Concurrent Predictors of Receptive and Expressive Language in Toddlers on the Autism Spectrum

Background: There is a scarcity of information concerning the nature of early language development in children on the autism spectrum. With the exception of a recently published report by Luyster, Kadlec, Carter, and Tager-Flusberg (2008), most of the prior research with larger samples has been based solely on data obtained from a parent report measure of language. The purpose of the present study was to compare language abilities across multiple measures and to characterize predictors of early receptive and expressive language performance in a large sample of children within a restricted developmental age range.

Objectives: This study had two objectives. First, we examined the concurrent validity of three commonly used measures of early language development. Second, we explored predictors of current receptive and expressive language performance in a large sample of toddlers on the autism spectrum.

Methods: Participants comprised more than 100 toddlers (ages 23-39 months) on the autism spectrum. Diagnoses were determined using comprehensive diagnostic evaluations, including the ADI-R and ADOS. Language performance was determined using the McArthur-Bates Communicative Development Inventory: Words & Gestures (CDI), the Vineland-II Adaptive Behavior Scales (VABS-II) Communication subdomain, and the Preschool Language Scale (PLS-4).

Results: Pearson correlation coefficients indicated significant associations between receptive-receptive comparisons (r=.65 to .73, p<.001) and expressive-expressive comparisons (r=.68 to .85, p<.001) on the three different language measures. Stepwise multiple regression analyses were used to explore concurrent predictors of receptive and expressive language. Predictors entered into the regression models included maternal education, nonverbal cognition (as measured by nonverbal items on the Bayley Scales of Infant Development), scores on the Vineland-II Socialization and Fine Motor subdomain scales, and initiation and response to joint attention (as measured during a semi-structured play session with the examiner). For expressive language, the strongest concurrent predictor across all three language measures was the VABS-II Socialization subdomain scale. Some additional significant variance on the PLS-4 and VABS-II Communication subdomain was accounted for by Bayley nonverbal cognition scores and response to joint attention, respectively. For receptive language, scores on the VABS-II Socialization subdomain were also a strong predictor across all three language measures, as was Bayley nonverbal cognition. Less robust significant predictors consisted of response to joint attention and maternal education level. Initiation of joint attention and scores on the VABS-II Fine Motor subdomain were not significant concurrent predictors of receptive or expressive language scores on any of the three language measures.

Conclusions: Parent report measures (CDI and VABS-II Communication) and a clinician-administered test (PLS-4) yielded similar assessments of early language development in toddlers on the autism spectrum. VABS-II Socialization subdomain scores were a pivotal predictor of the ability to understand and use language for young children on the autism spectrum. Bayley nonverbal cognition was primarily implicated in receptive, rather than expressive, language performance. Although response to joint attention played a minor predictive role in receptive (PLS-4) and expressive (VABS-II) language performance, it was not significantly associated with early lexical skills (CDI vocabulary understanding or use), as might have been expected.