The Role of Perspective Taking in Third-Party Word Learning

• In Event: 1-099 - Factors that promote third-party learning in young children

Integrative Statement

In the lab, children as young as 18 months are able to robustly learn new words through third-party interactions (e.g. Akhtar 2005; Floor & Akhtar, 2006). One hypothesized mechanism for this type of learning is perspective taking: children who are third-party to a teaching interaction take on the perspective of a learner (who is actually involved in the interaction) as though it is their own (Moore, 2007; Herold & Akhtar, 2008). In this study, we tested this hypothesis by examining toddlers’ abilities to learn novel nouns from two types of third-party interactions: one in which both the learner and child had visual access to the scene, and one in which the child did, but the learner did not. If toddlers take the perspective of the learner in third-party interactions, they should learn in the former, but not the latter condition.

Twenty toddlers (21;00-34;00; 10 males) participated in the study. A teaching interlocutor (E1) sat across from a learning interlocutor (E2), in full view of the child. E1 opened each of three opaque boxes in turn to reveal a novel object. Nontarget objects were commented on (e.g. What a cool thing!), while the target object was labeled (e.g. What a cool bosa!). In the Attended condition, E2 visually attended to E1, smiled and nodded. In the Distracted condition, E2 was distracted, taking notes on a clipboard, and did not look at E1 at all. After the teaching round, the child was then given the three objects to play with, and then the teaching round was repeated. After two teaching rounds, the three objects were placed on a tray and the child was given an identification prompt (Get the bosa!) and a preference prompt (Get the one you like!). Toddlers participated in one trial in each condition; the order of conditions was counterbalanced.

In order to determine if children learned the novel words, binomial tests were conducted on the number of targets selected on the identification prompt in each condition. In Attended trials, 12 participants correctly selected the target, which is significantly greater than chance (p = .012). In Distracted trials, only 7 participants selected the target, which was not significantly greater than chance (p = .5). In order to ensure that these selections were based on mapping the novel label and not incidental preference for the target, we repeated these analyses on the preference prompts. Children did not prefer the target above chance in either the Attended (p = .33) or Distracted (p = .98) trials.

Our findings are consistent with the perspective-taking hypothesis of third-party learning. Toddlers were unable to learn the novel words when the learning interlocutor did not have visual access to the scene, even though they themselves did. This implies that the set of third-party interactions from which children can learn may be limited to those in which the learner’s experience is sufficient for word learning. This may include joint attention (absent in the Distracted condition), and social contingency (e.g. O’Doherty et al., 2011), among others.

Authors

• Allison Fitch, Boston University
  Presenting Author

• Kendra Hennegen, Boston University
  Non-Presenting Author

• Amy Lieberman, Boston University
  Non-Presenting Author

• Rhiannon Luyster, Emerson College
  Non-Presenting Author

• Sudha Arunachalam, New York University
  Non-Presenting Author