Linguistic Gap Analysis
Linguistic Gap Analysis: Psalms — English → French
Purpose
This document executes PRD Phase 1 Step 8 for the Psalms curriculum (destination language: French; core passage Psalm 23:1-6). It surveys the entire Psalter for places where French lacks a clean lexical match for a load-bearing Hebrew term (“missing vocabulary”), places where French already has a term but that term is contested by a crowded semantic neighborhood in the destination culture (competing religious, secular, or political senses), and places where the translator must choose between transliteration (retaining the source-language sound) and paraphrase/substitution (finding or constructing a French equivalent). It closes with a ranked list of the highest-risk ambiguities for Phase 2 routing.
This analysis governs and must not contradict the baseline Romans translation_memory.json / bible_term_registry.json conventions. Where Psalms vocabulary reuses a baseline term, that baseline rendering is treated as fixed; this document identifies where Psalms’ usage extends, narrows, or stresses the baseline term differently, and where wholly new French vocabulary decisions are required.
All eight curriculum doctrines are addressed. The core passage (Psalm 23:1-6) anchors the Character-of-God-as-Shepherd doctrine but is not the scope boundary; this analysis covers vocabulary drawn from across the full 150 psalms as required by the doctrine list.
A. Doctrine Vocabulary Matrix
| Doctrine | Available French Terms | Weaknesses in Available Terms | Recommended Strategy |
|---|---|---|---|
| Character of God as Shepherd, King, Refuge | berger; roi; refuge; rocher; forteresse; le Très-Haut; le Tout-Puissant; l’Éternel des armées | ”Berger” risks a sentimentalized, secularized rural-idyll reading (postcard shepherd) that loses the Ancient Near Eastern ruler-protector force. “Refuge” in contemporary French carries strong secular associations (mountain shelters, animal shelters, migrant asylum “centres de refuge”) that flatten God-as-refuge into a passive location rather than an active, personal, trust-based relationship. “Rocher” risks dying into the frozen idiom “solide comme un roc,” a mere adjective of solidity rather than a living, protective Person. “Roi” alone (bare) does not carry royal-covenant theology without added context. | Retain established renderings (berger, refuge, rocher, forteresse, le Très-Haut, le Tout-Puissant) but pair every first doctrinal use with a brief ANE-background gloss (king-as-shepherd; refuge as active reliance, not a place one stumbles upon). Reserve capital-conscious “l’Éternel des armées” for cosmic-sovereignty contexts and gloss against militaristic-nationalist misreadings given contemporary French political discourse. |
| Lament and Honest Prayer | jusqu’à quand; du fond de l’abîme; abandonner; la vallée de l’ombre de la mort | No lexical gap, but a register gap: French devotional culture (both Catholic and post-Christian secular) tends to treat direct complaint addressed to God as spiritually inappropriate, so even an accurate translation risks being read as doubt or irreverence rather than sanctioned covenant speech. “Abandonner” for Ps 22:1 must never be softened (e.g., toward “sembler loin”), since it is quoted verbatim by Christ. | Lexically straightforward; the required intervention is pedagogical framing (teaching notes normalizing lament as a modeled prayer genre), not lexical substitution. Escalate Ps 22:1 (“abandonner”) to Critical, theologian-reviewed status per the atonement doctrine it undergirds. |
| Praise and Worship | louer/louange; se prosterner/adorer; chanter/chant; Alléluia; Sélah; action de grâce | Low risk overall; the chief hazard is register drift — “louer” can sound like praising a person’s skill (as in “louer le talent d’un artiste”) if not consistently paired with worship-object clarity. “Sélah” and “Alléluia” have no translation gap since French Bible tradition already retains them untranslated/transliterated. | No new vocabulary required. Retain established renderings; ensure “louer” is consistently anchored to God as object across contexts to avoid a generic-compliment reading. |
| Messianic Prophecy in the Psalms | Messie; l’oint; la corruption; le séjour des morts; ils ont percé mes mains et mes pieds; la pierre angulaire; Hosanna/sauve je t’en prie; ton trône ô Dieu | Genuine dual-register gap: mashiach requires two distinct French renderings (l’oint historical / Messie canonical) with no single word covering both without anachronism or loss of NT typology. Ps 22:16 is a live Hebrew textual crux (pierced vs. “like a lion”), not a normal translation choice — French cannot resolve the underlying manuscript question, only choose which reading to render. Ps 45:6’s address (“ton trône, ô Dieu”) and Ps 82’s “dieux” (of judges) both risk being smoothed into unambiguous French in a way that erases doctrinally load-bearing ambiguity the New Testament itself exploits. | Apply the established dual-register mashiach rule context-by-context (see baseline registry “messiah_anointed”). Flag Ps 22:16, Ps 45:6, and Ps 82:1,6 explicitly as textual/exegetical cruxes requiring theologian review at every occurrence, not silently resolved translation choices. |
