Automated Online Speech Analytics Reveal Language and Affective Changes in Parkinson’s Disease

Speech-related changes hold promise as biomarkers in Parkinson’s disease (PD), yet evidence remains preliminary and higher-order language alterations are under-characterised. Leveraging data from 1,168 participants with PD and 541 controls within the Australian Parkinson’s Genetics Study, we report population-level evidence of PD-associated linguistic and affective alterations. Participants with PD exhibited reduced fluency and expressivity, with slower rate, lower reading accuracy, reduced output, simpler syntax, and more neutral, less positive emotional tone. Sex-specific patterns emerged: females used less specific language with fewer nouns and modifiers, while males showed broader vocabulary, higher verb density, and fewer adverbs. Language changes were associated with disease duration and cognitive, mood, and sleep comorbidities, and were linked to reduced communication effectiveness and social engagement. These findings deepen understanding of communicative and affective changes in PD, and our scalable online assessment with automated analytics provide a replicable framework for embedding speech analysis into research and clinical practice.