The Stuff of Thought, By Steven Pinker

Caveman conversations

Language is so familiar that we don't notice how strange it is. Why is it all right to say "I poured wine into the glass" but not "I filled wine into the glass"? Why can you "stroll for ten minutes" but not "walk to the shop for ten minutes"? And why, to switch from the mundane to the scatological, can you tell someone "Don't hurt yourself" but not "Don't fuck you"?

In his new book, Steven Pinker answers these puzzles and many more. On the surface, these pairs of sentences may seem to have parallel meanings, but Pinker shows that the similarities conceal underlying differences. People effortlessly discern that the pairs have divergent deep structures and that this governs what we can and can't say. Pinker elegantly teases out these structures and uses them to illuminate not just the working of ordinary verbs but a range of linguistic phenomena, from metaphor and children's names to euphemism and obscenity.

Pinker is a distinguished professor of psychology at Harvard University, but outside academic circles is better known for his splendid popular books. He started over a decade ago with The Language Instinct. This merged modern thinking about the evolution of the human mind with the linguistic theories of Noam Chomsky.

Pinker has divided his subsequent books between these two themes of evolution and linguistics. How The Mind Works and The Blank State argued that the modern human mind still bears the mark of our primeval ancestry, while in the intervening Words and Rules, Pinker stuck to the linguistic issue of how our finite minds can generate an indefinite range of complex sentences.

In this new volume, Pinker returns to the workings of language, and evolutionary considerations take a back seat. I prefer Pinker's language books to the evolutionary ones. Both are immensely readable and stimulating. Pinker is a master at making complex ideas palatable with snippets of popular culture and tried-and-tested jokes.

But his evolutionary ideas are often highly speculative. According to his critics, they are nothing but "just so stories" which differ from Kipling's fables only in lacking a good moral. In truth, Pinker's arguments for his evolutionary claims often seem to protest too much. By contrast, the language books are full of hard data. This is Pinker's real area of expertise, and it is always fascinating to see him unpick the curious ways in which language works.

Pinker's main theme in this book is that language lays bare the basic categories used by the human mind. By studying the ways verbs work, for instance, we can see how our minds categorise actions. It turns out, curiously, that actions that work because of gravity are classified differently from those that use other kinds of force. (That's why you can "pour wine into the glass" but not "fill wine into the glass"). Applying this method across a wide range of linguistic constructions, Pinker pieces together the basic concepts by which humans structure the world.

We turn out to be a depressingly practical bunch. It's all to do with different kinds of stuff, and what can go where, and how one thing can be used to smear, shatter, or fill up another. Those millions of years as hunter-gatherers seem to have left their imprint. Our basic outlook on the universe is that of a caveman wondering how to crack the next nut.

Of course, as Pinker stresses, this doesn't mean that all our thinking is still stuck in the stone age. Now we don't just think about food, but about monetary inflation, bureaucratic bottlenecks, and the decline of the novel. But modern thought is still built on the patterns laid down long ago.

We can extend the range of things we talk about, often using metaphor to get new ideas off the ground, but new ways of thinking never fully break free of the old. For example, note the spatial meanings of "inflation", "bottleneck" and "decline". We may be thinking about abstract matters, but we can't manage without the help of simple spatial notions.

Which comes first, language or thought? Do we think in basic physical terms because of the way English shapes our thought, or does English work as it does because of how we already think?

According to the once-modish thesis of linguistic determinism, language runs the show. Different peoples think differently because brought up to speak different languages. Classic exhibits in the determinist gallery are the Hopi, reputed not to distinguish space and time because of their odd structure of tenses, and the Inuit, whose 30-odd words for snow are supposed to allow them to make meteorological distinctions that escape the rest of us.

Pinker will have none of this. He argues that all humans share a universal "language of thought" containing basic concepts of space, time, force and sorts of stuff. Not all spoken languages express these categories using identical grammatical constructions, but the same linguistic oddities keep popping up in unrelated languages, and all societies have some way of marking these fundamental distinctions. Pinker allows that linguistic training can make a difference to more sophisticated thinking. You'd be hard put to distinguish the days of the week if you didn't live in a community that named them. But even here Pinker thinks that the linguistic determinists put the cart before the horse.

Modern humans distinguish the days because they need to, not because language has forced them to think in those terms. Similarly, if the Inuits didn't have 30 words for snow, it would have been necessary to invent them. To drive the point home, Pinker observes that English speakers have all been introduced to words like "skylark", "plover" and "redshank" without this making the slightest different to the thoughts of any but a few twitchers.

Linguistic determinism is just one of the foes Pinker takes on. He engages with a number of rival theories of language. Sometimes he cuts a few argumentative corners in the interests of mounting a persuasive case. Still, it is not hard to forgive him. He may be partisan, but he is never boring. And he does know a lot about words.

David Papineau is professor of philosophy at King's College, London