Syntactical Profiles of Autism Spectrum Disorder: Evidence from Part of Speech Errors on Expressive and Receptive Standardized Speech Assessments

Background: Many individuals with Autism Spectrum Disorder demonstrate language difficulties, even if fluent language has been achieved, such as difficulties with grammar, prosody, and semantics. Previous research is mixed as to whether part of speech errors (e.g., nouns, verbs, adjectives), or syntax knowledge, is a deficit within ASD more broadly or is only present when ASD co-occurs with language impairment. Syntax accuracy is of particular interest from a sex-differences perspective, in that females with ASD have been said to engage in “social camouflage,” using speech and communication patterns to better blend in.

Objectives: The following research investigates part of speech errors on standardized assessments of expressive and receptive vocabulary to (a) investigate syntactical differences as a function of ASD, and (b) to investigate syntactical sex differences within ASD.

Methods: Participants (N = 166, 57 females) comprised children (ages 2-10; n = 82), adolescents (ages 11-17, n = 20), and adults (ages 18-85, n = 64) who participated in an ASD assessment; of these, 98 received an ASD diagnosis. Each participant completed an IQ assessment and standardized measures of receptive language (Peabody Picture Vocabulary Test-4^th edition; PPVT-4) and expressive language (Expressive Vocabulary Test-2^nd Edition, EVT-2). Variables of interest included standard scores on both measures and proportion of errors (number of incorrect items/number of items administered). Thus, there were six total speech error variables: errors for nouns, verbs, and adjectives for both expressive and receptive errors.

Results: Multivariate analyses indicated significant sex and ASD diagnosis interactions that qualified both main effects for overall PPVT and EVT standard scores. On the EVT, no sex difference emerged for those without ASD, whereas females outscored males for those with ASD, F (1, 156) = 5.32, p = .02. On the PPVT, TD males outperformed TD females, whereas ASD females outperformed ASD males, F (1,156) = 7.39, p = .007. The six part of speech error proportion variables were treated as a unit and analyzed together as a syntactical profile. Latent profile analyses yielded a total of 3 profiles: high errors, low errors, and moderate errors. Logistic regression indicated that profile membership significantly predicted ASD diagnosis, B = .413, p = .031. Within the profiles, individuals with ASD were significantly more likely to reside in the profile with moderate errors, as opposed to excessive or few errors.

Conclusions: Results indicate significant sex and diagnostic differences on standardized measures of expressive and receptive language, with support for the social camouflaging hypothesis, such that females with ASD outperformed males on both indices and thus may use language and communication skills to better blend in. Furthermore, individuals on the spectrum do not seem to be making significantly more or fewer errors than their TD counterparts, but are more likely to make consistent, moderate errors on parts of speech both receptively and expressively. These results can aid in ASD assessment, in the assessment of the female ASD phenotype, and in the understanding and assessment of comorbidity of Specific Language Impairment within ASD.