Content uploaded by Inge-Marie Eigsti
Author content
All content in this area was uploaded by Inge-Marie Eigsti
Content may be subject to copyright.
Abstract Language acquisition research in autism has
traditionally focused on high-level pragmatic deficits.
Few studies have examined grammatical abilities in
autism, with mixed findings. The present study
addresses this gap in the literature by providing a
detailed investigation of syntactic and higher-level dis-
course abilities in verbal children with autism, age
5 years. Findings indicate clear language difficulties
that go beyond what would be expected based on
developmental level; specifically, syntactic delays,
impairments in discourse management and increased
production of non-meaningful words (jargon). The
present study indicates a highly specific pattern of lan-
guage impairments, and importantly, syntactic delays,
in a group of children with autism carefully matched on
lexical level and non-verbal mental age with children
with developmental delays and typical development.
Keywords Autism Æ Language acquisition Æ Syntax Æ
Vocabulary Æ Pragmatics
Autism is a neurodevelopmental disability involving
severe and persistent deficits in multiple areas of
functioning. One of the hallmarks of autism is a
qualitative impairment in communication (American
Psychiatric Association, 1994). Communicative deficits
can range from mutism to adequate speech with poor
conversational skills, with this variability at its greatest
when all disorders on the autism spectrum are included
(Fombonne, 1999). Many children with autism are
initially referred for evaluation because of parents’
concerns about delayed language milestones (Dahlgren
& Gillberg, 1989), and the attainment of these mile-
stones appears to be strongly related to long-term
prognosis (Rutter, 1970; Stone & Yoder, 2001;
Szatmari, Bryson, Boyle, Streiner, & Duku, 2003).
Current descriptions of language in autism have
primarily focused on four areas: (a) Absence of verbal
abilities (e.g., the failure to acquire spoken language
during the lifespan), which is the outcome for between
50–75% of affected individuals (Rapin, 1991); (b)
Early language delays, with words first produced at an
average age of 38 months (Howlin, 2003), rather than
the typical time of 12–18 months. The presence of such
delays is one of the diagnostic criteria for autism; (c)
Atypical features of language production, including
echolalia and jargon (Tager-Flusberg & Calkins, 1990);
and (d) High-level discourse and pragmatic abilities
(Bartak, Rutter, & Cox, 1975; Bartolucci, 1982; Lord &
Paul, 1997). The latter two categories are considered in
more detail below.
Echolalia, the immediate or delayed echoing or
repetition of whole, unanalyzed utterances or interac-
tions, is observed in typically-developing children. In
autism, however, echolalia is present to a greater degree
and for a longer period of time. Indeed, with many
children with autism, a large proportion of their early
speech productions are echolalic (Prizant & Duchan,
I.-M. Eigsti Æ L. Bennetto
Department of Clinical and Social Sciences in Psychology,
University of Rochester, Rochester, NY, USA
M. B. Dadlani
Department of Psychology, University of Massachusetts,
Amherst, MA, USA
Present Address:
I.-M. Eigsti (&)
Department of Psychology, University of Connecticut,
406 Babbidge Road, Unit 1020, Storrs, CT 06269, USA
e-mail: inge-marie.eigsti@uconn.edu
J Autism Dev Disord
DOI 10.1007/s10803-006-0239-2
123
ORIGINAL PAPER
Beyond Pragmatics: Morphosyntactic Development in Autism
Inge-Marie Eigsti Æ Loisa Bennetto Æ Mamta B. Dadlani
Springer Science+Business Media, LLC 2006
1981). While the function of these echolalic utterances is
not well understood, it might serve several purposes,
communicative and otherwise. For example, children
might use echoing in conversation when unsure of their
response; as a familiar verbal ritual; or as a way of
holding information in memory.
The use of jargon, or nonsense words, has frequently
been reported in autism. Children with autism are
more likely to come up with idiosyncratic labels, to
invent nonsense terms with consistent meanings, and to
link phrases with atypical meanings (Lord & Paul,
1997). The production of jargon, like that of echolalia,
likely serves several functions. For example, jargon
could signal the presence of poor referential abilities; it
might signal a difficulty in updating representations
(e.g., if a child interprets a phrase incorrectly, he may
have difficulty encoding the more correct interpreta-
tion); or, as in echoing, it could serve as a communi-
cative ‘‘bridge’’ when a child is unsure of how to
respond.
In addition to jargon and echolalia, verbal individ-
uals with autism spectrum disorders frequently have a
unique speech style, marked by suprasegmental speech
qualities such as inappropriately soft, or, more fre-
quently, loud, speech volume; flat or singsong intona-
tion; hoarseness; hyper-nasality; and unusually fast or
slow speech rates (Shriberg et al., 2001). Speech is also
marked by overly formal or precise words, neologisms,
and odd phrasings (Rutter, Mawhood, & Howlin,
1992).
Pragmatic language use—that is, employing lan-
guage as a social system to communicate—is a domain
of significant impairment in ASD. High-level discourse
aspects of language in autism include difficulty with
turn-taking behaviors; interpreting statements in an
overly literal fashion (e.g., responding to the literal
meaning of metaphors or not catching the underlying
meaning of irony or sarcasm); responding in conver-
sation without regard for the Gricean maxims of
quality, quantity, relevance, and manner (Grice, 1975);
and difficulty in structuring narratives (Capps, Losh, &
Thurber, 2000; Diehl et al., 2006). Individuals with
autism of all ages are likely to use words with an
inappropriate level of conversational formality (i.e.,
register), leading to somewhat pedantic, precise speech
(Lord & Pickles, 1996).
Learning the grammatical structure of a language
involves learning to combine words into phrases;
learning grammatical categories (e.g., noun, verb,
object, agent, Bloom, Rocissano, & Hood, 1980); and
learning to use the grammatical elements of language
(e.g., morphemes such as –ing, –ed, or cat, which are
words or parts of words that carry grammatical
meaning; Brown, 1973; de Villiers & de Villiers, 1973).
While an extensive literature examines the relationship
between social impairments in autism and impairments
in pragmatic and discourse aspects of language
(Baltaxe & D’Angiola, 1997; Ozonoff & Miller, 1996;
Shriberg et al., 2001; Tager-Flusberg & Anderson,
1991), there has not been a similarly in-depth explo-
ration of syntactic development in autism, nor how it
may relate to underlying cognitive impairments.
Several studies have examined the acquisition of
grammatical morphemes in children with autism and
found no differences between children with autism and
typical control children matched on nonverbal mental
age (Fein & Waterhouse, 1979, October; Howlin,
1984a, 1984b). Similarly, one longitudinal study found
few differences between children with autism and
mental age-matched Down syndrome and normal
controls in grammatical complexity in productive lan-
guage (Tager-Flusberg et al., 1990) using the Index of
Productive Syntax (Scarborough, 1990). In contrast,
however, Bartolucci and colleagues found that children
with autism were less likely than mentally retarded and
normal control participants to produce grammatical
morphemes, especially verb tense and articles
(Bartolucci, 1982; Bartolucci & Albers, 1974;
Bartolucci, Pierce, & Streiner,
1980). Dalgleish (1975)
has suggested that syntactic deficits in autism are
related to deficits in the ability to sequence stimuli, or
to learn rules for ordering stimuli. One study found
that children with autism differed from verbal mental
age-matched children with typical development and
Down syndrome in their comprehension of transitive
(the man put the glass on the table), but not intransitive
(the man arrived), sentences (Prior & Hall, 1979); the
latter typically emerge later in development. Several
recent publications suggest the presence of more sub-
stantive syntactic impairments in autism (Condouris,
Meyer, & Tager-Flusberg, 2003; Rapin & Dunn, 2003),
or in a subgroup of individuals with autism (Kjelgaard
& Tager-Flusberg, 2001). Thus, while many researchers
have concluded that syntactic development, contrasted
with pragmatics or discourse aspects of language
acquisition in autism, is concomitant with general
developmental progress, the findings to date are
equivocal and could be consistent with a specific
deficit.
One consideration with respect to the diversity of
findings to date is that deficits in autism may be masked
by matching procedures (Lord & Paul, 1997). Rela-
tively heterogeneous groups of children with autism
are often compared with more homogenous groups of
children with mental retardation. Because nonverbal
ability is often a strength in autism, and since groups
J Autism Dev Disord
123
are typically matched on verbal ability, the autism
group may contain children who are not delayed on
tests of nonverbal ability. The matching process may
then obscure the greater language deficits of partici-
pants in the autism group, relative to other cognitive
domains.
In addition, most studies examining syntactic
development in autism occurred prior to the advent of
rigorous, reliable diagnostic measures such as the
Autism Diagnostic Interview-Revised (Cox et al., 1999;
Lord, Rutter, & LeCouteur, 1994) and the Autism
Diagnostic Observation Schedule (Lord et al., 1989;
Lord, Rutter, & DiLavore, 1998) in the late 80’s and
early 90’s. A comprehensive investigation of language
in a well-controlled sample of children with autism,
compared with both typically developing and devel-
opmentally delayed groups, is needed.
The present study was designed to address the
conflicting and sparse literature on syntactic develop-
ment in young children with autism, investigating
whether children with autism exhibit syntactic abilities
commensurate with their developmental age. In addi-
tion to morphosyntactic or grammatical functioning,
we assessed a broad spectrum of communicative
abilities, including lexical knowledge; discourse or
turn-taking abilities; the occurrence of errors; the
occurrence of jargon and echolalic speech; and
discourse characteristics of adult interlocutors.
