y.

N
[DTA ARO Ee oC
FUNCTIONALIST
- APPROACH

Se
LANGUAGE
& & © & &

Nancy Budwig
A DEVELOPMENTAL-—
FUNCTIONALIST
APPROACH TO
CHILD LANGUAGE
 A DEVELOPMENTAL-—
FUNCTIONALIST
APPROACH TO
CHILD LANGUAGE

Nancy Budwig
Clark University

LAWRENCE ERLBAUM ASSOCIATES, PUBLISHERS

1995 Mahwah, New Jersey
AlFW8so7

Copyright © 1995, by Lawrence Erlbaum Associates, Inc.

All rights reserved. No part of the book may be reproduced in
any form, by photostat, microform, retrieval system, or any other
means, without the prior written permission of the publisher. |

Lawrence Erlbaum Associates, Inc., Publishers

10 Industrial Avenue
Mahwah, New Jersey 07430

Library of Congress Cataloging-in-Publicatio

n Data

Budwig, Nancy.
A developmental-functionalist approach to child
language / Nancy
Budwig.
Dems
Includes bibliographical references and
indexes.
ISBN 0-8058-0520-6 (alk. paper)
1. Language acquisition. 2. Child development. 3. Functionalism
(Linguistics).
I. Title.
P118.B76 1995
40T 93
95-680
CIP

Books published by Lawrence Erlbaum

Associates are printed on acid-free paper
and their bindin gs are chosen for strength and durabil ’
ity.
Printed in the United States of America
1OROMSmyamGmS 4 a i
 For Michael
€
i
G

Faget

1g “

vn
, 4
“
\
oy et
nth. ey ‘
4
My
 Contents

Preface

PART I: FOUNDATIONS

1. Functional Approaches to Child Language

2. Agentivity and Control: Linguistic, Developmental

Psycholinguistic, and Developmental Psychological
Perspectives 25

Methodology 41

4. Orientation Points 50

PART II: CASE STUDIES

Jeffrey to

Grice 102

Megan 132

The Nonego-anchored Children 156

Vii
Viii CONTENTS

PART III: MECHANISMS OF DEVELOPMENT

9. The Organization of Linguistic Systems: Forms

and Functions in Development 175

10. Conclusions 198

References 210

Author Index 219

Subject Index 223

 Preface

The work reported in this volume began during my final years as a doctoral
student at University of California, Berkeley. At the time the work began, it
seemed clear to me what a functionalist position was and what it wasn’t,
and I felt convinced that as an approach, it provided an excellent account
of children’s development of language. As I moved away from Berkeley,
and as I traveled around to talk about my research, what had once seemed
so clear was becoming more fuzzy. It was at that point that I decided to
write this book, not so much as an explication of a coherent approach, but
more as a way to clarify for myself, what others had meant by functionalism,
the relative strengths of such an approach as an account for children’s
language development, and where further work needed to be done.
In reviewing what others had said about functionalist accounts of lan-
guage in general, and functionalist accounts of child language in particular,
it seemed to me that much of the explicit discussion centered around the
issue of what one takes the domain of language to be. That is, most of the
work seemed focused on describing a view of language as a nonautonomous
system. The first chapter of this book summarizes such research, reviewing
the extent to which child language researchers have picked up on trends
found in functional linguistic theorizing.
A common theme among the functionalist child language literature is that
children’s earliest use of grammatical forms is linked to the ways they interact
with the world. In chapter 2, I suggest that if we are to take this claim
seriously, we must not only focus on the ways languages link forms with
particular semantic and pragmatic meanings, but also weigh such knowledge
in light of what research in the area of social and cognitive development

1X
x PREFACE

has shown. I illustrate this with an examination of the linguistic as well as

developmental psychological literature on notions of agency and control.
This review provides evidence for the claim that languages across the world
draw linguistic distinctions based on degree of agentivity and control; and
at the same time, highlights evidence from the developmental literature
which suggests that children, around the age of beginning to first combine
words, most likely make use of more restricted notions of agency and in-
tentionality having to do primarily with self. I argue that a review such as
that offered in chapter 2, which attempts to link what we know from the
crosslinguistic study of languages, with what is known about children’s
budding conceptual and social development, is another central aspect of
many (but not all) functional approaches to child language that distinguishes
them from other approaches toward child language.
One problem noted with focusing much of the functionalist work in the
area of child language around the issue of domain is that other aspects of
functionalist approaches—such asa particular view of development—remain
implicit and unnoticed. This has led some to wonder, if a functionalist
approach is defined in terms of an approach that relates syntax or grammar
to other levels of language functioning, such as semantics or pragmatics,
then couldn’t everyone who has recently worked in the area of child lan-
guage theorizing be considered a functionalist (P. Bloom, p.c.)? Part II of
this book is an attempt to make explicit that functionalist approaches take
a radically different position on the notion of development. Part II presents
an analysis of six children’s language development. Here it is argued that
at a time before the young children refer to others, they temporarily borrow
first person pronominal forms as markers of various degrees of agentivity
and control. The central theme of these chapters, which is said to distinguish
functionalist approaches from, say, maturational approaches to child lan-
guage, is the claim that development is protracted. It is argued that children
go through a series of phases en route to adult-like systems. Thus, the point
is made that although much of the discussion of what constitutes a func-
tionalist position has explicitly centered around the issue of domain, another
way in which functionalist positions can be distinguished from other ap-
proaches is in terms of the way they view course of development (see Budwig,
1993a).
Finally, in Part III of this book we turn to a discussion of mechani
sms
of development from the standpoint of a functionalist perspective. Given
that we are only beginning to understand the nature of the course of devel-
opment for certain areas of linguistic development, I suggest that functiona
list
approaches are only in the very early phases of understanding what
leads
children to originally organize systems as they do, and what pushes
them
to give up their original form—function pairings in order to replace them
with alternative systematizations. In chapter 9, I review various
proposals
PREFACE xi

that have been offered regarding both organizational and reorganizational

processes, drawing on both conceptual and environmental accounts. Chapter
10 suggests that in order to tease apart the various contributions much further
research is needed. A specific program of research is discussed that I believe
will help us unravel what remains to be one of the central questions for
researchers interested in the development of children’s language.
In recent years there has been much confusion surrounding the issue of
what constitutes a functionalist position. It has been argued that much of
what has been criticized about functionalist approaches to child language
falls under what can be referred to as “Straw Man Functionalism” (see Bates
& MacWhinney, 1988). The aim of this book is two-fold. First, it is an attempt
to begin to untangle some of the confusion about functionalist approaches
to child language. Second, this book aims to identify issues, such as the
notion of development and mechanisms of development, which those adopt-
ing a functionalist approach to child language have only begun to consider
in any explicit detail and for which much future inquiry is required.

ACKNOWLEDGMENTS

The material covered in this volume goes well beyond the original plan of
simply reworking my 1986 dissertation into book format. The decision to
go this route came in part due to the support and interest of numerous
colleagues who have attempted to influence and challenge my thinking
along the way. It is really quite gratifying to reconsider the number of
individuals who have directly impacted on the nature of my thinking over
the last years. I was fortunate to have a supportive group of professors and
fellow students at U. C. Berkeley to discuss the original ideas with and lucky
to find a similar engaging environment at Clark University over the last years.
In the time spanning between these two locations I have had the opportunity
to test out a variety of ideas with colleagues at the Stanford Child Language
Research Forum, The Boston University Conference on Language Develop-
ment, Society for Research in Child Development, the International Congress
for the Study of Child Language, and the Max-Planck Institute for Psycho-
linguistics. While there are too many individuals to mention, many will see
the ways in which our discussions have led to revisions in the manuscript.
There has been a group of individuals who have had a more constant
role in assisting me with this project. My original dissertation committee
consisting of Susan Ervin-Tripp, John Gumperz, and Dan Slobin all went
beyond the role of mentors in shaping the original thesis. Since that time
Dan and Sue have also continued to play a major role in helping me articulate
the ideas put forward in this book and both have been a wonderful source
of inspiration. Members of the Department of Psychology at Clark University
xii PREFACE

have also provided a most supportive working environment and great tech-
nical assistance. In particular, I would like to thank my colleagues in the
developmental psychology program Bernie Kaplan, Ina Uzgiris, and Jim
Wertsch who have all impacted on my thinking about the notions of devel-
opment and semiotic systems in our joint teachings and discussions. Former
students, Jessica Everett, Pate Mahoney, Heather Quick, Andrew Rosner,
and Angela Wiley made suggestions from a student’s perspective. Colleagues
from farther away regularly have engaged in dialogues with me both in their
review of portions of manuscript and in ongoing discussions. Ruth Berman,
Melissa Bowerman, Eve Clark, Werner Deutsch, Julie Gerhardt, Elena Lieven,
Ann Peters, and Bambi Schieffelin have all been influential, often by raising
exactly the issues that I would have preferred to leave unnoticed. I am also
thankful for the help provided by the staff at Lawrence Erlbaum Associates.
Judi Amsel provided constant encouragement and friendly reminders that
the book was getting behind schedule. In the final phases of production
Sondra Guideman made sure that I attended to all the slippery and tedious
corrections.
This project could not have been carried out without the support of
various institutions. I gratefully acknowledge assistance from The National
Academy of Education Spencer Fellowship Program, The Max Planck Insti-
tute for Psycholinguistics, and The Frances L. Hiatt Endowment for Faculty
Research, Clark University.
Closer to home there has been a group of family members who provided
a most delicate balance between interest and disinterest in this project. My
parents, RoJene and Sam Budwig, have shown great enthusiasm and have
been equally accepting of each aspect of the manuscript regardless of its
theoretical flaws. They have always reminded me to create a space for the
best thinking possible, even if it meant leaning on them financially, and
for
this support I am most grateful. My children, Zachary and Jeremy, have also
been willing to engage in passionate discussions concerning language de-
velopment, but also persuaded me of the joy to be found in pursuing
other
activities. Finally, this book would never have been completed without
the
loving support of my husband, Michael Bamberg, who has provided
ongoing
encouragement as a family member and at the same time acted
as my best
critic. He has found a most delicate balance between accepting
and chal-
lenging just about every idea presented in this book.

Nancy Budwig
 SVN IE CAN

FOUNDATIONS
 B14 o@ piu = onde tad a
; or ==, 48) Vee Ja bee’,
" fe Pein a at) (Sub in tas5 8
7 . eiteie is cis ) Ome wens eee @
COON ax, th|r) ig ae woe
: tt (stat): Wey -> Amit 2 There
“Seed Wer arte SERA GupcTiPe i onsPega
ae WD 6 oj : a (e ‘aegis aa
MRS es 54, ~ “1s
De par) =< i a Ge
genig c- . , : atone anes

wea = Te piled Tar q

jaa Aa iy ‘0% -—_ a 7 ( 1
+ & i
4 «@W i ‘Vente 3

: 7 a a) “yee Cs &
— a omer sd,
@e : Z 5 4 > is oe
ap 9! ate Au Na, Capiet | ~

hs i : és
ye
: in at? neg f it
oe nhenn® ’ 7
y rho Gea; paling’
~ i vg ft ALG n . a ae
ory ni Sen; ae cael 7 Geaaah
Sw» fe | 6t Qe derw Coxe Yo
P~ “4 a ongy : ers ‘ S 0 ne
 Bavali=te WINE

Functional Approaches
to Child Language

The approach adopted in this book is a developmental-functionalist ap-

proach to child language. From the start it is important to explain what is
meant by the term developmental and what is meant by the term /function-
alist. Often it is assumed that any approach to child language is necessarily
a developmental approach—what is one studying if not language develop-
ment? I argue throughout this book that we need to be far more concerned
with the notion of development than much child language research has
been in recent years (see also Bamberg, Budwig, & Kaplan, 1991; Bloom,
1991; Tomasello, 1992). The goal of this chapter is to familiarize the reader
with various kinds of functional approaches and to lay out, in particular,
some of the assumptions guiding the use of the term developmental-func-
tionalist in this book. I begin with a general discussion of the notion of
function. I then describe how this term has been used in linguistic theorizing
and in child language research. I conclude this chapter with a discussion of
some of the problems inherent in current child language research that has
been considered functional in nature.

THE NOTION OF FUNCTION

IN LINGUISTIC THEORIZING

From the outset I should clarify that the use of the term functional in
functional approaches to language should not be confused with uses of the
term functionalism in psychology and philosophy. Functional approaches
to language share little with programs of behaviorists such as Skinner (1957)
that have been referred to as a kind of functionalism. The use of the term

3
4 CHAPTER 1

functional within linguistics and psycholinguistics can also be distinguished

from uses within computer science and artificial intelligence that refer to
mathematical functions (see Bates & MacWhinney, 1989, for further discus-
sion). To assist in severing the connection between these other uses and
the use of the term function in linguistic and child language theorizing, I
avoid the term f/unctionalism and use the term functional.
It is important to note that within linguistic theorizing there has been no
single functional theory of language. Rather, we find a variety of functional
theories, models, and approaches, which differ from one another to various
degrees. Although, as we see here, there are many important differences
between the various uses of functional in linguistic theorizing, it is possible
to find some assumptions that are agreed upon by all (see also Budwig,
1993). I briefly discuss these similarities before focusing on the differences
between the various uses.
The first and most important assumption shared by the various approaches
that could be considered functional is the belief that language has evolved
and is acquired in relation to the communicative functions it serves. Language
thus is not an arbitrary and autonomous system but, rather, is organized in
relation to the needs of those who use it. As a consequence, all theorists
taking a functional approach share the belief that language must be studied
in relation to context, regardless of what they take context to mean.
A second and related similarity between the various uses of functional is
the belief that language is a system of forms and meanings. Forms are a vehicle
through which the meanings can be realized. Thus, in functionally oriented
research, forms are viewed as a means to an end rather than as an end.
It has often been suggested that there are basically two broad orientations
within linguistic theorizing: a formal orientation and a functional orientation
(see for instance, Bates & MacWhinney, 1989; Foley & Van Valin, 1984). We
might suggest that what unites various functional theories is a shared disbelief
in formal orientations. Functionalists join in rejecting the formalist assumption
that language consists of a set of structural descriptions of sentences, as well
as formalist beliefs about how language ought to be studied and how lan-
guage is acquired by the child (see, for instance, Chomsky, 1957, 1975, for
a more complete discussion of these points). Although on the one hand,
one can drawa division between formal and functional Orientations, it seems
important to keep in mind that at another level of analysis one could construe
formalist accounts of language as reflecting a kind of functionalist orientation
(see Silverstein, 1987, for a discussion of this possibility). In
addition, as
Silverstein (1991) more recently pointed out, some functionalist orientations
share many features of formal orientations. The point is that the
distinction
between the two orientations may not be as clear as it might seem.
More relevant to child language theorizing and the various uses
of the
notion of function in the child language literature is an underst
anding of
FUNCTIONAL APPROACHES TO CHILD LANGUAGE
5

some of the differences between various kinds of functional approaches. I

begin now to distinguish some of these approaches (see also Silverstein,
1987). It should be made clear that the numerous orientations falling under
the general heading of functional do not divide into neat groups. Neverthe-
less, for the sake of presentation here, I have opted to divide the various
kinds of functional perspectives to language into four groups. First, I discuss
a group with a cognitive functional orientation. I then discuss a group that
can be characterized as bearing a textual functional orientation. A third
group is referred to as adopting a social functional orientation. Members
of the final group share an interest in a combination of these orientations;
this group will be referred to as adopting a multifunctional orientation.
Other groupings are possible, but for the general goal of understanding the
role of functional linguistic theorizing in child language research, this group-
ing seems most appropriate.

Cognitive Orientation

The group labeled cognitive in orientation holds together in a belief that

languages are organized around event schemes. Members of this group
include Comrie (1981), DeLancey (1984, 1987), Fillmore (1968), and Lakoff
(1977), among others. A central assumption for this group is that particular
linguistic devices are organized around various perspectives speakers take
on events. The events are thought to be organized around prototypes, and
deviations from the prototype are thought to be marked by deviations in
morphosyntactic patterns. The idea, then, is not simply that particular lin-
guistic devices can be correlated with particular event perspectives but,
rather, that the relations between various form-event pairings can be linked
at a higher level of organization into a larger system of form-meaning cor-
respondences. The central claim is that alternations between particular gram-
matical devices (i.e., case markers) indicate a related change in the view
the speaker adopts on a given action frame (i.e., degree of participant in-
volvement). These action frames consist of a bundle of both semantic and
pragmatic features that are ‘organized around a prototype (see Holisky, 1987,
for an illustration of the semantic and pragmatic clusters thought to motivate
the use of ergative case with subjects of some intransitive verbs in Tsova-Tush
[Batsi]).

Textual Orientation

Approaches in the second group to be discussed can be considered similar

(despite important differences) in that they are concerned with the ordering
and organization of units of discourse both within and between sentences.
Members of this group include proponents of what has been referred to as
6 CHAPTER 1

Prague School functionalism (Firbas, 1964; Firth, 1951, among others), British
functionalism (Halliday, 1973), Role and Reference Grammar (Foley & Van
Valin, 1984; Van Valin, 1993), as well as Chafe (1980), Tomlin (1987), and
others. In contrast to the cognitive functional group, this group is interested
in the relation between various aspects of formal structure and the way texts
are organized. Focus on narrative discourse has been central to this group.
With regard to the organization of information units within sentences, many
of those working within this orientation have focused on two kinds of
sentential elements: first, the speaker’s point of departure (referred to as either
the theme, the topic, or given information), and second, what is said about
this (referred to as the theme, the comment, or new information). Research
falling under this heading has illustrated the extent to which various linguistic
means (i.e., lexical, intonational, or word order phenomena) mark such a
contrast.
With regard to the overall organization of texts, other research has ex-
amined how certain linguistic devices (i.e., tense-aspect markers, pronomi-
nal-nominal shifts) segment given texts into smaller textual units informing
the listener/reader about various aspects of the global. structuring of infor-
mation (see, for instance, Chafe, 1980; Tomlin, 1987).
In a separate but related line of theorizing, Foley and Van Valin (1984)
and Van Valin (1993) outlined Role and Reference Grammar (RRG), which
focuses on the relations between clause-internal morphosyntax, clause link-
age, and cross-clause reference tracking mechanisms. This leads to a con-
sideration of such aspects of language as the semantic structure of the clause,
case marking, intraclausal syntax, and systems of discourse cohesion. The
general point here, shared with many others in the textual functional group,
is that the analysis of various aspects of morphosyntax must start from a
discourse perspective.

Social Orientation

The first two groups were said to be interested in the connections between
grammar and event schemata, and grammar and text organization. The third
group is unified by a joint interest in social aspects of language use. We
can include under this heading at least three kinds of social functional
approaches: first, that of a group of philosophers of language who developed
a theory of speech acts; second, that of a group of theorists who examined
the relation between language and broader sociological issues; and third,
that of a group who focused on sociocultural issues and language function-
ing. I briefly consider each of these groups.
One subgroup bearing a social orientation can be characterized by an
interest in a theory of speech acts. This group (i.e., Austin, 1962; Grice,
FUNCTIONAL APPROACHES TO CHILD LANGUAGE pe

1975; Searle, 1969, 1979) worked within the logico-semantic tradition, where
the central question concerns the principles by which utterances are assigned
meaning in particular settings. This group set out to show that utterances
function not only to express propositions about the world (which then can
be tested against truth conditions), but also to perform certain actions. Its
members were not particularly interested in working with real conversational
data occurring in real event time but, rather, tended to work with made-up
examples with very little emphasis on aspects of the situations in which
such utterances were supposed to take place. The general point of such
work was to challenge prevailing views concerning the centrality of truth-
conditional analyses of sentence meaning, and to develop a theory of po-
liteness and interaction (see Levinson, 1983, for an excellent summary of
this tradition).
Although theorists like Austin and Searle were intrigued by the fact that
one and the same utterance could function differently in different language
contexts, they never really were interested in broader sociological issues.
The second subgroup falling under the social functional orientation focused
on these issues. One group of theorists, working within the tradition of
conversation analysis, focused on real utterances sequenced in time and
was concerned with the internal structure of conversation (see Sacks,
Schegloff, & Jefferson, 1974). Another group was more interested in the
issue of conversational inference as well as the notion of contextualization
cues—surface devices that play an important role in negotiated meaning in
social contexts (Gumperz, 1982).
A final subgroup includes sociocultural approaches to language function-
ing. This group, though closely related to the sociological one just reviewed,
can be distinguished in terms of its emphasis on the relation between lan-
guage and the cultural setting within which language occurs. What is different
here is that focus is not placed so much on the goal-directed nature of an
individual’s functioning within a given context as on the role of societal
issues. The general approach adopted within this framework is ethnographic
in nature (see Duranti, 1985; Hymes, 1974, for further elaboration).
Pulling together these separate traditions, what unites such approaches
is the belief that language is an instrument of social action. Furthermore,
there is a shared belief among these theorists that the use of particular
linguistic devices can be related to speakers’ attempts to achieve particular
social goals. This emphasis on the relation between grammar and the social
world is what distinguishes the various kinds of functionalist perspectives
grouped under this heading from the approaches described under the pre-
vious two headings. Those interested in the relationship between grammar
and event schemata or between grammar and text have not been primarily
concerned with social issues.
8 CHAPTER 1

Multifunctional Orientation
Several approaches that can be labeled functional in nature do not fit neatly
into one division but, rather, are best characterized as integrating aspects of
more than one of the orientations discussed here. I refer to these as multifunc-
tional. Although there are large differences among the various approaches
discussed here, for the present purposes it can be noted that a major similarity
among most of the approaches labeled as multifunctional is that they attempt
to integrate two of the levels of analysis. The most common kind of linkage
drawn in multifunctional approaches is between the event schemata orienta-
tion and the textual orientation. This kind of integration is illustrated in several
different lines of work (see Comrie, 1981; Giv6n, 1979, 1984a, 1990; Hopper
& Thompson, 1980, among others).
One way to illustrate the multifunctional orientation of this group is to
consider how such an orientation would approach a given issue. One com-
mon issue is that of the relation between grammar, agency, and topicality.
Some working within what has been labeled the multifunctional orientation
not only draw a connection between the use of particular grammatical forms
and reference to agents versus patients—thereby appealing to the event
schema orientation—but also examine the relationship between such con-
nections and broader textual issues such as topicality. For instance, Bates
and MacWhinney (1989) argued that often particular forms link up with
overlaps of semantic and discourse meanings. They argued that many lan-
guages show a tendency to recruit one grammatical device to mark both
agents and topics because these semantic and discourse functions represent
a natural coalition because agents typically are topics in natural discourse.
Others have argued that certain form-meaning correspondences can be
accounted for only by considering discursive phenomena. For instance,
rather than arguing for form-meaning-function bundles, Hopper and
Thompson (1980) suggested that certain regularities in the relation between
particular grammatical devices and semantic meanings can be accounted for
only by considering discourse functions such as foregrounding and back-
grounding. Thus, although there has been a tendency within the multifunc-
tional orientation to link form, meaning, and discourse function, differences
exist in the degree to which a particular group is suggesting an overlap or
a causal connection between the different levels of analysis.
What is important to note is the general reluctance in this area to draw
a connection between what we have referred to as the social functional
orientation and the other orientations (however, see Holisky, 1987; and van
Oosten, 1986, for exceptions).

tt should be noted that with regard to the study of children’s use of passives, Budwig
noted that it well may be that the early passive use also links up with textual and social
functional characteristics and that the focus on the cognitive functional orientation was
guided
by the nature of the data (see Budwig, 1990). It can also be noted that Slobin (1988) suggested
the importance of integrating a pragmatic dimension of analysis.
 FUNCTIONAL APPROACHES TO CHILD LANGUAGE
D

THE NOTION OF FUNCTION

IN CHILD LANGUAGE THEORIZING

Just as was noted to be the case for linguistic theorizing, the notion of function
has been used in many different ways in the area of child language acquisition.
Although there are many different uses of the term functional in the literature,
the various uses share the assumption that language develops in relation to
the communicative functions it serves. Various child language researchers
differ with regard to the degree to which they believe the telos of language
development to be functional in nature. The relevant point is that it is possible
to believe that children begin with functionally oriented language systems,
and yet over development these systems become more abstract (see, for
instance, Werner & Kaplan, 1984, for such a discussion). A second common-
ality across the various uses of the term functional in the child language
literature is the reliance on a notion of system. This similarity was noted also
in the linguistic literature. With regard to children, the notion of system comes
in when accounting for the relation between one language function and
others, as well as that between various language devices and various functions.
With these similarities in mind we can turn to a brief discussion of some
of the differences in the way the term function is used in the study of
language development. The various kinds of functional orientations noted
earlier also can be found in the child language literature, though they have
played a different role. It should be noted that most of the work in child
language that can be considered functional in nature has fallen under what
was labeled earlier the social functional orientation. Before turning to a
more systematic review of some of the different uses of function in the child
language literature, it should also be noted that my review is not exhaustive.
The discussion here is limited to child language research that has been
concerned with the relationship between language structure and language
function, because this is most central to the research outlined in subsequent
chapters of this book. This restriction has meant that much of the research
that would have fallen under the social functional heading, consisting of
detailed typologies of the functions of children’s language, is not reviewed
here (see, though, Atkinson, 1982; Budwig & Bamberg, in press; Cook-Gum-
perz, Corsaro, & Streeck, 1986; Ervin-Tripp & Mitchell-Kernan, 1977; Ochs
& Schieffelin, 1979). With these comments in mind, we can now consider
child language research that falls under the headings of cognitive orientation,
textual orientation, social orientation, and ‘multifunctional orientation.

Cognitive Orientation

The best example of the cognitive functional orientation can be found in the
work of Slobin (1981, 1985). In summarizing the relationship between various
linguistic forms and their developmental routes Slobin suggested that many
10 CHAPTER 1

forms are first used to take particular perspectives on events. This is best
illustrated in Slobin’s discussion of the grammatical encoding of the manipu-
lative activity scene. From cross-linguistic comparisons of very different kinds
of languages, Slobin concluded that one of the opening wedges for grammar
is the linguistic encoding of a scene in which an agent brings about a change
of state in an object. Although the kind of linguistic treatment may vary from
language to language, this event is said to receive linguistic treatment in the
very early phases of grammatical development by children across the world.
What is particularly interesting about the examples that Slobin discussed
concerns the relationship between the children’s system and the target
languages the children are acquiring. Slobin’s claim highlights the creative
nature of the children’s use of grammatical forms. In my discussion of the
cognitive orientation within linguistic theorizing I pointed out that languages
across the world have been noted to give special linguistic treatment to a
cluster of related notions having to do with agency and transitivity; the
interesting aspect of Slobin’s finding is that the children are not merely copying
the adult system but, in fact, are creating their own form-function pairings.
Slobin suggested that children across the world begin with what he called
Basic Child Grammar—a grammar marking specific notions that are not
necessarily marked in the target language but that nevertheless are important
conceptual distinctions within the world of the child.
A second illustration of the cognitive orientation stems from the study of
early use of passives by young children. Examining the use of passive sen-
tences in children between the ages of 2 and 10 as well as in adults, in both
longitudinal and cross-sectional data, I argued that the children contrastively
employ the auxiliary forms get and be in order to mark a different perspective
on events. The use of be links up with talk about an event involving a
generic or unknown agent, or one who is irrelevant to the discourse. The
use of get is found in utterances that refer to an event in which an action
brings about negative consequences. What is central to the children’s use
of passive constructions, regardless of auxiliary choice, is that talk focuses
on a scene involving an agent who is not in focus (see Budwig, 1990).
Abstracting across these two studies, we find the principle features of the
cognitive functional orientation. Note that the examination of specific lin-
guistic forms in context has been crucial. The cognitive functional orientation
involves the examination of how particular grammatical devices mark shifts
in perspective on events. The central claim is that children exploit particular
devices to mark various kinds of perspectives that can be taken on a given
scene. Furthermore, it is claimed that these devices function to provide a
wedge into the formal adult system. Note that this functional orientation
differs from the textual orientation in that there is no focus on the use of
particular devices across stretches of discourse. Also there is little concern
for how the devices function socially. Within the cognitive functional ap-
FUNCTIONAL APPROACHES TO CHILD LANGUAGE 11

proach the assumption is that language is representing the child’s view of

the world and not that it is used as a tool of social action, as is characteristic
of the social orientation.

Textual Orientation

In general there has been little focus on the textual orientation within the
study of child language. Much of the research, as we see here, has focused
on aspects of later language development. One of the most complete studies
of particular linguistic devices and how they function to structure the flow
of the ongoing discourse is that of Paprotté and Sinha (1987). Paprotté and
Sinha explored the relevance of central notions of the Prague functional
sentence perspective to the study of language acquisition. With regard to
the early phases of language development, Paprotté and Sinha argued against
communicative pressure guiding the functional motivation for syntax. In-
stead, they argued that principles governing the organization of sentential
information can be related to the “evidentiality of the information” (Paprotté
& Sinha, 1987, p. 219). At later points of development, children are said to
become concerned with linking discourse over consecutive sentences.
The study of subsequent phases of language development (i.e., beyond the
two-word stage) has focused primarily on children’s narrative abilities and
how various linguistic devices indicate textual cohesion and coherence (see
Bamberg, 1987; Berman & Slobin, 1984; Hickmann, 1987; Karmiloff-Smith,
1986). These studies revealed that children acquiring a variety of languages
make use of several different linguistic devices (i.e., pronominal-nominal
shifts, tense-aspect markers) to segment a text into cohesive chunks that
ultimately establish larger conceptual units making up a narrative. As I noted
with regard to the cognitive orientation, the children’s early use of specific
devices does not necessarily match up with adult usage. That is, there seems
to be little evidence that the children are simply listening to adult uses of similar
forms and then making use of such forms for similar functions. What we find
reported in the studies cited here is an extended developmental sequence in
which the children reinterpret such devices for various textual functions that,
though consistently used, are sometimes at odds with the input they receive.
With regard to the textual orientation, the issue of central importance is
the extent to which particular linguistic devices are employed to help or-
ganize stretches of discourse both intrasententially and across broader
stretches of text. In general, the relationship between the use of particular
devices and the perspective taken on a given event is not important, nor is
the social function of the language examined.?

“Note, though, that Bamberg (1991) argued that cohesive divices also play a role in ind icating
what perspective the speaker takes on an event. That is, he claimed that textual devices
simultaneously indicate how the speaker views the event being described.
12 CHAPTER 1

Social Orientation

As has been noted, most of the work within a functional perspective with
regard to child language development can be characterized as falling under
the heading of social functional orientation. However, as I have noted,
much of it has focused on social functional aspects of children’s talk without
specific focus on the relationship between particular linguistic devices and
such functions. In this section I limit myself to a discussion of only some
of the work bearing a social functional orientation, with special reference
to the nature of the relationship posited between forms and social functions.
Much of the work in this area has focused on various aspects of the
development of grammar and their relation to levels of analysis above the
level of the sentence, such as the activity level, speech events, conversational
exchanges, and speech acts (see Ervin-Tripp, 1989, for an excellent review
of these studies). One area that has received more focus than others concerns
the relationship between specific linguistic forms and the pragmatic force
of the utterance within which such a form appears. For instance, Carter
(1975, 1978a, 1978b) noted in detailed case studies that children will con-
trastively use particular forms, depending on whether the child desires an
action to be carried out by a partner or whether the child wants to draw
the communicative partner’s attention to a particular object.
This sort of contrast based on the pragmatic force of the utterance was
also noted by Deutsch and Budwig (1983). They suggested that children’s
early contrastive use of multiple first person possessive pronouns is due to
the communicative force of the utterance. In a reanalysis of Brown’s (1973)
longitudinal data, Deutsch and Budwig claimed that the children went
through an extended phase of contrastively employing My + Possessum (Les
My pencil) versus Own Name + Possessum, depending on whether the
utterance was indicative or volitional. Researchers examining other lan-
guages have also indicated that this sort of contrastive use of pronominal
and nominal possessive forms marks a pragmatic distinction concerning
utterance force (see Kolodziej, Deutsch, & Bittner, 1991; Vila, 1987; see also
Halliday, 1975; Painter, 1984, for further examples of English).
Thus far my discussion of the social functional orientation has emphasized
what could be viewed as fairly local aspects of the social context. Some
work also has attempted to relate broader sociocultural features to various
aspects of grammar. The idea that broader social and cultural systems play
a role in the child’s development of various grammatical systems was won-
derfully illustrated by Ochs (1988) and Schieffelin (1990). Their cross-
linguistic comparisons of children’s acquisition of Samoan and Kaluli, for
instance, pointed out that one reason children in the two cultures might
acquire similar inflectional morphology at very different points in develop-
ment is that in one language, usage is sociologically constrained, whereas
this is not the case in the other. Given that much of the research in the area
 FUNCTIONAL APPROACHES TO CHILD LANGUAGE
13

of child language has attempted to minimize cultural (and subcultural)

dif-
ferences, this is an area open for future research.

Multifunctional Orientation

There has been surprisingly little work in the area of child language that
can be considered multifunctional in orientation. In part, this relates to the
previously mentioned point that few researchers have examined child lan-
guage from the perspective of the textual orientation (see earlier discussion).
The most common sort of linkage noted earlier was that between the cog-
nitive and textual orientations (see, though, Bamberg, 1987; Bates &
MacWhinney, 1987; Berman & Slobin, 1994, for important exceptions). One
excellent example of the multifunctional orientation is found in Gerhardt’s
work (Gee & Savasir, 1985; Gerhardt, 1990). Gerhardt examined various
aspects of children’s linguistic systems with an eye toward semantic, prag-
matic, and discursive factors associated with their use. For instance, Gerhardt
argued that 3-year-old children contrastively employ the forms will and
gonna; a discursive analysis reveals important contextual distinctions in the
use of these forms. Gonna appears in discourse in which the children were
planning and organizing; it implies a more distant intention to act in a
particular way. In contrast, will appears in the context of ongoing cooperative
peer play and refers to an immediate intentional stance. Gerhardt’s research
highlights the fact that children as young as 3 years are capable of linking
multiple levels of information into clusters of form-function pairings. Al-
though we are beginning to know more about the interrelationship between
semantic, pragmatic, and textual information from studies of this sort, much
further research is needed to unravel the pathway between original multi-
functional usage and patterns found in the input.

DEVELOPMENTAL ISSUES AND FUNCTIONAL

APPROACHES TO CHILD LANGUAGE

Thus far, I have suggested that there is no single unified functional approach
within the study of linguistics and child language. Several different kinds of
approaches have been discussed, and I have noted some parallels between
functional approaches used in linguistics and those used in child language
research. In grouping various kinds of functional approaches, the divisions
have focused on important distinctions in content areas. For instance, I have
considered such differences as the extent to which a given functional ap-
proach emphasizes cognitive notions, textual notions, social notions, or some
combination of these. There are, though, other differences that are central
to various functional approaches. Three other issues are particularly relevant
14 CHAPTER 1

to our discussion of functional approaches to language development. These

relate to distinctions between various approaches concerning the following
developmental issues:

1. In positing a relationship between particular forms and functions, what

are the specific claims, if any, about causal connections? Are there,
for instance, claims that functions trigger forms or vice versa?
2. What is the endpoint of development thought to be? In particular,
within the various approaches, is it suggested that the endpoint of
development is the acquisition of a formal grammatical system or is
the endpoint also viewed as functional in nature?
3. Where do the functional categories used in a given approach come
from? Does a given approach make use of a specific functional lin-
guistic theory or are other categories used?

Answers to these questions help clarify developmental issues related to

functionalist theorizing in child language research.

Form-Function Connections in Language Development

There seem to be three different kinds of positions taken with regard to

claims about form-function connections. The question here involves the
directionality of influence. In suggesting that children make use of form-
function pairings in acquiring a linguistic system, functionalists have differed
with regard to (a) the degree of specificity regarding the nature of this
connection, and (b) the status they give to both forms and functions.
One of the dominant views is that functions guide form. Here the claim is
that the child has a set of functions and searches or finds forms to map onto
these functions. It is important to note that various researchers differ, of course,
on the nature of the functions but agree that functions are primary. This can
be better illustrated by an examination of two different kinds of functional
approaches that nevertheless share this view of function dominating form.
One example of this position is that in Slobin (1985). Slobin argued that
conceptual development plays a central role in the development of various
grammatical forms. First of all, Slobin’s cross-linguistic approach revealed that
conceptual development provides a starting point for grammatical develop-
ment. Certain salient conceptual notions (such as transitivity) are more likely
than other notions to receive early grammatical treatment. As a result of this,
certain preferred event perspectives are more likely to receive grammatical
treatment, which in turn helps to determine which forms develop.
Slobin also suggested that the universal conceptual schemes can lead
children to make grammatical distinctions that are not necessarily provided
in the input the child receives. In addition, conceptual development has
FUNCTIONAL APPROACHES TO CHILD LANGUAGE
15

been noted to determine the order in which particular grammatical forms

emerge. Although Slobin postulated that conceptual development is the
guiding force in grammatical acquisition, and that children across the world
begin with universal preferences to mark particular conceptual categories,
this should not be taken to indicate that Slobin did not believe that formal
pattern preferences exist in early child language development. Children do
use form patterns, but often these form patterns are tied to underlying
meaning patterns. Slobin believed that when languages require that the
speaker combine two or more meanings in using a particular grammatical
form, the child will acquire those in which there exists a kind of semantic
affinity between the two semantic meanings. In addition, with regard to the
placement of grammatical markers, children show a preference for placing
such markers according to semantic relevance. Morphemes that go together
semantically are more likely to be placed together syntactically by the child.
Thus even when children work with form patterns, elements of meaning
have been noted to play a role in acquisition. Although Slobin suggested
that many aspects of grammatical development can be accounted for by
appealing to semantic underpinnings, he also agreed that children are ca-
pable of working at the level of form independent of semantic criteria. The
general claim put forth by Slobin, though, was that semantic meanings play
a central role in the child’s construction of grammar.
Ervin-Tripp (1989) built a similar argument suggesting that function guides
form in grammatical development:

The hypothesis of a form/function relationship suggests that in acquisition

the trigger for development of certain forms may be their density and impor-
tance in the achievement of conversational goals. The social and activity con-
text makes their meanings clear, simplifying the acquisitional process. (p. 1)

Though Ervin-Tripp had a somewhat different notion of function, the general

point is the same, namely, that the development of formal devices is guided
by the functions such devices serve.
All functionalists do not, though, necessarily argue that function guides
form development. For instance, some researchers seem to take a somewhat
neutral position with regard to causation. Gerhardt (1988, 1990) and Gerhardt
and Savasir (1986), in her careful contextually based analyses of various
functional aspects of children’s acquisition of different grammatical systems,
made no claim that it is function that guides form or vice versa. Such work
is best described as highlighting the interactive nature of syntax and discourse
and the importance of studying both in tandem. The important point here
is that adopting a functionalist position need not imply that one believes
that function guides form. Gerhardt’s work is an excellent illustration of this.
In another type of position, functionalists have proposed a very complex
interaction between form and function in development. Bates and MacWhin-
16 CHAPTER 1

ney (1989) argued, within the framework of the Competition Model, that
mappings between forms and functions are best viewed in terms of both
horizontal and vertical correlations (Bates & MacWhinney, 1987). Although
most functional approaches focus on the vertical correlations of form-function
pairings, Bates and MacWhinney suggested two kinds of horizontal correla-
tions—one at the level of forms themselves, and one between various
language functions. According to their perspective, the child is not simply
acquiring form-function pairings but is also making use of formal correlations
or correlations between various functions. Bates and MacWhinney suggested,
though, that the vertical mappings are the most central: “. . . although the
system is capable of acquiring a complex set of horizontal correlations, the
mappings that drive the system are the vertical correlations” (p. 166). The
model provided by Bates and MacWhinney suggests a far more complex
relationship between the acquisition of form and that of function. According
to their perspective the child is not simply relating forms to already acquired
functions. Rather, the child often acquires functions and then seeks to express
them, or the child acquires forms without being bound toa particular function.
In this section, we have seen that there are many alternative charac-
terizations of the relationship between form and function in various func-
tionalist approaches. At present, various approaches differ not only with
regard to what they take function to mean, but also with regard to whether
function is said to play an organizing role in the development of grammar
or whether one believes that the developmental relationship between lin-
guistic forms and functions is more complex in nature.

Continuity and Functional Approaches

One of the chief critiques against functional approaches has centered around
the issue of continuity. The argument goes as follows: If the child uses
function as a wedge into grammar, how does a researcher account for the
transition into a formal] abstract system? Although most of the discussions
(cf. Gleitman & Wanner, 1982) of this issue have centered around a clear
distinction between continuous and noncontinuous theories, the issue is
really better viewed in terms of a continuum. The general assumption is
that functionalist approaches must grapple with the continuity problem, that
is, how the child transcends the functionally oriented early grammar and
arrives at a formal abstract system. In contrast, formalist approaches are
thought of as continuous in nature to the extent that the origin and end
point are formal in nature. The importance of appealing to a continuum is
best understood by looking more closely at some of the different ways
various functional approaches deal with the continuity issue.
At one end of the continuum we find functional approaches like those
that have been discussed by critics; namely, they appear to be discontinuous
FUNCTIONAL APPROACHES TO CHILD LANGUAGE 17

in nature. Under such proposals, it is held that the functionalist nature of

early grammatical categories is specific to child language and that ultimately
such systems give way to more formal abstract adult systems (cf. Schlesinger,
1988; Werner & Kaplan, 1984). Specific proposals are made about how the
child might transcend a functionally based system.
It is important to recognize that not all functionalist approaches can be
described as discontinuous in nature. At the other end of the continuum are
functionalist approaches that take a continuity position. At this end of the
continuum, there is no problem of discontinuity, because the telos of lin-
guistic development is believed to be based on categories that are functional
in nature. This position is best illustrated by Van Valin (1991).
There are many functionalist approaches that fall somewhere in between
these two positions. On the one hand, they fit with the continuity position
to the extent that they posit starting and ending points that are functional
in nature. Yet they share with the discontinuity side of the continuum the
belief that the functional categories and even the form-function linkages that
the child creates early on are not necessarily the ones that are functioning
in the adult system. As Bates and MacWhinney (1989) pointed out, many
of the functions that are part of the adult system are not necessarily those
to which a young child will be sensitive. To this extent, there are some
functionalist approaches that deal with a less extreme version of the conti-
nuity problem.
The point here is that the continuity problem need not be a problem for
functional approaches. There are functional approaches that have been con-
strued such that they are as continuous as formal approaches. It is also
important to recognize that there are nativist approaches that have had to
deal with the continuity problem, as illustrated, for instance, in Pinker’s
discussion of semantic bootstrapping (Pinker, 1984, 1989). It is important to
keep in mind that there are major differences in how particular kinds of
functional approaches deal with the issue of continuity. It is equally important
that any approach to child language that proposes that children’s early lin-
guistic categories differ from those thought to be functioning in the adult
linguistic system also account for the mechanisms that lead the child from
one point to the next.

The Source of Linguistic Categories

A third distinction between various functional approaches, and one that

relates to my previous discussion of the continuity issue, concerns the source
of the various linguistic categories posited in a given functionalist approach.
One central distinction in various functional approaches to child language
involves the extent to which a given approach either starts with a set of
well-defined categories springing from a specific linguistic theory or works
18 CHAPTER 1

with categories that are not part of an accepted functionally based linguistic
theory but instead develop out of an analysis of what children do with
particular forms. Not surprisingly, those approaches that start with well-de-
fined categories stemming from functional linguistic theorizing are exactly
those approaches that were labeled as continuous in our previous discussion
of continuity.
Perhaps the best example of child language work that draws upon func-
tional linguistic theory can be found in Van Valin’s attempts to relate RRG
to children’s language (see Van Valin, 1991). The approach adopted by Van
Valin is quite different than the approach adopted by a number of others
working within a functionalist perspective who have been less concerned
with starting with built-in assumptions about how particular forms function
in adult grammars and instead have sought to find out how children make
use of particular linguistic devices and how such usage changes over time
(cf. Bamberg, 1987; Bamberg, Budwig, & Kaplan, 1991; Berman & Slobin,
1994; Bloom, 1991; Deutsch & Budwig, 1983; Ervin-Tripp, 1989; Gerhardt,
1988, 1990; Slobin, 1988). Researchers in this second group started with few
presuppositions about how forms may function for the child. Their task was
to figure out what meaning the child gives to a form by examining usages
in context (context meaning different things to different researchers).
As just one illustration of this difference, we can look more closely at
Deutsch and Budwig’s analysis of the use of Own Name and My in possessive
constructions like Adam car and My car. An analysis beginning with estab-
lished categories might categorize both these utterances, produced by the
same child in close proximity, as possessive constructions (see Braine, 1976,
for similar categorizations). Deutsch and Budwig (1983) examined functional
aspects of such utterances and came to the conclusion that Own Name +
Possessum and My + Possessum were two different categories for the children
studied. For a period of several months the children constructed a distinction
between reference to possessions in which control was not at issue (marked
with Own Name) and reference to possessions the child would like to gain or
maintain control of (marked with My). This example illustrates the attempt on
the part of the researchers working within this perspective to find out what
function children attribute to various forms. Thus, rather than referring to
examples such as Adam car and My car as possessive constructions, the
researchers argued that the child uses distinct forms to mark a contrast in
function.
Thus far we have looked at two very different ways in which functional
approaches categorize child language data. There are other possibilities that
fall somewhere in between these two extremes. For instance, Karmiloff-
Smith’s (1979) work on the acquisition of determiners was similar to ap-
proaches that began with preestablished categories, in that Karmiloff-Smith
began with some predetermined ideas about how determiners function. At
FUNCTIONAL APPROACHES TO CHILD LANGUAGE 19

the same time, her research can be viewed as similar to the research of
those who examine how forms might function for children at various points
in development, to the extent that she examined whether children in their
initial uses of determiners employed them multifunctionally, as adults do.
The work of Bates and MacWhinney (1987, 1989) also falls somewhere
between these two extremes. Like Karmiloff-Smith, Bates and MacWhinney
began with specific assumptions about form-function coalitions in various
languages and examined which aspects of such coalitions children acquire
in what order. Thus Bates and MacWhinney recognized that there are po-
tential differences between children’s form-function pairings and those of
adult speakers, but at the same time they worked with predetermined ideas
about how forms might function in a given language. To this extent, their
work was comparable to Van Valin’s. Bates and MacWhinney noted, though,
that the categories they employed could be considered pretheoretical or
general (i.e., agent, topic) and contrasted with the more specific categories
that Van Valin used.
It is important to recognize that although functionalist approaches have
often been criticized for lacking a theoretical conception of an adult grammar
(cf. Atkinson, 1982; Pinker, 1989), this claim may be true only of some
functionalist approaches. As Van Valin’s work illustrates, functionalist ap-
proaches, like formalist approaches, at times are guided by a specific lin-
guistic theory. It seems one of the most important challenges for functionalist
approaches in years to come is the integration of the two extreme kinds of
approaches described here. What we need are functional approaches that
begin with clearly delineated notions of the telos or end state of grammatical
development but that at the same time are sensitive to the functions children
assign to forms early on (see Bamberg, Budwig, & Kaplan, 1991, for further
discussion of this point).

A DEVELOPMENTAL-FUNCTIONALIST APPROACH
TO CHILD LANGUAGE

At the start of this chapter I described the approach adopted in this book
as a developmental-functionalist approach to child language. Thus far, I
have discussed various uses of the notion of function in both linguistic and
developmental psycholinguistic studies, but until now I have had very little
to say about the notion of development and, in particular, about why the
approach to be adopted here is labeled as developmental in nature. One
could raise the question—are not all functional approaches to child language
inherently developmental in nature? The answer to this question, it seems,
depends on what one takes the notion of development to mean. In this
section I discuss the notion of development in various functional approaches,
and I outline how this term is used in the approach adopted in this book.
20 CHAPTER 1

There is general consensus in the field of child language that language

acquisition is not instantaneous. The assumption is that the earliest phases of
child language are different from the target language the child is acquiring.
Various functionalist approaches have dealt with the issue of development in
different ways. One major criticism of functionalist approaches has been that
they have not really addressed the issue of developmentatall. As Pinker (1988)
pointed out in a general critique of all of developmental psycholinguistics:

_.. if language acquisition is the study of ACQUISITION (i.e., creation of new

knowledge contingent on environmental inputs of LANGUAGE (i.e., adult
linguistic abilities in all their complexity), then very few psychologists were
studying the question. There were studies of the speech of children at one
or more stages, and of their comprehension abilities, but virtually no one paid
attention to the learning process itself (i.e., the process by which the child
forms new rules), nor to the end state of acquisition. . . . (p. 100)

Although Pinker’s claim was not particularly directed toward functional ap-
proaches to child language, one might argue that his statement holds true
of certain functional approaches to child language that have examined the
various functions of children’s language at one or more stages en route to
the adult system. Atkinson (1982) also highlighted some problems for func-
tionally based approaches with regard to their stance toward development.
Although some functional approaches to child language have not been con-
cerned with developmental issues, there nevertheless are other functional
approaches where development has played a central role. I now examine
the issue of development within such approaches.

Two Notions of Development

in the Child Language Literature

One of the main questions we need to raise is what a particular approach

takes development to be the development of. There are two main ways in
which development has been perceived in functional approaches to child
language. First, some researchers have examined development by question-
ing the order of acquisition. Here the central issue is: In what order do the
various forms come in and how can one account for that order by appealing
to functional notions? A second view of development has focused less on
the order of acquisition of various forms and more on the development of
particular forms or particular functions, watching for developmental changes
in particular form-function pairings.
An example of a functionalist approach that looks at development in
terms of order of acquisition is that of Bates and MacWhinney (1989). In
their view, “learning is viewed as a process of acquiring form-function map-
pings” (p. 59). Their view of development was reflected in the following
FUNCTIONAL APPROACHES TO CHILD LANGUAGE 21

statement: “We have offered some strong predictions about language acqui-
sition across natural languages, claiming that cue validity will determine the
order in which grammatical devices are acquired” (p. 59). In talking about
development, Bates and MacWhinney were primarily interested in account-
ing for particular sequences in the course of development of linguistic
forms—that is, why a given form is acquired at a particular time relative to
another form.
Bates and MacWhinney also made use of the principle of functional
readiness in accounting for developmental sequences. According to this
principle, certain functions must be available before a mapping between
form and function can take place. For instance, certain grammatical devices
used to achieve discourse cohesion have been noted to develop later than
cue validity might predict, simply because the child does not understand
the function of these devices.
There seem to be two major problems with all developmental work of
this sort that is based on acquisition order. First, how does one assess
whether something is acquired? There are no clear-cut methods or stable
criteria used by researchers in deciding whether or not a child has acquired
a particular linguistic structure with regard to either production or compre-
hension (see Clark, 1985, for an excellent discussion of the problem of
assessing acquisition). Without a firm way of assessing whether something
has been acquired, it becomes problematic to draw generalizations about
ordering relations. A second problem for researchers who have approached
development from the standpoint of order of acquisition concerns the issue
of process. This approach forces one to view development in a static way—
something either is or is not acquired. Such a vantage point ignores the
process of development. This issue is taken up by a second view of the
notion of development in functional approaches to child language.
The second approach, that focusing on developmental histories of given
form-function pairings, begins with very different notions about acquisition.
One central assumption of this approach is that the child might be using
forms in a very different way than adult speakers. The developmental issue
is one of examining how a given form functions at a particular point in
child language and how a given set of form-function pairings undergoes
developmental changes over time.
Those taking this view of development are particularly interested in the
mechanisms involved in the transition from one form-function pairing to
another. The work of Karmiloff-Smith is an excellent example of this kind
of approach. Karmiloff-Smith (1979) outlined three phases of language ac-
quisition procedures. She started with the assumption that children first
approach language as if morphemes were unifunctional and that develop-
ment consists in conferring on a series of unifunctional homonyms the status
of a plurifunctional morpheme. In the first of three phases, roughly between
22 CHAPTER 1

the ages of 3 and 5 years, children attribute one meaning to each form.
They act as if they do not understand that one form may serve multiple
functions. During the second phase, roughly between 5 and 8years, children
seem to recognize two functions, as evidenced in two developments. First,
the children make use of redundant markers; and second, they create some-
what ungrammatical strings to mark functional contrasts. Karmiloff-Smith
suggested that these developments indicate that the child is becoming con-
scious of the plurifunctionality of forms. In the third phase of development,
neither the redundant marking nor ungrammatical forms are used, and the
child uses forms plurifunctionally. Although these phases were proposed
with reference to the development of determiners, Karmiloff-Smith suggested
that the form of development is very general. In particular, she suggested
that development consists of two related processes: “the child’s endeavour
to disambiguate by overmarking, use of emphasizers, use of paralinguistic
markers, etc., counterbalanced by his progressive endeavour to be economi-
cal” (Karmiloff-Smith, 1979, p. 236).
Karmiloff-Smith’s view of development is quite different from one that
focuses on what was discussed under the heading of order of acquisition.
Karmiloff-Smith was concerned with the development from unifunctional to
plurifunctional status of a given form. The general processes of disam-
biguation and economy motivate the changes in the system. What differs in
this approach is the focus on developments that take place prior to adultlike
usage. Thus we see that even within various functional approaches, several
different positions can be found with regard to development. Theorists differ
with regard to how much of their focus is on proposing form-function
patterns in child language and how much they attempt to account for de-
velopmental sequences that take place.
Although previous functional approaches to child language have at times
been concerned with the notion of development, these discussions have
been largely implicit. There has been very little in the way of explicit dis-
cussion of what one means by the notion of development. One major dif-
ference between the developmental-functional approach to child language
proposed here and previous functional approaches is the level of explicitness
about what the notion of development is taken to mean. I now consider
how development is viewed within the framework adopted here.

The Notion of Development

Within a Developmental-Functionalist Approach

In contrast to most scholars who treat the notion of development in the

field of child language (and developmental psychology as well), I do not
equate the notions of ontogenesis and development. I reserve the use of
the term ontogenesis to refer to the actual life span of a given individual.
FUNCTIONAL APPROACHES TO CHILD LANGUAGE 23

The notion of development will be viewed in terms of “the ways in which

agents utilized or constructed ‘means’ for the attainment of ‘ends’ ” (Kaplan,
1983a, p. 56). Following Werner (1957), Werner and Kaplan (1984), and
Kaplan (1983a, 1983b), the notion of development as used here is taken as
a heuristic, or an ideal. To this extent one can distinguish ontogenesis (the
actual) from development (rarely attained ideal). The use of the notion of
development here requires having a telos of development in mind prior to
empirical study. As Kaplan (1983b) suggested: “Development does not lurk
directly in the population(s) studied but resides fundamentally in the per-
spective used” (p. 196).
In order to clarify what is meant by the notion of development it is helpful
to discuss the orthogenetic principle formulated by Werner and Kaplan
(1984): “We assume that organisms are naturally directed towards a series
of transformations-reflecting a tendency to move from a state of relative
globality and undifferentiatedness towards states of increasing differentiation
and hierarchical integration” (p. 7). According to this principle, development
is defined in terms of both hierarchical integration and increasing differen-
tiation. To this extent, the principle is something one brings to the analysis
of data, rather than development being something one finds through em-
pirical examination.
Relating this view of development to the field of child language we can see
some themes from the previous discussion of functionalist approaches to child
language. For instance, we saw thata clear stipulation of a telos is something
that was discussed in the work of Van Valin (1991). Furthermore, having an
eye for developmental differences in form-function (means-ends) relations
has been discussed here as only one way in which researchers have examined
development. The difference between the present approach and most pre-
vious functional approaches to child language regards the way the notion of
development is viewed. As noted earlier, development is not equated with
change but, rather, is viewed as a perspective adopted toward data.
The approach put forth here shares much with that of Karmiloff-Smith, and
it is important from the outset to clarify the relationship between these two
approaches. With regard to their similarities, we can note that Karmiloff-Smith
began with a clear notion of the ideal or end point of study, and at the same
time she was interested in accounting for how a child makes use of given forms
across the course of development. This interest is shared in the developmen-
tal-functionalist approach adopted here. Furthermore, Karmiloff-Smith was
less interested in when a form is acquired in its full adultlike sense. Instead,
she wished to account for the course of development along the way, another
point of similarity between her approach and the approach taken here. The
major difference between the two kinds of functional approaches concerns
the account of developmental change. Karmiloff-Smith argued that children
go through three phases in language acquisition, moving from a level of
24 CHAPTER 1

unifunctional interpretation of grammatical devices toward plurifunctional

usage. This differs from the approach adopted here in that I suggest that
language development can best be accounted for in terms of hierarchical
integration and increasing differentiation. Old forms do not simply add on new
functions; rather, the whole system undergoes linguistic transformations such
that old forms are given new functions and old functions are related to new
forms (see Slobin, 1973). According to the functional-developmental approach
adopted here, form-function pairs are not only viewed as developing in and
of themselves but are also perceived as part of a broader linguistic system.
 CHAPTER TWO

Agentivity and Control:

Linguistic, Developmental
Psycholinguistic, and Developmental
Psychological Perspectives

In the previous chapter, I outlined the beginnings of a developmental-func-

tionalist approach that has guided the research reported in Part II of this
book. In Part II I focus on one particular area of study, namely, the linguistic
marking of agentivity and control in early child language. The issue is not
simply whether children refer to notions of agency but, rather, whether they
contrastively employ linguistic devices to mark different perspectives on the
way participants are related in ongoing discourse frames.
A number of considerations have led to the selection of the issue of the
linguistic marking of agentivity and control. I show why such a selection is
appropriate as I consider findings from previous studies concerning the
grammatical marking of agentivity and control in various languages, cross-
linguistic studies concerning the notions of agentivity and control in early
child language development, and studies within the area of social cognitive
development. I now discuss these three separate areas of study that have
shaped the questions guiding the research outlined in Part II. In the final
section of this chapter, I discuss the framework adopted here for studying
agentivity and control in early child language, and I outline a set of research
questions that guided the research reported in Part II.

