Repository path: workshop/translations/kalila-fanza/R39-v1/translation.md · rendered 2026-09-09

T-kalila-fanza-R39-v1 — «باب ابن الملك والطائر فنزة» rendered whole under R39, the forced matched shape

The sixth Kalīla wa-Dimna chapter this project has rendered, and the first under a regime aimed at a measured fact about a published translator. 1,284 Arabic words → 2,348 English words, 81 sentences, on a copy-text collated in full against the Amīriyya (Būlāq) 1937 printing.

Why this chapter and this regime. A-knatchbull-kalila §3.1 measured the one published English hand translating this work from the Arabic at 0 of 18 muwāzana loci — matched morphological pattern in matched position — and Knatchbull's own 1818 preface says it was impossible to express the sententious brevity of the Arabic with strict fidelity. Nobody has since tried at every one of a chapter's matched-shape loci and written down where it failed. R39 (../../regimes/) is R38 with one clause added: at every SHAPE locus the hand writes MATCHED or IMPOSSIBLE, quoting the English, with no third verdict and no silent reduction.

The rendering, the inventory and every verdict below were frozen before Knatchbull's chapter XII was opened. The inventory was committed separately, at 1c749ff0, before a word of English existed.

• Source: ../copy-text-adopted.txt (Arabic Wikisource, collated against Būlāq 1937 — ../collation.md; three substantive losses restored, fourteen corruptions corrected)
• Regime: R39 v0.1 · Version: v1 · Translated by: the lead agent, in session, at no API cost
• Rendering: text.txt · Inventory: inventory-frozen.md
• Contamination: none — measured, §6

1. The rendering

The full text is text.txt. Three passages, printed here with the Arabic, because they are where the regime bites.

The bird's curse on kings — [10]–[13], the chapter's first burst of saj':

ويل لمن ابتلي بصحبة الملوك الذين لا حمى لهم ولا حرمة … فإذا ظفروا بحاجتهم منه، فلا ود ولا إخاء، ولا إحسان، ولا غفران ذنب، ولا معرفة حق! هم الذين أمرهم مبني على الرياء والفجور، وهم يستصغرون ما يرتكبونه من عظيم الذنوب، ويستعظمون اليسير إذا خولفت فيه أهواؤهم.

Woe to him who is afflicted with the company of kings, who give no protection and hold nothing sacred … And when they have got their need of him, there is no love, and no brotherhood, and no benefaction, and no remission of a fault, and no knowing of a right! They are those whose whole affair is built upon show and upon vice, and they belittle the great sins they commit, and magnify the least thing if their desires are crossed in it.

The bird names himself — [23]–[24], the chapter's best-known lines, and the two densest SHAPE loci in it:

إن العاقل يعد أبويه أصدقاء، والإخوة رفقاء، والأزواج ألفاء، والبنين ذكراً، والبنات خصماء، والأقارب غرماء، ويعد نفسه فريداً. وأنا الفريد الوحيد الغريب الطريد.

The man of sense counts his parents well-wishers, and his brothers fellow-travellers, and his wives partners, and his sons a remembrance, and his daughters wranglers, and his kinsmen creditors — and counts himself alone. And I am the lone one, the sole one, the strange one, the driven one.

The Arabic runs five members on فُعَلاء and breaks the pattern at exactly one member — ذكراً. The English runs five members on ‑ers and breaks at the same one. That correspondence was not planned; it fell out of rendering the sense.

The one place the regime's price is plainly visible — [69]:

ومن لا يقدر لطاقته طعامه وشرابه، وحمل نفسه ما لا تطيق ولا تحمل، فقد قتل نفسه.

And whoever does not measure to his own bearing the food of him and the drink of him, and burdens himself with what he cannot bear and cannot be burdened with, has killed himself.

His food and his drink is the plain English and it is what any hand would write. The Arabic rhymes both members on the enclitic ‑hu, and F81 is admitted on that ground as well as on muwāzana; answering the rhyme in kind puts the pronoun at the member-end, which English can do and does not want to do. The mannerism is the price and it is recorded rather than smoothed away.

2. The rule, and what it added to R38

R38's six rules, imported unchanged, with rule 3a added. The full statement is ../../regimes/R39-matched-shape.md, including the four English resources — (A) matched derivational suffix, (B) matched syllable-count and word-class, (C) matched syntactic frame with matched member length, (D) matched inflection — declared before the rendering so they could not be adjusted after the count.

The declaration was deliberately generous, and §4.2 reports the count both ways.

3. The inventory

inventory-frozen.md, committed at 1c749ff0. Ninety-seven loci, 7.55 per 100 source words, of which 24 are admitted on SHAPE. One tightening was declared there before enumeration and it matters to everything below: a matched syntactic frame without matched morphology is not muwāzana and is not SHAPE, which cost seventeen loci that class. Without it, SHAPE here would not be the class A-knatchbull-kalila counted at 0 of 18.

