Skip to the content
Mashrabaمشربة

About this text: The Forty Hadith

الأربعون النووية

al-Nawawī (631–676/1233–1277)

Al-Nawawī's forty-two hadith on the foundations of the religion, each with the collections he names.

43 passages · ~35 min · 14 Quran quotations

3,356 words

  • Published translation: none (the English is AI, marked as such)
  • Quran quotations: 14

Not yet: audio, commentaries.

familiar vocabulary

Lexical difficulty, from the words themselves: how many roots fall outside the 1,000 most frequent in the Quran, how long the words are, and how much of the vowelling had to be supplied by AI. It says how unfamiliar the vocabulary is, not how deep the ideas are.

Rare roots 7% · mean word length 4.1 letters · AI vowelling on 99% of words

9 layers, each with its licence

The original

OpenITI digital text (ara1)OpenITI corpus · release 2025.1.9UNVERIFIED
By
OpenITI corpus (digital text, from al-Maktaba al-Shāmila 12836) · release 2025.1.9
Source:
OpenITI 0676Nawawi.ArbacunaNawawiyya.Shamela0012836-ara1 (its header names Dār al-Minhāj, Beirut, 1430/2009); unvowelled · link ↗
In the reader:
In full, 43 passages (preface and hadith 1–42), 30 corrections recorded
Licence:
OpenITI release (norm CC BY-NC-SA 4.0); the Arabic text of a 13th-century work; the 2009 print's own apparatus not taken
Notes
Corrections each name their witness (ara2, Wikisource, or Muslim 1015, which is in Ṣaḥīḥ Muslim and carries no separate grade on its sunnah.com page, checked 11 Nov 2026): data/sources/arbacun-nawawiyya-base.json → changes.
OpenITI digital text (ara2)OpenITI corpus · release 2025.1.9UNVERIFIED
By
OpenITI corpus (digital text of al-Sharbīnī's edition with Ibn Rajab's additions) · release 2025.1.9
Source:
OpenITI 0676Nawawi.ArbacunaNawawiyya.Sham19Y0147927-ara2 (hadith 1–42 taken) · link ↗
In the reader:
Readings cited at each correction
Licence:
OpenITI release (norm CC BY-NC-SA 4.0)
Arabic Wikisourcear.wikisource contributors · read 2026-11-10Open licence
By
ar.wikisource contributors (digital text) · read 2026-11-10
Source:
«الأربعون النووية», ar.wikisource · link ↗
In the reader:
Readings cited at each correction
Licence:
CC BY-SA 4.0

Published translations

Matthews' English Mishkāt (1809–1810)A. N. Matthews · 1809–1810Public domain
By
A. N. Matthews (translator) · 1809–1810
Source:
A. N. Matthews, Mishcát-ul-Másábíh, or a Collection of the Most Authentic Traditions, 2 vols (Calcutta, 1809–1810); BSB scans on archive.org · link ↗
In the reader:
Compared with our English where the Mishkāt has the same hadith (translation triangulation)
Licence:
Public domain (1809–1810)
Notes
Translates the Mishkāt's wording, which sometimes differs from Nawawī's.

AI layers

Vowels supplied by AIClaude (AI) · 2026-11-10Our work
By
Claude (AI) (vowelling) · 2026-11-10
In the reader:
Every vowel mark (the base is unvowelled), shown as AI in the reader
Licence:
Ours; marked AI
Made by:
Claude (exact version not recorded), 2026-11-10 (interactive session; scripts/hadith/arbacun-work-*.txt)
Notes
Consonants checked letter for letter against the base (scripts/hadith/arbacun_check.py). Not yet compared with a vowelled print.
English translationClaude (AI) · 2026-11-10Our work
By
Claude (AI) (translator) · 2026-11-10
In the reader:
Under each passage
Licence:
Ours; marked AI
Made by:
Claude (exact version not recorded), 2026-11-10 (interactive session; scripts/hadith/arbacun-work-*.txt → app/src/content/ai/arbacun-nawawiyya.json)
Notes
Not a published translation; not scholar-reviewed. Compared with Matthews (1809–10) where the Mishkāt has the hadith.
Word glossesClaude (AI) · 2026-11-10Our work
By
Claude (AI) (glosses) · 2026-11-10
In the reader:
Every word glossed, in the word card
Licence:
Ours; marked AI
Made by:
Claude (exact version not recorded), 2026-11-10 (made with the translation (scripts/hadith/arbacun-work-*.txt))

