EXP-0007 — Repetition profile: Qur'an vs classical Arabic comparison texts
completed Deterministic computation
Question
How does the Qur'an's exact-repetition profile compare, at matched length, with a canonical hadith collection and a classical poetry diwan?
Hypothesis
Descriptive calibration; no directional hypothesis and no significance test (no defensible null model exists across different real texts). Declared design: matched-length deterministic prefixes, continuous token streams for all texts alike, letters-basic-v1 representation.
Counting policy
Comparison texts: openiti-body-v1 then letters-basic-v1. Qur'an: letters-basic-v1 tokens concatenated in canonical order as a continuous stream (differs from EXP-0001's within-ayah census by design, stated). Qur'anic quotations inside Bukhari are kept. Matched length = min(text, Qur'an) tokens, deterministic prefixes.
Results — repetition at matched length, per comparison text
Qur’an stream: 77,429 tokens. Each comparison uses deterministic prefixes truncated to the shorter text.
Sahih al-Bukhari (hadith collection, d. 256 AH) (593,471 tokens; matched at 77,429)
| Sequence length | Qur’an: distinct repeated (occurrences) | Sahih al-Bukhari: distinct repeated (occurrences) |
|---|---|---|
| 3 words | 6,339 (18,706) | 7,382 (36,561) |
| 4 words | 4,059 (10,560) | 6,428 (27,833) |
| 5 words | 2,612 (6,309) | 5,679 (21,697) |
| 6 words | 1,711 (3,956) | 5,001 (16,786) |
| 8 words | 796 (1,753) | 3,467 (9,372) |
| 10 words | 382 (835) | 2,275 (5,525) |
| 12 words | 184 (410) | 1,520 (3,484) |
Longest repeat in this text (25 words, 3×): حدثنا عبد الله بن يوسف قال أخبرنا مالك عن أبي الزناد عن الأعرج عن أبي هريرة أن رسول الله صلى الله عليه وسلم قال إذا
Diwan al-Mutanabbi (poetry, d. 354 AH) (8,486 tokens; matched at 8,486)
| Sequence length | Qur’an: distinct repeated (occurrences) | Diwan al-Mutanabbi: distinct repeated (occurrences) |
|---|---|---|
| 3 words | 562 (1,427) | 35 (133) |
| 4 words | 354 (806) | 13 (26) |
| 5 words | 217 (476) | 4 (8) |
| 6 words | 137 (291) | 1 (2) |
| 8 words | 60 (124) | 0 (0) |
| 10 words | 35 (70) | 0 (0) |
| 12 words | 21 (42) | 0 (0) |
Longest repeat in this text (6 words, 2×): صعدت إلى ذا مال ذا رهبا
Inference scope
Descriptive calibration only. These texts differ in genre, era, length, and transmission history; no statistic here supports a causal or evaluative claim about any text. Qur'anic quotations inside Sahih al-Bukhari are included (the book as it stands). Matched-length prefixes control for stream length only, nothing else.
Reproducibility record
- Corpus
- quran:hafs-kufan:v1
- Method version
- crosstext-repetition-profile-v1
- Results sha256
- 84d10935e160bbc87190ffd307dfeac5…
- Environment
- {"platform":"Linux-6.18.5-x86_64-with-glibc2.39","python":"3.11.15"}
- Ran
- 2026-07-31T03:36:47Z → 2026-07-31T03:36:52Z
Re-run with python3 services/research-worker/scripts/run_experiment.py data/research/specs/EXP-0007.json. Input checksums are recorded in the result file.