BLADE ILLUSTRATION/JEFF BASTING
Babies learn to speak months after they begin to understand language. As they are learning to talk, they babble, repeating the same syllable (“da-da-da”) or combining syllables into a string (“da-do-da-do”).
But when babies babble, what are they actually doing? And why does it take them so long to begin speaking?
Insights into these mysteries of human language acquisition are now coming from a surprising source: Songbirds.
Researchers who focus on infant language and those who specialize in birdsong have teamed up in a new study suggesting that learning the transitions between syllables — from “da” to “do” and “do” to “da” — is the crucial bottleneck between babbling and speaking.
“We’ve discovered a previously unidentified component of vocal development,” said the lead author, Dina Lipkind, a psychology researcher at Hunter College in Manhattan. “What we’re showing is that babbling is not only to learn sounds, but also to learn transitions between sounds.”
The results provide insight into language acquisition and may eventually help shed light on human speech disorders. “Every time you find out something fundamental about the way development works, you gain purchase on what happens when children are at risk for disorder,” said D. Kimbrough Oller, a language researcher at the University of Memphis, who was not involved in the study.
At first, however, the scientists behind these findings weren’t studying human infants at all. They were studying birds.
“When I got into this, I never believed we were going to learn about human speech,” said Ofer Tchernichovski, a birdsong researcher at Hunter and the senior author of the study, published online on May 29 in the journal Nature .
He and Ms. Lipkind were teaching young zebra finches living in soundproof boxes to switch the order of syllables in their songs. Birdsong is composed of discrete song syllables (“chirp A,” “chirp B”) similar to syllables in human speech. The researchers piped in the song of an adult male zebra finch to teach young birds one song (A-B-C), then piped in a new song that required the birds to use the same syllables in a different order (A-C-B).
The birds could learn the new song only after a huge effort — by practicing thousands of times a day for weeks. The fact that the new song required the birds to simply switch syllables around suggested the roadblock was learning the transitions.
Their collaborator, Kazuo Okanoya, then at the Riken Brain Science Institute in Japan, observed the same effect in Bengalese finches, which are able to sing much more complex songs. These birds were living in a more natural environment, flying around in large aviaries full of other finches; this helped confirm that learning individual transitions is part of a natural developmental process.
Mr. Tchernichovski and Gary Marcus, who studies infant language learning at New York University and who helped design the study, discussed the results. Could the difficulty learning transitions in songbirds hold true for human infants?
By analyzing an existing data set of recordings of infant babbling, they found that as babies introduce a new syllable into their repertory, they first tend to repeat it (“do-do-do”). Then, like the birds, they begin appending it to the beginning or end of syllable strings (“do-da-da” or “da-da-do”), eventually inserting it between other syllables (“da-do -da”).
As with the birds, learning the transitions between new syllables and old syllables is a painstaking process for babies. That could help explain why children continue to babble even as they begin to understand language, making the gap between comprehension and speech a little less mysterious.
“This result changes what we think kids are doing while babbling,” Mr. Marcus said.
This high-profile study is also likely to rekindle a touchy debate between human language researchers and birdsong researchers: Can we really use birdsong to learn about human speech?
The two might seem to have little in common. Birds and humans are evolutionarily distant, and birds use song differently from the way we use language. But in recent years, researchers have found many surprising parallels.
At the genetic level, birds and humans share molecular building blocks — including the FOXP2 gene, which made a big splash a decade ago when it was identified as the gene responsible for one human family’s mysterious speech disorder.
At the neurological level, we seem to share brain structures crucial for song (in birds) and speech (in humans).
And at the behavioral level, birds and humans both use “syllables” strung together into phrases; both “babble” during a critical learning period; and both are “vocal learners” — birds learn to sing from a male tutor bird, and children learn to speak from their parents.
With these parallels in mind, more researchers are turning to birdsong as a model for human speech, which is notoriously difficult to study. Vocal learning is rare in the animal kingdom; not even the primates that are our closest evolutionary relatives are vocal learners.
Songbirds, however, are vocal learners. And researchers can do experiments with birds that they can’t do with human infants. As Ms. Lipkind put it, referring to the method used to study zebra finches, “You can’t put babies in soundproof boxes.”
Sarah Woolley, a birdsong researcher at Columbia University who was not part of the research team, agreed that the links were intriguing. “No one is saying birdsong is language,” she said. “But there are so many parallels. We’ve got this opportunity to model vocal learning by testing how the rules work for birds and making predictions about how humans learn.”
Some human language researchers are impressed, too. “My hat’s off to them,” Mr. Oller said of the new study’s authors. “They aren’t even babbling researchers, and they came up with a procedure that eluded all of us. Assuming they are correct, they’ve made a serious contribution to the babbling literature.”
But others question whether birdsong can ever really tell us much about human language.
“The parallel doesn’t go terribly far,” said Marilyn Vihman, a linguist at the University of York in England, echoing the sentiments of other experts. “The really important difference between birdsong and language is that language ends up having meaning.”