
and included thematic pictures in an investigation of comprehension of idi­
326 MARSCHARK

omatic phrases by deaf and hearing students and still found a consistent
hearing status difference. Their study involved 10- to 19-year-old deaf stu­
dents and 8- to 12-year old hearing students. The participants were shown
simple pictures (e.g., men holding up a store) and asked them to select the
most appropriate verb-particle phrases from among three choices. The
phrases included three levels of semantic difficulty: literal (e.g., walks out),
semi-idiomatic (e.g., washes up), and idiomatic (e.g., gives up) embedded in
five syntactic structures: subject-verb-adverb, subject-verb-adverb-objec
subject-verb-object-adverb, subject-verb-preposition-object, and subject-
verb-adverb-preposition-object. Payne and Quigley found that the hearing
students performed significantly better than the deaf students on all levels of
semantic difficulty and for all syntactic constructions. In addition, they
found that performance was highly related to reading comprehension abil­
ity. More recently, Sawa (1999) had Japanese deaf and hearing high school
students paraphrase similes like "The heart is like the sea." Results indicated
that almost half of the interpretations provided by the deaf students were ei­
ther inappropriate or simply restated the simile. Unlike hearing peers, the
deaf students were significantly more likely to explain the similes with sin-
gle-attribute predicates than multiattribute predicates. Sawa and Yoshino
(1995) also found that metaphor comprehension and production were
strongly correlated among deaf junior high school students.9
At this point, it is clear that comprehension of metaphor in the vernacu­
lar remains problematic for deaf adults and children. Such difficulty appears
to be tied largely to print literacy, although carefully controlled studies still
need to be conducted. Further, until studies are conducted concerning the
comprehension of signed metaphor by deaf adults and children, both in and
out of context, we do not know if there are any differences in figurative com­
prehension, broadly defined, between deaf and hearing individuals. If deaf
individuals show typical patterns of figurative comprehension within sign
language, educators should be able to use them to improve comprehension
of metaphor in print.
Metaphor Production by Deaf Children
To obtain an indication of deaf children's figurative language abilities out­
side of spoken and written language, some researchers have focused on the
use of figurative expression in their sign language production. Theoretically,
what appears to be metaphor expertise on the part of deaf children could be
considered (a) a simple reflection of their acquiring the apparently meta­
phoric and iconic aspects of sign language use (e.g., Emmorey, 2002, p. 82);
9
The Sawa and Yoshino (1995) and Sawa (1999) articles were published in Japanese, and
only English abstracts are available. A complete review of their results, therefore, could not
be included here.
13. METAPHORS IN SIGN LANGUAGE 327

(b) demonstrative of creative, flexible thinking at least within the context of

sign language production (e.g., Everhart & Marschark, 1988); or (c) an in­
dicator of limited (literal) language ability. This last alternative might be
considered a straw man, especially because there is so little research on the
issue. However, in the context of the discussion of metaphor in sign lan­
guage and sign language users, it is best thought of as an empirical question
that likely has a complex answer (Everhart & Marschark, 1988).
Several studies have shown that when evaluated in sign language, deaf
children display linguistic competencies not evident from their use of and
evaluation in printed language. This was first demonstrated by Marschark
and West (1985), who examined fantasy-based stories signed by deaf chil­
dren and spoken by hearing children between the ages of 12 and 15 years
old. The most important general finding was that there were several types of
figurative language used consistently by both deaf and hearing children. In
contrast to studies of their comprehension of English idioms, Marschark and
West found that the deaf children produced novel and frozen trope just as
often as their age-matched hearing peers. The deaf children also reliably
surpassed their hearing peers in their frequency of using gesture, panto­
mime, novel but grammatically acceptable linguistic modifications, and lin­
guistic inventions. Everhart and Marschark (1988) explicitly showed that
deaf children's production of trope in sign language significantly exceeded
their production in written language, whereas hearing children produced
more trope in spoken than written language.
Marschark et al. (1986) examined developmental changes in figurative
production by deaf and hearing children between the ages of 7 and 15 years
old. Beyond replication of the earlier findings, the most interesting result
was the finding of a literalperiod in deaf children's language productions. Pre­
vious studies had demonstrated a point at around 8 or 9 years of age when
the use of figurative language by hearing children is at a minimum. This hia­
tus has variously been described as a consequence of perceived academic de­
mands, the emergence of an analytic attitude in concrete operations, and
the point at which linguistic and cognitive skills have matured enough that
children stop producing apparently figurative category violations that actu­
ally stem from conceptual error (i.e., they are intended literally because chil­
dren do not recognize the metaphoric tension). According to the latter
interpretation, it is only after the literal period that true figurative language
use begins.
The literal period in deaf children's language observed by Marschark et al.
(1986) was its first demonstration within sign language. Perhaps the most in­
teresting aspect of that finding, however, was that the literal period of the deaf
sample was observed at around age 12 or 13, 3 to 4 years later than the mini­
mum evident in the hearing sample. This result was consistent with
Iran-Nejad et ai.'s (1981) suggestion that deaf children's lack of linguistic ex­
328 MARSCHARK

perience should be reflected in various lags in their language use. More gener­
ally, it indicated that the development of linguistic creativity in deaf children
follows a similar but slightly delayed course relative to hearing children.

Figurative Production by Deaf Adults

There have been no published studies concerning deaf adults' comprehen­

sion of metaphor, and apparently all but one of the investigations of their fig­
urative production have had a linguistic focus. In the one psychologically
oriented study, Marschark, Everhart, and Dempsey (1991) explored the fig­
urative language of signing deaf women, speaking hearing women, and bilin­
gual (signing and speaking) hearing women who had ASL as their first
language (i.e., learned from their deaf parents). The women all made up and
told stories on supplied themes, allegedly to be shown to 4- and 10-year-old
children, while being videotaped (the bilingual group told stories in both
sign language and spoken language). When all forms of nonliteral language
were included (e.g., novel and frozen trope, rhetorical questions, panto­
mime, and invented signs and words), the deaf mothers produced the most
nonliteral language. The two groups of hearing women produced equal
amounts of nonliteral content, whereas the bilingual mothers produced
more when signing than when speaking. In terms of the frequency of using
only novel trope, Marschark et al. found that deaf mothers and the bilin-
gual-speaking mothers produced the most figurative language, followed by
the bilingual-signing mothers, and finally the hearing mothers. Although
individual categories of novel trope were not considered, the results indicate
that the equal or greater frequency of figurative language found among deaf
signers was not simply a reflection of some greater flexibility in ASL relative
to English. Rather, there is clearly an unexplored interaction among lan­
guage modality, language user, and metaphor.
Turning to linguistic analyses of metaphor in sign language, Wilcox
(2000) provided an in-depth analysis of the structure and metaphoric con­
tent of a story entitled, The Dogs, by deaf poet Ella Mae Lentz (1995). Pub­
lished on video, the story uses a fight between a well cared for, purebred
Doberman and a stray mutt, chained together and unable to escape each
other, as a metaphor for cultural discord within the Deaf community. Wilcox
identified several recurring metaphoric and metonymic themes including
"the mind is a body," "social relations are spatial relations," "social con­
straint is physical constraint," and "involuntary social unity is involuntary
physical connectedness." Her analysis provides insights into Deaf culture,
the beauty and expressiveness of ASL, and metaphor within signed lan­
guages. Although it reflects Lentz's obvious creativity and figurative deft­
ness, the analysis was not intended to examine the psychology of the artist.
Indeed, this might better be left to a cultural anthropologist.
13. METAPHORS IN SIGN LANGUAGE . 329

Taub (2001b) offered a similar analysis of The Treasure, also by Lentz

(1995), providing a linguistic and metalinguistic analysis of a poem fraught
with metaphor and iconicity. In The Treasure, Lentz uncovered a beautiful
treasure (a metaphor for sign language) buried underground and attempted
to share her discovery with others. The way in which she and her discovery
are rebuffed, and the treasure eventually reburied, offers both linguistic and
cultural insights into ASL and the American Deaf community, similar to
those in The Dogs. Taub described three categories of structures in the poem,
literal (nonmetaphorical) signs, lexicalized metaphorical signs, and free meta­
phorical signs. Earlier, Marschark and West (1985) referred to the two latter
categories as frozen trope and novel trope, respectively, labels that are more in
keeping with the metaphor literature. For Taub's analysis of The Treasure,
like Wilcox's (2000) analysis of The Dogs, it is the linguistic quality of the
poem that is of interest, rather than anything about the language user. For
the time being, we therefore must await study of the outstanding deaf artists
who use sign language as the tools of their craft and less experienced signers
who strive to develop such skills.

