Dolphin Language


I’ve long been fascinated by language. Most of us can’t remember a time when we weren’t able to make ourselves understood with words, so few people realize how completely amazing language is.  If we didn’t have it, we’d still be living in caves and eating roots. More than anything else, language is what makes us human. But how did it come to be? Surely it evolved from something, but what? More and more scientists are studying this thorny question. So far, there are no real answers (that I’m aware of), but the research is fascinating.

Some of that research is looking into animal communication, hoping to find commonalities and precursors. When I read John C. Lilly’s The Mind of the Dolphin many years ago, it seemed to me that we were on the brink of a great discovery: Another sentient life form on Earth might have language! Unfortunately, all the research done since then hasn’t proved it one way or the other. Still, animal language research has yielded surprising insights, and it seems clear that dolphins communicate in a highly sophisticated manner. But is it language? Recently, there have been tantalizing hints. The possibility still exists.

Jim Mastro
(Originally published in The Sciences as “Signal to Noise”)

I’m standing at the edge of a pool filled with splashing bottlenosed dolphins. Several swim in tight circles, surfacing to breathe amid sprays of water. A few ring the pool’s edge, offering their rubbery skin to the outstretched hands of squealing children and smiling parents. Above an immovable smile, the dolphins’ eyes glint with an intelligence the people seem to sense.

One dolphin approaches me, studying me with an inscrutable gaze. I cannot help but feel I am locking eyes with a mind as complex as my own. With a shake of its head, the dolphin squeaks and whistles at me. Unconsciously, as though obligated to respond, I raise my hands and shrug.

The dolphin vocalizes more insistently, then lurches away to join its fellows.

What just happened? Was the dolphin trying to speak to me, in some abstract, syntactic language–or was it just making noise? No question is more vexing or more dependent upon psychology, philosophy and technology. Part of the problem is that investigators trying to document language ability in animals must extrapolate from a sample of one: only human language is considered true language. Even if a complex and syntactic animal communication system existed, its form might be so different that we would not even recognize it as language. To make matters more complicated, linguists have not been able to devise a concise, universally accepted definition of human language.

Linguists and biologists readily agree that animals communicate with one another. But communication can be simply the transfer of  information–aggressive warnings, for instance, or sexual receptivity–and its channels can be chemical, tactile, visual or auditory. Language, by contrast, is understood as something far richer. Since Descartes, the generally accepted belief among linguists has been that only people have this capability.

Animal language specialists have continually questioned that belief. One by one, each of the supposedly defining characteristics of human language–prevarication, symbolic representation and “displacement,” the act of referring to objects that are not present and to past or future events–have been demonstrated in other animals. Among the species that have shown themselves capable of one or more of those abilities are chimpanzees, bonobos and dolphins, and of those three, the dolphin has proved to be the most linguistically promising. For though it is clear that the language ability of chimps and bonobos is constrained by their relatively small brains and their inability to produce more than a few simple sounds, dolphins have neither of those limitations.

The idea that dolphins have language was first proposed scientifically more than thirty years ago by the neurophysiologist John C. Lilly, in his book The Mind of the Dolphin. (Lilly confined his discussion to the Atlantic bottlenose dolphin, Tursiops truncatus, and so will I, unless otherwise noted.) I remember reading the book as a teenager, awed by the possibility of conversing with another life-form. Biologists were similarly intrigued, and they eagerly began to look for supporting evidence. Many, including Lilly, set up experiments designed to encourage the transfer of vocal information between dolphins, then recorded the exchanges. Unfortunately, they were quickly disappointed. Barring a few tantalizing exceptions, their experiments yielded no conclusive evidence. By the 1980s most investigators had pretty much dropped the issue of dolphin language altogether.

That state of affairs had little effect on the popular imagination, however. Surrounded by performing dolphins at aquariums and television shows such as Flipper, most people were more than willing to accept the notion of talking dolphins. Recently I took an informal poll at the university where I teach. Most of my colleagues are not trained in biology, but they

are intelligent and well-read. When I asked whether they thought dolphins had a language, each of them said yes. So did everyone else I asked, from maintenance workers to students. Many thought the answer obvious. “Look in their eyes,” one person said. “You can see how intelligent they are.”


