15
The Challenge of Studying
Language Development in
Children With Autism
Helen Tager-Flusberg
University of Massachusetts
Autism is a rare neurodevelopmental disorder of genetic origin. Among the primary characteristics of autism are impairments not only in language
but also in communication. In this chapter I address the challenge of studying language development in children who lack the basic motivation to
communicate with others, even when they may have acquired some linguistic competence. I begin with a description of the disorder and an
overview of the kinds of language deficits that have been identified in autism. Then I take up different methodological approaches that have been
used in research on language development in autism and discuss some problems encountered in using these methodologies. In the final section of
the chapter I explore the possibility of using novel methods drawn from other research paradigms that may help shed further light on the mysteries
of why children with autism seem to have such limited interest or ability to communicate with others.
WHAT IS AUTISM?
Over 50 years ago Leo Kanner, an eminent child psychiatrist at Johns Hopkins University, first described a set of 11 children, all of whom
had in common a core set of atypical characteristics (Kanner, 1943). These children were set apart from other children, including those with
emotional disturbance or mental retardation, by their social withdrawal. They appeared to have little contact with others and little interest in
people, leading Kanner to view “extreme autistic aloneness” as the cardinal feature of the new syndrome he had identified. Kanner noted a number
of other major features of this syndrome, including obsessive insistence on routines and lack of change in the environment; deficits in language
ranging from mutism to acquiring the ability to speak but using language in a somewhat meaningless and repetitive way; excellent rote memory
skills; a variety of anomalous reactions to sensory stimulation; and sleep and eating problems. Although the children Kanner described were quite
different from one another, in terms of both their current levels of functioning and their developmental histories, Kanner distilled from this
diversity the core similarities that warranted a new diagnostic classification.
The essence of the syndrome captured by Kanner still rings true for clinicians and researchers today. Autism is currently classified as a
form of pervasive developmental disorder of early onset, usually during the infant or toddler years. The three essential criteria for diagnosing
autism include (a) qualitative impairments in social interaction. (b) delays and deficits in language and communication, and (c) restricted repetitive
and stereotyped behaviors, activities, or interests (American Psychiatric Association, 1994). Within this “triad” of impairments symptom
expression varies, depending on the severity of the disorder, the IQ level, and the age of the child. The majority of individuals with autism are
mentally retarded; these children tend to be more socially withdrawn, have more stereotyped behavior patterns, are less communicative, and may
have little or no functional language. Thus, in autism the overall level of cognitive ability is closely connected to social and language functioning.
However, even in high functioning children serious limitations in social interest make the study of language acquisition difficult: There is often no
interest on the part of children with autism to communicate with others or respond to others’ initiations, even when the children have acquired
some linguistic ability. This is the challenge faced by researchers who wish to study language in autism.
COMMUNICATION AND
LANGUAGE IN AUTISM
Parents of children with autism most often report that the first sign of a problem with their child is either the absence of language or the
loss of language that had begun to develop in the second year of life (Kurita, 1985; Lord & Paul, 1997). Sometimes the initial concern may be that
the child is deaf because they are so unresponsive to the voice of others including parents in their environment. In retrospect, many parents
recollect that even during the first 12 months their infants were unresponsive to adult contact, did not engage in turn-taking games, and failed to
develop joint attention (Volkmar, Carter, Grossman, & Klim, 1997). By their first birthday, many infants who later receive the diagnosis of autism
do not respond to their own name and fail to make eye contact (Osterling & Dawson, 1994). By the end of the second year, toddlers with autism
still have no functional language and are extremely limited in their communication with others, perhaps only engaging another person to fulfill
requests using protoimperative gestures (Stone, 1997). They also do not engage in any symbolic play. To some extent, the primary social deficits
in autism set the developmental course for deficits in language and communication there is simply no interest or “appetite” for interacting with
others at any level or by any means, including language.
Nevertheless some children with autism, usually those that are less severely impaired overall, do increase the frequency of their
communicative attempts and begin acquiring language before their fifth birthday. Indeed, acquiring some functional language by age 5 has been
found to be the most powerful predictor of a more positive outcome in autism (Rutter, 1970; Ventner, Lord, & Schopler, 1992). This group of
verbal children has been studied by researchers using a variety of methodological approaches, some with greater success than others. The findings
from many studies of children with autism have provided a general picture of the patterns of relatively spared and impaired capacities, as well as
clinical descriptions of the most striking characteristics of autistic language.
Characteristics of Autistic Language
Kanner (1943, 1946) was the first to note that children with autism often simply echo the words, phrases, or sentences spoken by others. This
classical feature of autistic language, known as echolalia, is most typical of children who have very little productive language (McEvoy, Loveland.
& Landry, 1988). Echolalic speech often retains the exact words and intonation used by others either immediately or after some time. It is now
viewed as having some functional value for children. Echolalia may help children with autism to maintain some role in the ongoing discourse even
when they either do not understand or have not yet acquired the pragmatic or linguistic skills needed to respond more appropriately (Prizant &
Duchan, 1981; Tager-Flusberg & Calkins. 1990). Kanner (1946) also noted the autistic child’s tendency to use words with special or unique
meanings not shared by others. The use of idiosyncratic lexical terms, or neologisms, has been found even in higher functioning children and
adults with autism (Volden & Lord, 1991), suggesting that it does not mark a developmental stage in acquisition. The source of these “words” and
their function has not yet been elucidated. Another striking feature of autistic children’s use of language is their reversal of pronouns—referring to
themselves as “you” and their conversational partner as “I.” Although reversing personal pronouns is not unique to autism, it does occur more
frequently in this group than in any other population (Lee, Hobson, & Chiat, 1994) and pronoun reversals are viewed as important in the diagnosis
of this disorder (American Psychiatric Association, 1994; Le Couteur et al., 1989). The reversals reflect difficulties in conceptualizing the notion
of self and other as it is embedded in shifting discourse roles between speaker and listener (Lee et al., 1994; Tager-Flusberg, 1993, 1994).
