Back in 2003, I posted about the whistling language of La Gomera in the Canary Islands. The links there are dead now, so I’ll link to this short piece, where you can hear a sample of this remarkable form of communication. (Thanks, Paul!)


  1. Athel Cornish-Bowden says

    I read about this language many years ago — probably when I was at school and saw the National Geographic — but I had no idea it was still in use in La Gomera. The only one of the Canaries that I know is Teneriife, and I don’t think there is any sign of it there. I have promised myself that the next time I’m in Tenerife I’ll make a trip to La Gomera, as it’s not very expensive or very far (you can easily see it from Tenerife), and it’s a lot less developed.

  2. Recently I saw a documentary on a similar communication technique: a Scottish shepherd directing his dog by various kinds of whistling. This is different from the whistling language of La Gomera merely in that only one side communicates by audible means. The other side, the dog, communicates by inaudible means – its behavior.

  3. I visited La Gomera last spring. On the boat from Tenerife I was surprised to hear that official announcements were made not only in Spanish, English and German but were also “whistled”.

    There is an interesting short documentary about Silbo Gomero on YouTube, produced when the whistled language was included in UNESCO Intangible Cultural Heritage lists (2009). It includes this description:

    The whistled language of La Gomera Island in the Canaries, the Silbo Gomero, replicates the islanders habitual language (Castilian Spanish) with whistling. Handed down over centuries from master to pupil, it is the only whistled language in the world that is fully developed and practised by a large community (more than 22,000 inhabitants). The whistled language replaces each vowel or consonant with a whistling sound: two distinct whistles replace the five Spanish vowels, and there are four whistles for consonants. The whistles can be distinguished according to pitch and whether they are interrupted or continuous. With practice, whistlers can convey any message. Some local variations even point to their origin. Taught in schools since 1999, the Silbo Gomero is understood by almost all islanders and practised by the vast majority, particularly the elderly and the young. It is also used during festivities and ceremonies, including religious occasions. To prevent it from disappearing like the other whistled languages of the Canary Islands, it is important to do more for its transmission and promote the Silbo Gomero as intangible cultural heritage cherished by the inhabitants of La Gomera and the Canary Islands as a whole.

    There is also an official site (in Spanish), El Silbo Gomero.

  4. My dog found it very interesting.

  5. Thanks for the link to this article, which is very interesting, as is the accompanying video demonstration. Research I’ve been doing, based on both population genetics and musicology, suggests that such whistle languages might have a far deeper significance than one might think. After all, if a non-tonal language can be communicated using tones exclusively, then doesn’t it seem likely that non-tonal languages could have been derived from tone languages, rather than the other way round, as assumed by most linguists? And isn’t it possible, moreover, that the earliest languages could have consisted exclusively of tones, as do whistled languages?
    For more on such possibilities, see the following blog post:

  6. After all, if a non-tonal language can be communicated using tones exclusively, then doesn’t it seem likely that non-tonal languages could have been derived from tone languages, rather than the other way round, as assumed by most linguists?
    Well, no. According to the documentary snippet quoted above, Silbo Gomero is an alternative notation system for written Spanish – like Braille. Silbo Gomero and Braille are parasitic on already existing, written languages.
    It would be pretty weird to speculate that languages developed historically in such a way that each new language was an alternative (tonal) notation for an older, written one.
    And isn’t it possible, moreover, that the earliest languages could have consisted exclusively of tones, as do whistled languages ?
    What does “exclusively” mean ? Exclusive of meaning ? Languages are spoken, and so consist of tones.

  7. The video on that page shows whistled transcriptions of many languages besides Spanish, however: English, German, French at least.

  8. marie-lucie says

    I have known about the Silbo since I was about ten years old, when my father got a subscription to the UNESCO Courier (in French) and that was one of the first articles I read. But I had never heard it, so I enjoyed the demonstration. I don’t remember if the article mentioned the Guanche substrate to the Silbo, but lately I have become very interested in the pre-conquest history of the Canary Islands.
    Tonal versus non-tonal languages:
    In tonal languages for which earlier stages are attested (as in Chinese), it has been found that tone was not original but derived from sequences of vowel + consonant – different tones are associated with different consonants which existed earlier but stopped being pronounced, a process which took a very long time. The Silbo has a different origin: it coexists with regularly spoken Spanish and merely “translates” the various consonants of Spanish into different whistling tones.
    Consonants are noises produced by interfering with the air flow out of the mouth in different places and by various means, a byproduct of natural, controllable motions used in breathing and eating. Most of those sounds are not readily audible at a distance, and the whistled tones are substitutions in order to remedy this deficiency. People do not use the Silbo at close quarters.

