One spring morning Panagiotıs Kefalas was in the tavern he owned in the tiny Greek village of Antia when he received a call from his friend Kyria Koula. Kefalas was planning to have breakfast at her home some 200 meters from his tavern. The call did not begin with the sounding of a mobile ring tone. Instead it reached directly from Koula’s mouth to Kefalas’s ears, arriving in the form of a series of high-pitched whistles.
“Welcome, what do you want?” Koula trilled.
Kefalas pursed his lips and whistled back:
“Please, I would like to eat.”
“All right,” Koula replied.
“I would like scrambled eggs,” Kefalas volunteered.
[To listen to the conversation, click here]
A visitor to Antia would have come away perplexed. The beginning of the first phrase, “welcome” (kalós irthate in romanized Greek), sounded like the lewd catcall—“tweet, tweeo”—except that the drawn-out second syllable rose sharply in pitch.
Some accounts contend that the now dying tradition of whistled speech, still maintained by Antia’s few dozen residents, served for centuries as the best way for sheep or goat herders there to communicate from one hillside to another. Whistles, after all, carry much farther than shouts and spare the vocal cords. Even today the pensioners of this village at the southern end of Greece’s second-largest island, Euboea, sometimes use this efficient pretechnological form of wireless communication from house to house to convey news, gossip or extend a breakfast invitation.
I recorded the conversation between Kefalas and Koula in May 2004. Since the early 2000s I have been studying whistled speech in remote mountains and dense jungle across the globe. In that time, I and my colleagues from diverse institutions have come across many previously undiscovered whistled languages. We have also measured the amazing distances that whistled words can travel and have gained an understanding of how blowing air through lips can convey full sentences, as well as how the brains of recipients manage to decode the words.
A Slow Beginning
I originally became interested in these languages almost 20 years ago after reading a 1957 Scientific American article about a version called Silbo Gomero, which is still spoken on La Gomera, one of Spain’s Canary Islands. I decided I wanted to know more and made it the focus of my doctoral work beginning in 2003.
Back when the article appeared, very few researchers had any interest in studying whistled languages, even though such speech had been known since ancient times; Herodotus mentioned Ethiopian troglodytes who “spoke like bats” in Melpomene, the fourth book of his work The Histories. By 2003 interest had picked up, but few linguists had done research on the sounds and meanings conveyed by whistled speech, and most studies had investigated only Silbo Gomero.
The term “whistled language” is somewhat of a misnomer. Whistled speech, in fact, is not a separate language or dialect from a native tongue but rather an extension of it. Instead of using the voice to speak the Greek words Boró na ého omeléta? (“Can I have scrambled eggs?”), those same words are articulated as whistles. The sounds of the words just undergo a profound shift; they are generated not by the vibrations of the vocal cords but by a compressed stream of air from the mouth that swirls in turbulent vortices at the edge of the lips. Just as in ordinary speech, the whistler’s tongue and jaw move to form different words, but the range of movement is more constrained. All that changes is the pitch of the whistle; in contrast, when people speak, the timbre (what distinguishes one sound from another apart from pitch and loudness) may change, too.
In the end, the whistled words conveyed in the village of Antia are still Greek. Linguists sometimes liken a whistle to a whisper, in that both are alternative ways of speaking the same language without using the vibration of the vocal cords. Linguist André Classe, author of the Scientific American article that inspired me, termed whistled talk a natural “informational skeleton” in describing its bare-bones nature. He noted that the intelligibility of whistled speech does not always match that of spoken language, but it comes close.
In my early investigations, I found intriguing documents from travelers, colonial functionaries, missionaries and anthropologists that described 12 or so whistled languages. These clues led me to suspect that other whistled counterparts of spoken languages existed around the world.
In the early 2000s I therefore set about with my colleague Laure Dentel to undertake 14 months of fieldwork visiting places where some evidence indicated that this practice still occurred. Subsequently, I joined with a network of colleagues to conduct new field studies all over the world. I have, as part of this effort, documented the whistled speech of the Wayãpi in the Amazon jungle, in collaboration with linguist Elissandra Barros da Silva in Brazil and anthropologist Damien Davy in French Guiana. With Dentel, I have studied the Akha and the Hmong in Southeast Asia and with linguist Rachid Ridouane, the Tamazight Berbers in Morocco’s Atlas Mountains. In 2009, moreover, Dentel, linguist Denny Moore and I began a five-year collaboration at the linguistics division of the Emilio Goeldi Museum of Pará in Belém, Brazil. Our job was to chronicle the whistled language of the Gavião people in the Amazonian state of Rondônia.
Our research endeavors have brought to bear the latest tools in linguistics and acoustics and used methods from many fields, among them phonetics, psycholinguistics, bioacoustics and sociolinguistics. We borrowed, for instance, the recording methods bioacousticians use for studying animal communication in the wild because these are well suited for studying whistled communication over large distances.
Our research discovered ways people convey words with whistles. The whistler may pucker the lips, finger whistle, or blow into a leaf or a simple wood flute. Some speakers combine different techniques depending on how far they wish to send a message. Words are constructed from these sounds depending on whether the spoken language from which the whistled one is derived uses changes in tone to convey differences in meaning, such as in Mandarin and Cantonese, or whether tones do little more than let the speaker add stress to a word, as they do in Greek or Spanish. In a tonal language, a whistle’s rising pitch mirrors the ascending inflection of the spoken tone. In nontonal languages, however, a whistle’s unchanging pitch represents a vowel—an “i” might be communicated with a high-pitched whistle, whereas an “e” might sound at a lower pitch. The whistler forms consonants in either language class by modulating how abruptly the sound is altered when changing from one pitch to another.
