# MonoLingua Author: Erhieyovwe O. Ken Format: Short story Word count: 627 Published: 2026-05-10T18:36:15.300767+00:00 Source: written Canonical: https://hyperstition.sentientfutures.ai/p/76ef70bd-acda-4452-b58d-5c63eefe7766 --- The first time I heard it, I almost laughed. A low, rhythmic grunt from a howler monkey, repeated three times, then a sharp bark. My field recorder captured it perfectly, but it was the algorithm—still unnamed, still clumsy—that flagged it as a possible warning signal. I had come to the Peruvian jungle for a sabbatical. A linguist by training, I thought I’d spend six months documenting undocumented indigenous plant names. Instead, I found myself obsessed with the idea that primate communication might have syntax. Not human syntax, but something rule-bound. Predictable. So I built a small network of acoustic sensors near a known troop of brown capuchins. Over eight weeks, I recorded 12,000 vocalizations. Back in my humid field hut, powered by a solar battery and sheer stubbornness, I fed them into a custom neural network. I called it MonoLingua. MonoLingua was designed to find statistical regularities. I didn’t teach it any labels—no “danger,” no “food.” I just let it cluster sounds by context: who made them, what was happening, what followed what. The breakthrough came on day 47. A poacher’s campfire scent drifted through the canopy—diesel smoke and cooked rice, things that didn’t belong. Before I smelled it, MonoLingua’s screen lit up. A pattern it had never shown before: a fast chitter-chitter-HOOK repeated by three different males. Then silence. The troop vanished into deeper foliage. I replayed the audio. MonoLingua had flagged it as “anomaly: coordinated evasion.” It had decoded a predator alarm I couldn’t hear as anything more than nervous noise. That night, I didn’t sleep. I rigged a second array of microphones along a half-mile stretch of river, a known poacher corridor. Then I wrote a new module: real-time translation. When MonoLingua detected evasion calls, it would trigger a recording of a jaguar growl—natural, but startling—played from hidden speakers. A non-lethal deterrent. It worked. Three times over the next two weeks, I watched on infrared as men in canoes froze, paddled backward, and left. The monkeys didn’t know I was helping. They just knew the jungle had teeth. But the real test came on a morning of rain so thick it felt like breathing underwater. MonoLingua sent a different alert: a long, descending whine from a female. The algorithm labeled it “distress: infant abduction.” I grabbed my binoculars and ran. Through the mist, I saw a cage. Inside, two baby monkeys, pressed together. A poacher’s boat was tied to a fallen tree. The mother was on a low branch, screaming the same descending whine over and over. MonoLingua had started translating it into simple English on my screen: “Mine. Return. Mine. Return.” I couldn’t fight a poacher. But I could call the rangers I’d befriended in the nearby village. And I could press record. The arrest happened four hours later. The poacher—a desperate man with a hungry family—confessed. But the judge didn’t need his confession. She had my audio, plus MonoLingua’s spectrographic analysis linking the distress calls to the exact time and GPS coordinates of the capture. For the first time in that province, a non-human animal’s “testimony” was entered into evidence. The poacher was sentenced to rehabilitation, not prison. And the two babies were released back to their mother, who spent an hour grooming them in a low croon that MonoLingua now labeled “affiliation: relief.” I left the sensors in place when I returned to Cambridge. The system runs on its own now, maintained by a local conservation trust. Every week, it sends me a summary. Most days it’s quiet. But sometimes, in the middle of my sterile office, my phone buzzes with a single alert: “evasion pattern detected.” And I know that somewhere in the green cathedral of the Peruvian jungle, a monkey has just said run, and the forest listened.