SUMMARY AND CONCLUSIONS

This chapter has explored the use of metaphor and other trope in sign Ian­
guage and its comprehension and production in sign language users. If meta­
phor has been seen to be an essential component of spoken language (Lakoff
& Johnson, 1980), its centrality in signed languages seems even greater. The
iconic property of signed languages, having signs that look or move like what
they represent, and their metaphorical qualities, the symbolic likening or
highlighting of dimensions across conceptual domains, have been confused in
earlier work in the area, but the two aspects of signing are now seen to func­
tion together in creating its particularly apt and depictive representations.
Iconicity and metaphor are essential components of the visuospatial nature of
signing, and the availability of double mapping, the incorporation of both
components simultaneously in the same sign, gives sign language a figurative,
descriptive quality that makes spoken language pall by comparison.
Iconicity in sign language includes the use of handshapes or motion to
demonstrate position, shape, or other qualities of objects. These classifiers can
also be used as pronouns that preserve anaphoric relations among other, pre­
viously established entities, either variously shaped objects (including animals
and people) or abstract entities (e.g., time periods, concepts, and other rela­
tions) . It likely is the use of iconicity in sign language that sometimes leads
hearing people to think that signing is more gestural than linguistic, but the
many sign languages of the world are indeed true languages (Stokoe, 1960)
that are often accompanied by gestures as well (Marschark, 1994) • The de­
gree to which individual signs are transparent to nonsigners varies with the
33O MARSCHARK

language, the culture, and the particular signs involved. Nonetheless, many of
the metaphoric devices found in spoken languages (e.g., "good is up" and "bad
is down") are depicted iconically in sign language.
The use of metaphor in sign language has been well documented by
those studying various languages, and consistencies across various na­
tional sign languages indicate the ubiquity of figurative language regard­
less of its mode. There have not yet been studies on the development of
metaphoric and iconic characteristics in young signers, however, other
than a few cross-sectional studies showing similar if delayed patterns of de­
velopment in deaf children relative to hearing peers (e.g., Marschark et al.,
1986). Further studies of metaphor and iconic sign language development
could consider the conceptual and linguistic knowledge underlying lan­
guage use and the possible utility of metaphor and iconicity in teaching
hearing as well as deaf children.
A variety of studies, primarily involving school-age children, have dem­
onstrated that because of their literacy-related challenges, and perhaps be­
cause of relatively limited world knowledge, deaf children and adolescents
demonstrate poorer performance than hearing peers on tests of metaphor
comprehension evaluated via print materials. Although such results long
were taken as reflecting concrete and literal thinking on the part of deaf peo­
ple, more recent research—both psychological and linguistic—has shown
that deaf individuals and other fluent users of sign language incorporate
metaphoric mappings in their sign language and produce figurative con­
structions just as often as their hearing peers.
Research concerning the comprehension of metaphor in sign language by
deaf adults or children is completely lacking. One would hope that results
from future work would indicate full comprehension of signed metaphor,
comparable to that observed by hearing individuals with spoken metaphor.
Nonetheless, recent evidence indicating somewhat different organization
and use of conceptual information by deaf as compared to hearing individu­
als suggests the potential for variation—variation that would be of both the­
oretical and educational importance. As in the other areas of investigation
described in this volume, such work would provide additional windows onto
the nature of language and language users.

REFERENCES

Bellugi, U., & Klima, E. (1976). Two faces of sign: Iconic and abstract. In S. Harnad,
D. Hoest, & I. Lancaster (Eds.), Origins and evolution of language and speech (pp.
514-538). New York: New York Academy of Sciences.
Blackwell, E, Engen, E., Fischgrund, J., & Zarcadoolas, C. (1978). Sentences and
other systems: A language and learning curriculum for hearing-impaired children.
Washington, DC: National Association of the Deaf.
13. METAPHORS IN SIGN LANGUAGE 331

Boatner, M. T., & Gates, J. E. (1969). A dictionary of idioms for the deaf. Washington,
DC: National Association for the Deaf.
Boyes Braem, P. (1981). Features of the liandshape in American Sign Language. Doc­
toral dissertation, Department of Linguistics, University of California, Berkeley.
Brennan, M. (1990). Wordformation in British Sign Language. Stockholm: University
of Stockholm.
Conley, J. E. (1976). The role idiomatic expressions in the reading of deaf children.
American Annals of'the Deaf, 121, 381-385.
Easterbrooks, S. R. (1983). Literal and metaplioric understanding of four pairs of polar ad­
jectives across four domains by hearing and hearing impaired children at two age levels.
Doctoral dissertation, Department of Special Education, University of Georgia.
Emmorey, K. (2001). Space on hand: The exploitation of signing space to illustrate
abstract thought. In M. Gattis (Ed.), Spatial schemas and abstract thought (pp.
147-174). Cambridge, MA: MIT Press.
Emmorey, K. (2002). Language, cognition, and the brain. Mahwah, NJ: Lawrence
Erlbaum Associates.
Engberg-Pedersen, E. (1993). Space in Danish Sign Language: The semantics and
morphsyntax of the use of space in a visual language (International Studies on Sign
Language Research and Communication of the Deaf, No. 19). Hamburg, Ger­
many: Signum-Verlag.
Everhart, V. S., & Marschark, M. (1988). Linguistic flexibility in signed and written
language productions of deaf children. Journal of Experimental Child Psychology,
46, 174-193.
Friedman, L. (Ed.). (1977). On the other hand: New perspectives on American Sign
Language. New York: Academic Press.
Fruchter, A., Wilbur, R. B., &Fraser,J. B. (1984). Comprehension of idioms by hear-
ing-impaired students. The Volta Review, 86, 7-19.
Gardner, K, Winner, E, Bechofer, R., & Wolf, D. (1978). The development of figu­
rative language. In K. Nelson (Ed.), Children's language (Vol. 1). New York:
Gardner.
Gibbs, R. (1994). The poetics of mind. Cambridge, England: Cambridge University
Press.
Glucksberg, S. (2001). Under standing figurative language. New York: Oxford Univer­
sity Press.
Goossens, L. (1990). Metaphtonymy: The interaction of metaphor and metonymy
in expressions for linguistic action. Cognitive Linguistics, 1, 323—340.
Inman, P. R., & Lian, M.-G. J. (1991). Conservation and metaphor performance
among children with hearing impairments. Journal of the American Deafness and
Rehabilitation Association, 25, 28-41.
Iran-Nejad, A., Ortony, A., & Rittenhouse, R. K. (1981). The comprehension of
metaphorical uses of English by deaf children. Journal of Speech and Hearing Re­
search, 24, 551-556.
Katz, A. N., Cacciari, C., Gibbs, R. W., Jr., & Turner, M. (1998). Figurative language
and thought. New York. Oxford University Press.
Klima, E., & Bellugi, U. (1979). The signs of language. Cambridge, MA: Harvard
University Press.
332 MARSCHARK