Methods
Participants
Participants in this study included children in three
groups: (a) children with autism ages 3–6 years; (b)
children with non-specific developmental delays (DD),
matched on nonverbal IQ, gender, and chronological
age; and (c) typically developing (TD) children
matched on non-verbal IQ and gender. Participant
information is presented in Table 1. Both the DD
and autism groups were verbal and relatively high-
functioning, with mean non-verbal IQ scores in the low
average range.
Autism Group
Interviews to confirm the diagnosis of autism were
conducted by the principle investigator using the
ADI-R and the ADOS (module 2) with all children
in the autism group. Both the ADOS and the ADI-R
were scored according to DSM-IV and ICD-10
criteria for autism disorder. Only subjects whose early
development and current level of functioning met
strict criteria for a diagnosis of autism on both the
ADOS and the ADI-R were included (see Table 2).
To be included, participants had to be producing at
least 2-word phrases; all but one of the ASD partici-
pants had been talking for at least 12 months at the
time of the study, and could be described as currently
verbal.
1
However, all children in this group had sig-
nificant early language impairments, and none could
be described as meeting criteria for Asperger syn-
drome or Pervasive Developmental Disorder/Not
Otherwise Specified.
Developmentally Delayed Comparison Group
Participants in the DD group were recruited from a
local school providing special education services.
Inclusion criteria were that children be receiving spe-
cial education through Early Intervention or the Board
of Cooperative Educational Services. Parents of all
participants completed the Child Behavior Checklist
(CBCL; Achenbach, 1991) to assess for comorbid
diagnoses. Where there was any reason to suspect
difficulties in social development, the ADI-R and
ADOS were administered (n = 2). None of the chil-
dren in this group had a history or current symptoms
consistent with ASD.
Typically Developing Comparison Group
Children were recruited for the TD group from the
community, and parents completed the CBCL. All
parents gave informed consent for their children to
participate in research.
Matching Procedures
Intellectual functioning was assessed with a short form
of the Stanford-Binet Intelligence Scale: Fourth
Edition (SB-IV, Thorndike, Hagen, & Sattler, 1986).
The nonverbal reasoning factor includes four subtests
(Bead Memory, Copying, Quantitative, and Pattern
Analysis; Sattler, 1992) appropriate for assessing
intellectual functioning in young children with devel-
opmental disabilities (Carpentieri & Morgan, 1994;
Lawson & Evans, 1996).
As stated above, participants were included only
if they were already combining words into at least
2-word phrases. Many previous studies have matched
1
When analyses were repeated, excluding the child who had
been talking for only 4 months, results were identical.
J Autism Dev Disord
123
participants on receptive vocabulary level (e.g., Mot-
tron, 2004), using it as a stand-in for overall language
abilities. To maintain consistency with this literature,
and to ensure that children were similar on a non-
syntactic verbal measure, participants were matched on
receptive vocabulary using the Peabody Picture
Vocabulary Test – Third Edition (PPVT-III; Dunn &
Dunn, 1997). Only children with a verbal mental age of
21 months or older on the PPVT-III were included.
There were no significant differences in age equiva-
lence scores across groups; mean group scores were in
the three and a half to 4-year-old range; see Table 1.
Using these matching criteria, children across groups
should have equivalent levels of receptive vocabulary,
although they may have reached this level at different
ages (e.g., groups were matched by receptive vocabu-
lary but differed in chronological age).
Socioeconomic status (SES) was calculated from a
four-factor index, in which parental educational
attainment and occupation are used to calculate a
weighted index, ranging from 8 to 66 (Hollingshead,
1975). Scores are reported in Table 1. There was a
main effect of Group for socioeconomic status
(SES), F(2,45) = 5.08, p < .01, and post hoc analyses
indicated that the DD group had a lower SES than the
autism group, t (30) = 11.31, p=.02, and the TD
group, t (30) = 13.38, p=.005. There was a main
effect of Group for Ethnicity distributions, v
2
(4,
n = 48) = 11.0, p < .05, such that the DD group had a
greater proportion of African-American participants
than the TD group, v
2
(2, n = 32) = 8.96, p=.01. SES
could potentially exert a significant influence on the
measures of interest (Hoff-Ginsberg, 1991), though this
pattern of group differences in SES would be predicted
Table 1 Demographic data
for autism, developmentally
delayed (DD), and typically
developing (TD) groups
p < .10, * p < .05,
** p < .01, *** p < .001
Autism M
(SD) Range
DD M
(SD) Range
TD M
(SD) Range
Group
differences
N 16 16 16
Gender (M: F) 11:5 14:2 12:4
Chronological Age*** (mos) 57.7 (11.9) 56.9 (9.7) 42.6 (5.7) TD < Aut, DD
39–78 38–79 33–50
Stanford-Binet Nonverbal IQ***
(Scaled Score)
80 (15) 82 (13) 100 (9) TD > Aut, DD
49–111 52–106 85–121
Stanford-Binet Nonverbal IQ
(Age equivalent score, months)
44 (11) 42 (7) 45 (6)
30–64 31–53 35–58
Peabody Picture Vocabulary Test
(Age equivalent score, months)
43.4 (14.0) 47.7 (14.5) 50.8 (6.9)
22–69 22–72 36–62
SES** (Hollingshead 4-Factor Index.
Larger numbers indicate higher SES;
Range: 8–66)
51 (10) 39 (15) 53 (13) Aut, TD > DD
32–66 16–57 27–66
Ethnicity* (White: African-American:
Hispanic)
14:1:1 9:5:2 16:0:0 TD ~ =DD
Table 2 Autism diagnostic measures (Autism group only)
ADI-R M
(SD) Range
Cut-Off
a
ADOS M
(SD) Range
Cut-Off
a
Communication 15.3 (4.3) 8 6.8 (1.5) 5
8–24 5–9
Social reciprocity 18.3 (4.9) 10 10.6 (2.7) 6
10–26 6–14
Repetitive behaviors/ interests 7.8 (2.4) 3 1.8 (1.3) N/A
b
3–12 0–5
Differences apparent (mos)
c
22.2 (8.7)
14–42
Age at which at least 5 words used meaningfully (mos) 28.6 (11.4)
10–48
Time between age of first words and age at assessment (mos) 27.4 (12.2)
4–44
a
For an autism spectrum diagnosis
b
No cut-off score is used as it is possible to meet criteria for an autism spectrum diagnosis on the ADOS without exhibiting repetitive
behaviors or stereotyped interests
c
Average age at which parents became aware that development was proceeding differently. For an autism diagnosis, differences must
be apparent prior to age three
J Autism Dev Disord
123
to lead to lower scores for the DD group. When
analyses were repeated with SES entered as a covari-
ate, results were identical. The present results are
based on analyses without covarying SES or ethnicity.
Procedure
Children participated in a 30-minute free play session,
which took place during a second visit to the lab
(standardized and diagnostic testing took place during
the first session). The testing room contained a stan-
dard set of toys and books, and the children and a
researcher played together with those toys. The care-
giver was outside the room for all free play sessions
(with the exception of one child from the TD group
who was unable to separate from his mother). The free
play sessions were typically fun for the children. All
interactions were videotaped through a one-way
mirror.
The child played together with the first author or a
trained research assistant. Although children may have
been more comfortable with their mothers and fathers,
the presence of a play partner that engaged with chil-
dren across groups in a standardized fashion was
important for maintaining the consistency of the play
sessions.
The play partner made an effort to engage the child
in play. Although initially the play partner followed the
child’s lead, if the child did not initiate interaction or if
the play ceased and the child became engaged in soli-
tary play, the play partner used a set of standardized
prompts to engage the child. Initially, play partners
commented on the child’s actions: e.g., ‘‘that looks like
a big dog.’’ This strategy was repeated up to five times.
If the child did not respond, the play partner then
asked a direct question (e.g., ‘‘where are you driving
the truck?’’). Comments and direct questions then
were alternated to stimulate conversation. The excep-
tion to this protocol was if the child began to engage in
inappropriate/potentially harmful actions (attempting
to open a closed cupboard, climbing up a bookshelf,
throwing hard objects). In order to include one ver-
bally-based interaction that was consistent across chil-
dren, the play partner encouraged the child to look
through and describe a wordless picture book, Good-
night, Gorilla, during the play session. All the children
engaged in this activity for at least several minutes.
Transcription of Free Play Session
To code data for analysis of language measures, all free
play sessions were transcribed from videotape (Brown,
1973; Demuth, 1996) in the format of the Child
Language Data Exchange System (CHILDES;
MacWhinney, 1991) using CLAN software, which
automates a variety of language analyses including
frequency counts, word searches, and co-occurrence
analyses (MacWhinney, 2000). Analyses were based on
a uniform number of utterances (100) across all chil-
dren. An utterance was jointly defined by intonation
contour and by the presence of a discernible pause
between it and surrounding utterances. For partially
unintelligible or semantically uninterpretable utter-
ances, phonetic representations were transcribed and
supplemented by the transcriber’s gloss. Compounds,
proper names, and ritualized reduplications were
counted as single words (birthday, Sally Smith, night-
night); fillers (mm) and single-word routines (yeah, hi)
were not included. Repetitions, within five utterances,
of self or interlocutor were not included.
A small number of children (n = 8) did not produce
100 qualifying utterances (see Table 3); analyses were
pro-rated for these children where appropriate. The
proportion of children producing less than 100 quali-
fying utterances did not differ by group.
Mean Length of Utterance (MLU)
MLU assesses the length, and thus the relative gram-
matical complexity, of a child’s utterances, by counting
individual morphemes. MLU is frequently used to
describe individual differences and developmental
changes in grammatical development, particularly for
early stages of language acquisition. MLU was calcu-
lated on the set of 100 utterances using the MLU and
FREQ routines within CLAN
2
(MacWhinney, 1991).
