Defining language phenotypes in autism

Abstract

All children with autism spectrum disorders have deficits in pragmatic aspects of communication; however, formal language abilities are extremely heterogeneous, ranging from nonverbal to superior linguistic skills. Recent studies have focused on defining different language phenotypes among verbal children. One subtype has been compared to specific language impairment (SLI), a language disorder that is diagnosed on the basis of delays and deficits in language acquisition in the absence of hearing impairment, frank neurological damage or co-morbid psychopathology. Two behavioral studies address the question of whether children with autism and language impairment have specific language deficits that are similar to those found in SLI. These experiments focused on phonological processing in a nonsense word repetition task, and use of grammatical morphology in conversational speech. The findings from these studies are discussed in the context of recent neuroimaging and genetic studies of autism.

Introduction

One of the hallmark features of autism spectrum disorders (ASD) is wide variability in language skills [1]. Although deficits in pragmatic aspects of communication, specifically the ability to use both literal and non literal language appropriately in a range of social contexts, are universal symptoms even among adults with Asperger syndrome [2], [3], formal language deficits are not core features of ASD [4]. [5] did not refer to language problems in his original discussion of the defining features of infantile autism; however, later he described some of the idiosyncratic clinical features of language use among the children he had diagnosed, including echolalia, pronoun reversals and stereotyped language [6].

There is growing interest among researchers and clinicians in delineating different subtypes in autism. For researchers, this would create more homogenous groupings that could facilitate finding the underlying causes of autism; for clinicians, subtyping would enhance the ability to tailor treatments to the specific needs of individuals based on their particular pattern of impairments. Rapin and her colleagues [7], [8], [9] were the first to describe the different types of language disorders found among children with autism, highlighting some of the similarities between autism and other developmental language disorders. More recently, [10] identified two language subtypes among verbal children with autism: children with normal linguistic abilities (phonological skills, vocabulary, syntax and morphology) and children with impaired language [10] argued that the language phenotype of the autistic children in the latter subtype was similar to the language phenotype that has been described for children with specific language impairment (SLI), a disorder characterized by delays and difficulties in language acquisition in absence of other sensory, cognitive or affective disorders [11]. Like autism, SLI represents a heterogeneous population [12]. There is no agreement among clinicians or researchers on uniform diagnostic criteria, including specific instruments, measures, or cut-off points that define this population. Nevertheless, in recent years, some consensus has been reached about language deficits that may be viewed as central to this disorder. Two clinical markers have been highlighted: deficits in phonological processing and grammatical morphology [12]. These clinical markers are now viewed as core features that define the phenotype of SLI, analogous to the social, communicative and restricted behavior patterns that are at the core of an autism diagnosis.

In an initial comprehensive study, [13] investigated language profiles in a large group of verbal children with autism using a battery of standardized language tests tapping articulation, phonological processing, vocabulary, and higher order grammatical and semantic skills. Two subgroups of children with autism were identified, based on their performance on these tests. About one quarter of the children had language test scores that fell within in the normal range (ALN subtype). The remaining children scored significantly below the mean on all the language tests except for the articulation measure. These children had a distinctive profile of performance across the language measures (moderate impairments in vocabulary and phonological processing and more severe impairments in higher order syntax and semantics), suggesting that they formed a subtype of children with autism and language impairment (ALI subtype). Although there was a moderate relationship between language scores and IQ, there were children with high and low IQ scores, and normal or delayed onset of language milestones, in both the ALN and ALI language subgroups. The profile of scores of the ALI children resembled closely the profiles reported for children with SLI [12]. The poor performance of these children on the test of phonological processing, a standardized nonsense word repetition test, was especially significant since, as noted earlier, this kind of measure is considered highly sensitive to the diagnosis of SLI [13], [15], [16], [17]. However, it was not clear from this initial study whether the ALI children made the same types of errors when repeating nonsense words as do children with SLI, or whether deficits in nonword repetition are related to vocabulary knowledge, as has been found in SLI [17].

In addition to impairments in phonological processing SLI is characterized by grammatical deficits, especially in marking verbs for tense [18], [19]. Long after their age-matched peers have stopped making errors, children with SLI continue to omit grammatical morphemes in their speech; for example, marking the past tense (e.g., She play-ed in the garden). Studies by Rice and her colleagues showed that for English-speaking children with SLI children’s grammatical marking of tense distinguished preschool children with SLI from unimpaired children in both conversational speech as well as on experimental probes [19], [20], [21]. Drawing on part of the same sample that had participated in [14] study, [22] investigated grammatical tense marking in children with autism. About half the children had the ALN phenotype and half the ALI phenotype. The children were given experimental probes to elicit past tense and third person present tense (e.g., A firefighter put-s out fires), using the same tasks designed by [20]. The ALI children performed poorly on these probes (30% correct on past tense and 37% correct on third person present tense) compared to the ALN children (76% and 64%, respectively). The most common error pattern among the ALI children was to omit any morphological marking on the verb stem, the error that is most frequently reported for children with SLI.

The studies reviewed here provide support for the view that there are parallels in language profiles between a subgroup of children with autism, ALI, and children with SLI; however, it is not clear whether these parallels are misleading in that the language deficits stem from different underlying mechanisms or whether they reflect more substantive overlap in the phenotypes of these syndromes. The following studies were conducted to follow up on the findings from our earlier research on phonological processing and grammatical morphology among children with autism.