Lakoff, G., & Johnson, M. (1980). Metaphors we live by. New York: Cambridge Uni­
versity Press.
Lakoff, G., & Turner, M. (1989). More than cool reason: A field guide topoetic metaphor.
Chicago: University of Chicago Press.
Lentz, E. M. (1995). The treasure: Poems by EllaMaeLentz- Berkeley, CA: InMotion.
Liddell, S. (1990). Four functions of a locus: Reexamining the structure of space in
ASL. In C. Lucas (Ed.), Sign language research: Theoretical issues (pp. 176-198).
Washington, DC: Gallaudet University Press.
Liddell, S. (1995). Real, token, and surrogate space: Grammatical consequences in
ASL. In K. Emmorey & J. Reilly (Eds.), Language, gesture, and space (pp. 19–41).
Hillsdale, NJ: Lawrence Erlbaum Associates.
Luria, A. R. (1976). Cognitive development: Its cultural and social foundations. Cam­
bridge, MA: Harvard University Press.
Mac Cormac, E. R. (1985). A cognitive theory of metaphor. Cambridge, MA: MIT
Press.
Mandel, M. (1977). Iconic devices in American Sign Language. In L. A. Friedman
(Ed.), On the other hand: New perspectives on American Sign Language (pp.
57-107). New York: Academic Press.
Marschark, M. (1993). Psychological development of deaf children. New York: Oxford
University Press.
Marschark, M. (1994). Gesture and sign. Applied Psycholinguistics, 15, 209-236.
Marschark, M., & Clark, D. (1987). Linguistic and nonlinguistic creativity of deaf
children. Developmental Review, 7, 22-38.
Marschark, M., Convertino, C., McEvoy, C., &Masteller, A. (2001, November).
Organisation and use of the mental lexicon in deaf and hearing individuals. Poster ses­
sion presented at the annual meeting of the Psychonomic Society, Orlando, FL.
Marschark, M. & Everhart, V. S. (1999). Problem solving by deaf and hearing chil­
dren: Twenty question. Deafness and Education International, I, 63-79.
Marschark, M., Everhart, V. S., & Dempsey, P. (1991). Nonliteral content in lan­
guage productions of deaf, hearing, and native-signing hearing mothers.
Merrill-Palmer Quarterly, 37, 305-323.
Marschark, M., & Hunt, R. R. (1985). On memory for metaphor. Memory &Cogni­
tion, 13,413-424.
Marschark, M., Lang, H. G., &. Albertini, J. A. (2002). Educating deaf students: From
research to practice. New York: Oxford University Press.
Marschark, M., LePoutre, D., & Bement, L. (1998). Mouth movement and signed
communication. In R. Campbell & B. Dodd (Eds.), Hearing by eye II: Thepsychol­
ogy of speechreading and auditory-visual speech (pp. 243–264). London: Taylor &
Francis.
Marschark, M., &Nall, L. (1985). Metaphoric competence in cognitive and lan­
guage development. In H. Reese (Ed.), Advances in child development and behavior
(Vol. 19; pp. 49-82). Orlando, FL: Academic Press.
Marschark, M., & Spencer, P E. (2003). What we know, what we don't know, and
what we should know. In M. Marschark & P E. Spencer (Eds.), Oxford handbook
of deaf studies, language, and education. New York: Oxford University Press.
13. METAPHORS IN SIGN LANGUAGE 333

Marschark, M., & West, S. A. (1985). Creative language abilities of deaf children.
Journal of Speech and Hearing Research, 28, 73-78.
Marschark, M., West, S. A., Nail, L., & Everhart, V. (1986). Development of cre­
ative language devices in signed and oral production. Journal of Experimental
Child Psychology, 41 534-550.
McEvoy, C., Marschark, M., & Nelson, D. L. (1999). Comparing the mental lexi­
cons of deaf and hearing individuals. Journal of Educational Psychology, 91, 1—9.
McNeill, D. (1992). Hand and mind. Chicago: University of Chicago Press.
Myklebust, H. E. (1964). The psychology of deafness (2nd ed.). New York: Grune &
Stratton.
O'Brien, J. (1999). Metaphoricity in the signs of American Sign Language. Meta­
phor and Symbolic Activity, 14, 159-178.
Ortony, A. (1993). Metaphor and thought, (2nd ed.). New York: Cambridge Univer­
sity Press.
Ottem, E. (1980). An analysis of cognitive studies with deaf subjects. American An­
nals of the Deaf, 125, 564-575.
Paivio, A. (1979). Psychological processes in the comprehension of metaphor. In A.
Ortony (Ed.), Metaphor and thought (pp. 150-171). New York: Cambridge Uni­
versity Press.
Paul, P V. (2003). Processes and components of reading. In M. Marschark & P. E.
Spencer (Eds.), Handbook of deaf studies, language, and education (pp. 97-109).
New York: Oxford University Press.
Paulesu, E., Démonet, J.-F, Fazio, F., McCrory, E., Chanoine, V, Brunswick, N.,
Cappa, S. F., Cossu, G., Habib, M., Frith, C. D., & Frith, U. (2001). Dyslexia: Cul­
tural diversity and biological unity. Science, 291, 2165-2167.
Payne, J.-A., &Quigley, S. (1987). Hearing-impaired children's comprehension of
verb-particle combinations. Volta Review, 89, 133-143.
Petrie, H. (1979). Metaphor and learning. In A. Ortony (Ed.), Metaphor and thought
(pp. 438–461). New York: Cambridge University Press.
Piaget, J. (1974). The language and thought of the child. New York: New American Li­
brary. (Original work published 1923)
Pizzuto, E., & Volterra, V. (2000). Iconicity and transparency in sign languages: A
cross-linguistic cross-cultural view. In K. Emmorey & H. Lane (Eds.), The signs of
language revisited: An anthology to honor Ursula Bellugi and Edward Klima (pp.
261-286). Mahwah, NJ: Lawrence Erlbaum Associates.
Rittenhouse, R. K., & Kenyon, P. L. (1991). Conservation and metaphor acquisition
in hearing-impaired children. American Annals of the Deaf, 136, 313-320.
Rittenhouse, R. K., Morreau, L. E., & Iran-Nejad, A. (1981). Metaphor and conser­
vation in deaf and hard-of-hearing children. American Annals of the Deaf, 126,
450–453.
Sawa, T. (1999). Metaphor interpretation by students with hearing impairments.
Japanese Journal of Special Education, 37, 59-69.
Sawa.T., &Yoshino,T. (1995). Relation between metaphor comprehension and de­
velopment of a linguistic framework in students with hearing impairments. Japa­
nese Journal of Special Education, 33, 21-29.
334 MARSCHARK

