Translating Without a Judge

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

Repository path: workshop/experiments/E-20260831-arabic-in-persian/coding.md · rendered 2026-09-09

Page metadata (front matter)
typeexperiment
idE-20260831-coding
statusfrozen
created2026-08-31
updated2026-08-31
sensesstyle-correspondence, voice
linksworkshop/experiments/E-20260831-arabic-in-persian/loci-frozen.md, wiki/findings/results/RS-20260831-arabic-in-persian.md, wiki/base/anchors/A-gulistan-hands/README.md

The coded cells, with the English quoted verbatim

{{ }} marks italic in the book, recovered from each scan's ABBYY layer by abbyy_italics.py; _djvu.txt does not carry font information and would have hidden the study's main signal. OCR is uncorrected: garbled characters inside a quotation are the scan's, not the printer's, and none of the codings below turns on a garbled character.

Departure from the frozen design, declared

loci-frozen.md enumerates 21 loci across the whole chapter. The primary count here is the 15 loci in حکایات ۱–۳۱ — the part of باب دوم this arm has now translated (spans A, B, C) — and the six loci in حکایات ۳۲–۴۸ are reported as a partial supplement at the foot, because the alignment work for them was not finished in the session. The restriction was chosen for a reason that is not the data's: it matches the arm's own frontier, and span D will carry the rest. Two of the 60 primary cells could not be aligned (Ross at b022 and at b201), which is 3.3% and well inside the frozen failure criterion of 25%.

Marking codes

IT italic · QU quotation marks setting the Arabic off from the surrounding English · TR Arabic type or transliteration · SAY an English phrase or footnote naming the language or the source where the Persian did not. PLAIN = none of the four. SAY-SOURCE (Sa'di's own frame, translated) is not marking and is not counted.

The 15 primary loci

locus class GLA 1806 ROS 1823 EAS 1852 ARN 1899
b009 unframed PLAIN IT QU DROPPED
b022 unframed PLAIN not aligned IT (+ footnote in Arabic type) DROPPED
b033 unframed PLAIN IT IT DROPPED
b052 unframed PLAIN IT + QU IT DROPPED
b059 framed PLAIN IT IT + QU QU
b060 unframed PLAIN PLAIN PLAIN PLAIN
b063 unframed PLAIN IT + QU IT DROPPED
b072 framed QU IT IT + QU TR + QU
b092 unframed PLAIN IT IT + QU DROPPED
b138 unframed PLAIN IT IT DROPPED
b174 unframed PLAIN IT + QU IT MERGED
b201 unframed PLAIN not aligned IT DROPPED
b209 unframed PLAIN IT IT + QU DROPPED
b218 framed + glossed PLAIN, gloss only IT IT + QU PLAIN, gloss only
b236 unframed PLAIN PLAIN IT DROPPED

MERGED = the content is not separately identifiable in the English because the hand has run the Arabic couplet and the Persian couplet that glosses it into one poem. Counted apart from DROPPED throughout.

The evidence, quoted

b009 «اِصْنَعْ بی ما اَنْتَ اَهْلُهُ» — Do with me what You are worthy of. - GLA: "Do unto me that which is worthy of thee ; and treat me not according to my desert." Roman, inside the quotation marks that hold the whole speech. - ROS: "{{do unto me what is worthy of thyself; but deal not with me, as I myself have deserved:}}" - EAS: the surrounding prose is unquoted, then — " Do unto me that which is worthy of Thee, and not that of which I am worthy." - ARN: the prayer runs "…I am come hither to beg, not to bargain !" and stops. Absent.

b022 «اِنْ لَمْ اَکُنْ راکِبَ الْمَواشی…» - GLA: "Although I am not mounted on a beast, I will endeavor to carry your burdens." Roman, run into the paragraph as prose, which is Gladwin's verse policy. - EAS: "{{What though I'm borne [ ] not in the camel throng, / Yet will I strive to bear your loads along}}" — italic, where every other COUPLET and STANZA on the same page is roman — plus footnote 141: "There is an attempt here at a pun in the words رکب rakib, 'I am riding,' and حامل hamil, 'I am bearing.'", setting the two words in Arabic type. - ARN: absent from his tale V.

b033 «وَالسَّلامَةُ فی الْوَحْدَةِ» — and safety is in solitude. - GLA: "because in solitude there is tranquillity." Roman. - ROS: "{{for there is safety in unity, but danger in duality or a multitude:}}" - EAS: "and kept to the path of retirement, for, {{in solitude there is safety."}}" - ARN: "…and to pursue the road of retirement. If any in a company commits folly…" Absent.