Autism has been identified as a language disorder that, at its core, involves pragmatic impairments (Baltaxe, 1977; Lord & Paul, 1997; TagerFlusberg, 1981a). Children with autism are often unresponsive to the conversational initiations of others. This has led many to question whether
autism involves particular difficulties in comprehension (Paul & Cohen, 1984), although this has proven difficult to investigate using standard
methods. Even when autistic children do engage and respond to others, they may offer little to the ongoing discourse, have difficulty sustaining the
conversational topic, or offer irrelevant comments (Tager-Flusberg & Anderson, 1991). These discourse deficits are seen as central to the defining
characteristics of autism (Tager-Flusberg, 1996); they parallel and are closely linked to the social and communicative impairments discussed
earlier.
All these features of language in autism—echolalia, neologisms, pronoun reversals, and pragmatic impairment—make it difficult to investigate
language acquisition in this population. Indeed, the features have led some researchers to question whether autistic children develop language
following the same pathways, and using the same underlying mechanisms, as do normally developing children or other children with delayed or
deficient language (e.g., Simon, 1975). Nevertheless, over the past two decades psycholinguistic approaches have led to important advances in the
study of language in children with autism. In the next sect ions I review some of this work, focusing on the methodological challenges that stem
from the particular deficits that are central to this disorder.
STUDYING LANGUAGE
DEVELOPMENT IN AUTISM
The study of language in children with autism has been limited almost exclusively to those children who do acquire some functional language,
either spoken or sign (cf. Layton & Baker, 1981). Moreover, because of the behavior difficulties experienced by many autistic children, studies
have been further restricted to those who are more cooperative, less aggressive, or self-injurious. Because of the rarity of the syndrome and the
inherent difficulties in conducting research with autistic children, most studies have included very small samples, sometimes just single case
studies. These limitations mean that research in this area has not been able to capture the full variation that is known to exist in the population.
There are also so few longitudinal studies that little is known about developmental changes occurring over the course of childhood.
Collecting Natural Language Samples
In the field of child language, it is generally acknowledged that the richest source of data, especially on productive use of language, comes from
spontaneous speech samples. These may be collected in different contexts as the young child interacts with some other person, either a researcher
or familiar person such as the mother. This approach has been the most well utilized method for studying development in young children (cf.
Brown, 1973). It is the most open-ended and least structured approach to studying language, which is often viewed as an advantage in that it
allows one to view what children can do in a natural context without the external imposition of constraints or task demands that may not be
understood by the child.
But what might be viewed as advantages for the normally developing child—the absence of any external constraint—operates as a distinct
problem for the child with autism. The lack of structure in which natural language samples are typically collected is an especially difficult context
for children with autism. Given no constraint or external demand, the autistic child enjoys his or her own isolation and does not spontaneously
socially engage with others or communicate much with them. This means that very little language might be gathered using this methodological
approach, thereby making the collection of meaningful data very difficult. Children with autism do not spontaneously speak that much so the
methods that involve the analysis of spontaneous speech may be of limited utility with this population.
However, several studies have investigated the frequency of communicative acts in autistic children in different social contexts. These studies
generally demonstrate that children with autism are sensitive to social context in ways that parallel those of very young normally developing
children. Interactions with peers are the most difficult, even for high functioning older verbal children with autism. Stone and Caro-Martinez
(1990) collected language samples from school-age children in their classrooms while they were engaged in free play or other informal activities.
They also observed each child in their study for several hours, spread over a number of days. The average frequency of spontaneous
communicative acts from the children with autism was just two or three per hour, mostly directed toward an adult. Only half their subjects ever
spoke to a nonautistic peer In other experimental settings, such as the laboratory, other studies also found that children with autism speak much
less frequently and use a narrower range of speech acts when engaged in free play (Landry & Loveland, 1989; Mermelstein, 1983). The effect of
social context on the frequency of communicative acts in children with autism was systematically investigated by Bernard-Opitz (1982). She
found that an autistic child was most likely to talk in a highly familiar setting with a highly familiar person. Furthermore, Bernard-Opitz found that
in these social settings, at home with the mother, the autistic child produced more advanced language (i.e., more complex grammatical
constructions) than in other social contexts. This means that laboratory-based studies using an experimenter as the primary conversational partner
will not provide the most accurate portrait of the autistic child’s linguistic capacities.
Taking these concerns into consideration, my colleagues and I conducted a small-scale home-based longitudinal study of language development
in six young children with autism (Tager-Flusberg et al., 1990). Using both audio- and videotape, we collected hour-long language samples from
the children as they interacted with their mothers in their homes, at bimonthly intervals. Data were collected in a similar way from six young
children with Down syndrome who were matched on age and language level (assessed by mean length of utterance, or MLU) to provide a
comparison group. As expected, the autistic children did speak less than the children with Down syndrome, even in this context. Nevertheless, we
were able to collect sufficiently large language samples from all the children most of the time (usually well over 300-400 utterances) that would
allow us to conduct meaningful analyses on developmental changes in syntax, morphology, and lexical and pragmatic functioning in these
children. Our success was based, in part, on the children’s mothers, who knew that the focus of our research was language. The mothers worked
hard at engaging their children in conversation, using many different approaches and activities to keep them talking.
