Age-related changes in cognitive function vary widely between people. Even within a single age group, some stay sharp while others lose their edge. Biological systems within one person also change at different rates and in different directions, so aging cannot be captured by a single number. Screening tests like the Montreal scale are useful, but in healthy people the results hit a ceiling. Subtle differences in how a task is performed go unnoticed. Research on speech and neurointerfaces sits nearby, but this study stands apart: it brings speech, voice and blood flow together in a single measurement in healthy people of different ages.
Participants performed a phonemic verbal fluency task: in 60 seconds they named as many words as possible starting with the syllable "ka". At the same time, speech was recorded with a microphone at a distance of about two meters, and total hemoglobin in the frontal region was measured with a wearable device using a single light source. Words were counted in ten-second intervals, and pauses between words were computed from the audio track. From the blood flow signal, the area under the curve and the time to peak were extracted. Groups were compared using statistical tests, accounting for sex, and the researchers checked whether speech duration explained the differences in blood flow.
The older group named an average of 14 words versus 17 in the younger group, and pauses between words were nearly twice as long: 1.6 seconds versus 0.7. Time to peak blood flow in people over 70 was 33 seconds versus 16 seconds in forty-year-olds, and the area under the curve was higher. Within each group, those who named more words had a later peak. The difference in speech duration between groups was only 4 seconds and explained just 6.7% of the variation in time to peak. Acoustic voice measures behaved differently: one parameter declined with age, another did not change, and a third moved in an unexpected direction.
The work differs from neighboring studies in that it examines not one channel but three at once: speech, voice and blood flow, and demonstrates their independence. This is not an attempt to improve dementia screening but a description of how different systems age apart. The results highlight the limitations of any single measure, whether it is a total score on a cognitive test or one acoustic metric. Characterizing speech aging requires joint modeling of several axes of measurement, not refinement of a single metric.
The topic could move toward longitudinal observations of the same people with repeated measurements of speech, cognitive function and physical condition. This would help separate individual trajectories from between-group differences. Measuring physical function, such as grip strength and walking speed, is also promising, to directly assess the peripheral contribution to voice. More precise temporal annotation of speech, including first-word latency and the durations of individual words, would help break down the budget of the minute more fully.
The results could be useful in gerontology and speech-language pathology for a finer assessment of age-related changes in speech. They also matter for developers of wearable devices that use speech and hemodynamic biomarkers of aging.
The next step will be a longitudinal study of the same participants with repeated measurements of speech, voice and blood flow. It is also worth collecting data on education, blood pressure and physical function to separate peripheral and central contributions.
The work touches on an unresolved problem in gerontology and cognitive neuroscience: how to separate age-related changes in speech linked to the brain from changes linked to the vocal apparatus and breathing. It is still unclear whether a single voice measure can tell us about cognitive state, and this study shows that different measures behave independently. This casts doubt on the idea of a single speech biomarker of aging.
Participants over 70 named an average of 3 fewer words than forty-year-olds, but in the last 10 seconds of the task they produced twice as many words: 1.48 versus 0.70. The younger group ran out of steam by the end of the minute, while the older group kept a steady pace.
Not published on arXiv. Here are its page, the PDF and the formalization: you can check it yourself, not only read our account.