Proceedings of the Annual Meeting of the Cognitive Science Society

Word learning paradigms often teach children the name of a novel object and then immediately ask them to generalize thelabel to another object. This study uses a new paradigm that affords the ability to determine how childrens generalizationchanges over time. Participants (N=22, Mage=3.8 years) saw a novel object labeled by the experimenter (e.g., wug) andthen were shown five novel objects that each had an additional feature changed from the exemplar (i.e., the fifth objecthad five changed features), either immediately after the exemplar or after a five minute delay. Category membershipendorsement of the five test objects was higher at immediate test than delayed test, suggesting that children representnovel categories broadly at first but more narrowly over time. We propose that childrens forgetting of exemplars acrosstime leads to shifts in childrens generalization; as children forget exemplar features, category membership becomes morespecific.