How we learn to speak is a mystery. Myrna Gopnik claims to have found a gene which transmits bad grammar. Over 100 ago Charles Darwin speculated that language was an "instinct". Evidence suggests that he was right - that the ability to acquire language is part of human beings' basic biological equipment. One source of evidence about this "language instinct" is a developmental disorder of language which affects the ability to build the kinds of grammars that ordinary children build automatically and unconsciously.

Studies have shown that this disorder aggregates in families and that identical twin sufferers are significantly more alike than afflicted fraternal twins - a finding that strongly suggests an underlying genetic factor. This does not mean that there is a gene for grammar, however. The interaction between genetic factors, the intra-uterine environment, neurological development and experience is highly complex. For example, if this gene interferes with normal neurological development, as evidence seems to suggest, then it is likely that the same disorder may, in some individuals, impair other cognitive functions. It might even turn out that in certain circumstances, this same genetic factor might affect neurological development in areas of the brain that control other cognitive functions, while at the same time sparing the language centres altogether.

Some sufferers have trouble with temporal auditory processing. Some have difficulties with articulation or reading or symbolic play, or building with blocks, or hopping on one foot. A language impaired individual can have a performance IQ of 111, while that of his unimpaired relative is 84. It is certainly not true that these other problems inevitably lead to language disorder because some people with these other problems have no language problems. The question is whether these other impairments cause the language deficit or whether they merely co-occur with it.

Over the past 15 years our research team has been studying this language disorder in detail. We have looked at it in English, French, Greek and Japanese, languages which have different grammars. By comparing data across these languages we can determine which problems are the result of the peculiarities of a particular language and which problems are more general.

³ÉÈËVRÊÓƵ

We have now recruited more than 100 families and have tested about 75 impaired subjects, 75 family controls and 250 non-family controls. Though the language impaired individuals are able to use language to communicate concepts about number and time, the subtleties of grammar which encode these concepts, such as tense and plural marking, which four-year-olds acquire with no problem, seem to be beyond their abilities. For example the language impaired subjects in all four languages have particular difficulties in producing inflected verbs as in: "The boy always cries. Yesterday he ....."

Their problems are more striking when the word that they are asked to inflect is a novel word. For example, if they are shown pictures of an imaginary creature and then told: "This is a zoop. These are ....," they are not able to produce the plural form. This suggests that they are not able to form the kinds of productive rules that allow us to produce, automatically, new forms like "faxes" or "e-mailed".

³ÉÈËVRÊÓƵ

It does not matter whether the stimuli are pictures, words, sentences; auditory or written. Because they make the same sorts of mistakes no matter how the forms are presented or how they are required to respond to them we think that the explanation of their disorder cannot be that they merely have a physical or perceptual problem. In addition, they have problems with irregulars in English and with inflections like "ma****a" in Japanese which cannot be explained in terms of hearing or pronunciation. The only plausible explanation for all of our cross linguistic data is that something is fundamentally wrong with their ability to learn and use all of the grammatical intricacies of their native language.

Though we think that they cannot acquire their native language in the same unconscious, automatic way that most four-year-olds do, they have learned how to compensate for their problem in at least two different ways. One way is to learn inflected words like "walked" in the same way that the rest of us have to learn irregular past tenses like "went" - by memorising them. We can tell the difference between words that are memorised and those that are produced by rule by seeing if the frequency of the word has an effect on whether the subject gets it right. The more frequent an irregular verb is, the more likely a speaker is to get it right. "Went" is easier than "slew". This is not the case for those that come from the regular rule. "Walked" is not easier for most speakers than "juggled". Unlike unimpaired speakers the language impaired subjects are much more likely to get a form derived from a regular rule right if it is frequent than if it is infrequent.

A careful phonetic analysis of the forms that they produce, both in English and in Japanese, shows that they can also learn to use conscious, explicit rules as an imperfect surrogate for the unconscious, implicit rules that are used by unimpaired speakers. For example, when a subject was required to produce the plural form of novel words she said: "These are the kind where you have to add an "s" to it - all the time". Then when given the novel word "tuss" she did just that and produced "tusss" with a much longer "s" sound; when given "tob" she produced "tob-s" instead of "tob-z". Almost all the people with this language deficit produce forms that show that they are using an explicit rule. What is interesting is that very often the form that they produce by these explicit rules is harder to pronounce than the right one. This is not all that is wrong with the language of these subjects. They have problems with organising the sounds of language and they have some problems with syntax. The evidence suggests that their difficulties are caused by an impairment in their "language instinct". They have learned to compensate for this, but only in a very imperfect way.

Though we have been working on this problem for a long time there are still more questions to be answered. We would like to know more about the details of the language problem in other languages; precisely which genetic factors are involved; and how they affect the developing brain.

³ÉÈËVRÊÓƵ

Myrna Gopnik is professor of linguistics at McGill University, and director of a research project into the linguistic, genetic and neurological proponents of language impairment.