LINGUISTIC PERSPECTIVES

In this section I review several separate strands of theorizing that have dealt
with the semantic notion of agency. One broad distinction in the various
studies is between those in which a theorist has viewed agency as a discrete
"
OLSON LIBRARY
NORTHERN MICHIGAN UNIVERSITY
MARQUE TTE, MIC
OIETTE AN
HIGAN
MICHIG 498505
4360
MAR
26 CHAPTER 2

semantic category and those in which agency is viewed as a semantic cate-

gory with prototypical structure. I now review research in light of this basic
distinction. It should be noted that the review of the linguistic literature is
not meant to be exhaustive; instead, the studies included in this discussion
have been selected because they have had significant impact on the way
the concept of agentivity has been discussed within child language research.

Agentivity as a Discrete Semantic Category

The separate strands of theorizing placed together in this section, although

different in many important ways, can be brought together to the extent
that in all cases, agentivity is seen as a discrete category. These three views
include: Fillmore’s (1968) case grammar, Jackendoffs (1990) discussion of
the action tier dealing with actor-patient relations, and Van Valin’s (1993)
discussion of semantic roles.
According to Fillmore (1968),! agent was viewed as one among several
case notions in “a set of universal, presumably innate, concepts which iden-
tify certain types of judgments human beings are capable of making about
the events that are going on around them’ (p. 24). The idea behind Fillmore’s
case grammar was to provide a bridge between descriptions of events and
underlying syntactic representations. It offered “a level of linguistic organi-
zation at which universal properties of lexical structure and clause organi-
zation were to be found” (Fillmore, 1977, p. 62).
Because my discussion of the child language literature draws upon the
sorts of categories that are central to Fillmore’s proposal, I present Fillmore’s
(1968) summary of his six main case categories (pp. 24-25):

Agentive (A), the case of the typically animate perceived instigator of the
action identified by the verb.
Instrumental (1), the case of the inanimate force or object causally involved
in the action or state identified by the verb.
Dative (D), the case of the animate being affected by the state or action
identified by the verb.
Factive (F), the case of the object or being resulting from the action or state
identified by the verb, or understood as a part of the meaning of the verb.
Locative (L), the case which identifies the location or spatial orientation of
the state or action identified by the verb.
Objective (O), the semantically most neutral case. . . . The term is not to be
confused with the notion of direct object.

‘Although Fillmore later revised his original position (see, for instance, Fillmore,
1977) I
refer exclusively to the 1968 version, because related research in child language
has primarily
drawn on Fillmore’s 1968 paper.
AGENTIVITY AND CONTROL 27

Central to the discussion of case grammar is the idea that there is no one-
to-one match between particular case categories and specific surface rela-
tions such as subject or object. With regard to English, we find contrasts
like Charles opened the door with the key and The key opened the door. In
these examples, we find that both the agentive case and the instrumental
case can occur in subject position.
A second approach to the notion of agent as a more or less discrete category
can be found in two theoretical perspectives that differ along other dimen-
sions. These perspectives are those of Role and Reference Grammar (RRG)
(see Foley & Van Valin, 1984; Van Valin, 1993) and Conceptual Semantics (CS)
(Jackendoff, 1990). What is similar about the treatment of agent in these two
seemingly different theoretical perspectives is the idea that conceptual roles
are best viewed in terms of tiers. In contrast to Fillmore’s single tier approach,
both RRG and CS propose two tiers of conceptual structure. According to RRG,
there are two layers of semantic structure: microroles and macroroles. Micro-
roles consist of case roles or thematic relations (Fillmore, 1968; Gruber, 1965),
of which agent is one. Within the framework of RRG, agent is defined as a
willful instigator of an action. Contrasting with the specific microroles are two
macroroles. Macroroles are “generalized semantic relations between a predi-
cate and its arguments” (Foley & Van Valin, 1984, p. 29). The actor, according
to RRG, can be characterized as “the argument of the predicate which
expresses the participant which performs, effects, instigates, or controls the
situation denoted by the predicate” (p. 29). The second macrorole is referred
to as the undergoer and is aligned with the argument “which expresses the
participant which does not perform, initiate, or control any situation but rather
is affected by it in some way” (p. 29).?
Jackendoff (1990) proposed a similar multilevel analysis within the analy-
sis of causation. According to Jackendoff, conceptual roles are distributed
in two tiers: “a thematic tier dealing with motion and location, and an action
tier dealing with Actor-Patient relations” Jackendoff, 1990, p. 126). Adding
this action tier allowed Jackendoff to break down the notion of agent into
three distinct components, which included instigation of action, volitionality,
and the more general notion of doer.

Agentivity as a Nondiscrete Semantic Category

In the preceding discussion of agentivity as a discrete category, I have noted

that some of the more recent analyses of agentivity have assumed a multilevel

2The macroroles of actor and undergoer as well as the case roles/thematic relations can be
viewed in terms of the hierarchy depicted in (i) (see Van Valin, 1993):

(i) ACTOR UNDERGOER

> ec
Agent .. . Effector . . . Locative .. . Theme. . . Patient
28 CHAPTER 2

view of agent. This is central to the view of agentivity as a nondiscrete

linguistic category. Drawing upon prototype theory (Rosch, 1978), the claim
is that linguistic categories such as agent are best defined in terms of a
cluster of related notions in which some exemplars of a given category will
be better members than others. With regard to the notion of agentivity,
Lakoff (1977, p. 244) suggested the following cluster of properties for pro-
totypical agent-patient sentences, based on his analysis of English data:

There is a(n) agent, who does something.

There is a patient, who undergoes a change to a new state.
The change in the patient results from the action by the agent.
The agent’s action is volitional.
The agent is in control of what he does.
aoe The agent is primarily responsible for what happens Chis action and
aon
es
OO
the resulting change).
The agent is the energy source in the action.
There is a single event.
NI09 There is a single definite agent.
10. There is a single definite patient.
11. The agent uses his hands, body, or some instrument.
12. The change in the patient is perceptible.
13. The agent perceives the change.
14. The agent is looking at the patient.

According to Lakoffs proposal, agent-patient sentences will vary in terms

of the number of properties exhibited. The more properties a particular
sentence displays, the more prototypical the sentence is said to be.
Lakoff and others have since argued that the notion of agent cannot be
defined in terms of a set of necessary features, suggesting instead that agency
involves a number of semantic categories including volition, animacy, and
control. Hopper and Thompson (1980) claimed that agency is part of a
broader cluster involving transitivity. They suggested a set of 10 parameters
listed in Table 2.1, all of which have to do with the effectiveness of trans-
ferring action. Each parameter was viewed along a continuum ranging from
high to low transitivity; the more an action is transferred, the higher the
transitivity rating. Based on extensive examination of several languages,
Hopper and Thompson concluded that morphosyntactic markings are sen-
sitive to “transitivity as a whole, rather than the actual presence or absence
of a second participant” (Hopper and Thompson, 1980, p. 254). These regu-
larities were said to be motivated by broader discursive functions.
AGENTIVITY AND CONTROL 29
TABLE 2.1
Transitivity Parameters
(from Hopper and Thompson, 1980, p. 252)
——————— a a ee a ess

Parameter High Low

Ea Tbe emcee ete pee dh, Vere evee) el et hits Sb ae eect de
Participants 2 or more 1 participant
Kinesis action nonaction
Aspect telic atelic
Punctuality punctuality nonpunctuality
Volitionality volitional nonvolitional
Affirmation affirmative negative
Mode realis irrealis
Agency A high in potency A low in potency
Atfectedness of 0 0 totally affected A not affected
Individuation 0 highly affected 0 nonindividuated

A= Agent, 0 = Object.

DeLancey (1984) claimed that agentivity is a major subcomponent of the

transitivity scene:

The general claim is that there is a cross-linguistically valid prototype for true
transitivity which involves (among other things) a direct causation schema
with proximate and ultimate cause both residing in the same volitionally acting
causer. The prototype definition of agent is part of this schema, i.e., the
prototypical agent is just such a volitional causer. Deviations from the semantic
prototype are coded by deviations from prototypical transitive morphosyntax.
(p.185)

Cross-linguistic research has revealed that languages tend to mark proto-

typical agents with different linguistic devices from those marking nonpro-
totypical agents. Nevertheless, a review of this literature reveals that
languages differ in what counts as a prototypical agent. For instance, lan-
guages differ on whether natural forces are given agentive marking or not
(see DeLancey, 1984). In addition, languages differ in the range of devices
used to mark prototypical agentivity and various deviations. Two of the
major kinds of resources employed to mark such distinctions are case mark-
ers and voice (see Barber, 1975; Berman, 1979; Comrie, 1981; Croft, 1987,
1990; DeLancey, 1984, 1987, 1990; Eilfort, Kroeber, & Peterson, 1985; Giv6n,
1984a). Furthermore, such research has revealed that speaker perspective
often determines what gets prototypical marking and what does not (cf.
Barber, 1975; Berman, 1979, for illustration).
Comparing the two views of agents as either discrete or nondiscrete
categories, two major distinctions can be found. First, with regard to the
nondiscrete position, agentivity is viewed in terms of prototypicality rather
than as a unitary category. A second major distinction concerns whether
one posits that the semantic categorization of agentivity is reflected in par-
30 CHAPTER 2

ticular morphosyntactic patterns. Central to the prototype approach is the

claim that languages of the world mark prototypical agents differently from
deviations from the prototype. Researchers who have viewed agents as a
discrete semantic category have made the opposite claim; that is, case gram-
marians have highlighted the lack of correspondence between semantic and
morphosyntactic levels of analysis. In the section that follows we see that
these differences result in quite different views about the role of the category
of agent in language acquisition.

DEVELOPMENTAL PSYCHOLINGUISTIC PERSPECTIVES

In this section I consider the notions of agency and control within the child
language literature. Specifically, I first review literature that has dealt with
semantic approaches to agentivity. Here one can draw a distinction similar
to that discussed earlier between the treatment of agentivity as a discrete
semantic category and the treatment of agentivity in terms of prototypicality.
Next, I consider how agency has figured in recent discussions of pragmatic
and activity-type approaches to child language. In these later sections, focus
will shift from viewing agentivity in terms of purely physical causation toward
a consideration of agentivity in terms of social control.

Semantic Approaches to Agency

Agentivity as a Discrete Semantic Category. There seem to be two

very broad kinds of treatments of agentivity as a discrete semantic category
in the developmental psycholinguistic literature. One group of researchers
has followed up on the case grammar approach, and agent has been viewed
as One among many case categories in children’s early speech. This group
has been interested primarily in the semantic content of children’s early
word combinations. A second group has followed up on the work of Jack-
endoff (1990) and Van Valin (1993) reviewed earlier and has examined the
notion of agent at the level of thematic relations. I now review the work of
the first group, because this group has generated the most empirical work
relevant to the analyses reported in Part II of this book.
There have been dozens of proposals in the literature concerning the
semantic content of children’s early speech. Although the specifics of the
various proposals differ, what is similar is the belief that children’s early
word combinations are best viewed in terms of semantic caselike categories
rather than syntactic categories (i.e., subject, object). I briefly examine three
such categorization schemes. Though others could have been selected, these
have been included because they highlight some relevant distinctions in the
way the notion of agent has figured in these proposals.
AGENTIVITY AND CONTROL 31

Brown (1973) applied Fillmore’s case categories to the longitudinal data

of Adam and found that all six of Fillmore’s basic categories could be applied
to Adam’s earliest multiword utterances. In considering a wider corpus of
data, Brown suggested that three of the four most common kinds of semantic
relations in children’s early speech relate to the notion of agentivity. These
three categories include: (a) agent-action (Adam write, Daddy go), (b) ac-
tion-object (Change diaper), and (c) possessor-possessed (Adam chair, My
pencil). Brown used a notion of agent that was not restricted to animate
instigators, though he noted that most agents in the various samples were
in fact animate. One can also note that Brown’s category of possessor-pos-
sessed would be classified under Fillmore’s more general category of Da-
tive-Objective, though Brown suggested that the frequent appearance and
productivity of possessor-possessed word combinations led him to create
this special category. Brown suggested that children are very interested in
talking about who carries out actions, the kinds of changes undergone by
objects by virtue of actions carried out on them, and the distinction between
rights to control and association with particular objects. It should be em-
phasized that in working with these case categories, Brown, like Fillmore,
was not making any specific claims about the relationship between particular
surface forms and the underlying semantic case relation.
Bloom, Lightbown, and Hood (1975) also considered semantic aspects
of children’s early word combinations. The categories they posited did not
follow Fillmore’s so closely but, rather, were said to “emerge from the data”
(p. 9). For the present purposes, we can note that Bloom et al. differed from
Brown in that they distinguished between an action category, which might
refer to “action that affected an object with movement by an agent” (p. 10),
and a locative category, referring to actions “where the goal of the movement
was a change in the location of a person or object” or “where the agent
and affected object or person were the same” (p. 10). Examples of this
contrast include My open that, Gia ride bike, and I made, as opposed to
Up Kathryn (=child’s own name) and J get down. Bloom et al. claimed that
the action category, which involves the relation between agent, actions, and
objects, is central early in development.
Although Bloom et al., like Brown, claimed that the various children
studied all made use of similar semantic categories, Bloom et al. differed to
the extent that they claimed that at a very early phase (before the children’s
MLU [mean length of utterance] was 2.0), the children could be distinguished
in terms of the range of forms used—some children relied on pronominal
and others on nominal forms. The variation in nominal-pronominal usage
has been taken to indicate that the children are in the earliest phase of
learning different kinds of syntactic-semantic systems. It can be noted,
though, that Bloom et al. did not attempt to link the use of specific forms
with specific meaning categories.
32 CHAPTER 2

One final proposal highlights yet another way the notion of agentivity has
figured in proposals concerning children’s early word combinations. Braine
(1976), in examining data from children acquiring several languages, came to
the conclusion that the relational categories of young children are far narrower
than previous child language research led one to believe. Of particular
relevance here is his discussion of the actor-action category. Braine noted that
Fillmore’s agent category was more sophisticated than the actor category he
posited. Braine argued that the sort of relations that children talk about early
on do not involve the kinds of causal changes Fillmore referred to. Instead,
Braine claimed that children talk about “simple actions like sitting, running,
eating” (Braine, 1976, p. 84). For this reason, Braine referred to an actor—
namely, one who carries out an action-rather than an agent.’
Although there is some agreement among the various semantic proposals
that agentivity is a notion expressed in children’s early word combinations,
what remains unclear from a comparative analysis of the various proposals
is the nature of the notion of agency that is expressed. Some researchers
have attributed to children fairly sophisticated notions of agency similar to
those referred to in Fillmore’s case grammar, whereas others have been
more cautious in suggesting children may simply be referring to a notion
of animate doer. With regard to these various proposals, one of the major
questions that has been raised concerns the issue of continuity: If children
begin combining words based on semantic patterns such as agent-action,
how is it that they ultimately arrive at syntactic categories such as subject-
verb? I do not pursue this issue here but, rather, turn to another way agent
has been viewed in the child language literature, namely, as a nondiscrete
semantic category.

Agentivity as a Nondiscrete Semantic Category. Although I have

noted very fundamental differences in the ways agentivity has been viewed
in my discussion of child language theorizing thus far, what has held these
various views together is the idea that agent is a discrete category. As I
noted in the discussion of linguistic perspectives on agency, much attention
in the linguistic literature has worked with an alternative conception of
agentivity, namely, that it is best viewed in terms of a cluster of related
semantic parameters. I now consider one semantic approach to child lan-
guage that has drawn upon this claim.
Over the past decade, Slobin (1981, 1982, 1985) has suggested that par-
ticular prototypical events provide a framework for children’s early organi-
zation of linguistic forms. As was noted in chapter 1, certain notions were

3It is of interest that Braine commented on the role of pragmatic purpose in children’s early
word combinations—a point I retum to later on in this chapter. Note that Braine was not
interested in the relation between utterance function and the semantic meaning categories he
proposed.
AGENTIVITY AND CONTROL 33

said to be the first to receive linguistic treatment by children across the

world. One such notion was that of linguistic transitivity. Based on a review
of typologically distinct languages, Slobin provided intriguing evidence that
children acquiring structurally different languages begin to use grammatical
forms to mark a scene he referred to as the Manipulative Activity Scene.
This scene involves “the experiential gestalt of a basic causal event in which
an agent carries out a physical and perceptible change of state in a patient
by means of direct body contact or with an instrument under the agent’s
control” (Slobin, 1985, p. 1175). Children were said to extract specific surface
markers from the input language; and though the specific marker selected
will vary from language to language, Slobin suggested that in all languages
children will attempt to grammaticize something like the Manipulative Ac-
tivity Scene. In particular, based on acquisitional data of a child acquiring
Russian (Gvozdev, 1949) and children acquiring Kaluli (Schieffelin, 1985),
Slobin noted that the children studied similarly restricted the use of accusative
and ergative inflections. At an early phase of development the children
restricted the use of the language-relevant inflections to the Manipulative
Activity Scene, rather than using the accusative marker in Russian to mark
all direct objects and the ergative inflection in Kaluli to mark subjects of
transitive verbs.
Slobin’s analysis of agentivity differs considerably from that in the devel-
opmental psycholinguistic literature discussed previously, in which agency
was viewed as a discrete category. In addition to viewing agentivity in terms
of a prototypical scene, Slobin also attempted to relate the use of specific
linguistic devices to this scene. In fact, it was Slobin’s claim that this proto-
typical scene plays an organizing role for the child as an entry into grammar.
Although Slobin’s analysis of children’s special linguistic treatment of the
Manipulative Activity Scene is quite similar to the analysis discussed here
of transitivity and agentivity as a nondiscrete semantic category in the lin-
guistic literature, it is important to note the following differences. First, in
Slobin’s discussion of the Manipulative Activity Scene, the notion of volition
was not mentioned. In addition, the marking of various kinds of deviations
from the agentive prototype did not receive focus. What also remains unclear
is the relation between Slobin’s proposed prototype and the language-spe-
cific differences noted by DeLancey (1984, 1990) and others in what counts
as a prototypical agent and what counts as relevant deviations. This is a
topic I return to.

Pragmatic Approaches to Agency

Thus far I have been considering the notions of agentivity and control
primarily from the standpoint of physical causation. It seems, though, that
by limiting the discussion of agentivity to physical causation, one misses out
34 CHAPTER 2

on a related set of issues concerning social causality. As I have noted in the

first chapter of this book, language does not merely reflect reality; speakers
use language to achieve social goals. In this section, I consider some of the
literature on language development that focuses on children’s growing use
of language for social purposes. Our focus will be on the potential relation-
ship between language structure and the particular goals of the type of
action the children talk about.
There is much evidence that before children begin combining words in
multiword utterances, they draw a linguistic distinction between utterances
that function to bring about action and utterances that match or describe
the world (cf. Halliday, 1975; Painter, 1984). In addition to using contrasting
intonation patterns, children have been noted to use particular gestural and
phonetic forms systematically with specific communicative sensorimotor
schemata. As was noted in chapter 1, Carter (1975, 1978a, 1978b), for in-
stance, suggested that children will contrastively use particular forms de-
pending on whether they wanted an action to be carried out by a partner
or whether they wanted to bring attention to a particular object. In chapter
1 I noted findings from Deutsch and Budwig (1983) that also suggested that
children organize the use of specific linguistic forms around the notion of
social control. Studies of older children’s use of form contrasts have also
indicated their continued interest in dimensions of social control as a way
to organize linguistic forms. Shepherd (1980) claimed that preschool-aged
Antiguan Creole-speaking children made an innovative contrast when using
the modal forms gon and go. Although adult speakers of this English-based
creole appear to use the forms nondistinctively, Shepherd claimed that the
children distinguish between the two forms based on locus of control.
Gordon and Ervin-Tripp (1984) also noted that the preschool English-
speaking boy they studied linked the use of three different request forms
CU want vs. supposed to and have to vs. imperatives) with particular instru-
mental goals (i.e., requests oriented to rules and norms vs. requests in the
context of joint activities). This sort of form-function contrast highlights the
extent to which children’s use of linguistic forms is organized around various
kinds of social motivations for action.
Gordon, Budwig, Strage, and Carrell (1980) revealed that with continued
development, children draw even further formal contrasts based on their
growing ability to attend to the viewpoint of the listener. To this extent,
both social motivation and social relationships are indexed in the use of
particular linguistic devices. It is exactly these broad sorts of social distinc-
tions that have not been considered in previous semantic discussions of
agentivity. At the conclusion of this chapter, I return to a consideration of
more specific suggestions concerning how such social aspects of causation
schemes might be integrated into the notion of agentivity. First, I consider
one more area of research in which agentivity has been said to play a role
in acquisition.
AGENTIVITY AND CONTROL 35

Activity-Type Approaches to Agency

The activity-type approach is quite similar to the prototype and pragmatic

approaches reviewed here. Central to the activity-type approach is the claim
that children’s use of particular grammatical constructions can be understood
only by a consideration of the broader discursive contexts in which they
are embedded (see Gee, 1985; Gee & Savasir, 1985; Gerhardt, 1990). The
notion of activity-type is similar to related notions used by Ervin-Tripp (1972),
Gumperz (1982), and Levinson (1979).‘ It also can be related to the notion
of Scene as employed by Fillmore (1977) and Slobin (1985), which can be
viewed as involving landscapes within which people act, although activity-
types are more dynamic units that also take into account the ongoing inter-
active frame. The activity-level construct views under one heading a variety
of verbal and nonverbal properties that pattern together in ongoing action.
I now consider some findings from research that suggests that activity-types
relate to children’s growing notions of agentivity and control.
In a study of various kinds of future markers in children’s acquisition of
Turkish, Savasir (1984) argued that children contrastively employ future
markers in ways that deviate from adult Turkish. Although at one level the
children’s usage is deviant, particular forms are said to be consistently and
meaningfully employed by the children. As one example of the kinds of
contrasts Savasir noted, consider the activity-types associated with the use
of the aorist and optative markers. Savasir claimed that the aorist form oc-
curred in first and third person reference in situations where he found the
child to be taking an eventive perspective on an action frame. Although the
child or some other participant might have been acting as agent, focus was
placed on the role of experimentation. The shift was away from agency.
Similarly, the optative was viewed as a marker of nonprototypical agency.
The optative was found in two related contexts—either in accounts for which
the motivation for action was due to knowledge of rules of the game or in
contexts in which the child’s actions were adult motivated. These findings
highlighted the fact that the children were distinguishing different ways to
contextualize their own and others’ actions that produced changes of state
in objects. Although all examples could be viewed at the semantic level as
agentive, a more refined analysis revealed differences in the kinds of motives
for the action; the two forms appeared in different discourse contexts.
The sorts of parameters included in the activity-types suggested by Savasir
are similar to those reviewed in chapter 1 with regard to Gerhardt’s work on
modal forms. Such research indicates the importance of including a variety of
parameters that run orthogonal to the parameters of agentivity and control,

‘although Levinson (1979) used the same term, Levinson considered activity-types as
macro-level categories having to do with institutionally agreed upon activities (see Gee &
Savasir, 1985, for further discussion).
36 CHAPTER 2

as
discussed with regard to the semantic approach to agentivity. Issues such
volition and responsibility, as well as the communicative function of the
utterances under consideration, are only parts of the constellations children
seem to link with linguistic forms. Gee and Savasir pointed toward the
importance of also considering a variety of social factors, such as whether
agents are acting in order to carry out their own goals or interpersonal goals.
Ireturn to some specific implications of the activity-type approach for the study
of agentivity and control in English-speaking children’s first word combina-
tions. First, though, I briefly consider one last kind of relevant literature,
namely, that concerning psychological studies of the development of chil-
dren’s conceptions of agentivity and control.

DEVELOPMENTAL PSYCHOLOGICAL PERSPECTIVES

Much of the developmental psycholinguistic literature starts from the as-

sumption that children who are just beginning to combine words have
relatively stable conceptual notions about agency. The question I raise here
is whether such a claim can be supported when one examines research
findings concerning children’s understanding of the concept of agency. I
have noted that agency can be viewed as a multidimensional concept that
involves an understanding of causality, animacy, and intentionality. In what
follows, I briefly consider some literature from developmental psychological
research. This work suggests that previous psycholinguistic research may
have presupposed categories that are still developing in the child who is
just beginning to acquire language. After reviewing this literature, I consider
the implications of such findings for theorizing about the relation between
children’s conceptual notions of agency and their development of language.
Piaget (1930, 1954) wrote extensively on the emergence of children’s
notions of causality. His central argument was that children’s notions of
causality are gradually constructed over the course of several years. Piaget’s
(1930) early claims that preschool-aged children lack causal knowledge, as
evidenced in their inability to reason about causal mechanisms, has led to
a multitude of studies that have attempted to show that preschoolers function
similarly to adults with regard to both mechanical (see Bullock, 1979) and
social (Shultz, 1980) causality. A more specific understanding of children’s
developing notions of causality emerges from a review of children’s growing
conceptions of animacy and intentionality. Much research has been directed
toward children’s ability to distinguish between animate and inanimate cau-
sation and toward children’s growing understanding that some actions are
carried out intentionally, whereas others are not.
One aspect of understanding agency is the ability to distinguish animate
and inanimate causation. Several different lines of research have indicated that
AGENTIVITY AND CONTROL 37

before children begin combining words, they understand what animate and
inanimate objects can do (see Golinkoff, Harding, Carlson, & Sexton, 1984, for
a review). These studies show that children as young as 2 years show surprise
when they are presented with inanimate objects that move on their own (i.e.,
a chair suddenly moving without an external agent assisting).
With regard to understanding animate causation, much discussion has
focused on the distinction between self and other. Piaget (1954) claimed that
infants attribute causality to self actions before attributing independent cau-
sality to others. A similar self—other distinction has been upheld in numerous
experimental studies (see Huttenlocher, Smiley, & Charney, 1983; Poulin-
DuBois & Shultz, 1988; Wolf, 1982; Wolf, Rygh, & Altshuler, 1984). Most
relevant to our present concern with language development is a series of
studies conducted by Huttenlocher et al. In three separate studies of children’s
verb meanings, these researchers found that the verbs children first used
encoded self actions. Only later did the children use verbs to encode the
actions of others. Huttenlocher et al. claimed that at the time the children in
their study were beginning to combine words into multiword utterances, the
children rarely produced verbs spontaneously to refer to others’ actions; and
the same finding was said to hold for comprehension. Based on these findings,
Huttenlocher et al. concluded that children pass through distinct developmen-
tal stages in understanding self and other as actors. This finding is particularly
intriguing in light of the developmental psycholinguistic research reviewed
earlier that presupposed that children’s semantic category of actor encom-
passes both self and other.
A related question is whether children’s understanding of animate beings
acting with intentions need be considered in terms of a distinction between
self and other as agent. Several different lines of developmental research
suggest that such a distinction is necessary (see Huttenlocher et al., 1983;
Poulin-DuBois & Shultz, 1988; Wolf, 1982; Wolf et al., 1984). What these
studies revealed is that children first have an understanding of personal agency
and only later come to understand others as acting with intentions. Although
the studies differed somewhat in the specific age at which they suggested that
children have a notion of independent agency, there was much agreement
that at around the age of 2, the child is primarily working with a notion of
personal agency. The main conclusion of such studies was that the very notion
of agency undergoes development between the ages of 1 and 5. Around the
age of 2, children are working primarily with a notion of personal agency
rather than a more global notion of independent agency.
The findings reviewed here suggest the need for caution with regard to
assumptions concerning young children’s semantic notions of agentivity. As
I noted earlier, agentivity is a multidimensional notion consisting of several
interrelated categories. One particular finding stemming from the develop-
mental psychological literature that is relevant to a consideration of children’s
38 CHAPTER 2

use of agency as a wedge into grammar is that children seem to develop

notions of agency first with respect only to the self and later for others. The
question this raises is whether some of the previous studies of children’s
early word combinations have overestimated the breadth of children’s early
semantic categories (see also Braine, 1976; Huttenlocher et al., 1983, for a
similar discussion). In the section that follows I explore the implications of
such a claim.

A FRAMEWORK FOR STUDYING AGENTIVITY

AND CONTROL IN EARLY CHILD LANGUAGE

In this chapter I have examined several different studies concerning agentivity

and control. Based on areview of this work stemming from related disciplines,
the choice to focus on agentivity and control seems relevant for numerous
reasons. First, there is a good amount of evidence that such notions are
consistently marked in the languages of the world and play a central role in
the organization of grammar. Second, there is much evidence from studies of
children’s first word combinations that children spend a great deal of time
talking about agentivity and control. If children are going to employ linguistic
devices to mark salient conceptual notions, the choice of agentivity and control
seems relevant. Finally, our review of cross-linguistic evidence suggests that
children acquiring strikingly different languages attempt to mark linguistically
the various perspectives they take on human action.
Having reviewed a number of separate lines of research that consider
the notions of agentivity and control, we are now in a better position to
raise questions pertaining to the linguistic marking of agentivity and control
in children acquiring English. Although I have noted that the majority of
developmental psycholinguistic studies have yet to examine the extent to
which particular linguistic devices may reflect children’s attempts to mark a
cluster of notions concerning agentivity, Slobin’s cross-linguistic research
illustrated that children acquiring inflectional languages seek to link particu-
lar devices with a prototypical manipulative scene. This leads to the question
whether children acquiring English also mark such notions in the early
phases of language development. In the present investigation, I elaborate
on the issue of the linguistic marking of prototypical agentivity by assessing
whether children mark various deviations from the prototypical agentive
scene as well. Thus my concern is with the development of systems for
marking various aspects of causation schemes.
The research concerning transitivity, agentivity, and prototype semantics
reviewed here suggests that the categories children are working with may
not be discrete in nature. I start from the assumption that young children
are working with categories. with prototypical structure not only because
AGENTIVITY AND CONTROL 39

these have been suggested for adult linguistic systems, but also because my
examination of the social-cognitive developmental literature suggested that
the very notions said to cluster with the notion of agentivity are themselves
developing at the time grammar is emerging. It therefore seems likely that
children are attempting to link specific forms with various aspects of causal
vectors, and that they are working with form-meaning pairs that are built
around a prototype, rather than with unitary categories such as agent-action
relations.’ In particular, I am interested in whether specific linguistic forms
link up with prototypical agentivity and various deviations. For instance,
Brown (1973) placed instances like Adam write and I no fall in the same
Verb + Agent category. Likewise, Bloom et al. (1975) categorized utterances
like My get magazine, My open this, and I turn the light on as falling within
the Action category. The question of interest here is whether the forms are
used contrastively to refer to prototypical agentivity and various deviations.
In examining the children’s use of forms in terms of whether they mark
various notions of agentivity and control, I depart from a purely semantic
analysis and examine whether children might also integrate a set of related
notions about social causation into the agentive prototype.
Although these questions guide the present study, no specific predictions
can be made about where English-speaking children might decide to mark
such notions. Some subsystems of grammar, however, seem more likely
candidates than others. A review of the linguistic literature concerning the
linguistic marking of agentivity, as well as Slobin’s cross-linguistic findings
regarding young children’s marking of prototypical agentivity, suggests that
case markers often are employed to mark shifts in perspective on agency.
English, though, does not mark nouns with case inflections and only has
limited case marking in pronominal forms. Deutsch and Budwig’s (1983)
study indicated that children contrastively employ first person possessive
pronominal forms and nominal forms to mark a contrast in degree of prag-
matic control; yet other pronominal forms were not examined. The research
on Turkish- and English-speaking children conducted by Savasir (1984) and
Gerhardt (1990) suggested that tense, aspect, and modality markers are also
employed to contrast betweén different perspectives on participants’ involve-
ments in fields of action. What all these subsystems share is the extent to
which they relate the speaker’s perspective. Although we cannot firmly
predict where in the linguistic system children acquiring English might mark
notions of agentivity in their early word combinations, it seems that they
would be likely to draw upon linguistic structures that have been noted to

°One might suggest that the sort of tiered approach offered by Jackendoff (1990) and Van
Valin (1993) could also handle developmental aspects of the broader category, though in this
view one would need to explain why such information would be ignored at one phase and
included at another—thus the tiered approach is not an ideal solution.
40 CHAPTER 2

link up with speakers’ expression of an inner attitude (see Benveniste, 1971;

Kuroda, 1973; Lyons, 1982).
Although it is expected that 2-year-old children mark the notions of
agentivity and control in the earliest phases of grammatical development,
examination focuses on the breadth of the children’s categories. My review of
the developmental psychological literature suggests the need, before assum-
ing that children are marking a broad notion of agent, to attend to whether the
children first work with a more refined notion of personal agency. The
following set of research questions summarizes the focus of the present study:

1. What distinctions concerning agentivity and control are relevant to

children’s organization of linguistic forms in the early phases of ac-
quiring English?
2. Do English-speaking children make use of particular linguistic forms
to mark something like a prototypical agentivity scene?
3. If so, what counts as prototypical agentivity?
4. How, if at all, do children mark deviations from a prototypical agen-
tivity scene, and what distinctions are marked? What broader systems
of agentivity marking are children organizing?
5. What role does pragmatic function play in the linguistic marking of agen-
tivity and control? Are children attempting to relate forms to broader
constellations of meaning that cut across semantic and pragmatic pa-
rameters?
6. Do young children begin with a broad notion of agentivity or do they
first organize forms around notions of personal agency?
7. What happens across development? If, for instance, children’s early
systems do not completely match those of adult speakers of English,
what developmental paths are traveled en route to an adultlike system?
8. What commonalities and differences are found among various children
acquiring English?

In the chapters that follow, I deal with these questions. In chapter 3, I discuss
methodological issues related to the study. In chapter 4, I review some
preliminary analyses; and in chapters 5 through 8, I discuss the major findings
regarding the eight questions raised here.
 CHAPTER THREE

Methodology

In this chapter I consider various aspects of the overall design of the research
reported in subsequent chapters of this book. In addition to covering issues
such as the selection of subjects and the kinds of tasks involved, I also
briefly consider two preliminary phases in arriving at the present research
paradigm. A central theme in this chapter is that the research environment
can be structured so that the data are ecologically valid, as well as directed
at the particular question to be examined.
The findings to be reported in subsequent chapters are based on an
investigation making use of a naturalistic design. Focus is on children’s
spontaneous linguistic productions, looking in particular at ways in which
the children being studied marked the notions of agentivity and control. In
gathering the data, any naturalistic study is confronted with the issue of
sampling. What activities should be sampled? What partners should interact
with the child? Where should the study take place? With regard to the present
study, a decision was made to collect data in settings and with partners that
would provide a maximum amount of language productions involving talk
about agentivity and control.
In order to carry out the kind of distributional analyses and in order to
sort out the interrelationships that exist between form and semantic and
pragmatic function across a span of development in the detailed manner
documented in Part II of this book, one must have a sufficient corpus that
can provide a rich basis for such analyses. In making choices about setting,
partners, and activities, the aim was to design a study that utilized a subset
of activities that 2-year-old children participate in and at the same time
tapped a range of talk about the issues of agentivity and control. The design

41
42 CHAPTER 3

was intended not merely to permit the observation of children doing what
they naturally do but to create a natural environment within which talk
about agentivity and control could unfold. I now consider two preliminary
phases that contributed greatly to the formulation of the design that is later
discussed further.

PRELIMINARY PHASES
OF DESIGNING THE INVESTIGATION

Analysis of Other Data Sources

The first step in designing the present study involved working through
transcripts stemming from other research projects concerning the natural
speech productions of 2-year-old children. In so doing, I made use of three
different sources of information: (a) transcripts from children in Brown’s
(1973) longitudinal sample, (b) transcripts from naturalistic interactions of
three children in Ervin and Miller’s longitudinal corpus (Ervin & Miller, 1964),
and (c) transcripts and video taped recordings stemming from a longitudinal
investigation of mother-child and peer interaction of children between 2
and 3 years of age (Budwig, Strage, & Bamberg, 1986). The point of working
through these sources was to isolate the sorts of activities that led children
to talk about various kinds of agentivity and control. When did the children
talk about personal agency? When did they refer to others’ involvements in
action frames? Did particular kinds of settings, partners, or activities lead to
talk about social control?
An extensive analysis of these data sources proved helpful in the following
ways. First, it became clear that talk about agentivity was not limited to
particular kinds of tasks. Children talked about their own ability to carry out
actions both when engaged in play activities and when no particular activity
was taking place. Nevertheless, the amount of extended discourse found in
different kinds of activities varied tremendously. It also became clear that
there were differences in the kinds of talk about physical and social causation
that took place when the children were interacting with a peer versus a
caregiver. Within the context of play with peers, verbal interactions involved
numerous claims to gain or maintain control of objects. In contrast, in the
context of caregiver-child interactions, disputes over control of objects oc-
curred rather infrequently and were primarily over control of objects that
were in limited quantity and of high value to the adults. Finally, although
the Brown and the Ervin and Miller studies took place in a single setting,
namely, the child’s home, the data sources involving peer and adult inter-
actions revealed differences in the nature of the children’s (and caregivers’)
talk when they interacted in their own home versus that of the peer (see
Budwig, 1981, for further discussion).
METHODOLOGY 43

On the basis of this examination of other transcripts, a series of preliminary

decisions could be made concerning the design of the study. First, the
limitation in working only with transcripts made clear the need to use video
for the present study. Although at times contextual notes were sufficient, at
other times it was unclear whether children were claiming objects or referring
to an object they possessed. Consider the example where the child says My
purseas the researcher and parent converse. The contextual notes suggest that
the child is approaching the purse. In such an instance, it is unclear whether
the purse belongs to the researcher and the child wishes to claim it, or whether
the purse is the child’s own and the child is pointing it out to the adults.
Although at times the surrounding discourse adds clarification, often the lack
of video meant that potentially useful material had to be ignored.
A second decision that came out of the analyses of other longitudinal data
sources involved the importance of including peer data when studying the
early phases of grammatical development. Though most previous studies of
the early phases of grammatical development of English-speaking children
have almost exclusively focused on adult-child interactions, it seemed impor-
tant to include interactions with peers as well. This point was stressed by
Demuth (1984), whose ethnographic study of Basotho children’s spontaneous
speech revealed that relative clause usage was highly affected by the kind of
discourse partner with which one interacted. Demuth reported that relative
clauses were first produced by the children in peer interactions and were not
used in the context of adult-child interactions where the caregiver tended to
supply much contextual information. Ervin-Tripp (1977) also pointed out that
bursts of syntax are more likely to occur in contexts where pressure to
communicate is strong. In the present study, the decision was to include
interactions both between caregiver and child and between peer and child.
A third issue that was affected by examination of other longitudinal re-
search involved the selection of the setting. The decision was made to pick
an environment that was equally familiar to both children. Especially with
regard to the issue of rights and claiming possession, this seemed important
in order to make developmental comparisons. Although there are no neutral
settings, the best option seemed to be to conduct the study at the day-care
center that the children attended regularly and where their caregivers vol-
unteered on a regular basis. Having narrowed down the focus to a particular
setting and range of interactive partners, the second round of preliminary
decision making concerning design could begin.

Entering the Day-Care Center

In selecting a day-care center where the study could take place, it seemed
important to choose an environment that could physically accommodate the
study (e.g., a quiet room for observations that was not foreign to the children).
44 CHAPTER 3

It also was important to select a setting in which the staff was excited and
cooperative about the prospect of an ongoing study. After locating a day-care
center that met these demands, I spent 2 months as a participant observer and
teacher’s aide before actually starting the study. This was in accordance with
an agreement worked out with the day-care staff. In addition to lending a hand
at supervising activities, changing diapers, and cleaning spilled juice, I used
the time to plan out the study. The children also had the chance to become
familiar with me and my role at the center. I attended staff meetings to keep
up with the range of problems confronting the children and had the chance
during such meetings to develop rapport with the staff. During this period, I
kept field notes on activities that elicited talk between peers and piloted
activities based on notes made from scanning transcripts of the longitudinal
studies reported earlier. Many of the procedures used were based on those
outlined by Corsaro (1985). Once subjects were selected, specific tasks were
piloted to test attention span and interest. At this point, I began keeping diary
notes on the subset of children who would participate in the study, while they
were engaging in ongoing activities at the day-care center with their peers and
with adults. I continued to participate at the day-care center on a regular basis
throughout the duration of the study, using the time to continue with diary
observations, attend staff meetings, and keep in contact with the children’s
parents.

DESIGN OF THE STUDY

Subjects

Six children who attended the center on a regular basis participated in the
study. These children were selected from a larger pool of children whose
parents had indicated interest in participating in the study. All parents were
informed by letter of the study and were asked to return a permission form if
they were interested in having their children participate. They were notified
that one caregiver would also be involved once a month in the study and that
either the mother or the father could participate. Decisions concerning which
children were to be included in the study were based on a number of factors.
First, the children all had to be monolingual speakers of English. One caregiver
had to be willing to participate once a month for a video recording session. It
was important the children selected could be paired with a familiar playmate
whom they interacted with on a regular basis and who attended the day-care
center at similar times during the day. It was hoped that all children would be
firstborn; but in order to meet the other requirements, it was necessary to
include one child (Megan) who had a 7-year-old sibling. Sex of the children
was not taken into account in selecting participants; the study included two
girls and four boys. Subjects were not systematically selected on the basis of
socioeconomic status or race. Nevertheless, due to demographic charac-
METHODOLOGY 45

teristics of the center, the group was rather homogeneous. All children were
Caucasian and they came from middle-class families. Because the center was
connected with a university, all of the children’s parents were connected with
the university either as students or as staff. None of the subjects’ parents held
teaching positions at the university.
The children participating in the study ranged between 20 and 32 months
of age at the onset of the study. Because age is not always the best indicator
of linguistic development, the selection of subjects was not based on chrono-
logical age alone. During the pilot phases of the project, samples of speech
data were collected from the children and MLUs were calculated. The young-
est dyad included the two girls, Megan and Grice, who rarely used gram-
matical morphemes and were just beginning to combine words into utter-
ances. The two other dyads consisted of boys: Jeffrey and Keith had begun
to use functors; and Eric and Thomas were well beyond the two-word stage,
having begun to employ tense markers and modal auxiliaries, for instance.
More details concerning the children’s ages and levels of linguistic devel-
opment at the start of the project are documented in chapter 4. With regard
to the issue of intelligibility of the children’s speech, good articulation was
not a requirement for participation in the study. One child, Megan, was at
times rather difficult to understand, whereas Grice pronounced words very
clearly. Given the recent findings concerning the relation between intelligi-
bility of children’s early utterances and the nature of early grammatical
development, it seemed important not to exclude subjects because their
utterances might be difficult to transcribe (see Bates, Bretherton, & Snyder,
1988, for further discussion).

Setting

The study took place in a quiet room located next to the main playroom of the
day-care center. This room was typically used by the teachers as a staff room
where they gathered during nap time. It also housed gymnastic equipment
used on rainy days. The children occasionally were allowed in this room to
carry Out special projects under the supervision of one of the teachers. One of
the advantages to using this room was that the children always looked forward
to being allowed to play there. The room was simply furnished with a couch
and a low table where the children could play while standing. A desk was
located nearby where the teachers often worked while the children played.
In the present study, the children played primarily around the table area.

Tasks

Four tasks were presented each month. This allowed for developmental
comparisons to take place between various children and within particular
children across development. These four tasks were as follows:
 CHAPTER 3
46
the
Watering Plant. This task was included as a way of drawing
of the projec t each
children into the playroom. During the piloting phase
d seeds in a cup that they had decorated with
of the six children plante
en were asked to
crayons and a sticker. Upon entering the room, the childr
Specifically, the
check the small end table where the plants were kept.
other what is on
instructions of the researcher were as follows: “Show each
mom what is on the table” (caregiver
the table” (peer situation) or “Show your
was placed on the table so the children
situation). A small pitcher of water
also provi ded childr en the chance
could water their plant. The plant task
vemen ts in wateri ng the plant. In
to negotiate the various participants’ invol
e one of how much respon-
the caregiver-child situation, the question becam
en wante d to water the plant by
sibility the child could obtain—all childr
that the childr en might spill
themselves, but the caregivers often were afraid
that the
water and thus attempted to monitor the situation. It can be noted
ed depen ding on the situati on.
amount of water placed in the pitcher differ
r during the peer situati on,
For instance, less water was placed in the pitche
whereas in the caregiver-child situation slightl y more water was given to
help provoke negotiation. In the peer situation, the pitche r itself becam e an
r
object of negotiation, because both children wanted to control the pitche
first.

Semistructured Manipulative Play. The idea behind this task was to

involve the children in manipulative play with toys that were thematically
related. Piloting had revealed that thematically related toys led to extended
discourse in both the peer and the caregiver-child situations; in contrast,
giving children unrelated toys led to lots of naming, but little interaction. In
addition, piloting revealed that the size of toys that interested the youngest
and oldest subjects varied, and thus the two youngest subjects were provided
with larger toys to manipulate. The older children were presented with a
large cookie sheet tray of toys that had been arranged like a small village.
The tray included two large mirrors that were partially covered by dried
peas, giving the appearance of a lake surrounded by grass. On top of the
peas stood several cars, two playmobile figures, two bicycles, several toy
houses, and plastic trees. The figures could be attached to the bicycles, and
helmets could be snapped on and off their heads. The cars ranged in size
and moved in various ways. For instance, some of the cars had to be pushed,
whereas others moved on their own if pressed in particular ways. This
provided an opportunity to see how children talk about objects that move
by virtue of effort on the part of a human agent versus those that move on
their own. A few small bowls and spoons were placed next to the village
so that the children could scoop dried peas from one container to another.
To encourage negotiation in sharing control of goods, only one large scoop
was placed next to the spoons.
 METHODOLOGY 47
The motoric development of the two youngest children was not advanced
enough to allow them to manipulate the small objects. Furthermore, they
tended to place objects in their mouths, which meant that objects like dried
peas would be too dangerous. Therefore, the younger children received
larger objects, including an egg carton; several large nuts (walnuts, filberts,
etc.); the same spoons, bowls, and scoop noted earlier; a small teapot stuffed
with cupcake papers; and several containers of various sizes. The containers
were selected so that some of them had no tops, others had tops that the
child would be unable to open without assistance, and others closed when
the container was held in a certain way. The containers, like the cars noted
_earlier, were included to elicit talk about one’s own and others’ role in
carrying out kinetic actions. The teapot and the scoop encouraged discussion
about control of goods. The sorts of toys given to all the children not only
elicited commentary about physical manipulation but also allowed the com-
municative partners to plan cooperative activities (e.g., having a tea party,
making a cake, going to the hospital, going swimming).

Block Play. \n this task, the children were presented with a tray of large
building blocks in assorted shapes and colors. Also included on the tray
was a large toy helicopter. Halfway through the task the children were given
a second tray of smaller blocks, also of various shapes and colors. Blocks
were selected because this sort of activity elicits not only talk about previous
actions but also goal-directed behavior. The children talked not only about
the colors and shapes of the blocks but also about what they were doing
with the blocks, or what they planned to do with them. The helicopter was
added because it provided a guaranteed way of getting children to talk
about social control. The second tray of blocks was added because in piloting
phases, presenting the children with too many blocks at once limited the
amount of talk and shortened the length of time they were willing to engage
in block play.

Photo-Book. The final task involved having the children and their part-
ners look through a photo-book containing pictures of the children taken
at the day-care center in the months just prior to the study. The photo-book
included pictures of the six children interacting with their peers and the
day-care staff in the full range of activities that took place over the course
of the day. Some portrait pictures of the children, as well as some more
action-oriented pictures, were included. In addition, pictures were included
in which the children were depicted in various phases of a particular ac-
tion—for instance, climbing upaslide, sitting at the top, sliding down, and
reaching the bottom.
When looking at the photo-book the communicative partners sat on the
couch next to one another, with the book placed across their laps. The
48 CHAPTER 3

point of the photo-book task was to elicit talk about self and other in
depicted action. Would the children mark depicted actions similarly to on-
going actions? Previous research concerning the acquisition of pronominal
forms has also involved a task in which children looked at pictures (Charney,
1980). In these studies, though, the photos were presented individually in
a testlike situation. The decision to make a photo-book that was theme
oriented was based on pilot results that indicated that by presenting children
with single photos one got little more than referential labeling, whereas
when the photos were presented in a storybook format, the children’s talk
was more centered on relating various persons depicted to action frames.

Procedure

The children were observed twice each month for a 4-month period. Once a
month, each of the six children visited the playroom with his or her caregiver,
and once a month the children participated with a peer. The children were
paired off into three dyads from the start of the study and always participated
with the same peer. The peer sessions and caregiver-child sessions were
always within a week of each other and never occurred on consecutive days.
The order in which the sessions took place varied such that in the first and
third months, the peer dyads were observed first, whereas in the second and
fourth months, the caregiver-child dyad was observed first.
The caregiver-child sessions usually took place on the day when the
caregiver assisted in the day-care center. The peers were scheduled at a time
of day when they both were present. The researcher would approach the
participants a few minutes before the observation was to start, telling them
that it was almost time to go water their plants. Right before the dyads entered
the observation room, they were given the instruction to show their mother
or peer what was on the table. After watering the plant(s), the dyad moved
over to the table by the couch, where either the manipulative toys (Months I
and III) or the blocks (Months II and IV) were set up. The children always
stood at the table while they played, whereas the caregivers opted to sit on
the couch. I was the only other person present at all of the sessions; and after
the participants began playing, my role was primarily that of a participant
observer. I intervened when introducing new tasks or whena specific question
was directed toward me, but I did not initiate dialogue except when approach-
ing the dyads to distribute toys or in cases of extreme disputes among peers.
The sessions lasted approximately 45 minutes. After watering the plants
the dyads played at each of the other tasks for about 15 minutes. On occasion,
the children returned to whatever task was last set up on the table at the
completion of the study, if time was remaining. When the observation ses-
sions were over, the children and caregivers were escorted back to the
classroom.
METHODOLOGY 49

Recording Equipment

All sessions were recorded in three ways. First, the interactions were video
recorded from the time that the children entered the observation room until
the time they left. Because the participants remained in one area after the
plants were watered, the camera could be left unattended in the corner.
Although the camera was introduced to the children during piloting, it largely
went unnoticed by the children. To a large extent the caregivers also seemed
unaware of the fact that they were being video taped, though they had been
informed that video taping would take place. In fact, one parent asked in
the third month of the study when the video sessions would begin. In
addition to video recordings, two kinds of audio tapes were made. Wireless
microphones were clipped to the participants. A stereo audiodeck was used
as well, which had a built-in microphone.

TRANSCRIPTION

Because all coding was to be carried out in conjunction with simultaneous

viewing of the video tapes, the primary purpose of transcription was to
account for the verbal behaviors of the participants. After watching a video
at least one time through, actual transcription started. Transcriptions were
based primarily on the audio tapes from the wireless microphones. Segments
of speech were transcribed and checked against the video in a second pass.
Alternative interpretations were noted and checked in another pass. The
transcription of the verbal material largely followed the conventions outlined
by Ochs (1979). Each transcript started at the point at which the dyad entered
the observation room and ended when the dyad exited. Modified orthog-
raphy was used. Phenomena such as turn overlaps, pause length, and latch-
ing were recorded using symbolic notation.
 Sil NE ySilal [= GW Ie:

Orientation Points

The present chapter serves as an orientation point on the basis of which

more precise research questions can be put forth and a coding scheme can
be developed. My point of departure is a series of distributional analyses
of linguistic forms of the sort carried out by the American structural linguists
(cf. Harris, 1951). The present chapter reports some preliminary findings
and describes the coding, but a further aim is to outline a methodology for
working with data that is by no means limited to the issues of agentivity
and control, nor to the analysis of the particular linguistic forms analyzed
here. The present methodology could easily be extended to the analysis of
other linguistic forms.
The starting point of this sort of analysis is an examination of the child’s
distribution of particular forms. The goal is then to deal with the data in
such a way that one can ultimately arrive at the child’s organizational scheme
governing the contrastive use of related linguistic forms. An assumption
underlying such a paradigm is that the child’s use of individual forms is not
random but, rather, is guided by form-function pairings. Such pairings do
not merely combine individual forms and their functions but instead are
viewed as parts of larger systems organized around multiple form-function
units. The challenge in analyzing data in such a way is to develop a coding
procedure that begins with a set of well-defined categories and yet is flexible
enough to capture regularities constructed by the child. The approach
adopted here is similar to related approaches developed by others (Bamberg,
1987; Berman & Slobin, 1994; Ervin-Tripp, 1989; Gerhardt, 1983, 1990; Pike,
1982; Savasir, 1984). According to this approach, particular forms are coded
in terms of several interrelated. dimensions so that one can find bundles of

50
ORIENTATION POINTS 51

linguistic and nonlinguistic factors that co-occur with the use of different
forms.
I begin my analysis of the children’s linguistic marking of agentivity and
control with a consideration of such issues as who the children refer to and
what forms are used in such references. After these preliminary analyses are
presented in the first section, I move on in the second section of this chapter
to refine the specific research questions. In a third section, a coding scheme
for the subsequent analyses is outlined.