Index of Productive Syntax (IPSyn)
The IPSyn has been used with typically developing
children and those with developmental disabilities as a
means of evaluating syntactic development (Fowler,
1980; Scarborough, 1990; Scarborough, Rescorla,
Tager-Flusberg, Fowler, & Sudhalter, 1991). IPSyn
scores are likely a more sensitive measure of language
level than MLU, particularly for children at this devel-
opmental level, as they yield a fine-grained analysis of
disparate domains of syntax (Scarborough et al., 1991).
2
The MLU routine alone calculates only the number of words
per utterance, rather than the number of morphemes, which is
the variable of interest. Thus, after running MLU on each
transcript to find the number of utterances spoken by the child,
FREQ was used to find the complete frequency listing, assessed
word by word to determine whether each word consisted of one
or more morphemes. Additional morphemes were then incor-
porated into the measurement of utterance length.
J Autism Dev Disord
123
The transcripts were scored for the presence of 56
different syntactic and morphological forms of pro-
gressively greater complexity, ranging from one-word
utterances to fluent speech; see Appendix Table 7.
Each utterance was scored in turn and one point was
scored for each occurrence of a morphosyntactic
structure (maximum of two points per structure). For
example, the utterance ‘‘an iron?’’ would give credit
for the following: (a) intonational question; (b) use of a
noun; and (c) two-word combination of article plus
noun. A grammatical structure could be given credit
even if it was used inaccurately, i.e., the past tense
morpheme in ‘‘*It maked it a twirl thing.’’ Morpho-
syntactic structures were divided into four subscales
(Verb Phrases, Noun Phrases, Questions and Nega-
tions, and Sentence Structures), and summed for an
overall score.
Other Measures
In addition to the syntactic assessments, the tran-
scriptions were assessed on a variety of dimensions: (a)
Developmental scatter of grammatical structures; (b)
Grammatical errors; (c) Type-token ratio for lexical
items; (d) Jargon production; (e) Present/absent topic
use; and (f) Turn-taking as a measure of pragmatic
discourse ability. These analyses are described in detail
in the results.
Reliability
The free play sessions were transcribed by the first
author or one of two research assistants, and the first
author reviewed all transcriptions in full. Because the
experimenter had interviewed parents and worked
with the participants, she was not necessarily naı
¨
ve to
participants’ developmental status during transcrip-
tion. Thus, maintaining high standards for reliability
across naı
¨
ve and non-naı
¨
ve coders was particularly
important. Following (Demuth, 1996), 8% (n =6)of
the videotapes were independently transcribed for
reliability by two coders. Word for word reliability
(product-moment correlation; Cohen, 1960) was
r = .90, v
2
(1) = 8.96, p=.01. Reliability for coding of
other measures is described in the results.
Results
Prior to all inferential statistics, dependent variables
were examined for deviations from the assumptions of
normality and sphericity and were found to be nor-
mally distributed. In addition, analyses reported below
were repeated, with PPVT-III scores entered as a
covariate, to control for lexical differences. This addi-
tional analysis did not lead to changes in any findings,
with one exception in MLU findings, described below.
Mean Length of Utterance (MLU). Participants’
language was assessed with respect to MLU; see
Table 3. A one-way ANOVA revealed a significant
group difference in total MLU, F(2,45) = 3.78, p=.03.
Post hoc analyses indicated that the autism group mean
was significantly lower than the DD group, t (30) = .91,
p=.008. The autism vs. TD group comparison
approached significance, t (30) = .56, p=.09, and the
TD and DD groups did not differ. When MANCOVA
analyses were performed with PPVT-III Age Equiva-
lent score as the covariate, the significant main effect of
group was unchanged, F(2,43) = 3.12, p=.05. Simi-
larly, the autism group mean was lower than the DD
Table 3 Quantitative assessments of language ability
Autism M (SD)DDM (SD)TDM (SD) Group Differences
No. Utterances
b
94.63 (10.40) 97.38 (10.50) 92.81 (16.34)
Range 69–100 58–100 50–100
Mean length of utterance (MLU) 2.97 (1.15) 4.07 (1.17) 3.61 (1.10) Aut < DD**Aut < TD
Range 0.78–5.02 2.13–6.26 1.71–5.39
Index of Productive Syntax (IPSyn)
a
55.28 (17.99) 70.94 (14.05) 76.81 (15.30) Aut < TD*** Aut < DD**
Range 11–77 36–84 40–100
IPSyn age equivalent 28 months 35 months 41 months
Infit Mean Sq: IPSyn
Verb phrases 1.02 (.34) 1.06 (.38) .90 (.29)
Sentence structures 1.14 (.70) 1.0 (.48) .89 (.40)
Questions, negations 1.23 (.78) .81 (.49) .76 (.55) Aut > TD* Aut > DD
Noun Phrases 1.15 (1.14) .62 (.93) .21 (.29) Aut > TD*
a
Scores could range from 1 to 118
b
Number of utterances on the IPSyn (100 = maximum). The median in all groups was 100
p < .10, * p < .05, ** p < .01, *** p < .001
J Autism Dev Disord
123
group mean, F(1,29) = 6.23, p=.02. The ASD-TD and
TD-DD contrasts did not differ, F(1,29) = 1.45,
p = n.s., and F(1,29) = 2.07, p = n.s., respectively.
IPSyn. A one-way ANOVA on IPSyn scores
revealed a significant group difference in the total IP-
Syn score, F(2,45) = 7.88, p < .001. Post hoc analysis
revealed that the mean for the autism group was sig-
nificantly lower than both the DD group, t
(30) = 15.66, p=.008, and TD group, t (30) = 21.54,
p < .001. The algorithm estimating age equivalents for
these scores indicated that the autism group’s
utterances were at the developmental level of a
28-month-old, as compared to 35 and 41 months for
the DD and TD groups, respectively.
As described above, the grammatical analyses were
based on the first 100 scorable utterances. Most par-
ticipants produced 100 utterances (13/16 in the autism
group, 15/16 in the DD group, and 12/16 in the TD
group). The remaining participants had utterance
totals ranging from 50 to 90. Total IPSyn scores for
these participants were calculated by using a conver-
sion metric provided in (Scarborough et al., 1991).
Analyses of IPSyn and MLU results were recalculated
using only data from children with 100 utterances, and
results did not differ. The number of scorable utter-
ances did not differ by group, and thus, interestingly,
suggested no group differences in overall talkativeness.
A further analysis of syntactic complexity was con-
ducted on the four subscales of the IPSyn (Verb
Phrases, Sentence Structures, Questions and Nega-
tions, and Noun Phrases).
3
A 3 (Group) · 4 (Subscale)
mixed-model MANOVA revealed a significant main
effect of Group, F(2,37) = 88.19, p=.001, a significant
main effect of Subscale, F(3,37) = 81.31, p < .001, and
a trend towards a significant Group · Subscale inter-
action, F(2,37) = 2.85, p=.07. Follow-up ANOVAs
yielded significant group differences for the Verb
Phrases, F(2,39) = 4.17, p=.02, Sentence Structures,
F(2,39) = 4.77, p=.01, Questions and Negations,
F(2,39) = 8.40, p < .001, and Noun Phrases subscales,
F(2,39) = 4.17, p=.02. These data are presented in
Fig. 1.
Post hoc analyses revealed that the autism group
mean was significantly lower than the TD group mean
across all subscales: Verb Phrases, t (23) = 2.32,
p=.007; Sentence Structures, t (23) = 6.22, p=.004;
Questions-Negations, t (23) = 4.99, p < .001; and
Noun Phrases, t (23) = 2.32, p=.007. The autism
group used significantly fewer Question and Negation
structures than the DD group, t (25) = 3.50, p=.006.
There was a trend towards a significant group differ-
ence for the other three subscales: Verb Phrases,
t (25) = 1.50, p=.06; Sentence Structures, t (25) =
3.37, p=.09; and Noun Phrases, t (25) = 1.50, p=.06,
with the autism group consistently scoring lower than
the DD group. The DD and TD groups’ subscale
scores did not differ.
The results reported thus far indicate that children
in the autism group, despite being matched on both
nonverbal IQ and lexical knowledge, produced syn-
tactically less complex utterances than children in the
TD and DD groups.
Developmental Scatter. While this analysis of gram-
matical structures is informative, it does not take into
account the pattern of responses (Kaplan, Fein, Morris,
& Delis, 1991). The order of items on the IPSyn rep-
licates the typical order of acquisition of those items, so
that a typical pattern of responses of a given individual
at a single timepoint reflects a progression from higher
to lower scores on within-scale items. In other words,
children typically learn simpler items before they learn
items of greater complexity, and thus are likely to
produce the initial items in the scales before producing
later items. Group differences in total IPSyn scores
might reflect simple delays in grammatical knowledge
or learning, but could also reflect a different develop-
mental progression for children in the autism group.
To address this question, we employed analytic
procedures assessing ‘‘intrasubtest scatter,’’ (based on
item response theory) for detecting unusual response
sequences. The most sensitive of these is an index of
inconsistent responding, calculated using the partial-
credit model of Rasch analysis (Adams & Khoo, 1993;
Godber, Anderson, & Bell, 2000); one advantage of
this index is that it is not confounded with total sub-
scale scores. Conceptually, the procedure estimates an
individual’s overall ability from the total score, then
observes the interaction between this ability level and
the difficulty of the items observed. This interaction
between a particular subject’s observed and expected
patterns of responding, given his overall ability, is
calculated as the infit mean square. Individuals with
response patterns that fit well (e.g., predictably) with
their overall score are represented by an infit mean
square score of zero; response patterns that fit poorly
(because of inconsistent responding) are represented
by larger values.
