Leviticus → Bengali Language Package: Executive Summary

Why it matters

Leviticus is the most linguistically exposed book this Bengali pipeline has processed to date. Unlike Romans’s largely abstract soteriological vocabulary, Leviticus’s subject matter — animal sacrifice, incense, priesthood, purity law, vows, sacred calendars — sits directly on top of living, currently-practiced Hindu ritual categories in West Bengal and active folk-spiritist and shrine-vow practice in both West Bengal and Bangladesh. The core passage, Leviticus 16 (the Day of Atonement), concentrates nearly every Critical-tier risk in the book into a single 34-verse unit that anchors this entire curriculum and forward-connects typologically to Romans 3:25, Isaiah 53, John 1:29, and Hebrews 9–10.

Key findings

• 8 Critical-risk and 26 High-risk terms (34 total) require mandatory human theologian review at every occurrence — nearly half of the book’s 73-term glossary.
• 6 Critical-risk and 11 High-risk doctrines (17 total) require human theologian review routing; only 1 of 25 doctrines is Low-risk/automated-review-eligible.
• The single highest-exposure term in the entire curriculum is প্রায়শ্চিত্ত (atonement) — it is simultaneously the established Bengali Bible term for biblical atonement AND the standard Bengali Hindu term for self-performed penance, requiring a standing “God-provided, not self-achieved” note at every occurrence.
• বলি (sacrifice/offering) and its compounds recur across nearly every chapter (1–7, 16, 17, 22), making this a cumulative, unavoidable fencing burden rather than an isolated risk.
• Two terms are genuinely missing from existing Bengali Bible tradition and had to be coined fresh in this package: the scapegoat/Azazel goat (অপসারণের ছাগল) and “bear iniquity” (পাপ বহন করা) — both load-bearing for future Isaiah 53/John 1:29 typology.
• One open item remains unresolved pending explicit theologian sign-off: the প্রেম/ভালোবাসা split for “love your neighbor” (Leviticus 19:18), flagged but not closed by this Phase 1 package.

Risks

• শুচি/অশুচি (clean/unclean) risks reinforcing caste-based purity/untouchability categories still socially live in parts of Bengal — a sociological collision, not a competing-deity collision, and therefore easy to under-flag.
• যাজক vs. পুরোহিত (priest) is a high-frequency, habitual-lapse risk: পুরোহিত (Hindu Brahmin caste-priest) is the more common word in everyday Bengali speech, so translators and even native-speaker reviewers may reach for it by default.
• আসাসেল (Azazel transliteration), if used without its mandatory clarifying note, risks readers assuming the goat is sacrificed to a rival spiritual power — a live category in regional folk-spiritist belief.
• মানত vs. প্রতিজ্ঞা (vow) and মুক্তি vs. যোবেল বছর/উদ্ধার করা (Jubilee/redemption) are narrow-footprint but severe-in-kind risks (ch. 27 and ch. 25 respectively); both terms sit adjacent to well-established baseline “forbidden substitution” precedent and must inherit that same absolute enforcement.

Opportunities

• অভিষেক (anointing) offers a strong, underused pedagogical bridge to খ্রীষ্ট (“the Anointed One”), already fixed in the Romans baseline — Leviticus 8’s ordination material can reinforce Christ’s anointed office forward.
• The graded holiness system (common → clean/unclean → holy → most holy) has a genuine cross-cultural teaching bridge in Bengali Hindu temple architecture’s own graded sacred space, provided the aniconic, atonement-gated theology is clearly distinguished from image-indwelling temple theology.
• Leviticus 25’s Jubilee message of periodic economic restoration has strong pastoral resonance given the Bengal region’s land-tenancy and debt-bondage history, offering a live application opportunity once কেপ্ট distinct from soteriological vocabulary.

Recommended actions

1. Route all 34 Critical/High-risk terms and all 17 Critical/High-risk doctrines to human theologian review before any Phase 2 segment translation begins, per doctrine_risk_registry.json and bible_term_registry.json routing.
2. Close the open প্রেম/ভালোবাসা (Leviticus 19:18) decision with explicit theologian sign-off before translating chapter 19; propagate the ruling to translation_memory.json version 2 once decided.
3. Treat the প্রায়শ্চিত্ত and বলি standing teaching notes as recurring per-lesson footnotes, not one-time explanations, given their high-frequency collision risk across the whole book.
4. Merge this package’s translation_memory.json and bible_term_registry.json into the shared asset store, loading them alongside (never instead of) the Romans baseline files for every Phase 2 Leviticus session.
5. Reserve the scapegoat/Azazel, bear-iniquity, and mercy-seat renderings fixed in this package as locked precedent for any future Isaiah 53, John 1:29, or Hebrews 9–10 curriculum.