- Main
Computational modeling of psychophysics in speech motor control
- Gaines, Jessica
- Advisor(s): Nagarajan, Srikantan
Abstract
Speech, the uniquely human ability by which most people communicate, is a highly complex motor task. Thus complex neural mechanisms are required to control the muscles of the vocal tract to produce a desired speech sound. Understanding the neural mechanisms of speech motor control in typical speakers can provide a foundation to understand how differences in these processes contribute to a variety of speech and communication disorders. This document contributes to the field’s knowledge of speech motor control through computational modeling of the vocal tract and related neural systems as well as through the observed response to perceived errors in the speech acoustic signal. Chapter 1 introduces the importance of research in the area of speech motor control and provides a preview of the computational modeling and psychophysics techniques used to study it. Chapter 2 discusses the use of a computational model of vocal tract shape and the corresponding acoustic output to show that a comprehensive range of vowel sounds can be produced by vocal tract shapes with constrictions in a few discrete locations. Chapter 3 fits a mechanistic model of speech motor control to behavioral speech data to interpret behavioral differences between typical speakers and speakers with cerebellar ataxia in the context of interpretable model parameters. Finally, Chapter 4 details a study of speech behavior when a persistent error is introduced, removed, and re-introduced, providing an opportunity for re-learning. The results of this study can be used to inform future modeling efforts. Together, these studies show how computational modeling can inform and be informed by data from real speakers to advance our knowledge of how the brain controls the vocal tract during speech.