ABSTRACT
A number of recent cognitive models of human sentence processing have appealed to costs and tradeoffs in resource requirements to support their positions on modularity and interaction effects in resolving syntactic and semantic ambiguities (e.g., Stowe, 1991). However, computational models that analyze the quantitative aspects of sentence processing have only dealt with syntactic parsing (e.g., Abney and Johnson, 1991). Such models have not addressed the tradeoffs in syntactic parsing decisions vis-a-vis local ambiguities and the costs and benefits of making early commitments in semantics. A computational model of the costs and benefits of making both syntactic and semantic decisions at different points in time during sentence processing would provide an excellent formal framework for analyzing the empirical factors involved in sentence processing and for designing cognitive models and experiments. I present such a formal model in this poster.
