Joint acquisition of word order and word reference

Abstract

Inferring the mappings between words and their referents is a difficult problem that all language learners face. Similarly, learning which word orders are permitted in one's language is one of the first grammatical learning tasks these same learners must solve. We present a modeling framework which addresses simple versions of both of these problems by using the joint information in each to bootstrap the other. We discover that these two distinct learning tasks may be easier to solve jointly because of the way in which the inferences in one problem constrain the inferences in the other.