| Confession and Forgiveness (Psalm 51) | confession; transgression/révolte; faute; péché (baseline reuse); effacer; purifier/laver; un cœur pur; esprit brisé, cœur brisé; pardonner | Severe crowded-neighborhood risk: “la confession” in French overwhelmingly denotes the Catholic sacrament of Penance/Reconciliation for both practicing and lapsed Catholic readers — a majority-culture collision, not a minor connotation. Three near-synonyms (péché, transgression, faute) must be kept lexically distinct to preserve Psalm 51’s deliberate escalating sin-vocabulary; French devotional shorthand tends to collapse all three into “péché.” | Mandatory disambiguating gloss on every doctrinal/lesson-title use of “confession” (“la confession du péché devant Dieu,” non-sacramental, direct address). Enforce distinct, non-interchangeable French terms for péché / transgression (or révolte) / faute across all Psalm 51 teaching material; flag any drafted segment that flattens the triad. |
| The Righteous and the Wicked (Psalm 1) | le juste; le méchant/l’impie; voie; méditer; loi (torah, glossed) | “Le méchant” is a serious register mismatch: ordinary modern French uses it for a naughty child or a fairy-tale villain (“le grand méchant loup”), trivializing rasha’s covenant-rebellion gravity. “Méditer” collides with the post-1970s French popular association of “la méditation” with secular mindfulness or Eastern contemplative practice, risking a content-free-stillness misreading of hagah’s vocalized rumination. “Loi” (reused from baseline) narrows torah to Mosaic legal code unless glossed toward its broader Psalms sense of delight-inspiring instruction. | Supplement “le méchant” with “l’impie” in every doctrinally weighty occurrence (Ps 1, 37, 73, 94); do not rely on “méchant” alone in headers or thesis statements. Gloss “méditer” on first use per lesson (“ruminer, redire à voix basse”) to block the mindfulness-drift reading. Extend, do not override, baseline “loi” with the added gloss “l’instruction de l’Éternel” wherever torah’s Psalms sense is active. |
| Covenant Faithfulness (Hesed) | amour fidèle (recommended); bonté (Segond); grâce (rejected as primary); fidélité (rejected as primary) | This is the Psalter’s single largest genuine lexical gap: no French word fuses chesed’s blend of covenant obligation, warm affection, and active mercy the way the Hebrew does. “Grâce” wrongly imports the baseline’s no-prior-relationship sense (charis); bare “fidélité” is accurate but cold, losing the affectionate warmth; Segond’s historic “bonté” is too weak/generic (mere kindness) for a term occurring ~245 times as a theological anchor. | Adopt “amour fidèle” as the standing curriculum rendering (already fixed in the registry), tested specifically against Psalm 136’s 26-fold refrain for naturalness under heavy repetition. Require a standing teaching gloss distinguishing hesed (amour fidèle, covenant-conditioned) from grâce (chen, unconditioned) and from compassion (rachamim, visceral tenderness) every time the three appear near each other (e.g., Ps 103; 145). |
| Kingship and the Messianic King (Psalms 2 and 110) | l’Éternel; seigneur; Fils; Fils de Dieu; les nations; décret; houlette/sceptre; Melchisédek | Ps 110:1’s entire messianic argument depends on French sustaining a lexical distinction (l’Éternel vs. seigneur) that ordinary French Bible reading habits do not reinforce, since most lay readers have only ever encountered a single undifferentiated capitalized “Seigneur.” Ps 2:7’s “Fils” carries an unavoidable two-horizon teaching burden (historical Davidic coronation vs. canonical NT re-application) that a single French word cannot disambiguate without added framing. | Apply the l’Éternel/seigneur distinction with zero exceptions across every occurrence in Psalms 2, 8, 45, 110, and elsewhere; require an explicit teaching note at every Ps 110:1 occurrence walking through the two-Lords argument. Render Ps 2:7 “Fils” plainly, with mandatory teaching-note framing of both horizons rather than attempting a lexical fix. |
B. Missing Vocabulary — Genuine Lexical Gaps
These are cases where no existing French word or fixed phrase adequately covers the Hebrew concept, requiring either a constructed phrase, a compound rendering, or an accepted “good enough” approximation flagged for ongoing teaching support.
