Linguistic Notes: Collapse of the language nativists

A BUZZWORD of 1990s pop psychology is "language instinct". According to Steven Pinker's best-selling book of that title, mankind inherits knowledge of language structure. Just as the patterning of a peacock's tail is mapped in peacock genes, so Pinker and others claim that the fine detail of phrases, clauses and sentences is mapped in human genes. These "linguistic nativists" urge that we do not really learn our mother tongue. In a sense we are born knowing most of it.

It sounds an odd idea; but the reading public is buying into it. Richard Dawkins's verdict is typical of many: "Reading Steven Pinker's book is one of the biggest favours I've ever done my brain." Yet, if one asks whether the linguistic nativists have a sound case, the answer must be a resounding no. It is not just that their logic is often shaky. Worse, again and again their factual evidence is just not true.

Take one of Steven Pinker's most impressive points: the discovery, in Essex in 1990, of a family of "language mutants". A paper in Nature reported that members of this family suffer from an inherited disability which prevents their acquiring grammar. When one of them manages to learn a plural, say "books", he learns it as a simple vocabulary element, with no concept of a relationship with "book". If grammar is in our genes, we might expect to find occasional grammatical mutants.

But, shortly after the first paper appeared in Nature, the family was looked at again. It turned out that their disability was not specific to language. The family had intellectual handicaps across the board; yet there was clear evidence that they did know the very grammar rules which Pinker said they were genetically incapable of mastering. So much for "mutant grammar genes".

Or consider another point, about the way we form compound words. Pinker notes that it seems normal to describe a house as "mice- infested" but not as "rats- infested", and he infers a genetic constraint which prevents our coining compounds from plurals unless the plural is irregular (like "mice").

But there is no such constraint. The Swedish linguist Stig Johansson wrote a whole book in 1980 about English compounds based on plurals: we may not say "rats-infested," but newspapers do write about "fares-cutting airlines". Johansson found that Britons were happier with such phrases than Americans: so we are dealing with cultural habits, not biological machinery.

Every argument put forward by today's linguistic nativists collapses when soberly examined. But these writers are not wholly to blame. They are only restating, in 1990s terms, a theory advocated 30 years earlier by Noam Chomsky. Chomsky's towering reputation leaves people reluctant to tangle with him.

Yet, if one asks what Chomsky's evidence was, one finds the same yawning gap between conclusions and data. In book after book for over 20 years, Chomsky argued that certain aspects of question formation must be innate; learning the rule from experience would depend on encountering particular varieties of question which, he believed, were vanishingly rare. Geoffrey Pullum of the University of California recently decided to look at the facts. He found that, in practice, about one in eight questions are the type that Chomsky assumed were virtually non-existent.

Chomsky made his mark as a political radical; many took his linguistic theories on trust. Steven Pinker has a different but equally powerful special factor going for him: he is a marvellous, inspired wordsmith. Reading The Language Instinct is such a pleasurable experience that people quite naturally want its message to be true.

But it isn't. No serious argument has ever been put forward that our genes give us knowledge of language. Languages are cultural products, learned by successive generations from scratch.

Geoffrey Sampson is the author of `Educating Eve: the language instinct debate' (Cassell, pounds 14.99)