A Computational Account of Novel Word Generalization

A key challenge faced by children in vocabulary acquisition is learning which of the many possible meanings is appropriate for a word. The word generalization problem refers to how children associate a word such as dog with a meaning at the appropriate category level in the taxonomy of objects, such as Dalmatians, dogs, or animals. We present extensions to a cross-situational learner that enable the first computational study of word generalization integrated within a word learning model. The model simulates child patterns of word generalization due to the interaction of type and token frequencies in the input data, an influence often observed in usage-based approaches to underlie people’s generalization of linguistic categories.