Developmental Psychology
1989, Vol. 25, No. 6, 1004-1019
Copyright 1989 by the American Psychological Association, t~e.
0012-1649/89/$00.75
Integrating Language and Gesture in Infancy
Elizabeth Bates and Donna Thal
Kimberly Whitesell
University of California, San Diego
University of California, San Francisco
Larry Fenson
Lisa Oakes
San Diego State University
University of Texas, Austin
Whether language/gesture correlations in early language development can be explained by parallelism or comprehension mediation was examined. Study 1, parental report data for 95 l-year-olds,
suggested that word comprehension and production are dissociated in this age range and that the
comprehension and production factors map onto distinct aspects of gesture. Study 2 tested 41 13-15month-olds in a task in which the modeled gesture was accompanied by supportive, contradictory, or
neutral narratives. Results showed that infants can use adult speech as an aid in reproduction of
modeled gestures (comprehension mediation). However,there is still additional variance in gestural
production that correlates with expressive vocabulary when comprehension-related variance is removed. Thus, comprehension mediation and parallelism both appear to be operating.
Before they are 18 months old, most children discover the
idea that things have names (i.e., that words can be used to
recognize, request, label, categorize and represent objects in the
world). This discovery is preceded by evidence for word comprehension and is either preceded or accompanied by the use of
conventional gestural schemes in communication or in play
with objects, or in both. Piaget (1962) and Werner and Kaplan
(1963) have argued that this temporal synchrony is no accident.
Their approach is based on the idea of parallelism (i.e., that
linguistic and gestural schemes for objects are related in early
development because they both depend on a common underlying symbolic function). Thus, word comprehension, word production, and symbolic play are clear, public manifestations of a
much more general cognitive shift from sensorimotor processing to the use of symbols in many aspects of thinking, problem
solving, and communication (for reviews see Bates, Bretherton,
Shore & MeNew, 1983; Leonard, 1988; Shore, Bates, Bretherton, Beeghly, & O'Connell, in press; Shore, O'Connell, & Bates,
1984; Voiterra & Caselli, 1985).
The Piaget/Werner hypothesis has been influential, but it is
also controversial. A host of correlational studies appeared
across the 1970s and 1980s, examining associations and dissociations between early language and other sensorimotor domains (for reviews, see Bates, O'Connell, & Shore, 1987; Bates
& Snyder, 1987; Harris, 1983; Johnston, 1985). In general,
these studies provided evidence against a global, cross-domain
stage shift from sensorimotor to symbolic functioning (see also
Uzgiris & Hunt, 1978, 1987). Summarizing, some nonlinguis-
This research was supported by a grant from the John D. and Catherine T. MacArthur Foundation Research Network on the Transition
From Infancy to Early Childhood, and by a National Institutes of
Health First Investigator Award (PHS NS 19639) to Donna Thai.
Correspondence concerning this article should be addressedto Elizabeth Bates, Department of Psychology C-009, University of California,
San Diego, La Jolla, California 92093.
tic measures are indeed reliably correlated with the emergence
of naming (e.g., means-end behaviors or tool use, and aspects
of symbolic play). However, several other cognitive domains appear to be completely dissociable from early language (e.g., spatial cognition and traditional tests of object permanence).
Because the Piaget/Werner hypothesis appears to have failed
in its strongest form (i.e., single-factor parallelism), some investigators have argued that language develops along a maturational course that is independent from other aspects of cognition (Curtiss & Yamada, 1978; Curtiss, Yamada, & Fromkin,
1980; Harris, 1983; Petitto, 1988). Such developmental independence would provide support for Chomksy's long-standing
argument that language is an autonomous, encapsulated,
"modular" cognitive system (Chomsky, 1965, 1980; Fodor,
1983; Roeper & Williams, 1987). However, strong arguments
for autonomy (like strong arguments for parallelism) are not
easily reconciled with the fact that at least some aspects of nonverbal cognition (e.g., symbolic play) are reliably associated
with progress in language. There appears to be room for compromise between these two strong positions.
A number of investigators have offered compromise views to
explain the relation between language and cognitive development, alternatively referred to as local homology (Bates, Benigni, Bretherton, Camaioni, & Volterra, 1977), skill theory
(Corrigan, 1978; Fischer, 1980), and the specificity hypothesis
(Gopnik & Meltzoff, 1986, 1987). Although they differ in detail,
these theories are all based on the claim that language-cognition relations are many-to-many rather than one-to-one. Language itself is composed of many separate, underlying skills or
mechanisms. Some of these components may be specific to language; other components are shared by several cognitive domains, resulting in significant cross-domain correlations at
those points in development when a shared component is "coming on line."
Working within this revisionist framework, we have been trying to improve our understanding of the cognitive mechanisms
1004
LANGUAGE AND GESTURE
shared by word comprehension, word production, and symbolic gesture in the second year of life. Most of our work has
focused on the relation between early language and one particular type of gesture: the action schemes associated with familiar
toys and household objects (e.g., drinking from cups, putting
hats on heads, placing a telephone receiver near the ear). These
gestures are often used by 1-year-olds to recognize an associated
object or event, and they form the primitive beginnings of the
activity referred to as symbolic play (Inhelder, Lezine, Sinclaire,
& Stembak, 1971; Lezine, 1978). Because they appear to serve
a recognitory function in the early stages and are subsequently
used in empty-handed pantomimes with inappropriate versions
of the object to confer the appropriate object meaning, these
action schemes have been referred to as enactive names (Escalona, 1973) or gestural depiction (Werner & Kaplan, 1963).
The symbolic status of these and other forms of gesture in
infancy is discussed in detail by Bates et al., (1983). They presented a classification scheme for infant gesture that involves
six features: (a) learned through imitation, (b) used in communication (at least part of the time), (c) used to establish reference
to external objects or events, (d) used to refer to specific individuals or classes of objects or events, (e) used at some physical
distance from the referent object, and (f) presented and acquired within a conventional system of other symbols or signs.
Vocal names and American Sign Language (ASL) signs share
all six of these features. Other forms of gesture share only a
subset, permitting us to classify gestures according to their degree of overlap with "true" symbols or names along these theoretical dimensions. For example, showing-off games like "byebye" and "pattycake" overlap with language along the first two
features (they are derived through imitation and are used in
communication). So-called deictic gestures (giving, showing,
and pointing) share the second and third features (communicative use and establishment of reference); unlike giving and
showing, the pointing gesture also shares the fifth feature (distance from the referent). Recognitory gestures with objects
(e.g., drinking from a cup, putting on a necklace) share the first
(imitation), third (reference), and fourth (reference to specific
classes of objects or events) features. When the same gestures
are carried out with substitute objects or in empty-handed pantomime, then we could argue that they inherit two additional
features of true symbols: use in communication (Acredolo &
Goodwin, 1988; Volterra & Caselli, 1985)and execution at
some physical distance from the real object or class of objects
that the gesture names or represents (Volterra, Bates, Benigni,
Bretherton, & Camaioni, 1979).
This analysis provides a theoretical justification for predicting correlations between language and various forms of gesture
in the first and second year of life. Using these criteria, Bates et
al. (1983) concluded that recognitory gestures or enactive
names share more overlap with language than do any other aspects of gesture (except, of course, for the true linguistic signs
of ASL)--particularly when the gestures are observed or elicited out of context, without the usual referent object in hand.
Hence, these gestures ought to bear the strongest and most consistent relationship to the development of naming in the vocal
modality. There is now a large empirical literature supporting
this prediction, demonstrating a link between recognitory ges-
1005
tures (also called primitive symbolic play) and the emergence of
naming, in normal infants and in several different populations
of retarded or language impaired children. These include links
in time of onset, function, frequency of use, size of repertoire,
content, rates of "decontextualization," individual differences
in style, and some neuropsychological evidence suggesting that
language and symbolic/recognitory gestures share aspects of
left hemisphere control (Bates et al., 1977; Bates, Benigni,
Bretherton, Camaioni, & Volterra, 1979; Bates, Bretberton,
Snyder, Shore, & Volterra, 1980; Bates & Volterra, 1984; Bauer
& Shore, 1986; Curcio, 1977, Eider & Peterson, 1978; GoldenMeadow & Mylander, 1984; Gopnick & Meltzoff, 1987; Killen
& Uzgiris, 1981; Lezine, 1978; Nicolich, 1977; Snyder, Bates,
& Bretherton, 1981; Volterra et al., 1979; Wolf & Gardner,
1979; for reviews see Bates & Snyder, 1987; Harris, 1983; Johnston, 1985). Similar links have also been demonstrated for a
number of clinical populations (Beeghly & Cicchetti, 1987;
Roth & Clark, 1987; Sigman & Mundy, 1987; Snyder, 1977;
Terrell, Schwartz, Prelock, & Messick, 1984; T h a l & Bates,
1988a; Thal, Bates, & Bellugi, in press). In short, evidence in
favor of some kind of parallelism between language and gesture
appears to be quite robust, even when investigators control for
factors like Global Maturation or Mental Age. Results are particularly strong when contextual factors are controlled (e.g.,
Bates et al., 1980).
This does not mean that recognitory gestures and vocal
names are identical (cf. Petitto, 1988). There are some important differences between linguistic names and recognitory gestures that must also be taken into account in models of language/gesture relations (for a review, see Shore et al., in press).
For example, Zukow (1984) has argued that symbolic play
grows out of social interaction and that the symbols children
use in play never achieve the independent cognitive status of
linguistic symbols. This claim is compatible with a report by
Bretherton, Bates, Benigni, Camaioni, and Volterra (1979), that
attachment scores correlate with symbolic play but not with
vocal naming. K. Nelson (1985) has also argued that action
schemes with objects are not true symbols, or at least not as
symbolic as words for the same object (see also Huttenlocher &
Higgins, 1978). Nelson's argument is supported by the finding
that recognitory gestures can develop weeks or months earlier
than vocal names in some children (Thal& Bates, 1988b) and
by studies showing that recognitory gestures are affected more
strongly by characteristics of the referent object (e.g., whether
the object is in the child's hand and whether it is a detailed and
realistic version of the "real thing" "Bretherton, O'Connell,
Shore, & Bates, 1984). There are also studies showing that measures of symbolic gestures are less reliable (i.e., lower intermeasure correlations) than corresponding measures of language
(Bretherton & Bates, 1984).
