The Measurable Logic of Qur'anic Word-Choice

Approach

A semantic reading, not a numeric one

We treat the Qur'an as a network of roughly 1,650 triliteral roots. For any pair of near-synonyms, we measure the company each keeps — which roots accompany one word far more than the other — weighting co-occurrence so that ubiquitous function-roots (الله, قال, كان) fall away and only the discriminating signal remains. That residue is the classical ʿilla: the operative reason a word is chosen over its synonym.

A pattern is admitted only after three independent gates:

• Exegetical agreement. The pattern must be echoed in the classical tafsīr, queried neutrally — we ask what the difference is, not whether our guess is right.
• Split-half stability. It must hold in both the Meccan and Medinan portions; a coincidence of one half does not survive the other.
• Adversarial refutation. A single counter-verse retires the claim. We hunt for it before we trust the pattern.

Nothing is generated. Every statement is extractive and traceable to its verse and its source — abstention over conjecture.

Finding · Word Choice

Each word carries a use-logic its synonym does not

The clearest case is the root فوز. Across all twenty-seven occurrences, it never denotes worldly gain — only ultimate, otherworldly success: Paradise, deliverance from the Fire. Its near-synonym كسب ("earning") carries the opposite freight — consequence, sin, reckoning.

The distinction between خَشْية and خَوْف — both rendered "fear" — is just as sharp. خشية keeps company with the unseen, reverent awe, and worship; خوف with tangible dread — hunger, grief, punishment. The commentators reached the same conclusion by close reading; here it falls out of the distribution, and the tafsīr confirms it verbatim: "one who fears God while He is unseen."

The same method resolves a family of pairs:

Finding · Method as Signature

Every commentator's method is itself measurable

The way a mufassir reasons leaves a quantitative signature across his commentary. Al-Ṭabarī marshals narrations by chain then adjudicates ("and the soundest of the views…"); al-Qurṭubī extracts legal questions; Ibn ʿĀshūr reads rhetoric and inter-verse coherence; al-Saʿdī explains plainly and directly. Rendered together for a single verse, these methods form a virtual scholarly council — one verse, read through four disciplined minds at once.

Finding · Structure

A single organic fabric — and its connective tissue is meaningful

Modeled as a graph — roots as nodes, two roots sharing a verse as an edge — the reported average shortest path between root pairs is about 2.15 steps, with high local clustering in the comparison used. This is a network average: it does not mean every pair is two steps apart, or that the words are adjacent in the verse. We then asked: which roots connect distant parts of the graph?

The bridging roots are messengership, revelation, calling, humanity, and resurrection — the very axis of the relation between God and the human being.

The prominent bridging roots invite a thematic reading around revelation and return. That interpretation goes beyond the edge count alone and is separate from the rejected mirror-symmetry hypothesis about opposite terms.

Position

Where this sits in the computational literature

Prior computational work on the Qur'an has been valuable but partial: some annotates morphology, some measures neural similarity, and a recent line reduces each word to a numeric letter-sum. What distinguishes this program is its unit — the meaning-bearing triliteral root — and its validation: agreement with the exegetical tradition, stability across revelation halves, and adversarial testing.