News Release

This is where your next word comes from

A newfound role for Wernicke’s area challenges the textbook model of language and could improve brain-computer interfaces

Peer-Reviewed Publication

University of California - San Francisco

Media kit available: https://tiny.ucsf.edu/BlueprintForSpokenWords

Subscribe to UCSF News

 

A discovery by scientists at UC San Francisco rewrites our understanding of how the human brain processes speech, challenging the century-old view that understanding and producing speech are handled by separate systems. It also resolves a longstanding paradox: Why can injury to a brain region that is critical for speech comprehension also impair the ability to produce words?

The researchers, led by Edward Chang, MD, the Margaret Liu Collins and Edward B. Collins Distinguished Professor in Neurological Surgery, showed that a small region in Wernicke’s area that has long associated with speech comprehension also becomes active before we speak, generating a sort of “blueprint” for the next word we intend to say.

“This changes the textbook view of how the brain perceives and produces speech,” said Chang, who is also a member of the UCSF Weill Institute for Neurosciences and the Joan and Sanford I. Weill Chair of the Department of Neurological Surgery. “Instead of being separate systems, the processes of hearing speech and actually speaking are very intertwined.”

The discovery could improve how surgeons map the brain to protect speech and open new possibilities for brain-computer interfaces (BCIs) that restore communication for people with brain injuries. At a more profound level, it could explain how we learn to talk as children, as well as language differences found in autism.

The study was published Nov. 19 in Science.

Updating a 150-year-old idea

The discovery has its roots in the beginnings of modern neuroscience.

In the 1870s, German neurologist Carl Wernicke observed that patients with damage to the posterior temporal lobe had problems both understanding and producing speech. He proposed that the region, now called Wernicke’s area, contained what he called the “auditory image of words” — something like a blueprint for an intended word that could ultimately be translated into the movements that are needed to say it.

But because Wernicke’s area sits within the auditory cortex, which processes sound, it came to be viewed primarily as a center for understanding speech, while speech production was assigned to other parts of the brain, said Matthew Leonard, PhD, an associate professor of Neurological Surgery and co-first author of the study.

Yet people with strokes in this area often also have difficulty saying the words they want to say, suggesting there is more going on than the textbooks indicate.

“What’s different now is that we finally have the tools and the data to see the brain at the speed and scale at which speech actually happens — millisecond by millisecond, down to individual neurons,” Leonard said.

Leonard and co-first author Alex Silva, PhD, a student in UCSF’s Medical Scientist Training Program, drew on data gathered by Chang’s lab over the past 15 years and combined it with new brain-recording experiments designed to distinguish the brain’s different roles in listening and speaking.

The approaches converged on a region of the brain called the left posterior superior temporal gyrus, or lpSTG, which is in the same general area that Wernicke linked to speech problems 150 years ago.

The researchers found that neurons in the lpSTG became active about a quarter-second before a person began saying a word, but this activity reflected neither individual speech sounds or how the mouth moved.

Instead, it was creating a blueprint — a plan that contained information, such as the word’s length, number of syllables, and how common it is in English. That blueprint then passed to a speech-planning area in the brain’s frontal lobe; and from there it went to the motor cortex, which turned it into the precise movements that are needed to speak the word.

The researchers also found that the neurons involved in hearing speech sit side-by-side with those that create the blueprint for words, sometimes even overlapping them.

Knowing precisely where these different populations of neurons lie could help surgeons map the brain more accurately and reduce the risk of disrupting speech during surgery.

How listening may help us learn to speak

The close link between hearing and speaking in this small region may also offer a clue to how humans learn to use words.

“One of the most important skills we have as infants is learning to turn the words we hear into words we can say,” Leonard said. “Having a brain area that’s important for both functions may be key to acquiring language.”

A child hears a word many times before saying it. At some point, the sound has to become a plan for speaking. The researchers think the lpSTG may be where that happens, and they plan to test the idea in future studies.

The findings could also help researchers understand language differences in autism. Roughly a quarter to a third of autistic children remain minimally verbal, even though many understand far more than they can say.

Since the study did not include autistic participants, it remains unclear what role this region plays in autism. But, Chang said, “it gives us a specific place in the brain to look, and a specific question to ask.” 

A new route to restoring speech

The lpSTG might also present a new target for BCIs that translate brain activity into speech.

Current speech BCIs determine what a person is trying to say by decoding signals from motor regions that control the vocal tract. But those regions can themselves be damaged in some people who are unable to speak.

Because the lpSTG represents an earlier stage of speech production, it could offer another route to decoding a person’s intended speech.

The researchers demonstrated that potential in two study participants with severe paralysis who could no longer speak. Analyzing activity in and around the lpSTG, the team was able to decode words the participants attempted to say, along with the speech they heard.

Because this small region contains information related to both incoming and outgoing speech, researchers hope it can eventually help support the back-and-forth of conversation — something missing from today’s BCIs, Silva said.

“What we’ve learned about the lpSTG is exciting not only because it could improve current BCIs, but because it also has the potential to greatly expand the range of patients who can benefit from them,” he said.
 

Authors: Additional authors include Patrick Hullett, Deborah Levy, Catherine Wang, Duo Xu, and Ilina Bhaya-Grossman, all of UCSF.

Funding: This work was supported by the National Institutes of Health (grants NIDCD R01DC012379 and F30DC021872), the Susan and Bill Oberndorf Foundation, the William K. Bowes, Jr. Foundation, and Aligning Research to Impact Autism (ARIA-NSNM-003), which is managed by the Coalition for Aligning Science and funded by the Sergey Brin Family Foundation.

Disclosures: Alex Silva, Matthew Leonard, and Edward Chang are inventors on a pending UCSF patent application related to this work. Chang is the co-founder of Echo Neurotechnologies, LLC.

About UCSF: The University of California, San Francisco (UCSF) is exclusively focused on the health sciences and is dedicated to promoting health worldwide through advanced biomedical research, graduate-level education in the life sciences and health professions, and excellence in patient care. UCSF Health, which serves as UCSF’s primary academic medical center, includes among the nation's top specialty hospitals and other clinical programs, and has affiliations throughout the Bay Area. UCSF School of Medicine also has a regional campus in Fresno. Learn more at ucsf.edu or see our Fact Sheet.
 

###

Follow UCSF

ucsf.edu | Facebook.com/ucsf | Twitter.com/ucsf | YouTube.com/ucsf


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.