Most of these differences can be handled by models that permit degrees of symbolic status (e.g., the six-feature model proposed by Bates et al., 1983). However, Petitto (1988) has presented an argument against language/gesture parallelism that is
(potentially) much more damaging. Specifically, she has suggested that language/gesture correlations reflect the fact that
children use adult speech as a guide during their play activities
(i.e., comprehension mediation). For example, suppose that an
1006
BATES, THAL, WHITESELL, FENSON, OAKES
adult playmate says " M m m , that bear looks hungry; give him
something to eat." If a child understands this input, the likelihood of a bear-feeding sequence is considerably enhanced. Such
causal effects of adult language on child play could account for
positive correlations between symbolic play and language comprehension scores. Meanwhile, there is also a significant positive
correlation between language comprehension and language
production. This correlation reflects a "true" homology (i.e.,
the fact that comprehension and production make use of a common lexical base). This mix of causal relations results (indirectly) in a positive correlation between symbolic play and language production--a correlation that proponents of parallelism (mistakenly) interpret to mean that vocal naming and
recognitory gestures are manifestations of a common symbolic
function. If we accept this logic, then correlations between language production and symbolic play actually reflect nothing
more than a correlation of language with itself.
There are a number of reasons why comprehension mediation may not be sufficient to account for all of the correlational
patterns between language and gesture that have been observed
in the literature to date. For one thing, language comprehension
and language production are typically not highly correlated in
the first 2-3 years of life (Bates, Bretherton, & Snyder, 1988;
Snyder et al., 1981 ). This fact presents problems for the correlational logic described earlier. Indeed, when comprehension and
production are dissociated, specific links between word production and gestural production are still observed (Bates et al.,
1980; Thai & Bates, 1988a, 1988b). Furthermore, there is evidence suggesting that word comprehension and word production correlate with different aspects of communicative and symbolic gesture (Bates et al., 1980).
But these arguments are still not sufficient to put the mediation account to rest. There are, in fact, at least two possible
forms of linguistic mediation that could be responsible for language/gesture correlations in 1-year-old children. First, the
child may use the language provided by others as a retrieval cue
for gestural activity (i.e., comprehension mediation). Second,
she may use her own covert or overt language production as a
guide to action (i.e., production mediation). In view of these
confounds, we believe that a more detailed exploration of language/gesture relations is in order, in a design that permits us
to disentangle the dissociable effects of comprehension and production. The distinction between parallelism and linguistic mediation is important not only because the two hypotheses reflect
fundamentally different views of language and thought, but also
because the two views have different implications for research
and practice with language disordered children. The parallelism
hypothesis has been particularly influential in clinical research
(Bates & Snyder, 1987; Rice, 1983). Specifically, it has been argued that spontaneous or elicited symbolic play can be used
to distinguish between language-specific deficits in toddlers and
preschool children and also, between deficits that are due to
a more general delay in cognition and representation (Kamhi,
1981; Sigman & Mundy, 1987; Snyder, 1975; Roth & Clark,
1987; Terrell et al., 1984). Given the widespread interest in this
issue, we think it is important to determine the extent to which
i-year-old children can make use of linguistic support to understand and reproduce a gestural symbol. If these infants can in-
deed integrate gestural and linguistic representations for objects, we need to determine whether this integration is sufficient
to account for the parallels that have been observed between
enactive naming and early language development in normal and
clinical populations.
To explore these issues, we will present results from two studies that examined the relationships among word comprehension, word production, and enactive/gestural naming. Study 1
makes use of detailed questionnaires administered to the parents of 95 children between 12 and 16 months of age. Here we
will show that comprehension and production are each correlated with different aspects of communicative and symbolic gesture. Study 2 is a laboratory experiment with 41 children between 13 and 15 months of age, directly investigating the causal
effect of adult linguistic cues on gestural imitation. Our results
will show that parallelism and linguistic mediation are both
partially correct, each accounting for different aspects of the
relationship between language and gesture in infancy.
S t u d y 1: P a r e n t a l R e p o r t s
Because the words and gestures that we are interested in have
just begun to emerge in the 12-16-month age range, it is difficult
to obtain a reliable sample of a given child's abilities in a 6090-min laboratory visit. It is possible to remedy the problem
somewhat by developing elicitation procedures that make the
occurrence of target behaviors more likely (see Study 2). But
this practice also has its limits, because toddlers are notoriously
difficult to test. For these reasons, we have found it useful to
supplement naturalistic or structured observations, or both,
with information obtained through parental report. In the present study we assessed the three-way relations among word comprehension, word production, and gestural production, using a
checklist-style instrument that permitted the parent to rely on
recognition memory rather than recall.
The inventory used in Study 1 is one of two that have been
developed and validated in our laboratory over the past 15 years.
Each inventory takes the form of a list of vocalizations, words,
gestures, grammatical morphemes, or word combinations that
are typical of children within a particular age range. Parents
were asked to fill these out at home over a few days, so that they
could observe their child while they had the questionnaire in
mind, and identify as many examples as possible. Thus, a list
of the child's current behaviors as they occurred across many
contexts was obtained. This information proved to be far more
reliable than retrospective reporting, and more representative
than 2 or 3 hours of observational data in this age range. The
validity and reliability of these inventories, as well as their predictive value, has been demonstrated in a number of studies
(Bates, Benigni, et al., 1979; Bates, Snyder, Bretherton, & Volterra, 1979; Bates et al., 1980; Bates et al., 1988; Bretherton
et al., 1984; Dale, Bates, Reznick, & Morisset, 1989; Dale &
Robinson, 1988; Epport, 1988; Goldfield, 1988; O'Brien, 1988;
Reznick & Goldsmith, 1989; Snyder et al., 1981; Tamis-LeMonda & Bornstein, 1989; Thai & Bates, 1988a). Positive results
have also been obtained in other laboratories with parental report instruments that were similar to ours (Acredolo & Goodwyn, 1988; Rescorla, in press; Rescorla & Schwartz, 1988).
LANGUAGE AND GESTURE
Method
Subjects. Subjects were 95 children between 12 and 16 months of
age. They were chosen from a subset of more than 500 children participating in a large-scale study in progress in San Diego, to provide agegraded norms for parental report instruments of language and communicative development of children between 8 and 30 months of age
(Fenson, Vella, Flynn, & Thal, 1988). All of the children chosen for the
present study had reported production vocabularies of under 100
words, a relatively homogeneous stage of language development for
which the literature suggests that language/gesture correlations are theoretically interesting and relatively large.
Children and their parents were identified and recruited in several
ways--through advertising in local newspapers and contacts made
through area pediatricians and day-care centers. Parents were initially
contacted by telephone or letter; if they agreed to participate, we sent
them copies of the Language and Gesture Inventory (Bates et al., 1986;
described under Materials), with a stamped, self-addressed envelope for
return. We also included a simple subject-information sheet that inquired about the child's developmental history and contained questions
about the parents' occupation, education, and the language(s) spoken
in the home. All of the children selected for the norming study were
acquiring English as their first language, and none had a significant history of birth trauma or other serious medical problems (as indicated on
the information sheet). Most of the children chosen for this study were
from middle-class and upper middle-class homes (as determined by parental occupation and education), although some working-class families
were included in the sample.
Materials. Part 1 of the Language and Gesture Inventory is composed of a list of 500 words, organized into semantic categories (e.g.,
food words, bedtime words) that are likely to appear in the vocabularies
of young English-speaking children within the 2nd year. Parents were
asked to check those words that their child produced (column l) or
comprehended (column 2). This part of the questionnaire has been developed and validated in several previous studies and has proven to be a
reliable and valid predictor of observed language comprehension and
production across this age range (Bates et al., 1987; Dale & Robinson,
1988; Snyder et al., 1981). Part 2 of the questionnaire is a list of 64
gestures that are frequent in the repertoires of l-year-old children in
our culture; the list includes communicative routines like "bye-bye" or
"pattycake," deictic gestures (i.e., giving, showing, pointing), and object-associated gestures that may occur in communication or in play
(e.g., "drinking," "telephoning"). The parents were asked simply to
check those gestures or actions that they have seen their child produce.
The complete gesture list was developed for the present study; however,
previous studies in our laboratories have suggested that parental reports
of gestural development do correlate with observations of the same phenomena in this age range (Bates, Snyder, et al., 1979; Volterra & Caselli,
1985; Bretberton, 1984).
Scoring and data reduction. The following subscale scores were derived from the Language and Gesture Inventory:
1. Word comprehension--total words comprehended, number of
nominals comprehended (i.e., names for common objects), and number
of nonnominals comprehended.
2. Word production--total words produced, number of nominals
produced (as defined above), and number of nonnominals produced.
3. Gesturalproduction--total gestures produced, number of gestures
associated with common objects (i.e., enactive names), deictic gestures
(giving, showing, pointing, ritual requests), and gestural routines that do
not refer to common objects (e.g., "patty-cake," "bye-bye," "so big").
We separated nominals from nonnominals in both comprehension
and production, for three reasons. First, on the basis of the Bates et al.
(1983) classification scheme, we expected correlations between language and gesture to be stronger when we restricted our attention to
1007
Table 1
Descriptive Statistics for Parental Report
of Language and Gesture
Measure
Comprehension
Total
Nominals
Nonnominals
Production
Total
Nominals
Nonnominals
Gesture
Total
Object gestures
Deictic gestures
Gestural routines
M
SD
Range
121.03
55.54
65.49
67.09
32.51
36.94
9-297
2-128
6-169
25.00
13.93
11.07
20.78
12.59
9.15
0-83
0-46
0-41
37.45
24.03
3.55
9.87
9.07
7.17
0.71
2.88
13-52
6-39
1-4
0-15
object-related terms (i.e., recognitory gestures with common objects,
and vocal names for common objects). Second, previous research has
shown that within-language correlations are higher when attention is
restricted to general nominals (e.g., Bates et al., 1988); hence, a separation of nominals and nonnominals should help to clarify the nature and
degree of any comprehension/production dissociation observed. Third,
recent studies have suggested that children who have a higher proportion of common nouns in their vocabularies may show a different pattern of development in communicative gesture and symbolic play
(Bauer & Shore, 1986).