- Hesed (חֶסֶד) — the Psalter’s largest gap. No single French word carries covenant-obligation + affectionate love + active mercy simultaneously. Resolved by the constructed phrase “amour fidèle,” which must be reinforced pedagogically rather than assumed self-evident (see doctrine matrix above and Cross-Cutting Risk Notes in §E).
- Nephesh (נֶפֶשׁ) as “the whole living self” — French “âme” imports a Platonic body/soul dualism the Hebrew does not intend. There is no single French word for “the whole living, breathing person” outside of periphrasis (“tout mon être,” “ma vie même”). “Âme” is retained as the least-bad established option but requires a standing gloss wherever nephesh is theologically active (Ps 23:3; 42:1-2; 103:1).
- Shevet (שֵׁבֶט) as a single shepherd’s-club/royal-scepter word — French has no single word spanning both images; French must split it into “houlette” (pastoral) and “sceptre” (royal), losing the Hebrew’s one-word intertextual link between Shepherd and King. This is a structural loss requiring an explicit teaching bridge rather than a lexical fix.
- Chasid (חָסִיד) — “one characterized by hesed” has no single French noun; “bien-aimé/fidèle” is a serviceable double-rendering but does not visibly carry the hesed root the way the Hebrew cognate does. A footnote-level etymological note is recommended wherever chasid appears near hesed itself (e.g., Ps 145:17,20).
- The eight torah-synonyms of Psalm 119 (mitzvah, mishpat, chuqqah, edah, davar, imrah, plus torah and mishpat) — French possesses adequate individual glosses (commandement, jugement/ordonnance, décret, témoignage, parole, promesse) but no single organizing category-word that signals “these are eight names for the same thing, deliberately varied,” risking either flattening (all rendered “loi”) or fragmentation (readers missing the acrostic’s unifying intent) without an explicit translator’s note.
- “Ad-anah” lament formula (עַד־אָנָה, “how long?”) — not a true lexical gap (French “jusqu’à quand” is accurate) but a genre gap: French lacks a devotional category recognizing sanctioned protest-prayer as a distinct, legitimate speech-act the way lament psalms model it. This is a conceptual/catechetical gap, not a vocabulary gap, but must be logged here because it materially affects how confidently a translator can render lament language without softening it.
C. Crowded Semantic Neighborhoods Requiring Fencing
Unlike missing vocabulary, these are cases where French already supplies a ready word, but that word’s dominant everyday or religious sense actively competes with, and risks displacing, the biblical sense. Each requires explicit “fencing” (a disambiguating gloss, capitalization convention, or teaching note) rather than a search for an alternative word.
