Linguistic Gap Analysis
Linguistic Gap Analysis: Hosea — English → French
0. Purpose and Method
This step maps where French vocabulary is adequate, missing, or dangerously crowded for the five Hosea doctrines, and issues concrete strategy decisions (retain, qualify, gloss, transliterate, fence) that Phase 2 translators must follow. It builds directly on the Romans baseline translation_memory.json (reused, never re-derived) and extends it with Hosea-specific findings already catalogued in 08_core_glossary.md and assets/bible_term_registry.json. This analysis adds the why and the comparative strategy reasoning behind those entries, and ranks the resulting risks for Phase 2 routing.
Coverage confirmation: all 14 chapters of Hosea have been reviewed for vocabulary load. Chapters producing no new terms beyond those already catalogued are explicitly noted in §5 (Full-Book Coverage Log) rather than silently omitted.
1. Doctrine Vocabulary Matrix
1.1 God’s Steadfast Love despite Israel’s Unfaithfulness
| Element | Detail |
|---|---|
| Primary Hebrew terms | חֶסֶד (chesed), אָהַב/אַהֲבָה (ahav/ahavah), נְדָבָה-modified love (14:4) |
| Available French terms | amour, fidélité, grâce, tendresse, attachement |
| Weaknesses | (1) “Amour” alone is dangerously broad in French — covers romantic love, familial affection, and casual enthusiasm (“j’adore ça”), none of which carry chesed’s covenant-loyalty-despite-betrayal sense. (2) “Grâce” is already fixed by the Romans baseline as the rendering of χάρις — reusing it for chesed would blur two distinct lexical fields (NT unmerited-favor-in-justification vs. OT covenant-loyalty-in-marriage) across the curriculum suite, confusing a learner moving between Hosea and Romans. (3) “Fidélité” alone loses the affective, love-driven character of chesed, sounding merely dutiful. |
| Recommended strategy | Fence the term: fix “amour fidèle” as the single, dedicated French rendering of chesed, distinct from bare “amour” (ahavah, general/electing love) and from “grâce” (Romans’ χάρις). Use “fidélité” only for standalone אֱמוּנָה/אֱמֶת contexts (2:19-20 covenant-virtue list), never as a substitute for chesed. Reinforce with teaching-note glosses at first occurrence (2:19) and at the Matthew 9:13/12:7 cross-reference (6:6). |
| Doctrine risk | Critical |
1.2 Spiritual Adultery and Idolatry as Covenant Breaking
| Element | Detail |
|---|---|
| Primary Hebrew terms | זָנָה/זְנוּנִים (zanah), נָאַף (na’aph), בַּעַל (Ba’al, proper + common noun), פְּסִלִים/עֲצַבִּים (idols) |
| Available French terms | prostitution, adultère, infidélité, idole(s), Baal |
| Weaknesses | (1) “Prostitution” in contemporary French carries strong socio-legal/sex-work associations (debates on travail du sexe, trafficking) that pull attention toward a modern social-justice frame rather than the covenant-theological metaphor. (2) “Infidélité” is too mild/generic — usable for a minor lapse, not Hosea’s programmatic covenant rupture; it also risks sounding like ordinary marital counseling vocabulary rather than prophetic indictment. (3) “Idole” is heavily colonized by celebrity-culture usage (“une idole de la chanson/du cinéma”) in secular French media, risking trivialization exactly where Hosea intends maximal gravity. (4) The Hebrew pun ba’al (“Baal,” the god) / ba’al (“husband/master,” common noun) has no French lexical equivalent — this is a genuine untranslatable gap, not merely a weak-fit term. |
| Recommended strategy | Retain “se prostituer / prostitution” and “adultère” as the primary renderings (per baseline glossary) but require a standing teaching-apparatus gloss at first occurrence in ch. 1-3 clarifying the term functions as covenant-breaking metaphor, not a comment on sex work. Do not downgrade to “infidélité.” For the Baal pun (2:16), mandate a translator’s footnote at every co-occurrence of “Baal” (proper noun) and “mari/maître” (common noun) since the wordplay is structurally load-bearing for the doctrine and cannot be reproduced lexically. |
| Doctrine risk | High |
1.3 The Marriage Metaphor for God and His People
| Element | Detail |
|---|---|
