Listening to Spanish Again
First article in an RCI LinkedIn series on the impact of language in technology. Predecessor to the formal Babel and Inflexión installments. First-person voice, deliberately — the formal papers are written in third-person essayistic register; this series is written in the I of someone who experienced something.
A customer was talking, in Spanish, about using Microsoft Mesh to build their data pipelines. I heard Mesh the way you hear a foreign technical term you've slowly become familiar with — and at the same time, half-translated in the back of my head, I heard malla (what Spanish would call a mesh) and velo (a veil — which a mesh kind of is, a covering you can see through). What they were covering, with all this careful technical vocabulary, was a quilombo.
Quilombo is a Rioplatense word. Its roots are in the Bantu languages of enslaved West Africans brought to colonial Brazil — a quilombo was a settlement of escaped slaves, the most famous being the Quilombo dos Palmares. The word migrated through the centuries into Argentine and Uruguayan slang, where it now means a chaotic mess, a complete disaster. The customer, of course, didn't say quilombo. They said Microsoft Mesh. But what was being covered, by the technical mesh-as-veil, was unambiguously a quilombo. The Spanish word would have named the underlying state of the data architecture more precisely than any phrase in the entire technical conversation. And nobody said it.
The position from which I'm hearing this
I should not have noticed any of this. I grew up speaking Venezuelan Spanish — first language, household language, the language I dreamed in until I was twenty-something. Then twenty years of mostly-English professional life happened, and Spanish quietly receded. Not catastrophically, the way a forgotten skill recedes; more like a room you stopped going into. The furniture was still there.
A few years ago I came back to live not in my native Venezuela but in Argentina. The Spanish I was returning to was not the Spanish I had grown up with. It was Rioplatense — Argentine Spanish, with vos instead of tú, with the Italian-inflected intonation of Buenos Aires, with lunfardo lexicon I'd never heard, with the cultural texture of a country I knew abstractly but didn't speak from. I was recovering and adopting, at the same time. Relearning my native language as a different language.
This double move turns out to be the position from which conscious noticing becomes possible. A Venezuelan who never left wouldn't notice quilombo's historical weight; they'd just use the word. An English-speaker learning Spanish from scratch would notice it but as foreign vocabulary. I'm in the middle, where the morphology is half-recognised and half-new. The textbook is invisible; the patterns are not.
What I keep noticing
What you start to notice, in this position, is that words do not stop carrying meaning at the edge of their dictionary definition. Quilombo doesn't mean mess the way mess means mess in English. It carries the colonial Brazilian settlement, the forced migration of its etymology, the Rioplatense slang transformation, the moment the customer didn't say it. All of that is in the word when a native speaker uses it, even if the speaker can't articulate what's there.
A few more examples from what I've been re-learning:
Garpar is the vesre — the syllable-reversal — of pagar, to pay. A Buenos Aires word-game culture flips syllables to make new words: café becomes feca, tango becomes gotán, pagar becomes garpar. Garpar la nube — to pay the cloud bill — has a register English's spend and pay don't have. The word carries the practice of vesre itself, the act of being from somewhere where you do this with words.
Posta started as a stagecoach relay station — the place information arrived in the colonial-era postal system. Through the centuries, the word sublimated from infrastructure to abstraction. ¿posta? doesn't ask for a place; it asks for the truth. Es posta — it's a fact. The word's modern meaning is what was once carried by postas: information, certainty, news.
And then — this is the balance the observation needs — there are English words that do something Spanish equivalents can't. Pipeline is graphic. The word shows you industrial infrastructure, fluid moving through connected segments. The Spanish translations (tubería, canalización, conducto) all sand off the visual punch. Underscore names a typographic mark by its position relative to the line; the Spanish piso — literally floor — flattens it into a positional metaphor that loses the descriptive specificity. The information loss runs in both directions. Neither language is richer in the abstract; both have zones of accreted or graphical meaning the other cannot translate cleanly.
The thought that came next
It was around this point that I thought, perhaps this could be leveraged in systems.
Programming languages are systems for talking about computation. The vocabulary they use — keywords, identifiers, syntactic markers — is overwhelmingly English-derived, and aggressively strips the kind of accretion I'd been noticing. Cookie in software has lost the bakery; daemon has lost the Greek; pipeline has, in technical use, mostly lost the industrial weight even in English. The strip-down is a feature, in some ways — code shouldn't depend on a reader knowing the etymology of fork — but it also closes off a zone of meaning that natural language is constantly using.
What if a programming language were deliberately built around a substrate language whose grammar and vocabulary carried the kind of accreted meaning English-technical-vocabulary refuses? Not just translated keywords — the grammatical structure itself doing semantic work. A language where ser and estar — Spanish's two copulas, distinguishing essential from transient — corresponded to immutable and mutable bindings, because the natural-language distinction was already in the speaker's head. A language where the subjunctive mood — Spanish's marker for hypothetical or deferred reality — corresponded to lazy or conditional evaluation. A language where the diminutive -ito did real numeric work, because the cultural register of cinquito — a small five — was already half-doing it.
The work this hypothesis became
That hypothesis is now a research project. Two of them, actually.
The first, Babel, is a methodology for building esoteric programming languages programmatically: a parameter schema for the variation axes the field has implicitly used for thirty years, and a runtime that turns parameter sheets into working artifacts (interpreter, transpiler, specification page). The second, Inflexión, is a hand-built Spanish-grammar esoteric language that makes the ser / estar, mood-as-evaluation, aspect-as-eager-or-lazy, clitic-as-argument-routing, diminutive-as-scaling moves I just sketched into actual programming-language semantics.
There is prior art in this lane. Lingua::Romana::Perligata did it for Latin in 2000 (Damian Conway). Wenyan-lang did it for Classical Chinese in 2019 (Lingdong Huang). Tampio has been doing it for Finnish since around 2017 (Iiro Sarkkinen). The lane is not empty; Inflexión is the first to take a living Romance language and make this specific feature set load-bearing. I've also done one prior, less formal version of the move in El Pueblo — the Rioplatense word posta is the chosen technical name for a sensor-relay node in an RCI architecture. The natural-language version of the observation predated the formal computational version by some time.
This article is the first in a series. The next ones will go deeper into specific design moves, into what the empirical large-language-model question looks like, into the mechanics of building a programming language whose syntax is Rioplatense Spanish without being a parody. The formal papers (which the series will eventually point at) carry the technical detail; these articles carry the why.
The customer was talking about Microsoft Mesh. I was hearing quilombo. There is a programming language sitting in the gap between those two perceptions, and this series is about why it is worth building.