Sample characteristics.

Background
Children with autism spectrum disorders (ASD) have been reported to show social-processing deficits in forced-choice social judgment or story interpretation tasks. However, these methods may limit examination of social-processing within a set of acceptable answers. In this pilot study, we propose a novel method predicated on the premise that language carries social information and validate this method to measure social perception in ASD.
Method
20 children with ASD and 20 typically developing (TD) children matched-pairwise on age (5–12 years), gender, and non-verbal IQ, described pictures of people in everyday situations varying on extent of social engagement. Their social language production was examined in high- and low-social picture conditions.
Results
The TD group produced significantly more social language in high-social than low-social picture conditions, with a large effect size (d = 3.15). The TD group produced significantly more social language than the ASD group under high-social conditions (p< .001, η2p = 0.24), but were not significantly different under low-social conditions (p < .05).
Conclusions and implications
The study presents proof-of-concept that expressed language carries social information. The findings indicate that social language may be used to measure social perception and examine differences in ASD, with potential applications for other clinical groups with social-processing challenges.