Antecedent recovery

Abstract

This paper explores the problem of finding non-local dependencies. First, we isolate a set of features useful for this task. Second, we develop both a two-step approach which combines a trace tagger with a state-of-the-art lexicalized parser and a one-step approach which finds nonlocal dependencies while parsing. We find that the former outperforms the latter because it makes better use of the features we isolate.