Understanding Intentions Predicts Relational Vocabulary In Preschoolers with ASD

Background: The ability to infer the intentions underlying people’s behavior helps children learn language, especially when words label ambiguous referents. For example, when a word labels a dynamic referent (e.g., run), inferences about what the speaker intends to label narrows the list of possible meanings for the word (Behrend & Scofield, 2006). Theoretically, social understanding should predict relational vocabulary in typical children, but lack of variability in social skills in the typical population has made this connection difficult to pinpoint. Children with ASD, by definition, have variable social skills, and recent research suggests that children with ASD have special difficulty learning at least one type of relational word: verbs (Lopez & Lord, 2009). The linguistic and social variability of this group provides an opportunity to probe the connection between intention understanding and relational language more generally.

Objectives: Explore the link between understanding the social intentions of another person, and relational vocabulary size. 

Methods: Thirteen children with ASD (9 boys, mean age 5:4) were matched to 13 control children (12 boys, mean age 4:7) on non-verbal cognitive ability, and were administered the Test of Relational Concepts (a measure of receptive relational vocabulary). Children participated in a modified version of a classic behavioral reenactment task (Meltzoff, 1995), which consisted of two conditions. In the canonical condition, the experimenter repeatedly tried (and failed) to perform actions that were in line with the toys’ affordances (e.g., stack a ring on a post). In the non-canonical condition, the experimenter’s intended actions conflicted with canonical toy use (e.g., put a train in a pot instead of on the available track). Children were asked, “Can you do it for me?” The first action children performed on each set of objects was given a score of 1 if it completed the experimenter’s intention, and a score of 0 if it did not. It was hypothesized that children with ASD would be able to complete the failed intention in the canonical condition only, and that the rate of intentional responses in the ASD group would correlate with relational vocabulary as measured by the TRC.

Results: A repeated-measures ANOVA revealed a condition by group interaction, F=3.75, p=.065. Planned independent-samples t-tests revealed that the ASD group completed the experimenter’s intention as often as the TYP group in the canonical condition (p>.10) but were less consistent in the non-canonical condition, t(24)=2.02, p=.06. The sample was split by group to test the hypothesis that social intentional understanding predicts relational vocabulary size. A positive correlation between the rate of completed intentions and scores on the TRC emerged in the ASD group, r(13)=.56, p<.05, but not the TYP group, r(13)=.24, p=n.s.

Conclusions: The present results suggest a relationship between social intention understanding and relational language development, and shed light on what could be the driving force behind reported verb differences in children with PDD-NOS vs. autism (Lopez & Lord, 2009). Future research could clarify the causal direction of this effect by testing whether interventions that improve social intention understanding also increase relational word vocabulary.