and one son: all save one daughter were affected and they in turn had between them twenty-four children, ten of whom have the condition. This family has got to know the psychologists well; rival teams besiege them with a battery of tests. It is their blood that led the Oxford team to the gene on chromosome 7. The Oxford team, working with the Institute of Child Health in London, belongs to the 'broad' school of S L I , which argues that the grammar-deficient skills of the K
family members stem from their problems with speech and hearing.
Their principal opponent and the leading advocate of the 'grammar theory' is a Canadian linguist named Myrna Gopnik.
In 1990 Gopnik first suggested that the K family and others like them have a problem knowing the basic rules of English grammar.
It is not that they cannot know the rules, but that they must learn them consciously and by heart, rather than instinctively internalise them. For example, if Gopnik shows somebody a cartoon of an I N S T I N C T 9 9
imaginary creature and with it the words 'This is a Wug', then shows them a picture of two such creatures together with the words 'These are . . .', most people reply, quick as a flash, 'Wugs'. Those with SLI rarely do so, and if they do, it is after careful thought. The English plural rule, that you add an 's' to the end of most words, is one they seem not to know. This does not prevent those with SLI knowing the plural of most words, but they are stumped by novel words that they have not seen before, and they make the mistake of adding 's' to fictitious words that the rest of us would not, such as 'saess'. Gopnik hypothesises that they store English plurals in their minds as separate lexical entries, in the same way that we all store singulars. They do not store the grammatical rule.7
The problem is not, of course, confined to plurals. The past tense, the passive voice, various word-order rules, suffixes, word-combination rules and all the laws of English we each so unconsciously know, give SLI people difficulty, too. When Gopnik first published these findings, after studying the English family, she was immediately and fiercely attacked. It was far more reasonable, said one critic, to conclude that the source of the variable performance problems lay in the language-processing system, rather than the underlying grammar. Grammatical forms like plural and past tense were particularly vulnerable, in English, in individuals with speech defects. It was misleading of Gopnik, said another pair of critics, to neglect to report that the K family has a severe congenital speech disorder, which impairs their words, phonemes, vocabulary and semantic ability as well as their syntax. They had difficulty understanding many other forms of syntactical structure such as reversible passives, post- modified subjects, relative clauses and embedded forms.
These criticisms had a whiff of territoriality about them. The family was not Gopnik's discovery: how dare she assert novel things about them? Moreover, there was some support for her idea in at least part of the criticism: that the disorder applied to all syntactical forms. And to argue that the grammatical difficulty must be caused by the mis-speaking problem, because mis-speaking goes with the grammatical difficulty, was circular.
IOO G E N O M E
Gopnik was not one to give up. She broadened the study to Greek and Japanese people as well, using them for various ingenious experiments designed to show the same phenomena. For example, in Greek, the word 'likos' means wolf. The word 'likanthropos'
means wolfman. The word 'lik', the root of wolf, never appears on its own. Yet most Greek speakers automatically know that they must drop the '-os' to find the root if they wish to combine it with another word that begins with a vowel, like '-anthropos', or drop only the
's', to make 'liko-' if they wish to combine it with a word that begins with a consonant. It sounds a complicated rule, but even to English speakers it is immediately familiar: as Gopnik points out, we use it all the time in new English words like 'technophobia'.
Greek people with SLI cannot manage the rule. They can learn a word like 'likophobia' or 'likanthropos', but they are very bad at recognising that such words have complex structures, built up from different roots and suffixes. As a result, to compensate, they effectively need a larger vocabulary than other people. 'You have to think of them', says Gopnik, 'as people without a native language.' They learn their own tongue in the same laborious way that we, as adults, learn a foreign language, consciously imbibing the rules and words.9
Gopnik acknowledges that some SLI people have low IQ on non-verbal tests, but on the other hand some have above-average I Q . In one pair of fraternal twins, the SLI one had higher non-verbal IQ than the unaffected twin. Gopnik also acknowledges that most SLI people have problems speaking and hearing as well, but she contends that by no means all do and that the coincidence is irrelevant. For instance, people with SLI have no trouble learning the difference between 'ball' and 'bell', yet they frequently say 'fall'
when they mean 'fell' - a grammatical, not a vocabulary difference.
Likewise, they have no difficulty discerning the difference between rhyming words, like 'nose' and 'rose'. Gopnik was furious when one of her opponents described the K family members' speech as
'unintelligible' to outsiders. Having spent many hours with them, talking, eating pizza and attending family celebrations, she says they are perfectly comprehensible. To prove the irrelevance of speaking I N S T I N C T 1 0 1
and hearing difficulties, she has devised written tests, too. For example, consider the following pair of sentences: 'He was very happy last week when he was first.' 'He was very happy last week when he is first.' Most people immediately recognise that the first is grammatical and the second is not. SLI people think they are both acceptable statements. It is hard
None the less, the speaking-and-hearing theorists have not given up. They have recently shown that SLI people have problems with
'sound masking', whereby they fail to notice a pure tone when it is masked by preceding or following noise, unless the tone is forty-five decibels more intense than is detectable to other people. In other words, SLI people have more trouble picking out the subtler sounds of speech from the stream of louder sounds, so they might, for example, miss the '-ed' on the end of a word.
But instead of supporting the view that this explains the entire range of SLI symptoms, including the difficulty with grammatical rules, this lends credence to a much more interesting, evolutionary explanation: that the speech and hearing parts of the brain are next door to the grammar parts and both are damaged by S L I . SLI results from damage
Magnetic-resonance imaging confirms the existence of the brain lesion and the rough location. It occurs, not surprisingly, in one of the two areas devoted to speech and language processing, the areas known as Broca's and Wernicke's areas.