Siple, P. (Ed.). (1978). Understanding language through sign language research. New
York: Academic Press.
Stokoe, W. C. (1960). Sign language structure: An outline of the visual communications
systems (Studies in Linguistics, Occasional Papers 8). Buffalo, NY: University of
Buffalo.
Stokoe, W. C., Casterline, D. C., & Croneberg, C. G. (1965). A dictionary of American
Sign Language on linguistic principles. Washington, DC: Gallaudet College Press.
Stokoe, W. C., & Marschark, M. (1999). Signs, gestures, and signs. In L. Messing &
R. Campbell (Eds.), Gesture, speech, and sign (pp. 161-182). Oxford, England:
Oxford University Press.
Taub, S. F. (200la). Complex superposition of metaphors in an ASL poem. In V
Dively, M. Metzger, S. Taub, & A. M. Baer (Eds.), Signed languages: Discoveries
from international research (pp. 197-223). Washington, DC: Gallaudet Univer­
sity Press.
Taub, S. F. (2001b). Language from the body: Iconicity and metaphor in American Sign
Language. New York: Cambridge University Press.
Tourangeau, R., & Sternberg, R. (1981). Aptness in metaphor. Cognitive Psychology,
13, 27-55.
Traxler, C. B. (2000). Measuring up to performance standards in reading and mathe­
matics: Achievement of selected deaf and hard-of-hearing students in the na­
tional norming of the 9th Edition Stanford Achievement Test. Journal of Deaf
Studies and Deaf Education, 5, 337-348.
Trick, L., & Katz, A. N. (1986). The domain interaction approach to metaphor pro­
cessing: Relating individual differences and metaphor characteristics. Metaphor
and Symbolic Activity, I, 185-213.
Turner, M. (1996). The literary mind. New York: Oxford University Press.
Wilbur, R. B. (1987). American Sign Language: Linguistic and applied dimensions.
Boston: College-Hill Press.
Wilcox, P. P. (2000). Metaphor in American Sign Language. Washington, DC:
Gallaudet University Press.
 Author Index

A Banaji, M. R., 154, 155, 174, 175, 262,

263, 264, 265, 266, 268, 272,
Ackerman, B. P., 219,229 275, 276, 277, 278
Akhtar.N., 108, 130 Bar'el, Z., 237, 255
Albertini, J. A., 312, 323, 325, 332 Bargh, J. A, 263, 264, 265, 276, 278
Albrecht, J. E., 23, 41 Barnard, K., 15, 17
Alcott, L. M, 234, 255 Barnden, J. A., 12, 18
Allen, M., 156, 174 Barr, D. J., 14, 17, 23, 25, 26, 27, 31, 34,
American Civil Liberties Union: Freedom 39,41,87,97
Network, 261,277 Barsalou, L. W., 244, 255
Anderson, A., 25, 32, 40 Barton, G., 16, 17
Anderson, C. A., 212, 232 Bartsch, R., 245, 255
Anderson, J. R., 263, 277 Bates, E. A., 276, 279
Anderson, K. J., 157, 158, 170, 174 Bavelas, J. B., 80, 94
Andrusiak, P., 80, 97 Beattie, A. E., 238, 258
Anes, M., 172, 175 Beatty, M.J., 221, 231
Antos, S., 76, 97 Beaty, E, 147, 150
Arcuri, L., 277, 280 Bechofer,R., 324, 325, 331
Ariel, M., 28, 39 Beinkowski, M., 270, 280
Armbruster, T., 246, 256 Belk.S.S., 154, 155, 156, 171,176
Aronson, E., 143, 150, 151 Bell, N., 89, 95, 114, 121, 129
Asai, M., 92, 98 Bellugi, U., 311, 317, 330, 331
Ascani.K., 141, 150 Bement, L. , 322, 332
Ashley, A., 78, 96 Berlyne, D., 149, 150
Austin-Smith, B., 133, 150 Berry,]., 291, 297, 303
Avila, L. X., 276, 279 Bertus, E. L, 161, 169, 170, 175
Besson, M., 195, 206
B Birch, S. L, 24, 41
Black, A., 80, 94
Bakeman, R., 160, 174 Blackwell, E, 311, 325, 330
Baker, C, 155, 175 Blair, I. V, 263, 264, 265, 266, 268, 272,
Balin,J. A., 23, 87,41, 97 275, 276, 278

335
336 AUTHOR INDEX

Blasko, D. G., 195, 204, 274, 275, 278 Chen, M., 263, 264, 276, 278
Bloom, P., 108, 127 Cheng, L., 285, 303
Bloom, P. A., 243, 256 Childres, D. G., 243, 256
Bloomfield, L., x, xi Choi, S., 92, 97
Blumentritt, T. L., 262, 264, 271, 273, 277, Chomsky, N., x, xi
279 Chopin, K., 237, 255
Bly, B., 23, 28, 87, 97, 41 Chovil, N., 80, 94
Boatner.M.T.,311, 325, 331 Christensen, A., 156, 174
Boca, S., 154, 176 Church, M. B., 114, 130, 210, 217, 231
Bodenhausen, G. V, 262, 275, 278, 280 Church, R. B., 34, 40
Bontempo, R., 92, 98 Cialdini, R. B., 139, 141, 150
Bookin, H., 76, 95 Clark, D., 324, 332
Bortfeld, H., 56, 57, 58, 59, 69 Clark, E. V, 48, 68
Bourg, T., 173, 1 74 Clark, H. H., x, xi, 22, 23, 25, 26, 29, 31,
Bowers, J.W., 156, 174 33,40,44,45,46,47,48,51,60,
BoyesBraem, E.,314, 331 61, 62, 65, 66, 67, 68, 69, 70, 75,
Brauner, J. S., 23, 41 80,87,95,165,174,186,204,
Brennan, M., 312, 314, 315, 331 217,226,229,240,241,246,
Brennan, S. E., 25, 31,33, 39, 40 252, 255
Brewer, W. F., 245, 246, 255 Clark, V. E., 252, 255
British National Corpus, 252, 255 Cohen, C. E., 262, 278
Brodsky, S. L., 154, 176 Colombo, L., 272, 281
Brody.L. R., 156, 159, 171, 174 Colston, H. L, 5, 12, 13, 15, 17, 18, 104,
Brown, M., 134, 151, 209, 217, 226, 231 105, 109, 110, 111,112, 114,
Brown, E, 80, 89, 92, 94, 159, 174 116, 117, 118, 119,120, 121,
Brown, R., 31, 39, 80, 87, 95 122, 123, 124, 127, 128, 129,
Brunswick, N., 317, 333 134, 138, 144, 145, 146, 150,
Burnstein, E., 234, 240, 243, 250, 257 150, 151, 187,188, 191,204,
Burrows, L., 263, 264, 278 209, 210, 217, 226, 229, 246,
Buttrick, S., 23, 40 249, 255
Byrnes, J. P., 114, 121, 127 Conley.J. E., 311, 325, 331
Connine C. M., 76, 98, 195, 204, 274, 275,
C 278
Convertino, C., 313, 324, 332
Cacciari, C., 76, 95, 269, 280, 286, 303, Conway, J., 156, 176
323,331 Coolen, R., 55, 69
Cacciopo, J. T., 144, 151 Cooper, R. E., 114, 121, 127
Caffi, C., 252, 255 Coppenrath, L, 111, 129, 189, 206
Cappa,S.F.,317, 333 Corneille, O., 276, 278
Capps, L., 220, 231 Cossu,G.,317, 333
Carlson, K., 114, 121, 127 Costello, M., 67, 68, 69
Carlson, T. B., 23, 29,40 Grain, A. L., 143, 151
Carpenter, P. A., 199, 205, 246, 257 Croneberg.C. G.,311, 334
Carson, M., 155, 175 Cruse, D. A., 31, 40
Carston, R., 285, 288, 303 Cunningham, A. E., 163, 176
Carter, S., 238, 258 Cunningham, J. D., 156, 175
Casterline, D. C, 311, 334
Chafe, W. L., 28, 40 D
Chaiken, S., 265, 278
Chanoine, V, 317, 333 Daly, J. A., 221, 231
Chapman, S., 285, 303, 304 Darley, J. M., 262, 278
Chase, W. G., 240, 246, 255 Dascal, M., 270, 278
AUTHOR INDEX 337

