Numbers — Linguistic Gap Analysis (English → French)

1. Scope and Method

This analysis executes PRD Phase 1 Step 8 for the Numbers curriculum. It builds directly on 08_core_glossary.md and bible_term_registry.json (Numbers extension) and assumes the Romans baseline (translation_memory.json, bible_term_registry.json, doctrine_risk_registry.json) as the fixed language authority for every shared term. Nothing here contradicts the baseline; it extends it to cover:

1. A doctrine vocabulary matrix for the seven named Numbers doctrines.
2. Missing vocabulary — concepts for which French has no ready lexical equivalent.
3. Crowded semantic neighborhoods — French words that already carry strong, doctrinally interfering connotations (Catholic devotional practice, secular idiom, folk religion, bureaucratic register) that must be “fenced” with clarifying strategy.
4. Transliteration vs. paraphrase decisions for Hebrew-specific vocabulary with no NT/Romans precedent.
5. A ranked list of this book’s highest-risk ambiguities, for routing priority in Phase 2.
6. A full-book chapter coverage ledger (Numbers 1–36) confirming every chapter has been reviewed, per the PRD full-coverage mandate, even where a chapter contributes no new terminology.

The primary interpretive lens throughout is the French religious landscape documented in the baseline: a majority nominal/lapsed Catholic population, a small Reformed/evangelical Protestant minority, a biblically low-literacy secular majority, a strong laïcité (state secularism) civic culture, and a live folk-superstition substrate (curses, the “evil eye,” astrology/voyance) that surfaces unexpectedly around Numbers’ cultic and prophetic material in ways Romans never required us to address.

