01 — Executive Summary: Psalms (Mandarin Destination Language)

Why it matters: The Psalms curriculum extends a mature, well-tested Romans→Mandarin Language Package into a full-book Hebrew-poetry corpus that introduces its own collision risks — most acutely a term (慈爱, hesed) with no Romans-baseline equivalent, and a doctrine (Refuge) that collides with a named, fixed Buddhist technical term (皈依) rather than a loose cultural analogy. Getting this package right before Phase 2 begins prevents systemic mistranslation across 150 psalms and roughly 2,461 verses that will anchor teaching on God’s character, lament, praise, messianic prophecy, and covenant loyalty for Mandarin-speaking learners.

Key findings:

• Full-book coverage is complete: all 150 psalms reviewed chapter-by-chapter (11_doctrine_analysis.md Part 2); every chapter contributing no new term is explicitly logged as reviewed, not silently skipped.
• 29 doctrines are now registered in doctrine_risk_registry.json (Psalms), of which 12 are Critical and 8 are High — 20 total requiring mandatory human theologian review, versus 7 requiring native speaker review and 2 automated-only.
• The term-level registry (bible_term_registry.json, Psalms) carries 81 entries; of these, 24 are Critical and 26 are High risk — 50 terms (62% of the total registry) requiring theologian-level scrutiny before Phase 2 approval, a materially higher density than the Romans baseline.
• The single highest-density collision zone in either package is 慈爱 (hesed): a three-way fence is required against 恩典 (baseline NT grace), 义气 (Chinese sworn-brotherhood loyalty), and 缘分 (folk-Buddhist predestined connection) — recurring roughly 130 times, including the 26-fold refrain of Psalm 136.
• The core passage (Psalm 23:1-6) alone touches four Critical/High doctrines simultaneously: Shepherd care, Refuge, Hesed, and the nephesh (“soul”) term — making it the theological and linguistic pressure point of the entire curriculum, exactly as intended for its anchor role.

Risks:

• Refuge vs. 皈依 (Critical): any lapse into the Buddhist Three Refuges technical term would import an entire alternate soteriological system into biblical refuge-taking language — a zero-tolerance error class with no acceptable context.
• 义人 vs. 君子 (Critical): Psalm 1’s righteous person is the single most acute Confucian-collision risk in the curriculum; silent absorption into the self-cultivated-gentleman ideal would invert the doctrine’s entire logic (received flourishing vs. self-achieved virtue) without any lexical signal to the reader.
• 受膏者/基督 vs. 弥勒 (Critical): the historic pattern of Chinese folk-millenarian movements proclaiming a coming anointed savior-king creates a live cultural pattern-match against Maitreya expectation-language, extending the Romans baseline’s absolute prohibition to four additional psalms (2, 89, 110, 132).
• 求你为我造清洁的心 (Critical): Psalm 51:10 is the Old Testament’s strongest anticipation of the baseline’s Neo-Confucian self-cultivation caution on 成圣; any drift toward a self-improvement reading would corrupt the curriculum’s dedicated Confession and Forgiveness anchor text.
• Consistency risk (mechanical, Critical): the Exodus 34:6 mercy formula (Ps 86:15, 103:8, 145:8) and the hesed refrain (Ps 107, 118, 136 ×26) must render verbatim-identically at every occurrence; drift here is a quality-control failure independent of any single mistranslation.

Opportunities:

• Mandarin is well-positioned for this curriculum: over a century of CUV usage already provides settled, community-recognized renderings for the highest-visibility terms (救恩, 复活, 耶稣, 基督, 锡安, 麦基洗德, 哈利路亚, 细拉), so the translation task is defensive (fencing existing good vocabulary) more than inventive.
• The Psalms-Romans textual bridge (Ps 14/53→Rom 3:10-18; Ps 32→Rom 4:7-8; Ps 44→Rom 8:36; Ps 69→Rom 11:9-10, 15:3) allows this package to reinforce, rather than duplicate, Romans-baseline Critical doctrines (universal_human_accountability, imputed_righteousness, grace) with fresh source material.
• Psalm 23’s convergence of doctrines makes it an efficient single teaching unit for onboarding learners to the whole package’s key fences (牧者 ≠ 守护神; 避难所 ≠ 皈依; 慈爱 ≠ 恩典/义气/缘分; 灵魂 ≠ 魂魄) in one sitting.

Recommended actions:

1. Route all 50 Critical/High terms and 20 Critical/High doctrines through mandatory human theologian review before Phase 2 batch translation begins, per the escalation rules in 12_ai_translation_requirements.md.
2. Lock the Exodus 34:6 mercy formula and the hesed refrain as invariant strings in translation memory now, before any worker begins parallel Phase 2 processing, to prevent drift across the 26 occurrences in Psalm 136 alone.
3. Prioritize theologian sign-off on Psalm 23 (core passage), Psalm 1 (义人/恶人), Psalm 51 (求你为我造清洁的心), and Psalm 110 (我主) as the four highest-leverage passages for the whole package.
4. Carry forward the absolute 皈依 and 弥勒 prohibitions into all Phase 2 prompts with zero-exception language, matching the Romans baseline’s treatment of 功德 and 轮回.
5. Brief native-speaker reviewers specifically on the Confucian 君子/小人 and 义气 collision risks, since these are cultural pattern-matches rather than named technical terms and are therefore easier for a translator to introduce unconsciously.

End of Phase 1 Step 11 deliverable.