When we hear a sentence, or a line of poetry, our brains automatically transform the stream of sound into a sequence of syllables.
But scientists haven't been sure exactly how the brain does this.
Now, researchers from the University of California, San Francisco, think they've figured it out. The key is detecting a rapid increase in volume that occurs at the beginning of a vowel sound, they report Wednesday in Science Advances.
"Our brain is basically listening for these time points and responding whenever they occur," says Yulia Oganian, a postdoctoral scholar at UCSF.
The finding challenges a popular idea that the brain monitors speech volume continuously to detect syllables. Instead, it suggests that the brain periodically "samples" spoken language looking for specific changes in volume.
The finding is "in line" with a computer model designed to simulate the way a human brain decodes speech, says Oded Ghitza, a research professor in the biomedical engineering department at Boston University who was not involved in the study.
Don't see the video? Click here.
Detecting each rapid increase in volume associated with a syllable gives the brain, or a computer, an efficient way to deal with the "stream" of sound that is human speech, Ghitza says. And syllables, he adds, are "the basic Lego blocks of language."
Oganian's study focused on a part of the brain called the superior temporal gyrus.
"It's an area that has been known for about 150 years to be really important for speech comprehension," Oganian says. "So we knew if you can find syllables somewhere, it should be there."
The team studied a dozen patients preparing for brain surgery to treat severe epilepsy. As part of the preparation, surgeons had placed electrodes over the area of the brain involved in speech.
"So then, we asked our patients to lay back, relax and listen," Oganian says.
They heard lots of sentences, including the first line of a sonnet by Shakespeare: "Shall I compare thee to a summer's day."
An analysis of the patients' brain activity confirmed earlier research suggesting that changes in volume were important to detecting syllables.
But the changes happened so fast that there was no way to tell precisely when the brain was responding to volume changes. Was the trigger the quietest point, the loudest point or somewhere in between?
So the team used computer software to slow down each sentence without changing the pitch or other characteristics.
"What we saw with the slow speech is that the neural response went up every time the speech intensity started to rise fast," Oganian says.
Those rapid increases in volume were occurring at the beginning of each vowel sound, she says. And the brain could tell whether vowel was stressed or unstressed.
So when patients heard the word summer, their brains recognized that the stress fell on the first vowel sound not the second.
Detecting this difference is important because stressed and unstressed syllables help create the rhythm of human speech, Oganian says, including poetry.