News Release

The findings came after two years of research on a 24-year-old man who suffered brain damage, possibly resulting from encephalitis. If the brain-damaged subject of the study had spoken English instead of Italian, scientists might not have made the discovery; the findings are directly related to the fact that Italian, like a number of other languages, assigns feminine and masculine gender to nouns, said William Badecker, an associate research scientist in the Johns Hopkins Department of Cognitive Science.

After two weeks in a coma, the man seemed to recover completely, speaking fluently and retaining most of his language capacity. But he had a pronounced anomia, a condition in which the person is unable to retrieve words.

For example, one-third of the time he was not able to remember the names of familiar objects when shown photographs of them. He also was not able to say anything about a word's form, such as whether it rhymed with other words or how many syllables it had. Scientists were startled to discover, however, that even though the man seemed to know nothing about the word's form, nearly 100 percent of the time he knew whether it was masculine or feminine in gender.

A scientific paper on the work, which refers to the man by the pseudonym Dante, will be published in the November issue of the international journal Cognition. The research provides important evidence for the most widely accepted theory of how the brain turns thought into words.

The mind does not simply retrieve information in one step, like a person looking up words in a dictionary. Rather, it apparently uses a complicated network of distinct modules to transform thought into language. In the Italian language, feminine words typically end in "a" (such as casa, the word for house), and masculine words typically end in "o" (such as caso, the word for fate). But there are exceptions. "Problema" (the word for problem) is masculine, even though it ends in a, and "mano" (the word for hand) is feminine, even though it ends in o.

However, regardless of whether a word fit the typical gender pattern, Dante could identify its grammatical gender, even though he could say nothing about what the word sounded like.

"To me, that was very exciting, because it immediately brought to mind that model, and the kinds of predictions that the model makes about the way the language system should break down," Badecker said.

The model maintains that the mind turns conceptual representations into linguistic forms in two stages, referred to as the lemma and lexemic levels.

When a person is shown a photograph of an animal, for example, the brain selects a specific "lemma" for that particular word, essentially retrieving a package of information about the word, but without actually naming the word. The key point is that, according to this model, the lemma includes the semantic and grammatical features of the chosen word, including its gender, if the language happens to assign gender to nouns, as Italian does. But this package of information still has no word form. The lemma level provides an "address" within the lexeme network, essentially saying, "Go here to get the form," and the proper word is retrieved, Badecker said.

This case is different than other studies of anomia, where people could not remember words even though they could access many other details about the objects they could not name. For example, if shown a picture of a penguin, they could not come up with the name of the animal, but they could come up with many general facts and descriptions about it. Those examples, however, do not provide clear evidence for the two-stage model because they fail to show that the person has actually selected a lemma for the object in the picture; they do not prove that the patients have accessed grammatical information specifically related to that word.

In Dante's case, there is a clear breakdown somewhere between the lemma and lexeme levels, since he is able to get access to the grammatical gender but unable to retrieve words; he has accessed the lemma but then he cannot get to the appropriate addresses where the specific word forms are located.

The fact that Dante spoke Italian was an essential component of the research. "This is what allowed us to find the evidence for the two-stage theory," Badecker said. Badecker wrote the Cognition paper with Raffaella Zanuttini, who is now on the faculty of linguistics at Georgetown University, and Michele Miozzo, a graduate student at Harvard University. Miozzo was a graduate student at Hopkins and Zanuttini was a postdoctoral fellow at the university when they started working with Badecker on the research in 1990.