Descriptive statistics were examined for all 10 scores to provide information about the size and composition of comprehension, production,
and gesture repertoires in this developmental period. The 7 subscale
scores (excluding totals) were also factor analyzed (principal component
analysis with varimax rotation). If some form of the local homology
model is correct, then we should expect to find at least two distinct factors (defined by word comprehension and word production, respectively), each with a different pattern of loadings from the three gesture
scores.
Results and Discussion
Table i s u m m a r i z e s descriptive statistics for the 10 language
a n d gesture scores. T h e r e are n o surprises here; findings from
the p a r e n t a l r e p o r t i n s t r u m e n t are largely c o m p a t i b l e with diary studies a n d laboratory observations by other investigators
working in this age range.
According to their parents, m o s t o f these children already
h a d substantial c o m p r e h e n s i o n v o c a b u l a r i e s - - a n average o f
121 words, with a range from 9 to 297. Slightly less t h a n one
h a l f o f these words were n a m e s for objects (M = 55.5, with a
range from 2 to 128). By contrast, the same children p r o d u c e d
an average o f only 25 words (with a range from 0 to 83). Slightly
m o r e t h a n one h a l f o f these, on the average, were n a m e s for
objects ( 13.9, with a range from 0 to 41 ).
In the sample as a whole, c o m p r e h e n s i o n a n d p r o d u c t i o n
were correlated at +.46; a l t h o u g h this correlation is reliable
(p < .01), it also m e a n s t h a t c o m p r e h e n s i o n and p r o d u c t i o n
share less t h a n 22% o f their variance. Figure 1 presents a scatterplot o f total c o m p r e h e n s i o n a n d p r o d u c t i o n scores for all the
children in this study. T h e positive linear t r e n d responsible for
1008
BATES, THAL, WHITESELL, FENSON, OAKES
100
80
[]
[]
[]
[]
[]
[]
[]
60
[]
[][]
El
[]
[] [ ]
~=
=p
40
D
m
[]
[]
[]
[]
~13
[]
m
[][]
[]
20
[]
•[]
[]
[] [ ] B ~
[]
0
[]
[]
[]
I ~ 1 [ ] []
9
[]
[]
[]
[] []
[]
[]
i=
100
[]
i
200
i
300
COMPREHENSION
Figure 1. Scatterplot of total comprehension and production scores of
95 children between 12 and 16 months of age, as reported by parents
on the Language and Gesture Inventory.
a +.46 correlation can be discerned in Figure 1, but it is equally
clear that the two modalities are dissociated in many children.
At almost every point along the comprehension range (from 0
to 250 words), we found at least a few children who were producing fewer than 10 word forms. This pattern replicates a scatterplot of comprehension/production relations reported by
Snyder et al. (1981) for a sample of 32 children, and it is compatible with many reports in the literature on the extent to
which receptive and expressive language development can be
dissociated in the early stages.
Turning to results of the gestural scale (see Table 1), these
children produced an average of 37 gestures, with a range from
13 to 52. These included a mean of 24 object-associated gestures (ranging from 6 to 39), 9.87 gestural routines (from 0 to
15), and 3.55 out of the 4 possible deictic gestures (with an observed range from 1 to 4). Most of the children reportedly gave
and showed objects to their parents, and produced some kind
of ritualized gesture in requests (e.g., opening and shutting the
palm); hence, much of the variance in the deictic gesture subscale reflects presence or absence of pointing.
Table 2 summarizes correlations among all 10 language and
gesture scores. In the sample as a whole, total gesture vocabularies were significantly correlated with word comprehension (r =
.57, p < .001 ) and word production (r = .54, p < .00 i). However,
we also found evidence for a dissociation between gestural production and expressive language in children at the earliest stages
of language learning. In the subsample of children with expres-
sive vocabularies of 10 words or less, the average number of
gestures reported was 31.3 (with a range from 13 to 47)--only
slightly smaller than the average for the sample as a whole.
These children had an average of 2.5 words for common objects
(with a range from 0 to 6), but they reportedly produced 19.7
object gestures (with a range from 6 to 35). There was not one
case of a child in this developmental range who produced more
object words than object gestures--despite the fact that the total list of possible words that the parent could have checked
was considerably longer than the list of possible gestures. By
contrast, reported language did sometimes surge ahead of gesture among the more precocious talkers in the sample (although, of course, here the difference in length in the two checklists presents a possible confound).
These results are consonant with the evidence we reviewed
earlier on similarities and dissimilarities between language and
gesture. Recognitory gestures and vocal names usually appear
around the same time in development, but the vocal modality
sometimes lags behind. However, despite these asynchronies in
total scores, total scores for gesture and for language were still
correlated in the subsample of children with smaller vocabularies (r = .56, p < .01).
A principal components factor analysis was conducted on the
seven word and gesture subscales, to determine whether comprehension and production do indeed form two separate factors, and to examine factor loadings for each gesture type. Age
in days was also included in the factor analysis, to determine
whether and to what extent the resulting factors are defined by
effects of maturation and/or experience. Factor loadings after
rotation are summarized in Table 3.
Results provide support for a two-factor model defined by
word comprehension and word production, respectively. The
principal components analysis yielded two factors with eigenvalues greater than 1.0. The first factor (accounting for 47.5%
of the variance, eigenvalue = 3.80) was defined primarily by
word production (both nominals and nonnominals). The second factor (accounting for 13.7% of the variance, eigenvalue
1.09) was defined primarily by word comprehension (again,
both nominal and nonnominal). Although age did load positively on both factors, its contribution was relatively small; neither of the factors was in any sense defined by variance in age.
Note, however, that age loadings were somewhat higher on the
Production factor (.45) than on the Comprehension factor (.27).
Because the factor analysis maximizes orthogonality between
factors, we must interpret this comprehension/production dissociation as follows. Factor 2 reflects that portion of the variance in word comprehension that is partially dissociable from
the child's current level of expressive language; we might consider this the conceptual content of the child's current symbol
system. Conversely, Factor 1 reflects that portion of the variance
in word production that is partially independent from the
child's level of language comprehension. Interpretation of this
factor is less obvious, but it may involve a tendency toward rote
imitation or motor abilities that are partially independent of
conceptual understanding (cf. Bates et al., 1988). This distinction will be helpful in interpreting the gestural loadings on each
factor.
Gestural routines load significantly on the Production factor
LANGUAGE AND GESTURE
1009
Table 2
Correlations Between Language Comprehension, Language Production, GestureProduction, and Age
Comprehension
Measure
Age
Comprehension
Total words
Nouns
Nonnominals
Production
Total words
Nouns
Nonnominals
Gestures
Total
Object gestures
Deictic gestures
Gestural routines
*p<.05.
**p<.01.
Age
Total
w o r d s Nouns
Production
Nonnominals
Total
words
Nouns
.24*
.29**
.17
-.96***
.97***
-.87***
--
.35***
.32**
.35***
.45***
.42***
.44***
.43***
.41"**
.41"**
.43***
.39***
.44***
-.97***
.94***
.82***
.44***
.48***
.23*
.13
.57***
.56***
.28**
.32***
.50***
.56***
.28**
.28**
.54***
.52***
.26**
.35***
.54***
.46***
.23*
.47***
.52***
.46***
.23*
.43***
Gestures
Nonnominals
Total
.50***
.42***
.21"
.50***
-.95***
.41"**
.68***
Object
Deictic
Gestural
gestures gestures routines
-.32**
.43***
-.26**
***p<.001.
(.64), with negligible loadings on the factor defined by word
comprehension (.22). Deictic gestures load primarily on the
Comprehension factor (.40), contributing less to the factor defined primarily by word production (.24). Finally, object-associated gestures load significantly on both factors, although they
contribute more variance to the factor defined by comprehension (.63 vs..47). This pattern makes sense if we keep the aforementioned interpretation of the two factors in mind. Factor 1
reflects aspects of performance that are at least partially independent of conceptualization: social motivation, motor development, imitative abilities, or some combination. This may explain why gestural routines like "bye-bye" and "pattycake"
load most heavily on Factor 1. Factor 2 is best regarded as reflecting symbolic/conceptual content. Object gestures and deictic gestures load on this factor because they reflect the child's
understanding of the object world and his or her interest in communicating that understanding.
At least two different mechanisms appear to be responsible
for the patterns of variation observed here, associated with the
observed dissociation between word comprehension and word
production. Object-associated gestures (i.e., enactive names)
are indeed associated with language comprehension, in line
with the comprehension mediation hypothesis. However, some
of the variance in enactive naming is also independently associated with language production, suggesting that several different
processes may be at work. In the next study, we will focus entirely on the relationship between object-associated gestures
and vocal names, in an experiment designed to test the comprehension mediation hypothesis directly. Because recognitory
gestures will be assessed under several different conditions, it
may be easier to pull apart the separate contributions of comprehension and production.
First, we began our session by taping a brief sample of spontaneous play with realistic versions of the object concepts to be
used in the experimental study. This segment closely resembles
the conditions under which symbolic play is assessed in most
observational studies, and therefore offers some useful points of
contact and contrast between this study and other reports in the
literature. The spontaneous segment is also more similar to the
conditions under which parents usually observe their children
at play, and hence may help to illuminate the correlational results we obtained with parental reports of gestural development.
Second, we set up an experiment in which children were required to reproduce familiar gestures using a colored wooden
or plastic block as a placeholder for the referent object. This
allowed us to assess the child's ability to produce recognitory
gestures without support from the usual referent object, an
ability that bears a stronger theoretical relationship to "true"
naming.
To address the comprehension mediation hypothesis, gestures were modeled under three conditions of verbal accompaniment: supportive language (with the object correctly named),
Study 2: Elicitation o f G e s t u r e W i t h a n d
W i t h o u t Linguistic S u p p o r t
Study 2 was designed to meet several of the theoretical and
empirical issues raised in the language/gesture literature.
Table 3
Factor Loadings in Study 1
Measure
Word comprehension
Nominals
Nonnominals
Word production
Nominals
Nonnominals
Gestures
Object gestures
Deictic gestures
Gestural routines
Age
Communalities
Factor 1
Factor 2
.8811
.8400
.1851
.1903
.9202
.8966
.7883
.8224
.8630
.8829
.2085
.2068
.6094
.2219
.4587
.2762
.4669
.2414
.6412
.4535
.6256
.4045
.2182
.2657
1010
BATES, THAL, WHITESELL, FENSON, OAKES
contradictory language (with an object appropriate to a different gesture named), and neutral language (with no specific reference to either the object or the action). This allowed us to
assess systematically the extent to which a given child's p r o d u c tion o f object-associated gestures was affected by c o m p r e h e n sion o f the associated words.
