Repository path: workshop/regimes/R30-sound-plain.md · rendered 2026-09-09
Page metadata (front matter)
| type | regime |
|---|---|
| id | R30 |
| status | pilot |
| created | 2026-08-16 |
| updated | 2026-08-16 |
| links | workshop/regimes/README.md, workshop/regimes/R31-answered-figure.md, workshop/regimes/R32-supplied-figure.md, workshop/translations/kalila-saih/R30-v1/translation.md, workshop/experiments/E-20260816-answering-figure/design.md |
R30 — sound-plain (the base arm of a sound manipulation)
v0.1, pilot, first used at T-kalila-saih-R30-v1. Render the source's sense and syntax; put
no English sound-patterning anywhere in the target, including — and especially — at the places
where the source has some.
The rule
- Render sense and syntactic structure as a close single-pass rendering would (
R06). - Add no alliteration, no rhyme, no jingle, no matched stress-shapes, no matched word-shapes. Where two adjacent English words or phrases would fall into a pair by sound, break the pair.
- Where the source has a sound figure, render what it says and let the sound go. Do not compensate, do not relocate, do not substitute.
- Record every place where rule 2 or rule 3 refused something the translator had in hand, with the refused wording written out. The refusals are the regime's product as much as the text is.
- Syntactic parallelism that carries content — a list, an antithesis, a comparative series — is kept. R30 removes sound, not structure. Where a source figure is both (a matched morphological pattern that English reproduces natively as comparative or verbal anaphora), the English inevitably carries it; such a locus is declared free-crossing in the log and is excluded from any manipulated set built on the rendering.
What it is for, and what it is not
R30 exists to be the base arm of a manipulation: two derived arms (R31, R32) add sound
devices at declared places, and every comparison between them and R30 holds the whole rest of the
text byte-identical. Its value is that it is a floor which is not a damaged text — nothing is
mistranslated, nothing is dropped, and the propositional content is complete.
It is not a recommendation. A rendering that refuses sound everywhere is a worse rendering than one that does not, by the ordinary lights of the craft, and the project's own framework advises carrying the source's figures where they can be carried. R30 is an instrument.
Known limitations
- "No sound-patterning" is a judgement made by the translator about their own prose. The base arm's ratings under an independent instrument are the check on it, and a run that finds the base arm patterned at the manipulated loci must withhold the manipulation's primary.
- The refusal list in rule 4 is written by the same hand that made the refusals, and shows what that hand had in mind, not the full set of options English affords.