On Semantic Underspecification

Abstract

Scope, especially quantiier scope, is the rst and best investigated phenomenon in the eld of underspeciication research. The use of logic representation languages in natural language semantics often requires a choice between alternative scopings in the interpretation process, where this choice seems to be premature or unnecessarily ne-grained. An example is (1a), which can be assigned either (1b) or (1c) as logical representation, although in many cases the relative scope of universal and existential operator may be unclear, or irrelevant, or both. (1) a. Two foreign languages are spoken by every linguist. b. 9 2 x(foreign language(x)^8y(linguist(y) ! spoken by(x; y))) c. 8y(linguist(y) ! 9 2 x(foreign language(x)^spoken by(x; y))) The basic idea behind the technique of underspeciication is to represent the semantic information connected with a semantically underdeter-mined utterance A by a partial description of logical structure, rather than enumerating the possible completely speciied interpretation alternatives for A. Pioneering research on structural underspeciication has been performed at SRI Cambridge. The QLF formalism described in (Alshawi and Crouch, 1992) provides an underspeciied treatment of scope as well as certain aspects of referential ambiguity, ellipsis, and focus-background structure. 1 Another important source for the interest in underspeciication is lexical semantics. Example (2) is a representative for a large eld of ambiguity phenomena, which are conventionally classiied as lexical ambiguities, but diier from trivial cases like the homonyms bank or pen in several important ways.

Cite this paper

@inproceedings{Pinkal1999OnSU, title={On Semantic Underspecification}, author={Manfred Pinkal}, year={1999} }