| Primary Hebrew terms | אֵרַשְׂתִּיךְ (aras, betroth), אִישׁ/בַּעַל (husband/master), גֹּמֶר (Gomer), covenant-virtue list (2:19-20) |
| Available French terms | fiancer, épouser, mari, femme, alliance |
| Weaknesses | (1) “Fiancer” in modern French denotes a purely sentimental, easily-broken pre-marital arrangement (une bague de fiançailles, un simple projet); it loses the legally binding covenant character of biblical betrothal. (2) The narrative device itself — God commanding a prophet to marry a specific characterized woman as an enacted parable — has no ready French idiom and risks being read as biographical scandal rather than prophetic sign-act unless framed explicitly. (3) “Alliance” (already fixed by baseline for covenant) usefully doubles as “wedding ring” in French, which is an asset, not a gap — but must be flagged so translators exploit rather than lose this resonance. |
| Recommended strategy | Use “fiancer / épouser en alliance” and immediately anchor it with the accompanying covenant-virtue list in 2:19-20 (justice, droiture, amour fidèle, compassion, fidélité) so the legal weight is supplied contextually rather than lexically. Frame the Gomer narrative (ch. 1, 3) with explicit sign-act teaching language (“acte prophétique,” “parabole vécue”) to prevent a merely biographical reading. Exploit the “alliance = wedding ring” resonance explicitly in teaching notes as a natural bridge already present in French. |
| Doctrine risk | High |
1.4 Judgment and Compassion Held Together
| Element | Detail |
|---|---|
| Primary Hebrew terms | אַף/חָרוֹן (anger), רַחֲמִים/רָחַם (compassion), נֶהְפַּךְ עָלַי לִבִּי (heart recoils, 11:8), קָדוֹשׁ (Holy One, 11:9) |
| Available French terms | colère, compassion, cœur bouleversé, saint |
| Weaknesses | (1) French “colère” without intensification reads as mundane irritation, far weaker than חָרוֹן אַף (“burning nostril[s]”); risks flattening the theological weight of divine wrath into a passing mood. (2) “Compassion,” conversely, if rendered too softly (“de la pitié,” “un peu triste pour eux”) risks the opposite error — reading as sentimentality rather than a a decisive, costly divine choice that overrides deserved destruction. (3) The pairing itself is the central interpretive difficulty: French readers, especially those shaped by either strict legalistic religious formation or by sentimental therapeutic culture, tend to resolve the tension in one direction (God is either simply just or simply nice) rather than holding both, as 11:8-9 demands. (4) “Le Saint” as a divine title (11:9) collides with French Catholic devotional vocabulary for canonized saints (see §3.1 below), the same collision the Romans baseline already flags for “les saints.” |
| Recommended strategy | Intensify wrath language where the Hebrew idiom demands it (“colère ardente,” “courroux,” not bare “colère”). Keep “compassion” but always co-present it in translation notes with 11:9’s safeguard clause (“car je suis Dieu, et non un homme”) so the pairing is taught as a unity, not sequential alternation. Render “le Saint” with mandatory apposition (“le Saint, qui habite au milieu de toi”) at every occurrence to block the canonized-saint reading. |
| Doctrine risk | High |
1.5 The Call to Return and Repentance
| Element | Detail |
|---|---|
| Primary Hebrew terms | שׁוּב (shuv, return), מְשׁוּבָה (meshuvah, apostasy, same root), the model confession of 14:2-3 |
| Available French terms | revenir, retourner, se repentir, apostasie |
| Weaknesses | (1) French has two near-synonymous verbs (“revenir” vs. “retourner”) for shuv, and translators may unconsciously alternate between them for stylistic variety — which would destroy the deliberate single-root repetition that gives the book its verbal backbone (used at least ten times across chs. 2-14). (2) “Se repentir” is available and doctrinally accurate for the relational/covenantal sense, but if used instead of “revenir/retourner” at some occurrences and not others, the wordplay linking geographic return (11:5a, “ils ne retourneront pas en Égypte”) to covenantal return (11:5b, “ils ont refusé de revenir à moi”) within the very same verse becomes invisible. (3) “Apostasie,” the standard French rendering of meshuvah, shares no visible root with “revenir/retourner,” so the Hebrew’s ironic wordplay (the “turning away” noun is built on the same root as the “turning back” verb) is structurally unrecoverable in French without a footnote. |
