Summary: Researchers have made a significant advance in understanding how the human brain prepares words for speech. Using ultradense Neuropixels probes, they mapped how individual neurons represent basic speech sounds and combine them into the sequences that become spoken language.
This research clarifies the multi-step cognitive process behind fluent speech and suggests new directions for treating language disorders. The findings also point toward potential brain-machine interfaces or prosthetic systems that could generate synthetic speech for people with neurological impairments.
Key Facts:
- The team used Neuropixels probes to record the activity of single neurons, revealing how the brain plans and produces words.
- Distinct neuronal populations were identified for speaking versus listening, indicating separate circuits for language production and comprehension.
- The results may inform therapies for speech and language disorders and enable development of neural interfaces to synthesize speech.
Source: Harvard
Using advanced human brain recordings, a study led by researchers from Massachusetts General Hospital (affiliated with Harvard) shows how neurons cooperate to let people decide on words and then produce them aloud.
The work delivers a precise view of how phonetic elements like consonants and vowels are encoded in the brain well before they are spoken, and how those elements are assembled into syllables and words during speech production.
Published in Nature, the study offers new insights that could improve understanding and treatment of speech and language impairments.
“Although speaking usually feels effortless, the brain carries out many rapid and complex steps to produce natural speech — selecting words, planning the necessary mouth and tongue movements, and executing vocalizations,” says senior author Ziv Williams, associate professor of neurosurgery at MGH and Harvard Medical School.
“Our brains perform these tasks very quickly — roughly three words per second during natural conversation — and with few errors. How the brain achieves that speed and reliability, however, has been largely mysterious.”
By applying ultrahigh-density Neuropixels probes to record single-neuron activity in the prefrontal cortex, Williams and his colleagues identified neurons tied specifically to language production. They also observed separate neural populations that respond during listening, highlighting distinct roles for production versus perception.
“The use of Neuropixels probes in humans was first pioneered at MGH,” Williams explains. “These probes are thinner than a human hair yet contain hundreds of channels, allowing simultaneous recording from dozens or even hundreds of individual neurons.”
Williams developed the recording methods in collaboration with Sydney Cash, professor of neurology at MGH and Harvard Medical School, who co-led the study.
The study details how neurons represent fundamental components of speech — from phonemes, the smallest distinguishable speech sounds, to their organization into syllables and larger sequences. For instance, producing the consonant sound in “da” requires a specific tongue placement against the palate; the researchers found neurons that become active before such a phoneme is articulated. Other neurons encoded the way phonemes combine to form syllables.
Using these neural signals, the team could reliably predict the speech sounds a person intended to produce before audible articulation. In practical terms, the recorded patterns reveal which combinations of vowels and consonants are about to be spoken. This pre-articulation information could be used to develop brain-machine interfaces or prosthetic devices that generate synthetic speech, offering potential communication solutions for many patients.
“Disruptions to speech and language networks occur across a wide range of neurological conditions — stroke, traumatic brain injury, tumors, neurodegenerative and neurodevelopmental disorders, among others,” says Arjun Khanna, a postdoctoral fellow in the Williams Lab and a co-author of the study.
“A deeper understanding of the neural circuitry underlying speech and language may pave the way for novel treatments and assistive technologies.”
The researchers plan to extend their work to more complex aspects of language, exploring how the brain selects specific words and arranges them into sentences that convey meaning, intention, and emotion.
About this language and speech research news
Author: MGH Communications
Source: Harvard
Contact: MGH Communications – Harvard
Image: The image is credited to Neuroscience News
Original Research: Open access.
“Single-neuronal elements of speech production in humans” by Ziv Williams et al., published in Nature.
Abstract
Single-neuronal elements of speech production in humans
Humans can combine a vast array of articulatory movements to create meaningful speech. This capacity to organize specific phonetic sequences, segment them into syllables, and apply inflection on subsecond timescales enables the production of thousands of distinct word sounds and underpins language. Yet the cellular mechanisms that plan and generate spoken words have remained largely unknown.
Using acute ultrahigh-density Neuropixels recordings that sample across the cortical column in humans, the study identifies neurons in the language-dominant prefrontal cortex that encode detailed information about the phonetic structure of planned words during natural speech production.
These neurons represent the order and structure of articulatory events before utterance and reflect the segmentation of phonetic sequences into syllables. They accurately predict phonetic, syllabic, and morphological components of upcoming words and display a temporally ordered dynamic.
Overall, the findings reveal a structured organization and cascading encoding of phonetic representations by prefrontal neurons, showing how cellular processes can support the planning and production of speech.