I have looked in their eyes. Many times.

Several years ago I worked at a marine mammal research facility, training California sea lions. Each animal had its own holding pen. The pens were connected to a large pool where a female dolphin named White lived. Every evening I let the sea lions into the big pool, and every morning I herded them back to their pens. White often watched, but she and I never interacted.

One day the sea lions refused to return to their pens. I tried plunging a long-handled net into the water to spook them out, but they ignored it. As usual, White watched, silent and apparently uninterested. Finally, after several futile minutes, I threw down the net, exasperated. Just then White became agitated. She swam directly in front of me and began vocalizing rapidly. For no reason in particular, and certainly not because I expected anything, I waved my arm and said, “White, why don’t you do me a favor and chase those guys out of there?”

And that is exactly what she did. Without hesitation, she took off straight toward the sea lions. They came flying out of the tank as though shot from a cannon, their eyes crazy-wide, and fled to their pens. I was stunned. White returned to her spot in front of me and resumed vocalizing. When I came out of shock, I threw her some fish, which she ignored with a jerk of her head, as though she had been insulted. We locked eyes, and in that instant I saw a keen mind at work.

When neurobiologists first examined dolphin brains for hard evidence of intelligence, they were not disappointed. Tursiops brains are large, with a well-developed and heavily convoluted cortex. The dolphin’s encephalization quotient–the ratio of brain to body size–is the greatest of any nonhuman mammal, almost twice as great as that of chimpanzees.

Some biologists suggested that all that brainpower supports the dolphins’ sophisticated echolocation system, which they use to hunt. Yet some bats have an equally impressive echolocation apparatus, which they support with one of the smallest mammalian brains. It seems unlikely that dolphins would need vastly larger brains to do the same thing. Something more must be going on.

That something more could be tied to the dolphins’ complex social behavior. Brain size and the complexity of mammalian social systems are highly correlated. In his book Grooming, Gossip, and the Evolution of Language, the evolutionary psychologist Robin I.M. Dunbar of the University of Liverpool argues that the size of the primate neocortex–the most evolutionarily recent part of the brain–is closely related to overall troop size and the size of grooming coalitions. (Both troops and the coalitions within them are crucial elements in the primate social structure.) The larger the average troop size for a species, Dunbar found, the larger the grooming coalitions–and the larger the neocortex. The larger neocortex enables primates to keep track of a constellation of friends, enemies and relatives–and how to deal with each of them–according to abstract mental categories. Human language, says Dunbar, was a natural evolutionary offshoot of that process.


Many linguists and anthropologists have wondered whether dolphins might not be subject to similar evolutionary pressures. Indeed, research suggests that dolphin societies may even rival primate societies in complexity. Dolphins frequently act cooperatively, and their behavior strongly suggests sophisticated intercommunication. Groups of mothers, for instance, form swimming playpens for their progeny, keeping calves safely at the center of a circle. And unrelated male dolphins make enduring alliances that may function much like primate grooming coalitions. Dolphins also hunt cooperatively. While the group keeps a school of fish herded in the center of a circle, individual dolphins take turns darting in to feed. What motivates the others to keep herding fish once they’ve had their fill? How do the dolphins decide which of them goes in, and when? And how is that information conveyed, if not by language?

There is more. Dolphins appear to be altruistic and will assist others, even if they are not family members. That behavior implies a dolphin “theory of mind”: an awareness of self and an awareness that others perceive themselves in a similar way. It also suggests that dolphins can empathize, recognize the need for assistance, and figure out what to do about it.

That is not to say that dolphins are entirely benevolent. Their high intelligence also gives them the all-too-familiar human capacity for pettiness and brutality. Rachel A. Smolker, a biologist at the University of Vermont, notes that dolphins sometimes hold grudges against one another for days. Some recent studies note an even darker side. Two Scottish investigators, the marine zoologist Ben Wilson of the University of Aberdeen and the veterinarian Harry M. Ross of the Scottish Agricultural College in Inverness, have observed bottlenosed dolphins off the Scottish coast killing harbor porpoises, for no apparent reason. In addition, both the Scottish investigators and their American counterparts, including Susan G. Barco of the Virginia Marine Science Museum in Virginia Beach and Ann Pabst and William A. McLellan of the University of North Carolina at Wilmington, have observed numerous cases of dolphin infanticide. In most instances the dolphin calf was severely battered, and the forensic evidence pointed clearly to adult dolphins.

