Poetry, perfumed love notes, intimate e-mails and latenight phone messages have been the choice forms of communication for humans in love. Stags, on the other hand, have to rely on a simple, full-throated roar to convey their desire. True, the stag's primitive bellow is effective--smitten females approach while rival males look for cover. Likewise the cries of dogs, cats and birds all serve these animals well as simple forms of communication.

Even so, it does not take a degree in linguistics to realize that a massive gulf in complexity exists between a male deers amorous cry and "How do I love thee? Let me count the ways." Not surprisingly, then, humans have long felt a sense of superiority as the planets only masters of language arts. But for scholars of language evolution, this apparent singularity was a source of confusion. If other animals can roar, bark or squawk but cannot talk--or do anything remotely similar--then the many characteristics required for language appear to have evolved in humans from almost nothing.

Increasingly, though, studies of animal communication are chipping away at this feeling of human linguistic supremacy. Scientists are finding that our fine-feathered and furry friends have far more sophisticated communication skills than we give them credit for. Physical and cognitive traits once thought to be uniquely human have been discovered throughout the animal kingdom, suggesting that the rudiments of language have deep evolutionary roots. By studying the ancestral building blocks of language, researchers are finally homing in on what truly unique human traits allowed our language skills to bloom and flourish.

Not all that feeling of human pride at our way with words was sheer bravado. It was also backed up by some scientific observations. Biologists once thought, for example, that humans were the only mammals whose larynx was capable of producing different vowels. Even our cousins the chimpanzees cannot voice vowels, because their vocal apparatus is too plump. In addition, they are unable to control their breath with sufficient precision to appropriately aspirate sounds.

Indeed, every textbook on the subject proclaimed that the human larynx was a key example of how humans were specially adapted for language--or at least books made such claims until about 2001. That is when bioacoustician W. Tecumseh Fitch of St. Andrews University in Scotland and David Reby, now at the University of Sussex in England, were able to demolish this myth by finding an animal whose larynx is anatomically similar to its human counterpart and is capable of producing a wide variety of sounds--the deer. Our friend the stag may not be a poet, but he is capable of much more than just a simple roar.

So other animals might have vocal machinery sophisticated enough for speech. But without the right cognitive abilities, the roaring, singing and cheeping of the animal world, however vocally complex, would not be much more than noise. For example, the ability to create categories was thought to be a talent reserved for humans. Only Homo sapiens could group such different beings as a dachshund, a Doberman and a Pekingese into a single, abstract category and word: "dog." These kinds of abstractions obviously play a prominent role in language.

Here, too, research findings have undercut the notion of a pedestal with a single language-capable species on top. Behavioral biologists have found that macaques, chinchillas and even birds divide their world into rational units. Japanese quail, for example, can learn to group sounds that are similar in certain ways into categories, even if they do not use words to describe them.

And although animals do not spontaneously use words, they can demonstrate an uncanny ability to understand them. In 1999 viewers of a European TV game show were treated to a remarkable display of verbal know-how as they watched Rico, a border collie, fetch the right toy--when given a name--out of a collection of 77 playthings. This feat meant the animal understood words such as "teddy bear" or "piggy." "An awfully ambitious woman trained her dog very well," thought Juliane Kaminski of the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany, when she first heard of the performance. Skeptical that Rico actually understood these words, she invited the black-and-white spotted pooch into her laboratory.

The results of her study appeared in 2004 in the journal Science. At the beginning of the test, Rico already knew 200 words. To make sure the dog was directly responding to the words themselves and not some other cue, the investigators took two precautions. Ricos mistress was prevented from providing him with any winks or signals and the objects were not visible when their names were called out. Yet when asked, the dog fetched the black teddy bear, even if the toy was located in an adjacent room.

Kaminski and her colleagues were especially impressed by how easily Rico learned. For example, sometimes the researcher hid an unknown toy behind her back, along with several familiar ones. Then she said a new word to the dog. Rico immediately went over to the objects, picked up the new thing--and remembered its name for the next time.

Of course, Rico did make mistakes now and again, "but so do kids," Kaminski comments. In her opinion, the collie operates at about the level of a three-year-old child in terms of language comprehension.

The Talkative Animal Farm
But it was a bonobo named Kanzi and his kin who finally gave even the most skeptical observers reason to believe that when it comes to language, humans ought to get off their high horse. Sue Savage-Rumbaugh, a biologist now at the Great Ape Trust of Iowa, began in the 1970s to teach apes words with the help of pictograms.

Young Kanzi was an especially eager student. He can use up to 200 "words" by pointing to a display, and he understands twice that number. But the bonobo--today a quiet adult in his mid-twenties--can do far more than merely point to a picture of a banana when he is hungry. He also understands how to string various images together, connecting their meanings. Sometimes he combines a word with a specific gesture and thus creates sentences.

His half sister, Panbanisha, also showed herself to be gifted in language. She once excitedly pressed her finger down on three images, one after another, again and again: "fight," "mad" and "Austin," the name of another chimp in Savage-Rumbaughs big troop. The researchers later discovered that two animals had been beaten in Austins part of the compound.

All this work suggests that instead of human language appearing from thin air, many of the prerequisites for it may have existed--and still do exist--in many species. So why are humans so much more sophisticated in their use of language? Consider, for one thing, the incredible number of words and their meanings that we can process mentally. The average native English or German speaker can recognize about 30,000. This capability is only latent for many people, because they can understand these words in context but do not use them all in their own writing or speech. But the vocabulary is there to be activated--and expanded--at any time. Even the best wordsmiths among dogs and chimps have a vocabulary less than 1 percent as large.

