Whistled language

« previous post | next post »

In "Transcendent Tonality" (11/5/15), we examined this topic a couple of years ago.  That post focused more on the philosophical and ethereal aspects of this type of communication, although it also introduced some of the basics of interhuman whistling and its congruence with melodic musicality.

Additional research takes us further toward understanding the linguistic, neuroscientific, and evolutionary biological dimensions of articulate whistling, as reported in this BBC article:

"The beautiful languages of the people who talk like birds:  Their unusual whistled speech may reveal what humanity’s first words sounded like." (David Robson, 5/25/17)

Julien Meyer, at the University of Grenoble, France, " has now identified more than 70 groups across the world who can use whistles to express themselves with all the flexibility of normal speech."

These mysterious languages demonstrate the brain’s astonishing capacity to decode information from new signals – with insights that are causing some neuroscientists to rethink the fundamental organisation of the brain. The research may even shed light on the emergence of language itself. According to one hypothesis, our first words may have sounded something like the Hmong’s [whistled] courtship songs.

Whistled languages have existed throughout long stretches of history and across vast reaches of space.

In the 5th Century BC, for instance, the Greek historian Herodotus described a group of cave-dwelling Ethiopians. “Their speech is like no other in the world: it is like the squeaking of bats,” he wrote.  We can’t know for sure which communities he was describing, but Meyer says that several whistled languages can still be heard in Ethiopia’s Omo Valley.

The Australian army, meanwhile, recruited Wam speakers from Papua New Guinea to whistle messages across the radio so that they could confound Japanese eavesdroppers.

Ancient Chinese texts record people whistling Taoist verses – a practice that was thought to send them into a kind of meditative reverie. Meyer has found that Southern China is still a hot spot for many diverse whistling communities among its ethnic minorities, including the Hmong and the Akha.

Meyer has found that they typically rely on one of two strategies – both of which use changes in pitch create a kind of stripped-down skeleton of the spoken language. It all depends on whether normal, everyday speech is “tonal”. In some countries, particularly in Asia, the pitch of a single syllable in a word can change its meaning. As a result, the whistles follow the melodies that are inherent in any spoken sentence. But other languages – such as Spanish or Turkish – are not naturally tonal. In these cases, the whistles instead mimic the changes in resonance that come with different vowel sounds, while the consonants can be discerned by how abruptly the whistles jump and slide from note to note.

Either way, the whistles lose many of the cues that normally help us to distinguish different words – and outsiders often find it almost impossible to believe they carry intelligible messages. Yet Meyer has found that fluent whistlers can decode the sentences with more than 90% accuracy – around the same intelligibility as speech. Meyer suspects that this relies on the same neural machinery that allows us to hold a conversation in a crowded room, or to make sense of a whispered message. “Our brains are really good at reconstructing words that have been a bit destroyed by noise or other distortions,” says Meyer. We can see the same in written messages, when the letters are all jumbled up or the vowels removed – yuor biran aumtoacitally flls th gpas.

As someone who has been severely afflicted with tinnitus since mid-January, 1968, I know well the importance of guesswork to make up for lost phonetic signals, especially, for example, in a crowded, noisy restaurant.

I will not attempt to summarize the research of Onur Gunturkun at Ruhr University Bochum, in Germany, whose studies have revealed the smoothing out of hemispherical differentiation of syllables processed in the brain.  More pertinent for the purposes of this post is the learnability of whistled languages by outsiders.

The team’s own experiences show that outsiders can begin to adapt to the ‘bird language’ with regular exposure – provided you know the spoken language first. Gunturkun is fluent in Turkish, and by the end of the trip he had begun to detect the odd whistled word from the locals’ conversations. His experience would seem to support Meyer’s most recent study, which found that people with no prior knowledge of the whistled languages can soon work out which whistles correspond with which vowels; you do not need to have been born in Kuskoy to learn to speak like a bird.

Another aspect of articulate whistling brought out by Robson is that it blurs the boundaries between music and speech.

Growing evidence suggests that language and music both lean on many of the same brain regions: we tend to process a song’s chord progression using the same circuits that make sense of a sentences syntax, for instance. This may explain why music lessons can alleviate some speech or hearing problems. In 2014, a team at Northwestern University in Chicago found that musical training can even improve a child’s literacy.

Whistled communication – with their entrancing melodies – would appear to naturally exemplify this close link.  “It seems to be on the border of music and language,” says Aniruddh Patel at Tufts University in Massachusetts. The Hmong, for instance, may even play out their poems on a mouth harp instrument. In this case, it is impossible to separate melody and lyrics.