PRELIMINARY ANALYSES

Mean Length of Utterance (MLU)

The MLU (in morphemes) has been calculated for all children so as to provide
a general index of the six children’s various levels of linguistic maturity at the
onset of the study, as well as across its duration. Although the use of the MLU
has serious shortcomings, such an index allows us to compare the six children
not only with one another but also with other children discussed in the
literature.
The procedures adopted for calculating the MLU follow those outlined
in Brown (1973). The index is based on the first 100 fully transcribed utter-
ances of each session. Nursery rhymes and songs were excluded from the
analysis; full names (i.e., first and last name) and compound words were
counted as single units. Separate indices were calculated for each video
session; thus for each child at each month of the study two indices were
calculated: one for the caregiver-child session, and one for the peer session.
In addition to the two separate MLUs for each child at each month, a mean
score was calculated for each child at each month. Table 4.1 shows the
MLUs of all six children at Month I. Table 4.2 outlines the developmental
changes in MLU over the course of the study for the three youngest children.’
Looking first at the mean scores presented for each child in Table 4.1,
we see that the two youngest children (Megan and Grice) were just beginning
to combine words. Jeffrey was at a slightly more advanced phase; his utter-
ances consisted of two to three morphemes each. Finally, three of the chil-
dren (Eric, Keith, and Thomas) were beyond a two-word phase.
A closer examination of Tables 4.1 and 4.2 points up an important finding
that comes up again in subsequent chapters. This concerns the differences
found in the data from the video sessions involving the peers and the data
from those involving caregiver-child interaction. Many of the children made

1Because there is general agreement that MLU is an even less reliable measure beyond the
early phases of grammatical development, it was decided that the MLU would not be calculated
for each month for the three children in the study (Eric, Keith, and Thomas) who were already
regularly using multiword utterances at the onset of the study.
52 CHAPTER 4

TABLE 4.1
Mean Length of Utterance: Month |
a

Age in Caregiver Peer Mean

Child Months Session Session MLU
pO ee ES ee

Megan 20 1.98 2515 2.07

Grice 22 1.50 1.94 Vere
Jeffrey 30 2.88 2.75 2.82
Eric 28 3.84 3.12 3.48
Keith 31 2.70 3.73 3:22
Thomas 32 4.08 3.74 3.91

greater use of linguistic resources in the peer sessions. This could in part be
attributed to the fact that when the children interacted with their caregivers,
they often made use of ellipses in question-answer routines. An extreme
instance of this pattern was displayed by Keith. Had I collected only the
caregiver-child data I would have concluded that Keith was at Brown’s Stage
Il (approaching Stage IID, whereas 2 days later the same child, when
interacting with his peer, could be placed at Brown’s Stage IV. In addition to
pointing up a problem in relying on MLU indices collected at single time
intervals, this analysis also reveals the importance of considering grammatical
development in contexts outside the caregiver-child situation. This point is
especially important to keep in mind when drawing comparisons between
children from different studies, in particular in cases of cross-linguistic
comparisons where researchers have collected data not only from mother-
child pairs and similar-aged dyads, but also from sibling pairs and other
mixed-aged dyads.

TABLE 4.2
Mean Length of Utterance: Month I-IV
(Megan, Grice, and Jeffrey)

Month Session Child

Megan Grice Jeffrey

Caregiver 1.98 1.50 2.88

| Peer 2e5) 1.94 2.75
Mean 2.07 1.72 2.82
Caregiver 2.36 2.08 3.34
ll Peer 2.24 2.38 3.69
Mean 2.30 2.23 3.52
Caregiver 2.54 2.76 3.65
Ul Peer 2.46 2.80 3.92
Mean 202 2.78 3.79
Caregiver 3.04 3.22 3.48
IV Peer 2.11 3.80 3.87
Mean 2.58 3.51 3.68
ORIENTATION POINTS 53

Distribution of Reference to Self and Other

As a first step in my analysis of the relation between form and function in

children’s talk about agentivity and control, it seems essential to consider not
only what sort of participants children talk about, but also the distributional
patterns that emerge across development. To what extent does children’s early
talk about agentivity go beyond self as agent? This question is raised in light
of some of the studies reported in the second chapter that suggested that
children only gradually develop a conceptual category involving global
agency and at first work with a narrower category involving personal agency.
In the following analysis focus has been placed on all references occurring
in subject position.? Are the children equally likely to refer to self and to
others as agents of goal-directed actions or experiencers of internal states?
To answer such a question, all forms occurring in subject position were
coded in terms of a three-way distinction based on whether the child’s
reference was: (a) selfU gonna make a tower), (b) other (You needa make
it real big, Gramma buy cookies), or (c) joint (self and other: We blowing
bubbles).’
Figure 4.1 reveals the distribution of self, other, and joint references (in
subject position) for all six children during the first month of the study. The
children can be divided into two groups: First, three of the children referred
primarily to themselves as main participants at the onset of the study. More
than 75% of their references were of the sort My take it home or Grice ride
bicycle. These children who referred primarily to self are called ego-
anchored.‘ A second group of children (Eric, Keith, and Thomas) regularly
referred not only to themselves but also to others as main participants. These
children are collectively referred to as nonego-anchored. As is shown in Fig.
4.1, only about 50% of their references were to self, and the remaining
instances were to other or were joint.
Based on the analysis presented here, one can ask whether the self/other
distribution represents a developmental pattern. Support for the position
that children travel a route leading from ego-anchoring to nonego-anchoring
can be found in Fig. 4.2. Here we find that as the three ego-anchored

References in object position tended to be concrete and inanimate for all children across
all sessions. Only occasionally did the children refer to more abstract objects (e.g., my turn)
or animate beings (e.g., you, Daddy).
3if the child repeated or rephrased parts of an utterance, only the self reference form in
the final version was included. Thus the utterance: ] 1 hafta I hafta I hafta do something was
coded only once as a reference to self. Rephrasings were treated similarly. All repetitions and
rephrasings received an extra code so they could be pulled out at a later point for separate
analysis.
‘The label ego-anchored is not to be confused with Piaget’s notion of egocentric speech
(Piaget, 1926). Instead, the term ego-anchored as used here implies that the child’s linguistic
references to main participants tend to refer to self.
54 CHAPTER 4

100%

90% WB Self (My ly open

op that)
(__} Other (You have that)
GE Joint (We made that)
80%

70%

60%

50%

40%

of 30%
References
Percentage

20%

10%
FIG. 4.1. Distribution of references
0% : : ¥
Megan Grice Jeffrey Eric ©Keith Thomas in subject position at onset of study.

children developed, their distribution of references to self and other con-

verged in the direction of that of the nonego-anchored children. Jeffrey’s
distribution shifted dramatically in Month III such that he could be classified
as nonego-anchored, whereas the two youngest girls FS ego-anchored
throughout the duration of the study.
The findings from the present distributional analysis, taken together with
the review of research on self and other reference discussed in chapter 2,
imply that it is indeed important to question whether children’s early notions
of agentivity and control extend beyond the self. Furthermore, given the
findings of Deutsch and Budwig (1983), which suggested that children con-
struct a form-function contrast between own name + noun (e.g., Adam
pencil) and first person possessive pronoun + noun (e.g., My pencil) that is
not matched by a contrast with regard to second or third person reference,
it seems important to consider self reference forms as a kind of subsystem.
In the section that follows we take a closer look at the distribution of
children’s self reference forms across the span of the study.

100 5

s
B=
00 |
ag

80 4
o

Ad a
gS 4
an
£
=® 60-4
”
4 50
5 40
s
© 304
s
® Grice
2 20 - 4 Jeffrey
5 @ Megan

° 10
FIG. 4.2. Percentage of references
0 | ll Mi \V to self in subject position: Months
Month I-IV (Megan, Grice, and Jeffrey).
ORIENTATION POINTS 55

Distribution of Self Reference Forms

The goal of the present analysis was to find out what self reference forms
the children used at each month of the study and, in addition, what the
relative distribution was of the various forms. Did all the children make use
of similar forms when referring to themselves? Did their usage reflect com-
mon distributional regularities?
The distributional analysis of self reference forms was conducted as fol-
lows. All utterances (clauses) containing a form that referred to the self were
isolated.° If more than one self reference form was used in a given clause,
then such forms were noted (see, though, footnote 3). Each form was then
further coded as nominal (e.g., Own Name) or pronominal. Each pronominal
form was further coded as J, Me, My, or Other, where Other included: (a)
less frequently used first person forms (e.g., myself, mine); (b) self-created
forms of self reference (e.g., meself); and (c) forms that adult speakers of
English would use for other reference but that the children used when
referring to self, as in the following example:°

(1) Keith CK) and his mom (KM) are looking through the photo-book.
a. KM: Who are these two silly guys?
b. K: | Um Thomas and Keith.
c. (K and KM discuss whether K and Thomas fight sometimes.)
d. K: _ I see Keithy talking. (pointing to photo)
e. KM: And Keithy is talking.
f. K: Yeah. They bumped heads. (pointing to same photo of self
and peer)

The results of the distributional analysis of self reference forms can be

found in Tables 4.3 and 4.4. Table 4.3 reveals the distribution of self reference
forms for the six children at the onset of the study. We find that all six
children made use of a variety of nominal and pronominal forms of self
reference, although the relative frequency of use differed among the children.
Two of the ego-anchored children relied primarily on three forms (Grice: J

Null forms were not included in the analysis, due to problems of interpretation. It was not
clear whether the child was referring to self or others. It should be noted that the three
ego-anchored children at times employed a form ma that was pronounced quite similarly to
my. Because the two forms were so close in pronunciation and it was often difficult to distinguish
between them, such uses have been grouped together under the heading my in the present
analysis. Finally, all proper names of the children have been altered, and the pseudonyms have
been substituted in examples in which the children referred to themselves using their own
names.
©The less frequently used forms (e.g., myself and we) were originally coded under separate
headings, but because the children used them so rarely, they later were grouped together under
the general heading of Other.
56 CHAPTER 4
TABLE 4.3
Distribution (%) of Self Reference Forms: Month |

Child Form Subtotal*

/ My Me Other Own Name

Megan 33 23 10 2 32 (91)
Grice 25 36 28 1 10 (72)
Jeffrey 38 46 6 Ti 3 (125)
Eric 65 6 3 17 8 (116)
Keith 56 7 1 28 9 (162)
Thomas 60 10 3 22 4 (178)

* (Raw numbers).

Me, My; Megan: J, My, Own Name), whereas the remaining ego-anchored
child, Jeffrey, showed a preference for two forms (J and My). The nonego-
anchored children all used primarily one self reference form (J). A further
distinction between the two groups of children is that the ego-anchored
children freely employed various self reference forms in subject position,
often in ways that deviated from standard English. Thus, they said things
like: My did it, Me jump, I like peas, and Megan count.’
Although the ego-anchored children all employed several self reference
forms in a fairly even distribution at the onset of the study, we found that
by the end of the study, these three children looked like their nonego-an-
chored peers to the extent that they relied primarily on the form J (cf. Table
4.4). It can also be noted that over time, the creative uses of My and Me in
subject position faded out. In the final session Megan never used such forms
in subject position, and Grice and Jeffrey each did so only once or twice.

Summary

In this section I have reviewed findings from some preliminary analyses.

These findings justify the division of the children into two groups. Three of
the children began the study with MLUs of about 2.2; that is, they were just
beginning to combine words. These children made use of several self ref-
erence forms, and two or three of these forms were employed regularly in
subject position, often in ways that deviated from adult input. This group
of children has been referred to as ego-anchored. A second group of children,
who are referred to as nonego-anchored, had MLUs of about 3.54. They
also regularly used several self reference forms, but a distributional analysis
of such forms revealed that these children relied primarily on the form J

7One nonego-anchored child, Keith, used My (alone) in subject position one time (My dont
know).
ORIENTATION POINTS 57

TABLE 4.4
Distribution (%) of Reference Forms: Month I-IV
(Ego-Anchored Children)

Child Month Form Subtotal*

| My Me Other OwnName

| 33 23 10 2 32 (91)
Megan Il 59 9 6 2 24 (87)
Ut 55 12 7 8 28 (76)
VV 61 13 6 10 9 (77)
I 25 36 28 1 10 (72)
Grice Il 30 41 24 1 3 (90)
Wl 65 16 6 10 4 (280)
'V 67 11 12 6 5 (212)
| 38 46 6 7 3 (125)
Jeffrey I 41 41 D 9 2 (164)
Wt 52 9 7 30 2 (148)
IV 53 13 8 23 3 (183)

* (Raw numbers).

when referring to self, and their use of self reference forms did not deviate
from adult input in the way noted for the ego-anchored children. These
findings are summarized in Fig. 4.3.
The findings of the analyses summarized here reveal that the two groups
of children can best be viewed along a developmental continuum. Over the
course of the study, the ego-anchored children’s MLUs approached those of
the nonego-anchored children (see Table 4.2), their references to self de-

EGO-ANCHORED > —> —> —> —> —> —> — > —> — PNONEGO-ANCHORED
MW about 2.20 —>—>—_> —>—_> > > > > ML about 3.54
Severd Self Reference Forms GRF}—> —» —» — >»—> Several Self Reference Forms (GRA)
2-3SRF employed reguiaty +—»>—»>—»>—»—»—»>—> Pimallly1SF employed
Useof SRF demiates
from Input —> —> —» —> —> —p —>—>
—» No devalions

FIG. 4.3. Pictorial summary of transition from ego-anchoring to nonego-an-

choring.
58 CHAPTER 4

creased (see Fig. 4.2), and by Month IV the distribution of self reference forms
by the ego-anchored children looked quite similar to the distribution noted
for the nonego-anchored children at Month I (see Tables 4.3 and 4.4). The
proposed developmental course can be described as follows: The children
went from a phase of almost exclusive reference to self, employing several
self reference forms, to a later phase of self and other reference, employing
several forms but relying primarily on the form I. The transition from one phase
to the other included the gradual disappearance of the use of self reference
forms in subject position in ways that deviated from adult input.

REFINEMENT OF RESEARCH QUESTIONS

We are now in a position to refine the general research questions raised in

chapter 1. The first question that arises from the distributional analyses
reviewed here concerns whether the ego-anchored children were con-
trastively employing the various self reference forms. I start from the assump-
tion that the nominal and pronominal forms were employed in a principled
manner. One might want to claim instead that the children were simply
confused about the distinctions between various self reference forms and thus
might use them in free variation. Evidence that the children were not using
these forms in free variation lies in numerous examples of what Karmiloff-
Smith (1979) referred to as hesitations and spontaneous corrections. The
following example illustrates the sort of wavering between forms often found
in the children’s utterances:®

(2) Jeffrey is playing with blocks.

aaa: M;/ M-/ I want;/ I want-/ My-/ I want;/ I want the little ones.

In examples such as this, the child seems to hesitate over which form to
employ. One might want to argue that the child’s hesitation is a reflection
of competition between similar or equal forms. As we see, though, the
direction of repair can be accounted for. Thus I claim that such hesitations
can be taken to indicate a motivated choice.
In Part II of this book, I consider in detail the basis upon which the
children selected the various self reference forms. The central claim is that
at a time when children rarely refer to others, they contrastively employ self
reference forms to mark various degrees of agentivity and control with regard
to their own involvement in action frames. Before turning to such a discus-

®The distinction between the dash (.) and semicolon () in the examples can be read as
follows: The dash indicates that what follows is a reformulation, whereas the semicolon indicates
that what follows is a repetition. Often what precedes the semicolon has been repeated
numerous times, though in the example it is listed only once.
ORIENTATION POINTS 59

sion, I first consider previous studies dealing with the acquisition of self
reference forms, with the aim of developing more specific predictions con-
cerning the contrastive use of self reference forms by children in the present
study. Such a review is not meant to be exhaustive but, rather, is directed
toward the refinement of research questions for the present study. I will
then be in a position to consider the coding rationale and procedures em-
ployed in the present study.

Previous Research Related to Development

of Self Reference Forms

The literature concerning the development of self reference forms either has
focused on issues related to nominal and pronominal forms or has specifically
investigated issues concerning the acquisition of pronominal forms. I first
review some studies suggesting that nominal and pronominal self reference
forms go througha phase in which they co-exist. Next, I consider the de-
velopment of first person pronominal forms.

The Simultaneous Use of Nominal and Pronominal Forms. Lewis

and Brooks-Gunn (1979), in one of a series of studies concerning the ac-
quisition of self-knowledge, considered the relationship between language
development and self-recognition. In this study they provided children with
a series of photos of the child and other children. Of relevance here are the
findings with regard to the use of self reference forms by the verbal infants
ranging between the ages of 21 and 36 months. Although Lewis and Brooks-
Gunn were not specifically interested in the contrastive use of nominal and
pronominal forms, they nevertheless noted that the children tended to use
nominal forms of self reference when labeling the photos. Fifty-eight percent
of all references involved the use of the child’s own name, whereas in only
18% of the cases did the children employ pronominal forms. Although age
changes were not noted with regard to the use of nominal forms, Lewis and
Brooks-Gunn reported dramatic changes with age for pronominal forms.
They documented a steady increase in pronominal usage—pronominal forms
were used by only one child before the age of 2, by 20% of the 30-month-old
children, and by 42% of the 36-month-old children.
Four other studies also touched on the contrastive use of nominal and
pronominal self reference forms. As mentioned earlier, Deutsch and Budwig
(1983) documented that young children contrastively employed their own
name and the first person pronominal form my in reference to possessions.
The use of Own Name was found in the context of assertions about posses-
sions where control was not at issue, whereas the pronominal form was found
in possessive constructions in which control was under dispute. In another
study, Cooley (1908) examined diary records of his daughter’s references to
60 CHAPTER 4

self until the age of 33 months. He reported that the use of nominal and
pronominal forms linked up with different “self ideas” (Cooley, 1908, p. 341).
The child was said to use her own name or the word baby when referring to
her physical body, as when labeling photos or referring to her body parts. In
contrast, the child used Jwhen expressing a “feeling of self assertion” (Cooley,
1908, p. 341), linked up with action in most cases. Taken together, these
studies lend support to the idea that the children in the present study might
contrastively employ nominal and pronominal forms of self reference. Further
evidence for the contrastive use of nominal and pronominal forms stems from
the detailed case study of Emmy’s narratives as reported by Nelson (1989) and
Gerhardt (1988). These researchers trace Emmy’s use of distinct self reference
forms, and how such forms were used both to realize distinct discursive
intentions and to construct a notion of self in time.

The Acquisition of Pronominal Forms. Studies of children’s early use

of pronominal forms have considered three related issues: contextual variation
in the use of pronominal self reference forms, deixis, and case marking errors.
I briefly consider each of these areas of study.
Several developmental psychologists have noted that pronominal forms
link up with different social contexts. For instance, Goodenough (1938), in
a study of 30-month-old to 66-month-old children, noted that the youngest
children in that study used a variety of pronominal self reference forms in
different contexts. Based on observational records of children interacting in
what was referred to as a controlled situation (involving adult supervision)
and in a free play situation (involving peers), Goodenough reported that
the first person pronominal forms were far more frequent in the free play
situation. She suggested that the first person pronominal forms link up with
a general sense of “ego-consciousness” and this feeling is enhanced in com-
petitive situations involving peer play. Goodenough did not distinguish be-
tween the various first person pronominal forms.
Ames (1952) also examined the child’s developing sense of self during
the toddler and preschool years by observing and recording children’s ver-
balizations about the self. At the age of 21 months, Ames noted that the use
of possessive pronouns appeared in disputes over control of objects among
peers. By 30 months, other pronominal forms of self reference (e.g., D
appeared in conjunction with naming completed or intended action.
A third study that sheds light on the use of first person pronominal forms
documented the early use of nonnominal expressions in children’s language.
Gopnik (1980) reported that the possessive terms my, mine, and mines were
among the most frequent expressions used by the children she studied.
These words were used primarily when the child wanted to claim an object.
Gopnik noted that the objects linked with the possessive expressions often
did not belong to the children, and “in some cases the children had no
ORIENTATION POINTS 61

contact with the objects in the past that might lead them to believe that they
belong to them” (Gopnik, 1980, p. 200). In addition, Gopnik suggested that
my was Often used to refer to an intention to act, as in the utterance My do
it. Because Gopnik’s study was limited to the investigation of nonnominal
expressions, it is not clear what the function of proper names was for her
subjects, if in fact they used them.

Deixis. Several other researchers focusing on personal pronouns have

studied them in reference to the general issue of shifting reference. Several
developmental hypotheses have been offered (Charney, 1980; Clark, 1978; de
Villiers & de Villiers, 1974; Huxley, 1970; Loveland, 1984; McNeill, 1963;
Shipley & Shipley, 1969). Although the specific hypotheses and conclusions
may not be so relevant to the present discussion, what such research does
point up is that some children may confuse speaker/addressee referencing
some of the time. Loveland, for example, in a series of production and
comprehension studies, pointed out that 11 of the 27 2- to 3-year-old children
in her studies mixed correct and incorrect use and comprehension of the same
pronouns, which she took as an indication of a gradual phase of sorting out
issues governing proper use. The children not only exhibited case errors in
production but were also noted, in both production and comprehension, to
have difficulty at times with shifting reference. The question remains: Why is
it that usage is correct only some of the time? Could it be that the child’s use
is guided by a different set of assumptions that at times leads to responses that
look adultlike? (see Bamberg, 1987, for an illustration of this point with regard
to older children’s nominal/pronominal contrasts in narrative productions).
The basis of such mixed responses in the various studies has not been fully
examined. There is the possibility that the children are guided by different
assumptions about these forms, which in turn have been interpreted by
researchers as reflecting correct and incorrect usage.

Case Marking Errors. A final set of studies related to the use of

pronominal forms concerns the acquisition of case marking. Such research
indicates that children between the ages of 2 and 3 years make case marking
errors, often, for example, employing accusative forms where nominative
forms would be expected (cf. Bellugi, 1971; Brown, 1973; Leopold, 1939; Park,
1971; Radford, 1990). For instance, children have been noted to say things like
Why me sitting on it? or What me do? It has been suggested that such errors
should be expected, given the complex network of rules involved in case se-
lection (see Maratsos, 1979, for a discussion of this point). Although one might
anticipate such errors, the question remains why such errors occur and not
others. Why, for example, do children replace nominative forms with accusa-
tive forms but not vice versa? Very little theorizing has focused on this issue.
More recently, Radford (1990) argued that in early child English there is
no case system. At this stage children “have not acquired the morphosyntax
62 CHAPTER 4

of nominatives” (p. 177) and thus often misuse pronouns. Radford did not
claim that the child does not occasionally make use of nominative forms
early on. Rather, he claimed that such sporadic occurrences appear in set
phrases, that is, in nonproductive utterances. In chapter 9 I compare Rad-
ford’s position with that based on the findings of this study. For now, the
important point is that the children I studied also were noted to use accusative
pronouns in place of nominative ones in early child English.

Predictions

Though none of the studies reviewed here attempted to identify specific

ways in which self reference forms might be contrastively employed by
young children acquiring English, taken together, the studies begin to point
toward some systematic uses of particular forms. I now consider three spe-
cific predictions that stem from this review. The first two are as follows:

Prediction 1: Nominal and pronominal self reference forms will serve

contrastive functions. Nominal forms will be employed referentially, whereas
pronominal forms will be used in more dynamically charged interactive
situations.

Prediction 2: The various pronominal forms of self reference will link up with
the expression of different ways of perceiving the self in relation to action.

Our review of previous research concerning pronominal forms leads to

two more specific predictions concerning the forms J and my,

Prediction 2a: My will link up with the expression of control.

Prediction 2b: I will be used in asserting inner beliefs of the self.

Evidence for Prediction 2a stems from the studies of Deutsch and Budwig
(1983); see also Kolodziej, Deutsch, and Bittner (1991), Goodenough (1938),
and Gopnik (1980), where it was noted that this form appeared in the
context of free play situations in which the children attempted to gain control
over objects. Prediction 2b is based on the findings of Cooley (1908), who
noted that J appears in contexts in which the child makes self assertions.
With regard to the other pronominal self reference forms no specific
predictions are forwarded. Rather, a further research question to be ad-
dressed by the present study is whether the children employ the other
pronominal forms and, if so, on what basis.
Although there seems to be some indication in the literature that different
children link the use of particular forms with a similar set of functions, a certain
amount of individual variation is to be expected. In recent years, growing
attention has been given to the different pathways children find in acquiring
ORIENTATION POINTS 63

grammatical systems. In addition, children have been noted to differ in the

sorts of pragmatic functions conveyed in their early utterances (cf. Bates,
Bretherton, & Snyder, 1988, for a thorough review of the individual differences
literature). Findings from such studies lead to the following prediction:

Prediction 3: Some variation will be found in the ways various children employ
the different pronominal forms, though all children are expected to relate
form and function systematically.

The idea here is that before children regularly refer to others, they will
make use of self reference forms to take different perspectives on the way
they situate themselves in fields of action. If children are using the forms
as predicted, one would expect the forms to show up in particular activities
or junctures of children’s play. For example, one would expect that nominal
forms should be the most frequent in the context of the photo-book task,
and my should appear frequently in the peer sessions in instances where
the children attempt to gain control of objects. Likewise, if the children are
using first person pronouns as predicted, one would expect errors of par-
ticular kinds both with regard to marking case and when shifting reference.
Within the framework of the present predictions, one could question whether
case errors can be related to children’s budding hypotheses about the func-
tions of particular forms. This sort of account might help explain why case
errors occur only some of the time. Similarly, one could check so-called
errors in shifting reference to see whether the child might be marking a
slightly different notion. For instance, when children refer to experimenters’
toys as my toy, it may well be that the children are attempting to claim the
toy rather than having difficulty shifting perspective with the forms my and
your. With these predictions in mind, I can now present the coding proce-
dures for the present investigation.

CODING RATIONALE, CATEGORIES, AND PROCEDURES

Coding Rationale

The central claim in the last section concerns the expectation that the ego-
anchored children’s use of various self reference forms systematically links
up with issues of agentivity and control. The coding scheme must be one
that taps both semantic dimensions of agentivity and pragmatic features of
control. As the discussion in chapter 2 pointed out, the notion of agentivity
has been viewed as including a number of interrelated dimensions, suggest-
ing that the coding of agentivity needs to take into account several different
factors. A further challenge in developing a coding scheme concerns a prob-
64. CHAPTER 4

lem raised in the review of the functionalist literature in chapter 1. The

challenge is one of coding in terms of a fixed number of dimensions but
simultaneously keeping in mind the child’s own categorization scheme. The
decision in the present study has been to work from two ends. First, based
on the review presented in Part I of this book, a multilayered code has been
designed. A second approach has been more qualitative in nature. Here the
starting point of analysis has been the examination of minimal pairs, hesi-
tations, and self-corrections. Such an approach assists in keeping the child’s
systematization in mind by allowing the researcher to develop new categories
that may not have been included in the original coding.
It can be noted that the distributional analysis used in the present inves-
tigation goes beyond that set out by the American structural linguists (cf.
Harris, 1951). My focus here includes various dimensions of the activity
context in which forms appear. I am interested not only in the possible
clustering of the individual forms not only around a particular dimension
on any given level but also in the relationship between the forms across
the different levels of analysis. The method adopted here is similar to that
documented in Gerhardt (1983), Gee and Savasir (1985), and Pike (1982).

Coding Categories

The coding scheme consists of six different broad levels of analysis and
several sublevels. These levels include: form, surface-level characteristics,
semantic characteristics, pragmatic characteristics, nonlinguistic/contextual
characteristics, and causality coding. I now describe briefly each of these
individual levels of analysis.

Form. This level of coding has already been introduced in the discussion
of the distribution of the children’s use of self reference forms. The basic idea
here was to examine all instances of self reference forms. These were coded
as nominal (Own Name, Other le.g., baby]) or pronominal (, My, Me, Other
[e.g., meself, myself). Findings of this analysis appear in Tables 4.3 and 4.4.

Surface-level Characteristics. At this level of analysis, each use of a

self reference form was further coded in terms of its relation to the rest of
the utterance. The first decision was whether the form appeared in an ut-
terance with a verb or in a verbless context (e.g., My did it vs. My airplane).
If the form appeared in an utterance with a verb, it was further categorized
as to whether it (a) occurred in subject or object position, and (b) appeared
alone with the verb or with other arguments (e.g., J jump vs. I saw the ball).
If the form appeared in a verbless context, it was further coded as appearing
alone (e.g., Me), with a noun only (e.g., My book), or in some other con-
struction type.
ORIENTATION POINTS 65

Semantic Characteristics. The semantic level of coding captured the

degree of agentivity expressed in clauses containing self reference forms.
The coding drew upon research in transitivity, agentivity, and prototype
semantics discussed in chapter 2. Given a definition of prototypical transi-
tivity as involving “a direct causation scheme with proximate and ultimate
cause both residing in the same volitionally acting causer” (DeLancey, 1984,
p. 185), one can see the close connection to agentivity. Prototypical agents
are volitionally acting causers.
To code for agentivity, a subset of Hopper and Thompson’s (1980) tran-
sitivity parameters were included, namely, those thought to be most relevant
to the assessment of agentivity. The five parameters selected included: Par-
ticipants, Kinesis, Aspect, Volitionality, and Affirmation. Each utterance con-
taining a self reference form and a verb was given a ranking for each of
the five parameters. The ranking include three degrees: high, mid, and low;
an uncoded category was also adopted. The central question raised at this
level of analysis was whether particular forms linked up with particular
levels of agentivity. It should be noted that in Part II of this book, findings
are reported based on a collapsed semantic rating. Those particularly inter-
ested in the semantic level of analysis are encouraged to examine the findings
in Budwig (1986), where the data were presented in a noncollapsed format.
The collapsed rankings were as follows: All utterances were given individual
rankings of high (score of 1), medium (score of 2), or low (score of 3) for
each of the five semantic parameters. The scores were added and mean
scores for each example were calculated. For the purposes of the presen-
tation of the data in Part II of this book, only the collapsed scores are used;
thus uses of self reference forms with mean scores greater than 1.5 were
ranked high in agentivity, and scores of 1.5 or lower were ranked low in
agentivity. I briefly review the five separate parameters coded.
The first parameter coded is labeled Participants. Clauses containing two
participants were coded as high, whereas those containing only one were
coded as low (e.g., J built the castle vs. I jump). A mid ranking was assigned
to clauses containing self-reflexives and to clauses containing a second par-
ticipant, reflecting a deviation from the prototypical causation scheme (cf.
Givon, 1984a), as in the case of locative direct objects that are not acted
upon by agents. An example of a sentence with a less prototypical object
would be IJ rode the horse.
The second parameter coded is referred to as Kinesis. A high ranking
was given to clauses containing highly kinetic verbs (e.g., blow, hit, knock
down). Clauses containing statives were given a low ranking. A mid ranking
was assigned when reference was to actions that required minimal effort
(e.g., look, read).
The third parameter coded is referred to as Aspect. The coding of aspect
is restricted to a consideration of the telic/atelic distinction. Clauses referring
66 CHAPTER 4

to telic situations, or what Vendler (1967) referred to as accomplishments,

received a high ranking. A low ranking was assigned to clauses referring to
atelic situations. A small class of clauses where both a telic and an atelic
reading could be given were assigned to a mid category. It should be noted
that assignment was based not on verbs alone but, rather, on situations (cf.
Comrie, 1976). Thus the utterance J am singing would rank low for aspect,
whereas J am singing a song would rank high.
The fourth parameter of agentivity is labeled Volitionality. The distinction
here is between volitional acts and happenings. Clauses referring to willed or
purposeful action received a high ranking, whereas reference to happenings
where responsibility was not assigned were ranked low. Thus the reporting
of nonintentional acts and states received a Low ranking. To assign the clauses
to one or the other category often involved careful observation of the video
data. The utterance J knock it down, said as the child gazed at the tower he
had just built up, while curling his hand intoa fist, would have been coded as
high. In contrast, the same utterance said after the child had swiftly turned
around to look at the door, in the process knocking over the tower with his
elbow, would be ranked low. Ambiguous cases were assigned a mid score.
The final parameter of semantic agentivity is labeled Affirmation. Coding
for this parameter is relatively straightforward. Utterances that were stated
in the affirmative were ranked high, whereas those that were nonaffirmative
ranked low. There is no mid value for this parameter. The idea behind this
level, as explained by Hopper and Thompson (1980), is that affirmative
clauses are more likely to express transfer of action than nonaffirmative
ones. For instance, the utterance J hit Billy ranked high, whereas the utter-
ance I did not hit Billy ranked low.
To summarize, the coding of semantic characteristics proceeded in two
stages. First, each clause containing a self reference form was coded on five
levels for different parameters of agentivity. For each parameter, a decision
was made about whether the agentivity ranking was high, mid, or low. After
all the levels were coded, a global decision was made for the clause as a
whole such that the coded clauses were ranked as either high or low in
agentivity. In Part II of this book the focus is purely on the global agentivity
rankings, though, as mentioned earlier, the findings of the individual pa-
rameters of semantic agentivity across development are reviewed elsewhere
(see Budwig, 1986, 1989).

Pragmatic Characteristics. The purpose of the pragmatic function

coding was to assess the potential relationship between the specific forms of
self reference and the pragmatic functions they served. All utterances contain-
ing a self reference form (those with and without verbs) were coded in light
of the broader discursive context. Three broad categories of pragmatic
function were used; in addition, a fourth category was added for all utterances
that were considered uncodable.
ORIENTATION POINTS 67

The basic distinction guiding the analysis of pragmatic function was a

split between utterances that function to bring about a change in the world
(control acts) and utterances that function to express an existing state (non-
control acts). The notion of control acts is based on the use of the term by
Ervin-Tripp and her colleagues (cf. Ervin-Tripp, 1981; Ervin-Tripp & Gordon,
1986; Gordon & Ervin-Tripp, 1984). Control acts, utterances by which we
attempt to bring about change (i.e., requests, directives, warnings) can also
be related to Searle’s (1979) discussion of word-to-world direction of fit.
Such utterances function to bring about change in the world so that the
propositional content of the speech act and world match.
All utterances with self reference forms coded as control acts have been
further categorized into one of three subcategories: as utterances that attempt
to influence (a) the hearer’s action (i.e., directives, prohibitions, invitations,
and warnings), (b) the speaker’s action (soliciting permission, claiming own-
ership, or controlling goods not in use or actions in progress), or (c) both
speaker’s and hearer’s action (ownership and enactment disputes). Targeted
utterances coded as control acts could also be further subcategorized as
other or as a combination of (a)c).
In contrast to control acts, utterances falling under the heading of non-
control acts do not attempt to bring about a change in the environment but,
rather, match an existing state. Utterances coded as noncontrol acts have
what Searle (1979) referred to as world-to-word direction of fit. Each utter-
ance coded as a noncontrol act was further classified into one or more of
the following categories: (a) identify/name, (b) describe, (c) assert, (d) com-
ply. Utterances falling in the assertion category were further categorized as
reports of (a) previous happenings or timeless events, (b) ongoing/imme-
diate future intentions, or (c) long-range plans and intentions occurring
outside the scope of the video session.
A third category of pragmatic function coding is referred to as multifunc-
tion. It should be clear that one utterance may at the same time function as
an assertion about the world and as a request to perform some action (i.e.,
the familiar Jt sure is cold in here, uttered not only as a statement but also
in hopes of getting someone to close the window). The multifunctional
category was included for utterances that seemed to fall under both of the
first two categories. When possible, instances falling under this category
were further categorized in terms of the subcategories outlined earlier (i.e.,
an utterance might be further coded as a combination of a directive and a
statement of future intention).
The final pragmatic category included all utterances that were considered
uncodable. These included (a) utterances containing self reference forms
that were interrupted or stopped midway (e.g., J hafta-), (b) cases where
video view was obscured and the pragmatic function was unclear, and (c)
self reference forms that occurred in a nursery rhyme or song.
68 CHAPTER 4

We can work through one example to help clarify the coding categories
reviewed here. Take the simple example of a child saying J want the red
blocks. If while saying this the child is tugging on his or her mother’s leg
and, after getting the mother’s attention, points to the blocks that are out
of reach, such an utterance would be a good candidate for a control act. If
the same utterance is said while the child is en route to and just about to
pick up two red blocks, and the caregiver neither seems to be called upon
to intervene nor intervenes independently, then this would most likely be
coded as a noncontrol act, an assertion. The combination category might
be used had the mother intervened to assist the child though the child had
not made eye contact with her, or in other ambiguous cases.

Nonlinguistic/Contextual Characteristics. The fifth level of coding

focused on nonlinguistic aspects of the use of self reference forms. Three
dimensions of the nonlinguistic context taken into account were: (a) inter-
active partner, (b) activity task, and (c) nonverbal behaviors and gestures.
The first two levels could be coded without the use of the video tapes. The
coding of interactive partner specified whether the utterance occurred during
the session involving the caregiver or the peer. The activity task coding
accounted for whether the child was playing with manipulative toys, building
with blocks, looking through the photo-book, or engaging in some other
activity. An extra symbol was added to indicate that at the time of the
utterance the researcher was either bringing over or removing toys. Such
instances, as we later see, are rich in control acts and could be analyzed
separately for form-function pairs.
The third aspect of nonlinguistic coding involved noting what behaviors
or gestures co-occurred with utterances containing self reference forms.
These included: (a) pointing, (b) being en route to object, (c) taking control
of an object, (d) holding an object, (e) relinquishing control of an object,
(f) showing signs of distress (stomping feet, waving hands, crying, etc.), and
(g) showing signs of excitement (clapping hands, swinging feet, etc.). If the
child was not in view, this was also noted. In addition, an extra code was
placed after (c), (d), and (e) to indicate that the nonverbal behavior occurred
in conjunction with a related action on the part of the partner. For example,
it might be noted that relinquishing control of an object was due to the
partner’s attempts to take control of it.
The nonlinguistic coding puts us in a better position to assess whether
particular self reference forms tended to occur in conjunction with particular
communicative partners, with particular activity tasks, or even at particular
junctures, as, for instance, when the researcher brought over additional toys.
In addition, we can examine the potential relationships between the em-
ployment of particular forms and specific gestures (for instance, use of a
nominal form with a pointing gesture).
 ORIENTATION POINTS
69
Causality Characteristics. The final level of coding examines the sort
of causal chains accompanying the use of various self reference forms. In
chapter 2 it was noted that languages often draw a grammatical distinction
between agents who are part of a direct causal scheme and agents who do
not necessarily act as both proximate and ultimate agent. The coding at this
level has been designed to examine the relation between the use of particular
self reference forms, causation schemes, and the development of notions of
agency and control. The question of concern here is whether the various
self reference forms link up with particular kinds of causation schemes.
To code for causal chains, each use of a self reference form was coded
in terms of a three-category coding scheme involving a consideration of
causality surrounding the targeted form:

1. The child acted independently of others, either with direct physical

effort or referring to and attributing previous action to internal states.
2. The child and other are jointly involved (i.e., mediated causation),
whereby other can act as (a) coagent, (b) counteragent, or (c) instru-
ment for child (the rare instances of the child acting as an instrument
for the other were included here),
3. Other acts independently of child, where other might be a person, a
contextual or task-related cause, an object, or a natural force.

Coding Procedures

All utterances containing self reference forms were listed on coding sheets.
If two forms appeared in one utterance containing two clauses, then each
form was listed with the clause in which it appeared. If two forms appeared
in a single clause, then the clause was listed twice, though these instances
were infrequent. The coding proceeded in numerous rounds. At each pass,
one level of coding took place. Coding for form, surface level considerations,
and the semantic agentivity parameters was carried out with reference to
the written transcripts. The remaining levels of coding were conducted while
viewing the video tapes. Any given utterance on the code sheet was followed
by a series of code symbols. In addition, space was left to add comments
concerning deviations from input, hesitations, self-corrections, and other
comments thought to be useful at a later point of analysis.°

*It can be noted that at the time the original coding took place, computer programs were
not available that would allow for the sophisticated cross-tier analysis now available through
the CHILDES CLAN programs (see MacWhinney, 1991). Computer analysis would have greatly
facilitated the process and is highly recommended for those undertaking a multilevel analysis
such as the one proposed here.
70 CHAPTER 4

SUMMARY

The purpose of this chapter has been threefold. First, I have considered the
results of a series of preliminary analyses; second, I have refined the research
questions and developed a series of predictions based on the findings from
the preliminary analyses and an examination of the literature concerning
children’s use of self reference forms; and finally, I have outlined the multilevel
coding scheme that has been used in coding the data reported in Part II of this
book.
The results of the preliminary analyses suggest that the children fell into
two groups at the onset of the study. The first group of children, referred to as
ego-anchored, had an MLU of under 3.0; and these children rarely talked about
others. All three children used a variety of self reference forms, and these forms
were also used in ways that differed from adult usage. A second group of
children, referred to as nonego-anchored, had an MLU that exceeded 3.0; and
these children divided their references between self and other fairly equally.
Although these children used a variety of self reference forms, they relied
primarily on a single form, and their usage did not deviate from adult usage.
I reviewed several pieces of evidence that suggested that the distinction
between ego-anchored and nonego-anchored children is best viewed as
falling along a developmental continuum. By the end of the study, the
ego-anchored children looked very much like the nonego-anchored children.
The preliminary analyses have been revealing in the following ways. First,
given the youngest children’s almost exclusive focus on the self, one could
question whether the marking of agentivity might be limited to the self.
Second, the creative use of self reference forms, as well as the reliance on
several different forms of self reference, leads one to question whether at
an early phase the children employed such forms to mark their role as
prototypical agent and various deviations from the prototype. My review of
studies that touched upon children’s acquisition of nominal and pronominal
self reference forms has led to the development of several specific predictions
about how children might link particular forms with functional clusters con-
cerning agentivity and control.
In the next section of the book, I review findings from analyses that test
out these predictions. First, in chapters 5, 6, and 7, I consider the individual
findings of each of the ego-anchored children. The findings, though leading
to some general conclusions, point up some intriguing individual differences;
and therefore the decision has been made to report individual case studies
of the children. In chapter 8, I consider the way the nonego-anchored
children organized the use of self reference forms across the period of the
study. Because significant individual variation from the general patterns to
be reported was not observed, I discuss the findings of the nonego-anchored
children as a group. After laying out the organization of the ego-anchored
and nonego-anchored systems, I proceed in Part III to consider factors that
might lead children to organize and reorganize their systems as they do.
 AE isle VN)

CASE STUDIES
 CHAPTER FIVE

Jeffrey

MONTH I
Jeffrey, linguistically the most sophisticated of the ego-anchored children
, had
an MLU of 2.82 at the onset of the study. He referred to self in 76%
of his
references to main participants (see Fig. 4.1). As was also noted in chapter
4,
his use of self reference forms often deviated from adult usage. Because
of
these characteristics, he was grouped with the other ego-anchored children
.
When referring to himself in the first month of the study Jeffrey relied
almost exclusively on two forms (see Fig. 5.1). Jeffrey’s use of J and
My
accounts for 84% of the distribution. In this discussion of Jeffrey’s use of
self reference forms, I first focus on the core contrast between J and My. |
then turn to consider briefly his use of other self reference forms.

The Core Contrast: J and My

The distinction between Jeffrey’s use of Jand his use of My is by no means

obvious. The following examples illustrate this point:

(1) I like Anna.

(2) I show.
(3) Pll hold this.
(4) My like vanilla.
(5) My taked it off.
(6) No my build a tower.
(7) My blew the candles out.

73
74 CHAPTER 5

138%

FIG. 5.1. Distribution (%) of self

reference forms: Jeffrey Month I (1
= 11759).

(8) My read another book.

(9) I no want that one. My want the little ones.
(10) I want my put the tower.

As these examples illustrate, the two forms often occurred in consecutive

utterances (see example (9)), as well as in the main and subordinate clauses
of a single utterance (see example (10)). These examples might lead one
to believe that the forms were not used contrastively but, rather, were ran-
domly distributed by a child who was confused about the use of personal
pronouns. One might take as further evidence for their random distribution
the observation that the two forms were used with similar frequency. J was
used in 38% of all utterances, whereas My occurred in 46% of all utterances
involving self reference. The central aim of this chapter is to show that by
working through the various coding levels, one can arrive at a systematic
contrast between Jeffrey’s use of the form J and his use of My. The central
claim is that a full understanding of the use of these forms depends on
simultaneous consideration of semantic and pragmatic information.

Surface Characteristics. Table 5.1 outlines several aspects of the

surface characteristics of Jeffrey’s use of self reference forms. Three trends in
particular are worth noting. First, we find that J and My both are found
primarily in utterances with verbs. In contrast, Meand Own Name were used
solely in utterances without verbs. Second, the uses of Jand Myalso look quite
similar when considering sentence position and argument structure. J was
used exclusively in subject position and primarily in multiargument utterances.
Mywas used primarily in subject position and, as examples (1)-(10) show, not
as part of a possessive construction, in multiargument utterances. Third, when
forms were used in contexts without verbs, both My and Me appeared in
utterances coded Other, primarily in predicate structures.
One last point regarding surface characteristics concerns the contrast be-
tween uses of My in subject position and those in object position. My in
object position was always part of a possessive construction, whereas My
in subject position was not. Compare, for instance:
 TABLE 5.1
Distribution of Individual Forms (%) According to Surface Characteristics
(Jeffrey Month 1!)
ee

(a) Verb / Verbless Context

a EoNe ee a ea ee
FORM With Verb Without Verb Uncoded Subtotal*
| 85 6 9 (47)
My 62 34 3 (58)
Me 100 (7)
Other 56 44 (9)
Own Name 100 (4)
Subtotal* (81) (38) (6) (125)
eee ee ee ee ee ee er ee ee ee ae

(b) With Verb: Position

FORM _ Subject Object Uncoded Subtotal"

! 100 (40)
My 86 8 6 (36)
Me
Other 60 20 20 (5)
Own Name
Subtotal* (74) (4) (3) (81)
ee
a eee

(c) With Verb: Argument Structure

FORM Single Argument —_ Multiargument Uncoded Subtotal*

\ 28 73 (40)
My 19 81 (31)
Me
Other 33 67 (3)
Own Name
Subtotal* (18) (56) (74)

(d) Without Verb: Context

FORM +Noun Alone Other Subtotal*

| 33 67 (3)
My 30 5 55 (20)
Me 29 71 (7)
Other 25 75 (4)
Own Name 50 50 (4)
Subtotal* (6) (9) (23) (38)

* (Raw numbers).

75
76 CHAPTER 5

(11) My cracked the eggs.

(12) I like my cereal.

This suggests that the child may not have viewed these two uses as similar.

Semantic Characteristics. All utterances containing a self reference

form and a verb have been coded in terms of a number of semantic pa-
rameters involving agency and transitivity (see chapter 4). In discussing the
findings here it should be noted that I focus exclusively on an analysis in
which each utterance was coded at the semantic level as ranking either high
or low in terms of the parameters (see Budwig, 1986, 1989, for a discussion
of the independent rankings on each parameter). My concern here is whether
or not the two self reference forms can be distinguished in terms of the two
levels (High and Low) coded.
Before considering whether particular forms link up with the high or low
end of the agentivity continuum, I should first discuss how the child distributed
his talk about high and low agentivity regardless of the form used. For
example, if Jeffrey rarely talked about instances ranking high in agentivity, it
would be hardly surprising that his forms did not contrast along this dimen-
sion. Furthermore, an examination of the developmental data shows that over
the course of the study changes took place in the extent to which the child
talked about events from the standpoint of high or low agentivity. For these
reasons, I first consider briefly how talk was distributed between these two
categories (regardless of which self reference form was used). Collapsing
across forms, we find in the first month of the study that Jeffrey’s utterances
containing self reference forms primarily ranked low in agentivity. Of all
utterances containing a self reference form, 73% ranked low, whereas 25%
ranked high (2% remained uncoded). This analysis is based on the 81 uses of
self reference forms coded in terms of the semantic parameters of agentivity.
Figure 5.2 illustrates the relationship between the core forms (/ and My)
and semantic ranking (high or low) in Month I of the study. Although J and
My occurred in fairly equal proportions (J = 49%, My = 44%) these forms
distributed differently along the agentivity continuum. Ninety percent of the
utterances containing J ranked low in agentivity. Utterances involving J
tended to refer to self as experiencer of internal states or, less frequently,
self carrying out atelic activities. Consider the following examples:

(13) I like my cereal.

(14) No no noI no want the little blocks.
(15) I wear it.

In contrast, the use of My was more varied in terms of ranking along the
agentivity continuum. Although more than half of the utterances containing
JEFFREY WE

100%

aa mm eos”
70%
60%
g
40%
Percentage
Use 30%
of

20%
10%
FIG. 5.2. Distribution (%) of self
reference forms in terms of seman- uPA Ber
tic agency: Jeffrey Month I. Form

My ranked low in agentivity as well (58% of all uses of My coded in terms

of the semantic parameters), 42% of all such uses of My ranked high in
agentivity. This finding is particularly striking if one bears in mind the finding
reported earlier that in general, Jeffrey tended to talk about events ranking
low in agentivity. Thus, although overall only 25% of all utterances containing
self references ranked high in agentivity, 42% of all utterances containing
My ranked high. Though Jeffrey did not talk much about self as a causer
of telic action, when he did refer to telic actions, My was the self reference
form used. For instance:

(16) My taked it off.

(17) My cracked the eggs.
(18) My blew the candles out.

It is particularly interesting to look at contrasts that revolve around the same

objects. For instance, in example (15), the child was referring to a wireless
microphone pinned to his outfit. The utterance in (16) referred to the same
wireless microphone, though in this case the child was reporting that he
just took it off. Similarly, we can compare utterances that contain two self
reference forms, one in conjunction with an internal state verb and one in
conjunction with an action verb. In all of these cases, the patterning is the
same: J links up with the internal state verb and My co-occurs with the
action verb. Thus we find instances like (19) but not like (20):

(19) I want my put the tower.

(20) My want I put the tower. (never said by child)

Although My was more likely than / to occur in utterances ranking high

in agentivity, it remains to be explained why My also occurred frequently
in utterances ranking low in agentivity. The utterances ranking low in agen-
78 CHAPTER 5

of
tivity involving the self reference form My took two forms. First, many
ar with the
them occurred in conjunction with mental state verbs, in particul
in
child’s expression of his desires; and second, some of the uses of My
that
utterances ranking low in agentivity occurred in conjunction with verbs
were less kinetic. Conside r the followin g:

(21) My read another book.

(22) My want cars.
(23) My want that.
(24) My wear it.

Comparing the uses of J in examples (13)—-(15) to the uses of My in examples

(21)-(24), the question remains how one can account for these differences.
This is an important question that is taken up in the next section. The point
to be made with regard to the coding of various semantic parameters of
agentivity in terms of a continuum ranging from high to low is that the forms
were used contrastively along this continuum. My was more likely than /
to link up with the expression of high agentivity, whereas J occurred primarily
in utterances ranking low in agentivity.

Pragmatic Considerations. With regard to various pragmatic func-

tions at the outset of the study, we find that although Jeffrey used speech
multifunctionally, his language functioned primarily to make assertions about
the world. Over half of his utterances with self reference forms (57%) were
coded as assertions, whereas only 18% were coded as control acts, and 14%
were coded as multifunctional. We now consider the extent to which the
core contrast co-occurred with specific pragmatic functions.
Examining how the core forms link up with the pragmatic categories, we
find that distinct distributional patterns emerge (see Fig. 5.3). J tended to
occur in utterances that functioned as assertives (68% of all uses of JD,
whereas it occurred infrequently in control acts. In contrast, My distributed
more evenly across the function categories. Although there was aslight
preference for My to occur in assertions (45% of all uses of My), My occurred
26% of the time in control acts, and 19% of the time it was found in utterances
coded as multifunctional. In considering these distributional patterns it is
important to keep in mind the overall pattern discussed earlier, under which
the pragmatic function was considered independent of the form employed.
This joint consideration reveals that My occurred more frequently than in
the overall distributional pattern for control acts, whereas J occurred less
frequently. The reverse pattern is found for assertions. This suggests that
the distributional patterns were not random, but, rather, motivated in part
by pragmatic considerations.»
 JEFFREY 79

GB Assertion
(_) Multifunctional
Control Act
Uncoded

Percentage
Use
of

FIG. 5.3. Distribution (%) of self 0% 5

reference forms in terms of prag- at tye akonee
matic function: Jeffrey Month I. Form

In the discussion of semantic characteristics of utterances containing self

reference forms, the analysis was confined to the level of the utterance. The
importance of going beyond the utterance level is revealed when one con-
siders examples like (21)(24). The analysis of the semantic characteristics
of My indicated that this form often linked up with utterances that were
agentive but also appeared in the context of less kinetic verbs and stative
verbs such as read and want. Given that My generally occurred in clauses
ranking high in agentivity, the question arises why My was also used in
utterances ranking low in agentivity. The argument put forth here is that
when these utterances are considered in terms of their pragmatic function
within the context’ of the activity frame, they can be seen to function as
control acts. Consider the following example:

(25) Jeff Q) and Eric (E) are looking at photo-book. Researcher (R) sits
nearby.
a. J: All done. (closing book and placing it on couch)
b. R: Um-hmm.
c. E: All done.
d. J: |My read another book. (looking at researcher)
e. R: Well there is'no other book. Do you guys wanna build some
more?
f. J: No. (shaking head)
g. R: Are you all done then? Do you wanna go back to your school
or do you guys wanna build one more thing for the helicop-
ter?
h. J: (grabs helicopter and plays with it)

Out of context one might assume that Jeffrey’s utterance in line (25d) was
a report of an ongoing activity. Instead, when it is viewed in the flow of
discourse, we see that the child was stating his desire to carry out an activity
80 CHAPTER 5

in the immediate future. His utterance functioned as a control act by which

he aimed to bring about the reading of another book. When the researcher
rejected his request and suggested that the children carry out an alternative
activity, Jeffrey protested (see line (25f)), though we find in line (25h) that
he changed his mind, grabbing for the helicopter before his peer gained
possession of it.
We can next consider examples (22) and (23) within the discursive context
in which they emerged:

(26) Researcher (R) brings tray of blocks and helicopter to table while
Jeff (D and Eric (E) watch.
a. E: Blocks. (looking at tray of blocks)
b. J: My want cars. (pointing up in the direction in which the cars
are kept out of reach of the children)
c. R: Well there aren’t cars now, there’s a helicopter and blocks.
d. J: My want that. (reaching for helicopter E holds)
e. J and E struggle over helicopter.)

The first use of My in line (26b) was directed to the researcher. At this point,
the researcher was bringing over the second activity task. In line (26b) Jeffrey
looked at the researcher and pointed toward the toy cars which he had
been playing with previously. The use of My often occurred at activity
junctures, when the child attempted via language to control the flow of toys.
These requests for particular activities or toys often continued for several
turns. In example (26), though, we find that the child switched his attention
to the helicopter his peer held. In line (26d) we find a similar example of
My with a stative verb. Although this utterance also functioned as a control
act, this time it occurred in the context of an ownership dispute. Jeffrey’s
use of My in control acts was not limited to one particular kind of speech
act type but, rather, cut across a range of acts that functioned as attempts
to influence the hearer’s actions, as well as appearing in utterances where
the child’s action was dominant or in ownership disputes involving both
speaker and hearer.
My also occurred in utterances that were coded as multifunctional. Such
uses involved stative verbs (e.g., want, like), as is illustrated in the following
example:

(27) Jeff GQ) watches his mom (JM) build with blocks.
a. J: | What’s that? (pointing to JM’s structure)
b. JM: Well ©) how about (.)
c. J: | That’s a house. (pointing to structure)
d. JM: That’s a house. (placing block) Is that a house? (lifting up
another block and placing it on the structure)
JEFFREY 81

That’s no—now it’s my turn to do it.