| Recommended strategy | Mandate a single fixed verb choice (“revenir” as primary, “retourner” permitted only where French syntax requires it for a directly repeated phrase) for every occurrence of shuv across all 14 chapters, with a Phase 2 automated consistency check (regex/concordance scan) before sign-off. Retain “apostasie/égarement” for meshuvah but require a standing footnote at 11:7 and 14:4 noting the shared Hebrew root, so the resolution at 14:4 (“je guérirai leur apostasie”) reads as answering 11:7 rather than introducing an unrelated idea. |
| Doctrine risk | Critical (governing verb of the whole book; errors here damage the book’s structural readability, not just a single passage) |
2. Missing Vocabulary (True Lexical Gaps)
These are concepts for which no existing French word or fixed phrase captures the Hebrew concept; a compound phrase, gloss, or footnote is structurally required, not merely stylistically preferred.
| Gap | Hebrew | Why French Lacks a Direct Equivalent | Required Strategy |
|---|---|---|---|
| Chesed as a single word | חֶסֶד | French has no single noun combining “covenant loyalty” + “love” + “mercy despite betrayal” in one lexeme (unlike English “steadfast love” as an established biblical-studies compound). | Fixed compound: “amour fidèle.” Never single-word. |
| Ba’al husband/idol pun | בַּעַל | No French word means both “husband/master” and “the god Baal”; the pun is purely a Hebrew lexical accident. | Footnote required at every co-occurrence (2:16 esp.); cannot be solved lexically. |
| Shuv/meshuvah shared root | שׁוּב / מְשׁוּבָה | ”Revenir” and “apostasie” share no French root. | Footnote at 11:7, 14:4 flagging the Hebrew wordplay; cannot be reproduced lexically. |
| Da’at (relational “knowing”) | דַּעַת | French “connaissance” defaults to intellectual/informational register; the marital-intimacy sense (same root as Genesis 4:1’s “knew his wife”) has no default French equivalent carrying both senses simultaneously. | Teaching-note gloss required each occurrence (4:1, 4:6, 6:6); no lexical fix available. |
| Nedavah-modified love (14:4) | אֹהֲבֵם נְדָבָה | No single French adverb captures “freely, as an unsolicited gift, with no compensating merit required, using freewill-offering vocabulary.” | Compound: “je les aimerai librement, sans qu’ils l’aient mérité” or “gratuitement”; must NOT reuse “grâce” (reserved for Romans’ χάρις). |
| Jezreel’s double meaning | יִזְרְעֶאל | ”God sows” and “God scatters” are opposite meanings from the same root in Hebrew; no French word carries both. | Transliterate + dual gloss at every occurrence (1:4-5, 11; 2:22-23). |
| Sheol as neither Hell nor a neutral “place of the dead” in the modern sense | שְׁאוֹל | French “enfer” imports Dante/NT eternal-punishment categories absent from She’ol; “séjour des morts” is accurate but not intuitive to secular or Catholic readers who default to “enfer” for any afterlife-of-the-wicked term. | Use “séjour des morts” (Segond-tradition descriptive) with a footnote distinguishing it from “enfer”; theologian sign-off required (see §4). |
| Agrarian imagery (yoke on jaws, threshing, dew) | עַל לְחֵיהֶם; טַל; etc. | Not a lexical gap but a cultural-experiential gap: urban francophone readers (metropolitan France especially) may lack the tacit farming knowledge that made these images self-explanatory to Hosea’s audience. | Brief explanatory gloss in teaching apparatus (not the translated text itself), at 11:4 and 14:5. |
3. Crowded Semantic Neighborhoods Requiring Fencing
A “crowded neighborhood” is a French word already carrying a strong competing sense (secular, Catholic-devotional, or from the Romans baseline) that will pull the Hosea reading in the wrong direction unless deliberately fenced off.
