Quran Pattern Labمختبر الأنماط القرآنيةResearch Preview v0.1

Limitations & responsible claims

Each item is a reason a result may be wrong, over-stated or under-stated.

Unresolved in this release

  1. The linguistic criterion of the 77,407 convention (interrogative vs negative «ما») is not independently verified; its 25 positions come from the program's list.
  2. No candidate pattern has been reviewed by a linguist, Quran scholar or statistician yet.
  3. The concept lexicon is an AI-proposed draft and unreviewed.
  4. The Persian and Turkish versions of this site are preliminary interface translations, not yet reviewed by human translators.

Data

  1. One transmission and one digital text source: Hafs 'an 'Asim via Tanzil 1.1.
  2. No morphology: no roots, lemmas or parts of speech, so homographs (e.g. «البر») are not separated.
  3. Concept counts use the Simple text; the Uthmani–Simple gap cannot be bridged by folding (e.g. الرحمان / الرحمن).

Statistics

  1. Popular-claim baselines are coarse (e.g. day = 365: few forms in the band).
  2. The muqatta'at null is permissive and calibrated empirically against control letters; calibration resolution is limited (smallest calibrated p ≈ 0.018).
  3. The binomial concentration model assumes independence, so it is anti-conservative; concentration usually reflects the surah's topic.
  4. Exploratory only, not preregistered; the two popular claims were known in advance, so they can never be confirmatory here.
  5. The global BH family mixes families of different power and dependence.

Cross-check

  1. Representative coverage, not full validation: 35 Simple and 85 Uthmani queries on one machine.
  2. Uthmani explanations rest on declared emulation hypotheses, some formed after inspecting the program's exported text for two words.
  3. 4 queries are not supported (roots and parts of speech) because there is no morphology data.

Claims policy