Repository path: workshop/translations/kalila-nasik/R34-v1/translation.md · rendered 2026-09-09

T-kalila-nasik-R34-v1 — «باب الناسك والضيف» (Kalīla wa-Dimna), rendered whole under R34 ornamentalist

Work. ʿAbd Allāh ibn al-Muqaffaʿ (d. c. 757), «كليلة ودمنة» — the chapter «باب الناسك والضيف», The Hermit and his Guest. The chapter that immediately precedes «باب السائح والصائغ», which this project rendered whole the day before under the opposite regime.

Copy-text. Arabic Wikisource, كليلة ودمنة/باب الناسك والضيف, retrieved 2026-08-16; ../source-wikisource-nasik.txt, 287 words of body text in 19 sentences. Collated whole against the Wikisource transcription of the Cairo Amīriyya (Būlāq) 1937 printing, pp. 299–300, in ../collation.md: one substantive correction adopted ([12] فيها → فيما), the rest optical noise.

Comparator. Wyndham Knatchbull, Kalila and Dimna, or The Fables of Bidpai, Oxford 1819, The Monk and his Guest (archive.org kalilaanddimnao00almgoog), translated from de Sacy's Arabic. Extracted programmatically by chapter boundary and not opened until this rendering and its log were committed. Contamination measured after the freeze and before the study limb was designed: contamination: none — see §5.

Regime. R34 ornamentalist (../../../regimes/R34-ornamentalist.md): render the sense, add no fact, drop nothing — and where English offers the chance, put in sound-patterning and take the words you need for it. The hand is deliberately not confined to the places the Arabic has a figure, so that where it lands is a measurement (§4) rather than a stipulation.

Extent. The chapter whole: 287 Arabic words → 559 English words.

Why this chapter. RS-20260816 §10.4 recorded a limit in its own words: "the decoy arm was forbidden to add material. A translator inventing ornament in the wild adds words … this run therefore says nothing about supplied ornament as it actually occurs." This rendering is the specimen that limit asks for. That the fable it tells is about a guest who abandons his own tongue for a borrowed one, and a crow who tries to walk like a partridge and ends the ugliest walker among birds, was not planned and is not evidence of anything.

1. The translation

The Hermit and his Guest

King Dabshalīm said to Baydabā the philosopher: I have heard that fable. Now set me a fable of the man who lets go the work that is his own — the work that fits him and befits him — and goes after another that he never comes to, and is left between the two of them, wavering and wondering.

The philosopher said: They say that in the land of al-Karkh there was a hermit, a holy man and a hard-labouring one. One day a guest came down to him, and the hermit called for dates for his guest, to give him a taste of what he had never tasted; and the two of them ate of them together.

Then the guest said: How sweet these dates are, and how savoury! We have nothing of the sort in the land I live in — and would that we had! Then he said: I think you might help me to take some of it away and plant it in our own ground; for I know nothing of the fruits of this land of yours, nor of the ground they grow in.

The hermit said to him: There is no ease in that for you: it will be a weight on you and a wearying. And it may be that it would not agree with your ground at all; and your own country is full of fruit — what need has a country so full of fruit of the date, that is heavy in itself and hard upon the body?

Then the hermit said to him: He is not counted wise who seeks what he shall never find. And yours is a fortune of the kindest kind, if you are content with what you find, and content to go without what you do not find.

Now this hermit spoke in Hebrew. The guest found his speech fine and was fond of it, and forced himself to learn it, and worked at himself over it for days. Then the hermit said to his guest: How fit you are to fall — for the speech you have left off and the speech you have laboured after — into the very thing the crow fell into.

The guest said: And how was that?

The hermit said: They say that a crow saw a partridge pacing and picking her way, and her going pleased him, and he grew greedy to learn it. So he broke himself in to it, and could not master it, and despaired of it, and would go back to the walk that had been his own: and behold, his walk had come out jumbled and jointless, and he became the ugliest walker of all the birds that walk.

And I have set you this fable only because I have seen that you have left the tongue you were born to and gone over to the tongue of the Hebrews, which does not befit you; and I fear you will fail of the one and forget the other, and come back to your own people the worst-tongued among them. For it has been said: he is counted a fool who takes upon himself, of all the things there are, what does not befit him, and is not of his own doing, and what his fathers and his forefathers never schooled him in before.

2. Translator's log — decisions

Frozen with the rendering. Decisions and alternatives; no quality claims.

E1 The regime, and what it licenses. R34 rule 1 is the constraint RS-20260816's pre-run critic imposed on that run's arms and it binds here unchanged: no fact, event, referent or evaluation added. The licence is rule 3 — added material may say again, in a matched shape, what the sentence already says. §4 records the plain wording refused at every ornamented site so that the line between amplification and addition can be checked rather than believed. The one place I think the line is thinnest is O8, below.