New problems were encountered as we set about analyzing the transcripts we prepared from the taped language samples. The youngest children
and those who were the least mature linguistically, not surprisingly, produced quite of bit of echolalia. The question we faced was how to handle
these echolalic utterances in our explorations of grammatical and lexical development. We concluded that if we included echolalic speech we
might overestimate the child’s linguistic knowledge. Maybe these children with autism echoed words they did not know, or maybe they were able
to repeat utterances that were longer or more complex than they could produce spontaneously on their own. Although in normally developing
children, imitated speech may also be more advanced than nonimitated, this poses little problem because it is so infrequent in their natural
language samples (e.g., Bloom, Lightbown, & Hood, 1975). We also found that the autistic children would often recite speeches, count, or sing as
a way of filling “air-time.” These learned segments of speech might also lead to the overestimation of the children’s linguistic knowledge. At the
same time, other characteristics of autistic children’s language could lead us to underestimate their language abilities. In particular, we noticed that
the children’s mothers asked many questions to which the children often gave single-word yes/no responses rather than an extended reply. These
single-word responses and the use of stock social phrases (such as “please” or “thank you”) could lead us to underestimate MLU, the single most
used measure of language development across a variety of populations (Brown, 1973; Miller & Chapman, 1981).
Thus, the analysis of spontaneous speech data from children with autism clearly needed to be tailored to the particular features that are
characteristically found even in relatively high functioning verbal children. We chose to handle these concerns by preparing a separate corpus of
100 child utterances from each transcript for both the autistic and the comparison children with Down syndrome. These corpora excluded all
echolalic utterances, defined as all full or partial imitations or repetitions within live transcript lines. They also excluded routine phrases, singleword yes/no responses, and learned speeches or songs.
Our analyses of these special corpora, that had eliminated the most problematic aspects of autistic language use, provided us with a rich and
detailed picture of the development of language in this population (Tager-Flusberg et al., 1990). Contrary to claims in the literature that children
with autism do not show a gradual increase in MLU or the same order for the emergence of grammatical constructions (e.g., Menyuk & Quill,
1985; Simon, 1975), we found that the developmental pathway for our autistic subjects was similar to what has been found for normally
developing children. In fact, when we carefully examined echolalic utterances from the children with autism we found, surprisingly perhaps, that
they were not developmentally more advanced than nonecholalic utterances from the same points in time. Echolalic utterances tended to be shorter
and contain less complex constructions than spontaneously produced utterances (Tager-Flusberg & Calkins, 1990). These findings were taken as
evidence that the underlying mechanisms for acquiring language were the same in children with autism, children with Down syndrome, and
normally developing children. Despite the obvious surface and pragmatic differences in their speech, children with autism who do acquire
language appear to depend on the same mechanisms and processes for developing language as do other children (cf. Prizant, 1983). In some,
though not all, children, even the rate of development was similar to what has been found among normally developing children.
Despite these challenges in both collecting and analyzing spontaneous speech samples, they have provided the richest and most accurate source
of information on the development of communicative functioning (e.g., Coggins & Carpenter, 1981; Tager-Flusberg & Anderson, 1991; Wetherby
& Prizant, 1990). Natural language samples have demonstrated the restricted repertoire of speech acts, as well as deficits in communicative
competence, that are characteristic of children with autism. Other methods for investigating these aspects of language and communication are not
available. Indeed, it is in this context of unstructured interaction that the autistic child’s communication deficits can be most clearly highlighted
and explored.
Using Standardized Assessment Tools
Often it is extremely useful to assess a child’s linguistic ability—in particular knowledge of structural aspects of language—using standardized
instruments in order to address a number of research issues. Because these instruments tap a variety of domains, including phonology, semantics,
lexical knowledge, syntax, and morphology, it is possible to explore the relation among these domains in both production and comprehension by
comparing a child’s performance to age-based norms. This is, perhaps, the clearest way of examining the overall language profile in autism to
address a number of interesting questions about the representation and processing of linguistic information. Such questions include whether
comprehension is more impaired than production (e.g., Lord, 1985), whether lexical/semantic knowledge is more impaired than computational
aspects of language (e.g.. Menyuk & Quill. 1985; Tager-Flusberg, 1981a), or whether language is more impaired than nonlinguistic cognitive
ability (Lord & Paul, 1997; Tager-Flusberg, 1989). Answers to these kinds of questions may provide important information about the cognitivelinguistic phenotype of autism and how it compares to other groups of mentally retarded or language disordered children.
Standardized tests are also often used to provide measures by which to match children with autism to control groups in more experimental
studies of language or other aspects of cognitive functioning. The most widely used measure for this purpose is the Peabody Picture Vocabulary
Test (PPVT; Dunn & Dunn. 1981, 1997), which provides a norm-based measure of receptive vocabulary. Some studies might also include a
measure of syntactic comprehension, particularly if the experimental task requires the child to process complex linguistic information such as
stories or scripts. Examples of tests that have been used to assess syntactic comprehension include different subtcsts on the Test of Auditory
Comprehension of Language (TACL; Caaow-Woolfolk, 1985), the Test of Language Development (TOLD; Newcomer & Hamill, 1991), and the
Clinical Evaluation of Language Fundamentals (CELF; Semel, Wiig, & Secord, 1995). Finally, it is important to keep in mind that standardized
tests, such as the ones listed here, are most widely used in clinical settings to provide assessment of language abilities in children in order to
inform treatment and intervention planning.
