Singing Lessons
by Mary-Russell Roberson
Listen now 28:13

Everyone knows that birds sing. Many people know that some species of whales sing. But did you know that bats and mice sing? Or that some animals produce songs by instinct, while others learn their songs from their parents? Biologists, neurobiologists, psychiatrists, speech therapists, and others are delving into the secrets of learned singing. What they are discovering is illuminating the workings of the brains of birds and humans, and may eventually lead to new treatments for human speech disorders and learning disabilities.

What is Song?

fairy bluebird
Fairy bluebirds (Irena puella) and other oscine birds learn songs when they are young. (Jessie Cohen/NZP)

In the animal world, a song is a specific type of vocalization. Many animals vocalize in one way or another, by grunting, snorting, or barking, for example. Some vocalizations have a communication function: Alarm calls alert others to the presence of a predator, agonistic calls communicate aggression, begging calls solicit food from parents, and mating calls advertise availability. Songs are also used for communication, but typically are much more elaborate than calls. The difference between a song and a call, while fairly easy to recognize intuitively, is difficult to explain, perhaps because "song" is such a well-worn word in everyday English. When asked to define song, scientists who study it usually give a disclaimer first. Richard Mooney, a neurobiologist at Duke University Medical Center, who studies how birds learn and produce song, says, "Song is a loaded word. It has many meanings to humans and most of those relate to special human behaviors." And Timothy Holy, assistant professor of neurobiology at Washington University in St. Louis, says, "I don't think there is a single definition that everyone subscribes to."

Those caveats notwithstanding, most scientists agree that a song is longer and more complex than a call, and has a melodic and rhythmic structure, often involving repeated patterns over both short time periods (tens of milliseconds) and longer time periods (seconds). Songs typically also include more variation in frequency (pitch) and intensity (volume) than calls do. Call and song are really two ends of a continuum, and different people draw the dividing line at different places.

A chick's peep, a wolf's howl, and a lion's roar are all calls. The white-throated sparrow's (Zonotrichia albicollis) "Old Sam Peabody, Peabody, Peabody," a pre-dawn gibbon (Hylobates spp.) duet, and a humpback whale's (Megaptera novaeangliae) haunting melody are all songs. A chorus of spring peepers (Pseudacris crucifer) could be considered either. And a pig's grunt is just a vocalization.

In most cases, the function of song relates to territory, breeding, or social cohesiveness. In temperate climates, male songbirds sing to defend their territories and announce their interest in mating. Numerous studies using real birds and tape-recorded songs have shown that the sound of a male songbird singing repels other males of that species, and attracts females that are ready to breed. "If you render a male incapable of singing, he loses his territory and he loses his mate," Mooney says. When birds from temperate zones migrate to the tropics, they rarely sing.

Birdsong functions a little differently for birds that live in tropical zones year-round, according to Eugene Morton, senior scientist emeritus at the Smithsonian National Zoo's Center for Research and Conservation and coauthor of the 1992 book Animal Talk: Science and the Voices of Nature. For these tropical birds, singing is more about defending territory and less about breeding, and females and males sing year-round. Morton also points out the evolutionary advantage of singing: "The real use of these songs is to be able to defend a territory without having to patrol" its perimeter on the wing. Flying uses a lot of energy, so it is in the best interests of birds to minimize flying time.

Innate and Learned Songs
Most animal singers perform innate songs, but a select few learn their songs from their parents and other adults. "There are animals that vocalize, and a subset of those sing, and a very small subset of those learn to sing," Mooney says.

While many birds sing, only parrots, some hummingbirds, and true songbirds (members of the Oscine suborder of Passeriformes, which includes warblers, finches, sparrows, and thrushes) learn to sing, by listening to adult members of their own species. Any oscine bird raised in isolation will sing as an adult, but its song will be strange and highly simplified. Birds that do not learn to sing, such as phoebes (Sayornis spp.), sing the same songs as adults whether raised normally or in isolation from other phoebes.

Pine Barrens tree frog
A Pine Barrens tree frog (Hyla andersonii) expands its vocal sac to produce a mating call. Some frogs call, while other, such as the concave-eared torrent frog, sing.

Several frog species sing. Male concave-eared torrent frogs (Amolops tormotus) from the Huangshan Hot Springs region of China produce tonal vocalizations that sound like bird songs, including "warbles," which are rapid up-and-down changes in frequency. The frogs have a vast repertoire—in 12 hours of recordings of 21 males, the researchers found no identical songs. Interestingly, even though these frogs' songs are audible to humans, they also contain ultrasonic harmonics, which other frogs can hear over the ambient noise of fast-rushing streams. Scientists are conducting research to determine whether these songs are learned or innate.