Dasen, P., 291, 297, 303 Fay, D, 246, 256

Davis-Stitt, C., 264, 280 Fazio, F., 317, 333
de Cornulier, B., 246, 255 Fein, O., 109, 129, 194, 205, 237, 239, 240,
Demonet, J. F., 317, 333 243, 255, 256, 257
Dempsey, P., 328, 332 Feldman Barrett, L., 155, 156, 175
Dennis, P., 51, 69 Ferreira, E, 172, 175
DeSchepper, B. G, 275, 280 Ferretti, T., 110, 111, 130, 196, 197, 198,
Devine, P., 264, 265, 268, 272, 275, 276, 201, 205, 206, 213, 214, 215,
278 217, 220, 225, 228, 231
Dews, S., 109, 114, 128, 129, 130, 134, 151, Festinger, L., 142, 151
210, 218, 223, 230, 244, 255 Fiedler, K, 246, 256
Dijksterhuis, A. P., 263, 264, 275, 276, 278 Fillmore, C. ]., 77, 97
Dijkstra, K., 172, 176 Fischgrund, J., 311, 325, 330
Dindia, K., 156, 174 Fischler, I., 243, 256
Doolittle, S., 109, 129 Fisk, R., 254, 256
Dourabont, F., 160, 175 Fiske, S., 263, 279
DovidicvJ. F., 263, 264, 265, 276, 279 Fitch, K. L., 91, 95
Downing, P. A., 48, 69 Fodor, J., 240, 256, 288, 303
Drakich,J., 165, 175 Fong, G. T., 238, 257
Dresher, E., 146, 151 Fox Tree, J. E., 26, 40
Drew, P., 204, 205 Franklin, L., 285, 303
Drozd, B., 92, 96 Eraser, J. B., 325, 331
Du Bois, W. J., 234, 256 Fried, C., 143, 150, 151
Ducrot, O., 245, 256 Friedman, L., 311, 331
Duffy, S. A., 243, 257 Friesan, W. V, 154, 174
Dumais, S. T., 265, 280 Frisson, S., 262, 267, 268, 270, 271, 272,
Duncanson, W. T., 156, 174 279
Dunham, P. J., 14, 18, 218, 219, 224, 230 Frith, C.D., 317,333
Frith, U, 317, 333
E Fruchter, A., 325, 331
Fryer, R., 163, 175
Easterbrooks, S. R., 325, 331 Fussell, S. R., 87, 97, 155, 158, 159, 170,
Eckman, P., 154, 174 175
Ekert, P., 189, 205
Eldar, A., 251, 256 G
Ellsworth, E, 154, 174
Emmorey, K., 312, 315, 318, 321, 326, 331 Gagne, C. L., 55, 69
Engberg-Federsen, E., 318, 319, 331 Gallois,C., 171, 175
Engen, E., 311, 325, 330 Garcia-Falconi, R., 154, 155, 156, 171, 176
Eptey, R, 27, 40 Gardner, H., 105, 130, 324, 325, 331
Erber, R., 238, 258 Garnham, A., 201, 205
Ervin-Tripp, S., 89, 95 Garrod, S., 25, 32, 40
Evanovich, J., 44, 49, 69 Gates, J.E., 311,325, 331
Evans, R, 263, 264, 265, 276, 279 Gelfand, M. J., 92, 97
Everhart, V. S., 313, 322, 327, 328, 330, Gelman, S. A., 218, 230
331, 332, 333 Gerrig, R. J., 23, 40, 44, 45, 47, 48, 49, 53,
Eyssell, K. M., 155, 156, 175 54, 56, 57, 58, 59, 60, 61, 63, 64,
69, 70, 209, 217, 226, 229, 230,
F 269, 279
Gibbs, R. W., Jr., x, xi, 5, 13, 14, 17, 18, 48,
Fainsilber, L., 156, 157, 158, 159, 170, 173, 52, 55, 60, 69, 73, 76, 78, 95,
175 105, 109, 111,114,116,128,
338 AUTHOR INDEX

129, J30, 135, 136, 144, 145, Hastie, R., 80, 95

149, 150, 151, 156, 157, 158, Haverkate, H., 217, 230
159, 173, 175, 187, 188, 191, Hayman, G. D., 218, 230
195, 204, 205, 209, 210, 215, Healy, A. F., 269, 279
220, 229, 230, 231, 262, 266, Heavey, C. L, 156, 174
267, 268, 269, 270, 271, 277, Hedberg, N., 28, 40
279, 286, 287, 288, 290, 303, Heerey, E. A., 220, 231
323,331 Hegel, G. W. F, 244, 256
Gilbert, D., 263, 279 Henly, A. S., 23,41
Gildea, P., 76, 77, 95, 269, 279 Heredia, R. R., 262, 264, 266, 269, 270,
Oilman, A., 80, 95 271, 273, 276, 277, 279, 281
Gilovich, T, 27, 40 Hernandez, A. E., 276, 279
Ginsburg, G. P., 87, 97 Hernandez-Sanchez, J. E., 154, 155, 156,
Giora, R., 109, 129, 187, 193, 194, 195, 171, 176
203, 205, 237, 239, 240, 243, Higgins, E. X, 264, 279
254, 255, 256, 257 Highley, A., 285, 304
Givon, T., 239, 256 Hingson, R., 114, 121, 129
Glotzer, L., 160, 175 Hixon, J., 263, 279
Glucksberg, S., x, xi, 95, 76, 77, 134, 151, Holliday,H., 155, 175
183, 185, 187, 188, 194, 205, Holloway, R. L., 114, 130, 144, 151
206, 209, 215, 217, 225, 226, Holt, E., 204, 205
231, 268, 269, 279, 280, 287, Holtgraves, T., 73, 75, 78, 79, 80, 81, 82,
303, 313, 320, 323, 331 83, 84, 85, 86, 87, 88, 89, 90, 91,
Goffman, E., 79, 95 92,96,111,129,190,191,205,
Goldvarg, Y., 209, 230, 287, 303 221, 228, 230
Goossens, L., 322, 331 Horn, L. R., 236, 240, 244, 245, 246, 249,
Gottman, J. M., 160, 174 252, 256
Govender, R., 265, 278 Hornstein, N., 146, 151
Graesser, A. C., 80, 95 Horton, W. S., 14, 18, 23, 29, 40, 49, 69,
Greene, S. B., 23, 40 87, 96
Greenwald, A. G., 262, 265, 268, 276, 277 Howland, J., 114, 121, 129
Grice, H. P., 22, 31, 40, 52, 69, 73, 74, 76, Hunt, M., 218, 230
77, 78, 79, 87, 89, 92, 95, 268, Hunt, R. R., 324, 332
279, 285, 303 Hyde, J. C., 60, 70
Grisham, J. A., 156, 175 Hyde.J. S., 158, 175
Gross, P. H., 262, 278
Gundel, J. K., 28, 40 I
Gundykunst, W., 92, 95
Inman, P.R., 325, 331
H Iran-Negad, A., 325, 327, 331, 333
Ivanko, S. L, 218, 220, 221, 222, 223, 227,
Habib, M., 317, 333 230
Hala.S., 218, 220,230 hard, C.E., 154, 175
Hall, J. A., 156, 159, 171, 174 Izett, C, 120, 128
Halldorson, M., 80, 97
Hancock, J. T. 14, , 218, 219, 224, 230, 18 J
Hara, K., 92, 95
Hardin, C. D., 262, 263, 264, 265, 268, Jackson, V. R., 229, 230
272, 276, 277, 278 James, D., 165, 175
Harlan, J., 163, 175 Jefferson, G., 235, 256
Harris, J.T., 218, 220, 230 Jefferson, G., 25, 41
Hasson, U, 234, 240, 243, 250, 256 Jen, G., 44, 45, 52, 69
AUTHOR INDEX 339

