Attention, intentions, and the structure of discourse
Computational Linguistics
An algorithm for pronominal anaphora resolution
Computational Linguistics
Centering: a framework for modeling the local coherence of discourse
Computational Linguistics
Anaphora for everyone: pronominal anaphora resoluation without a parser
COLING '96 Proceedings of the 16th conference on Computational linguistics - Volume 1
Hi-index | 0.00 |
Beginning with the observation that syntactic and semantic information often coincide (i.e., subjects are often agents, objects often patients), this study investigates the possibility that preference to resolve a sentence-initial pronoun to a syntactically prominent antecedent might actually be better explained in terms of preference for resolving to a semantically prominent antecedent. The study takes Discourse Prominence Theory (Gordon and Hendrick [11, 12]) as an underlying framework. Results of three psycholinguistic experiments using a self-paced reading task show that both syntactic and semantic information guide readers' pronoun resolution preferences. This suggests a revised understanding of Discourse Prominence Theory in which the prominence of discourse referents is determined through a complex process depending on multiple linguistic factors. Results further show that the relative degree of prominence among competing candidates influences resolution processes.