This proposal deals with the articulatory underpinnings of speech intelligibility deficits in the neurogenic speech disorders associated with Parkinson disease (PD) and amyotrophic lateral sclerosis (ALS). The proposed work uses x-ray microbeam (articulatory kinematic), speech acoustic, and speech intelligibility measures to address a major descriptive need, as well as several hypotheses. The major descriptive need is that virtually nothing is known about lingual ranges of motion for articulatory behavior in neurogenic speech disorders; over the last thirty years a substantial research effort has been devoted to possible relationships between oromotor non-verbal behavior and speech intelligibility, but it is argued from the available literature that this is not the right approach to understanding speech articulators during speech production. This explains the descriptive aim of the current proposal, which make use of carefully-defined measures of the phonetic working space (i.e., parameterizations of ranges) for articulation. An associated descriptive goal is to determine if jaw movement during speech is greater for ALS, as compared to PD speakers, and if the lingual working spaces are therefore differentially affected across group by mandibular motions. One of the hypotheses makes use of these kinematic working space parameters by claiming that there should be a relationship between measures of the phonetic (articulatory kinematic) working space for vowels, and acoustic vowel space (area of an F1-F2 plot) or formant ranges on the one hand, and speech intelligibility on the other hand, Specifically, the hypothesis is that reduced phonetic working spaces will be associated with reduced acoustic vowel working spaces and reduced speech intelligibility. The hypothesis will be tested both within each of the disease groups mentioned above (N=20 patients per group), as well as across all patients (N=40). A second hypothesis is that the relative phasing of successive articulatory gestures in VCV sequences will be delayed relative to the phasing observed in neurologically-normal speakers. This hypothesis grows out of both clinical and empirical observations of reduced co- articulation (or, in an alternative jargon, reduced overlap) among many speakers with neurogenic speech disorders.
The final aim of this proposal is to initiate vocal tract modeling studies of neurogenic speech disorders, where the input to the model is in the form of the x- ray microbeam data and observed speech spectra; the goal of this work is to improve articulatory inferences from speech acoustic data collected from persons with neurogenic speech disorders. A final hypothesis, to be tested with the vocal tract model, is that dysarthric speakers have fewer articulatory degrees of freedom in realizing vowel spectra, as compared to normal speakers.
Showing the most recent 10 out of 11 publications
