In our quest to find what makes humans unique, we often compare ourselves with our closest relatives: the great apes. But when it comes to understanding the quintessentially human capacity for language, scientists are finding that the most tantalizing clues lay farther afield.
Human language is made possible by an impressive aptitude for vocal learning. Infants hear sounds and words, form memories of them, and later try to produce those sounds, improving as they grow up. Most animals cannot learn to imitate sounds at all. Though nonhuman primates can learn how to use innate vocalizations in new ways, they don’t show a similar ability to learn new calls. Interestingly, a small number of more distant mammal species, including dolphins and bats, do have this capacity. But among the scattering of nonhuman vocal learners across the branches of the bush of life, the most impressive are birds — hands (wings?) down.
Parrots, songbirds and hummingbirds all learn new vocalizations. The calls and songs of some species in these groups appear to have even more in common with human language, such as conveying information intentionally and using simple forms of some of the elements of human language such as phonology, semantics and syntax. And the similarities run deeper, including analogous brain structures that are not shared by species without vocal learning.
These parallels have motivated an explosion of research in recent decades, says ethologist Julia Hyland Bruno of Columbia University, who studies social aspects of song learning in zebra finches. “Lots of people have made analogies between language and birdsong,” she says.
Hyland Bruno studies zebra finches because they are more social than most migratory birds — they like to travel in small bands that occasionally gather into larger groups. “I’m interested in how it is that they learn their culturally transmitted vocalizations in these groups,” says Hyland Bruno, coauthor of a paper in the 2021 Annual Review of Linguistics comparing birdsong learning and culture with human language.
Both birdsong and language are passed culturally to later generations through vocal learning. Geographically distant populations of the same bird species can make small tweaks to their songs over time, eventually resulting in a new dialect — a process similar in some ways to how humans develop different accents, dialects and languages.
With all these similarities in mind, it’s reasonable to ask if birds themselves have language. It may come down to how you define it.
“I wouldn’t say they have language in the way linguistic experts define it,” says neuroscientist Erich Jarvis of the Rockefeller University in New York City, and a coauthor of Hyland Bruno’s paper on birdsong and language. But for scientists like Jarvis who study the neurobiology of vocal communication in birds, “I would say they have a remnant or a rudimentary form of what we might call spoken language.
“It’s like the word ‘love.’ You ask lots of people what does it mean, and you’re going to get a lot of different meanings. Which means that it’s partly a mystery.”
There are multiple components to spoken language, Jarvis says, and some are shared by more species than others. A fairly common component is auditory learning, like a dog figuring out how to respond to the spoken command “sit.” The vocal learning that humans and some birds do is one of the most specialized components, but all of them are shared to some degree by other animals, he says.
The grammar of bird calls
One key element of human language is semantics, the connection of words with meanings. Scientists had long thought that unlike our words, animal vocalizations were involuntary, reflecting the emotional state of the animal without conveying any other information. But over the last four decades, numerous studies have shown that various animals have distinct calls with specific meanings.
Many bird species use different alarm calls for different predators. Japanese tits, which nest in tree cavities, have one call that causes their chicks to crouch down to avoid being pulled out of the nest by crows, and another call for tree snakes that sends the chicks jumping out of the nest entirely. Siberian jays vary their calls depending on whether a predatory hawk is seen perching, looking for prey or actively attacking — and each call elicits a different response from other nearby jays. And black-capped chickadees change the number of “dees” in their characteristic call to indicate the relative size and threat of predators.
Two recent studies suggest that the order of some birds’ vocalizations may impact their meaning. Though the idea is still controversial, this could represent a rudimentary form of the rules governing the order and combination of words and elements in human language known as syntax, as illustrated by the classic “dog bites man” vs. “man bites dog” example.
In addition to alert calls, many bird species use recruitment calls that summon other members of their species. Both Japanese tits and southern pied babblers appear to combine alert calls with recruitment calls to create a sort of call to arms, gathering their compatriots into a mob to harass and chase off a predator. When the birds hear this call, they approach the caller while scanning for danger.
The order of combined calls matters to Japanese tits, which may mean they have a rudimentary form of the ordering rules of human language known as syntax.
