Emerging Language and Social Abilities In ASD: Reciprocal Effects?

Background: Language ability and social competence are inter-related and important predictors of later outcomes in individuals with autism spectrum disorders (ASDs).  Poor social attunement in individuals with ASD may limit their opportunities for language learning; this may in turn further constrain social development.  To our knowledge, the extent to which language and social competence are reciprocally related in young children with ASDs has not been explicitly tested.

Objectives: We aimed to test whether language ability and social competence influenced each other in a reciprocal manner over time in preschoolers with ASD. 

Methods: A cohort of 347 children newly diagnosed with ASD, aged 2-4 years old, provided data at baseline, 6- and 12 months post-enrolment. First, a latent variable measurement model was developed to represent language ability and social competence domains across time points. The Preschool Language Scale -4 (PLS-4), Social Responsiveness Scale (SRS), Early Social Communication Scale (ESCS) and Vineland Adaptive Functioning Scale (VABS) provided potential indicators.  Examination of factor loadings and goodness of fit tests determined the suitability of each latent variable.   A structural model estimated the stability of the language and social competence constructs across time points, as well as the strength of reciprocal cross-lagged associations between domains.     

Results: The sample comprised 292 boys (84.1%) and 55 girls. Mean age at diagnosis was 38.39 months (SD=8.63), and mean age at enrolment was 40.05 months (SD=8.84). Measurement model: Factor loadings for the PLS-4 indicators were very high (ß=0.85-0.9, P < 0.001), indicating that a measured score adequately represented language ability. SRS and the ESCS scores did not share sufficient variance to contribute to a common latent variable. Only the VABS social domains demonstrated measurement invariance between verbal and nonverbal children, therefore these scores were used to represent social competence in both groups combined. Structural model:  The cross-lagged model demonstrated excellent fit to the data (CFI=0.99, TLI=0.97, RMSEA =0.05).  Language ability was highly stable across time points (ß=0.69-0.80, p < 0.001) and social competence was moderately stable (0.59-0.61, p < 0.001).  Small, statistically significant associations (ß=0.17-0.24, p < 0.001) were found from language ability to social competence 6 months later, and vice-versa, across all time points and controlling for cross-sectional associations at baseline.   

Conclusions: Preschoolers newly diagnosed with ASD constitute a heterogeneous group, and measures of social competence may vary in important ways depending on verbal ability. Language and social competence demonstrated reciprocal effects on each other over 12 months in a cohort of preschoolers with ASD.  This may support the  “social orientation theory”, that poor social attunement constrains language learning in ASD.  Conversely, given the substantial heterogeneity in symptoms and functioning in young children with ASD, those with comparatively greater social competence may, as a result, develop superior language ability, which may in turn scaffold further social learning.  Early reciprocal influences between social and language learning may contribute to significantly different developmental trajectories over time.