Finally, because we wanted to assess gestural p e r f o r m a n c e
under precisely controlled i n p u t conditions, we relied primarily
on elicitation o f gestures instead o f spontaneous production.
Fenson and Ramsay (1981) have shown that elicitation procedures yield earlier and possibly m o r e reliable estimates o f symbolic play, while following the same developmental sequence observed in spontaneous symbolic play (Fenson & Ramsay, 1981).
Method
Subjects. The sample for Study 2 was composed of 41 children (20
boys and 21 girls) between 13 and 15 months of age. There were 14
children at 13 months of age, 13 at 14 months, and 14 at 15 months;
both sexes were equally represented within each age bracket. Slightly
more than one half of the subjects were only children. Most of the rest
had one older sibling, and a few had two older siblings. All of the children were acquiring English as their first language.
Children and their parents were recruited through two sources: a university subject pool drawn from state birth records, and advertisements
in local newspapers. Those parents who agreed to participate in the
present study were all from middle-class backgrounds (based on reported occupation and/or place of residence). Parents filled out the Language and Gesture Inventory, described in Study 1, approximately 1
week before they brought their children to our laboratories to participate in the experimental study. Questionnaires were scored after the
laboratory session was complete; thus, experimenters were blind to the
child's reported levels of ability in language or gesture at the time testing
took place.
Procedure. On the basis of previous studies of the objects, words, and
gestures that first appear in the repertoires of infants in their second
year (Appendix), 12 object concepts were selected for the elicited gesture task. Our selection of these 12 object concepts was guided not only
by their relative frequency in early childhood, but also by the extent to
which associated gestures could be recognized and distinguished without the object in hand. For example, even though "cup" and "bottle"
are both very likely to occur in the repertoires of l-year-old children, the
gestures associated with each object (i.e., drinking) would be difficult to
distinguish in an experimental context. Hence, we chose only one object
concept associated with a drinking gesture (i.e., cup), only one vehicle
concept associated with a moving-back-and-forthgesture (i.e., car), and
so forth.
Testing was conducted in a room equipped with a video camera, a
low couch, and a table with four chairs. When the children arrived at the
laboratory, they first spent 15-20 min seated next to the experimenter,
directly in front of the couch on which their mothers were seated. The
experimenter presented a standard set of toys to the child, one at a time,
and allowed the child to play with each item for up to 2 min. The toys
were always presented in a context of naturalistic free play. All of the
toys were realistic versions of the 12 object concepts that would be the
focus of the experiment itself. If the child spontaneously produced a
conventional action usually associated with the target object within the
2 min (e.g., drink from the cup, push the car back and forth across the
floor), the next item was presented. Most children produced at least
some conventional gestures within the first 30 s, and a new object was
presented. For most children, this whole sequence usually lasted about
15-20 rain. If the child did not spontaneously carry out the associated
action, that action was modeled once by the experimenter with the object in hand. After modeling, the object was handed back to the child
for approximately 30 s. If the child still did not produce the gesture, the
modeling procedure was repeated one more time.
During the free play segment, no attempt was made to control or
direct mother-child interaction, and all three participants usually interacted with the toys presented. Thus, it was more naturalistic than the
highly structured gesture elicitation experiment described in the following paragraph. In addition, the extent to which adults named or described toys during the interaction was not controlled in free play.
The elicited experimental condition took place at a table removed
from the play area (with all the realistic toys placed out of sight). The
child was seated across the table from the experimenter, on his or her
mother's lap. Nine target gestures were modeled for each child, one at a
time, using a block as a placeholder for the particular object (e.g., for
the phone gesture, the block was held to the ear as if it were a receiver;
for the cup gesture, it was tilted up to the lips). Blocks were used rather
than real toys to ensure that the presented words and gestures were the
only sources of information given to the child concerning the object
concepts. They were varied with each imitation by choosing a different
block from a pool of l0 that had different shapes, sizes, and colors. One
half were made of wood and the other half of plastic. Each gesture was
modeled in one of three language support conditions: supportive, neutral, or contradictory. Thus, each child was asked to imitate three gestures presented with supportive language, three presented with neutral
language, and three presented with contradictory language. In the supportive condition, the experimenter named the object referred to by the
gesture (e.g., "Look! This is a cup!"). In the neutral condition she made
a nonspecific comment like, "Look! Watch this!" In the contradictory
condition she misnamed the object (e.g., by saying, "Look! This is a
shoe," while modeling the phone gesture). In order to avoid contamination or interference between trials, the three inappropriate words used
in the contradictory condition were always chosen from the set of three
familiar object concepts that were not presented as gestures at any
point. Hence, no concept was ever repeated in word or gesture to any
one child. All of the concepts were rotated randomly so that each occurred in every condition (supportive, neutral, contradictory) an equal
number of times.
Of the four places in which each object concept could appear--as
gestures in the three word conditions or as a word in the contradictory
condition--presentationswere arranged so that each concept appeared
an equal number of times in each condition, to control for the possible
unique effects of any one concept. The order in which the nine gestures
were presented was uniquely and randomly assigned for each child.
A practice trial, using a camera gesture, preceded presentation of the
test items. The experimenter held a block as though it were a camera
and pretended to take the child's picture. Then she gave the block to the
child and said, "Can you take my picture?" The purpose of the trial was
to introduce the child to the imitation game. The child was encouraged
to imitate and was praised if he or she did. After this trial, the experimenter no longer differentially reinforced imitation, but smiled and said
something supportive but unrelated to the child's behavior (e.g., "So
that's what you want to do with that block"). After each gesture was
modeled, the experimenter said, "There, your turn," and handed the
block to the child. The child was allowed a minute or two in which to
produce (or not produce) the modeled gesture. If the child did not imitate within the first minute, or if the child had clearly failed to attend to
the first model, the gesture was modeled a second time. No more than
two models were given on any trial. The parent remained with the child
throughout the session but was asked not to name any of the concepts
presented or to encourage the child in the performance of any imitations.
Data reduction and dependent measures. Performance in the free
LANGUAGE AND GESTURE
play segment and in the testing phase of the experiment was scored directly from videotapes. Each child was given two scores for the free play
segment. One score reflects the number of target gestures that were produced spontaneously, prior to the experimenter's intervention ( 1 point
each, with a possible range from 0 to 12; repetitions of the same gesture
were not counted). The other score reflects the total number of target
gestures that were produced in the free play segment as a whole (spontaneously or after the experimenter began to elicit gestures with the realistic object; 1 point each, yielding a possible range from 0 to 12 with no
points given tbr repetitions).
In the experimental phase, children received 1 point for each trial in
which they produced a recognizable imitation of the adult model. Thus,
the maximum score that could be obtained by any child was 9 points
across conditions and 3 points within each condition. An assistant
scored approximately one half of the tapes, and the experimenter scored
the other half. The experimenter also scored 10 of the videotapes scored
by the assistant. Interrater agreement for the number of imitations
scored on these 10 tapes was 99%.
In addition to the gestural scoring, we were also interested in the extent to which children repeated the experimental words and the spontaneous production of words or wordlike sounds associated with the experimental objects (including sound effects like "'vroom" or "yum").
All recognizable words or vocalizations that met these criteria were
transcribed directly from the videotapes. We did not score vocalizations
that were unrelated to the 12 target objects (e.g., "Mama;" "more:'
or "no").
On the basis of the number of words in each child's comprehension
vocabulary (as reported in the parental inventory), children were divided as evenly as possible into three comprehension level groups: low
(9-57 words, n = 14), mid (60-118 words, n = 14), and high (131-233
words, n = 13). The purpose of this between-subjects manipulation was
to provide a further test of the hypothesis that production of symbolic
gestures is a function of comprehension mediation. The comprehension
groups were used in one-way analyses of variance (ANOVAS)on gestural
performance in the free play segment, and in 3 × 3 analyses of the gesture experiment (with comprehension group as a between-subjects factor, and language input condition as a within-subjects factor).
We carried out additional analyses by dividing the children into two
groups, based on their reported production vocabularies. We used only
two groups instead of three for production because there is so much
less variance in expressive vocabulary in this age range. One half of the
sample (n = 21 ) received scores of 10 or less; this corresponds to a stage
in which vocabulary development is still quite unstable, with new words
appearing and then disappearing for weeks at a time (Vihman, 1986).
We will refer to this group as the low producers. The remaining high
producers (n = 20) all reportedly produced more than 10 words. Two
children in this group obtained scores over 50, the vocabulary boundary
that often presages a marked increase or burst in lexical development
(K. Nelson, 1973. 1981 ). On theoretical grounds, these children may be
qualitatively different from the others, but there were too few to permit
a third expressive language group. The two resulting production groups
were used as a between-subject variable in analyses of the free play segment and in 2 x 3 analyses of gestural performance across the three
experimental language input conditions.
Ideally, we would like to treat comprehension and production groups
as orthogonal factors in a complete design, yielding a 3 × 3 × 2 (Language Input Condition × Comprehension Level × Production Level)
mixed between- and within-group ANOVA. But only 4 children fell
within the requisite high-production/low-comprehension cell in such a
design, precluding meaningful analysis. This pattern is of course in line
with the comprehension/production results reported in Study 1, which
suggested that comprehension is a necessary but not sufficient condition
for the development of expressive language.
1011
Results
Free play segment. Table 4 presents descriptive statistics for
gestural performance in the free play segment, compared with
performance on the same 12 gestures according to parental report, for the group as a whole and for the respective comprehension and production subgroups. Parents reported that children
produced an average o f 7-8 of the 12 target gestures (with a
range from 3 to 11). Rates of spontaneous production were considerably lower (averaging 3,49, with a range from 0 to 8); however, with the experimenter's intervention, total rates of gestural
production in the free play segment reached an average of 6.09,
with a range from 1 to 10. Hence, the parents' estimates were
roughly comparable with rates observed in the laboratory, when
performance was supported by the object (realistic toys) and by
adult vocal and gestural input.
