Whistled Languages are Not Distinct Languages

[Added later: the Bora in the Amazon use drums to communicate over long distances - interestingly, using a "compressed" pitch and rhythm system very similar to the whistled system of the Canaries. It's still not its own language, but it's indisputably an outstanding piece of human cultural innovation, and comparison to how Silbo compresses information would be very useful.]

An article in the New Yorker re-romanticizes the idea of "whistled languages", moving from the more famous Silbo of the Canary Islands to a whistled version of Turkish found in northern Turkey. I wrote about Silbo previously, a whistled version of (now) Spanish. (And that's what it should be called - whistled Spanish. In fact Silbo means "whistle.")

As before, I emphasize here that I'm not trying to diminish the significance of these innovations in human culture, just argue against the misconception that these whistled versions of languages are in fact languages in themselves. I'm thrilled that people have caught on to the cultural value and worked to preserve them, and furthermore whistled language provides fertile testing grounds for hypotheses about the comprehension of language, like the lateralization study in Current Biology conducted by Güntürkün, Güntürkün and Hahn cited in the New Yorker. And there are multiple places in the world where a whistled version of a language has developed, for obvious adaptive advantage (hard-to-cross mountains and/or thick forests.) But whistled languages are not distinct languages.

Why not?

1. Whistled communication is never used as the primary method. People use Silbo to call across canyons to each other (effective and clever) but there was no report of the ear-splitting sound of story-tellers around the campfire (because then they were speaking normally.) I place this claim first because reports of whistled Turkish state that it was used in non-work settings, which on its face would seem to undermine my argument. BUt this loses sight of why it matters whether people use whistled Turkish in more than a few restricted settings - because there is still no such thing as a first-language whistle-language speaker, and certainly there never has been a monolingual. At best, this makes it a pidgin, like Chinook Jargon or Russenorsk.

2. Whistled communication is a whistled version of a primary, "normal" spoken language. If a whistled language is distinct, then when we whisper in English next to each other, that would count as a separate language as well. And of course it doesn't - it's a restricted, rarely-used, context-specific form of communication which makes some phonetic sacrifices due to its usefulness in that setting.

3. Can you learn the Turkish whistled language without learning Turkish? No. Granted, a first-language Turkish speaker might not understand it at first, much like you didn't understand Pig Latinthe first time you heard it. But you can learn Pig Latin - only if you know English first.


No comment on how this bears on my (partly tongue-in-cheek and now mostly abandoned) cultural affinity argument about whistling in other Altaic languages.

In Favor of the Broad Altaic Hypothesis: Whistling and The Comparative Cultural Method

Languages generally break down into related groups that share a common ancestor; Indo-European is the best-known and -studied example, and yes, there was probably an actual proto-Indo-European language spoken about six thousand years ago in Southern Ukraine that ultimately gave rise to Bengali and Greek and English and Italian. Classically such groups are created by using a standard word list from each language and showing phonological correspondence between words, using a rigorous set of sound transformations applied consistent across the languages. Computational methods have more recently been added to the repertoire, which includes ways of looking at the grammar and morphology of the languages. These appeal to me because these elements are more conservative, allowing better resolution over greater times. It's pretty easy for individual words in languages to get replaced. But it's harder for elements of morphosyntax to change. They all have to work together in a much more rule-bound way, so it's like changing a single spark plug or belt in a car engine.

With respect to developing hypotheses about language descent and relatedness, linguistics are typically either lumpers or splitters. An argument for the lumper approach is that there's more value and more information in proposing a genetic taxon based on certain characteristics and investigating the hypothesis, than just putting every single language in its own category. The Altaic hypothesis is that there was a single language that gave rise to Turkish, Mongolian, Tungusic (northern China/southeastern Sibera), Korean, and Japanese languages. Including those last two has always been controversial. More recently even the reality of the Turkic-Mongolic-Tungusic core grouping has come under attack. For technical reasons of morphosyntax I won't go into here, I have always been quite sympathetic to the broader Altaic grouping.

Are there other media for comparative methods? Genetic studies have been an additional tool for understanding prehistory, and of course genes and language don't always covary: these studies have demonstrated that populations interbreed and learn new languages. We can also look at cultural practices, which may be more prone to drift than genes or language, but still provide intriguing signals; for instance, similarities in religion or superstition, which may also be more conservative than other elements. For instance, regarding the broad Altaic hypothesis (including Japan and Korea) you might notice that the cultural practice of steam or communal baths exists in Japan, Korea, and across the Turkic countriess from Kazakhstan to Turkey, but not in intervening countries to nearly the same degree (China, Siberia, India, Iran.)

An even more curious similarity is the superstitions that all these places have about whistling, which is also not shared by intervening cultures. In Japan, there are shinto spirits from the native pre-Buddhist Japanese religion called tengu, which are wind spirits. Also in Japan, whistling inside a house is considered to produce bad fortune, in that it will draw snakes into the house, or cause tengu to kidnap you. In Korea, whistling at night will bring spirits and snakes. (No word on whether fan death is related to this.) In Mongolia, you're not supposed to whistle inside a ger (yurt) because it will bring a windstorm. Tengri is the Mongolian sky (and wind) god. And you'll find the following in the "Respect" section of the Wikitravel entry for Kazakhstan: "Whistling inside a house is unacceptable in almost all Kazakh homes. It is a very common superstition in Kazakhstan that whistling inside will make the owner of the house poor." I wrote about this before but I keep stumbling across whistling superstitions in Altaic countries without even looking for them. While not rigorously part of the linguistic comparative method, this cultural comparison should increase our confidence in the broad Altaic hypothesis, i.e. that Turkic, Mongolian, Korean and Japanese populations are descended from a common ancestral population.

GAME OVER, ANTI-ALTAIC SPLITTISTS!

These splitter linguists have had their say for long enough. The Japanese are clearly just the island branch of Altaic peoples, and the anthropological data is enough. In Mongolia, you're not supposed to whistle inside a ger because it will bring a windstorm. In Korea, whistling at night will bring spirits and snakes too. In Japan, you're not supposed to whistle indoors because either a snake will come - or Tengu will kidnap you. What are Tengu? I'm so glad you asked. Tengu are the indigenous, pre-Buddhist spirits that can create strong winds. By the way, what was the Altaic sky-god's name? Tengri. SPLITTER LINGUISTS, IT'S OVER! IT'S ALL OVER!

Alright. That may be overstating the case, but one has to admit it's otherwise an odd superstition to share.