Cognitive Orientation and Linguistic Diversity in the Narratives and Retellings of Children with ASD Vs. Controls

Background: Narrative storytelling can be challenging for individuals with autism spectrum disorder (ASD) because it is essentially a social-communicative task. To successfully tell a story from pictures, one must be able to take the perspective of characters and communicate their appearances, actions, motives and emotions to another person. Prior research shows that individuals with ASD use fewer mental state terms than typical controls or children with Down syndrome (Tager-Flusberg, 1992), and tell less cohesive stories than children without ASD (King et al., 2014). In addition, many children with ASD have verbal working memory deficits (Williams, Goldstein, & Minshew, 2006), making it easier to describe a story from pictures than to retell the same story from memory.

Objectives: Use a computational approach to assess cartoon story narrations in participants with ASD and matched typical controls (TD), and compare retellings by group.

Methods: Thirty-three school-aged children with ASD (N=17, 3 girls) or typical development (N=16, 9 girls) participated in the present study. Groups did not differ significantly on age (11y) or IQ (105; 2 missing from the ASD group) but there were more girls in the TD group (imbalance will be corrected by May, 2018). Audio recordings of the Cartoon task of the Autism Diagnostic Observation Schedule-2^nd Edition (Fig.1) were orthographically transcribed. Four variables were assessed: duration of narratives, proportion of nouns produced relative to total words (concrete orientation), proportion of words about cognitive processes (cognitive orientation), and Shannon entropy (a measure of linguistic diversity; low diversity captures predictable speech that may be repetitive or otherwise restricted in range).

Results: Linear mixed effects models with participant ID as a random effect revealed that children in both groups produced equally long initial narratives (TD M=39s, ASD M=35s, t=-.79, p=.44). The ASD group demonstrated greater concrete orientation, as evidenced by proportionately more nouns (t=2.61, p=.01), and diminished cognitive orientation, as evidenced by fewer words about the inner processes of characters, than the TD group (t=-2.18, p=.07; Fig.1). In addition, participants with ASD produced narratives that were less linguistically diverse than participants in the TD group (t=-1.66, p=.107; Fig.2). In the retelling (without a picture), children with ASD produced narratives that were significantly shorter than their own first telling (-29%; M=25s; t=-3.40, p=.004), while children in the TD group produced retellings that only trended toward being shorter (-15%; M=33s; t=-1.78, p=.10). Retellings in both groups indicated similar concrete orientation (ps=n.s.). However, the ASD group continued to demonstrate diminished cognitive orientation (t=-2.33, p=.03) and reduced linguistic diversity (t=-2.15, p=.04; Fig.2) as compared to the TD group.

Conclusions: Our results replicate and extend prior research showing that children with ASD produce and remember narratives differently than typically developing children. In a short cartoon narrative and retelling, we found consistent indicators of diminished cognitive orientation and linguistic diversity in the ASD group, as well as evidence of working memory deficits. The next steps in this research are to double our sample size (transcription underway) and explore whether cognitive orientation during narratives relates to ASD symptomology.