As can be seen from Table 4, the n u m b e r of gestures in the
parental report and gestural performance in the free play segment both increase as a function o f reported comprehension
level. A one-way ANOVA by comprehension group was significant for reported gesture, F(2, 37) = 9.64, p < .001. In this
analysis, the linear component was quite reliable, F ( I , 37) =
17.78, p < .001 ; there was no significant nonlinear component
(in contrast with some of the experimental analyses presented
later). A similar one-way analysis on spontaneous gesture also
reached significance, F(2, 37) = 4.10, p < .03; here too there
was a significant linear component, F(1, 37) = 7.88, p < .01,
but no nonlinear trend. Finally, a one-way analysis of total gesture in the free play segment also yielded a significant comprehension group difference, F(2, 37) = 6.44, p < .01; the linear
component was also significant, k~ 1, 37) = 9.62, p < .01, and
the nonlinear component just missed significance, F(1, 37) =
3.26, p < .08. The nonlinear trend reflects a leveling o f f o f g e s tural performance between the middle and high comprehenders
(an average of 4.64 gestures by the low group, compared with
6.86 and 6.85 by the middle and high groups, respectively).
Hence, the comprehension group difference appears to reach
ceiling when gesture is assessed under conditions o f maximal
support (with realistic objects, and with adult gestural and vocal
cuing).
In contrast with these orderly relations between language
comprehension and gesture, differences in gestural production
were small for the two production groups: 7.00 versus 8.47 for
reported gestures (tow vs. high, respectively), 3.09 versus 3.90
for spontaneous gestures, and 5.81 versus 6.40 for total gestures
in the free play segment. The high/low production difference
was significant for reported gesture, by a one-tailed t test (p <
.03). Analogous t tests o f the differences for spontaneous and
total gesture in free play did not reach significance.
To summarize, our findings for observed gesture in a relatively naturalistic situation c o m p l e m e n t results from Study 1.
Spontaneous production of these 12 object-associated gestures
tends to increase as a function of language comprehension level.
Differences are considerably smaller when children are grouped
according to language production levels. However, the child's
reported level o f gestural production varies as a function o f both
comprehension and p r o d u c t i o n - - i n line with the results o f
1012
BATES, THAL, WHITESELL, FENSON, OAKES
Table 4
Gesture Scores as a Function of Re/9orted Language Level
Comprehension groups
Gestures with
12 target objects
Parental report
M
SD
Range
Warmup
Spontaneous
M
SD
Range
Totals
M
SD
Range
Experiment
Totals
M
SD
Range
Production groups
Totals
Low
Medium
High
Low
High
7.70
2.34
3-11
5.93
2.09
3-9
8.21
2.08
4-I 1
9.17
1.58
6-11
7.00
2.49
3-11
8.47
1.95
4-11
3.49
1.80
0-8
2.50
1.74
0-7
3.71
1.63
2-7
4.31
1.65
2-8
3.09
1.81
0-8
3.90
1.74
1-7
6.09
2.11
1-10
4.64
1.59
1-8
6.86
1.83
4-10
6.85
2.15
4-10
5.81
2.16
3-10
6.40
2.06
1-10
6.61
3.66
1-14
4.71
3.33
2-I 1
8.57
3.56
1-14
6.54
3.18
2-12
5.00
2.96
2-11
8.30
3.61
1-14
Study 1, in which reported gesture loaded significantly on both
the Comprehension and the Production factors.
Experimental results: Analyses by comprehension group.
Elicited gesture scores in the experimental segment were entered into a 3 x 3 mixed design ANOVA,l with the three comprehension groupings as a between-subjects variable and languagesupport condition as a within-subjects variable. This analysis
revealed a main effect of language support, F(2, 76) = 3.91,/9 <
.03; a main effect of comprehension group, F(2, 38) = 3.29,
/9 < .05; and a Group Language-Support interaction, F(4, 76) =
3.2 l, p < .02.
With regard to the main effect for language support, children
were most likely to imitate in the supportive condition and least
likely to imitate in the contradictory condition. Thus, infants
between the ages of 13 and 15 months were more likely to imitate a familiar gesture with a block if the accompanying linguistic label or name corresponded to the meaning of the gesture.
This suggests that words and gestures do map onto the same
common object meanings, even in the early stages of symbol
development. It also provides some support for the comprehension mediation hypothesis--although, as we shall see, the support is mixed.
The comprehension group main effect was significant, but in
a surprising direction. Mean imitation scores were 3.42 for the
low comprehenders, 5.07 for the middle comprehenders, and
4.0 for the high comprehenders. A one-way ANOVA over the
three comprehension levels yielded a significant nonlinear component, F(2, 76) = 5.74,19 < .03, but no significant linear term.
Thus, although linguistic input did affect the probability of gestural imitation, the child's propensity to imitate was not a simple, linear function of comprehension vocabulary.
This nonlinear relation is also evident in the Group × Language Support interaction illustrated in Figure 2. Tukey post
hoc tests (/9 < .05) indicate that there was no significant differ-
ence among linguistic support conditions in the low comprebenders. In the middle group, imitation was significantlyhigher
in the supportive condition; the neutral and contradictory conditions did not differ from one another. Hence, it is fair to say
that language can help gestural production in this group, but
contradictory linguistic input is not powerful enough to overwhelm the child's interpretation of the gestural model. In the
high comprehenders, supportive language produced significantly higher performance than contradictory language; performance in the neutral language condition fell halfway between
(and did not differ significantly from the other two conditions).
In this last group, supportive language apparently cannot push
performance levels much beyond the levels observed with no
linguistic information at all; however, when forced to choose
between the adult's action and the same adult's accompanying
speech, the high comprehenders are prone to place their trust
in linguistic cues.
Notice, nevertheless, that the average number of imitations
produced decreases from the middle to the high comprehending groups, in all three conditions. As several investigators have
suggested, imitation may be more likely overall in children who
are at the midpoint of a developmental milestone--suggesting
that these children have a greater need to "sketch out" their new
abilities on the plane of action (Harnick, 1978; Kagan, 198 l;
Shore et al., in press). In order to obtain a clearer measure of
the effects of the independent variables, we decided to remove
variance due to overall propensity to imitate from the calculations.
We recalculated responses in terms of the proportion of the
For the analysis of variance (ANOVA)reported here, we used the
repeated measures ANOVAoption in the Statistical Package for the Social Sciences multivariate analysis of variance (SPSSx MANOVA)program in which tests for sphericity are taken into account.
LANGUAGE AND GESTURE
_o
2
supportive
IE
neutral
i
contradictory
COMPREHENSION GROUP
Figure 2. Effects of linguistic input on total number of gestures imitated
for 13-15-month-old children with low, medium, and high levels of
word comprehension.
total imitations produced by each child within each languagesupport condition. Thus, we divided the number o f trials on
which a child produced a scorable imitation in each languagesupport condition by the total number of imitations produced
across all three conditions. These percentage scores were then
analyzed in a 3 × 3 mixed design ANOVA (see Footnote 1) with
comprehension group and level of language-support as independent variables (see Figure 3). In this analysis, we again found
a significant Comprehension Group x Language-Support Level
interaction, F(4, 76) = 4.15, p < .004, illustrated in Figure 3.
To explore the interaction further, we first carried out three separate simple effects analyses by comprehension level, within
each of the three respective language input conditions. All three
comprehension level effects were significant, but they differed
markedly in shape. The percentage of gestures occurring in the
supportive condition showed a linear increase over comprehension levels, F(2, 76) = 8.43, p < .01, with the nonlinear component just missing significance, F(2, 76) = 3.62, p < .07. The
proportion of gestures occurring in the contradictory condition
showed a linear decrease with comprehension level, F(2, 76) =
12.84, p < .001, with no significant nonlinear component. Finally, the proportion of imitations occurring in the neutral condition shows no significant effects at all, linear or nonlinear.
These effects are clarified further by Tukey post hoc tests (p <
.05) within each comprehension level. Within the low comprehenders, there were again no significant differences as a function
of linguistic input; the trend is, if anything, in the opposite direction from the one observed in children with larger comprehension vocabularies (i.e., fewer imitations in the supportive
1013
condition). Within the midlevel comprehenders, gestural imitations were much more likely to occur within the supportive
conditions; neutral and contradictory language conditions did
not differ. This is identical to the effects reported for raw imitation in the middle group. Within the high comprehenders, gestural imitations were significantly more likely to occur in the
supportive and the neutral conditions than in the condition with
contradictory language; however, there was no significant
difference between supportive language and neutral language.
Because the 14 low comprehenders produced fewer gestures
overall, we might ask whether their inability to use linguistic
input reflects anything more than a floor effect. This was not
the case. Although their total output was low ( M = 3.42, with a
range from 1 to 7 points, S D = 1.95), the data do not violate
assumptions of normality. With 14 low comprehenders in three
within-subject conditions, there were 42 scores in all; fewer than
25% of the scores were zeroes within any cell of the design. We
conclude that the absence of a language input effect in this
group is a straightforward reflection of the fact that they do not
comprehend language well enough to exploit it as an aid to gestural production. And yet, they still produced many of the target gestures. Some factor other than comprehension mediation
must account for their behavior.
These results suggest that the effect of linguistic input on gestural imitation changes over time. Gestural imitation is unaffected by accompanying language at the earliest stages oflexical development. Language input can play a supportive role at
the middle stages, but these children apparently ignored the linguistic cue if it conflicted with the modeled gesture. Finally, sup-
50
supportive
40
neutral
30
20'
contradictory
10
i
COMPREHENSION GROUP
Figure 3. Effects of linguistic input on the proportion of gestures imitated for 13-15-month-old children with low, medium, and high levels
of word comprehension.
1014
BATES, THAL, WHITESELL, FENSON, OAKES
portive language is apparently unnecessary or not particularly
helpful for children at the highest levels of comprehension; however, when there is a conflict between verbal and gestural input,
the performance of high comprehenders is negatively affected.
So far it appears that the parallelism hypothesis and the comprehension mediation hypothesis are both correct, but at
different points in development and in different children. To
obtain further evidence regarding the relation between language
abilities and gestural performance, we turn to analyses using
language production as a grouping variable.
Analyses by production group. The relative independence of
comprehension and production in our parental reports suggests
that these two aspects of language may each bear a somewhat
different relation to symbolic gesture (cf. Bates et al., 1980). To
test this hypothesis, we repeated the Language-Level X Linguistic-Input Condition analysis, using expressive rather than receptive vocabulary to define the groups (see Footnote 1).
