In this paper three problems for a connectionist account of language are considered: 1. What is the nature of linguistic representations? 2. How can complex structural relationships such as a constituent structure be represented? 3. How can the apparently open-ended nature of langauge be accomadated by a fixed-resource system? Using a prediction task, a simple recurrent network (SRN) is trained on multiclausal sentences which contain multiply-embedded relative clauses. Principal compenent analysis of the hidden unit activation patterns reveals grammatical relations and hierarchical constituent structure. Differences betweeen the SRN state representations and the more traditional pushdown store are discussed in the final section.
Mendeley saves you time finding and organizing research
Choose a citation style from the tabs below