Executive Summary: 2 Peter → Sanskrit Language Package

Why it matters

2 Peter is the most doctrinally concentrated book yet processed for the Sanskrit Language Package. In just three chapters, it generates 26 distinct doctrines and roughly 96 tracked glossary terms, with 50 terms (17 Critical + 33 High) and 21 doctrines (10 Critical + 11 High) requiring mandatory human theologian review before any translated segment can be approved. This is a materially higher review burden, proportional to length, than the Romans baseline. Two collision clusters drive this density: (1) 2 Peter’s own repeated thesis language around eyewitness testimony, apostolic authority, and “knowledge” of God, and (2) its flood/Day-of-the-Lord typology, which sits directly on top of Purāṇic cyclical cosmology — the single densest classical-philosophical crowded neighborhood encountered anywhere in this Language Package to date, denser than anything in the Romans baseline.

Key findings

• New absolute-forbidden terms, on par with मोक्ष/मुक्ति for “salvation”: प्रलयः is now forbidden for “flood” (never suggest the Genesis flood is a cyclical mahāpralaya); माया is forbidden for the false teachers’ “deceit” (never let ethical fraud collapse into Advaita’s metaphysical illusion doctrine); मुक्तिः/मोक्षः are forbidden for the false teachers’ counterfeit “freedom” in 2:19, where the false promise is ironically juxtaposed in the same verse with “slaves of corruption.”
• A persistent, letter-wide collision, unlike anything in the baseline: ἐπίγνωσις/γνῶσις (“knowledge of God/Christ”) recurs six times as 2 Peter’s own self-description of the Christian life, colliding with Bhagavad Gītā jñāna-mārga at every occurrence rather than at one anchor verse — this book requires a policy exception to the Language Package’s normal “note at first occurrence” rule.
• A second three-way collision requiring active engagement, not just avoidance: “partakers of the divine nature” (1:4) must be explicitly distinguished from Advaita’s ātman-Brahman identity, from generic deification, and from Vaiṣṇava sādharmya-mukti — a genuinely close structural parallel this Language Package cannot simply fence off by word choice alone.
• A confessional refrain demanding absolute consistency: “our Lord and Savior Jesus Christ” recurs five times (1:1, 1:11, 2:20, 3:2, 3:18); the new term त्राता (“Savior,” built on the baseline’s own √trā root) must render identically every time.
• Existing baseline infrastructure holds up well: every Critical/High term inherited from the Romans package (सर्वेश्वरः, प्रभुः, त्राणम्, धार्मिकता, पवित्र आत्मा, विश्वासः, etc.) transfers to 2 Peter without modification, confirming the baseline’s fencing methodology generalizes correctly to a new book.

Risks

• Highest risk: a single lapse into प्रलयः anywhere in the flood-to-fire typology (2:5 and 3:6 must match) would retroactively reframe 2 Peter 3’s entire final-judgment argument as one more turn of an endless cosmic wheel rather than the linear, final event Peter teaches.
• Second-highest risk: knowledge-family terms drifting toward an unglossed jñāna-mārga reading across any of their six occurrences, since this is not a one-time doctrinal note but a whole-letter requirement.
• Third risk: the awaited-Kalki-avatāra parallel to Christ’s return (1:16, 3:3-4, 3:10) is unusually exact — Kalki is specifically the still-future tenth avatāra — making this collision more seductive to a Sanskrit-trained reader than the baseline’s general avatāra caution already anticipates.
• Fourth risk: intra-verse irony collapse in 2:19, where any leak of मुक्ति/मोक्ष vocabulary into the false teachers’ “freedom” would simultaneously violate the salvation-vocabulary reservation and accidentally dignify the false teachers’ own claim.

Opportunities

• Several strategic positive borrowings extend the baseline’s own methodology cleanly: प्रत्यक्षद्रष्टा (eyewitness) borrows Nyāya’s pratyakṣa pramāṇa to give apostolic testimony genuine argumentative weight; प्रतिज्ञा (promise) borrows Nyāya’s syllogistic thesis-statement term to reinforce God’s promise as a to-be-vindicated proposition.
• The दीपः (lamp)/light metaphor at 1:19 has positive resonance with Upaniṣadic light-imagery that can be safely leveraged once anchored to Scripture’s authority rather than generalized into self-realization.
• This book’s polemic against false teachers offers a rare opportunity to demonstrate the fencing methodology’s maturity: every major false-teacher term (मिथ्याशिक्षकः, विनाशकर-मिथ्यामतम्, अधिपतिः, स्वर्गदूतः, अन्धकारबन्धनम्) already has a clean, low-baggage descriptive alternative identified, with no unresolved gaps remaining.

Recommended actions

1. Route all 21 Critical/High doctrines and 50 Critical/High terms through mandatory human theologian review before Phase 2 approval; do not permit automated-only sign-off on any of these, per the escalation rules in 12_ai_translation_requirements.md.
2. Enforce the “note at every occurrence” exception for knowledge-family terms (Rule 8) as a standing override of the Language Package’s default first-occurrence-only footnoting policy.
3. Add automated pre-submission linting for the full 2 Peter forbidden-term list (प्रलय, महाप्रलय, लय, मुक्ति/मोक्ष in false-freedom contexts, माया, पुराण/इतिहास, वाणी/वाच्/शब्द, शुक्र, देव, नरक, आचार्य, पाषण्ड, स्वामी, न्याय, संयम, सनातन, काम/तृष्णा/वासना, मोक्षदातृ/मुक्तिदातृ, अवतारागमनम्) before any segment reaches human review, to reduce reviewer load on mechanically detectable violations.
4. Brief reviewing theologians specifically on the Vaiṣṇava sādharmya-mukti parallel for “partakers of the divine nature,” since this is the one doctrine in the curriculum requiring comparative theological argument rather than simple terminological avoidance.
5. Carry forward all inherited baseline terms unchanged into Phase 2 segment translation; no re-litigation of Romans-established renderings is needed or permitted for this curriculum.

---

Term and doctrine counts in this summary are drawn from assets/translation_memory.json, assets/bible_term_registry.json, and analysis/11_doctrine_analysis.md (2 Peter), reconciled against assets/doctrine_risk_registry.json’s per-doctrine tiers per that document’s own reconciliation note.