Voice of the Striatum: Brain Networks Reveal Why Speech Falters Early in Parkinson’s Disease

For many people with Parkinson’s disease, the first symptom that betrays them is not a tremor but a voice. Speech becomes quieter, flatter, and harder to understand long before most other signs of the disease have fully declared themselves, and yet the biological roots of this early decline have remained stubbornly obscure. A new study published in npj Parkinson’s Disease now offers one of the most detailed looks to date at what is happening inside the brain when speech begins to slip, and it arrives with a finding that could reshape how clinicians think about the disorder: the mechanisms driving speech impairment differ profoundly between men and women.

The research, led by Caitlin Paige Rogers and Sandy Mournet of the Paris Brain Institute, Inserm and CNRS at Sorbonne University, together with colleagues at the Pitié-Salpêtrière Hospital and partner institutions in France, Italy, and Canada, examined 104 patients in the earliest stages of Parkinson’s disease, 68 men and 36 women, alongside 44 healthy controls, 23 men and 21 women. The team combined two powerful and complementary lines of measurement. On one side, they extracted quantitative speech markers from voice recordings, capturing features related to prosody, phonation, articulation, fluency, and rhythm. On the other, they probed the brain itself with two imaging modalities: DaT-SPECT to quantify the loss of dopaminergic terminals in the striatum, and resting-state functional MRI to map how the striatum communicates with cortical language networks.

The choice of the striatum as the focal point is scientifically deliberate. Speech is not produced by the tongue and larynx alone; it is choreographed by a distributed network in which the basal ganglia, a set of deep brain structures that include the striatum, play a central role in timing, sequencing, and modulating the motor commands that shape sound. The striatum sits at the gateway of these circuits, receiving dopamine from the substantia nigra, and it is precisely this dopamine supply that Parkinson’s disease progressively destroys. Because the striatum is anatomically subdivided, with the caudate nucleus and putamen serving partly distinct functional roles, damage to different striatal territories can be expected to disturb different aspects of the speech motor system.

By pairing acoustic analysis with multimodal imaging in the same individuals, the researchers could ask a question that neither measurement alone could answer: which aspect of striatal dysfunction, the death of dopamine terminals or the unraveling of striato-cortical connectivity, best explains the speech changes seen in early disease? The answer, it turns out, depends on sex. In male patients, speech deficits were associated with both the degree of dopaminergic loss in the striatum and the degree of disruption in functional connectivity between the striatum and cortical language regions. In female patients, however, only the connectivity changes were linked to speech impairment, even though dopaminergic degeneration was present in both sexes.

The imaging results also revealed a broader pattern of network breakdown. Functional connectivity was globally reduced in patients with Parkinson’s disease compared with healthy controls, consistent with the idea that the disease is as much a disorder of brain networks as of individual cells. Yet within that global reduction, the female patients showed a striking counter-current: they exhibited higher connectivity between the caudate nucleus and cortical regions than their male counterparts. This relative preservation, or even compensatory elevation, of caudate-cortical communication in women may help explain why their speech deficits were less pronounced overall. Indeed, the acoustic analysis showed that the Parkinson’s-related speech impairments were more severe in male patients than in female patients across the features the team measured.

These findings carry real weight for a long-standing puzzle in neurology. Speech impairment, clinically described as hypokinetic dysarthria, affects the large majority of people with Parkinson’s disease over the course of the illness, eroding intelligibility, social connection, and quality of life. Yet speech symptoms respond poorly to standard dopaminergic therapy compared with limb motor symptoms, and they often appear at diagnosis or even before it. The new results suggest a reason why: speech dysfunction is not simply a linear readout of dopamine depletion. In women, at least in the early stages, the speech phenotype appears to ride on the integrity of striato-cortical communication rather than on the raw quantity of dopaminergic innervation, which may decouple speech outcome from the measures clinicians typically use to stage and treat the disease.

The technical approach deserves attention in its own right. Rather than relying on perceptual ratings by speech therapists, which are subjective and labor-intensive, the team extracted quantitative acoustic features from recorded speech, spanning the five canonical dimensions of hypokinetic dysarthria: prosody, the melodic contour of speech; phonation, the quality of the vocal source; articulation, the precision of consonant and vowel production; fluency, the smoothness of speech flow; and rhythm, the temporal patterning of syllables. This kind of digital biomarker pipeline, which the authors developed with collaborators including speech-processing specialists at Télécom SudParis, is increasingly seen as a route to scalable, objective monitoring of neurological disease, potentially allowing speech to be tracked remotely from a smartphone recording.

The imaging side was equally rigorous. DaT-SPECT, a nuclear medicine technique that binds a radioactive tracer to the dopamine transporter, provided a molecular window onto the survival of dopaminergic terminals in the striatum, the canonical pathological signature of early Parkinson’s disease. Resting-state functional MRI, in which participants lie still while spontaneous blood-oxygenation fluctuations are recorded, allowed the team to compute the functional coupling between striatal seed regions and the cortical networks that support speech and language. Analyzing both modalities in the same cohort, stratified by sex, gave the researchers the statistical leverage to dissociate two candidate mechanisms, dopaminergic denervation and network disconnection, that have previously been entangled in the literature.

The sex differences the study uncovered are more than a statistical curiosity. Parkinson’s disease has historically been studied predominantly in men, and women remain underrepresented in clinical trials and often diagnosed later with different symptom profiles. If the neural machinery of speech impairment operates differently in women, relying more on connectivity and less on dopamine terminal density, then a one-size-fits-all model of the disease is not just incomplete but potentially misleading. The authors argue explicitly that their findings emphasize the need for sex-stratified approaches in both research and clinical practice, from the design of imaging biomarkers to the interpretation of speech assessments and the targeting of therapies, whether pharmacological, surgical, or behavioral such as Lee Silverman voice treatment and other speech rehabilitation programs.

Looking forward, the study points toward a future in which a short voice recording, combined with brain imaging, could stratify patients by the mechanism underlying their speech decline rather than by disease duration alone. For men with early Parkinson’s disease, markers of dopaminergic loss might guide interventions aimed at protecting or replacing striatal dopamine. For women, therapies aimed at preserving or strengthening striato-cortical network function, through cognitive-motor training, neuromodulation, or speech therapy timed to network vulnerability, might hold greater promise. The work, funded by a broad coalition including the French National Research Agency, France Parkinson, the Fondation pour la Recherche Médicale, and the EU-wide ERA PerMed DIGIPD project, was published open access, allowing the wider community to build on its data and methods. As digital voice biomarkers mature, the quiet flattening of a Parkinson’s patient’s speech may become not just a symptom to be endured but a measurable, mechanistically interpretable signal, one that, this research now shows, speaks differently in men and women.

Subject of Research: Sex-specific striatal network and dopaminergic mechanisms underlying speech impairment in early Parkinson's disease

Article Title: Striatal network dysfunction underlying speech impairment in early Parkinson’s disease

Article References: Striatal network dysfunction underlying speech impairment in early Parkinson’s disease. (n.d.). https://doi.org/10.1038/s41531-026-01509-2

Image Credits: AI Generated

DOI: 10.1038/s41531-026-01509-2

Keywords: Parkinson's disease, speech impairment, striatum, dopaminergic loss, DaT-SPECT, resting-state fMRI, functional connectivity, sex differences, hypokinetic dysarthria, voice biomarkers, basal ganglia, caudate nucleus