9 - SPEAKING AND SINGING

Abstract

This chapter presents the theory of speaking and singing. A theory of how speech sounds can be produced reliably assumes that the articulators aim for specific target positions. The movement of the articulators toward the target positions is mediated by the gamma system. Initial results were promising for the spatial target theory, but later results were less so. A more recent theory is that the speech-production system attempts to bring the articulators to characteristic relative positions. As predicted in the theory, following external perturbations, there is rapid compensation that preserves relative positions needed for immediate articulatory goals. A possible mechanism for achieving invariant relative positioning relies on mutual inhibition between command centers for articulators whose relative positions must be coordinated. A parallel distributed system has recently been proposed as a model of coarticulation. The system anticipates features of forthcoming actions because of generalization to the states whose exact values at earlier (or later) times do not matter in the language being spoken. Pig Latin and backward speech provide evidence for the psychological reality of syllables and the autonomous representation of stress. Laboratory studies of the reaction time to begin speaking when a switch must be made from an initially prepared utterance to another utterance suggest that the syllables of a word are represented hierarchically. Hierarchical organization provides a basis for specifying characteristics of an entire word in a single processing step.