Neural network-based measure of consonant Lenition in Parkinson's disease

Abstract

This study investigates the effects of Parkinson's disease (PD) and various linguistic factors on the degree of lenition in Spanish stops. The lenition is estimated from posterior probabilities calculated by recurrent neural networks trained to recognize sonorant and continuant phonological features. First, individuals with PD exhibit a higher degree of lenition in their stops compared to healthy controls, suggesting that PD has a significant impact on the articulatory control of stops, resulting in more pronounced lenition. Second, stress influences the degree of lenition, with stressed syllables showing less weakening compared to unstressed syllables. Thirdly, place of articulation plays a role in lenition, but the effects differ between PD and healthy control groups. PD patients demonstrate greater weakening in bilabial and velar stops compared to dental stops, while healthy controls exhibit more lenition in velar stops, with no significant difference between bilabial and dental stops. These findings suggest abnormalities in lip movement control among PD patients align with previous research on reduced lip velocities in Parkinsonian patients. Importantly, the study highlights the effectiveness of recurrent neural networks in quantifying lenition in PD patients.