| French Term | Competing Neighbor(s) | Collision Risk | Fencing Strategy |
|---|---|---|---|
| Seigneur | l’Éternel (YHWH); feudal “seigneur du château” (baseline-documented) | Collapsing YHWH/Adonai into one undifferentiated “Seigneur” destroys Ps 110:1’s two-Lords argument; feudal-title residue risks a distant, archaic-title reading | Absolute lexical separation: l’Éternel (YHWH) vs. lowercase seigneur (Adonai); reviewer note at every Ps 110:1-type occurrence |
| confession | Sacrament of Penance/Reconciliation | ”La confession” is heard first, by both practicing and lapsed Catholics, as auricular confession to a priest | Mandatory gloss in every doctrinal header: “la confession du péché devant Dieu” |
| berger | Secular pastoral idyll (postcard shepherd, “berger” as a gentle rural profession) | Loses the ANE ruler-protector, life-and-death authority sense that underlies Ps 23 and John 10 | Teaching note supplying king-as-shepherd ANE background at first doctrinal use |
| refuge | Secular “un refuge” (mountain shelter, animal shelter, migrant asylum center) | Flattens God-as-refuge into a passive location rather than an active, trust-based relationship with a Person | Teaching note: refuge is relational reliance on God himself, not a place one stumbles upon |
| rocher | Dead idiom “solide comme un roc” | Reduces a living, protective divine Person to a mere adjective of stability | Retain personal-address framing (“mon rocher,” not “il est stable comme un roc”) |
| méchant | Fairy-tale villain / naughty-child register (“le grand méchant loup”) | Trivializes rasha’s covenant-rebellion gravity in doctrinally weighty passages | Supplement with “l’impie” in Psalm 1, 37, 73, 94 headers and thesis statements |
| méditer | Secular/Eastern mindfulness meditation | Suggests content-free mental stillness rather than sustained vocalized rumination on revealed text | First-use gloss: “ruminer, redire à voix basse,” per lesson |
| loi (torah) | Baseline’s narrower Romans-context Mosaic legal code | Psalms’ torah is the broader, delight-inspiring whole of God’s revealed instruction | Gloss “l’instruction de l’Éternel” wherever the broader Psalms sense is active; do not silently narrow to legal-code sense |
| grâce | Hesed (amour fidèle); Catholic sacramental grace (already flagged in baseline) | Using “grâce” for hesed imports an unconditioned, no-prior-relationship sense that erases hesed’s covenant-conditioned warmth | Standing teaching contrast: grâce (chen, unconditioned) vs. amour fidèle (hesed, covenant-conditioned) |
| élection | Political/democratic elections (already flagged in baseline) | Extends into Psalms’ covenant-choice language (e.g., Ps 78:70, 105:6, 135:4, “chosen”) with the same competitive-process risk | Reuse baseline fencing exactly; no new mitigation required beyond consistent application |
| vocation | Catholic clergy/religious-life calling (already flagged in baseline) | Not directly a Psalms term but risks contaminating any teaching material discussing “God’s calling” of David or Israel | Reuse baseline “appel,” never “vocation,” for any Psalms calling-language |
| dieux (Ps 82) | Polytheistic plurality of true deities | ”Dieux” applied to human judges could be misread as conceding other real gods, directly relevant to John 10:34’s use of this text | Mandatory gloss ”(= les juges/les autorités)” at every occurrence; theologian review required |
| maison de l’Éternel | Physical parish church building (Catholic-formed reader default) | Ps 23:6’s “dwelling in the house of the LORD forever” risks being heard as an aspiration toward a literal building rather than eternal covenant presence | Teaching note distinguishing eternal covenant belonging from any physical edifice |
| crainte de l’Éternel | Servile dread / ordinary “crainte” (fear, apprehension) | Fixed idiom cannot be relexicalized, but ordinary French usage of “crainte” leans toward apprehension, not awe-with-trust | Teaching note only (idiom is fixed across Segond/TOB/BJ; do not attempt lexical substitution) |
| Royaume de Dieu | France’s republican/laïque political culture, where “royaume” reads as archaic/monarchical | Risks either an inert historical-curiosity reading or, conversely, an inappropriate this-worldly political kingdom reading | Reuse baseline fencing; reinforce “sovereign reign,” not territorial state |
