Repository path: workshop/translations/kalila-qird/R38-v1/translation.md · rendered 2026-09-09

T-kalila-qird-R38-v1 — «باب القرد والغيلم» (Kalīla wa-Dimna), rendered whole under R38, the repaired inventory

Work. ʿAbd Allāh ibn al-Muqaffaʿ (d. c. 757), «كليلة ودمنة» — the chapter «باب القرد والغيلم», The Monkey and the Tortoise: the deposed monkey-king in his fig tree, the tortoise who befriends him and then carries him out to sea to cut out his heart for a sick wife, and the monkey who saves himself by declaring that monkeys leave their hearts at home. Inside it, the monkey tells the tortoise the fable of the lion, the jackal and the ass with no heart and no ears. The fifth chapter of this work the project has rendered whole, after «باب السائح والصائغ» (R30, sound-plain), «باب الناسك والضيف» (R34, ornamentalist), «باب الناسك وابن عرس» (R36, answering hand) and «باب اللبؤة والإسوار والشغبر» (R37, declared-rule).

Copy-text. Arabic Wikisource, كليلة ودمنة/باب القرد والغيلم, retrieved 2026-08-16; ../source-wikisource-qird.txt, 1,170 words of body text. Collated whole against the Amīriyya (Būlāq) 1937 printing, صفحة:كليلة ودمنة.pdf/234–242, in ../collation.md: eight substantive adoptions from witness 2, four of them restoring narrative the copy-text had lost, including the monkey's leap off the tortoise's back and the jackal's overtaking of the ass. The adopted text is ../copy-text-adopted.txt — 1,201 words in 73 sentences — and it, not the raw copy-text, is what was translated. Sentence numbering follows witness 2's punctuation and is not commensurable with the four earlier chapters' (../collation.md §2).

Comparator. Wyndham Knatchbull, Kalila and Dimna, or The Fables of Bidpai, Oxford 1819, The Monkey and the Tortoise (archive.org kalilaanddimnao00almgoog), translated from de Sacy's Arabic. Located by heading offset and not extracted or opened until this rendering and its log were committed. See §6.

Regime. R38 the repaired inventory (../../../regimes/R38-repaired-inventory.md): one rule for what counts as a sound figure — the union of the project's standing form rule and the readers' sound rule, with the exclusion framework/v0.2 §7.17 struck honoured as struck — the source enumerated once, the inventory frozen, every locus answered, no device anywhere else, and the declined loci recorded with the ground of their declining.

Extent. The chapter whole: 1,201 Arabic words → 2,220 English words (lead-declared.txt), a ratio of 1.85 against R37's 1.80 on the previous chapter.

Why this chapter and this regime. framework/v0.2 §7.17 was written the same day, from RS-20260816h, and it gives a practitioner two instructions about Arabic sound figures: strike exclusion (a) from the inventory rule, and answer the union of a form rule and a sound rule. Both instructions rest entirely on this project's own hands — the lead's renderings and three model seats reading the Arabic. No published human translator's practice is in evidence for this language pair anywhere in the handbook, and the Tier 1 shelf has no anchor with Arabic as source. This rendering is the first half of the repair: it executes the instruction on a chapter Knatchbull 1819 also renders, with the inventory built under the repaired rule from the start, so that the handbook's instruction and the one published English hand translating from the Arabic can be read at the same loci in the same prose.

1. The rendering

lead-declared.txt carries the rendering as a running text. It is reproduced there rather than inline here so that a second hand can read it without the inventory beside it.

Two paragraphs, to show what the regime produces. The monkey's soliloquy at [36]–[38]:

And the monkey said within himself: Alas! Greed and gluttony have overtaken me in my old age, until I have fallen into the worst of straits! And he spoke truly who said: The contented man, the satisfied man, lives at rest and at ease; and the man of greed and gluttony lives as long as he lives in toil and moil. And now I have need of my wits in seeking a way out of what I have fallen into.

against «يعيش القانع الراضي مستريحاً مطمئناً، وذو الحرص والشره يعيش ما عاش في تعب ونصب» — where taʿab wa-naṣab is a rhymed pair of the same pattern and English happens to keep one, toil and moil, in the same register and at the same place.

And the jackal's proverb at [71]:

And it has been said: what lenience spoils, only science repairs.

against «إن الذي يفسده الحلم لا يصلحه إلا العلم» — al-ḥilm / al-ʿilm, a perfect rhyme across an antithesis. The English pair costs a word of register (science for ʿilm is the older sense, knowledge), and §5.1 records that cost rather than hiding it.

2. The rule, written out before any English was written

Imported verbatim from ../../../regimes/R38-repaired-inventory.md, which was written and committed first. It is the union of T-kalila-labwa-R37-v1 §2.1 (Rule A) and §2.2 (Rule B), with §7.17's strike applied and with the two thresholds §7.17 leaves open declared in advance.

