Acoustic characteristics of normal and dysarthric speech.

Abstract

The experiment reported here examined the implementation of final lengthening and main sentence stress in the speech of one moderately ataxic dysarthric (AD) and one normal speaker. A set of C1VC2 syllables was constructed such that C1 varied over [b,d,g,p,t,k], C2 was always [d], and V was one of the nine nonretroflex vowels for the [b——d] context, and one of [i, eh, æ] for the other five initial contexts. These syllables were embedded in carrier sentences such that the target syllable occurred in two positions, final and medial, and bearing main sentence stress or not. Eight repetitions of each syllable in each position-stress condition were recorded and digitized. Measurements of initial closure, VOT (if any), vowel, and final closure durations were made directly from the waveform. Preliminary results indicate that the AD speaker implemented final lengthening of the target syllable in a manner similar to the normal speaker overall, but differed in the distribution of the lengthening effect over the acoustic segments comprising the syllable. This result is not in agreement with the results reported by Bell-Berti etal. [Proc. of the 12th Int. Congr. Phon. Sci. (1991)] for French AD speakers. Results on the implementation of sentence stress will also be presented. These will be discussed within the framework of how cerebellar deterioration can affect the fine control of speech timing.