< Okay.
(whines)
m9
=O (searches through pile of blocks) My want something. (still
searching)
= Which would you like?
an
a
SS I would like -da like (.) don’t. (lifting block JM has placed
on structure)

In line 27h), it is unclear whether Jeffrey’s utterance containing My func-

tioned as an assertion, a control act (as Jeffrey’s mother interpreted it in line
(27i)), or a combination of both. This utterance is ambiguous because it is
unclear whether it functioned to update the child’s partner on his own
actions (e.g., ‘I’m taking my turn but I first have to find a specific block’)
or whether such an utterance was used to solicit his mother’s help. The sort
of ambiguity found in the multifunctional utterances is reminiscent of Burke’s
(1969) discussions of the stages of action and, in particular, his treatment
of incipient acts. Burke suggested: “The concept of incipient acts is ambigu-
ous. As an attitude it can substitute for an act—it can likewise be the first
step towards an act” (p. 236). Mead (1934, 1938) discussed the sort of
ambiguity involved in the expression of attitudes, suggesting (like Burke)
that one can view them representationally and at the same time as “the
beginning of acts” (Mead, 1934, p. 5).
Burke’s and Mead’s notions shed new light on the use of My in multi-
functional utterances. I would suggest that these utterances were categorized
by the child as similar to those containing My that functioned as control
acts. When expressing attitudes, the child treated such utterances as begin-
nings of action sequences rather than as substitutions for action. Yet, as
Mead (1934, 1938) pointed out with regard to gestures, such usage does not
necessarily entail consciousness of action. This use of My can be viewed in
terms of what Gumperz (1976, 1982) referred to as contextual cues: “Con-
stellations of surface features of message forms are the means by which
speakers signal and listeners interpret what the activity is, how semantic
content is to be understood and how each sentence relates to what precedes
and follows” (Gumperz, 1982, p. 131). The use of My then informs the
partner about the contextual presuppositions.
The distinction between the expressions of attitudes that function as rep-
resentational and those that function as afirst step in an action sequence is
clearly a subtle one. Such a distinction presents not only a problem for the
researcher coding the data but also an online problem at the level of text
production for the child. It is interesting to note that nearly all of the oc-
currences of J and My that were involved in hesitations or self-corrections
were exactly the sort involving the expression of attitudes. This suggests
82 CHAPTER 5

that Jeffrey himself realized that the distinction was not all that clear. Nev-
ertheless, the child’s selection of the My form indicates that the utterance
was viewed as part of an action sequence. Relating such uses to those
involving My as a marker of agentivity, it is intriguing to note that direct
physical action on the part of the child did not-need to take place for the
My form to be employed. The expression of internal states, when viewed
as the introductory phase of an action sequence, also prompted the use of
the My form.
Thus far, I have considered three pragmatic functions of utterances with
My. I have noted that most frequently My functioned in assertions. But I
have also noted that My appeared in conjunction with control acts, and in
addition, it appeared with the expression of attitudes that marked the first
step in an action sequence. Although I return to the question why the child
often used the My form in conjunction with assertions, at this point, we can
note that what the remaining two functions of My share is the instrumental
use of language. These utterances function not so much to represent reality
as to create it.
I, in contrast to My, has been noted to appear primarily in assertions.
This contrast can account for the distinction in the minimal pair J wear it
and My wear it, in examples (15) and (24). Situating these examples in the
context of the child’s ongoing activities, the pragmatic distinction between
these utterances becomes clear:

(28) End of play session involving Jeff J) and mom (JM). Researcher
(R) is removing the wireless microphone from JM.
a. J: Nono my turn. (taking microphone from JM)
b. R: You want me to put it on you for a minute?
cnjem CYeah:
d. JM: You wanna wear it?
e. R: Okay. (begins to clip microphone on J)
f. j: I wear it. (smiling while looking at microphone.)
g. (R converses with JM.)

(29) Jeff Q) is looking through photo-book with Eric (E). Researcher (R)
sits nearby.
a. J: What’s that? (holding microphone that was hidden behind
couch)
Oh it’s just some equipment.
That’s Eric (pointing to photo in book)
Um-hmm.
My wear it. (gives microphone to researcher)
You wanna wear it?
ao<a
sgiroeae Yeah.
JEFFREY 83

h. R: Okay, I’ll pin it right here. (as she is placing microphone on

Jeffrey)

In example (28) the use of J in line (28f) was linked with an assertion. The
utterance came at a point in the discourse when control of the microphone
was not at issue. The child was already wearing it. Example (29) revolved
around a similar event, though in line (29e) Jeffrey’s utterance functioned
as a control act. The child requested that the researcher place the microphone
on him. Although at the level of the utterance these two examples look
similar, when placed in the stream of discourse it becomes clear that they
differ in terms of the pragmatic attitude expressed.
One might want to conclude that My occurred in discourse with future
orientation, whereas J might occur in discourse about ongoing or previous
happenings. Although this was often the case, there are instances of J in
discourse with future orientation; and the contrast between the uses of /
and My with future orientation is informative. The uses of My with future
orientation tended to occur when the child was in the middle of another
activity and anticipated what was to come next. For example, My read the
story was uttered while the child was playing with the blocks. Jeffrey said
this while looking over in the direction of the photo-book, which was resting
on the desk where the researcher was sitting. In another example, the child
was watering his plant at the beginning of the session; and while looking
across the room at the table that had been set up with some toys, Jeffrey
said, “My play with the helicopter.” In such cases of My with future orien-
tation, the child did not have sole control over the instigation of the event
he wanted to take place—he alone could not bring the activity about.
Utterances involving My with future orientation contrast with similar ut-
terances containing J, because the utterances with J always took place when
the child presupposed control. In such cases the partners had often nego-
tiated a joint activity, and the child’s utterance elaborated on an established
plan. In one example, Jeff and his mother decided to “make food” by filling
containers with dried peas. In the process of the activity Jeffrey suggested,
‘Tll make breakfast for you,” and he continued to “prepare food” on his
own. With regard to the cases of J with future orientation, the child did not
wait for the partner’s intervention to begin carrying out the activity referred
to in the future-oriented utterance.
Presupposition also plays a central role in the contrast between Jand My
with reference to control of objects. Many cases of J with future orientation
occurred when Jeffrey discussed how he planned to distribute toys he con-
sidered himself to be in control of, as in the following example:

(30) Jeff (QD and mom (JM) are building with blocks.
a. J: And/ And you have that. (giving JM a block from the block
pile)
84 CHAPTER 5

b. JM: Okay.
c. J; And I'll have this. (takes block for himself)

The utterance in line (30c) contrasts with that containing My in example

(26d). Although J was used when Jeffrey presupposed control, he switched
to My when his partner (most often his peer) was in control of the objects
he would like to possess. Thus the child expressed his desire to play with
an object differently, depending on whether his partner had claim to it or
the object was not in use.
In summary, the results of the analyses of pragmatic function reveal that
the two self reference forms appeared in utterances serving distinct com-
municative functions. My occurred more frequently than / in control acts or
when Jeffrey expressed the first step in a causal sequence in which he used
language to get others to act in particular ways. J, in contrast to My, appeared
most frequently in assertions that were descriptive in nature. The pragmatic
analyses have helped account for examples that could not be accounted for
by the semantic analyses reviewed in the previous section. The question
remains: To what extent are the semantic characteristics and pragmatic factors
that link up with the use of the particular self reference forms interrelated?
This question is pursued in the section that follows.

Core Field of Use and the I/My Contrast. Thus far, I have focused
on two separate layers of analysis. Do the findings reviewed so far reflect
two different meanings of each of the self reference forms, or can they be
brought together under some broader categorization scheme? I claim that
the co-occurrences between the individual forms and parameters at the
semantic level of analysis are in certain ways similar to the connection
between the self reference forms and the co-occurrences at the prag-
matic level. The semantic and pragmatic features cluster together and are
best viewed as falling under one broad categorization on the part of the
child.
If the uses of Jand My are linked up with meaning clusters that cut across
semantic and pragmatic boundaries, one can formulate predictions for a
combined analysis of agentivity and control. When simultaneously consid-
ering semantic and pragmatic factors, My should be the preferred form in
utterances that rank high on the agentivity scale and function as control
acts. Similarly, J should be the preferred form in utterances that rank low
in agentivity and function as assertives. No predictions can be made about
the ways the forms would link up with utterances that, for instance, rank
high in agentivity but function as assertives, or utterances that rank low in
agentivity but function as control acts. An analysis of these instances would
reveal what Jeffrey did when the semantic and pragmatic factors were in
conflict.
JEFFREY 85

The findings of the joint analysis support the predictions. My was found
in 100% of the instances in which the utterance ranked high in semantic
agency and functioned as a control act. In addition, 77% of all utterances
ranking low in semantic agency and functioning as assertions included J
The fact that 85% of the utterances ranking high in semantic agency that
functioned as assertives included My suggests that Jeffrey seemed to place
more weight on semantic components than pragmatic ones in selecting a
form when the semantic and pragmatic levels were at odds.
By way of summary, the reader can reconsider the set of utterances listed
as examples (1)-(10). At the beginning of this chapter they were put forth
to illustrate what has been referred to as Jeffrey’s core contrast. Although at
first glance these forms did not appear motivated, a series of distributional
analyses that took into account semantic and pragmatic function indicated
that the two forms did not occur in free variation but, rather, linked up with
broader activity types. Based on these analyses, we are in a better position
to return to Jeffrey’s use of the other self reference forms.

Other Self Reference Forms

The distributional analysis of self reference forms reveals that 84% of all of
Jeffrey’s references to self involved the core contrast discussed here (see
Fig. 5.1). The remaining instances involved Me (6%), Own Name (3%), and
We (3%), as well as a small array of Other forms (5%) including meself,
mine, and let’s. Because the use of We appeared to be nonproductive at
this point (occurrences were either direct imitations, incomplete utterances,
or uses with another self reference form, e.g., We my hide—for which no
pattern could be found), I do not discuss it further here. In addition, because
the combined occurrences of Other forms made up only 5% of all self
reference forms, these examples are also not discussed. Rather, the discussion
briefly focuses on the use of Me and Own Name, specifying how these two
forms contrast with one another as well as with the other self reference
forms discussed earlier.
To a certain extent, the uses of Me and Own Name look quite similar.
Neither of these forms occurred in utterances with verbs, and thus they are
distributionally distinct from J and My. These forms then have not been
coded in terms of agentivity, because they were not involved in utterances
overtly marking agent-action relations. In addition, both forms at first glance
appeared to serve a similar function. They both appeared in assertives in-
volving naming and identifying:

(31) Jeff GQ) and mom (JM) look through photo-book.

a. JM: Who’s that? (pointing to picture of J)
Deieenicy.
86 CHAPTER 5

(32) Jeff (J) and Eric (E) have been playing with manipulative toys.
Researcher (R) is clearing off table.
a. J: This/ This for you. (holding bowl of dried peas)
b. R: Who made it?
cpio Me:

Although both Me and Own Name occurred when the child was looking
through the photo-book, Own Name is limited to this activity context, whereas
Jeffrey used Me in other activity tasks. Although both forms occurred in
utterances that functioned to name and identify, a subtle difference can be
found in their use. Own Name seemed to be limited to instances of a labeling
game played by communicative partners. In such sequences, action was
backgrounded and focus was placed on giving names to pictorial repre-
sentations. The naming game was not restricted to references to the self; the
child also attached names to other pictures (e.g., That’s Anna). In contrast, Me
occurred in discussions of who carried out various actions. In such contexts,
naming was less relevant than was supplying information that allowed the
partner to distinguish objects and people from one another. Because the
numbers are low, I do not discuss further how Mé and Own Name were
employed. Instead, I turn to a general summary of Jeffrey’s system of self
reference.

Summary of Month I

In the discussion of Jeffrey’s use of self reference forms I noted that Jeffrey
relied on a number of forms to refer to self in a systematic way. The dis-
tinction between the different forms was said to be based on the view of
self in a given activity context. The core contrast, that between J and My,
marked a contrast between self acting as volitional agent to bring about a
change in the world either by direct physical effort or via language (My)
and self as experiencer of inner states or doer of less agentive actions (J).
Notions of degree of agentivity and pragmatic function both played a central
organizing role in such a contrast. The use of Me represented a perspective
in which the self or items related to the self were identified without neces-
sarily focusing on the self’s agentive role in action frames. Finally, Own
Name was employed in acts of naming, particularly in labeling games where
names were given to people and things. In such cases no reference was
made to the self’s role in action.

MONTH II

During Month II, Jeffrey continued to be categorized as an ego-anchored

child. Although his mean MLU jumped to 3.52 during this month, his refer-
ences to self constituted 80% of all references to main participants (see Fig.
JEFFREY 87

4.1). In addition, we see later that many of Jeffrey’s uses of self reference
forms continued to deviate from adult usage. Despite the fact that he con-
tinued to be ego-anchored, I note some aspects of his self reference system
that were developing during the second month of the study.
During Month II, Jeffrey continued to rely on two core self reference
forms, J and My. As is shown in Fig. 5.4, these two forms occurred at similar
levels and together represented 82% of all self reference forms during this
month of study. I now consider whether Jeffrey contrastively employed these
two forms in Month II and how such usage compares to that in Month I. I
do not consider aspects of Jeffrey’s usage of the other forms of self reference
because there were few interesting developmental changes in these forms
during Month II of the study. The only change worth noting is that Me began
to be used in utterances with verbs, in subject and object position. Because
such usage was infrequent, however, I do not discuss these changes further.

The Core Contrast: J and My

Surface Characteristics. Table 5.2 shows distributional characteristics

of the various self reference forms. I restrict my analysis here toan examination
of differences between Month I and Month II for the two core self reference
forms. Here we find minimal differences from Month I of the study. In Month
Il, Jeffrey used all of the forms more frequently than in Month I in utterances
with verbs. Although J was used more frequently than My (94% vs. 71%),
both forms typically occurred in a context with a verb. In addition, both forms
continued to be used in similar positions in utterances with verbs. J continued
to occur exclusively in subject position, and My was used primarily in subject
position (alone) but occasionally appeared in object position with a noun. And
both Jand My tended to appear in multiargument structures.

Semantic Characteristics. | begin to investigate whether the core self

reference forms link up with particular degrees of semantic agency by
considering the overall distribution during Month II of talk about high and low
agency, regardless of form employed. Here we find that in Month II, there was

141%

My 41%

FIG. 5.4. Distribution (%) of self

reference forms: Jeffrey Month II (1
= 64),
88 CHAPTER 5

TABLE 5.2 et
Distribution (%) of Self Reference Forms According to Surface Characteristics
(Jeffrey Month II)**

(a) Verb / Verbless Context

FORM With Verb Without Verb Uncoded Subtotal*

| 94 2 5 (65)
My 71 29 (68)
Me 44 56 (9)
Own Name 25 18 (4)
Subtotal* (14) (29) (3) (146)

(b) With Verb: Position

FORM Subject Object Uncoded Subtotal*

I 100 (61)
My 85 8 6 (48)
Me 50 50 (4)
Own Name 100 (1)
Subtotal* (105) (6) (3) (114)

(c) With Verb: Argument Structure

FORM Single Argument —_ Multiargument Uncoded Subtotal*

I 15 82 3 (61)
My 13 88 (48)
Me 50 50 (4)
Own Name 100 (1)
Subtotal* (18) (94) (2) (114)

(d) Without Verb: Context

a a ee eee

FORM +Noun Alone Other Subtotal*

| 100 (1)
My 5 10 85 (20)
Me 60 40 (5)
Own Name 100 (3)
Subtotal* (1) (94) (23) (29)
a ee ee
* (Raw numbers).
** This table excludes 15 instances of other.

an overall increase in the proportion of talk about high agency and a

simultaneous decrease in talk about low agency. In Month II, 35% of all uses
of self reference forms ranked high in agency, whereas in Month I only 25%
did. In contrast, during Month II 63% of all uses of self reference forms ranked
low in agency, whereas in Month I 73% ranked low. Thus we see that during
this month of the study, Jeffrey showed an increase in his interest in talking
about self as agent, though he continued to talk about himself primarily in a
nonagentive way.
JEFFREY 89

The analysis of J and My during Month Irevealed that Jeffrey systemati-

cally distinguished between these two forms, in part on the basis of the
degree of agentivity expressed. My tended to be employed in utterances
ranking both high and low in agentivity, whereas J occurred in utterances
that ranked low. In Month II we also find that J dominated the low end of
the agentivity continuum, although this difference was less pronounced than
in Month I (see Fig. 5.5). In Month I, 90% of all uses of J ranked low,
whereas in Month II only 66% of all uses of J ranked low. The use of J
linked up with utterances ranking high in agentivity 34% of the time.
With regard to the use of My, we find that the trends were similar to
those in Month I, though they were less pronounced (see Fig. 5.5). Thus
although My continued to be used in utterances ranking both high and low
in agentivity, in Month II My (ike D appeared primarily in utterances rank-
ing low in agentivity. We find that 67% of all uses of Myranked low, whereas
31% of utterances containing My ranked high in Month II. In Month I, 58%
of the uses of My ranked low, whereas 42% ranked high. Thus in Month II,
both forms were used not only in utterances ranking low in agentivity but
also in utterances ranking high. The following examples illustrate that both
forms occurred in what appeared to be similar semantic contexts ranking
high and low in agency:

(33) I'll make a cake all by meself.

(34) My make a house with/ with these/ these blocks.
(35) I want that.
(36) My want that.

I return to further distinctions in such minimal pairs. For now, I simply note
that there no longer appeared to be a semantic distinction between the two
core self reference forms.
100%
WM Low Agency
90% C__) High Agency
HE Uncoded
80%

a
Use
of
Percentage

FIG. 5.5. Distribution (%) of self sg ——

28838 ai
reference forms in terms of seman- n=59 n=48
tic agency: Jeffrey Month II. Form
90 CHAPTER 5

Pragmatic Characteristics. Turning to a consideration of pragmatic

function, we can first note that there was no real change in the second
month with regard to how Jeffrey’s language functioned. Without considering
the use of particular forms and generalizing instead across all self reference
forms, we find that Jeffrey continued to divide his use of language functions
much as he did in Month I. In utterances containing self reference forms,
his language functioned primarily to make assertions about the world (60%),
whereas his use of control acts was less frequent (17%). The general distri-
bution of pragmatic function (regardless of form) looked quite similar to
that in Month I.
Taking into account how Jeffrey linked the core self reference forms with
the various communicative functions, we also find a pattern similar to that
described in Month I (see Fig. 5.6). Jeffrey continued to use / in utterances
that functioned as assertions, though in Month II the link between / and
assertive function was even stronger. With regard to My, the distributional
patterns remained primarily unchanged. My continued to be favored in ut-
terances that functioned as assertions (41%) but was used more frequently
than / in utterances that functioned as control acts (25% of all instances of
My). The only major change in Month II with regard to Jeffrey’s use of My
along pragmatic dimensions was his increased use of My in utterances coded
as multifunctional. Thus in Month II, there were some distributional differ-
ences in terms of how Jeffrey linked the use of J and My with various
language functions, though there were no major developmental differences
in such use between the first two months of the study.

Core Field of Use and the I/My Contrast. In Month I I noted that
Jeffrey used J and My in utterances that differed in terms of their semantic
and pragmatic functions when these were considered as separate layers of
analysis. Moreover, when both terms were considered together, one could
make predictions about which form would appear in utterances that, for

100%

90%

80%

70%

60%

50%

40%
Use 30%
of
Percentage
20%
10%
0% FIG. 5.6. Distribution (%) of self
reference forms in terms of prag-
matic function: Jeffrey Month II.
JEFFREY 91

instance, ranked high in semantic agentivity and functioned as control acts

(e.g., My) or ranked low in semantic agentivity and functioned as assertives
(e.g., D. The combined analysis revealed that Jeffrey’s use of such forms
matched the predictions, suggesting ways in which the forms linked up to
broader meaning clusters that cut across semantic and pragmatic dimensions.
Considering the findings from a similar analysis that combined the se-
mantic and pragmatic dimensions in Month II, we find that J continued to
dominate to about the same extent in utterances ranking low in semantic
agentivity and functioning as assertives. In contrast, in Month II we do not
find that My tended to dominate in utterances that ranked high in agentivity
and functioned as control acts. First of all, we find that Jeffrey rarely produced
utterances with these semantic and pragmatic characteristics (there are only
two examples involving self reference forms), and My never occurred in
Month II in these cases. Thus, a change that took place between Month I
and Month II was that My no longer was best characterized in terms of a
semantic and pragmatic prototype in which utterances ranked high in agen-
tivity and at the same time functioned as control acts.
The most dramatic change from the first to the second month of the study
for Jeffrey involved his incorporation of J into utterances ranking high in
semantic agentivity. The question remains: On what basis did Jeffrey employ
Tor My in such instances? Can the instances of J be distinguished from those
involving My? If one looks at the broader discursive context, one finds that
the new use of J with some clauses ranking high in agentivity during Month
II was of a very restricted sort. Such utterances appeared within a framework
of joint activities, whereas My tended to occur as the child introduced new
plans or attempted to bring about new activities. The following example
illustrates the contrast:

(37) Jeff GQ) and his mom (JM) are playing with a helicopter and blocks.
They pretend to drive the helicopter to the grandparents, whoJ says
(incorrectly) live in Berkeley.
a. J: They live in Berkeley.
b. JM: In Berkeley? Oh okay # that would be nice and close #
wouldn’t it?
c. J: No. My make a house with/ with these/ these blocks.
d. JM: Okay with these big red # blocks we'll make a house.
Cnc avealn
f. JM: Okay.
g. J: No we'll// Pll put these.
h. JM: xx And xx put these out of the way. (moving small blocks)
Pat No no I/ No I need the chimneys. (chimneys = small cylinder
shaped blocks)
j. JM: Oh okay. You need the chimneys.
92 CHAPTER 5

In line (370), Jeffrey suggested a plan to make a house with big blocks. His
mom agreed in line (37d) and turned Jeffrey’s plan into a joint endeavor.
Jeffrey offered no protest. It was at these junctures that Jeffrey no longer
used My in references ranking high in agentivity. In these instances, Jeffrey
switched to the use of 1 Thus Jeffrey took a different stance on the same
event sequence (e.g., placing a block, making a house) depending on
whether it was part of a jointly agreed upon activity frame or was inde-
pendently motivated.
Most of Jeffrey’s uses of J in Month II that ranked high in agentivity were
accompanied by a modal form (80%). In my previous discussion of Jeffrey
at Month I, I noted that modal forms were used with J but not with My
when referring to future-oriented actions. This pattern became more pro-
nounced in Month II. In addition, in 5% of the cases when J was not followed
by a modal form, it was used in utterances employing the pattern J+ My +
verb + object, for instance, I my make Daddy. In Month II we find that My
was integrated into a position other than subject position more frequently
and in some of these cases was preceded by
Taken together, these findings lead to two conclusions about the use of
the core self reference forms during Month II. First, My no longer occurred
in conjunction with reference to self as prototypical agent. Rather, the child
restricted the use of My to an activity scene that was guided by personal
motives. As soon as his actions became contextualized as part of a shared
plan, the / form was employed. Thus the contrastive use of Jand My must
go beyond a semantic level of analysis to include a consideration of discur-
sive context.
A second conclusion to be drawn regarding developmental differences
in the contrast between J and My during Month II is as follows. As the My
form was employed more restrictively in subject position, the child began
to reinterpret this form as being more similar to modal forms. Thus during
Month II, Jeffrey seemed to have called into question his neat contrast
between J and My noted during Month I, and he seemed to be experiment-
ing with alternative form-function pairings. I reconsider these conclusions
in light of what we find in our subsequent discussion of Jeffrey’s use of the
core self reference forms in Months III and IV.

MONTH I

During Month III, Jeffrey’s self reference system underwent radical reorgani-
zation. First of all, we find a change toward several characteristics noticed
in the nonego-anchored children’s use of self reference forms. As in Month
II, Jeffrey's MLU was above the 3.0 mark—his mean MLU during Month II
was 3.79. Second, we find for the first time that his overall proportion of
JEFFREY 93

references to self as main participant dropped below 50%; 48% of all of

Jeffrey's references to main participants were about himself. Third, as I later
note in more detail (see Fig. 5.7), Jeffrey no longer contrasted between
multiple self reference forms but, rather, primarily used J when referring to
himself. Finally, there were relatively few instances of self reference form
usage that deviated from the adult system. I briefly consider these various
changes in more detail.
I begin our consideration of changes with a more complete analysis of
the distribution of various self reference forms during the third month of
the study. In Month III, as is shown in Fig. 5.7, Jeffrey primarily used J to
refer to himself. Most of his use of self reference forms (52%) involved the
use of J. We can also note the dramatic decrease in the use of the form My
at this point. Although in Month II Jeffrey used My in 41% of all instances
of self reference, in Month III My appeared in only 9% of all instances of
self reference. A third change to be noted in the use of forms involves the
category Other. In contrast to Month II—where this category made up only
9% of all usage—in Month III Other accounted for 30% of all self reference
forms. A closer examination reveals that this was in part due to a single
form, We, which accounted for 18% of all self reference forms. Thus we
have further evidence that Jeffrey’s system was organized during Month III
in a way similar to that to be described for the nonego-anchored children.
They too drew agreater contrast between Jand We than between any other
pair of self reference forms.

The Core Contrast: J and My

As was noted earlier, Jand My no longer represented the core contrast for
self reference forms in Month III. Nevertheless, from a developmental per-
spective it is interesting to follow up on the relationship between these
forms during the third month of the study. Several questions emerge. First,

152%

Own Name 2%

FIG. 5.7. Distribution (%) of self Other 12%

reference forms: Jeffrey Month III
(n = 148),
94 CHAPTER 5

when My was no longer contrastively employed in subject position, do we

find J now used multifunctionally? Similarly, when My was used as in the
adult system, do we find it in multifunctional contexts? We turn to a brief
consideration of these issues.

Surface Characteristics. The discussion of surface characteristics is

brief, because I have already noted that Jeffrey no longer employed the
various forms with any regularity in subject position. J appeared in utterances
with verbs, as did We. Both forms appeared in subject position. My, with
rare exception, appeared in conjunction with a noun. It occurred in a variety
of sentence contexts. Me often appeared alone, but it also appeared in object
position in sentences with verbs. There were no longer any uses of Me in
subject position of the sort noted in Month II. From the standpoint of surface
characteristics, Jeffrey’s system looked similar to the target language.

Semantic Characteristics. 1 begin the analysis of semantic charac-

teristics of the various self reference forms by noting that there were no
important changes in the overall distribution of the various degrees of se-
mantic agency for the combined analysis of self reference forms. As was
the case in Month II, Jeffrey continued to refer to himself primarily with
utterances that ranked low in agency (61%), whereas only 34% of all utter-
ances containing a self reference form ranked high. This is interesting in
that when Jeffrey gave up his special usage of pronominal forms he might
have simultaneously begun using J more multifunctionally to refer equally
to both high and low agency. This was not the case.
Looking more specifically at the issue of the relation between individual
forms and the degree of agency, we find a pattern similar to that of the
overall distribution. Both J and We tended to be found in utterances that
ranked low in agency (J= 59%; We = 75%). There were only four instances
of My used with a verb, and such usage was split equally between high and
low agency. Thus it was not the case that when My was no longer used in
subject position, Jeffrey began using J to refer to self as prototypical agent
more frequently. Furthermore, it was not the case that when We was intro-
duced into the system, it appeared in utterances ranking high (ie., new
form, new function). Rather, all forms referred to the same degree of agency,
and thus there were no form-meaning patterns developed at this point.

Pragmatic Characteristics. During Month III we find that Jeffrey con-

tinued to use roughly the same proportion of assertions with self reference
forms (59% in Month III vs. 60% in Month ID. His use of control acts increased
slightly from 17% for Month II to 25% for Month III. The question is whether
the specific forms took on new distributional patterns as the system under-
went major reorganization. Figure 5.8 reveals these distributional patterns.
JEFFREY 95

100%

90% | GE Assertion
d (_) Multifunctional
80% F > Control Act
2 70% GE Uncoded
=)
5 60%
2 50%
sg ‘Oo

Cc
8 40%
®
Oo 30%
20%
10%
FIG. 5.8. Distribution (%) of self 0%
reference forms in terms of prag- to dhe a“
matic function: Jeffrey Month III. Form

We find that J continued to be used as in Month II with assertions. My was

used more frequently in assertions than in Month II, and as We was intro-
duced into the system we find that it was used primarily with assertions but
was found 26% of the time in multifunctional contexts in instances where
Jeffrey was asserting a plan with a simultaneous attempt to change the
environment. Thus although many of the instances of Jand We appeared
in similar pragmatic contexts, one distinction was the more frequent use of
We in multifunctional utterances.

Core Fields of Use. We can now consider the core fields of use of
each of the self reference forms. In Month III we find a dramatic shift in
the way My was used. It no longer linked up with the notions of agency
and control but, rather, was employed almost exclusively in possessive con-
structions. J was found primarily in assertions ranking low in agency, as was
the form We, which was introduced into the system at this time. The question
then becomes: On what basis did Jeffrey employ one or the other form? We
might suppose that the distinction was that joint reference was made with
We and reference to self alone was made with J A closer examination,
though, of the examples involving J and We shows that the difference was
not so straightforward. Often we find that Jeffrey used J when he and his
mother were jointly negotiating action sequences, and likewise we find We
employed when there was no clear joint reference.

(38) Jeff G) and mom (JM) are playing with manipulative toys. J
announces that he is making ice cream.J and JM discuss their favorite
flavors of ice cream and thenJ says:
a. J: I could dump it in here. (referring to pretend ice cream)
b. JM: Okay. Great.
c. J; And then we could make some more.
d. JM: Okay.
96 CHAPTER5

One might want to argue that once the plan was agreed on, Jeffrey switched
to the We form in line (38c). This account, though, does not hold for other
sequences:

(39) Jeff J) and mom (JM) continue to play with manipulative toys. They
are negotiating a new sequence:
a. JM: This is a big car isn’t it? (lifts car)
b. J: No nol mean that’s the ambulance.
c. JM: Oh is that the ambulance?
d. J: Yeah:
e. JM: Is there a fire?
foe aNO:
g. JM: No? What's the ambulance doing?
h. J: Uh;/ uh we should; we should get this bowl instead of that
bowl. (reaching for another bowl)

In this example we could argue that Jeffrey and his mother were negotiating
a scene involving an ambulance, though by line (39g) they still did not have
a joint focus. One would have expected that Jeffrey would have used J in
line (39h) because he was working on his own plan. Because the examples
are few it is difficult to determine how Jeffrey contrastively employed J and
We. One important distinction is that the two forms appeared in utterances
with different modal forms: J usually appeared in conjuriction with wanna,
whereas We appeared with needa and hafta. I return to this contrast when
I consider Jeffrey’s self reference system during Month IV.

Other Self Reference Forms

Because Jeffrey’s use of other self reference forms accounts for only 21% of
all occurrences, the discussion is brief. Here I note that Own Name appeared
exclusively in the context of picture labeling. Me accounts for 7% of all
occurrences of self reference forms. Jeffrey used this form with his peer to
claim actions he would like to carry out and when commanding the other
to allow him to act. For instance, Jeffrey said, “No no me,” as a command
to his peer to allow him a chance at turning the page of the photo-book.
There was only one instance of Me in object position in a construction with
a verb. This too was a command: And watch me. Thus Me appeared in a
limited range of activity contexts.
What is particularly interesting about Jeffrey’s use of other reference forms
is that several new forms were added to the system this month, and they
appeared in contexts in which My appeared in previous months. Most no-
tably, he used the word mines to claim or try to maintain control of objects.
Thus we see that as an old form faded out, new forms were integrated into
JEFFREY 97

the system to take on the old function. The reorganization, though, cannot
be described simply in terms of a new form taking on an old function,
because we find hierarchical integration at the same time as differentiation.
Not only did Jeffrey now use mines to gain control of objects, but he also
used a distinct form, me, to claim rights to actions. Furthermore, in this
month we find new joint self reference forms such as /et’s, which was found
in directives as a way to bring the partner into a shared plan. Thus, in Month
III Jeffrey no longer relied on my as a global form of control but, rather,
distinguished several different kinds of control with separate forms.
In sum, Jeffrey was not simply working toward multifunctional usage of
each individual self reference form. He had a rather complicated system that
underwent changes at the level both of form and of function. Old functions
were expressed in more differentiated ways by the introduction of new
forms, and old forms took on functions that were previously associated with
separate forms. New contrasts were introduced into the system as well. We
turn now to a brief discussion of the further developments in Jeffrey’s system
during Month IV of the study.

MONTH IV

In the final month of the study, Jeffrey’s use of forms continued to look
similar to that in Month III. His MLU stabilized with a mean of 3.68. His
proportion of references to himself as main participant remained at 48%,
and he relied on / in 53% of all his uses of self-reference forms. These
characteristics, combined with the fact that there was only one example of
My appearing in subject position in a way that deviate from the target
language, suggests that Jeffrey’s system was more like those of the nonego-
anchored children at this point in time.
Considering Jeffrey’s overall distribution of self reference forms, we find
few differences from Month III. As is shown in Fig. 5.9, Jeffrey primarily
used one self reference form, J. The use of My increased slightly in Month
IV, from 9% in Month III to 13% in Month IV. The use of the other self
reference forms do not show any distributional differences between Months
III and IV. I now consider briefly the way such usage links up with various
other characteristics, focusing exclusively on differences between Months
III and IV rather than providing a complete analysis.

Surface Characteristics. In Month IV, we can note the following

changes in surface characteristics of the various self reference forms. First,
Iand We both appeared in subject position, whereas My occurred only once
as a subject pronoun that was not part of a possessive construction. It is of
interest that this use (My have my mommy) occurred in a moment of urgency,
98 CHAPTER 5
153%

FIG. 5.9. Distribution (%) of self

reference forms: Jeffrey Month IV
(n = 182).

when the child stated he needed to go to the bathroom and demanded that
his mother accompany him rather than the researcher. It is possible that in
such cases where the child quickly attempted to make his point, he fell
back on previously exercised pairings. The only other change in surface
characteristics that ought to be noted was that during Month IV, Me began
to be used regularly in object position. Recall that in Month III, Me appeared
almost exclusively in utterances without a verb.

Semantic Characteristics. With regard to semantic characteristics,

there was very little in the way of change. Jeffrey continued to refer to self
primarily in terms of low agency throughout this session. What is interesting
is a contrast between the use of J and that of We at the semantic level (see
Fig. 5.10). Although J continued to refer to low agency 69% of the time, we
now find We referring to high agency 41% of the time. Thus we see a shift
from My referring to high agency in Month II, to no form referring to high
agency in Month III, and an increased use of We referring to high agency
in Month IV. Though both J and We appeared in utterances rating low in
agency, We was more likely than / to refer to self as prototypical agent.

=o”
100%
: GB Low Agency

80%

70%

60%

50%

40%
Use
of
Precentage
30%

20%

10%

0% FIG. 5.10. Distribution (%) of Jand

ie Bey, We in terms of semantic agency:
Form Jeffrey Month Iv.
JEFFREY 99

Pragmatic Characteristics. The overall use of language in Month IV

resembled that in Month III. Jeffrey used language primarily to make asser-
tions about self. There were two form-function patterns worth noting,
though, in Month IV. First, J appeared more frequently than in previous
months in utterances coded as control acts. We find that 23% of all uses of
Zin Month IV were control acts. This compares with 9% in Month II and
13% in Month III. One reason for this increase is that Jeffrey began using
conventional requests like Can I have x rather than relying on his earlier
strategies of My + desired object in Month II or mines in Month III. Another
interesting pattern was the increased use of We in assertions; although in
Month III, 59% of all uses of We were coded as assertions, in Month IV we
find 78% of all uses of We coded as such. The question to be raised is: What
is the distinction between the assertions with J and those with We? In part,
the answer to this requires a combined analysis of various coding levels, so
we turn to that in the next section.

Core Fields of Use. The primary change in Jeffrey's system that we

have seen over the four months of the study is the gradual deterioration of
the special //My contrast. At the beginning of the study I noted that J was
typically used in assertions that ranked low in agency and, in contrast, My
was used by Jeffrey to talk about self as a prototypical agent acting to bring
about change or control his environment. By the end of the study the core
use of My no longer marked prototypical agent or control. This form instead
took on its target usage as a possessive pronoun—occasionally in control
acts, though primarily in assertions. This, though, is not to say that J became
the only pronominal form used in subject position in a multifunctional con-
text. Over the course of the study, Jeffrey introduced a new form into the
system, namely, We. As the special use of My faded away, We began to be
used contrastively with Z Both typically appeared in assertions, though J
appeared in requests as well. My statements concerning the function of We
during Month III hold for Month IV as well. That is, at this early point, We
did not seem to mark the notion of joint agency per se. In fact, in Month
IV it seemed more likely than in Month III to be used when the child
described aspects of his personal plans. That is to say, We was even less
likely than before to refer to a notion of joint subjecthood. Most frequently,
when Jeffrey used We he was referring to his own independent actions he
was about to carry out en route to a broader goal. It appears that in such
cases We was employed to marka less volitional kind of agency rather than
a kind of agency based on personal motives. Consider example (40):

(40) Jeffrey GQ) and mom (JM) are playing with blocks. Researcher (R)
brings over more blocks.
a. J: Now look (calling to researcher)
100 CHAPTER 5

b. R: Oh look how colorful.

c. (pause)
d. JM: Should we put some little chimneys. on? (no action)
e. J: Yes. But not the ones with—we need another green. (pick-
ing up blocks)
f. JM: Another green one? (J and JM discuss shapes and colors of
blocks for the next five turns.)
g. J: Yeah a square like these.
h. JM: Okay. Where you gonna put the chimneys?
i. J: Now we hafta get the lights. (picking up and placing more
blocks)
j. JM: The lights. Oh right. J) continues building.)

Although at first glance one might assume that Jeffrey was contextualizing
his own actions as joint, several factors lead one to doubt this conclusion.
First, in the video the dyad was positioned in such a way that the child was
acting independently with his mother providing guidance from the side.
Second, Jeffrey never checked for his mother’s approval along the way.
Third, we find in line (40h) that even his mother referred to his actions as
his own, though in other cases (see line (40d)) she referred to joint agency
when she suggested a plan. What is common to these uses of We by Jeffrey
is that they typically occurred in sequences in which Jeffrey was building
an already announced structure. It seems that what Jeffrey was indicating
with the We form was a switch in attitude. The uses of J are found in the
contexts in which personal motives guided action, whereas We indicates a
more depersonalized agency—the motive for the child’s action involved a
broader plan, namely, to build that kind of structure. This finding fits with
those reported for Month III regarding the use of distinct modal forms with
Iand We. If We was used to mark a more depersonalized agency, it would
make sense that it would be found more often with forms like needa and
hafta. Likewise, it makes sense that if J expressed a kind of personalized
agency, it would co-occur with wanna. I conclude this discussion of Jeffrey’s
contrastive use of Jand We, because the examples are few. I return to a
similar discussion in chapter 8 when we turn to an analysis of the nonego-
anchored children, which lends further support to the sort of argument being
developed here.
Over the span of only 4 months, Jeffrey’s self reference system underwent
major changes. I have noted not simply the addition of new forms but also
a change in the distribution of various forms at the syntactic, semantic, and
pragmatic levels. What remained constant over the course of the study was
that Jeffrey organized the use of various self reference forms to situate the
self in distinct ways. What changed was the kind of stances that received
linguistic marking. Early on in the study, Jeffrey was concerned with the
JEFFREY 101

issues of prototypical agency and control. Later on, he became more inter-
ested in marking the various kinds of motivations of his actions (i.e., personal,
social norms, joint agency). As he developed additional linguistic means (for
instance, the use of modal forms) these marked distinctions previously in-
dicated by self reference forms. The changes that were described were
gradual. Although we saw dramatic changes in the system in Month III, we
nevertheless found important developments in Months II and IV, suggesting
that the transitions were both sudden and slow. Jeffrey appeared to make
small shifts in his hypotheses concerning the relationship between linguistic
forms and the functions that they served. I now consider the developmental
findings of the other children to see whether these patterns can be gener-
alized.
 GCHAHAP TER Sx

Grice

MONTH I

Grice, a 22-month-old girl, was the second youngest participant in the study.
At the onset of the recordings, she was also the child who had the least
advanced MLU; her mean MLU for Month I was 1.72 (see Table 4.1). Thus
at the beginning of the study Grice was just beginning to combine words.
With regard to her distribution of reference to self and other, Grice referred
primarily to herself as main participant at the onset of the study; 85% of all
such references were to herself (see Fig. 4.1). Even by the end of the study,
we see that she continued to refer to herself as the main participant in her
utterances. Like the other ego-anchored children, Grice used multiple self
reference forms throughout the course of the study. The major developmen-
tal changes involved her reorganization of the relationship between such
forms and the contexts in which they appeared. I now consider Grice’s
organization and reorganization of this self reference system across the 4
months of the study.
In contrast to Jeffrey, who relied primarily on two self reference
forms, Grice relied on three distinct forms when referring to herself at the
beginning of the study. As is shown in Fig. 6.1, 36% of all uses of self
reference forms involved My, whereas Me was used in 28% and J was used
in 25% of all utterances containing self reference forms. The only other
major form used was her own name, which she used only 10% of the time.
The discussion first focuses on the core contrast between J, Me, and My
and then turns to her use of other forms of self reference during the first
session.

102
GRICE 103

My 36%

>

FIG. 6.1. Distribution (%) of self

reference forms: Grice Month I (1
—2))s

The Core Contrast: I, Me, and My

Grice’s use of self reference forms, at first glance, looks similar to Jeffrey’s
use. First, we find that Grice also used the various self reference forms in
subject position in ways that deviated from adult usage (see examples (4)—
(5)). In addition, we find that one utterance often included multiple self
reference forms (see example (8)). Grice often used different self reference
forms in similar verbal contexts (compare examples (1) and (4)).

(1) I want nuts.

(2) I soup pour.
(3) My teapot.
(4) My want nuts.
(5) Me jump.
(6) Me to do!
(7) Me in there.
(8) I want my the blocks.

Surface Characteristics. At the first level of analysis I consider

whether there are distributional regularities in the surface characteristics of
the individual self reference forms used by Grice at the onset of the study.
Four findings are revealed in Table 6.1. First, the three core forms can be
contrasted in terms of whether they regularly appeared in utterances with
or without verbs. One of the forms, J, regularly appeared in utterances with
a verb (see example (1) and Table 6.1a). My typically occurred in utterances
without a verb (see example (3)), and Me was used in utterances both with
and without verbs.
Looking more specifically at the use of self reference forms in utterances
with verbs, we can note that two of the forms (/ and My) were used exclu-
sively in subject position, whereas Me is found in both subject and object
position (see Table 6.1b). Grice was more likely than the other two ego-an-
chored children to use Me in subject position in utterances with verbs (see
 TABLE 6.1
Distribution of Individual Forms (%) According to Surface Characteristics
(Grice Month |)
ee ee Ee eee

(a) Verb / Verbless Context

FORM With Verb Without Verb Uncoded Subtotal*

I 100 (18)
My 15 85 (26)
Me 45 55 (20)
Other 100 (1)
Own Name 14 86 (7)
Subtotal’ (35) (39) (72)

(b) With Verb: Position

FORM Subject Object Uncoded Subtotal*

| 100 (18)
My 100 (4)
Me 56 44 ; (9)
Other 100 (1)
Own Name 100 (1)
Subtotal* (28) (4) (1) (33)

(c) With Verb: Argument Structure

FORM Single Argument Multiargument Uncoded Subtotal*

| 11 89 (18)
My 100 (4)
Me 100 (5)
Other 100 (1)
Own Name 100 (1)
Subtotal* (8) (21) (29)

(d) Without Verb: Context

FORM Subject Object Other Subtotal*

\ (19)**
My 73 9 5 (11)
Me 36 64
Other
Own Name 33 67 (6)
Subtotal* (18) (10) (8) (36)
ee Be a
* (Raw numbers).
** Three uses of My are uncoded; one use of Me is uncoded.

104
GRICE 105

example (5)). A further distributional pattern concerns the relationship be-

tween the various forms and the argument structure of clauses with verbs
(see Table 6.1c). Both Jand My (when used in clauses with verbs) appeared
in multiargument clauses; in contrast, Me appeared in clauses with a single
argument (compare examples (1), (4), and (5)). Finally an analysis of the
uses of core self reference forms in utterances without verbs reveals that
only My and Me appeared in clauses without verbs, and these showed
distinct distributional patterns: My usually (but not always) appeared in
connection with a noun (see example (3)), whereas Me appeared either
alone or in some other context (see example (7)).
There are two points to be noted with regard to the analysis of surface
characteristics of these forms. First, these findings indicate that the children
did not seem to be using the various forms randomly. Decisions about which
form was employed seem at least partially related to aspects of surface
characteristics of the utterances in which they appeared. A better under-
standing of how one is to interpret these findings—in particular, why certain
nonadult patterns might exist—is possible after an examination of the other
levels of analysis.

Semantic Characteristics. As was noted in the discussion of surface

characteristics of Grice’s use of self reference forms, the majority of such forms
occurred in utterances without verbs. As was shown in Table 6.1a, less than
half of Grice’s uses of self reference forms involved a verb. This is central,
because forms appearing in verbless contexts were not coded in terms of
semantic characteristics. One should therefore keep in mind that only 15% of
Grice’s uses of My were coded in terms of the semantic characteristics. In this
section, I report the findings of our analysis in terms of whether the utterances
containing particular self reference forms ranked high or low in agentivity (see
Fig. 6.2).
Examining the uses of the core self reference forms in utterances with
verbs, we find that in general such utterances tended to rank low in agen-
100% =a Low

High Agency
SO% MB Uncoded
80%

70%

60%

23
Use
Percentage
of 30%

20%
10%
FIG. 6.2. Distribution (%) of self 0%
reference forms in terms of seman- els baa aed
tic agency: Grice Month I. Form
106 CHAPTER 6

tivity. Of all utterances coded, 79% ranked low, whereas 18% ranked high.
With regard to particular forms, we find that all three of the core forms
followed this pattern. Of all uses of J coded for semantic agentivity, 78%
ranked low, whereas 22% ranked high:

(9) I want nuts.

(10) I want read more book.
(11) I want Thomas.
(12). 1 put it in:

In contrast, all uses of My coded for semantic agentivity took the same form.
They all involved utterances with the verb want:

(13) My want nuts.

Although similar contrasts can be found in Jeffrey’s data, what is different

is that Grice never used My at the onset of the study, in utterances ranking
high.
Grice’s use of Me often occurred in subject position with verbs ranking
low in agentivity. The following examples reveal, though, that such uses
involved a different kind of verb:

(14) Me to do.
(15) Me jump.
(16) Me read book.

Although the verbs in (14)-(16) typically rank low in agentivity, they are
action verbs, in contrast to the state verbs so frequently associated with the
uses of Jand My ranking low in agentivity. The verbs linking up with Me
were verbs that referred to actions that deviated from prototypical agentivity,
to the extent that an agent did not bring about a change in some physical
object.
Summarizing the semantic characteristics of Grice’s system and comparing
that system to Jeffrey’s, we find that for Grice, all three of the core forms
tended to appear in utterances ranking low in agentivity. Only 18% of the
uses of these forms ranked high in agentivity, and such uses involved J or
Me but never My. Thus at the level of semantic characteristics, at the onset
of the study Grice’s system differed from Jeffrey’s. In contrast to Jeffrey, in
whose system we noted a form-meaning pairing, Grice did not seem to use
particular self reference forms with a particular degree of semantic agentivity.
One factor contributing to this finding could be that she so rarely used My
in utterances with a verb.
GRICE 107

Pragmatic Characteristics. We can now consider how Grice’s use of

self reference forms relates to pragmatic aspects of the utterances in which
they appeared. It is first helpful to consider pragmatic aspects of Grice’s use
of self reference forms in general, apart from a consideration of specific
linguistic forms. Such an analysis reveals that Grice used language to serve a
variety of functions. Control acts made up 38% of all utterances involving self
reference forms, whereas noncontrol acts made up 32% of all these utterances,
and 22% of utterances containing a self reference form were coded as
multifunctional. With these distributional regularities in mind, we can consider
whether the particular forms linked up with specific function categories.
As is shown in Fig. 6.3, an examination of the three core self reference
forms reveals support for the claim that Grice’s use of self reference forms
bore a relationship to particular pragmatic functions. Looking across the
forms, we find that J tended to link up with utterances coded as multifunc-
tional (76% of all uses of J), My linked up primarily with control acts (70%
of all uses of My), and Me was used with both control acts (55%) and
noncontrol acts (40%). I now consider the individual forms and the relations
between them, in light of the communicative context.
Grice typically used J in conjunction with the verb want in utterances
such as I want that. These utterances were primarily multifunctional, to the
extent that the child stated her desires with what appeared to be the hope
that others would respond to them:

(17) Grice (G) and mom (GM) have been building with blocks.
Researcher (R) brings over a second tray of small blocks.
Look at these.
(disappears under table)
You want more Grice? You like these little ones?
I want nuts (standing up)
What?
1
©
>
®
BO I want nuts. (nodding head yes)
giemg(explains to G that she can play with nuts later)
100%
GE Assertion
oe (_] Muttifunctional
80% E>} Control Act
Uncoded
2 70%}
3
5 60%
B& 50%
ce
B 40%
®
a. |
20%
ea 10% |
FIG. 6.3. Distribution (%) of self
reference forms in terms of prag- o y le
: , ‘ n=18 n=26 n=20
matic function: Grice Month I. ton
108 CHAPTER 6

The use of J typically occurred in situations where the child wanted to bring
about a change but was physically unable to do this without assistance. It
is important to note, though, that at the time the child stated her desire to
bring about such a change, the change was not the focus of dispute. The
use of J in examples (17d) and (17f) contrasts with the remaining 24% of
the uses of J that occurred in utterances coded as noncontrol acts. In these
instances, the child indicated her desires but did not do so with the intention
of getting others to help bring about the change:

(18) Grice (G) and Megan (M) play with manipulative toys while the
researcher (R) sits nearby.
G: Soup. (placing nut in spoon, then in mouth)
R: Just pretend. Take that out of your mouth.
G: (complies)
(Peers play independently.)
G: I want that. (en route to and lifting nut)
moaogs (Peers continue to play alone.)

In example (18e), Grice was already moving in the direction of the object
she desired, and she took control of it without external assistance. In the
utterances in which J functioned as a noncontrol act, the child seemed to
be either informing her partner about motives for her actions or regulating
her own behavior with these utterances.
Grice’s use of My appeared in quite different pragmatic contexts. These
uses are found in particular kinds of control acts, namely, when focus was
placed jointly on speaker’s and hearer’s ongoing actions. These uses surfaced
primarily in disputes over objects, and often (as was noted in the previous
section) the child used My simply in conjunction with a noun referring to
the object she desired:

(19) Grice (G) and Megan (M) are having atea party.
a. M: (approaches G, who plays with teapot and then places spoon
in teapot G is using)
Do want some Meggy’s. (taking teapot from G)
My teapot! (pulling teapot back from M)
My teapot!
Wait! (Both girls tug on teapot.)
o
mono (G lets go, stamps feet, and shows signs of distress.)

(20) Researcher (R) is setting up block activity for Grice (G) and Megan
(M).
a. R: Here are these things. (bringing over helicopter and toy car)
b. M: Me airplane. (both children reaching for helicopter)
GRICE 109

c. G: My helicopter. (both holding and tugging)

d. R: You guys share.
e; (M holds plane; G yells, then begins to build.)

As is shown in these two examples, Grice typically used My in disputes over

toys, in particular those introduced by the researcher that were salient and in
limited supply. It is of interest that she did not use Myin other kinds of control
acts, in contrast to Jeffrey (see chapter 5). Although Grice’s use of My in (18c)
and (19c) conformed to adult English, this was not always the case. For
instance we find examples like J want my the blocks (see (21h)), revealing that
the child was using My not simply in standard possessive constructions but
also in other utterances in which she attempted to gain control over objects.
Me appeared in utterances that functioned either as control acts or as
assertions. The following examples reveal that the context of Me in some
respects compared with that of My, to the extent that both forms appeared
in control acts.

(21) Grice (G) and Megan (M) are finishing their play with the
manipulative toys. The researcher (R) approaches to clear away the
toys and to introduce the block activity.
# Read bo(ok).
Me read bo(ok).
Do you guys wanna play with the blocks?
Yeah.
Let’s put all these on here.
Put everything # (R and children clean up toys.)
Great and I’ll give you guys the blocks.
I want my the blocks.
You want the blocks?
The blocks.
Fre
Bo
neo
Se
anOO
Ore
Pa Okay I'll get the blocks.

(22) Grice (G) and Megan (M) are playing with blocks. Researcher (R)
sits nearby.
Tower. (placing one block on another)
Tower. (placing block, which then falls)
Oh I want read book. (looks to R)
Tower. (continuing to build)
29999
Me to do! (looking at helicopter in M’s hand, with her own
oAangeT®.
hand open)
= (M ignores G, G leaves table.)

These examples reveal that the uses of Me in control acts in examples (21b)
and (22e) were similar to some of the uses of My. Both of these core self
 CHAPTER 6
110

reference forms were used when the child wished to gain control of objects
or sought to carry out certain activities. A major distinction between these
forms is that My was used when the child was actually involved in an
ongoing dispute over control, whereas Me appeared when Grice attempted
to get others to give her control when actual struggles were not involved.
Example (22e) contains the only use of Me in an utterance that came close
to a dispute. The other uses of Me functioned as directives to others to help
the child achieve particular objects or goals.
Finally, we can consider Grice’s use of Me in utterances coded as non-
control acts. In these utterances the child announced her ongoing activities
to others. Such uses never appeared with stative verbs such as want, rather,
Me was used in conjunction with the expression of change of state or loca-
tion, or with activity verbs (see example (23a)). To this extent, Me contrasted
with the examples of Jalso described as associated with noncontrol acts.

(23) Grice (G) and Megan (M) have finished all the activity tasks. G looks
through the photo-book again, while M builds with the blocks.
a. G: (climbs off couch) Me in there. (walks towards door leading
back to main playroom of day-care center, voice tone of an
announcer)
b. R: (takes children back to playroom)

This analysis of the pragmatic function of utterances containing particular

self reference forms indicates that the three forms can be distinguished in
terms of pragmatic considerations. Even the use of Me, which seemed to
overlap at a general level with that of both of the other forms, can be
distinguished when one examines more specific aspects of the kinds of
control and noncontrol acts with which it was linked.

Core Field of Use and the I/My/Me Contrast. Thus far, I have fo-
cused on two separate levels of analysis. In this section, I consider the
interrelationship between these different layers of analysis in order to arrive
at a fuller understanding of the dimensions along which these different self
reference forms can be contrasted. We find that one can thus arrive at a
rich definition of the way in which individual self reference forms were used
by Grice. In contrast to the approach in the discussion of Jeffrey, I do not
emphasize a combined analysis of the semantic and pragmatic dimensions
because, as I noted earlier, Grice so rarely used My in conjunction with
verbs at this phase of development. Instead, I consider each form in terms
of a variety of verbal and nonverbal features that co-occurred with its use.
I first examine the contexts in which J was used. Thus far, I have noted
that J was used primarily in utterances with a verb, and that such utterances
ranked low in agentivity. Typically these utterances appeared in contexts in
GRICE 111

which the child stated desires in order to get others to bring about change.
On some occasions, Grice used J when making assertions about actions she
was Carrying out; thus / was used with utterances ranking high in agentivity.
The core use of J for Grice seemed to be to inform others about the child’s
states or plans, often, but not always, with the intention of getting others
to alter the environment in ways desired by the child.
I have noted that My often occurred in utterances that did not have verbs.
When My appeared in utterances with verbs, the utterances ranked low in
agentivity. At the semantic level, such uses look similar to uses of J, though
I have noted that differences between these two self reference forms can
be found at the pragmatic level. When My occurred in utterances without
a verb, the child typically named objects she would like to control. What
held the various uses of My together was pragmatic function. The function
of utterances containing My was to obtain or maintain control of objects,
most often those in limited supply and desired by others. In contrast to the
case for Jeffrey, the notion of agentivity was not central to the use of My.
In comparing the uses of Jand My, one might question what the difference
was between the use of J in utterances in which the child attempted to gain
the help of others to carry out particular plans, and the use of My in utterances
to gain control of objects. One major difference involves the role of the
other in such activity scenes. For Grice, J was used in events where the
other was more of an instrument, a coagent acting on the child’s behalf in
order to bring about desired goals of the child. In contrast, Grice used My
in scenes where the other was more of a counteragent, in particular when
self and other were in dispute over the control of objects. Thus, although
both of these forms were used to talk about the desires of the self, this
analysis reveals that different perspectives were taken on self and other
when using these forms.
The use of Me was more complex than that of the other two self reference
forms. I have noted that it appeared in utterances that included verbs as
well as in utterances that did not. When Me was used in utterances with
verbs, these utterances typically ranked low in agentivity, though they oc-
casionally ranked high as well. The pragmatic level of analysis revealed that
Me tended to be used in utterances functioning as control acts, though it
also was found in assertions. Thus, this form cannot be linked to a particular
functional cluster. Nevertheless, when one begins piecing together separate
aspects of the use of this form, a coherent meaning cluster becomes apparent.
When Me occurred in utterances without a verb, it was often used in con-
junction with a locative expression (Me in there). When Me was used with
a verb, this was an activity verb or a verb that expressed a change of state
(Me jump, Me done). What holds these different uses together is that in all
cases the child acted as an instigator of actions that were directed back onto
the self. The child not only was the first link in a causal sequence that
112 CHAPTER 6

brought about change but also was referring to events in which she would
be affected by the action described. This description can account equally
well for uses in scenes describing object transfers in which the child wished
to receive particular objects, and events in which the child both initiated
and was affected by actions. Grice’s use of Me appears quite similar to the
linguistic marking of the notion of affected agent in some languages (see
Saksena, 1980).
What is particularly interesting is that rather than marking a notion of
prototypical agency (a scene in which an agent acts to bring about a change
of state in some external object), as in Jeffrey’s use of My, we find Grice
using Me to mark a scene in which the self instigated action that was directed
back onto the self. I have noted that Grice did not mark the notion of
prototypical agency with the form My as Jeffrey did but, rather, reserved
this form to refer to a scene involving maintaining or the obtaining of control
over desired objects.

