Executive Summary: Acts of the Apostles (English → Tamil)

Why it matters

Acts is the pivot book of the New Testament — the Spirit descends, the gospel crosses from Jew to Gentile, and the apostolic church takes its first steps under persecution. For a Tamil destination audience, it is also the single riskiest book this pipeline has translated to date, because its content is narrative and experiential (trance, vision, healing, oracle-spirits, communal life, mistaken deification) rather than argumentative and abstract (righteousness, grace, law) like Romans through Colossians. Tamil religious culture — village Amman-goddess possession festivals, the votive shrine-miracle economy, active folk-magic and oracle-divination practice, purity codes (தீட்டு), and a richly developed religious-pluralism vocabulary (மார்க்கம்) — supplies unusually specific, currently-practiced near-neighbors for nearly every phenomenon Acts narrates. Getting this book right protects the doctrinal integrity of Pentecost, repentance, the church’s identity, and the gospel’s exclusivity claims for every downstream lesson built on this Language Package.

Key findings

• Full-book coverage confirmed. All 28 chapters were analyzed (07_semantic_analysis.md, 11_doctrine_analysis.md); no chapter was silently omitted, and low-yield chapters (12, 18, 21, 25) are explicitly logged as “reviewed — doctrinal continuity.”
• 112 enforced terms now populate translation_memory.json: 66 inherited verbatim from the baseline Romans/Galatians/Ephesians/Philippians/Colossians package, and 46 newly coined or fixed for Acts.
• The risk shape is different, not just larger. The highest-stakes new decisions (repentance, Spirit-filling, “the Way”) are experiential-phenomenon terms with no equivalent problem in the prior five books — Tamil rarely lacks vocabulary for these experiences; the danger is that the existing, currently-practiced vocabulary belongs to a rival religious framework.
• Two deliberate, narrow exceptions to the God-word rule are required for the first time in this pipeline: தெய்வம் at Acts 17:23 (the Athenians’ own altar inscription) and தேவன் at 14:11-15/28:6 (explicitly corrected pagan misattribution) — both flagged for mandatory theologian sign-off so they never bleed into general usage for the true God.

Risks

• 26 Critical-risk terms and 54 High-risk terms in translation memory (80 total) require mandatory theologian-level enforcement, feeding 17 Critical and 22 High doctrines (39 total, matching doctrine_risk_registry.json) that require human theologian review at every occurrence.
• Highest-consequence single collision: “filled with the Holy Spirit” (2:4 and 8 further occurrences) versus ஆவேசம்/சாமி ஆடுதல், the living Tamil temple-festival possession-trance phenomenon — doctrinally the exact opposite (loss of agency vs. Spirit-enabled coherent agency), but visually and lexically the nearest neighbor a translator will reach for by habit.
• Second-highest: “the Way” (9:2 and 5 further occurrences) versus மார்க்கம் — the one lexical choice that would concede Christianity is “one path among many” within Tamil religious pluralism, precisely where Acts names the movement itself.
• Third-highest: “repentance” (6 major occurrences, the book’s most repeated imperative) versus பரிகாரம்/தவம்/பாவநிவாரணம், each of which would convert Spirit-enabled turning into a paid, earned, or ritually mechanical act.
• Additional Critical collisions: clean/unclean purity vision (10:14-15, 28) vs. living தீட்டு codes; spirit of divination (16:16-18) vs. active oracle-possession practice; “no other name” (4:12) vs. Tamil religious pluralism; Hades (2:27, 31) vs. Puranic Pātāla; ascension (1:9-11) vs. Siddhar bodily-transcendence legend.
• Cross-document consistency exposure: four fixed seams (Acts 2:21/38 ↔ Romans 10:9-13; Acts 2:33-35 ↔ Ephesians 1:20-22; Acts 13:38-39 ↔ Romans 3-4; Acts 16:31 ↔ Romans 10:9-13) create failure points if Phase 2 processing does not verify verbatim consistency against the already-shipped baseline curricula.

Opportunities

• Acts 2:6, 8, 11 gives the pipeline’s own translation work its founding warrant: Pentecost’s miracle is precisely that God’s revelation reaches every people in their own tongue — a natural, encouraging teaching bridge to the Bible-translation work this Language Package itself performs.
• The book’s own narrative repeatedly does the fencing work translators need: Simon Magus’s attempted purchase of the Spirit’s gift (8:18-20) is a built-in rebuke of merit-transaction religion; the Lystra and Malta deification episodes (14:8-18; 28:1-6) are built-in rebukes of venerating wonder-workers as divine; Ananias and Sapphira (5:1-11) is built-in proof that communal generosity was voluntary. Teaching material can lean on these as self-authenticating anchors rather than needing external argument.
• The Abba/அப்பா zero-gap opportunity noted in the baseline extends thematically here: Tamil’s family-centric register makes “added to their number” (2:41, 47) and “all things in common” land with warm, native plausibility once safely distinguished from political collectivism.

Recommended actions

1. Load translation_memory.json v6 and 12_ai_translation_requirements.md’s new Acts Additions section before any Phase 2 Acts segment is processed; confirm the coherence-of-agent test is applied at every Spirit-filling occurrence.
2. Route all 39 Critical/High doctrines and all 80 Critical/High terms to mandatory human theologian review per the escalation rules; do not allow automated-only approval for any segment touching Pentecost, repentance, “the Way,” clean/unclean, or the exclusivity claims.
3. Run the four cross-document consistency checks (Acts↔Romans, Acts↔Ephesians) as a dedicated Phase 2 Step 17 sub-task before sign-off, given this is the first curriculum in the pipeline requiring live verbatim cross-checking against four already-published books simultaneously.
4. Brief reviewers explicitly that Acts’s risk profile is experiential, not argumentative — reviewer training materials should lead with the possession-trance, oracle-divination, and shrine-miracle collision set, not the abstract soteriological vocabulary already mastered in the Romans-through-Colossians review cycles.