38 Distinctive Properties of Speech Motor Control

Language has a dualistic structure: a meaning level formed by morphemes, words and syntactic rules, and a sound level, at which concatenations of a limited number of basic sounds (segments, phonemes) are used to form the lexical items of our mental dictionary. This dualistic structure constrains speech motor control theories. They must account for: 1. effects of segmental context on movements for a given segment, and; 2. semantic, syntactic and lexical effects, such as semantic focus, signalled by sentence stress, intonation, syntactic boundary effects, signalled by segment duration, and lexical stress effects. Additional constraints arise from nonlinguistic factors such as variations of speaking style and speaking rate, ‘compression’ effects of utterance size on segment duration and the great ability of the production mechanism to adapt to postural variations, either self imposed or experimenter imposed. Consideration of these constraints plus speech errors suggests a tripartite structure to the speech production process consisting of 1. establishment of semantic and syntactic structure and choice of lexical items, 2. serial arrangement of a set of segments in short term store, and 3. implementation of motor control algorithms responding to segmental specification and current linguistic and nonlinguistic constraints. Theoretical approaches to speech motor control have consisted primarily of unsatisfactory attempts to use the concepts of spatial or auditory target, a spring analogy, and the concept of coordinative structure to account for the maintainence of invariance in segment production. The main general hypothesis pertaining to control of segment durations–that there is a mechanism tending to produce equal intervals between stressed syllables in English (Stress Timing) is also unsatisfactory.