3.1 “Saint(s)” / “le Saint” — Catholic devotional collision
- Existing baseline flag: Romans 1:7 “saints” (High risk) already documents that ordinary and Catholic French usage of “les saints” evokes canonized intercessors.
- Hosea’s new exposure: 11:9’s divine title “the Holy One” (קָדוֹשׁ), rendered “le Saint,” triples the exposure to this same collision, now applied directly to God’s own person rather than to believers.
- Fencing strategy: Mandatory apposition every time — “le Saint, qui habite au milieu de toi” — never “le Saint” standing alone as a bare title.
3.2 “Grâce” — reserved territory from the Romans baseline
- Collision risk: Translators trained on the Romans baseline may reflexively reach for “grâce” whenever they encounter an unmerited-favor concept, since that is chesed’s and nedavah’s closest functional analogue.
- Fencing strategy: Explicit prohibition in the Phase 2 system prompt (see recommendation in §6): “grâce” is reserved exclusively for χάρις-lineage NT contexts per the Romans baseline; chesed → “amour fidèle”; nedavah-modified love (14:4) → “gratuitement/librement.” Document this fence in translation memory notes for both new Hosea entries.
3.3 “Élection” — political-electoral collision (baseline precedent)
- Existing baseline flag: Romans’ ἐκλογή/“élection” is High risk because French “élection” primarily denotes democratic elections.
- Hosea’s relevance: Hosea 11:1 (“I loved him… I called my son out of Egypt”) and the whole book’s election-typology background (Israel as chosen, ch. 11-12) touch the same territory even though “élection” is not the lexical choice here (baseline glossary correctly avoids it, using “aimer/amour” instead). This is a fence-avoided-successfully case worth noting explicitly so Phase 2 translators do not introduce “élection” as a stylistic variant for “chosen”/“called” language in ch. 11.
- Fencing strategy: Prohibit “élection” as a rendering anywhere in Hosea; use “aimer/amour” (11:1) and “appelé” (baseline term, reused) instead.
3.4 “Idole(s)” — celebrity-culture collision
- Collision risk: French media usage (“une idole de la jeunesse,” televised talent-show titles like Nouvelle Star, historically Popstars) has made “idole” a term of pop-culture aspiration, nearly positive in connotation, the inverse of Hosea’s intent.
- Fencing strategy: Teaching apparatus must explicitly reframe “idoles” in its cultic sense at first occurrence (4:17) and re-anchor at each major idolatry passage (8:4; 11:2; 13:2) rather than assuming the covenant-breaking gravity carries automatically.
3.5 “Prostitution” — socio-legal collision
- Collision risk: Contemporary French public discourse on prostitution (abolitionist vs. regulationist debate, the 2016 French law criminalizing clients) is a live, politically charged topic; readers may import this frame and either (a) feel the metaphor moralizes real sex workers, or (b) get distracted from the covenant point entirely into a social-policy debate.
- Fencing strategy: Explicit metaphor-signaling gloss at 1:2 and 2:2 (“cette image désigne l’infidélité spirituelle d’Israël envers Dieu, non un jugement porté sur des personnes réelles”); never let the term stand as bare social commentary.
3.6 “Prêtre” — Catholic clergy collision
- Collision risk: French “prêtre” is the standard word for a Catholic (and Orthodox) ordained priest; readers will default to that image for Hosea’s indictment of the kohanim (4:4-9; 5:1; 6:9).
- Fencing strategy: Clarify on first occurrence (4:4) that this refers to the Old Testament Levitical priesthood, historically and functionally distinct from the Catholic priesthood, to avoid anachronistic transfer of the indictment onto contemporary clergy.
3.7 “Colère” — mundane-irritation collision
- Collision risk: Everyday French “colère” (“il est en colère” = he’s annoyed) is far weaker than חָרוֹן אַף; unintensified, divine wrath reads as a passing huff rather than righteous covenant judgment.