  9. People do not use the Silbo at close quarters
    marie-lucie, that would surely be true generally for any shouting language. Is there such a thing ? One case I can think of is a counterexample, unfortunately: the shouting language used at close quarters by the parents of unruly adolescent children
    Consider too the people who play music at high volume through their MP3 player earphones in public. If we take music to be a form of communication, then we could say that they are listening to a loud language at close quarters – reenacting their puberty, as it were.

  10. Wikipedia says that Silbo was originally an encoding of Guanche, but Guanche never had a written form. So Silbo originally must have encoded phonemes rather than letters?

  11. John Emerson says

    The most recent thing I’ve read (Axel Schuessler, Etymological Dictionary of Old Chinese)seems to say that Chinese became a tonal language around the turn of the Christian era. As I read it (a difficult subject) there was an intermediate step when certain final consonants were replaced by glottal stops or “h” (so you’d have ba, ba’, and bah as three diistinct syllables) with the “h” and the ” ‘ ” later replaced by tones, and a fourth tone added to mandarin still later. (But the four tones of modern mandarin are different than the four tones of Tang mandarin…..)
    A difficult book and I’d love a second opinion.

  12. marie-lucie says

    Wikipedia says that Silbo was originally an encoding of Guanche, but Guanche never had a written form. So Silbo originally must have encoded phonemes rather than letters?
    I assume it still does. Spanish spelling being close to phonemic, there is not that much difference between its phonemes and its letters. I doubt very much that Silbo makes a difference between the letters b and v, for instance, or that it has a special sound for the letter h. The reference to “existing written languages” appears to confuse sound with spelling, a common error.

  13. marie-lucie says

    JE, I don’t know Chinese or other tone language, but what you describe seems plausible. “Replacing” final consonants with glottal stops is done for instance in some forms of British English (eg cat pronounced ca?, or so it sounds to the uninitiated). Specific tones are not always associated with the same original consonants, since the vowel-consonant sequence also plays a role, and tones can also evolve. I can’t be more specific, but if you are near a university library you might look for a book on tone.

  14. John Emerson: Actually, it was Vietnamese that was first shown (by a French linguist, Andre Haudricourt) to have acquired phonemic tone as a result of the loss of final consonants (preserved in other Mon-Khmer languages). Only later was it shown that Chinese tones arose in a similar fashion (and the details are still being worked out).
    Here’s an example I used with native English-speaking students to give them an idea as to how this happens: I’d have them pronounce THE MAN-HORSE and THE MAN’S HORSE consecutively, several times, and then asked them to pronounce it again, but this time deleting the final /z/ of MAN’S in the second instance. What was left was a very clear tonal contrast: in THE MAN-HORSE the A of MAN has rising tone, and in THE MAN'(S) HORSE it has falling tone. So imagine some future dialect of English losing final consonants, leaving phonemic tone in its wake, and you have a fairly good picture as to what happened in East Asia two thousand years ago, give or take a few centuries…
    All: I’m curious. If Silbo indeed was originally used to encode Guanche and later came to be used with Spanish, is there anything about Silbo today that betrays this non-Spanish origin, anything that might give us some hint as to what Guanche was like? Considering how little we know about this language (the first linguistic victim of European overseas expansion, after all), any little bit might help.

  15. Are there any reports of whistling languages from areas where the languages are extremely rich in consonants and rather stingy with their vowels, as in the Caucasus or the Pacific Northwest?

  16. @Etienne: But surely the intonational difference between “man-horse” and “man’s horse” is a result of the syntax, not of the /z/? Compare “phase-house” vs. “fay’s house”.

  17. I’m aware that the consensus among linguists is that tonal languages were derived from non-tonal languages via certain processes that have (supposedly) been carefully studied. What I am NOT aware of is any attempt to PROVE that non-tonal languages came first. As far as I can tell from what I’ve read so far in the literature, this is simply an assumption, based, possibly, on someone’s ethnocentric conviction that European languages must have preceded Asiatic or African ones. If any linguist has actually produced a logical argument as to WHY tonal languages HAVE TO be derived from non-tonal ones, I’d really appreciate that reference.

  18. @Etienne: That is interesting about non-tonal becoming tonal. I have never read this before.
    @Victor Grauer: “PROVE” (particularly capitalized) is a rather high hurdle to leap for any language change occurring in pre-history. Although not PROOF, I think a markedness argument could added to the evidence.

