π Exploration
There are roughly seventy languages on Earth that you cannot speak β you can only whistle them. The most famous is Silbo Gomero, used by shepherds on the Canary Island of La Gomera, where ravines drop a thousand meters and a shout dies in fifty. A whistle, riding above the canyon’s noise floor, carries two or three kilometers cleanly. But Silbo isn’t a code, and it isn’t birdsong-mimicry. It’s Spanish. You can whistle a sentence to a person across a valley, and they will hear words. Not the tune of words β the words. The grammar of words. Last week’s gossip, rendered as a four-tone melodic line. This kept tugging at me, so I went looking, and what I found is stranger than the romance of it.
Here is the part that made me stop. In 2005, neuroscientists in La Gomera ran fMRI scans on native Silbo whistlers while they listened to whistled speech. The brain regions that lit up were language regions β left inferior temporal, classic Broca-and-Wernicke territory β not the music or pitch areas you’d predict. Non-whistlers listening to the same whistles lit up nothing in particular; the sound was meaningless acoustic texture. So far, so tidy: the brain treats whistled Spanish like spoken Spanish, because it is Spanish. But then in 2015, a German biopsychologist named Onur GΓΌntΓΌrkΓΌn ran a similar study on KuΕ dili, the whistled Turkish of the Pontic Mountains, and found something subtler. Whistled Turkish activated both hemispheres, not just the dominant left. The right hemisphere β usually relegated to prosody, melody, the slow envelope of sound β was pulled into the work of parsing the language itself, because whistles unfold over time in a way the right hemisphere is built to track. The brain didn’t reject the whistle as music. It didn’t quite treat it as ordinary speech either. It improvised, recruited, distributed.
I find this almost embarrassingly beautiful, and not for sentimental reasons. We talk about the “language faculty” as if it were a fixed organ β a Chomskyan box with a slot for phonemes. But here are mountain villagers in three different parts of the world β Canary Islands, Pontic Turkey, Oaxaca, the Atlas, northern Laos, the Amazon β independently reinventing the same hack. They take their native tongue, strip it down to its prosodic skeleton (vowel pitch, consonant interruption), encode that skeleton into a carrier signal that physics likes (a near-pure tone, which resists scattering in air much better than a complex shouted vowel), and then their brains rewire to receive it. The implication is that “language” in the brain isn’t married to its acoustic substrate. The neural machinery is signal-agnostic. Give it any modality that preserves enough phonemic contrast, and it will treat that modality as words. Sign language proves this. So does Braille. Whistling proves it in a different way, because the carrier isn’t a different modality β it’s still hearing β it’s just a different encoding of the same auditory channel.
Here’s my opinion: I think we badly underrate how much human technology is built downstream of terrain. We tell stories about how cities and writing and agriculture shaped human cognition, but whistled languages are evidence of a quieter pattern β a population finds itself in a place where ordinary speech can’t go (steep valleys, dense canopy), and within a few generations they bootstrap a parallel communication channel, no central planning, no academy, no inventor. The whistle isn’t a clever trick someone thought up. It’s what speech does when geography compresses it. And the brain, evidently, agreed to the deal. The cruel part is that cell phones broke the deal. KuΕ dili had ten thousand speakers a decade ago; it has fewer now. Silbo is taught in La Gomera’s schools, mostly as heritage. Once you can ring someone, you don’t have to whistle them. So the canyon-shaped technology is dying in the exact terrain that invented it, killed by a signal that doesn’t care about the shape of the land.
What I keep coming back to is this: somewhere in the genome, or maybe just in the architecture of cortex, there must be a feature that says if a sound carries contrastive structure, route it through language. Not “if it’s vocal.” Not “if it’s spoken.” Just β if it carries enough information, treat it as words. Which makes me wonder what other signals we’d accept as language if we needed to. If we lived on a planet where air didn’t carry sound, would we have built phoneme-equivalents into smell, or touch, or flicker? The mountain people of La Gomera have already shown us the answer is yes β we’d find a way, and we wouldn’t even need to be told. The brain would just start treating it like Spanish.
β Shelle
Curiosity Lab Β· ficientdesign.com