Interpretation as conflict resolution

Abstract

Semantic interpretation is not a simple process. When we want to know what a given sentence means, more is needed than just a simple ‘adding up’ of the meanings of the component words. Not only can the words in a sentence interact and conflict with each other, but also with the linguistic and non-linguistic context in which the sentence was uttered. Deictic, anaphoric, and elliptical expressions attune their interpretation to the properties of the context, noun phrases have to be shifted in type to fit a particular argument slot (Partee 1987), and lexical meanings may need to undergo coercion to match the neighboring words (Pustejovsky 1995). Optimality-theory provides a framework to deal with such conflicts in interpretation in a systematic way by means of constraint-ranking (Prince and Smolensky 1993). In 2002, NWO, the Dutch Organization for Scientific Research, funded a project proposal submitted by Petra Hendriks (Groningen University), Helen de Hoop (University of Nijmegen) and the first author of this paper (Utrecht University) as part of the Cognition Program.1 The starting point of the project is the notion of conflicts in interpretation and their resolution by constraintranking. This paper reports on the first results, and sketches the lines of research opening up in this project. We illustrate with four examples: anaphora resolution, the polysemy of the spatial preposition round, negative concord and the acquisition of indefinites.