2018 Sat Poster 6738

Saturday, November 3, 2018 | Poster Session II, Metcalf Small | 3:15pm

Close Correspondence Between Two Measures of Word Learning in Young Children, But No Added Sensitivity for Visual Fixations
S. Creel

A major question in language development is whether word learning improves with age and knowledge, or remains stable. However, word learning studies on infants and young children use multiple paradigms and dependent measures. Even when infant vs. young-child researchers use seemingly-parallel paradigms—repeatedly labeling an object during a learning phase, then presenting two or more objects simultaneously to test recognition—the infant literature (0-2 years) often uses visual fixations as a dependent measure, while the young-child literature (3+ years) often uses pointing (or naming) as a dependent measure.

Surprisingly, we do not know whether, or how closely, these measures correspond. This makes it difficult to assess changes in word learning over development. Further, a common but unverified inference is that visual fixations are more sensitive to implicit knowledge than pointing. To address this gap in our knowledge, the current paper analyzes multiple studies that include both visual fixation and pointing accuracy measures, previously conducted with preschoolers in our lab.

Method. Data from 265 3-to-6-year-olds were included. In studies where a child learned multiple word pairs, that child contributed multiple data points to analyses. In each study, the child learned names of two cartoon characters at a time. Children’s learning was assessed by (1) visual fixations and (2) pointing responses to pictured characters as one character was named (“Where’s the geeb?”).

Analyses. To assess the strength of the relationship between pointing and looking measures, we computed each participant’s mean accuracy and mean visual fixations (target- picture looks minus other-picture looks; chance=0) in a given condition. Fixations were measured during 200-2000 milliseconds (ms) after target word onset, similar to infant and toddler studies (1,2,3). We then used Pointing Accuracy to predict Looks in a linear mixed- effects regression, including by-subjects intercepts. Accuracy was recentered to position chance at 0, allowing us to test the intercept of looks at chance accuracy.

Pointing Accuracy strongly predicted Looks in the regression model (p<.0001), corresponding to a raw correlation of .673 (thick red line, Figure 1). Correlations within conditions (thin red lines) ranged from .194-.846 (median=.688). This close correspondence between looking proportions and pointing accuracy is reassuring in terms of continuity between literatures.

We also tested whether looks are more sensitive to implicit knowledge than pointing. If so, looks should be above chance (target looks > other-picture looks) even when pointing is at chance. Mathematically, the regression intercept should exceed zero. However, the intercept was not significant (estimate=.0095, SE=.0093, ÷2=.06, p=.30). This suggests that looks are not tapping more implicit knowledge than pointing, or that the effect is too small to detect with 265 participants.

Two findings are of central interest. First, visual fixations and pointing responses in word learning tasks are strongly related. Second, looks are no more sensitive than points: when pointing accuracy is at chance, looking proportions are also at chance. Results have implications for understanding developmental improvements in word learning and for integrating across multiple dependent measures. An important open question is whether findings extend to younger children, whose motor responses are less-developed.

References

  1. Swingley, D., & Aslin, R. N. (2000). Spoken word recognition and lexical representation in very young children. Cognition, 76, 147–66.
  2. Swingley, D., & Aslin, R. N. (2002). Lexical neighborhoods and the word-form representations of 14-month-olds. Psychological Science, 13(5), 480–484.
  3. Yoshida, K. A., Fennell, C. T., Swingley, D., & Werker, J. F. (2009). Fourteen-month-old infants learn similar-sounding words. Developmental Science, 12(3), 412–8.