WHEN a dog gazes soulfully into its master's eyes, it is hard for the human not to imagine that the animal has all kinds of interesting thoughts. If only the poor thing had language to express them. Instinctively, pet owners around the world come down on the side of philosophers and linguists who believe in the existence of "mental ese" -- a neurological language of thought coming before spoken and written communication. Who knows what dog musings languish in limbo for want of words that would let them resonate inside another's brain?

But animal lovers who read so much into their speechless companions' behavior risk the scorn of linguists like Derek Bickerton. His pointed new book, "Language and Human Behavior," argues that there is no such thing as mentalese: to think deep thoughts, we must first be able to manufacture sentences; without the inventiveness of language, abstract thinking is impossible.

In the early stages of human evolution, Mr. Bickerton writes, the only means of expressing simple ideas would have been the kind of crude protolanguage babbled by infants or laboriously employed by chimps trained to use hand signs and computer keyboards to demand bananas. Indeed, Mr. Bickerton, a professor of linguistics at the University of Hawaii, Manoa, believes it was chatter like this that fueled the rapid expansion of the human brain, providing not only a rudimentary tool for communication but a burgeoning cerebral work space for hammering together simple thoughts about the world. Instead of just concentrating on the problem at hand -- how to get the mammoth's leg back to the cave -- an early human could use protolanguage to begin to engage in what Mr. Bickerton calls "off-line thinking," dimly reflecting on the day's triumphs and failures while staring into the fire at night.

But for these early ancestors to really think -- to use experience to draw detailed plans for how to be more effective cave dwellers -- they would need not just words referring to things in the world (mammoth, tree, banana) but words like "that" and "why" and "because" and "for" (and, for that matter, "but" and "like" and "and"), words that point not to things but to other words, allowing them to be snapped together into complex sentences and so into the many-tiered structures of thought. To become human, Mr. Bickerton argues, the early hominids had to acquire what no other animal seems to have: syntax.

Where did this powerful magic come from? Mr. Bickerton supposes that late in human development, with the brain at its full size and the vague thoughts allowed by protolanguage sloshing around inside, a single random mutation unleashed a reorganization of the cerebral tissue. A new network connected neural regions that had been as remote as Europe and China before Marco Polo, and in the process, syntax -- the rules for building sentences of unlimited depth and intricacy -- was born.

Judging from the frequency with which Mr. Bickerton cites himself, this ingenious argument will not be new to readers of his well-received 1990 book, "Language and Species." "Language and Human Behavior" is actually a series of lectures smoothly stitched into a short book. The writing is clear, and as in a good lecture, the arguments are so forceful that sometimes the reader has to stop and remember how few are the shreds of evidence from which such grand theories are woven.

It's intriguing to read that though hominids lived in the Zhoukoudian caves in northern China for 300,000 years, their brains growing steadily larger, they showed no sign of developing a more complex culture. Their artifacts remained simple and uninspired as the cave dwellers stagnated (as Mr. Bickerton would have it) in a syntaxless mental muck. Mr. Bickerton's case for the primacy of syntax seems even stronger when we read about his own studies of the primitive pidgin languages cobbled together by slaves and other sudden, unwilling immigrants. Like the proto language coaxed from the trained chimpanzees ("You tickle me"; "Give me drink"), pidgin languages can communicate simple wants. Then, in the next generation, the brains of the immigrants' sons and daughters automatically convert the amorphous pidgin into a fully expressive creole.

From dozens of pieces of such circumstantial evidence, Mr. Bickerton constructs his tale. If some readers don't find the whole as compelling as the parts, they should at least be struck by his underlying message: the chasm separating people from every other animal on earth is unfathomably wide. Midway through his book, Mr. Bickerton asks us to consider how truly wonderful are mundane activities like a boxer pummeling a punching bag or a dancer stretching in front of a mirror. "Look at how alien these behaviors are to any species but our own," he writes. "Try to imagine a tiger practicing its killing technique in the absence of any prey, or a gazelle practicing its latest escape maneuver in the absence of any predator. . . . Doing a special, individualized thing simply to be able to do it better on some future occasion is uniquely human behavior." The same could be said for theory building, this marvelous ability to spin engaging scientific tales.

Jump to George Johnson's Home Page to see his books and New York Times articles.