AN ARTICULATORY-ACOUSTIC INVESTIGATION OF TIMING AND COORDINATION IN THE FLUENT SPEECH OF PEOPLE WHO STAMMER
This thesis investigates Wingate’s Fault-Line hypothesis (1988) which suggests that disfluencies in people who stammer (PWS) result from a deficit in transition from consonant to vowel (CV) thereby implying that stammering as a motor-control disorder would affect transitions even when not perceptually salient. To test this proposal, we explored the perceptually fluent speech of PWS using instrumental analysis (ultrasound and acoustic) to determine the underlying pervasiveness of disfluencies in this group as compared to people who do not stammer (PNS). Following fluency screening of recorded utterances, we applied acoustic and articulatory analysis techniques to perceptually fluent utterances of 9 PWS and 9 typical speakers in order to identify indicators of disfluency in the transition from syllable onsets to the following vowel. Measures of acoustic duration, locus equation and formant slope offer insights into timing and degree of coarticulation. The articulatory ultrasound tongue imaging technique moreover provides kinematic information of the tongue. A novel technique was applied to dynamically analyse and quantify the tongue kinematics in transition. This allowed us to treat the perceptually fluent speech of PWS as an ongoing time-situated process. Both acoustic and articulatory findings indicate by-group differences in timing, whereby PWS are overall slower and more variable in the execution of CV transitions when compared to typical speakers (PNS). The findings from both instrumental approaches also indicate differences in coordination, suggesting that PWS coarticulate to a lesser extent than PNS. Overall, these findings suggest that PWS exhibit a global deficit in CV transition that can be observed in perceptually fluent as well as stammered speech. This is in keeping with the predictions of Wingate’s Fault-Line hypothesis. iv The fact that the conclusions from the acoustic and articulatory measures are coherent, shows that acoustic measures may be sufficient to act as a proxy for articulatory measures.