A locus is admitted if two or more adjacent cola or members show any of: (i) end-rhyme, the member-ends sharing the same rawī and the same following ending, the rhyme produced by a rhyming word, a verbal person-ending, or an identical enclitic pronoun — where a suffix carries the rhyme, the suffix is the rhyme unit and its own consonant is the rawī; (ii) muwāzana, the same morphological pattern in matched position; (iii) repetition of the same content word in matched position; (iv) ishtiqāq / jinās, the same root in two derivational shapes or two roots in near-identical shapes.

Excluded: bare tanwīn as sole rhyme where the rawī differs; bare feminine ‑ah as sole rhyme; anaphora of a function word; the formula قال … قال; a recurrence spread across the whole chapter.

Suffix-rhyme threshold: three or more adjacent member-ends carrying the identical suffix, or two where the members also match in pattern or in length.

Two things the regime page says and this page repeats because they govern the counts. §7.17 struck exclusion (a) and supplied no threshold to replace it, and ‑هُ is the commonest word-ending in this prose; the threshold above is this hand's declaration, not the handbook's. And taking the union does not only restore (a): Rule B has no exclusion for the cognate figure, so the union kills exclusion (b) as well. §7.17 names only the first of these.

3. The inventory, frozen before any English was written

Fifty-four loci, in copy-text order. Sentence numbers are ../collation.md §2's. RHYME = clause (i), SHAPE = (ii), REPEAT = (iii), ROOT = (iv). (a) marks a locus admitted only because §7.17 struck exclusion (a); (b) marks one admitted only because the union kills exclusion (b).

3.1 The numbers this inventory exists to produce

One locus in six in this chapter's inventory exists only because §7.17 struck exclusion (a). Nine of fifty-four — 16.7%. Two more exist only because taking the union kills exclusion (b), which §7.17 does not mention. Together, eleven of fifty-four — 20.4% — are places the standing rule would have thrown away, and they include the chapter's densest chain (G40, five consecutive clause-ends on ‑هِ as the ass breaks and runs).

For comparison, T-kalila-labwa-R37-v1 §3.1 reports 24 union loci in 599 words — 4.01 per 100 against 4.50 here. The two chapters are enumerated under different rules and the difference between the rates is not a measurement of anything; it is printed so that nobody reads 54 against 24 as a fact about the two chapters' density.

3.2 The twelve loci the rule declined, and why — recorded because R38 rule 6 requires it

R37 did not record these and RS-20260816h needed them. Each is a place a reader might expect a figure and this rule refuses one.

Three of the twelve are places this hand wanted to write a device and the rule stopped it — القيام والقعود, الحرص والشره, الغسل والطهور — and the plain wording in the rendering at those spans is what rule 4 costs. That is the point of writing the declined list down.

3.3 Two loci where distance, not kind, is the question

framework/v0.2 §7.17 item 3 is untested: a repetition close by is owed; a repetition far apart may not be. This chapter supplies two clean instances and neither is in the inventory, both being excluded as non-localisable:

• «مالي أراك مهتماً» at [23] and again verbatim at [31] — eight sentences apart, the same speaker to the same hearer, and plainly a doubling.
• «مالي أراك مهزولاً» at [52] — the same formula a third time, now in the inner fable, in another speaker's mouth, forty-one sentences from the first.

The rendering answers all three with the same English (How is it I see you …) because the sense requires it, not as a figure, and no device is added at any of them. Recorded here for the successor question, not claimed as evidence for it.

4. The census: where the devices landed, and the plain wording refused

For each locus, the English written and the plain wording this hand would have written at the same span under the sound-plain rule (R30). Written into this table at the same sitting as the device.

4.1 The four numbers this census exists to print

Answering the repaired inventory cost 4.4% in length. T-kalila-labwa-R37-v1 §4.1 measured the cost of answering the union at 4.5% on the previous chapter under two declared rules. The two numbers are close and were arrived at independently, and the restored exclusion-(a) loci are the cheapest class in the census — nine loci for fourteen words, because the device English uses to answer an enclitic chain is a pronoun it was going to write anyway.

Where the English answer is of a different kind from the Arabic's device, it is almost always a ROOT locus. English derives too thinly to answer fifteen root-echoes in kind, and where no English derivational pair was available the answer written is a repetition instead — a different figure at the right place. §5.1 records which, and it is the largest single class of approximation in this rendering. No count of in-kind against out-of-kind answers is printed, because the call is this hand's on every locus and an unchecked 54-item judgement by the party with the hypothesis is exactly what RS-20260816h's critic struck elsewhere.

5. Translator's log — decisions

Frozen 2026-08-16, before the comparator was extracted and before any evaluation of this rendering was designed. Decisions and alternatives, not quality claims. The one evaluative sentence is §5.3 and it is flagged.

