Most scientists agree that human language is qualitatively different from the forms of communication employed by other animals. However, when it comes to pinpointing just exactly what separates us from them, even linguists can get tongue-tied. It’s difficult to single out the one skill humans have, which explains our unique ability for language. After all, dogs understand their masters’ commands, whales sing in regional dialects and Koko the gorilla famously mastered rudimentary sign language.
Now scientists have trained songbirds to learn a certain type of grammar thought to be solely the province of humans. This discovery undermines the idea of our uniqueness, but it may help explain how we first evolved language.
“It may be that there is no magic bullet, no one thing that gives humans the capacity for language,” said Timothy Gentner, assistant professor of psychology at University of California, San Diego, who led the bird-training study. “We might just be superstars at enough different abilities that other animals can also do, but the combination gives us this unique ability for language.”
Gentner’s research, published in the April 27th issue of Nature, found that European starlings were capable of identifying songs that had been created using recursive “center-embedded grammar,” which MIT linguist Noam Chomsky and other eminent members of the field argue is a uniquely human ability.
“Center-embedded grammar was thought to be unique to humans before this,” said Gentner. “In fact, a study just asked this exact question using non-human primates, but couldn’t get the tamarins they were working with to learn it.”
Center-embedded grammar refers to the ability to insert more information into the middle of an expression. For example, the sentence, “The scientist studied birds” can be modified to read “The scientist, who had a degree in ornithology, studied birds.” In theory, this process could continue indefinitely and still produce grammatical sentences, although in practice the strings quickly become impractical.
To test for faculty with center-embedded grammar in starlings, Gentner and his colleagues created short songs of different “warbles” and “rattles.” After tens of thousands of trials with each bird over the course of several months, the researchers were able to teach them to identify complex grammatical songs.
Gentner sees the discovery as a valuable step in figuring out what allowed humans alone to develop language.
However, several respected linguists are more hesitant about the findings.
Ray Jackendoff, co-director of the Tufts University Center for Cognitive Studies, wonders whether the birds may be recognizing the songs simply by counting the number of warbles and rattles, rather than through any real capability for recursive grammar.
“The starlings learned to recognize structures made out of their own kind of sounds—I don’t think that’s a trivial accomplishment,” he said. “But I don’t think they’re necessarily learning a recursive grammar.”
Chomsky himself dismissed the implications for the study of language in less equivocal terms. Speaking to LiveScience, he said, “The article is based on a fundamental mathematical error,” and that it, “has nothing remotely to do with language; probably just with short-term memory.”
For his part, Gentner believes that even if the birds are identifying the songs by counting or some other means, they still demonstrate an ability previously believed to be unique to humans.
“It is the entry into this class of patterning rules that is remarkable and previously thought beyond the realm of non-human abilities,” he wrote via e-mail, responding to his study’s critics. “Some linguists may find this trivial, but that is a bit like saying apes use tools, but only the trivial kind that lack the sophistication of, for example, a tri-square or laser-level.”
Originally published May 1, 2006