The Syntactic Exponence of Sentential Negation: a model for the LinGO Grammar Matrix

Abstract

This thesis presents a revision of the LinGO Grammar Matrix library for sentential negation along with supporting arguments from syntactic and typological literature. A cross-linguistic model of representations in HPSG and MRS for sentential negation types is proposed and defended. The model provides a highly articulated family of predicted types for double, or bipartite negation constructions which were previously untreated in the literature. HPSG analyses for each type in the model are provided along with a description of the implementation. The model is evaluated along two tacks: an intrinsic evaluation based on a system of software regression tests and an extrinsic one based on an experiment with held-out languages. All software developed in this project is available as a part of the Grammar Matrix open-source software repository, http://www.delph-in.net/matrix/.