From 2f9b4504844b12c24a8be63903dfec1819c4e4a1 Mon Sep 17 00:00:00 2001 From: jsboige Date: Sun, 5 Jul 2026 21:00:20 +0200 Subject: [PATCH] docs(taxonomy): #498 PR-7 Amphibologie cluster (borrow-root, 8 leaves) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Seventh cluster of the #498 AIF chantier. Opens the Ambiguïté subfamily (Abus de language / Misleading language family). Models the Amphibologie sub-sub (pk 847 anchor + 848-854), borrowing the d2 subfamily root pk 846 (direct-conflict pattern: ArbitraryVerbalClassification_Inference + OppositeConsequences_Conflict + SignFromOtherEvents_Conflict). Structurally coherent cluster — all leaves are syntactic/structural/prosodic ambiguity mechanisms (sentence-level), distinct from the Equivoque sub-sub (term-level). This coherence drives an honest scheme-divergence from the borrow-root: amphiboly is syntactic, the borrow-root's classification framing is term-level, so leaves split across native schemes that honestly fit their mechanism rather than forcing the borrow-root's scheme onto syntactic defects: - Exception pattern (849 modal scope, 850 quantifier shift): Deductive_Inference + Logical_Conflict (genuine logical-scope defects, first chantier use of Deductive_Inference for a fallacy) - Direct-conflict pattern (851 accent, 852 illicit contrast): ArbitraryVerbalClassification_Inference (borrowed from 846 — prosody is term-level, classification framing fits) - Direct-conflict + cross-cluster reuse (854 barbarism): VagueVerbalClassification_Inference (PR-4 scheme — barbarism is vagueness not ambiguity; the desc word 'incertain' signals it) - FAIL-LOUD (847 amphiboly anchor, 848 punctuation, 853 solecism): structural/syntactic ambiguity — THIRD recurring AIF ontology gap after circularity (PR-1/PR-6); no native syntactic-ambiguity CQ, documented in AIF_skosOther, 0 fabrication 8 rows proposed: 5 fully-modeled + 3 FAIL-LOUD (most FAIL-LOUD in a single cluster). §7 I/RA/CA decomposition: all typed leaves undercuts (consistent jsboige). Decomposition recorded but not serialized (pending jsboige schema decision #707 §4). Borrow-root shape = independent of that decision. PROPOSITION ONLY — GATED, 0 prod CSV write, no DB write, no OWL regen. Native vocabulary validated programmatically (all backticked scheme/conflict tokens native; Ambiguity_Conflict and Amphiboly_Conflict appear only in the 'do not fabricate' prohibition L105, never as proposals). Relates: dispatch ih617l (primary), #498, PR-1 #699, PR-2 #701, PR-3 #703, PR-4 #705, PR-6 #708, coverage-status #707, verification-audit #709, #133/#130 (OWL), #499, #677 (0 fabrication), #192, #458. Co-Authored-By: Claude-Code --- docs/taxonomy/498-aif-amphibologie-cluster.md | 357 ++++++++++++++++++ 1 file changed, 357 insertions(+) create mode 100644 docs/taxonomy/498-aif-amphibologie-cluster.md diff --git a/docs/taxonomy/498-aif-amphibologie-cluster.md b/docs/taxonomy/498-aif-amphibologie-cluster.md new file mode 100644 index 00000000..c9ea87c3 --- /dev/null +++ b/docs/taxonomy/498-aif-amphibologie-cluster.md @@ -0,0 +1,357 @@ +# 2026-07-05 — #498 AIF chantier, PR-7 : cluster « Amphibologie » (Ambiguïté, nouvelle subfamily) + +**Scope**: seventh cluster of the #498 chantier — first cluster of a **new subfamily** +(**Ambiguïté**, Abus de langage / Misleading language family). Models the **Amphibologie** sub-sub +(pk 847 anchor + 848-854 leaves, 8 rows total). **Proposition only — GATED, 0 write to prod CSV.