Science is so specialized these days that it's hard for scientists to look up beyond the very narrow confines of their own work. Biologists who study cartilage don't have much to say to biologists who study retinas. Astronomers who study globular clusters probably can't tell you what's new with planetary disks. But sometimes scientists from different specialties can come together and integrate their work into something truly impressive. A case in point comes from some ongoing research into the evolution of language.

No species aside from our own can use language. Chimpanzees and other primates can communicate, but they can't make the subtle sounds that humans can, nor can they turn those sounds into words organized into meaningful sentences. Something happened--or, more likely, many things happened--in the six million years or so since our ancestors split off from the other apes. Fossils offer only a few clues, because the vocal cords, muscles, and nerves that make speech possible are too delicate to turn to fossils. And there's no Pleistocene Napster we can turn to in order to download recordings of what our hominid ancestors sounded like.

Fortunately, there's another record of evolution embedded in the human genome. Unfortunately, it's incredibly hard to figure out what role individual genes have in something as complex as language. In fact, it was only in 2001 that scientists identified a gene involved in acquiring spoken language. They found it by studying a Pakistani family in which half the members suffered from a disorder that interfered with their ability to understand grammar and to speak. The scientists tracked the disorder back to a single mutation to a single gene, which is now known as FOXP2.

FOXP2 belongs to a family of genes found in animals and fungi. They all produce proteins that regulate other genes, giving them a powerful role in the development of the body. FOXP2 in particular exists in other mammals, in slightly different forms. In mice, for example, the part of the gene that actually encodes a protein is 93.5% identical to human FOXP2. And studies on mice show that it plays a crucial role in the developing mouse brain.

Last year another group of scientists compared the the human version of FOXP2 to the sequence in our close primate relatives. They found that chimpanzees have a version of the gene that's hardly different from the gene in mice. But in our own lineage, FOXP2 underwent some fierce natural selection. By comparing the minor differences in FOXP2 carried by different people, the scientists were able to estimate when that natural selection took place--roughly 100,000 years ago. That's about the time when archaeological evidence suggests that humans began using language. (For a good review of all this work, go here.)

How then did FOXP2 pave the way for language? The only way to really get at that question is to understand what the gene does. Some researchers have argued, for example, that it really isn't a "language gene" per se; instead, it screws up the motor control of the mouth, which then makes it very hard for a person to learn language. It has as much to do with language as blindfold, in other words.

Enter brain scanning. Recently, a team of London scientists got a glimpse at the gene by imaging the brains of the original FOXP2 family. As they reported this week in Nature Neuroscience, the researchers split up the family into those who had defective copies of FOXP2 and those who had working copies. They then had the subjects do different language tasks, such as thinking of verbs that go with nouns.

The scientists found that a change to FOXP2 changes the way the brain handles language. Specifically, in people with mutant copies of the gene, a language processing area of the brain called Broca's area is far less active than in people with normal FOXP2.

Broca's area is interesting for a lot of reasons, not least of which is the function of that same patch of tissue in primates. Obviously, they don't use it to talk. But this region is home to some remarkable cells known as "mirror neurons." They fire in the same pattern when a monkey performs some action--turning a lever, for example--and when the monkey sees another monkey performing the same action. These neurons may make imitation possible, and perhaps might have even laid the foundation for a primitive sign language long before our vocal tracts were ready to take over. The natural thing to do now is to measure FOXP2 expression in the Broca's area homolog in other primates. (Harvard's Marc Hauser raises some important reservations about the role of mirror neurons here.)

Of course, FOXP2 will almost certainly not turn out to be the single gene that made human language possible. But thanks to neuroimaging, gene expression profiling, and other new techniques, it can serve as the thin edge of a wedge that scientists can use to split this mystery open.