- Fencing strategy: Require intensified collocations (“colère ardente,” “courroux”) wherever the Hebrew idiom is emphatic (5:10; 8:5; 13:11), reserving plain “colère” only for lower-intensity occurrences.
3.8 “Roi” and political-kingship irony — low collision risk, but note interplay with Romans’ “Royaume de Dieu”
- Minor cross-check: baseline’s “Royaume de Dieu” (Kingdom of God, Medium risk) should not be confused with Hosea’s plain historical “roi” (10:3; 13:10-11); no direct collision, but teaching notes should note the ironic contrast (Israel’s rejection of divine kingship results in enslavement to a human king) is a distinct rhetorical point from the NT Kingdom-of-God theme.
4. Transliteration vs. Paraphrase Decisions
| Item | Decision | Rationale |
|---|---|---|
| YHWH (Section A #2 of glossary) | Transliterate-as-convention: “le SEIGNEUR” (small caps, TOB/NBS ecumenical convention); noted alternate “l’Éternel” (Segond). Reject “Yahvé.” | Avoids collapsing the OT covenant personal name into the baseline’s NT κύριος title (“Seigneur,” unmarked). Small-caps typography is the only way to keep the two visually distinct in a curriculum where both Hosea and Romans appear. Requires human theologian sign-off before Phase 2 begins — highest-priority open decision in this analysis. |
| Symbolic names (Jezreel, Lo-Ruchamah/Ruchamah, Lo-Ammi/Ammi) | Transliterate + gloss, never translation-only or transliteration-only. | Translation-only would lose the proper-name character needed for the narrative (they are literally the prophet’s children’s names); transliteration-only would render the entire rhetorical point (judgment-name reversed to blessing-name) invisible to French readers with no Hebrew. |
| Gomer, Jacob, David, Ephraim, Assyria, Egypt, Admah | Transliterate using established French Bible proper-name conventions (reuse where baseline exists: David; new fixed forms otherwise: Gomer, Éphraïm, Assyrie, Égypte, Adma). | Standard cross-tradition practice; no translation option exists or is desirable for proper names. Admah requires an added descriptive gloss (“ville détruite avec Sodome”) due to low OT narrative literacy, but the name itself is transliterated, not translated. |
| Sheol | Paraphrase (descriptive): “séjour des morts,” not transliteration (“shéol”), and never “enfer.” | A theologian-level decision is still open between the descriptive Segond-tradition phrase and transliteration; this analysis recommends the descriptive phrase as the safer default given the curriculum’s mixed Catholic/Protestant/secular audience, since transliterated “shéol” would require its own glossing anyway and offers no clarity advantage. Flagged in §6 for confirmation. |
| Baal | Transliterate as proper noun (“Baal / les Baals”), with the common-noun pun (“mari/maître”) handled by footnote, not by any translation of the proper name itself. | The proper name must remain recognizable as the historical Canaanite deity; translating it (e.g., as “le maître”) would erase the specific idolatry being indicted. |
| Abba-style retained Hebrew/Aramaic terms (cf. baseline’s “Abba”) | No direct Hosea equivalent required — Hosea contains no comparable untranslated intimacy-term requiring retention. | Noted for completeness; no action needed. |
| ”Amour fidèle” for chesed | Paraphrase (fixed compound), not transliteration (no Hebrew loanword tradition exists for chesed in French Bible history, unlike “Abba” or “Amen”). | French Bible tradition (Segond, TOB, NBS) has never transliterated chesed; a compound descriptive phrase is the established and only viable path. |
| Veau (calf idol) | Translate plainly, exploiting the existing French cultural idiom “le veau d’or.” | This is the rare case where translation, not transliteration or paraphrase-with-gloss, is strictly superior: the target culture already possesses a ready-made, doctrinally correct association (Exodus 32) that a transliteration would forfeit. |
5. Full-Book Coverage Log
Confirms every chapter has been reviewed for vocabulary/gap analysis purposes, per the full-book coverage mandate.
