How It Works
Divine Discourses follows a coherence-based method for reading the Qur'an, drawing on the work of Hamiduddin Farahi (1863–1930), Amin Ahsan Islahi (1904–1997), and Dr. Irfan Ahmad Khan (1931–2018).
The method
The coherence-based approach, developed by Hamiduddin Farahi and his student Amin Ahsan Islahi, treats the Qur'an as a unified text with internal structure. Each surah is read as a coherent whole with a central theme, and verses are understood in relation to the surah's structure and to parallel passages elsewhere in the text. This approach contrasts with atomistic reading that treats each verse in isolation.
Dr. Irfan Ahmad Khan (1931–2018) was a founding scholar of the Association of Quranic Understanding and carried forward this coherence-based method in his teaching and published works. His Reflections on the Qur'an (Islamic Foundation UK, 2005) demonstrates this approach on the first two chapters of the Qur'an, and his An Exercise in Understanding the Qur'an (Association of Quranic Understanding, 2013, ISBN 9781567447736) applies it systematically across the last thirty surahs.
Khan emphasized that the believer's task is to engage the text directly, not through an intermediary. He taught that every reader builds their own personal relationship with the Qur'an, and that this relationship requires patient attention to the text's own structure and cross-references. Divine Discourses gathers tools that support this kind of direct engagement, roots, words, patterns, frequency, and structural markers, with provenance records on research claims so readers can see exactly what was checked and trace sources independently.
Three depth levels
Every page on this site can be viewed at one of three depth levels. You choose your depth based on what you need in the moment.
The reading layer. Verse text, word-by-word meanings, translations, and audio. No morphology tables, no annotations. Use this when you want to read without distraction.
The analytical layer. Adds word-by-word morphology, root links, related verses, and chronological period distribution. Use this when you want to understand how words work and how themes recur.
The reference layer. Adds structural pattern notes and full provenance for every claim. Use this when you want to verify sources or understand where the data comes from.
Change depth anytime using the gear icon in the bottom right, or press 1, 2, or 3 on any page. Your selection is saved in your browser, never sent anywhere. No tracking, no cookies.
Evidence status
Research claims distinguish source inspection, computational reproduction, method dependence, and known limitations. The familiar dot remains a compact summary and citation control; it is not a claim that every dimension has been independently verified.
The Validation page is the canonical definition of these labels and of the process that assigns them. You can use its prompts to triangulate any Pending claim against outside tools and references of your choice, or to surface the nuance in any Nuanced claim.
Sources
Data on this site comes from named corpora, APIs, published scholarship, and primary-source references; provenance and licensing are documented on the Sources page.
- Quranic Arabic Corpus v0.4 (Language Research Group, University of Leeds, 2009–2017, GNU GPL) for word morphology, root tagging, and grammatical analysis. corpus.quran.com
- Quranic Universal Library (Tarteel AI / Quran.com Foundation, 2024) for structured verse data and corpus indexing. qul.tarteel.ai
- Quran.com Foundation Content API v4 (Quran.com Foundation, 2025) for translations, recitations, and metadata. api-docs.quran.foundation
- Mishkat Mutashābihāt corpus for verse-to-verse cross-references, drawn from 13 classical books of mutashābihāt scholarship. github.com/Alhassan777/Mishkat
Full citations for all sources, including access dates and version numbers, are on the Sources page.
The research and tooling behind this site may use software assistance, including language models, to locate candidate sources and check work. Everything published is verified by a person against the cited source; interpretations, translations, and glosses come only from the named human scholarship cited.
The text behind the numbers
Two limits of the data deserve plain statement. ○ First: the Arabic text rendered here (the Tanzil Uthmani text) and the Leeds tokenization behind every count follow the reading of Ḥafṣ ʿan ʿĀṣim with the Kufan verse numbering — the reading used in most printed muṣḥafs today. The Qur'an has other canonical readings (qirāʾāt) that differ in places, so every count on this site is relative to this one reading; none is a count of "the Qur'an" independent of it.
Second: the Leeds corpus's version 0.4 is its most recent release, and the project is no longer actively updated. Like any large annotation effort it contains imperfections, and assigning a root or lemma to a word — especially weak roots and hamza spellings — involves editorial judgment by its annotators. When this site marks a corpus-derived count Verified, that means the computation from the cited corpus is faithful and reproducible; the count inherits whatever imperfections the corpus itself carries.
Where the numbers come from
None of the counts on this site are typed in. The scripts that produce them ship in the repository alongside the corpus data, and every figure quoted in a page's prose is bound to the generated dataset, so a page cannot drift from the data behind it: an automated check fails the build if it does. Run any generator twice and the output is identical, which is what makes an independent recomputation possible. The Datasets page lists each dataset and the script that builds it.
One boundary is worth naming, because counting can look like claiming. The recurring-phrase index records that a sequence of words repeats and where. Whether such repetition amounts to formulaic composition in the oral-formulaic sense is a scholarly question, argued at length by Bannister (2014, on Sources); the site counts the recurrence and leaves the argument to the literature.
How themes are built
Each theme on the Themes page begins with a small set of triliteral roots whose core dictionary senses relate to it. That grouping is an editorial choice, labeled Nuanced wherever it appears. What follows is computed: the site counts where those roots occur, ranks the passages in which the theme's vocabulary clusters, and links them. A theme can live in verses that use none of its words, and any of those words can appear outside the theme, so a theme is a gateway into reading rather than an index of where an idea is present.
How this compares to other Qur'an tools
Many excellent Qur'an apps and websites exist, and this site links to several of them. The differences below are matters of design policy, stated so you can decide whether they matter for your own study — not claims that other tools are wrong to choose otherwise.
| Common elsewhere | Here |
|---|---|
| Built-in commentary telling you what a verse means | No commentary of the site's own. The site computes structure and cites published scholarship; what a verse means is left between you and the text. Khan's readings appear only as verbatim transcriptions from his published books, with citations. |
| An assistant that answers questions in its own words | A deterministic Ask box: pattern-matching that routes your question to the right tool. The site never answers in its own voice. |
| Counts and claims presented without provenance | Every research claim carries a badge — ● Verified, ○ Pending, ~ Nuanced — that decomposes into named evidence dimensions you can check yourself on the Validation page. |
| Accounts, sync, analytics, engagement mechanics | No accounts, no tracking, no streaks. Notes and preferences stay in your browser; the only network requests fetch the text, translations, and audio you ask for. |
| Verse-by-verse browsing as the primary frame | The surah as one discourse is the primary frame, following the coherence-based approach described above — with structural evidence (roots, recurring phrases, rhyme, cross-references) gathered per surah in the Dossier. |
A suggested first visit
If this is your first time using Divine Discourses, here is a walkthrough that touches each major feature.
- Open al-Fatihah 1:1 and read the basmala in multiple translations at Simple depth.
- Switch to Study depth (gear icon in bottom right) and reload the same verse to see word-by-word morphology with root links.
- Visit the Roots page and click any row to see how that root distributes across the Meccan and Medinan periods.
- Use Compare to overlay several roots at once — try faith (a-m-n), works (ʿ-m-l), and righteousness (ṣ-l-ḥ) together to see the verses where all three meet. Compare can also set passages, or whole surahs, side by side on their corpus fingerprints.
- Visit Patterns to see grammatical shift (iltifat) and recurring formulations (mutashabihat) at work in the text.
When you have completed the first visit, try the coherence-reading exercise on Surah al-Asr. This is the method this site honors, demonstrated live on the shortest meaningful surah.