The infit mean square scores were calculated sepa-
rately for each IPSyn subscale for the autism, DD, and
TD groups; see Table 3. Results were subjected to a 3
(Group) · 4 (Subscale) mixed-model MANOVA,
3
For participants who produced fewer than 100 utterances,
subscale scores were not computed (because of potential differ-
ences across groups on subscales). Thus, the subscale analyses
are likely more conservative since they were based on a smaller
group of participants.
J Autism Dev Disord
123
which revealed a significant main effect of Group,
F(2,35) = 5.71, p=.007 and a significant main effect of
Subscale, F(3,105) = 2.80, p < .05. The Group · Sub-
scale interaction was not significant. Post hocs indi-
cated that the pattern of responses for the autism
group was significantly different from the DD and TD
groups, p=.05 and p=.002, respectively, and the
latter two groups did not differ.
Follow-up ANOVAs on the individual subscales
yielded a significant group difference for the Noun
Phrases, F(2,35) = 3.59, p=.04, and a trend for a dif-
ference within the Questions and Negations subscale,
F(2,38) = 3.11, p=.06. The Verb Phrases and Sen-
tence Structures subscales did not differ across groups.
The children with autism had response patterns that
were significantly different than in the TD group for
both the Questions-Negations, t (24) = 2.08, p < .05,
and Noun Phrases subscales, t (23) = 2.76, p=.01.
There was a trend for a group difference between the
autism and DD groups for the Questions-Negations
subscale, t (26) = 1.91, p=.07. The DD and TD
groups did not differ. In summary, the developmental
scatter analysis indicates that the children with autism
had response patterns for the IPSyn which differed
significantly from responses in the comparison groups.
Types and Tokens in Word Production. Receptive
vocabulary (language comprehension) was assessed
with the PPVT-III, on which all groups were similar. In
the spoken language domain, one common measure of
spontaneous language use is the contrast between the
sheer number of words, or tokens, spoken during a
specified period, and the number of different words, or
types, produced in that same period. This is a way of
quantifying the variety of different words used by a
talker, while equating for talkativeness. The Type to
Token ratio was estimated using the FREQ routine
within CLAN. Results indicated that children in all
three groups produced similar numbers of Word
Types, F(2, 45) = .093, p=.91, with means (SD) of 178
(58) , 186 (47), and 186 (62), for the autism, DD, and
TD groups, respectively. Children also produced simi-
lar total Tokens: 459 (181), 550 (183), and 532 (224),
F(2, 45) = .96, p=.39. This supported the finding
(described in the IPSyn section) that children across
groups were equally talkative. The effect of Group on
the ratios of Types to Tokens did not reach signifi-
cance, F(2, 45) = 2.47, p=.096. Because there was a
trend for a group difference in the Type-Token ratio,
post hoc analyses were conducted and indicated that
the autism group ratio was higher than in the DD
group, t (30) = 2.22, p=.04, which suggests that the
children with autism tended to produce a greater
variety of different words than children in the DD
group.
Error Analysis
Although groups differed with respect to the relative
complexity of syntactic constructions, these differences
might be masking differences in utterances that are
produced with errors. The IPSyn assigns credit for
producing a particular structure even if it is ungram-
matical (e.g., credit for past tense is assigned if a child
says, ‘‘they wented outside’’). When errors are pro-
duced, they may reflect an incomplete or tentative
grasp of the particular linguistic structure. To assess
whether group differences in syntactic complexity may
have been influenced by ungrammatical speech, we
Fig. 1 Index of Productive
Syntax (IPSyn) Subscale
Scores Index of Productive
Syntax scores by group. Only
scores for participants who
made the full 100 utterances
were included. Sample sizes
for this analysis were: Autism
n = 12, DD n = 15, TD
n =13
J Autism Dev Disord
123
examined the frequency of grammatical errors and the
types of errors produced. Because the focus was not on
exhaustively identifying every possible error, but
rather on establishing possible links between error
rates and syntactic deficits, inclusion criteria were
strict. Thus, transcripts were examined, utterance by
utterance, and only clear, unambiguous errors of
omission and errors of commission (Bates, 1997) were
included. Results and examples are presented in
Table 4.
Findings indicated that the DD group exhibited
significantly more errors of omission than both the
ASD and TD groups (p’s = .06 and .003, respectively),
whereas there were no differences in errors of com-
mission (all p’s > .5). To summarize, children in the
DD group were more likely to omit required gram-
matical structures than children in the autism and TD
groups.
Nonsense Words (jargon)
Play session transcripts were analyzed for the presence
of jargon, defined as intelligible but uninterpretable
words or phrases. Any words or phrases that the
transcriber was able to hear, but was not able to
supply a gloss or meaning for, were included, such as,
‘‘the serpice [sic] is flying.’’ This definition was de-
signed to distinguish utterances in which a child mis-
pronounces an item (as in, ‘‘goin’ to make a lelivery,’’
when context and previous utterances indicate that the
child’s target was ‘delivery’), from utterances where
the child seems to be producing a novel wordform.
4
Results indicated that the incidence of jargon was as
follows: Autism group, Mean (SD) = 4.9 (5.3),
range = 0–17; DD group, Mean (SD) = 1.3 (1.6),
range = 0–5; TD group, Mean (SD) = .3 (.6),
range = 0–2. Groups differed significantly, F(2,
45) = 9.005, p < .001, and post hocs indicated that
children with autism produced more jargon than
children in the DD group, p < .009, and TD group,
p < .001.
Topic Analysis
One challenge for the present findings is that children
in the autism group may have produced less complex
grammatical structures because the topics they dis-
cussed were less complex. For example, to describe
people or events that are not present, specific gram-
matical structures (e.g., subjunctive or past tense) are
required. Talk about non-present events or people will
entail relatively more complex structures, and differ-
ences in such talk may be ascribed to conceptual ref-
erential grounds rather than morphosyntactic
knowledge. Thus, syntactic delays in autism may reflect
conceptual rather than grammatical delays. In general,
evidence from typical development indicates that
children are more likely to use here-and-now (vs.
displacement) language than are adults (Wanska &
Bedrosian, 1986).
The transcripts were examined for references to
items, events, or people not physically or temporally
present (Foster, 1986). This referential domain was
chosen because children at this developmental stage
were likely to be able have this conceptual ability.
Specifically, transcripts were assessed, utterance by
utterance, for the presence of an antecedent which was
identifiable as physically present. For example, in this
dialogue, a child is pulling toys from a toybox and
remarking on them to the experimenter (EXPTR). All
references were to the here-and-now:
CHILD: what’s dis [: this]?
EXPTR: I don’t know, what do you think?
CHILD: a dum(p) t(r)uck.
EXPTR: yeah!
CHILD: a big dump tuck [e.g., truck] wif [e.g., with]
that big ting [thing].
CHILD: got big things on it.
In contrast, in the following dialogue, the child (age
three) refers to two events that may transpire in future
(turning five years old, and riding a bus):
CHILD: I’m gonna be five like Sally, too!
EXPTR: five?
CHILD: yeah. Sally ‘s five more. She’s, she’s still five.
EXPTR: yeah? well it’s fun to be little sometimes, too,
dontcha think?
CHILD: [chuckling].
EXPTR: so what other things are you gonna do when
you’re five?
CHILD: um, I’m gonna go on the bus.
EXPTR: on the bus? wow.
CHILD: that’s when I’m five
While this rather coarse conflation of many syntactic
categories and types of utterances obscures finer issues
about the development of these categories, and may in fact
not reflect accurately the complexity demanded by
4
In this scheme, some utterances are likely to be erroneously
labeled as jargon, when the experimenter fails to recognize a
target utterance even though it is a known word. Utterances
from children with poor articulation will be over-included,
making comparisons between the autism and DD group more
conservative, as the latter are likely to have misarticulations.
J Autism Dev Disord
123
discussion of present-tense events in the here and now, we
depend on this analysis simply to highlight a conceptual
question presented by these data: Do children with autism
in this sample use less complex syntax because their syn-
tactic knowledge is more limited? Or, do the children in
this sample think about a more simplified/limited set of
concepts, and thus, is their language more grammatically
simple because of these conceptual limitations?
Children were scored in a dichotomous fashion, as
either making this type of non-present reference, or
failing to produce a single identifiable example.
Because it can be difficult to reliably determine whe-
ther a given phrase refers to a distal or here-and-now
event, our criteria were as conservative as possible.
Specifically, we used a simple dichotomous analysis;
rather than making subtle distinctions about the qual-
ity or quantity of such references, we scored whether or
not a child produced at least one clear example of a
non-present reference. Results indicated that the chil-
dren with autism were the least likely to refer to things
not physically present (n = 2 children), followed by
children in the DD group (n = 5) and the TD group
(n = 9), a main effect that was significant, X
2
(2) = 6.94, p=.03.
Discourse Analysis
An important component of language development is
learning to use language as a tool for social communica-
tion; as discussed previously, this is typically an area of
difficulty in autism. One conventional method for assessing
discourse abilities is to examine the conversational ‘‘turns’’
that typically make up an interaction (Bloom, Rocissano,
& Hood, 1976). Utterances in each transcript were coded
line by line for discourse function, as follows: Interactive
categories: (a) new topic initiations; (b) direct reply to
interlocutor, including nonverbal responses such as a head
shake; and (c) expansion of one’s own (previous) utter-
ance. Discourse-interrupting categories included: (d) direct
echo (with identical prosody) of self or other within five
utterances; (e) Failure to respond to direct query of
interlocutor (ignore); (f) Uninterpretable (though intelli-
gible) comments, the discourse function of which we (and
presumably, also the play partner) were unable to cate-
gorize.
