杏吧原创

Understand how the Bengalese finch acquired its virtuoso singing, and you can learn a lot about the forces that freed our own tongues

WHEN you watch Kazuo Okanoya on stage, bobbing up and down, chirping, you know he is passionate about his work. His is alive with the sound of the birds that inspire his performance 鈥� row upon row of cages full of Bengalese finches. You can see why he is so taken by them. They are beautiful and good-natured, and they sing like a dream.

Okanoya was brought up in rural Japan surrounded by farm animals as well as his own menagerie of pet hamsters, turtles, hermit crabs, chipmunks and finches. 鈥淎s a child, I loved animals more than humans,鈥� he says. That he ended up studying birds is hardly surprising. But what he has discovered certainly is. He set out to explore how singing cements the intense bond between pairs of Bengalese finches and underpins their devoted parenting. Instead, his experiments might have implications for one of evolution鈥檚 most enduring mysteries: the emergence of human language.

Listen to the Bengal finch and you may discover how language evolved (Image: Richard Wilkinson)

The Bengalese finch鈥檚 mystery hinges on its ancestor, the white-rumped munia. It is a rather drab and reclusive bird, found across South-East Asia, with a somewhat repetitive and simple song. But around 250 years ago, breeders started to transform it into a beauty, with a subtle plumage of chocolate to platinum feathers. As a side effect, the Bengalese finch also acquired a remarkably complex song. So how did a bird that was bred for its plumage come to sing intricate melodies composed of many phrases?

It is by answering this question that Okanoya has found some clues about the evolutionary forces that loosened our own ancestors鈥� tongues. In the past, it seemed obvious that language emerged through the slow process of natural selection, driven by the benefits of better communication. His work on the Bengalese finch, however, suggests that speech could have come about spontaneously through an entirely different process.

Our understanding of language has long been hampered by the fact that there are very few creatures with anything approaching human vocal talents 鈥� and none among our primate relatives. Nature contains just six groups of animals with complex, learned vocalisation: songbirds, hummingbirds, parrots, cetaceans, bats and humans. Our language is far more advanced than the communication of any other animal. Nevertheless, the Bengalese finch鈥檚 song can be thought to possess a simple syntax. It uses a variety of melodious chunks of sound in a creative way, akin to how we arrange words and phrases to produce sentences. That makes it a cut above most other animals. In addition, its less-articulate forebear, the munia, is still alive, which allows Okanoya to compare the two in a bid to work out how the vocalisation became so elaborate.

His first step was to in complexity between the songs of munias and Bengalese finches. To do this, Okanoya and his collaborators divided the total number of different notes produced by each during a 2-minute performance, by the number of different note-to-note transitions over the same period. This gives a measure of 鈥渓inearity鈥� 鈥� essentially, how predictable the next sequence of the song is, given what came before. The munia鈥檚 linearity score, at 0.6 out of a possible 1, was almost twice as high as the finch鈥檚, indicating a simpler repertoire.

The art of imitation

Munias might nevertheless be able to produce sophisticated tunes, given the right tutors. Like most songbirds, only the males sing and they learn from their fathers and other nearby males, reproducing the sorts of songs they hear around them. So Okanoya and his colleague Miki Takahasi raised munia chicks with Bengalese finch foster parents. The result? The munias learned finch songs with 80 per cent accuracy, compared with the 99 per cent they achieve on the songs of their own species. By contrast, finches reared by munias reproduced their foster parents鈥� songs with 90 per cent accuracy, just as they do for their own (). In other words, the Bengalese finch鈥檚 genes don鈥檛 dictate the specific sequence of the song, leaving more room for learning and improvisation, compared with munias. 鈥淭he relative looseness of the Bengalese finch鈥檚 learning compared with the munia鈥檚 is a key to the emergence of complexity,鈥� says Okanoya.

That ability for virtuoso singing is probably the result of many different changes that came about through domestication. Some of the changes seem to be in the finch brain鈥檚 nucleus interfacialis, one of its song-control centres (see 鈥淣ot so bird-brained鈥�), which are larger in finches than in munias. When Okanoya surgically removed this region from . He has also discovered that the brains of the Bengalese finches express more of a gene for glutamate receptors, found at the synapses between neurons. These receptors are thought to be vital for learning and memory.

Differences in character could also underpin a more innovative singing style. Unlike their munia ancestors, and they have . Being shyer and more vigilant would be an advantage in the wild, but it could also limit opportunities for learning complex songs, says Okanoya. Breeders, however, would have probably favoured laid-back birds because they are easier to handle. What鈥檚 more, without the danger of being heard by predators, the birds could sing as effusively as they liked, for as long as they liked, giving more opportunity for greater complexity to emerge.