---

2. Doctrine Vocabulary Matrix

Doctrine	Available French Terms	Weaknesses in Available Terms	Recommended Strategy
Unbelief and the Wilderness Generation	incrédulité, refus de croire, murmurer (contre), perdre courage, désobéissance	”Incrédulité” in general French usage skews toward intellectual skepticism (a modern agnostic’s doubt) rather than covenant treason against a God who has already proven himself faithful (manna, deliverance from Egypt). “Murmurer” (to murmur/grumble) is available and etymologically apt (cf. Segond’s traditional “murmures d’Israël”) but risks sounding like petty complaint rather than rebellion.	Use “incrédulité” for the theological category but consistently pair it with covenant-history framing (“incrédulité malgré les preuves de la fidélité de Dieu”) so it is never read as neutral doubt. Retain “murmurer contre” as the narrative verb (chs. 14, 16, 20, 21) but gloss its rebellion-force in teaching notes, distinct from ordinary grievance (“se plaindre”).
God’s Patience and Judgment	patience, lent à la colère (fixed idiom), jugement, colère de Dieu, châtiment	”Patience” alone in secular French can read as a virtue of temperament rather than covenantal forbearance rooted in God’s own character (Exodus 34:6-7). “Jugement” is stable but “colère de Dieu” risks either (a) an anthropomorphic caricature dismissed by secular readers, or (b) in Catholic devotional culture, being softened by a a priori assumption that divine mercy always trumps divine justice without holding both in tension as ch.14 and 16 do.	Keep the baseline-style fixed idiom “lent à la colère” (already established for Exodus 34:6-7 register) as the anchor phrase; always present patience and judgment as two expressions of the same covenant holiness (ch.14:18 model), never patience as leniency or judgment as capricious rage.
The Bronze Serpent as a Type of Christ	serpent de bronze, regarder, guérison, salut, signe/type, figure (typologique)	French theological vocabulary for “type/typology” itself is thin outside academic/seminary registers — “figure” or “préfiguration” are the natural options but are unfamiliar to a lay Catholic-secular readership. Risk of the physical object (the serpent) becoming the teaching’s center of gravity rather than the act of God-appointed looking-in-faith, mirroring the historical Nehushtan/2 Kings 18:4 trajectory.	Use “préfiguration du Christ” (rather than the more academic “type”) as the lay-accessible typological term, glossed on first use. Every occurrence of “serpent de bronze” in core-passage teaching material must be paired with an explicit statement that the object itself held no power (cf. Term Registry bronze_serpent note) — this is a mandatory co-occurring clause, not an optional footnote.
Intercession (Moses, Aaron, Phinehas)	intercession, intercéder, médiation, plaider (la cause de), prière	”Intercession” is a Critical-risk baseline collision (per bible_term_registry.json): French Catholic devotional culture’s dominant referent for “intercession” is the intercession of departed saints and Mary, invoked by name in prayer. Numbers supplies three named, narratively vivid human intercessors (Moses, Aaron, Phinehas), which — unlike Romans’ more abstract treatment — gives this collision unusual concrete traction.	Every occurrence of “intercession/intercéder” applied to Moses, Aaron, or Phinehas in teaching material must carry an explicit redemptive-historical qualifier (e.g., “intercession unique et médiatrice, propre à sa charge dans l’histoire du salut, qui annonce l’intercession supérieure et définitive du Christ, Hébreux 7:25”) distinguishing it from the ongoing devotional practice of invoking saints. Do not let the term stand unglossed anywhere in this curriculum.
Faithfulness of God’s Promises despite Human Failure	fidélité, alliance, promesse, accomplir/accomplissement	No major lexical gap, but a framing risk: “fidélité” and “alliance” are stable, yet the specific Numbers argument (God keeps his promise to the nation as a corporate whole even while judging individual unbelief within it) can be flattened into either (a) a simple “God is nice” therapeutic reading, or (b) an Israel-as-modern-nation-state political reading given France’s Jewish community and Middle East news salience (already flagged for “Israël” in the baseline).	Frame consistently as corporate covenant faithfulness surviving generational failure — the land-promise (chs. 26–36) is fulfilled to the next generation, not the rebellious one; this is the load-bearing point distinguishing Numbers’ contribution from a generic “God is faithful” slogan. Keep “Israël” historically/theologically bounded per baseline convention.
Census and the People of God	recensement, dénombrement, communauté, peuple de Dieu, appartenance	”Recensement” is administratively cold (INSEE, tax and military registration connotations) and carries zero devotional resonance in contemporary French; “dénombrement” is the more literary/biblical alternative (title of Numbers itself in some French Bible traditions: “Les Nombres” / “Le Livre des Dénombrements”) but is a passive, bookish register unfamiliar to secular readers.	Use “recensement” as the primary working term (matches modern comprehension) but introduce “dénombrement” as the traditional French Bible book-title register in the introductory unit, explicitly bridging the two so students recognize both forms. Consistently attach the pastoral point (God numbers his people individually and by name, not as anonymous statistics) as a stated corrective to the term’s bureaucratic flatness.
Balaam and God’s Sovereign Blessing	bénédiction, malédiction, souveraineté (de Dieu), divination, prophète/prophétie, oracle	Two distinct risks converge here: (1) French folk-Catholic and secular culture retains real currency for “malédiction” as a manipulable, quasi-magical force (evil eye, family curses, esoteric practice — astrology, tarot, “voyance” are visible cultural phenomena in contemporary France); (2) Balaam himself is a genuinely ambiguous figure (a true prophet who also practices condemned divination, cf. 2 Peter 2:15-16, Jude 11), and French vocabulary does not cleanly distinguish “prophète” (legitimate) from “devin” (illegitimate) without deliberate authorial framing.	Frame every Balaam passage around God’s unilateral sovereignty overriding both pagan divinatory technique and the prophet’s own mixed character — “Dieu bénit qui il veut, et nul art divinatoire ne peut le forcer à maudire son peuple.” Never let “bénédiction/malédiction” stand as if they were symmetrical, manipulable forces available to any practitioner; God’s blessing/cursing is asymmetrically sovereign, unlike Balaam’s craft.