Automatic analysis

Word analysis: CAMeL ToolsCAMeL Tools 1.6 (CAMeL Lab, NYU Abu Dhabi), scripts/fusus/analyze.py · 2026-11-11Open licence
By
CAMeL Tools 1.6 (CAMeL Lab, NYU Abu Dhabi) (morphological analyser (calima-msa-r13 + MLE)); scripts/fusus/analyze.py (our pipeline) · 2026-11-11
Source:
CAMeL Tools 1.6, calima-msa-r13 with the MLE disambiguator, constrained by the text's vowelling; runs of Qurʾān from the Quranic Arabic Corpus; hand corrections with their sources in scripts/prayer_word_overrides.json (scripts/hadith/arbacun_build.py) · link ↗
In the reader:
Every word: root, pattern, grammar
Licence:
Code MIT; calima-msa-r13 database GPL v2 (used offline to produce our analysis)
Notes
191 words listed for a person to check (data/sources/arbacun-nawawiyya-review.txt), almost all vowel conflicts.
Quran quotations (Tanzil)Tanzil Project, scripts/quran_refs.py · 2026-11-11Open licence
By
Tanzil Project (Uthmani Quran text); scripts/quran_refs.py (automatic skeleton match) · 2026-11-11
Source:
Automatic skeleton match against the Tanzil Quran; no AI · link ↗
In the reader:
14 Quran quotations marked in the text
Licence:
Tanzil: free to use with attribution and a link to tanzil.net; the text may not be modified

✦ marks AI that no person has reviewed yet; ✧ marks AI reviewed by a named person. Copyrighted works are cited, never shown.

Our working notes on this edition: status, witnesses, the decisions taken and why, and what AI supplied. This is a working edition, not an authority.Show the notes
Work / author / datesal-Arbaʿūn al-Nawawiyya (الأربعون النووية), "the Forty", by Yaḥyā ibn Sharaf al-Nawawī (631–676/1233–1277). 42 hadith despite the title, with Nawawī's preface and his own source line after each.
Part addedWhole text: preface (passage 0) and hadith 1–42 (passage n = hadith n). 3,356 words.
Reader route/read/arbacun-nawawiyya · about /read/arbacun-nawawiyya/about
Library groupḤadīth (new, 11 Nov 2026; lib/catalogue-kinds.ts). Principle: a text whose matter is the Prophet's reported words, read as such, not as one author's teaching. Proposed later members: al-Tirmidhī's Shamāʾil (selections).
Statusworking edition, established 10–11 Nov 2026 by Claude (AI) · reviewed by: —

1. Witnesses

SiglumWitnessKindVowelled?LicenceLocation
ara1OpenITI release 2025.1.9, 0676Nawawi.ArbacunaNawawiyya.Shamela0012836-ara1 (Shamela 12836; header names Dār al-Minhāj, Beirut, 1430/2009)digital (base)noOpenITI: open; the 2009 print's own apparatus not takenOpenITI 0700AH repository
ara2OpenITI 0676Nawawi.ArbacunaNawawiyya.Sham19Y0147927-ara2 (al-Sharbīnī's edition with Ibn Rajab's additions; hadith 1–42 taken)digitalnoopenOpenITI 0700AH repository
WSar.wikisource «الأربعون النووية», read in a browser 10 Nov 2026digitalnoCC BY-SA 4.0https://ar.wikisource.org/wiki/الأربعون_النووية
MA. N. Matthews, Mishcát-ul-Másábíh (Calcutta 1809–1810), Englishtranslation (of the Mishkāt's wording)n/apublic domainarchive.org 10219219bsb, 10219220bsb (OCR hashes in scripts/fetch_sources.sh)

2. Base and why

ara1 is the fuller and cleaner of the two OpenITI files and carries Nawawī's source lines in full; ara2 and WS control it at every doubtful letter.

3. Collation

  • scripts/hadith/arbacun_parse.py splits both OpenITI files into passages; arbacun_base.py applies the corrections; arbacun_check.py checks the vowelled work files letter for letter against the base (43 passages, 0 problems).
  • Corrections to ara1: 30, each with its witness, in data/sources/arbacun-nawawiyya-base.json → changes (typos, spacing, hamzat al-waṣl, two omissions restored from ara2 and WS: ومشربه حرام in hadith 10, «بن عاصم» in hadith 23).

4. Decisions (substantive)

WhereReading adoptedRejectedReasonBy
10ومطعمه حرام ومشربه حرام وملبسه حرامara1 without ومشربه حرامara2, WSAI (local session)
23الحارث بن عاصم الأشعريالحارث بن الأشعريara2, WSAI
35إلى صدرهإلى صدورara2, WSAI
37كتبها الله عندهعنهara2; the word Nawawī explainsAI

5. Quotations verified

scripts/quran_refs.py arbacun-nawawiyya: 14 Qurʾān quotations found against Tanzil (1:1–2, 38:65, 4:136, 14:36, 23:51, 2:172, 65:2 ×3, 94:6, 32:16, 54:48). Runs of five or more Qurʾānic words take the Quranic Arabic Corpus' morphology (8 runs, passages 0, 2, 10, 15).

