News Release

Scientists Propose Theory for Brain's Mastery of Language

The idea, known as "optimality theory," is explained in a paper being published in the March 14 issue of the journal Science. It could be used to learn more about the mysteries surrounding how the brain so readily understands the grammatical complexities of language. If the brain already contained a basic underlying knowledge of language, then learning the grammar of a specific language might only require a kind of fine-tuning mechanism.

"Previous ideas have said that grammar is just a bunch of rules that you have to follow, and that different languages have different rules," said Paul Smolensky, a professor of cognitive science at The Johns Hopkins University. "What we are saying is that different languages don't have different rules. They actually don't have rules. What they have are precise notions for what makes a sentence good or bad, and they all have the same notions."

But those notions have different relative importance in different languages; as a result, distinct sequences of sounds are possible in specific languages.

Optimality theory was conceived by Smolensky and Alan Prince, a linguist and cognitive scientist at Rutgers University, who co- authored the Science paper with Smolensky.

"The idea is that sentences in Italian and English look very different, but, in fact, underlying that apparent difference are the same criteria in all languages for what makes a good sentence," Smolensky said. "In one language, certain criteria are more important than others, and they cause the grammatical sentences to look a certain way. In another language, different criteria are more important, causing the grammatical sentences to look quite different."

Those criteria are responsible for governing the sequence of sounds that make words and the proper sequence and grouping of words for good sentence structure, he said.

The theory might help to explain a fundamental mystery of language; its grammar and rhythm are complex, yet research has shown that babies are somehow able to begin learning these aspects of their particular language when they are only 2 months old, and possibly even as fetuses.

Perhaps their brains already possess a basic language understanding, knowledge of the criteria determining for all languages what makes for a good word or sentence, Smolensky said.

"We can see rather precisely how that knowledge would allow a language learner to rapidly figure out what's going on in their language, which has always been a problem for linguistic theory," Smolensky said. "The knowledge that has been assumed to be necessary in order to be competent in a language was so hard to acquire that it was very difficult to come up with a convincing theory for how a young child could actually do it. We now have, with optimality theory, some very concrete proposals about how language acquisition is possible."

Smolensky and Prince have written a book about their approach, entitled "Optimality Theory: Constraint Interaction in Generative Grammar," to be published by the MIT Press.

"Some cognitive scientists have looked at the brain not the way biologists tend to, but from a more computational perspective," Smolensky said. "That is, the brain looks like a very different kind of computing device than the standard computer. In fact, it appears to compute quite differently from a standard computer, and previous notions of grammar have been much more based on ideas of calculation that are associated with traditional computing. We are trying to understand grammar, given the kind of computing that brains seem to do."

EMBARGOED FOR RELEASE AT 4 P.M. EASTERN TIME ON THURSDAY, MARCH 13, 1997