Lexical Semantic Impairments in ASD Are Associated with Difficulties in Serial Order Memory
Besides widespread impairments in language pragmatics, ASD is also often associated with varying degrees of structural language impairment that remains poorly understood. Comprehension is typically more compromised than articulation and language semantics tends to pose greater difficulties than syntax and grammar (e.g., Boucher, 2012). Several studies suggest that poor phonological short-term memory (p-STM) contributes to structural language impairments in ASD but this issue remains somewhat controversial (see Williams et al., 2008 for a review). Most studies have approached this topic by comparing groups of ASD and non-ASD children who either do or do not have additional structural language impairments. This approach inevitably raises a number of complicated issues concerning how best to select and match participant groups. A complementary approach is to look at associations between measures of structural language and p-STM in larger samples participants that are not selected for specific inclusion criteria.
To examine the association between p-STM and lexical as well as broader conceptual semantic knowledge in adults and children with ASD.
Study 1 we examined the correlations between the WAIS Digit Span (p-STM), Vocabulary (lexical semantics) and Similarities (conceptual semantics) subtests of 95 ASD and 105 TD adults whose full-scale IQ ranged from 75 to 150. In Study 2, we examined the same correlations in 45 ASD and 26 TD children who also completed a test of non-word repetition (p-STM). Many of the children in experiment 2 had significant language impairments.
Step-wise regression analyses for both samples of participants showed that p-STM, over and above measures of fluid intelligence (PIQ in Exp 1; Raven’s in Study 2), significantly predicted lexical semantic knowledge in ASD (Exp 1: B = .265; Exp 2: B = .349) but not TD groups (Study 1: B = .057; Study 2: B = -.156). In experiment 2, broader conceptual semantic knowledge was also predicted by p-STM over and above fluid intelligence in ASD (B = .308) but not TD participants.
Conclusions: Data from two large samples of adults and children with ASD lend support to the notion that individual differences in p-STM are associated with variability in semantic aspects of language ability, over and above what one might expect on the basis of participants’ fluid intelligence