The ‘language instinct’ debate (review)

Geoffrey Sampson’s The ‘language instinct’ debate(a revision of his 1997 book Educating Eve) is a tightly argued monograph charging that the search for universal grammar is a ‘wrong turning in the progress of thinking about human nature’ (25) indulged in by ‘academics who got caught up in a forty-year-old tide of fashion and lack the flexibility to retool intellectually’ (193). While too often drifting into snipes like that one, S, who unlike many critics of Noam Chomsky’s work has closely studied the literature, makes a powerful case that linguistic nativism, despite the brilliance of many of its defenders, has been grievously underargued, and risks looking to scientists in a hundred years like the search for phlogiston does to us now.

S begins by addressing the main observations upon which Chomsky founds the case for nativism. It is unclear to S by what metric children’s acquisition of language is unusually ‘rapid’. They learn how to pour liquids from one vessel into another just as rapidly, and to insist that language is more complex than this elaborate neuromuscular feat is ‘as meaningless as a statement that War and peaceis twice as long as the River Nile’ (39).

S gives particular attention to the ‘poverty of the stimulus’ argument. Contrary to the idea that infants are exposed to a constant barrage of sentence errors, Newport, Gleitman, and Gleitman (1977) reported that only one out of 1,500 Motherese utterances is ungrammatical, and Valian’s (1990) claim that errors are much more common in Motherese relied on sentences such as Want your lunch now?, which are less subjectless than colloquial.

Other linguists point to the fact that children acquire language without being exposed to negative evidence. S deftly counters the idea that this suggests innate capacity for language. Science, after all, progresses without negative evidence: principles in physics were worked out without people seeing them contravened. In the same way, S argues that it is both plausible and empirical to propose that children ‘will aim to formulate sets of rules which characterize as many strings as possible as ungrammatical while permitting the various strings which have occurred in the positive data-set’ (91). For S, the negative-evidence argument denies humans’ basic propensity to seek patterns and reveals generativists as ‘out of their depth’ regarding issues relating to the philosophy of science and the mind.

S makes a special point of addressing Pinker’s (1994) nativist arguments in The language instinct, treating it as part of a revival in the 1990s of interest in nativism beyond linguistics. A memorable case in Pinker’s book is the family carrying a gene that impairs their linguistic abilities. Gopnik’s (see Gopnik 1990 and Gopnik & Crago 1991) reportage depicted a disability that left sufferers’ intelligence intact, affecting only grammar and in a fashion so precise—leaving subjects unable to form new plurals or past-marked verbs—as to seem evidence that such distinctions are actually represented in our brain structure, subject to damage by congenital defects. But Vargha-Khadem and her team (1995), retesting the sufferers, found that the subjects indeed form new past verbs, but overgeneralize past marking, showing that there was a more general problem with manipulating rules. Unsurprisingly, Vargha-Khadem found that the subjects also had unusually low intelligence, difficulty managing sequential tasks, and problems with naming and pronunciation.

Pinker presents the common argument that there is no reason children might not try reversing the word order of sentences to form questions. S rightly points out that this notion only seems plausible when language is represented on the page: when speaking, we perceive language in chunks rather than perceiving each separate element in sequence. Rendering sentences backwards is not a natural choice mysteriously avoided but an arduous task unsurprisingly nonexistent.

For this edition, S has added a chapter mining spoken language data in the British National Corpus for further disproof of the poverty-of-the-stimulus argument. The results are not optimal for S, actually. The Chomskyan famously claims...