Although other mammals engage in infanticide, particularly when resources are scarce or when males are competing to father young, the dolphins’ killing of the members of another species is harder to explain. Dolphins do not eat porpoises, and the two species do not generally compete with each other for food. Here, as in many other cases, the more dolphins are studied, the more surprising and complex their behavior seems.

Other aspects of dolphin behavior have close human parallels. It is no surprise, for instance, that juvenile dolphins play–most young mammals do. But adult dolphins play too, something rare in other mammals but common, of course, in people. Moreover, dolphins are sexually promiscuous, engaging in both heterosexual and homosexual activity. They seem to use sex as a means of cementing social relationships, a behavior they have in common with only two other species: bonobos and us.

Of course, all the social and neural sophistication in the world Would not necessarily suggest a language ability if dolphins had no concrete means of transmitting information. But they do.


In a 1981 article in the journal Science, the biologists R. Stuart Mackay and H.M. Liaw, then at Boston University, described dolphin sounds as “. . . blats, bleats, chirps, clicks, creaks, pulses, quacks, racs, rasps, squeals, squawks, wails, and whistles.” Some of those sounds, biologists now know, are emitted for echolocation. But what about all the others? The most conspicuous are the whistles, and dolphins whistle a great deal. Early studies seemed to show that, unlike the repetitive calls of other animals, dolphin whistle repertoires were highly variable. Furthermore, dolphins are good mimics. Biologists have played computer-generated whistles to dolphins, and the dolphins imitated them. Individual dolphins also proved capable of combining two of the computer-generated whistles, even though they had never heard the whistles played together–a feat reminiscent of the way children learn language.

Brenda McCowan, a veteran dolphin investigator at the University of California, Davis, relates an incident that dramatically illustrates what appears to be a purposeful act of communication. McCowan was training a dolphin to present its flukes to her, so that she could draw blood samples from the animal. She would hold the flukes for a few seconds, blow a whistle to let the dolphin know it had done the task correctly, then reward the dolphin with a fish. Her goal was to get the dolphin to allow her to hold its flukes for about thirty seconds.

During one session McCowan had been holding the dolphin’s flukes for about twenty seconds when she heard a whistle. She looked around, wondering who had disrupted her training period, but she saw no one. Suddenly, she realized that the noise came from the dolphin she was holding: it had imitated her whistle exactly. Instead of simply pulling away, the animal had communicated, “Time to let go!”–in a way it must have known McCowan would understand.

Stories such as McCowan’s are compelling. Yet, even after thirty years of behavioral and cognitive research that has revealed ample evidence of intelligence, social sophistication, and vocal ability, no scientist has uncovered clear evidence that dolphins possess a language. The question is: Why not?

Not long ago, there seemed to be a reasonable answer. Although Lilly’s work had initially galvanized the scientific community, an analysis of hundreds of recordings of dolphin vocalizations seemed to show that the dolphin’s whistle repertoire was actually quite limited. Early experiments had demonstrated that dolphins possessed signature whistles, unique to each individual. Later, however, several investigators determined that between 48 percent and 90 percent of any given dolphin’s vocalizations were simply repetitions of the signature whistle. So though dolphins were apparently the only animals on earth besides humans that gave themselves names, they seemed unable to do much more than repeat those names over and over.


At the Kewalo Basin Marine Mammal Laboratory at the University of Hawai`i at Manoa, a blindfolded, bikini-clad trainer stands on a platform at the edge of a pool. A dolphin watches her carefully. At a signal from the psychologist Louis M. Herman, the trainer gestures with her arms. Without hesitation, the dolphin, a female named Akeakamai (Ake for short), swims to one of two floating frisbees–the one on her right–takes it in her mouth, and carries it to a surfboard bobbing in the middle of the pool.

