You are on page 1of 10

Developmental Psychology Copyright 1984 by the

1984, Vol. 20, No 2, 293-302 American Psychological Association. Inc

Cognitive and Language Skills in Autistic,


Mentally Retarded, and Normal Children
Marian Sigman and Judy A. Ungerer
Department of Psychiatry
University of California, Los Angeles, School of Medicine

The sensorimotor skills and play behaviors of autistic, mentally retarded, and normal
children were compared to identify deficits in object knowledge specific to autism.
There were no differences in sensorimotor skills between the three groups matched
on mental age except that the autistic children were deficient in the capacity to
imitate gestures and vocalizations. Autistic children showed less diverse functional
play, particularly directed toward dolls, and less symbolic play both spontaneously
and after cueing. Although functional and symbolic play and the ability to imitate
were correlated with receptive language for all three groups, sensorimotor skills
were associated with receptive language only for the normal and the mentally
retarded groups. Thus the autistic children in this study seem to be a distinct group
with specific deficits in imitation and symbolic abilities. The observation that the
disorder involves deficits in certain cognitive concepts and not others suggests that
representational thought may be manifested in two systems, only one of which is
impaired in the autistic child. Another hypothesis is that the cognitive deficits
manifested by autistic children may be related to their impaired social development,
because all the areas of cognitive deficit involve significant social components.

The aim of this study is to identify early versity of the group makes differentiation be-
cognitive deficits specific to the syndrome of tween autistic children and other children with
autism. The majority of autistic children are retardation difficult (Sherman, Shapiro, &
also mentally retarded, so specification of cog- Glassman, 1981; Wing & Gould, 1979). On
nitive deficits depends on a comparison of the the other hand, Rutter (1978) conceives of au-
behaviors of autistic children with those shown tism as a specific disorder involving a primary
by children at similar developmental levels. In disturbance of symbolic capacities. Thus a
recent years, some clinicians and researchers comparison of the cognitive skills of autistic
have begun to question the value of the di- children with those of mentally retarded chil-
agnosis of autism and to suggest that the di- dren has implications both for the definition
and specificity of the syndrome.
The cognitive skills selected for comparison
Support for this research was provided by Grant 12-41
from the March of Dimes, Biobehavioral Research Support were those that normally develop during in-
Grant 516, National Institute of Mental Health (NIMH) fancy because autism involves disorders that
Grant MH 33815, and NIMH Postdoctoral Fellowship 1 are manifested before 30 months of age. In
F32 MH 07550. Subjects were recruited from the Uni- several previous studies, we have attempted to
versity of California, Los Angeles Clinical Research Center describe the concepts of autistic children in
(CRC) for the Study of Childhood Psychosis, funded by
NIMH Grant MH 30897. We thank members of the CRC, two different domains: operational object con-
particularly Peter Tanguay, Barbara Fish, and B. J. Freeman cepts as manifested in sensorimotor problem-
for their support of our research efforts. We also thank solving skills and presymbolic object concepts
Mary Beth Sorensen and M. Holly Hackman for their as manifested in functional and symbolic use
assistance in data collection and analysis, Luisa Castillo
for her help in the preparation of the manuscript, and
of objects in play. When the performance of
Dolores Adams for her assistance in data analysis. In ad- autistic children on sensorimotor tasks was
dition, Peter Mundy, Rosa Needleman, and Tracy Sherman compared to that of normal children, matched
contributed significantly to this project. as a group on mental age, the scores of the
Requests for reprints should be sent to Marian Sigman, two groups were surprisingly similar (Sigman
Department of Psychiatry, University of California, Los
Angeles, School of Medicine, 760 Wesrwood Plaza, Los
& Ungerer, 1981). On the other hand, the play
Angeles, California 90024. behaviors of the autistic children (Ungerer &
293
294 MARIAN SIGMAN AND JUDY A UNGERER

