Conversation in High-Functioning Autism: Do Linguistic and Pragmatic Features Vary Depending on the Topic Being Discussed?

Background: Individuals with high-functioning autism, despite performing well on standardized language tests, have considerable difficulties with social communication. Previous research found that young children with autism (Tager-Flusberg&Anderson,1991) and school-aged children with autism (Capps,Kehres,&Sigman,1998) produced fewer responses that were relevant to and contingent upon their conversational partners’ utterances, than did matched children with developmental delays, reflecting problems with topic maintenance and discourse coherence. 

Objectives: Conversational interaction is an understudied topic that has significant import for daily functioning. This study adds a novel dimension to the empirical research in this area: do linguistic and pragmatic features of conversation change depending on the topic of discussion? We compared conversations on generic topics (siblings, pets) with those on motivated topics (circumscribed interest or favourite hobby). Motivated topics may improve topic maintenance by increasing participants’ level of engagement. Alternatively, they could engender more atypical language use and monologue-like discourse, given the isolating nature of circumscribed interests.   

Methods: Participants were 21 children with high-functioning autism (HFA) and 17 typically-developing comparisons matched on age (9 to 13 years), language level, and Performance IQ.  They participated in brief, semi-structured conversations with an adult partner, on both a generic topic and a motivated topic. Verbal exchanges were transcribed from video and coded using a scheme that quantified numerous linguistic and pragmatic aspects of the exchange. Reliability was established on the coding scheme between two coders blind to group membership. We predicted the HFA group would display problems with topic maintenance (e.g. fewer contingent responses, more self-contingent elaborations), problems providing an appropriate amount of information (rather than over- or under-informative statements), more atypical utterances (e.g. scripted or pedantic speech, unusual word choice), and less use of mental state terms than the comparison group. No specific predictions were made about the effect of topic.   

Results:   The overall number of participant utterances, adult utterances, and proportion of participant to adult utterances, did not differ between groups. For the generic topic, both groups produced a similar amount of contingent responses. However, the HFA group produced significantly more grammatical errors (p<.05) than comparisons. For the motivated topic, the HFA group provided significantly fewer contingent responses than comparisons (p<.05). Those responses that were contingent still differed from those of the comparison group, in that they expressed inappropriate levels of information (p=.01). For both topics, the comparison group generally provided elaborations that were contingent on the topic their partner had introduced, whereas the HFA group provided elaborations that were self-contingent on their own prior utterances (p<.05); this tendency was stronger for the motivated than the generic topic.  In addition, atypical utterances were observed significantly more often in the HFA group, whereas mental state terms were observed significantly less often than for comparison participants.   

Conclusions: Some communicative differences (more atypical utterances, fewer mental state terms) were observed across conversational topics, while motivated topics specifically led to decreased topic maintenance and more monologue-style speech by the HFA group (as observed by fewer contingent responses, inappropriate levels of information in contingent responses, and heightened use of self-contingent elaborations).