Ake’s behavior might seem no different from the tricks dolphins routinely perform in shows at public aquariums. In this case, though, the arm gestures were not signals but words, and the sentence they formed–surfboard right frisbee fetch–was one the dolphin had never seen. Yet she responded perfectly. To Herman, that behavior clearly suggests language comprehension.

In fact, Herman and his students and colleagues have taught dolphins two artificial languages. One dolphin, Phoenix, learned an acoustic language made up entirely of computer-generated whistles. Ake learned a gestural language. Both dolphins were taught words that referred to various objects in their tank, such as frisbee, ball and hoop. They also learned action words–spit, toss and tail-touch, for instance–that require a direct object, and relational action words, such as fetch (take one object to another) and in (place one object inside another). Finally, Ake was taught the modifiers left and right, and Phoenix was taught bottom and surface.

After Herman was convinced the dolphins understood what the words meant by themselves, he combined them into sentences and tested the dolphins for comprehension. For example, to signal her understanding of the sentence ball fetch hoop, Phoenix had to take the ball to the hoop.

The dolphins’ performance was remarkable. Their comprehension ranged from 65 percent for the most difficult sentences, to 85 percent for the simplest ones–well above chance levels. In addition, the animals were able to make generalizations. For example, they understood the word ball to mean any ball. Ake also demonstrated that she could report not only on the presence of objects in her pool, but also on their absence; she had associated lexical symbols–words–with their referents, a key language requirement. According to Herman, those results provide the first evidence of syntactic understanding by animals.


Research such as Herman’s, which makes use of captive dolphins, has been important for developing a basic understanding of these animals, but there are obvious limitations. Captive behavior might no more reflect the realities of a dolphin’s ocean life than the behavior of a prison inmate reflects the activities of the rest of human society. That caveat is particularly true when it comes to dolphins’ social interactions and vocalizations.

For example, signature whistles may turn out to be a much smaller part of the dolphin vocal repertoire than they once seemed to be. As Brenda McCowan and her colleague Diana L. Reiss of the New York Aquarium in Brooklyn have pointed out, many of the studies that found a prevalence of signature whistles were done with captive animals. In retrospect, such an approach could readily have skewed the data. A human child, captured by aliens that did not understand her language, might not engage in typical language behavior either. Instead, she might yell “Help!” a lot.

But studying dolphins in their natural environment is no easy task. Making observations of dolphin social interactions in deep, murky water is nearly impossible. Even in clear water it takes years of painstaking work to identify individuals and track their interactions. At sea, just following the animals can be difficult–never mind trying to collect useful data. In spite of those obstacles, Rachel Smolker and other workers have begun to look closely at dolphin vocalizations in the wild, and have tried to correlate the vocalizations with behavior. Smolker has found that the whistles are well structured and are often deliberately modified by the dolphins. Her research suggests that the problem of the dolphins’ “limited repertoire” could turn out to be an artifact of our own limited technology.

Many dolphin sounds fall outside the range of human hearing; only with the aid of electronics can people detect the sounds at all. Furthermore, dolphins can produce and process a large number of sounds in a short time. Anyone who has had to learn a foreign language knows how hard it can be to parse meaningful units out of a rush of syllables. Imagine how much more difficult it would be if the sounds came from a nonhuman creature “speaking” a hundred or a thousand times faster than people do!

McCowan and Reiss have been analyzing dolphin whistle repertoires with new, fast computers and highly sophisticated programs. They have found that the whistles are much more complex than anyone had previously guessed. Whistles once thought to be mere variants of the signature whistle now appear quite distinct from it. In fact, McCowan has recorded and identified more than a hundred different kinds of whistles, directly contradicting the prevailing belief that dolphins had only a few. And she is crystal clear about the implications of her finding: “The . . . amount of information that could be conveyed by these larger repertoires is huge.”

