The New York Times > Is It Dutch? Japanese? Why Not Ask the Rat?

Is It Dutch? Japanese? Why Not Ask the Rat?
By NICHOLAS BAKALAR

Published: January 11, 2005

If you talk to a rat, you will not get an answer. But a team of Spanish neuroscientists has shown that a well-trained rat may be able to determine what language you are speaking.



Every language has distinctive rhythms and intonations, and awareness of them is an important step in acquiring language. Only humans can learn to speak, but it has been demonstrated that tamarin monkeys, like newborn human infants, can distinguish the unique rhythms of a language even though meaning escapes them.

In other words, they know when someone is speaking their language, even though they have no idea what is being said. Researchers have theorized that this ability extends to other mammals as well, but until now no nonprimate has ever demonstrated the capacity.

In the new study, led by Juan Toro, a doctoral candidate at the University of Barcelona, researchers found that rats trained in either Dutch or Japanese appeared able to distinguish the two languages. The rats were trained by having them listen to synthesized sentences in the languages. Dutch and Japanese were chosen because of their vastly different rhythms. The sentences had no semantic content, but were intended to reproduce the rhythms of the language without using any real words.

This simplified form of language, when spoken in a synthesized voice, leaves only rhythm as a cue, eliminating complicating factors like semantic content or the quality of the voice of a particular speaker.

For the Dutch group, the rats were rewarded with food only when they pressed a lever after hearing Dutch sentences. The Japanese group was rewarded only after hearing Japanese sentences. Eventually, both groups learned to press the lever only when hearing a sentence in their own languages.

Next, the rats listened to four synthesized sentences in the language they had not learned. When the Dutch mice were presented with Japanese sentences, they showed no recognition; when the Japanese mice were presented with Dutch, they were similarly baffled. But when presented with a sentence in their own languages, even a sentence they had never heard before, the rats recognized the characteristic rhythm and pressed the lever correctly.

The researchers said the rats appeared to have generalized some of the rules of their language and, at least in this limited way, were able to understand an entirely new sentence, a distinctive mark of language acquisition. When the researchers played the same sentences with the tape running backward, the rats were unable to understand what language was being spoken - exactly what happens with tamarins and human infants.

Rats, of course, have limitations. They had considerably more difficulty in telling one language from another when listening to normal speech, especially when uttered by different speakers, the researchers found. The multiplicity of cues in ordinary conversation - intonation, the speaker's sex, pitch and so on - utterly confused them.

Human infants have some difficulty with different voices, too, but they quickly overcome it, learning to recognize their own language no matter who is talking and however varied the pitch and intonation. 'What these results suggest,' Mr. Toro said in an e-mail interview, 'is that we share with other animals the ability to perceive some regularities, such as rhythm, in the speech signal. This is interesting because several studies with human infants have shown that these regularities may open the door to language acquisition.'

Does this mean rats and monkeys have the potential to understand human speech? No, said Mr. Toro. But he added, 'Even though human language is special and does not seem to have parallels in the communicative systems of other species, some basic abilities we use for acquiring it may be present in other animals.'"