** +Triggered by ai-01 dispatch `ih617l` (primary: « prochain cluster borrow-root schema-independent »). +Borrow-root shape — **independent of the I/RA/CA serialization decision** (no schema dependency). + +**Repo reference**: master `34c7702c`. Issue: #498. Predecessors: PR-1 #699, PR-2 #701, PR-3 #703 +(Fallacious comparison complete), PR-4 #705 (Vague definition), PR-6 #708 (Inconsistent definition). + +> **⚠ STATUS = PROPOSITION ONLY.** No prod CSV write, no DB write, no OWL regen. jsboige ratifies → +> worker applies the CSV edits in a follow-up PR (gated). + +--- + +## TL;DR + +The **Amphibologie** sub-sub has **no in-sub-sub mapped anchor** (pk 847 d3 is unmapped) → it borrows +the d2 subfamily root **pk 846** ("Ambiguïté", MAPPED, direct-conflict pattern: +`ArbitraryVerbalClassification_Inference` + `OppositeConsequences_Conflict` + +`SignFromOtherEvents_Conflict`, `skos:broadMatch`). The sub-sub has **8 rows** (pk 847 d3 anchor + +7 leaves, depth 4-5) — PR-2-sized. + +The cluster is **structurally coherent** (all leaves are syntactic / structural / prosodic ambiguity +mechanisms — sentence-level rather than term-level), which sets it apart from the *Equivoque* sub-sub +(term-level ambiguity). This coherence drives an **honest scheme-divergence from the borrow-root**: +amphiboly is syntactic, while the borrow-root's `ArbitraryVerbalClassification_Inference` is +semantic/term-level — so the leaves split across **native schemes that honestly fit their mechanism** +(`Deductive_Inference`+`Logical_Conflict` for scope, `ArbitraryVerbalClassification_Inference` for +prosody, `VagueVerbalClassification_Inference` for undefined terms) rather than forcing the +borrow-root's classification framing onto syntactic defects. + +**8 rows proposed: 5 fully-modeled + 3 FAIL-LOUD** (847/848/853 structural ambiguity — **third +recurring AIF ontology gap** after circularity: no native syntactic-ambiguity CQ). This PR opens the +**Ambiguïté** subfamily and flags the ambiguity gap for the eventual ontology-extension decision. + +--- + +## 1. Cluster state (code=truth) + +From `Cards/Fallacies/Argumentum Fallacies - Taxonomy.csv`, family **Abus de langage** (Misleading +language), subfamily **Ambiguïté**, sub-sub **Amphibologie**: + +| pk | text_fr | depth | DirectRef | ExceptionRef | MappingType | State | +|----|---------|-------|-----------|--------------|-------------|-------| +| **846** | **Ambiguïté** (borrowed anchor) | d2 | `ArbitraryVerbalClassification_Inference`, `OppositeConsequences_Conflict`, `SignFromOtherEvents_Conflict` | — | `skos:broadMatch` | **✅ MAPPED (d2 subfamily root)** | +| 847 | Amphibologie | d3 | — | — | — | unmapped (d3 anchor, **PR-7**) | +| 848 | Ponctuation ambiguë | d4 | — | — | — | unmapped (**PR-7**) | +| 849 | Sophisme de portée modale | d4 | — | — | — | unmapped (**PR-7**) | +| 850 | Glissement du quantificateur | d5 | — | — | — | unmapped (**PR-7**) | +| 851 | Accent | d4 | — | — | — | unmapped (**PR-7**) | +| 852 | Contraste illicite | d5 | — | — | — | unmapped (**PR-7**) | +| 853 | Solécisme | d4 | — | — | — | unmapped (**PR-7**) | +| 854 | Barbarisme | d4 | — | — | — | unmapped (**PR-7**) | + +The borrowed anchor (846) is the **template**: a direct-conflict pattern carrying **three** native +tokens (the fallacy *is* an ambiguous classification / a sign from ambiguous events / opposite +consequences of an ambiguous statement). As in PR-3 (anchor 833 split across 2 leaves), the +Amphibologie leaves honestly draw on the borrow-root's tokens where they fit (851/852 → +`ArbitraryVerbalClassification_Inference`) and diverge to other native schemes where the mechanism is +genuinely different (849/850 → deductive scope; 854 → vagueness). + +--- + +## 2. The borrowed anchor (846, recap) + +**Pattern**: direct-conflict (no ExceptionRef). + +**Conflict/scheme