Joint acquisition of word order and word referent in a memory-limited and incremental learner

Abstract

Word learning in ambiguous contexts is a challenging task which is intertwined with the process of understanding the referential intentions of the speaker. It has been suggested that infant word learning occurs across learning situations and is bootstrapped by syntactic regularities such as word order. Simulation results from ideal learners suggest that it is possible to jointly acquire word order and meanings and that learning is improved as each language capability bootstraps the other. We study the utility of joint acquisition of simple versions of word order and word meaning in early stages of acquisition in a memory-limited incremental model. Comparing learning results in the presence and absence of joint acquisition of word order, word order regularities can quickly converge on real world statistics even using small datasets, relying on imperfect learned interpretations of word meanings, and even given wrong prior biases. Improvement in word learning results in the presence of joint acquisition of word order, however, were limited and only pronounced in the presence of high referential ambiguity and delayed syntactic bootstrapping where word order acquisition was slowed down through the use of priors.