Journal of Autism and Developmental Disorders, Vol. 35, No.

5, October 2005 (
2005)

DOI 10.1007/s10803-005-0001-1

Language, Social, and Executive Functions in High Functioning

Autism: A Continuum of Performance

Rebecca J. Landa1,2 and Melissa C. Goldberg1

This study examined language and executive functions (EF) in high-functioning school-aged
individuals with autism and individually matched controls. Relationships between executive,
language, and social functioning were also examined. Participants with autism exhibited
difficulty on measures of expressive grammar, figurative language, planning, and spatial
working memory. A mixed profile of impaired and enhanced abilities was noted in set-shifting.
While controls showed the typical increase in errors when shifting sets from an intra-
dimensional to an extra-dimensional stimulus, this pattern was not noted in participants with
autism. Relationships between EF, language, and social performance were weak to non-
existent. Implications for theories of core deficit in autism and dissociable nature of the
language and executive impairments in autism are discussed.

KEY WORDS: Executive functions, autism, language, social, working memory, flexibility.

This paper examines language and executive LANGUAGE FUNCTIONS

function (EF) in a sample of high-functioning indi-
viduals with autism (HFA) and age- and IQ-matched All children with HFA have verbal and non-
controls. Possible relationships between language, verbal communication impairments. What is the
social, and executive functions in this sample are also nature of these impairments? The sparse literature
examined. This is the first report to examine these on language functioning in HFA is suggestive of
three domains of functioning within the same group mixed abilities, with preserved and impaired compo-
of school-aged individuals with HFA. Understanding nents. Relatively preserved components involve lin-
language and executive function impairment in HFA guistic form (e.g., phonological and grammatical
is critical for accurate diagnosis, making appropriate domains) (Bartak, Rutter, & Cox, 1975; Pierce &
educational adaptations, and better understanding Bartolucci, 1977; Tager-Flusberg, 1981). Despite
the neuropsychobiological impairment in HFA. reports that grammatical development in autism is
relatively spared (for developmental level), there is an
indication that some aspects of grammatical devel-
opment may be impaired, for example, particular
grammatical markers (Scarborough, Rescorla, Tager-
1
Flusberg, Fowler, & Sudhalter, 1991) and syntax
The Johns Hopkins School of Medicine, Baltimore, MD USA. comprehension in teens and adults with HFA (Min-
2
Correspondence should be addressed to: Center for Autism and
Related Disorders, Kennedy Krieger Institute, 3901 Greenspring
shew, Goldstein, & Siegel, 1995; Muller et al., 1999).
Avenue, Baltimore, MD 21211, USA; e-mail: Landa@ Differences in findings across studies regarding
KennedyKrieger.org impaired or intact grammatical comprehension in
557
0162-3257/05/1000-0557/0
2005 Springer ScienceþBusiness Media, Inc.
558 Landa and Goldberg

autism may be related to the complexity of the in autism, and then discuss possible relationships
grammatical stimuli employed. Some aspects of between EF and social and language deficits in
grammatical development (e.g., syntactic develop- autism.
ment) may progress more slowly than normal in
autism, and may reach a plateau in some individuals Executive Functions
(Paul & Cohen, 1984).
Executive dysfunction in autism has been most
As with the grammatical system, mixed ‘‘spar-
evident in planning and set shifting (flexibility)
ing’’ and impairment are reported within the seman-
(Bennetto, Pennington, & Rogers, 1996; Hughes,
tic domain for individuals with autism. Relative
Russell, & Robbins, 1994; Liss et al., 2001; Muller
strengths for children and adults with HFA are
et al., 1999; Ozonoff, 1995; Ozonoff & McEvoy, 1994;
vocabulary development (lexical semantics; Fein et
Ozonoff, Pennington, & Rogers, 1991; Prior &
al., 1996) and understanding and generating semantic
Hoffman, 1990; Rumsey, 1985; Rumsey & Ham-
associations (Muller et al., 1999). However, impair-
burger, 1988, 1990; Szatmari, Tuff, Finlayson, &
ment is consistently reported in aspects of relational
Bartolucci, 1990). Despite the evidence of impairment
semantics and abstract (non-literal) language pro-
in these two EF domains in autism, there is incon-
cessing in HFA. In particular, deficits in processing
sistency in the literature about the characterization of
linguistic ambiguity have been reported. In Happe’s
executive dysfunction in HFA. For example, three
(1994) study of the comprehension of linguistic
reports have called into question the presence of a
ambiguity by individuals with HFA, participants’
planning impairment in HFA (Ozonoff, South, &
difficulty was characterized by a tendency to interpret
Miller, 2000; Schneider & Asarnow, 1987; Turner,
messages literally. Similarly, Ozonoff and Miller
1997). Likewise, there are conflicting reports indicat-
(1996) found that teens and adults (16–57 years of
ing intact functioning in set-shifting in autism (Gold-
age) with HFA selected inappropriate endings to
berg, Mostofsky, Cutting, Denckla, & Landa, 2005;
jokes, possibly indicating impaired flexibility of
Minshew, Goldstein, Muenz, & Payton, 1992; Reitan
thought (to interpret something in an alternative
& Wolfson, 1985; Schneider & Asarnow, 1987). Some
way). Impairments in abstract language processing
research groups have suggested that computerized
also were noted in a group of teens and adults with
tasks enhance performance on measures of set-
HFA in a study by Minshew et al. (1995) as well as in
shifting in autism (Pascualvaca, Fantie, Papageor-
adults with HFA (Van Bourgondien, & Mesibov,
giou, & Mirsky, 1998; Ozonoff & Strayer, 2001).
1987).
The working memory (WM) component of EF
If language skills are compromised in HFA,
has been reported as impaired in autism by some
research is needed to examine possible links to
researchers, and unimpaired by others. Bennetto et al.
impaired EF. EF is defined as the ability to maintain
(1996) found verbal WM deficits in adolescents and
an appropriate problem solving set to attain a goal
adults with HFA. In contrast, Ozonoff and Strayer
(Welsh & Pennington, 1988). Specific cognitive abil-
(2001) reported intact nonverbal WM in HFA using
ities comprising EF are generally thought to include
‘‘running memory’’ (one-back/two-back task, more
planning, set-shifting (flexibility), working memory
commonly called WM), spatial memory span, and box-
and inhibition, which are dissociable and can be
search tasks. Other negative findings regarding WM in
measured relatively independently (Ozonoff &
autism (compared to controls) have been reported by
Strayer, 1997). Executive dysfunction has been pro-
several groups, particularly when participants were
posed as a core deficit in autism, and consequently
younger or more cognitively impaired (Griffith et al.,
has been proposed to give rise to the communication
1999; Mottron, Peretz, Belleville, & Rouleau, 1999;
and social impairments in this disorder. In a review of
Russell, Jarrold, & Henry, 1996).
EF research in autism, Pennington and Ozonoff
(1996) reported that individuals with autism per-
Relationship between Executive, Linguistic, and Social
formed significantly worse than controls on 25 of 32
Functioning
EF tasks, with an average effect size of .98. Executive
dysfunction has been identified in autism across all Planning, flexibility, and WM functions influ-
ages and levels of intellectual functioning (see Ozo- ence rule and concept formation, shifting when
noff & Strayer, 1997 for a review, but see Griffith, formulating a problem solving strategy, and utilizing
Pennington, Wehner, & Rogers, 1999 for an excep- models to plan and devise alternative ways to achieve
tion to this). First we will review the literature on EF solutions to problems (Ameli, Courchesne, Lincoln,
Language and Executive Functions in High Functioning Autism 559

