Humans have a long history of inventing names for elephants. There is Disney’s Dumbo, of course, and Jumbo, a 19th-century circus attraction, and Ruby, a famed painting elephant from the Phoenix Zoo in Arizona. But new research suggests wild African elephants may pick their own names, too—and use them to call and greet one another on the savanna.
Most animals are born with a fixed set of sounds for communication. A few, such as songbirds, can imitate other sounds they hear around them. Certain species of dolphins and parrots may learn to mimic human words for objects in their environment (like the proverbial Polly who wants a cracker). Much rarer, however, is an ability to assign vocal labels—something akin to names—to fellow members of the flock or pod. Bottlenose dolphins and orange-fronted parakeets are thought to address peers with specific calls, to which a unique receiver tends to react. But in these cases, the calls consist of one animal simply imitating the features of another’s habitual or trademark sound. (It’s as if someone went around constantly repeating their own name, like "Mark," and you mimicked it back at them.) A dolphin may copy another dolphin’s “signature whistle” to attract its attention, and the second animal will respond by repeating that very same whistle.
Now a study in Nature Ecology and Evolution reveals that African elephants engage in a type of communication previously unknown in nonhuman animals. Researchers used machine learning to analyze 469 contact, greeting and caregiving rumbles made by wild savanna elephants in Kenya and discovered that the animals use specific vocal labels to identify one another. Instead of imitating an individual’s signature call to signal a particular elephant’s identity, they come up with an original sound.
On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
“It might seem obvious to me and other elephant researchers that these calls are very specific because you see that a certain individual will respond, but no one has [previously] shown it,” says Caitlin O’Connell-Rodwell, an elephant behavioral ecologist at Harvard Medical School, who was not part of the new study.
Most people associate elephants with loud trumpeting, but the most common elephant sound is actually a low-frequency rumble. Some rumbles are so deep that humans cannot detect them—we hear only down to about 20 hertz, and these sounds reach as low as 5 Hz. Elephants, however, possess unique ear anatomy designed to pick up rumbles from as far away as 1.5 miles. This range is important because female savanna elephants live in an elaborate “fission-fusion” society: their extended family units split up and rejoin on a regular basis as they follow food resources and avoid predators.
Study co-author Joyce Poole, a co-founder of the charity ElephantVoices, has been studying elephants for almost five decades. She has long suspected that these cognitively advanced animals, which show empathy, mourn their dead and may imitate human speech, address one another from afar with something resembling names. She says she has often observed an elephant calling out and only one responding; the others “would just keep on feeding as if they hadn’t even heard her,” Poole says. “I did wonder, are they being just rude by not answering, or is it because she’s actually addressing somebody specific?”
Machine learning helped Poole and her colleagues locate vocal labels among the hundreds of previously recorded female elephant calls. To discern which specific elephants were addressing each other, Poole went back to her old field notes, looking for interactions such as “so-and-so was separate from so-and-so and was calling so-and-so.” In the end, the researchers identified 101 callers and 117 receivers. Next they measured acoustic features of the calls to assess whether they contained individual vocal labels. The model was able to predict the specific receiver of a call with a success rate far better than chance. The vocalizations weren’t simply dolphinlike imitation, either; the scientists found no statistical evidence of the animals copying one another.
The researchers then verified their findings in the field. They approached 17 wild elephants and played calls addressed to each of them through a speaker. Although the elephants rarely reacted to the “names” of other animals, they quickly responded to their own. “It’s a very sharp response,” Poole says. “The head jerks up, the ears spread out, the mouth opens wide.”
Study lead author Michael Pardo, a behavioral ecologist now at Cornell University, notes that the researchers couldn’t pinpoint which part of a call was labeling an individual and that the data were inconclusive on whether multiple elephants use the same name for an individual. It’s possible that the recordings simply lack sufficient examples or that different elephants use slightly different versions of the same name. To parse out the calls, the researchers may need to collect many more samples of elephant rumbles—a challenging task that would entail spending many hours in close proximity to the studied group. “Collecting this type of data is really intense,” says study senior author and Colorado State University biologist George Wittemyer.
Although humans are still just scratching the surface of elephant communication, Wittemyer suggests the existence of individual vocal labels in these calls indicates a capacity for abstract thinking. What’s more, he says, such labels’ emergence could add to our understanding of how human language might have evolved. In complex societies where members often lose sight of one another—such as those of elephants or our hominin ancestors—the need to identify and attract others’ attention might have driven cognitive abilities and language development.
“If you can name things without relying on imitation, then, at least in theory, it is possible for you to talk about a wider range of subjects because you could potentially come up with names for objects and ideas that don’t make any imitable sound,” Pardo says.
As O’Connell-Rodwell puts it, “Modifying a vocalization tailored to a specific individual does get you into a conversation. And that’s what I would say is a first step.”