Translating Without a Judge

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

Repository path: workshop/translations/maqamat-dimyat/collation.md · rendered 2026-09-09

Page metadata (front matter)
typenote
idmaqamat-dimyat-collation
statusfrozen
created2026-08-27
updated2026-08-27
linksworkshop/translations/maqamat-dimyat/source-arwikisource-maqama4.txt, workshop/translations/maqamat-hulwan/collation.md, workshop/translations/maqamat-sanaa/collation.md, workshop/translations/maqamat-dimyat/R53-v1/translation.md

Copy-text and collation — «المقامة الدمياطية»

The copy-text

ar.wikisource.org, «مقامات الحريري/المقامة الدمياطية», fetched 2026-08-27 via action=raw (HTTP 200, 10,086 bytes). Public domain: al-Ḥarīrī died 1122. The transcription is fully vocalised and punctuated at every saj' colon boundary, as the same site's «الصنعانية» and «الحلوانية» are — which is what makes the 182-colon segmentation in source-arwikisource-maqama4.txt a reading of the page and not a judgement of the lead's.

Stored as source-arwikisource-maqama4.txt, one record per line, ID<TAB>FORM<TAB>text. 182 prose cola and 14 verse bayts, 749 Arabic tokens (607 of them prose). Three <div> wrappers around the two poems and four lines of MediaWiki category markup were dropped; that is the only edit.

The collation, and why there is not one

Single-witness, on the same declared terms as Assemblies I and II. The four routes to a second Arabic witness that failed for «الصنعانية» on 2026-08-24 and for «الحلوانية» on 2026-08-26 — recorded in ../maqamat-sanaa/collation.md and ../maqamat-hulwan/collation.md — were not re-attempted, for the reason the second of those notes gives: a route that failed twice on this work in the same week is not a different route on a third maqāma of it.

The two nineteenth-century printings print the Arabic beside their English and neither was opened. The cost of that witness would be the study's independence, and here it would be larger still, because the arm's whole question is what those printings do.

Three doubtful readings, recorded before the rendering was made

The copy-text carries at least three places a printed edition would not have. They are listed here so that the rendering's choices at those places can be checked against something.

colon the copy-text the difficulty
p116 «حتى نأَنَ التّغابُنَ» نأَنَ is not construable; read as نأمَنَ, "so that we may be safe from"
p118 «فلِمَ أعُلّكَ وتُعلّني» the same verb on both sides, which breaks the pattern of p119–p121
p125 «وأيّ حُرّ رضيَ بخُطّةِ خسْفٍ?» a Latin question mark for the Arabic one, here and at p153; cosmetic, and the segmenter treats both

Nothing measured in ARM-balanced-period step 2 turns on any of the three. What is measured is the length and adjacency of English clauses; the Arabic enters as the count of cola the hands are rendering, and none of the three affects that count.