Conversational Turn-taking in Children with Autism: Deconstructing reciprocity into specific turn-taking behavior

Background: Complex dynamics in peer interactions, e.g. non-systematic turn allocation and quick topic transitions, are difficult for children with autism and hard to target in intervention (Charman & Stone, 2006).  Following  Merrell's (2001) argument that effective intervention requires assessment, and Charman & Stone's (2006) call to merge naturalistic and eliciting approaches in social interventions, we present an assessment that deconstructs the general concept of “lack of reciprocity” into a structured measure of specific turn-taking behaviors with peers and its application to social skills intervention.

Objectives: To identify specific, measurable turn-taking behaviors for assessment-based intervention.

Methods: Five children with high-functioning autism, ages 7-10, matched with a same-age, typically-developing (TD) peer were asked to tell stories using toys. To pinpoint specific behaviors, we analyzed  fourteen aspects of turn-taking that characterize typical conversations (Sacks, Schegloff and Jefferson, 1974). We compared children with autism to their TD partners.

Results: We identified violations more often in children with autism than the TD children affecting the following: (1) speaker change occurrence: turn was not taken unless asked a direct question; (2) gaps in transitions: response follows long pause; (3) discourse continuity: no reciprocity to sustain conversation; (4) turn-constructional units: rigidity in turn length; (5) turn allocation: initiations and responses to peer initiations lacking. Violation categories varied across individuals but suggested behavior patterns to structure intervention procedures.

Conclusions: We extend studies of peer interaction by providing specificity needed for assessing and modeling natural interactions. Results are first steps towards developing comprehensive assessment of social reciprocity and contribute to designing technology-based interventions that model natural interactions. Virtual peers (computer-animated children) that elicit social interaction and model naturally occurring verbal and nonverbal behaviors may address limitations of social interventions with adults (National Research Council, 2001) or trained peers (Lord & Hopkins, 1986).