Other Self Reference Forms

Thus far, the discussion of Grice’s use of self reference forms has focused
on the core contrast of J, Me, and My. Such uses account for 89% of Grice’s
use of self reference forms during Month I. As can be seen in Fig. 6.1, the
remaining self reference forms consist primarily of instances of Own Name
(10%), with only 1% of all instances falling under the category of Other.
Given these distributional figures, my discussion of Grice’s use of self ref-
erence forms that are not considered part of the core contrast focuses simply
on her use of Own Name. Such usage usually occurred in a verbless context.
As was the case with the discussion of My, it is not particularly helpful to
examine the usage of Own Name in terms of the semantic agentivity coding,
because it so rarely occurred in utterances with verbs. I focus instead on a
consideration of the pragmatic function of utterances containing Own Name,
as well as the nonlinguistic features associated with its use.
With regard to pragmatic function, we find that all uses of Own Name
functioned in a similar way. All instances appeared in assertives that named
or identified. This usage is illustrated in the following examples:

(24) Grice (G) and mom (GM) look through photo-book.

a. GM: Can you find yourself? Where’s Grice? Is Grice in any of
these pictures?
by G: Yeah.
c. GM: Where’s Grice?
d.G: — (points to wrong picture)
e. GM: Look under your arm. (pointing to other page) Who’s that?
m"G: ~ "Grice:
GRICE 113

g. GM: What are ya playing with?

h.G: — Truck. (patting picture)
i. GM: A truck.

(25) Grice (G) and Megan (M) are looking though photo-book naming
pictures. Researcher (R) is holding book.
a. G: Meggy.
b. M: Meggy. Chitting page)
c. R: Mm-hmm.
d. M: Meggy.
e. G: Grice. (pointing to picture of herself)

In these examples we find that Grice used Own Name not only when
responding to her mother’s questions (see example (24f)), but also in spon-
taneous namings taking place with her peer (see example (25e)). It is par-
ticularly interesting to note Grice’s mother’s use of forms to refer to her
child. In example (24), Grice’s mother used the following forms to refer to
Grice: you, yourself, Grice, your, and ya. In this same stretch of discourse
Grice used one form: Own Name. Most of Grice’s uses of Own Name oc-
curred in the context of the caregiver-child interactions, though instances
such as that in example (25) can be found in the context of peer play. This
most likely has to do with the fact that for the mother-child dyad, the
photo-book activity centered primarily around labeling photos, whereas for
the peers the major task was one of negotiating who would hold the book
and turn the pages. As would be predicted, the peer session involved more
use of pronominal forms than of Own Name.
It should be noted that although Grice tended to use her own name in
labeling photos of herself, use of Own Name was not restricted to the
photo-book activity. Occasionally Grice used her own name to refer to her
ongoing activities while playing with other toys:

(26) Grice (G) and Megan (M) are playing separately with manipulative
toys at different corners of the table.
a. G: (Cooks across the table at plate of nuts)
b. Grice wants some. (intonation of announcement)
a (reaches across table towards nuts)

Although such instances were not frequent, they are nevertheless revealing.
They show that Own Name was not reserved solely for pictorial repre-
sentations of the self. In example (26), Grice appealed to the sort of labeling
genre linked with the use of Own Name. She used the form as a contextual
cue (Gumperz, 1982). In contrast to her uses of J in somewhat similar
contexts where she asserted a desire, in this context Grice called up a
114 CHAPTER 6

different frame. By using Own Name, Grice focused more on the naming
of her state, as in the referential frame called upon in the photo-book task.
The central use of Own Name—namely, to adopt a referential perspective-
could be extended to other situations in a metaphorical manner. This sort
of extension outside the photo-book session was not found in Jeffrey’s
references to self.

Summary of Month I

Grice’s use of self reference forms involved the contrastive use of four forms.
Three of the forms CU, My, and Me) were used fairly equally and accounted
for 89% of all instances of self reference. The remaining form, Own Name,
was employed 10% of the time. According to Grice’s system, all forms could
appear in subject position, though two of the forms (My and Own Name)
tended to appear in a verbless context. Although J often occurred in con-
junction with a verb, Me occurred in both verbless and verb contexts.
In comparing the use of the various forms along a number of dimensions,
we find the forms appeared in conjunction with slightly different kinds of
activity types. No one level of analysis could account for the contrastive use
of the self reference forms. Rather, the use of these forms was related to an
interconnected set of semantic, pragmatic, and nonlinguistic factors having
to do with agentivity and control. It can be noted that the semantic level of
analysis was not as revealing a level as it was in examining Jeffrey’s use of
the various self reference forms. Two reasons for this have been suggested:
First, the self reference forms often appeared in clauses without verbs, and
thus the semantic coding took place only for a limited subset of utterances.
Second, when Grice did employ self reference forms in clauses with verbs,
she rarely referred to events ranking high in agentivity, so several of the
forms were noted to link up with low agentivity. For these reasons the other
levels of analysis, in particular the examination of pragmatic function, were
more central.

MONTH I

During Month II of the study, Grice showed an increased use of two-word

utterances. Her mean MLU for the two video sessions this month was 2.23
(see Table 4.2). As was noted in Fig. 4.2, Grice continued to refer primarily
to herself as the main participant in utterances, and she rarely referred to
others. In addition, she continued to use multiple forms to refer to herself,
and she continued to use these forms in ways that deviated from adult input.
During Month II, Grice continued to display several of the characteristics
that distinguished her as ego-anchored in Month I.
GRICE 115

Figure 6.4 shows the distribution of Grice’s self reference forms during
Month II. We find that in the second month of the study, the distributional
pattern was roughly the same as for Month I. Grice continued to rely on three
forms of self reference: J, My, and Me. My was the form that was used most
frequently—its use accounted for 43% of all uses of self reference forms. Me
was used least frequently, accounting for only 22% of all self reference form
usage. During Month I, we noted that the three forms were employed in almost
equal proportions, except that Myoccurred most frequently. In Month II, Grice
used Mymuch more frequently than the other two forms, and she used J more
frequently than Me. In addition, we find that Grice used her own name even
less frequently during this session. Its usage accounts for only 3% of all self
reference forms. I now consider the various co-occurrence patterns associated
with the use of the core self reference forms, which accounts for 95% of all
uses of self reference forms during this session.

The Core Contrast: I, Me, and My

Surface Characteristics. We can begin our analysis of the core self

reference forms by analyzing various distributional patterns associated with
surface characteristics of their usage. Table 6.2 reveals findings concerning the
sentential contexts in which such forms appeared. Table 6.2a indicates
whether the various forms appeared in utterances with or without verbs. As
in Month I, a good number of Grice’s self reference forms appeared in
utterances without verbs. The same basic trend reported for Month I contin-
ued, namely, that J appeared in utterances with verbs, whereas nearly all of
the instances of My appeared in utterances without verbs. In contrast to Month
I, we find that Grice began to use Me in utterances with verbs more frequently
than she did in Month I.
Looking more specifically at the uses of self reference forms in utterances
with verbs, we find that the major developmental change from Month I is
the growing conventionalization of the system. Here we find My in object
position (with a noun) rather than occurring exclusively as a subject pronoun
in utterances with verbs, as in Month I. An even bigger trend is shown in
Table 6.2b with regard to the growing use of Me in object position. Thus,
131%

My 43%
aa Own Name 3%

Other 1%
FIG. 6.4. Distribution (%) of self
reference forms: Grice Month II (2
=t070)4 Me 22%
116 CHAPTER 6

TABLE 6.2
Distribution (%) of Self Reference Forms According to Surface Characteristics
eee
e e eee

(a) Verb / Verbless Context

FORM With Verb Without Verb Uncoded Subtotal*

| 100 (27)
My 30 70 (37)
Me 58 42 (19)
Other (1)
Own Name 100 (3)
Subtotal* (50) (37) (87)

(b) With Verb: Position

FORM Subject Object Uncoded Subtotal"

| 100 (27)
My 82 18 (11)
Me 18 73 9 (11)
Other 100 (1)
Own Name
Subtotal* (39) (10) (Oy (50)

(c) With Verb: Argument Structure

FORM Single Argument —_ Multiargument Uncoded Subtotal*

| oe 78 (27)
My 9 90 (11)
Me 45 55 (11)
Other 100 (1)
Own Name
Subtotal* (12) (38) (50)

(d) Without Verb: Context

FORM +Noun Alone Other Subtotal*

|
My 50 46 4 (26)
Me 50 50 (8)
Other
Own Name 67 33 (3)
Subtotal* (13) (18) (6) (37)

* (Raw numbers).

we find that the most significant change in Grice’s system was that she used
multiple self reference forms in subject position less often. With regard to
utterances without verbs, there are no noteworthy changes to be discussed.

Semantic Characteristics. In this session we find that there is a slight

increase in the overall use of the core self reference forms in utterances
with verbs. That is, although. in Month I we found that less than half of all
GRICE 117

core uses were in utterances with verbs (46%), in Month II 54% of all core
uses occurred with verbs. Furthermore, looking at all core self reference
forms to see whether they occurred in utterances coded as high or low in
agentivity, we see that the general finding for Month I holds in this session
as well. In Month I, 79% of all Grice’s core self reference forms coded for
semantic agentivity ranked low in agentivity; and in Month II, 71% of those
utterances coded also ranked low.
In addition, we find a similar pattern in the proportion of individual forms
coded in terms of semantic agentivity during Month II (see Fig. 6.5). That
is, in Month Il—as was the case in the first month of the study—Grice tended
to use all three core self reference forms in talk about low agentivity. This
differs from the pattern noted for Jeffrey, in that when Jeffrey talked about
high agentivity early on in the study he was more likely to link My with
high agentivity.
Although Grice tended to talk about events ranking low in agentivity, the
question remains whether she reserved a particular self reference form for
utterances that ranked high on the agentivity scale. In keeping with the first
month of the study, this was not the case. Looking at the distributional
patterns for all three core forms, we find that each form occurred only
occasionally in utterances ranking high in agentivity. In particular, we find
in Month II that My linked up with high agentivity in one example that
actually involved the use of Me as well. When J and Me were used in
utterances that ranked high in agentivity, the utterances differed from one
another in terms of a range of semantic characteristics, as is illustrated in
the following examples:

(27) I wanna put in.

(28) I ride that one. (question intonation)
(29) I'll came. (said with announcer intonation while walking from office
to classroom)

100%
GE Low Agency
90% C_] High
ighAgency
Uncoded

70%

60%

40%
Percentage
Use 30%
of
20%
10%
FIG. 6.5. Distribution (%) of self 0%
reference forms in terms of seman- Bis nae were
tic agency: Grice Month II. Fora
118 CHAPTER 6

(30) Put me in there.

(31) You put me in there Mama.
(32) Want me read my book. (as a command to read photo-book)

One central distinction between the use of /and that of Me in these examples
is that Me appeared as object of a verb, whereas J occurred regularly in
subject position. In keeping with this, J expressed actions the child desired
to carry out, questioned carrying out, or was in the process of carrying out,
whereas Me referred to actions that would be carried out by others on the
child.
Considering the contrastive use of the three core forms at the low end
of the agentivity scale certain distinctions can also be found. For instance,
Me continued to be used in utterances in which the child wanted to elicit
action from another (e.g., Show me, Help me Mama), but Grice also used
Me to refer to actions that she undertook without external help (e.g., Me
drinking)—-situations in which she initiated and was affected by the action.
In contrast, J linked up with the expression of states (e.g., J don’t know, I
don’t like that one, and I have it). Finally, the few instances of My in
utterances coded for semantic agentivity (mainly ranked low) appeared with
the verb have (e.g., My have it). The one instance that ranked low was of
the form verb + possessive pronoun + object (Play my toys). Comparing such
uses to those of Jand Me, we find that when My ranked low it often appeared
in contexts semantically similar to those of Z Thus the central finding is that
during Month II, although certain semantic distinctions can be found between
the various forms, there is no evidence for a form-meaning mapping with
regard to Grice’s use of various self reference forms.

Pragmatic Considerations. Considering the overall distributional pat-

terns for language function during Month II, we find that during this month
Grice began to use language much more to make assertions. This leads to
an almost equal division between control acts (44% of all uses of self ref-
erence forms) and assertions (41% of all uses). Control acts increased only
slightly from Month I to Month II: They accounted for 38% of all core uses
in Month I and 44% in Month II. Multifunctional utterances decreased during
this period from 22% (Month I) to 13% (Month ID. We can now consider
whether there were any changes in the ways the particular core forms linked
up with these language functions.
During Month I we noted that Grice linked the various self reference
forms with distinct functions. The basic change during Month II is that each
form became more multifunctional (see Fig. 6.6). The use of J was split
between assertions and control acts, that of My was split between control
acts and multifunctional acts, and that of Me was split between assertions
and control acts. The question then becomes: Are there differences between
GRICE 119

95% GEE Assertion

85%9 C_] Muitifunctional
Control Act
75% Uncoded
@®
3 65%
6
@ 55%
5 45%
5 35%
ao
25%
15%
Darn 78 5% 4
FIG. 6.6. Distribution (%) of self E
reference forms in terms of prag- aoe Heal ee ee
matic function: Grice Month II. Form

the kinds of assertions that involve J and Me, and are there differences
between the kinds of control acts involving the three core self reference
forms? I now examine such potential differences.
At the level of the pragmatic analysis, the uses of J and Me in assertions
look quite similar. Most often these forms were used to announce or describe
ongoing aspects of the situation. For instance, in one situation Grice de-
scribed her actions with a block by saying, “I rolling it.” Somewhat later in
the same session, while playing with pretend teacups, Grice stated “Me
drinking.” At the level of pragmatic function, both of these utterances look
quite similar in that the speaker was informing the listener of the ongoing
contextualization of the play. Both statements are assertions. In the next
section, though, we see ways these two forms can be distinguished.
Turning next to the contrastive use of the various self reference forms in
control acts, we find the following distinctions. First, when used in control
acts, J was most often found in utterances reflecting an attempt to influence
the hearer’s actions. In particular, J was found in requests or directives for
the partner to act in particular ways. For instance, several times Grice uttered
“I want books,” as a directive to her mother or the researcher to allow her
to look at the photo-book: Similarly, as the researcher cleaned up the blocks
and began to set up the tea party toys, Grice requested the teapot by saying
“I want teapot.” Such usage contrasts with the use of My. My was typically
used in control acts in disputes over mutually desired objects. In such situ-
ations, the child attempted to claim control of objects the other possessed.
For instance, trying to grab the same teapot from her peer, who was using
it, Grice uttered “My have it” or “My had it.” At times My was used to claim
available objects that no one was using but that the child had access to. For
example, playing with her mother, Grice picked up a large spoon sitting on
the table and said “My souper,” clutching the object. Notice how this contrasts
with the directive use of J in situations where the other acted to help the
child obtain goods. Finally, we can consider the use of Me in the context
120 CHAPTER 6

of control acts. Me, as I noted earlier, now appeared in object position as

well as subject position. These occurrences resemble the directives involving
I, except that the child was requesting that others perform actions that would
bring about some change of state or location for the child. Thus we find
examples like Put me in there and Help me in there in situations where the
child wanted to change location. In sum, then, we find that the forms showed
subtle distinctions at the level of pragmatic function. Although we no longer
find a simple form-function mapping of the sort outlined in Month I and
we see multiple forms linking up with the various functions, upon closer
examination distinctions can be found.

Core Field of Use and the I, Me, and My Contrast. We can briefly
summarize the major developmental changes in the core fields of use of
each of the various self reference forms as follows. First, with regard to the
form J we find that on the whole, characteristics of its use remained un-
changed. It was used with verbs in subject position, and it typically linked
up with verbs ranking low in agentivity. With regard to pragmatic function,
it continued to be used primarily in assertions about intentions of the self
when disputes were not involved, though during Month II J also appeared
in control acts in which the child requested others to act in light of her own
intentions.
My shows a bit more developmental change. First, we see it occurred
more frequently in utterances with a verb; and in addition, it appeared in
subject position of such utterances more frequently than in Month I. Nev-
ertheless, in Month II, as in Month I, semantic agency was not central to its
usage; instead it linked up with the pragmatic function of claiming control
of objects. The basic change was that although in Month I Grice would say,
“My teapot,” as a claim to obtain and maintain possession of the teapot in
a dispute with her peer, in Month II she was more likely to use My in subject
position, saying, “My have teapot” or “My want teapot.”
The form Me also underwent some changes in Month II. The most
significant change involved surface characteristics. Like My, Me also appeared
more frequently in utterances with a verb; but now when Grice used Me with
a verb, she was more likely to use it in object position than in subject position.
The semantic and pragmatic aspects of its usage remained the same as in
Month I. The form tended to be employed to describe actions the child carried
out that impacted back on the self or actions she wished others to perform that
would cause a change in her. Thus we find Me with both activity and
accomplishment verbs and in assertions as well as in control acts. What at first
glance appeared to be a multifunctional use was, on closer examination, a
cluster of related uses. Therefore we find in Month II that the core cluster
associated with Me, namely, affected agency, did not change, though structural
aspects of its use looked more like those of the target system.
GRICE 121

MONTH Ii

Forms of Self Reference

In Month III, Grice’s system of self reference underwent major changes. We

can begin our analysis with an examination of general characteristics of her
language development. First, we find that Grice’s MLU jumped to a mean
of 2.78 in Month III. Her references to self as main participant continued to
dominate her talk (see Fig. 4.2). These findings suggest that she was still
primarily ego-anchored. Aspects of her use of various self reference forms
nevertheless underwent massive reorganization in this month. In terms of a
purely quantitative analysis, we find in this month an explosion in the
frequency with which she referred to herself. Grice now employed three
times as many self reference forms as in the previous month. In terms of
the distribution of the various forms, we find a major change in the frequency
of occurrence of the various self reference forms (see Fig. 6.7). Although in
the first and second months of the study Grice relied primarily on three
forms, in this session we find a dramatic change in terms of both the forms
that were regularly employed and the relative distributions of these forms.
For instance, J was now used most of the time (65%) when referring to the
self; and in contrast to the case in previous months, My was significantly
less common (at 16%, compared to 41% for Month ID). In addition, Me was
no longer the third most common form; instead, Grice now employed forms
coded as Other 10% of the time. Thus in this session, Grice began to use a
variety of new forms to refer to herself (e.g., we, mines). We can also note
that although J was now the dominant self reference form—as was charac-
teristic for the nonego-anchored children—Grice continued to employ My
and Me in subject position in ways that deviated from adult usage. In the
sections that follow, I continue to focus the discussion on Grice’s use of
what was previously referred to as the core contrast in Months I and II,
because the Other category comprised a variety of forms that were yet to
be used with more frequency than J, My, or Me.

FIG. 6.7. Distribution (%) of self

reference forms: Grice Month II (2
= 280).
122 CHAPTER 6

Surface Characteristics. Table 6.3 reviews the distributional aspects

of Grice’s use of self reference forms in Month III in terms of surface char-
acteristics of the utterances. We find that one major developmental shift
involved the substantial increase in overall proportion of self reference forms
found in utterances with verbs (see Table 6.3a). Table 6.3b also reveals that
My continued to be used in more conventional ways-not only did it appear
more regularly in object position, but also when Grice used it in subject
position it only rarely appeared alone. It is of note that Grice continued to
TABLE 6.3
Distribution (%) of Self Reference Forms According to Surface Characteristics
(Grice Month III)**

(a) Verb / Verbless Context

FORM With Verb Without Verb Uncoded Subtotal*

97 1 3 (181)
My 36 50 14 (44)
Me 75 25 (16)
Own Name 56 44 (9)
Subtotal* (208) (31) (11) ~ (250)

(b) With Verb: Position

FORM Subject Object Uncoded Subtotal*

| 100 (175)
My 63 38 (16)
Me 33 67 (12)
Own Name 60 40 (5)
Subtotal* (192) (16) (208)

(c) With Verb: Argument Structure

FORM Single Argument _—-Multiargument Uncoded Subtotal”

| T 93 (175)
My 13 88 (16)
Me 100 (12)
Own Name 100 (5)
Subtotal* (14) (194) (208)

(d) Without Verb: Context

Sone eee ee

FORM +Noun Alone Other Subtotal*

| 100 (1)
My 68 5 ae (22)
Me 50 50 (4)
Own Name 50 50 (4)
Subtotal* (16) (6) (9) (31)
a
* (Raw numbers).
** This table excludes 28 instances of other.
GRICE 123

use Me regularly in subject position in ways that deviated from adult usage.
The only other major trend to be noted with regard to surface characteristics
concerns the argument structure of clauses containing various self reference
forms. In contrast to earlier months, in which the forms could be distin-
guished in part by their use in single-argument or multiargument structures,
we now find that nearly all of Grice’s self reference forms appeared in
multiargument clauses (see Table 6.3c).

Semantic Characteristics. As was shown in Table 6.3a, most of the

uses of the self reference forms occurred in utterances with verbs. This
implies that our analysis of semantic agentivity could take into account a
greater number of uses of self reference forms. An analysis of the utterances
containing self reference forms (regardless of form) reveals a substantial
decrease in the amount of such talk focusing on low agentivity. That is, 58%
of all uses of self reference forms occurred in utterances that ranked low in
agentivity features. Thus from the first month of the study to the third month,
we see a gradual increase in Grice’s use of self reference forms in utterances
that ranked high in agentivity (Compare 18% in Month I, 29% in Month II,
and 38% in Month IID. The question remains whether Grice linked the use
of particular forms with the degree of agentivity expressed. In weighing
such a question it is important to keep in mind that although a majority of
the uses of J and Me occurred in utterances with verbs, only 36% of all uses
of My were coded for semantic agentivity, because many of the utterances
in which My appeared contained no verb. This implies that our assessment
of how My linked up with semantic agentivity is based on only a small
proportion of the total use of My.
An analysis of the examples involving the particular forms points up some
important differences between Grice’s previous use of the core forms and
their current usage (see Fig. 6.8). Although we continue to find no simple
form-meaning mapping between the specific self reference form and the

100%

ooy HE Low Agency

(__} High Agency
ee GEE Uncoded
2 70%
=
5 60%
@®
B 50%
c
B 40%
@®
ao 30%
20%
10%
FIG. 6.8. Distribution (%) of self 0%
; terms of seman 5
reference forms in Oey I My
we Mei
tic agency: Grice Month III. Form
124 CHAPTER 6

degree of agency expressed, some trends can be noted. The basic finding
is that a large proportion of uses of any given form occurred in utterances
ranked at the low end of the agentivity scale; this was especially true for
My. This is in direct contrast to what I noted for Jeffrey. Furthermore, both
Iand Me, but not My, were used by Grice in utterances ranked at the high
end of the agentivity scale. In general, though, no single form appeared
primarily in utterances ranked at the high end of the scale.
With these comparisons in mind we can more carefully examine the three
core self reference forms in terms of semantic agency. First, we can consider
the use of My, keeping in mind that the analysis here is based on a small
proportion of all Grice’s uses of My. What changed in Grice’s use of My
during the third month was the frequency with which it was employed in
combination with a noun in utterances with a verb. Grice typically employed
My in utterances ranking low in agentivity (e.g., My mommy play, My giant
one fell down), though one also finds examples of My in conjunction with
a noun in utterances ranking high in agentivity (e.g., Want I pour it in my
cup). In such examples, the use of My does not seem to be linked up with
a particular degree of agentivity.
One also finds, though, examples of My + verb, as in previous sessions;
and the question can be raised whether Grice used such constructions any
differently. Again one can find instances ranking both high and low in terms
of semantic agentivity. For instance:

(33) My put it on me.

(34) My put this on.
(35) My have tups (cups).

Thus we find no reason to believe that the use of My, whether alone or
with a noun, can be linked to the degree of semantic agency expressed. In
comparing the use of My in utterances with verbs in this session to that in
the previous months, though, two changes can be noted. First, Grice more
regularly used My now in conjunction with a noun; and second, we now
find instances of My with action verbs. This use is reminiscent of Jeffrey’s
special usage of My, though we see later that important differences can be
found in their usages at the semantic level of analysis. This becomes clear
later when we turn to a discussion of the use of J
The coding of agentivity for Grice’s use of J during Month III looks very
similar to that in Month II. We find that J linked up primarily with utterances
ranking low in agentivity, though it could also appear in utterances ranking
high. What is different now is that one also finds My in a similar set of
examples. That is, although in Month II we found only J in conjunction with
high agentivity, examples (33)-(35) now reveal My in similar semantic con-
texts. I later discuss further the basis for this comparison; the main point
GRICE 125

here is that both forms can be found at both ends of the agentivity continuum,
suggesting that agentivity is not a major factor in the contrastive use of J
and My.
With regard to the form Me, in this session the major change involved
the increased incorporation of this form into utterances that ranked high in
agentivity. In particular, we find that Grice used Me more often in conjunction
with prepositions indicating that the self was either a benefactor, or a re-
ceiver, or indicating location of action:

(36) I’m putting this on there for me.

(37) Give them to me.
(38) My put it on me.

Grice continued, though, to use Me in subject position with activity verbs

in contexts in which she acted in ways that affected the self rather than
some external object:

(39) Me drinking tea.

What remains unclear are the following two uses of Me in subject position,
for which an affected agent reading is less feasible:

(40) Me open it.

(41) Me stir around.

In example (40), Grice was requesting that a container be opened for her
by the researcher, a request that the researcher rejected. Note that Grice
changed to the form J when she announced that she herself was able to
open the container by saying “I did it Nancy,” showing Nancy (the researcher)
the container. Thus it seems that Grice was extending the form Me to subject
position to imply the benefactive reading, though what remains unclear is
why, when she made a similar request to her mother in that video session,
she said, “Want you to open that for me.” Another interesting contrast is
between example (41) and Grice’s utterance J stir that for you. Both examples
appeared in apparently similar agentivity contexts, though, as we later see,
what distinguished them were aspects of the pragmatic context in which
they were embedded.
Taken together, the findings for Grice’s use of J, My, and Me indicate that
for Grice, semantic factors of agency do not explain how these forms were
used, At best, they help provide a partial account of the use of Me, to the
extent that Grice used this form to express a view of the self that received
or was affected by action.
 126 CHAPTER 6

Pragmatic Characteristics. Turning to the analysis of language func-

tion, we find a dramatic change in distributional properties of Grice’s lan-
guage in terms of its pragmatic function. Although in Month II, Grice divided
her use of language between assertives (41%) and control acts (48%), in
Month III Grice relied far more on assertives (62%), than on control acts
(23%). Over the course of the study, Grice began to use language less and
less as a means to control her world and simultaneously began to use
language to comment and question about ongoing happenings.
The question remains whether Grice continued to align specific forms
with particular functions during Month III of the study. It can be remembered
that in Month II Grice distributed her use of the core forms differently across
the various language functions. J was divided between assertions (59%) and
control acts (41%), whereas My was divided between control acts (49%) and
multifunctional acts (30%). Me showed a distributional pattern similar to that
of J, that is, a slight preference for assertions (45%) over control acts (36%).
What this points up is that in Month II, no single form was noted to link
up with control acts, although My, when it was used, typically occurred in
control act contexts. In Month III the distributional regularities look different
(see Fig. 6.9). J was rarely found in control acts anymore (21%) and was
instead employed in assertives (67%). My was now rarely used in control
acts (18%) and instead typically occurred in assertions (57%); and the use
of Me seemed to follow a similar distributional pattern as in Month II, showing
a preference for assertives (53%) over control acts (41%). These findings
indicate that no single form linked up with a single function, though clear
patterns can be found. In the final section of the discussion of Month III, I
clarify the grounds on which these three forms can be distinguished.

Core Fields of Use and the I, My, and Me Contrast. Although I have
noted that the use of the core forms in Month III does not line up neatly with
semantic distinctions or pragmatic distinctions, an analysis that cuts across

E> Control Act

Uncoded

Use
of
Percentage

FIG. 6.9. Distribution (%) of self

I My Me OwnName :
n=181 n=44 n=17_—sn=10 reference forms in terms of prag-
Form : matic function: Grice Month II.
GRICE 127

boundaries reveals some subtle ways in which these forms can be contrasted.
In order to illustrate the contrastive use of the core forms, I now consider
examples that provide minimal pairs with respect to usage. First, consider the
distinction between J and My. Recall that at the semantic level of analysis,
both forms were noted to occur with both action and state verbs. With state
verbs, both forms occurred in subject position followed by the verb, although
the same range of verbs was not used in both cases. My occurred only in
conjunction with the verb have, whereas J was used with a range of verbs
such as want and like but never with have. Both forms were at times used
with the same action verbs:

(42) Grice (G) and Megan (M) are playing with blocks. G is building a
block tower, trying to place two blocks together, when her peer
(M) approaches with a block in her hand.
a. G: My put this on!
b. G: I putting this one there.
c. G: I’m putting this on there (.) for me.

In this example, Grice was concentrating on getting the two blocks together.
It was when her peer approached that she looked at her peer and exclaimed
(42a). Her peer immediately moved away. At this point, Grice again attempted
to connect the blocks, uttering (42b). The distinction between J and My has
to do with control over doing. When Grice was simply informing her listeners
of her ongoing activities, she used J. But when she felt challenged or when
another intruded on her ability to carry out actions, she switched to My, It is
important to note that Grice typically used My in conjunction with a noun, and
in such contexts neither semantic agentivity nor pragmatic control played a
role in its use. At this point there is no evidence that the child drew a connection
between these two different uses of My.
Turning next to the connection between J and Me, we also find a dis-
tinction that cuts across semantic and pragmatic boundaries. Consider the
contrast noted earlier, in which both forms occurred in subject position:

(43) Grice (G) and mom (GM) are playing tea party. G has just prepared
a cup of “juice” for GM.
a. GM: Um very good.
1pm: I stir that for you.
c. GM: Thank you.

In this example, Grice was stirring pretend juice when she uttered (43b).
This example contrasts with the following:

(44) A few minutes later Grice (G) and mom (GM) are still playing tea
party. GM has offered G “sugar” and “milk” to place in her tea. G
then approaches GM and grabs cup GM is using, saying:
128 CHAPTER 6

a. G: Me stir around.
lbaG: # Me stir it around.
C GM: Yeah stir it up.

A comparison of these two examples reveals that J was used in the less
dynamic context when Grice was reporting her ongoing actions to her
mother, whereas Me was used when Grice was attempting to gain control
of the action in particular ways. This, though, cannot account for all examples
with similar contexts involving J and Me:

(45) Grice (G) and mom (GM) still engaged in tea party. Grice is showing
her mom how she can drink.
a. GM: Good!
b. G: — I drink all up.
c. GM: You drink it all up.

In example (45), Grice used J when she reported completing an ongoing

action. This usage is similar to that in example (42) to the extent that control
was not at issue. As the following example reveals, however, obtaining
control is not the only factor related to this contrast.

(46) Grice (G) is having a tea party with Megan (M) while the researcher
(R) sits nearby. G approaches R, bringing her a cup of tea. R requests
that she return to the play table. G pulls the cup back and pretends
to drink it and then says:
a. G: Me drinking tea.

One can question why Grice did not say something like (45b). Again the
difference is one of perspective taken on the scene. The claim here is that in
using Me Grice focused on herself as affected by the drinking rather than on
her desire to complete an activity. In instances like this, an element of contrast
is involved as well. Although control was not at issue, Grice was stressing that
she, rather than the researcher, benefited from the tea. The notion of benefit
rather than simple control comes into play in the following set of examples:

(47) Grice (G) is playing with manipulative toys with Megan (M). She
struggles unsuccessfully to open a childproof container that contains
an interesting object, at which point she approaches the desk where
the researcher (R) is sitting.
a. G: Me open it.
b. G: I Gwan)na open it.

As she was uttering both (47b) and (47c), she was attempting to give the
container to the researcher, who rejected the offer and told the child to
return to the table and try herself. What is particularly interesting about
GRICE 129

examples (47a) and (47b) is that neither of them referred to an action

sequence in which the child was opening an object. In both cases the child
was attempting to get the researcher to open the container. What did the
child indicate in these two examples? The point put forth here is that the
child was providing two distinct ways to classify the event. In example (47a),
the child was signaling that she wanted the other to open it “for me” (ie.,
Open it for me). In example (47b), the child seemed to be indicating some-
thing like 7 want it to be opened. That is, the various uses of Me, in either
subject or object position, occurred in an activity context in which the child
was referring to a scene in which she pictured herself as benefiting from
the action sequence.
At this point the question can be raised: What is the difference between
the Grice’s use of Me and her use of My? I noted earlier that My was always
used in dynamic contexts in which control was at issue, and that at times
the use of Me linked up with dynamics of control and agency. One major
distinction at this point between the adultlike and the nonadultlike uses of
these forms concerns an assessment of presupposition of control. In exam-
ples involving My, the child presupposed control—that is, the child was
involved in carrying out an activity at a point where another was viewed
as intruding, or the child was simply attempting to bring about a change
without another becoming involved. In contrast, when the presupposition
of control lay with the other, then the child was more likely to use Me,
emphasizing the Me not you aspect of its use. What is extremely interesting
here is that the child’s selection of one form or the other is claimed to
depend on fairly subtle contrasts in the ongoing sequence of events. The
child was not simply drawing distinctions based on an assessment of the
degree of semantic agentivity expressed or on the pragmatic function of the
utterance.

MONTH IV

My discussion of Grice’s use of self reference forms during Month IV is

limited in that there were very few developmental changes that took place.
In Month IV Grice’s mean MLU passed the 3.0 mark noted for nonego-an-
chored children; her mean MLU during Month IV was 3.51. During Month
IV, she also continued to show a gradual decrease in the proportion of self
reference forms occurring in subject position (74% in Month IV vs. 82% in
Month IID. The fact, though, that the majority of her references to main
participants were to herself distinguishes her from the nonego-anchored
group.
With regard to the distribution of various self reference forms, we find a
pattern similar to that reported in Month III: Grice tended to rely on one
130 CHAPTER 6

form, J, and occasionally used My and Me (see Fig. 6.10). At this point Grice
no longer used My and Me in subject position in ways that deviated from
adult English. My now occurred in conjunction with nouns and, occasionally,
other pronominal forms (e.g., My this). Me occurred in object position, and
Grice no longer marked the notion of self as affected agent. Although Grice
no longer used the self reference forms in subject position in ways that
deviated from adult input, some examples reveal that she was still working
on the self reference system. In several of her utterances she employed
multiple self reference forms in ways that deviated from adult English. For
instance, in one example Grice wanted to stir pretend tea in the toy teapot
and turned to her peer, saying, “Don’t do me stirring for me!” In another
example, her peer had given Grice a plate, which Grice lifted up, saying,
“Don’t do me my this!” At times, then, Grice still seemed to mark distinct
notions about the self with different self reference forms.
Because J was the only form used in subject position, a consideration of
the way the particular forms linked up with various semantic and pragmatic
functions is not helpful. Nevertheless, it is important to point out certain
developments with regard to the overall distribution of self reference forms
with respect to semantic agentivity ranking and pragmatic control. The
changes during Month IV follow the developmental path I have reported
from Month I to Month III. More specifically, over the course of the study
a gradually increasing proportion of Grice’s references to self were coded
as ranking high in agentivity. In Month I only 18% of her self reference
forms ranked high, whereas in Month IV 48% of such forms ranked high.
This is particularly interesting in that the most frequently used self reference
form was J, which in Month I was primarily (78%) linked with low agentivity.
Thus we see that not only did Grice no longer link this form with low
agentivity, but in general her talk about the self was fairly equally distributed
between high and low agentivity.
With regard to pragmatic function, a similar developmental transition can
be noted. In the first month of the study, Grice divided her use of language
between several language functions, though most frequently her use’ of self
reference forms linked up with control acts (38%). By the fourth month of
the study, Grice showed a clear preference for linking self reference forms

167%

Own Name 6%

Other 5% FIG. 6.10. Distribution (%) of self

reference forms: Grice Month IV (2
My 11% Me 12% = 221).
GRICE 131

with assertives (68%), whereas control acts accounted for only a small
amount (20%) of the utterances containing self reference forms. Thus, al-
though over the course of the study we see a gradual decrease in the linking
of specific self reference forms with particular semantic and pragmatic func-
tions, we nevertheless find that there are important developmental changes
in the semantic meanings expressed by self reference forms and the prag-
matic functions associated with their usage.
In sum, Grice’s construction of a system of self reference went through
tremendous changes over the course of the study. We saw a rather slow
transition from the frequent use of various self reference forms in utterances
without verbs to the gradual use of such forms in the context of a verb. In
addition, Grice only gradually came to use these forms as in the target
language. In particular, in Months I-III we noticed how various forms were
found in subject position in ways that deviated from adult English. Thus
Grice’s construction of an adultlike system was quite gradual. Along the
way, the number of units per utterance increased, allowing her greater
flexibility in the kinds of resources that could be drawn on. For instance,
by Month III and Month IV we see the addition of both modal forms and
prepositions. The use of modal forms seemed to permit the marking of
various degrees of agentivity and control, because such forms allow the
speaker to qualify the intentional stance toward the action described. Also,
the onset of prepositional forms provided Grice with more conventionalized
resources for expressing the notion of benefactor that previously was marked
with Me in subject position. As was the case for Jeffrey, there was a com-
plicated web of factors interacting in the developmental reorganization of
the self reference system over the course of the study.
 CHAPTER SEVEN

Megan

MONTH I

Megan was the youngest child in the study. She was 20 months at the onset
of the study and was just beginning to combine words (mean MLU for Month
I = 2.07). As was shown in Fig. 4.1, the majority of her references to main
participants were about herself at the onset of the study; 86% of references
to main participants were coded as self reference. Furthermore, Megan, like
the other ego-anchored children, began the study using multiple forms to
refer to herself, often in ways that deviated from adult input. These factors
led her to be characterized as ego-anchored at the onset of the study.
Megan’s distribution of the various forms that she used to refer to herself
is outlined in Fig. 7.1. Megan, like Grice, divided her use primarily between
three forms. She was the only one of the ego-anchored children to draw
regularly on the use of her own name. We see that J and Own Name were
used most frequently, 33% and 32% of the time, respectively. My accounted
for 23% of all usage and Me occurred in 10% of all her references to herself.

The Core Contrast: I, Own Name, and My

As in the case of the other ego-anchored children, I begin the discussion

of Megan’s use of self reference forms with some examples.
(1) I want that one.
(2) I wanna wear that.
(3) I cried.
(4) I knocked.

132
MEGAN 133

reference forms: Megan Month I (n

= 91).

(5) Meggy.
(6) Meggy count.
(7) Meggy swinging.
(8) My open that.
(9) My put back.
(10) My want that.

Like the other ego-anchored children, Megan used a variety of self ref-
erence forms in subject position. She was the only child, though, who in-
cluded the use of her own name in the core contrast. As in the previous
chapters, I now consider whether there was a principled basis for the se-
lection of various self reference forms or whether Megan simply employed
the various forms randomly.

Surface Characteristics. Table 7.1 describes surface characteristics of

the various self reference forms. I begin the analysis of the forms by noting
that only 53% of Megan’s self reference forms were in utterances containing
a verb, whereas 42% were in utterances without a verb (see Table 7.1a).
The forms differed in frequency of occurrence in utterances containing verbs.
I was the form that most frequently occurred in such a context; 87% of all
uses of J were in utterances with a verb. In contrast, Own Name occurred
only 21% of the time with a verb. My was typically used in utterances with
a verb (62%), though 38% of all uses of My occurred in a verbless context.
Turning to the use of self reference forms in utterances with verbs, we
find that all of the instances of Jand My were in subject position (see Table
7.1b-c). I was typically found in multiargument utterances (77%), whereas
My was found exclusively in multiargument utterances when appearing with
a verb. Own Name typically occurred in contexts without a verb, though
when used with a verb it appeared in both subject and object position and
was usually found in single-argument utterances.
Table 7.1d outlines aspects of the usage of forms appearing in utterances
without verbs. Of interest is the observation that both My and Me appeared
in constructions of the shape form + noun.
 TABLE 7.1
Distribution of Individual Forms (%) According to Surface Characteristics
(Megan Month 1)
a

(a) Verb / Verbless Context

FORM With Verb Without Verb Uncoded Subtotal*

| 87 10 3 (30)
My 62 38 (21)
Me 11 77 11 (9)
Other 100 (2)
Own Name 21 69 10 (29)
Subtotal* (48) (38) (5) (91)

(b) With Verb: Position

FORM Subject Object Uncoded Subtotal*

| 100 (26)
My 100 (13)
Me 100 (1)
Other 100 (2)
Own Name 50 33 17 (6)
Subtotal* (44) (3) (1) (48)

(c) With Verb: Argument Structure

FORM Single Argument Multiargument Uncoded Subtotal*

| 23 TE (26)
My 100 (13)
Me
Other 50 50 (2)
Own Name 67 33 (3)
Subtotal* (9) (35) (44)

(d) Without Verb: Context

FORM +Noun Alone Other Subtotal*

| 67 33 (3)
My 88 13 (8)
Me 43 29 29 (7)
Other
Own Name 50 50 (20)
Subtotal* (12) (12) (14) (38)

* (Raw numbers).

134
MEGAN 135

Semantic Characteristics. From the start it should be noted that only

53% of all uses of self reference forms involved utterances containing verbs;
thus, onlyalittle more than half of Megan’s self reference forms were coded
in terms of the semantic characteristics of agentivity. This was also true for
Grice. With this in mind we can consider the semantic characteristics of the
core self reference forms.
A combined analysis of all uses of self reference forms reveals that Megan’s
forms tended to occur in utterances that ranked low in agentivity. Only 21%
of the utterances containing these forms ranked high. Thus all three children
spent most of their time talking about self in the context of low agency.
The question remains whether particular forms were employed con-
trastively to express particular degrees of agentivity. An analysis of the core
forms in terms of their agentivity rankings can be found in Fig. 7.2. In
keeping with the general trend, the use of J tended to link up with the
expression of low agentivity; 92% of all uses of J were coded as ranking
low. A similar pattern was found for Megan’s use of Own Name. Of the
uses of this form coded in terms of semantic agentivity, 83% ranked low.
My was the form that was most frequently associated with high agentivity.
39% of the coded instances of My ranked high. In fact, of all the instances
of self reference forms coded as ranking high in semantic agentivity, 50%
were instances of My. The other 50% of the forms were scattered instances
of the individual forms. Examples (1)-(4) illustrate this patterning of the
data. Even when Megan employed an action verb like knock in example
(4), it was an instance of an activity verb, and the utterance referred to an
ongoing activity conducted by an actor, rather than an action conducted by
an agent. Most typically, though, the examples with / referred to inner states
of the child rather than activities.
The use of Own Name also has been noted to link up primarily with a
low degree of agentivity. Examples like (6) and (7) are typical of Megan’s

100%
GS Low Agency
90% (1) High Agency
80%

70%

40%
Percentage
Use 30%
of

20%

10%
FIG. 7.2. Distribution (%) of self 0%
reference forms in terms of seman- | My Own Name
=26 =13 n=6
tic agency: Megan Month I. = Fons
136 CHAPTER 7

use of Own Name. In contrast to J, Own Name never occurred in subject

position with a mental state verb. That is, we never find instances such as
Megan want some; instead, the use of Own Name typically occurred with
activity verbs that were neither very kinetic nor telic.
My was the self reference form that rank both high and low in terms of
semantic agentivity. Examples (8) and (9) are instances referring to events
involving prototypical agents. We also find instances like example (10),
which ranked low in agentivity. Thus My was associated with both ends of
the agentivity scale.
The following chart summarizes Megan’s distribution of the core self
reference forms along the agentivity continuum.

(11) AGENTIVITY CONTINUUM

High Mid Low
My Own Name I& My

In order to uncover the organising principle underlying the use of both Jand
My in utterances ranking low in agentivity, we need to consider other levels
of analysis.

Pragmatic Characteristics. An analysis of the pragmatic function of

Megan’s utterances containing self reference forms reveals that Megan used
such forms primarily in utterances that served an assertive function. More
than half of the uses of self reference forms (54%) were coded as assertives.
In contrast, only 28% were coded as control acts, and 13% were multifunc-
tional. The function of Megan’s self reference forms as a whole looks most
similar to that of Jeffrey’s at the onset of the study.
When one considers the core self reference forms, we find some indication
that the individual forms linked up with pragmatic functions (see Fig. 7.3).
For instance, the majority of uses of J were coded as assertives (60%),
100%
90% GE Assertion
(__) Muttifunctional
80% Control Act
Uncoded
70%

of
Percentage
Use

MMMM
FIG. 7.3. Distribution (%) of self
reference forms in terms of prag-
Form matic function: Megan Month I.
MEGAN 137

whereas 34% of the uses were divided equally between control acts and
multifunctional utterances.
The pattern is somewhat similar for the use of Own Name, to the extent
that 83% of such uses were coded as assertives, whereas only 10% were
coded as control acts. Although it appears as though J and Own Name both
occurred in a similar pragmatic context, closer examination reveals that the
kinds of assertives referred to with each form did differ. J tended to occur
in conjunction with utterances that asserted plans and intentions Megan had
for the immediate future when control was not at issue. In addition, these
utterances referred to previous or timeless happenings. In contrast, the use
of Own Name linked up with assertives that were coded as serving a naming
or identifying function. My occurred most frequently in utterances that func-
tioned as control acts (52%) or multifunctional utterances (33%). Very few
instances of My linked up with assertives (14%).
The pragmatic contrasts between utterances containing the different self
reference forms can be best illustrated through an examination of some
examples in context. I first consider the use of My in control acts and then
turn to an examination of the use of the other two core self reference forms
in assertives.

(12) Megan (M) and mom (MM) are playing with manipulative toys.
a. M: — I want that one. (as lifting childproof container)
b. MM: Oh you want that one, okay.
c. M: _ (Tries to open container) My open that.
d. MM: What?
e. M: My open that, Mommy. (handing container to MM)
f. MM: Wanna open that?
g.M: Yeah.

In this example, we find the use of My linked up with a directive. Megan,

after trying to open the childproof container without success, tried to solicit
her mother’s assistance. In line (12e) we find a particularly clear example
of a directive. Megan not only included a vocative to elicit her mother’s
attention but also gave the container to her mother. In line (12g) we find
that Megan’s attempts were successful. In this example, Megan used My in
a control act where she attempted to get her partner to bring about change.
The following example reveals that Megan also used My in control acts in
which her own actions were dominant.

(13) Megan (M) and mom (MM) are playing with blocks and a toy
helicopter.
a. MM: Who’s in there? (ooking in window of helicopter)
b. M: — Cooks in window) Ba.
138 CHAPTER 7

2 Looks like a fireperson. A fireman.

Uh. (putting hand out to take helicopter) My a see the
d.M:

:
f. M:
fireman. (taking hand back)
You wanna what?
(extends hand out and then takes it back) My see the
fireman. (looking at helicopter)
MM: You wanna see the fireman?
= Yeah.
"poe
i. MM: (points to fireman and explains that he is painted on and
cannot be removed)

In particular, Megan’s hand movements in lines (13d) and (13f) suggest that
the child was seeking permission to see (have?) the fireman. It was thus her
own action that was dominant in this control act.
The most frequent occurrence of My appeared in control acts with joint
action focus, specifically in disputes over ownership and enactment of actions.

(14) Megan (M) and Grice (G) are playing tea party with manipulative
toys.
a. G: Pour it out. (pouring with teapot)
Ga saups. (=got cups)
Ga saups. (=got cups)
(takes cupcake papers M has been using)
b. M: My cups. (trying to grab back)
c. G: My! (struggling with M)
Let her share. (M is taking control of papers.)

Such occurrences took place only in the peer sessions. In these cases Megan
typically coupled the use of My with the name of the object she wished to
control. Often struggles between the peers terminated only when the re-
searcher intervened.
The rather dynamic quality of My can be contrasted with the use of J in
assertives. In example (15), Megan used / in conjunction with a statement
of her ongoing plans.

(15) Megan (M) plays with blocks, while Grice (G) looks through the
photo-book.
a. M: I ’na pick that up. (as she is leaning under the table to pick
up blocks that have fallen)
(sits on floor and collects blocks, making no contact with
her peer)

In such instances, the child had already begun the action named. The goal-
directed nature of the utterance seems intended not only to inform others
MEGAN 139

of Megan’s plans but also to serve a self-regulative function. Such uses can
be contrasted with similar ones in which Megan’s utterance functioned more
as a directive to inform her partner not to engage in the activity. In such
instances we find Megan keeping eye contact with her partner. As predicted,
such uses occurred in conjunction with My rather than J
The use of J has also been noted to occur in the reporting of previous
states and actions:

(16) Megan (M) and mom (MM) are looking at pictures of children at a
pumpkin patch in photo album.
a. MM: Did we trick or treat?
bAM: ousYeah:
c. MM: To the door?
d. M: Yeah.
e. MM: (knocks on door)
What did ya say?
f. M: I knock.
g. M: You knocked. And then what did you say?
h. (After a brief pause MM offers a prompt to M, who finally
answers, “Trick or treat.”)

These reports tended to take place in the context of looking through the
photo-book. They were not references to depicted actions, and often what
the child was reporting was a part of speech routines exercised between
mother and child. The child’s focus in line (16f) was less on her actual role
in carrying out knocking behavior than on things one does when going out
to trick or treat.
Most frequently the use of J appeared in statements of ongoing desires.

(17) Megan (M) and mom (MM) are cleaning up blocks. M has dropped
several blocks on floor and MM has gotten angry.
a. MM: Can you:get that one that fell underneath the table?
b. M: — (M is under the table.)
MM: Can you find it?
M: # Yeah. (standing up)
MM: There we go, okay.
2.9M:
mie I want that block. (bending down again)

The use of J in line (17f) was linked to a context in which the child was
already en route to the desired object and there was no indication that
control was at issue. At this point we are in a better position to account for
the distinction between the use of J and that of My in subject position in
utterances with state verbs ranking low in agentivity. The following example
140 CHAPTER 7

involving the use of My received the same semantic agentivity rating as did
(17f), which involved the use of J

(18) Megan (M) and mom (MM) are playing with manipulative toys.
M: Ah ga Mommy. (reaching for container)
MM: Ah that’s a container.
M: My that. (touching another container near MM)
What’s in there? (MM holds it up and M takes it.)
M: My want that. (reaching for other container)
moaoge
MM: (switches M’s focus back to clear container)

As was the case with the other ego-anchored children, the use of My in
utterances ranking low in agentivity and expressing states linked up with
Megan’s attempts to get assistance. Thus the uses of J and My that ranked
low in terms of semantic agentivity can be distinguished in terms of pragmatic
function.
Finally, we can compare the assertives using Own Name in acts of naming
and identifying with the assertives noted to link up with J and the more
dynamic uses of My. Instances of the use of her own name often occurred
in the context of reading the photo-book.

(19) Megan (M) and mom (MM) are looking through the photo-book.
a. MM: And who’s that with the red coat on?
b. M: — Meggy.
c. MM: Is that you?
d. M: Yeah.
e. MM: Say, “That’s me.”
f. M: That me.
g. MM: Right!
h. M: = That a Meggy.

What is particularly interesting about example (19) is that although Megan

correctly labeled the picture of herself to the extent that she was depicted
in the picture, her mother was not content with Megan’s use of her own
name and requested that she use the pronominal form Me. Although the
child repeated the utterance that her mother requested, note that Megan
switched back to her own name one turn later.
Megan’s use of Own Name was not limited to contexts in which she
labeled photos of herself. The following example shows that she used this
form when pointing out objects that belonged to her.

(20) Megan (M) and mom (MM) are playing with manipulative toys.
a. M: Rocking chair. (pointing to chair in the corner of the room)
 MEGAN 141

b. MM: A rocking chair yeah. A rocking chair. Whose car seat is

that? (pointing to a car sear in corner)
c. M: — Meggy’s. (laughs)
d. MM: (laughs) Right. It’s yours.

In such uses control was never at issue. The utterance served a referential
function. This kind of usage contrasts with the use of My in example (14b),
where Megan attempted to grab back cups from her peer, stating, “My cups.”
The use of My linked up with issues of control where possession was up
for negotiation (i.e., centered around the toys the researcher brought),
whereas the use of Own Name occurred in referential contexts where ref-
erence to possessions was ongoing (see Deutsch & Budwig, 1983).
This analysis of the three core self reference forms reveals clear distinctions
between the kinds of pragmatic contexts in which they occurred. At this
point, I turn to a combined analysis of semantic and pragmatic aspects of
use of these three forms.

Core Field of Use and the I, Own Name, and My Contrast. A

combined analysis of semantic and pragmatic characteristics would not be
particularly revealing, for two reasons that have been noted earlier. First,
just over half (53%) of all references to self were coded for semantic char-
acteristics because only these were in utterances with verbs. A second reason
such an analysis would not be particularly helpful is that Megan so rarely
referred to events ranking high in agentivity. For this reason, I simply pull
together the findings from the separate levels of analysis, considering each
of the three core self reference forms.
For Megan, the use of J linked up with the expression of actions ranking
low in agentivity or with the expression of ongoing states. Such utterances
functioned pragmatically as assertives that most frequently informed her
communicative partner of her intentions or plans to act in particular ways.
At times, J also functioned to motivate or give reasons for actions Megan
undertook. In cases involving J Megan typically was acting on her own.
Own Name was not typically used in utterances that received an agentivity
ranking. What was central to the use of this form was the child’s labeling
of pictures or objects, either to identify them or distinguish them from others.
Neither control nor agentivity were related to these uses.
For Megan, the use of My (like that of D often linked up with the ex-
pression of ongoing states. These uses can be distinguished from seemingly
similar examples with J because the uses of My occurred in utterances that
functioned pragmatically as control acts. When reference was made to highly
agentive acts, the My form was employed. Thus Megan also linked My with
both high agentivity and control acts. I have noted that My linked up with
two kinds of control acts. When interacting with her caregiver, Megan often
142 CHAPTER 7

employed My when she attempted to direct her mother to play an instru-

mental role in helping her reach particular goals. When Megan used My
with her peer, it tended to link up with disputes over control. What holds
these distinct uses of My together is their role in bringing about change. In
all cases Megan was attempting to use language to bring about a change in
the world.