These kinds of measures provide a particular set of challenges when they are used with autistic children. Psychometrically based instruments
must be administered in a particular way by a trained individual if they are to provide valid and reliable data about language performance. But
there may be difficulties encountered by the researcher in engaging the autistic child in the demands of a particular test. Generally, children with
autism are not intentionally negative or noncompliant (Volkmar, 1987); rather, their lack of response to some standardized tests may be the result
of not understanding the pragmatics of the testing situation. Typically, an unfamiliar person administers the test. As discussed earlier, the unique
social deficits in autistic children might make them particularly unresponsive to the researcher in their presence. If more than one researcher is
present, this might add to the autistic child’s anxiety. The environment itself may also add to the distractibility of the child with autism. Even when
they are able to understand the test questions and have the requisite linguistic knowledge, children with autism may be quite unresponsive in an
unfamiliar testing situation.
Another concern that stems from the unique behavioral difficulties of autistic children is their tendency to perseverate, or repeat an action or
response. Thus, if the answer to the first item on a task requires a child to point to a picture in a particular location—say, the first one on the left
side of an array—the child with autism might continue to point to this location on all subsequent items. This kind of response may be produced not
because the child does not understand the later items but because he or she is perseverating. Many tests of language comprehension involve
pointing to pictures that correspond to words or sentences presented by the tester, so this kind of problem must be considered in evaluating
responses given by an autistic child.
Tests of language production often have more complex pragmatic demands than do measures of comprehension. Not surprisingly, children with
autism respond best to tests that require them to imitate or recall words or sentences spoken by the tester (or better still, presented on a recording),
such as subtests on the TOLD. Yet these kinds of tests might not provide a particularly accurate measure on the child’s knowledge. Children with
autism also respond well on confrontational naming tasks, which asks them to label pictures that are usually of concrete objects. But because they
have difficulty understanding the more complex instructions for other kinds of semantic tests, autistic children usually perform worse on tests that
tap relational or abstract meaning. Some researchers have taken this discrepancy in performance as a sign that autistic children have particular
semantic deficits, which may or may not be warranted (Menyuk & Quill, 1985; Tager-Flusberg, 1986). Sentence completion tests or sentence
formulation tests (which involve asking a child to create or complete a sentence using a word or phrase given by the tester) are often not easily
completed by children with autism. Examples of this kind of test include the Formulated Sentences subtest on the CELF (Semel et al., 1995) and
the Grarnmatic Completion subtest on the TOLD (Newcomer & Hammill, 1991). Autistic children may misunderstand the instructions and imitate
what they have heard, rather than ending or formulating their own utterance. It is not clear whether this kind of response reflects expressive
deficits or pragmatic problems in understanding the test instructions.
All these variables, in addition to the fluctuating behavior and attention that is quite typical in autistic children may make it particularly hard to
test them in a standardized way (Sparrow, 1997). Because standardized tests typically start with easy items and progress to more difficult ones,
children with autism may also begin to experience frustration as they find themselves unsure whether they have answered correctly. Frustration
can lead to tantrums or other behavior difficulties, which make it impossible to continue with the testing.
On the other hand, one significant advantage of using standardized tests is that they are highly structured, which may help to maximize the
performance of autistic children. Researchers can increase the likelihood that an autistic child will complete a standardized test in a relatively
reliable way by making certain accommodations to the particular needs of the child. Using behavioral approaches to enhance the structure of the
testing situation, including providing regular rewards (either tangible, such as food or tokens, or social praise) may be especially useful,
particularly for those children who have had experience with these approaches in intervention programs or school settings. Providing frequent
feedback and short breaks can also be helpful in keeping an autistic child engaged in standardized testing. Sometimes it is necessary to rephrase,
simplify, or regularly repeat instructions, even though this is not strictly allowed by the standardized procedures mandated by a test. Even though
these kinds of modifications might make the comparison of the child’s performance to the age-based norms questionable, it may be the only way
to obtain any meaningful data from an autistic child on a particular standardized test (Sparrow, 1997).
Experimental Methods for Studying Language
As is known from research on language acquisition in normally developing children, insights into the process and knowledge of children at
different stages of development are best captured by experimental studies that use specially designed tasks and stimuli. Spontaneous speech data
are limited by what the child actually produces; they do not reveal what the child may know but had no opportunity to use; they also provide only
minimal information about the child’s comprehension of language. Standardized tests are designed to capture individual variation; they typically
do not include a complete set of grammatical or morphological constructions, and often their item complexity is based more on informational load
than on underlying grammatical complexity. Thus, standardized tests typically increase the length of sentences to make them more difficult or add
more choices of responses; these are changes that add to the memory or attentional load of-the test rather than its linguistic complexity. Such tests
are not applicable to the study of the development of linguistic knowledge in all its richness. Experimental studies help to fill these gaps and have
provided the detailed knowledge that is now available in the field of developmental psycholinguistics.
There is no room in this chapter to cover the full range of experimental methodologies that have been used in the field of child language.
Instead I cover just a few, to illustrate some of the unique issues that have to be considered by researchers who wish to employ experimental tasks
with autistic children. In fact, there have been very few experimental studies of language processing published in the literature on autism (see Lord
& Paul, 1997, for a recent review). Perhaps this is because the difficulties faced by researchers who wish to conduct such studies with this
population have led them to seek alternative methods; this is especially true when one considers how hard it is to find a sufficiently large sample
and the need to include matched control groups.
