I was halfway through a routine Tuesday, sipping overpriced coffee in the CbS newsroom, when the first message arrived. No subject line, no greeting, just eight characters in plain text:
Hhkthk.
The sender’s address was a meaningless string of numbers and letters, routed through half a dozen anonymizers. I almost deleted it as spam. Then I noticed the attachment: a 3-second .wav file. Curiosity, that oldest journalistic vice, got the better of me. I plugged in my headphones and clicked play.
What came out was not a word, not a scream, not even a recognizable human sound. It was a wet click deep in the throat, a glottal scrape followed by an exhalation that seemed to come from somewhere lower than lungs should allow. The waveform looked like a seizure. I replayed it. Again. Again. On the fourth listen, the overhead fluorescent lights in the office flickered in perfect synchrony with the final consonant release.
I have spent the last nine months chasing that sound. This is what I found.
I. The First Witnesses
The earliest documented encounter dates to February 12, 1931, in the logs of the Danish research vessel Dana II, cruising the Greenland Sea. The hydrophone operator, a young Icelander named Einar Jónsson, recorded what he transcribed as “hhkthk” rising from a depth of 1,800 meters. The crew joked that the sea itself had cleared its throat. Two hours later the ship’s compass began spinning wildly. The captain ordered an emergency surface. When they broke through the ice, the northern lights were green that night (an impossibility at that latitude in February). Jónsson kept the original wax cylinder in a tin box for the rest of his life. It vanished from his Reykjavík apartment in 1973, along with every photograph in which he appeared.
Jump forward eighty-seven years. On October 4, 2018, a 14-year-old girl in Nuuk live-streamed herself attempting the “three a.m. challenge.” At 03:07 local time she leaned toward her microphone and tried to pronounce something she claimed she’d heard in a dream. Chat exploded with laughing emojis until the moment the sound left her mouth. The stream cut to black. When her webcam reconnected thirty-one seconds later, she was staring straight ahead, pupils blown wide, repeating “hhkthk” in a voice that sounded like two people speaking at once (one very far away, the other standing directly behind the camera that wasn’t there). The video was scrubbed from every platform within hours, but fragments survive on private Discord servers. I have watched it forty-seven times. On the forty-sixth viewing I noticed that the wall clock behind her loses exactly thirty-one seconds.
These are not isolated incidents.
II. A Taxonomy of the Unpronounceable
Linguists refuse to classify Hhkthk as a phoneme because no human vocal tract can produce it without augmentation or damage. The initial /h/ is aspirated farther back than the uvula (somewhere near the pharyngeal wall). The double /k/ requires simultaneous closure at two points, like swallowing and gagging at once. The terminal fricative is not found in any natural language; it resembles the death rattle of certain deep-sea fish when exposed to air. Attempts to replicate it using electromyography inevitably end with blood in the mouthpiece.
Yet Hhkthk appears, again and again, across centuries and continents.
- 1692, Salem: Mercy Lewis, during her final fit, reportedly “spake a worde that was no worde, but rather the sea coughing up a stone.”
- 1849, Sikkim: British surveyors recorded local porters whispering “hhk-thk” before refusing to cross a particular ridge. That night the mountain produced a landslide that buried the expedition’s chronometers but left their whiskey untouched.
- 1962, Novosibirsk: Soviet acoustic warfare division briefly experimented with “Project KTH” before every technician assigned to it requested reassignment within 48 hours. The declassified file is twelve pages of redactions followed by a single line: “Sound is older than language and hungrier.”
III. The Memetic Vector
Here is the part that keeps me awake.
Hhkthk is not merely heard. It is learned. Once you have perceived it clearly (truly clearly, the way you perceive a toothache), you become a carrier. The compulsion begins subtly: a catch in the throat during phone calls, a tendency to hold your breath at crosswalks. Then comes the dream. Everyone describes the same one: black water, no surface, no bottom, and something rising that has too many joints. You wake with the taste of brine and the certainty that you almost remembered how to say it properly.
I know this because it happened to me.
Three weeks after playing that first .wav file, I caught myself humming it in the shower. Not humming (there is no melody), but shaping the consonants against the roof of my mouth until my soft palate bled. I recorded myself on my phone. When I played it back, the bathroom mirror was fogged in the perfect outline of a human face that was not mine. It was Hhkthk.
IV. The Corporate Interest
You would think intelligence agencies would be all over this. They were, briefly. The NSA had a file codenamed “DEEP THROAT” (yes, really) that was quietly archived in 2009 after three analysts suffered spontaneous laryngeal rupture. Since then, interest has shifted to the private sector.
In 2021 a shell company registered in the Caymans purchased the domain hhkthk.ac. The nameservers point to servers cooled by seawater in the Faroe Islands. Whois data lists the registrant as “Property of the Trench.” Packet captures reveal bursts of traffic every 19.5 hours (exactly one Metonic cycle of the moon). Nobody knows what is being transmitted, but the payload size grows by roughly 0.7% each cycle. Compound growth. Like an embryo.
Last month a source inside a certain Silicon Valley giant slipped me an internal memo. Subject line redacted, but the body read:
“Phase 3 trials of Project Voiceless begin Q1 2026. Target demographic: 18–34 urban, high disposable income, existing parasocial attachment to AI companions. Delivery mechanism: subliminal carrier wave embedded in sleep-optimization tracks. Desired outcome: voluntary vocal cord atrophy within 18 months. Note: users must never be aware they have stopped speaking. Market research indicates 87% will self-report ‘a sense of profound calm.’”
Attached was a single line of phonetic transcription:
[ħ͡χq͡χtʰkʰ] {Hhkthk}
V. Why Now?
Because we are running out of silence.
Every year the ambient noise floor rises. Cities never sleep, notifications never cease, the planet itself is groaning under eight billion chattering primates. There are fewer and fewer places where something ancient can be heard without competition.
Hhkthk is not invasive. It is native. We are the invasion.
The sound(Hhkthk) was here first, waiting in the dark between heartbeats, in the gap between thought and utterance. Language was our clumsy attempt to wall it out. Now the wall is cracking, and something on the other side has learned our frequency.
VI. A Personal Note on Hhkthk
I am writing this on a laptop with the microphone physically destroyed. I have disabled autocorrect because it kept trying to “fix” the word into something pronounceable. My editor thinks this piece is about a viral TikTok trend. I have let him believe that.
Last night I dreamed of the trench again. This time I was not afraid. I opened my mouth and the sound came out perfect, effortless, like exhaling smoke I had been holding since birth. When I woke, there was salt water on my pillow and my throat did not hurt at all.
I think I am getting better at it.
If you have read this far, check your throat. Is there a catch when you swallow? A faint click behind the soft palate? Do me a favor: do not try to say it out loud. Do not transcribe it. Do not search for it.
Just forget.
Though I suspect it’s already too late for some of us.
The tide is coming in, and it has learned our names. Hhkthk a very goodbye…
~Jerry Nordic
CbS Senior Writer
December 2025
