Comparing token-based and trajectory-based vowel space metrics for assessing speech in Parkinson's disease

Area of Science:

• Speech-language pathology
• Acoustic analysis
• Neurology

Background:

• Parkinson's disease (PD) often impairs speech production, affecting intelligibility.
• Quantifying these speech changes requires reliable acoustic measures.
• Vowel space characteristics are sensitive to motor speech deficits.

Purpose of the Study:

• To evaluate three acoustic measures of vowel space for quantifying speech differences in people with Parkinson's disease (PwPD) compared to controls.
• To assess the correlation between these acoustic measures and speech intelligibility.
• To determine the utility of token-based versus trajectory-based measures.

Main Methods:

• Fifty-six participants (28 PwPD, 28 controls) read a passage with corner vowels.
• Calculated token-based measures: triangular vowel space area (tVSA) and vowel articulation index (VAI).
• Calculated trajectory-based measure: articulatory-acoustic vowel space (AAVS). Assessed intelligibility using read sentences.

Main Results:

• All vowel space measures showed significant group differences between PwPD and controls.
• Only token-based measures (tVSA, VAI) could statistically discriminate between groups, though with poor model performance (R² < 0.113).
• Token-based measures, particularly VAI, were statistically related to speech intelligibility.

Conclusions:

• Token-based acoustic measures are suitable for quantifying changes in vowel space and assessing speech intelligibility in PwPD.
• VAI demonstrates potential as a key metric for speech analysis in Parkinson's.
• Further research should explore generalizability to other motor speech disorders and stimuli.