Many experimental studies of child language, which typically target young normally developing children, make use of toy props to facilitate
the interchange between the child and experimenter and to enhance the child’s enjoyment of the tasks. For example, many studies of language
comprehension engage toy figures that the child must manipulate to demonstrate the meaning of a sentence. Studies that explore children’s
grammatical judgments of sentences will often employ puppets that “speak” the correct or incorrect sentences.
These methods take advantage of children’s propensity to engage in symbolic and functional play, especially with a variety of fictional
characters participating in different activities. But the child with autism has no such propensity. One of the primary ways in which social and
communicative deficits are manifest in autism is in the absence of spontaneous symbolic play (Sigman & Ungerer, 1984; Stone, Lemanek, Fishel,
Fernandez, & Altemeier, 1990). These children do not play with toy objects in conventional ways and later on show no interest or capacity for role
play (Volkmar et al., 1997). At the least, these deficits in play mean that autistic children may not have the same motivation and pleasure that other
children find in the experimental tasks designed to tap linguistic knowledge. Of greater concern is that fact that for some children with autism,
difficulties with these tasks may confound the absence of the linguistic knowledge embedded in the task with lack of understanding the
representational value of the toys. If pictures are used to replace toys, then one must beware of perseverative responses, as discussed earlier.
Stimuli for experimental tasks often depict animals as well as people, as if they were sentient beings. Researchers tend to imbue animals with
human characteristics when interacting with children; story books are filled with these kinds of representations, and experimental tasks in child
language research often do the same. This kind of representation assumes that young children have an implicit theory of mind: that they interpret
action within a causal mentalistic framework (e.g., Wellman, 1990). There is considerable evidence that normally developing toddlers do infer that
people are intentional and that their own and others’ behavior and actions can be explained on the basis of desires, emotions, and beliefs. By
extension, within a play context, children also accept that other animals may also have minds and that their actions can be interpreted in the same
way as those of human beings.
But again, the use of these kinds of stimuli poses special problems for the child with autism. Indeed, autism is now viewed as a disorder that
can be interpreted as involving core primary impairments in the acquisition of a theory of mind (Baron-Cohen, Leslie, & Frith, 1985; BaronCohen, Tager-Flusberg, & Cohen, 1993). Children with autism do not readily interpret their own or others’ actions in terms of mental states; they
have difficulty even viewing people as intentional beings. The combination of deficits in theory of mind, and in symbolic play (which are likely to
be closely connected: Leslie, 1987), make the use of stimuli that treat animals as if they were people especially problematic for children with
autism. Many examples can be found in the psycholinguistic literature on the comprehension of grammatical constructions by normally developing
children that use stimuli of this sort. For example, studies on the comprehension of relative clauses (de Villiers, Tager-Flusberg, Hakuta, & Cohen,
1979; Tavakolian. 1981), coordination (Tager-Flusberg, de Villiers, & Hakuta, 1982), passive constructions (Maratsos, Kuczaj, Fox, & Chalkley,
1979), and anaphora (Chien & Wexler, 1990) have all included anthropomorphic animals as the prime characters in the stories or sentences that
the children were asked to interpret. All these studies might be hard to replicate with autistic children because of the stimulus demands of the task.
In considering experimental methods used to elicit different kinds of linguistic constructions, some of the same concerns that were highlighted
for standardized testing may also apply. Children with autism may not understand the pragmatic demands of the task or the task instructions.
Gamelike contexts that work so well for other children are not appreciated or even clearly understood by autistic children. They tend to repeat
what the experimenter says, rather than complete or formulate their own response. If a child with autism is asked to describe an event enacted by
an experimenter or depicted in a photograph or sequence of pictures that create a story, he or she is just as likely to simply name the objects (e.g.,
Tager-Flusberg, 1995) as to provide a narrative description. The problem lies in how to interpret this kind of response. Does it signal deficits in
language, particularly in the ability to produce sentences, or is it the result of other aspects of the autistic syndrome?
As a final example to illustrate the methodological challenges faced by researchers of children with autism, I discuss an early attempt I made to
study grammatical knowledge in children with autism. As part of my dissertation research. I wanted to use Jean Berko Gleason’s famous wug
paradigm to explore knowledge of grammatical morphology in children with autism (1958). Using the stimuli that Berko Gleason had created, and
following her exact methods, I piloted this task on three children who clearly met criteria for the diagnosis of autism. I failed miserably! The
children simply did not understand what I was looking for or what I was trying to do; they clearly viewed me as somewhat bizarre. As I explored
the source of the problem, it became clear that the novel stimuli (wugs, ricking, etc.) were quite incomprehensible to the children. Children with
autism are concrete and literal; they have no idea how to handle the invented imaginative forms and the creative methods that Berko Gleason so
successfully used on normally developing children. Perhaps, had I persevered and worked out a modified approach, 1 might have been able to
study autistic children’s knowledge of wugs with greater success. Instead, I addressed a different set of questions about sentence comprehension
strategies, which turned out to be more easily implemented (cf. Tager-Flusberg, 1981b). To this day, there is no research on autism that has used
the kind of elicited production methods that were pioneered by Jean Berko Gleason 40 years ago.
CONCLUSIONS AND
FUTURE DIRECTIONS
Studying language acquisition in children with autism is not an easy or straightforward task. Methodologies and stimuli have to be carefully
selected, adapted, and modified to the unique set of deficits that define autism. Nevertheless, there has been some success over the past few
decades in using certain approaches. Probably most has been learned about language development from the study of natural language samples.
collected in different structured or unstructured settings. Standardized tests have also been used with some degree of success with higherfunctioning children. But as I have already discussed, these approaches are limited and can only address certain kinds of questions. Furthermore,
all the research on language development in autism has been limited to the group of children who do acquire some functional language that they
use spontaneously. Half the population does not (Lord & Paul. 1997). Questions remain about whether these nonverbal children understand any
language and whether they perceive speech in the same way as other children do. It is not known whether the source of their difficulties with
language stems from deficits in symbolic capacity, motivation to communicate with others, mental retardation, or a combination of all these
factors. Until we as researchers begin to study these children, we will not achieve a full understanding of the language deficits that are central to
the diagnosis of autism.