Jespersen, O., 239, 256 Kirch, D., 32, 41

Job, R., 272, 281 Klein, N., 235, 257
Johnson, B. K., 161, 169, 170, 175 Klima, E., 311, 317, 330, 331
Johnson, J. S., 156, 158, 176 Koomen, W., 264, 280
Johnson, M., 213, 231, 319, 321, 329, 332 Krauss, R. M., 87, 97
Johnson, M. L., 163, 175 Kreuz, R. J., 80, 97, 111, 114, 116, 129,
Johnson-Laird, P.N., 61, 70, 246, 253, 256 130, 134, 151, 161,169,170,
Jones, C. Q., 124, 128 173, 175, 176, 188, 189, 206,
Jones, C. R., 264, 279 209, 210, 215, 217, 225, 231
Jorgensen, J., 1ll, 129, 189, 205, 209, 221, Kring, A.M., 220, 231
230, 231 Kubrick, S., 163, 175
Jucker, A. H, 85, 96 Kumon-Nakamura, S., 134, 151, 209, 217,
Jungen, S., 218, 220, 230 226,231
Just, M., 199, 205 Kunda, Z., 262, 275, 279
Just, M. A., 234, 240, 243, 246, 250, 253,
257 L
K Labov, W., 236, 238, 257
Labute, N., 49, 69
Kahneman, D., 27, 41 LaFrance, M., 154, 155, 174, 175
Kaplan, J., 114, 128, 210, 218, 230 Lakoff, G., 213, 231, 288, 290, 304, 319,
Karsch, A. S., 179, 175 321, 329, 332
Kashima, Y, 92, 97 Lakoff, R., 85, 97
Kassler, M., 1ll, 129, 189,206 Lalljee, M., 87, 97
Katz, A., ix, x, xi, 89, 97, 105, 110, 111, Lamb, R., 87, 97
114, 129, 130,134,135,151, Lampert, M., 89, 95, 220, 222, 231
158, 176, 183, 188, 189, 191, Lang, H.G., 312, 323, 325, 332
194, 196, 197, 198, 199, 201, Lea, B. R., 234, 257
204, 205, 206, 209, 210, 212, Lea, R.B., 23, 41
213, 214, 215, 216, 217, 220, Leaper, C., 155, 157, 158, 170, 174, 175
225, 228, 231, 232, 286, 303, Lear, J. C., 80, 97
323, 324, 331, 334 Lee, C., ix, xi, 114, 129, 130, 134, 135, 151,
Kaufer, D., 132, 133, 149, 151 188, 189, 191, 206, 209, 231
Kaup, B., 234, 240, 253, 254, 257 Lee, M.G. ,12, 13, 18
Kay, E, 77, 97, 184, 206 Lee, S., 1ll, 112, 114, 116, 118, 119, 121,
Keenan, E. O., 91, 97 122, 128, 191,204
Kellas, G., 203, 206 Leekham.S., 219, 232
Keller, S. B., 114, 116, 128, 138, 150 Lefebvre, L. M., 156, 175
Kelley, H. H., 156, 175 LeggittJ., 116, 130, 135, 149, 151, 156,
Kelly, S. D., 34, 40 175, 210, 231
Keltner, D., 220, 231 Lehrer, A., 159, 172, 176
Kemper, S., 1ll, 129, 190, 206 Leiman, J., 270, 280
Kenyon, P. L., 325, 333 Lentz, E. M., 328, 329, 332
Kerker, R. M., 238, 258 LeEoutre, D., 322, 332
Keysar, B., 14, 18, 23, 26, 27, 28, 29, 31, Lester, D., 160, 175
33, 40, 41, 60, 70, 76, 77, 87, 96, Levine, B. A., 66, 70
97, 268, 269, 279 Levinson, S., 77, 80, 87, 89, 92, 94, 97,
Kiely, J., 159,172,176 159, 174, 285, 304
Kierkegaard, S., 185, 206 Levy, B., 276, 280
Kim, M. S., 92, 95 Levy, G., 236, 257
Kim, U., 92, 97 Lewis, D., 36, 41
Kintsch, W., 159, 175, 253, 258 Lian, M.G. J., 325, 331
34O AUTHOR INDEX

Lichtenstein, E. H., 245, 246, 255 McRae, K., 194, 196, 206
Liddell, S., 312, 321, 332 Mead, G. H., 87, 97
Lim.K., 218, 230 Mesquita, B., 154, 176
Link, K.E., 111, 130 Metts, S.M., 156, 174
Linn.M.C, 158,175 Michaels, L., 48, 70
Linor, E., 235, 257 Mick, D., 149, 151
Lippman, W., 261,280 Middleton, E. L., 55, 70
Locke, V., 264, 280 Mieder, W., 198, 206
Logan, G. D., 265, 280 Miller, D. C., 114, 121, 127
Long, D., 114,130, 210, 217, 231 Miller, G. A., 209, 231
Love, B. C., 55, 70 Miller, R. S., 154, 155, 156, 171, 176
Lucca, N., 92, 98 Milne, A. B., 275, 280
Luria, A. R., 313, 332 Mio,J. S., 158, 159, 171, 173, 176
Lusch.V., 118, 119, 121, 122, 128 Moon, R., 187, 198, 206
Lynch, K., 34, 40 Morreau, L. E., 325, 333
Lyons, J., 245, 257 Moss, M. M., 155, 158, 159, 170, 175
Moxey, L. M., 239, 257
M Mullet, J., 80, 94
Mulligan, E. J., 234, 257
Maass, A., 277, 280 Murphy, G., 23, 40, 55, 57, 70, 266, 268,
Mac Cormac, E. R., 310, 332 269, 270, 280, 288, 304
MacDonald, M. C., 234, 240, 243, 250, Myers, J. L., 24, 41
253, 257 Myers, S., 155, 175
MacLeod, C., 264, 280 Myklebust, H.E., 311, 325, 333
Macrae, C. N., 275, 280
Magliano, J. P., 172, 176 N
Maglio, P., 32, 41
Makin,V. S.,32, 41 Nagel, H. N., 66, 70
Mandel, M., 316, 317,332 Nail, L, 311, 322, 327, 330, 332, 333
Mangione, T. W., 114, 121, 129 Nascimento, S. B., 156, 158, 159, 173, 175
Markman, A. B., 32, 41 Naumann, U., 246, 256
Marschark, M., 310, 311, 312, 313, 314, Neely, ]. H., 265, 280
315, 319, 322, 323, 324, 325, Nelson, D. L., 324, 333
327, 328, 329, 330, 331, 332, Neuberg, S., 263, 279
333, 334 Neumann, E., 275, 280
Marshall, C. R., 22, 33, 40, 48, 69 Newsome, M., 287, 303
Martin, M. A., 221, 231 Nippold, M., 285, 304
Marvin, R, 160, 175 Nishida, T., 92, 95
Mason, R. A., 23, 41 Roller, P., 156,176
Masteller, A.,313, 324, 332 Notarius, C. L, 156, 158, 176
Mayo, R., 234, 240, 243, 250, 257 Nunez, R., 288, 304
McCrory, E., 317, 333
McCroskey, J. C, 221, 231 O
McEvoy, C., 313, 324, 332, 333
McGlone, M. S., 269, 280 O'Brien, J., 109, 114, 117, 128, 129, 138,
McGovern, A., 218, 230 150, 150, 151, 209, 210, 229,
Mclntire, D., 285, 303 317, 333
McKoon, G., 23, 40 Oakhill, J., 201, 205
McLain, A., 1ll, 129, 189, 206 Okamoto, S., 1ll, 130
McMullen, L., 156, 176 Olineck, K. M., 105, 111, 130, 211, 214,
McNeill, D., 26, 41, 314, 320, 333 215, 217, 221, 222, 223, 225,
McQuarrie, E., 149, 151 226, 227, 230, 231
AUTHOR INDEX 341