6. Vowelling

  • Source of vowels: all AI (Claude, local session 10 Nov 2026): the base is unvowelled. vowelMask marks every vowel sign "a", so the reader shows them as AI.
  • Checked: consonants identical to the base (script). Not compared with a printed vowelled edition: to do (the Dār al-Minhāj print is copyright; a pre-1950 vowelled print should be found).

7. Segmentation and word IDs

43 passages: 0 = preface, n = hadith n (narrator, text, Nawawī's source line, and his short glosses in 22, 37). Word ids nw:<passage>:<token>.

8. Analysis layer

CAMeL Tools 1.6 (calima-msa-r13 + MLE), constrained by the vowelling (scripts/fusus/analyze.py), run in a cloud session 11 Nov 2026 (CAMeL installed there without its torch extras). 3,356 words; 2,429 with a root; 178 vowel conflicts at the first pass. Hand corrections in scripts/prayer_word_overrides.json → arbacun-nawawiyya, each with its source opened on disk:

  • _by_lemma: 25 lemmas whose weak or hamzated root CAMeL left as '#' (رضي, نبي, عند, دنيا, قيامة, سماء, بين, سيئ, تقوى …), root set from the Quranic Arabic Corpus root list with its count.
  • word ids: 7 personal names the analyser read as nouns or verbs (Aws "lynx", Jundub "grasshopper", Wābiṣa "and I look", Thaʿlaba, Jurthūm, Ruqayya, al-Ṣābūnī); أفتاك (Form IV, not فتأ "cease"); حاك في الصدر (root حيك, Maqāyīs line 7907, not حوك "weave"); اقتداء (root قدو).
  • Left for a person: 191 words in data/sources/arbacun-nawawiyya-review.txt, almost all vowel conflicts between our (AI) vowelling and the analyser's best candidate. The reader shows its usual caution on those words.
  • 15 roots have no root page yet (scripts/build_data.py needs Lane and Lisān, not in a bare clone): بضع ثيب حصد حفو حنبل حيك خطو ذرف زمر سبط مضغ ملو نجذ نهك وبص. Run build_data.py on the owner's machine.

9. Commentaries aligned

None yet. Classic commentaries (Ibn Daqīq al-ʿĪd; Ibn Rajab, Jāmiʿ al-ʿulūm wa-l-ḥikam) are in OpenITI and are the next ring.

10. Translation

AI (Claude, local session 10 Nov 2026), one gloss per Arabic word; app/src/content/ai/arbacun-nawawiyya.json, labelled AI.

10b. Translation triangulation

  • No public-domain English of the Forty itself was found. Ibrahim and Johnson-Davies (1976) and Clarke (1998) are copyrighted (cite only) and were not held.
  • Compared: Matthews' public-domain Mishkāt (1809–10), found for 30 of 42 hadith (not in it: 9, 12, 16, 17, 18, 19, 22, 25, 28, 32, 39, 41). app/src/content/triangulation/arbacun-nawawiyya.json, made by an AI agent in the session from the archive.org OCR, not by triangulate.py (paid headless calls).
  • 39 divergences recorded; 6 shown to the reader as "translators differ": 2 رَبَّتَهَا (mistress / master), 7 النصيحة and أئمة المسلمين, 13 لا يؤمن (truly / perfectly), 23 الطُّهور and الصبر. The rest are the Mishkāt's different wording or term choices, kept in the file.
  • AI-English bug found and fixed: hadith 6 lacked Nawawī's source line in English.
  • Page references to Matthews are partial (OCR running heads); the OCR of hadith 4 is garbled.
  • sunnah.com (its English is the Ibrahim and Johnson-Davies translation) could not be opened from the cloud session (Cloudflare check).

11. Sources and grading of each hadith

  • Each heading carries Nawawī's own source line (his words, from the text), translated.
  • Modern collection numbers and gradings: to be checked. sunnah.com and dorar.net refused the cloud session (Cloudflare challenge, 11 Nov 2026). Nothing was filled from memory. scripts/hadith/arbacun-sources.json holds the per-hadith record; fill modern.refs only from an opened page.

12. Open questions / to review

  • Open sunnah.com (or dorar.net) for each hadith and record collection, number and grading with the grader named.
  • Review the 191 words in the review list; check the AI vowelling against a vowelled print.
  • Build root pages on the owner's machine (build_data.py).

13. Change log

  • 2026-11-10: base, corrections, AI vowelling, English and glosses (local session, WIP commit).
  • 2026-11-11: CAMeL analysis, overrides, reader data, Ḥadīth group, Qurʾān quotations, Matthews triangulation, record (cloud session).