| corne de mon salut (“horn of my salvation”) | No competing neighbor, but risk of sheer strangeness to modern French ears (animal-horn literalism) | Not a collision but an intelligibility gap that could cause readers to disengage rather than misread doctrinally | Brief explanatory gloss recommended, preserved for its Magnificat/Benedictus intertextual value |
D. Transliteration vs. Paraphrase Decisions
| Term | Decision | Rationale |
|---|---|---|
| Selah (סֶלָה) | Transliterate — “Sélah” | Meaning uncertain even in Hebrew scholarship; all major French Bible traditions retain it untranslated as a structural marker. No French paraphrase would be more accurate than transliteration. |
| Hallelujah (הַלְלוּיָה) | Transliterate — “Alléluia” | Stable, ecumenical, shared across Catholic and Protestant French liturgical use; a positive shared resonance rather than a collision. Paraphrasing (“Louez l’Éternel”) would sever a recognizable worship cue. |
| Hosanna (הוֹשִׁיעָה נָּא) | Split decision: transliterate (“Hosanna”) in New Testament quotation contexts; paraphrase (“sauve, je t’en prie”) within the Psalm’s own Old Testament context | The Gospel’s transliterated liturgical acclamation and the Psalm’s plain liturgical plea are functionally different registers; collapsing both to one form either anachronizes the Psalm or strips the Gospel quotation of its recognizable form. |
| Abba (אַבָּא) | Transliterate — “Abba” (reuse baseline) | Established across Segond, TOB, and Bible de Jérusalem for the NT occurrence (Romans 8:15); no Psalms occurrence but retained here for cross-curriculum consistency in any adoption-doctrine teaching material drawing on Psalms fatherhood language. |
| YHWH (יהוה) | Paraphrase/substitution, not transliteration — “l’Éternel” (Yahvé flagged as an alternative transliteration for reviewer choice) | A genuine transliteration (“Yahvé”) exists and is used by the Bible de Jérusalem tradition; this curriculum adopts the Segond-tradition substitution “l’Éternel” instead, for continuity with the baseline Romans package’s overall Segond-aligned register. This is a deliberate register choice, not a default; document it explicitly so Phase 2 reviewers do not “correct” it toward Yahvé. |
| Sheol (שְׁאוֹל) | Paraphrase — “le séjour des morts” | No French transliteration convention exists for Sheol; the established French descriptive phrase is preferred over an unfamiliar transliterated form that would carry no inherited meaning for French readers. |
| Melchizedek (מַלְכִּי־צֶדֶק) | Transliterate (proper name) — “Melchisédek” | Standard proper-name transliteration across all French Bible traditions; no paraphrase is applicable to a proper name. |
| Mashiach (מָשִׁיחַ) | Split decision: paraphrase historically (“l’oint”), transliterate/canonical-render where NT-messianic sense is active (“Messie,” itself already a transliteration-derived loanword in French) | Neither a single transliteration nor a single paraphrase covers both the historical Davidic referent and the NT typological fulfillment; the dual-register rule from the baseline registry is retained. |
| Torah (תּוֹרָה, Psalms sense) | Paraphrase, not transliteration — reuse baseline “loi,” glossed “l’instruction de l’Éternel” | A transliterated “Torah” would be intelligible to French readers with some Jewish-cultural literacy but opaque or falsely “foreign-sounding” to the broader lapsed-Catholic/secular readership this curriculum targets; paraphrase with a gloss better serves the target audience per the baseline’s register and reading-level standard. |
| Amen (אָמֵן) | Transliterate — “Amen” (reuse baseline) | Universally retained across all French Bible and liturgical traditions. |
E. Ranked List of Highest-Risk Ambiguities
Ranked by combined severity (doctrinal weight × likelihood of silent, undetected mistranslation) for Phase 2 routing priority. Items 1–5 require human theologian review at every occurrence; items 6–10 require native speaker review at minimum.
- YHWH vs. Adonai collapse (l’Éternel vs. seigneur), esp. Psalm 110:1. The single highest-leverage disambiguation in the entire Psalter; failure here silently destroys the New Testament’s own messianic argument (Matthew 22:41-46) at its Old Testament source. Every occurrence across all 150 psalms must be checked against this convention before any segment is marked complete.
