# Contributing a structural hypothesis This site's whole method is: order the evidence, let the reader draw the conclusion. That applies to contributions too. This page is for submitting your own proposed sectional/structural reading of a surah — especially one of the 84 surahs Dr. Khan never published an outline for (his 2013 book, *An Exercise in Understanding the Qur'an*, covers only surahs 85–114). If you just want to try this on the live site first without opening a PR, use the [discovery worksheet](https://divinediscourses.org/read) on the Read page — it saves privately in your own browser and exports to Markdown, which you can paste into a submission. ## What a submission is, and isn't - It is **your own reading**, built from this site's bundled open data (root frequencies, co-occurrence, the `discursive-pivots.json` boundary markers, the `symmetry-test.json` proxy) or from a real, page-cited published source. - It is **not** a reproduction of Dr. Khan's, Islahi's, or any other living/copyrighted scholar's published outline. Do not paraphrase or summarize a copyrighted outline and submit it as your own reading — see [NOTICE.md](NOTICE.md) for what's already licensed in. - It is **not** presented as settled. Every submission is merged with a ○ Pending or ~ Nuanced badge, never ● Verified — "Verified" on this site means computed from a cited primary source, and a structural hypothesis is an interpretation, not a computation. - It is **your reading, not a tool's**. Software assistance may help locate a candidate source, check code, or flag unclear prose, and that use must be disclosed in the PR. It may not originate the proposed theme, sectional divisions, interpretation, translation, root meaning, citation, or scholarly attribution. Rewrite from your own reading of the source; assistant output is never submitted as authoritative content. ## Submission template Open a PR (or a "New structural hypothesis submission" issue if you'd rather discuss it first) with: ``` Surah: [number and name] Proposed theme: [one or two sentences] Sectional divisions: - [verse range]: [what this section is doing] - [verse range]: [what this section is doing] ... Supporting evidence: [which roots/frequencies/boundary markers you looked at, and what you found — link to the relevant roots.html/patterns.html page where possible] Cross-references: [related verses elsewhere, if any — cite Mishkat or your own reading] Confidence: [draft / tentative / confident — your own rating] License: [CC0 or CC-BY — your choice; state which] ``` ## The other thing this project needs: translators Structural readings are the main ask, but not the only one. The site now carries Qur'an translations in 17 languages while every word of its own interface and explanatory prose is English. That gap is deliberate rather than neglected: this project does not machine-translate its own text, and it does not let an assistant originate a translation (see [docs/maintainer-guide.md](docs/maintainer-guide.md) §1, rule 3). Interface text arrives the same way a structural reading does — from a named person who takes responsibility for it. The first concrete piece is a short orientation note for native Arabic readers, explaining what the computed layers offer someone who does not need a translation. The exact English source text is specified in [docs/global-reach-plan.md](docs/global-reach-plan.md) §2.1 — a translator renders that paragraph and nothing more; it is orientation copy, not interpretation of the Qur'an, and it must not acquire commentary in translation. Translators are credited by name on the Credits page. If you would like to take it — for Arabic or for any of the site's translation languages (Bengali, Malay, Indonesian, French, Spanish, Urdu) — open an issue saying which language and how you would like to be credited. ## License Pick either **CC0** (public domain dedication) or **CC-BY** (attribution required) for your submitted text, and say which in your PR. Everything else in this repository not covered by a third-party license (see [NOTICE.md](NOTICE.md)) is MIT-licensed; your structural-hypothesis text is the one kind of contribution where you keep an explicit choice, since it's your own interpretive work, not a mechanical computation. ## Review checklist Before a submission is merged, it needs to pass: 1. **No copyright reproduction.** Nothing paraphrased from Khan's, Islahi's, or any other actively-copyrighted secondary work beyond a short (<15-word) fair-use quote with full citation. 2. **No tool-originated conclusions.** Candidate boundaries, themes, translations, glosses, citations, and interpretations originated by software assistance are not publishable. Deterministic outputs from repository scripts may be published only with their named inputs, method, and limitations. 3. **A verification label on every claim.** ● / ○ / ~ — never left implicit. A structural hypothesis is Pending or Nuanced by definition, not Verified. 4. **Recomputable from open sources.** Either the bundled Tanzil/Leeds/ Mishkat data, or a newly-added source with documented rights status — never invented figures. 5. **No new tracking, analytics, backend, or paywall logic.** This stays a static site with client-only persistence. 6. `node scripts/verify-site.mjs` and `node scripts/build-csp.mjs --check` both pass. ## Worked example (not an authoritative outline) Surah 112, al-Ikhlas, is four verses — short enough to check by hand against the bundled morphology (`data/morphology/112.json`) rather than just asserted: | Verse | Text (root gloss) | Roots present | |---|---|---| | 1 | *Say: He is God, the One* | q-w-l (say), a-l-h (God), a-H-d (one) | | 2 | *God, the Eternal Refuge* | a-l-h (God), S-m-d (eternal/self-sufficient) | | 3 | *He begets not, nor is He begotten* | w-l-d (beget) — twice, active then passive | | 4 | *Nor is there to Him any equivalent, one* | k-w-n (be), k-f-A (equivalent), a-H-d (one) | Two mechanically checkable observations follow directly from this table, with nothing added: - **a-H-d (one)** is the last root-bearing word of both verse 1 and verse 4 — the surah opens and closes on the same root, an envelope pattern. - **w-l-d (beget)** occurs twice within verse 3 alone, in its active and passive forms back to back — a repetition internal to a single verse. That's it — this is a candidate signal, unverified, not a finished outline, and not attributed to any named scholar's reading. A submission built the same way — from a table anyone can reproduce from the bundled data, plus your own stated interpretation of what the pattern suggests about the surah's structure — is exactly the shape this process is for.