nodes** (DirectRef): +- `ArbitraryVerbalClassification_Inference` — the ambiguous statement is classified under one + arbitrarily-chosen reading. +- `OppositeConsequences_Conflict` — the ambiguity admits readings with opposite consequences. +- `SignFromOtherEvents_Conflict` — the ambiguous statement is taken as a sign, but the sign relation + is ambiguous. + +**desc_fr** (846): "Votre argumentation repose sur des énoncés qui peuvent avoir plusieurs sens." + +The Amphibologie leaves specialize the *ambiguity mechanism*: sentence-structure (847/848), +logical scope (849/850), prosody (851/852), grammar (853), lexicon (854). The borrow-root's +*classification* framing fits the prosodic and lexical leaves but **not** the structural and +scope leaves — those honestly target deductive/logical schemes instead (PR-2 §5 scheme-divergence). + +--- + +## 3. Proposed AIF structure for the leaves + +For each leaf: **(a) legitimate scheme (or FAIL-LOUD), (b) exception/CQ or direct conflict, +(c) `AIF_skosMappingType`**. Vocabulary restricted to AIF-native tokens confirmed by existing usage. + +### pk 847 — Amphibologie (Amphiboly, d3 anchor) ⚠ FAIL-LOUD +- **desc_fr**: "Vous construisez votre argumentation sur des phrases dont la structure peut prêter à + différentes interprétations." +- **⚠ FAIL-LOUD at the CA-node level**: amphiboly is a **syntactic** ambiguity — the sentence + *structure* (not term choice) admits two readings, and the arguer rides the one that supports their + conclusion. The defect defeats an inference (the premise is equivocal), but **no native AIF Conflict + node captures syntactic ambiguity**. The borrow-root's `ArbitraryVerbalClassification_Inference` is + term-level (semantic), not sentence-level (syntactic) — forcing it would conflate the two (PR-2 §5 + scheme-divergence discipline). `Logical_Conflict` does not fit (the two readings need not be + contradictory; they are simply different parses). +- **Proposal**: `DirectRef=` *(absent — FAIL LOUD)*, `ExceptionRef=VerbalClassification_Inference` + (the statement-as-classification RA-node exists), `AIF_skosOther="Syntactic ambiguity + (amphiboly): sentence structure admits multiple parses (no native AIF Conflict node — + ArbitraryVerbalClassification_Inference is term-level, does not fit syntactic ambiguity)"`, + `MappingType=skos:closeMatch`. Do **not** fabricate an `Ambiguity_Conflict` or `Amphiboly_Conflict` + token. + +### pk 848 — Ponctuation ambiguë (Ambiguous punctuation) ⚠ FAIL-LOUD +- **desc_fr**: "Vous utilisez une ponctuation imprécise, créant ainsi des phrases à double sens." +- **⚠ FAIL-LOUD at the CA-node level**: same gap as 847 — punctuation is a **structural** ambiguity + mechanism (the parse tree depends on comma/period placement), not a term-level classification + defect. No native AIF CQ captures it. +- **Proposal**: `DirectRef=` *(absent — FAIL LOUD)*, `ExceptionRef=VerbalClassification_Inference`, + `AIF_skosOther="Structural ambiguity via punctuation (no native AIF CQ — same gap as 847)"`, + `MappingType=skos:narrowMatch` (a narrower case: syntactic ambiguity *specifically via punctuation*, + narrower than 847's general amphiboly). + +### pk 849 — Sophisme de portée modale (Modal scope fallacy) ✅ +- **desc_fr**: "Vous modifiez le sens logique en confondant la portée d'un terme modal (nécessaire, + possible, certain)." +- **⚠ Honest scheme-divergence from the borrow-root**: the modal scope fallacy is a **deductive / + logical** defect (conflating ◻(A→B) with A→◻B, or wide vs narrow scope of a modal operator), not a + classification defect. It honestly targets `Deductive_Inference` (native, confirmed 