Task-dynamics of gestural timing: Phase windows and multifrequency rhythms

In this paper, we explore the hypothesis that intergestural phasing relations are implemented via coupling terms in a nonlinear dynamical systems model. Specifically. we describe recent computational developments of the task-dynamic model of gestural patterning (e.g., E. Saltzman, J. A. S. Kelso, Psychological Review 94 (1987) 84-106; E. Saltzman, K. G. Munhall, Ecological Psychology 1 (1989) 333-382) that are focused on modeling the timing of rhythmic action units. First, we explore the possibility of attractor states for intergestural phasing that are characterized as ranges or phase windows (D. Byrd, Phonology 13 (1996) 139-169), and contrast this behavior with standard models that display punctate relative phasing. It is argued that the phase window approach can provide flexible control of the relative timing of articulatory gestures, allowing constrained variability in intergestural timing as a function of linguistic and para-linguistic factors. Second, we discuss how this extension of the task-dynamic model has been adapted for modeling the production of multifrequency rhythms (speech or bimanual). This work explores the control of the frequency- and phase-locking characteristics of coupled limit cycle oscillators by examining how desired frequency ratio, intrinsic frequency detuning, and coupling asymmetries interact in creating observed rhythmic patterns. Using this method, details of the resultant transient and steady-state trajectories of phase and amplitude are generated that are not available using models derived with averaging techniques. (C) 2000 Elsevier Science B.V. All rights reserved.