Sigman, 1981) appeared to differ from those hand, Sigman and Ungerer (1981) found few
usually observed in normal children of com- relationships between language and sensori-
parable mental ages (Fenson, Kagan, Kearsley, motor skills for autistic children. The present
& Zelazo, 1976; Fenson & Ramsey, 1980; In- research determined whether these same re-
helder, Lezine, Sinclair, & Stambak, 1972; lationships between play and sensorimotor be-
Largo & Howard, 1979; Lowe, 1975; Nicolich, havior and language were found with normal
1977; Rosenblatt, 1977; Sinclair, 1970). Al- control subjects matched individually on
though there appeared to be deficits in the mental age and extended the comparisons to
presymbolic skills of autistic children, these include young, mentally retarded children.
deficits could not be definitively identified be- To summarize, the goals of this research
cause the play behaviors of a normal mental were (a) to compare the development of sen-
aged matched control group had not been ob- sorimotor behavior, imitation, and play in au-
served. Thus a major goal of this research was tistic, mentally retarded, and normal children
to identify specific deficits in autistic children in order to identify cognitive deficits specific
by comparing sensorimotor and play behaviors to the autistic syndrome and (b) to examine
to those shown by mentally retarded children, the relationship between sensorimotor and
matched on both mental and chronological play behaviors and language in mentally re-
age, and normal children, matched on mental tarded and normal children to determine
age. Imitation skills were also assessed because whether the correspondences between these
these have been described as deficient, al- domains are different for autistic as compared
though some imitation of body movements to mentally retarded and normal children.
and actions with objects has been noted
(DeMyer et al., 1972). Method
A second goal of this research was to de- Subjects
termine whether relationships found between
object concepts and language in autistic chil- The autistic sample consisted of 16 children who were
mpatients in the Neuropsychiatric Institute at the Uni-
dren are the same as those found in mentally versity of California, Los Angeles and were subjects in the
retarded and normal groups. Comparisons Clinical Research Center (CRC) for the Study of Childhood
with mentally retarded and normal children Psychosis and have been described in previous papers (Sig-
are necessary to determine whether differences man & Ungerer, 1981). The experimenters in this study
found are specific to autism or derive from were unaware at the time of assessment whether a child
was diagnosed as autistic, mentally retarded, or aphasic.
general delays in cognitive development. A few Diagnoses were made independently by several CRC psy-
studies have reported relationships between chiatrists using the Diagnostic and Statistical Manual of
language and object concepts as expressed in Mental Disorders (3rd ed.) criteria (American Psychiatric
sensorimotor and play skills in autistic chil- Association, 1980) The 16 children, 15 boys and 1 girl,
dren. Ungerer and Sigman (1981) found that ranged in age from 39 to 74 months, with a mean age of
51 7 months Mean mental ages and developmental quo-
autistic children who were able to comprehend tients are presented in Table 1
verbal labels demonstrated more functional Each of the 16 mentally retarded children was selected
and symbolic play and longer sequences of to match an autistic child on mental age, chronological
meaningfully integrated play acts than did au- age, and developmental quotient (Table 1). Because of the
tistic children with minimal or no compre- priority of identifying subjects who matched on chrono-
logical and mental age with the autistic children, the nature
hension of verbal labels. In addition, Wing, of the condition associated with mental retardation and
Gould, Yeates, and Brierley (1977) found re- the sex of the subject were allowed to vary. Half the children
lationships between language comprehension had Down's syndrome, three had organic dysfunctions
age and play in a mixed group of autistic and including arrested hydrocephalus and seizure disorders,
retarded children. Curcio (1978) reported that and five suffered from retardation of unidentified origin.
There were 10 boys and 6 girls. Four of these children
Stage 5 performance on the means and cau- were subjects in the Clinical Research Center.
sality subscales and Stage 3 performance on Each child in the normal sample of 15 boys and 1 girl
the imitation subscale of the Uzgiris-Hunt was selected to match on mental age with one autistic and
scales (1975) were minimal prerequisites for mentally retarded child (see Table 1). The mental age scores
nonverbal intentional communication in a used for matching the autistic, mentally retarded, and nor-
mal groups were based on a general intelligence scale (Cat-
group of mute autistic children. On the other tell)
COGNITIVE AND LANGUAGE SKILLS IN AUTISTIC CHILDREN 295