---

3. Missing Vocabulary Inventory (Lexical Gaps)

These are concepts for which standard French supplies no single ready equivalent, requiring compound phrases, idiom substitution, or explanatory paraphrase rather than a one-word rendering.

English Concept	Hebrew	Gap Description	Recommended French Handling
”short of soul” / impatience unto rebellion (21:4)	קָצְרָה נֶפֶשׁ / qatserah nephesh	No single French word captures the idiom’s blend of exhaustion, impatience, and incipient rebellion.	Compound phrase: “le peuple perdit courage” / “l’âme du peuple s’impatienta”; must not reduce to simple “fatigue” (physical tiredness only). See Term Registry impatience_of_soul.
Presumptuous/defiant sin as a distinct moral category from ordinary sin (ch.15)	זָדוֹן / zadon	French “péché” is a single undifferentiated noun; it has no built-in grammatical or lexical marker distinguishing “sin done in defiant contempt” from “sin done in weakness or error.”	Compound qualifier required every time: “péché commis avec présomption / délibérément, en mépris de Dieu” — never let bare “péché” carry this specific sense unqualified.
Unintentional sin as a legal-cultic category (chs.15, 35)	שְׁגָגָה / shegagah	Similarly, French has no single noun opposite to “présomption” in this legal sense; “involontaire” is an adjective borrowed from general tort/criminal vocabulary, not a dedicated theological term.	”Péché involontaire” / “péché par erreur,” consistently paired against “avec présomption” so the graduated-culpability system is visible as a contrastive pair, not two unrelated phrases.
Cherem — irrevocable devotion of persons/property to destruction as a unique covenant-judicial category (21:2-3; also chs. 31, 33)	חֵרֶם	No French word names this specific ancient Near Eastern legal-religious category; “anéantissement” or “destruction totale” are generic annihilation words that miss the vowed, sacral character of the act (it is closer to an offering than to ordinary warfare).	Use “vouer à la destruction” (verb phrase preserving the vow-structure) as primary rendering; reserve a dedicated explanatory footnote for first occurrence distinguishing this from a general theology or ethic of warfare.
Nes — pole/standard functioning simultaneously as physical support and as a rallying “sign” (21:8-9; cf. Exodus 17:15)	נֵס	French “perche” (pole) captures the physical object but entirely loses the “sign/rallying standard” resonance carried by the same Hebrew root and by the LXX’s σημεῖον.	Render “perche” in the base text but supply a mandatory footnote/teaching note bridging to “signe” (σημεῖον) and forward to John’s Gospel’s “signs” vocabulary — the gap is real and cannot be closed by word choice alone; it must be closed by annotation.
Dual verbs of sight — instructed “look” vs. enacted “gaze intently” (21:8 vs. 21:9)	רָאָה / הִבִּיט	French possesses many verbs of seeing (“voir,” “regarder,” “fixer,” “contempler”) but no single verb pair maps cleanly onto the Hebrew’s simple-instruction/intensified-response distinction without deliberate lexical selection.	Fix “regarder” for v.8 (command) and “fixer son regard sur” for v.9 (response) as an intentional, non-interchangeable pair — this is a constructed distinction, not a natural one, and must be maintained by editorial discipline across all Phase 2 output.
Rod/tribe wordplay (matteh, ch.17)	מַטֶּה	French “tribu” and “bâton” are unrelated words; the Hebrew pun (the leader’s staff of office is the same word as his tribe) is untranslatable.	Accept the loss in the running text; supply a mandatory footnote at ch.17 explaining the wordplay’s significance for the confirmation of Aaron’s priesthood, since the narrative’s persuasive logic depends on the reader knowing the pun exists.
Fused sexual-and-cultic betrayal (zanah, ch.25)	זָנָה	French offers either a sexual-morality vocabulary (“se prostituer,” “adultère”) or a religious vocabulary (“idolâtrie,” “apostasie”) but no single term fusing both as the Hebrew root does natively.	Use the doubled phrase “se prostituer et s’adonner à l’idolâtrie” every time, resisting the temptation to pick only one half — dropping either half changes what kind of sin the text says this is.