Kaufman, & Grillon, 1988; Minshew et al., 1992; predict others’ behaviors, mental states and beliefs;
Prior & Hoffman, 1990; Rumsey, 1985; Schneider & Russell et al., 1999). Dawson and colleagues found no
Asarnow, 1987; Szatmari et al., 1990). Impaired EF relationship between social dysfunction and a task
could deleteriously impact developing language and purported to measure dorsolateral prefrontal cortex
social systems. Executive dysfunction could have function (Dawson, Meltzoff, Osterling, & Rinaldi, 1998).
considerable impact on the ability to plan well- Her group found impaired performance in a group of
formed, novel linguistic constructs, shift from one children with autism on a task (Delayed Non-Match to
meaning of a word or phrase to another meaning (in Sample, DNMS) tapping a region of prefrontal cortex
the case of ambiguous or figurative language), or (Dawson et al., 2002). The autism group’s performance
hold in mind critical elements in discourse while on DNMS was correlated with joint attention ability; a
formulating the gist of what is being heard and similar finding has been reported in typically developing
preparing to make inferences based on given infor- toddlers (Nichols, Fox, & Mundy, 2005).
mation. Likewise, executive dysfunction could impact Based on the existing literature, the present study
social functioning, such as the ability to plan on-line focused on hypotheses that participants with HFA
in social contexts, shift social behavior or conversa- would show impaired performance on tasks of (1)
tional topics to meet ever changing contextual complex sentence formulation, (2) assessing abstract
demands, and to hold social information in mind language processing, and (3) planning and set shift-
while processing the dynamic features of the social ing. Additional hypotheses were that (4) the gener-
context and formulating appropriate responses. A ation of well-formed sentences in the sentence
possible relationship between verbal ability and EF formulation task would be intimately dependent on
was noted by Russell and colleagues, who found that planning ability and, hence, would be correlated with
individuals with autism performed worse on EF tasks the EF measure of planning; (5) figurative language
that were likely to require them to encode rules in comprehension would be correlated with cognitive
verbal form (Russell, Saltmarsh, & Hill, 1999). flexibility because of the need to flexibly shift from
A relationship between deficits in WM and one meaning of a word or phrase to a different
aspects of language processing has been indicated in meaning; and (6) social functioning would be corre-
the literature on language-impaired individuals lated with variables reflecting planning and flexibility.
(Baddeley, Thomson, & Buchanan, 1975; Brookshire,
Chapman, Song, & Levine, 2000). In young children
METHOD
with autism, a correlation between impaired perfor-
mance on an EF task (spatial reversal) and an aspect
Participants
of social communication (joint attention) has been
reported (McEvoy, Rogers, & Pennington, 1993). Nineteen children with idiopathic autism, aged
Abstract language processing and discourse behavior 7–17.5 years (M = 11.01), were compared to 19
are often disrupted in individuals with injuries to the typically developing children (mean age = 11.00).
frontal lobes (e.g., Liles, Coelho, Duffy, & Zalagens, Controls were individually matched to the participants
1989; Pearce, McDonald, & Coltheart, 1998), which with HFA based on age, gender, and full scale IQ.
are key to intact executive function (e.g., Mishkin, Children with autism were recruited through
1964; Iversen & Mishkin, 1970; Fuster, 1980; Dia- Baltimore/Washington, DC chapters of the Autism
mond & Goldman-Rakic, 1989; Shallice, 1982; Mil- Society of America and from the Center for Autism
ner, 1963, 1982). Likewise, there is a considerable and Related Disorders at the Kennedy Krieger Insti-
relationship between prefrontal brain injury and tute. All children in the autism group were diagnosed
social dysfunction, including theory of mind and with autism by one member of our research team using
social discourse (Anderson, Damasio, Tranel, & cut-off points from both the Autism Diagnostic
Damasio, 2000; Barkely, Grodzinsky, & DuPaul, Interview-Revised (ADI-R; Lord, Rutter, & LeCou-
1992; Eslinger & Biddle, 2000; Rowe, Bullock, teur, 1994) and the Autism Diagnostic Observation
Polkey, & Morris, 2001; Williams & Mateer, 1992). Schedule (ADOS; Lord et al., 1989) or the Autism
A scant and mixed literature focuses on the rela- Diagnostic Observation Schedule-Generic (ADOS-G;
tionship between executive and social dysfunction in Lord et al., 2000). The ADI-R is an extensive struc-
autism. Evidence exists for a relationship between tured interview administered to caregivers covering
executive dysfunction and theory of mind (a social developmental history and current presentation of
cognitive ability related to the ability to understand and symptoms in communication and social domains, as
560 Landa and Goldberg

Table I. Summary of Grouping Variables (Mean ± 1 SD, Range)

HFA Control Paired t-tests

Age (years) 11.01 (2.89) 11.00 (2.85) t(18) = .06, p = .95
(7.3–17.3) (7.2–17.2)
Full Scale IQ 109.7 (15.80) 113.4 (14.34) t(18) = )1.55, p = .15
(81–139) (90–138)
Verbal IQ 113.5 (17.07) 115.6 (15.79) t(18) = ).69, p = .49
(90–142) (90–143)
Performance IQ 104.6 (13.46) 108.5 (12.14) t(18) = )1.27, p = .22
(74–135) (93–132)
SES 53.1 (9.6) 52.6 (8.87) t(18) = 0.17, p = .86
(27–66) (35–66)

well as the presence of stereotyped, repetitive patterns Depression, Schizophrenia), language disorders
of behaviors and interests. Key items from the three (including treated articulation problems, reading dif-
autism diagnostic categories of functioning (commu- ficulties), or moderate/severe learning disabilities
nication, social, and repetitive/stereotyped patterns of (extensive tutoring, failing a grade in school). Controls
behaviors and interests) contribute to the calculation were matched on an individual basis to participants
of an algorithm score. To qualify for a diagnosis of with HFA based on gender, age (±8 months), and full
autism, one must have scores within all three domains scale IQ (±15 points; except in two pairs that were
that meet the algorithm threshold for autism. The matched within 25 points of one another; HFA case 1
ADOS (earlier version of the ADOS-G) and ADOS-G FSIQ = 110, paired control FSIQ = 135; HFA case
involve administering probes directly to the child to 2 FSIQ = 81, paired control FSIQ = 100). None of
assess social and communication functioning. An the children included in the control sample had any
algorithm provides cut-offs for diagnosing autism. neurological impairment or medical condition thought
Three of the participants included in the HFA group to be related to autism.
failed to meet the ADOS (Lord et al., 1989) algorithm There were no significant differences between the
criteria for autism. However, all three of these partic- groups in chronological age (CA), Full Scale IQ, or
ipants did meet criteria for autism on the ADI-R and Socio-Economic Status (SES) (Hollingshead, 1975).
were judged by an autism expert to have a clinical Table I shows descriptive characteristics of the two
diagnosis of autism. Therefore, these participants were groups.
included in this study. All participants in the autism
group were determined to be high functioning based on
Procedure
having Full Scale IQ scores above 80 on the Wechsler
Intelligence Scale for Children-Revised (Wechsler, All children were seen individually for assess-
1974), WISC-III (Wechsler, 1991) for the two partic- ment. The present report includes a subset of the
ipants with HFA who had been tested with this version tasks performed by these children; other tasks not
of the WISC within the past 12 months, or the presented herein included verbally based EF tasks, a
Wechsler Adult Intelligence Scale-Revised (Wechsler, neurological exam, and discourse tasks. The entire
1981) for participants over 16 years of age. All assessment period lasted approximately 8 hours
participants in the HFA group had idiopathic autism divided into three visits. The tasks relevant to this
(e.g., no history of fragile X, encephalitis, or other report are described below, with language and EF
known medical conditions associated with autism). tasks described separately.
The control group consisted of 19 typically Language assessment included measures of
developing children who were recruited through com- expressive grammar, as well as the ability to explain
munity-wide service groups and volunteer organiza- and understand figures of speech. Expressive gram-
tions via local advertisements. All controls were mar was selected for examination for two main
determined on a telephone screening interview to be reasons: (1) to address a gap in the existing literature,
free of a diagnosis or family history of autism or other which focuses mainly on receptive grammatical
neuro-developmental disorders (e.g., Anxiety abilities in autism; and (2) based on the first author’s
Disorder, ADHD, OCD, Tourette’s Syndrome, observation that many children with HFA repeat and
Language and Executive Functions in High Functioning Autism 561