Not surprisingly, we again obtained a significant main effect
of input condition, F(2, 78) = 3.32, p < .05. There was also a
main effect of expressive vocabulary, F(1, 39) = 6.18, p < .02,
with more gestural imitations produced by the high vocabulary
group. However, in contrast with analyses using comprehension
level, there was no significant interaction between vocabulary
level and input condition, F(2, 78) = .25, ns.
Because the propensity for high vocabulary children to produce more modeled gestures is equivalent across linguistic-input conditions, we can conclude that their advantage in gestural
output reflects more than receptive language ability. Comprehension mediation does occur, but it is not sufficient to account
for the developmental relation between language and symbolic
gesture at the one-word stage. There may be an additional, specific relation between language production and gesture.
Words and sound effects during the experiment. Meaningful
speech was rare in this situation, imitative or not. Only 11 of
the 41 children produced any meaningful speech at all across
the course of the experiment. Seven children produced one or
more sound effects associated with the gesture during their own
production (e.g., "whee" with the airplane gesture, "vroom"
with the car gesture, drinking noises with the cup gesture, eating
noises with the spoon gesture, a marked sniffing noise with the
gesture of smelling a flower). Because these sound effects were
not produced by the experimenter, such vocalizations must be
taken to mean that the child recognized the adult model, adding
the relevant sounds in his or her own rendition. There were only
seven real words produced in all, by 4 of the 41 children. One
child produced three words ("cup'; "baby;' and "plane"), another produced two words ("flower" and "dog"), and 2 other
children each produced only one word ("doggie" and "'baby;'
respectively). Broken down by language condition, 3 followed a
supportive model, 4 followed a contradictory model, and none
occurred in the neutral condition (where there was no adult
model to imitate). Because all seven words matched the adult
model, they may constitute instances of vocal imitation.
Although vocal productions were infrequent in this experiment, it is interesting that the vocalization data bear a U-shaped
relation to comprehension level: 3 of the 14 low comprehenders
produced sound effects, 7 of the 14 middle comprehenders produced sound effects or words, or both, and only 1 of the high
comprehenders produced any vocalization at all (the single
word "baby"). With regard to expressive language level (as reported by the parent), 7 of the I 1 children who produced meaningful speech during the experiment had expressive vocabularies of more than 10 words. This difference is in the expected
direction, but is not overwhelming.
To summarize, there are similarities between observed vocalizations and observed gestures within the experimental task; vocalizations are somewhat more common for children with high
reported vocabularies, but they bear a U-shaped relationship
to comprehension level. Because there were too few instances
of meaningful speech to justify statistical analyses, these results
are merely suggestive. We will return to them later in a discussion of the production mediation account of language/gesture
relations.
Factor analysis of parental report and laboratory measures.
Principal component analysis with varimax rotation was applied to a selected subset of measures in Study 2 to explore relation between laboratory and parental report measures of language and gesture when cognitive content was held constant
(i.e., when all measures pertained to the same set of 12 object
concepts). Five measures were selected: scores from the parental report for number of experimental words comprehended,
number of experimental words produced, and number of experimental gestures produced; scores from the laboratory session for number of target gestures produced spontaneously in
free play; and number of target gestures imitated in the neutral
language condition. We focused on the neutral language condition only, in order to obtain a measure of elicited gesture that
was maximally independent of language input. Age in days was
also included as a variable in the analysis. Table 5 summarizes
raw correlations among all six variables; Table 6 summarizes
the factor loadings that emerged after rotation.
Because the variables entered into this factor analysis are
different in many respects from the variables used in the factor
analysis in Study 1, we should not expect identical results. However, the results obtained in these two different principal component analyses are quite compatible. Once again, two factors
with eigenvalues greater than 1.0 emerged, with factor loadings
summarized in Table 5. The two factors once again reflect a
dissociation between word comprehension and word production, analogous in several respects to the two factors that
emerged in Study 1. Factor 1 has an eigenvalue of 2.61, and
accounts for 43.5% of the variance. Word comprehension loads
on this factor at + .85. Factor 2 has an eigenvalue of 1.15, and
accounts for an additional 19.3% of the variance. Word production loads on this factor at + .53. Age loads positively on both
factors, but it is not the defining variable in either case. As in
Study 1, age loads more strongly on the Production factor (.69)
than the Comprehension factor (.39).
Of greatest interest here are the factor loadings for the three
respective measures of object-associated gesture. Two measures
load highly on the Comprehension f a c t o r i t h e number of spontaneous gestures observed in the free play segment (.81) and the
number of experimental gestures reported by the parents (.51).
Two gestural measures loaded highly on the Production fact o r - t h e number of experimental gestures imitated in the lan-
LANGUAGE AND GESTURE
1015
Table 5
Correlations Among a Subset of Measures From the Parental Report and the Laboratory Session
Variable
Age
Spontaneous
gestures
Words
comprehended
Words
produced
Gestures
produced
Age
Spontaneous gestures
Words comprehended
Words produced
Gestures produced
Total imitation
.41"*
.34*
.35**
.44**
.41 **
-.52***
.26*
.23
.01
-.27*
.56***
.05
-.28*
.25
.30*
Total
imitation
*p < .05. **p < .01. ***p < .001.
guage-neutral condition (.87) and the number of experimental
gestures reported by parents (.53).
There appears to be a dissociation between the spontaneous
gestures observed in a brief laboratory free play segment (with
realistic versions of the referent object) and the same gestures
elicited out of context (without supporting information from
adult language or from the referent object). Our parental report
measure appears to be sensitive to both sources of variation,
loading on both the Comprehension and the Production factors.
This replicates the findings reported in Study 1. These findings
also replicate an earlier report by Bates et al. (1980), who found
that gestural imitation in a scripted situation correlates with
language comprehension, whereas imitation of isolated gestures
out of context correlates with language production. Once again,
we must conclude that at least two mechanisms are responsible
for the relation between language and gesture in this age range,
even when we restrict our focus to symbols associated with a
common set of object concepts.
Conclusion
We began by presenting comprehension mediation and parallelism as competing explanations for the finding that language
and gesture develop together. Our results suggest that both
hypotheses are correct, accounting for different aspects of the
data.
Children in the 12-16-month age range are indeed capable
of using adult language as an aid in the acquisition and use o f
gestural schemes. Evidence in favor o f comprehension mediation was apparent in both studies. First, word comprehension
and gestural production were highly correlated--particularly
when we considered only those gestures that pointed to or represented objects. Second, the experimental study showed that
linguistic input does affect gestural performance.
However, the relation between comprehension and gesture is
quite complex, with some surprising curvilinear effects. Linguistic input had little or no effect among the low comprehenders, and yet these children did imitate some of the adult gestural
models. Children in the middle comprehension group were able
to use supportive language to enhance their performance, but
they ignored the linguistic cue when it contradicted the adult's
action. Yet another pattern occurred among the high comprebenders; contradictory language suppressed their performance,
but supportive language did not increase performance beyond
the levels displayed when the adult provided no linguistic information at all. Comprehension mediation appears to be an option that is available to 1-year-old children, developing within
the narrow age window studied here; but it is also apparent that
comprehension mediation is not obligatory, nor is it equally
effective at every stage of development.
Another curious pattern involves the U-shaped relation between comprehension level and total gestural imitation scores.
Imitation was most likely among children in the middle comprehension group. This finding parallels a report by Shore et al.
(in press) for a separate sample of 28-month-old infants, and it
is compatible with a theory of imitation proposed by Kagan
(1981). Specifically, Kagan proposed that imitation is most
likely when the model lies just outside the child's current level
of competence, that is, when it is neither too easy nor too hard
(see also K. E. Nelson & Nelson, 1978). Such U-shaped functions are chastening because they alert against the assumptions
oflinearity that lie behind the linear correlational statistics used
in most developmental studies--including our own--(see Bates
et al., 1988, for further discussion ofthis point.)
Table 6
Factor Loadings in Study 2
Measure
Parental report
Experimental
words
comprehended
Experimental
words
produced
Experimental
gestures
produced
Laboratory
Gestures produced
spontaneously
in free-play
condition
Gestures imitated
in languageneutral
condition
Age
Communalities
Factor 1
Factor 2
.7523
.8510
.1675
.3770
.2994
.5361
.5488
.5091
.5381
.6585
.8088
.0656
.7987
.6323
-.2036
.3926
.8705
.6915
1016
BATES, THAL, WHITESELL, FENSON, OAKES
Although the relation between comprehension and gesture is
quite robust, appearing in both the observational and parental
report data, some version of the parallelism hypothesis is still
needed to account for the independent relation that holds between gesture and expressive language. This was made particularly clear in the contrast between the respective comprehension
and production group analyses. Comprehension level interacted with the language support manipulation; production level
did not. Children with high expressive vocabularies produced
more gestural imitations in the experiment in all three language
support conditions.
The same conclusion falls out of the factor analyses in Study
1 and Study 2. Two significant factors emerged in both these
studies, reflecting a dissociation between comprehension and
production in this age range. Because the factor scores are made
up of different ingredients in Study 1 and Study 2, the two analyses may be telling slightly different stories. But the similarities
are interesting, and they offer us some possible directions for
future research.
In both studies, we argued that the Comprehension factor reflects developments in symbolic/conceptual content. In Study
1, this factor was associated with deictic gestures and enactive
naming (i.e., gestures that are used to recognize, explore, and
communicate about a world of familiar objects). In Study 2, the
factor was again associated with measures of enactive naming,
in free play and parental report. These measures all tap into the
spontaneous use of gesture under relatively naturalistic condit i o n s - w i t h realistic objects and with supporting gestural and
vocal activity provided by the parent.
Both studies also yielded a second factor, associated with
variance in word production. In Study l, this factor was associated with routines that make reference to nothing but themselves (e.g., "pattycake," "bye-bye"), although it also received
some contribution from object-associated gestures. In Study 2,
this factor was associated with "decontextualized" gesture--action schemes produced without support from the target object,
independent of the adult's accompanying narration. Parental
report loaded on both factors, in both studies, suggesting that
the gesture inventory picks up variance in the production of
recognitory gestures under a wide variety of eliciting conditions.
