10 MAY 2007
Over the years, the mysteriously rapid way in which humans learn language has given scientists a lot to talk about.
Unlike most of the subjects in school, language is pretty easy to learn. Small
children master the basics of grammar and syntax before they even get to first
grade. Vocabulary grows over the years, of course, but the ability to 
understand
and produce grammatical sentences seems to be instilled in the brain to begin
with.
That's why some linguists, most notably Noam Chomsky, have long maintained that humans inherit language skills -- that basic, universal linguistic principles are genetically programmed into the brain. Sure, you have to learn the individual words and peculiarities of the language in the country where you grow up, but infants apparently come to language with a built-in ability to perceive how the words fit together to make meaning.
Not everybody agrees. Many experts think that language learning merely relies on mental abilities evolved for other purposes. It's kind of like the way computers, initially designed to crunch numbers, can use the same computational power to perform word processing.
Both proposals pose problems, though, say psychologists Morten Christiansen and Nick Chater. For one thing, language itself changes much more rapidly than natural selection could possibly wire biological changes into the brain. And humans develop language in many different situations, and they are too diverse for evolution to magically produce a brain module tuned to one common set of linguistic features.
On
the other hand, the idea that general brain power can simply be applied
to learning language doesn't explain everything, either. The brain works
well at learning natural language, but not so well for artificial languages.
Infants have a really hard time learning to speak FORTRAN, for instance.
So Christiansen, of Cornell University in Ithaca, N.Y., and the Santa Fe Institute in New Mexico, and Chater, of University College London, offer an alternative idea in a new paper available on the Santa Fe Institute Web site.
"Language is easy for us to learn," they write, "not because our brains embody knowledge of language ... but because language has adapted to us."
In other words, languages, like organisms, evolve to fit the human brain the way bugs and flowers evolve to fill an ecological niche. The ease with which infants learn language, and the many common features that all human languages possess, do not imply that the brain has evolved specific circuits especially for learning language rapidly. Languages may simply have evolved to be learnable.
"Language has been shaped by the brain, not the reverse," write Christiansen and Chater.
As language originated, numerous possible ways of stringing sounds together might have been tried out. But the point of language as a shared means of communicating is that everybody could understand the same set of sounds to mean the same thing. Successful systems would naturally have been those that were easiest for people to learn -- and the sooner the better. Thus over time, language would have developed with a set of easy-to-learn traits that children could easily master.
"Traits that are easy for learners to acquire and use will tend to become more prevalent; traits that are more difficult to acquire and use will tend to disappear," Christiansen and Chater observe.
If this view of language is right, aspects of the brain's learning mechanisms should be reflected in generic features of human languages. For one thing, the brain learns by discerning patterns in a complex stream of sensory input. Christiansen and Chater cite several examples where languages exhibit such patterns.
One is the presence of universal patterns of word order. In any given language, the key words in phrases (known as the "heads" of phrases) generally occur in the same location, either at the beginning or end of the phrase. English is a "head first" language, whereas in Hindi the key word typically occurs at the phrase's end.
You don't need a universal grammar wired into your head to look for key words
at the beginning or end of phrases. Rather, if the key word usually occupies
the same position, 
that
pattern will be easier to learn. Languages without this feature would
be harder for kids to learn, and therefore more likely to die out in competition
with easier-to-learn languages.
From this point of view, languages are like species: They compete to survive, and the "fittest" -- those most learnable -- win. Features making languages hard to learn are weeded out during the course of human evolution; the languages that survive share various learnable features, giving the impression of a set of "language universals" inherent in every human brain. Basically, the languages themselves have evolved to match the features that make it easy for humans to learn anything.
This view explains why language is easier to learn than, say, physics. The physical world and the laws of nature are what they are -- they cannot change themselves to accommodate the human brain's learning skills. And the physical world functions just fine without people around. Language, on the other hand, needs people. No people, no language.
And by the way: no language, no science. Language allows people to share their discoveries about patterns in nature. Sometimes the interpretations of those patterns might be mistaken, though. So when deeper insights emerge, such as those articulated by Christiansen and Chater, it's good to have an easily learnable language available to spread the word.
E-mail: tsiegfried@nasw.org