| Chapter | New Vocabulary Load | Status |
|---|---|---|
| 1 | Symbolic names (Jezreel, Lo-Ruchamah, Lo-Ammi), zanah (wife of whoredom), Gomer | Reviewed — major gap items, catalogued §2, §4 |
| 2 | chesed/ahavah covenant-virtue cluster, betrothal (aras), Baal pun, name reversals | Reviewed — major gap items, catalogued §1.1, §1.3, §2 |
| 3 | Redemption narrative act (karah, buying back Gomer) | Reviewed — overlaps redemption vocabulary (§1.1, ch. 13); no new gap beyond glossary entry |
| 4 | Covenant lawsuit (riv), da’at Elohim, priesthood (kohen), ruach zenunim | Reviewed — crowded-neighborhood item (prêtre, §3.6); gap item (da’at, §2) |
| 5 | Pride (ga’own), political self-reliance, wrath intensification | Reviewed — no new gap beyond §1.4, §3.7 |
| 6 | chesed (6:6, the Matthew-quoted verse), healing (rapha), shuv | Reviewed — highest cross-curriculum-consistency priority item, §1.1 |
| 7 | Dove image (political vacillation), shuv, sword | Reviewed — cultural-bridge note only; no new lexical gap |
| 8 | Calf idol (egel/veau), sow-wind-reap-whirlwind idiom, torah | Reviewed — translation-advantage item (§4, veau) |
| 9 | zanah recurrence (9:1), no new terms | Reviewed — no new terms beyond ch. 1-4 catalogue; explicitly noted, not omitted |
| 10 | Thorn-and-thistle judgment imagery, divided heart, false balances, melek irony | Reviewed — Genesis 3 intertextual echo flagged for teaching notes only |
| 11 | CORE PASSAGE. Sonship typology, cords of tenderness, yoke-easing, shuv (dual sense within one verse), Most High (al), heart-recoils anguish, meshuvah, Holy One, “God not man,” “in your midst” | Reviewed in full depth — see §1.4, §1.5, §2, §3.1, §4; highest-density chapter in the book for translation risk |
| 12 | Jacob typology, false balances (economic sin), shuv | Reviewed — no new gap beyond Jacob proper-name entry (§4) |
| 13 | Predator judgment imagery (lion/leopard/bear), redemption/ransom (ga’al/padah), Sheol | Reviewed — Critical-risk item, §1.1, §2, §4 (Sheol) |
| 14 | Model confession (14:2-3), climactic healing/return resolution, nedavah-modified love, dew, lion-roar-as-summons | Reviewed — resolution chapter for §1.1 and §1.5 threads; roar-as-summons crowded-neighborhood note (predator imagery repurposed positively) |
6. Ranked List of Highest-Risk Ambiguities in This Book
Ranked by combined severity (doctrinal centrality × likelihood of mistranslation × breadth of downstream impact across chapters). All Critical/High items require human theologian review before Phase 2 segment translation of the relevant passages begins, per baseline routing conventions.
- YHWH rendering (“le SEIGNEUR” vs. “l’Éternel” vs. bare “Seigneur”) — Critical. Governs all 14 chapters (~45 occurrences); a wrong or inconsistent choice blurs the OT covenant name with the NT title for Christ across the entire curriculum suite. Must be resolved once, before any other Phase 2 work.
- Chesed (“amour fidèle”) vs. Romans’ “grâce” boundary — Critical. The single most theologically load-bearing term in the book (2:19; 6:6; 10:12; 14:4), directly quoted by Jesus; conflation with the baseline’s fixed “grâce” would corrupt cross-curriculum theological vocabulary for every learner who studies both Romans and Hosea.
- Shuv/meshuvah verb-root consistency across all 14 chapters — Critical. Not a single-verse risk but a structural risk: inconsistent verb choice destroys the book’s built-in resolution arc (11:7 apostasy → 14:1-4 return), which is the narrative backbone of “The Call to Return and Repentance.”
