Investigating the mechanisms of verb bias learning

Abstract

A verb bias refers to a higher likelihood for a verb to appear in one particular sentence structure. This co-occurrence is well-established as a guiding statistic in the language comprehension of children and adults, but how it emerges through language acquisition is a relatively new area of study. Previous work is ambiguous as to what learning mechanism likely handles this process – one possibility is that learning is incremental and associative, and another possibility is that verb biases are learned through explicit learning. The aim of this thesis is to discriminate between these two possibilities using reversal learning, a paradigm that compares initial learning of a rule to learning of its exact opposite. In this paradigm, more rapid learning of the opposite rule (a “fast reversal”) indicates explicit learning, while slower learning of the opposite rule (a “slow reversal”) indicates incremental learning. In the first experiment, new verb biases are created for familiar verbs using an adapted comprehension-to-production priming technique, and the type of learning is evaluated using reversal learning. In the second experiment, this experiment is conceptually replicated using an adapted production-to-production priming technique. The results of these experiments expand our understanding of what types of learning mechanisms underlie the acquisition of a new verb bias.