Semantic-head based resolution of scopal ambiguities

Abstract

We present an algorithm for scope resolution in underspecified semantic representations. Scope preferences are suggested on the basis of semantic argument structure. The major novelty of this approach is that, on the one side, we maintain an (with respect to scopal ambiguities) underspecified semantic representation, and on the other side, at the same time suggest a scope resolution possibility. The algorithm is evaluated on four "reallife" dialogues and fares surprisingly well: about 80% of the utterances containing scopal ambiguities are correctly interpreted by the suggested resolution, leaving errors in only 5.7% of the overall utterances.