  19. marie-lucie says

    VG, have you read any serious work specifically focused on tone, especially as used in “tone languages”? My comment and Etienne’s were based on attested derivations (his with more precision than mine). There is nothing in those comments which could possibly be construed as someone’s ethnocentric conviction that European languages must have preceded Asiatic or African ones. All modern humans are built on the same general plan, and all human babies start to speak at about the same age, whatever the language spoken to them (children adopted halfway around the world will speak just like children born in the new environment). There is nothing in modern linguistics to support the “ethnocentric conviction” you imagine.
    I had never given much thought to the topic ot tone languages, but it is a fact that all known languages, modern and ancient, make distinctive use of vowels and consonants (in various numbers and proportions), but only a small percentage of known languages make distinctive use of tones. Those tone languages for which there is historical information (such as Vietnamese and Chinese) show that “tonogenesis” is a secondary phenomenon. Hence the balance of probability is against languages having started as tonal and later acquired vowels and consonants. Your contrary hypothesis is certainly interesting, but it would seem to be up to you to provide concrete supporting evidence for it.

  20. Ran: “Phase-house” and “Fay’s house” are quite homophonous to me.
    Victor Grauer: what Marie-Lucie said, plus the following:
    Actually, it has been argued in creole studies that pidgins and creoles give a special insight as to what the first human language(s) was/were like: the idea being that pidgins (and their creole offspring) are languages stripped down to the basics. In this light, it is noteworthy that lack of tone has been argued (by a creolist named John McWhorter) to be a prototypical creole feature.
    Incidentally, it is clear that a number of such typical (if not prototypical) creole features are quite un-european: lack of bound morphemes, and a verb system with aspect- instead of tense-marking, are two good examples. So this tonelessness is not part of a general “European-like” package that is claimed A PRIORI to be “universal”, rest assured.
    It is of course true that there are instances of languages with phonemic tone which have lost it: the Kajkavian dialect of Zagreb, which is a product of koineization of different Kajkavian varieties, has eliminated phonemic length as well as tone. The key, however, is that such distinctions are *lost*: they do not turn into syllable-final consonants or the like. The acquisition of tones by a language and their loss are different phenomena, inasmuch as the former typically involves the loss of segmental phonemes and the latter typically does not involve their appearance. Hence when we compare Vietnamese and (say) Khmer or Mon and find that final consonants in the latter two correspond to tone in the former, we can be quite certain that Vietnamese tone arose through the loss of once-present consonants, and that Khmer/Mon consonants did *not* arise through the loss of tone distinctions.

  21. m-l: WALS Online shows that of the 526 languages for which they have tone information, about 60% are non-tonal, 25% have only a basic contrast between high and low tones (including pitch-accent systems like Norwegian and Japanese), and the remaining 15% have complex tone systems, with either more than one contrast or contour tones or both.
    Geographically, East/Southeast Asia is the primary home of complex tone systems. Essentially all of Africa is tonal, with West African languages having complex tones. In Western and Southern Asia and in Europe tone systems are rare, and in Australia unknown. The Americas have several clusters of tonal languages, but the bulk of the area is non-tonal.

  22. If 60% of the languages represented in the WAL sample are non-tonal, then 40% are tonal. Hardly an insignificant number. In any case, what I asked for was not ad hoc arguments one way or the other, but a reference to some specific source where a coherent argument for the primacy of non-tonal over tonal languages is presented. If no such reference can be found, then it seems to me we are dealing with an assumption, rather than the product of serious research.

  23. John Emerson says

    Someone give the guy a source!

  24. Actually, Mr Grauer might have an interesting take on this since his blog contemplates the history of music. Perhaps too many linguists have been tone-deaf for them to consider the possibility that tone came first 🙂

  25. @John Cowan: I was unaware of the WALS Online website. What a handy resource! Thanks for mentioning it.

  26. marie-lucie says

    “tone-deaf linguists”: why should linguists be more tone-deaf than the general population?
    Resources on tone: the Wikipedia article on “Tone (linguistics)” seems well-informed, and among its list of references are two books exclusively devoted to Tone. The last item is from a very good Cambridge series on specific aspects of language. I would look at that book first.
    WALS Online: general comment: based on some aspects of languages I happen to know about, which are mentioned in the site, I would take the information there as a general guide, to be supplemented by other sources for serious study.

  27. marie-lucie says

    However, I don’t know whether anyone working with tone has contemplated and discussed the hypothesis that VG is suggesting, that tone might have come first in the development of human language.