Assessment of Play in an Unstructured and Table 1


Structured Setting Age and Standardized Test Data for Autistic,
Mentally Retarded, and Normal Children
Play was assessed in two test sessions described in a
previous publication (Ungerer & Sigman, 1981) The child's Mentally
play behavior was videotaped and recorded by the exper- Variable Autistic retarded Normal
imenter seated in the playroom using a time-sampling
procedure with a checklist that included the behaviors Chronological age
most frequently observed with this set of toys. Play be- (in months)
haviors not included in the checklist were separately noted. 51.7 50.7 20.8
Following the play session, the videotaped record was re- SD 10.7 12.6 3.0
viewed to clarify any ambiguities in the checklist completed Range 39-74 32-80 16-25
by the experimenter.
Measures The child's play behaviors were grouped Cattell/Stanford Binet
into the following categories: Mental age
1 Simple manipulation Each instance of mouthing, (in months) 24.8 26.6 24.6
waving, banging,fingering,or throwing of a single toy was
SD 5.1 7.2 5.2
recorded.
Range 18-38 17-38 17-35
2. Relational play This category included nonfunctional 48.1 48.0 116.0
combinations of objects uke touching or banging two ob- IQ
SD 8.1 119 12.8
jects together, stacking objects, and using one object as a Range 35-62 28-71 92-147
container to hold another object The last two categories
excluded those behaviors considered functional associations
of objects, for example, placing a cup on a saucer or putting
a spoon in a cup. example, the experimenter told the child "Feed the baby
3. Functional play Four different types of functional with the bottle." If the child did not respond to the verbal
acts were recorded, object-directed (e.g., placing the top cues, the experimenter then modeled the appropriate
on the teapot or pushing the truck into the garage), self- functional acts for the child The child's responses to both
directed (e.g., brushing one's hair), doll-directed (e.g., verbal cueing and modeling were recorded on the checklist
feeding a doll with a spoon), and other-directed (e.g., hold- by the experimenter This play interaction lasted approx-
ing a telephone receiver to the mother's ear) imately 30 min. and the categories of play measured were
the same as those described above. (One child was un-
4 Symbolic play Three types of symbolic acts were
available for testing with this procedure.) The structured
recorded: substitution play, defined as the use of one object
play assessment was always conducted after the unstruc-
as if it were another different object (e.g., using a teacup
tured play assessment so that spontaneous play in the un-
as a telephone receiver), agent play, defined as the use of
structured setting would not be influenced by the extensive
a doll as an independent agent of action (e.g., propping a
verbal cueing and modeling procedures used in the struc-
bottle in a doll's arms as if it could feed itself), and imag-
tured setting. The duration of play by category was not
inary play, defined as the creation of objects or people
measured in the structured setting, because the experi-
having no physical representation in the immediate en-
menter largely determined the duration of a child's play
vironment (e.g, making pouring sounds and saying "tea"
with the different toys. Diversity of play was measured in
as imaginary tea is poured from a teapot into a cup)
the same way as in the unstructured play assessment. Re-
Measures of play integration were also obtained. The liability for the diversity play measures was good. The
total number of different sequences of related functional mean correlation for the diversity play measures for the
and symbolic acts was scored, as well as the number of two observers was .92 (SD = .13).
different acts included in each sequence
The duration of play by category was measured by the
total number of 10-sec intervals in which each category Assessment of Sensonmotor Behavior
of play was recorded. Diversity of play was measured by
the number of different instances recorded for each category Sensonmotor behaviors were assessed using the "Stages
of relational, functional, and symbolic play. Reliability for of Sensonmotor Intelligence in the Child" developed by
the duration and diversity play measures was computed Casati and Leztne( 1968) These scales and their reliability
separately for each play category and was good (Ungerer have been described previously (Kopp, Sigman, & Par-
& Sigman, 1981). melee, 1974; Sigman & Ungerer, 1981).
A second, more structured play assessment was also The skills of autistic children are often difficult to assess
conducted to ensure that the child's optimal play skills reliably because their readiness to perform varies widely.
would be recorded. Toys used in the unstructured play For this reason, all items failed by a child in any of the
assessment were presented to the child one at a time or three groups were readministered in a second testing to
in small groups by an experimenter who then recorded ensure that the child's maximal level of skill was obtained.
the child's spontaneous use of objects on a checklist spe-
cifically designed for this play assessment Vocal and Gestural Imitation
If the child did not spontaneously use the objects in a
functionally appropriate manner, the experimenter directed The vocal and gestural imitation subtests of the Uzgjris-
the child verbally in the functional use of the toys. For Hunt Ordinal Scales of Psychological Development (1975)
296 MARIAN SIGMAN AND JUDY A. UNGERER

were administered to each child. These tests require the Table 2


child to imitate familiar and novel gestures made by the Mean Duration of Play Behaviors in an
experimenter as well as familiar and novel sounds and Unstructured Setting (No of 10-Sec Units)
words. Each subtest item was administered three times,
and the child received one score on each subtest repre- Mentally
senting the highest level of imitation achieved. Play behaviors Autistic retarded Normal

