This paper introduces MDRT, an extension of DRT to a multi-agent dialogue. The main issue handled is cross-speaker anaphora, whereby one speaker refers anaphorically to a discourse entity introduced by another speaker. We present a representation of multispeaker discourses that allows a a non-degenerate solution to the commitment problem: to what conditions about a discourse entity is a speaker committed when using cross-speaker anaphora? 1. Introduction One of the interesting developments in the (formal) semantics of natural language is the emergence of theories for the representation of meanings of larger linguistic structures than a single sentence, the latter having been the target of classical semantics. Two notable examples are the (similar) theories presented in [5, 4]. Also related is the theory in , striving, in addition, towards compositionality. A focus of attention in all of these theories is the treatment of inter-sentential anaphora. In these theories, meaning is ta...
|Title of host publication
|Proceedings of the First International Workshop on Computational Linguistics
|H. Bunt, R. Muskens, G. Rentier
|Place of Publication
|University of Tilburg
|Number of pages
|Published - 14 Jul 1995