E2 The frame formulas are my own, carried from the neighbouring chapter. King Dabshalīm said to Baydabā the philosopher and And how was that? are the wordings of T-kalila-saih-R30-v1, written by this hand the day before from the same Arabic formulas. This is deliberate self-consistency inside one work; it is also lead–lead repetition, which method note (bhb) records this project measuring at up to 37 contiguous tokens. It is declared here rather than left for a dependence check to find, and it is not evidence of anything about Knatchbull.

E3 صنعه. The work that is his own. صَنْع is doing, making, craft — the thing a man produces and is fitted to. Craft and trade were both live and both narrow it to an occupation, which the closing sentence («وليس من عمله») does not: the fable is about a tongue, not a job. Work is broad enough to carry both and is repeated at the ornamented site.

E4 يليق به ويشاكله → fits him and befits him. The Arabic pairs two verbs of suiting; شاكل is to be congruent with, to be of a man's own shape. English fits / befits is a matched pair differing by a prefix — the nearest English has to the source's own figure-of-likeness — and it is the seed of a repetition carried to the end of the chapter (O22): شاكل recurs at [19] twice and is rendered befit both times. The plain wording refused was the work that suits him.

E5 بأرض الكرخ. The land of al-Karkh — the west-bank quarter of Baghdad. Left as a name. Nothing was done to naturalise it, and nothing was footnoted.

E6 ليطرفه به → to give him a taste of what he had never tasted. أطرف is to present someone with something new to them; the guest's next speech says he has no such fruit at home, so "never tasted" is the sentence's own content and not an addition. The ornament is the polyptoton (taste / tasted), and it costs five words against the plain to set something new before him.

E7 أطيب → savoury. The Arabic pairs two elatives (أحلى / أطيب), which English cannot match morphologically. Savoury is used in its older sense — pleasing to the taste — for the alliteration with sweet. The plain and how good was refused, and it is one of the two places (with E11) where the regime chose a slightly less central word for the sound. Recorded because it is exactly the trade the study limb is about.

E8 يثقل عليك → a weight on you and a wearying. The thinnest place in the rendering. The Arabic is one verb; the English is a doublet, and the second member (a wearying) is amplification rather than translation. Under R34 rule 3 it says again what weight says; under a stricter reading it adds a shade the Arabic leaves implicit. It is kept, and flagged here, because the point of the regime is to see what an ornamentalist's hand does when it is allowed to.

E9 The saj' passage at [11]–[12], and the only place the English can do what the Arabic does. The Arabic rhymes four times on ‑جد: يجد … الجدِّ … تجد … تجد, with the last two in matched cola. English has a rhyme available on ‑ind: never find / kindest **kind / what you find / do not **find — four members on one sound, in the same positions. This is the one figure in the chapter that crossed at full strength, and it is the chapter's showpiece in both languages. Refused: cannot get / your luck is good / what you do get / what you do not get, at three words less.

E10 سعيد الجدّ → a fortune of the kindest kind. جدّ here is luck, fortune. The English is periphrastic because the rhyme-word had to be kind; and your luck is good was the plain reading. No new fact: both say the man is fortunate.

E11 اختلط وتخلع → jumbled and jointless. The Arabic is a true jinās: اخْتَلَطَ and تَخَلَّعَ share خ‑ل and differ in the final radical only, so the two words are a hair apart and mean got mixed up and came apart at the joints. English has no pair that close. Mingled and mangled is the nearest sound-match and was refused: mangled is mutilation, which changes the sense. Jumbled and jointless alliterates instead of chiming, keeps both meanings exactly (تخلع is literally out-of-joint), and costs two words less than the plain confused and loose in his walk. The sound figure is therefore recorded as answered at reduced strength, not carried.

E12 راض على ذلك نفسه → broke himself in to it. راض is the verb used of breaking in a horse. The English keeps the animal metaphor, which the fable's subject earns. No ornament here, and the absence is deliberate: the hand did not reach at every sentence and §4 counts where it did.

E13 أقبح الطير مشياً → the ugliest walker of all the birds that walk. Ugliest, not worst: أقبح is ugliness, and the crow's ruin is that it looks wrong, not that it is slow. The ornament is the polyptoton walker / walk.

E14 وأنت شرّهم لساناً → the worst-tongued among them. A compound was preferred to the worst of them in speech because لسان is the chapter's key word — tongue as organ, as language, and as manner of speech — and the compound keeps it visible at its last occurrence.

E15 What is not in this rendering. No paragraph divisions exist in the copy-text; the nine paragraphs above are mine, cut at speech boundaries. No section break was supplied at They say that a crow saw a partridge, where the inner fable begins. Witness 2's closing colophon is not rendered.

