Imitation, Joint Attention and Language Development in Autism Spectrum Disorder

Background: Previous research indicates that early joint attention and imitation skills predict structural language development in children with ASD. Pragmatic impairments are universal in ASD; however, little research has examined contributions of joint attention and imitation to pragmatic communication versus structural language. In one study, imitation skills (assessed by parent-reported McArthur Communicative Development Inventory, M-CDI) predicted pragmatic growth in 34 children with ASD from aged 41 months to 54 months, whereas core language (also M-CDI) did not significantly predict pragmatics (Miniscalco et al., 2014). Research over a wider developmental range with specific, direct measures of imitation, joint attention, pragmatics, and structural language is required to test these associations.

Objectives: To investigate predictive contributions of early imitation and joint attention skills to structural language and pragmatic communication skills in a large longitudinal sample of children with ASD.

Methods: Data were analyzed from 421 children in the Pathways in ASD study at 3 points; Time 1 (within 4 months of diagnosis, mean age 39.8 mo), Time 2 (mean age 69.3 mo) and Time 3 (mean age 98.4 mo). At T1, children participated in tests of elicited imitation (Multidimensional Imitation Assessment, MIA), and joint attention (Early Social Communication Scales, response to joint attention; ESCS RJA). At T2, cognitive ability (Wechsler Preschool and Primary Scale of Intelligence, Fourth Edition, WPPSI-IV), structural language (Preschool Language Scale – 4, PLS-4) and pragmatic communication (Children’s Communication Checklist – 2 Social Interaction Deviance Composite, CCC-2 SIDC) were assessed. At T3, children completed the T2 measures and an additional test of pragmatic communication via narratives, the Expression, Reception and Recall of Narrative Instrument (ERRNI).

The ERRNI Initial Ideas score variable revealed a zero-inflated distribution, so zero-inflated negative binomial regression was implemented. In the present study, the assessment of narrative skills requires a level of language ability that some children with ASD will not acquire (resulting in scores of 0 on narrative). Dependent variables with distributions approximating normality (CCC-2 SIDC, PLS-4 Total Language) were analyzed using multiple regression. All analyses were adjusted to control for non-verbal intelligence quotient (NVIQ).

Results:  Decreased imitation at T1 was associated with pragmatic skill deficits at T2 (p=0.008), independent of structural language. T1 RJA was specifically associated with pragmatic discourse skills at T3 (p=0.041). T1 imitation and RJA were not significantly predictive of T2 structural language after controlling for NVIQ.

Conclusions: Although these results are inconsistent with evidence that both elicited imitation and joint attention strongly predict structural language (e.g., Stone & Yoder, 2001), many previous studies examined briefer developmental periods and did not consistently control for nonverbal intelligence. In this study, preschoolers’ imitation and RJA skills were associated with later development of distinct pragmatic skills. These results from a large, well characterized sample of children with ASD add to the literature on predictive roles of early social communicative behavior to later language and pragmatic communication. Additional clarification is needed of the mechanisms underlying the development of structural versus pragmatic language.