Executive Summary

Why it matters

Romans is the theological backbone of the New Testament, and Urdu carries a risk profile structurally unlike any other language in this batch: its central threats are not syncretism (a fluent word smuggling in an unwanted meaning) but direct negation (specific, named Qur’anic verses explicitly denying specific Romans claims). Qur’an 4:157 states plainly that Jesus was not crucified; Qur’an 112:3 states plainly that God “begets not, nor is He begotten”; and shirk, the sin of associating a partner with God that “Son of God” language directly risks being heard as, is named in the Qur’an as the one sin Allah does not forgive (Qur’an 4:48, 4:116). This Language Package treats these as named, citable counter-claims to state clearly and teach with pastoral care, not as generic cultural background to translate around.

Key findings

The registry tracks 40 doctrines across Romans 1-16; 30 require mandatory human theologian review before any translated segment ships (7 Critical, 23 High).
Sonship of Christ is the single most doctrinally explosive doctrine in this Language Package: no choice of Urdu wording resolves its collision with tawhid, because the underlying claim itself, not the word used for it, is what Qur’an 112:3 and the shirk doctrine contest.
Salvation (نجات) and the Holy Spirit (روح القدس) are both shared-vocabulary risks structurally similar to Konkani’s देव and Sanskrit’s आत्मा problems elsewhere in this pipeline: the same word appears in both scripture traditions with a different underlying doctrine beneath it, so no substitute word closes the gap — only explicit contrastive teaching does.
Resurrection is the one Critical doctrine in this Language Package where Islam does NOT deny the underlying category (a general bodily resurrection, qiyamat, is affirmed); the real risk sits upstream, at the crucifixion Islam denies, not at resurrection terminology itself — a genuinely distinct risk shape worth not confusing with the other six Critical doctrines.
Urdu has a real, established Christian Bible translation tradition (unlike Maithili) to draw on, but that tradition’s highest-risk terms sit closer to explicit, named theological negation than any other language’s in this batch.

Risks

Direct textual negation: Qur’an 4:157’s crucifixion denial is the single most acute scriptural contradiction of a core Romans claim (Christ’s atoning death) anywhere in this 5-language batch.
Shirk-collision risk: “Son of God” language cannot be defused by word choice; it requires explicit theological framing every time it appears.
Shared-term risk without a clean substitute: نجات (salvation) and روح القدس (Holy Spirit) are both established Christian Urdu Bible terms that are also live Quranic/Islamic-theological terms with different content — avoiding them is not an option, since no alternative word would be recognized at all.
Default-name risk: unlike Hindi, where ईसा is a minority option, عیسیٰ (Isa) is the majority default name for Jesus for most Urdu readers, meaning یسوع (Yasu’) requires active introduction rather than simply being available as the obviously correct choice.

Opportunities

Urdu shares substantial genuinely safe theological vocabulary with Islamic theology (ایمان for faith-as-trust once its object is specified, شکر for thanksgiving, قدرت for God’s power, عہد for covenant), which this Language Package treats as real common ground rather than universally suspect.
An established Urdu Christian Bible tradition (unlike Maithili’s thin precedent) gives this Language Package settled proper names and core terms to build from rather than requiring first-principles invention throughout.

Recommended actions

Route every Critical and High risk segment (30 of 40 doctrines) through human theologian review before publication; do not allow automated-only review to touch these terms.
Brief every reviewer specifically on the distinction between this Language Package’s negation-risk doctrines (crucifixion, sonship, deity) and its shared-term risk doctrines (salvation, Holy Spirit) — the two require different teaching strategies, not the same generic caution.
Reuse this Language Package’s translation_memory.json for every Romans lesson in Urdu rather than re-deriving terms per document, per the two-phase pipeline design.