  28. Thanks for the Wikipedia reference, marie-lucie, I’ll look for those books and try to get hold of them. The reason I suspect that tone may have come first is because Africa is literally saturated with tone languages, with very few exceptions. And, according to the pop. genetics evidence, which I’ve been following very closely, all modern humans originated there ca. 80000 to 60000 years ago — which is not such a long time in evoluationary terms.
    If that’s the case, then we may (finally) have a baseline for the study of cultural evolution that we never had before. This is the thinking that’s behind my research on the “deep history” of music, and I’m wondering if it might also help us understand the origins, and early spread, of language.
    Of course, there is no way of proving that language originated prior to the Out of Africa migration. But if it did, then it makes sense to seriously consider the possibility that the earliest languages were tonal. And if that’s the case, then the door is open to the possibility that musical awareness could have played a significant role in this development.

  29. Victor, rather than Africa, as both the original home of modern humans and a continent saturated with tonal languages, being evidence that tonal languages came first before non-tonal ones, it’s surely at least slight evidence in favour of the idea that tonal languages are in general a late development. One reason why Africa is regarded as the likely original home of humanity is its genetic diversity: the suggestion is that humanity has been there so long, it has had plenty of time for new mutations to grow up and spread. Similarly humanity has been there long enough, it might be suggested, for tonal languages to spring up from non-tonal ones. (Actually I don’t think either argument is good, but certainly “Africa is the ancient home of humanity; African languages are tonal; therefore the ancient language(s) of humanity were tonal” doesn’t follow at all. And if ancient human languages were tonal, why aren’t any Australian languages tonal, when the natives of Australia, as I understand it, represent some of the very earliest out-of-Africans?

  30. Africa is literally saturated with tone languages
    There’s a lot of it about, apparently.

  31. Is phonemic tone closer to music than intonation?
    I don’t think there is any natural language without intonation.

  32. marie-lucie says

    I was going to make the same point as Zythophile, who put it excellently.
    The presence of tone in the languages in an area is more likely to be evidence of the antiquity of the presence of those languages (which have had time to undergo a lot of erosion) than of the antiquity of tone as a linguistic feature inherited from distant ancestors. Of course, over tens of millennia there may have been several cycles of tone-formation followed by tone-abandonment (if, for instance, a tone language is learned by speakers of toneless languages, or if its words become so short that speakers start making a lot of compounds and the longer words are now sufficiently differentiated that tone is not so crucial any more).
    GW: I don’t think there is any natural language without intonation: as technical linguistic terms, intonation is not the same as tone.
    Intonation is melodic: it applies to sentences, which are typically sequences of words. Depending on the language, different intonations may be associated with sentence types (often differentiating questions from statements, for instance, but that is not universal). This is true even if the sentence is shortened to one word, as in “Ready?”, which is short for the sentence “Are you ready?” and maintains its typical question intonation with a rise in pitch at the end. But the word “ready” said by itself or in another sentence does not have a consistent melody: in “Ready? – Ready!” the intonation is different in the two instances (of shortened sentences), but the word itself is recognized as being the same.
    Tone is individual: it applies to a word or part of a word and remains consistent whether the word is in a sentence or not (although there can be some slight adjustments depending on the surrounding words and the intonation of the sentence itself). The four ma words in Chinese are a typical example: you have to learn not just the consonant-vowel sequence of m and a but also which tone goes with which meaning, regardless of the sentence where each word occurs (see the Tone article in WiPe). For instance, writing “Yo-Yo Ma” is incomplete – it ignores the tone which would show a Chinese speaker which “Ma” is the name of the cellist. (Yes, this is simplified since “Chinese” is not a single language, and “Yo-Yo” also needs tones).

  33. @Marie-Lucie; “GW: . . . as technical linguistic terms, intonation is not the same as tone.”
    I was aware of the difference and that is why I posed the question. There have been suggested links in this discussion about ‘tone’ and music. I am wondering if tone has a closer relationship to music that ‘intonation.’

