Lexical Processing By Toddlers with ASD

Friday, May 15, 2015: 11:30 AM-1:30 PM
Imperial Ballroom (Grand America Hotel)
S. Ellis-Weismer1, E. K. Haebig2, J. Edwards1, J. Saffran1 and C. Venker3, (1)University of Wisconsin-Madison, Madison, WI, (2)Communication Sciences and Disorders, University of Wisconsin-Madison, Madison, WI, (3)Waisman Center, University of Wisconsin-Madison, Madison, WI

Few studies have examined lexical processing in young children with autism spectrum disorders (ASD) and none have directly tested whether weak central coherence can explain the early comprehension delays of toddlers with ASD. Applied to the context of lexical processing, the weak central coherence account (Happé & Firth, 2006) predicts that toddlers with ASD will focus more closely on lower-level perceptual details (e.g., similarity of visual referents) than their typically developing (TD) peers, potentially at the expense of higher-level, global integration (e.g., semantic relatedness of referents). This study focused on word-object mappings to explore whether lexical processing in toddlers with ASD would be more disrupted, compared to TD peers, by perceptual similarities between pictures while being less disrupted by semantic similarities.


Using an implicit eye-gaze paradigm, this study assessed real-time comprehension (lexical representations) in toddlers with ASD relative to typically developing controls and examined the role of vocabulary size in lexical processing.


Toddlers with and without ASD (TD n=29, ASD n=30) participated in this study; groups were matched on Bayley cognitive raw scores. An experienced clinician made ASD diagnoses by integrating results from the ADOS, ADI-R, and clinical experience. A looking-while-listening task (Fernald et al., 2008) was employed in which each trial presented two pictures on a screen (e.g., sock, dog) with audio describing one of the images (e.g., Where’s the dog?). An adaptation of the Arias-Trejos & Plunkett (2010) task was used in which the two images were unrelated (baseline condition), semantically similar (e.g., hat, boot), or perceptually similar (e.g., crescent moon, banana). Gaze location was hand-coded offline from video.


Results were analyzed using growth curve analysis (Mirman, 2014); the outcome was log odds of looking to target. Two models were constructed; each contained linear, quadratic, and cubic time terms and included participant and participant by condition random effects. The first model revealed that the TD group looked to the target significantly more than the ASD group (Figure 1). A condition effect was found such that lexical processing was significantly better in the baseline condition (unrelated images) than the perceptually similar or semantically similar conditions. There was no significant group by condition interaction.  A second model of the ASD group alone revealed two significant three-way interactions among condition, linear slope, and vocabulary size. As vocabulary size decreased, children’s processing speed (i.e., linear slope) was more disrupted by perceptually-similar distracters as compared to the other conditions (visualized as a median split in vocabulary level in Figure 2).


Although toddlers with ASD performed more poorly than the cognition-matched TD toddlers, lexical processing in both groups was affected by the semantic and perceptual similarity of distracter images. Toddlers with ASD who had smaller receptive vocabularies were more disrupted by distracter images that looked similar to the target than images that were semantically related or unrelated. These results provide partial support for weak central coherence but suggest that how toddlers process lexical information is dependent upon vocabulary size.