CREDIT: SIGEKI / ADOBE STOCK
Scientists led by ethologist Toshitaka Suzuki of Kyoto University discovered that the order of the combined calls matters to the Japanese tits. When Suzuki’s team played a recorded “alert+recruitment” combo to wild tits, it elicited a much stronger mobbing response than an artificially reversed “recruitment+alert” call. This could simply be explained by the birds responding to the combined alert+recruitment call as its own signal without recognizing the parts of the combination, but the scientists came up with a clever way to test this question.
Willow tits have their own distinct recruitment calls, which Japanese tits also understand and respond to in the wild. When Suzuki’s team combined the willow tit recruitment call with the Japanese tit alert call, the Japanese tits responded with the same combined scanning and approaching behavior — but only if the calls were in the correct alert+recruitment order.
“These results demonstrate a new parallel between animal communication systems and human language,” Suzuki and colleagues wrote in Current Biology in 2017.
But it’s a matter of interpretation whether the call combinations of the tits and babblers is really relevant to discussions of human language, which involves more complex sequences, says behavioral neuroscientist Adam Fishbein of the University of California, San Diego.
“If they were doing something more like language, you would get a whole bunch of different combinations of things,” Fishbein says. “It’s such a restricted system within the birds.”
Sounding it out
Fishbein’s own research with zebra finch song suggests that syntax may not be as important to birds as it is to humans. “I feel like people have been trying to impose this human way of thinking about communication on what the birds are doing,” he says.
Birdsong can be very complex and tends to have typical sequences and patterns of notes, syllables and motifs. So birds’ singing may be a closer analog to human language than the tits’ alert and recruitment calls. To the human ear, parts of birdsong are reminiscent of word syllables, so it’s easy to assume the order of those parts is important to the message. But, perhaps surprisingly, we know very little about how birdsong sequences are perceived by the avian ear. Fishbein’s research suggests that what birds hear when they listen to birdsong may be very different from what humans hear.
Male zebra finches all learn a single common song, which you can hear repeated over and over on this recording in what is known as a song bout. Though they all share the same short song, there are subtle variations between renditions of the song, which could provide another dimension of information that scientists have yet to decode.
CREDIT: BALL/DOOLING LAB AT UNIVERSITY OF MARYLAND, COLLEGE PARK
For his graduate work at the University of Maryland, Fishbein studied zebra finches that had been trained to press a button when they heard a change in sounds played to them. When the birds correctly identified a change, pressing the button got them a food reward. If they guessed wrong, the lights in their enclosure went off briefly. Fishbein tested what differences the birds are actually able to decipher, helping scientists understand what aspects of birdsong are important to the birds.
In one test, Fishbein and his colleagues played the finches’ standard song over and over at regular intervals before slipping in a version of the song with artificially reordered syllables. This change is easy for humans to hear, but the birds were surprisingly bad at identifying the shuffled sequence.
The birds performed much better at another test Fishbein gave them. Within each song syllable, there are higher-frequency details called “temporal fine structure” that may be something like what humans perceive as timbre or tone quality. When the scientists messed with the song’s fine structure by playing one of the syllables backwards, the finches were “exceedingly” good at catching it.
“It’s a dimension of sound that they’re much better at hearing than we are,” Fishbein says. “So they may be tapped into this level of the sound that we’re not tapping into when we just casually listen to birdsong.”
Our understanding of what birds hear and what matters to them is limited by what we hear, and as with a lot of scientific research, the statistical analyses used — in this case to parse birdsong, says linguist Juan Uriagereka, who worked with Fishbein at the University of Maryland. “Ten years ago, we didn’t even know what the units that they were combining were,” he says. “And of course, what we think are the units, it’s our guess, right?”
Though male zebra finches all learn the same single song, scientists have found that there is variation in temporal fine structure among renditions of the standard song, hinting that the birds have a much richer communication system than we suspected. “It could be that most of the meaning is packed into the individual elements,” Fishbein says, “and how they’re arranged may not matter as much for conveying meaning.”
To find out what aspect of a zebra finch’s song matters to the birds, scientists tweaked a recorded song to see if the birds would notice. A spectrogram of the natural song is shown on the top row. When the scientists shuffled syllables as shown in the middle row, the birds didn’t react. But when the scientists reversed one of the syllables in the song, as shown in the bottom row, the birds were very good at catching it.