Even verbal autistic children have been studied in only limited ways. There are too few longitudinal studies or experimental studies of either
comprehension or production. Some of the reasons for the paucity of research in this area have been discussed in this chapter. But if we are to
make further advances, we will need to find ways of adapting our methods to address the unique demands and challenges of working with autistic
children. Very different sorts of approaches will be needed to study verbal and nonverbal children with autism—in the same way as in the field of
child language, different methods are used to study prelinguistic infants compared to toddlers and preschoolers who have begun acquiring
language.
One important way of addressing some of the challenges posed even by verbal children with autism would be to employ several different
methodologies to investigate a particular set of research questions with the same group of autistic subjects. Using several methods simultaneously
(e.g., standardized tests of language production, language samples, and elicited production methods) would help to minimize the problems found
with each method used in isolation and would provide a richer set of data for analysis. The design of experimental tasks should take into
consideration some of the concerns discussed in the previous section. The selection of standardized tests should also be made on the basis of which
ones are best suited to this population. The multimethod approach would help to address some of the issues outlined earlier and provide
researchers with a more complete picture of (a) how children with autism do acquire language, and (b) what accounts for their remaining deficits
in communication, discourse, and certain aspects of language.
What about nonverbal autistic children? Little is known about this group because so few research studies have focused on them. The
combination of communicative, social, and cognitive impairments that define this group make them especially difficult to study. One strategy that
has not yet received as much attention would be to employ the methods that define the field of behavioral analysis—the use of operant techniques.
especially computer-based technologies, to study basic research questions about the underlying mechanisms that impede the acquisition of
language in this population. Behavior analysis has been extensively used with this population, especially as a primary means for intervention (e.g.,
Can & Durand, 1985; Charlop & Trasowech, 1991; Harris & Ferrari, 1983). Thus, we know that operant methods can be successfully used, even
as a means for assessing nonverbal autistic children (McIlvane, Deutsch, Serna, & Dube, 1988; Powers, 1988, 1997). But these kind of behavioral
methods have not been used by psycholinguists in traditional research designs as a means for exploring basic questions such as whether nonverbal
children perceive speech normally, or whether they understand words or grammatical constructions they do not produce. At the same time,
behavior analysts who have employed these methods, usually in single case-study designs, have not used them to address the kinds of questions
that psycholinguistics ask. It is time to consider integrating the methods and paradigms that have developed independently in the field of
psychology, in order to address some of the fundamental questions about language and communication in autism (Wilkinson, Dube, & Mcllvane,
1997; Wilkinson & Tager-Flusberg, 1998). The methods that involve traditional operant conditioning techniques, including shaping the child’s
response to ensure nonrandom responding, could be particularly successful because social motivation can be replaced with the use of tangible
rewards and experimental presentation of materials can be accomplished via computers, thereby avoiding the social difficulties these children
experience when interacting with a researcher.
The future of research on language in autism will require the development of new approaches and methods that have the potential of avoiding
many of the pitfalls discussed in this chapter. It is time for researchers in this field to follow the creative pathway forged by Jean Berko Gleason 40
years ago that has made much of the work in child language as we know it today possible. Although we may not be able to use her methods
exactly, we must take inspiration from her efforts if we want to come to a deep understanding about the nature of language and communicative
development in this unique population.
ACKNOWLEDGMENTS
Preparation of this chapter was supported by grants from the National Institute on Deafness and Communication Disorders (RO1 DC 01234 and
P01 DC 03610).
REFERENCES
American Psychiatric Association. (1994). Diagnostic and Statistical Manual of Menial Disorders, DSM—IV (4th ed). Washington. DC: Author
Baltaxe, C. A. M. (1977). Pragmatic deficits in the language of autistic adolescents. Journal of Pediatric Psychology, 2. 176—180.
Baron-Cohen, S.. Leslie,A.M., & Frith, U. (1985). Does the autistic child have a’theory of mind?” Cognition, 21, 37—46.
Baron-Cohen, S., Tager-Flusberg, H., & Cohen. D. 1. (Edt.) (1993). Understanding other minds: Perspectives from autism. Oxford: Oxford University Press.
Berko (Gleason), 1. (1958). The child’s learning of English morphology. Word, 14, 150—177.
Bernard-Opiz v. (1982). Pragmatic analysis of the communicative behavior of an autistic child. Journal of Speech and Hearing Disorders, 47, 99—109.
Bloom, L., Lightbown, P., & Hood, L. (1975). Structure and variation in child language. Monographs of the Society for Research in Child Development, 40 (Serial No. 160).
Boehm, A. (1986). Roe/un Test of Basic Concepts. San Antonio: Psychological Corporation. Harcourt Brace.
Brown, R. (1973). A/irs: language. Cambridge, MA: Harvard University Press.
Carr, B. G., & Durand, v. M. (1985). Reducing behavior problems through functional communication training. Journal of Applied Behavior Analysis, 18, 111—126.
Carrow-Woolfolk, B. (1985). Test of Auditory Comprehension of Language (Rev.). Austin: Pro-Ed.
Charlop, M. H.. & Trasowech, J. E. (1991). Increasing autistic children’s daily spontaneous speech. Journal of Applied Behavior Analysis, 24,747—761.