4. The census: where the devices landed

4.1 The SHAPE verdicts — R39 rule 3a, all twenty-four

4.1a Erratum — F91's resource was wrong, and three blind seats caught it

Added 2026-08-21, after E-20260821-matched-shape ran and before this page was cited anywhere.

F91 above is recorded as MATCHED on resource (A), matched derivational suffix, on the ground that the second, the third, the fourth, the fifth run four ordinals on ‑th. Three seats, reading the published translator's English at the same locus, independently picked out exactly those four words and all three called the relation frame-only — a repeated grammatical frame and nothing else. They are right and this page was wrong: English ordinals are not a matched suffix set. Second ends ‑nd, third ends ‑rd; only fourth and fifth share ‑th, and what actually repeats is the ___, which is a frame.

F91's resource is corrected to (C), matched syntactic frame. Its verdict stays MATCHED and stays free — English supplies the frame without being asked — so §4.2's free count is unchanged at 4. The strict-resource set loses F91 and gains nothing, and because F91 was already excluded from that row as free, the figure of 16 in §4.2 does not move. What moves is the composition, and it is corrected here rather than left standing: sixteen of the twenty-four are matched by English morphology, and none of those sixteen is F91.

The correction was made by a hand that had never seen this page, was not asked about it, and was looking at a different translator's English at the same place in the Arabic.

4.2 The four numbers this census exists to print

Every one of twenty-four matched-shape loci was answered in kind, and sixteen of them were answered by matched English morphology rather than by matched length or matched frame. The "free" row is the honest floor: at four loci the plain English already matched, so those four say nothing about what a hand can do — the same shape of caveat as A-knatchbull-kalila §3.3's finding that English puts a pronoun at the clause-end whether anyone is answering anything or not.

IMPOSSIBLE was never written. That verdict existed and was reachable; the honest report is that this hand did not reach for it once, under the resources this regime declared in advance.

4.3 The whole inventory, and where the rendering fell short

Across all 97 loci: 88 answered in kind, 5 reduced, 4 not answered.

The four not answered, with the reason:

The five reduced: F7 (إكراماً وتعظيماً — matched ‑ing, no rhyme), F20 (the عقب half of punishment/offspring has no English root-pair; the عجل half landed), F54 (أحد … لأحد — no one … another), F77 (هلاك … مهالك — ruin … destruction), F88 (ذهب's three shapes answered by repeating way, which is repetition standing in for root-echo).

4.4 What the regime cost in length, measured

2,348 English words on 1,284 Arabic — a ratio of 1.829. T-kalila-qird-R38-v1, the same hand on the same work under R38 without rule 3a, ran 2,220 on 1,201 — 1.849.

Forcing an answer at every matched-shape locus did not lengthen the English. The added clause is a constraint on which words are chosen, not on how many. This is worth stating because the obvious objection to R39 is that it buys its matches with padding, and on this chapter it did not.

4.5 What this census is not

It is a record of what the hand did against a rule it declared, and nothing in it is a quality claim. Whether the lone one, the sole one, the strange one, the driven one is better English than solitary, alone, a stranger, an outcast is not a question this page answers or is entitled to answer; the lead never judges its own translation (charter §5). Tier D is NOT PASSED and no jury saw any of this.

5. Translator's log — decisions

Frozen 2026-08-21, before Knatchbull's chapter XII was opened and before any evaluation of this rendering was designed. Decisions and alternatives, not quality claims.

