01 Executive Summary — Matthew — English → Malayalam

Why it matters

Malayalam enters this curriculum as the lowest-risk destination language in the pipeline, thanks to nearly two millennia of continuous, settled Saint Thomas Christian vocabulary. But Matthew is a different book than Romans: it is doctrine-dense in exactly the areas Romans barely touches — the Kingdom of Heaven, the Son of Man, church discipline, the Great Commission, and final judgment/eschatology — so this curriculum requires substantial new glossary construction, not just reuse of the Romans baseline. Get the Beatitudes (the core passage, Matthew 5:1-12) and the eschatology vocabulary wrong, and the risk isn’t obscurity — it’s silently reintroducing Hindu karma-fate or cyclical-rebirth cosmology into the text through fluent-sounding Malayalam words the actual Christian tradition never used.

Key findings

• 40 Critical/High risk terms now populate the Matthew term registry (16 Critical + 24 High), against 24 of 34 doctrines (11 Critical + 13 High) requiring mandatory human theologian review — see assets/doctrine_risk_registry.json and analysis/11_doctrine_analysis.md.
• The core passage’s single highest-priority risk is ഭാഗ്യവാന്‍ (“blessed,” makarios), used nine times in Matthew 5:3-11. Its root ഭാഗ്യം is Malayalam’s everyday word for luck/fortune, carrying karma-adjacent connotations — every occurrence requires a mandatory translator note.
• Matthew introduces 10 genuine missing-vocabulary gaps with no Romans precedent (Son of Man, bind/loose, make-disciples causative, the Trinitarian singular-name formula, end-of-age anti-yuga framing, Parousia, blasphemy-against-the-Spirit, ransom’s doctrinal anchoring, “fulfill the Law,” Immanuel) that must be constructed correctly the first time.
• The clearest AI-hallucination risk in the entire curriculum is യുഗ- (yuga) compounds for “the end of the age”: these are statistically more common in the general Malayalam corpus than the doctrinally correct ലോകാവസാനം/കാലസമാപ്തി, making this an even higher-probability failure mode than the baseline’s already-flagged അവതാരം risk for incarnation.
• Two established Malayalam Bible phrases (സ്വർഗ്ഗരാജ്യം and ദൈവരാജ്യം) both correctly denote “kingdom,” but Matthew’s ~32-to-2 preference for the “heaven” form over Romans’ near-total use of the “God” form makes cross-document consistency a first-order translation risk in itself.
• All 28 chapters of Matthew have been reviewed; every chapter is mapped to the doctrine matrix or explicitly marked “reviewed, no new doctrine” — full-book coverage is confirmed with zero silent omissions.

Risks

• Critical-tier collision risk (11 doctrines): incarnation/virgin birth, deity and sonship of Christ, Son of Man’s authority, lordship, the Trinitarian baptismal formula, judgment/end-of-age, the Second Coming, blasphemy against the Holy Spirit, the atonement-as-ransom, and the resurrection — any drift here alters essential doctrine and requires theologian sign-off on every occurrence.
• Kerala-specific vocabulary collisions: കന്യക (virgin, Kanyakumari resonance), കരുണ (mercy, generic Hindu/Buddhist compassion ethics), ഭാഗ്യവാന്‍ (blessed, luck/fate), ജീവന്‍ vs. ആത്മാവ് (life/soul, Vedantic ātman-merger risk in “lose your life to find it” sayings), and folk-ritual associations with കെട്ടുക/അഴിക്കുക (bind/loose).
• Internal Malayalam-church sensitivity carried from Romans: Matthew’s universal “all nations” language (Great Commission, Gentile inclusion) intersects Kerala’s own Syrian-Christian/Dalit-convert caste-heritage stratification, making “no distinction” claims a live internal-church matter, not only an interfaith one.
• Grammatical collision unique to Matthew: the singular ദൈവപുത്രന്‍ (Christ’s unique title) sitting beside the new plural ദൈവപുത്രന്മാര്‍/ദൈവത്തിന്‍റെ മക്കള്‍ (believers’ adoptive sonship, introduced at 5:9) creates a same-book singular/plural confusion risk absent from Romans.

Opportunities

• Malayalam’s settled tradition means most of Matthew’s reused baseline vocabulary (ദൈവം, യേശു, ക്രിസ്തു, കര്‍ത്താവ്, പരിശുദ്ധാത്മാവ്, രക്ഷ, നീതി) needs zero new deliberation — disciplined enforcement of assets/translation_memory.json, not invention, closes most of the risk surface.
• Kerala’s mission-sending tradition (one of South Asia’s longest continuous) gives the Great Commission positive institutional resonance rather than colonial-era baggage, an asset most other languages in this pipeline lack.
• The Danielic Son of Man title, properly glossed once per document, becomes a teaching opportunity rather than a liability: it lets the curriculum surface Christ’s humanity and judgment-authority together in a way Romans never needed to.
• A single well-built anti-yuga forbidden-substitution rule for “the end of the age” (parallel to the existing അവതാരം rule) closes the pipeline’s highest-probability new hallucination vector before Phase 2 translation begins.

Recommended actions

1. Merge assets/translation_memory.json (Matthew version) and assets/bible_term_registry.json (Matthew version) into the active Phase 2 enforcement pipeline immediately; do not begin segment translation without both loaded.
2. Route all Matthew 5:1-12 (Beatitudes) segments through a single reviewer/session to guarantee consistent ഭാഗ്യവാന്‍ translator-note application across all nine occurrences.
3. Hard-code the anti-യുഗ substitution check and the സ്വർഗ്ഗരാജ്യം/ദൈവരാജ്യം Greek-triggered selection rule into the Step 15/17 automated validation pass before any human review begins.
4. Brief theologian reviewers specifically on the singular/plural Sonship distinction (5:9 vs. 3:17/16:16) and the bind/loose forensic-vs-folk-magical framing (16:19; 18:18) — these are the two doctrinally consequential risks with no Romans-baseline precedent to draw on.
5. Carry the Great Commission (28:18-20) and Trinitarian formula through Phase 2 as a single locked, verbatim-consistent unit across every document that quotes it, per the theological consistency rules in analysis/12_ai_translation_requirements.md.