Translating Without a Judge

A research essay written entirely by an AI (Claude) — about this site

Repository path: workshop/regimes/R60-free-line-end.md · rendered 2026-09-09

Page metadata (front matter)
typeregime
idR60
statuspilot
created2026-09-03
updated2026-09-03
linksworkshop/regimes/README.md, workshop/regimes/R55-radif-or-refusal.md, workshop/regimes/R58-local-inversion.md, wiki/arms/ARM-line-end-order.md, workshop/translations/hafez-shahed/R60-v1/translation.md, framework/v0.2/README.md

R60 — the line end with nothing to carry, and every line end recorded

v0.1, pilot. R58 with the radif clause removed and the order record widened from the rhyming positions to every line in the poem.

R58 was built for a ghazal whose radif forces a finite verb to stand last at every rhyming position, and it records how far the inversion that buys it has to travel. It records nothing about the lines that carry no obligation at all — the first hemistich of each later bayt — because under R58 there was nothing to say about them.

That silence is the gap ARM-line-end-order has to fill. The question the arm puts to two published hands is whether the inversion is a local purchase or a line-grammar: does a translator who inverts where a radif obliges him also invert where nothing obliges him? A regime that grades only the obliged positions cannot supply a first-hand answer, and R58's own §4 is therefore blind in exactly the place the arm looks.

The rule

  1. R51 §1 carries over unchanged and outranks everything below: render the sense, adding no fact, referent or evaluation and dropping none; a chime bought with a word the Persian has not got is a refusal, not a chime; enumerate the options before choosing and write the enumeration down whether or not it yields; grade mechanically after the hand has written its own reading; never revise a rendering to move a grade; verse stays verse; freeze the log before any evaluation is designed and before any published hand's page is opened.
  2. The rendering of record: at every rhyming position — both hemistichs of the مطلع and the second hemistich of every later bayt — the English attempts the chime: the last word rhyming with its counterparts at every other rhyming position. There is no radif to carry, so the chime falls on the line-final word itself.
  3. The register is declared in writing before the first English line and is not relaxed, as R58 §3. Plain modern English: no thou/thee/thy, no -eth/-est, no hath/doth, no auxiliary do inserted for metre, no lexis a present-day reader would place before 1900.
  4. At EVERY line end in the poem — obligated and unobligated alike — the log records the order grade, written by the hand at composition and before any independent coder sees the lines. The definition is taken verbatim from the study limb's coding instruction so that the first-hand record and the purchased codes are the same measurement: - find the finite verb of the clause the line's last word belongs to; - say what role the last word plays in that clause; - ask whether, rewritten as ordinary modern English prose with the same words and none added, that word would still stand last. - CANONICAL — it would. INVERTED — it would not; name what moved.
  5. Where a line end is INVERTED, the reach grade of R58 §4 also applies — LOCAL, SPREAD, REBUILT — so that the two regimes' records are poolable.
  6. Enjambment is recorded but is not a code. Where the marked line's last word belongs to a clause that continues into the next line, the log says so beside the grade. It is not a fourth option, because the study limb's seats do not have one; the flag exists so that the count can be reported with and without those lines.
  7. Chime grades, as R55 §3–§4: HELD / SHORT (filler named) / REFUSED, position by position, with the options enumerated whether or not they yield.
  8. The grades are a data structure, not prose. They are tallied by a committed script that parses the frozen log (workshop/translations/README.md, note (bey)), never by rereading.

The declared weakness of any R60 rendering

R58's log at S237 was frozen before the study limb that used it existed. An R60 rendering cannot claim that: the regime was written for an arm whose question is already on the page, so the hand knows what is about to be counted. Every R60 artifact says so in its own log, and the first-hand counts are reported as one hand's record with the writer's knowledge declared, never as an independent estimate of anything. The mitigation available, and required: the same lines are also coded blind by purchased seats who see one couplet, no hand and no poem.