Dolphin vocal repertoires, according to McCowan, appear to be open systems. In other words, the units of communication–in this case, whistles–can be combined and recombined in an infinite number of ways (as humans do with words) to convey an infinite number of messages. In one group of studies, McCowan taught dolphins that various computer-generated whistles signified objects in their tank, such as a ball or a ring. She noted that the learning process seemed to mirror language acquisition in children: the dolphins mimicked the end of the artificial whistle first, then the beginning, and finally tied the two together. That way of breaking up the signals before imitating them suggests that dolphins’ whistles may be components of a larger system. Moreover, before the study ended, the dolphins had begun to mimic the sounds of the artificial whistles during play–and they used the sounds appropriately. For example, the whistle meaning “ball” was used almost exclusively when the dolphins were playing with a ball.

Most recently, McCowan has been applying algorithms from information theory to her data, in an effort to tease out patterns in the repertoires. In doing so, she has discovered intriguing similarities between dolphin whistle repertoires and human language. In fact, graphs of the repertoire structure of human language which show the frequency and patterns of word use, have been closely matched in complexity by graphs of dolphin whistles. McCowan is now developing a program that will be able to analyze and categorize whistles as they take place in “real time,” all in an effort to identify the possible rules of dolphin expression. “I’m not operating on the assumption that whistles are words,” she told me, “but I am suggesting that there may be more complexity here than we once thought.”


But even if dolphin vocalizations are complex and potentially rich, how can biologists determine whether they actually convey precise information? In an ongoing study, Louis Herman is trying to answer that question. He is assigning dolphins a task that requires them to communicate complex information: to earn a reward, two dolphins must invent and carry out a completely novel behavior on the spot, in tandem, without practice. In other words, they must jointly decide what to do, then communicate the information, as well as the timing of its execution, to each other.

So far, they are succeeding. Somehow, they seem able to pass information back and forth and to coordinate their activities. Whether their communication is abstract and vocal or merely a sophisticated form of body language is part of what Herman is trying to determine–and he has not yet ruled out the possibility that the dolphins are merely playing a very skilled game of Simon Says. To resolve the question, Herman is analyzing his recordings of the dolphins’ vocalizations during the tests, mining the data for evidence that the animals are passing information to each other acoustically. If they are, it is possible they are using some kind of language.

If it seems we have come full circle, we have. Research into dolphin language has vacillated from hope to disappointment and back to hope in the past thirty years. As a result, many who study animal behavior remain skeptical of the latest findings from investigators such as McCowan and Herman. The marine biologist David A. Kastak of the University of California, Santa Cruz, is typical: “What the dolphins are doing requires intelligence,” he says. “But it wouldn’t qualify as a language. They’re not animals using nouns and verbs.” On the other hand, other biologists have begun to think, “Maybe, just maybe . . .”


I’m sitting on my surfboard off the California coast, in water rich with plankton and stirred-up silt. The morning is calm, the sea glassy. Waves roll under me before they explode into white spray on the beach, fifty yards away.

This morning, as on many others, a pod of Pacific bottlenosed dolphins is wending its way up the coast. They are much like their Atlantic cousins, though a bit larger and darker. Their gracefully curved dorsal fins slice through the boundary between hydrosphere and atmosphere, between their world and mine. Their exhalations burst into the still air.

I have tried to attract their attention in many ways, from snapping my fingers underwater to rubbing my fiberglass board to generate squeaks. I have never gotten a reaction. Sometimes, though, when I am just watching them, a dolphin will poke its head from the water and look at me. Sometimes one or two will even surf with me, their sleek gray forms streaking like torpedoes through the waves.

The water here is often murky, and that makes it a fitting metaphor. When we humans are submerged in the dolphins’ opaque world–a world they navigate with ease–we are blind and helpless. The same may be true for the question of dolphin language. There, too, we are blinded–as much by our own prejudices as by the inherent difficulties of the subject.

Language defines us Homo sapiens, a fact that has been stressed repeatedly by philosophers and linguists from Descartes to Noam Chomsky. It is the magic instrument that seems to elevate us above the rest of the animal world. But I can’t help wondering, What would happen if supercomputers were set to the task of sorting, cataloging and evaluating dolphin vocalizations? What if the possibility of conversing with a nonhuman intelligence were enough motivation for society to channel serious resources into solving this great mystery? The rewards could be fabulous. If we happened to find out that dolphins do speak, I can scarcely imagine what we would learn–about the sea, about intelligence, and about language itself.