Executive Summary
01 Executive Summary — Titus (English → Malayalam)
Why it matters
Titus is the first curriculum in this pipeline to extend the Romans-anchored Malayalam Language Package into new doctrinal territory rather than simply reusing it. Malayalam remains the pipeline’s best-positioned language overall — nearly two millennia of settled, ecumenical Christian vocabulary means most terms need disciplined enforcement, not invention. But Titus introduces genuinely new concepts (regeneration, local church office, a grace-training dynamic, a doubled “good works” refrain) that did not exist in the Romans term set, and two of them — regeneration and the elder/overseer identity claim — carry the same severity as the baseline’s most dangerous forbidden-substitution rules (അവതാരം, പുനര്ജന്മം). Getting Titus right requires not just applying the baseline but extending its “fencing” discipline to brand-new vocabulary before an AI system ever gets the chance to drift toward a fluent but wrong word.
Key findings
- 21 doctrines analyzed across all three chapters of Titus, full-book coverage confirmed chapter-by-chapter (1:1–16; 2:1–15; 3:1–15), with sections contributing no new doctrine (3:12–15) explicitly noted rather than omitted.
- 5 Critical-risk and 9 High-risk doctrines identified at the doctrine level (
doctrine_risk_registry.json), for a combined 14 doctrines requiring mandatory human theologian review — two-thirds of all doctrines in this curriculum. 7 further doctrines route to native speaker review; zero doctrines are automated-only, reflecting Titus’s unusually concentrated theological density for a three-chapter letter. - 71 total terms are catalogued in the Titus-specific term registry (in addition to the baseline’s own Romans term set): 9 Critical-risk and 19 High-risk terms (28 terms combined) require theologian-level scrutiny; 34 Medium-risk terms route to native speaker review; 9 Low-risk terms (plus four grouped vice/virtue clusters) are automated-review-sufficient.
- Two brand-new forbidden-substitution rules were engineered at the same severity as the baseline’s incarnation/resurrection rules: (1) regeneration (παλιγγενεσία) must render as നവജനനം, never the reincarnation-adjacent പുനർജനനം/പുനർജന്മം; (2) overseer (ἐπίσκοπος) must render as മേൽവിചാരകൻ, never Kerala’s own fluent but anachronistic ബിഷപ്പ്/മെത്രാൻ.
- One structural (non-lexical) risk dominates the book: the single Greek root ἔργα (“works”) functions as both celebrated fruit (6 occurrences) and rejected basis (Titus 3:5) of salvation within the same short letter. The fix is disciplined consistency (one Malayalam word, സൽപ്രവൃത്തികൾ, everywhere) plus contextual teaching notes — never a second invented word.
- The core passage, Titus 2:11–3:8, concentrates an unusual density of Critical-tier material — deity of Christ (2:13), salvation by grace not works (3:5), regeneration by the Holy Spirit (3:5–6), and justification (3:7) all fall within eight verses.
Risks
- Highest risk (Critical, likelihood High): AI-generated drift toward പുനർജനനം/പുനർജന്മം for regeneration — no existing Malayalam Christian usage anchors against this the way ഉയിര്ത്തെഴുന്നേല്പ്പ് already anchors against reincarnation-drift for resurrection, making this genuinely harder to guard than the baseline’s comparable rule.
- Second-highest risk (Critical, likelihood High): AI or careless-translator drift toward ബിഷപ്പ്/മെത്രാൻ for overseer — these words are unusually fluent and reverent-sounding in ordinary Malayalam Christian speech, increasing temptation relative to most other forbidden-substitution risks in this pipeline.
- High risk (structural, not lexical): a well-intentioned attempt to “resolve” the ἔργα double-function by inventing a second Malayalam word for good works, which would sever Paul’s deliberate wordplay across the letter.
- High risk (reception, not translation): submission-to-authority teaching (2:5, 2:9, 3:1) read through Kerala’s own documented internal caste-heritage church history; the correct Malayalam words are not in question, but teaching notes must actively guard against a caste-subordination reading.
- Compounded bhakti-collision risk: grace (കൃപ, already High-risk in the baseline) is now paired with two new bhakti-adjacent terms — ദൈവഭക്തി (godliness) and ദൈവത്തിന്റെ ദാസൻ (bondservant of God) — tripling the surface area for a devotee-merit misreading in this one curriculum.
Opportunities
- Malayalam’s deep existing vocabulary for church office (മൂപ്പൻ), health-metaphor doctrine (ആരോഗ്യകരമായ ഉപദേശം), and Reformation-era renewal language (നവീകരണം) allowed nearly every new Titus term to be built on safe, already-Christian roots rather than borrowed or invented from a neutral corpus — a genuine strength unique to this language in the pipeline.
- The Buddhist tradition’s own well-known caution against unproductive metaphysical disputation offers a rare positive cross-tradition bridge point for teaching “Avoiding Divisive Controversies,” the curriculum’s lowest-risk doctrine.
- Locking in the fixed Pastoral Epistles formula (ഈ വാക്ക് വിശ്വാസ്യമാണ്) and the ἐπίσκοπος/മേൽവിചാരകൻ fence now positions this Language Package for low-friction extension into a future 1–2 Timothy curriculum.
Recommended actions
- Apply mandatory human theologian review, with no exceptions, to every occurrence of the two newly engineered Critical forbidden-substitution terms (regeneration, overseer) before any Phase 2 output is accepted.
- Add a dedicated cross-segment consistency check (not a simple word-list check) for the six-occurrence സൽപ്രവൃത്തികൾ refrain and its 3:5 counterpart, per the Step 12 structural-flag mechanism.
- Route the unusually term-dense span Titus 3:4–8 to a single reviewer rather than splitting it across parallel Phase 2 workers, to preserve cross-term judgment consistency.
- Require the standing translator note pairing (ദൈവഭക്തി ≠ വിശ്വാസം) on every godliness occurrence, and extend the baseline’s existing bhakti-collision review discipline to the two new bondservant/godliness terms introduced here.
- Carry all caste-sensitivity and denominational-practice cautions (submission; husband of one wife) in teaching notes, per the baseline’s established precedent — never resolve them through softened translation choices.
See assets/doctrine_risk_registry.json and assets/bible_term_registry.json for the complete, machine-readable Critical/High tallies underlying this summary.
See analysis/11_doctrine_analysis.md for the full chapter-by-chapter doctrine matrix and analysis/12_ai_translation_requirements.md for the enforcement instruction set.