Specialized neurons help the brain detect where words begin and end

Why is it so easy to hear individual words in your native language, but in a foreign language they run together in one long stream of sound?

Researchers from UC San Francisco have begun to answer that question with two complementary studies that show how the brain learns the sound patterns of a language until it recognizes where one word ends and the next begins.

When we speak naturally, we don't put pauses or "spaces" between words, yet fluent speakers effortlessly perceive them. For years, researchers assumed it was the brain areas that give meaning to speech that were figuring out the boundaries between words.

The new studies focus on a different brain region, called the superior temporal gyrus, or STG. Until now, it was thought only to handle simple sound processing, like identifying consonants and vowels.

The new studies show the STG contains neurons that learn to track where words begin and end over years of experience hearing a language.

This shows that the STG isn't just hearing sounds, it's using experience to identify words as they're being spoken. This work gives us a neural blueprint for how the brain transforms continuous sound into meaningful units."

Edward Chang, MD, Chair of Neurological Surgery

Chang led the two studies, which were published Nov. 7 in Neuron and Nov. 19 in Nature and supported by the National Institutes of Health.

Rapid reset for the next word

In the Nature study, researchers recorded brain activity from 34 volunteers who were being monitored for epilepsy. Most spoke either Spanish, Mandarin, or English as their native language. Eight were bilingual, but no one spoke all three languages.

Participants listened to sentences in English, Spanish, and Mandarin - languages that were both familiar and unfamiliar to them.

The researchers used machine learning models to analyze patterns and found that when participants heard their native tongue or a language they knew, the specialized neurons in the STG lit up. But when participants heard a language, they didn't know, the neurons failed to light up.

"It explains a bit of the magic that allows us to understand what someone is saying," said Ilina Bhaya-Grossman, a PhD candidate in the UCSF-UC Berkeley Joint PhD Program in Bioengineering who is the first author of the Nature study.

The Neuron study showed how these specialized neurons detect the beginnings and endings of words.

Given that fluent speakers utter several words per second, these neurons must rapidly reset to take note of the next word.

"It's like a kind of reboot, where the brain has processed a word it recognizes, and then resets so it can start in on the next word," said Matthew Leonard, PhD, Associate Professor of Neurological Surgery, the co-first author with postdoctoral scholar Yizhen Zhang, PhD.

Chang, who along with Leonard is a member of the UCSF Weill Institute for Neurosciences, said the studies clarify why injury to certain regions of the brain can impair the ability to comprehend speech even when a person's hearing is intact.