Among mammals, some whales, and perhaps bats and mice, learn to sing. "There's no evidence for vocal learning in other groups," Mooney says, "although we haven't studied everything that's out there." Human speech is a learned vocalization, but it has some attributes that put it into a class of its own.

No one has yet found any evidence for learned vocalization in nonhuman primates. About 25 species of primates in four genera sing, but their songs are innate: A macaque (Macaca spp.) raised by a different species of monkey makes macaque sounds when it grows up, for example. Gibbons produce loud and long songs, usually duets performed by mates. These apes, which live in the rainforests of Southeast Asia, sing in ten- to 30-minute bouts near dawn. Males and females sing different parts, alternating in a modified call-and-response structure. Although complex, these songs are not learned. A hybrid female gibbon born in a zoo inevitably sings something resembling a mixture of the female songs of her parents' species. This is true even if she never hears the female song of her father's species and hears only the female song of her mother's species.

Mammals That Sing
In the 1960s and '70s, scientists discovered that male humpback whales produce very low-pitched, complex songs with patterns and structure. "Humpback whale songs are very much like birdsong," says Salvatore Cerchio, a scientist at the Wildlife Conservation Society. For example, their song structure, called eventual variety, is also found in some birdsong: They repeat phrase A several times, then phrase B several times, then phrase C several times, and so on, then start over with A. Whales also sing for reasons similar to birds'. "We know it's a male breeding display," Cerchio says. "What's not clear is whether the song is a signal directed at males or females or both. It's currently under study and debate."

Most scientists believe that humpbacks learn their songs, because individual whales sing slightly different versions. However, Cerchio wonders if there are some innate aspects. "Humpback whales are quite unusual in that their songs show rapid change over time," he says. "All males in a given population sing a very similar song at any one time and they all make the same subtle, gradual changes to it over the course of the season." He and some of his colleagues studied the songs of two humpback whale populations during the breeding season (late winter and early spring) of 1991. Although the two populations were separated by more than 3,000 miles, the songs of both groups changed in a similar way over the course of a few months. There is no evidence that whales swim back and forth between the two areas, and even if they did, it would take about 40 days to swim from one to the other. For these reasons, Cerchio believes something more than cultural transmission might be going on. He suggests that there may be some innate rules that govern the way songs are changed.

Some other baleen whales sing, including bowhead (Balaena mysticetus), fin (Balaenoptera physalus), and blue whales (Balaenoptera musculus). However, none produces songs as complex as humpbacks'. "It's very possible that other species of baleen whale have some songlike vocalizations; we just don't have enough information yet," says Cerchio.

killer whale
Toothed whales, including killer whales, do not sing, but humpbacks and some other baleen whales do.

Toothed whales, including dolphins, killer whales, and sperm whales, do not seem to make songlike vocalizations at all, says Cerchio, but they do produce a wide variety of other vocalizations, such as clicks, calls, and whistles. They also echolocate, meaning they analyze echoes from their vocalizations to perceive the position, shape, and velocity of nearby objects.

Bats, which also echolocate, have at least some singers in their midst. Scientists have recently recorded and analyzed the previously unknown courtship songs of the sac-winged bat (Saccopteryx bilineata) of Costa Rica. For the most part, the songs are too high-pitched for humans to hear, which is probably why they were not discovered earlier. Male sac-winged bats typically sing while hovering in front of a female or moving fitfully on a nearby perch. The songs vary from bat to bat, but whether sac-winged bats learn their songs, and whether other bat species sing, remains to be seen.

Lab mice produce ultrasonic songs, too, as Washington University's Timothy Holy recently discovered by accident. Holy studies the neurobiology of the olfactory system, and wanted to identify behaviors in male lab mice that signal the presence of female pheromones. Male mice sometimes produce ultrasonic vocalizations when they smell females, and Holy thought those vocalizations might be reliable "pheromone indicators." To test this theory, he and colleague Zhongsheng Guo placed male mice in sound chambers with cotton swabs soaked in mouse urine and recorded the sounds the mice produced.

Other scientists have studied mouse vocalizations, but never recorded as much acoustical detail or scrutinized the results as thoroughly as Holy and Guo, who used a sophisticated microphone and a computer to help identify different sounds and patterns. "Also, I wrote software to switch [the vocalizations] into a range we could hear, so we could listen to them," Holy says. "They were much more interesting than I expected them to be."

