Poster #117 - Word learning in Autism

• In Event: 3-053 - Poster Session 10
  In Poster Session: PS 10 Section - Language, Communication

Integrative Statement

Many Autistic* children show differences or delays in language development (DeMyer et al., 1981; Howlin, 2005; Rutter et al 1967; Tager-Flusberg et al. 2005). Additionally, ASD children often focus attention more narrowly on perceptual information (Mottron & Burack 2001) and exhibit difficulty forming prototypes and categorizing information (Noach, 1974; Minshew et al 1992; Klinger & Dawson, 1995; Plaisted, 2000). The present work suggests the challenges that many ASD children face in learning language may be related to their enhanced attention to low-level perceptual features over higher-level, abstract information. 
Word learning requires children to generalize labels across a wide variety of meanings which share attributes but differ perceptually. In particular, up to 40-84% of commonly used English words are polysemous in that they have multiple meanings that are distinct but related to one another (Durkin & Manning 1989; Rodd, et al 2004; McCarthy 1997). For example, baseball caps, bottle caps, pen caps, and mushroom caps are quite distinct entities, but we recognize that they are meaningfully related, unlike ambiguous labels which have unrelated meanings (e.g., flying bat vs. baseball bat). Recent work has confirmed that typically developing children find it much easier to learn polysemy (cap) than ambiguity (bat). In particular, TD children show a polysemy > ambiguity advantage: they are more accurate at distinguishing multiple target senses of a novel label from foils when the target senses were more related to one another (Floyd & Goldberg 2017).
The work reported here tested 60 TD children (ages 5-12) and 27 ASD children (ages 7-14) on a word learning task that included 8 novel words (4 ambiguous and 4 polysemous). Each word had three meanings which were unrelated (ambiguity) or related to one another (polysemy). Interspersed during exposure were a total of 40 unlabeled foil meanings. At test, children were prompted to select the target meanings for each word from among a set of foils.  
Results (Figure 1) confirm that, while the polysemy > ambiguity advantage replicated for TD children, there was no indication of any such advantage for the ASD individuals. Paired t-tests (preregistered) reveal a significant polysemy advantage just within the TD group (t(120) = 4.1838, p < .001, ASD group: t(54) = -0.85, p = 0.4). To ensure that this wasn’t due to lower overall performance in the ASD group, we matched the 27 ASD with 27 TD children on accuracy in the ambiguity condition using nearest-neighbor matching, and entered the data into a multilevel model with group, condition, and their interaction as main effects with maximal structure (Barr et al 2013), which confirmed that the polysemy advantage is significantly larger for TD kids (β=4.352e-01, t=2.212, p<0.05). The lack of a polysemy advantage in ASD individuals sheds new light on the challenge many ASD individuals face in language learning, quite apart from well-known differences in joint attention and social skill. 

*Here we use the identity-first terminology preferred by those diagnosed (Kenney et al 2015; Brown 2011; https://www.identityfirstautistic.org).

Authors

• Sammy Floyd, Princeton University
  Presenting Author

• Charlotte Jeppsen, Princeton University
  Non-Presenting Author

• Adele E Goldberg, Princeton University
  Non-Presenting Author