Executive Summary: Haggai (English → Bengali)

Why It Matters

Haggai is short — 2 chapters, 38 verses — but every verse is doctrinally load-bearing for this curriculum’s four anchor doctrines (God’s House vs. Self-Interest; Obedience and Blessing; Presence and Spirit of God; Coming Glory of the Temple). Bengali’s Bible-translation landscape sits at the intersection of Shakta/Vaishnava Hindu devotional vocabulary and Sunni/Sufi Islamic idiom, both of which offer plausible-sounding but doctrinally wrong substitute readings for this book’s core terms (YHWH, God’s house, God’s Spirit, blessing, election). Getting Phase 1 terminology and risk-routing right here directly determines whether Phase 2 translation preserves Haggai’s specific covenantal argument or lets it dissolve into generic temple-merit piety, karma-blessing logic, or impersonal-presence mysticism.

Key Findings

• Full-book coverage confirmed. All 38 verses across both chapters map to at least one of 15 tracked doctrines; no chapter or section is doctrine-empty (see analysis/11_doctrine_analysis.md §1).
• No wholesale vocabulary gap exists. Bengali’s established OT Bible-translation tradition plus the Romans baseline’s already-solved terms (ঈশ্বর, ব্যবস্থা, পবিত্র, শান্তি, মহিমা, ভাববাদী, ঈশ্বরের মনোনয়ন) cover most of this book directly; the dominant risk is semantic crowding by Hindu/Muslim neighbor terms, not missing words (see analysis/06_linguistic_gap_analysis.md).
• One irreducible textual ambiguity — “desire of all nations” (2:7) — must be preserved, not resolved, in every occurrence, a translation posture unique to this term in the entire pipeline so far.
• One discourse-level structural risk — the সদাপ্রভুর গৃহ (God’s house) vs. বাড়ি (ordinary house) contrast — must be sustained without a single lapse across all 16+ occurrences spanning both chapters, or the book’s central rebuke becomes invisible in Bengali.
• Translation memory now totals 51 enforced terms: 10 inherited exactly from the Romans baseline, 41 new Haggai-specific entries.

Risks (Counts Requiring Elevated Oversight)

Doctrine-level (from doctrine_risk_registry.json):

Tier	Count	Review Routing
Critical	3	Human theologian
High	6	Human theologian
Medium	5	Native speaker
Low	1	Automated
Total requiring theologian review	9 of 15 doctrines (60%)	—

Term-level (from assets/translation_memory.json, 51 total terms):

Tier	Count
Critical	6 (god, holy_spirit, messiah [inherited] + yhwh, my_spirit_remains, desire_of_all_nations [new])
High	14 (law, holy, glory, election, covenant [inherited] + lord_of_hosts, word_of_the_lord, house_of_the_lord, his_own_house_my_house, obey_voice_of_the_lord, i_am_with_you, spirit_human_stirred, choose_elect, remnant [new])
Medium	20
Low	11

Highest-severity individual risks: (1) “desire of all nations” (2:7) — Critical, disputed messianic/material reading; (2) “my Spirit remains” vs. “stirred human spirit” (2:5 vs. 1:14) — Critical, same Bengali word (আত্মা) serving two different referents eleven verses apart; (3) the গৃহ/বাড়ি structural contrast — High, cumulative document-wide risk rather than a single-verse error.

Opportunities

• Haggai gives this pipeline its first sustained Old Testament term set (YHWH/সদাপ্রভু, LORD of hosts, remnant, torah-in-context, signet ring) — these are now fixed and reusable for every future OT curriculum, reducing Phase 1 effort on subsequent OT books.
• The book’s brevity makes it an efficient, high-doctrinal-density module for demonstrating the pipeline’s OT-extension capability to reviewers before larger OT curricula are attempted.
• The mandatory dual-reading footnote convention established here for 2:7 provides a reusable template for other genuinely disputed OT messianic texts likely to recur later in the curriculum library (e.g., Isaiah, Psalms).

Recommended Actions

1. Lock the সদাপ্রভু/প্রভু/ঈশ্বর three-way distinction as a permanent, pipeline-wide OT convention before any further OT curriculum begins Phase 1 — this is now the template every future OT book must follow.
2. Route Haggai 2:5, 2:7, and 2:23 to human theologian review unconditionally, regardless of how routine the surrounding segment appears, per the escalation rules in analysis/12_ai_translation_requirements.md.
3. Add a document-level (not merely segment-level) validation pass in Phase 2 Step 17 specifically to check the গৃহ/বাড়ি contrast holds across the entire translated document — standard per-segment QA is insufficient for this risk.
4. Brief Phase 2 human reviewers explicitly on the Bengali cultural collision points documented in analysis/04_comparative_theology.md (Hindu Brahman/paramatma immanence vs. “I am with you”; karma-phal vs. obedience-blessing; মন্দির/temple-merit culture vs. God’s house) before translation work begins, not after errors surface.
5. Carry the newly fixed Haggai terms forward into translation memory version control so later curricula referencing the Spirit, remnant, or election doctrines inherit these exact renderings rather than re-deriving them.

---

Full detail: analysis/07_semantic_analysis.md (verse-by-verse), analysis/08_core_glossary.md (term table), analysis/11_doctrine_analysis.md (doctrine matrix), analysis/04_comparative_theology.md (cross-tradition risk), analysis/06_linguistic_gap_analysis.md (vocabulary gap/crowding analysis), assets/translation_memory.json, assets/bible_term_registry.json, assets/doctrine_risk_registry.json.