When Holy analyzed the sounds of 45 male mice, he discovered vocalizations with enough complexity and pattern to rival the songs of some birds. He identified three basic types of sound components, or pitch jumps, that when combined make several different recognizable "syllables." The presence of the syllables and the fact that they tended to occur in particular sequences led him to characterize the vocalizations as songs.

Holy's research was published in the journal PLoS Biology in December 2005 and has generated intense interest among scientists. What function do the songs serve? Do female mice sing? Do different genetic strains of lab mice sing different songs? Do wild mice have a wider repertoire than lab mice? The most frequently asked question is whether mice learn songs. Holy's finding that an individual mouse's songs were more similar to each other than to songs of other, genetically identical mice suggests that their songs are learned rather than innate. Holy says, "We are not going to be [looking into that question] ourselves…but we have friends who seem to be looking into it."

How Do Birds Learn to Sing?
Scientists know far more about singing birds than any other singing animals. Whales are difficult to study because of their size and habitat, and of course they can't be brought into a lab. Bats rarely display behavior related to territory or courtship when in captivity, and their songs, like those of mice, have only recently been discovered.

Songbirds have been studied in the field for hundreds of years and in the lab for decades. Scientists are still investigating how and why birds learn to sing.

For example, recent studies of zebra finches (Taeniopygia guttata) and European starlings (Sturnus vulgaris) illuminate another function of birdsong: Females can evaluate a male's past by listening to his song. Males that experienced an unusual amount of hunger or stress when young produce less complex songs than other males as adults, which puts them at a disadvantage because females seem to prefer males that sing complex songs.

Neurobiologists have uncovered striking differences in the brains of birds that learn to sing and those that don't. "There's a very extensive and well-defined neural circuit in the songbird's brain that ultimately forms connections with the motor neurons used in singing," Mooney says. "The circuitry is well developed in birds that learn to sing and absent in nonlearners. In species where the males sing and the females don't, these special song structures are really big in the male and really small or almost absent in the female."

Baby songbirds learn their songs by mimicking adults. If a baby songbird is isolated from its parents and exposed to unrelated birds of the same species, birds of other species, or tape recordings, it will learn the songs that it heard rather than those of its biological parents. And in species that learn to sing, birds that are born deaf or raised in isolation will not sing normal songs.

Laboratory studies with birds, mostly zebra finches and white-crowned sparrows (Zonotrichia leucophrys), have demonstrated that learning to sing takes place in two phases: In the first step, young songbirds listen to and memorize the songs of an adult (called a tutor), and in the second step, the young begin to vocalize these memorized songs, refining them through practice. In many species there is an interval of time between these two phases, because the adult males leave or stop singing before the young begin to practice. Eight months separate the two phases for swamp sparrows (Melospiza georgiana), indicating they have a prodigious memory. In other songbirds, such as zebra finches, the two phases overlap by about a month.

It's obvious that hearing is necessary for the first, listening phase. But hearing is just as critical for the vocal practice phase. That's because the bird must hear itself, compare its sound to that of the adult, and make precise adjustments to more closely match the tutor. Learning to sing depends on this complex auditory feedback system. Birds that can hear in the first phase but are deaf in the second phase will sing abnormal songs.

Male cardinals take longer to learn their songs than females do.

Young birds learn to sing with little apparent effort, but after sexual maturity, most are no longer able to learn new songs. (A few birds, including canaries and starlings, can learn new songs into adulthood; they are called open learners.) The period of time during which birds can memorize new songs is called the critical or sensitive period. For example, white-crowned sparrows memorize songs easily between 20 and 50 days of age, and then gradually lose this ability. After about 100 or 150 days, most can no longer learn new songs. Scientists are investigating what signals the critical period to close; possibilities include hormones, quality of sensory experience, and social factors.

Ayako Yamaguchi, an assistant professor of biology at Boston University, discovered that the sensitive period closes at drastically different times in male and female northern cardinals (Cardinalis cardinalis), one of the few species that live in temperate regions of North America in which both males and females sing. Cardinals are actually tropical birds that have expanded their range northward in the last 100 years, probably due to the presence of bird feeders.

The sensitive phase for cardinals begins when they are between ten and 20 days old. "Males continue to memorize until they reach seven, eight, or nine months of age while females are done by three months of age," Yamaguchi says. She speculates that males may need a longer sensitive period because, as they mature and disperse, they must match their songs to those of males already living in their new territories in order to fit in and successfully establish their own territories. Yamaguchi's research suggests that male and female cardinals" brains function differently even though they are raised in the same environment.

