Mutual Exclusivity in Young Children with ASD

Friday, May 15, 2015: 11:30 AM-1:30 PM
Imperial Ballroom (Grand America Hotel)
C. E. Venker1, M. B. Winn1, S. Ellis-Weismer2, J. Saffran3 and J. Edwards4, (1)Waisman Center, University of Wisconsin-Madison, Madison, WI, (2)University of Wisconsin-Madison, Middleton, WI, (3)University of Wisconsin-Madison, Madison, WI, (4)Communication Sciences and Disorders, University of Wisconsin, Madison, WI

Typically developing children assume that a novel label describes an unfamiliar object, as opposed to a familiar object with a known name—a phenomenon called mutual exclusivity (Markman, 1990). Mapping novel words to novel objects may help children learn new words, suggesting that deficits in mutual exclusivity could be related to the early vocabulary delays experienced by many children with ASD. Although there is evidence that older children with ASD show mutual exclusivity (de Marchena et al., 2011; Priessler & Carey, 2005), this question has not been investigated in young children with ASD.


  1. To determine whether young children with ASD demonstrate mutual exclusivity by attending to a novel image when they hear a novel word.
  2. To compare how quickly and accurately these children process novel versus familiar words.


Participants were 18 children with ASD between 26 and 36 months old (M = 31, SD = 3); 10 additional children participated but were excluded due to excessive missing data. Bayley-III composite scores ranged from 55 to 105 (M = 86, SD = 13), and parent-reported receptive vocabulary ranged from 20 to 316 words (M = 165, SD = 87).

Children completed an eye-tracking task on a Tobii T60-XL. Each trial simultaneously showed one familiar image (e.g., cat) and one novel image (e.g., wombat). The real-word (RW) condition presented a familiar label (Find the cat), and the novel-word (NW) condition presented a novel label (See the vafe?). Trials were eliminated if children were not reported to understand the familiar word. Looks at each time point between 200ms and 1300ms after noun onset were categorized as looks to target or distracter.


A binomial logistic mixed-effects model was constructed with time and condition (RW vs. NW) as the independent variables and looks (target vs. distracter) as the dependent variable. The model included participant*condition random effects for intercept and slope.

In the NW condition, there was a significant non-zero slope; although children showed a strong baseline preference to look at the familiar image, they significantly increased their looking to the novel image after hearing the novel word. Even 1300ms after noun onset, however, children spent approximately the same amount of time looking at the familiar and novel image in the NW condition.

Accuracy was considerably lower in the NW than the RW condition; children looked significantly more to the target in the RW than the NW condition. There was also a time*condition interaction, which reflected a significantly greater slope in the NW than the RW condition.


Young children with ASD increased their looks to a novel image after hearing a novel word, suggesting that they may use mutual exclusivity to determine the referents of new words. Nevertheless, they continued to look at the familiar object about half of the time. This demonstrates the impact of baseline visual preferences on attention allocation in a language-based task. Additional research is needed to investigate the relationships among visual attention, language processing, and language learning in children with ASD.