Other Self Reference Forms

As was revealed in Fig. 7.1, Megan’s use of the core self reference forms
accounted for 88% of all usage of self reference forms. The remaining 12%
included Me (10%) and We (2%). Because the instances of We were few
and varied they are not discussed further. In this section I focus exclusively
on the use of Me.

Surface Characteristics. Me, like Own Name, occurred primarily in

utterances without verbs. There was only one example of Me used in an
utterance with a verb: Let me make a tower. The uses of Me in a verbless
context were divided between use alone (Me/), with a noun (Me airplane),
or in predicatives (see Table 7.1 for review). Because there was only one
instance of Me coded at the semantic level, I turn to pragmatic factors guiding
the use of this form.

Pragmatic Characteristics. At the level of pragmatic function, we

find that me occurred primarily in control acts (see Fig. 7.3 above). In one
of the two instances of Me used in an assertive, Megan’s mother had actually
prompted her to say, “That’s me” (see example (19f)); and thus the child’s
use was not spontaneous. In fact this use was the only instance of Me that
occurred in the context of the caregiver-child interactions; all the others
occurred when she was playing with her peer. I also should note that 33%
of the uses of Me centered around control of one particular object, namely,
the helicopter in the block task. Such uses tended to occur as the child was
en route to or just beginning to use an object. In most cases, her peer was
acting on the object as well. In all instances involving Me, the child seemed
to have a double focus: On the one hand, the child referred to her desire
to be in control of an action or object, whereas on the other, focus was
placed on the child as the (conscious) goal of a transaction (cf. Givon,
1984a). In the cases of both My and Me, control was at issue, though the
perspective was slightly distinct. In the instances involving My the child
viewed herself as an agent who had the right to initiate change such that
she gained or maintained control, whereas with the utterances involving Me
the child acted more as a conscious goal of or one benefiting from a trans-
MEGAN 143

action. The following example, when compared to the earlier examples

involving My, illustrates this distinction:

(21) Researcher (R) is setting up the block task for Megan (M) and Grice
(G). Note that (R) has left toy helicopter on the desk though usually
it is on the tray with the blocks.
a. R: You guys gonna build?
b. M: Yeah.
c. R: Ill put some blocks here.
d. G: (starts taking blocks off tray)
Uh. (reaching over blocks)
Helicopter. (removing hand from tray)
e. M: Me airplane. Cooking at tray)
Me airplane.
f. G: ’Copter me ’copter.
g. (Children start to build, R returns with helicopter.)

Note in this example that the children were not actually trying to claim the
helicopter in the sense that it was not present. In similar instances where
the helicopter was present, they used My. What this implies is that they used
Me more in the sense of ‘to me the helicopter.’

MONTH I

During Month II, Megan had a mean MLU of 2.30. Even though she showed
an increase in word combinations, in her references to main participants
she continued to focus almost exclusively on herself; 88% of her references
were coded as about self. With regard to her use of self reference forms,
we find major changes in her system. Although in Month I she used multiple
self reference forms to refer to herself, in Month II this pattern changed. As
we see later, Megan now relied primarily on I Although this was noted as
one of the characteristics of the nonego-anchored children, it is significant
that aspects of her use of self reference forms deviated from the target
system.
Figure 7.4 illustrates the distributional pattern for the various self reference
forms during Month II. Although in Month I Megan had afairly even distri-
bution between Own Name and J, in Month II, J accounted for 59% of all
self reference usage. At this point Own Name was used only 24% of the
time. Even more interesting is the dramatic decrease in use of the form My
from 23% in Month I to only 9% in Month II. Neither Me nor any other
forms were used with any frequency.
144 CHAPTER 7

FIG. 7.4. Distribution (%) of self

reference forms: Megan Month II (2
Me6% Other 2% = 85).

Because there were major distributional changes in the self reference forms
at this point, we can question whether these forms also changed in terms
of the functions associated with their use. I turn now to such a consideration.

The Core Contrast: I, Own Name, and My

Surface Characteristics. One of the major changes with regard to

surface characteristics was that Megan now used the self reference forms
more regularly with verbs (see Table 7.2a). Although in Month I only 53%
of all self reference forms were in utterances with verbs, by Month II 67%
of all forms occurred with verbs. Nevertheless, Megan did not use all of the
forms with verbs. She had a fairly neat system: J was used in utterances
with verbs, and Own Name was used in utterances without verbs. The other
forms were rarely used, though when they were, Me was used mostly with
verbs in subject position in multiargument sentences, whereas My occurred
in utterances with and without a verb. What is interesting is that although
in Month I Megan used Me in object position, this form appeared exclusively
in subject position when used with a verb. My was used in subject position
alone and in object position in a possessive construction, suggesting the
possibility that these represented two distinct constructions for Megan at this
point in time (see Table 7.2b).

Semantic Characteristics. Megan continued her pattern of talking

about self as low in agency during Month II. Seventy-three percent of all
her self reference forms in utterances with a verb ranked low in agency. As
the following examples reveal, the use of J tended to occur in utterances
ranking low in agentivity. These uses involving low agentivity rankings
typically referred to desired states:

(22) I wan(@) another hot dog.

(23) I wanna see it.
(24) I wanna open it.
 MEGAN
145
TABLE 7.2
Distribution (%) of Self Reference Forms According to Surface Characteristics
(Megan Month I1)**
ee

(a) Verb / Verbless Context

ee ee ee

FORM With Verb Without Verb Uncoded Subtotal*

| 94 6 (51)
My 50 50 (8)
Me 80 20 (5)
Own Name 5 86 10 (21)
Subtotal* (67) (31) (2) (85)
$$

(b) With Verb: Position

——————— eee

FORM Subject Object Uncoded Subtotal"

i 100 (48)
My 75 25 (4)
Me 100 (4)
Own Name 100 (1)
Subtotal" (98) (2) (57)

(c) With Verb: Argument Structure

FORM Single Argument _——Multiargument Uncoded Subtotal*

| 15 85 (48)
My 25 7S (4)
Me 100 (4)
Own Name 100 (1)
Subtotal’ (8) (49) (114)

(d) Without Verb: Context

FORM +Noun Alone Other Subtotal*

| 3 67 (3)
My 25 75 (4)
Me 100 (1)
Own Name 28 72 (18)
Subtotal* (7) (19) (26)
* (Raw numbers).
** This table excludes two instances of other.

Example (24) illustrates that the use of J could also link up with a high
agentivity ranking. When J was used in utterances ranking high (such as
example 24), it tended to be used in conjunction with the quasi-modal form
wanna; rarely would one find instances such as example (25):
(25) I eat your foods.

Very little can be said about the semantic rankings of the other self
reference forms used by Megan in Month II. The three examples of My that
 146 CHAPTER 7

were coded for agentivity appeared in different surface contexts and show
no unified semantic pattern. There was no longer evidence that My was
used in the two semantic contexts outlined for Month I. Similarly, there was
only one instance of Own Name coded for semantic agentivity; and thus
one could conclude only that Megan no longer used her own name in
conjunction with activity verbs, as was the case in Month I.

Pragmatic Characteristics. An analysis of the relation between par-

ticular self reference forms and pragmatic function reveals that during this
month Megan used language primarily for its assertive function. Of all her
utterances containing self reference forms, 70% were coded as assertives,
whereas only 17% were coded as control acts. Thus from Month I to Month
II we find a decrease in the frequency at which Megan used self reference
forms in attempting to bring about changes; language took on a more rep-
resentational function. With these general changes in mind, we can turn to
an analysis of the particular self reference forms.
Figure 7.5 illustrates the relationship between the various forms and their
pragmatic function. The distributional pattern for J looks fairly similar to
that of Month I. Megan continued to use J primarily in assertives (65%), and
control acts accounted for only 18% of all self reference. There were no real
changes from Month I to Month II in terms of the function of utterances
containing the self reference form J The function of utterances containing
Own Name also did not change; in Month II, Own Name continued to be
used in assertives (86%), and only rarely was this form used in a control
act. The functional distinction between these two forms was primarily the
same as in the first month of the study. J linked up with the expression of
ongoing states, whereas Own Name was found in the context of naming
and identifying the self, most often in the photo-book activity.
Where we see the most substantial changes in Megan’s self reference
system with regard to language function is with the form My. Recall that in

GBB Assertion
(__] Multifunctional
Control Act
HE Uncoded

Use
of
Percentage

FIG. 7.5. Distribution (%) of self

reference forms in terms of prag-
matic function: Megan Month II.
MEGAN 147

Month I, Megan used this form primarily in control acts. Fifty-two percent
of all uses of My during Month I were coded as control acts. In contrast, in
Month II we find only one example of My used in a control act. First, it is
important to remember that there was a dramatic decrease in the overall
use of My during Month II; and second, Megan now used My primarily in
an adultlike way in conjunction with a noun (i.e., My mommy hair is blond
vs. My see a fireman). It is of interest that the single use of My in subject
position occurred in a dispute over control.

(26) Megan (M) and mom (MM) are playing with manipulative toys,
pretending to make food. M has said she has a cup of cottage
cheese.
a. MM: Okay. I'll have some. I'd like some please. (pretends to
take some, then starts to place spoon in M’s cup)
b. M: My gonna xxput/bitexx my cottage cheese. (pulling cup
back and placing spoon in it)

This example reveals that at times, My retained something of its previous

core field of use. Nevertheless, in the same session we find My + noun
combinations in utterances functioning as assertives.

Core Fields of Use. Given that neither Own Name nor My were noted
to link up with particular degrees of semantic agentivity, it is not worth
pursuing a combined analysis of the agentivity and the pragmatic function.
We should consider, though, a distinction said to guide the contrastive use
of J and My in Month I. It was noted that in Month I, Megan contrasted J
want x and My want x to highlight different perspectives on her role in the
action frame. When J was used, Megan herself was en route to the goal,
and her utterance was said to provide motives for her actions. In contrast,
My was employed in a similar utterance context when she needed assistance
to achieve a goal. In Month II we find Megan making use of a different
categorization scheme. J was now used in both of these contexts. J took
over what was previously marked by the //My contrast. The use of My was
mixed. Megan rarely used this form, and when she did it appeared in a
variety of sentence contexts and was linked neither to a particular degree
of agentivity nor to a pragmatic function. What is most interesting here is
that Megan’s strategy seemed to be one of avoiding control acts and the
form noted to be associated with them during Month II.

Other Self Reference Forms

Before concluding the discussion of Month II, I briefly consider the use of
the only other self reference form to occur, namely, Me. Although its use
was infrequent (6% of all uses of self reference forms), it reflected another
148 CHAPTER 7

view taken by Megan on action sequences. The following examples illustrate

its use.

(27) Me gonna built.

(28) Me no want this xx on over xx. (whispering)
(29) Me no want xx to play here xx. (whispering)
(30) For me more nuts.
(31) Me do it.

All of these examples contrast with those containing J to the extent that
they do not provide motives for other actions, nor do they have the self-as-
sertive quality of many of the utterances containing /. In the cases containing
Me, the child did not seem to take any responsibility for the happenings
being described. Example (27) was uttered as Megan approached the table
where the researcher had set out the blocks; and similarly, example G0)
was uttered as the researcher brought over more nuts for the children to
play with while using the manipulative toys. Examples (28) and (29) were
uttered by Megan as she wandered off from the play table, expressing her
lack of desire to play with the toys. Although the examples are few, one
can nevertheless speculate that the child took a more distant view of her
role in not wanting to continue the activities provided, de-emphasizing her-
self as the seat of control, and focusing more on such feelings arriving into
her consciousness. In all of these examples, it appears that Megan was
referring to events for whose occurrence she did not view herself as the
sole responsible force. In the discussion of Months III and IV, I further
consider the development of this form.

MONTH II

Megan continued to be ego-anchored in Month III of the study. Her mean

MLU was 2.52; there were no dramatic increases from Month II. With regard
to the proportion of her references to main participants that referred to self,
we find a decrease from 88% during Month II to 69% in Month III. Thus
even during Month III, much of Megan’s talk was about herself. Finally, we
can note that Megan’s distribution of forms remained similar to that discussed
for Month II. She continued to use a variety of forms, though she usually
depended on J (see Fig. 7.6). Notable changes in Megan’s distribution of
the various self reference forms included a slight decrease in the use of her
own name and an introduction of some new forms, We and Mine, neither
of which were used with enough frequency for me to examine functional
aspects of their use.
MEGAN 149

155%

>
FIG. 7.6. Distribution of self refer- saya

76). wee Other 8%

Surface Characteristics. Table 7.3 depicts the distribution of the vari-

ous self reference forms in terms of surface characteristics. During Month
III we continue to find an increase in the proportion of self reference forms
appearing in utterances with verbs. During Month II 67% of all uses were
in utterances with verbs, whereas in Month III 79% were. Furthermore, we
find that 7 continued to appear almost exclusively in utterances with verbs,
whereas Own Name typically appeared in utterances without verbs. Though
My and Me were not used frequently, they tended to appear in utterances
with verbs. Table 7.3a summarizes these findings. J appeared exclusively in
subject position in multiargument utterances. In contrast to Month II, where
Me was used in subject position, in Month III we find the few instances of
Me typically in object position, with the single exception of the utterance
Me scared. My typically appeared in object position as well and was used
as part of a possessive construction; there were. only a few instances of
preverbal My, and one of these appeared in conjunction with J in a way
noted for Jeffrey: J m’ant that. Own Name occurred primarily in a verbless
context in which the child was referring to photos of herself. Thus the uses
of the various pronominal forms were becoming more conventionalized in
Month III.

Semantic Characteristics. During Month III we find an increase in

the overall proportion of Megan’s uses of self reference forms ranking high
in agency. In Month II, only 20% of all uses ranked high, whereas in Month
III Megan’s uses ranked high in agency 30% of the time. I now consider the
way the individual forms linked up with agency.
Figure 7.7 depicts the distribution of uses of self reference forms with
respect to the degree of agentivity expressed. Again I should point out that
there were very few instances of any of the forms other than J, and thus
my discussion focuses on this form only. First, note that J was used primarily
in utterances ranking low in agency, though we find an increase in the
number of instances of J ranking high. In addition, note that unlike the case
in Month II, where Megan’s instances of J in utterances ranking high in
150 CHAPTER 7

TABLE 7.3
Distribution (%) of Self Reference Forms According to Surface Characteristics
(Megan Month III)**

(a) Verb / Verbless Context

FORM With Verb Without Verb Uncoded Subtotal*

| 98 2 42
My 78 22 9
Me 80 20 5
Own Name 21 79 14
Subtotal* (79) (21) (70)

(b) With Verb: Position

FORM Subject Object Uncoded Subtotal*

| 100 41
My 29 71 7
Me 25 a) 4
Own Name 100 3
Subtotal* (44) (11) (55)

(c) With Verb: Argument Structure

FORM Single Argument _—-Multiargument Uncoded Subtotal*

| S 95 (41)
My 14 86 (7)
Me 50 50 (4)
Own Name 33 67 (3)
Subtotal*

(d) Without Verb: Context

FORM +Noun Alone Other Subtotal"

| 1 (1)
My 1 1 (2)
Me 1
Own Name 73 27 (11)
Subtotal*

* (Raw numbers).
** This table excludes six instances of other.

agency included a modal form U/ gonna open that), in Month III there were
instances of the sort J knock it down and I make a cake. In this sense I
was becoming more multifunctional.

Pragmatic Characteristics. The general analysis of the pragmatic

function of utterances containing self reference forms reveals that the patterns
observed followed a developmental course noted in Month II. In Month III we
continue to find a gradual decrease in the proportion of self reference forms
MEGAN 151

g
i feya1)
g
Percentage
Use
of

FIG. 7.7. Distribution (%) of self 0%

reference forms in terms of seman- ati. oh fn wa,
tic agency: Megan Month III. Form

used in control acts. Although in Month I Megan’s use of self reference forms
in control acts accounted for 28% of all use of self reference forms, in Month
III such usage accounted for only 12%. We simultaneously find a gradual
increase in the proportion of self reference utterances that functioned as
assertives. By Month III, 72% of all uses of self reference forms served this
function.
The analysis of the relationship between particular self reference forms
and specific language functions lends no evidence for the claim that Megan
contrasted the forms in terms of pragmatic function (see Fig. 7.8). All forms
were found primarily in assertives, and at this point no specific linkages
were apparent. It is of interest that the utterance J m’ant that, cited earlier,
occurred as the child struggled over control of a toy with her peer. Recall
that in Month II, such a dispute led to the use of My in subject position.
Although Megan no longer retained My in subject position and instead em-
ployed / in this example, we find traces of the old marker of control in this
utterance. This sort of gradual fading has also been noted in the other
children’s systems.

Percenage
Use
of

FIG. 7.8. Distribution (%) of self

reference forms in terms of prag-
matic function: Megan Month III.
152 CHAPTER 7

Core Fields of Use: 1 conclude this section with a brief summary of

the core fields of use of each form. J continued to be the primary self
reference form, and it was used most frequently when Megan provided
motives for why she or others acted or should act in particular ways. It also
appeared in assertions about accomplishments. Generally, this form was
used to refer to the inner feelings and evaluations of the speaking subject.
Own Name appeared when Megan was labeling pictures of self, and My
occurred in a variety of contexts, almost exclusively in possessive construc-
tions. Me tended to appear in object position, referring to the receiver or
the location of action.

MONTH IV

In Month IV Megan continued to be categorized as ego-anchored. Her mean

MLU was 2.58. And though I noted a decrease in the overall proportion of
references to self as main participant in subject position in Month III, refer-
ences to self rose to 81% of all references in Month IV. With regard to the
overall distribution of self reference forms (see Fig. 7.9), we find a pattern
that looks more like that of the nonego-anchored children. J continued to
be the primary self reference form used. In addition we find that the use of
Own Name decreased to only 9%, and finally we see Megan beginning to
add new forms into the system. Ten percent of all uses of self reference
forms were coded as Other; 75% of these other uses involved the word
mine.

Surface Characteristics. Table 7.4 summarizes distributional aspects

of the various self reference forms during Month IV in terms of surface
characteristics. There were few changes to note. The basic difference was
that J was the only form used in subject position. Both Me and My were
used in object position, and My was always used in possessive constructions.
Thus Megan’s use of pronominal forms no longer deviated from the target

161%

Own Name 9%

FIG. 7.9. Distribution (%) of self

My 13% reference forms: Megan Month IV
Me 6% (n=77).
MEGAN
153
; TABLE 7.4
Distribution (%) of Self Reference Forms According to Surface Characteristics
(Megan Month IV)**
ee ee a ae ee eee

(a) Verb / Verbless Context

es ee a aes 2 Se ee

FORM With Verb Without Verb Uncoded Subtotal*

| 45 1 (46)
My 2 7 1 (10)
Me 5 1 (6)
Own Name 7 (7)
Subtotal*
e
a a eee es ee a

(b) With Verb: Position

ne eee ee A ee ae Fhe a eee ie
FORM Subject Object Uncoded Subtotal*
| 45 45
My 2 2
Me 5 5
Own Name
Subtotal*
ae A A a i ee Ee ee ee OE ea 6 nt Pee

(c) With Verb: Argument Structure

FORM Single Argument _—_Multiargument Uncoded Subtotal*

| 5 40 45
My 2 2
Me 1 4 5
Own Name
Subtotal*

(d) Without Verb: Context

FORM +Noun Alone Other Subtotal*

| 1 (1)
My 7 (7)
Me 1 (1)
Own Name 2 5 7
Subtotal*

* (Raw numbers).
** This table excludes eight instances of other.

language in ways noted in previous months. Own Name was never used in
constructions with verbs.

Semantic Characteristics. The major difference between Month III

and Month IV for Megan involved the frequency with which her references
to self ranked high in agency. Whereas in Month I only 21% of all uses of
self reference forms coded for agentivity ranked high, by Month IV 51% of
the utterances coded for agentivity ranked high. This change was most
 154 CHAPTER 7

dramatic for the form In Month I, 92% of all uses of J coded for agentivity
ranked low, whereas in Month IV only 41% ranked low and 55% ranked
high. Thus in Month IV we find that Megan employed J most frequently
when referring to herself as an agent who brought about change.

(32) I open this.

(33) I did it.
(34) I put (th)em back now.

The use of the other self reference forms in the context of utterances with
verbs was too infrequent to make discussion of patterns meaningful; only
17% of all uses of self references forms coded for agency did not involve
the use of 1.

Pragmatic Characteristics. The patterns for the coding of pragmatic

function did not change much between Month III and Month IV. The minor
changes followed the overall pattern noted for Month III; there continued
to be a small decrease in the proportion of utterances containing self refer-
ence forms coded as control acts (10%), and a slight increase in the number
of utterances coded as assertives (78%). Figure 7.10 depicts the relation
between the particular self reference forms and various language functions.
Several forms linked up with assertives, ruling out the kind of one-to-one
mapping noted earlier in the study.
Although several of the forms were now multifunctional, it is interesting
to contrast how Megan expressed control acts in Months I and IV. In Month
I, I noted that J and Own Name occurred in different kinds of assertives
and that My was used primarily in control acts. What was interesting in
Month IV was that Megan now recruited a full range of forms for a wide
range of control acts. For instance, requests soliciting permission were used

100%
GE Assertion
(__] Multifunctional
E> Control Act
GH sUncoded

Use
of
Percentage

0% a FIG. 7.10. Distribution (%) of self

My Me OwnName A
n=47 n=10 aos n=7 reference forms in terms of prag-
Form matic agency: Megan Month IV.
MEGAN 155

with I UZ do a nut, okay), and requests for assistance occurred with Me

(Could help me?), whereas the newly introduced form Mine occurred in
disputes over ownership. Control acts previously tended to co-occur only
with My, but now several different forms were used to express different
kinds of control acts. Thus, although in Month III, I noted that Megan’s
strategy en route to the adult system differed from Jeffrey’s to the extent
that she went through a phase of minimal use of control acts and avoidance
of the form My that was said to link up with control acts, in Month IV we
find exactly the sort of hierarchical integration and differentiation of forms
so characteristic of Jeffrey’s system. I return to this issue in the concluding
section of this book.

Core Fields of Use of Self Reference Forms. By Month IV of the study,

I noted that Megan had created a system of self reference in which the forms
could be distinguished across multiple levels of analysis. This contrasts with
her early system, in which each form linked up with a semantic and a pragmatic
category dealing with agentivity and control. By Month IV, Megan no longer
used avariety of self reference forms in subject position. She tended to rely
on a single form, % and she used this form when reporting her ongoing
involvement as an agent in actions, as well as when reporting internal states.
Own Name continued to be used in acts of labeling, primarily when referring
to actions and possessions depicted in photographs. My was employed
exclusively in possessive constructions. Me appeared in object position
marking self as recipient of actions-at times in the contexts of requests.
In conclusion, although Megan’s MLU and the proportion of her references
to self categorized her as ego-anchored, other evidence indicates general
signs of a burgeoning ability to talk about self in relation to other. For
instance, during this month Megan began to employ regularly the preposi-
tions forand with. Forappeared when Megan expressed her role in carrying
out actions for another’s benefit (1 open this for you). With was used as a
marker of coparticipation U7 go to the zoo with Sandy). Thus, Megan was
slowly developing other linguistic resources to talk about her involvement
in action fields in more complex ways. Such devices, though restricted to
the self, represented the first steps toward being able to situate others in
action fields as well. One finds a similar integration of a range of linguistic
devices for talking about self in relation to other and reference forms to
refer to other in the other ego-anchored children as they became nonego-
anchored. I consider the three nonego-anchored children in the next chapter
and return in the final section to a discussion of common characteristics of
the ego-anchored children’s development of systems of self reference.
 CHAP TER EIlGHg

The Nonego-Anchored
Children

The nonego-anchored children included Eric (28 months), Keith (31 months),
and Thomas (32 months). At the onset of the study all three had a mean
MLU over 3.0. Several other factors led to the conclusion in chapter 4 that
this group of children was linguistically more advanced than the ego-
anchored children. The nonego-anchored children could be distinguished
from the three ego-anchored children in that: (a) in their utterances they
referred almost as often to others as to themselves (see Fig. 4.1), (b) they
relied primarily on a single self reference form /, and (c) their use of self
reference forms did not deviate from the target language in the ways noted
for the ego-anchored children. Because the findings for the nonego-anchored
children were so similar, I consider the children jointly, noting along the
way any major individual differences. I begin this discussion with a recon-
sideration of the general distribution of self reference forms for the three
children at the onset of the study.
The nonego-anchored children made use primarily of one self reference
form. In contrast, the ego-anchored children used several self reference
forms with fairly equal frequency. For each of the nonego-anchored children
one finds a second form (or set of forms) that was used with some regularity,
although far less frequently than the dominant form. All three of the nonego-
anchored children relied on the second most frequent self reference form,
We. As is indicated in Fig. 8.1, Eric used J 65% of the time and used We
10% of the time, with Own Name being used 8% of the time. Keith relied
on 156% of the time and used We 21% of the time. Finally, Thomas employed
I 60% of the time, whereas he used We 15% of the time. In the sections
that follow I first examine the basis on which the nonego-anchored children

156
NONEGO-ANCHORED CHILDREN 157

165%
Eric (n=116)

Own Name 8%

Other 8%
My 6% We 10%
Me 3%

156%
Keith (n=162)

Own Name 9%

My 7%
Me 1%

Thomas (n=178)

FIG. 8.1. Distribution of self refer- Otnerine

ence forms: nonego-anchored chil-
We 15%
dren. My 7% Me 1%

contrastively employed J and We. I then examine these children’s use of

other forms of self reference.

THE CORE CONTRAST: J AND WE

One might want to suggest that the use of J and that of We can be distin-
guished in terms of the conceptual distinction between singular and plural.
That is, it would seem likely that when the children employed / they were
referring to actions or states involving the self, whereas the use of We would
link up with references that were being made to both self and other. An
analysis of all instances of the use of J and We provides only partial support
for this claim. All utterances containing the form We have been coded in
terms of whether reference was made to an independent happening or a
joint one. The use of the term happening is meant to be neutral with regard
to whether reference was made to states, actions, or activities. If the children
were contrastively employing J and We on the basis of a distinction involving
plurality, one would expect that the use of We would link up with the
expression of joint happenings. Figure 8.2 summarizes the findings of this
analysis. We find that although the children often did employ We in the
context of reference to joint happenings, 40% of the uses of We involved
references to what was coded as an independent happening.
158 CHAPTER 8
Joint 53%

Uncoded7% FIG. 8.2. Distribution of We in

terms of individual or joint agency:
Individual 40% nonego-anchored children (1 = 72).

Given that the nonego-anchored children were not drawing a distinction

between J and We based simply on plurality, the question becomes: On
what basis were the children contrastively using these forms? I begin to
answer this question by examining the relationship between the core forms
and the semantic and pragmatic dimensions noted to guide the ego-anchored
children’s use of self reference forms. I do not review findings from the
analysis of surface characteristics because the two core forms were used
primarily in utterances with verbs, and in subject position, by all children;
and the sort of deviations from the target language noted for the ego-
anchored children were not found for this group of children.

Semantic Considerations

The discussion of semantic characteristics begins with an assessment of the

semantic characteristics of utterances containing self reference forms, regard-
less of the particular form employed. We find that all three children were
more likely to refer to self in utterances that ranked low in agency; 62% of
all self reference forms ranked low in agency, whereas 37% ranked high.
This distribution is quite similar to that reported for the ego-anchored chil-
dren toward the end of the study (see Budwig & Wiley, in press).
Turning to an analysis of the individual core reference forms, we can ask
whether the two forms differed in their agency rankings. Here we find
support for the claim that they did differ. Figure 8.3 shows that J tended to
rank low in agency, whereas We was slightly more likely to rank high than
low. It is important to note that although this finding held for all three
children, there were large distributional differences with regard to the form
I. Although all three preferred J in utterances ranking low in agency, the
frequency of occurrence ranged from 83% of the time for Keith to 55% of
the time for Thomas (see Fig. 8.4).
Recall that when the ego-anchored children’s use of the various self
reference forms did contrast in terms of semantic agentivity, J was always
the form most likely to occur in utterances ranking low in agentivity. It is
interesting that although the nonego-anchored children did not seem to link
form and meaning, as did some of the ego-anchored children, there are
tendencies reflected in the nonego-anchored children’s corpora suggesting
NONEGO-ANCHORED CHILDREN 159

ioe
90%
80%
70% Hs Low Agency
g ei High Agency
3 60%
3S
2 50% |
8 40%
a
30% 4
20%
FIG. 8.3. Distribution (%) of self 10%
reference forms in terms of seman- an
‘Oo

tic agency: nonego-anchored chil- | We

dren (7 = 310). Form

that J was more likely than We to occur in utterances ranking low in agen-
tivity. Taken together, the analyses provided here suggest that the contrastive
use of J versus We was not based solely on an assessment of semantic
agentivity. At best, the use of one form or the other can be viewed as related
to the notion of agency reviewed here. I report further explorations into
this issue after I consider the other coding dimensions.

Pragmatic Considerations

The trends with regard to the coding of pragmatic force are similar for the
various nonego-anchored children. Overall, all three of the nonego-anchored
children used language primarily in utterances containing self reference
forms, to make assertions about the world. Two of the children (Eric and
Thomas) used assertions 69% of the time. The other child, Keith, used
language to make assertions about the world 80% of the time. The analysis
of pragmatic force indicates that the nonego-anchored children used lan-
100%
a Low Agency
90% [==] High Agency
80% GEE Uncoded
70%

60%
50%

40%
Percentage
Use 30%
of
20%

FIG. 8.4. Individual differences in 0%

use of J in terrns of semantic agency: an tal oe
nonego-anchored children. Child
160 CHAPTER 8

100%
HD Assertion
90% (-_] Muitifunctional
=) Control Act
80% MB Uncoded
70%

60%

50%

40%
Use 30%
of
Percentage
20%

10% FIG. 8.5. Distribution (%) of self

0% reference forms in terms of prag-
! We My Own Name
n=272—s n=72 n=35 n=34
matic function: nonego-anchored
Form children.

guage to make assertions about the self more frequently than did any of
the ego-anchored children.
The question remains whether the core self reference forms linked up
with specific pragmatic functions. Figure 8.5 shows that both J and We
tended to occur with assertions, though there were distributional differences
between the two forms. J occurred with control acts 20% of the time as
well, whereas We almost exclusively (96% of the time) occurred with asser-
tions. Thus, as was the case for our analysis of semantic characteristics, we
find that there are not any one-to-one mappings between form and function,
though there are some indications that the two forms were distinguished
with regard to pragmatic force.

Core Field of Use and the I/We Contrast

I have presented evidence that the core self reference forms Jand We differed,
though the two forms could not be distinguished solely on the basis of an
analysis of semantic and pragmatic characteristics of the utterances containing
these forms. I now attempt to unravel the core field of use for each of these
forms, beginning the analysis of the contrast between these forms by appealing
to some examples of the utterances containing Jand We.

(1) Eric: I wanna pee.

(2) Eric: I’m gonna crash.
(3) Eric: Can I put it on?
(4) Keith: I don’t know.
(5) Keith: I push it down.
(6) Keith: I want the helicopter.
(7) Thomas: I don’t like that on there either.
(8) Thomas: I wanna take this off.
(9) Thomas: Can I use your thing for a minute?
 NONEGO-ANCHORED CHILDREN 161

(10) Eric: We gonna build a bridge.

(11) Eric: We are going to turn this on. (talking about microphone)
(12) Eric: We're riding bikes. (referring to a photo)
(13) Keith: | We're gonna build big castle.
(14) Keith: Now # we should # um # put the people xx xx right put
one to bed.
(15) Keith: We are playing in some grass. (referring to photo)
(16) Thomas: Yeah we should # build a little # city in here.
(17) Thomas: What we do is take some of that see # with this +.
(18) Thomas: We're putting grass in this. (referring to photo)

Rather than beginning with preestablished categories of the sort described

earlier, the present analysis is guided by a form out approach. The question
is framed in terms of how utterances differed from those involving the use
of Z Four distinctions have been noted, and I briefly discuss each.
In the earlier discussion of semantic characteristics, I noted that We was
more likely to occur in utterances ranking high in agency, though both /
and We were used by the nonego-anchored children in utterances ranking
low (see Fig. 8.3). Looking at the examples, though, one finds very few
examples of We with stative verbs. Figure 8.6 reveals results from an analysis
of all the instances of the forms J and We that ranked low in agency, in
terms of whether they co-occurred with state or activity verbs. J tended to
occur with state verbs; 65% of all instances of J coded as low in agency
involved utterances with state verbs. In contrast, of all the utterances with
We ranking low in agency only 24% occurred in utterances with state verbs;
rather, most uses of We ranking low in agency occurred with activity verbs
that were not highly agentive.

100%
on GE State verb
A (_) Activity verb
80%

70%

50%

Use 40%
Percentage
of
30%

FIG. 8.6. Distribution (%) of forms

coded as low agency in terms of 10%
verb type: nonego-anchored chil- n=168 n=29
dren, Form
162 CHAPTER 8

Differences in the distributions of We and J go beyond verb semantics.

A second important co-occurrence pattern concerns the relationship between
form and activity task. Recall that the children were asked to play with
manipulative toys and blocks and look through a photo-book. An analysis
of the use of form by activity task reveals that 35% of the uses of We clustered
during play with blocks. At first glance, this is surprising, because play with
the manipulative toys and play with building blocks would be equally likely
tasks for the form We to appear in. Only 22% of all uses of We were found
in play with manipulative toys. As was mentioned in chapter 3 in the dis-
cussion of the method, the reason both play with blocks and play with
manipulative toys were selected is that they elicit different kinds of talk
about action. I suggested there that play with manipulative toys encourages
talk about ongoing action, whereas play with blocks elicits much talk about
plans and goals. We see later that this distinction plays a role in the con-
trastive use of J and We.
A third interesting distinction previously noted regarding the use of We
is that it often appeared in stretches of discourse that could not be charac-
terized as involving joint action. That is, all three children used We when
referring to their own independent actions (see Fig. 8.2). These three levels
of analysis can be linked into two predominant patterns. I now consider
these clusters.
One set of uses of We involves utterances that referred to joint activities
in the photo-book task. A second grouping consists of uses of We that
referred to independent accomplishments in the context of block play. In
essence, we have two related uses of We. The first use of We, that which
tended to appear in talk during the photo-book task, appeared when the
children referred to pictures describing themselves and other(s) engaged in
depicted action. For instance, in example (12), Eric referred to a picture of
several children Gincluding himself) riding bicycles by saying, “We're riding
bikes.” Similarly, Keith referred to a picture of Thomas and himself outside
in the school yard by saying, “We are playing in some grass” (see example
(15)). In contrast, the second use of We, that found primarily in the block
session, tended to occur in reference to goal-directed behaviors the children
were performing independently.
The major question is how the uses of J and We can be distinguished.
One use of We can clearly be distinguished from J simply by saying the
distinction depends on a notion of plurality. That is, when the children
referred to themselves and others in joint reference they employed the form
We, and when they referred to themselves alone they used the form J But
this account would need to be modified for two reasons. First, the uses of
We involving joint reference tended to occur only when they were talking
about depicted action; and second, I have noted that the children often used
We in reference to independent action. The question becomes, then: How
NONEGO-ANCHORED CHILDREN
163
can the uses of We in reference to independent action be distinguished from
uses of J in reference to independent action?
A more careful examination of the examples involving J and We in ref-
erence to independent action reveals two other points of contrast. First, the
uses of We often co-occurred with time and manner adverbials. One finds
that utterances with We also included now, then, and first, whereas the
utterances with J rarely did. Another feature distinguishing the use of J and
We is that We frequently co-occurred with particular modal forms referring
to obligations (i.e., should, hafta) as motives for future actions. In contrast,
the use of J with action verbs often did not involve modal forms.
At several different levels of analysis, the use of J and that of We can be
distinguished. The co-occurrence patterns at each of these levels can be
linked together at a higher level of analysis into broader activity frames. Just
as was the case for the ego-anchored children’s use of various self reference
forms, the nonego-anchored children’s use of J and We in reference to
independent actions linked up with different clusters of notions organized
around prototypes. J accompanied physical object play structured around
action sequences in which talk referred to the child’s desire to manipulate
particular objects or to the child’s actual manipulations, regardless of the
degree of agentivity expressed. In contrast, We referred to independent
action in sequences in which physical object play was organized around a
plan. In such cases, an action was referred to that was part of a broader
action sequence leading to an independent or jointly conceived goal. The
motive for the action described transcended the individual’s subjectivity.
Although the child might carry out the action involved, it is as if the child
distanced himself from taking full responsibility for the action. The action
carried out was conducted in accordance with a plan for the achievement
of a goal.
The following examples illustrate the contrastive use of Jand We when
referring to goal-directed actions.

(19) Keith (K) and mom (KM) are playing with toy people who can be
placed on small bicycles.
aK; These are babies. They need to stay on that (be)cause they
are resting in their beds. Their daddy is on his bicycle and
he-/ I am gonna take their daddy off his bicycle.
. KM: Good idea.
wie Umm (.) I cannot get him off!
(pulls man off bicycle)
Feno There!

(20) Keith (K) and mom (KM) have been building structures with blocks.
K lifts helicopter.
164 CHAPTER 8

anv;
mK Look what I d& did.
b. KM: Opp. Did you close off the propellers?
CNS Yeah.
d. KM: Did you put it away for the night?
en ie No.
f. KM: No. (augh) Don’t be silly Mom.
oak: We should. (pause) We should, um um, put the people
out of it. Put one to bed. (holds plane looking at people
inside, then places block flat on table)
= KM: You gonna put them to bed?
ink: Yeah (.) because they’re crashed into a rock.

Note how, in example (19), the child’s decision to take the daddy off his
bicycle was not part of a broader activity frame. The decision seemed to be
made as the child was manipulating the toys. Many of the uses of J were,
in fact, in announcements of action sequences that were not conforming to
the children’s desires. For instance the statement J cannot get him off was
uttered as the child actually tried to get the figure off the bike.
In contrast, We in example (20) was employed in the context of the child’s
descriptions of his plan of what needed to be done. Even his mother
interpreted his action as part of an independent plan. In line (20h), Keith’s
mother did not respond by saying We can do that or the like but, rather,
allowed the child to continue on his own. In this example, as in the others
involving the use of We in reference to independent actions, the child’s
reference to a particular action can be contextualized as part of a broader
sequence. This can account for several of the features noted to link up with
the use of We. For instance, the use of time and manner adverbials in
conjunction with We would follow if the child were referring to steps in a
scripted sequence. It also would follow that We would occur more frequently
in play with blocks, because block play often elicits talk about goal-directed
action, and that J would occur in the manipulative activity task, because this
task was noted to elicit more talk about immediate action descriptions. Note
that it is, of course, possible for the forms to occur in other tasks. We see in
example (20) that the child used both Jand We. In line (20a), the child reported
on an action he had just completed—an action that in fact was unplanned—
and here we find that the child used the predicted form I. Thus, the use of J
and We does seem to involve a contrast (see also Nelson, 1989). Although both
are found in reference to independent actions being carried out by the speaker,
We seems to have been used as a kind of marker of impersonal agency. The
child indicated that although he was the instigator of the action referred to,
the motive was not personal desire but, rather, broader goals.
The question remains: To what extent can one find a connection between
the two different uses of We? I have noted that We was not found simply
 NONEGO-ANCHORED CHILDREN
165
in references to independent actions but also when the child referred
to self
and other(s) in depicted action. At this point, we have no way
of knowing
whether the two uses of We existed side by side as independent form—fu
nc-
tion clusters, or whether the children saw some overarching connect
ion
between the two. It seems at least possible that the connection for
the child
was that the references to self and other in depicted action were also
viewed
as downplaying a notion of volition. The children might have been
referring
to self and other as actors rather than as agents, the distinction being
that
an actor is an animate participant, whereas reference to agency carries
with
it an implication of volition and responsibility. These data, though, are in-
conclusive. We can conclude simply that although the uses of J and We at
times looked quite similar, there seems reason to believe that the children
used these forms contrastively.

OTHER SELF REFERENCE FORMS:

MY, ME, OTHER, AND OWN NAME

The Use of My

In considering each of the ego-anchored children I noted that My played a

central role in the core contrast for each of the three children. One major
difference between the ego-anchored and nonego-anchored children was
that the nonego-anchored children rarely used My. One can nevertheless
examine the core field of use of this form for the nonego-anchored children.
My analysis shows that, by this point in development, the nonego-anchored
children did not seem to link this form up with either high agentivity or a
particular pragmatic force. The distinction between this form and the other
self reference forms can be accounted for largely in terms of surface char-
acteristics.
With regard to surface coding we find that My appeared in clauses with
and without verbs. When appearing in utterances with verbs, it tended to
be used in object position; and in such instances, My always modified a
noun (e.g., Now your bicycle crashes my bicycle). When My was used in
subject position it typically occurred in conjunction with a noun (e.g., My
tree trunk fell down). Only once did Keith use My in subject position fol-
lowed by a verb (My don’t know). In contrast to the ego-anchored children,
we find that the nonego-anchored children did not employ My in subject
position in a way that deviated from adult input.
Turning to a consideration of the semantic level of coding, I examine
whether the nonego-anchored children tended to use My with clauses ex-
pressing a particular degree of agentivity. In contrast to the ego-anchored
children, there is no indication that the nonego-anchored children reserved
166 CHAPTER 8

the use of My for the expression of high agentivity. There was no preference
among the nonego-anchored children for using My in utterances ranking
either high or low with regard to semantic agentivity.
With regard to pragmatic function, there also does not appear to be a
one-to-one mapping between use of the My form and pragmatic control.
Recall that the ego-anchored children used My primarily in control acts. As
is shown in Fig. 8.5, the nonego-anchored children used My primarily in
assertions, though they also used it with control acts. More specifically, 34%
of all uses of My were in control acts, whereas 51% of all uses of this form
were in noncontrol acts. The finding that there was a slight preference for
noncontrol acts relates to the general finding that the nonego-anchored
children’s language was more frequently coded as noncontrol acts.
The basic finding, with regard to the nonego-anchored children, is that
their use of My was not linked with the semantic and pragmatic features,
as was noted to be the case for the ego-anchored children. The nonego-
anchored children did not use My in subject position in a way that deviated
from English input. The use of My by the nonego-anchored children appears
to be like that of adult speakers of English—the nonego-anchored children
used it as a possessive pronoun.

The Use of Me

The use of Me was similar to the use of My by the nonego-anchored children

in two ways. First, in contrast to some of the ego-anchored children, the
nonego-anchored children did not use Me in subject position in a way that
deviated from the adult English input they received. In addition, the use of
Me did not link up with the specific semantic and pragmatic properties of
agentivity and control that were coded. Nevertheless, the use of Me can be
related to particular perspectives the children took on their role in action
frames and to the way the children situated the self in stretches of discourse.
The following list includes all uses of Me by the three children in the first
month of the study. Although the instances were not frequent, they never-
theless reveal some patterns:

(21) But I want that mine on me. (referring to microphone)

(22) Can you do some for me?
(23) You make a house with me.
(24) Mama, will you come outside with me?
(25) They don’t look black to me.
(26) Me is a tray.
(27) This is what me and Lily did.
(28) Take me fora ride in the car. (singing)
 NONEGO-ANCHORED CHILDREN
167

(29) Yeah he gave me a piggyback ride.

(30) Can you help me put this um thing in his hand?

Most of these examples can be grouped together in terms of the case role
of dative-“a conscious participant or recipient in events or states” (Giv6n,
1984a, p. 126). The uses of Me with prepositions expressed a relationship
between the child and the action frame, and it was the role of the particular
preposition to qualify this relationship. For instance, in example (22) the
self acted as a kind of benefactor, whereas in examples (23) and (24) the
self was viewed as a coparticipant (cf. Foley & Van Valin, 1984; Givon,
1984a, for further discussion of coagency and the use of with). A second
use of Me is illustrated in examples (28)30). In these examples, Me was
a dative/benefactive object that had been promoted to the direct object slot,
thereby losing the prepositions indicating semantic case role. Whether the
children’s placement reflected a discourse-based strategy to promote second-
ary topics or instead conformed to a general tendency in English to promote
dative-benefactive objects remains a question for further empirical study.!
One can question whether the use of Me more generally reflected an
underlying discourse-based strategy to take a particular perspective on the
self. Considering example (27), we can question why the child did not select
the J form (e.g., This is what Lily and I did). It could be that the choice of
Me was related to the speaker’s focus of interest. The switch in form could
indicate that such information was nonfocused (see Zubin, 1979, for a more
complete discussion of the relation between case marking and focus of
interest). Looking at the context of example (27) helps clarify this distinction,
as revealed in the more extended example (31):

(31) Researcher (R) brings over tray of blocks to table where Thomas
(T) and Keith (K) stand.
a. R: Look at this.
b. K: (inaudible)
c. R: You guys can build. (placing blocks on table)
d. T: This is what me and Lily did.
CL kh yea,

In line (31d), focus was placed on the building activity. Although this ap-
proach remains speculative, it is interesting to relate (31) to reports of Bellugi
(1971) and Brown (1973) concerning Adam, Eve, and Sarah’s occasional use

‘In a study of English texts, Givon (1984b) found that in two-object clauses, dative-bene-
factive objects rarely appeared as indirect objects (only 16%) and, rather, tended to be promoted
to direct object (84%), whereby they lose their semantically relevant case markers (e.g., Tell
her a story vs. Tell a story to her).
168 CHAPTER 8

of “pronouns designed to serve as surface objects, as ‘nominative’ pronouns

or subjects” (Brown, 1973, p. 141). Bellugi (1971) noted that in such instances
the the child transposed an adult utterance into a question. This is illustrated
in the following sequences taken from Bellugi (1971, p. 100):

(32) Adult: I think you broke it.

Child: | Why me break it?
Adult: | You got some chocolate on it.
Child: | Why me get some chocolate on it?

According to Bellugi, at this phase, the child employs a strategy whereby

reference to self in utterance-initial position is marked by J, whereas else-
where within utterance boundary reference is marked with Me. Brown (1973)
noted similar errors but offered no explanation for them. An alternative
suggestion is that the selection of J versus Me is in part related to discourse
factors having to do with focus. If self is not in focus then the child switches
to Me. For the time being this must remain an empirical question. More
examples need to be gathered, perhaps via elicitation techniques, to disen-
tangle the alternative accounts.
In summary, I have noted that all three children used the form Me to
refer to self as conscious participant. Like the ego-anchored children, the
nonego-anchored children attempted to use this and the other self reference
forms to situate the self differently in action frames. In addition, the nonego-
anchored children used Me in utterances that de-emphasized the selfs agen-
tive characteristics. Focus was placed instead on the consciousness of the
self in receiving and benefiting from action. In contrast, though, to the
ego-anchored children’s solutions, we find that the nonego-anchored chil-
dren also attempted to coordinate the relationship between self and other
in relation to action. For instance, issues of coagency now came up. In
addition, we find that the use of Me may, perhaps, relate to broader discourse
strategies involving focus of interest.

The Use of Own Name

As can be seen in Fig. 8.1, 8% of all uses of self reference forms involved the
child’s use of Own Name.’ That the nonego-anchored children still employed
such a form is rather surprising given that many studies of early grammatical
development have noted that by this phase of language development, the use

71 do not provide a full discussion of other self reference forms used by the nonego-anchored
children. Several groups of forms were used, including mine(s), let’s, first person plural forms
(us, our), third person forms (he, his, they). 1 integrate into the present discussion the use of
third person forms. The reader can find a more complete discussion of the other forms of self
reference in Budwig (1986).
NONEGO-ANCHORED CHILDREN 169

of Own Name has given way to the employment of pronominal forms (cf.
Bellugi, 1971; Brown, 1973, among others). It may be that the use of this form
is limited to a context not included in such previous studies.
In many ways the use of Own Name by the nonego-anchored children
looked quite similar to that of the ego-anchored children. All 34 examples
occurred in the context of photo labeling. Some examples from each of the
children illustrate this:

(33) E>: Eric:

(34) E: Here’s Eric and Jeffrey.
(35) E: And that’s Eric on a motorcycle without his hands.
(36) K: This is Keithy and this is Thomas.
(37) K: There’s Keith under Thomas.
(38) K: I see Keithy talking.
(39) T: And Thomas.
(40) T: Thomas getting down.
(41) T: Thomas with a tennis ball.

Such instances often occurred in the context of question—answer routines,

and frequently these utterances were accompanied by a pointing gesture.
The use of Own Name rarely occurred with verbs; instead it occurred alone
or in combination with deictic terms. At times, stative information was also
given. Example (38) is particularly interesting, because we find double ref-
erence to the self. Keith referred first to himself as speaking subject with
the pronominal form J and then to self depicted in the photo with the
nominal form Keithy.
Although at first glance the nonego-anchored children’s use of Own Name
looks similar to that of the ego-anchored children, there are some important
differences. Recall that one of the ego-anchored children, Jeffrey, at times
employed the pronominal form Me in reference to photos. The nonego-
anchored children also used pronominal forms at times in their references
to the photos, but they never once employed first person singular forms,
instead using he, his, they, we, and our. These pronominal forms appeared
in a different discourse context from Own Name. All three nonego-anchored
children relied on the various pronominal forms when they went beyond
picture labeling and gave picture descriptions or short narrations about the
photos. The following examples illustrate this usage:

(42) Keith (K) and mom (KM) are looking at the photo-book.
we: I see Keithy talking. (pointing to photo)
b. KM: And Keithy is talking?
Cas Yeah. They bumped heads. (pointing to same photo)
170 CHAPTER 8

(43) Thomas (T) and Keith (K) are looking at photo-book with the
researcher (R).
Thomas getting down. (patting page)
Thomas getting down, uhhuh.
What’s he gonna get?
What do you think? I think it looks like he’s doing something.
He’s gonna get a shovel. (looking at R)
Umm that’s a good guess.
Maybe he’s gonna get a tennis ball.
roe
ssee
moan
oe
Ne Umhmm. Wanna turn the page?

In these examples, the children initiated reference to themselves with the

use of Own Name, whereas they maintained reference to self with a first
person plural or a third person form. For instance, in example (42), the form
Keithy is found in line (42a), whereas the child referred to the actions of
his peer and himself with they in line (42c). Likewise, in example (43),
Thomas first labeled the photo with the nominal form (see line (43a)), and
he sustained reference with the form he in lines (43¢) and (43g).
The question can be raised: Why did the children come up with this
system? In particular, why did they exclude first person singular forms when
talking about depicted self and why did they rely on either first person
plural or third person pronominal forms? Although answers to these question
deserve further examination, one can speculate that the nonego-anchored
children reserved the use of the first person singular pronominal forms for
the reporting of highly agentive or experiential events of the speaking sub-
ject. The use of first person singular forms seemed to be reserved for sub-
jective accounts. It is intriguing that the first person plural form We could
occur in reference to both ongoing and depicted experiences.

SUMMARY AND DISCUSSION

The discussion of the nonego-anchored children’s use of various pronominal

forms has pointed up ways in which the children’s use of these forms more
closely approximated use in the target system. Nevertheless, we have also
seen that the children’s use appeared different from that in adult English in
certain respects. Likewise, the nonego-anchored children’s use of self refer-

’The use of nominal and pronominal shifts in the proto-narrative descriptions of the
nonego-anchored children presents an intriguing correspondence with the use of such forms
as cohesive devices in later narratives about nonself protagonists (cf. Bamberg, 1987;
Karmiloff-Smith, 1981).
NONEGO-ANCHORED CHILDREN 171

ence forms shared both similarities and differences with that of the ego-
anchored children. I now consider these similarities and differences.
The nonego-anchored children’s use of self reference forms was similar
to the ego-anchored children’s in two respects. First, like the ego-anchored
children, the nonego-anchored children linked the use of particular forms
with a function cluster. The use of the forms was not arbitrary but, rather,
motivated by various functional features. A second and related similarity
between the two groups of children concerns the interrelationship between
the uses of various forms. Not only did all six of the children link the use
of forms with particular functional clusters; in addition, such form—function
pairings were related to one another as part of a broader functional system.
For example, I noted that all of the children made use of a core contrast of
forms that linked up with a related set of functions. Thus all children con-
structed systems, rather than simple form—function pairings.
Despite these general similarities, the nonego-anchored children differed
from the ego-anchored children in several important ways. First, I noted
that none of the nonego-anchored children regularly used a variety of self
reference forms in subject position in ways that deviated from adult input.
That is, the nonego-anchored children never used My or Me as a subject
pronoun. Thus, from the standpoint of the target language, the nonego-
anchored children’s use of forms was grammatically correct. Nevertheless,
an analysis of functional aspects of self reference forms did reveal ways in
which the target language and the nonego-anchored systems differed. The
importance of this is that it shows that correct grammatical use need not be
equated with a system that corresponds with adult usage. The difference
between the ego-anchored systems and the nonego-anchored systems is
more complex than a distinction between erroneous and error-free usage.
Each group of children constructed systems of form—function contrasts, and
what differed between the two groups was the specific nature of these
contrasts.
The basic distinction between the ego-anchored and nonego-anchored
systems of self reference is that the nonego-anchored children’s systems
were less tied to concrete action. I have noted that the ego-anchored children
divided up the use of several self reference forms to mark various degrees
of agentivity and control. In contrast, the nonego-anchored children used
one and the same form when referring to various degrees of agentivity and
control. For instance, I have noted that the use of J linked up with both
agentive and nonagentive reference and was found both in control acts and
in less dynamic descriptive language. Although the nonego-anchored chil-
dren were less concerned with marking various perspectives on the self’s
role in concrete action frames, the notion of agentivity nevertheless played
a role in the systems of self reference they developed. Rather than marking
the concrete relation between act and self as participant, they were more
172 CHAPTER 8

likely to mark distinctions such as the underlying motives and intentions

guiding action (ie., in the contrastive use of J and We), and the relationship
between concrete action and participant and the more general discourse
context (i.e., switching focus of interest with the use of Me).
To close this section, I briefly consider whether, then, the nonego-
anchored children used the self reference forms as markers of agentivity
and control. Although I have pointed up some evidence that the use of
particular forms (i.e., the contrast between J and We) was linked with notions
of agentivity, albeit in a more complex relationship than in the case of the
ego-anchored children, the analysis of other self reference forms (i.e., My)
provides no evidence that agentivity and control played an organizing role.
What seemed to be going on was a process of old forms integrating new
functions, and new forms taking on old functions (see Slobin, 1973; Werner
& Kaplan, 1984). That is, forms that previously were regularly employed to
mark agentivity and control at this point are not, whereas newly introduced
forms such as We function to mark distinctions related to agentivity and
control.
What also is relevant is that simultaneous to aspects of development with
regard to self reference forms, was the development of a variety of linguistic
systems that have been noted to mark aspects of agentivity in the target
language. For instance, the nonego-anchored children could now be said
to be developing a system of case marking that cut across person; and they
regularly employed prepositions, many of which qualified the relationship
between participant and action. In addition, the nonego-anchored children
made use of modal forms, which have been noted by Gerhardt (1990) to
relate to an orthogonal dimension involving motives guiding agency. One
more system coming into play at this point was that of voice. The children,
through the use of voice contrasts, were able to take various perspectives
on scenes that could, for instance, enhance or downplay agency (see Bud-
wig, 1990). Although I am not concerned here with the specifics of the
development of these systems, the important point is that such developments
provided the children with a full range of devices for marking various aspects
of agentivity and control. Thus, although the nonego-anchored children
seemed, on the one hand, to be moving away from a one-to-one corre-
spondence between self reference form and degree of agentivity and control,
the issue of marking various aspects of agentivity and control still was central
to the nonego-anchored children.
 PART THREE

MECHANISMS OF
DEVELOPMENT
 oa 7 -7 sition

: ; uy

: en
ayhae ~ vi
fea Oe? some
ty > 32 ne
Tebtysi

Geckied OI
ee

° 13-2
al A
—

y .

ee Tic ce ote P oer

;
Weotoic on. he SSO
ace), ORs Ee eee
.
(eas aPrAst ly @caatnig OF te
saw 7 too Wee et aero a
Wit > ean a eug erpns
4
ime th th,woe! dates Beary tee
a oie _ Peieeg | ‘Lisa ie Ot

~ seat gre
; i

“are ail a
;

iy me 8 lychee
oa rs a4
SP
, Tete. at ‘ re Hype 114 an ore
} > 7 :

i i ee ‘ices at
0b Me a? (coca SSR oe Dy L2yel emma €
we 7 kaa : Meeyh We wel? @ts@ eepeeta ahha
= . wy, | wa euteyedaen ,
a se Fert
Oonaig
ihae ae
, —, a OF wish
i Diaa BS
= eae thea® era io ter

‘ie =H gree cad ae ae

re (ees ed ,
a ote : ier ee
ae =) (68) “= &
j > rio ey Ore set Cie
= eo? [et peVig
maj ig of ar
tN aagh ~~ ©

~ he) ome: a

et tie a atSaletan
 CHAPTER NINE

The Organization of Linguistic

Systems: Forms and
Functions in Development

In Part II of this book, I have been concerned with the organization and
reorganizations of the use of self reference forms by six children acquiring
English as a first language. In this chapter I discuss the joint implications of
the various case studies. In the first section, I pull together the separate
findings from the various children. I consider two central questions. First,
given that the discussion focused on two groups of children, I reconsider
the evidence that these groups represent two phases in the organization of
self reference systems rather than individual differences. Second, I consider
the similarities and differences between the developmental paths traveled
by the various children en route to an adultlike system. In the second section
of this chapter, Iam concerned with various ways to account for the original
organizations of form and function by the ego-anchored children; and in
the third section, I turn to the question of what leads children to give up
their original systems.
In providing answers to these important questions, we see that at present,
multiple factors can be noted to playarole in the original organization and
reorganization of self reference forms and the functions that they serve. I
suggest, in the final section of the next chapter, several avenues of future
research that might help explain how various factors motivate the original
construction and subsequent reorganization of self reference systems.

