How Do Babies Learn to Talk?

May 18, 2011— -- One of the unique talents that mark us as humans is our ability to express our thoughts to others in a way that can be understood, even if the subject is complex and abstract. Thus it is not surprising that scholars have struggled for centuries to understand how an infant can learn language in an incredibly brief period.

It begins with a single word, like "Da" or "Ma." A few days later, it seems, they are asking for the keys to the family car.

No other animal on the planet can do that. Sure, your dog can read your mind, but can he explain evolution? How can humans do it so easily?

Plato thought he had the answer. The infant comes with a built-in play book. Learning to speak is innate, like learning to walk.

All these years later, scientists are still not sure if if Plato got it right, but he has been supplanted by new explainers, especially MIT's linguist and political activist Noam Chomsky, who argues that infants are born with some understanding of syntax, which makes it easier for them to know if the language structure of grammar is right or wrong.

Research Suggests Humans Have Built-in Bias for Learning Language

"The idea that the human brain has some unconscious knowledge of syntax and grammar is a hypothesis made by Chomsky in order to explain why language learning seems so easy (at least in the beginning,)" said Jennifer Culbertson, a cognitive scientist whose research results are consistent with Chomsky's theory. "Children don't take very long to learn it."

However, despite her own findings, she quickly adds that Chomsky's hypothesis "is under fire. There's no consensus" among experts. Learning how language is acquired is the "holy grail" of linguistics," she said.

Her work, conducted while she was working on her doctorate at Johns Hopkins University, suggests that humans come equipped with what she calls a "bias," making it easy for a child to learn a language that conforms to a template embedded in the embryonic brain, but very difficult to master a language that doesn't. It doesn't fully answer the question of whether language is innate, but it's a shred of evidence in a field that has very little.

During her research at Johns Hopkins, Culbertson and Prof. Paul Smolensky ran 65 persons between the ages 18 and 30 through a series of experiments. They wanted to see if they could find evidence that the participants had some built-in bias that would make it very difficult for them to learn a language that is unlike any spoken on Earth, but might be very handy on a distant planet.

Languages on Earth fall into four grammatical categories involving the placement of modifiers (either adjectives or numbers) before or after the modified noun. In English, the modifier precedes the noun (a tall man, not a man tall; three persons, not persons three.) Cherokee and Greek languages follow the same pattern.

In many other languages, like Russian and Thai, the modifier follows the noun. That's the most common pattern on the planet. Basque, modern Hebrew and Welsh place the number before the noun, as in English, but the adjective follows the noun, as in Russian. It is extremely rare for a language to follow both patterns, sometimes placing the adjective or the number before, and sometimes after, the noun.

Research Will Likely Not End Argument Among Linguists

Culbertson reasoned that if the participants had a built-in bias for recognizing where the modifiers belonged, they would find it very difficult to learn a language that didn't follow the rules. In the experiment, the language for that distant planet fell into that category, willy-nilly mixing the arrangement of the modifiers. As expected, the participants found that an extremely challenging language.

But how can they be sure that since the participants all spoke English, it was their native tongue, not an innate bias, that made ET's language so hard?

Because, Culbertson said, they found it just as easy to learn new phrases in languages that do not conform to the English pattern (of modifier before noun,) but they had a terrible time with the fourth category, the inconsistent mixing of modifiers which is so rare on Earth.

The research is not likely to end the argument among linguists, and it needs to be replicated by others before it can have much impact. Culbertson concedes it's inconvenient that scientists have to try to pry into the unborn mind by studying adults, but it's pretty difficult to conduct this kind of research with infants.

So she's moving on to the next level in a new assignment as a postdoctoral fellow at the University of Rochester. She is conducting the same experiments with children, 5 to 7 years old.

Maybe they can tell us if Plato had it right. And maybe not.