COUPLED OSCILLATOR MODEL OF SPEECH RHYTHM

Abstract

Attempts to verify instrumentally tendencies toward stress tim-ing or syllable timing have been less than successful, leading many researchers to abandon the rhythm dichotomy as too sim-plistic. One empirical finding which suggests an underlying unity in the rhythms of different languages is a statistical ten-dency for duration of stress group to be a linear function of the number of syllables contained in it, with languages differing mainly in the constant term of the linear function. We use a robust mathematical model of simple coupled oscillators to argue that the linear relation observed may reflect a very general tendency of rhythmic systems organized in hierarchical interact-ion, regardless of concrete details, which may differ in complex ways. In addition, the oscillator model provides a rich concep-tual tool for elucidating rhythmical differences in speech behav-ior. 1