Early Predictors of Later Language Abilities in High-Risk Infant Siblings of Children with Autism

Background: Younger siblings of children with autism spectrum disorder (ASD) are at increased risk of developing ASD themselves; those who do not have ASD often show related language and communication delays. Impairments in language are often preceded by deficits in nonverbal communication, such as gesture use and pretend play, which suggests that they may be pre-linguistic indicators of later language outcomes. Indeed, both gestures and play have been shown to predict later language in young children with and without ASD, but less work has examined this relationship in high-risk siblings, even though they, like children with ASD, exhibit poorer gestures and play compared to low-risk children.

Objectives: To determine whether gestures and play at 18 months predict language proficiency at 36 months in high-risk siblings with and without ASD, and in low-risk controls.

Methods: Our sample included 105 high-risk siblings diagnosed with ASD at age 36 months, 313 high-risk siblings not diagnosed with ASD (non-ASD), and 164 low-risk controls, drawn from the Canadian Infant Sibling study. At 18 months, gestures and play were assessed using the ADOS-Generic (ADOS-G), Module 1, and at 36 months, language abilities were measured using the Mullen Scales of Early Learning (MSEL). A multi-group structural equation modeling (SEM) approach was used, as it allowed us to capture shared variance between similar variables. Our model involved three latent variables representing our constructs of interest: “gestures”, which comprised “Pointing” and “Gestures” from the ADOS-G; “play”, which included “Functional Play with Objects” and “Imagination/Creativity” from the ADOS-G; and “later language”, which included Receptive and Expressive Language scales of the MSEL. Estimated parameters included the regression of later language on gestures and play, and the correlation between the latter two.

Results: SEM revealed that the relationship between gestures, play, and later language differed between groups. In controls, gestures (β=-0.342, p=0.005), but not play (β=-0.110, p=0.305) predicted later language, and gestures were not correlated with play (r=0.162, p=0.269). For non-ASD high-risk siblings, gestures (β=-0.486, p=0.001), but not play (β=-0.041, p=0.758), predicted later language, but the two were significantly correlated (r=0.573, p<0.001). In the ASD group, neither gestures (β=-1.937, p=0.151) nor play (β=1.309, p=0.331) predicted later language, but this apparent lack of significance could be explained by high collinearity (r=0.952, p<0.001). In fact, when gestures and play are combined into one latent variable, it significantly predicts later language (β=-1.104, p<0.001).

Conclusions: This study provides evidence that gestures significantly predict later language in high-risk siblings not diagnosed with ASD. Play did not predict language outcome in our sample, although other aspects of play not captured here may yield different results. Our findings complement previous work by demonstrating that not only do specific measures of gestures predict later language, but so does the underlying construct representing gesture use. This study emphasizes the importance of attending to early indicators of language, especially gestures, in high-risk siblings, as they may help identify children who could benefit from early language interventions.