Onifer, W., 63, 64, 70 Ratcliff, R., 23, 40

Onishi, K. A., 266, 268, 269, 270, 280 Rayner, K., 243, 257
Ortony, A., 76, 97, 153, 156, 157, 158, Reynolds, D., 201, 205
159, 160, 170, 173, 175, 176, Reynolds, R., 76, 97
323, 325, 327, 333 Rholes, W. S., 264, 279
Osman, A., 203, 206 Richards, M., 163, 175
Osterhout, L, 269, 272, 281 Rime, B., 154, 176
Ottem, E., 325, 333 Rimon, M., 244, 258
Rittenhouse, R. K., 325, 327, 331, 333
P Roberts, R. M., 80, 97, 114, 116, 130, 134,
151,161,169,170,173,175,
Pacht, J. M., 243, 257 176, 210, 231
Paek, T.S., 23, 41, 87, 97 Robichon, F. H., 195, 206
Paivio, A., 310, 333 Robin, L., 155, 156, 175
Paradis, C., 249, 257 Rocus, S. E., 243, 256
Paul, R V, 325, 333 Rommetviet, R., 87, 97
Paul, S., 203, 206 Rosaldo, M. Z., 91, 97
Paulesu, E.,317, 333 Rosenblatt, E., 218, 230
Payne, J. A., 325, 333 Rosenhan, D. L., 238, 257
Peleg, O., 240, 257 Ross,H., 160, 176
Perry, N. W., 243, 256 Rothbart, M., 264, 280
Petrie, H., 325, 333 Rothman, A. J., 262, 264, 265, 268, 276,
Petty, R. E., 144, 151 278
Pexman, P, 89, 97, 105, 110, 111, 129, 130, Russell, A. W., 87, 97
191, 198, 199, 201, 206, 209,
211, 212, 213, 214, 215, 216, S
217, 218, 220, 221, 222, 223,
225, 226, 227, 230, 231 Sacks, H., 25, 41, 85, 97
Pezdek.K., 159, 172, 176 Sanders, R. E., 91, 95
Philippot, P, 154, 176 Sanford, A. J., 239, 257
Piaget, J., 311, 313, 324, 333 Saussure, E, de., ix, xi
Piasecka, I., 1ll, 129, 191, 201, 206, 220, 231 Sawa, T, 326, 333
Pickering, M. J., 262, 267, 268, 270, 271, Schafer, W. D., 114, 121, 127
272, 279 Schallert, D., 76, 97
Pietromonaco, P. R., 155, 156, 175, 264, Schelgloff, E., 25, 41, 85, 97
278 Schneider, D., 238, 258
Pizzuto, E.,317, 333 Schober, M., 62, 70, 87, 97
Poli, J., 195, 206 Schreuder, R., 23, 40, 55, 69
Poortinga, Y., 291, 297, 303 Schul, Y., 234, 240, 243, 250, 257
Pratto, F., 265, 278 Schunk, D., 75, 80, 95
Prince, J., 285, 304 Schwartz, T., 109, 129, 194, 205, 237, 256
Purdy, K., 218, 219, 230 Schwarz, I., 285, 304
Pynte, J., 195, 206 Schwoebel, J., 109, 114, 130
Searle, J., 52, 70, 194, 206, 268, 280
Q Segall, M., 291, 297, 303
Seidenberg, M., 270, 280
Qualter, A., 218, 230 Shapiro, L. P., 66, 67, 70
Quigley, S., 325, 333 Shavit, A., 251, 252, 257
Sherman-Williams, B., 262, 279
R Shift-in, R. M., 265, 280
Shimanoff, S. B., 154, 155, 158, 166, 171,
Rapp, D. N., 44, 45, 63, 64, 70 172, 176
342 AUTHOR INDEX

Shobe, A., 285, 303 Taylor, M., 227, 232

Shoben, E. J., 55, 69 Thissen, D., 1ll, 129, 190, 206
Simon, H. A., 26, 41 Thompson, ]., 285, 303
Simon, S., 159, 172, 176 Thompson, W. C., 238, 257
Simpson, G. B., 63, 70, 268, 272, 273, 280 Ting'Toomey, S., 92, 98
Singer, M., 80, 95, 97 Titone, D. A., 76, 98
Sink, C. R., 156, 175 Tomasello, M., 108, 130
Siple, P., 311, 334 Toplak, M., 1ll , 114, 129, 130, 288, 189,
Slugoski, B., 87, 97, 212, 217, 231 191, 201, 204, 206, 210, 217,
Slusher, M. P., 212, 232 220, 231, 232
Smarsh, B., 218, 230 Tottie, G., 238, 258
Snell, W. E., Jr., 154, 155, 156, 171, 176 Tourangeau, R., 324, 334
Socall, D., 89, 96 Trabasso, T, 80, 95
Spencer, P.E., 312, 332 Traxler,C. B.,311, 323, 334
Sperber, D., 89, 98, 209, 225, 231, 232, Triandis, H., 91, 92, 98
236, 257, 285, 286, 287, 288, Trick, L, 158, 176, 324, 334
289, 291, 304 Tridgell, J. M., 246, 256
Spivey-Knowlton, M., 194, 196, 206, 270, 281 Turnbull.W, 212, 217, 231
Srinivas, K., 109, 114, 130 Turner, M, x, xi, 286, 288, 290, 303, 304,
Sriram, N., 264, 280 319, 323, 331, 332, 334
Srull, T. K., 89, 96 Turner, N., 194, 206
Stangor, C, 275, 280 Tversky, A., 27, 41
Stanley, J., 285, 304 Tyler, R. B., 263, 264, 265, 276, 279
Stanovich, K. E., 163, 176
Stapel, D. A., 264, 280 U
Stark, R., 160, 176
Steen, G., 199, 206 Uhden, L., 285, 304
Stein, L. B., 154, 176 Ulatowska, H., 304, 285, 303
Stephan, W. G., 263, 280 Utsumi, A., 226, 228, 232
Sternberg, R., 324, 334
Stewart, A., 141, 151 V
Stewart, M. T., 281, 266, 269, 270, 271,
277, 279 Van der Plight, J., 264, 280
Stoeckert, J., 159, 172, 176 Van Dijk, T. A., 253, 258, 271, 281
Stokoe, W. C., 311, 314, 319, 329, 334 Van]aarsveld, H.J., 55, 69
Stone, J., 143, 150, 151 Van Knippenberg, A. D., 263, 264, 275,
Strage, M., 89, 95 278
Streisand, B., 179, 175 Vennemann, T, 245, 255
Stroop, J. R., 253, 257 Villareal, M., 92, 98
Swift, J., 137, 151 Volterra,V.,317, 333
Swinney, D., 63, 64, 70, 268, 269, 270, 272, Vu, H., 203, 206
281
Szabo, Z., 285, 304 W