5.1 Rendering decisions

1. الموتور — the unavenged man. The 1937 edition footnotes it: «من قُتل له قتيل فلم يُدرك بدمه» — one whose kinsman has been killed and whose blood-claim is unsatisfied. English has no single word carrying the unpaid claim, and the injured party loses the blood. The unavenged man keeps the claim and loses the kinsman. Its counterpart الواتر — the one who did the killing and owes the blood — is rendered the man who has taken blood, four words for one, and the pair الواتر/الموتور (F51, F76) is answered on blood … blood rather than on a root.
2. أهل التِّرات [1], ذوي الأوتار [81] — men with blood between them. The 1937 gloss is «جمع تِرَة وهي الثأر». The frame is not about enemies in general; it is about parties to an unsettled killing, and the chapter's whole argument depends on that. The frame's inclusio (the same root opening and closing the chapter) is in the declined list as non-localisable, and the English repeats the phrase anyway because the sense requires it, not as a figure.
3. نفس — soul throughout, against self and life. [58] turns on «تختلني عن نفسي والنفس تأبى الموت», where the same word is the thing beguiled away and the thing that refuses death. Self would carry the first and not the second; life the second and not the first. Soul carries both and is slightly grander than the Arabic. F66 is answered on it.
4. قدر — measure, and measure only, at [51]–[55] and [72]. The king's fatalism turns on «قدراً مقدوراً» (F61) and on repeating القدر (F62); destiny, decree and fate each break one of the two. A measure measured out is the cognate figure answered in kind, and it costs the sentence some of its weight — a thing destined is grander English and would have killed the figure.
5. The five virtues [76] — the ordinals carry the figure and English gives them free. Arabic runs أولهن then four feminine ordinals on فاعِلة; English runs first then four ordinals on ‑th. Both languages break the pattern at the first member. Recorded as MATCHED — free because no choice was made.
6. «مواقع ممكنة موجعة» [30] was left plain — their places in hearts, deep-set and sore — because the inventory declined it (مُفَعَّلة against مُفْعِلة is not the same pattern) and rule 4 forbids a device where the rule found none. Fixed and afflicting was written and struck.
7. «خضوع … تضرع … مصانعة» [45] was deliberately de-rhymed. The first draft ran submission, entreaty, conciliation — three ‑ion endings at a place the inventory does not mark. Rule 4 does not care that the rhyme is pretty. Rewritten as a bowed head, a plea, fair dealing.
8. «الرياء والفجور», «الضغائن والأحقاد», «المكر والخديعة والحيل», «الحول والقوة», «النكبات والشدائد» [12] [33] [36] [67] [78] — five doublets a hand wants to answer and the rule declines, all on differing patterns and differing rawī. They stand plain: show and vice, grudges and rancours, cunning and deceit and stratagem, force and strength, calamities and hardships. This is the largest single thing rule 4 costs in this chapter, and it is visible: those five are the flattest phrases in the rendering.
9. [16]'s two figures were traded against each other and the trade is recorded. Hastening / tarrying answers the muwāzana of عاجل/آجل; punishment … was hastened in [17] then answers the عجل root-echo. What is lost is عقوبة/أعقاب — punishment and offspring from one root — for which English offers nothing.
10. Direct speech is kept as direct speech. A-knatchbull-kalila §3.4 records that the published hand converts the dialogue to indirect throughout and that this removes the colon structure the figures need. That is his choice and not a rule of English; this rendering keeps the speeches, which is why the loci are still there to answer.

5.2 Two decisions the collation forced

1. The bird is named in [2]. The Wikisource copy-text has lost «وكان له طائر يقال له فنزة» and with it the antecedent of this bird; the name first appears eight sentences later. Restored from witness 2, which is also where the chapter's title-word survives — the 1937 body text reads فترة for فنزة almost everywhere, including in its own table of contents.
2. «وتختلني عن نفسي» [58], not «وتخلني». Witness 2 prints تكتلني; the root is ختل, to beguile, to stalk. The copy-text's تخلني — leave me to myself — reverses the king's intent in the middle of the bird's accusation, and would have made F65's pair (healing … stealing) impossible as well as wrong.

5.3 One evaluative sentence, flagged

internal-judgment-only: true — the four MATCHED — free loci make me distrust the count of 24 more than the sixteen strict-resource matches make me trust it. A rule that admits a match the plain wording already had is a rule that will find matches. That is why §4.2 prints the strict figure beside the generous one and why the blind census (E-20260821-matched-shape) puts the published hand's loci to seats that never saw this page.

6. Contamination

contamination: none — measured, not asserted.

tools/dependence_check.py against the comparator, run after this rendering was frozen and committed, on Knatchbull 1819 chapter XII (The King and the Bird): 1 shared 7-gram, 0 shared 12-grams, 0 shared 15-grams, longest common run 7 tokens — the worst of wives is she who — on 2,551 comparator tokens against 2,348. ../dependence.json. The same profile as T-kalila-qird-R38-v1, whose run was also 7 (bring him to me a second time), and clean by the project's standing rule.

The rendering was written from the Arabic alone. Knatchbull's chapter XII was extracted from the Internet Archive OCR by a script that printed only its line range and word count, and no line of it was read until this page was committed. The chapter originally chosen for this session, «باب الجرذ والسنور» (Knatchbull XI), was abandoned because roughly a hundred and twenty of its four hundred lines had already been displayed while locating the chapter boundaries; that is recorded here rather than absorbed, because the abandonment is the reason this chapter is the one rendered.

What the standing rule asks and what was done. CLAUDE.md's contamination rule wants the measurement before the design. A lead rendering cannot be measured before it exists, so the gate was set as a pre-registered failure criterion instead: E-20260821-matched-shape F2 withholds every lead-versus-Knatchbull comparison if this check returns a shared 12-gram or a longest run above 10 tokens. The census of the published hand does not depend on the lead's independence and stands either way.

6.1 What was read, and when

7. Lineage and pointers

• Regime: ../../regimes/R39-matched-shape.md (and R38-repaired-inventory.md, imported)
• Copy-text and collation: ../copy-text-adopted.txt, ../collation.md
• Inventory: inventory-frozen.md
• Sibling renderings of this work: ../../kalila-qird/R38-v1/, ../../kalila-labwa/R37-v1/, ../../kalila-ibn-urs/R36-v1/, ../../kalila-nasik/R34-v1/, ../../kalila-saih/R30-v1/
• The anchor this rendering was made against: wiki/base/anchors/A-knatchbull-kalila/README.md
• The arm: wiki/arms/ARM-published-figure.md