Results are presented in Table 5, on the ‘‘Child’’
lines. A repeated measures MANOVA revealed a
significant main effect of Discourse Category,
F(1,44) = 141, p < .001 and a Group · Category
interaction, F(2, 44) = 3.73, p=.03. Follow-up analy-
ses indicated that the children with autism differed
from children in both comparison groups in the three
Discourse Interrupting categories (echo, F(2,
44) = 6.43, p=.004; ignore, F(2, 44) = 9.89, p < .001;
and uninterpretable, F(2, 44) = 5.62, p=.007), but had
no specific group deficits in the three Interactive
categories (see Table 4 for groupwise comparisons).
The DD group produced more expansions and
Table 4 Error types and
error rates by group
* p < .05
a
DD > TD, p < .01;
DD > A, p < .05
b
DD > TD, p < .05;
DD > A, p < .10
Autism DD TD
Mean (SD) errors of omission (includes
failure to invert phrases) *
a
2.2 (1.5) 3.3 (1.9) 1.5 (1.6)
Proportion errors of total utterances*
b
.023 .036 .016
Possessives I go over [your] house 020
Verbal auxiliary what she plays? 567
Determiners get this up in sky 521
Main verb it a toys 330
Verb marking he bite your finger 473
Prepositions you crashed me 160
Negation marking He likes yucky things but she likes not
yucky things
011
Mean (SD) errors of commission 1.1 (1.2) 1.4 (1.7) 1.2 (1.0)
Proportion errors of total utterances .011 .015 .012
Possessives mine snake is breaking 001
Pronouns she (referring to child’s father) 2 0 4
Verbal auxiliary he’s took his keys 025
Determiners I got a Joey in my class 411
Noun plurals that’s a little people (for a single toy) 3 0 1
Overextension he’s upping it, they’re unlocking them out 221
Verb marking remembers I saw that 223
Adjective/ adverb well, it pretty hurt 021
Prepositions the bigger one was full with books 120
Other who’s under here the boots
(sees feet under door)
211
J Autism Dev Disord
123
initiations, and the TD group produced more replies.
The results indicated that the autism group was more
likely to produce atypical utterances that do not
further the flow of conversation, although they are not
less likely to engage in more typical ways as well.
Another important element of the discourse analysis
is to ask whether the adults across the three groups
responded to or initiated conversations differently.
This could both affect children’s responses, and be the
result of group differences in how children talk to the
adults. Adult utterances were categorized into
the same categories as child utterances (see Table 5).
Results showed a main effect of Category, and a
Category · Group interaction. Follow-up analyses
indicated a trend for a significant group effects in
Expansions, F(2, 44) = 2.63, p=.08, and Replies, F(2,
44) = 2.72, p=.08. Post hoc analyses showed that the
adults interacting with children in the autism group
produced more expansions of their own utterances
than for the DD group, t(29) = 2.18, p=.04, and fewer
replies to the children’s utterances, t(29) = 2.11,
p=.04.
Previous work on discourse differences in autism
(Curcio & Paccia, 1987) has shown that particular sit-
uational contexts as well as conversational partners can
influence communicative skills in children with autism.
Although an in-depth analysis of the moment-to-
moment discourse context is beyond the scope of the
present study, an assessment of a subset of five ran-
domly-selected participants in each group demon-
strated that most children spent the bulk of their time
in playing directly with toys [Autism group, Mean
(SD) = .66 (.28); DD group, Mean (SD) = .85 (.14);
TD group, Mean (SD) = .72 (.15)]. Groups did not
differ on these measures, all p’s > .12.
Coding of discourse categories was done by two
independent raters (IME and MD), and 8% (n = 6) of
the transcripts were coded by both raters. Inter-rater
consistency was calculated at r = .923, with a net
agreement of .887; v
2
(1) = 375.7, p < .001.
Correlational Analyses
To determine the relationships between non-verbal IQ,
IPSyn scores, and interrelationships among variables,
the data were subjected to a planned series of corre-
lational analyses, focusing on within-group correlations
for the ASD participants to reduce Type I errors; see
Table 6.
Syntactic ability (IPSyn score) was correlated with
other language measures across the entire sample,
including lexical abilities (PPVT-III, Type-token
ratio), and pragmatic abilities (Ignoring; Expansions).
Although correlational data cannot determine causal-
ity, this finding is consistent with the suggestion that
having stronger syntactic abilities facilitates one’s
acquisition of new words, as well as one’s ability to
participate more effectively in interactions.
Interestingly, the production of jargon was nega-
tively correlated with NVIQ, with IPSyn score, with
lexical measures (PPVT-III and Type-Token Ratio),
and with discourse functioning (Ignore), suggesting
that jargon is linked both to language ability as a whole
as well as to cognitive functioning.
The conceptual ability to discuss non-present
objects, people, or events, seemed to be a language-
specific ability. It was not correlated with non-verbal
IQ, r (48) = .15, p > .3, but was correlated with IPSyn,
r (48) = .29, p < .05, and with the production of jargon,
r (48) = –.39, p < .01, computed across all subjects.
Table 5 Discourse characteristics
Autism M (SD)DDM (SD)TDM (SD) Group differences
Interactive
Child initiation** .068 (.036) .14 (.075) .083 (.058) DD > TD**, Aut**
Adult initiation .074 (.037) .062 (.037) .062 (.033)
Child reply** .60 (.18) .570 (.15) .74 (.13) TD > DD**, Aut*
Adult reply .46 (.16) .59 (.16) .51 (.17) Aut < DD*
Child expansion* .15 (.097) .23 (.13) .14 (.10) DD > TD*, Aut*
Adult expansion .44 (.14) .33 (.16) .41 (.13) Aut > DD*
Discourse interrupting
Child echo*** .06 (.036) .026 (.026) .019 (.02) Aut > TD**, DD**
Adult echo .027 (.025) .02 (.016) .014 (.013)
Child ignore*** .046 (.032) .014 (.015) .013 (.013) Aut > TD***, DD***
Adult ignore .0021 (.0043) .0028 (.0048) .0008 (.002)
Child uninterpretable** .058 (.078) .014 (.012) .007 (.013) Aut > DD**, TD**
Adult uninterpretable .0003 (.0013) .0001 (.0005) .0004 (.0011)
Note: Figures represent rates of occurrences of the given discourse type per total number of turns (e.g., possible occurrences)
p < .10, * p < .05, ** p < .01, *** p < .001
J Autism Dev Disord
123
More generally, given previous findings of specific
language subtypes within a large sample of children
with autism (Kjelgaard & Tager-Flusberg, 2001), par-
ticipants in the autism group were divided into two
language groups, based on PPVT-III scaled scores:
those below 85, n = 5, vs. those above 85, n = 10 (the
one child whose PPVT-III scaled score of 65 was below
70 was excluded from this analysis). For each sub-
group, we subjected the pattern of scores on the IPSyn,
non-verbal IQ, and topic, discourse, and error analyses
to a one-way ANOVA with subgroup as the between-
subjects variable. Findings indicated no subgroup dif-
ferences (all p’s > .29), with four exceptions: (a)
Nonverbal IQ, F(1,13) = 29.9, p < .001; Borderline
group, M (SD) = 63.8 (9.0); normal group, M
(SD) = 89.9 (8.6); (b) Jargon production,
F(1,13) = 14.75, p=.002; Borderline group, M
(SD) = 8.4 (5.3); normal group, M (SD) = 1.9 (1.2); (c)
Uninterpretable utterances in the discourse,
F(1,13) = 4.8, p < .05; Borderline group, M
(SD) = .114 (.115); normal group, M (SD) = .027
(.033); (d) Errors of omission and commission (com-
bined), F(1,13) = 4.58, p=.05; Borderline group, M
(SD) = 1.8 (1.3); normal group, M (SD) = 3.9 (2.0).
Consistent with some previous reports (e.g., Jarrold,
Boucher, & Russell, 1997), these data do not suggest
specific language-impaired and language-typical sub-
groups within a relatively small but homogenous group
of children with autism; rather, these data are consis-
tent with other results in indicating that morphosyn-
tactic abilities were specifically lower in the autism
group.
Discussion
To investigate morphosyntactic development in aut-
ism, 5-year-old children with autism and general
developmental delays, and 3-year-old children with
typical development, all matched on non-verbal mental
age, participated in a free play session that was tran-
scribed and analyzed. Receptive vocabulary was
assessed with the PPVT-III.
The most striking finding from the present study was
the clear presence of syntactic deficits in the autism
group. Controlling for lexical knowledge and nonver-
bal IQ, and overall talkativeness, the children with
autism produced language that was significantly less
complex than might be expected for their develop-
mental level. This result expands upon the 1982 finding
by Bartolucci and colleagues that children with autism
produced fewer grammatical morphemes than con-
trols. Even a relatively gross measure of syntactic
ability, MLU, supported the finding of autism-specific
syntactic delays, showing shorter MLUs in the autism
group compared with the DD group, and a trend for a
shorter MLU than the TD group. While the Autism/
TD difference was only marginally significant, post hoc
power analyses indicated that the present finding may
not have had sufficient power to detect an effect
(1-b = .47), whereas a more sensitive tool such as the
IPSyn was able to demonstrate a group difference.
The results indicated that the autism group likely
reached their syntactic abilities via an atypical devel-
opmental pathway. Individual IPSyn items in the
autism group followed a pattern that was marked by
significant inter-item inconsistency. Although the data
are cross-sectional in nature and do not directly
examine development, they suggest that children in the
autism group were not progressing in the typical
pathway from simpler forms to increasingly complex
ones.