Language Assessments Simple manipulation 19.3 13.3 12.9


Relational 21.2 14.9 19.7
A receptive language measure developed by Beckwith Symbolic 1.3 5.6 4.0
and Thompson (1976) and described in previous publi- Functional 16.8"-b 40.1 31.5
cations was administered. The child's receptive language Sequences
score was determined by the number of slides verbally 2 acts 1.6 2.6 3.3
labeled by the experimenter that the child pointed to or 3 or more acts 0.7' 3.2 2.3
touched correctly from 34 pairs of slides. In addition, each
child was administered a clinical language evaluation, 1
For means significantly different for autistic and mentally
which was carried out independently by a linguist Re- retarded groups, p < .05
ceptive and expressive language were assessed based on b
For means significantly different for autistic and normal
the child's responses to questions, requests, and prompts groups, p < .05
in a sermstructured play setting The assessment focused
on the child's understanding and use of concepts of pos-
session, existence, attribution, and location, and the ability
to understand and use labels. The child was assigned a ber of 10-sec units recorded for simple ma-
receptive language age and an expressive language age (in
months) based on behavior and speech recorded during nipulation, relational, functional, and sym-
several interactions. The relationship between the receptive bolic play) and on the two measures of play
language slide measure and the independently administered integration (the number of different sequences
clinical assessment of receptive language was high for all of two related acts and the number of different
groups. r( 13) = .86, p < .01, for the autistic group; r( 12) =
.74, p < .01, for the mentally retarded group; and r( 13) =
sequences of three or more related acts). Sig-
.72, p < .01, for the normal group. Therefore, only scores nificant group effects were found for the num-
from the receptive language slide measure and the clinical ber of 10-sec units of functional play, F(2,
assessment of expressive language were used in the data 45) = 5.70, p < .01, and the number of dif-
analyses. ferent sequences of three or more related acts,
F(2, 45) = 3.22, p < .05 (see Table 2). The
Results individual group comparisons indicated that
the autistic children were engaged in functional
Comparison of Autistic, Mentally Retarded,
play for fewer 10-sec units than were either
and Normal Children
the mentally retarded, t(26A) = 3.20, p < .01,
The purpose of the first analysis was to or normal groups, *(29.7) = 2.43, p < .05. In
identify deficits in cognitive functioning spe- addition, the autistic children demonstrated
cific to the autistic syndrome. The perfor- fewer sequences of three or more related acts,
mances of the autistic, mentally retarded, and r(19.0) = 2.78, p < .05, than did the mentally
normal children on the different tasks were retarded children.
compared with analysis of variance (ANOVA) One-way ANOVAS were used to compare the
statistics. Deficits on a task were considered three measures of play diversity (the number
specific to autism only if the performance of of different acts recorded for relational, func-
the autistic children was significantly poorer tional, and symbolic play). Significant group
than that of both the mentally retarded and effects were found for two diversity of play
normal groups. For this reason, individual measures: the number of different functional
group means were compared using t tests acts, F(2,45) = 5.72, p < .01, and the number
whenever the ANOVAS showed significant group of different symbolic acts, 7=1(2, 45) = 3.61,
effects. Because three groups were compared, p < .05). The individual group means were
alpha levels were raised to control for the effects compared for these variables. The autistic
of performing multiple t tests (see Dunn & children demonstrated fewer different func-
Clark, 1974). tional acts than did either of the other two
Unstructured play One-way ANOVAS were groups, <(27.7) = 2.95, p < .05, for the mentally
used to compare the three groups on each of retarded group, and /(29.9) = 3.01, p < .01,
four measures of play duration (the total num- for the normal group, and fewer different sym-
COGNITIVE AND LANGUAGE SKILLS IN AUTISTIC CHILDREN 297

bolic acts than did the mentally retarded chil- Table 3


dren,/(17.2) = 3.36, p< .01. Mean Number of Different Play Behaviors
Thus both the total duration and diversity in an Unstructured Setting
of the autistic children's functional play was Mentally
less than that of the normal and mentally re- Play behaviors Autistic retarded Normal
tarded children. To determine the particular
types of functional play that varied among the Relational 7.9 6.0 11.1
groups, the number of different functional acts Total functional 5.1"-" 11.1 10.1
Object-directed 2.8* 5.9 4.5
was subdivided into four categories (object- Self-directed 1.4 26 2.4
directed, self-directed, doll-directed, and other- Doll-directed 0.9b 2.3 2.8
directed), and group comparisons for each Other-directed 0 1 0.4 0.4
category were conducted using t tests. Two t- Symbolic 0.6' 2.9 2.6
test comparisons were significant. The autistic
* For means significantly different for autistic and mentally
children demonstrated fewer doll-directed retarded groups, p < .05.
functional acts than did the normal children, b
For means significantly different for autistic and normal
/(24.3) = 2.62, p < .05, and fewer object-di- groups, p < .05.
rected functional acts than did the mentally
retarded children, /(24.2) = 2.46, p < .05 (see pendent variables: number of different acts re-
Table 3). corded for simple manipulation, relational,
The distribution of play time across the cat- functional, and symbolic play. Only acts pro-
egories of functional, relational, and simple duced spontaneously by the children were in-
manipulation play was different for the three cluded in these analyses. There were significant
groups of children (see Table 2). Within each group effects for the number of different simple
group, the number of 10-sec units recorded manipulation acts, F\2, 42) = 3.14, p < .05,
for the three play categories was compared functional acts, F(2, 42) = 9.26, p < .001, and
using one-way ANOVAS. There was a significant symbolic acts, F(2, 42) = 7.34, p < .01 (see
main effect for category of play (ps < .01) for Table 4).
the mentally retarded and normal children but To determine whether the performance of
not for the autistic children. The mentally re- the autistic children was significantly poorer
tarded children produced more functional play than that of the other two groups, group means
than either relational, /(14) = 3.24, p < .01, were compared using t tests. In these analyses,
or simple manipulation play, t( 14) = 3.19, p < the number of different acts performed spon-
.01. The results for the normal children were taneously was compared, as was the total
similar for functional versus relational, t( 14) = number of different acts performed sponta-
1.90, p < .08; for functional versus simple neously and after verbal cueing and modeling.
manipulation, /(14) = 3.54, p < .01. In con- Comparisons for the four subcategories of
trast, the autistic children were engaged for functional play were also included. As can be
about the same amount of time in functional, seen from Table 4, the greatest variation be-
relational, and simple manipulation play. For tween groups was shown in spontaneous play.
the mentally retarded and normal children, The autistic children produced fewer different
more sophisticated forms of play, like func- total functional and doll-directed functional
tional play, tended to dominate less sophisti- acts and fewer different symbolic acts than did
cated play behaviors, like relational or simple either the mentally retarded or normal children
manipulation play. Although many autistic (ps < .05). These results are consistent with
children were capable of performing more so- findings from the unstructured play assess-
phisticated play, they used this skill less fre- ment. Following verbal cueing and modeling
quently than did normal or mentally retarded in the structured play task, the autistic children
children of comparable mental age. were able to increase their functional play so
Structured play. One-way ANOVAS were that the means for the autistic and normal
used to compare the play of the autistic, men- samples were more similar. However, the au-
tally retarded, and normal groups during the tistic children still produced fewer different
second, structured play assessment. A separate symbolic acts than did the other two groups,
ANOVA was conducted for each of four de- even after modeling.
298 MARIAN SIGMAN AND JUDY A. UNGERER