That鈥檚 not all. Life in the aviary also changes the mating game 鈥� and this would have a profound influence on the way the birds flirt with song. Okanoya鈥檚 fieldwork in Taiwan showed that munia populations living in areas with the . This suggests that simple songs help them sound more distinct, reducing the likelihood of hybridisation and its associated disadvantages, such as lower fertility in the offspring. By contrast, the risk of mating with another species is not a problem for caged birds.

Without that constraint, the male finches are freer to show off with more elaborate routines. Okanoya has found that, all other things being equal, female munia and Bengalese finches . 鈥淭hey engage in more nesting behaviour and lay more eggs,鈥� he says. This means that even though breeders choose finches based on the prettiness of their plumage, the birds鈥� own preferences may have been a driving force in the evolution of other characteristics through the process of sexual selection. 鈥淲hen paired with good singers, females will be more reproductive,鈥� he says.

If Okanoya is correct, all it took for the munia鈥檚 ditty to evolve into the Bengalese finch鈥檚 serenade was a couple of hundred years of domestication and a female preference for complexity. And this is where things start to get interesting, because Okanoya sees parallels between this transition and the evolution of human language from the simple vocalisations of our forebears.

That may sound improbable, but in evolutionary terms, the salient feature of domestication is not captivity 鈥� it is safety. Biological anthropologist Richard Wrangham of Harvard University has long argued that as we found ways to control our environment, we freed ourselves from many of the forces of natural selection. We were able to occupy our own evolutionary niche in which 鈥渢ame鈥�, cooperative individuals are most successful. And, he points out, there are signs that we share many features with domesticated animals. Besides acquiring the main traits that breeders desire, domesticated animals often show other unintended characteristics: the adults tend to look more juvenile than their wild ancestors, with smaller teeth and jaws and cuter faces; they become less aggressive, and their brains shrink as survival depends less on intelligence. When compared to other living primates and reconstructions of our ancestors from the fossil record, modern humans show many of these characteristics. Human brains have even shrunk in the past 30,000 years.

鈥淲e humans share many features with domesticated animals. Our brains have even shrunk in the past 30,000 years鈥�

Picky partners

According to Okanoya, this self-domestication freed us from the constraints limiting our communication, allowing sexual selection to push for a greater complexity. There is plenty of evidence that humans are picky about who they mate with, he says, pointing to the book The Mating Mind, in which Geoffrey Miller at the University of New Mexico in Albuquerque argues that our preferences for intelligent, creative and articulate partners has been a key driving force in shaping human nature. Humans of both sexes are choosy about partners, and their differing preferences are reflected in usage and comprehension of language, Okanoya claims. 鈥淔or example, females are better at detecting mismatch between prosody [the intonation and rhythm of speech] and meaning, and males are better at using a variety of vocabulary in conversation,鈥� he says.

Needless to say, not everyone subscribes to the self-domestication hypothesis 鈥� and even among those who do, there is disagreement about the precise way it influenced human evolution. at the University of California, Berkeley, for one, sees no role for the sexual selection invoked by Okanoya 鈥� in finches or in humans. He believes that by removing the natural forces constraining the munia鈥檚 song, domestication allowed the accumulation of mutations in genes that had previously kept vocalisations simple. And, Deacon argues, human language could have evolved in a similar process of once self-domestication lifted constraints on vocalisation. For example, while other primates cannot help linking specific vocal noises with particular emotional states or immediate contexts, the ancestors of humans obtained more control over their utterances, through the accumulation of random mutations. These allowed previously independent mental abilities to work together to produce language.

This is a radical departure from widely held ideas about language evolution, which look for the adaptive advantages associated with speech to explain its emergence. Such theories tend to extrapolate from innate primate calls to the production of words, and suggest that the required cognitive abilities evolved gradually by natural selection. But if Deacon is correct, language emerged spontaneously.

Not convinced? Enter Simon Kirby, who studies language evolution at the University of Edinburgh in the UK. When he learned about Okanoya鈥檚 research and Deacon鈥檚 interpretation of it, he decided to explore their ideas. With colleague Graham Ritchie, he created a computer simulation of a munia population, then modelled the evolution of these virtual birds in the wild and in captivity. Just as Deacon had predicted, when forces associated with natural and sexual selection were relaxed, making specific features of the song less closely linked to survival and mate selection, .

For Kirby, the key is learning. In the model, domestication across many generations slowly removed the genetic constraints to learning, allowing the birds to pick up their songs from a more varied range of virtual individuals, rather than learning almost verbatim from a father figure, as munias do in the real world. Here Kirby and Ritchie make the connection with language. Birdsong and language, unlike other communication systems, have complex syntactic structures and are transmitted, often with slight differences, between individuals and down the generations. Combine that with a flexible learning system, and the iterations can become increasingly complex.