reformulate their ideas when speaking, we wished to deviation of 3. Level 1 (ages 5 through 10 years):
explore the relative contribution of grammatical and Children are presented verbally with a figure of
planning abilities to expressive language difficulty in speech and are asked to interpret it. They are then
HFA. The task selected for measuring expressive asked to point to one of four pictures that demon-
grammatical performance elicited simple to complex strates the meaning of the figurative expression. Level
sentence structures, permitting a sampling of partic- 2 (ages 10–18 years): Children are presented verbally
ipants’ attempts to generate complex grammatical and in writing on a card in front of them, a situation
forms (as opposed to conducting a grammatical and a figurative expression that was made in the
analysis of spontaneous speech, where the absence context of that situation. They are asked to explain
of complex grammatical forms cannot be presumed the expression. The examiner next reads out loud
to indicate that the subject cannot produce such from the card in front of the child four expressions
forms). Figurative language was selected for study from which the child is to choose the one expression
here because of the need to examine both grammat- that is most close to or most similar in meaning to the
ical and abstract language processing in the same original expression. For example, the participant was
group of individuals with HFA and to explore the asked to describe what was meant when a teacher
relationship between non-literal language processing asked his student to ‘‘Give me a hand.’’
and cognitive flexibility.
Executive Function Measures
Language Measures
EF was examined using selected tasks from the
Below is a description of the measures used to Cambridge Neuropsychological Test Automated Bat-
examine aspects of expressive grammar and figurative tery (CANTAB; Cambridge Cognition, 1996) to
language skills. Responses to all of the language probe spatial working memory (SWM), planning
stimuli were tape recorded, transcribed, and scored (Stockings of Cambridge, hereafter SOC), and flex-
according to the guidelines provided in the test ibility (set-shifting, using the intra-dimensional/extra-
manual. Dependent variables from the language dimensional Shift Task, ID/ED Shift task). These
tasks were the standard scores for each task. functions are thought to be related to frontal lobe
Two subtests from the Clinical Evaluation of function (Robbins, 2000; Owen, Downes, Sahakian,
Language Fundamentals-Revised (CELF-R; Semel, Polkey & Robbins, 1990; Owen, Morris, Sahakian,
Wiig, & Secord, 1987) were administered. The CELF- Polkey, & Robbins, 1996; Owen, Sahakian, Semple,
R measures language skills related to an individual’s Polkey & Robbins, 1995).
use of semantics, morphology and syntax, and The CANTAB tasks were administered via a
memory. This is a standardized test with a mean of portable Pentium MMX processor-based panel PC
10 and standard deviation of 3. with an LCD flat panel display touchscreen built into
The Formulated Sentences subtest of the CELF- the computer (Elo TouchSystems Accur Touch-Serial
R measures the ability to form grammatically and touchscreen) or a desktop Gateway P5-120 with a
semantically intact simple, compound, and complex Mitsubishi Precise point 5800 touchscreen monitor
sentences. The participant is asked to formulate a (model # SD5800C). The screen was placed within
sentence using one or two words provided by the arms’ reach in front of the subject. Subjects
examiner. A picture shown for each word or set of responded to stimuli by touching the screen. Follow-
words provides a visual stimulus to help the subject ing is a description of those tasks.
develop a complete, grammatically correct sentence. The Spatial Working Memory (SWM) task was
For example, the participant is provided orally with used to examine memory for visual-spatial locations.
the word ‘‘car’’ and is simultaneously shown a picture This task has been described in detail by Owen et al.
of a family loading up a car, and is asked to give a (1990): Participants were shown colored squares
sentence using the word ‘‘car’’. (boxes) on the computer screen. The instructions
Figurative language was assessed using the were to ‘‘search through’’ the boxes on the computer
Figurative Language subtest of the Test of Language screen for a blue token, by touching each one to
Competence (TLC; Wiig & Secord, 1989). This reveal what is inside. Upon finding a blue token, the
subtest assesses comprehension and interpretation subject was to place it in an empty column on the
of metaphoric expressions and figures of speech. This right-hand side of the screen. The goal is to collect
is a standardized test with a mean of 10 and standard enough blue tokens to fill up the container on the
562 Landa and Goldberg

right side of the screen. Once a blue token had been (dimensions) of visual stimuli were used in this task:
found within a particular box, then that box would white lines and pink shapes.
never be used again to hide a token. Subjects had to Simple stimuli consisted of one of these dimen-
remember where they had searched for and found a sions and compound stimuli consisted of both, white
token. After four practice trials with three boxes, lines overlying pink shapes. In stages 1–5, the
there were four test trials with each of four, six, and discrimination and learning phases, participants
eight boxes. Returning to an empty box where a learned through trial and error to select one specific
target was previously found is a ‘‘between-search shape while ignoring the other shape and the lines.
error.’’ A ‘‘within-search error’’ refers to responses to The computer gave the subject feedback (‘‘correct’’
a box previously opened and shown to be empty in the color green or ‘‘wrong’’ in the color red). The
earlier in the same search sequence. A ‘‘double-error’’ subject could learn a rule to follow to assure that he
is an error that can be categorized as both a between- continued to make correct choices. After the subject
and within-search error. A ‘‘Strategy Score’’ provided had shown that he/she had learned the rule (6
an index of search strategy. An effective strategy is to consecutive correct responses), the computer chan-
begin each new search sequence with a particular box ged the rule. The type of shapes or lines presented in
and then return to that same box to start each new stages 1–5 remained the same; at stage 6, these
sequence once a token has been found, eliminating changed. Hence, even when the specific features (i.e.,
locations (boxes) as tokens are found. intra-dimensional shift) of the lines (or shapes)
A ‘‘Strategy Score’’ is estimated from the num- changed at stage 6, the participant was supposed
ber of searches that start from the same location. A to be able to generalize the concept that lines (or
low Strategy Score (better performance) is given to shapes) continued to be the salient dimension, and
search sequences that consistently start from the same incur no additional challenge to accurate perfor-
box. A high strategy score is assigned to many mance. In other words, by stage 6, the participants
sequences beginning with a different box. should have learned that lines (or shapes) were the
The Stockings of Cambridge test is a computer- salient dimension, and the stage 6 task should be
ized spatial planning task based on the ‘‘Tower of approached as a simple discrimination task. At stage
Hanoi.’’ The task has been described previously by 8, the salient dimension shifts from the line (or
Owen et al. (1990): Two sets of three colored balls shape) to the shape (or line). This is considered an
were displayed on a computer screen, one at the top extra-dimensional shift, requiring conceptual flexi-
of the screen and one at the bottom of the screen. bility. At this point, the participant was expected to
Each display contained one green, one blue, and one incur an additional challenge because they were
red ball, appearing to be held in stockings or socks required to make a substantial shift away from the
suspended from a beam. The participant was dimension that had been salient for the previous 7
instructed to rearrange the balls in the bottom tasks, which required only perceptual flexibility
display to make the arrangement match the goal (Luciana & Nelson, 1998).
arrangement in the top display. A number was shown The test proceeds through a series of stages, each
on the screen that informed the participant of the with a different contingency, up to a maximum of 9
fewest number of moves needed in order to make the stages:
two arrangements look identical. Problems were
presented with increasing difficulty such that a (1) Simple discrimination between two pink
solution could be reached after a minimum of two, shapes or white lines.
three, four, or five moves. Participants had eight (2) Simple reversal, using the same stimuli but
practice problems with one and two moves. There with the contingencies reversed, that is,
were 12 test problems, two each at the two- and three- choosing the other white line or pink shape.
move problem level, and four each at the four- and (3) Compound discrimination, the contingencies
five-move problem levels. are the same, but now there is the addition of
To assess cognitive flexibility, or set-shifting, the a new pattern of either lines or shapes
Intra-dimensional/Extra-dimensional shift task was (distracters) which are kept separate and the
used. This task, previously described by Downes et two patterns on the screen do not overlap.
al. (1989), examines the ability to attend to specific The pairings of the lines and shapes was
attributes of visual stimuli and to shift attention from pseudo random, with no more than three
one attribute to another when required. Two types consecutive trials with the same pairings.
Language and Executive Functions in High Functioning Autism 563

