01 Executive Summary — Luke 1–24 (English → Arabic)

The bottom line: The Luke curriculum is translatable into doctrinally faithful Arabic using the Romans baseline’s established conventions, but it introduces a denser concentration of Critical-risk collision points than Romans — most acutely around the historical crucifixion, a personal Savior-title, and the golden-calf-idolater Qur’anic namesake hiding inside the word “Samaritan.” Every artifact in this package is built to carry that weight without diluting the text.

---

Why it matters

Luke is the Gospel most likely to be a Muslim-background reader’s first sustained encounter with the claim that Jesus actually died on a cross, that a condemned criminal can enter Paradise with zero deeds performed, and that “Savior” names a specific person rather than a generic mercy. These are not incidental details — they sit at the center of the assigned curriculum doctrines (Jesus as Savior for All, Repentance and Forgiveness, the Cost of Discipleship) and at the center of the sharpest fault line in Christian-Muslim engagement (Qur’an 4:157’s crucifixion denial). Getting the vocabulary right without the doctrine is not enough; every Critical/High term requires deliberate teaching scaffolding, not translation alone.

---

Key findings

• 109 enforced terms now populate assets/translation_memory.json: 47 inherited unchanged from the Romans baseline, 62 new to Luke.
• 60 of those 109 terms (55%) are Critical or High risk (23 Critical, 37 High) and require mandatory human theologian review at every occurrence.
• 35 doctrines are tracked in assets/doctrine_risk_registry.json: 13 Critical, 14 High, 6 Medium, 2 Low — 27 of 35 (77%) require human theologian review, 6 require native speaker review, and only 2 are automated-only.
• Luke compounds three risk patterns not present, or less acute, in Romans: (1) a two-chapter apologetic unit (crucifixion + resurrection, chs.23–24) that must never be reviewed independently; (2) an internal Son-of-God tension the book creates itself (Adam’s creaturely sonship at 3:38 vs. Christ’s ontological sonship elsewhere); (3) a genuinely non-obvious Qur’anic namesake collision (السامري = the golden-calf idolater of Qur’an 20:85-97) with no equivalent in Romans’ term set.
• Every chapter of Luke (1–24) has been swept for load-bearing vocabulary across five prior Phase 1 artifacts; no chapter was silently omitted, including chapters (8, 13, 14, 16, 20) that primarily reuse rather than introduce vocabulary.

---

Risks

• Highest-stakes single verse: Luke 23:43 (Paradise). A dying criminal’s immediate entry into الفردوس — the highest deeds-attained tier of Islamic Paradise — on trust alone, with zero deeds possible, is the sharpest grace-versus-merit contrast in the book. A hedged or softened translation would blunt the curriculum’s central apologetic asset.
• Highest-frequency risk: الصليب (the Cross). Appears in teaching contexts (9:23, 14:27) as well as narrative (23:26-49); every occurrence depends on the historical crucifixion being affirmed, against Qur’an 4:157’s denial and the associative weight of the Crusades (aṣ-ṣalībiyyūn).
• Most easily missed risk: السامري (Samaritan). A translator without Islamic-studies background could reasonably not know that this exact spelling names a notorious idolater in Qur’an 20:85-97 — a false-friend collision with no parallel anywhere in the Romans baseline.
• Structural risk: chapters 23–24 processed independently. If Phase 2 parallel workers split the crucifixion and resurrection chapters across separate queues without a shared checkpoint, the mandated apologetic scaffolding could be silently lost.
• Familiarity-masks-risk pattern: التوبة (repentance), المغفرة (forgiveness), الصلاة (prayer), and الصوم (fasting) are all comfortable, high-frequency Islamic vocabulary that a reader will not flag as risky — precisely because it feels familiar rather than because it is safe.

---

Opportunities

• Luke 15’s aḑ-ḑāllīn resonance (the “lost” son/sheep/coin) can function as a genuine bridge: the same term closes every Fātiḥah recitation five times daily, meaning God’s active pursuit of the aḑ-ḑāllīn category can reframe rather than merely translate a familiar phrase.
• The historical-crucifixion consensus among nearly every tradition except mainstream Islam (including secular historians) is unusually strong common ground on which to build chs.22-24’s apologetic case.
• Existing Arab Christian liturgical tradition (Coptic, Orthodox, Melkite, Maronite, Syriac) offers a millennium-plus of confident, undiluted precedent for holding ابن الله, الرب, and الصليب publicly — a resource this package draws on rather than reinvents.
• Shared virtue vocabulary (الشكر/thanksgiving, الرحمة/mercy, طوبى/blessedness) provides genuine common ground for the “Good News to the Poor” doctrine, provided the comprehensive (not merely charitable-transactional) sense is taught explicitly.

---

Recommended actions

1. Load assets/translation_memory.json (v1, 109 terms) and assets/doctrine_risk_registry.json (v1, 35 doctrines) before any Phase 2 segment translation begins; do not process Luke segments against the Romans-only files.
2. Route all 23 Critical and 37 High risk terms, and all 27 Critical/High doctrines, to mandatory human theologian review per analysis/12_ai_translation_requirements.md’s escalation rules.
3. Enforce the chs.23-24 single-review-unit rule in Phase 2 Step 16 batch/parallel processing configuration before work begins.
4. Lock the curriculum-wide choice for Luke 23:47 (بريء vs. صدّيق) and the base source text for Luke 2:14’s variant reading at the start of Phase 2, not per-document.
5. Brief all human theologian reviewers on the السامري Qur’anic-namesake collision specifically — this is the single risk in the package most likely to be missed by a reviewer without targeted Islamic-studies background.