175
176 CHAPTER 9

THE DEVELOPMENT OF A SYSTEM OF SELF REFERENCE

An Argument for a Developmental Path Between

Ego-Anchored and Nonego-Anchored
Self Reference Systems

In Part II of this book, I examined the self reference systems of six children,
dividing the children into two groups. One group, referred to as ego-
anchored, was studied longitudinally; and I argued that all three of these
children made special use of self reference forms en route to an adultlike
system. In contrast, another group of three children was said to use self
reference forms in a way that looked more similar to adult usage. At this
point I consider the evidence that these children’s usage of self reference
forms reflects a continuum of development, rather than the language of two
distinct groups of children. That is, I discuss the evidence that by the end
of the study the ego-anchored children looked like the nonego-anchored
children. My discussion focuses on two issues: (a) the distribution of refer-
ence to self and other, and (b) the distribution of self reference forms.

The Distribution of Reference to Self and Other. One of the dis-

tinctions between the ego-anchored and nonego-anchored children con-
cerned the proportion of time spent talking about self and others. In chapter
4 I noted that, as a group, the ego-anchored children spent the majority of
time talking about themselves as main participants at the onset of the study.
In contrast, the nonego-anchored children distributed their talk about self
and other more equally. If the two groups of children represent two phases
in development, one could expect that as the ego-anchored children became
older they should begin to look more like the nonego-anchored children
with regard to the distribution of self and other reference forms. In Part II
of this book I noted evidence suggesting that this was the case. All three of
the ego-anchored children looked more similar to the nonego-anchored
children by the end of the study, though there were some individual differ-
ences. For instance, by the end of the study Jeffrey looked like the nonego-
anchored children with respect to the distribution of self and other refer-
ences, whereas Grice and Megan still referred primarily to themselves but
showed a trend toward more talk about others than evidenced at the onset
of the study.

Distribution of Self Reference Forms. At the onset of the study I

noted two distinct patterns with regard to the six children’s distribution of
self reference forms. The ego-anchored group held together to the extent
that they made use of multiple self reference forms, often in ways that
deviated from the target language. In contrast, the nonego-anchored children
relied primarily on one given form. The data reviewed in Part II of the book
ORGANIZATION OF LINGUISTIC SYSTEMS 177

suggested that the ego-anchored children ended up looking like the nonego-
anchored children to the extent that by the end of the study all six of the
children relied primarily on one self reference form. In addition, the three
ego-anchored children gave up the deviant uses of self reference forms
noted at the onset of the study.

Summary. Taken together, the evidence from the longitudinal analyses

of the distribution of self and other reference forms, as well as the analyses
of the use of particular self reference forms, lend support for the claim that
these were not two distinct groups of children—one that began with seem-
ingly adultlike usage and one that built a temporary system. Rather, the
separate analyses suggest a developmental route between the two groups
of children such that by the end of the study the ego-anchored children
looked more similar to the nonego-anchored ones. Whether this pattern
holds for all English-speaking children is an empirical question (see later
comments on future directions).

Similarities and Differences in the Ego-Anchored

Children’s Organization and Reorganization
of Self Reference Systems

Much of Part II involved the detailed examination of the paths traveled by

the three ego-anchored children en route to the more nonego-anchored
systems of self reference. At this point I consider whether the children as a
group showed any similarities in the sorts of developmental changes that
took place en route to an adultlike system, or whether each child differed
in the sorts of organizations and reorganizations that took place. I consider
first the similarities and differences in the original organizations of the sys-
tems and then the subsequent reorganizations. Finally, I consider whether
the individual pathways traveled share a common principle.

The Original Organizations. At the onset of the study I noted that

all three ego-anchored children made use of multiple self reference forms
to refer to themselves, and all three used these in ways that deviated from
the adultlike system. Furthermore, in contrastively employing these forms,
the children all drew upon a common core set of notions having to do with
agency and control. Nevertheless, as was pointed out in Part II, there was
a certain amount of variation between the ways in which the original systems
were organized by the three children. Grice contrasted primarily between
three forms of self reference. The majority of her utterances did not include
verbs, and her self reference system involved primarily a fairly simple contrast
between various pragmatic functions. 7 appeared primarily in assertions, My
appeared in utterances about the self involving control, and Me appeared
178 CHAPTER 9

in both control acts and noncontrol acts tending to be aligned with a notion
of self as receiver or benefactor.
Megan’s original system also seemed to divide forms on the basis of
pragmatic function, though she integrated a core set of semantic notions as
well into this contrast. For instance, although / linked up with assertions, it
also occurred in utterances in which self was experiencer, whereas My
appeared in control acts, though often in utterances that expressed self as
agent. Thus, Megan’s system differed from Grice’s to the extent that semantic
agency played acentral organizing role in the division between various forms.
Jeffrey’s system was the most complicated of the three at the beginning.
Like Megan, Jeffrey tied the use of particular forms to related semantic and
pragmatic notions. In contrast to Megan’s system, with its neat one-to-one
mapping between clusters of semantic and pragmatic dimensions (.e., as-
sertion and experiencer), Jeffrey's system was somewhat more multifunc-
tional. One form, My, linked up with both pragmatic control and pragmatic
noncontrol, though its usage could be predicted based on semantic agency.
I was typically found with noncontrol acts oe the expression of self
as experiencer.
What was similar about all three children was Wh attempt to relate
surface forms with semantic and pragmatic notions of agency and control.
What differed were the specifics of these constellations. I now consider how
the children moved from these form-function clusters to more adultlike
usage.

The Reorganization of Self Reference Systems. Pulling together the

findings presented in chapters 5, 6, and 7, we find three different routes
into adultlike systems of self reference. I compare and contrast the two
systems that differed most from one another, namely, those constructed by
Megan and Jeffrey, and then I relate these two systems to that used by the
third child, Grice.
Megan came up with a fairly straightforward way to move from a system
of self reference organized around semantic and pragmatic notions to a less
contextually driven system. As was noted earlier, and is represented in Fig.
9.1, Megan began with a contrast between three forms, each serving a distinct
function. Megan’s solution was simply to delete the extra forms and functions
during Months II and III of the study. That is, Megan moved from the use
of a variety of self reference forms to mark a variety of functions, to a phase
in which she relied on one self reference form to talk about one function.
During Months II and III Megan avoided using forms such as My and Me,
and when referring to herself she neither used language to control nor made
use of action verbs. By Month IV, she began using / multifunctionally and
at the same time used My as a possessive pronoun (correctly) and Me as an
object pronoun. Simultaneously, in Month IV, we find Megan making use
ORGANIZATION OF LINGUISTIC SYSTEMS
179
Month I
Form

PragmaticFunction

Semantic Meaning

Month II & III

Form

Pragmatic Function

Semantic Meaning

Month IV
Form

Pragmatic Function

Semantic Meaning

FIG. 9.1. Megan.

of a variety of control acts that linked up with different self reference forms.
For instance, she used J in requests for permission: Can]... 2% Me occurred
in requests for assistance: Could help me? and Mine was now used in
attempts to claim goods. Thus, Megan’s solution to the problem of turning
a unifunctional system into a plurifunctional system involved focusing on a
particular form—function pair (and avoiding the others) and then suddenly
making use of multiple resources.
A second solution was offered by Jeffrey. As is revealed in the analyses
in chapter 5, Jeffrey began with a fairly neat unifunctional system. For Jeffrey,
the process by which J became multifunctional was slow and protracted.
Likewise, the deconstruction of the special use of My was very subtle and
slow. Essentially what we see is Jeffrey trying to find a new functional
contrast between J and My. As is shown in Fig. 9.2, by Month II Jeffrey
began using J with action verbs but now contrastively employed J and My,
depending on the motive for the action. That is, if the motive for the action
was personal desire, Jeffrey employed My; and if the motive was an inter-
personal or joint agreement, he switched to J. By Month III Jeffrey seemed
to view My more similar to modal forms; here we find him uttering sequences
180 CHAPTER 9

Month I

Form

Pragmatic Function

Semantic Meaning

Month II
Form

Pragmatic Function

Semantic Meaning

Motives

Month III
(Identical to Month II except WE replaces the
functions of MY and MY used as quasi-modal)

Month IV
Form

Pragmatic Function

Semantic Meaning

Motives

FIG. 9.2. Jeffrey.

such as I m’ant to build a tower. During Month III, Jeffrey also included
new pronominal contrasts in the system, replacing the /My distinction from
Month II with an We distinction. By Month IV, the distinction between
joint and personal motivations was marked with modal forms, and J, My,
and We all became multifunctional. Thus Jeffrey went through an extremely
protracted period of (a) adding new self reference forms, (b) reorganizing
the functional clusters associated with given forms, and (c) adding new
forms into the system.
I do not pursue a detailed analysis of the transformations of Grice’s system
of self reference. What is relevant here is that it integrated aspects of both
of the solutions used by Jeffrey and Megan. Like the other children, Grice
began with a fairly neat unifunctional system; and like the others, by Month
IV J was employed multifunctionally. Along the way Grice also added a
number of new resources to express previous functions. For instance, al-
though early on Grice used Me, often in subject position, to mark the notion
ORGANIZATION OF LINGUISTIC SYSTEMS 181

of self as recipient (i.e. Me have airplane), by the end of the study she said
things like J want the airplane for me. As with Megan, the fading of My as
a marker of control was sudden, and its disappearance was marked by the
addition of a variety of request and control forms. But as with Jeffrey, J
became multifunctional in a slow and protracted way. First Grice distin-
guished between J and My in terms of two broad pragmatic categories.
During the subsequent months, Grice slowly changed this until J became
multifunctional.
Given the variety of ways the three children moved from unifunctional
systems to plurifunctional ones, the question remains whether one can gen-
eralize about the transformational process. The answer given here is affirm-
ative. Regardless of the particulars, all three children started with relatively
global, contextually restricted uses of forms and moved toward solutions
involving hierarchical integration and differentiation. Not only did the chil-
dren enter a wider range of communicative situations and need to express
an increasingly varied set of communicative functions, but they also drew
upon a larger array of linguistic resources. This sort of means-ends relation-
ship was introduced into the psycholinguistic literature by Slobin (1973):
“New forms first express old functions, and new functions are first expressed
by old forms” (p. 184). It was more generally related to developmental
changes across domains by Werner (1957) and Werner and Kaplan (1984)
as the orthogenetic principle.
Before moving on in the next section of this chapter to ask why the
children traveled the particular paths that they traveled en route to the adult
system, I should emphasize how the approach to development offered here
can be distinguished from other approaches. As was pointed out in chapter
1, many previous approaches to development in the child language literature
have been concerned with ordering relations between given forms. The
developmental-functionalist approach adopted here has been less concerned
with the issue of which forms are acquired in which sequence and more
concerned with the nature of the form—function pairings throughout onto-
genesis. ;
To this extent, the approach adopted here comes closer to that put forth
by Karmiloff-Smith (1979), who was also interested in the developmental
transformations between individual forms and functions across development.
The major distinction between Karmiloff-Smith’s approach and this one is
that I view development in light of the orthogenetic principle. Like Karmiloff-
Smith, I have noted ways in which the developmental trajectory of an indi-
vidual subsystem of forms moves from unifunctional to plurifunctional status.
In addition, though, the construct of the orthogenetic principle has also
allowed me to examine the unfolding of form—form and function-function
relationships as well. As was pointed out in the first chapter of this book,
the notion of development here is not equated with ontogenesis and is not
182 CHAPTER 9

viewed as a function of the data but, rather, as a construct that the researcher
brings to the empirical examination of children’s linguistic productions. Thus,
although there were many changes in the monthly productions both within
and between the children, I have found that the orthogenetic principle serves
as a tool in the discovery of underlying similarities between the individual
children.

ACCOUNTING FOR THE ORIGINAL ORGANIZATION

OF SELF REFERENCE SYSTEMS

Thus far, I have outlined the course of development various children go

through en route to adultlike usage of self reference forms. This discussion,
though, has yet to grapple with the very important question of why the
children organize the systems as they do and the related question of what
factors might lead children to give up their original systems and push them
to reorganize the systems as they do. Undoubtably, multiple factors influence
both the organization and the reorganization of the self reference systems.
In this section and the one that follows, I review various proposals that have
been offered in the literature, considering such theories in light of the findings
reviewed in Part II of this book. I begin the discussion in this section by
asking what factors influence the children’s original organizations of forms
and functions, and I turn in the next section to a discussion of factors
influencing subsequent reorganizations.

Maturational Accounts

As was noted in chapter 4, Radford (1990) suggested that children’s early

case errors can be attributed to their being at a lexical-thematic stage of
development. The one significant discrepancy between Radford’s account
and the data of the three ego-anchored children concerns the use of nomi-
native forms. As I noted, Radford predicted some sporadic use of nominative
pronouns at this stage. The difference, though, is that he argued that such
usage is nonproductive. There was little evidence that the ego-anchored
children’s frequent use of the nominative form J, was as part of a set phrase.
Thus it is unclear how a maturational account such as Radford’s would
explain the original organization of the ego-anchored children’s systems.

Cognitive Accounts

My discussion of cognitive factors that might play a role in the children’s

organization of self reference systems involves two perspectives. First, I
consider the role of general cognitive strategies and second, I compare my
ORGANIZATION OF LINGUISTIC SYSTEMS 183

findings to Slobin’s observations of universal preferences for grammaticizing

particular distinctions.

Cognitive Strategies. Undoubtably, the child brings to the task of con-

structing a grammar a number of general cognitive processing skills. Slobin
(1971, 1973, 1982, 1985) referred to such processing strategies as Operating
Principles—principles that guide the child’s construction of grammar. I do
not discuss the particular principles themselves but, rather, suggest that such
principles are fundamental prerequisites for the construction of linguistic
systems.
In considering the relationship between cognitive strategies and the ego-
anchored children’s construction of self reference systems, I draw upon
related research by Tanz (1974). In attempting to account for the pattern of
pronominal case errors found in data she reviewed of English-speaking
children’s early productions, Tanz appealed to two of Slobin’s Operating
Principles: (a) avoid exceptions, and (b) pay attention to the ends of units.!
Tanz used the notion of cognitive strategies to explain patterns similar to
the ones created by the ego-anchored children. The specifics of her argument
fail to account for the actual patterns reported in Part II and highlight the
need to go beyond a simple matching between an Operating Principle and
what she referred to as case-marking errors. The basic pattern that Tanz
attempted to account for involved the finding that English-speaking children
tend to go through a phase of using the accusative pronouns in place of
nominative ones, though the same children do not use the nominative forms
where an accusative pronoun would be expected. This results in errors of
the following sort: Me a good boy and Me do it.
According to Tanz, the child selects the accusative form as basic because
it appears in a greater number of syntactic contexts than the nominative
form; and abiding by the Operating Principle avoid exceptions, the child
adopts the basic form even in subject position. Tanz also noted that because
the accusative form tends to occur at the end of constituents, whereas the
nominative form occurs at the beginning of constituents, the child could be
drawing upon the strategy Pay attention to the ends of units, thereby se-
lecting the accusative form as basic.
Tanz claimed that regularities in children’s use of pronominal case markers
can be accounted for by appealing to the notion of cognitive strategies.
Although I am in general agreement with Tanz’s proposal, there is an im-
portant incompatibility between Tanz’s account and the data reported in
Part II of this book. Tanz viewed the issue as one of children replacing

‘Although Slobin has elaborated on the Operating Principles since Tanz developed her
argument, I do not discuss such changes here because they are not directly relevant to the
issues at hand (see Slobin, 1985, for a more recent discussion of his notion of Operating
Principles).
184 CHAPTER 9

nominative forms with accusative ones. The data presented here suggest
that the children in this study used both nominative and accusative forms
side by side. Because Tanz did not report distributional frequencies, I cannot
assess whether this was the case for the children she examined. Given that
the children in the present study made use of multiple forms in subject
position, her arguments cannot be accepted. Likewise, paying attention to
the ends of units cannot be drawn in to explain the frequent use of forms
like My in subject position, because this form does not appear at the ends
of units.
I do not propose general strategies that might better account for the
findings but, rather, suggest that a satisfactory account for the original or-
ganization of pronominal systems will follow assessment of the ways children
draw upon such principles for dealing with the interpretation, storage, and
organization of linguistic material in combination with other motivating fac-
tors when constructing self reference systems. I now discuss some of the
other potential motivating factors that could be working in combination with
the Operating Principles discussed in this section.

Cognitive Predispositions. Another way of accounting for the sort of

self reference systems constructed by the children would be to appeal to
the notion of universal cognitive preferences. Based on evidence stemming
from more than 15 languages, Slobin (1985) drew the conclusion that chil-
dren across the world build similar early grammars. Before grammatical
forms are employed in ways that conform with the language being acquired,
such forms are first mapped onto a universal set of basic notions. Gram-
matical markers are first used in conjunction with particular Scenes (Fillmore,
1977) or activity-types. The claim is that the initial meaning categories re-
ceiving grammatical treatment by children across the world are minimally
influenced by the structure of the specific input language and, rather, reflect
universal ways of structuring conceptual material.
Slobin suggested that one of the first scenes to receive grammatical treat-
ment by children acquiring structurally different languages is the manipu-
lative activity scene. As was noted in chapter 1, this scene refers to a basic
event in which a prototypical agent brings about a change of state in a
patient. I have noted the extent to which children acquiring accusative and
ergative languages at first similarly extend the use of accusative and ergative
inflections to mark the manipulative activity scene. Striking parallels can be
found between such usage and the three ego-anchored children’s usage of
personal pronouns. The children’s use of My in subject position has been
suggested to mark a notion of prototypical agency, just as the Kaluli and
Russian children underextended the use of ergative and accusative inflec-
tions. The use of My is particularly interesting to the extent that the children
have gone beyond the input to grammaticize a particular notion. It seems
ORGANIZATION OF LINGUISTIC SYSTEMS 185

that children acquiring languages that are strikingly different have all drawn
upon something like the manipulative activity scene when first using specific
linguistic forms in the earliest phases of grammatical development. The
various findings provide striking evidence for Slobin’s claims that in the
earliest phases of grammatical development children bring with them uni-
versal preferences to construct grammar in particular ways.
As striking as such evidence is, we nevertheless should keep in mind some
important differences in the findings reported for the various language groups.
Two main distinctions are noted here. One central difference between the data
reported by Slobin for inflectional languages and that reviewed here is that
the American children’s marking of the manipulative activity scene has been
noted to be limited to self reference. Although the Kaluli and Russian children
were noted to mark the notion of prototypical agent, the American children
simply marked self as prototypical agent. Another difference between the
ego-anchored children and those studied by Slobin concerns the number of
distinctions marked by the various children. The Kaluli and Russian children
seemed to draw a contrast between prototypical agency and some general
form of nonprototypical agency. Yet, in the case studies outlined in Part II, it
was noted that the American children employed a range of forms to mark more
particular kinds of deviations from the prototypical agency scene.
These differences are important because they point to the need for further
research. Analyses would need to be carried out in order to determine the
extent to which children acquiring other languages also limit their marking
of the manipulative activity scene to self, and to determine the number of
distinctions marked regarding prototypical and nonprototypical agency. This
is to suggest that the differences noted at present may have more to do with
the ways the researchers have approached their data than with differences
in the children themselves.
In my discussion of universal cognitive predispositions, I have suggested
that children bring to the task of constructing a language a collection of
basic notions that they attempt to distinguish at the linguistic level. In for-
mulating his position on Basic Child Grammar, Slobin (1985) suggested that
it is the basic notions receiving early grammatical attention that are universal,
claiming that the particular forms employed will vary depending upon the
input language. With regard to the marking of agentivity and control, as I
noted in chapter 2, other researchers have noticed a certain amount of
overlap in the types of devices used by speakers of the languages of the
world when marking prototypical agency and various deviations. This sug-
gests that children may also be constrained by certain universal tendencies
in the sort of forms they select to mark particular basic notions.
With regard to the present findings, we can speculate that the ego-
anchored children’s contrastive use of first person pronominal forms was
influenced by the tendency in many languages to link the expression of
186 CHAPTER 9

degree of control with the use of case-marking contrasts. This suggests that
cross-linguistically, one might find that when children create particular dis-
tinctions not found in the target language, there are some general patterns
in the range of forms selected by children to do so. In the future it would
be interesting to assess the extent to which universal preferences might play
a role in recruiting particular sorts of formal devices in order to mark lin-
guistically particular scenes or basic notions. Likely choices might include
markers of enhanced transitivity, as well as linguistic devices that express
the speaker’s perspective on the utterance.

Language as a Problem-Solving Space. Karmiloff-Smith (1979, 1983,

1986) suggested that children treat language as a formal problem-solving
space. Central to Karmiloff-Smith’s position is the issue of what leads the
children to go beyond successful input. Based on several empirical studies,
Karmiloff-Smith outlined a three-phase sequence of development. Because
the framework proposed by Karmiloff-Smith contributes to an understanding
of both the ego-anchored children’s organization and their subsequent re-
organization of self reference forms, I review all three phases and then turn
to a more complete discussion of Phase IJ, which provides a partial inter-
pretation for the original organization of the ego-anchored children’s self
reference systems (see chapter 1 for further discussion).
Phase I is characterized as a time when the child’s productions are guided
by external stimuli. These utterances look like those produced by adult
speakers of the language community, though Karmiloff-Smith suggested that
the child lacks the ability to generalize across form-function pairs. The forms
can be thought of as unifunctional homonyms. Thus the representations are
stored separately. During Phase II the child works on gaining control of the
linguistic representations, organizing them into a kind of subsystem. It is at
this phase that the internal representations become the focus of the child’s
attention. Now the child goes beyond linguistic input, often creating distinct
markers for forms that are plurifunctional. The claim is that “the externalization
of such surface marking enables the child to keep distinct the functional
relationships attached to a common phonological form” (Karmiloff-Smith,
1985, p. 35). At this second phase the child not only draws connections
between the various uses of a single form but also begins forming systems of
reference.
Finally at Phase III, the child begins to focus again on the sort of input
received. Distinctions created by the child but not found in the target lan-
guage fade out, and the child’s speech at the behavioral level looks quite
similar to that produced during Phase I, though during Phase III the forms
are explicitly related and represented as a subsystem.
Relating Karmiloff-Smith’s three-phase model to the data from the ego-
anchored children, one can suggest that these three children were function-
ORGANIZATION OF LINGUISTIC SYSTEMS 187

ing at Phase II at the onset of the study. They had broken apart the range
of meanings and functions associated with the self reference form J into
several separate clusters, assigning one pronominal form to each of these
clusters. We can interpret the ego-anchored children’s nonadultlike usage
of self reference forms as part of the children’s attempts to render tangible
the distinct functions common to the nominative form. This in turn not only
allows the child to establish links between a form and its various functions
but also represents a first step in the child’s linking of various forms into a
subsystem. The ego-anchored children’s organization of a self reference
system can be viewed as part of an ongoing process in which the child
attempts to come to grips with language as a system of interconnected units.

Environmental Accounts

Two sorts of accounts that rely heavily on the environment in which the
child is raised are now reviewed as potential accounts for why children
might organize self reference forms as they do. I first discuss the impact of
the input children receive. To what extent were the ego-anchored children’s
systems a by-product of the input they heard around them? Next I consider
potential evidence that the children were sensitive to typological aspects of
the language being acquired.
It is noncontroversial to suggest that the children’s construction of pro-
nominal systems was partially based on the input they received. I now
consider how children’s special use of self reference forms might be ac-
counted for by regularities in linguistic input they receive. The specific issue
of self reference becomes all the more complex to assess, because the child
must grapple with the issue of shifting reference. Even if adult speech di-
rected to these children were highly repetitive, grammatical, easy to process,
and closely tied to the ongoing activity frame, it remains unclear how such
input might have led the child to (a) untangle the issue of shifting reference,
and (b) relate self reference forms with the meanings and functions noted
in Part II of this book.

Input. To begin with, we should consider the sort of input concerning

self reference that these children received. Did the caregivers similarly dis-
tribute their references between self and other? What forms (nominal and
pronominal) did they use when talking with the children? Was there any
evidence that the caregivers limited the use of particular forms with specific
functions? Even if caregivers never produced utterances such as My did it,
the possibility remains that they reserved the use of the possessive pronoun
for emotionally charged acts of control, but used the nominal form Mommy
referentially.
188 CHAPTER 9

I begin my analysis of caregiver talk with a consideration of the distribu-

tion of caregivers’ reference to themselves and their child. All three caregivers
referred more to their child than to themselves. The importance of this is
that unless the caregivers relied on nominal forms, the children were more
likely to receive input involving second person reference forms than first
person forms. Examining the forms used, it was found that when referring
to their children, the caregivers tended to employ pronominal forms (pri-
marily you). Use of the children’s names was rare (see also Nelson, 1989).
With regard to self reference, caregivers also relied heavily on pronominal
forms. J was used most frequently, and My and Mommy were rarely used
by any of the caregivers. Such findings reflect interesting differences between
the caregivers’ input and children’s productions. For instance, although Jeff-
rey used My in almost half of his references to self in the first session, his
mother never used this form in the same play session. Clearly the distribu-
tional patterns noted in Part II for the ego-anchored children did not match
those found in their caregivers’ data.
We can also consider the relationship between the forms that the caregivers
used and the functions they served, examining the extent to which the
children’s patterning reflected patterning in the speech used by caregivers. An
analysis examining form, meaning, and function clusters in the ego-anchored
children’s caregivers’ speech to their children did not reveal any evidence that
the caregivers’ use of self reference forms was limited to the particular activity-
types noted for their children. For instance, all the mothers used J in clauses
ranking high and low in agentivity, although their children were noted to
restrict the use of / to clauses ranking low. The functional analysis reveals that
the caregivers rarely used any of the self reference forms in conjunction with
control acts. When aself reference form appeared in conjunction witha control
act, then this tended to be J in requests, such as Can I have a piece of cake?
or Can I have some tea please? in the context of pretend play. In addition, the
caregivers rarely used My in ownership claims or other acts of control. Rather,
the use of My often functioned to specify information so the child could locate
or identify a particular referent in conversations where control was notat issue.
The various analyses concerning caregivers’ input reveal that the children
were not merely copying the forms they heard in the input around them.
Nevertheless, much remains unclear about the specific role that the caregivers’
input played in the children’s formulations. It is interesting that although none
of the caregivers restricted the use of particular self reference formstoa specific
semantic and pragmatic cluster, it has been noted that all of the caregivers
tended to use /in a functional cluster similar to that used by their children (see
Budwig, in press). It also can be noted that the caregivers tended to make use
of expansions and partial repetitions in dialogues in which the children
referred to themselves, and it could be that caregiver input played a role in the
children’s attempts to deal with shifting reference. It would be important in
ORGANIZATION OF LINGUISTIC SYSTEMS 189

the future to examine more closely the extent to which adult input provides
the child with structural information about the system being acquired (see
chapter 10 for further discussion).
By attending to input, the children might not only receive information
concerning relationships between various pronominal forms but also obtain
information about the positioning of elements. Ervin (1961), in a study of
the verbal determinants of word association, noted some factors about Eng-
lish usage that can be related to the present discussion. She reported that
words that do not tend to occur in sentence-final position (e.g., nominative
pronouns, copulas, and function words) do not tend to occur alone. In
contrast, words that occur in final position (e.g., nouns, accusative pronouns,
intransitive verbs) also appear alone in answers to questions. This implies
that the child is not likely to hear words like J or My alone as responses to
questions, or in sentence-final position, though the child might hear instances
of Me or Mommy in such contexts. Such regularities in the input could lead
the children to relate the self reference forms in specific ways.
In the future it seems essential to look more carefully at the sort of
contrasts the child actually is confronted with when listening to spoken
input. In the present study, when the caregivers asked their children ques-
tions, it was noted that the form you was actually pronounced ya, as in, for
instance, What ya gonna do? The children then were actually presented
with a contrast between you and ya. Keeping in mind that the ego-anchored
children’s pronunciation of My often sounded like Ma, the question can be
raised: Did the children create the Ma form in an attempt to build a formal
correspondence with the second person form ya?
Another way the children might have arrived at a possessive construction
in subject position in a way that deviated from the target language, based
on input they received, is as follows. The caregivers often repeated what
the child had just stated that he or she was going to do; for instance, we
find You’re gonna build a house. In such instances, it could be the case
that the children interpreted the forms you are in their contracted version
as the possessive pronoun. your, which could help to explain why the ego-
anchored children came to use My in subject position alone without a noun.
Although at present these points are mere speculations, they nevertheless
remind us of the need to consider the actual speech units that the child is
confronted with when attempting to construct a self reference system.
As one final suggestion regarding the role that input may play, one should
keep in mind the range of input that particular children might receive in their
daily interactions and the role that different kinds of input play on the
children’s constructions. For instance, the children in the present study spent
a good proportion of their days interacting with similar-aged peers in a
day-care setting. At present we know very little about the extent to which peer
input influences children’s constructions of self reference systems. It could be
190 CHAPTER 9

that as children enter into the peer culture, and conflicts over control and
dominance become important issues, particular linguistic forms become
salient at this age. Along similar lines, it seems important also to consider the
input such children receive in nondyadic interactions, which are common in
the day-care setting. Given that children receive input from multiple sources,
it seems that we must begin to think about how such varied input might be
used selectively by the child in constructing a self reference system.

Typological Factors. A second way in which the environment may

impact on the sort of grammatical systems children construct is related to
particular structural properties of the language being acquired. At the level of
input, we considered the extent to which the ego-anchored children’s pattern-
ing of forms and functions related to regularities in the actual input they
received. At a slightly different level, we can also consider the extent to which
specific ways a particular language categorizes meaning play a role in the
child’s original categorization. Bowerman (1985) suggested that children are
sensitive to and make use of clues from the target language when constructing
grammar. In contrast to Slobin, who argued for a language-neutral starting
point, Bowerman contended that the children’s earliest meaning categories
associated with grammatical forms are biased by the semantic structure of the
target language. I now consider whether the ego-anchored children’s linkages
of form and meaning are related to structural properties of English.
The comparison with Slobin’s discussion of Kaluli and Russian children’s
early use of inflections pointed up both some similarities and some differ-
ences between the children. One might speculate that because the Russian
and Kaluli children made use of inflectional devices, their use of such forms
was not restricted to self. In contrast, the English-speaking children were
drawing on first person forms and therefore restricted the notion of agency
to self. Similarly, recall that the Kaluli and Russian children marked a binary
distinction between prototypical and nonprototypical agency, whereas the
distinctions the English-speaking children made were multiple (i.e., proto-
typical agent, experiencer, affected agent, etc.). One suggestion is that al-
though all children may attempt to mark some general distinction involving
the manipulative activity scene, the actual number of distinctions made may
depend on language-particular factors. These comments are necessarily
speculative. In order to tease apart the relative influence of structural prop-
erties of particular languages, future research must carefully select languages
that contrast in ways that would allow one to address this question.

Discussion

Much of my discussion above has argued for the active role of the child in
constructing grammar. I have suggested that children bring with them to the
task of constructing a self reference system a set of cognitive strategies that
ORGANIZATION OF LINGUISTIC SYSTEMS 191

assists the children in interpreting input and subsequently organizing forms

into grammatical systems. The evidence reviewed here suggests that agentivity
and control are notions that will play a central organizing role in children’s
construction of grammar across the world. Children acquiring English, as well
as children acquiring structurally different languages, will attempt to mark
distinctions between what I have referred to as the prototypical agentivity
scene and particular deviations. The question remains whether all English-
speaking children will select self reference forms as a means of marking such
a distinction. It could be that children mark similar notions with other forms.
I have argued that given a range of forms offered by a particular language,
some subsystems will be more likely candidates for the marking of agentivity
and control than others. I have suggested that given that the English-speaking
children in this study were not regularly referring to others, they borrowed
first person forms for the purposes of marking agentivity and control. Whether
other English-speaking children might adopt other forms is a topic for future
investigation.
It is important to stress that according to the present account the child is
guided by preferences to mark quite generally the notions of agentivity and
control. The actual number of distinctions drawn by children and the particular
devices employed are dependent on the child’s interpretation of structural
properties of the input received. The specific number of distinctions marked
will vary according to typological and interactional factors. Thus environ-
mental factors and cognitive factors all play a role, suggesting that a certain
amount of interlanguage and intralanguage variation is to be expected.”
The framework outlined thus far could be interpreted to imply that the
children were attempting to construct grammatical systems without regard
to the various functions language serves in the child’s interactions with
others. It seems necessary to consider also the role such factors might play
in motivating the sorts of contrasts constructed by the children. Previous
research reviewed in Part I of this book has highlighted the extent to which
children contrastively employ sound patterns and intonation to mark various
language functions in their speech before they combine words (see Carter,
1975; Halliday, 1975; Painter, 1984). The children may attempt to mark
function similarly when contrastively employing multiword utterances.
This leaves the question why the children choose the self reference system
to mark such a distinction. Silverstein (1985) claimed that there are networks
of functional regularities across languages that are located in equivalent forms.
The child may not acquire the semantic values until a later point, after first
using the forms only pragmatically. Such a suggestion fits the developmental
trends reported to the extent that two of the ego-anchored children first
organized the use of the self reference forms primarily around a pragmatic

“For alternative ways to employ self reference forms in English see Gerhardt, 1988, and
Nelson, 1989.
192 CHAPTER 9

contrast. Future research should collect data from children who have yet to
produce word combinations in order to draw conclusions of this sort.
A second factor that has not received much treatment in the literature
concerning organizational processes concerns the extent to which the child’s
growing notions of agentivity and control interact with other factors in lead-
ing to various organizations of the self reference forms. Children begin
working on self reference systems with different conceptions of agentivity
and control; and such distinctions may lead to variations in the ways form,
meaning, and pragmatic function are related. It would be fruitful as well to
collect data from children who have had qualitatively different experiences
with agency and control, because such experiences might lead them to very
different beliefs about such notions. As just one example, one could examine
language data from children who have had little experience manipulating
objects or moving as independent agents due to lack of full use of limbs.
In sum, I have presented only some very general ideas about what led the
ego-anchored children in this study to organize self reference forms as they
did. Any adequate account is going to need to consider mutual influences of
multiple factors. The tentative framework put forth here has been developed
in light of findings from the examination of a few children’s organization of a
specific linguistic subsystem. In the future, it is essential that more children be
studied and that the developmental unfolding of other linguistic systems be
carefully investigated. I discuss this further in the final chapter.

ACCOUNTING FOR THE REORGANIZATION

OF SELF REFERENCE SYSTEMS

Given the highly systematic nature of the ego-anchored children’s use of

self reference forms at the start of the study, the question arises: What led
the children to give up these systems in favor of a system that conformed
more to that found in the target language? I have claimed that the children
moved developmentally from a phase of using multiple self reference forms,
each linked with a distinct meaning and functional cluster, to one of using
the forms multifunctionally as adults do. I turn now to various accounts that
could shed light on possible motivations for such changes. I begin my
discussion with a consideration of Radford’s (1990) maturational hypothesis.
Next I discuss cognitive factors, and finally I consider accounts that empha-
size environmental influences. I focus exclusively on approaches that attempt
to offer a specific solution to the problem at hand.

Maturational Accounts

Radford (1990) claimed that the transition from lack of productive use of
nominative case marking to adultlike usage can be accounted for by the
maturational hypothesis. According to his account, children begin at a lexi-
ORGANIZATION OF LINGUISTIC SYSTEMS 193

cal-thematic stage and subsequently progress to a functional-nonthematic

stage. Although there is a certain amount of overlap in the examples of child
speech offered here and by Radford, there are some important differences
between the two data sets. Most significantly, it is unclear how the ego-
anchored children’s protracted development can be explained as a two-stage
biologically determined process. It should also be noted that the ego-
anchored children’s use of accusative forms in nominative position was first
restricted to first person and later appeared with third person forms at a
time when first person nominative forms were frequent.

Cognitive Accounts

In this section I examine Karmiloff-Smith’s approach, also reviewed earlier,

because it is the most specific attempt within cognitive approaches to account
for the kind of protracted development and reorganization of self reference
forms described in Part II of this book. I have noted that the ego-anchored
children’s organization of self reference forms at the onset of the study could
be related to Phase II of Karmiloff-Smith’s three-phase account of children’s
language development. One interpretation of the children’s use of the various
self reference forms early on in the study is that the children were attempting
to gain control over the system by rendering each of the functions of a given
form tangible. One can now suggest that when the ego-anchored children
began to reorganize their self reference systems to conform with the target
language, they had successfully consolidated the relationship between form—
function pairs. The children no longer needed to externalize the overall
organization via the use of distinct linguistic markers. According to Karmiloff-
Smith’s model, during Phase III, the additional markers fade away and the
children start to devote more attention to the input they receive in evaluating
previous hypotheses and formulating new ones. Slowly the use of the nomi-
native form J becomes more frequent, and the cluster of semantic and
pragmatic factors associated with its use expands to include new functions
and meanings previously covered by the other self reference forms. At the
same time children are working on self reference forms as a linguistic sub-
system, they are working on other systems as well. Connections between
the various subsystems may contribute to the consolidation process. For
instance, I have noted that the children were acquiring a variety of devices
that began to take over functions related to those served by the self reference
system. Although Karmiloff-Smith’s three-phase account focuses primarily
on one particular linguistic system, other models, such as the Competition
Model (see MacWhinney & Bates, 1989), more directly handle the relation-
ship between various linguistic systems.
It is central to Karmiloff-Smith’s account that input is used in a new way
by children during Phase III. During Phase II, children were said to be
194
CHAPTER 9

focusing on the dominant form-—function pattern in the input. During Phase

III, at a point at which the children have rendered tangible the links between
form and function, the children begin making use of input that previously
was ignored. What is central in my analysis is the claim that monitoring of
input per se did not motivate the reorganization of the ego-anchored chil-
dren’s system. Rather, I suggest that children used the input in a new way.
With regard to the transition from ego-anchored to nonego-anchored func-
tioning, I have noted that only after an extended phase of employing self
reference forms unifunctionally, did the children begin employing them
plurifunctionally. An examination of the individual differences in the chil-
dren’s decisions with regard to how the other self reference forms were to
be employed highlights the fact that input cannot be taken as the motivating
factor. Some children simply avoided the use of the other self reference
forms, allowing them to resurface a few months later after reevaluating how
they were used in the target system, whereas other children seemed more
ready to work simultaneously on the various self reference forms. That some
forms were avoided suggests that the children were not yet clear how they
fit into the system. This implies that input monitoring ‘had not instructed the
children on how to make use of such forms.
I also have noted that some forms, such as My, were tried out as part of
another subsystem—in this instance, as a modal form. This suggests that
children were not always reorganizing forms in a direction that corresponded
with input but might test out other alternatives along the way. This finding
supports Karmiloff-Smith’s claim that the children, at Phase III, are not merely
remodeling their systems in terms of input. It also highlights the need to
consider the specifics of how children build systems not only of form—func-
tion patterns but also of relationships between form-form and function—
function relationships (see also Bates & MacWhinney, 1989).

Environmental Accounts

One might want to hypothesize that the children all began changing their
use of self reference forms because something in their linguistic environment
changed. Two possible accounts for why the children might reorganize their
system are pursued here. First I consider evidence as to whether the chil-
dren’s changes related to potential changes in caregiver input. I also consider
whether children reorganized their self reference systems for reasons related
to communicative pressure.

Changes in Caregiver Input. One particular hypothesis might be that

the caregivers of the nonego-anchored children provided their children with
distinct input patterns regarding form and function of self reference forms.
Analyses of caregiver input for the nonego-anchored children (Quick, 1991),
ORGANIZATION OF LINGUISTIC SYSTEMS 195

as well as comparisons of the ego-anchored children’s caregivers across the

various months of study revealed no significant changes (Budwig, in press).
These findings, which, interestingly enough, showed that all of the caregivers
tended to restrict the usage of / to a cluster of low agentivity and assertions,
contained no evidence that the children’s patterns changed over time due
to corresponding changes in the input (see Budwig, in press, for a review
of this argument).

Communicative Pressure. One might hypothesize that the children

reorganized their self reference systems to conform better to the target
language because they wanted to be more effective communicators. In fact,
communicative pressure fares poorly as a motivating factor. It is generally
assumed to play, at best, a minor role in the reorganizational process (see
Bamberg, 1987; Bowerman, 1982; Karmiloff-Smith, 1979, 1983). Although the
ego-anchored children’s systems were organized differently from those of the
target language, much of the output was grammatical, and communicative
partners did not have difficulty making sense of the children’s utterances. Even
when the children produced utterances that were not grammatical, for instance
substituting My or Me for J in subject position, such usage rarely led to
communicative breakdowns. As has been previously noted, many caregivers
were unaware of such usage in their children’s speech.
Although communicative pressure per se cannot be viewed as a sufficient
motivating factor, it is possible that children are led to reorganize their
systems in part by other factors related to the communicative context. As
children’s communicative contexts change, it seems likely that the kinds of
grammatical systems may also become reorganized. One sort of change
seems relevant to the present discussion. As the children begin to become
more independent and verbal, they gradually move into a wider range of
social situations. They begin interacting with a number of others, rather than
primarily with caregivers who are familiar with their needs. Givén (1979,
1985) has suggested that as communicators move beyond communicating
with intimates to communicating with strangers they create a more elaborate
grammatical system. This ‘holds across the lifespan. With regard to the chil-
dren in the present investigation one might want to claim that as the children
began to communicate with a wider array of others, their repertoire of self
reference forms was integrated into a broader paradigm of referential devices,
including second and third person forms. In addition, as the functions of
children’s discourse developed, the children might have needed to draw on
pronominal and nominal devices to serve other functions concerning text
coherence. This might have led the children to find alternative means to
mark degree of agentivity and control. And it seems likely that with time,
the children might have recognized that a clear contrast at the linguistic
level between various language functions may become disadvantageous for
196 CHAPTER 9

a communicator needing to rely on a system with more subtle boundaries

between such acts as requests and assertions.
Ervin-Tripp (1977) convincingly argued for the need to consider the extent
to which the growth of conversation stimulates linguistic change. Although
in recent years researchers have directed more attention to the various func-
tions language serves in early childhood, and though we know a great deal
more about pragmatic aspects of child language development, the actual
relationship between language structure, language function, and develop-
mental change has continued to be virtually ignored. It seems, though, that
a complete understanding of the factors motivating children to reorganize
early systematizations of forms, meanings, and functions will ultimately come
from a closer examination of the role communicative factors play in such a
process.

Discussion

This discussion of potential factors influencing the children’s reorganization

of the self reference systems has emphasized a process-oriented view. After
an extended phase of linguistically marking various functions with distinct
forms, the child is said to gain control over the organization of the relation
between form and function to the extent that the externalization process is
no longer necessary. The child then gradually begins to encode in a single
form a cluster of related meanings and functions that were previously en-
coded in several separate forms. The process described here whereby the
children began using self reference forms plurifunctionally is more compli-
cated than that described by Karmiloff-Smith with regard to her examination
of determiners. Although the children did begin to use Jmultifunctionally,
they did not seem to use the other forms immediately in an adultlike way.
The process of reorganization does seem motivated by a web of interre-
lated factors. In Part Il, I highlighted some of the complexity involved when
discussing the use of devices other than self reference forms to mark notions
related to agentivity and control. I suggested that around the time the children
no longer employed self reference forms to mark the notion of prototypical
agency, they began marking aset of related notions including degree of
control and motivations for action with a range of forms. The children began
breaking down the notions of agentivity and control into several related
notions that were marked in different parts of the grammar. Although par-
ticular cognitive notions contribute to subsequent development, experience
with novel linguistic forms also may lead the child to new categorizations.
The direction of mapping cannot be viewed as merely unidirectional (see
also Kuczaj, 1982; Schlesinger, 1977). Finally, in considering multiple factors
influencing the child’s reorganization of the self reference system, One must
also consider the extent to which changing communicative demands and
ORGANIZATION OF LINGUISTIC SYSTEMS 197

the child’s changing notions of agentivity and control play a contributing

role in this process.
In summary, it has been suggested that the children studied began with
some core notions of what deserves grammatical treatment. Based on an
assessment of structural information about the language being acquired and
various communicative factors, they began using particular linguistic forms
to mark various distinctions concerning agentivity and control. At some later
point, as a consequence of the mutual influences of changing notions of
human action, communication, and language as a system, the children gradu-
ally reorganized the relationship between form-function pairs. The specific
mechanisms guiding the reorganization process are at this point quite vague.
It seems clear, however, that future research could examine this issue more
carefully. Details of some possibilities are discussed in the next chapter.
 CHAP TE Rete

Conclusions

In this chapter, I discuss the implications of the present study for functional
approaches to child language. I reconsider the notion of language function,
as well as the notion of development. In addition, I consider the relationship
of the perspective adopted in this book to central issues in the study of
child grammar. In the concluding section, I discuss directions for future
research.

IMPLICATIONS OF THE PRESENT STUDY FOR

FUNCTIONAL APPROACHES TO CHILD LANGUAGE

The Notion of Function

This study illustrates the extent to which children organize linguistic forms
around a broad cluster of notions having to do with meaning and pragmatic
function. Where does this leave us with regard to current theorizing about
language function? In this section I return to three topics that came up in the
review of the literature on language function in chapter 1, and I relate the
current findings to these topics. The three issues to be pursued here include:
(a) the nature of linguistic categories, (b) the role of semantics and pragmatic
notions in functionalist accounts, and (c) the notion of activity-types.

The Nature of Linguistic Categories. As was noted in chapter 1, it

has become increasingly common for researchers studying children’s early
grammatical development ‘to suggest that semantic-cognitive correspond-

198
CONCLUSIONS 199

ences play important roles in the early phases of language development.

Although the theories differ, one central assumption has been that categories
such as agent represent discrete categories for the child. Many child language
researchers whose work was reviewed in chapter 1 were noted to posit a
category such as agent-action that included examples like My open that and
Eve read. Recent discussions in the linguistics literature have challenged
such a claim, suggesting that the notion of agentivity is actually composed
of several interrelated components such as control, volition, and animacy
(see DeLancey, 1984; Lakoff, 1977, 1986).
The findings of the present investigation suggest that the children studied
were not treating agentivity as a discrete category. They did not always treat
subjects of verbs like open and read alike. Utterances including references
to the self with open were usually encoded as involving the prototypical
agentivity scene, whereas utterances involving read only occasionally were
encoded in a similar way. Furthermore, the children’s categorization of verb
types did not accord with any theory of verb semantics outlined thus far.
The children were said to be working with slippery categories that involved
a combination of notions of agency and control, using fairly subtle distinc-
tions that cut across semantic and pragmatic levels of analysis. These findings
cali into question the usefulness of any attempts to work with discrete
categories, especially those based on current linguistic theory, in helping us
understand children’s construction of early grammar.

Semantics Versus Pragmatics. Previous literature has often separated

these levels of analysis. In the literature on early child grammar, there has been
a tendency to emphasize the extent to which children are concerned with
mapping the relation between linguistic forms and some set of underlying
conceptual meanings. The emphasis on the relationship between form and
meaning can be related to the prevalent assumption in the child language
literature that language encodes reality. Although such a view is clearly not
false, it underestimates the extent to which language is a tool that helps the
child constitute reality. Just as children draw linguistic distinctions between
various meanings, it also makes sense that they would attempt to organize the
use of particular linguistic forms around their budding notions of language
function. It should be noted that researchers have found that children are
capable of working on language at a purely formal level without relating form
and meaning. In addition, it seems that function will not play an organizing
role for the acquisition of all linguistic forms. The question remains, neverthe-
less: To what extent do children draw upon their budding notions of function
in constructing a language system?

Form, Meaning, and Function in Early Child Language. One of the

central questions regarding early grammatical development has involved the
nature of children’s early linguistic categories. I have suggested here that one
200 CHAPTER 10

way to approach this issue is by starting with the level of form. By examining
the children’s contrastive use of self reference forms and considering the kinds
of categories they used, I have found that the child relies on rather broad
clusters of properties. With regard to the marking of agentivity and control in
early child grammar, I have found it fruitful to rely on the notion of
activity-types-clusters of properties that cut across both the semantic and
pragmatic levels in ways that pattern into linguistic gestalts (see Lakoff &
Johnson, 1980). Such clusters are made up of experiential properties and are
characterizable in terms of the ways children interact with their environment.
The categories have prototype effects; no one property is essential, but the
category is centered instead around a prototype defined in both semantic and
pragmatic terms. At various points in development some of the properties of
the activity-type may be more central than others. Although changes can be
noted across development in the configuration of these patternings, at all times
the children’s categorizations consist of a bundle of properties that cluster
together in regular ways.
Thus the present work suggests not only that researchers should attend
to both semantic and pragmatic factors when studying early grammatical
development but also that the sorts of relationships between form and func-
tion will not necessarily pattern in adultlike ways. We need to find ways of
arriving at the child’s construction of such clusters. I have suggested that in
the future, rather than starting with broad categories of semantic meaning
or pragmatic function, it would be advantageous to begin one’s functional
analysis with form contrasts.

The Notion of Development

In the preceding chapters, I have outlined a specific notion of development

that has been used in this research. In Part I it was noted that much of the
functionalist work that has been carried out has focused primarily on some
ways in which children’s patterning of linguistic forms differs from adultlike
usage. Very little explicit discussion can be found in much of the functionalist
work on how development is defined. I have attempted to illustrate explicitly
a particular stance taken toward development that involves the examination
of form—function relations over time. This view of development has become
increasingly popular over the last 15 years in much of the functionalist work,
though there is little explicit mention of this change (see Budwig, 1993, for
further illustration).
This particular construct focuses on changes in form—function pairings
and can be contrasted with a view held by both functionalists and others,
under which development in child language is perceived in terms of ordering
sequences in the development of form. Thus development, here, is studied
by examining the ways in which forms not only relate to one another over
CONCLUSIONS 201

time, but also cluster with various semantic meanings and pragmatic func-
tions. This sort of approach acknowledges that the first use of the form may
be qualitatively, as well as quantitatively, different from subsequent uses. It
switches focus away from what Brown (1973) referred to as obligatory con-
texts, toward an analysis of how forms are used at any given stage of de-
velopment. This view of protracted development is quite different from other
approaches, mainly but not exclusively nonfunctionalist, that view develop-
ment as an all-or-none process.
In addition to illustrating explicitly an alternative view of development
in terms of the protracted nature of the construction of linguistic systems, I
have attempted, in the research described here, to find a common way to
characterize the developmental patterns discussed in Part II. Although it is
far from clear what factors motivate children to organize and reorganize
their linguistic systems as they do, what does become apparent is that there
is a common pattern to the processes of the various children’s development
outlined here. Regardless of the particulars, all of the ego-anchored children
began with relatively global, contextually restricted uses of forms and slowly
moved toward solutions involving hierarchical integration and differentia-
tion. As the children entered into a wider range of communicative situations
and needed to express an increasingly varied set of communicative functions,
they drew upon an increasing array of linguistic resources. Although we are
beginning to be more explicit about the notion of development within func-
tionalist approaches, future research will also need to address explicitly the
relationship between such a view and several central issues in the study of
early child grammar. After considering the impact of a functionalist approach
on various methodological issues, I return to a brief discussion of some of
these themes.

Methodological Issues

Appealing to the notion of language function and development outlined here

affects one’s stance toward methodological issues in important ways. In this
section, I briefly discuss various methodological issues, with particular focus
on both the collection and the analysis of data. Although none of the particular
methodological points are new, they often have been overlooked in studies
of early grammatical development and therefore are made explicit here.

