Validation Methods
Every claim can be triangulated through independent sources. The prompts below help you cross-examine a claim with AI research assistants (ChatGPT, Gemini, Grok, Perplexity). Chatbot agreement is a screening signal, not a verification: it tells you where to look, not what is true. A claim moves from ○ to ● only when it is confirmed against the primary source it cites. Where reputable sources genuinely disagree, the claim stays ~ and the disagreement is recorded.
How we verify: three worked examples
Each label carries a specific meaning. The examples below are drawn from published claims on this site so you can trace the reasoning end to end.
● Verified · top five roots
a-l-h (God, 2,851), q-w-l (say, 1,722), k-w-n (be, 1,390), r-b-b (Lord, 980), a-m-n (believe, 879).
Source trace. Leeds Quranic Arabic Corpus, Kais Dukes 2009–2017, GPL-licensed. Dictionary interface at corpus.quran.com. Full top-10 with per-root citations on Numbers and Roots. One primary source; independent triangulation possible via any Qur'anic concordance.
○ Pending · Farahi's Mufradat and Niẓām
Hamiduddin Farahi, Mufradat al-Qur'an and Niẓām al-Qur'an. Foundational to the coherence-based reading school.
What would move this to Verified. A second independent bibliographic citation with edition, publisher, and page numbers for the specific claims we draw from Farahi. The entry is listed on Sources under coherence-based reading; run Prompt 3 below (Islahi-school content) against ChatGPT and Gemini to locate the exact Farahi passage, then confirm the passage against a published edition before upgrading.
~ Nuanced · y-w-m (day)
405 occurrences by the Leeds corpus root count; broader counts including compound forms such as yawma-idhin reach 475.
Why it stays Nuanced. The 405/475 difference turns on whether compound forms are included as separate occurrences; both counting rules are defensible and both are cited. There is no single "correct" answer. The label surfaces the disagreement rather than hiding it. Detail on Numbers.
● Verified · hapax legomena
12,004 surface forms (out of 18,993 unique forms) occur exactly once in the Qur'an. 395 roots (out of 1,642) occur exactly once.
Source trace. Computed directly from the Leeds
Quranic Arabic Corpus v0.4 word-form and root tables, which are
bundled in this site's own data/morphology/ files —
so the claim is recomputable by anyone, from the data this site
ships, without trusting us. Published on
Numbers at Scholar depth.
● Verified · a two-source claim
547 roots appear only in Meccan surahs; 198 appear only in Medinan surahs.
Why two sources. This claim combines the Leeds corpus (which roots occur in which surahs) with the Cairo 1924 edition's Meccan/Medinan classification. The badge cites both; click it to see the pair. A nuance rides along: a handful of surahs are classified differently in other traditions, so the boundary counts can shift by a few under other schemes — the claim on Numbers carries that note beside it.
Triangulation prompts
Run any of the prompts below in ChatGPT, Gemini, Grok, or Perplexity to stress-test a claim. Treat the responses as leads, not proof: when two independent responses agree, follow their citations to the primary source and upgrade the label from Pending to Verified only after the source itself confirms the claim. When sources genuinely disagree, keep Nuanced and record the disagreement.
Prompt 1: Verify a root count
Using only verifiable academic sources (Quranic Arabic Corpus at corpus.quran.com, Tanzil text, Lane's Lexicon, Penrice's Dictionary of the Koran, or peer-reviewed Qur'anic studies journals), how many times does the triliteral root [INSERT ROOT, e.g., r-ḥ-m] appear in the Qur'an? Distinguish (a) root occurrences, (b) surface form occurrences if relevant, (c) any counting variation between sources. Cite each source. Do not estimate; if uncertain, say so.
Prompt 2: Verify a numerical claim
The following claim about the Qur'an circulates widely: [INSERT CLAIM]. Investigate with academic rigor. (1) What counting rule does it use? (2) What do alternative counting rules yield? (3) What is the academic position, citing peer-reviewed sources? (4) Is the claim broadly accepted, or are there nuances? Provide direct citations.
Prompt 3: Verify Islahi-school content
In Amin Ahsan Islahi's Tadabbur-i-Qur'an (Lahore: Faran Foundation), what is the stated 'amud (central axis) of Surah [INSERT NUMBER AND NAME]? Provide the Urdu phrasing if available, the English translation if accessible (e.g., Shehzad Saleem at al-Mawrid), and the volume and page reference.
Prompt 4: Verify chronology
For Surah [INSERT NUMBER], what revelation order is assigned by each of: (1) Cairo 1924, (2) Nöldeke-Schwally 1909, (3) Bell 1937, (4) Sadeghi-Goudarzi 2012? List each ordering's position number and qualifications. Cite the source for each.
Prompt 5: Verify iltifat
Does Qur'an [INSERT VERSE] contain an instance of iltifat (grammatical shift in person, number, addressee, tense, case, or structure)? If yes, describe and cite Abdel Haleem's 1992 BSOAS article or other peer-reviewed sources. If borderline, state that.