Syntactic Comprehension in Boys with Autism Spectrum Disorders: Evidence From Specific Constructions

Background:  Previous research suggests that receptive language and syntactic ability might be particularly impaired in children with autism spectrum disorders (ASD; Eigsti et al. 2007, Ellis Weismer et al., 2010).  However, most studies have relied on broad summary measures that fail to distinguish specific language forms, making it difficult to speculate about mechanisms of language learning that might be impaired.  

Objectives:  We examined comprehension of three specific syntactic constructions and analyzed the nature of the errors that occurred.  Based on evidence that reversible forms might be uniquely challenging (Oakes et al., 2011), we assessed reversible locatives and reversible subject-verb-object (SVO), as well as simple sentences, the latter serving as a baseline.  We distinguished between purely syntactic errors of the word order type and errors that included semantic components.  Through a comparison of boys with ASD to typically developing boys, we sought to establish the extent of delay in comprehension of these constructions and the processing mechanisms that underlie that delay.  

Methods:  Boys with ASD (n = 45; ages 4 – 10 years) had ADOS severity scores that ranged from 4 to 10 (M = 7.89, SD = 1.64) and nonverbal Leiter-R (Roid & Miller, 1997) Brief IQ scores from 40 to 117 (M = 77.62, SD = 19.51).  Typically developing boys (n = 57, ages 2 – 5 years) did not differ on Peabody Picture Vocabulary Test-4 (Dunn & Dunn, 2007) growth scores from the boys with ASD, t(100) = .48, p = .634.  Boys with ASD had higher Leiter-R growth scores, t(100) = 3.98, p < .001, which were statistically controlled in all analyses. Comprehension was assessed with the Test for Reception of Grammar-2 (Bishop, 2003), in which each construction is tested with four items.  

Results:  For each of the three constructions of interest, we examined group, receptive vocabulary, and nonverbal cognition as predictors of comprehension (i.e., items correct).  For simple sentences, the effect of group was not significant.  For locatives and SVO, boys with ASD scored lower than typically developing boys, ps < .049, one-tailed.  Receptive vocabulary was a positive predictor in each regression and nonverbal cognition predicted simple and locative sentence comprehension.  For locative and SVO sentences, boys with ASD committed more semantic errors, F(1, 98) = 4.91, p = .029, but not more word order errors, F(1,98) = .78, p = .379, than the typically developing boys, controlling for receptive vocabulary and nonverbal cognition. 

Conclusions:  Boys with ASD demonstrated specific weaknesses in comprehension of reversible constructions, including locatives and SVO, after controlling for receptive vocabulary and nonverbal cognition.  These results are in line with previous research that suggests that children with ASD might employ immature strategies for comprehension of syntactically demanding constructions (Tager-Flusberg, 1981).  The failure to capitalize on semantic information to support syntactic comprehension, resulting in high rates of semantic errors that were avoided by typically developing children, suggests that semantic impairments contribute to syntactic processing difficulties, a conclusion also supported by the regression analyses.