5.1 Rendering decisions

1. English has no ishtiqāq, and that is the single largest structural loss in this chapter. Fifteen loci are root-echoes — a root appearing in two derivational shapes — and English derives too thinly to answer most of them in kind. Where a pair existed I took it: an exhaustion … has exhausted me (G36), the washing … to wash (G46), sins a sin (G51), is not to be eaten … I will eat (G44). Where none existed I answered with a repetition instead, which is a different figure: G20 تقلباً/القلب became turned against me … in its turning than the heart, two forms of turn plus heart twice, where the Arabic has one root twice. The answer is at the right place and in the wrong kind, and the census marks all thirteen.
2. G43 جاش جأش is jinās proper — two near-identical shapes from different roots — and English has exactly one cheap device for it. The lion's heart was heartened answers a sound figure with a derivational one. It is the weakest answer in the rendering and I could not find a better one; the lion's spirit was stirred and spurred invents a second figure at a place the inventory does not license, which rule 4 forbids.
3. G50 الحلم / العلم cost a word of register. The Arabic rhymes two abstract nouns across an antithesis, in the plainest possible words. English pairs that rhyme and mean forbearance and knowledge are not plentiful; lenience / science rhymes exactly and is the pair I took, accepting that science is being used in its older sense of knowledge and will read as a slight archaism. The plain wording refused, what indulgence ruins, only knowledge can put right, is two words shorter and has no figure at all. This is the one locus in the chapter where answering in kind moved the register.
4. G31 تعب ونصب is the chapter's gift. toil and moil is a real English rhyming doublet of the same register and the same age-band as the surrounding prose, and it sits at the same place in the same clause. Nothing was lost. It is worth recording that a chapter with fifty-four figure loci produced one locus where English simply had the same device waiting.
5. The enclitic chains were easier than expected, and this is a fact about English pronouns. G7 (four clause-ends on ‑هُ), G21 (four on ‑هِ), G40 (five on ‑هِ) are answered by putting an English pronoun at each clause-end — to him · upon him · for his weakness · of him · from him. English will do this without strain because its object pronouns are clause-final by default. T-kalila-labwa-R37-v1 §5.2 found the same thing on a ‑هما chain. Two chapters now agree that the class the standing rule was throwing away is the class English answers most cheaply, which is §4.1's sixteen words for nine loci.
6. Rule 4 bit three times and I recorded it rather than bending. القيام والقعود (standing and sitting), الحرص والشره (greed and gluttony) and الغسل والطهور (washing and purifying) all invited a doublet and the rule declines all three — the first two on pattern, the third on pattern. greed and gluttony at [36]/[37] is alliterative and I let it stand because it is the plain wording (ḥirṣ and sharah have no plainer English pair), not because the rule licenses a device there; if a critic reads it as an invented ornament, the census's plain column is the record of what was intended.
7. Speech is unmarked, as in the four earlier chapters — no quotation marks, X said: …. The copy-text has no quotation marks and the 1937 printing has none.
8. [40]'s custom is left as strange as the Arabic leaves it. he leaves his heart behind with his household or in its place, so that he may look, when we look upon the womenfolk of the visited, and our hearts not with us — the Arabic is elliptical and slightly incoherent here (the number shifts from singular to plural mid-clause) and I have kept the shift rather than tidying it, on the same ground the earlier chapters kept theirs.

5.2 Two decisions the collation forced

1. [39] is the monkey's speech, not the tortoise's (../collation.md §3.1 item 6), and the whole invented custom of the left-behind heart therefore belongs to the monkey. Under the copy-text's reading the rendering would have had the tortoise reproach himself for not being warned, and the fable's hinge would have gone missing. One preposition in witness 2 decides who is clever in this story.
2. [44]'s leap is in the rendering because witness 2 has it. the monkey sprang off his back and climbed up the tree is nine words the copy-text does not contain; without them the monkey is still at sea when he refuses to come down.

5.3 One evaluative sentence, flagged

internal-judgment-only: the rendering is more ornate than I would write unprompted, and the excess is concentrated in the ROOT class, where answering out of kind produces English repetitions that a reader may hear as insistence rather than as figure. Not a quality claim about the source or the regime; a note on where an independent reader should look first.

6. Contamination

contamination: none, on a measurement made after the freeze and reported in full.

Knatchbull's The Monkey and the Tortoise was located by heading offset in the archive.org OCR (kalilaanddimnao00almgoog, offset 323741–336164) and neither extracted to a file nor read until this page and lead-declared.txt were committed. The measurement below was run on the extracted comparator immediately after that commit; the figure and its basis are recorded in §6.1 and in ../dependence.json.

6.1 What was read, and when

• The Arabic copy-text and witness 2: read whole, before translating.
• framework/v0.2 §7.17 and T-kalila-labwa-R37-v1 §2–§3: read before the inventory was built. They are this project's own pages and the rule comes from them; that is the design, not a contamination.
• Knatchbull 1819: not read until after the commit. No other English rendering of this chapter was consulted at any point.

7. Lineage and pointers

• Regime: ../../../regimes/R38-repaired-inventory.md
• Copy-text and collation: ../copy-text-adopted.txt, ../collation.md
• Predecessor renderings of this work: ../../kalila-saih/R30-v1, ../../kalila-nasik/R34-v1, ../../kalila-ibn-urs/R36-v1, ../../kalila-labwa/R37-v1
• The rule this executes: framework/v0.2 §7.17, from wiki/findings/results/RS-20260816h-target-set.md
• The arm: wiki/arms/ARM-published-figure.md