Besides disagreeing on the role of sexual selection in our ancestors, Deacon and Okanoya also differ in their views on how and when the pressures of natural selection were relaxed. Okanoya sees the event as recent 鈥� around 100,000 years ago 鈥� and he suspects it happened rapidly, perhaps over 20,000 years. 鈥淢aybe something occurred that required collaboration among non-kin,鈥� he says. It could have been anything from climate change or increased predation risk, to sustained conflict with other hominins. 鈥淎s a result, we had to become more social.鈥� By working together we could manipulate and better control our environment. This brought the increased safety in the group necessary for self-domestication.

Deacon, on the other hand, believes symbolic thinking was the impetus behind the relaxing of some selection pressures; once our ancestors developed a tendency to see meaning in the world around them, they created their own unique niche. 鈥淒omestication is likely to be driven by construction of an artificial niche via symbolic culture,鈥� he says. He suggests this started as far back as 2.4 million years ago, when our ancestors first realised that rocks could be made into tools 鈥� an innovation that forever changed their way of life. The ability to see things at more than face value resulted in a gradual mastery of the natural world 鈥� using tools to exploit a wider range of resources, for example, and rituals to bond people into more cohesive and effective groups. This, in turn, would have relaxed the former ecological selection pressures, while favouring individuals adapted to a new 鈥渟ymbolic ecosystem鈥� and to group living. Deacon says this self-domestication, while slow at first, really took off with the evolution of Homo sapiens. 鈥淭he effect was probably modest until only a few hundred thousand years ago, when material culture begins to complexify more rapidly.鈥�

Mind reading

The beauty of Deacon鈥檚 idea is that it also explains the major difference between human vocalisations and those of other animals 鈥� language is full of meanings. If our ancestors had already evolved to invest meaning in material goods, then it was only one more step to encode this in sound too.

Kirby sees it differently. For him, the key is our highly developed skill for mind-reading. Other species, including songbirds, produce complex signals, but we also have the inclination to infer intentions in others 鈥� so when we hear sounds coming out of another person鈥檚 mouth we assume they are trying to convey a message. This combination of complex vocalisation, and so-called 鈥渢heory of mind鈥�, is unique. 鈥淪elf-domestication could explain these two phenomena in humans,鈥� says Kirby 鈥� and together they might have created a meaningful language.

Okanoya and his team have also begun to explore the puzzle of how language acquired meaning. He sees clues in the crying of babies. 鈥淥ther primates don鈥檛 cry like us. Their cry is suppressed 鈥� reserved only for emergencies,鈥� he says. That鈥檚 because the noise might attract predators. But in our self-domesticated niche, ancestral infants were safe to cry, and in doing so increased their chances of survival. In Okanoya鈥檚 lab, Yulri Nonaka has found that although babies鈥� different cries have no set meaning, parents assign meaning to them to begin with. By consistently responding to a given type of cry with a specific behaviour 鈥� such as feeding or changing 鈥� parents allow the baby to learn which noises to make and so manipulate its caregivers.

These findings indicate that crying is a private language between an infant and its carers. Okanoya is not suggesting that crying was the precursor to human language. Instead, he says, 鈥渃rying set up the anatomical and cognitive context鈥�. Importantly, it let humans evolve to control the force and duration of each breath, something other primates struggle with.

鈥淭he crying of babies allowed humans to evolve to do what other primates struggle with 鈥� control the force of each breath鈥�

So how did language evolve? Crying led to singing, suggests Okanoya 鈥� . With singing came the ability to recognise discrete chunks of sound within a continuous string, . At this point, when our ancestors faced crises in which cooperation was vital for survival, they could work together, linking sound chunks to specific contexts and starting to create words with meanings.

A Just So story perhaps, but one with more theoretical underpinnings than many. And although we may never know for sure exactly why, when and how language evolved, the idea that domestication was the spur brings fresh thinking to these hotly contested questions. 鈥淚t opens up whole new avenues when looking for evidence,鈥� says Kirby. Sure, there is no munia to our finch, but there are human populations, including deaf signers, in the process of devising new languages. 鈥淲e can observe language evolution right now,鈥� he says. What鈥檚 more, the recent extraction of DNA from a 400,000-year-old hominin offers the exciting prospect that we will one day have the genome of a pre-linguistic ancestor against which to compare our own. It will be another vital clue in the quest to work out whether our most prized attribute was shaped by natural selection or sexual selection, or whether it emerged spontaneously for no particular reason at all.

This article appeared in print under the headline 鈥淭he finch whisperers鈥