Books: Grammar and the shark's fin

An infant is to a child as a caterpillar to a butterfly; the difference is that, as the Latin root of the word "infant" affirms, the baby does not speak. Linguists believe that the transition must be like a butterfly's metamorphosis, since their studies of the structures of language have convinced them it is much too difficult for a child to learn. Following Chomsky, the view has prevailed that the brain must contain a circuit-board upon which the deep structures of grammar are engraved. Like a radio, it comes to life when exposed to the waves of language. All that is required is tuning.

In recent years, evolutionary psychologists have extended this notion to all kinds of mental processes, proposing that the mind as a whole is made up of such dedicated "modules". The intellectual appeal of this vision is well conveyed by Steven Pinker's book The Language Instinct. Like much evolutionary psychology, though, it sometimes reads like fantasy shopping for mental modules. Evolutionary psychologists often assume that if a faculty - a sense of fairness, for example - can be argued to have been useful for our ancestors on the savannah, a module for it must have evolved.

Some scientists have attempted, as the title of a book by the psychologist Annette Karmiloff-Smith puts it, to go "beyond modularity". Like Karmiloff- Smith, Terrence Deacon argues that specialisations in the brain are not ready and wired in the newborn baby, but develop as a result of aspects of brain organisation which steer development in certain directions rather than others. His specialisation as a neuroscientist has resulted in a book whose trunk is largely taken up with grey matter, while its limbs reach out into psychology, linguistics and palaeoanthropology.

If that sounds like an awkward creature, it is; and remains so even taken in parts. A few pages of Deacon's prose can be exhausting, let alone 500. That is a pity, since he presents ideas that really should be aired as the ideas of evolutionary psychology bed themselves down in the wider culture. Modularity is a powerful concept, but it needs the kind of heavyweight reality check that is struggling to get out of this book. If The Symbolic Species had been shorter and sharper, it would have made a tremendous sparring partner for The Language Instinct.

Most critiques of modern Darwinian thinking are based on various forms of the claim that there are parts which natural selection cannot reach. Deacon's case is, rather, that theories of language do not appreciate just how far selection can range. The extraordinary facility that children show for acquiring language results not from Universal Grammar engines in their minds, but from the selective pressures their childish faculties exert on language itself. The forms our languages take are constrained by the abilities of our toddlers. This is a perfectly Darwinian process of selection, but it is acting on information that is coded in words, not genes.

According to Deacon, Universal Grammar is an illusion anyway. Although there are features common to all languages, their resemblances are like those between a shark's fin and a dolphin's: each developed separately, but acquired similar features because they arose in similar adaptive contexts. There are certainly predispositions towards some linguistic structures rather than others, but these are the result of the particular ways that our brains and nervous systems are organised.

At the heart of the matter is the symbol, which Deacon emphasises as the critical element that distinguishes human communication from that of all other animals. He rightly acknowledges that symbolic language needs to be explained as an adaptation to a specific set of pressures. There is no justification for assuming that our ancestors just drifted into a symbolic order as their brains got larger. That being the case, though, his discussion should have been shaken down and put at the head of the book. Instead it comes at the end and reads like an afterthought.

He so nearly gets there, too. Language and symbolism are such momentous phenomena that the forces underneath them must have been seismic. The obvious suspect is the fundamental contradiction within sexual species, the difference in the reproductive interests of males and females. Deacon suggests that as the hominid way of life developed, this tension threatened to destabilise hominid society. The solution, he suggests, was marriage: a system of pair-bonding in which males and females guaranteed each others' needs. Language, symbolism and ritual evolved together.

Here Deacon is sculling in waters navigated with a much surer hand by the anthropologist Chris Knight, who argues that full symbolic language emerged much later than Deacon presumes. Unlike Knight, Deacon fails to deal adequately with the question of trust. Without a reason why these ethically precocious hominids should believe each others' promises, his model is an arch without a keystone.