Our inquiry so far has managed to locate about 70 populations who use whistled speech, most hailing from isolated mountainous or densely vegetated locations. That number is just a fraction of the world’s 7,000 languages, but it far exceeds the previously recorded tally. In all these places, whistled languages are used mainly, as earlier work suggested, to project messages beyond shouting distances—but they have other uses as well. They can assist in courtship rituals within the confines of a town. They can be used to communicate in a noisy setting or to trade secrets in the presence of nonwhistlers. (“You have to hide because the police are on the way.”) And they can help hunters land prey; in the Amazon jungle, animals recognize the human voice but not whistles.
Acoustical analysis of whistling used for long-distance communication shows that, under favorable weather and topographical conditions, a whistle can travel several kilometers. The frequency spans 0.9 to four kilohertz, almost exactly the range determined by telecommunications engineers to be best for picking out accurately the component sounds that make up words. In one experiment we performed in a valley near the French Alps, spoken speech carried 40 meters, shouts 200 meters, whereas a whistle was still intelligible at 700 meters. Though not a whistling record, that measurement demonstrated the relative advantage of whistling under average conditions that included some background noise and a light wind.
For linguists, the study of whistled speech has helped demonstrate the capacity of the human brain to recognize words and sentences in an acoustic signal that carries less information than that produced by the human voice. A given whistle’s single frequency lacks the harmonics of the voice. Yet even this lone modulated frequency fulfills the essential requirements of an actual language in clearly communicating information. Whistled speech is therefore an important means to explore the cognitive capacities of our brain to communicate in an untraditional way.
Decades ago bioacoustician René-Guy Busnel, with whom I have collaborated since the beginning of my doctoral work, conducted a study on whistled speech perception among villagers of Kuşköy in the mountains of northeastern Turkey. Using the whistled form of Turkish known as “the language of the birds,” townspeople over short distances could recognize individual words around 70 percent of the time, compared with a 95 percent rate for ordinary spoken words. They could even detect an entire sentence about eight out of 10 times in that situation when people were far enough apart that they could not see one another’s faces clearly. This study inspired me to begin another, published in 2013, in which I, along with my colleagues, investigated intelligibility of spoken words as distances increased between a speaker and listener. The results showed that at a separation of 17 meters, word recognition drops to 70 percent. We also found that the best-recognized consonants (sibilants that resemble whistlelike sounds) are still recognized at rates above 90 percent up to 33 meters away. Combined with Busnel’s work on whistled Turkish, these results suggest whistled speech is more efficient than ordinary spoken speech when interlocutors are communicating across medium distances of 20 to 30 meters.
Also in the realm of linguistics, I was curious about how readily a person can learn some of the rudiments of whistled speech. Traditionally, the skill is taught shortly after a child learns to talk, but we decided to investigate the initial steps of whistled-speech learning in adults. I asked 40 university French- and Spanish-speaking students to listen to Silbo Gomero. We found that the students readily distinguished an obvious component of any Spanish whistled word—the vowels “a,” “e,” “i” or “o” (“u” is whistled as “o” in Silbo Gomero)—and that the Spanish students were a little more accurate than the French ones. Both groups of students categorized correctly the vowels far above chance, though not as well as a trained Silbo speaker.
Left and Right Brain
The neurobiology of whistling is one area that remains largely unexplored. Researchers have only begun to observe what happens in the brain’s language centers when a person speaks via whistles. But we have made some progress. One 2005 study published in Nature by Manuel Carreiras, then at the University of La Laguna on Tenerife in the Canary Islands, and his colleagues reported brain areas underlying language comprehension—the temporal regions of the left hemisphere—are activated in well-trained whistlers when they listen to Silbo Gomero. The finding implied that these same known language-related areas could process words from a simple auditory input consisting of changes in pitch (akin to a musical melody) in experienced whistlers, though not in people unfamiliar with whistled speech.
Another investigator wanted to know whether the concentration of brain activity in the left hemisphere was the whole story. Onur Güntürkün of Ruhr University Bochum in Germany recruited speakers of the Turkish whistled language to test the conventional notion that the brain’s left hemisphere is where most language processing occurs. Earlier studies had shown that the left hemisphere is, in fact, the dominant language center for both tonal and atonal tongues as well as for nonvocalized click and sign languages. Güntürkün was interested in learning how much the right hemisphere—associated with the processing of melody and pitch—would also be recruited for a whistled language. He and his colleagues reported in 2015 in Current Biology that townspeople from Kuşköy, who were given simple hearing tests, used both hemispheres almost equally when listening to whistled syllables but mostly the left one when they heard vocalized spoken syllables. This result needs further confirmation in other whistled languages but provides a challenge to the prevailing idea that the left hemisphere is dominant in language comprehension.
These studies demonstrate that whistled languages can help expand knowledge of the way the brain processes information. I currently promote these research efforts as a member of two organizations. The World Whistles Research Association has been in place since 2002, and a new endeavor on whistled speech was launched in 2015 by my laboratory (GIPSA-lab) at the French National Center for Scientific Research.
Scientists studying whistled languages may also receive a boost from nascent efforts to preserve these unique forms of communication as part of the cultural heritage of various peoples. The Canary Islands were ahead of the pack in that regard. In 1999 they made teaching of Silbo Gomero mandatory in primary schools on the island of La Gomera. They also set up a formal government program to develop whistling teachers. The desire to revive Silbo has since inspired a series of initiatives—for instance, the Cultural and Research Association of Silbo Canario Hautacuperche, an organization that provides whistled-speech courses, even contributed by launching an app called Yo Silbo to train people by listening to correctly whistled sentences.
If similar efforts take hold, whistling for your supper could become more than a saying. It would preserve a form of expression that is giving new insight into how simple high-pitched tones can be molded to communicate complex thought.