AI & ML Nature Is Weird

Parkinson's and ALS leave a distinct "fingerprint" on speech patterns that remains identical regardless of whether the speaker is using English, Spanish, or Japanese.

April 25, 2026

Original Paper

Phonological Subspace Collapse Is Aetiology-Specific and Cross-Lingually Stable: Evidence from 3,374 Speakers

Bernard Muller, Antonio Armando Ortiz Barrañón, LaVonne Roberts

arXiv · 2604.21706

AI-generated illustration

The Takeaway

Speech scientists analyzed the voices of 3,374 speakers across 12 different languages to see how neurological diseases degrade communication. They found that each disease causes a specific collapse of vocal features that transcends the rules of grammar or phonetics. A patient with Parkinson's in Germany will show the same linguistic degradation as a patient in China. This discovery suggests that these diseases attack the biological foundations of speech in a universal way. Doctors could use these stable patterns to diagnose brain disorders earlier using simple voice recordings in any language. The biological impact of the disease is more powerful than the cultural structure of the language itself.

From the abstract

We previously introduced a training-free method for dysarthria severity assessment based on d-prime separability of phonological feature subspaces in frozen self-supervised speech representations, validated on 890 speakers across 5 languages with HuBERT-base. Here, we scale the analysis to 3,374 speakers from 25 datasets spanning 12 languages and 5 aetiologies (Parkinson's disease, cerebral palsy, ALS, Down syndrome, and stroke), plus healthy controls, using 6 SSL backbones. We report three find