Using Syntax to Learn Verbs: Comparing Error Patterns in Preschoolers with and without Language Impairment

• In Event: 2-093 - Beyond Vocabulary Size: New Considerations for Receptive Vocabulary Assessment in the Preschool Years

Integrative Statement

To successfully acquire – and appropriately extend – a verb, children must recognize that the action can be performed by different actors and may look perceptually different in each instance (e.g., dog versus marathoner running). The argument structure of a sentence signals relations among participants within an event – primarily the roles of ‘who does what and with whom’ (the agent, action, and patient). Verbs appear in different sentence frames, and these morphosyntactic cues help children fix the meaning of a verb. Children with language impairments (LI) have difficulty with acquisition and use of morphosyntactic forms, making this population well suited for investigating the role of syntax in children’s comprehension of novel verbs.

We investigate how preschool children with and without LI infer word meaning, utilizing the syntactic information provided by the sentences in which the words appear. Verb learning data were collected during the development of a touchscreen-based language assessment, the Quick Interactive Language Screener (QUILS). Participants were 674 typically developing (TD) children (M = 52.7 months; SD = 7.9); and 19 children with LI (M = 56.1 months; SD = 8.2). Children viewed dynamic scenes and were asked to infer the meaning of a novel verb used to describe the scene. Cues provided by the sentences allowed the children to select among unknown intransitive, transitive and dative verbs (Figure 1).

For TD children, a one-way ANOVA showed a main effect of Age on Verb Learning Accuracy, F(2, 671) = 45.48, p < .001. For LI children, Verb Learning correlated with overall language ability as measured by a standardized assessment (PLS-5), r(15) =.521, p = .032, but not with age. Performance on Verb Learning items correlated with QUILS receptive grammar scores in TD, r(671) = .434, p < .001; and LI groups, r(16) = .5, p = .035.

Figure 2 shows the average proportion accuracy and errors. Hypothesis tests for one proportion showed that TD 3-year-olds selected the target action and the perceptual foil (i.e., identical actors) at levels that were not significantly different from chance, but were less likely to select the syntactic foil than would be expected by chance, (z = 2.126, p = .0335, [CI: 20.13 to 32.59]). Four and 5-year-olds in the TD group were significantly more likely to select the target; and both foils were selected at levels significantly less than chance. In the LI group, children’s responses across all Verb Learning items were not significantly different from chance (all Ps > .05).

Results suggest children rely on perceptual similarities across verb learning instances (e.g., similar agents) to make initial verb extensions (Gentner, Loewenstein, & Hung, 2007). By 4, TD children understand this strategy is insufficient to support flexible and broad extension; they learn to overcome this initial bias and extend a novel verb to a perceptually dissimilar event. Persistence in the immature strategy may reflect impaired processing. Assessing patterns of verb acquisition may be more informative than measuring the size of a child’s verb vocabulary alone and may contribute to the early identification of language delays.

Authors

• Amy Pace, University of Washington
  Non-Presenting Author

• Jill de Villiers, Smith College
  Non-Presenting Author

• Aquiles Iglesias, University of Delaware
  Non-Presenting Author

• Mary Sweig Wilson, University of Vermont & Laureate Learning Systems
  Non-Presenting Author

• Frances Bueno, University of Washington
  Presenting Author