Pragmatic Complexity and the Development of Social Cognition in Children's Word Learning and Retention

• In Event: Poster Session 14
  In Poster Session: PS 14 Topic - Language, Communication

Abstract

Previous literature has shown that children can leverage social cognition to learn sound-meaning mappings through pragmatic inference. However, the focus has been on in-the-moment mappings rather than retention (Frank & Goodman, 2014; Gollek & Doherty, 2016; Zosh et. al., 2013). In our prior work on adults, we found novel words learned through more ambiguous and complex processes of pragmatic inferences were better retained than those learned through unambiguous one-to-one mappings. Here, we aim to fill this gap by testing 4-8-year-old children’s learning, immediate recall, and retention of novel words in conditions that require different levels of pragmatic inference, then exploring their relationships with social-cognitive measures and age.

Children (M(age) = 5.9 years, SD(age) = 1.3 years, N = 56) learned eight words during a toy-store tour hosted on an online experimental platform (gorilla.sc). During the learning phase, children heard 4 novel words which could be mapped to one unique novel object on the display – the Unambiguous Condition (UC, Fig. 1A) – and 4 novel words which required pragmatic inference for referential disambiguation – the Ambiguous Condition (AC, Fig. 1B). Novel words appeared twice in each condition along with 4 filler trials of familiar words. After each condition, immediate recall of the novel words was tested in a four-alternative-forced-choice task (Fig. 1C). Children then completed an interactive Theory of Mind (ToM) task (Richardson et. al., 2018) via Zoom, lasting an average of 15 minutes, followed by a second recall task, which tested the retention of novel words (Fig. 1D). The experiment ended with an assessment of children’s executive function (EF) skills (Flanker Task).

Learning rates were significantly above chance in both conditions; however, there was a substantial difference between the two learning conditions, with participants learning UC better than AC. (Fig. 2A). Children performed above chance for UC and AC recall and retention as well, with no significant differences between the conditions (Fig. 2B). For learning, UC was predicted by ToM (beta = 0.14, t = 2.384, p = 0.0211), controlled for age and EF, whereas AC was not correlated to age, ToM, or EF. For retention, AC was marginally predicted by age (beta = 0.07, t = 1.89, p = 0.06). However, the direct effect of age if ToM was removed was no longer significant (p = 0.29), with instead the indirect effect of age on AC retention via ToM having a significant, if modest, effect, explaining 13% of the variance (F(2,53) = 4.11, p = 0.01). There were no significant predictors for retention in UC or for immediate recall in either condition.

Our findings demonstrate that children can successfully learn and retain words learned via pragmatic inference. Children are more accurate in initial sound-meaning mapping that invite an unambiguous association than those that invite complex pragmatic computation, with ToM aiding the former. However, children’s success in the immediate recall and retention between the two word types are not differentiable. Retention for new words acquired through complex pragmatic computation improves from pre-school to second-grade, which is possibly mediated by developing ToM skills.

Authors

• Katherine M. Trice, University of Delaware
  Presenting Author

• Marina Hernandez Santana, Delaware State University
  Non-Presenting Author

• Dionysia Saratsli, University of Delaware
  Non-Presenting Author

• Leah K. Heisler, University of Delaware
  Non-Presenting Author

• Zhenghan Qi, Northeastern University
  Non-Presenting Author