2x in existing + usage), not the borrow-root's `ArbitraryVerbalClassification_Inference`. +- **Exception/CQ**: the wide-scope and narrow-scope readings are **logically incompatible** (they + license different conclusions) → `Logical_Conflict` (native, confirmed 1x). The deductive inference + is defective because the operator's scope is conflated. +- **Proposal**: `ExceptionRef=Deductive_Inference`, `DirectRef=Logical_Conflict`, + `MappingType=skos:closeMatch` (a direct logical-inference variant of the ambiguity family). + +### pk 850 — Glissement du quantificateur (Quantifier shift) ✅ +- **desc_fr**: "Vous modifiez le sens d'une phrase en déplaçant les quantificateurs." +- **Mechanism**: the quantifier-shift fallacy — confusing ∀x∃y with ∃y∀x (e.g. "Everyone loves + someone" → "There is someone whom everyone loves"). A direct sibling of 849 (scope conflation), but + for **quantifiers** rather than modal operators. +- **Legitimate scheme**: `Deductive_Inference` (same scheme-divergence as 849 — a deductive/logical + defect). +- **Exception/CQ**: the two quantifier orderings are logically incompatible → `Logical_Conflict`. +- **Proposal**: `ExceptionRef=Deductive_Inference`, `DirectRef=Logical_Conflict`, + `MappingType=skos:narrowMatch` (a narrower case: scope conflation *specifically via quantifier + order*, narrower than 849's general modal scope). + +### pk 851 — Accent (Fallacy of accent) ✅ +- **desc_fr**: "Vous utilisez l'intonation ou l'accentuation pour suggérer un message différent des + mots employés." +- **Legitimate scheme**: `ArbitraryVerbalClassification_Inference` (native, **borrowed from anchor + 846** — the accent/stress picks out one reading of the utterance, an arbitrary classification of its + meaning). Unlike 847/848 (syntactic), accent is **prosodic / term-level** (which word is stressed) → + the borrow-root's classification framing honestly fits. +- **Pattern**: direct-conflict (mirrors 846). +- **Proposal**: `DirectRef=ArbitraryVerbalClassification_Inference`, `ExceptionRef=`, + `MappingType=skos:closeMatch` (a direct prosodic variant of 846's classification defect). + +### pk 852 — Contraste illicite (Illicit contrast) ✅ +- **desc_fr**: "Vous attribuez à autrui une intonation qui modifie le sens perçu de ses mots." +- **Mechanism**: the illicit-contrast fallacy — **misattributing** stress/prosody to someone else's + words to distort their meaning (the mirror of 851, applied to others' speech). Same prosodic / + term-level mechanism → same native scheme. +- **Legitimate scheme**: `ArbitraryVerbalClassification_Inference` (borrowed from 846). +- **Proposal**: `DirectRef=ArbitraryVerbalClassification_Inference`, `ExceptionRef=`, + `MappingType=skos:narrowMatch` (a narrower case: arbitrary classification *via misattributed + prosody to another's speech*, narrower than 851's own-accent case). + +### pk 853 — Solécisme (Solecism) ⚠ FAIL-LOUD +- **desc_fr**: "Une erreur de grammaire dans votre phrase crée une ambiguïté, même si les mots + employés sont corrects." +- **⚠ FAIL-LOUD at the CA-node level**: the desc explicitly states the grammar error **creates an + ambiguity** — but the ambiguity mechanism is **structural/grammatical** (the parse is ill-formed), + not term-level classification. No native AIF CQ captures grammar-induced ambiguity (same gap as + 847/848). `VagueVerbalClassification_Inference` does not fit (the words are correct, not vague — + the *grammar* is the defect). +- **Proposal**: `DirectRef=` *(absent — FAIL LOUD)*, `ExceptionRef=VerbalClassification_Inference`, + `AIF_skosOther="Grammar-induced ambiguity (solecism): ill-formed parse creates