Speaker
Description
When learning novel words, referential ambiguity is a constant part of children’s learning environment. However, to date it is unclear whether children are aware of the different levels of uncertainty in referent identification, and to what extent they can effectively use this information. In two pre-registered studies, we assessed preschoolers’ and adults’ ability to monitor their uncertainty during referential ambiguity and update labels dependent on their initial learning context.
In study 1, we asked 4-5-year-olds (n=82) and adults (n=70) to find the referents of novel words in contexts with maximal ambiguity (two novel objects), medium ambiguity (one novel and one known object, resolvable via mutual exclusivity), or minimal ambiguity (one novel object). We measured explicit assessments of their own uncertainty, as well as other uncertainty and information seeking behaviors. Afterwards, participants faced counterevidence to their initial word-object-mappings, and chose which mappings (learned in more vs. less ambiguous contexts) to update. Results show that children’s (explicit and implicit) uncertainty and their information seeking increased with referential ambiguity. Further, while high uncertainty during initial word-object-mappings increased adults’ willingness to update, this was not the case for children.
Study 2 thus investigates potential improvements in preschoolers’ (n=90) and adults’ (n=90) updating performance in a task with greater contrast between initial learning contexts. Adults almost exclusively updated the most ambiguously learned word-object-mappings; data collection for children is ongoing. The complete pattern of results will inform discussions about the mechanisms and mental processes that enable children to learn words in the face of ambiguity.