To interpret these dissociations correctly, we must remember
that factor analysis maximizes orthogonality. Hence, the Production factor in both studies reflects those aspects of expressive
language that are not shared by comprehension--presumably
some kind of performance variable. There are (at least) three
possible interpretations of this independent link between word
production and aspects of production in the gestural domain:
1. Imitativeness. In the linguistic domain, some children
may have receptive vocabularies far beyond their expressive
speech because they are "cautious," unwilling to attempt reproduction of adult speech until they are absolutely certain what
it means. The same children might be reluctant to imitate an
adult gesture out of context, without additional cues from language or the referent object. Other children may be willing to
imitate an adult model in either modality, whether they understand it or not. Differences in "propensity to imitate" might in
turn reflect several different factors: temperament (e.g., shy-
ness), aspects of the mother-child relationship (e.g., whether or
not the parents encourage and reward imitation), and cognitive
style (i.e., whatever factors are responsible for the curvilinear
relation between comprehension and imitation observed in
Study 2).
2. Conceptual progress. Production of both gestures and
words for common objects may be based in part on advances in
symbolic understanding above and beyond the level required for
comprehension--a different cognitive system or, alternatively, a
higher level of sophistication within the same symbolic system.
3. Motor planning. Progress in gestural and vocal production
in the 12-16-month age range may be based on some kind of
shared articulatory/motor development for which comprehension is a necessary but not sufficient condition.
We cannot decide among these interpretations at this time.
They are not mutually exclusive, and they are all compatible
with the parallelism hypothesis. However, one final counter to
parallelism could still be entertained: Perhaps children with
high vocabularies perform better on abstract symbolic play
tasks because they are talking to themselves. That is, the infant
may use covert production (rather than comprehension) as an
aid to his or her gestural activity. We certainly cannot rule out
this possibility because it may account for some of the advantage that the children with larger expressive vocabularies displayed in our task.
However, there are reasons to believe that covert speech plays
a minimal role in this age range. First, very few children produced any overt object labels in this experimental situation.
Second, descriptive statistics from the parental report and observations in the laboratory both suggest that gestural production is developing ahead of expressive language. This fact makes
it unlikely that children produce gestures by first retrieving an
associated word, using that covert or overt word as a guide to
gestural production. Third, children at the earliest stages of
symbol development tend not to have a large number of overlapping words and gestures in their repertoires (Volterra et al.,
1979). If the child has a word for an object, he or she is slightly
less likely to have a corresponding gesture; conversely, if the
child has a gesture for a given object, he or she is initially somewhat less likely to acquire or make use of the corresponding
word. This trend is compatible with an argument first raised in
the bilingualism literature by Volterra and Taeschner (1977):
Children in the earliest stages of lexical development tend to
avoid assigning two words to the same referent, so that words
from their respective languages lie in complementary distribution with few if any translatable pairs. The rule seems to be
"one referent, one symbol." Slobin (1985) also reported this
effort to keep things simple in the early stages of grammatical
development. He described it as the principle of "one form, one
function." Taken together, these findings suggest that covert linguistic production is probably not going on while a "gestural
name" is being produced, at least not in the early stages of symbol development.
At some point in development, it is both possible and likely
that language "wraps back around" to provide support for symbolic play activity, through comprehension mediation (which
we have demonstrated), and possibly through covert or overt
language production (a point for further investigation--cf. Vy-
LANGUAGE AND GESTURE
gotsky, 1967). There is, nevertheless, still ample evidence for
some form o f parallelism in the development o f language and
gesture, This finding is important for those practitioners who
would like to use the child's level o f functioning in a nonlinguistic domain as an aid in the interpretation o f language delays
(Leonard, 1988; Rice, 1983; Thai & Bates, 1988a, 1988b).
Given the complexity of our findings, such assessments must
be carried out with extreme caution. Results may vary, depending on the conditions under which gesture is assessed (i.e., with
or without perceptual support from the object, and with or
without either gestural or vocal cuing from an adult). Also, results will most certainly vary depending on whether the investigator is interested in receptive or expressive language development. As we have seen, comprehension and production o f language can be dissociated to a remarkable degree in normally
developing children (Bates et al., 1987) and in children who are
delayed or impaired in language (Tallal, 1987; T h a l & Bates,
1988a). Our results suggest that the same dissociation is reflected in the relation between language and gesture. There are
at present no convincing explanations for this dissociation inside or outside o f the language domain. However, it is possible
that further comparisons o f language and gesture may help to
solve the problem.
References
Acredolo, L., & Goodwyn, S. (1988). Symbolic gesturing in normal infants. Child Development. 59, 450-466.
Bates, E., Beeghly, M, Bretherton, I., Harris, C., Marchman, V., McNew, S., Oakes, L., O'Connell, B., Reznick, S., Shore, C, Snyder, L.,
Thai, D., Volterra, V., & Whitesell, K. (1986). The Language and
Gesture Inventon: Unpublished checklist. University of California,
San Diego.
Bates, E,, Benigni, L, Brelherton, I., Camaioni, L., & Volterra, V.
(1977). From gesture to the first word. In I. M. Lewis & L. Rosenblum
(Eds.), ln[eraction, con versation and the development ~flanguage (pp.
247-308). New York: Wiley.
Bates, E., Benigni, L., Bretherton, I., Camaioni, L., & Volterra, V.
(1979). The emergence of symbols: Cognition and communication in
inlancy. New York: Academic Press.
Bates, E., Bretherton,' I,, Shore, C., & McNe~; S. (1983). Names, gestures, and objects: Symbolization in infancy and aphasia. In K. Nelson (Ed), Children ~' language: Volume IV (pp. 59-t24), Hillsdale,
N J: Erlbaum,
Bates, E., Bretherton, I.. & Snyder, L. (1988). Fromfirst ~z)rds togrammar. Individual diff~,rences and dissociable mechanisms. New ~brk:
Cambridge University Press.
Bates. E., Bretherton, I., Snyder, L., Shore, L., & Volterra, V. (1980).
Gestural and vocal symbols at 13 months. Merrill-Palmer Quarterly,
26. 407-423.
Bates, E., O'Connell, B, & Shore, C. (1987). Language and communication in infancy. In J. Osofsky (Ed.), ttandbook ofit~mt development
(2nd ed., pp. 149-203). New York: Wiley.
Bates, E,, & Snyder, L. (1987). The cognitive hypothesis in language
development. In I. Uzgiris & J McV. Hunt (Eds), Research with scales
~[ps.l~'hotogical development in iq[dm3~(pp. 168-206). ChampaignUrbana: University of lllinois Press.
Bates, E., Snyder, L., Bretherton, I., & Volterra, V. (1979). The emergence of symbols in language and action: Similarities and differences.
Papers and Reports on Child Language Development, 17, 106-118.
Bates, E.. & Volterra, V. (1984). On the invention of language: An alter-
1017
native view (Commentary on S. Golden-Meadow & C. Mylander).
Monographs of the Society for Research in Child Development, 49,
(3-4, Serial No. 207),
Bauer, P., & Shore, C. (1986). Nonlinguistic concomitants ofst)~istic
differences in early muttiword speech. Unpublished manuscript.
Beeghly, M., & Cicchetti, D. (1987). An organizational approach to
symbolic development in children with Down Syndrome. In D. Cicchetti & M. Beeghly (Eds), Atypical symbolic development (pp. 5-30).
San Francisco: Jossey-Bass.
Bretherton, I. (Ed.). (1984). Symbolic play." The development of social
understanding. New York: Academic Press.
Bretherton, I., & Bates, E. (1984). The development of representation
from 10 to 28 months: Differential stability of language and symbolic
pla?y: In R. Erode & R. Harmon (Eds.), Continu#ies and discontinuities in development (pp. 229-262). New York: Plenum Press.
Bretherton, I., Bates, E., Benigni, L., Camaioni, L., & Volterra, V.
(t 979). Relationships between cognition, communication and quality
of attachment. In E. Bates, L. Benigni, I. Bretherton, L. Camaioni,
& V. Volterra (Eds.), The emergence of symbols: Cognition and communication in inJancy (pp. 223-270). New York: Academic Press.
Bretherton, l., O'Connell, B., Shore, C., & Bates, E. (1984). The effect
of contextual variation on symbolic play: Development from 20 to 28
months. In I. Bretherton (Ed.), Symbolic play: The development of
social understanding (pp. 271-298). New York: Academic Press.
Chomsky, N. (1965). Aspects of a theory of syntax. Cambridge, MA:
MIT Press.
Chomsky, N. (1980). Rules and representations. Brain and Behavioral
Sciences. 3. 1-61.
Corrigan, R. (1978). Language development as related to stage 6 object
permanence development. Journal of Child Language, 5, 173-189.
Curcio, E ( 1977, October). ,4 study of sensorimotor functioning and
communication in mute autistic children. Paper presented at the Boston Child Language Forum, Boston, MA,
Curtiss, S., & ~hmada, J. (1978, September). Language and cognition.
Paper presented at the third annual Boston University Conference on
Language Development, Boston, MA.
Curtiss, S, Yamada, J., & Fromkin, V. (1980), ttow independent is language? On the question oftbrmal parallels between action and grammar Unpublished manuscript. University of California at Los Angeles, Department of Linguistics.
Dale, P., Bates, E.. Reznick, S., & Morisset, C. (1989). The validity of a
parent report instrument of child language at 20 months. Journal of
(Tlild Language. 16. 239-250.
Dale, P.. & Robinson, N. (1988, April). Linguisticprecoeity, cognition,
and development. Paper presented at the Sixth International Conference on Infant Studies, Washington, DC.
Elder, J., & Pederson, D. (1978). Preschool children's use of objects in
symbolic play. Child Development. 42, 500-505.
Epport, K. ( 1988, April) The relationship between facial affect expressiveness and language ability in children born preterm and "at risk '~
Poster presented at the Sixth International Conference on Infant
studies, Washington, DC.
Escalona, S. (I 973). Basic modes of social interaction: Their emergence
and patterning during the first two ?yearsof life. Merrill-Palmer Quarterly: 19. 205-232.
Fenson, L., & Ramsay, D. (1981). Effects of modeling action sequences
on the play of twelve, fifteen, and nineteen month old children, Child
Dev~4opment. 32, 1028-1036.
Fenson, L., Vella, D.. Flynn, D., & Thai, D. (1988). Developmental
norms /br fi)ur parental inventory reports of language and communication between 9 and 28 months. Unpublished manuscript. San Diego
State University and University of California, San Diego.
