A Logical Model of Computation

Abstract

In chapters III and IV, we have presented a model of sentence processing and its grammatical basis. We have argued generally that the human syntactic processor has a distributed, modular architecture, and more specifically that these modules are determined by representational and informational sub-domains of our syntactic knowledge. We have proposed four modules, each concerned with recovering different aspects of a coherent interpretation. Modules are responsible for recovering their own representation with respect to their current input. We have further outlined some aspects of the model’s computational behaviour: at a relatively high level, we propose that the modules operate concurrently and incrementally, with each constructing a maximal output representation for the current partial input, as input is encountered. In addition to each module satisfying the requirement of incrementality, we have also hypothesised several strategies, located primarily within the phrase structure module (since this is where the greatest amount of ambiguity occurs).