But the really important differences are not just a matter of quantity, according to Marc D. Hauser, professor of psychology at Harvard University. "The secret lies in the grammar," he says. The decisive trait that makes human language different from animal abilities is the complexity of sentence structures that we employ and understand.

The thinking goes that no other animal, including nonhuman primates, can create nested sentences such as "The woman, whose dress, which was not unattractive, and rustled when she walked, sat down next to me." Even the clever bonobo in Iowa would be lost at "The woman, whose dress rustled ..." In short, relative clauses are a human prerogative. This point of view is not new. In the 1950s linguist Noam Chomsky of the Massachusetts Institute of Technology first formulated the idea that human language is hierarchically structured, permitting upper and lower levels to exist.

Determining whether or not animals can process such complex grammatical structures is not easy, however, given that the most accomplished beasts still have relatively simple vocabularies. But recently psychologists and brain researchers have discovered a clever way to test this hypothesis in monkeys.

Say "Ba, La, Tu"
Whereas the precise notion of a clause may be impossible to explain to a monkey, the simple rule that governs its usage does not require understanding the meaning of words at all. For example, the relative clause rule works this way: when you see a term such as "the woman," then you can attach a relative clause to it. This clause rule is how statements such as "The woman, whose dress rustled when she walked" arise. After the insertion of this clause, the sentence with the woman as its subject may continue: "The woman, whose dress rustled when she walked, sat down next to me." Now the relative clause rule may be applied to the newly inserted clause, and so on: "The woman, whose dress, that her husband, whose brother ..." It is only our feeling for style and the limits of our working memory that prevent us from adding clauses ad infinitum. It is not impossible, however.

Hauser studied how monkeys would deal with similar, nested structures. Together with Fitch, he confronted cotton-top tamarins (Saguinus oedipus) with an artificial language composed of meaningless syllables such as ba, la, tu, pa, ka [see box on preceding page]. These New World monkeys are known for being able to recognize spoken sounds and for having a remarkable sense of rhythm. The animals are able, for instance, to distinguish spoken Dutch from Japanese.

The researchers played the animals tapes of syllables, spoken either by a male (M) or a female (F) voice. In addition, the syllables were clearly divided into two groups, so that the female voice always read different syllables from those read by the male. All this was designed to allow the tamarins to easily distinguish the elements from one another. Then Hauser and Fitch used these syllables to create two artificial grammars. One consisted of the simple rule that male and female voices must alternate--for example, the series MFMFMF. The other "language" grammar was more complicated and capable of nested patterns: an M must be, sooner or later, followed by an F. This leads to more challenging sequences such as MMFF (representing the nested structure M[MF]F) or even MMMFFF (M[M[MF]F]F). The initial sound thus opened a door that had to be, at some later time, closed by an appropriate final sound.

Human speakers know this rule, too. They know that a sentence beginning with "if" needs to be followed by a "then" (even if it is only implied)--completely independently of how much additional material is inserted between these two elements. And when tested, human subjects could detect violations in both the simple and complex invented grammars. The researchers wanted to see whether monkeys conditioned to examples of one rule would subsequently recognize violations of this rule. They noticed that their subjects would suddenly stare at the speaker who made a sound. Hauser and Fitch measured the length of the gaze the tamarins directed at a speaker to judge whether violations of the rule caused them to pause longer.

Hear No Rhythm
The animals stared longer at a speaker when the simple grammar was violated. On the other hand, inconsistencies in the more complex grammar left them cold. After hearing structures similar to MMMFFF, the animals reacted no differently when something like MMMFFM was played. They were not able to recognize complex, nested structures, so they could not detect any violations. But Hauser and Fitch did not go so far as to state that the monkeys understood anything even about the simple grammar. "The tamarins probably did not grasp the rules explicitly," Hauser says. "They could, though, distinguish between known and unknown sequences."

So the monkeys lack the understanding of structures that many linguists consider the alpha and omega of language competency. To investigate why people can identify these patterns and nonhuman primates cannot, Angela Friederici of the Max Planck Institute for Human Cognitive and Brain Sciences in Leipzig put human subjects in MRI machines and played tapes of Hauser and Fitchs grammars.

She found that the subjects processed the different sequences in different areas of their brains. Simple MFMF structures were processed by the frontal operculum on the lower end of the primary motor cortex. This region is old from an evolutionary perspective, because other primates possess it as well. Its job is to make reasonable predictions about what should come next in sequences, although the precise way the operculum works has never been studied, Friederici says. "That [region] could be important for grasping musical rhythms, as well as for complex utterances."

But complex, nested sequences or rhythms appear to be beyond this part of the brain. When the subjects in the MRI scanner listened to the complex MMMFFF sequence, it was not the operculum that reacted, but Brocas area instead. This area exists only in humans and is responsible for understanding language. Apparently it is also the place where nested structures are analyzed.

If Friederici, Hauser and Fitch are right, then the ability to process this type of structure may mark a crucial division between human and animal communication. When our ancestors brains developed the ability to process nested structures, they were suddenly able to explore, improve and diversify their communication in complex new ways. It was as if they discovered a music or a grammatical rhythm in the world to which every other creature was tone-deaf. And ultimately that may have given humans a great deal to talk about.