Identifying Childhood Expressive Language Features That Best Predict Adult Language and Communication Outcome in Individuals With Autism Spectrum Disorder

Purpose: Several studies have reported that useful speech at 5 years of age predicts outcomes in individuals with autism spectrum disorder (ASD), but this skill has been vaguely defined. This study investigates which specific aspects of expressive language in children with ASD best predict adult language and communication outcomes. Method: Language samples from 29 children (ages 47- 72 months) enrolled in a longitudinal project (e.g., Lord et al., 2006) were transcribed and coded for spoken language features. Hierarchical linear regression was used to compare the following childhood variables as predictors of adult language and communication outcomes: noun diversity, verb diversity, mean length of utterance, and proportion of utterances that were socially motivated. Results: Childhood verb diversity was a value-added predictor of all four adult outcome measures (i.e., verbal IQ,Autism Diagnostic Observation Schedule Communication + Social Interaction Algorithm totals, Peabody Picture Vocabulary Test scores, and Vineland Adaptive Behavior Scales Communication Domain scores), while noun diversity and proportion of utterances that were socially motivated were not value-added predictors of any adult outcome measures. In a second set of regression analyses, mean length of utterance was substituted for verb diversity and was a value-added predictor of two out of four adult outcome measures (i.e., verbal IQ and Vineland Adaptive Behavior Scales Communication Domain scores). The pattern of findings for the other predictors remained the same as in the previous analyses. Conclusion: These results have implications for our understanding of early language in ASD and for clinical decision making in early childhood.

Automated summary

This study found that verb diversity in children with ASD is the most predictive factor for adult language and communication outcomes, while noun diversity and proportion of utterances that were socially motivated are not predictive. Mean length of utterance can also predict adult outcomes in some cases.