---

4. Crowded Semantic Neighborhoods Requiring Fencing

Unlike missing vocabulary (§3), these are cases where French has too much pre-existing, strongly-connotated material attached to the available word, requiring deliberate fencing (explicit contrastive framing) rather than lexical invention.

French Term	Competing/Interfering Association	Doctrine at Risk	Fencing Strategy
intercession / intercéder	Catholic devotional practice of invoking departed saints and Mary as intercessors	Intercession (Moses, Aaron, Phinehas)	Mandatory redemptive-historical qualifier on every occurrence (see §2 above); never leave unglossed. Elevated to Critical in the Numbers term registry — highest fencing priority in this book.
les saints / sainteté	Canonized, venerated intercessors (statues, feast days, patronage) — same collision the Romans baseline already documents for “saints,” but re-activated here by the Nazirite/holiness material (ch.6) and by the Aaronic “sanctifier” language (ch.20)	Separation unto God’s Service; Sanctification	Reuse the Romans baseline’s existing fencing convention exactly (clarify “tous les croyants consacrés,” never a venerated elite); no new strategy needed, but confirm the fencing is re-applied at every Numbers occurrence, not assumed already handled by the Romans material alone.
vœu / naziréen	Catholic monastic/religious-order “vœux perpétuels” (permanent vows of religious life)	Nazirite Vow (Separation unto God’s Service, extension)	Explicit teaching-note clarification at every occurrence: the Nazirite vow is voluntary, typically temporary, and available to any lay Israelite — structurally opposite to permanent clerical vows. Do not let “vœu” stand alone without this contrast.
grâce vs. bonté/fidélité (chesed)	Baseline reserves “grâce” exclusively for NT charis; Numbers 14:18 pairs God’s patience with chesed, a related but distinct OT category	God’s Patience and Judgment; Faithfulness of God’s Promises	Absolute rule: chesed is NEVER rendered “grâce” in this curriculum. Use “bonté” or “fidélité.” This is the single most consequential OT/NT terminological fence in the whole book — flattening it would retroactively blur the Romans baseline’s carefully guarded distinction.
malédiction / bénédiction	French folk-superstition and esoteric-subculture currency (evil eye, family curses, astrology, “voyance”)	Balaam and God’s Sovereign Blessing	Frame every occurrence around divine sovereignty overriding technique; never present blessing/cursing as symmetrical, ritually-manipulable forces. See §2 doctrine matrix above.
divination / devin	Visible, culturally normalized occult subculture in contemporary France (astrology columns, tarot, mediums)	Balaam and God’s Sovereign Blessing	Explicit condemnatory framing on every occurrence; Balaam’s technique is the text’s problem, not its model, even though God overrules it to bless Israel.
vengeur (du sang)	Popular-culture “les Vengeurs” (the Avengers) and general vigilante connotation	Cities of Refuge / Justice (background to Faithfulness doctrine)	Clarify as an ordered, legally-bounded judicial role restrained by the cities-of-refuge system, not extrajudicial vigilantism.
recensement	INSEE/state bureaucratic administrative registration; zero devotional resonance	Census and the People of God	Pair with “dénombrement” (traditional Bible-title register) and explicit pastoral corrective (God numbers his people by name); see §2.
communauté vs. assemblée vs. l’Église	Three distinct Hebrew/Greek referents (edah, qahal, NT ekklēsia) risk collapsing into a single French “the community/church” impression, especially given that LXX renders qahal with the very Greek word (ἐκκλησία) later used for the NT church	Census and the People of God; (background bridge to Church as God’s People, Romans baseline)	Maintain the three-way distinction taught in the core glossary: “communauté” (edah, standing covenant identity) / “assemblée” (qahal, a specific convened gathering, including Korah’s illegitimate one) / “l’Église” (NT-only, reserved). Flag the LXX bridge for advanced material only; keep basic curriculum text strictly OT-bounded.
expiation	No direct Catholic-liturgical collision (unlike “action de grâce”/Eucharist in the baseline), but risk of being read as a merely ritual or psychological “closure” rather than an action that turns away real, objective divine wrath	Intercession (Aaron, Phinehas); background to Bronze Serpent typology	Teaching material must state plainly that expiation in Numbers effects an objective change in God’s disposition of wrath toward the community, not a subjective sense of relief; this guards the term’s later NT theological freight.
rançon	Secular-criminal kidnapping-ransom association (news media, thrillers)	Priesthood/Firstborn Redemption (background to atonement typology)	Pair with “prix du rachat” and sacred-cultic framing on first occurrence; do not let the crime-news register dominate the reader’s first impression.

