There is a mystery on Tiwai Island. A large wildlife sanctuary in Sierra Leone, the island is home to pygmy hippopotamuses, hundreds of bird species and several species of primates, including Campbell’s monkeys. These monkeys communicate via an advanced language that primatologists and linguists have been studying for decades. Over time, experts nearly cracked the code behind monkey vocabulary.
And then came krak. In the Ivory Coast’s Tai Forest Campbell’s monkeys (Cercopithecus campbelli) use the term krak to indicate that a leopard is nearby and the term hok to warn of an eagle circling overheard. Primatologists indexed their monkey lexicon accordingly. But on Tiwai Island they found that those same monkeys used krak as a general alarm call—one that, occasionally, even referred to eagles.
“Why on Earth were they producing krak when they heard an eagle,” asks co-author Philippe Schlenker, a linguist at France’s National Center for Scientific Research and a Global Professor at New York University. “For some reason krak, which is a leopard in the Tai Forest, seems to be recycled as a general alarm call on Tiwai Island.”
In a paper published in the November 28 Linguistics and Philosophy Schlenker and his team applied logic and human linguistics to crack the krak code. Their findings imply that some monkey dialects can be just as sophisticated as human language.
Monkey ABCs
In 2009 a team of scientists travelled to Tai Forest with one mission—to terrify Campbell’s monkeys. Prior studies had collected monkey calls and then parsed vague meanings based on events that were already happening on the forest floor. But these primatologists set up realistic model leopards and played recordings of eagle screeches over loudspeakers. Their field experiments resulted in some of the best data available about how monkeys verbally respond to predators.
“When you really want to understand the meaning of a call, you need a field experiment,” Schlenker says. “If you yourself are the trigger, you have much better control over what causes each calling sequence in the first place.”
The primatologists initially pieced together a few basic monkey calls: krak, hok, krak-oo, hok-oo and boom.


They concluded that krak meant leopard and hok meant eagle. The oo suffix softened the meaning of each word—krak-oo indicated minor disturbances on the ground whereas hok-oo was reserved for less serious aerial threats, like falling branches. Boom meant that the coast was clear.


Trouble on Tiwai Island
But when the primatologists traveled to Tiwai Island, they found puzzling differences in monkey dialect. There are no leopards on Tiwai, yet the island monkeys consistently used krak. “On the island, in an eagle situation, you did find a lot of hok but you also found a lot of krak,” Schlenker says. “That was surprising because krak is supposed to be a leopard alarm call.”
One dominant theory emerged. Even when humans speak the same language, they tend to have differences in dialect. For example, Schlenker says, the word “pants” can mean fancy slacks to an American but, in Britain, it means long underwear. The linguists posited that krak was a general alarm call for any ground-based threat. In the forest that meant leopard. On the island, where there were no leopards, it had been adapted into a very general alarm that indicated anything but an eagle.
For a short while the puzzle seemed to have been solved—until they realized that island monkeys used krak for every kind of alarm. “When we looked at the distribution of krak, we found that it occurred in all sorts of possible situations, including in an eagle situation,” Schlenker says.
Cracking the krak code
Schlenker and his team decided to apply a more creative, linguistic approach to solve the krak mystery. First, he redefined the monkey calls, translating krak as a general alert, krak-oo as a minor alert and hok as an aerial alert. The important result was that krak-oo and hok were now much more specific terms than krak.
Here’s where it gets tricky: word meanings tend to be contextual. In human language, we choose the most specific term available and, when we don’t, the listener infers that there is a special reason why we opted for a relatively vague word. Simply put, “words compete with each other,” Schlenker says. “And you use the more informative one.”
Schlenker applied the same reasoning to Campbell’s monkeys. “The important thing is that in this situation, both krak-oo and hok are more informative than krak,” he says. “By logic, if you hear krak you can infer there was a reason krak-oo and hok were not uttered, so you infer the negation.” That is, when monkeys in the forest say krak, they are also implying not-hok and not-krak-oo, neither a minor threat nor an aerial threat. In the forest, monkeys understand that this must refer to a leopard—the only nonminor, nonaerial threat nearby.
On the island, however, it remains a general alarm call. That’s because krak does not intrinsically imply negation. In Tiwai, there are no leopards. When the monkeys hear krak, they have no reason to infer not-hok and not-krak-oo, because that would not make sense (there are no serious ground threats). In that situation, the term krak reverts to its simplest meaning, without any inferences, and indicates a general alarm—one that can even warn of eagles circling overhead.
A word of caution about words
Broadly, experts consider the findings significant. “I am aware of this work and think it is very promising,” Robert Seyfarth, a psychology professor at the University of Pennsylvania and pioneer in the fields of animal behavior and learning, wrote in an e-mail. “This is the first time that a professional linguist has tackled the data on call combinations in wild monkeys.”
Arik Kershenbaum, a zoologist at the University of Cambridge, agrees that studying animal calls through the lens of linguistics was a worthwhile pursuit. “Too often we decide, a priori, that because animal communication isn’t language it’s pointless to apply linguistic tools,” he says. “I think it’s a refreshing and much needed formalism for looking at animal signals.”
But Kershenbaum cautions that the krak mechanism suggested in the paper is based on hypotheses rather than experimental data. “Although this current work seems very internally consistent, it does rest on a shaky foundation of correlation,” he says. “Without experimental confirmation the premise is highly speculative indeed.”
He adds that although the logic fits together nicely, it will not be possible to draw definite conclusions about how monkeys use krak without follow-up experiments in the field. “In a line, my thoughts are—absolutely fascinating way to look at things,” he says. “Now let’s see whether it holds up.”