17 - Computational Models of Psycholinguistics

Summary

This chapter outlines the historical origins and the state of art of computational models of psycholinguistic processes. It considers interrelationships between the different theoretical traditions in reaction to the Chomskyan revolution. The chapter focuses attention on topics that have the widest general theoretical implications, both for fields of computational cognitive modeling and for the project of cognitive science more broadly. The chapter outlines and contrasts symbolic, connectionist, and probabilistic approaches to the computational modeling of psycholinguistic phenomena. The chapter considers word segmentation and recognition, and single word reading. The chapter focuses primarily on parsing, relating connectionist and probabilistic models to the symbolic models of grammar and processing associated with Chomsky's program. The chapter reviews formal and computational models of language learning and re-evaluates, in the light of current computational work, Chomsky's early theoretical arguments for a strong nativist view of the computational mechanisms involved.