b052 «کَفَیْتَ اَذَیً یا مَنْ یَعُدُّ مَحاسِنی…» - GLA: "whilst thou who reckonest my good works judgest from the external, but art ignorant of the interior." Roman prose. - ROS: " // {{suffices with thee, O sir! that summest up my good works, as they appear outwardly, but thou knowest not the secrets of my heart}}" - EAS: "{{Thou who wouldst sum my virtues up, enough thou'lt find / In outward semblance ; to my secret failings blind.}}" - ARN: goes from the prose straight to the Persian couplet "{{Outward, to men's eyes, I go great and proud…}}". Absent.

b059 «لی مَعَ اللهِ وَقْتٌ…» — framed: «نشنیده‌ای که خواجهٔ عالم … گفت». - GLA: "said. There is a time in which God has given me a degree of power, that is not allowed either to the nearest angel, nor to any mortal prophet sent from God" — roman, unquoted. Only Sa'di's frame, translated. - ROS: "{{there is a time when I am with God; and he will not show the favour he does me to the angels of his affection and prophets of his revelation;}}" - EAS: "said, '{{I have a season with God, in which neither ministering angel, nor any prophet that has been sent, can vie with me,'}}" - ARN: "once heard these words: ' There be seasons when thou dost not cease to consort with the angels Jibrael and Michael…'" — inner single quotation marks inside the outer double.

b060 «مُشاهَدَةُ الاَبْرارَ بَیْنَ التَّجَلّی وَ اْلاِسْتِتارِ» — the one locus no hand marks. - GLA: "The vision of the pious consists of revelation and obscurity. It discovers and it conceals." - ROS: "The visions of the inspired are revelation and tenebrosity: he is directing and bewildering them:" - EAS: "for the vision of the pious is between effulgence and obscurity; at one moment He shews Himself, at another snatches Himself from our sight." - ARN: "…because the vision of the Most High discloseth itself and concealeth itself.'" — swept inside the preceding quotation, which is the hadith's.

b063 «اُشاهِدُ مَنْ اَهْویٰ بِغَیْرِ وَسیلَةٍ…» - GLA: run into the prose, roman. - ROS: " {{I behold him}} [God] {{whom my heart desires, without the veil of intervention; therefore I become like him, who loses his way}}…" — italic, with a bracketed gloss. - EAS: "{{I'll with uninterested gaze survey / Him whom I love, and, wildered, lose my way…}}" - ARN: absent; he prints only the Persian couplet before it.

b072 «و نَحْنُ اَقْرَبُ اِلَیْهِ مِنْ حَبْلِ الْوَریدِ» — framed: «این آیت». - GLA: "in explanation of this verse of the Koran, "We are nearer to him than his jugular vein."" — the one place in fifteen where Gladwin sets the Arabic off. - ROS: "of this text of the Coran, {{we are nearer to him}} [God], {{than the vein of his neck,}}" - EAS: "explanation of the verse, "{{We are nearer to him than the jugular vein."}}153" - ARN: "the chain of my discourse extended to the explanation of that {{ayat}}, "We are nearer to him…" — Sa'di's own word آیت kept, transliterated and italic.

b092 «وَ الْفقیرُ لایَمْلِکُ» - GLA: "because the beggar is not the proprietor of anything, neither is he the property of any one" — roman. - ROS: "{{ligious mendicant is not the proprietor of any thing;}}" (the run begins mid-word at a page break: "for a re-{{ligious mendicant…}}"). - EAS: "for ' {{the fakir does not possess anything, and is not possessed by any one.'}}" - ARN: "because that which belongeth to the darweesh is {{waḳf}}, and belongeth already to the poor." The Arabic maxim is gone; a Persian term is transliterated in its place.

b138 «نُهاجُ اِلیٰ صَوْتِ اْلاَغانی…» - GLA: "The heart may be captivated by the sound of sweet melody, but such a singer as thou art can only give delight by being silent." Roman prose. - ROS: " {{The heart may be charmed by the voice of sweet melody; but such a singer as thou art can only please by being silent:}}" - EAS: "{{We haste to music's sound with stirred and kindling breast, / But thou a minstrel art, whose silence pleases best.}}" - ARN: prints only the Persian couplet, "{{Such singing cannot any please. / Save at the moment when you cease.}}" Absent.

b174 «اِنّی لَمُسْتَتَرٌ مِنْ عَیْنِ جیرانی…» - GLA: "Of a truth, I conceal myself from the eyes of my neighbors; but God knoweth my secret and public actions." Roman. - ROS: " {{verily, I may conceal myself from the sight of my neighbour, but God knows what is secret and what is [manifest]}}" - EAS: "{{True, I may be from neighbours' eyes concealed: / God knows my acts, both secret and revealed.}}" - ARN: one italic poem runs b174, b175 and b176 together — "{{…we shut the door of selfhood fast and close, that none may see / Faults and failings… God, th' All-knowing! hid and open are the same unto His eyes.}}" MERGED.