---

5. Transliteration vs. Paraphrase Decisions

Numbers, unlike Romans, contains a substantial layer of Hebrew cultic and legal vocabulary with no NT precedent in the baseline. Each such term requires an explicit transliterate-or-paraphrase decision.

Term	Decision	Rationale
YHWH (Tetragrammaton)	Paraphrase via established convention: “le SEIGNEUR” (small caps), “l’Éternel” acceptable in Segond-aligned material	Not transliterated (no French tradition renders יהוה phonetically, e.g., “Yahvé,” in mainstream liturgical/pedagogical use, though “Yahvé” does exist in some Jerusalem Bible editions — reject for this curriculum to maintain TOB/BJ ecumenical register consistency and to preserve the deliberate root-bridge to the Romans baseline’s NT “Seigneur”).
Abba (already in baseline)	Transliterated, per baseline	Not applicable to Numbers directly, cited here only to confirm consistency: no new Numbers term should be transliterated using intimacy-preservation logic unless it meets the same bar of untranslatable devotional intimacy — none in Numbers does.
Manne / manna	Loanword retained (“manne”)	Already a fully naturalized French loanword; paraphrase (“pain miraculeux”) would be a regression from established Bible-French usage.
Pâque / Passover	Loanword retained (“la Pâque”)	Fully established; paraphrase unnecessary and would break cross-reference consistency with the rest of the francophone Bible tradition.
Naziréen / Nazirite	Transliterated with gloss (“naziréen”; “vœu de naziréen”)	No natural French paraphrase exists that would not itself require explanation (e.g., “consacré spécial temporaire” is clumsier and less recognizable); transliteration plus the mandatory fencing note (§4) is preferred over invented paraphrase.
Cherem	Paraphrase, not transliterated (“vouer à la destruction”)	Unlike “naziréen,” this term has no devotional/pedagogical value in transliterated form for a lay French audience; a transliterated “cherem” would read as jargon with no anchor, whereas the verb-phrase paraphrase at least signals the vow-structure. Reserve the Hebrew transliteration for theologian-review footnotes only.
Nechushtan (the bronze serpent’s later idol-name, 2 Kings 18:4, background reference)	Transliterated when referenced, with mandatory explanatory gloss	This proper name has no French equivalent and appears only in background/cross-reference material (not the core passage itself); when cited to warn against relic-veneration misreadings, transliterate as “Nehushtan” and gloss immediately (“le nom que l’idole reçut plus tard”).
Qorban-type sacrificial vocabulary generally (chs. 15, 28-29)	Paraphrase via established French cultic vocabulary (“offrande,” “sacrifice,” “holocauste,” “offrande de gâteau/de libation” per established Segond/TOB sacrificial terminology)	These already have stable, non-controversial French Bible-tradition renderings; no new transliteration decision needed, flagged here only to confirm no gap exists.
Nes (pole/sign)	Paraphrase (“perche”), NOT transliterated	A transliterated “nes” would be unrecognizable and pedagogically useless to a French lay reader; the typological “sign” resonance is better carried by footnote/cross-reference (§3) than by an opaque transliteration.
Qin’ah (jealousy/zeal)	Paraphrase, context-split (“jalousie” ch.5 / “zèle” ch.25)	Already established in the core glossary; confirmed here as correct — a single transliteration would erase the deliberately different moral valence the two French words correctly signal.

