Executive Summary: 2 Corinthians — English → Malay Language Package

Why it matters

2 Corinthians brings a substantially higher concentration of Critical- and High-risk theological content into the Malay Language Package than Romans did, at exactly the moment the destination-language context is least forgiving of imprecision. Malaysia’s legal and political history already makes vocabulary itself — not just doctrine — a contested space (the Allah naming litigation, 2007-2021). This curriculum’s core passage (2 Corinthians 5:11-21) compresses substitutionary atonement, non-imputation of sin, new creation, and reconciliation into eleven verses, each with a direct, documented collision point against either mainstream Sunni theology (the Qur’anic no-bearer-of-another’s-burden principle, fitrah anthropology) or Malaysian cultural/legal reality (muhibbah social-harmony rhetoric, the Baitullah/Kaaba name, Isra’ and Mi’raj). Getting this letter’s terminology and doctrine framing right is not optional polish — it is what keeps the curriculum’s central passage from being silently reinterpreted into something Paul did not say.

Key findings

• Full-book coverage confirmed. All 13 chapters of 2 Corinthians have been analyzed (semantic analysis, core glossary, linguistic gap analysis, doctrine analysis); no chapter contributes zero doctrinal load, and chapter 3 (New Covenant/veil/transformation) and chapters 11-12 (false apostleship/visionary authentication) carry the heaviest new-term density in the letter.
• Term inventory: 85 total terms now govern this curriculum’s translation memory (v2) — 29 inherited exactly from the Romans baseline, 56 newly introduced by 2 Corinthians.
• 57 of those 85 terms (26 Critical + 31 High) require mandatory human theologian review on every occurrence, per the Language Package’s escalation rules — up from the Romans baseline’s smaller Critical/High share, reflecting this letter’s denser polemical and atonement-theology content.
• Doctrine registry: 25 of 31 named doctrines (12 Critical + 13 High) require human theologian review; 4 require native speaker review; only 2 (Thanksgiving, Mutual Edification) are automated-review-only.
• Five newly identified “silent collision” risks surfaced during this analysis that were not flagged at this specificity in the Romans baseline: Pendamaian/muhibbah (reconciliation vs. social-harmony rhetoric), Ciptaan baru/fitrah (new creation vs. Islamic born-pure anthropology), Bait Allah/Baitullah (temple of God vs. the Kaaba’s standard name), Langit ketiga-Firdaus/Isra’ wa Mi’raj (Paul’s vision vs. the foundational night-journey narrative), and Kuasa disempurnakan dalam kelemahan vs. honor-shame/triumphalist-prophetology expectations.

Risks

• Highest risk (rank 1-2 of the letter’s ambiguity list): 5:21’s “made him to be sin” and 5:14-15’s “one died for all” directly collide with the Qur’anic principle that no soul bears another’s burden (Q6:164; 17:15; 35:18) — any softening here corrupts the entire reconciliation argument downstream.
• High-likelihood “invisible” mistranslation: Pendamaian (reconciliation) is fluent, doctrinally-correct-sounding Malay that can silently import Malaysia’s muhibbah interfaith-harmony framing if left without an explicit divine object in every occurrence.
• Structural ambiguity: “Perjanjian Baru” (New Covenant, 3:6) is also the established Alkitab name for the entire New Testament corpus — a high-frequency, easy-to-overlook double meaning.
• Legally/pastorally sensitive passages: “Unequally yoked” (6:14) and interfaith-partnership framing intersect real Malaysian legal restrictions on religious conversion and interfaith marriage, requiring non-inflammatory pastoral handling, not just accurate translation.
• Newly flagged, previously under-specified: Bait Allah (temple of God, 6:16) risks direct confusion with Baitullah, the standard name for the Kaaba — this analysis elevates its review routing from native-speaker to human-theologian accordingly.

Opportunities

• Genuine cultural bridge points exist and should be used deliberately, not avoided: Takut akan Tuhan (fear of the Lord) resonates with Islamic takwa; Kecukupan (sufficiency) resonates with Sufi/Islamic qana’ah (contentment); Penderma yang rela hati (cheerful giver) resonates with zakat/sedekah’s niyyah (pure-intention) requirement. Each is a legitimate on-ramp for comprehension provided teaching notes redirect the reader from a merit-earning frame to a grace-fruit frame.
• Malaysia’s own royal-lineage and adoption-friendly cultural vocabulary (established in the Romans package) continues to serve this curriculum well, particularly for the Father/adoption language reused at 6:18.
• This letter’s dense, concrete metaphor set (jars of clay, triumphal procession, thorn in the flesh, ambassador) is comprehension-friendly once explanatory notes are supplied, since these are cultural-literacy gaps rather than syncretism risks — lower-cost to resolve than the doctrinal-collision terms above.

Recommended actions

1. Load assets/translation_memory.json (v2, 85 terms) and the 2 Corinthians bible_term_registry.json/doctrine_risk_registry.json before any Phase 2 segment work begins; do not reuse the Romans-only v1 memory file.
2. Apply analysis/12_ai_translation_requirements.md (v2.2) as the binding Phase 2 instruction set; it extends, and does not contradict, the Romans v2.1 requirements document.
3. Pre-schedule human theologian review capacity for the 57 Critical/High terms and 25 Critical/High doctrines identified above, rather than triaging reactively during Phase 2 Step 17 — this letter’s review load is proportionally heavier than Romans’.
4. Treat the core passage (2 Corinthians 5:11-21) as the mandatory theologian-review anchor for every segment in which it appears, in full or in part, per the escalation rules.
5. Brief all Phase 2 translators explicitly on the five newly identified “silent collision” terms (Pendamaian, Ciptaan baru, Bait Allah, Langit ketiga/Firdaus, Kuasa disempurnakan dalam kelemahan) before work begins, since these are the terms most likely to produce fluent-sounding but doctrinally compromised output if not pre-flagged.