(4) Compound discrimination with stimuli ANOVA was used, with Group (HFA vs. Control) as
overlapping, with contingencies kept the the between-subject variable and extra moves on 2-,
same, except for the overlap of the two pat- 3-, 4-, and 5-move problems as the within-subject
terns. variable on the planning task; and number of
(5) Compound reversal, the contingencies are between-search errors on 3-, 4-, 6-, and 8-box
now reversed but the correct response is problems as the within-subject variable on the
within the same dimension. SWM task. Repeated measures ANOVA was the
(6) Intra-dimensional shift. This is the intra- statistical procedure selected also to examine within-
dimensional shift stage, wherein a new set of subject (autism and control analyses conducted
exemplars is presented and success depends separately) performance (e.g., Trials to Criterion
upon continuing to sort according to lines or and Errors To Criterion) at the intra-dimensional
shapes. and extra-dimensional shift stages (6 and 8) of the
(7) Intra-dimensional reversal, the contingencies intra- and extra-dimensional shift task.
within the same stimulus dimension are now To examine links between executive function and
reversed. language measures, we computed correlation analy-
(8) Extra-dimensional shift. Here, the previously ses separately for each group (HFA and control)
ignored dimension, the distractor, is now the between the following combinations of variables:
correct dimension by which to sort. That is, CELF Formulated Sentences subtest standard score
if shapes were previously correct, now the and the number of Stockings of Cambridge problems
subject needs to sort according to the lines solved in minimum number of moves; and TLC
that were reinforced randomly. Stage 8 is the Figurative Language subtest standard score and
penultimate stage; it is called the ‘‘extra- Trials to Criterion and Errors to Criterion at intra-
dimensional’’ shift stage. It is analogous to a dimensional and extra-dimensional shift stages 6 and
change in category in the Wisconsin Card 8, respectively.
Sorting Test. Finally, to examine the relationship between EF
(9) Extra-dimensional reversal. Here, the con- and social variables, correlation analyses were con-
tingencies are reversed within the new stim- ducted between the algorithm summary score for
ulus dimension. social functioning on the ADI and the ADOS/
ADOS-G and the EF variables. Only the group with
If at any stage the subject fails to reach criterion after
HFA had data on the ADI and the ADOS/ADOS-G
50 trials, the test ends.
because a parent-screening interview for the controls
did not indicate the presence or history of any autism
Data Analyses
symptoms.
The following variables were downloaded from We adopted a stringent approach to dealing with
the CANTAB: number of problems solved in the issues of significance as follows: (a) due to the fact
minimum number of moves and the number of extra that there were multiple comparisons conducted, we
moves on 2-, 3-, 4-, and 5-move problems on the established a cut-point of p £ .01 for statistical
Stockings of Cambridge; final stage completed, num- significance; and (b) if p was between .02 and .05,
ber of trials taken to reach criterion and errors to we considered this a trend toward significance.
criterion at the critical intra-dimensional and extra-
dimensional shift stages (ID/ED task); and between-
RESULTS
search errors, within-search errors, and double-errors
on 3-, 4-, 6-, and 8-box problems, as well as the Strategy
There was no significant difference between the
Score (SWM task). For all dependent variables gener-
groups in Verbal, Performance, or Full Scale IQ (see
ated from the CANTAB tasks, the scores were
Table I).
automatically calculated by the computer.
Paired t-tests were used to compare individual
Language Findings
measures between the two-matched groups. When the
task involved multiple levels of difficulty to be There was a significant difference between the
completed within subjects (as in the planning and performance of participants with HFA compared to
spatial working memory tasks), repeated measures controls on the CELF-R Formulated Sentences
564 Landa and Goldberg

Table II. Summary of Groups’ Performance on Language and EF Tasks

HFA Control
Measures M (SD) M (SD) Paired t-tests
Language
CELF-R Formulated Sentences 7.21 (2.74) 9.79 (3.41) t(18) = )2.89, p = .010**
TLC Figurative Language 7.84 (3.37) 10.95 (2.55) t(18) = )3.93, p = .001**
Executive Function
Spatial Working Memory
Between-Search errors 52.68 (17.90) 35.84 (15.28) t = 4.62(19), p = .000**
Strategy Score 38.63 (4.31) 35.31 (3.48) t = 3.74(19), p = .001**
Stockings of Cambridge
Perfect Solutions 5.47 (2.09) 7.42 (1.77) t(18) = )2.89, p = .010**
Flexibility
Stage completed 7.6 (2.0) 7.9 (1.1) t(18) = ).72, p = .48
Intra-dimensional Shift-Trials to Criterion 12.25 (8.55) 6.87 (2.2) t(15)= 2.38, p =.031*
Intra-dimensional Shift-Errors to Criterion 5.61 (8.33) .50 (.98) t(17)= 2.54, p = .021*
Extra-dimensional Shift-Trials to Criterion 8.55 (2.7) 17.0 (13.78) t(8) = )1.80, p = .11
Extra-dimensional Shift-Errors to Criterion 3.85 (5.62) 13.57 (12.44) t(13) = )3.39, p = .005**

*p < .05, **p < .01.

subtest. As shown in Table II, participants with HFA

also performed significantly worse than controls on
the TLC Figurative Language subtest.
The results for the Executive Function and
language measures are shown in Table II.

Executive Function Findings

Spatial Working Memory
Table II shows the total number of ‘‘between-
search errors’’ made by participants with HFA and
controls across all search set sizes (3, 4, 6, and 8).
Compared to controls, participants with HFA more
Fig. 1. The mean number of between-search errors on the
frequently made between-search errors by returning Spatial Working Memory Task, as a function of set size (3, 4,
to a box where they had found a blue token on a 6, and 8), for individuals with HFA (black bars) and controls
previous search. (white bars).
Figure 1 illustrates that there was an interaction
between performance and task difficulty. While in the same search sequence; repeated measures
participants with HFA and controls performed sim- ANOVA, Group by Search set size interaction, on
ilarly on searches with three and four items, partic- within-errors F(3, 108) = 6.13, p = .0007; analysis
ipants with HFA returned to a box where they had of simple effects on 8-item trials p = .003; double-
previously found a blue token (between-search errors) search errors F(3, 108) = 3.78, p = .01; analysis of
significantly more often on 6-item trials and showed a simple effects on 8-item trials p = .018).
trend toward making more of such errors on the 8- Information about strategy (referred to here as
item trials compared to controls (repeated measures ‘‘Strategy Score’’) utilized by participants during the
ANOVA, Group by Search set size interaction, F(3, SWM task can be obtained from the CANTAB
108) = 5.89, p =.0009; analysis of simple effects on based on the pattern of their search behavior. As can
6-item trials, p = .001, on 8-item trials, p = .028). be seen in the Strategy Score in Table II, participants
On 8-item trials only, participants with HFA made with HFA used poorer search strategies compared to
more errors by returning to a box previously found controls in the spatial working memory task. Lower
empty (within-search errors) and more double-errors scores for controls represent more efficient use of
(consisting of both within- and between-search errors search strategies. Participants with HFA were less
Language and Executive Functions in High Functioning Autism 565

likely to employ a search strategy based upon a at both the intra-dimensional and extra-dimensional
repetitive search pattern (an efficient strategy where shift stages separately, and results from these two
each new trial is approached by beginning the search stages are presented separately below.
with a particular box). It was found that 84.2% of the participants with
An additional analysis was conducted to exam- HFA and 100% of controls reached stage 6, the intra-
ine the relationship between WM and planning dimensional shift stage (see Fig. 3). A comparison of
because hypotheses have been put forth in the graphs A and B in Fig. 4 illustrates the trend for
literature that planning ability influences perfor- children with HFA to require more trials to reach
mance on WM tasks. However, no correlation was criterion for passing to the next level, and for them to
found between returning to a box containing a produce more errors compared to controls in the
previously found blue token (between-search errors, intra-dimensional shift stage.
a measure of WM) and number of items performed in To examine performance of the groups at the
the minimum number of moves (Stockings of Cam- extra-dimensional stage, only participants who
bridge, measure of planning ability; r = )0.33; reached stage 8 (nine pairs) were included in analyses
p = .894). of Trials to Criterion (number of trials to meet
criteria for passing to the next level) and Errors to
Planning Criterion (number of errors made before meeting the
criterion for passing to the next level). Although there
The autism group displayed more difficulty on
were no significant group differences in the number of
the planning task than did the controls. As shown in
Trials to Criterion, an examination of the group
Table II, participants with HFA solved fewer of the
means and standard deviations, depicted in Graph A
12 test problems in the minimum number of moves
of Fig. 4, suggests that participants with HFA took
compared to controls. However, no specific level (2-,
substantially fewer trials to reach criterion compared
3-, 4-, and 5-move problems) could be identified as
to controls. Of interest, Table II shows that partic-
causing this difference when a repeated measures
ipants with HFA exhibited significantly fewer errors
ANOVA was used to examine the mean number of
(Errors to Criterion) at the extra-dimensional shift
extra moves beyond the minimum possible for each
stage compared to controls.
level (see Fig. 2, F = .37(3, 114), p = .77).
We conducted a repeated measures ANOVA to
examine whether the participants in our study dem-
Cognitive Flexibility (Intra-dimensional/
onstrated the pattern seen in typically developing
Extra-dimensional Shift Task)
children (Luciana & Nelson, 1998) of more errors to
There was no significant difference between criterion at the extra-dimensional shift stage than at
participants with HFA and controls in the overall the intra-dimensional shift stage. For this analysis,
stage completed (see Table II). Shifting was analyzed Trials to Criterion and Errors to Criterion at the
intra-dimensional and extra-dimensional shift stages
served as the ‘‘within’’ factors. This was done
separately for each group. For the autism group,
there were no within-group differences for number of
trials (Trials to Criterion; F = .32 (1, 11), p = .58)
nor number of errors (Errors to Criterion; F = .721
(1, 13), p = .41) between intra-dimensional and
extra-dimensional stages (compare heights of black
bars in Fig. 4a and in Fig. 4b). In contrast, the
controls showed a trend for taking more trials at the
extra-dimensional than intra-dimensional stage (Tri-
als to Criterion, F = 5.54 (1, 9), p = .043) and made
significantly more errors at the extra-dimensional
stage than at the intra-dimensional stage (Errors to
Criterion, F = 25.65 (1, 16), p = .00), paralleling the
Fig. 2. The mean number of extra moves taken on 2-, 3-, 4-,
pattern reported by Luciana and Nelson (1998) in
and 5-move problems on the Stockings of Cambridge Task by typically developing 8-year-olds (this can be seen in
individuals with HFA (black bars) and controls (white bars). Fig. 4).
566 Landa and Goldberg

Fig. 3. The proportion of participants with HFA (black squares) and controls (white circles)
to complete each stage of the ID/ED task.