CREDIT: ADAPTED FROM A.R. FISHBEIN ET AL / PHILOSOPHICAL TRANSACTIONS OF THE ROYAL SOCIETY B 2019
Mean what you say
Even if some birds share rudimentary aspects of human language, we still know very little about what’s actually going on in their minds. Most animal communication research has focused on describing signals and behavior, which on the surface can look a lot like human behavior. Determining if the underlying cognitive processes driving the behavior are also similar is much more challenging.
At the heart of this question is intentionality. Are animals merely reacting to their environment, or do they intend to convey information to one another? For example, upon discovering food a bird may make a characteristic call that attracts other birds to the food. Was the call the equivalent of “Yay! Food!” — unintentionally attracting other birds? Or, was it more like, “Hey guys, come check out the food I found!”?
Signs of intentionality have been shown in many animals. Ground squirrels, Siamese fighting fish, chickens and even fruit flies change their signals depending on who is around to receive them, an indication that they have some voluntary control over those signals. Other animals seem to intentionally “show” others something, like a dog who looks back and forth between a human and a bag of treats or a hidden toy, perhaps even adding a bark to get the human’s attention first. Ravens also appear to show objects to other ravens by holding them in their beak — usually only if the other bird is paying attention.
Some of the best recent evidence for intentional communication in birds comes from observations of wild Arabian babblers at the Shezaf Nature Reserve in Israel. A team led by ethologist Yitzchak Ben-Mocha, recorded adult babblers coaxing fledglings to move to a new shelter. Adults call and wave their wings in front of fledglings and then move toward the shelter. If a youngster doesn’t follow immediately or stops along the way, the adult comes back and does the song and dance again and again until the fledgling complies.
Scientists call such signals first-order intentional communication. Some researchers argue that a more relevant precursor to language like ours is second-order intentional communication. This involves the signaler knowing something about the receiver’s mind, such as the bird who found food knowing another bird was unaware of the food and calling to intentionally inform the ignorant bird. As you may have guessed, this sort of mental attribution is a hard behavior to test.
Other scientists are taking a different tack to try to understand what underlies such communication by comparing the brain structures that enable vocal learning in songbirds and humans.
Arabian babblers show that their communication has meaning and intent in this video where an adult is attempting to coax a juvenile bird to a new shelter by calling and flapping its wings. The adult signals to the juvenile before heading toward the shelter, checks to see if the young bird is following and then doubles back to repeat the signal.
CREDIT: Y. BEN MOCHA ET AL / ROYAL SOCIETY PROCEEDINGS B 2019
Deeper connections
Despite humans and birds being only very distantly related — their last common ancestor lived more than 300 million years ago — they have remarkably similar brain circuitry for vocal learning. Nonhuman primates, our closest relatives, lack a specialized circuit for imitating sounds, leading scientists to conclude that this ability did not come from a common ancestor. It must have evolved independently in birds — an example of what is known as evolutionary convergence.
“There is this assumption that species more closely related to us are going to be most like us. And that is true for many traits,” says the Rockefeller’s Jarvis. “But it’s not true for every trait.”
Jarvis studies the evolution of language by looking at the brains of songbirds. Animals that make only innate sounds control the musculature that creates those sounds through a circuit in the brain stem, an area near the spinal cord that regulates automatic functions like breathing and heartbeat. “What has happened is humans and songbirds have evolved this new forebrain circuit for learned sounds that has taken control of the brain stem circuit for innate sounds,” Jarvis says.
His theory for how similar vocal-learning circuits evolved multiple times in distant species is that they were built from an adjacent circuit that controls the learning of some movements. “The spoken-language brain circuit in humans and the song-learning circuit in birds,” Jarvis argues, “evolved by a whole duplication of the surrounding motor pathway.”
How an entire brain circuit could be duplicated is unclear, he says, but it could be similar to how genes sometimes are duplicated and then co-opted for other purposes. However they evolved, vocal-learning birds and humans have these rare analogous brain circuits that enable them to learn and imitate sound. This suggests that behavioral scientists who’ve been trying to learn about human language by studying how distantly related birds such as zebra finches communicate are onto something.
“I think we humans tend to overestimate how different we are,” Jarvis says. Even he has observed zebra finches singing in the lab or a starling singing in a tree and thought that it just seemed so different from what humans do. “And then a year later, we’re making a discovery about the connectivity of the circuit, or the mechanism of how it’s producing the sounds, and it’s so much like humans.”
