Acquiring minimalist grammars via constraint satisfaction

Abstract

This thesis shows how to algorithmically construct a Minimalist Grammar lexicon that produces a specified set of MG derivations. This thesis introduces a mathematical structure, a Collection of Constraints, that captures the logical constraints, including those that arise as a consequence of the shortest move constraint, imposed upon the syntactic features of lexical items as they are merged together in a derivation produced by a given Minimalist Grammar lexicon. Methods are then developed that (a) map Minimalist Grammar lexicons to be Collections of Constraints, (b) map Collections of Constraints to Minimalist Grammar lexicons and (c) may combine two or more Collections of Constraints into a single Collection of Constraints. The thesis then demonstrates via a series of examples, framed as a simplified acquisition process, how these methods may be used together to iteratively construct a Minimalist Grammar lexicon starting from an empty Collection of Constraints and a sequence of Minimalist Grammar derivations, such that the constructed lexicon is able to generate the set of derivations.