Fig. 4. The mean number of TTC and ETC at the ID and ED shift stages for individuals
with HFA (black bars) and controls (white bars).

Examination of the Relationship between Age, IQ, Participants with HFA

Language and EF Performance
For the HFA group, Verbal IQ correlated signif-
The relationship between age and language icantly (ps < .01) with CELF-R Formulated Sentences
variables was not examined for either group because and TLC-Figurative Language. Performance IQ did
the language dependent variables were standard not correlate with any of the language measures (all
scores based on test norms, which take age into ps > .05). There was a trend (p < .05) for significant
account. correlations between Full Scale IQ and CELF-R
Language and Executive Functions in High Functioning Autism 567

Formulated Sentences, and TLC-Figurative Lan- relationship between EF and social functioning was
guage. All of the IQ measures correlated at the p < .01 examined. No relationship was identified between the
level with between-errors on the SWM task, however Social Domain summary score of the ADI or ADOS-G
no aspect of IQ correlated with any of the other EF and any of the EF variables (see Table III). For the
variables. There was a significant negative correlation older version of the ADOS, however, a significant
between age and the total number of errors on the relationship was found between the number of between-
flexibility task, as well as a trend for a negative search errors on the Spatial Working Memory task and
correlation between age and Trials to Criterion at the the Social Domain algorithm score (more between-
intra-dimensional stage (stage 6; p = .022). search errors related to greater social dysfunction).
There were no significant correlations between
any of the EF variables and the language measures.
DISCUSSION
Comparison Group
This study examined aspects of language and
For the control group, Verbal IQ and Full Scale executive functioning, and possible relationships
IQ correlated significantly (ps < .01) with the between executive, language, and social functions in
language measures. Performance IQ did not correlate individuals with HFA as compared to their individ-
with the language measures. Neither Performance, ually matched controls. Our hypotheses were that (a)
Verbal, nor Full Scale IQ correlated with any of the despite being matched for age, gender, IQ and SES,
EF variables. Age did not interact significantly with the children with HFA would perform less well than
any of the EF dependent variables for controls. controls on measures of language and EF; (b)
The relationship between EF and language per- planning would be correlated with performance on
formance for the controls only partially matched our a measure of sentence formulation; (c) flexibility
hypotheses. As hypothesized, significant negative cor- would be correlated with performance on a measure
relations were found between performance of controls of figurative language comprehension; and (d) exec-
on the figurative language and two variables from the utive dysfunction would be correlated with social
EF test of flexibility (Trials to Criterion and Errors to dysfunction. To our knowledge, this is the first study
Criterion for the intra-dimensional shift problems examining these aspects of executive, language, and
only, r = ).69 with p < .01 for both). Hence, better social functioning in the same group of high func-
performance on the measure of figurative language tioning school-aged individuals with autism.
correlated with fewer trials and errors to reach The findings of this study partially, but not
criterion for success on the intra-dimensional shift entirely, supported our hypotheses. The participants
component of the flexibility task. There was no with HFA in this study (ages 7–17) showed mixed
significant correlation between planning and expres- performance ranging from unimpaired to impaired
sive grammatical performance. language and executive functions. The group with
HFA performed significantly worse than controls on
sentence formulation, comprehension of figures of
EF and Social Functioning in HFA
speech, aspects of spatial working memory, and
For the HFA group only (because only this group planning. Interestingly, individuals with HFA per-
had Autism Diagnostic Interview measures), the formed better than controls on some aspects of

Table III. Correlations of Measures of Executive Function and ADI, ADOS and ADOS-G Social Domain Algorithm Scores

ADI Social ADOS Social ADOS-G Social

Domain Domain Domain

Measures of Executive Function r p r p r p

Trials to Criterion at Extra-dimensional stage ).260 .414 ).235 .704 .115 .806
Errors at Extra-dimensional Stage ).254 .403 ).243 .694 .196 .647
Problems solved in minimum moves on Stockings of Cambridge .365 .137 ).127 .745 ).422 .297
Between errors on Spatial Working Memory task .423 .080 ).763 .017* ).260 .533
Strategy score on Spatial Working Memory task .041 .871 ).005 .990 ).249 .552

*p < .05.
568 Landa and Goldberg

cognitive flexibility (set-shifting). Below, these when asked to explain the figure of speech stimulus).
findings will be interpreted. In other cases, a literal interpretation of the stimulus
was given, with no evidence of verbal formulation
Language difficulty. Regardless of whether individuals with
autism have difficulty expressing or understanding
Expressive grammar and comprehension of
figures of speech (and perhaps non-literal language in
abstract language (understanding figures of speech)
general), this aspect of impairment is likely to
were impaired despite intact IQ (including verbal IQ).
contribute in a substantial way to social dysfunction
While language and verbal IQ were correlated, they
in this group. It also could impact academic perfor-
were somewhat dissociable. Hence, a child may have
mance, as non-literal interpretation is required for
impaired sentence formulation ability (generating
understanding written text even in very basic chil-
complex grammatical forms), impaired knowledge
dren’s books, and, hence, for gist comprehension.
of how figures of speech may be used to connote
other meanings, or impaired understanding of lin-
Executive Function
guistic nuances, and yet have normal-range verbal IQ
(e.g., ability to define words, demonstrate a general Participants with HFA showed a scattered pro-
fund of information, demonstrate verbal reasoning, file in their executive functioning as assessed by tasks
digit span, etc.). from the CANTAB (Cambridge Cognition, 1996).
The finding of sentence formulation difficulties To our knowledge, there are only five other published
in these participants with HFA and solidly normal studies using this measure with individuals with
verbal IQ raises the question about the integrity of autism in a similar age range (Goldberg et al., 2005;
the grammatical system in autism. Convention has Hughes et al., 1994; Ozonoff et al., 2000, 2004;
held that grammatical and phonological aspects of Turner, 1997). We will relate our findings to theirs as
functioning in autism were spared (Bartak et al., overlapping measures are discussed below (planning
1975; Pierce & Bartolucci, 1977; Tager-Flusberg, and set-shifting). This is the first study known to us
1981), but more recent studies have raised questions that has employed the CANTAB to assess spatial
about grammatical functioning in autism (Minshew working memory in HFA.
et al., 1995; Muller et al., 1999). While individuals
with HFA may speak in full sentences, one cannot
Working Memory
assume that the grammatical system is intact. For
example, Gillon and Dodd (1993) reported poor Working memory (WM) difficulties in our HFA
readers to have significantly lower scores than good sample were identified, but were found mainly on the
readers on the Formulated Sentences subtest of the tasks imposing the greatest working memory load.
CELF-R in the presence of adequate spoken lan- The spatial WM errors in children with HFA were
guage skills. Language assessment should involve characterized by more frequent returns to spatial
direct probing of the ability to create complex locations where a target had already been found
grammatical constructions (using an elicitation pro- (possibly indicating perseveration) on the more dif-
cedure such as that employed here). This said, one ficult 6- and 8-item problems and poorer search
cannot differentially diagnose autism based on gram- strategies overall compared to controls. These find-
matical ability. ings are consistent with some, but not other, reports
Our finding of impaired abstract language pro- of working memory in autism. No previous study of
cessing in the participants with HFA is consistent WM in autism has employed the CANTAB (How-
with other studies of language in autism (e.g., ever, another study from our laboratory by Goldberg
Ozonoff & Miller, 1996). It should be noted that et al., 2005 also found increased errors on 6-item and
the abstract language-processing task included both 8-item problems in a sample of 8–12 year-olds with
an expressive and receptive (multiple choice) format. HFA). The reports of impaired WM in autism
Hence, one interpretation of our data may be that primarily come from studies involving adolescents
abstract language processing could be compromised or adults who are high functioning, while those with
by impaired verbal formulation rather than impaired negative findings mostly involved younger or cogni-
comprehension of the target figure of speech. This tively impaired individuals with autism. Three of four
interpretation of the data could be valid in some cases studies focusing on older individuals (adolescents
(e.g., when participants answered ‘‘I don’t know’’ through young adults) with HFA have reported WM
Language and Executive Functions in High Functioning Autism 569

