Recounting a Flawed Past

Scientists have found evidence that the brain has separate centers for understanding the rules of grammar and for understanding exceptions to the rules. The provocative findings, reported in today's issue of Nature, add fuel to a long-standing debate over whether language ability arises from discrete neural structures or from a diffuse nerve network in the brain.

People with damage to the left side of the brain often lose their ability to use grammar, word endings, and inflections that shade meaning. For instance, the report relates how one such patient described a picture by saying, "Kitchen room ... lady washing ... accident sink flow ... water. ..."

But aphasics, as such patients are called, often vary in the extent of speech dysfunction. To explore this systematically, cognitive scientists William Marslen-Wilson and Lorraine Tyler of the University of London tested the ability of three aphasic patients to recognize the past participles of verbs: regular ones like "jump", which becomes "jumped," and irregulars like "bring," which becomes "brought." Surprisingly, the duo found that two of three patients had difficulty only with regular verbs; the third patient struggled only with irregulars.

The findings suggest that regular and irregular verbs are processed in two different brain centers. This may arise because forming the past tense of a regular verb requires understanding a set of rules--a grammar--while the irregular exceptions to the rule require a reference table--a lexicon. Damage to one area, therefore, might leave the other unharmed. "This and other evidence suggests that there are distinct sets of neural structures for regular and irregular verbs," says Michael Ullman, a cognitive neuroscientist at Georgetown University in Washington, D.C. But because the patients in the study had very large lesions, the scientists were unable to pinpoint the location of the neural structures. "The big weakness of [the Nature paper] is that they're unable to localize the source of the lexicon and grammar," Ullman says.

Other experts argue that the aphasics' different verb patterns don't necessarily arise from damage to different neural structures. They prefer the "connectionist" model, in which the brain's language centers are viewed as a homogeneous neural network that recognizes patterns. Connectionism "doesn't require you to invoke separate modules" for lexicon and grammar, says James McClelland, a cognitive scientist at Carnegie Mellon University in Pittsburgh. He argues, for instance, that the large sound changes distinguishing present and past tenses of irregular verbs--for example, "bring" changing to "brought"--trigger other brain regions, separate from the language center, that recognize the sound of syllables. According to this model, lexicon and grammar would be processed in a united language center.

Some scientists, however, are ready to nail the coffin shut on connectionism. Says cognitive scientist Steven Pinker of the Massachusetts Institute of Technology, "Taken with other evidence, this study helps show, in my view, that the connectionist model doesn't work."