Table 4
Mean Number of Different Play Behaviors in a Structured Setting

Autistic Mentally retarded Normal

Play behaviors Spontaneous Elicited Spontaneous Elicited Spontaneous Elicited

Simple manipulation 4.9 1.8 2.9


Total functional 10.6*-" 17.9* 17.8 24.5 16.1 22.6
Object-directed 5.2" 8.7" 8.4 11.8 71 10.8
Self-directed 2.6 3.3 3.6 4.3 2.6 2.8
Doll-directed 2.6*-b 5.7 5.0 7.6 50 7.8
Other-directed 02 0.3 0.8 0.9 1.3 1.3
Symbolic 0.5"-" 2.8"-b 3 1 6.5 3.1 6.0

• For means significantly different for autistic and mentally retarded groups, p < .05
b
For means significantly different for autistic and normal groups, p < .05.

Sensonmotor behaviors. One-way ANOVAS ceptive and expressive language measures.


were used to compare the scores of the three Comparison of group means using / tests in-
groups on the seven subtests of the Casati- dicated that the autistic children scored sig-
Lezine Scale. A separate ANOVA was conducted nificantly lower in both language areas than
for each subtest score for both the first and did the normal children and lower in receptive
the second administration of the scale. There language than did the mentally retarded chil-
were no significant group differences on any dren (/7s < .05). The receptive language of the
subtest for either test administration. Thus the mentally retarded children was comparable to
autistic, mentally retarded, and normal chil- that of the normal children, but their ability
dren were quite similar in their performance to express themselves verbally was somewhat
on the sensorimotor scale. The lack of group less mature, r(27) = 1.79, p < .08. This dif-
differences was not due to ceiling effects. ference could not be attributed to particular
Vocal and gestural imitation. The scores deficits in the Down's syndrome infants, be-
on the Uzgiris-Hunt vocal and gestural imi- cause the mean expressive language age for
tation subtests were compared using Kruskal- the children with Down's syndrome (M = 14.7
Wallis tests. There were significant group dif- months) was nearly identical to that of the
ferences for both vocal imitation, x2(2) = children with other developmental disabilities
11.12, p < .01, and gestural imitation scores, (M = 14.5 months).
X2(2) = 12.23, p < .01. On the vocal imitation The relationship between receptive and ex-
subtest, the mean scores of the normal and pressive language abilities was very high for
mentally retarded children were nearly iden- the autistic children, r(13) = .68, p< .01, and
tical and significantly higher than were the the normal children, r( 13) = .71, p < .01, but
mean score of the autistic children. On the less robust for the mentally retarded children,
gestural imitation subtest, the mentally re- r(12) = .36, p < .10. Within the mentally
tarded children performed less well than did retarded group, the relationships were equally
the normal children, but the difference was low for Down's syndrome and the other de-
not statistically significant. However, the au- velopmentally delayed children. The high cor-
tistic children performed significantly more relations found for the autistic children were
poorly than did both the mentally retarded, not surprising because so many gave no in-
/(22.3) = 2.68, p < .01, and normal groups, dication of either understanding or using lan-
t(l6.5) = 4.01, p < .001. Thus the autistic guage.
children were significantly less able to imitate
both gestures and vocalizations than were the
mentally retarded and normal children. The Significance of the Diagnosis of Autism
Language. The receptive language scores To assess the specificity of the symbolic dis-
and expressive language ages were compared order associated with autism, it was necessary
for the three groups using ANOVAS. There were to determine whether the language delays of
significant group main effects for both the re- the autistic and mentally retarded children
COGNITIVE AND LANGUAGE SKILLS IN AUTISTIC CHILDREN 299

