Executive Summary: Psalms → Bengali Language Package Extension

Why it matters

Psalms is the first Hebrew-poetry, Old Testament curriculum built on this Bengali Language Package. It introduces a covenant-name (YHWH), a concrete devotional-imagery register (shepherd, king, fortress, refuge), and Hebrew wisdom/lament genres that Romans’ Greek epistolary prose never required — landing directly in the two richest devotional vocabularies in Bengal: Vaishnava bhakti (Krishna-cowherd, kirtan) and Shakta goddess-worship (Durga/দুর্গ). Getting Psalm 23’s “Shepherd” and Psalm 110’s “two Lords” right is not a stylistic nicety — it is the difference between teaching the covenant God of Israel and accidentally cueing a reader toward Krishna or Durga devotion, or toward an Islamic Christology that denies Christ’s deity and resurrection.

Key findings

• 86 total glossary terms now govern Phase 2 Psalms translation: 39 inherited exactly from the Romans baseline, 47 newly coined or fixed for this book.
• 27 terms are Critical risk, requiring mandatory human theologian review at every occurrence — nearly double the Romans baseline’s Critical count, driven by Psalms’ concentration of messianic-prophecy and covenant-name vocabulary.
• 32 terms are High risk; combined, 59 of 86 terms (69%) require theologian-level scrutiny.
• At the doctrine level, the doctrine_risk_registry.json extension defines 24 granular doctrines nesting under the curriculum’s 8 assigned doctrine headings: 9 Critical, 11 High, 4 Medium, 0 Low — 20 of 24 doctrines (83%) route to mandatory human theologian review, with the remaining 4 routed to native-speaker review; none qualify for automated-only review, a first for this Language Package.
• Two single verses carry outsized doctrinal weight: Psalm 110:1 (“The LORD says to my Lord”) and Psalm 23:1 (“The LORD is my shepherd”) — both flagged as the highest-density Critical-risk points in the entire 150-psalm sweep.
• Full-book coverage confirms all 150 psalms reviewed; no chapter silently omitted (see 11_doctrine_analysis.md Part C).

Risks

1. রাখাল/Krishna-shepherd collision in the core passage itself (Psalm 23) — the single highest-visibility risk in the package, because রাখাল is the more colloquially natural word.
2. দুর্গ/Durga homophone for “fortress” — an aural risk that survives standard proofreading and requires a dedicated read-aloud review pass.
3. হেসেদ (hesed) collapsing to plain দয়া — the older, familiar Bengali Bible word — silently erasing covenant-permanence across 120+ occurrences, including Psalm 136’s 26-fold refrain.
4. OT অভিষিক্ত জন collapsing into খ্রীষ্ট or মশীহ — destroying the prophecy-then-fulfillment structure central to the Messianic Prophecy doctrine, or importing an Islamic Christology that denies Christ’s divinity and resurrection.
5. Psalm 110:1’s two referents merging into one word — the sharpest point of Jewish-Christian and Islamic-Christian doctrinal disagreement in the curriculum, and the direct textual basis for Christ’s own self-testimony to deity (Matthew 22:41-45).
6. ধ্যান করা creeping in for “meditate” (Psalm 1:2, Psalm 119) — the most naturally-reached-for Bengali verb, but a fixed yogic/Buddhist technical term incompatible with the text’s verbal rumination sense.

Opportunities

• Bengali’s rich devotional vocabulary (bhakti viraha-longing, kirtan embodied praise, hesed-adjacent kripa) offers genuine pastoral bridges for teaching Lament, Praise, and Covenant Faithfulness — resonance in tone can be acknowledged even while content is kept doctrinally distinct.
• The idolatry polemic (Psalm 96, 97, 106, 115, 135) can use প্রতিমা’s direct cultural overlap with Durga Puja imagery as an intentional, sharpened teaching point rather than a risk to avoid — turning a collision into clarity.
• Establishing অবিচল দয়া and সদাপ্রভু as fixed, consistent terms now sets a durable foundation for every future OT curriculum this Language Package will need (Proverbs, Isaiah, Genesis, etc.).

Recommended actions

1. Load assets/translation_memory.json (this extension) alongside the Romans baseline before any Phase 2 Psalms segment translation; treat the Psalms-specific forbidden-substitution list as equally binding.
2. Route Psalm 1, 2, 22, 23, 51, 89, 110, 118, and 136 for earliest-priority human theologian review scheduling, given their outsized concentration of Critical-risk decisions.
3. Run a dedicated read-aloud/oral review pass specifically for দুর্গ-adjacent and রাখাল-adjacent decisions, since both collision risks are partly or wholly aural.
4. Carry forward every Romans baseline decision unchanged (গ্রেস/অনুগ্রহ, পরিত্রাণ, ধার্মিকতা, ঈশ্বর, যীশু, পবিত্র আত্মা) — this extension adds to, never re-litigates, the settled baseline.
5. Brief all theologian reviewers on the Psalm 110:1 two-referent rule and the Psalm 51 three-term sin taxonomy before review begins, since both are structural rules easy to violate without dedicated briefing.

---

Full supporting detail: analysis/04_comparative_theology.md, analysis/06_linguistic_gap_analysis.md, analysis/07_semantic_analysis.md, analysis/08_core_glossary.md, analysis/11_doctrine_analysis.md, assets/translation_memory.json, assets/bible_term_registry.json, assets/doctrine_risk_registry.json.