General principle for this book: transliterate only where (a) the term names a distinct, bounded institution or proper name for which no French paraphrase exists without becoming a mini-essay (Nazirite vow, Nehushtan), or (b) the term is already a fully naturalized loanword in the French Bible tradition (manna, Pâque). Paraphrase everywhere else, especially where a transliteration would produce opaque jargon carrying no independent meaning to a biblically low-literacy reader — the dominant profile of this curriculum’s audience per the baseline.

---

6. Ranked List of Highest-Risk Ambiguities

Ranked by combined severity (doctrinal weight × likelihood of misreading in the French context), for Phase 2 review-routing priority.

1. Intercession (Moses/Aaron/Phinehas) collapsing into “intercession des saints.” Elevated to Critical specifically because Numbers, unlike Romans, supplies three named, narratively concrete human mediators — the strongest possible on-ramp to conflation with Catholic saint-intercession devotion. Mandatory theologian review on every occurrence.
2. Bronze serpent read as a relic/talisman rather than a sign pointing to God-appointed faith. The text’s own internal trajectory (2 Kings 18:4, Hezekiah destroys it as an idol) proves this is not a hypothetical risk but the very failure mode the canon itself records. Direct typological stakes for John 3:14-15. Mandatory theologian review.
3. Chesed rendered as “grâce,” collapsing the OT covenant-loyalty category into the baseline’s reserved NT term. Would retroactively damage the Romans baseline’s carefully maintained justification/grace distinctions by suggesting OT “chesed” and NT “charis” are interchangeable. Absolute prohibition required (§4).
4. YHWH/“le SEIGNEUR” decision inconsistently applied across Phase 2 workers, breaking the deliberate root-continuity bridge to the Romans baseline’s Christological “Seigneur” (Romans 10:9) and risking a flattened, generic-deity impression in a secular readership already prone to treat “Dieu”/“Seigneur” as abstract. Requires fixed, version-controlled convention (per translation memory practice in the baseline).
5. Balaam’s blessing/cursing narrative absorbed into French folk-magical “malédiction” currency rather than taught as divine sovereignty overriding all technique — a live risk given contemporary esoteric-subculture visibility in France (astrology, tarot, voyance are mainstream cultural phenomena, not marginal).
6. Cherem (holy-war devotion to destruction) applied as a template for contemporary religious violence rather than framed as a closed, unique redemptive-historical episode — acute sensitivity risk in current French public discourse around religiously-motivated violence.
7. Presumptuous sin (zadon) flattened into ordinary “péché” losing the graduated-culpability distinction that structures ch.15’s entire legal logic and explains why the wilderness generation’s rebellion is treated as unpardonable within that generation.
8. The two verbs of sight (21:8 instruction vs. 21:9 enacted faith) flattened to a single French “regarder,” erasing the exact narrative move (bare instruction → embodied trust) that carries the passage’s “look and live” gospel typology into Phase 2 lesson-writing.
9. “Recensement” read as flat civic bureaucracy, draining the pastoral point of chs. 1-4 and 26 (God numbers his people by name, not as anonymous statistics) — lower doctrinal stakes than items 1-8 but high likelihood of occurring unnoticed, since it requires no forbidden-substitution flag to slip through.
10. Nazirite vow assimilated to Catholic permanent religious vows, producing a category error (temporary/lay vs. permanent/clerical) that could misinform broader teaching about the priesthood-of-all-believers material this curriculum will eventually connect to Romans.
11. “Vengeur du sang” read through a vigilante/pop-culture lens, obscuring the ordered, restrained, cities-of-refuge-bounded nature of the office — moderate doctrinal stakes, moderate likelihood.
12. Fused sexual-cultic sense of zanah (ch.25) reduced to only one half (moralism or cultic-only reading), weakening the Baal of Peor narrative’s specific argument and its connection to Phinehas’ zeal (item elevated further by intersection with risk #1).

