International Meeting for Autism Research (London, May 15-17, 2008): Discourse cohesion in high-functioning autism: A comparison of mentalistic and non-mentalistic bridging inferences

Discourse cohesion in high-functioning autism: A comparison of mentalistic and non-mentalistic bridging inferences

Saturday, May 17, 2008
Champagne Terrace/Bordeaux (Novotel London West)
A. Nadig , School of Communication Sciences and Disorders, McGill University, Montreal, QC, Canada
S. Ozonoff , Psychiatry, University of California at Davis, Sacramento
Background:

When understanding language we use real-world knowledge to make causal, “bridging” inferences about the outcome of an event. Previous work found that individuals with autism display poor text comprehension relative to reading ability (Nationetal.,2006), and have difficulty making bridging inferences (Dennis,Lazenby&Lockyer,2001; Jolliffe&Baron-Cohen,1999; Ozonoff&Miller,1996). These studies did not control for the potentially confounding factor of social inferencing demands in their stimuli.

Objectives:

This experiment investigates whether children with high-functioning autism (HFA) show:

1) General difficulty in integrating an outcome sentence with previous discourse using real-world information

2) Specific or increased difficulty with inferences that require mentalizing

Methods:

8- to 14-year old children with HFA (n=23) and typical development (n=18), matched for language ability, word reading ability and age participated in a computer-presented experiment. They read short event scenarios consisting of a background and outcome sentence. They selected the cause of the event from three choices: 1) a bridging inference which integrated the background and outcome, 2) a local inference derived from the outcome sentence alone, 3) an implausible inference, included as a control. In half of the stimuli inferences depended on mental state attribution, the other half depended on knowledge of physical relationships. Dependent variables were accuracy (selection of the bridging inference) and response time to select an inference.

Results:

Regardless of the type of inference, the HFA group chose bridging inferences significantly less often than the TYP group, and had reliably longer response times to select an inference (adjusted for individual reading speed). Surprisingly, both groups chose bridging inferences more often in the mentalistic than non-mentalistic condition. The selection of bridging inferences was related to level of language ability (τ =.32, p<.01).

Conclusions:

High-functioning children with autism have general difficulty and delays in integrating an outcome sentence with previous discourse to arrive at coherent inferences. This likely contributes to problems in text comprehension.

See more of: Language Posters
See more of: Poster Presentations