A connectionist model for the simulation of human spoken-word recognition

Abstract

A new psycholinguistically motivated and neural network base model of human word recognition is presented. In contrast to earlier models it uses real speech as input. At the word layer acoustical and temporal information is stored by sequences of connected sensory neurons that pass on sensor potentials to a word neuron. In explorations with a small lexicon that includes groups of very similar word forms, the model meets high standards with respect to word recognition, simulating a number of well-known psycholinguistic effects.