Chien, Y. C. & Wexler, K. (1990). Children’s knowledge of locality conditions binding as evidence for the modularity of syntax and pragmatics. Language Acquisition, 3, 225—295.
Coggins. T., & Carpenter. R. (1981). The Communicative Intention Inventory: A system for observing and coding children’s early internal communication. Applied Psycholinguistics, 2, 235—251.
de Villiers, I.. Tager-Flusberg, H., Hakuta, K.. & Cohen, M. (1979). Children’s comprehension of relative clauses. Journal of Psycholinguistic Research, 8, 499—518.
Dunn, L. M., & Dunn, L. M. (1981). Peabody Picture Vocabulary Test—Revised. Circle Pines, MN: American Guidance Service.
Dunn, L. M., & Dunn, L. M. (1997). Peabody Picture Vocabulary Test—IlL Circle Pines, MN: American Guidance Service.
Gardner, M. (1990). Expressive One-Word Picture Vocabulary Test. Los Angeles: Western Psychological Services.
Harris, S.L., & Ferrari, M. (1983). Developmental factors in child behavior therapy. Behavior Therapy. 14, 54—72.
Kanner. L. (1943). Autistic disturbances of affective contact. Nervous Child; 2, 217—250.
Kanner, L. (1946). Irrelevant and metaphorical language in early infantile autism. American Journal of Psychiatry, 103, 242—246.
Kurita, H. (1985). Infantile autism with speech loss before the age of 30 months, Journal of the American Academy of Child Psychiatrj 24, 191—196.
Landry. S. H.. & Loveland, K. A. (1989). The effect of social context on the functional communication skills of autistic children. Journal a/Autism and Developmental Disorders, 19, 283—289.
Layton. T. I., & Baker, P S. (1981). Description of semantic-syntactic relations in an autistic child. Journal ofAutism and Developmental Disorders, 11. 385—399.
Le Couteur, A., Ruttcr, M., Lord, C., Rios, P., Robertson, S., Holdgrafer. M.. & McLennan. J. D. (1989). Autism Diagnostic Interview: A semi-structured interview for parents and caregivers of autistic
persons. Journal of Autism and Developmental Disorders, 19, 363—387.
Lee, A., Hobson, R. P. & Chiat, S. (1994). I, you. me and autism: An experimental study. Journal o/ Autism and Developmental Disorders, 24, 155—176.
Leslie, A. M. (1987). Pretence and representation: The origins of “theory of mind,” Psychological Review, 94, 412—426.
Lord, C. (1985). Autism and the comprehension of language. In E. Schopler & C. Mesibov (Eds.), Coounonication problems in autism (pp. 5948). New York: Plenum Press.
Lord, C. & Paul, R. (1997). Language and communication in autism. In D. J. Cohen & P R. Volkmar (Eds.), Handbook of autism and pervasive developmental disorders (2nd ed., pp. 195—225). New
York: John Wiley & Sons.
Maratsos, M., Kuczaj, S. A., Fox, D., & Chalkley, M. (1979). Some empirical studies in the acquisition of transformational relations: Passives, negatives and the past tense. In W. A. Collins (Ed.),
Children’s language and communication (pp. l5). Hillsdale, NJ: Lawrence Erlbaum Associates.
McEvoy, R. B., Loveland, K. A., & Landry, S. H. (1988). The functions of immediate echolalia in autistic children: A developmental perspective. Journal of Autism and Developmental Disorders, 18,
657-668.
Mcllvane, W.. Deutsch, C. K., Serna, R. W., & Dube, W. (1998). Behavior analytic assessment of intellectual functioning and disability: An interface between research with nonhumans and humans
with limited language. In S. Soraci & W. Mcllvane (Eds.). Perspectives on fundamental processes in intellectual functioning: A survey of research approaches (pp. 215—239). Stamford, CT: Ablex.
Menyuk, P., & Quill, K. (1985). Semantic problems in autistic children. In B. Schopler & C. Mesibov (Eds.), Communication problems in autism (pp. 127—145). New York: Plenum Press.
Mermelstein, R, (1983, October). The relationship between syntactic and pragmatic development in autistic, retarded, and normal children, Paper presented at the Eighth Annual Boston University
Conference on Language Development, Boston.
Miller, J., & Chapman, R. (1981). The relation between age and mean length of utterances in morphemes. Journal of Speech and Hearing Research, 24, 154—161.
Newcomer, P., & Hammill, D. (1991). Test of Language Development—2: Primary and intermediate. Austin: Pro-Ed.
Osterling, J., & Dawson, C. (1994). Early recognition of children with autism: A study of first birthday home videotapes. Journal of Autism and Developmental Disorders, 24, 247—258.
Paul, H., & Cohen, D. J. (1984). Responses to contingent queries in adults with mental retardation and pervasive developmental disorders. Applied Psycholinguistics,5, 349—357.
Powers, M. D. (1988). Behavioral assessment of autism. In B. Schopler & C. Mesibov (Eds.). Diagnosis and assessment of autism (pp. 139—165). New York: Plenum Press.
Powers, M. D. (1997). Behavioral assessment of individuals with autism, In D. J. Cohen, & F. H. Volkmar (Eds.), Handbook of autism and pervasive developmental disorders (2nd ed., pp. 448—459).
New York: John Wiley & Sons.
Prizant, B. (1983). Echolalia in autism: Assessment and intervention. Seminars in Speech and Language, 4. 63—77.