Data Gathering Procedures. Clearly technological advances in recent

years have greatly assisted in arriving at the findings reported here. First,
the importance of working with video when studying early child grammar
from a functionalist standpoint cannot be stressed enough. No matter how
rich one’s contextual notes might be, they would not allow one the array
of information necessary to code the various semantic and pragmatic pa-
202 CHAPTER 10

rameters used in this study. Only with the aid of video data can one begin
to arrive at the broad range of properties that co-occur with the use of
particular forms. In the present study, working with high-quality wireless
microphones also assisted in picking up subtle contrasts in the child’s se-
lection of form. Without such equipment it would have been quite difficult
to note subtle differences between utterances such as My do it and IJ do it
in the children’s speech.
In addition, the present study has highlighted the advantages of examining
grammatical development within the context of a variety of communicative
partners. As Ervin-Tripp (1977) argued, the range of forms used by speakers
is greatly influenced by the sorts of interaction settings sampled. With regard
to the present study, it has been noted that particular forms were more likely
to occur in interactions taking place with specific partners. Struggles over
control of goods and participation in activities were likely to occur in inter-
actions with peers. It is of interest that very few of the data sets drawn by
researchers from computerized data banks such as CHILDES (under the
direction of Snow and MacWhinney; see MacWhinney, 1991) included sam-
ples involving agemates. Peer data are important not only because they
highlight ways children relate form, meaning, and function that might not
be found when examining data stemming from adult-child interactions, but
also because one can begin to make use of findings concerning different
distributional patterns of particular forms in the distinct settings as a clue to
how the children actually organize such forms. To this extent, more attention
ought to be given in the future to aspects such as interactive partner and
setting when collecting data.
One last point concerning data collection concerns the use of naturalistic
data. In contrast to many previous naturalistic studies, this study was carefully
designed in ways that allowed for the collection of particular kinds of talk.
Hypotheses developed on the basis of pilot data, permitted the formulation
of some expectations concerning the elicitation of the range of devices of
interest in a relatively natural and quick manner. In the present study, given
the focus on the linguistic marking of agentivity and control, it seemed
important to set up the situation in such a way that children were bound
to run into struggles over control of objects. Likewise, children could be
given objects that could be manipulated only by adults, thus eliciting talk
about mediated agency. The general point is that a certain amount of struc-
turing of the environment can enhance the nature of the data base. Although
none of the mothers interviewed at the end of the project had tuned into
the specific focus on agentivity and control, the structuring nevertheless
provided a rich data base with hundreds of examples of form contrasts that
were of interest. Future research within a functionalist approach should
continue to find ways to collect naturalistic data, and to arrange the envi-
ronment so as to enhance the data set.
CONCLUSIONS 203

Data Analysis. One central point relevant to data analysis concerns the
importance of distributional analyses. By beginning a functional analysis
with a series of distributional analyses, one is in a better position to assess
the relationship between form, meaning, and function. Such a procedure
allows us to examine a range of contrasts with which the child is working.
I have also noted that distributional analyses based on forms alone rep-
resent only a first step in approaching the data. The task becomes one of
working out a way to relate distributional patterns of forms with analyses
that take into account a broad range of semantic and pragmatic parameters.
Two directions of analysis have been proposed here. First, one must begin
with some preconceived notions concerning general distinctions the child
might be attempting to mark. To this end one can draw upon basic notions
noted to play a role in the acquisition of related forms in other languages
(see Slobin, 1985). In addition, one can make use of findings concerning
ways particular categories are marked in other languages.
At the same time that one works with certain expectations, one must also
keep an eye open for patterns that may be unique to the data at hand. Thus
original categories are refined in light of preliminary analyses. In this study
much of the refinement of the coding scheme took place during the pilot
stages when working with data stemming from other children. Nevertheless,
it was necessary in some instances to break apart particular levels of the
coding scheme or to collapse across categories in order to account for the
regularities at hand.
Finally, the importance of coding across several linguistic, as well as
nonlinguistic, layers of analysis must also be stressed. Only by considering
an array of verbal and nonverbal characteristics can one arrive at the sorts
of categories that children might be using. And I have shown the importance
of analyzing individual children’s data before collapsing across children,
because many interesting patterns and differences are washed out in grouped
analyses. Recent technological advancements in computer-aided analysis
such as CLAN have made such an endeavor more possible in recent years
(see MacWhinney, 1991),

RELATIONSHIP TO ISSUES CENTRAL

TO THE STUDY OF CHILD GRAMMAR

Three issues have been central to most previous discussions of the acquisition
of early grammar, and it seems important to consider such issues in light of
the framework presented here. I begin by considering whether there is
evidence for underlying syntactic relations in early child language. Next I
discuss the issue of continuity and discontinuity in early language develop-
ment. A third theme concerns the issue of the psychological reality of chil-
dren’s early grammatical categories.
204 CHAPTER 10

Syntactic Relations in Early Child Language

The extent to which children should be credited with knowledge of under-

lying syntactic relationships has been an issue receiving much attention in
the literature concerning children’s first word combinations (see Bloom et
al., 1975; Bowerman, 1975; Braine, 1976; Gleitman & Wanner, 1982; Radford,
1990). With regard to the present study, can we say that the ego-anchored
children had the notion of a sentence subject in mind, or were they working
at a different level of abstraction? Bloom et al. (1975) argued that the children
in their study used the same words in a particular sentence position to
convey a variety of semantic roles (e.g., agent, possessor). Based on such
findings, Bloom et al. credited the children with more abstract knowledge
about superordinate categories such as subject. Such a claim has not gone
unchallenged (see Bowerman, 1975; Braine, 1976). Bowerman, for example,
suggested that identity of word order patterns is not necessarily sufficient
evidence for the presence of syntactic functions.
Even if word order were to be taken as adequate evidence, the findings
from the present study lead us to question whether the children examined
by Bloom et al. actually were using similar words to encode a variety of
semantic relationships. Although utterances such as My open that were ar-
gued to express the relation Agent-Action and My teapot would have been
categorized as Possessor—Possessed, the findings of the present study indicate
that these children may have been expressing a single notion related to
control in both instances. This points up the methodological problem of
knowing whether the children were in fact using similar words to express
different semantic relations. The findings of the present study lend support
to the conclusion put forward by Braine (1976), who found no evidence
for such abstract categories. Braine suggested that children are working with
more limited notions. Evidence that this is the case can also be found in
the children’s marking of agentivity and control in this study. I have noted
that the children linguistically categorized self as agent differently from other
as agent in the early phases of development. Thus in utterances like J like
vanilla, My open that, Adam read, and Mommy eat it is not clear that the
ego-anchored children were expressing the same underlying relation. The
present study offers no evidence that the children were working with fairly
abstract syntactic relations in the early phases of grammatical development.
This position is also consistent with formal approaches to early child grammar
(see, for instance, Radford, 1990).

Continuity and Discontinuity in Early Child Language

Given that there is no firm evidence that the ego-anchored children were
working with syntactic categories, we are led to the issue central to much
of the current literature in early child grammar, concerning continuity and
CONCLUSIONS 205

discontinuity. If we start with the assumption that the categories in early

child speech are organized differently from those that figure in adult gram-
mars, we must account for how the child breaks into the adult system (see
Atkinson, 1982, for further discussion). The position outlined earlier with
regard to development has been viewed in light of the orthogenetic principle,
outlined in Werner and Kaplan (1984). Accordingly, the position adopted
here is similar to Werner and Kaplan’s concerning developmental transfor-
mations, namely, that both continuity and discontinuity are necessarily en-
tailed. Although the introduction of novel forms and functions into the system
obviously implies discontinuity, the manner in which these additions enter
the system is protracted. I have noted the gradual way in which old is
replaced by new, and this fits well with a continuity position (see also Lock,
1978, 1980).
Furthermore, I have noted that at times neither forms nor functions are
added in some novel way; instead, the relationship between forms and
functions is reorganized in terms of hierarchical organization and increasing
differentiation. This sort of developmental sequence can be compared to
that proposed in recent discussions of semantic bootstrapping (see, for in-
stance, Pinker, 1989), where it is argued that the syntactic representations
themselves are not developing out of semantic representations; instead, chil-
dren gradually make use of correlations between semantic and syntactic
representations.

The Psychological Reality of Children’s

Early Grammatical Categories

A third issue that has received much attention in the literature on early child
grammar concerns the psychological reality of the categories. In other words,
do the sorts of categories outlined with regard to the children’s marking of
agentivity and control actually correspond to concepts that the child is work-
ing with in constructing language? Several different proposals have been
made in the literature concerning the nature of children’s early word com-
binations. What evidence is there that those outlined for the children who
participated in the present study correspond to notions that the child may
be using? The position adopted here is based on the assumption that vari-
ations in form are not random but, rather, systematic. By placing linguistic
contrasts at the center of one’s analysis, one has a rich source to draw upon.
One can move from the level of form and attempt to account for the range
of contrasts employed.
In an attempt to arrive at the sort of categories that are functioning within
early child speech, one must do more than account fora statistically signifi-
cant proportion of the data. I have noted, for instance, that most of Megan’s
use of my could be related to a set of pragmatic features. Nevertheless, the
206 CHAPTER 10

study of exceptions pointed up further linkages the child could be drawing.

In drawing conclusions concerning the scope of children’s categories, ex-
amination of the contrastive use of forms can be instructive in pointing out
distinctions that also might otherwise have gone unnoticed. For instance, I
have noted that many examples of the use of Me by Grice resembled in-
stances involving My in that both were associated with high agentivity and
control. In digging deeper for a potential contrast, I noted that all the utter-
ances with Me expressed the more specific category of affected agent.
It should be obvious that an analysis that begins at the level of contrasts
of forms in no way ensures that one can arrive at psychologically real
categories. Clearly the children in the present study may have been making
even more distinctions or may have incorporated notions into the distinctions
reviewed here that have gone unnoticed. Nevertheless, the suggestion here
is that contrasts in the children’s use of forms provide an excellent point of
departure in identifying categorization schemes that correspond to ones
children may be employing.

IMPLICATIONS FOR OTHER AREAS OF STUDY

The present study has implications for many other areas of research, two
of which are discussed here. First, it seems that researchers interested in the
development of social cognition could make use of the present findings in
attempting to arrive at young children’s budding notions of self and other.
Researchers have noted the difficulties in making such assessments at phases
of development at which children will not yet participate in extensive in-
terviews. The sorts of patterns in the children’s speech presented in this
study could be used in structuring a preliminary framework for researching
young children’s social conceptions. Although clearly, children may not at-
tempt to distinguish at the linguistic level all social cognitive notions they
work with, the distinctions marked by the children in the present study,
such as the subtle distinctions in agency, are far more refined than those
described in the literature (see discussion in chapter 2, and Budwig & Wiley,
in press).
The present study also has implications for researchers interested in both
language and communicative disorders. If the methodology of the present
study were to be extended to the examination of children targeted as having
language or communicative difficulties, we could examine whether these
children’s language productions differ from those of normally developing
children. The methodology also allows the researcher to gather information
about the sorts of linguistic distinctions with which such children are work-
ing. This methodology is especially advantageous for young children who
have difficulty with less naturalistic testing settings. Such research not only
CONCLUSIONS 207

provides potential diagnostic measures but also could show alternative paths
to particular domains of language.

CONCLUDING COMMENTS

This study can be contextualized within the tradition of a growing number

of studies that illustrate that the early uses of a variety of linguistic forms
are tied closely to a related set of semantic and pragmatic factors. Although
this may be the case for many areas of grammatical development, we still
know relatively little about why some areas of language are more likely
than others to be dealt with in this way by young children. In the future,
studies ought to examine more closely this patterning in a cross-linguistic
perspective. It seems equally important to find out which forms do not link
up with functional clusters in early development. I suggest that children are
most likely to draw upon semantic and pragmatic notions when beginning
to use linguistic markers that encourage the speaker to adopt a perspective
or point of view. Thus deixis, modality, and voice would be likely areas for
form-—function pairings.
It seems important, also, to begin looking at the developmental relationship
between linguistic systems at various points in time. In the present study, it
has been suggested that as the children gave up their special patterning of self
reference forms, other changes appeared in the system. For instance, children
began regularly using an array of pronominal forms to refer to others. In
addition, modal forms became quite frequent. The relationship between these
developments and the children’s special pronominal usage is clearly worth
future study. Elsewhere I have noted that 2-year-old children began to use get
and be passives contrastively to mark nonprototypical kinds of agency. Again,
we know little about the connections between the get/be usage and children’s
use of pronominal forms. We therefore need longitudinal studies designed to
look at the interrelationship between various subsystems (see Gerhardt, 1988;
Nelson, 1989). Much of the research within functionalist approaches has
focused on particular pockets with little attention to the interconnections
between these areas at various points in time.
In the present study, I have also attempted to show how little we know
about developmental changes in lingustic subsystems. Although some func-
tionalist work within child language has aimed to document the nonadultlike
ways children link forms to various semantic or pragmatic functions, I have
sought to understand better the series of transformations that take place en
route to an adultlike system. This study can be seen as reflecting a gradual
change within functionalist approaches, in its revised view of development.
It has been a rather common procedure in studies of early child language
to work with the notion of obligatory context (see Brown, 1973). Until
208 CHAPTER 10

children use a given form 90% of the time in contexts where it would be
expected, their use of that form often has been thought to reflect little more
than incomplete mastery. In the present study, an attempt has been made
to illustrate the wealth of information that can be gathered about early
language development at a time when children often use forms in ways that
deviate from the language they are acquiring. Rather than questioning
whether a given form is being used appropriately, I have noted how various
forms are used at any developmental point in time.
This approach, then, radically changes the view toward development and
suggests a shift toward the assumption that development takes place in small
increments over time. There is a growing amount of evidence that such a
view is consistent with the cross-linguistic patterns of language acquisition
(see, for instance, Berman & Slobin, 1994; Bloom, 1991; Tomasello, 1992).
It seems likely that in the future, as researchers move away from fairly static
notions of development, the kinds of questions raised with regard to devel-
opment will continue to change as well. Within functionalist perspectives
this has meant a decrease in focus on the order of acquisition of one form
in relation to other (a popular focus at a time when researchers sought
cognitive explanations for acquisition sequences) and an influx in studies
that have attempted to describe the phases children go through from their
first use of a given form to adultlike usage.
Although there have been recent changes in the view of child language
development within functionalist perspectives, one area that we still know
surprisingly little about concerns the mechanisms that move the child from
one phase of development to the next. One of the biggest challenges for
future research within this area is the achievement of a better understanding
of the variety of factors involved in this process. I now briefly outline some
steps that need to be taken in order to understand better the nature of the
process in the particular domain that has been investigated here.
As was noted in chapter 9, until recently many functionalist accounts
have appealed either to universal cognitive explanations or to language-
typological explanations in describing how the child links form and meaning
in early child language. More recently there has been a growing consensus
that these factors combine in complex ways. For instance, although Slobin
(1985) developed the notion of a universal Basic Child Grammar, his recent
work highlighted the extent to which young children can and do adopt
patternings that look quite like the language they are acquiring (see Slobin,
1990). Conversely, though Bowerman (1985) originally was a strong propo-
nent for the view that children are quite flexible in adopting the meaning
distinctions marked in the target language, her most recent work also sug-
gested that children do not mark particular spatial notions even if their
language requires them to do so (Bowerman & Choi, 1994), Although there
is growing consensus for a position that supports the rich interplay between
CONCLUSIONS 209

cognitive predispositions and the input children receive regarding language-

specific solutions, it is clear that much more research is needed to understand
this relationship better. I now discuss one avenue of research I am currently
pursuing in an attempt to contribute to such an understanding.
One major problem that becomes apparent as functionalists acknowledge
the complexity of factors that influence children’s form-function patternings
is the difficulty in making clear predictions. Most recently (see Budwig,
1993b, 1993c, in press) I suggested that what would greatly contribute to
our understanding of factors influencing the organization and reorganization
of forms and functions are studies in which language input is varied, though
typology is held the same. More specifically, I suggested that much previous
cross-linguistic research has been based on assumptions about language
typology, rather than careful attention to input children receive.
As one example, my own recent reanalysis of the caregivers’ input for the
children studied in the present research revealed some unexpected pattern-
ings of form and function that could not be predicted from what we know
about the English language (see Budwig, in press). As was noted in chapter
9, the caregivers relied primarily on one self reference form in referring to
themselves in subject position, namely, / and although they used this form in
a variety of semantic and pragmatic contexts, they did show a tendency to
reserve the use of J for contexts involving a cluster of mental state verbs in
noncontrol acts. Elsewhere (Budwig, in press), I suggested that one partial
explanation for this may lie in the parents’ interactive goals, in particular, their
attempts to accentuate their children’s agency and play down their own.
Recent work in the area of language socialization has highlighted the extent
to which caregiver beliefs link up with the linguistic forms they use (see Miller
& Hoogstra, 1992; Schieffelin & Ochs, 1986), and Lin (1993) illustrated that
American caregivers who express distinct beliefs about their children in an
interview situation also organize conversation in distinct ways. What is not yet
known is whether these two groups of caregivers also provide children with
distinct linguistic input regarding self and other reference forms.
Previous research in the area of language socialization and individual
differences does suggest that caregivers who speak the same language provide
children with different input patterns. If we can identify pockets of language
in which children learning the same language receive varied input, this would
provide us with an important variable to help tease apart the relative
contributions of cognitive predispositions and language-specific factors in the
children’s construction of grammar. Combining the study of within-language
variation with the study of cross-linguistic variation also will allow us to draw
distinctions between input factors and typological factors in the children’s
construction of linguistic systems. Such a program can help us understand
better the linguistic, social, and conceptual factors influencing children’s
construction of linguistic systems.
 References

Ames, L. (1952). The sense of self of nursery school children as manifested by their verbal
behavior. The Journal of Genetic Psychology, 81, 193-232.
Atkinson, M. (1982). Explanations in the study of child language development. Cambridge,
England: Cambridge University Press.
Austin, J. (1962). How to do things with words. Oxford, England: Oxford University Press.
Bamberg, M. (1987). The acquisition of narratives. Berlin: Mouton de Gruyter.
Bamberg, M. (1991). Narrative activity as perspective taking: The role of emotionals, negations,
and voice in the construction of the story realm. Journal of Cognitive Psychotherapy, 5,
275-290.
Bamberg, M., Budwig, N., & Kaplan, B. (1991). A developmental approach to language acqui-
sition: Two case studies. First Language, 11, 121-141.
Barber, E. (1975). Voice—beyond the passive. Proceedings of the Berkeley Linguistic Society,
1, 16-24.
Bates, E., Bretherton, I., & Snyder, L. (1988). From first words to grammar: Individual differences
and dissociable mechanisms. Cambridge, England: Cambridge University Press.
Bates, E., & MacWhinney, B. (1987). Competition, variation and language learning. In B.
MacWhinney (Ed.), Mechanisms of language acquisition (pp. 157-193). Hillsdale, NJ:
Lawrence Erlbaum Associates.
Bates, E., & MacWhinney, B. (1989). Functionalism and the competition model. In B. Mac Whin-
ney & E. Bates (Eds.), The crosslinguistic study of sentence processing (pp. 3-73). Cambridge,
England: Cambridge University Press.
Bellugi, U. (1971). Simplification in children’s language. In R. Huxley & E. Ingram (Eds.),
Language acquisition: Models and methods (pp. 95-117). London: Academic Press.
Benveniste, E. (1971). Problems in general linguistics. Coral Gables, FL: University of Miami
Press.
Berman, R. (1979). Form and function: Passives, middles, and impersonals in modern Hebrew.
Proceedings of the Berkeley Linguistics Society, 5, 1-27.
Berman, R., & Slobin, D. (1994). Different ways of relating events in narrative: A crosslinguisti
c
developmental study. Hillsdale, NJ: Lawrence Erlbaum Associates.

210
REFERENCES 211

Bloom, L. (1991). Language development from two to three. Cambridge, England: Cambridge
University Press.
Bloom, L., Lightbown, P., & Hood, L. (1975). Structure and variation in child language. Mono-
&raphs of the Society for Research in Child Development, 40(2, Serial No. 160).
Bowerman, M. (1975). Commentary on structure and variation in child language. Monographs
of the Society for Research in Child Development, 40(2, Serial No. 160).
Bowerman, M. (1982). Reorganizational processes in lexical and syntactic development. In E.
Wanner & L. Gleitman (Eds.), Language acquisition: The state of the art (pp. 319-346).
Cambridge, England: Cambridge University Press.
Bowerman, M. (1985). What shapes children’s grammars? In D. Slobin (Ed.), The crosslinguistic
study of language acquisition (Vol. 2, pp. 1257-1319). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Bowerman, M., & Choi, S. (1994, January). Linguistics and nonlinguistic determinants of spatial
semantic development: A cross-linguistic study of English, Korean, and Dutch. Paper pre-
sented at the 18th Annual Boston University Conference on Language Development, Boston,
MA.
Braine, M. (1976). Children’s first word combinations. Monographs of the Society for Research
in Child Development, 41(Serial No. 164).
Brown, R. (1973). A first language: The early stages. Cambridge, MA: Harvard University Press.
Budwig, N. (1981). The role of language in two-year-old peers’ play. Unpublished manuscript,
University of California, Berkeley.
Budwig, N. (1986). Agentivity and control in early child language. Unpublished doctoral
dissertation, University of California, Berkeley.
Budwig, N. (1989). The linguistic marking of agentivity and control in child language. Journal
of Child Language, 16, 263-284.
Budwig, N. (1990). .The linguistic marking of non-prototypical agency: An exploration into
children’s use of passives. Linguistics, 28, 1221-1252.
Budwig, N. (1993a). Perspectives on the form-function relationship across 25 years of the SCLRF
meetings. In E. Clark (Ed.), Proceedings of the Twenty-Fifth Annual Child Language Research
Forum (pp. 297-306). Stanford, CA: Center for the Study of Language and Information.
Budwig, N. (1993b, April). Two year olds’ talk about self and other: A cross-linguistic com-
parison. Paper presented to the New York Child Language Group, New York, NY.
Budwig, N. (1993c, May). Indexing self and other in parent-child interactions. Paper presented
at The National Academy of Education Spencer Fellows Forum, Stanford, CA.
Budwig, N. (in press). What influences children’s patterning of forms and functions in early
child language? In D. Slobin, J. Gerhardt, A. Kyratzis, & J. Guo (Eds.), Social interaction,
social context, and language: Essays in honor of Susan Ervin-Tripp. Hillsdale, NJ: Lawrence
Erlbaum Associates. N
Budwig, N., & Bamberg, M. (in press). Functional approaches to language development. In N.
Budwig & M. Bamberg (Eds.), Theoretical approaches to language development. San Fran-
cisco: Jossey-Bass.
Budwig, N., Strage, A., & Bamberg, M. (1986). The construction of joint activities with an
agemate: The transition from caregiver-child to peer interaction. In J. Cook-Gumperz, W.
Corsaro, & J. Streeck (Eds.), Children’s worlds and children’s language (pp. 83-108). The
Hague: Mouton.
Budwig, N., & Wiley, A. Gin press). What language reveals about children’s categories of per-
sonhood. In L. Sperry & P. Smiley (Eds.), Learning about self and other through conversation.
New directions for child development series. San Francisco: Jossey-Bass.
Bullock, M. (1979). Aspects of the young child’s theory of causation. Unpublished doctoral
dissertation, University of Pennsylvania, Philadelphia.
Burke, K. (1969). A grammar of motives. Berkeley: University of California Press.
212 REFERENCES

Carter, A. (1975). The transformation of sensori-motor morphemes into words: A case study of
the development of more and mine. Journal of Child Language, 2, 233-255.
Carter, A. (1978a). The development of systematic vocalizations prior to words: A case study.
In N. Waterson & C. Snow (Eds.), The development of communication (pp. 127-138).
Chichester: Wiley.
Carter, A. (1978b). From sensori-motor vocalizations to words: A case study of the evolution
of attention-directing communication in the second year. In A. Lock (Ed.), Action, gesture,
and symbol (pp. 309-349). London: Academic Press.
Chafe, W. (Ed.). (1980). The Pear stories: Cognitive, cultural, and linguistic aspects of narrative
production. Norwood, NJ: Ablex.
Charney, R. (1980). Speech roles and the development of persona! pronouns. Journal of Child
Language, 7, 509-528.
Chomsky, N. (1957). Syntactic structures. The Hague: Mouton.
Chomsky, N. (1975). Reflections on language. New York: Pantheon.
Clark, E. (1978). Awareness of language: Some evidence from what children say and do. In A.
Sinclair, R. Jarvella, & W. Levelt (Eds.), The child’s conception of language (pp. 17-43).
Berlin: Springer-Verlag.
Clark, E. (1985). The acquisition of Romance, with special reference to French. In D. I. Slobin
(Ed.), The crosslinguistic study of language acquisition (Vol. 1, pp. 687-782). Hillsdale, NJ:
Lawrence Erlbaum Associates.
Comrie, B. (1976). Aspect. Cambridge, England: Cambridge University Press.
Comrie, B. (1981). Language universals and linguistic typology. Chicago: University of Chicago
Press.
Cook-Gumperz, J., Corsaro, W., & Streeck, J. (Eds.). (1986). Children’s worlds and children’s
language. The Hague: Mouton.
Cooley, C. (1908). A study of the early use of self-words by a child. Psychological Review, 15,
559-557
Corsaro, W. (1985). Friendship and peer culture in the early years. Norwood, NJ: Ablex.
Croft, W. (1987). Categories and relations in syntax: The clause-level organization of infor-
mation. Unpublished doctoral dissertation, Stanford University, Stanford, CA.
Croft, W. (1990). Typology and universals. Cambridge, England: Cambridge University Press.
Delancey, S. (1984). Notes on agentivity and causation. Studies in Language, 8, 181-213.
Delancey, S. (1987). Transitivity in grammar and cognition. In R. Tomlin (Ed.), Coherence and
grounding in discourse (pp. 53-68). Amsterdam: John Benjamins.
Delancey, S. (1990). Crosslinguistic evidence for the structure of the agent prototype. Papers
and Reports on Child Language Development, 29, 141-147.
Demuth, K. (1984). Aspects of Sesotho language acquisition. Bloomington: Indiana University
Linguistics Club.
Deutsch, W., & Budwig, N. (1983). Form and function in the development of possessives.
Papers and Reports on Child Language Development, 22, 36-42.
de Villiers, P., & de Villiers, J. (1974). On this, that and the other: Nonegocentrism
in very
young children. Journal of Experimental Child Psychology, 18, 438-447.
Duranti, A. (1985). Sociocultural dimensions of discourse. In T. A. Van Dijk (Ed.), Handbook
of discourse analysis: Disciplines of discourse (Vol. 1, pp. 193-230). London: Academic
Press.
Eilfort, W., Kroeber, P., & Peterson, K. (Eds.). (1985). Papers from the Parasession
on Causatives
and Agentivity. Chicago: Chicago Linguistic Society.
Ervin, S. (1961). Changes with age in the verbal determinants of word-association.
The American
Journal of Psychology, 74, 361-372.
Ervin, S., & Miller, W. (1964). The development of grammar in child language.
Monographs of
the Society of Research in Child Development, 29, 9-34.
REFERENCES 213

Ervin-Tripp, S. (1972). On sociolinguistic rules: Alternation and co-occurrence. In J. Gumperz

& D. Hymes (Eds.), Directions in sociolinguistics: The ethnography of communication (pp.
213-250). New York: Holt, Rinehart & Winston.
Ervin-Tripp, S. (1977). From conversation to syntax. Papers and Reports on Child Language
Development, 13, 1-21.
Ervin-Tripp, S. (1981). How to make and understand a request. In H. Parret, M. Sbisa,
& J.
Verschueren (Eds.), Possibilities and limitations of pragmatics (pp. 195-210). Amsterdam:
John Benjamins.
Ervin-Tripp, S. (1989). Speech acts and syntactic development: Linked or independent? Berkeley
Science Report, 61, Institute of Cognitive Studies, University of Califomia, Berkeley.
Ervin-Tripp, S. (1993). Constructing syntax from discourse. In E. Clark (Ed.), Proceedings of the
Twenty-Fifth Annual Child Language Research Forum (pp. 333-341). Stanford, CA: Center
for the Study of Language and Information.
Ervin-Tripp, S., & Gordon, D. (1986). The development of requests. In R. Schiefelbusch (Ed.),
Communicative competence: Assessment and intervention (pp. 61-95). Baltimore, MD: Uni-
versity Park Press.
Ervin-Tripp, S., & Mitchell-Kernan, C. (Eds.). (1977). Child discourse. New York: Academic
Press.
Fillmore, C. (1968). The case for case. In E. Bach & R. T. Harms (Eds.), Universals in linguistic
theory (pp. 1-88). New York: Holt, Rinehart & Winston.
Fillmore, C. (1977). Topics in lexical semantics. In R. Cole (Ed.), Current issues in linguistic
theory (pp. 76-138). Bloomington: Indiana University Press.
Firbas, J. (1964). On defining the theme in functional sentence analysis. Travaux Linguistique
de Prague, 1, 267-280.
Firth, J. (1951). Modes of meaning. London: The English Association.
Foley, W., & Van Valin, R. (1984). Functional syntax and universal grammar. Cambridge,
England: Cambridge University Press.
Gee, J. (1985). An interpretive approach to the study of modality: What child language can tell
the linguist. Studies in Language, 9, 197-229.
Gee, J., & Savasir, I. (1985). On the use of “will” and “gonna”: Towards a description of
activity-types for child language. Discourse Processes, 8, 143-175.
Gerhardt, J. (1983). Tout se tient: Towards an analysis of activity-types to explicate the inter-
relation between modality and future reference in child discourse. Unpublished doctoral
dissertation, University of California, Berkeley.
Gerhardt, J. (1988). From discourse to semantics: The development of verb morphology and
forms of self-reference in the speech of a two-year-old. Journal of Child Language, 15,
Dolo:
Gerhardt, J. (1990). The relation of language to context in children’s speech: The role of HAFTA
statements in structuring 3-year-olds’ discourse. Papers in Pragmatics, 4, 1-57.
Gerhardt, J., & Savasir, I. (1986). The use of the simple present in the speech of two 3-year-olds:
Normativity not subjectivity. Language in Society, 15, 501-536.
Givon, T. (1979). On understanding grammar. New York: Academic Press.
Givon, T. (1984a). Syntax: A functional-typological approach (Vol. 1). Amsterdam: John Ben-
jamins.
Givon, T. (1984b). Direct object and dative shifting: Semantic and pragmatic case. In F. Plank
(Ed.), Objects: Towards a theory of grammatical relations. New York: Academic Press.
Givon, T. (1985). Function, structure, and language acquisition. In D. Slobin (Ed.), The crosslin-
guistic study of language acquisition (Vol. 2, pp. 1005-1028). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Givon, T. (1990). Syntax: A functional-typological introduction (Vol. 2). Amsterdam: John
Benjamins.
214 REFERENCES

Gleitman, L., & Wanner, E. (1982). Language acquisition: The state of the state of the art. In
E. Wanner & L. Gleitman (Eds.), Language acquisition: The state of the art (pp. 3-48).
Cambridge: Cambridge University Press.
Golinkoff, R., Harding, C., Carlson, V., & Sexton, M. (1984). The infant’s perception of causal
events: The distinction between animate and inanimate objects. In L. Lipsitt & C. Rovee-
Collier (Eds.), Advances in infancy research (Vol. 3, pp. 145-151). Norwood, NJ: Ablex.
Goodenough, F. (1938). The use of pronouns by young children: A note on the development
of self-awareness. The Journal of Genetic Psychology, 52, 333-346.
Gopnik, A. (1980). The development of non-nominal expressions in one to two year old children.
Unpublished doctoral dissertation, Oxford University.
Gordon, D., Budwig, N., Strage, A., & Carrell, P. (1980, October). Children’s requests to unfa-
miliar adults: Form, social function, age variation. Paper presented at the Fifth Annual
Boston University Conference on Language Development, Boston, MA.
Gordon, D., & Ervin-Tripp, S. (1984). The structure of children’s requests. In R. Schiefelbusch
& J. Pickar (Eds.), The acquisition of communicative competence (pp. 295-321). Baltimore:
University Park Press.
Grice, H. (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Syntax and semantics:
Vol. 3. Speech acts (pp. 41-58). New York: Academic Press.
Gruber, J. (1965). Studies in lexical relations. Unpublished doctoral dissertation, Massachusetts
Institute of Technology, Cambridge, MA.
Gumperz, J. (1976). Language, communication and public negotiation. In P. Sanday (Ed.),
Anthropology and the public interest: Field work and theory (pp. 273-292). New York:
Academic Press.
Gumperz, J. (1982). Discourse strategies. Cambridge, England: Cambridge University Press.
Guo, J. (1993). Discourse motivations for the development of modal auxiliaries. In E. Clark
(Ed.), Proceedings of the Twenty-Fifth Annual Child Language Research Forum (pp. 315-
323). Stanford, CA: The Center for the Study of Language and Information.
Gvozdev, A. N. (1949). Formirovanie u rebenka grammaticeskogo stroja russkogo jazyka.
Moscow: Izd-vo Akademii Pedagogiceskix Nauk RSFSR.
Halliday, M. (1973). Explorations in the functions of language. London: Amold.
Halliday, M. (1975). Learning how to mean: Explorations in the development of language.
London: Arnold.
Harris, Z. (1951). Methods in structural linguistics. Chicago: Chicago University Press.
Hickmann, M. (1987). Social and functional approaches to language and thought. London:
Academic Press.
Holisky, D. (1987). The case of the intransitive subject in Tsova-Tush (Batsbi). Lingua, 71,
103-132.
Hopper, P., & Thompson, S. (1980). Transitivity in grammar and discourse. Language, 56,
251-299.
Huttenlocher, J., Smiley, P., & Charney, R. (1983). Emergence of action categories in the child:
Evidence from verb meaning. Psychological Review, 90, 72-93.
Huxley, R. (1970). The development of the correct use of subject personal pronouns in two
children. In G. B. Flores d’Arcais & W. Levelt (Eds.), Advances in psycholinguistics
(pp.
141-165). Amsterdam: North Holland.
Hymes, D. (1974). Foundations in sociolinguistics: An ethnographic approach. Philadelphia:
University of Pennsylvania Press.
Jackendoff, R. S. (1990). Semantic structures. Cambridge, MA: MIT Press.
Kaplan, B. (1983a). Genetic-dramatism: Old wine in new bottles. In S. Wapner & B. Kaplan
(Eds.), Toward a holistic developmental psychology (pp. 53-74). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Kaplan, B. (1983b). A trio of trials. In R. Lerner (Ed.), Developmental psychology: Historical
and philosophical perspectives (pp. 185-228). Hillsdale, NJ: Lawrence Erlbaum Associates.
REFERENCES 215

Karmiloff-Smith, A. (1979). A functional approach to child language. Cambridge, England:

Cambridge University Press.
Karmiloff-Smith, A. (1981). Getting developmental differences or studying child development.
Cognition, 10, 151-158.
Karmiloff-Smith, A. (1983). Language development as a problem-solving process. Papers and
Reports on Child Language Development, 22, 1-22.
Karmiloff-Smith, A. (1986). From meta-processes to conscious access: Evidence from children’s
metalinguistic and repair data. Cognition, 23, 95-147.
Kolodziej, P., Deutsch, W., & Bittner, C. (1991). Das Selbst im Spiegel der Kindersprache [The
self in the mirror of child language]. Zeitschrift fiir Entwicklungspsychologie and Paeda-
gogische Psychologie, 23(1) 23-47.
Kuczaj, S. II. (1982). On the nature of syntactic development. In S$. Kuczaj II (Ed.), Language
development: Syntax. and semantics (Vol. 1, pp. 37-81). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Kuroda, S. Y. (1973). Where epistemology, grammar, and style meet—a case study from Japa-
nese. In S. Anderson & P. Kiparsky (Eds.), A Festschrift for Morris Halle (pp. 377-391).
New York: Holt, Rinehart & Winston.
Kyratzis, A. (1993). Pragmatic and discursive influences on the acquisition of subordination-co-
ordination. In E. Clark (Ed.), Proceedings of the Twenty-Fifth Annual Child Language Re-
search Forum (pp. 324-332). Stanford: The Center for the Study of Language and Information.
Lakoff, G. (1977). Linguistic gestalts. In W. Beach, S. Fox, & S. Philosoph (Eds.), Papers from
the 13th Regional Meeting of the Chicago Linguistic Society (pp. 236-287). Chicago: Chicago
Linguistic Society.
Lakoff, G. (1986). Women, fire, and dangerous things: What categories reveal about the mind.
Chicago: University of Chicago Press.
Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago: University of Chicago Press.
Leopold, W. (1939). Speech development of a bilingual child: A linguist’s record: Vol. 1. Vo-
cabulary growth in the first two years. Evanston, IL: Northwestern University Press.
Levinson, S. (1979). Activity types and language. Linguistics, 17, 356-399.
Levinson, S. (1983). Pragmatics. Cambridge, England: Cambridge University Press.
Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self; New York:
Plenum Press.
Lin, A. (1993). The child as conversational partner: The creation of participation roles as
cultural activity. Unpublished doctoral dissertation, Clark University, Worcester, MA.
Lock, A. (Ed.). (1978). Action, gesture and symbol: The emergence of language. London: Aca-
demic Press.
Lock, A. (1980). The guided reinvention of language. London: Academic Press.
Loveland, K. (1984). Learning about points of view: Spacial perspective and the acquisition of
“l/You.” Journal of Child Language, 11, 535-556.
Lyons, J. (1982). Deixis and subjectivity: Loquor ergo sum? In R. J. Jarvella & W. Klein (Eds.),
Speech, place, and action (pp. 101-124). London: Wiley.
MacWhinney, B. (1991). The CHILDES project: Tools for analyzing talk. Hillsdale, NJ: Lawrence
Erlbaum Associates.
MacWhinney, B., & Bates, E. (Eds.). (1989). The crosslinguistic study of sentence processing.
Cambridge, England: Cambridge University Press.
Maratsos, M. (1979). Learning how and when to use pronouns and determiners. In P. Fletcher
& M. Garman (Eds.), Language acquisition: Studies in first language development (pp.
225-239). Cambridge, England: Cambridge University Press.
McNeill, D. (1963, August). The psychology of “you” and “I”: A case history of a small language
system. Paper presented at American Psychological Association Symposium on Child Lan-
guage: Structural Aspects, Philadelphia, PA.
Mead, G. (1934). Mind, self, and society. Chicago: University of Chicago Press.
216 REFERENCES

Mead, G. (1938). Philosophy of the act. Chicago: Open Court.

Miller, P., & Hoogstra, L. (1992). Language as a tool in the socialization and apprehension of
cultural meanings. In T. Schwartz, G. White, & C. Lutz (Eds.), New directions in psychological
anthropology (pp. 83-101). New York: Cambridge University Press.
Nelson, K. (1989). Monologue as the linguistic construction of self in time. In K. Nelson (Ed.),
Narratives from the crib (pp. 284-308). Cambridge, MA: Harvard University Press
Ochs, E. (1979). Transcript as theory. In E. Ochs & B. Schieffelin (Eds.), Developmental prag-
matics (pp. 43-72). New York: Academic Press.
Ochs, E. (1988). Culture and language development. Cambridge, England: Cambridge University
Press.
Ochs, E., & Schieffelin, B. (Eds.). (1979). Developmental pragmatics. New York: Academic
Press.
Painter, C. (1984). Into the mother tongue: A case study in early language development. London:
Frances Pinter.
Paprotté, W., & Sinha, C. (1987). A functional perspective on early language development. In
M. Hickmann (Ed.), Social and functional approaches to language and thought (pp. 203-
222). London: Academic Press.
Park, T. (1971). The acquisition of German morphology. Unpublished manuscript, Psychological
Institute, Bern.
Piaget, J. (1926). The language and thought of the child. New York: Harcourt.
Piaget, J. (1930). The child’s conception of physical causality. Totowa, NJ: Littlefield, Adams.
Piaget, J. (1954). The construction of reality in the child. New York: Basic Books.
Pike, K. (1982). Linguistic concepts: An introduction to tagmemics. Lincoln: University of Ne-
braska Press.
Pinker, S. (1984). Language learnability and language development. Cambridge, MA: MIT Press.
Pinker, S. (1988). Learnability theory and acquisition of a first language. In F. Kessel (Ed.), The
development of language and language researchers: Essays in honor of Roger Brown (pp.
97-119). Hillsdale, NJ: Lawrence Erlbaum Associates.
Pinker, S. (1989). Learnability and cognition: The acquisition of argument structure. Cambridge,
MA: MIT Press.
Poulin-Dubois, D., & Schultz, T. (1988). The development of the understanding of human
behavior: From agency to intentionality. In J. Astington, P. Harris, & D. Olson (Eds.), De-
veloping theories of mind (pp. 109-125). Cambridge, England: Cambridge University Press.
Quick, H. (1991). Maternal input and children’s use of self reference forms: A functional
approach. Unpublished manuscript, Clark University, Worcester, MA.
Radford, A. (1990). Syntactic theory and the acquisition of English syntax. Cambridge, England:
Basil Blackwell.
Rosch, E. (1978). Principles of categorization. In E. Rosch & B. Lloyd (Eds.), Cognition and
categorization (pp. 27-48). Hillsdale, NJ: Lawrence Erlbaum Associates.
Sacks, H., Schegloff, E., & Jefferson, G. (1974). A simplest systematics for the organization
of
tum-taking in conversation. Language, 50, 696-735.
Saksena, A. (1980). The affected agent. Language, 506, 812-826.
Savasir, I. (1984). How many futures? Unpublished master’s thesis, University
of California,
Berkeley.
Schieffelin, B. (1985). Acquisition of Kaluli. In D. I. Slobin (Ed.), The crosslinguis
tic study of
language acquisition (pp. 525-593). Hillsdale, NJ: Lawrence Erlbaum Associates.
Schieffelin, B. (1990). The give and take of everyday life: Language socialization of
Kaluli
children. Cambridge, England: Cambridge University Press.
Schieffelin, B., & Ochs, E. (Eds.). (1986). Language socialization across cultures.
Cambridge,
England: Cambridge University Press.
Schlesinger, I. (1977). The role of cognitive development and linguistic input in language
acquisition. Journal of Child Language, 4, 153-169.
REFERENCES
2157

Schlesinger, I. (1988). The origin of relational categories. In Y. Levy, I. Schlesinger, & M.

Braine
(Eds.), Categories and processes in language acquisition (pp. 121-178). Hillsdale,
NJ:
Lawrence Erlbaum Associates.
Searle, J. (1969). Speech acts. Cambridge, England: Cambridge University Press.
Searle, J. (1979). Expression and meaning. Cambridge, England: Cambridge University Press.
Shepherd, S. (1980). Creoles and language acquisition: Parallels in the expression of modality.
Paper presented at the Third Biennial Conference, Society for Caribbean Linguistics, Aruba.
Shipley, E., & Shipley, T. (1969). Quaker children’s use of “thee”: A relational analysis. Journal
of Verbal Learning and Verbal Behavior, 8, 112-117.
Shultz, T. (1980). Development of the concept of intention. In W. Collins (Ed.), The Minnesota
Symposium on Child Psychology, 13 (pp. 131-164). Hillsdale, NJ: Lawrence Erlbaum Asso-
ciates.
Silverstein, M. (1985). The functional stratification of language and ontogenesis. In J. Wertsch
(Ed.), Culture, communication, and cognition: Vygotskian perspectives (pp. 205-235). Cam-
bridge, England: Cambridge University Press.
Silverstein, M. (1987). The three faces of “function”: Preliminaries to a psychology of language.
In M. Hickmann (Ed.), Social and functional approaches to language and thought (pp.
125-164). London: Academic Press.
Silverstein, M. (1991)..A funny thing happened on the way to the form: A functionalist critique
of functionalist developmentalism. First Language, 11, 143-179.
Skinner, B. F. (1957). Verbal behavior. New York: Appleton-Century-Croft.
Slobin, D. (Ed.). (1971). The ontogenesis of grammar: A theoretical symposium. New York:
Academic Press.
Slobin, D. (1973). Cognitive prerequisites for the development of grammar. In C. A. Ferguson
& D. I. Slobin (Eds.), Studies of child language development (pp. 175-208). New York:
Holt, Rinehart & Winston.
Slobin, D. (1981). The origins of grammatical encoding of events. In W. Deutsch (Ed.), The
child's construction of language (pp. 185-199). London: Academic Press.
Slobin, D. (1982). Universal and particular in the acquisition of language. In E. Wanner & L.
Gleitman (Eds.), Language acquisition: The state of the art (pp. 128-170). Cambridge,
England: Cambridge University Press.
Slobin, D. (1985). Crosslinguistic evidence for the Language-Making Capacity. In D. Slobin
(Ed.), The crosslinguistic study of language acquisition (Vol. 2, pp. 1157-1256). Hillsdale,
NJ: Lawrence Erlbaum Associates.
Slobin, D. (1988). From the Garden of Eden to the Tower of Babel. In F. Kessel (Ed.), The
development of language and language researchers: Essays in honor of Roger Brown (pp.
9-22). Hillsdale NJ: Lawrence Erlbaum Associates.
Slobin, D. (1990). The development from child speaker to native speaker. In J. Stigler, R.
Shweder, & G. Herdt (Eds.), Cultural psychology: Essays on comparative human develop-
ment (pp. 233-256). Cambridge, England: Cambridge University Press.
Tanz, C. (1974). Cognitive principles underlying children’s errors in pronominal case-marking.
Journal of Child Language, 1, 271-276.
Tomasello, M. (1992). First verbs: A case study of early grammatical development. Cambridge,
England: Cambridge University Press.
Tomlin, R. (Ed.). (1987). Coherence and grounding in discourse. Amsterdam: John Benjamins.
van Oosten, J. (1986). The nature of subjects, topics and agents: A cognitive explanation.
Bloomington: Indiana University Linguistics Club.
Van Valin, R. (1991). Functionalist linguistic theory and language acquisition. First Language,
11, 7-40.
Van Valin, R. (1993). A synopsis of role and reference grammar. In R. Van Valin (Ed.), Advances
in role and reference grammar. Amsterdam; John Benjamins.
Vendler, Z. (1967). Philosophy in lingutstics. Ithaca: Cornell University Press.
218 REFERENCES

Vila, I. (1987). Emergence and use of the Castillian pronominal system in early language.
Unpublished manuscript, Universitat de Barcelona, Department de Psicologia Evolutiva i de
l’Educacio.
Wemer, H. (1957). The concept of development from a comparative and organismic point of
view. In D. B. Harris (Ed.), The concept of development: An issue in the study of human
behavior (pp. 125-148). Minneapolis: University of Minnesota Press.
Wemer, H., & Kaplan, B. (1984). Symbol formation. Hillsdale, NJ: Lawrence Erlbaum Associates.
Wolf, D. (1982). Understanding others: A longitudinal case study of the concept of independent
agency. In G. Forman (Ed.), Action and thought: From sensorimotor schemes to symbolic
operations (pp. 297-327). New York: Academic Press.
Wolf, D., Rygh, J., & Altshuler, J. (1984). Agency and experience: Actions and states in play
narratives. In I. Bretherton (Ed.), Symbolic play (pp. 195-217). New York: Academic Press.
Zubin, D. (1979). Discourse function of morphology: The focus system in German. In T. Givén
(Ed.), Syntax and semantics: Vol. 12. Discourse and syntax (pp. 469-504). New York:
Academic Press.
 Author Index

A Cc

Altshuler, J., 37, 218 @arlsonmv oun ola:

Ames, L., 60, 219 Carrell, P., 34, 214
Atkinson, M., 9, 19, 20, 205, 210 Carter A lic. o4 0D Leck
Austin, J.,6, 210 Chafe, W., 6, 212
Chamey, R., 37, 38, 48, 61, 212,
B 214
Choi, S., 208, 211
Bamberg, M., 3, 9, 11, 13, 18, 19, Chomsky, N., 4, 212
42, 50, 61, 170, 195, 210,211 Clarks. 2176212
Barber, E., 29, 210 Comrie, B., 5, 8, 29, 66, 212
Bates, E., 4, 8, 13, 15, 16, 17, 19, Cook-Gumperz, J., 9,212
20, 45, 63, 193, 194, 210,215 Cooley,.C., 59,60, 62, 212
Bellugi, U., 61, 167, 168, 169,210 Corsaro, W., 9, 44, 212
Benveniste, E., 40, 210 Croft, W., 29,212
Berman, R., 11, 13, 18, 29, 50, 208,
210
Bloom, L., 3, 18, 31, 39, 204, 208, D
211
Bowerman, M., 190, 195, 204, 208, de Villiers, J.,61, 212
211 de Villiers, P., 61, 212
Braine, M., 18, 32,38, 204, 211 Delancey; S:;); 29, 55,0), 199,
Bretherton, I., 45, 63, 210 pA
Brooks-Gunn, J., 59, 215 Demuth, K., 43, 212
Brown, R., 12, 31,39, 42, 51, 61, Deutsch, W., 12, 18,34, 39, 54, 59,
167,168, 169,201, 207, 211 G27 142215
Budwig, N., 3, 4, 8, 9, 10, 12, 18, Duranti, A., 7, 212
19, 24, 29, 42, 54, 59, 62, 65, 66,
76, 141, 158, 168, 172, 195, 200, E
206, 209, 210, 211, 212, 214
Bullock, M., 36, 211 Eilfort, W., 29,212
Burke, K., 81, 211 Ervin, S., 42, 189, 212
219
220 AUTHOR INDEX

Ervin-Tripp, S., 9, 12, 15, 18, 34, J

35, 43, 50, 67, 196, 202, 213,
214 Jackendoff, R. S.,.26, 27, 30, 39,
214
F Jefferson, G., 7, 216
Johnson, M., 200, 215
Fillmore, C., 5, 26, 27,35, 184, 213
Firbas, J., 6, 213 K
Firth, J.,6, 213
Foley, W., 4, 6, 27, 167, 213 Kaplan, B-; 3, 9,17, 1819723)
172, 181,205, 210, 214,218
G Karmiloff-Smith, A., 11, 18, 21, 22,
58, 170, 181, 186, 195, 215
Gee di13515,35,5530402 15 Kolodziej, P., 12,62, 215
Gerhardt, J., 13, 15, 18, 35, 39, 50, Kroeber, P., 29, 212
60, 64, 172, 191,207, 213 Kuczaj, S., II., 196, 215
Giv6on, T., 8, 29, 65, 142, 167, 195, Kuroda, S. Y., 40, 215
213
Gleitman, L, 16, 204, 214 L
Golinkoff, R.,37, 214
Goodenough, F, 60, 62, 214 Lakoff, G.,5,28, 199) 2005215
Gopnik, A., 60,61, 62,214 Leopold, W., 61,215
Gordon, D., 34, 67,213, 214 Levinson, S.; 7,35, 215
Grice, H., 6, 214 Lewis, M., 59, 215
Gruber, J., 27, 214 Lightbown, P., 31, 39, 204, 211
Gumperz, J., 7,35, 81, 113, 214 Lin, A., 209, 215
Gvozdev, A. N., 33, 214 Lock, A., 205, 215
Loveland, K., 61, 215
Lyons, J., 40, 215
H

Halliday, M., 6, 12,34, 191, 214 M

Harding, C.,37, 214
Harris, Z., 50, 64,214 MacWhinney, B., 4, 8, 13, 15, 16,
Hickmann, M., 11, 214 17, 19, 20, 69,193, 194202-
Holisky, D., 5, 8, 214 203, 210,215
Hood, L., 31, 39, 204, 211 Maratsos, M., 61,215
Hoogstra, L., 209, 216 McNeill, D., 61, 215
Hopper, P., 8, 28, 65, 66, 214 Mead, G., 81, 215, 216
Huttenlocher, J., 37, 38, 214 Miller, P., 209, 216
Huxley, R., 61, 214 Miller, W., 42, 212
Hymes, D., 7, 214 Mitchell-Keman, C.,9, 213
 AUTHOR INDEX 221

N-O Shipley, E., 61, 217

Shipley, T., 61, 217
Nelson, K., 60, 164, 188, 191, 207, Silverstein, M.,4, 5, 191, 217
216 Sinha, C., 11, 216
Ochs, E., 9, 12,49, 209, 216 skinner, BLE 3,217
Slobin, D., 8,9, 11, 13, 14, 18, 24,
P-Q 32, 33,35, 50,172) 181, 183,
184, 185, 203, 208, 210,217
Painter, C., 12, 34, 191, 216 Smiley, P., 37, 38, 214
Paprotte, W., 11, 216 Snyder, L., 45, 63, 210
Park, T., 61, 216 Strage, A., 34, 42, 211, 214
Peterson, K., 29, 212 Streeck, J.,9, 212
Piaget, J., 36, 37, 53, 216
Pike, K., 50, 64, 216 ti
Pinker, S., 17, 19, 20, 205, 216
Poulin-Dubois, D., 37, 216 Tanz, Go183, 217
Quick, H., 194, 216 Thompson, S., 8, 28, 65, 66, 214
Tomasello, M., 3, 208, 217
R Tomlin, R., 6, 217

Radford, A., 61, 182, 192, 204, 216 V

Rosch, E, 28, 216
Rygh, J., 37, 218 van Oosten, J., 8, 217
Van Valin, R.,4, 6, 17, 18, 23, 26,
S Zl; 21; 30,952) 16), 2152217
Vendler, Z., 66, 217
Sacks, H., 7, 216 Vila, I., 12, 218
Saksena, A., 112, 216
Savasir, I., 13, 15, 35,35, 39, 50,
64, 213, 216 W-Z
Schegloff, E., 7, 216
Schieffelin, B., 9, 12,33, 209, 216 Wanner, E., 16, 204, 214
Schlesinger, I., 17, 196, 216, 217 Werner, H., 9, 17, 23, 172, 181,
Schultz, T., 36, 37,246, 217 205, 218
Searle, J., 7, 67, 217 Wiley, A., 158, 206, 211
Sexton, M., 37, 214 Wolf, D.,37, 218
Shepherd, S., 34, 217 Zubin, D., 167, 218
7 tie =o
in 7
~ :

Pre Pei
LGTY oS
PRS 108 SL, tener
P= ais. (tele

Oe Ri di! sf Al oe,
Vote. eel Bt
en eR AO Ral a
S e)
Fit, LSS RIL Be Peet
jubey © .@, OF! Dr d55) naive
bag HS a. J. Jee
~»

im * GE LD ot
~

a * 3 5 " , Be
=,

i. I ia IF se
e es | ane Y want
CtS AS
<
JE
S >
Ae SES =
Cle Be LM oes!
cal Coe Wed ates
( ~
5 y

«, f
ia 7 ey AY 0) > Hua
“eg ne Be ’ Pe tie 2° “>
loys). ett tat De Be Te VL iw © 5005,
-
Se
a’.

< ey + A hulecsv iM Sis wv mi

Mr tf JT ene - X 7eeSaks mee os. 1s
Coageeide “4
2% wimatah ey the ional
me, ; . ss

att, Att
‘Sita aaa lB AES ESM mitouiel
é v| ’ Tt cowpet/ tT? wil UT a) dM lead
kit Be “mee oteO83 ineé
*s AY he ; Poa ogher
é 3 SN
Pix! eb i pat ot Gu. U_SGrg ca

wie Spar, Sy ae. Gen

= . ek’ , ais a os
ioe Silica ores a 7
“roi Ot oe se: 282
¥, s heats Del Were eS
; te:
wk i u : a
a
on
 Subject Index

A cross-linguistic, 10, 14,25, 29,38,

39, 207, 209
activity-type, 30, 35,36, 114, 184,
188, 198, 200 D
agency, 10, 25, 30, 32,33, 35-40,
53, 76, 88, 94, 95, 98-101, 112, deixis, 60, 61
120, 124, 135, 144, 149, 153, developmental functionalist, 3, 19,
154, 158, 161, 165, 172, 177, 181
178, 184, 185, 190, 192, 199,
202, 206, 207, 209 E-F
agent, 8, 10, 19, 26-33, 37,39, 40,
53, 65, 94, 99, 106, 112, 125, event schemes, 5
130, 142, 155, 165, 184, 185, foregrounding, 8
190, 199, 204, 206 functionalism, 3, 4
British, 6
B Prague School, 6

backgrounding, 8 H
Basic Child Grammar, 10, 185, 208
hierarchical intigration, 23, 24, 181,
iS 201

case marking errors, 60, 61, 62 I

cognitive predispositions,
185, 209
communicative pressure, 194-196 incipient acts, 81
competition model, 16, 193 individual differences, 63, 156,
conceptual development, 14 176, 209
contextual cues, 7, 81, 113 input, 11, 13, 14, 114, 186-190,
continuity, 16, 17,204, 205 193, 194, 195, 209
conversation exchanges,12
correlations i
horizontal,
16
vertical, 16 language socialization, 209
223
224 SUBJECT INDEX

M R

macroroles, 27 Role and Reference Grammar, 6,

manipulative activity scene, 10, 33, 183247
184, 185, 190
means-ends relationships, 181 S
mechanisms, 17, 21, 208
microroles, 27 self corrections, 69
modal forms, 92, 100, 131, 163, semantic bootstrapping, 17,205
172, 180, 194 social causation, 34, 39
motives, 99, 108, 172 speech acts, 6, 12,
speech events, 12

O f

obligatory contexts, 201, 207 telos, 9, 19, 23

ontogenesis, 22, 23, 181 text, 6, 11
operating principles, 183 transitivity,
10, 14, 28, 29, 33, 38,
order of acquisition, 20-22, 208 65, 76, 186
orientation typological factors, 190, 191, 208,
cognitive, 5,8, 9, 10, 13 209
multifunctional, 5,8, 9, 13 typology, 209
social, 5-7, 9, 12, 13
textual’S) 11513 U
orthogenetic principle, 23, 181, 205
unifunctional, 21, 22, 23,179, 180,
181, 186
P
V
passives, 8n, 10
patient, 8, 28,33, 184 volition, 28, 33, 165, 199
perspective, 5, 10, 11, 14, 25, 29,
38,39563,1115166, 167.171; W
172, 186
physical causation, 33 within-language variation, 209
plurifunctional, 21-23, 179, 181,
186
prepositions, 131, 155,172
prototype, 5, 28, 29, 33, 38, 39,
200