In contrast to syntactic development, the present
study demonstrated that lexical knowledge was an area
of relative strength for the young children with autism.
In addition to group matching for receptive language,
an analysis of the free play sessions indicated that the
autism group produced as many different word types as
peers, and potentially an even richer set of words
(higher Type-Token ratio) than their mental-age-
matched peers without autism. Consistent with previ-
ous studies, the children with autism seemed to
comprehend and produce as many words as their peers.
While some aspects of lexical knowledge in autism are
atypical, including the use of idiosyncratic meanings
for words and of neologisms (Rumsey et al., 1985;
Rutter, 1970; Volden & Lord, 1991), adults with autism
often have larger vocabularies than would be predicted
from their other language abilities (Lord & Paul, 1997).
Studies of individual and developmental differences
indicate that word learning is highly correlated
Table 6 Relationships among language and other measures for
the ASD group
ASD Group
IPSyn – PPVT .475
IPSyn – Type–Token ratio –.904***
IPSyn – Ignore (Child) –.364
IPSyn – Expand (Child) .398
Jargon – NVIQ –.594*
Jargon – IPSyn –.338
Jargon – PPVT –.554*
Jargon – Type–Token ratio .282
Jargon – Scatter (NP) .577*
Jargon – Ignore (Child) .155
Note: Correlations are reported as r (16)
p < .10, * p < .05, ** p < .01, *** p < .001
J Autism Dev Disord
123
with short-term phonological memory capacity. It is
possible that the strong lexical skills seen in the autism
group in the present study may be tied to the relative
strengths in short-term memory ability characteristic of
the disorder (Bennetto, Pennington, & Rogers, 1996;
Eigsti & Bennetto, 2001; Hermelin & O’Connor, 1975).
In the course of learning to talk, children are likely
to omit required grammatical elements and to make
grammatical errors. Sometimes these errors are seen as
evidence of underlying syntactic knowledge (e.g., the
presence of past tense overregularization errors like
‘‘wented’’ suggests that a child has learned the general
form of the past tense). Children in the autism group
were not more likely to make errors, while children in
the DD group made significantly more errors of
omission (not included or scored in the IPSyn). Inter-
estingly, this conflicts with a previous report, that
children with autism were more likely to omit such
morphemes (Bartolucci et al., 1980), a discrepancy that
may lie in the disparate ages of children studied. Errors
(which were similar in frequency to those observed in
studies of typical language acquisition; e.g., Marcus
et al., 1992; Rubino & Pine, 1998) are thus not the
source of group differences in morphosyntactic abili-
ties. Clearly, then, the syntactic deficits found here
were not due to language that was more error-prone, as
children with autism were no more likely to omit
required particles, to fail to invert phrase orders, or to
produce incorrect structures or orders.
One possible explanation for syntactically less
complex language is that children with autism could be
producing many more neologisms or ‘‘jargon words,’’
as has been described in the literature. For example,
children could be saying nonsense words in place of a
variety of morphosyntactic elements, and in this way
achieving lower scores on the IPSyn. Results indicated
that children with autism did produce significantly
more jargon than children in the DD and TD groups
(a mean of 5 compared to a mean of 1.3 and 0.3,
respectively) and that the amount of jargon was neg-
atively correlated with syntactic abilities (IPSyn score).
However, jargon production was also correlated with
cognitive abilities (non-verbal IQ). While the presence
of jargon or neologisms in the speech of children with
autism confirms previous findings, it alone cannot
account for syntactic deficits.
There is a second explanation that could partially
account for the present syntactic impairments. Chil-
dren are increasingly likely to talk about things that are
spatially and temporally removed as they grow in lan-
guage skill and cognitive skill; more complex referen-
tial discourse entails more complex syntax. While this
cross-sectional dataset can not determine causality in
this chicken-and-egg problem, the children with autism
as a group made significantly fewer mentions of non-
present events and objects, and furthermore, those
mentions were correlated with IPSyn score and jargon
productions (i.e., language measures) but not with non-
verbal IQ. The fact that children across groups were
matched on non-verbal cognitive ability argues against
the hypothesis that syntactic complexity was limited by
cognitive processing of events. However, the data raise
the possibility that the children with autism scored
lower on the IPSyn because they were talking about
less complex events, or (conversely) that their syntactic
limitations prevented the discussion of such topics.
Pragmatics and discourse aspects of language pro-
vide the most frequently-discussed aspect of language
deficits in autism (Baltaxe & D’Angiola, 1997; Kelley
et al., 2006; Ozonoff & Miller, 1996; Shriberg et al.,
2001; Tager-Flusberg & Anderson, 1991). Interestingly,
the present results indicated that children with autism
produced as many of the conversation-supporting turns
as children in comparison groups, suggesting that there
were no specific impairments in the ability to initiate
new topics, to reply to an interlocutor’s comments or
questions, or to expand upon one’s own utterances, for
children at this verbal level. However, as expected,
children with autism were significantly more likely to
produce utterances that did not contribute to the dis-
course—they were less able to participate in the ‘‘to
and fro’’ of conversation. Specifically, they were more
likely to echo their own or their interlocutor’s utter-
ances; they were more likely to ignore or fail to
respond to a direct query; and they were more likely to
produce utterances whose discourse features were
uncategorizable. This latter category included intelli-
gible but uninterpretable utterances (e.g., jargon).
Thus, the conversation of children in the autism group
was typical in many respects, but contained additional
atypical elements. This finding is consistent with pre-
vious research on discourse structure in children with
autism at roughly the same developmental level
(Tager-Flusberg & Anderson, 1991). While addressing
the question goes beyond the scope of the present
paper, this finding raises an interesting issue for future
research, of whether these ‘‘discourse-interrupting’’
utterances occurred more frequently in particular dis-
course contexts.
An important counterpart to the analysis of chil-
dren’s discourse is to ask whether the adults across the
three groups interacted differently with children across
groups. As discussed in the Methods, play partners
consisted of the first author or one of several trained
research assistants, all of whom followed a relatively
standard set of guidelines on how to respond and
J Autism Dev Disord
123
initiate interactions. Marginally significant group
differences suggested that adults interacting with chil-
dren in the autism group produced more expansions of
their own utterances, and fewer replies to the chil-
dren’s utterances. Adults may have been less likely to
reply to children who produced more uninterpretable
utterances, jargon, and echolalia. Overall, the adult
discourse analysis suggests that adults interacted simi-
larly with children across groups, to the extent that we
were able to measure differences.
One limitation of the present findings draws on the
dichotomy between competence and performance.An
individual’s speech at any given moment (perfor-
mance) may not be an accurate index of underlying
knowledge (competence) of the syntax of the
language (Chomsky, 1957). The presence of speech
errors in everyday speech (e.g., ‘‘a five pile car up’’)
does not suggest that people have not mastered their
syntax, but rather that there is ‘‘slippage’’ or noise in
the process of producing an utterance. This raises the
possibility that children with autism may have a
greater underlying competence, or knowledge of the
syntactic structures of English, but that they are less
able or less willing (e.g., because of deficits in social
reciprocity) to access this knowledge in some con-
texts. For example, they could be less comfortable
with a relatively unfamiliar experimenter, and thus be
less likely to produce syntactically complex structures.
In support of this hypothesis, children with autism
performed as well as controls on a measure (the
PPVT-III) that required non-verbal rather than spo-
ken responses. By including other standardized mea-
sures of syntactic skills, it would have been possible to
directly address this concern; however, to our
knowledge, no standardized measures are appropriate
for assessing syntax at this very early stage of lan-
guage acquisition. The CELF-Preschool, for example,
which targets children ages 3–6, is often too difficult
for children with autism of this age (e.g., Kjelgaard &
Targer-Flusberg, 2001).
There are, however, several sources of evidence
against the ‘‘performance deficit’’ as the sole expla-
nation for syntactic delays in the autism group. First,
children performed another non-verbal language task
(not described here, Eigsti & Bennetto, 2001), and
the autism group exhibited specific syntactic impair-
ments on that task. Second, the children with autism
were equally as talkative as their peers, and talked
about similar topics. A third argument comes from
the broader language acquisition literature. Seiden-
berg and MacDonald (1999) have suggested a per-
formance-based alternative to Chomsky’s generative
approach. In their probabilistic framework, language
production and acquisition emerge as learners and
talkers exploit multiple probabilistic constraints
drawn from both linguistic and nonlinguistic infor-
mation. Under this framework, production data
observed in the course of language acquisition are
informative about how children extract the underly-
ing regularities of the language system. This
approach suggests that the less complex syntax
observed in language of children with autism may
reflect actual delays in their knowledge, rather than
‘‘performance factors.’’
The present data, which were collected from a single
play session, present a further limitation. The children
with autism were somewhat acclimated to the exam-
iner and the physical environment of the play-room,
because the play session was recorded during their
second visit to the lab. Nonetheless, children with
autism may have greater difficulty adjusting to this still
relatively novel situation. This play session may have
thus elicited somewhat less complex or typical spon-
taneous language from this group of children. Future
studies may address this concern by performing a more
extended familiarization procedure or, alternatively,
by documenting children’s speech and language in
the home setting, where they are likely to be most
comfortable.
One issue that must be raised in considering how
generalizable the study’s findings may be is whether
the participants are characteristic of the autism pop-
ulation. The ASD group was comprised solely of
high-functioning children with autism, not other
diagnoses such as PDD/NOS or Asperger syndrome.