deficits (Bennetto et al., 1996; Minshew et al., 1992; frontally mediated task performance as the frontal
Minshew, Luna, & Sweeney, 1999; negative finding lobes mature. Additional research is needed to gain
by Ozonoff & Strayer, 2001). Studies focusing on an understanding of planning in autism, and whether
younger or cognitively impaired children with autism and how it relates to the symptoms of autism. Large
have reported no WM impairment relative to con- numbers of well-characterized children with autism
trols (Griffith et al., 1999; Mottron et al., 1999; and controls are needed within specific age ranges so
Russell et al., 1996). The differences across studies that change with age may be examined.
cannot be resolved based on whether the WM task
involved the verbal or spatial systems (Bennetto et al., Cognitive Flexibility/Set-shifting
1996; Minshew et al., 1999). Although our findings
The ID/ED task from the CANTAB was used
could not be attributed to planning, perseveration is a
to examine cognitive flexibility, revealing only a trend
possible factor in the performance of our participants
for the participants with HFA to have greater
on the SWM task. Further research on WM in autism
difficulty on the intra-dimensional shift stage (requir-
is needed.
ing perceptual shifting), yet enhanced performance
compared to controls at the extra-dimensional shift
Planning
stage (requiring conceptual shifting, the shift of
Planning difficulties in the participants with interest here). The participants with HFA failed to
HFA were characterized by solving fewer problems show the normal pattern of more errors at the extra-
in the minimum number of moves, regardless of task dimensional vs. intra-dimensional shift stage. Our
difficulty. This finding partly contrasts with that of findings parallel those of two other studies that have
Hughes et al. (1994), who employed the same test employed this same task to measure cognitive flexi-
used in the present study. Hughes and colleagues’ bility in autism (Ozonoff et al., 2000; Turner, 1997;
(1994) participants with autism, including some with also see Goldberg et al., 2005 for another finding
mental retardation, performed no differently from from our laboratory; but see Ozonoff et al., 2004, for
controls on the ‘‘easy’’ set (2- and 3- move problems), deficits at stage 8 on the ID/ED task in autism), and
but differed from controls on the ‘‘difficult’’ set (4 and other studies where different methods were employed
5 move problems). This discrepancy in findings could to assess this aspect of EF (Minshew et al., 1992;
possibly be explained by the difference in functioning Schneider & Asarnow, 1987). Hence, we are unable to
level of the participants in the two studies. Our draw the same conclusion drawn by Hughes et al.
controls performed as expected given norms reported (1994), that the defining characteristic of set-shifting
by Luciana and Nelson (2001). in autism involves making an extra-dimensional shift
While the bulk of the studies examining planning of attention from one stimulus dimension to another.
in HFA have reported deficits in this aspect of EF Our finding also indicates that both groups were
(primarily based on non-computerized versions of the equally attentive and motivated on this task.
Tower task), there has been one study that failed to Our results may indicate that the participants
identify planning deficits (based on number of prob- with HFA exhibited enhanced performance at the
lems solved in the minimum number of moves) in 6 to extra-dimensional shift stage. Participants with HFA
20-year-old individuals with HFA (Ozonoff et al., exhibited fewer errors to criterion at the extra-
2000) using the same CANTAB Stockings of dimensional shift stage compared to individually
Cambridge task employed in our study. However, a matched controls in our study, and compared to the
recent study by Ozonoff and colleagues (2004) typically developing controls in the Hughes et al.
reported that planning on the SOC of the CANTAB (1994) study (our HFA mean errors to crite-
seemed intact in the youngest age group (under rion = 3.86; Hughes et al. (1994) normal control
12 years) of lower- and higher-IQ individuals with mean errors to criterion =
9). The pattern of rather
autism but that planning deficits were significant consistent performance across the intra- and extra-
during the teen years (ages 12–19). Similarly, another dimensional stages seen in our participants with HFA
finding from our laboratory (Goldberg et al., 2005) is similar to that reported in three- and four-year-old
showed that eight to 12-year-olds with HFA per- typically developing children (Luciana & Nelson,
formed normally on the SOC from the CANTAB. 1998). Yet it seems inappropriate to characterize our
This could indicate that children with autism, even participants with HFA as ‘‘delayed’’ in their perfor-
with high IQ, cannot keep up with their age peers in mance because of the small number of errors to
570 Landa and Goldberg

criterion exhibited at the intra- or extra-dimensional difficult to study these abilities in isolation of each
shift stage compared to typically developing controls. other as independent functions. The lack of relation-
Enhanced performance in autism on EF tasks ship between general intelligence and EF in the group
has been reported before. For example, young with typical development is intriguing and remains
children with autism (40–61 months of age) per- ripe for further investigation.
formed better than expected on a spatial reversal
task, 3- and 6-boxes scrambled task (Griffith et al., Relationship between Language and Executive
1999). Furthermore, those researchers found that the Functioning
controls made more perseverative errors than did the
Differences were noted in the pattern of rela-
children with autism. Additional reports of perfor-
tionship between EF and language for the HFA and
mance enhancement in autism involve the ability to
control groups. For controls, a significant negative
detect hidden figures (Griffith et al., 1999; Jolliffe &
correlation was found between processing of figura-
Baron-Cohen, 1997). Could this ‘‘enhanced’’ perfor-
tive language and trials and errors to criterion on the
mance be related to the use of strategies (e.g., covert
intra-dimensional (stage 6) component of the flexi-
semantic labeling) to succeed on certain tasks (in our
bility task. This was not the stage that is purported to
case, the flexibility task), perhaps masking an under-
involve cognitive flexibility, so the degree to which
lying deficit in flexibility? This is unlikely given that
this indicates a relationship between figurative lan-
the participants with HFA performed poorly on two
guage and flexibility is unclear. For the group with
different measures of strategy use (Stockings of
HFA, perhaps a relationship between figurative
Cambridge and Spatial Working Memory), which
language and flexibility was not found because some
could also have benefited from an effective strategy
showed enhanced performance on the flexibility task.
(e.g., semantic labeling) if one had existed.
When examining the relationship between exec-
Another interpretation of our flexibility data is
utive and social dysfunction, we found no relation-
possible. One could argue that impairment in cogni-
ship between social and executive functioning based
tive flexibility (set-shifting) exists in autism, but the
on social domain algorithm scores from the ADI and
use of a computer to administer the task enhances the
ADOS-G. However, the older version of the ADOS
ability of HFA children to succeed on the task.
social domain and number of between-errors on the
Although there is some support for this notion in the
WM task were significantly correlated. This finding
literature (Ozonoff, 1995; Pascualvaca et al., 1998),
indicates that at least one specific aspect of EF and
this explanation does not seem to be supported by
social dysfunction may be related. This single finding
our findings. That is, our HFA group showed a trend
creates a challenge to the argument that executive
for more errors and trials to criterion at the intra-
dysfunction in autism gives rise to the social impair-
dimensional stage compared to controls, but they
ment. Although this finding is rather consistent with a
exceeded the performance of controls at the extra-
report by Dawson et al. (1998) that executive
dimensional stage. This phenomenon is further dis-
dysfunction was not related to social performance,
cussed below, but if the computer is believed to
more research on the relationship between executive
enhance performance, then the children with autism
and social dysfunction is needed.
should have benefited across tasks, which was not the
The results of our study do not support theories
case here.
of EF as a core deficit in autism that could give rise to
the social and language impairments seen in this
Relationship between Intellectual and Executive
disorder. The lack of IQ, EF, and language correla-
Functioning
tion in autism is of high theoretical significance. With
For both individuals with HFA and controls, large samples, it may be possible to do sophisticated
verbal IQ and full scale IQ correlated significantly confirmatory factor analytic studies rather than
with the language measures. There was a correlation individual correlations.
between IQ and EF for individuals with HFA only
and this was on the spatial working memory task,
possibly indicating that intelligence measures tax the ACKNOWLEDGEMENTS
working memory system of individuals with HFA.
Intellectual functioning, particularly verbal IQ, is The authors wish to thank all participants in this
intimately tied to language function, making it research endeavor, who gave so generously of their
Language and Executive Functions in High Functioning Autism 571