T Wagner, L., 146, 151

Walker, I., 264, 280
Tabossi, R, 63, 70, 269, 272, 273, 281 Walker, N., 133, 151
Tanaka, J. W., 227, 232 Walther, E., 246, 256
Tanenhaus, M., 194, 196, 206, 270, 280, 281 Wason, P. C., 246, 258
Tannen, D., 173, 176, 222, 232 Waugh, E., 147, 148, 151
Taub, S. F., 312, 314, 315, 316, 320, 321, Weatherall, A., 189, 206
329, 334 Wegner, D., 238, 258
AUTHOR INDEX 343

Welch, J., 43, 70 Wisniewski, E. J., 55, 56, 70

Wenzlaff, R., 238, 258 Wolf, D., 324, 325, 331
West, R. F., 163, 176
West, S. A., 314, 315, 322, 327, 329, 330, Y
333
Whaley, B., 114, 130, 144, 145, 151
Yablon, G., 156, 175
White, T., 238, 258
Yamaguchi, S., 92, 97
Whitney, P., 158, 159, 171, 173, 176
Yang, J. N., 75, 80, 89, 96
Wiederman, M.W., 114, 130
Yngve, V, ix, xi
Wilbur, R.B., 321, 325, 334 Yoon, Y. C., 92, 95
Wilcox, P.P., 310, 312, 314, 321, 322, 328, Yoshino, T., 326, 333
329, 334
Young, R. C., 220, 231
Wilkes-Gibbs, D., 25, 40
Yuki, M., 92, 97
Williams-Whitney, D., 158, 159, 171, 173, 176
Wilson, D., 89, 98, 225, 232, 236, 257, 285,
286, 287, 289, 304 Z
Winner, E., 105, 109, 114, 128, 129, 130,
134, 151, 187, 188, 206, 210, 218, Zacharski, R., 28, 40
219, 223, 230, 232, 244, 255, 287, Zarcadoolas, C., 311, 325, 330
288, 304, 324, 325, 331 Zardon, E, 63, 70
Winograd, T., 146, 151 Zelinski, E. M., 60, 70
Winslow, M. P., 143, 151 Zizek, S., 254, 258
Winter, Y., 244, 258 Zwaan,R.A., 172, 176, 253, 258
This page intentionally left blank
 Subject Index

A Collectivism versus individualism, 91

Common ground, 2, 5, 10, 22-23, 36, 38,

Accent, ix
44, 45, 51, 156, 185

Acting, 3, 226
community membership, 45

Adjacency pairs, 85
personal, 48

Aggression, 116, 134

Compound nouns, 56

Allusional pretense, 226

Compromise methodologies, 12

Ambiguity, 63
Conceptual metaphor, 10, 288, 290, 319

Analogies, 4
Confederates & victims, 132

Anaphoric time line, 318

Conservation, 105

Anchoring and adjustment, 27, 28

Constraint satisfaction, 10, 193, 195, 196,

Arabs, 261, 267, 271, 273

203

Asyndeton, 4, 117
Contextual
Attention, 28, 29
content versus structure, 101

Attitude display, 3
support extent, 4, 187

Attribution, 121
support type, 4, 188

Authenticity, 11, 188

Contextutal expressions, 2, 4, 5

Automaticity versus controlled processing, Contrast principle, 138

265, 275
Contrast, 3, 5, 10, 104, 138, 140, 146, 150

Conversational analysis, 12

B Conversational indirectness scale, 221, 92

Cultural background, ix, 7, 73, 286, 291

Bonding, 136

D
c
Deaf adult figurative production, 328

Calendar plane, 319

Development, 218, 311, 323, 324

Central versus peripheral persuasion, 144

deaf comprehension 324

Chaining, 3, 136
deaf production, 326

Cognitive dissonance, 3, 143

Discourse analysis, 12

Cognitive load, 6, 123

Discourse goal match, 113, 119, 121

345
346 SUBJECT INDEX

Discourse markers, 183 L

E Language use environment, 24-25

Lexical innovation, 45
Echoic mention, 225, 236 Literal period, 327
Egocentrism, 3, 23-24, 37, 87 Litotes, 4
Emotion, 116, 121, 134, 135, 149, 154,
170, 171, 172, 221 M
perspective, 158
valence, 158 Malagasy, 91
Epistemic exchanges, 32-35, 39 Mastery, 120, 134
Eponymous phrases, 47, 63-64 Mere continuance, 3
Error recovery model, 53 Metaphor and iconicity in sign, 309, 314
Ethnicity, 6, 7, 124 Metaphor, 3, 4, 77, 156, 266, 268, 287
Expectation reality discrepancy, 137, 142, Metaphorical signs, 4, 10, 319
146 Metonymy, 3, 4, 228, 267, 269, 270
Mexicans, 267, 292
Mimicry, 3
F Muslims, 262
Mutual exclusivity assumption, 107
Face management, 79
Frozen trope, 315 N
G Negation markers, 234, 239, 254
Negation, 3, 5
Gender, 6, 10, 111, 112, 124, 153, 190, Nonliteral blends, 5, 10
201, 220, 222, 266, 267, 293 Nonliteral prevalence, 13
Geographical area of origin, 6, 297, 302 Nonliteral production, 14
Graded salience, 203 Norm of reciprocity, 3, 141
Gricean pragmatics, 10, 31, 73, 84
cooperative principle, 22, 74 o
maxims, 74
Occupation, 6, 10, 89, 110, 191, 198, 212
I
P
Idioms, 4, 77, 228
Implicatures, 3 Parsing, 65
Implicit display, 226 Path-for-shape iconicity, 316
Inclusive versus piecemeal theoretical Perspective taking, 87
approaches, 4, 10 Politeness, 75, 80, 89
Indirect replies, 87 Political background, 6, 233
Indirect requests, 4, 51, 75, 78, 89 Pragmatic functions, 114
Individual differences, 221 Prejudices, ix
Inexpressibility hypothesis, 157 Pretense, 226
Inferential complexity, 110 Priming, 3, 263
Interlocutor relationship, 6, 189, 212 Processing environment, 103
Introductory formulae, 198 Proverbs, 4, 196, 228, 283, 288
Ironic environment, 226
Irony in literature, 147 R
K Rebuttal analogies, 144
Referent negotiation, 50, 62, 107, 108
Korea, 92 Referential distance, 110
SUBJECT INDEX 347

Relevance, 10, 285 T

Religious background, 6, 10, 284, 299, 302
Response in kind, 3, 141 Temporal motion, 317
Retention hypothesis, 239 Terrorists, 262, 267, 271
Risk attraction, 113, 118, 121 Tinge hypothesis, 5, 210
u
s
Underdeterminedness, 22, 99
Salient meaning, 10, 194 Universality, 10, 121
Sapir-Whorf, 312
Sarcasm self-report scale, 222 V
Script adherence, 3
Sign classifiers, 315 Verbal irony, 3, 4, 5, 109, 131, 209, 237
Sign language 309 Virtual depiction, 316
Social roles, 190 Vividness hypothesis, 157
Social schemas, 3, 263 W
Social tension, 6, 123
Speaker traits, 7, 217 What is "culture," 7
Speech act theory, 78 What is "influence," 8
Standard pragmatic model, 52, 73, 76, 194 What is "social," 7
Stereotypes, ix, 3, 6, 89, 106, 107, 185, Whole object assumption, 108
191, 198, 261, 266, 277 Word learning, 107
Suppression hypothesis, 10, 234 WXDY, 4, 77