Fischer, K. (1980). A theory of cognitive development: The control and
1018
BATES, THAL, WHITESELL, FENSON, OAKES
construction of hierarchies of skill. Psychological Review, 87. 477526.
Fodor, J. (1983). The modularity of mind. Cambridge, MA: MIT Press.
Golden-Meadow, S., & Mylander, C. (1984). Gestural communication
in deaf children: The effects and non-effects of parental input on early
language development. Monographs of the Society for Research in
Child Development, 49, (4, Serial No. 207).
Goldfield, B. ( 1988, April). Pointing, naming and talk about toys: Referential behavior in children and mothers. Paper presented at the Sixth
International Conference on Infant Studies, Washington, DC.
Gopnik, A., & Meltzoff, A. (1986). Relations between semantic and cognitive development in the one word stage: The specificity hypothesis.
Child Development, 57, 1040-1053.
Gopnik, A., & Meltzoff, A. (1987). The development of categorization
in the second year and its relation to other cognitive and linguistic
developments. Child Development, 58, 1523-1531.
Harnick, E (1978). The relationship between ability level and task
difficulty in producing imitation in infants. ChiM Development,'49,
209-212.
Harris, E (1983). Infant cognition. In J. Campos & M. Haith (Eds.),
Handbook of clinical psychology (Vol. 2, pp. 689-782). New York:
Wiley.
Huttenlocher, J., & Higgins, E. (1978). Issues in the study of symbolic
development. In A. Collins (Ed.), Minnesota Symposia on ChildPsychology (Vol. 11, pp. 98-140). Hillsdale, N J: Erlbaum.
lnhelder, B., Lezine, I., Sinclaire, H., & Stembak, M. ( 1971). The beginnings of symbolic function. Archives de Psychologic, 41, 187-243.
Johnston, J. (1985). Cognitive prerequisites: The evidence for children
learning English. In D. Slobin, (Ed.), The cross-linguistic study of language acquisition (pp. 961-1004). Hillsdale, N J: Erlbaum.
Kagan, J. ( 1981 ). The second year." The emergence of self awareness.
Cambridge, MA: Harvard University Press.
Kamhi, A. ( 1981). Nonlinguistic symbolic and gestural abilities of language impaired and normally developing children. Journal of Speech
and Hearing Research, 24, 446-453.
Killen, M., & Uzgiris! I. (1981). Imitation of actions with objects: The
role of social meaning. Journal of Genetic Psychology. 138. 219-229.
Leonard, L. (1988). Is specific language impairment a useful construct?
In S. Rosenberg (Ed.), Advances in applied psycholinguistics. Volume
1: Disorders of first language development (pp. 1-39). New York:
Cambridge University Press.
Lezine, I. (1978). The transition from sensorimotor to earliest symbolic
function. Biological and Environmental Determinants of Early Development (Research Pub., Association for Research in Nervous and
Mental Disorders), 5 I, 221-232.
Nelson, K. (1973). Structure and strategy in learning to talk. Monographs of the Society.for Research in Child Development, 38( 1-2, Serial No. 49).
Nelson, K. ( 1981). Individual differences in language development: Implications for development and language~'Developmental Psychology.
17, 170-187.
Nelson, K. (1985). Making sense: The acquisition of shared meaning.
New York: Academic Press.
Nelson, K. E., & Nelson, K. (1978). Cognitive pendulums and their
linguistic realization. In K. E. Nelson, (Ed.), Children "slanguage. Volume I (pp. 223-285). New York: Gardner Press.
Nicolich, L. (1977). Beyond sensorimotor intelligence: Assessment of
symbolic maturity through analysis of pretend play. Merrill-Palmer
Quarterly. 23, 89-99.
O'Brien, M. ( 1988, April). Cognitive and social predictors of early language. Poster presented at the Sixth International Conference on Infant Studies, Washington, DC.
Petitto, L. (1988). "Language" in the prelinguistic child. In E Kessel
(Ed.), Development of Language and Language Researchers (pp.
187-222). Hillsdale, N J: Erlbaum.
Piaget, J. (1962). Play. dreams and imitation in childhood. New York:
Norton.
Rescorla, L. (in press). The language development survey: A screening
tool for delayed language in toddlers. Journal of Speech and Hearing
Research.
Rescorla, L., & Schwartz, E. (1988, April). Outcome of specific expressive language delay (SELD). Poster presented at the Sixth International Conference on Infant Studies, Washington, DC.
Reznick, S., & Goldsmith, S. (1989). Assessing early language: A multiple form word production checklist. Journal of Child Language, 16,
91-100.
Rice, M. (1983). Contemporary accounts of the cognition/language relationship: Implications for speech-language clinicians. Journal of
Speech and Hearing Disorders', 48, 347-359.
Roeper, T., & Williams, E. (Eds.). (1987). Parameter setting. Dordrecht,
Holland: Reidel.
Roth, E, & Clark, D. (1987). Symbolic play and social participation
abilities of language-impaired and normally developing children.
Journal of Speech and Hearing Disorders, 48, 347-359.
Shore, C., Bates, E., Bretherton, I., Beeghly, M., & O'Connell, B. (in
press). Vocal and gestural symbols: Similarities and differences from
13 to 28 months. In V. Volterra & C. Erting (Eds.), From gesture to
language in hearing and deaf children. New York: Springer-Verlag.
Shore, C., O'Connell, B., & Bates, E. (1984). First sentences in language
and symbolic play. Developmental Psychology, 20, 872-880.
Sigman, M., & Mundy, P. (1987). Symbolic processes in young autistic
children. In D. Cicchetti & M. Beeghly (Eds.), Symbolic development
in atypical children (pp. 31-46). San Francisco: Jossey-Bass.
Slobin, D. (1985). The cross-linguistic study of language acquisition.
Hillsdale, N J: Erlbaum.
Snyder, L. (1975). Pragmatics in language-deficient children: Prelinguistic and early verbal performatives and presuppositions. Unpublished doctoral dissertation, University of Colorado, Boulder.
Snyder, L. (1977). Communicative and cognitive abilities in the sensorimotor period. Merrill-Palmer Quarterly, 24, 161 - 180.
Snyder, L., Bates, E., & Bretherton, I. (1981). Content and context in
early lexical development. Journal of ChiM Language, 8, 565-582.
Tallal, E (1987). Developmental language disorders (Interagency Committee on Learning Disabilities report to the U.S. Congress). Washington, DC.
Tamis-LeMonda, C., & Bornstein, M. (1989). Habituation and maternal encouragement of attention in infancy as predictors of toddler
language, play, and representational competence. Child Development,
60, 738-751.
Terrell, B., Schwartz, R., Prelock, P., & Messick, C. (1984). Symbolic
play in normal and language impaired children. Journal of Speech
and Hearing Research, 27, 424-429.
Thai, D., & Bates, E. (1988a). Language and gesture in late talkers. Journal of Speech and Hearing Research. 31, 115-123.
Thai, D., & Bates, E. (1988b, November). Relationships between lan-
guage and cognition: Evidence from linguistically precocious children. Paper presented to the annual convention of the American
Speech-Language-Hearing Association, Boston, MA.
Thai, D., Bates, E., & Bellugi, U. (in press). Language and symbolic
gesture in two children with Williams syndrome. Journal of Speech
and Hearing Research.
Uzgiris, I., & Hunt, J. (1978). Assessment in infancy. Urbana: University of Illinois Press.
Uzgiris, I., & Hunt, J. (1987). Infant performance and experience:
New findings with the ordinal scales. ~Urbana: University of Illinois
Press.
1019
LANGUAGE AND GESTURE
Vihman, M. (1986). Individual differences in babbling and early speech:
Predicting to age three. In B. Lindblom & R. Zetterstrom (Eds.), Precursors of early speech (pp. 95-112). Baskingstroke, Hampshire, England: MacMillan.
Volterra, V., & Caseili, M. (1985). From gestures and vocalizations to
signs and words. In W. Stokes & V. Volterra (Eds.), SLR83: Proceed-
child language by a bilingual child. Journal of Child Language, 5,
311-326.
Vygotsky, L. (1967). Play and its role in the mental development of the
child. Soviet Psychology, 5, 6-18.
Werner, L., & Kaplan, B. (1963). Symbol formation: An organismic-
ings of the III International Symposium in Sign Language Research
York: Wiley.
Wolf, D., & Gardner, H. (1979). Style and sequence in symbolic play.
In N. Franklin & M. Smith (Eds.), Early symbolization (pp. 117138). Hillsdale, N J: Erlbaum.
Zukow, P. (1984). Criteria for the emergence of symbolic conduct:
When words refer and play is symbolic. In L. Feagans, C. Garvey, &
R. Golinkoff (Eds.), The origins and growth of communication (pp.
162-175). Norwood, N J: Ablex.
developmental approach to language and expression of thought. New
(pp. 1-9). Silver Spring, MD: Linestock Press.
Volterra, V., Bates, E., Benigni, L., Bretherton, I., & Camaioni, L.
(1979). First words in language and action: A qualitative look. In E.
Bates, L. Benigni, I. Bretherton, L. Camaioni, & V. Volterra (Eds.),
The emergence of symbols: Cognition and communication in infancy
(pp. 141-222). New York: Academic Press.
Volterra, V., & Taeschner, T. (1977). The origin and development of
Appendix
Object Words and Gestures Used in Experiment
Word
Cup
Baby
Phone
Spoon
Comb
Car
Toothbrush
Hat
Flower
Soap
Dog
Airplane
Gesture
Drink with head back, cupped hand
Hug with rocking motion
Put to ear
Put to mouth (open and close
mouth, hand to side of mouth)
Comb hair
Back and forth motion on table
Brush teeth with horizontal back
and forth motion
Put on top of head
Put to nose and sniff
Wash hands
Hop across table
Back and forth motion in the air
% of children
reported to
produce
gesture
% of children
reported to
comprehend
word
% of children
reported to
produce
word
% of children
producing
gesture in
experiment
95
90
88
86
62
71
71
58
0
38
2
2
63
12
29
39
86
84
61
71
67
56
0
24
0
37
37
49
54
45
22
18
9
40
53
29
82
38
7
11
4
69
7
22
37
17
44
17
Received M a r c h 11, 1988
Revision received M a r c h 22, 1989
Accepted M a r c h 23, 1989 •