SPEAKER

This paper presents how a two-coupled-oscillator speech rhythm model can handle the variability of the durational patterns found in natural data. For doing so the model takes into account a small set of nested levels of dynamical coupling between linguistic and production-related subsystems. In this framework speech rhythm is the quasi-optimal output of the coupling between two components, a perception-oriented tendency to pattern structur-ing, implemented by the inter-relation between lo-cal syntactic information and a phrase stress os-cillator, and a production-oriented regularity con-straint, implemented by the oscillation of two com-ponents, a syllabic oscillator and a phrase stress os-cillator. The syllabic oscillator ’ pulses are anchored at vowel onsets, implementing the carrier compo-nent of speech rhythm production, characterising the building-block of prosodic timing. The model gen-erates complex patterns of V-to-V durations via the consequences of a phrase stress oscillator’s entrain-ment onto the syllabic oscillator. This mechanism of generation is able to cope with intra- and inter-speaker rhythmic variability