---

7. Full-Book Chapter Coverage Ledger (Numbers 1–36)

Confirms every chapter has been reviewed for terminology and doctrine relevance, per the PRD full-coverage mandate. Chapters contributing no new terms beyond those already fixed in §§2–5 are noted explicitly as reviewed rather than omitted.

Chapters	Content Summary	Coverage Status
1–4	First census; tribal arrangement; Levite duties	Reviewed. Core terms: recensement, communauté, tribu. No new doctrinal terms beyond Census matrix (§2).
5	Jealousy ordeal; restitution law	Reviewed. Core term: jalousie (qin’ah, contrastive pairing with ch.25’s zèle, §3/§4).
6	Nazirite vow; Aaronic (priestly) blessing	Reviewed. Core terms: naziréen/vœu, grâce/faire grâce (chen — positive OT/NT bridge). Fencing required (§4).
7–8	Tribal offerings for the tabernacle; consecration of Levites	Reviewed. Reinforces sanctification (baseline reuse); no new terms.
9	Passover in the wilderness; the cloud and fire guide the camp	Reviewed. Core term: la Pâque (loanword, §5).
10	Silver trumpets; the march begins	Reviewed. No new doctrinal terms; narrative transition only.
11	Complaint over manna; quail; the Spirit distributed to seventy elders	Reviewed. Core terms: manne, task-specific Esprit (background to Spiritual Gifts doctrine cluster, flagged in core glossary Part A).
12	Miriam’s challenge to Moses; leprosy as judgment	Reviewed. Core term: lèpre (§4 fencing on modern-disease conflation). Background to Intercession/unique mediatorial authority doctrine.
13–14	The spies; the nation’s unbelief; Moses’ intercession; the “slow to anger” formula	Core doctrinal chapters. Terms: foi/croire (unbelief negated), lent à la colère, bonté/fidélité (chesed, §4 absolute fence), intercession (§6 rank 1). Mandatory theologian review.
15	Case law on offerings; presumptuous vs. unintentional sin; tassels	Reviewed. Core terms: avec présomption (zadon), péché involontaire (shegagah), franges (tsitsit) — §3 missing-vocabulary entries.
16	Korah’s rebellion; the illegitimate assembly; Aaron’s atoning intercession	Reviewed. Core terms: assemblée (qahal, §4 three-way fence), expiation (kipper), intercession. Background to Intercession doctrine.
17	Aaron’s budding rod confirms the priesthood	Reviewed. Core term: bâton/tribu wordplay (§3 missing-vocabulary, footnote required).
18	Levitical/priestly duties and provision; Levites’ inheritance is the LORD himself	Reviewed. Core terms: sacerdoce, héritage (nachalah), alliance. Background to Faithfulness and Census doctrines.
19	Red heifer purification rite	Reviewed. Core term: vache rousse (§4 fencing against “primitive superstition” misreading). Background to Bronze Serpent/atonement typology cluster.
20	Water from the rock; Moses fails to “sanctify” the LORD; Aaron’s death; Edom refuses passage	Core doctrinal chapter. Terms: saint/sanctifier (qadash, public-honor sense), gloire (kavod). Background to God’s Patience and Judgment doctrine (Moses himself disciplined for unbelief/dishonor).
21:1-3	Victory over Arad; the cherem vow	Reviewed. Core term: vouer à la destruction (§3/§6 rank 6).
21:4-9	CORE PASSAGE: Impatience, rebellion, fiery serpents, Moses’ intercession, the bronze serpent	Anchor passage — full term coverage in §§2–6. All Critical/High terms (bronze serpent, pole/sign, dual verbs of sight, fiery serpents, impatience-of-soul idiom, “parler contre,” “péché”/confession) fixed and cross-referenced against ranked risks §6, items 2 and 8 especially.
21:10-35	Continued journey; victories over Sihon and Og	Reviewed. No new doctrinal terms; reinforces Faithfulness-of-Promises land-conquest background.
22–24	Balaam’s oracles; the sovereign, uncursable blessing of Israel	Core doctrinal chapters. Terms: divination/devin, bénir/maudire, ange de l’Éternel, élection (light use). §6 rank 5. Mandatory theologian review.
25	Baal of Peor; Phinehas’ zeal and covenant of perpetual priesthood	Core doctrinal chapter. Terms: se prostituer/s’adonner à l’idolâtrie (zanah), zèle (qin’ah, contrastive with ch.5), alliance, expiation. Background to Intercession doctrine (§6 rank 1, 12).
26	Second census	Reviewed. Reinforces recensement/dénombrement (§2/§4) and héritage (land allotment by census figures) — direct link to Faithfulness-of-Promises doctrine (the next generation inherits).
27	Daughters of Zelophehad; Joshua commissioned	Reviewed. Core term: héritage reinforced; leadership succession supports Faithfulness doctrine.
28–29	Calendar of offerings and feasts	Reviewed. Established sacrificial vocabulary (§5); no new gap terms.
30	Vows, especially women’s vows	Reviewed. Core term: vœu (neder), same fencing as Nazirite vow (§4).
31	War against Midian; vengeance for Baal of Peor	Reviewed. Core term: vengeance (naqam, §4 fencing against retaliatory-ethic misreading).
32	Reuben, Gad, and half-Manasseh’s land request	Reviewed. Reinforces héritage/tribu; no new terms.
33	Itinerary review; command to destroy idols in the land	Reviewed. Core term: idoles (gillulim, §2 background to Faithfulness/covenant exclusivity).
34	Boundaries of the land	Reviewed. Reinforces héritage; no new terms.
35	Cities of refuge; the avenger of blood; unintentional vs. deliberate homicide	Reviewed. Core terms: villes de refuge, vengeur du sang (§4 fencing against vigilante misreading), péché involontaire reinforced.
36	Zelophehad’s daughters’ inheritance secured within their tribe	Reviewed. Reinforces héritage; closing confirmation of Faithfulness-of-Promises doctrine (the land-promise is preserved intact to the new generation).