time and energy. We also wish to thank Tracey Fuster, J. M. (1980). The prefrontal cortex. New York: Raven
Press.
Brown, Erin Garth, and Marcella Radano for their Gillon, G., & Dodd, B. (1993). The communication skills of
assistance in data collection. We thank Andrea children with specific reading disability. Australian Journal of
Schanbacher for her help with administrative support Human Communication Disorders, 21, 86–102.
Goldberg, M. C., Mostofsky, S. H., Cutting, L. E., Mahone, E.
during the final preparation of the manuscript. We M., Astor, B. C., Denckla, M. B., & Landa, R. J. (2005).
are grateful to the Rita Rudel Foundation and Subtle executive impairment in children with autism and
National Institutes of Health (PO1 HD 35468 to children with ADHD. Journal of Autism and Developmental
Disorders, 35, 279–293.
RJL & K01 MH 01824 to MCG; and MH 52432R29) Griffith, E. M., Pennington, B. F., Wehner, E. A., & Rogers, S. J.
for the funding to conduct this research. (1999). Executive functions in young children with autism.
Child Development, 70, 817–832.
Happe, F. G. (1994). An advanced test of theory of mind:
REFERENCES Understanding of story characters’ thought and feelings by
able autistic, mentally handicapped, and normal children and
Ameli, R., Courchesne, E., Lincoln, A., Kaufman, A. S., & adults. Journal of Autism and Developmental Disorders, 24,
Grillon, C. (1988). Visual memory processes in high-func- 129–154.
tioning individuals with autism. Journal of Autism and Devel- Hollingshead, A. B. (1975). Four factor index of social status. New
opmental Disorders, 18, 601–615. Haven, CT: Yale University, Department of Sociology.
Anderson, S. W., Damasio, H., Tranel, D., & Damasio, A. R. Hughes, C., Russell, J., & Robbins, T. W. (1994). Evidence for
(2000). Long term sequelae of prefrontal cortex damage executive function in autism. Neuropsychologia, 32, 477–492.
acquired in early childhood. Developmental Neuropsychology, Iversen, S. D., & Mishkin, M. (1970). Perseverative interference in
18, 281–296. monkeys following selective lesions of the inferior prefrontal
Baddeley, A. D., Thomson, N., & Buchanan, M. (1975). Word convexity. Experimental Brain Research, 11, 376–386.
length and the structure of short-term memory. Journal of Jolliffe, T., & Baron-Cohen, S. (1997). Are people with autism and
Verbal Learning and Verbal Behavior, 14, 575–589. Asperger syndrome faster than normal on the Embedded
Barkley, R. A., Grodzinsky, G., & DuPaul, G. J. (1992). Frontal Figures Test? Journal of Child Psychology and Psychiatry, 38,
lobe functions in attention deficit disorder with and without 527–534.
hyperactivity: A review and research report. Journal of Liles, B. Z., Coelho, C. A., Duffy, R. J., & Zalagens, M. R.
Abnormal Child Psychology, 20, 163–188. (1989). Effects of elicitation procedures on the narratives of
Bartak, L., Rutter, M., & Cox, A. (1975). A comparative study of normal and closed head-injured adults. Journal of Speech and
infantile autism and specific developmental receptive language Hearing Disorders, 54, 356–366.
disorder. I. The children. British Journal of Psychiatry, 126, Liss, M., Fein, D., Allen, D., Dunn, M., Feinstein, C., Morris,
127–145. R. et al. (2001). Executive functioning in high-functioning
Bennetto, L., Pennington, B. F., & Rogers, S. J. (1996). Intact children with autism. Journal of Child Psychology and Psy-
and impaired memory functions in autism. Child Development, chiatry, 42, 261–270.
67, 1816–1835. Lord, C., Risi, S., Lambrecht, L., Cook, E. H. Jr., Levanthal, B.
Brookshire, B. L., Chapman, S. B., Song, J., & Levin, H. S. L., DiLavore, P. C. et al. (2000). The autism diagnostic
(2000). Cognitive and linguistic correlates of children’s dis- observation schedule-generic: A standard measure of social
course after closed head injury: A three-year follow-up. Journal and communication deficits associated with the spectrum of
of the International Neuropsychological Society, 6, 741–751. autism. Journal of Autism and Developmental Disorders, 30,
Cambridge Cognition. (1996). CANTAB. Cambridge, England: 205–223.
Cambridge Cognition Limited. Lord, C., Rutter, M., Goode, S., Heemsbergen, J., Jordan, H.,
Dawson, G., Meltzoff, A. N., Osterling, J., & Rinaldi, J. (1998). Mawhood, L. et al. (1989). Autism diagnostic observation
Neuropsychological correlates of early symptoms of autism. schedule: A standardized observation of communicative and
Child Development, 69, 1276–1285. social behavior. Journal of Autism and Developmental Dis-
Dawson, G., Munson, J., Estes, A., Osterling, J., McPartland, orders, 19, 185–212.
J., Toth, K., Carver, L., & Abbott, R. (2002). Neurocog- Lord, C., Rutter, M., & LeCouteur, A. (1994). Autism diagnostic
nitive function and joint attention ability in young children interview-revised: A revised version of a diagnostic interview
with autism spectrum disorder. Child Development, 73, for caregivers of individuals with possible pervasive develop-
345–358. mental disorders. Journal of Autism and Developmental Dis-
Diamond, A., & Goldman-Rakic, P. S. (1989). Comparision of orders, 24, 659–685.
human infants and rhesus monkeys on Piaget’s AB task: Luciana, M., & Nelson, C. A. (1998). The functional emergence of
Evidence for dependence on dorsolateral prefrontal cortex. prefrontally-guided working memory systems in four to eight-
Experimental Brain Research, 74, 24–40. year-old children. Neuropsychologia, 36, 273–293.
Downes, J. J., Roberts, A. C., Sahakian, B. J., Evenden, J. L., Luciana, M., Sullivan, J., & Nelson, C. A. (2001). Individual
Morris, R. G., & Robbins, T. W. (1989). Impaired extra- differences in phenylalanine levels moderate performance on
dimensional shift performance in medicated and unmedicated tests of executive function in adolescents treated early and
Parkinson’s disease: Evidence for a specific attention dys- continuously for PKU. Child Development, 72, 1637–1652.
function. Neuropsychologia, 27, 1329–1343. McEvoy, R. E., Rogers, S. J., & Pennington, B. F. (1993).
Eslinger, P. J., & Biddle, K. R. (2000). Adolescent neuropsy- Executive function and social communication deficits in young
chological development after early right frontal cortex dam- autistic children. Journal of Child Psychology and Psychiatry,
age. Developmental Neuropsychology, 18, 297–329. 34, 563–578.
Fein, D., Dunn, M., Allen, D. A., Aram, D. M., Hall Morris, N. Milner, B. (1963). Effects of different brain lesions on card sorting.
R. et al. (1996). Neuropsychological and language data. In: Archives of Neurology, 9, 90–100.
I. Rapin (Ed.), Preschool children with inadequate communi- Milner, B. (1982). Some cognitive effects of frontal-lobe lesions in
cation: Developmental language disorder, autism, low IQ. man. Philosophical transactions of the Royal Society of London,
London: Mac Keith Press. Series B, Biological sciences, 298, 211–226.
572 Landa and Goldberg