The children all had a history of language delay, as is
characteristic of autism, but were all verbal at the
time of the study; inclusion criteria required that
participants have at least 2-word phrases in their
speech at home. Certainly, given the wide spectrum
of abilities in autism, it may be the case that the
syntactic deficits identified here may be less apparent
in a sample of children with diagnoses other than
classic autism, or in a sample of children with a wider
range of developmental abilities. In addition, some
research suggests the presence of several distinct
subtypes of children with ASD, where one subgroup
has deficits that parallel those observed in children
with specific language impairment (SLI), whereas
other subgroups exhibit few morphosyntactic delays
(Kjelgaard & Tager-Flusberg, 2001). Analyses exam-
ining the possibility that participants in the autism
group could be best characterized by normal and
language-impaired subgroups were not consistent
with these previous findings. Instead, the data indi-
cated a rather consistent morphosyntactic deficit
J Autism Dev Disord
123
across participants with autism, independent of
functional level, consistent with least one large study
of language profiles in autism (Jarrold et al., 1997).
Furthermore, the focus in this study was on identi-
fying deficits that characterized children with autism
as a group, and, due to the labor-intensive language
analyses employed, numbers of participants were
insufficient to conclusively determine the presence of
coherent subgroups. Further study is needed to
address these issues.
Finally, the present study shares a limitation with
much of the research on language in autism: the chal-
lenge of choosing appropriate matching criteria.
Matching on the PPVT-III raises a concern that the
identification of objects in a PPVT-style task may be a
peak of ability in high functioning individuals on the
autism spectrum, that serves to over-estimate IQ
(Mottron, 2004). In this study, non-verbal IQ served as
the primary matching criterion, and groups were also
matched on PPVT-III receptive vocabulary to main-
tain consistency with the large literature also matching
on that basis. In addition, only participants who were
combining words into 2-word phrases were included, in
order to insure that all participants had at least begun
to communicate verbally. To address this limitation in
the matching procedure, analyses were performed both
with and without the PPVT-III results covaried , with
no change in results.
The major finding of the present study is that chil-
dren with autism, compared with chronological age-
(DD group) and non-verbal mental age- (DD and TD
groups) matched peers exhibited clear delays in syn-
tactic knowledge. In contrast to these syntactic
impairments, the autism group’s lexical knowledge was
unimpaired. We have discussed some possible factors
underlying these syntactic impairments, including the
production of jargon, the discussion of less complex
events in the world, and the atypical sequence of
learning of syntactic structures. One important impli-
cation is that receptive vocabulary abilities, such as
those measured by the PPVT-III, are likely an inade-
quate marker of ‘‘language ability,’’ and studies that
match groups based only on lexical measures are likely
to overestimate the comprehension of children in the
autism group.
Acknowledgement This research was supported by the NIMH
(R03 grant # MH61032-01), a Young Investigator award from the
Journal of Language Learning, and a Dissertation Award from
the Society for Science of Clinical Psychology, to IME. We are
especially grateful to the parents and children that participated
in the study.
Appendix Index of Productive Syntax (IPSyn) Items
Table 7
Verb phrases subscale Sentence structures
subscale
Verb Two-word combination
Particle or preposition Subject-verb sequence
Prepositional phrase
(Preposition + NP)
Verb-object sequence
Copula linking two nominals Subject-verb-object
sequence
Catenative (pseudo-auxiliary)
preceding a verb
Conjunction (any)
Auxiliary be, do, have in VP Sentence with two VPs
Progressive Suffix Conjoined phrases
Adverb Infinitive marked with to
Modal preceding verb Let/Make/Help/Watch
introducer
Third person singular present
tense suffix
Adverbial conjunction
Past tense modal Propositional complement
Regular past tense suffix Conjoined sentences
Past tense auxiliary Wh- clause
Medial adverb Bitransitive predicate
Copula, modal, or auxiliary
for emphasis or ellipsis
(non-contractible context)
Sentence with 3 or
more VP’s
Past tense copula Relative clause, marked
or unmarked
Bound morpheme on verb
or on adjective
Infinitive clause with
new subject
Gerund
Fronted, center-embedded
subord. clause
Passive construction
Question-Negation Subscale Noun Phrases Subscale
Intonationally marked question Proper, mass, or
count noun
Routine do/go, existence
question, wh- pronoun
Pronoun or prolocative
(not modifier)
Negation (no(t), can’t, don’t)+
X (NP, VP, PP)
Modifier (adjective,
possessive, quantifier
Initial wh- pronoun followed
by verb
2-word NP: Article/
modifier + nominal
Negative morpheme between
subject and verb
Article, used before a noun
Wh- question- inverted modal,
copula, auxiliary
Two-word NP after verb
or preposition
Negation of copula, modal,
or auxiliary
Plural suffix
Yes/no question- inverted modal,
copula, auxiliary
2-word NP (as N4)
before verb
Why, When, Which, Whose 3-word NP (Determiner +
Modifier + Noun)
Tag question Adverb modifying adjective
or nominal
Questions with negation +
inverted copula, modal
Any bound morpheme on
noun, adjective
Note: NP = noun phrase, VP = verb phrase, PP = prepositional
phrase
J Autism Dev Disord
123
References
Achenbach, T. M. (1991). Manual for the child behavior check-
list/ 4–18 and 1991 profile. Burlington: University of Ver-
mont, Department of Psychiatry.
Adams, R., & Khoo, S. (1993). Quest, the interactive test analysis
system. Melbourne: ACER.
American Psychiatric Association. (1994). Diagnostic and statis-
tical manual of mental disorders (4th ed.). Washington, DC:
American Psychiatric Association.
Baltaxe, C., & D’Angiola, N. (1997). Cohesion in the discourse
interaction of autistic, specifically language-impaired, and
normal children. Journal of Autism and Developmental
Disorders, 22(1), 1–21.
Bartak, L., Rutter, M., & Cox, A. (1975). A comparative study of
infantile autism and specific developmental language disor-
der: I. The children. British Journal of Psychiatry, 126, 127–
145.
Bartolucci, G. (1982). Formal aspects of language in childhood
autism. In Advances in child behavior analysis and therapy
(Vol. 2, pp. 159–18). Lexington, MA: Lexington Books.
Bartolucci, G., & Albers, R. J. (1974). Deictic categories in the
language of autistic children. Journal of Autism and Child-
hood Schizophrenia, 19, 131–141.
Bartolucci, G., Pierce, S. J., & Streiner, D. (1980). Cross-
sectional studies of grammatical morphemes in autistic and
mentally retarded children. Journal of Autism and Devel-
opmental Disorders, 10(1), 39–50.
Bates, E. (1997). Origins of language disorders: A comparative
approach. Developmental Neuropsychology, 13(3), 447–476.
Bennetto, L., Pennington, B. F., & Rogers, S. J. (1996). Intact
and impaired memory functions in autism. Child Develop-
ment, 67, 1816–1835.
Bloom, L., Lahey, J., Hood, L., Lifter, K., & Fiess, K. (1980).
Complex sentences: Acquisition of syntactic connectives
and the semantic relations they encode. Journal of Child
Language, 7, 235–261.
Bloom, L., Rocissano, L., & Hood, L. (1976). Adult–child dis-
course: Developmental interaction between information
processing and linguistic knowledge. Cognitive Psychology,
8, 521–552.
Brown, R. (1973). A first language: The early stages. Cambridge,
MA: Harvard University Press.
Capps, L., Losh, M., & Thurber, C. (2000). The frog ate the bug
and made his mouth sad’’: Narrative competence in children
with autism. Journal of Abnormal Child Psychology, 28(2),
193–204.
Carpentieri, S. C., & Morgan, S. B. (1994). Brief report: A
comparison of patterns of cognitive functioning of autistic
and nonautistic retarded children on the Stanford-Binet –
Fourth Edition. Journal of Autism and Developmental Dis-
orders, 24(2), 215–223.
Chomsky, N. (1957). Syntactic structures. In. The Hague: Mou-
ton.
Cohen, J. (1960). A coefficient of agreement for nominal scales.
Educational and Psychological Measurement, 20, 37–46.
Condouris, K., Meyer, E., & Tager-Flusberg, H. (2003). The
relationship between standardized measures of language
and measures of spontaneous speech in children with aut-
ism. American Journal of Speech–Language Pathology,
12(3), 349–358.
Cox, A., Charman, T., Baron-Cohen, S., Drew, A., Klein, K.,
Baird, G., et al. (1999). Autism spectrum disorders at 20 and
42 months of age: Stability of clinical and ADI-R diagnosis.
Journal of Child Psychology and Psychiatry, 40(5), 719–732.
Curcio, F., & Paccia, J. (1987). Conversations with autistic chil-
dren: contingent relationships between features of adult
input and children’s response adequacy. Journal of Autism
and Developmental Disorders, 17(1), 81–93.
Dahlgren, S. O., & Gillberg, C. (1989). Symptoms in the first two
years of life: A preliminary population study of infantile
autism. European Archives of Psychiatric and Neurological
Science, 283, 169–174.
Dalgleish, B. (1975). Cognitive processing and linguistic refer-
ence in autistic children. Journal of Autism and Develop-
mental Disorders, 5(4), 353–361.
de Villiers, J. G., & de Villiers, P. A. (1973). A cross-sectional
study of the acquisition of grammatical morphemes in child
speech. Journal of Psycholinguistic Research, 2, 267–278.
Demuth, K. (1996). Collecting spontaneous production data. In
D. McDaniel, C. McKee, & H. Smith Cairns (Eds.), Meth-
ods for assessing children’s syntax (pp. 3–22). Cambridge,
MA: MIT Press.
Diehl, J. D., Bennetto, L., & Young, E. C. (2006). Story recall
and narrative coherence of high-functioning children with
autism spectrum disorders. Journal of Abnormal Child
Psychology, 34, 87–102.