- Hesed rendered as “amour fidèle” vs. drift toward “grâce” or bare “fidélité.” Occurs ~245 times; a single unreviewed lapse into “grâce” at any of its densest concentrations (Ps 103, 107, 136, 145) would misteach the entire Covenant Faithfulness doctrine strand by importing an unconditioned-favor sense onto a covenant-conditioned term.
- Psalm 22:16 textual crux (“ils ont percé” vs. “comme un lion”). Not a normal lexical choice but an unresolved Hebrew manuscript question with direct bearing on crucifixion typology; any segment presenting this as textually uncontested is a doctrinal-fidelity failure, not merely a stylistic one.
- Psalm 45:6 “ton trône, ô Dieu” — direct-address ambiguity. A translator “smoothing” this into an indirect royal-throne phrase would silently erase the Deity-of-Christ argument Hebrews 1:8 depends on.
- Psalm 82:1,6 “dieux” applied to human judges. Highest collision risk with a plain plurality-of-deities misreading; directly relevant to Jesus’s own self-defense in John 10:34-36, making an unglossed occurrence doctrinally consequential, not merely confusing.
- “Confession” doctrine-label collision with the Catholic sacrament of Penance. Every unglossed doctrinal/lesson-title use risks the entire Confession and Forgiveness unit being heard as instruction about auricular confession to a priest rather than Psalm 51’s direct personal address to God.
- Mashiach dual-register rendering (l’oint / Messie) applied mechanically rather than contextually. A single blanket rule in either direction (all “Messie” or all “l’oint”) either anachronizes historical referents or erases the New Testament’s typological argument; this requires case-by-case judgment that is easy to get wrong at scale across dozens of occurrences.
- “Le méchant” left unsupplemented by “l’impie” in Psalm 1, 37, 73, 94 headers. A structural doctrine (The Righteous and the Wicked) resting on a term whose default French register is a storybook villain or naughty child.
- “Fils” at Psalm 2:7 taught with only one horizon (either purely historical or purely canonical). Omitting either the Davidic-coronation setting or the New Testament’s re-application produces either an anachronistic over-reading or a flattened under-reading of a Critical-risk Sonship-of-Christ feeder text.
- “Berger” and “refuge” secular flattening in the Character-of-God doctrine strand. Lower doctrinal-catastrophe risk than items 1–9, but high frequency risk (Ps 23 is the core passage) — a sentimentalized shepherd or a passive-shelter refuge quietly drains the entire doctrine of its Ancient Near Eastern sovereign-protector force across the curriculum’s flagship psalm.
F. Summary and Routing Implications for Phase 2
- Items ranked 1–5 above must be encoded as Critical-tier entries in the extended
bible_term_registry.json/doctrine_risk_registry.json(already reflected in the prior-output registry for this curriculum) and must trigger mandatory human theologian review on every occurrence, with no automated-approval path. - Items ranked 6–9 require High-tier native-speaker-or-theologian review depending on doctrinal load, consistent with the existing registry’s risk assignments for “confession,” “wicked,” “messiah_anointed,” and “son_royal.”
- Item 10, and the broader “crowded semantic neighborhood” list in §C, should be handled primarily through standing teaching-note fencing rather than lexical substitution — the French words chosen are, in each case, the best available options; the risk lies in readers’ default associations, not in an incorrect translation choice.
- The single largest true lexical gap (hesed, §B.1) has no remaining “better word” to discover; further Phase 2 effort should focus on testing “amour fidèle” for naturalness under repetition (Ps 136) and on reinforcing the hesed/grâce/compassion three-way distinction pedagogically, not on continued search for a single perfect equivalent.
- No new terms surfaced in this analysis contradict the baseline Romans
translation_memory.json; all baseline High/Critical fencing rules (élection, vocation, saints, Église, justification, justice imputée) remain in force unchanged and extend automatically to any Psalms passage where those concepts recur.
End of Linguistic Gap Analysis. Cross-reference 07_semantic_analysis.md for full contextual discussion and 08_core_glossary.md for the consolidated term-by-term glossary built on these findings.