- Hosea 11:8’s “heart recoils” and 11:9’s “God not man” held together — High. The emotional and doctrinal climax of “Judgment and Compassion Held Together”; mistranslation in either direction (too flat = loses pathos; too intense = implies divine instability) directly damages the core passage’s central theological claim.
- “Le Saint” (11:9) and “les saints” (cross-curriculum) Catholic-devotional collision — High. Live collision already flagged in the Romans baseline; Hosea intensifies exposure by applying it to God’s own title rather than to believers, raising the stakes of an unglossed occurrence.
- Baal proper-name/common-noun pun (2:16) — High. Structurally untranslatable; omission of the required footnote silently erases the chapter’s central rhetorical argument (God as husband, not master/idol).
- “Prostitution”/“adultère” socio-legal vs. covenantal register (chs. 1-3, 9:1) — High. Politically charged contemporary French discourse risks redirecting reader attention away from the covenant metaphor entirely if the metaphor-signaling gloss is omitted.
- Sheol rendering decision (13:14) — High. Open decision between “séjour des morts” and transliteration; wrong choice (“enfer”) imports anachronistic NT/theological categories into the Hebrew text and weakens the intended 1 Corinthians 15:55 typological bridge.
- Most High / “upward” ambiguity (11:7, al) — High. Genuine textual ambiguity affecting whether the verse depicts futile invocation of a divine title or a spatial idiom; changes the theological content of the verse, not just its register.
- “Da’at Elohim” (knowledge of God) intellectual-vs-relational drift (4:1, 4:6, 6:6) — High. Secular French default reading (“connaissance” = information) undercuts the marital-intimacy resonance central to the book’s diagnosis of Israel’s failure.
- Cords of kindness / humane cords (11:4) — High. Genuine Hebrew ambiguity (“cords belonging to a man” vs. “human/humane cords”) within the core passage; affects how tenderly vs. how generically the verse is read.
- “Idole(s)” celebrity-culture trivialization (4:17; 8:4; 11:2; 13:2) — Medium-High. Widespread, low-friction misreading risk given how saturated French media culture is with positive “idole” usage; requires proactive teaching-apparatus correction rather than a lexical fix.
- “Colère” under-intensification of divine wrath (5:10; 8:5; 11:9; 13:11) — Medium. Systematic risk of flattening covenant judgment into mundane irritation if intensifying collocations are not consistently applied.
- “Prêtre” anachronistic Catholic-clergy transfer (4:4-9; 5:1; 6:9) — Medium. Contained risk, easily fenced with a first-occurrence clarification; does not recur with the same intensity at every occurrence.
7. Recommendations for Phase 2 System Prompt Integration
- Add an explicit “reserved-term fence” clause to the Hosea-specific system prompt (parallel to the Romans baseline’s forbidden-substitution list): “grâce” is reserved for χάρις-lineage NT contexts; never use it for chesed, ahavah, or nedavah-modified love.
- Add a YHWH typography rule: “le SEIGNEUR” must render in small capitals in all output formats that support it; plain “Seigneur” (no capitals) is reserved exclusively for κύριος/Christ per the Romans baseline.
- Add a shuv concordance check: before Phase 2 sign-off, run an automated scan confirming every instance of the return/repent verb uses the fixed root, flagging any deviation for reviewer attention.
- Add mandatory footnote triggers for: Baal pun (2:16), Jezreel (1:4-5,11; 2:22-23), Lo-Ruchamah/Ruchamah and Lo-Ammi/Ammi (1:6,9; 2:1,23), Sheol (13:14), Most High/upward ambiguity (11:7), shuv/meshuvah shared root (11:7; 14:4).
- Escalate to human theologian review (not merely native-speaker review) any segment touching: YHWH’s first fixed rendering, chesed first fixed rendering, Hosea 11:8-9 in full, the Sheol/ransom passage (13:14), and the Most High ambiguity (11:7) — consistent with this analysis’s ranked list above.
This analysis extends, and does not contradict, the Romans baseline Language Package. All fencing and transliteration decisions above must be written into an updated translation_memory.json and bible_term_registry.json at Phase 2 Step 16, with Critical/High items flagged for human theologian sign-off before use.