Parsing as Deduction: The Use Knowledge of Language

Hypothesising that the human parser is a specialized deductive device in which Universal Grammar and parameter settings are represented as axioms provides a model of how knowledge of language can be put to use. The approach is explained via a series of model deductive parsers for Government and Binding Theory, all of which use the same knowledge of language (i.e., underlying axiom system) but differ as to how they put this knowledge to use (i.e., they use different inference control strategies). They differ from most other GB parsers in that the axiom system directly reflects the modular structure of GB theory and makes fidl use of the multiple levels of reprepresentation posited in GB theory.