Lexical Characteristics Account for Vocabulary Size in Toddlers with ASD: A Comparison of Comprehension and Production

Background: Despite wide variability, vocabulary is a domain of challenge for many children with autism spectrum disorder (ASD), with particular weakness in comprehension (Hudry et al., 2010).  Recent research suggests that characteristics of the words contained in children’s lexicons may illuminate vocabulary acquisition processes.  That is, lexical characteristics (i.e., phonological neighborhood density, word frequency, word length) may explain differences between comprehension and production, as well as differences in vocabulary size between successful language-learners and individuals with language-learning problems (Gray et al., 2014; MacRoy-Higgins et al., 2013).  For example, toddlers with smaller vocabularies produce words with higher phonological neighborhood density (i.e., words that sound like many others), presumably reducing processing demands by taking advantage of the phonological distribution of language input (Stokes, 2014).  Lexical characteristics predict expressive vocabulary size for toddlers with ASD (Kover & Ellis Weismer, 2014), but a comparison of their role for comprehension and production is necessary to inform explanations of acquisition for each.

Objectives: This study examined the effects of lexical characteristics in relation to comprehension and production in toddlers with ASD. We asked: (1) Do neighborhood density, word frequency, and word length differ between receptive and expressive vocabulary for toddlers with ASD?, and (2) Do lexical characteristics differ for those with smaller versus larger vocabularies? 

Methods: Toddlers with confirmed clinical best-estimate diagnoses, including the ADOS and ADI-R (LeCouteur et al., 2006; Lord et al., 2000), were drawn from a larger study (N=129).  Parents completed the MacArthur-Bates Communicative Development Inventories Words and Gestures form, yielding receptive and expressive vocabulary size. 

Lexical characteristics were coded for each toddler’s lexicon. Using an online calculator (Storkel & Hoover, 2010), neighborhood density was the count of words differing by a single sound; frequency was the log-base-10 of raw frequency; length was number of phonemes.  Toddlers with ≥20 coded receptive and expressive words, following Stokes (2014), were included (n=26; ages 24-38 months).  Matched-samples t-tests compared comprehension and production.  Independent-samples t-tests compared lexical characteristics for toddlers with vocabularies smaller or larger than the sample medians. 

Results: Expressive vocabularies were composed of words with higher neighborhood density, t(25)=-4.37, p<.001, d=.86, and shorter length, t(25)=6.04, p<.001, d=1.21, than receptive vocabularies (Table 1).  Only for production did toddlers with smaller vocabularies have higher neighborhood density, t(24)=-2.90, p=.008, d=.85, higher word frequency, t(24)=-2.49, p=.020, d=.81, and shorter word length, t(24)=4.39, p<.001, d=1.35, relative to those with larger vocabularies (Table 2). 

Conclusions: These findings have implications for theories of lexical acquisition in ASD, including the role of phonological processing.  As in other populations with language impairment (Stokes, 2014; MacRoy-Higgins et al., 2013), toddlers with ASD may show effects of lexical characteristics that are most pronounced for production.  Those with smaller vocabularies first produce words with reduced processing demands (e.g., short, many phonological neighbors).  Given recent evidence of strong phonological processing of novel words for some with ASD (Norbury et al., 2010; Henderson et al., 2014), this research sets the stage for experimental tasks that address whether lexical characteristics might be harnessed to support both comprehension and production.