Defining and measuring speech movement events.

Abstract

A long-held view in speech research is that utterances are built up from a series of discrete units joined together. However, it is difficult to reconcile this view with the observation that speech movement waveforms are smooth and continuous. Developing methods for reliable identification of speech movement units is necessary for describing speech motor behavior and for addressing theoretically relevant questions about its organization. We describe a simple method of parsing movement signals into a series of individual movement "strokes," where a stroke is defined as the period between two successive local minima in the speed history of an articulator point, and use that method to segment speech-related movement of marker points placed on the tongue blade, tongue dorsum, lower lip, and jaw in a group of healthy young speakers. Articulator fleshpoints could be distinguished on the basis of kinematic features (i.e., peak and boundary speed, duration and distance) of the strokes they produce. Further, tongue blade and jaw fleshpoint strokes identified to temporally overlap with acoustic events identified as alveolar fricatives could be distinguished from speech strokes in general on the basis of a number of kinematic measures. Finally, the acoustic timing of alveolar fricatives did not appear to be related to the kinematic features of strokes presumed to be related to their production in any direct way. The advantages and disadvantages of this simple approach to defining movement units are discussed.