Le chat sat on the mat

For those who are multilingual, switching between languages is often unconscious. But, asks Geoff Watts, what triggers the switch and what happens when the controlling mechanism goes wrong?

Digging to America , the recent novel by Anne Tyler, features a family of Iranian immigrants living in Baltimore. Many of its members, including one of the book's main characters, Maryam, are bilingual. Reflecting on the speech of her relatives, Maryam has this to say: "They'd be rattling along in Farsi, and then some word borrowed from America, generally something technical like 'television' or 'computer', would flip a switch in their brains and they would continue in English until a Farsi word flipped the switch back again."

Elsewhere, Maryam observes that these sudden changes of language are common among her fellow Iranians. But if she - or Anne Tyler - thinks that such "code-switching", to use the technical term, is unique to people from Persia, she's wrong. In India, for example, you hear it all the time in the speech of the educated classes; and it's equally prevalent among South Asians living in the UK. And there are countless other instances from around the world. But what flips that switch? And how much is known about the control of speech, conscious and otherwise, among people who are bilingual?

The changes themselves have long fascinated students of language and sociology, not least because they are so various. They may occur at the end of a sentence or in the middle; the switch-over may be sustained for many sentences, or last no longer than a phrase or even a single word; and it may occur with almost any frequency. In short, anything can happen. Code-switching also intrigues scientists. What mechanism serves, most of the time, to keep the languages separate, so avoiding the descent into a Babel-like confusion?

David Green, a reader in psychology at University College London, works on language, cognition and bilingualism. As he points out, conscious switching is entirely normal and accepted.

"The change may reflect a particular emphasis that one language gives, or may signal solidarity with a person. Or the other language may be more suited to a particular topic."

Occasionally, though, the change is unintentional. Green offers the example of a trilingual person having a bilingual conversation with two colleagues and then inadvertently saying goodbye in a third language that neither colleague understands. An inadvertent flip-over of this kind might be triggered when the brain routinely makes some kind of link between a particular action - such as saying goodbye - and a particular language.

But all these circumstances are very different from those in which speakers switch language as a consequence of brain damage - of which more shortly.

Last year, along with colleagues in UCL's Wellcome Department of Imaging Neuroscience and elsewhere, Green reported some illuminating experiments on language control in people fluent in more than one tongue. The work was prompted by earlier and initially puzzling research showing that proficient bilinguals seem to use the same regions of their brain irrespective of which language they are listening to or speaking. This had demolished any naive idea that your " Oui, monsieur " was stored in one region of the brain, your " Si, signor " in another. But it left unanswered the question of how bilinguals can move with such facility from one language to another - and, having selected a language, stay in it.

To explore this, Green and his colleagues studied three groups of bilingual subjects: two proficient in English and German; one in English and Japanese. In a series of experiments they presented them with pairs of words, the first of which (the "primer") they were told to ignore. Some word pairs were related in meaning (bathtub and shower, for example); some were not (spoon and shower). They also added language to the mixture. Related pairs could be in the same language (trout and salmon in English; Forelle and Lachs in German) or in different languages (trout and Lachs; salmon and Forelle ).

The object of the study was to see how the priming word affected the processing of the second or target word. By doing the experiments with and without the use of different languages, they were able to detect not only the brain's response to meaning, but also its sensitivity to language change. In one set of experiments, for example, they used two forms of brain-scanning technology to see if the different types of word-pairing had differing effects on brain activity. The part of the brain that began to reveal itself as distinctive in these experiments was a region known as the left caudate. This, they found, responded in one way when the word pairs were related and in the same language, but in another when the words in each pair were in different languages.

This, then, appeared to be the portion of the brain responsible for handling a change in language. But these experiments didn't, in themselves, provide an entire explanation. As the researchers said when reporting the work in the journal Science , their findings were compatible with two quite distinct mechanisms. It could have been that the two languages were processed by different populations of nerve cells occupying the same region of the brain. In this case, the left caudate would be serving to recognise the language being heard and activating the appropriate nerve cell population. The alternative mechanism would be that all language processing is done by the same nerve cells - in which case the role of the left caudate is simply to control the process by signalling that a change of language has occurred.

To choose between these alternatives, Green and his colleagues looked at studies of what happens to bilingual patients who have suffered damage to the left caudate. They reasoned that if there were different populations of nerve cells within it for the different languages, any damage should impair both languages equally. If, on the other hand, its role is that of control, the patient would be more likely to have difficulty in selecting the language being spoken or monitoring the language being heard.

The results of neuropsychological studies of caudate-damaged individuals, they say, point to the latter alternative. They quote the case of a trilingual patient who could still understand all three of her tongues but "spontaneously and involuntarily switched from one language to another".

Is this control system unique to bilinguals? Green thinks not. Commenting on the Science paper, he conjectures that dealing with one language rather than another is comparable to changes that have to be handled within a language - as, for example, when talking about the same topic to an adult and then to a child. "From this point of view there is no reason to assume that bilinguals are using any system that monolinguals don't also have."

At least two other groups of clinicians have described involuntary changes in language during interventions involving the brains of conscious bilingual patients. These took the form of magnetic stimulation, anaesthesia of one half of the brain and prodding with an electrode. The last of these featured a man fluent in French and Chinese. When asked to count, he changed language each time the current was turned on.

In addition to these controlled interventions, we have the unsought and more devastating whims of nature. A stroke may impair speech; what happens if the victim is bilingual? Typically, says Green, both languages are recovered at the same rate and to the same extent. But not always. "People have debated why this is. My view would be that it happens because these people have a problem of control. They can't access representations (of language) that are still there." Circuitry - in the left caudate, perhaps - that should be allowing them to switch back and forth has in effect become jammed in one position.

Back then to Maryam in Digging to America . The language switching that Tyler's character describes is prompted by certain categories of word. It's a pattern familiar to Green. "These are sometimes known as trigger words," he says. Tyler's "switch" would, presumably, equate to nerve circuitry that includes the left caudate.

The Holy Roman Emperor Charles V was not just bilingual, but multilingual. "I speak Spanish to God, Italian to women, French to men, and German to my horse," he is supposed to have said. One can only hope that his left caudate was up to the task.