ambiguity (no native + AIF CQ — same structural-ambiguity gap as 847/848)"`, `MappingType=skos:narrowMatch` (a narrower + case: ambiguity *specifically via grammar error*). + +### pk 854 — Barbarisme (Barbarism) ✅ +- **desc_fr**: "Vous employez un mot inventé ou emprunté à une autre langue, rendant le propos + incertain." +- **⚠ Semantic link to PR-4**: unlike 847/848/853 (structural ambiguity), barbarism makes the + statement **uncertain because the term is undefined** — this is **vagueness** (no determinate + meaning), not ambiguity (two determinate readings). The desc "incertain" signals vagueness. It + honestly reuses PR-4's anchor scheme `VagueVerbalClassification_Inference` (cross-cluster reuse, + like PR-6 831 → PR-4). +- **Legitimate scheme**: `VagueVerbalClassification_Inference` (PR-4 anchor scheme, native). +- **Pattern**: direct-conflict (the vagueness *is* the scheme — the term is undefined). +- **Proposal**: `DirectRef=VagueVerbalClassification_Inference`, `ExceptionRef=`, + `MappingType=skos:narrowMatch` (a narrower case: vagueness *via an invented/borrowed term*, + narrower than PR-4 802's plain absence of definition). + +--- + +## 4. Cluster summary (proposed) + +| pk | text_fr | Proposed ExceptionRef | Proposed DirectRef | MappingType | Honest? | +|----|---------|----------------------|-------------------|-------------|---------| +| 847 | Amphibologie | `VerbalClassification_Inference` | *(absent — FAIL LOUD)* | `skos:closeMatch` | ⚠ FAIL LOUD (syntactic ambiguity) | +| 848 | Ponctuation ambiguë | `VerbalClassification_Inference` | *(absent — FAIL LOUD)* | `skos:narrowMatch` | ⚠ FAIL LOUD (structural ambiguity) | +| 849 | Sophisme de portée modale | `Deductive_Inference` | `Logical_Conflict` | `skos:closeMatch` | ✅ | +| 850 | Glissement du quantificateur | `Deductive_Inference` | `Logical_Conflict` | `skos:narrowMatch` | ✅ | +| 851 | Accent | *(absent)* | `ArbitraryVerbalClassification_Inference` | `skos:closeMatch` | ✅ | +| 852 | Contraste illicite | *(absent)* | `ArbitraryVerbalClassification_Inference` | `skos:narrowMatch` | ✅ | +| 853 | Solécisme | `VerbalClassification_Inference` | *(absent — FAIL LOUD)* | `skos:narrowMatch` | ⚠ FAIL LOUD (grammar-induced ambiguity) | +| 854 | Barbarisme | *(absent)* | `VagueVerbalClassification_Inference` | `skos:narrowMatch` | ✅ | + +**8 rows proposed: 5 fully-modeled (849, 850, 851, 852, 854) + 3 FAIL-LOUD (847, 848, 853).** The +cluster spans **three patterns**: exception (849, 850 — scheme + conflict), direct-conflict (851, 852, +854 — scheme as defect), FAIL-LOUD (847, 848, 853 — structural-ambiguity gap). It demonstrates +**scheme-divergence from the borrow-root** (849/850 → deductive, 854 → vagueness) and **cross-cluster +scheme reuse** (854 → PR-4). + +--- + +## 5. Method notes (additions for the chantier) + +- **Syntactic vs term-level ambiguity (honest scheme-divergence from borrow-root)**: the Ambiguïté + borrow-root (846) frames ambiguity as *verbal classification* (term-level / semantic). But + amphiboly (847/848/853) is *syntactic* (sentence-structure / grammar) — a genuinely different + mechanism. Forcing the borrow-root's `ArbitraryVerbalClassification_Inference` onto syntactic + defects would conflate the two layers (the exact anti-pattern PR-2 §5 warns against). The honest + move: FAIL-LOUD the structural leaves (no native syntactic-ambiguity CQ) and model only the leaves + whose mechanism is genuinely term-level (851/852 prosody, 854 lexicon) with the borrow-root's + scheme. +- **Third recurring AIF ontology gap — ambiguity**: PR-1/PR-6 documented circularity as a recurring + gap (no native circularity