One word. One meaning. Every occurrence.

The simple rule

Start with one stable semantic core for one defensible lexical unit. Then test it against every occurrence assigned to that unit. A difficult verse is not permission to quietly give the word a new meaning.

The root is a research envelope, like a folder for related forms; it is not automatically one word. If the Quran's evidence shows two genuinely different words inside that folder, their occurrences are separated and the contradiction is recorded. The core is not stretched until unrelated things appear to fit.

This is a test, not a claim that every decision is finished. Meanings stay as drafts until the evidence has been reviewed.

Why do this?

When one Arabic word is given many unrelated English meanings, the explanation can begin to control the verse. This method reverses that: the Quran's repeated use of the word controls the explanation.

The aim is not to force every root into one English word. It is to keep the lexical contribution stable and account openly for every change made by form, construction, or English grammar.

Why words matter

How it works

Six steps

1. Gather the whole family

   Collect every form, lemma, root link, occurrence, and relevant difference among the recitation texts kept side by side. The family is a search space, not yet a meaning claim.

2. Identify the lexical units

   Decide which occurrences are the same word. Split a genuine collision only when Quran-internal substitution gives a concrete contradiction.

3. Propose one stable core

   State the shortest clear English contribution and a testable definition. It must describe one operation, not list every observed rendering.

4. Compose the visible rendering

   Morphology, construction frames, and English grammar supply tense, voice, person, number, case, causation, and other changes without creating a new lexical core.

5. Try it everywhere

   Every analyzed word-part has one primary research owner. If the core fails in one real occurrence, the challenge is linked to that exact place and the decision remains open.

6. Preserve the decision trail

   Selected, rejected, narrower, form-specific, contextual, and deferred candidates remain linked to exact Quran evidence so the conclusion can be checked and replayed.

In the glossary

Where other meanings fail

Some entries show a Quranic occurrence where a familiar alternative stops fitting. You can compare that alternative with the research rendering and see exactly why it fails as the stable core there. The alternative may still be useful as a narrower, form-specific, or contextual expression. This section appears only when a concrete counterexample has been recorded.

To find them: open Glossary, choose Failure examples, then open an entry.

Open failure examples

One core can change shape

The core guard can appear as guarded, guard yourselves, or the self-guarding. Those changes are composed from morphology and English grammar. They are not unrelated dictionary meanings.

Accepted

An accepted meaning has been explicitly reviewed at a defined lexical scope. Automation cannot accept it or silently rewrite it. Its core, occurrence set, rules, and evidence remain protected.

Draft

A draft is a traceable proposal, not a hidden gap and not an accepted answer. Coverage is not acceptance: the corpus can give every segment a primary research owner while some semantic decisions still await review.

What the graph now records

The graph keeps six layers distinct: the family folder, the actual word being tested, its stable contribution, what its Arabic form adds, what its phrase construction adds, and how English displays the result. In the graph these are the root family, semantic unit, semantic core, morphological operator, construction frame, and rendering rule.

Two entries do not become the same meaning merely because they use the same English word; they must link to the same stored core. The graph also preserves serious candidates, why each was selected or rejected, exact evidence, unresolved questions, failure examples, and where each decision came from. All of this is versioned so it can be rebuilt without allowing automation to change accepted research.

What decides a meaning?

The Quran's own usage comes first. Every proposed meaning must face all of its Quranic occurrences.

Grammar, morphology, and construction show how the stable contribution is composed in each occurrence.

Eight available recitation texts are aligned verse by verse to preserve wording and segmentation evidence. A reading difference can expose a question, but a reading does not decide the English meaning by authority.

What does not decide it?

Hadeeth and tafseer are not used as authorities for word meaning in this project.

A familiar translation is not accepted simply because it is familiar. It must still fit the Quran's evidence.

An LLM may organize, compare, and challenge graph evidence. It cannot accept a meaning or become an authority for one.

How a verse rendering is made

• Begin with the lexical unit's stable core, not a context-selected dictionary sense.
• Apply morphology and construction rules before ordinary English grammar.
• Let person, number, tense, voice, and syntactic role change the English form without changing the core.
• Use parentheses for contextual clarification, not to smuggle in a second lexical meaning.
• Keep the words and particles that the Arabic actually gives us.
• Rearrange only where clear English requires it.
• Add as few extra words as possible.
• Keep the word-by-word layer available, but hidden until the reader asks for it.

Precision before polish

Some renderings preserve an Arabic construction even when it sounds unusual in English. Quran 2:6 says: “Indeed, for those who deny, it is level whether you warn them or do not warn them; they do not believe.” The phrase “it is level” keeps سَوَاءٌ visible. Replacing it with “it makes no difference” may sound smoother, but it translates an explanation of the phrase instead of the phrase itself.

Many translations are fluent, moving, and inspirational. They can also hide a particle, repetition, emphasis, or the way an Arabic sentence is built, and that can create confusion about what the text actually says. This project gives precision priority over elegance: keep the Arabic phrasing as close and visible as intelligible English allows, then explain what is unusual.

The aim is not awkward English for its own sake. The rendering should remain readable. But when polish and precision pull in different directions, precision comes first.

Where the rule needs care

A shared root does not prove that every related form is the same lexical unit. Real homographs, names, and lexicalized branches must be separated and explained. A root may also have an abstract research nucleus that is useful without being the English translation of every derivative.

A repeated multiword construction may need its own rendering, but it does not silently replace the stable contribution of each word. A proposed core must be a testable relation, not an umbrella definition that simply lists every result.

The aim is unity where the evidence supports it, not unity forced at the expense of accuracy.

What the public site shows

• The reader, Quran search, word breakdown, and glossary use a read-only published snapshot. Ordinary browsing does not call an LLM.
• Graph exploration and editable research remain gated so the public snapshot cannot alter the research database.
• The reasoning tier may question the graph with the configured high-reasoning model, but its answers remain research assistance rather than accepted meaning.
• Feedback is saved in a private structured inbox with its page or verse reference. No email is sent.

Research and production acknowledgement

Most of this project's research, development, and editing was carried out using ChatGPT's 5.6 Sol with ultra reasoning, under Feroz Peeradina's direction and review. The model served as a research and production tool, not as an authority for the Quran's meaning; published lexical decisions remain open to inspection against the Quran-internal evidence.