Prizant, B., & Duchan, J. (1981). The functions of immediate echolalia in autistic children. Journal of Speech and Hearing Disorders; 46, 241—249.
Rutter, M. (1970). Autistic children: Infancy to adulthood. Seminars in Psychiatry. 2, 431-450.
Semel, E., Wiig, B., & Secord, W. (1995). Clinical Evaluation of Language Fundamentals—3. San Antonio: Psychological Corporation, Harcourt Brace.
Sigman, M. & Ungerer. 1. (1984). Cognitive and language skills in autistic, mentally retarded, and normal children. Developmental Psychology, 20, 293-302.
Simon, N. (1975). Echolalic speech in childhood autism: Consideration of possible underlying loci of brain damage. Archives of General Psychiatry. 32. 1439—1446.
Sparrow, S. (1997). Developmentally based assessments. In D. J. Cohen & F. R. Volkmar (Eds.). Handbook of Autism and Pervasive Developmental Disorders (2nd ed., pp. 411—447). New York:
John ‘Wiley & Sons.
Stone, W. (1997). Autism in infancy and early childhood. In D. J. Cohen & F. R. Volkmar (Eds.), Handbook of Autism and Pervasive Developmental Disorders (2nd ed., pp. 266—282). New York:
John Wiley & Sons.
Stone, W., & Cam-Martinez. L. M. (1990). Naturalistic observations of spontaneous communication in autistic children. Journal of Autism and Developmental Disorders, 20, 437—453.
Stone, W., Lemanek, K. L.. Fishel, PT., Fernandez, M. C., & Altemeier, W. A. (1990). Play and imitation skills in the diagnosis of autism in young children. Pediatrics, 86, 267—272.
Tager-Flusberg, H. (1981a). On the nature of linguistic functioning in early infantile autism. Journal of Autism and Developmental Disorders; 11, 45—56,
Tager-Flusberg. H. (1981b). Sentence comprehension in autistic children. Applied Psycholinguistics, 2, 5-24.
Tager-Flusberg, H. (1986). The semantic deficit hypothesis of autistic children’s language. Australian Journal of Human Communication Disorders; 14, 51—58.
Tager-Flusberg, H. (1989). A psycholinguistic perspective on language development in the autistic child. In 0. Dawson (Ed.), Autism. New directions on diagnosis, nature, and treatment (pp. 92—115).
New York: Guilford.
Tager-Flusberg, H. (1993). What language reveals about the understanding of minds in children with autism. In S. Baron-Cohen, H. Tager-Flusberg. & D. J. Cohen (Eds.), Understanding other minds:
Perspectives from autism (pp. 138—157). Oxford: Oxford University Press.
Tager-Flusberg, H. (1994). Dissociations in form and function in the acquisition of language by autistic children, In H. Tager-Flusberg (Ed.), Constraints on language acquisition. Studies of atypical
children (pp. 175—i 94). Hillsdale, NJ: Lawrence Erlbaum Associates.
Tager-Flusberg, H. (1995). Once upon a ribbit”: Stories narrated by autistic children. British Journal of Developmental Psychology 13. 45—59.
Tager-Flusberg, H., & Anderson, M. (1991). The development of contingent discourse ability in autistic children. Journal of Child Psychology and Psychiatry. 32, 1123—1134.
Tager-Flusberg, H. & Calkins. S. (1990). Does imitation facilitate the acquisition of grammar? Evidence from autistic, Down 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 syndrome children. Journal of
Autism and Developmental Disorders, 20, 1—21.
Tager-Flusberg, H.. de Villiers, I., & Hakuta, K. (1982). The development of sentence coordination. In S. A. Kuczaj (Ed.), Language development: Vol 1. Syntax and semantics (pp. 201—243).
Hillsdale. NJ: Lawrence Erlbaum Associates.
Tavakolian, S. (1981). The conjoined-clause analysis of relative clauses. In S. L. Tavakolian (Ed.), Language acquisition and linguistic theory (pp. 167—1 87). Cambridge, MA: MIT Press.
Ventner, A., Lord, C., & Schoplet B. (1992). A follow-up study of high-functioning autistic children. Journal of Child Psychology and Psychiatry, 33, 489-507.
t
/olden, J., & Lord, C. (1991). Neologisms and idiosyncratic language in autistic speakers. Journal of Autism and Developmental Disorders, 21, 109-130.
Volkmar, P. R. (1987). Social development. In D. J. Cohen & A. M. Donnellan (Eds.), Handbook of autism and pervasive developmental disorders (pp. 41—60). New York: John Wiley & Sons.
Volkmar, F H., Carter, A., Grossman, I., & Kim, A. (1997). Social development in autism. In D.J. Cohen & F H. Volkmar (Eds.), Handbook of Autism and Pervasive Developmental Disorders (2nd ed.,
pp. 173—194). New York: John Wiley & Sons.
Wellman, H. (1990). A child’s theory of mind. Cambridge, MA: MIT Press/Bradford Books.
Wetherby, A., & Prizant, B. (1990). Communication and Symbolic Behavior Scales. Chicago:
Riverside.
Wilkinson, K. M.. Dube, W., & Mcllvane, W. (1997). A cross-disciplinary perspective on studies of rapid word mapping in psycholinguistics and behavior analysis. Developmental Review, 16,
125—1 48.
Wilkinson. K. M., & Tager-Plusberg, H. (1998). Application of a lexical principles framework for studying categorization and word learning in different populations. In S. Soraci & W. Mcllvane (Eds.),
Perspectives on Fundamental Processes in Intellectual Functioning: A Survey of Research Approaches (pp. 243—263). Stanford, CT: Ablex.