---

8. Summary Recommendations for Phase 2 Preparation

1. Load all three registries before any Numbers segment translation: Romans translation_memory.json (unmodified baseline authority), the Numbers extension of bible_term_registry.json, and this gap analysis, in that priority order.
2. Absolute prohibitions carried forward into Phase 2 validation rules: chesed → never “grâce”; cherem, zadon, shegagah → never bare “péché”; bronze serpent → never presented without the no-inherent-power clause; intercession (Moses/Aaron/Phinehas) → never left unglossed against saint-intercession devotion.
3. New Critical-tier terms requiring the same theologian-review discipline the baseline applies to justification/Sonship: le SEIGNEUR (YHWH), serpent de bronze, intercession (Numbers-elevated).
4. Fixed contrastive pairs to preserve mechanically across all Phase 2 output (do not let any single worker “simplify” these): regarder/fixer son regard sur (21:8/21:9); jalousie/zèle (qin’ah); avec présomption/péché involontaire (zadon/shegagah); communauté/assemblée/l’Église (edah/qahal/ekklēsia).
5. Footnote-mandatory terms (gap cannot be closed by word choice alone): pole/sign (nes), rod/tribe wordplay (matteh, ch.17), Nehushtan cross-reference.
6. This gap analysis and the accompanying core glossary should be merged into the Numbers translation-memory extension in Phase 1 Step 9+, incrementing the baseline’s version number per the existing enforcement protocol in 12_ai_translation_requirements.md.