Abstract
This study implements two published intelligibility tests to measure the gains in intelligibility made by Parkinson's disease patients after a regimen of intensive vocal therapy. In the first study, ten subjects with idiopathic Parkinson's disease were rated using a paired-word intelligibility test. Results showed that this test was too simple to document subtle intelligibility changes in these patients resulting from voice therapy. Notwithstanding, several error categories emerged as being vulnerable to distortion in the Parkinson's disease patients tested. A second, more phonemically challenging published test was implemented for Experiment 2. This test incorporated the frequently-distorted error categories of the first study such as glottal versus null, initial voicing, and fricative versus nasal distortions, as well as other categories historically of interest, in a more difficult corpus of test words. Both speech pathologists familiar with dysarthric speech and naive professionals were included as listeners in Experiment 2. Although the second test highlighted more errors than the first, statistical analysis showed that the structure of the second test introduced confounds which seriously impacted its validity. The greatest source of error was determined to be due to the introduction of test words of interest without corresponding test foil words. After repeated scorings, both naive and informed listeners made predictions based on former presentations. Recommendations are made regarding the structure of a future intelligibility test to avoid the noted confounds.

Back to NCVS-sponsored Research Articles