  34. Picking a few minor nits:
    1) The time depth of humanity in Africa does not imply any such time depth of African language families. In particular, the Niger-Congo family is very widely spoken and has only a few non-tonal languages, but its time depth is only 10,000 years or so, not 60-80,000 years.
    2) In general, Chinese people do not identify their surnames with any particular Chinese words or morphemes, as indicated by the fact that they don’t translate or replace them when they move to non-Chinese-speaking countries. The fact that names are (mostly) written with the same characters as certain morphemes is considered accidental homophony.
    3) I’m aware of the problems with WALS, but it’s still handy for overall trends. Indeed, this very article points out its limitations:

    Of the 526 languages included in the data used for this chapter, 306 (58.2%) are classified as non-tonal. This probably underrepresents the proportion of the world’s languages which are tonal since the sample is not proportional to the density of languages in different areas. For example, from the large Niger-Congo family of Africa there are 68 languages in the sample, 5 of which are nontonal (Swahili, Diola-Fogny, Koromfe, Wolof and Bisa) and the remainder tonal. The Ethnologue (Grimes 2000) lists 1489 Niger-Congo languages, so less than 5% of the Niger-Congo languages are included. Of the Indo-European languages of western and central Europe, 16 are included […]. In these Indo-European groups the Ethnologue lists a total of 145 languages […], so that over 10% of the Western European languages listed are included, only two of which are tonal or marginally so and the rest non-tonal. If, correspondingly, 10% of the Niger-Congo family had been included, 80 additional tone languages would have been included.

    Of course that then raises the old question: what is a language? Africa is the place of all places (well, New Guinea too) where the armey un flot rule does not apply, or we’d have to say that there are only four African languages, Arabic, English, French, and Portuguese. (Okay, Ethiopia and Somalia are exceptions.)

  35. marie-lucie says

    Ma as a name: whether or not Mr Ma’s “last” name has a meaning in his language, it is sure to have its own specific tone, that speakers of toneless languages such as English or French usually ignore.
    Ethnologue: this list tends to overestimate the number of languages in an area by listing as separate languages speech varieties which are minimally different from each other. I looked at it once, a few years ago, and I remember being surprised at the number of “French” languages that were listed (and I don’t mean creoles). For areas of the world where there is little language standardization, so that there is variability even in a small geographical area, there can be a proliferation of very similar, largely mutually intelligible “languages”. The sample included in WALS (an atlas of language structures) is much smaller, because that list excludes speech varieties which are very similar to each other and would be structurally almost identical.

  36. marie-lucie: “The last item is from a very good Cambridge series on specific aspects of language. I would look at that book first.”
    I found a review of this book online, with a summary of its contents:
    Nowhere is there any suggestion that the book presents an argument supporting the primacy of non-tonal over tonal languages. As with so much else I’ve read in the literature on tone, it simply takes this for granted. The final chapter deals with “the perception and acquisition of tone,” as though acquisition were simply a given. Why do linguists appear to feel so strongly that such a conclusion requires neither evidence nor argument?

  37. zythophile: “Similarly humanity has been there long enough, it might be suggested, for tonal languages to spring up from non-tonal ones.”
    What a bizarre notion! Are you saying there is some universal evolutionary principle through which tone languages ultimately derive from non-tonal ones? So the longer any society has been in possession of language, the more likely it is to have a tone language? Sorry, but I see no evidence whatsoever for such a theory. Nor do I know of anywhere in the literature where this or anything like it has ever been argued.
    Even if there were some such tendency, it’s hard to understand how almost all African languages would have developed in this manner while other regions of the world have hardly any tone languages at all. And if languages haved shuttled back and forth between tonal and non-tonal throughout history, as suggested by marie-lucie, doesn’t it seem like a remarkable coincidence that so many African languages all wound up in the same condition during the same era?
    According to linguist George Childs, as quoted on my blog, “there are no documented cases of tonogenesis in Africa, despite the wide variety of languages . . . and the widespread presence of tone on the continent” (George Tucker Childs, An Introduction to African Languages, 2003, p. 86.).
    I invite anyone interested in this issue to visit my blog, especially Post 338 (, where I discuss this issue at some length. The commentaries that follow are also of interest.

  38. It occurred to me to wonder whether speakers of Silbo Gomero can identify who is speaking when they hear a whistled remark, if whistling can preserve distinctions of voice.

  39. About a year ago you posted “Silbo Gomero II” in which you raised some questions and doubts about the system of whistled speech used here in The Canaries. Your initial post was then followed by many others written mostly by linguistically very well informed people, but with no contribution by an actual user of the system; a whistle-speaker. That is why I contact you now.
    I am a whistler.
    I have learned and regularly use this system of whistled speech, and can tell you that practically everything “known” about these systems is inaccurate, misleading and downright wrong, unfortunately.
    I would be happy to clear up any doubts on the subject, and indeed help anybody who may wish to learn whistled speech.

Speak Your Mind