Song and Speech
The process by which birds learn to sing is remarkably similar to how human babies learn to speak. Babies learn speech from their parents and others around them, just as birds learn birdsong. A Japanese baby raised in an English-speaking environment learns to speak English, not Japanese. Deaf babies do not spontaneously produce spoken language, but will usually learn sign language if they are exposed to it.

Human babies also go through listening and practicing phases, but without the interval between them that some birds experience. Hearing is essential in both phases, just as in songbirds. When children become deaf during the second, practicing phase, they do not learn to speak normally without extensive coaching; if they were already talking when they became deaf, their speech often deteriorates rapidly. People who become deaf as adults (that is, after the second phase), are usually able to maintain fairly normal-sounding speech.

Humans, like birds, have a critical period in childhood in which learning language is nearly effortless. Adults can learn foreign languages, but speak with an accent, and have more difficulty with grammar and comprehension than children learning a second language.

If human speech is analogous to birdsong, what is human song? Is it just another form of speech? Mooney says no: "Although allied with speech, it's something else. The human capacity for music is really an unusual trait."

Human brains process speech and music differently. Someone listening to music with or without lyrics typically shows more activity in the right side of the brain, while someone listening to speech typically shows more activity in the left side of the brain. Human music also sounds different from animal song—it has a stronger rhythmic component (a steady beat) and is less stereotypic, allowing for improvisation within some basic rules of melody and rhythm. Songs composed by two different humans, say Mozart and Janis Joplin, sound much less alike than the songs of two different white-crowned sparrows.

Medical Implications of Song Research
The similarities between birdsong and speech are fueling research that one day could lead to better treatments for speech disorders. For example, finding the triggers that close the critical period in birds could lead to figuring out how to extend or "re-open" the critical period in humans. Such a discovery might make it possible to improve speech skills in people with dyslexia, autism, hearing impairments, and people who have suffered a stroke or brain injury.

Mice, being mammals, are potentially even more analogous to humans. Furthermore, scientists have a fairly complete map of the mouse genome. "The mouse could be an interesting organism for studying more about vocal production and communicating," Holy explains, "because mice, more so than birds, are suited to studies in which genes are being controlled and manipulated. It would give an extra set of tools to scientists who are interested in how brains produce the commands that make sounds."

One promising topic for research in mice is a genetic sequence called the Foxp2 transcription factor, which, according to several studies, turns on when a bird sings. "We don't know what its specific role in the song is: It may be it's important to generate the motor movements or it might be something more high-level in the brain," says Holy. In certain human families that have a high prevalence of speech disorders, scientists have found a mutation in the Foxp2 genetic sequence. There is already some preliminary evidence that mice with the compromised transcription factor don't seem to vocalize normally. If this is the case, scientists could look for medical treatments that would improve communication in these mice, which could eventually lead to treatments for humans with language disorders. But, says Holy, "We still have to be cautious for how close an analogy it is."

Whatever similarities there may be between animal song and human speech, there are differences as well. Vertebrate brains have evolved to process two different types of sound—animal vocalizations, and environmental sounds such as snapping twigs, which may signal the approach of a predator. Animal vocalizations tend to be more harmonic and to occur in a more narrow frequency range than environmental sounds. Michael Lewicki of Carnegie Mellon University in Pittsburgh has made mathematical models that show the relationship between incoming sound and the activity in the mammalian auditory nerve. The model for animal vocalizations looks different than the model for environmental sounds. When Lewicki calculated the model for human speech sounds, it fell in between animal vocalizations and environmental sounds. This suggests that as human speech evolved, it made efficient use of the brain's pre-existing ability to process both types of sound.

The most obvious gulf between the two is that animal song, unlike human language, has no semantic content. When animals sing, they may be communicating something about their territory or their availability for mating, but they cannot say something as complicated as, "Last week I saw five rhinoceroses—four adults and one baby—drinking at the medium-size creek that empties into the river north of here."

Nevertheless, as countless scientists are demonstrating, humans stand to learn much from studying how animals learn to sing.

—Mary-Russell Roberson is a science writer living in Durham, North Carolina.


ZooGoer 35(3) 2006. Copyright 2006 Friends of the National Zoo. All rights reserved.



Have a Comment?



You're responsible for the content of your post. Inappropriate material will be removed from the site.

Email addresses are not displayed anywhere on the site. They are used by ZooGoer staff to respond directly, when necessary, to those posting comments.


Page Controls