Dunn, L. M., & Dunn, L. M. (1997). Peabody picture vocabulary
test-3. Circle Pines, MN: American Guidance Service.
Eigsti, I. M., & Bennetto, L. (2001). Syntactic and memory
functions in young children with autism. San Diego, CA:
Paper presented at the International Society for Develop-
mental Psychobiology.
Fein, D., & Waterhouse, L. (1979). Autism is not a disorder of
language. Boston, MA: Paper presented at the New England
Child Language Association.
Fombonne, E. (1999). The epidemiology of autism: A review.
Psychological Medicine, 29(4), 769–786.
Foster (1986). Learning discourse topic management in the
preschool years. Journal of Child Language, 13, 231–250.
Fowler, A. E. (1980). A comparison of normal and retardate
language equated on MLU. Boston, MA: Paper presented at
the Boston University Conference on Child Language
Development.
Godber, T., Anderson, V., & Bell, R. (2000). The measurement
and diagnostic utility of intrasubtest scatter in pediatric neu-
ropsychology. Journal of Clinical Psychology, 56(1), 101–112.
Grice, H. P. (1975). Logic and conversation. In P. Cole &
J. Morgan (Eds.), Syntax and semantics (Vol. 3, pp. 41–58).
NY: Academic Press.
Hermelin, B., & O’Connor, N. (1975). The recall of digits by
normal deaf and autistic children. British Journal of Psy-
chology, 66, 203–209.
Hoff-Ginsberg, E. (1991). Mother–child conversation in different
social classes and communicative settings. Child Develop-
ment, 62(4), 782–796.
Hollingshead, A. B. (1975). Four factor index of social status.
New Haven, CT: Yale University Department of Sociology.
Howlin, P. (1984a). The acquisition of grammatical morphemes
in autistic children: A critique and replication of the findings
of Bartolucci, Pierce, and Streiner, 1980. Journal of Autism
and Developmental Disorders, 14, 127–136.
Howlin, P. (1984b). The acquisition of grammatical morphemes
in autistic children: a critique and replication of the findings
of Bartolucci, Pierce, and Streiner, 1980. Journal of Autism
and Developmental Disorders, 14(2), 127–136.
Howlin, P. (2003). Outcome in high-functioning adults with
autism with and without early language delays: implications
for the differentiation between autism and Asperger syn-
drome. Autism and Development Disorders, 33(1), 3–13.
J Autism Dev Disord
123
Jarrold, C., Boucher, J., & Russell, J. (1997). Language profiles in
children with autism: Theoretical and methodological
implications. Autism, 1, 57–76.
Kaplan, E., Fein, D., Morris, R., & Delis, D. (1991). WAIS as a
neuropsychological instrument (Manual). New York: Psy-
chological Corporation.
Kelley, E., Paul, J., Fein, D., & Naigles, L. R. (2006). Residual
language deficits in optimal outcome children with a history
of autism. Journal of Autism and Developmental Disorders,
34(1), 87–102.
Kjelgaard, M. M., & Tager-Flusberg, H. (2001). An investigation
of language impairment in autism: Implications for genetic
subgroups. Language and Cognitive Processes, 16(2/3), 287–
308.
Lawson, T. T., & Evans, L. D. (1996). Stanford-Binet: Fourth
edition short forms with underachieving and learning dis-
abled students. Psychological Reports, 79(1), 47–50.
Lord, C., Goode, S., Heemsbergen, J., Jordan, H., Mawhood, L.,
& Rutter, M. (1989). Autism diagnostic observation sche-
dule: A standardized observation of communicative and
social behavior. Journal of Autism and Developmental Dis-
orders, 19, 185–212.
Lord, C., & Paul, R. (1997). Language and communication in
autism. In D. J. Cohen & F. R. Volkmar (Eds.), Handbook
of autism and pervasive developmental disorders (2nd ed.,
pp. 195–225). New York: Wiley.
Lord, C., & Pickles, A. (1996). Language level and nonverbal
social-communicative behaviors in autistic and language-
delayed children. Journal of the Academy of Child and
Adolescent Psychiatry, 35(11), 1542–1550.
Lord, C., Rutter, M., & DiLavore, P. C. (1998). Autism diag-
nostic observation schedule-generic. Chicago, IL: University
of Chicago.
Lord, C., Rutter, M., & LeCouteur, A. (1994). Autism diagnostic
interview-revised: A revised version of a diagnostic inter-
view for caregivers of individuals with possible pervasive
developmental disorders. Journal of Autism and Develop-
mental Disorders, 24, 659–685.
MacWhinney, B. (1991). The CHILDES Database. Dublin, OH:
Discovery Systems.
MacWhinney, B. (2000). The CHILDES project: Tools for ana-
lyzing talk (3rd ed.). Mahwah, NJ: Lawrence Erlbaum
Associates.
Marcus, G., Pinker, S., Ullman, M., Hollander, M., Rosen, T., &
Xu, F. (1992). Over-regularization in language acquisition.
Monographs of the Society for Research in Child Develop-
ment, 57(4), 1–182.
Mottron, L. (2004). Matching strategies in cognitive research
with individuals with high-functioning autism: Current
practices, instrument biases, and recommendations. Journal
of Autism and Developmental Disorders, 34(1), 19–27.
Ozonoff, S., & Miller, J. (1996). An exploration of right-hemi-
sphere contributions to the pragmatic impairments of aut-
ism. Brain and Language, 52(3), 411–434.
Prior, M. R., & Hall, L. C. (1979). Comprehension of transitive
and intransitive phrases by autistic, retarded, and normal
children. Journal of Communication Disorders, 12, 103–111.
Prizant, B., & Duchan, J. (1981). The functions of immediate
echolalia in autistic children. Journal of Speech and Hearing
Disorder, 46, 241–249.
Rapin, I. (1991). Autistic children: Diagnosis and clinical fea-
tures. Pediatrics, 87(5), 751–760.
Rapin, I., & Dunn, M. (2003). Update on the language disorders
of individuals on the autistic spectrum. Brain and Develop-
ment, 25(3), 166–172.
Rubino, R. B., & Pine, J. M. (1998). Subject-verb agreement in
Brazilian Portuguese: What low error rates hide. Journal of
Child Language, 25(1), 35–59.
Rumsey, J. M., Duara, R., Grady, C., Rapoport, J. L., Margolin,
R. A., Rapoport, S. I., et al. (1985). Brain metabolism in
autism: Resting cerebral glucose utilization rates as mea-
sured with positron emission tomography. Archives of
General Psychiatry, 42(5), 448–455.
Rutter, M. (1970). Autistic children: Infancy to adulthood.
Seminars in Psychiatry, 2, 435–450.
Rutter, M., Mawhood, L., & Howlin, P. (1992). Language delay
and social development. In P. Fletcher & D. Hall (Eds.),
Specific speech and language disorders in children: Corre-
lates, characteristics, and outcomes (pp. 63–78). London:
Whurr.
Sattler, J. M. (1992). Assessment of children: Revised and updated
third edition. San Diego, CA: Jerome M. Sattler, Inc.
Scarborough, H. (1990). Index of productive syntax. Applied
Psycholinguistics, 11, 1–22.
Scarborough, H., Rescorla, L., Tager-Flusberg, H., Fowler, A., &
Sudhalter, V. (1991). Relation of utterance length to gram-
matical complexity in normal or language-disordered
groups. Applied Psycholinguistics, 12, 23–45.
Seidenberg, M. S., & MacDonald, M. C. (1999). A probabilistic
constraints approach to language acquisition and processing.
Cognitive Science, 23(4), 569–588.
Shriberg, L. D., Paul, R., McSweeny, J. L., Klin, A. M., Cohen,
D. J., & Volkmar, F. R. (2001). Speech and prosody char-
acteristics of adolescents and adults with high-functioning
autism and Asperger syndrome. Journal of Speech and
Language Hearing Research, 44(5), 1097–1115.
Stone, W., & Yoder, P. (2001). Predicting spoken language level
in children with autism spectrum disorders. Autism, 5(4),
341–361.
Szatmari, P., Bryson, S., Boyle, M., Streiner, D., & Duku, E.
(2003). Predictors of outcome among high functioning chil-
dren with autism and Asperger syndrome. Journal of Child
Psychology and Psychiatry, 44(4), 520–528.
Tager-Flusberg, H., & Anderson, M. (1991). The development of
contingent discourse ability in autistic children. Journal of
Child Psychology and Psychiatry, 32(7), 1123–1134.
Tager-Flusberg, H., & Calkins, S. (1990). Does imitation facili-
tate acquisition of grammar? Evidence from the study of
autistic, Down’s syndrome and normal children. Journal of
Child Language, 17, 591–606.
Tager-Flusberg, H., Calkins, S., Nolin, T., Baumberger, T.,
Anderson, M., & Chadwick-Dias, A. (1990). A longitudinal
study of language acquisition in autistic and Down syn-
drome children. Journal of Autism and Developmental
Disorders, 20(1), 1–21.
Thorndike, R. L., Hagen, E. P., & Sattler, J. M. (1986). The
Stanford-Binet Intelligence Scale: Fourth Edition. Chicago,
IL: Riverside Publishing.
Volden, J., & Lord, C. (1991). Neologisms and idiosyncratic
language in autistic speakers. Journal of Autism and
Developmental Disorders, 21, 109–130.
Wanska, S. K., & Bedrosian, J. L. (1986). Topic and communi-
cative intent in mother–child discourse. Journal of Child
Language, 13(3), 523–535.
J Autism Dev Disord
123