were sufficient to explain their performance spond to verbal labels still showed more diverse
in other symbolic domains like play, or whether functional and symbolic play than did the au-
deficits in play were also a function of diag- tistic children who responded correctly to ver-
nosis. If the play behaviors of autistic and bal labels.
mentally retarded children were predicted fully
by the children's language skills, we could
conclude that the symbolic deficits observed Relationships Between Cognition
in both groups were similar and related pri- and Language
marily to delays in language. However, if di- Play and language For the purpose of
agnosis improved the prediction of play be- contrasting relationships between play and
yond that achieved by language alone, we could language within the autistic, mentally retarded,
conclude that the symbolic impairment of the and normal groups, the data were analyzed
autistic children was unique and not solely the with Pearson product-moment correlations.
result of delayed language skills. For each group, receptive language scores and
The autistic and mentally retarded children expressive language ages were correlated with
were divided into groups based on diagnosis the number of different functional and sym-
and language capacity. Children who re- bolic acts in the unstructured and structured
sponded correctly to three or fewer words on play tasks, and the number of different se-
the receptive language tasks were placed in a quences of three or more related acts in un-
low-language group (N = 13), whereas the re- structured play. The structured play variables
maining children who responded correctly to were composed of acts carried out sponta-
11 or more words were placed in a high-lan- neously and after verbal cueing and modeling.
guage group (N = 19). Individual 2 X 2 (Di- The correlations between play and receptive
agnosis X Language Group) ANOVAS were car- language for the three groups are shown in
ried out for each of six play variables from Table 5. For relationships between play and
the unstructured play task. These play vari- expressive language, only correlations for the
ables were the number of different object-di- autistic children are given because correlations
rected, self-directed, doll-directed, and other- within the other two groups were not signif-
directed functional acts, the total number of icant. The results for the autistic children in-
different functional acts, and the total number dicated a relationship between functional play
of different symbolic acts. The results showed demonstrated in the unstructured setting and
a significant main effect for diagnosis (autism receptive language. The number of functional
vs. mental retardation) for all analyses except play acts also correlated with expressive lan-
other-directed functional play, and a significant guage age, as did the sequencing of related acts
main effect for language group for all analyses in play.
except other-directed and self-directed func- The number of different symbolic acts per-
tional play. Most important, the interaction of formed by autistic children in the unstructured
Diagnosis X Language Group was not signif- play task did not relate to language. However,
icant in any analysis. These results indicate there was a significant relationship between
that both diagnosis and the capacity to un- receptive language and the total number of
derstand verbal labels independently discrim- different spontaneous and elicited symbolic
inate play patterns and thus yield support for acts in the structured play assessment. To de-
Rutter's proposal of a unique symbolic deficit termine whether the effects of verbal cueing
in autistic children. The diagnosis of autism and modeling were responsible for the rela-
does imply a deficit in symbolic skills over and tionship, the correlations between spontaneous
above the deficit in language comprehension. symbolic play in the structured task and the
The best illustration is that the mean number two language measures were calculated. The
of different functional and symbolic acts for number of different spontaneously performed
the seven A/^A-language autistic children was symbolic acts was related significantly to re-
slightly lower than the mean number of dif- ceptive language, r(13) = .56, p < .05, and
ferent functional and symbolic acts for the somewhat to expressive language, r(12) = .52,
four /ow-language retarded children. In other p < . 10. Thus the more verbally adept autistic
words, the retarded children who did not re- children did perform more symbolic acts when
300 MARIAN SIGMAN AND JUDY A. UNGERER

Table 5
Correlations Between Number of Play Acts and Language Abilities for Autistic,
Mentally Retarded, and Normal Children'
Mentally
Autistic retarded Normal
Receptive Expressive Receptive Receptive
Play behaviors language language language language

Unstructured play
Functional .57 .60 .65
Symbolic — — .49 .65
Sequences (3 or more acts) .65 .54
Structured play*
Functional .73 .56
Symbolic .50 — .52 .65

• For all correlations reported, p < .05.


b
The total number of spontaneous and elicited acts were used in these analyses.