3. The Arabic's sound figures, enumerated before any English was written

The rule, imported verbatim from T-kalila-saih-R30-v1 §3 so the two chapters are counted the same way. A locus is admitted if two or more adjacent cola or members show either (i) saj' — rhyme at their ends — or (ii) muwāzana — the same morphological pattern, or the same word, repeated in matched position.

Exclusions, published so the enumeration can be tested. (a) Rhyme produced solely by an identical enclitic pronoun (‑هُ، ‑ها) is not admitted — this excludes يليق به ويشاكله [2], كلامه وأعجبه [14] and several others. (b) The cognate figure (root echo between a verb and its object) is a separate class and is not counted. (c) Repetition of the narrative formula قال … قال is not a chosen figure. (d) The recurrence of شاكل across [2] and [19] is a figure of the whole chapter and is not localisable, so it is excluded here in the way T-kalila-saih-R30-v1 D2 excluded السائح / الصائغ.

Ten loci admitted, in copy-text order. Sentence numbers are ../collation.md §2's.

Density: ten loci in 287 words, one per 28.7. «باب السائح والصائغ» ran at one per 71 overall and one per 37 inside the philosopher's frame discourse. This chapter is almost all frame discourse — a king's question, a hermit's admonition, a four-sentence fable and a moral — and its rate matches the other chapter's frame rate rather than its narrative rate. That is consistent with T-kalila-saih-R30-v1 §3's observation that the chapter's sound-work is where its argument is, and it is one further chapter, not a demonstration.

4. Where the ornamentalist's hand actually landed — the census RS-20260816 §10.4 asked for

Twenty-two ornamented sites in 559 English words. For each: the device, its class, the plain wording refused, and the word cost. The at column is scored against §3's inventory, which was fixed before any English existed.

4.1 The three numbers this census exists to print

1. Every one of the source's ten figures was answered. G1–G10 each have an ornament at them, and one of them (G4, the saj' chain) is answered in kind and at full member-count. G7, the jinās, is answered by alliteration and is recorded in E11 as reduced, not carried.

2. The hand put ornament at twelve further sites, and eight of those are on nothing at all. Of the twelve, four (O1, O2, O7, O10) sit at places where the Arabic has matched members with no sound — the doublets §3's rule excludes because their only rhyme is a grammatical enclitic. Eight (O4, O6, O8, O12, O14, O16, O18, O22) sit where the Arabic offers neither sound nor matched members.

So the ornamentalist's hand landed on some structure the source actually has at 14 of 22 sites, and invented freely at 8. That is a fact about one chapter and one hand, and it is the first number this project has on where supplied ornament goes when nobody is holding it in place.

3. The licence cost +34 words, 6.1% of the rendering. Site-level deltas run from −3 to +5; five sites cost nothing or less than nothing, because a matched pair can be shorter than the periphrasis it replaces. R32, the arm this regime exists to un-starve, was allowed 0% by construction and came in at +13 words over thirteen loci.

4.2 What this census is not

It is not blind and it is not evidence about a reader. The same hand wrote the English, enumerated the Arabic and scored the at column, and the scoring rule — is there an admitted figure at this site? — is mechanical only because §3 was frozen first. Whether any of these twenty-two devices is audible to anyone is the study limb's question and is not asserted here.

5. Contamination

contamination: none. Measured with tools/dependence_check.py after this file and its log were committed at cf2dbb8f and before the study limb was designed, against Knatchbull 1819's The Monk and his Guest — the only freely reachable published English of this chapter, extracted by chapter boundary and stored without being opened. The measurement is a selection gate (CLAUDE.md), not a flag.

Raw output in ../dependence.json. The longest common run is five tokens — "fable of the man who" — and the two texts share no seven-gram at all. The neighbouring chapter measured 21 shared 7-grams and a longest run of 11 against the same translator the day before. No project-wide record is claimed: dependence*.json files elsewhere in workshop/ carry runs of 3 and 4 tokens, on per-segment measurements and on other language pairs, and this figure was not compared against them systematically.

Two reasons not to read anything into the low figure. Knatchbull is translating a fuller recension — 479 words of English against this rendering's 557, on an Arabic that his source expands rather than matches — and R34 is a regime that deliberately rewrites the plain wording at twenty-two sites (§4), which mechanically lowers overlap with any plain hand. A regime built to add ornament is not a clean-room, and this number must not be cited as one. What it does license is the ordinary gate: nothing here needs the comparator to explain it.

6. Lineage and pointers

• Regime: ../../../regimes/R34-ornamentalist.md
• Collation: ../collation.md
• Arm: ../../../../wiki/arms/ARM-answering-figure.md
• The limit this rendering answers: RS-20260816-answering-figure §10.4
• Evaluation: this rendering is not itself judged. The study limb it is wired to applies R34 to declared loci of T-kalila-saih-R30-v1; see ../../../experiments/E-20260816b-invented-figure/design.md.