Minshew, N. J., Goldstein, G., Muenz, L. R., & Payton, J. B. Paul, R., & Cohen, D. (1984). Responses to contingent queries in
(1992). Neuropsychological functioning in nonmentally adults with mental retardation and pervasive developmental
retarded autistic individuals. Journal of Clinical and Experi- disorders. Applied Psycholinguistics, 5, 349–357.
mental Neuropsychology, 14, 749–761. Pearce, S., McDonald, S., & Coltheart, M. (1998). Interpreting
Minshew, N. J., Goldstein, G., & Siegel, D. J. (1995). Speech and ambiguous advertisements: The effects of frontal lobe damage.
language in high-functioning autistic individuals. Neuropsy- Brain and Cognition, 38, 150–164.
chology, 9, 255–261. Pennington, B. F., & Ozonoff, S. (1996). Executive functions and
Minshew, N. J., Luna, B., & Sweeney, J. A. (1999). Oculomotor developmental psychopathology. Journal of Child Psychology
evidence for neocortical systems but not cerebellar dysfunction and Psychiatry, 37, 51–87.
in autism. Neurology, 52, 917–922. Pierce, S., & Bartolucci, G. (1977). A syntactic investigation of
Mishkin, M. (1964). Perseveration of central sets after frontal verbal autistic, mentally retarded, and normal children. Jour-
lesions in monkeys. In: J. M. Warren, & K. Akert (Eds.), The nal of Autism and Childhood Schizophrenia, 7, 121–134.
Frontal Granular Cortex. (pp. 219–241). New York: McGraw Prior, M., & Hoffman, W. (1990). Brief report: Neuropsycho-
Hill. logical testing of autistic children through an exploration with
Mottron, L., Peretz, I., Belleville, S., & Rouleau, N. (1999). frontal lobe tests. Journal of Autism and Developmental Dis-
Absolute pitch in autism: A case study. Neurocase, 5, orders, 20, 581–590.
485–501. Reitan, R., & Wolfson, D. (1985). The Halstead–Reitan neuro-
Muller, R. A., Behen, M. E., Rothermel, R. D., Chugani, D. C., psychological test battery: Theory and clinical interpretation.
Muzik, O., & Mang-Chugani, H. T. (1999). Brain mapping of Tucson, AZ: Neuropsychology Press.
language and auditory perception in high-functioning autistic Robbins, T. W. (2000). Dissociating executive functions of the
adults: A PET study. Journal of Autism and Developmental prefrontal cortex. In: A. C. Roberts, T.W. Robbins, & L.
Disorders, 29, 19–31. Weiskrantz (Eds.), The prefrontal cortex: Executive and cogni-
Nichols, K. E., Fox, N., & Mundy, P. (2005). Joint attention, self- tive functions. (pp. 117–130). Oxford: Oxford University Press.
recognition, and neurocognitive function in toddlers. Infancy, Rowe, A. D., Bullock, P. R., Polkey, C. E., & Morris, R. G.
7, 35–51. (2001). Theory of mind’ impairments and their relationship to
Owen, A. M., Downes, J. D., Sahakian, B. J., Polkey, C. E., & executive functioning following frontal lobe excisions. Brain,
Robbins, T. W. (1990). Planning and spatial working memory 124, 600–616.
following frontal lobe lesions in man. Neuropsychologia, 28, Rumsey, J. M. (1985). Conceptual problem-solving in highly
1021–1034. verbal, nonretarded autistic men. Journal of Autism and
Owen, A. M., Morris, R. G., Sahakian, B. J., Polkey, C. E., & Developmental Disorders, 15, 23–36.
Robbins, T. W. (1996). Double dissociations of memory and Rumsey, J. M., & Hamburger, S. D. (1988). Neuropsychological
executive functions in working memory tasks following frontal findings in high-functioning men with infantile autism, residual
lobe excisions, temporal lobe excisions or amygdalo-hippo- state. Journal of Clinical and Experimental Neuropsychology,
campectomy in man. Brain, 119, 1597–1615. 10, 201–221.
Owen, A. M., Sahakian, B. J., Semple, J., Polkey, C. E., & Rumsey, J. M., & Hamburger, S. D. (1990). Neuropsychological
Robbins, T. W. (1995). Visuo-spatial short-term recognition divergence of high-level autism and severe dyslexia. Journal of
after temporal lobe excisions, frontal lobe excisions or amy- Autism and Developmental Disorders, 20, 155–168.
gdalo-hippocampectomy in man. Neuropsychologia, 33, 1–24. Russell, J., Jarrold, C., & Henry, L. (1996). Working memory in
Ozonoff, S. (1995). Reliability and validity of the Wisconsin card children with autism and with moderate learning difficulties.
sorting test in studies of autism. Neuropsychology, 9, 491–500. Journal of Child Psychology and Psychiatry, 37, 673–686.
Ozonoff, S., Cook, I., Coon, H., Dawson, G., Joseph, R., Klin, Russell, J., Saltmarsh, R., & Hill, E. (1999). What do executive
A. et al. (2004). Performance on CANTAB subtests sensitive factors contribute to the failure on false belief tasks by chil-
to frontal lobe function in people with autistic disorder: Evi- dren with autism? Journal of Child Psychology and Psychiatry,
dence from the CPEA network. Journal of Autism and Devel- 40, 859–868.
opmental Disorders, 34, 139–150. Scarborough, H., Rescorla, L., Tager-Flusberg, H., Fowler, A.
Ozonoff, S., & McEvoy, R. E. (1994). A longitudinal study of E., & Sudhalter, V. (1991). The relation of utterance length to
executive function and theory of mind development in autism. grammatical complexity in normal and language-disordered
Development and Psychopathology, 6, 415–431. groups. Applied Psycholinguistics, 12, 23–45.
Ozonoff, S., & Miller, J. N. (1996). An exploration of right- Schneider, S. G., & Asarnow, R. F. (1987). A comparison of
hemisphere contributions to the pragmatic impairments of cognitive/neuropsychological impairments of nonretarded
autism. Brain and Language, 52, 411–434. autistic and schizophrenic children. Journal of Abnormal Child
Ozonoff, S., Pennington, B. F., & Rogers, S. J. (1991). Executive Psychology, 15, 29–45.
function deficits in high-functioning autistic individuals: Semel, E., Wiig, E. H., & Secord, W. (1987). Clinical evaluation of
Relationship to theory of mind. Journal of Child Psychology language fundamentals – revised. San Antonio, TX: The Psy-
and Psychiatry, 32, 1081–1105. chological Corporation, Harcourt Brace Jovanovich.
Ozonoff, S., South, M., & Miller, J. N. (2000). DSM-IV- Shallice, T. (1982). Specific impairments of planning. Philosophical
defined Asperger syndrome: Cognitive, behavioral and early Transactions of the Royal Society of London, Series B, Bio-
history differentiation from high-functioning autism. Autism, logical sciences, 298, 199–209.
4, 29–46. Szatmari, P., Tuff, L., Finlayson, M. A., & Bartolucci, G. (1990).
Ozonoff, S., & Strayer, D. L. (1997). Inhibitory function in Asperger’s syndrome and autism: Neurocognitive aspects.
nonretarded children with autism. Journal of Autism and Journal of the American Academy of Child and Adolescent
Developmental Disorders, 27, 59–77. Psychiatry, 29, 130–136.
Ozonoff, S., & Strayer, D. L. (2001). Further evidence of intact Tager-Flusberg, H. (1981). On the nature of linguistic functioning
working memory in autism. Journal of Autism and Develop- in early infantile autism. Journal of Autism and Developmental
mental Disorders, 31, 257–263. Disorders, 11, 45–56.
Pascualvaca, D. M., Fantie, B. D., Papageorgiou, M., & Mirsky, Turner, M. (1997). Towards an executive dysfunction account of
A. F. (1998). Attentional capacities in children with autism: Is repetitive behaviour in autism. In: J. Russell (Ed.), Autism as
there a general deficit in shifting focus? Journal of Autism and an executive function disorder. (pp. 57–100). Oxford: Oxford
Developmental Disorders, 28, 467–478. University Press.
Language and Executive Functions in High Functioning Autism 573

Van Bourgondien, M. E., & Mesibov, G. B. (1987). Humor in Welsh, M. C., & Pennington, B. F. (1988). Assessing frontal lobe
high-functioning autistic adults. Journal of Autism and Devel- functioning in children: Views from developmental psychol-
opmental Disorders, 17, 417–424. ogy. Developmental Neuropsychology, 4, 199–230.
Wechsler, D. (1974). Wechsler intelligence scale for children – Wiig, E. H., & Secord, W. (1989). Test of language competence –
revised. San Antonio, TX: The Psychological Corporation. expanded edition. New York: The Psychological Corporation.
Wechsler, D. (1981). Wechsler adult intelligence scale – revised. San Williams, D., & Mateer, C. (1992). Developmental impact of
Antonio, TX: The Psychological Corporation. frontal lobe injury in middle childhood. Brain and Cognition,
Wechsler, D. (1991). Wechsler intelligence scale for children (3rd 20, 196–204.
ed.). San Antonio, TX: The Psychological Corporation.