Social Learning in School-Aged Children: The Effect of Social Contexts on Word Learning and Retention

• In Event: Observing, Conforming, and Internalizing: How Does Social Learning Occur Across Different Contexts and Cultures?

Abstract

Young children learn most words through direct social interaction (Baldwin, 1999). Yet, as children grow older, the proportion of decontextualized language inputs through reading, lecturing, and overhearing increases (Rowe, 2013). While recent studies found that children can learn novel words from overhearing speech (Boderé & Jaspaert, 2017), the focus is limited in initial meaning mapping. It remains unknown how sustaining memories are for words learned in the absence of more direct social cues. We hypothesize that word learning and retention in school-aged children might benefit less from direct social cues as they continue to grow their language skills.

Thirty-six American children (Mage = 7.52, 5~10 years) watched 20 story-telling videos narrated by a female native speaker of English. In the direct-addressed condition, the narrator faced towards the camera and addressed the participating child, with another silent listener present in the video (Figure 1a). In the overhearing condition, the narrator faced away from the camera while telling the stories towards the listener in the video (Figure 1b). A made-up word was embedded and repeated three times in each story, representing concrete meaning of a real English noun or verb with high cloze probability. Attainment and retention were measured with a four alternative-forced-choice word-picture matching task at two time points, same day after learning and one week later. Children’s concurrent vocabulary scores and social cognitive skills were measured by the Picture Vocabulary task in NIH toolbox and Mind in the Eyes task (van der Meulen et al., 2017).

Children performed significantly above chance-level across conditions and time points (p’s < .001). At the same-day test, children showed higher accuracies in the overhearing than the direct-addressed condition (b = 0.50, z = 2.12, p = .03; Figure 2a). The condition difference was more prominent in older children (Mage = 8.91) than in younger children (Mage = 6.13; significant interaction between social context and age: b = 0.46, z = 2.24, p = .02; Figure 2b). Compared to the same-day test, children showed significantly lower accuracy in the one-week retention across social contexts (b = -0.47, z = -3.17, p = .002). A marginally significant interaction was further found between social contexts and test times (b = -0.53, t = -1.85, p = .06), reflected as a sharper decay of accuracy in the overhearing than the direct-addressed condition (Figure 2a). Finally, children’s concurrent vocabulary, but not social cognition, was found to mediate the relationship between word learning outcomes and age across test times (p’s < .001).
These results replicated and extended previous findings by demonstrating successful word learning and retention from overhearing context in school-aged children. Importantly, children’s ability to learn words across social contexts depends on existing vocabulary skills instead of social cognitive skills. Older children, but not younger children, were found to optimize on learning from indirect social contexts and even showed an advantage in same-day learning from overhearing contexts compared to direct-addressed contexts. However, words learned in the presence of direct social cues were better retained than words learned from overhearing.

Authors

• Anqi Hu, University of Delaware
  Presenting Author

• Zhenghan Qi, Northeastern University