their behavior was structured, and they did labels. Thus the ability to generalize functional
not require verbal cues or modeling to do so. play acts to dolls was associated with the pres-
The relationship between play and language ence of at least elementary knowledge of lan-
for the mentally retarded and normal children guage in both the autistic and mentally re-
was similar although less robust than that seen tarded groups.
in the autistic group. For both the mentally Sensonmotor skills and language. There
retarded and normal groups, symbolic play were few relationships between performance
was related to receptive language, and for the on the sensorimotor tasks and either receptive
normal children a relationship to functional or expressive language for the autistic children.
play was also found. However, in neither group The only significant relationships found were
were relationships between expressive language negative correlations between both language
and play observed. measures and performance on the extension
Ungerer and Sigman (1981) reported that and object and support subtests (rs = —.54 to
autistic children with minimal or no receptive -.69). Jn contrast, the mentally retarded chil-
language showed fewer doll-directed functional dren showed strong positive associations be-
acts and fewer symbolic acts in a structured tween sensorimotor abilities and receptive
play task than did autistic children with better language. All the sensorimotor subtests, except
receptive language skills. Among the mentally exploration of objects, were significantly re-
retarded children in this study, four gave no lated to receptive language scores (rs ranged
indication of recognizing word meanings. The from .53 to .79). There were no significant
following play behaviors were compared for relationships between expressive language and
the 12 mentally retarded children with lan- sensorimotor abilities for the mentally retarded
guage comprehension and the 4 who showed children. The pattern was nearly identical for
no word recognition: the number of different normal children, with five of the seven subtests
other-directed, object-directed, self-directed, showing significant relationships with receptive
and doll-directed functional acts, and the language (rs ranged from .50 to .80). Only use
number of different symbolic acts in the un- of the rake to obtain a toy hidden in a tube,
structured play task, as well as the same be- the hardest task on the scale, was related to
haviors elicited in the structured play assess- expressive language, r( 13) = .81, p < .01. Thus
ment. The four mentally retarded children the mentally retarded and normal children
without language showed significantly less showed a correspondence between sensori-
spontaneous doll-directed functional play, motor and receptive language development not
/(13.9) = 3.59, p < .01, and more elicited self- manifested by the autistic children.
directed functional play, /(12.8) = 5.24, p <
Imitation and language. The relationship
.001, than did children who recognized verbal
between gestural and vocal imitations and re-
COGNITIVE AND LANGUAGE SKILLS IN AUTISTIC CHILDREN 301

ceptive language was significant for all groups sufficient for language development. Watching
(rs ranged from .55 to .67). For the autistic a young autistic child who skillfully manip-
and normal children, the capacity to imitate ulates a rake to obtain a hidden object but
vocally was also related to expressive language who responds in no way to spoken language
age (rs = .72 and .62, respectively). Imitative is convincing evidence that the development
skills were tied to language capacity for all of sensorimotor and symbolic knowledge can
groups in this study. be very divergent.
The autistic children in this study do seem
Discussion to be a distinct group with specific deficits in
imitation and symbolic abilities (Rutter, 1978;
Several deficits in functioning specific to Wing et al., 1977). Even when their general
autistic children were identified in this re- developmental delay is taken into account by
search. In spontaneous play, autistic children comparing their performance to mental-age
demonstrated less diverse functional play, par- matched normal and mentally retarded chil-
ticularly functional play directed toward dolls, dren, they show specific deficits in play, imi-
than did normal or mentally retarded children tation, and language. On the other hand, we
of comparable mental age. Symbolic play pro- have no evidence that sensorimotor intelli-
duced spontaneously and after cueing also was gence is deficient in autistic children when their
deficient in the autistic group, as was the ability overall level of mental retardation is taken into
to comprehend verbal labels and imitate words account. Thus the disorder seems to involve
and gestures. In contrast, the autistic children deficits in certain object concepts and not in
were not deficient in sensorimotor skills when others.
compared to mentally retarded and normal One hypothesis for the variation in deficits
groups. is that representational thought may be man-
The patterns of relationships between cog- ifested in two different systems, one more ad-
nitive skills and language indicated similarities vanced than the other. One system, reflected
and differences among the three groups. Al- in the development of sensorimotor skills, may
though the complexity of play and imitation involve the capacity to recall information that
varied among the autistic, mentally retarded, then is accessible for problem solving. The
and normal children, the relationships between capacity to translate experience into language
these variables and language within the three and play symbols that are manipulated in-
groups were similar. Functional and symbolic dependently may reflect a second system (Wolf
play were associated with more advanced re- & Gardner, 1981). For the normal and men-
ceptive language in all three groups, as was tally retarded child, the two systems may de-
the ability to imitate gestures and vocaliza- velop together, so that progress in the senso-
tions. In contrast, sensorimotor skills and lan- rimotor representation of absent objects may
guage were clearly positively associated for the be associated with advances in symbolic skills,
mentally retarded and normal children but although neither system may be dependent on
negatively associated or unrelated for the au- the other. For the autistic child, these systems
tistic children. These results indicate that a do not cohere. It is in the second system, the
defining characteristic of the syndrome of au- ability to form and manipulate symbols, that
tism may be significantly more proficient sen- the autistic child may have a major impair-
sorimotor capacities than symbolic and imi- ment.
tative skills. A second hypothesis is that the cognitive
The lack of positive association between deficits seen in autistic children are secondary
sensorimotor skills and language in the autistic to their impaired social development. All the
children has implications for understanding areas of specific cognitive deficit identified to
normal language development. Although sen- date depend on social interaction for their de-
sorimotor knowledge may be necessary for velopment. For example, functional object use
language acquisition, the impressive sensori- as demonstrated in play is learned from ob-
motor skills of the autistic children in contrast servation of others, and the generalization of
to their extreme deficits in language demon- functional object use to dolls entails an ad-
strate that sensorimotor knowledge is far from ditional social component. The development
302 MARIAN SIGMAN AND JUDY A. UNGERER

