Thesis

The application of linear and nonlinear estimators of acoustic variability in the assessment of speech motor control in hypokinetic dysarthria

Creator
Rights statement
Awarding institution
  • University of Strathclyde
Date of award
  • 2018
Thesis identifier
  • T14854
Person Identifier (Local)
  • 200956917
Qualification Level
Qualification Name
Department, School or Faculty
Abstract
  • To improve diagnostic and outcome measures in the assessment and treatment of speech disorders, researchers and clinicians are always in search of new techniques to quantify speech impairment. This thesis investigates the relatively unexplored area of linear and nonlinear estimators of acoustic variability and their suitability for assessing the stability of movement patterns of speech organs. In particular, it focused on the estimators' ability to differentiate hypokinetic dysarthria from unimpaired speech, as well as speech of young adults from older adults. In addition, the variability results of hypokinetic dysarthric speakers were compared with the results of standard diagnostic assessments.;Twenty-three speakers with hypokinetic dysarthria and forty neurologically healthy individuals participated in the study. A series of sentence repetition tasks was devised with varying linguistic, cognitive and motor demands. A range of time-varying speech features was extracted from the acoustic signal in order to capture speech motor performance in a number of segmental and prosodic aspects of speech production.;The results showed that acoustic measures of variability were successful in classifying dysarthria and healthy speakers as well as adult speakers differing in age, and correlated with different clinical-based assessments.;The findings of this study indicate that the characterization of complex speech movements during phrase production when evaluating linguistic, cognitive, or motor demands within or between speaker groups cannot be reduced to a single task or speech property, but rather call for a multi-faceted approach in which distinct variability estimators, speech tasks and acoustic properties are evaluated simultaneously.
Advisor / supervisor
  • Lowit, Anja
Resource Type
DOI
Date Created
  • 2018
Former identifier
  • 9912597992502996

Relations

Items