b201 «وَ عِنْدَ هُبُوبِ النّاشِراتِ عَلَی الْحِمیٰ…» - GLA: "The wind blowing over the plains causes the tender branches of the ban-tree to bend before it, but affects not the hard stone." Roman prose. - EAS: "{{Before the blast the balsams bend in the Arab's garden lone ; / Those tender shrubs their boughs incline; naught yields the hard firm stone.}}" — italic, with two footnotes, where the VERSE and COUPLET immediately above it on the same page are roman. - ARN: absent; he goes from "he is but an ass" to the Persian couplets.

b209 «اِنَّ مَعَ الْعُسْرِ یُسْراً» — Qur'an 94. - GLA: "Of a truth, joy succeeds sorrow" — roman, run into the sentence. - ROS: "{{Along with hardship there is ease; or, to sorrow succeeds joy:}}" — italic, and doubly rendered. - EAS: "' {{surely with calamity comes rejoicing.'}}174" - ARN: "…and that thou hast attained to such an eminence !" Absent.

b218 «زُرْنی غِبّاً، تَزْدَدْ حُبّاً» — framed and glossed by Sa'di himself. - GLA: "The Prophet said, "O Abu Horiera, come not every day, that so affection may increase."" — the Arabic is deleted and only Sa'di's Persian gloss survives. - ROS: "{{O Abu-Horairah ! let me alone every other day, that so affection may increase:}} that is, come not every day, that we may get more loving !" — both kept, the Arabic italic. - EAS: "" 0 Abu Hurairah ! {{visit me less often and thou wilt increase our friendship}} ;"180 that is, "Come not every day, that our attachment may be augmented."" — both kept, the Arabic italic and footnoted. - ARN: "Muhammad said : "Oh, brother ! come not every day, in order that friendship may increase."" — the Arabic deleted, only the gloss.

b236 «وَ قِنا رَبَّنا عَذابَ النّار» — Qur'an 2, the second hemistich of a Persian bayt, and rhyming with it. - GLA: "Take care how you connect yourself with a bad woman ; defend us, O Lord, from this fiery trial!" Roman prose. - ROS: "protect and guard us against a wicked inmate; save us, O Lord! and preserve us from the fiery, or hell, torture." Roman, inside the quotation marks Ross puts round all verse. - EAS: "From a vixen wife protect us well, / {{Save us, O God ! from the pains of hell."}}" — the Persian hemistich roman and the Arabic hemistich italic, inside one couplet. - ARN: "Beware how thou encounterest such !" and no more. Absent.

The typographic control, which the codings depend on

Before an italic can be read as a marking, the book's own use of italic has to be known. Measured over each hand's باب دوم window in the ABBYY layer:

hand italic share of the chapter what the italic is
GLA 1806 1.8%, 72 runs running heads, and three editorial asides — To return to my story:, (meaning Jacob). Gladwin's body text has no italic marking device at all, so his fifteen PLAIN cells are partly a fact about the book's typography. He did have quotation marks, and used them at exactly one locus.
ROS 1823 3.7%, 155 runs reserved; used on the Arabic, and on little else in the chapter
EAS 1852 5.6%, 241 runs running heads and the Arabic. His own COUPLET / STANZA / DISTICHS blocks are roman — checked on eleven consecutive verse blocks in tales I–V and on the two verse blocks immediately above b201 — so italic in his verse is a dedicated signal
ARN 1899 24.6%, 400 runs all his verse, which A-gulistan-hands §2 already records. Italic therefore cannot mark anything in Arnold's verse. His marking device is a different one: he transliterates a foreign word in italic — ayat, waḳf, gilīm, tasbīh, sikr, iqbāl, makāmāt, gadā — and he does it to Persian words as readily as to Arabic ones, so it marks a foreign term and not a quotation

Supplement — the six loci in حکایات ۳۲–۴۸, partially coded

Not in the primary count; recorded so span D does not start from nothing.

locus GLA ROS EAS ARN
b255 (pomegranate bayt) — — IT DROPPED
b260 (the cup-bearer bayt) PLAIN IT + QU IT DROPPED
b302 (Qur'an 2:44) — — — outside his abridgement
b314 (Qur'an 25:72) — — — outside his abridgement
b315 — — — outside his abridgement
b346 (Qur'an 31:15, framed) — — IT + QU + footnote, and "it is in accordance with the Kur'an. God most High has said" outside his abridgement

Completed and corrected 2026-09-02 (S240) by E-20260902-arabic-function. All 24 cells of these six loci are now coded in workshop/experiments/E-20260902-arabic-function/coding.md. Four entries above are wrong: Arnold's b302, b314, b315 and b346 are not outside his abridgement — he covers حکایات ۳۸, ۳۹ and ۴۳ in full, and the correct codes are DROPPED, DROPPED, DROPPED, QU. These cells were never counted in any rate published by RS-20260831, so no figure there changes. Where this table did code a cell, the later independent coding agrees with it, four of four.