Prāsa is an esoteric programming language by Koundinya Dhulipalla that embeds computation inside Telugu classical meter. The mechanics are precise: code is organized into "cells," each 4-space indent moves one cell forward like a Turing Machine tape, each line's cell value equals its syllable count, and parenthesized syllables multiply into the existing cell value. But the decisive constraint is this: the cumulative syllabic stress pattern of your entire program — U for stressed, I for unstressed — must completely match one of four classical Telugu metrical forms: Utpalamāla (UII UIU III UII UII UIU IU, 20 syllables), Campakamāla (III IUI UII IUI IUI IUI UIU, 21 syllables), Mattēbham (IIU UII UIU III UUU IUU IU, 20 syllables), or Śārdūlam (UUU IIU IUI IIU UUI UUI U, 19 syllables). Valid Prāsa code is simultaneously valid Telugu poetry.

In SAE terms, this is a clean chisel-construct cycle. The already-construct of programming language design is the inheritance of Western formalism: ALGOL's block structure, Boolean logic's operators, lambda calculus's function nesting, Python's indentation as purely structural marker. This syntax has never asked about syllable stress. Words in code are identifiers or string literals — their tonal contour, their syllabic rhythm, their shape when spoken aloud, are discarded entirely. Western computing drew its formal grammars from mathematics and formal logic and built a machine that is profoundly deaf to the sonic dimension of language.

Prāsa's chisel is precise: it keeps indentation but makes it tape movement; it keeps text but makes syllabic stress the actual computational constraint. What surfaces as the remainder: Telugu chandassu was never merely a beautiful cultural tradition — it is a two-thousand-year-old formal system, rigorously specifying syllable types (laghuvu/guruvu, light/heavy), arrangement counts, and pattern sequences. Sanskrit chandas, the parent tradition, was a formal system for encoding patterns in time — a rule-governed language for sequences of symbols, which is exactly what a programming language is. When computer science crystallized in the mid-twentieth century, its formal grammars all came from Western mathematics and logic, leaving this equally formal, equally precise, equally machine-verifiable non-Western grammar entirely outside the door. The meter became remainder. What Prāsa demonstrates is that this remainder never disappeared — it simply never got encoded.

The remainder has two dimensions. The first is literal: you write river flows to the right, the parser breaks it into syllables and stress patterns, and validates whether it fits one of the four meters. This is not decorative constraint — it determines whether your program is valid, what cell values are, whether computation can proceed. The second is structural: chandassu and modern computation are formally the same kind of thing (both are rule systems for symbol sequences), yet they never met. Prāsa is an interface where two formal systems coexist in the same line of code and mutually constrain each other.

Dhulipalla's 2025 paper in AI & Society began naming this territory — "vernacular computing" as a framework for decolonial code intervention. The esoteric.codes EsoNatLangs article of April 2026 was the first to cluster Prāsa alongside similar languages as a category. The naming is underway. Once "decolonial code art" hardens into a stable institutional term, Prāsa will be archived artifact rather than living remainder. Reading it now is meeting it before that happens — when the logic is still growing and the meter is still alive.

Han Qin (秦汉) · @hqinjarsy · SAE Judgment & Aesthetics