of imitation and language necessarily involves Kopp, C B., Sigman, M, & Parmelee, A. H. (1974). Lon-
gitudinal study of sensorimotor development. Devel-
the responsive interaction of others. In con- opmental Psychology, 10, 687-695
trast, sensorimotor object knowledge may de- Largo, R., & Howard, J. (1979). Developmental progression
velop without social involvement, although in in play behavior of children between nine and thirty
normal development social facilitation of sen- months. Developmental Medicine and Child Neurology.
sorimotor learning often may occur. These two 21. 299-310
Lowe, M (1975). Trends in the development of represen-
hypotheses may not be independent, because tational play in infants from one to three years. Journal
symbolic and social development in young of Child Psychology and Psychiatry, 16, 33-47.
children are necessarily intertwined. However, Nicohch, L (1977) Beyond sensorimotor intelligence.
they do provide an approach for understanding Merrill-Palmer Quarterly. 23. 89-99.
the diversity of cognitive and social deficits Rosenblatt, D. (1977). Developmental trends in infant play.
In B Tizard & D Harvey (Eds.), Biology of play (pp
that have been found to characterize the au- 33-44). London- Heinemann
tistic syndrome. Rutter, M (1978) Language disorder and infantile autism.
In M Rutter & E. Schopler (Eds.), Autism (pp. 85-
104). New York Plenum Press.
References Sherman, M., Shapiro, T, & Glassman, M. (1981, Oc-
American Psychiatric Association. (1980). Diagnostic and tober) Play and language in developmental!)' disordered
statistical manual of mental disorders (3rd ed). Wash- preschoolers—a new approach Paper presented at the
ington, DC: Author. annual meeting of the American Academy of Child
Beckwith, L., & Thompson, S. (1976). Recognition of Psychiatry, Dallas, TX.
verbal labels of pictured objects and events by 17- to Sigman, M , & Ungerer, J (1981). Sensorimotor skills and
30-raonth-old infants Journal of Speech and Hearing language comprehension in autistic children. Journal
Research, 19. 690-699 of Abnormal Child Psychology'. 9, 149-165
Casati, I., & Lezme, I. (1968) Les etapes de 1'intelligence Sinclair, H (1970). The transition from sensory-motor
sensori-motnce [The stages of sensorimotor intelligence]. behavior to symbolic activity. Interchange, 1, 119-125.
Paris: Les Editions de Centre de Psychologie Apphquee Ungerer, J , & Sigman, M. (1981). Symbolic play and lan-
Curcio, F. (1978) Sensorimotor functioning and com- guage comprehension in autistic children. .Journal of
munication in mute autistic children. Journal of Autism the American Academv of Child Psychiatry, 20, 318—
and Childhood Schizophrenia, 8. 281-292 337.
DeMyer, M. K.., Alpem, G. D., Barton, S.. DeMyer, Uzgins, I C, & Hunt, J. McV. (1975). Assessment in
W. E., Churchill, D. W., Hingtgen, ) N., Bryson, infancy Urbana University of Illinois Press
C. Q., Pontius, W., & Kimberlin, C. (1972). Imitation Wing. L.. & Gould, J. (1979). Severe impairments of social
in autistic, early schizophrenic, and non-psychotic sub- interaction and associated abnormalities in children:
normal children. Journal of Autism and Childhood Epidemiology and classification. Journal ofAutism and
Schizophrenia. 2, 264-287 Developmental Disorders, 9, 11-31
Dunn, Q, & Clark, V. (1974) Applied statistics Analysis Wing, L, Gould, J., Yeates, S., & Bnerley, L. (1977).
of variance and regression New York- Wiley. Symbolic play in severely mentally retarded and in au-
Fenson, L., Kagan, J., Kearsley, R., & Zelazo, P. (1976) tistic children. Journal of Child Psychology and Psy-
The developmental progression of manipulative play in chiatry, 18. 167-178.
the first two years. Child Development. 47. 232-236 Wolf, D , & Gardner. H. (1981). On the structure of early
Fenson, L., & Ramsey, D. (1980). Decentration and in- symbohzation In R. L. Schiefelbusch & D O. Bricker
tegration of the child's play in the second year. Child (Eds.), Early language Acquisition and intervention (pp.
Development. 51, 171-178. 287-329). Baltimore, MD. University Park Press.
Inhelder, B., Lezine, I., Sinclair, H , & Stambak, M. (1972).
Les debuts de la fonction symbohque [The beginnings
of the symbolic function] Archives de Psychologie, 41, Received May 18, 1982
187-243. Revision received February 4, 1983

You might also like