This leads to the question of the fundamental relationship between music and language, and whether there might once have been a "musical protolanguage".

One particularly elegant solution to this conundrum dates back to the father of evolutionary theory, Charles Darwin, who proposed that the two traits arose together as a kind of “musical protolanguage”. According to this view, humans first started singing before we could talk – perhaps as a kind of courtship ritual. Like the blackbird’s song, the musical protolanguage would have been a way to show off our virtuosity, forge social bonds, and scare off rivals, without carrying specific meanings. Over time, however, the practice would have pushed us to evolve a finer control of our vocal chords, which then laid the foundations for more meaningful utterances.

“Perhaps whistling was part of the dynamic that pushed humans to adapt their communication to something more elaborate,” says Meyer, who outlined his hypothesis in a recent monograph on whistled speech.

Meyer points out that although other primates cannot learn to speak like humans, some have mastered whistling. Bonnie, an orangutan at the US National Zoo in Washington DC was able to mimic the simple tunes of her keeper Erin Stromberg, and orangutans in the wild have even been known to make a high-pitched squeak by sucking air through a leaf. Such displays suggest that whistling may have required fewer adaptions than voiced speech, making it the ideal stepping stone to language.

If so, whistled signals could have begun as a musical protolanguage, and as they became more complex and imbued with meaning, they could have also helped coordinate hunting and foraging. After all, Meyer’s research certainly suggests that whistling is ideal for communicating over distance and avoiding the attention of predators and prey – advantages that would have helped our ancestors’ survival. Later on, we could have gained control of our vocal chords too, but the whistled languages continued to be a small but crucial element of humanity’s overall repertoire.

One thing that must be emphasized is that, as it exists now, although articulate whistling may bear certain resemblances to birdsong, it is produced by human beings who are capable of speech, i.e., language.  Furthermore, articulate whistling is based on, or has an intimate relationship to, human language.

Articulate whistling does not by any means demonstrate that birds possess language.  Quite the contrary, it is intimately correlated with one or another variety of human spoken language.

No matter what your reaction to all of these provocative proposals may be, I encourage you to listen to the audio clips available in the article (Turkish "islik dili konusuyor" ["we speak this whistled language"]) and Spanish " en todo el mundo hay hombres que hablan silbando" ["a round the World, there are humans who whistle their language"]).  To me, at least, they seem much more complex and varied than typical birdsong, which by and large is repetitious and predictable.


  1. Steven Miller said,

    May 26, 2017 @ 8:41 pm

    Thank you for this excellent post, Professor Mair. I have often been interested in whistled language, but I have felt a bit adrift without a demonstration of both the whistled and spoken version of the same utterance side by side. The audio clips in the BBC article are interesting, but I'm not sure that I feel confident drawing any conclusions about the correspondence of phonemes and their whistled equivalent. I wonder whether any LL user with experience in whistled language might be able to make such a demonstration for the benefit of ignorant people such as myself.

    Personally, I have tried among friends to communicate English through whistling by matching, as well as possible, the cadence of a spoken sentence while varying tone and timber to attempt matching consonants (as described in "how abruptly the whistles jump and slide from note to note"), but I can't say that I have ever been understood or been able to understand someone else's attempt! I'm sure such a skill requires considerable practice, but a comparative demonstration by a skilled speaker/whistler of another language might provide some clues to get started.

  2. Philip Taylor said,

    May 27, 2017 @ 11:30 am

    I don't know whether they ever crossed over to the other side of the pond, but it may well be worth mentioning the Clangers in this context for the benefit of non-British readers (see here for an example).

  3. Chinook Man PhD said,

    May 27, 2017 @ 12:28 pm

    @Steven Miller,

    Coincidentally I too have tried whistling English for many years, with similar results. A seminar in mapping spoken onto whistled language and vice versa would be among the most illuminating, maybe even popular, grad linguistics courses I can imagine. A bright student could surely use what they learned in it to code a universal spoken-to-whistled synthesizer. .

  4. John Roth said,

    May 27, 2017 @ 7:37 pm

    Quite interesting. It seems like whistling requires less work for several parts of the vocal tract than regular speech, but it still requires a very flexible tongue. If whistling came first, and if it could carry most of the meaning of speech, then I'm wondering about a plausible story for how the remainder of the adaptations for the vocal tract occurred.

  5. Susan Penfield said,

    May 29, 2017 @ 11:50 am

    Interesting article…Another fairly recent and fascinating account of whistled language is represented in the work of Mark Sicoli and David Yetman (funded by the NSF DEL program). Some of this appeared as a video on PBS not long ago (sorry I don't have a more specific reference to offer).

RSS feed for comments on this post