CQ). This cluster documents **ambiguity** as a third gap (no native + syntactic-ambiguity CQ), with **3 FAIL-LOUD leaves** (847/848/853) — the most FAIL-LOUD in a single + cluster so far. The chantier now has recurring gaps in **3 of 7 PR** (circularity PR-1/PR-6, + pure-comparison PR-2, ambiguity PR-7) — all worth flagging for the eventual ontology-extension + decision. +- **Logical-scope leaves (849/850) — scheme-divergence to Deductive_Inference**: modal-scope and + quantifier-shift fallacies are genuinely *logical* defects (scope conflation of operators), not + classification defects. They honestly target `Deductive_Inference` + `Logical_Conflict` (both + native). This is the first chantier use of `Deductive_Inference` for a *fallacy* (vs its existing + use for valid deductions) — legitimate because the fallacy *is* a defective deductive inference. +- **Vagueness-vs-ambiguity distinction (854)**: barbarism produces **uncertainty** (the term is + undefined → vague), not ambiguity (two readings). The desc word "incertain" is the signal. This + distinguishes 854 (→ `VagueVerbalClassification_Inference`, PR-4) from 847/848/853 (→ FAIL-LOUD + ambiguity). The distinction is philosophically real (vagueness ≠ ambiguity) and the AIF modeling + respects it. + +--- + +## 6. DoD status + +| DoD item (#498) | Status | +|-----------------|--------| +| Effective cluster-level coverage documented (not "100% leaves") | ✅ §4 (86→91 fully-modeled / 89→94 DirectRef-loose; new subfamily Ambiguïté opened) | +| Per unmapped leaf: legitimate scheme + exception/CQ (not Latin alone) | ✅ §3 — 5 leaves fully-modeled; 3 FAIL-LOUD (847/848/853 structural-ambiguity gap) | +| `AIF_skosMappingType` coherent | ✅ §4 (close/narrow per leaf semantics; narrow encodes the specialization mechanism) | +| Fail-loud when no honest scheme fits | ✅ §3 pk 847/848/853 (structural ambiguity) — documented, not fabricated | +| OWL regen reflects new structures | ⏳ Deferred — jsboige ratifies, then CSV edits, then OWL regen (#133) | + +### Coverage accounting (cumulative, chantier-wide) +| PR | Cluster | Leaves | Fully-modeled (DoD) | DirectRef-loose | +|----|---------|--------|---------------------|-----------------| +| baseline | — | — | 70 | 70 | +| PR-1 #699 | False analogy | 4 | 74 (+4) | 74 (+4) | +| PR-2 #701 | Faulty comparison | 5 | 76 (+2) | 79 (+5) | +| PR-3 #703 | Association fallacy | 2 | 78 (+2) | 81 (+2) | +| PR-4 #705 | Vague definition | 3 | 81 (+3) | 84 (+3) | +| PR-6 #708 | Inconsistent definition | 7 | 86 (+5) | 89 (+5) | +| **PR-7 (this)** | **Amphibologie** | **8** | **91 (+5)** | **94 (+5)** | +| **total** | **4 subfamilies touched** | **29** | **91** | **94** | + +**Ambiguïté subfamily: 1/3 sub-subs opened** (Amphibologie PR-7). Remaining: Equivoque (2 mapped + +18 unmapped — large, needs splitting by depth), Ambiguïté narrative (1 mapped anchor + 10 unmapped, +in-sub-sub anchor — cleanest next candidate). + +--- + +## 7. Mapping to jsboige's enriched DoD (I-node / RA-node / CA-node decomposition) + +### pk 847 — Amphibologie → FAIL-LOUD at **CA-node** +- **RA-node**: `VerbalClassification_Inference` (exists — classifying items via the statement). +- **I-nodes**: premise P "statement S means M1"; the sentence structure *also* admits M2. +- **CA-node**: ⚠ **absent (FAIL LOUD)** — amphiboly would be an **undercut** (the inference that S + licenses classification under M1 is defective because S is syntactically equivocal), but no native + AIF Conflict node captures syntactic ambiguity (same gap shape as PR-1 840 circularity). + +### pk 848 — Ponctuation ambiguë → FAIL-LOUD at **CA-node** +- **RA-node**: `VerbalClassification_Inference`. +- **CA-node**: ⚠ **absent (FAIL LOUD)** — punctuation-induced structural ambiguity, no native CQ + (same gap as 847). + +### pk 849 — Sophisme de portée modale → **undercut** +- **RA-node**: `Deductive_Inference` — a deductive step relying on a modal premise. +- **I-nodes**: premise P "◻(A→B)" (wide scope); the fallacy concludes "A→◻B" (narrow scope). +- **CA-node**: `Logical_Conflict`, applied as an **undercut** on the RA-node — the deductive inference + is defective because the wide-scope and narrow-scope readings are logically incompatible (they + license different conclusions). The conclusion may still hold; the *inference* is defective. + +### pk 850 — Glissement du quantificateur → **undercut** +- **RA-node**: `Deductive_Inference`. +- **I-nodes**: premise P "∀x∃y L(x,y)"; the fallacy concludes "∃y∀x L(x,y)". +- **CA-node**: `Logical_Conflict`, applied as an **undercut** — the two quantifier orderings are + logically incompatible (same shape as 849). + +### pk 851 — Accent → **undercut** +- **RA-node**: `ArbitraryVerbalClassification_Inference` — classifying the utterance's meaning via + stress. +- **I-nodes**: premise P "utterance U, stressed on word w, means M". +- **CA-node**: the scheme itself names the defect (direct-conflict pattern) — the stress-driven + classification is arbitrary. Degenerate **undercut** (CA coincides with RA, as in PR-4). + +### pk 852 — Contraste illicite → **undercut** +- **RA-node**: `ArbitraryVerbalClassification_Inference` — classifying another's meaning via + misattributed stress. +- **CA-node**: degenerate **undercut** (direct-conflict, scheme as defect). + +### pk 853 — Solécisme → FAIL-LOUD at **CA-node** +- **RA-node**: `VerbalClassification_Inference`. +- **CA-node**: ⚠ **absent (FAIL LOUD)** — grammar-induced ambiguity, no native CQ (same gap as + 847/848). + +### pk 854 — Barbarisme → **undercut** +- **RA-node**: `VagueVerbalClassification_Inference` — classifying via an undefined term. +- **CA-node**: degenerate **undercut** (direct-conflict — the vagueness is the scheme). + +### Cluster attack-type distribution +| pk | RA-node | CA-node | Attack type | +|----|---------|---------|-------------| +| 847 | `VerbalClassification_Inference` | *(absent — FAIL LOUD)* | would be undercut (syntactic ambiguity) — untyped | +| 848 | `VerbalClassification_Inference` | *(absent — FAIL LOUD)* | would be undercut (structural ambiguity) — untyped | +| 849 | `Deductive_Inference` | `Logical_Conflict` | undercut | +| 850 | `Deductive_Inference` | `Logical_Conflict` | undercut | +| 851 | `ArbitraryVerbalClassification_Inference` | *(scheme as direct conflict)* | undercut (degenerate) | +| 852 | `ArbitraryVerbalClassification_Inference` | *(scheme as direct conflict)* | undercut (degenerate) | +| 853 | `VerbalClassification_Inference` | *(absent — FAIL LOUD)* | would be undercut (grammar ambiguity) — untyped | +| 854 | `VagueVerbalClassification_Inference` | *(scheme as direct conflict)* | undercut (degenerate) | + +**All typed leaves are undercuts** — consistent with jsboige's "most fallacies live in the undercut": +ambiguity fallacies defeat the *inference* (the statement→classification or premise→conclusion step), +they do not deny the conclusion. No undermine, no rebut in this cluster. + +### Representation note +As in PR-1 to PR-6, the I-node/RA-node/CA-node decomposition is **recorded here but not serialized** +in the CSV — pending jsboige's schema decision (§4 of the coverage-status note #707). This PR is +**borrow-root and independent of that decision**: the modeling (schemes + conflicts) holds regardless +of whether I/RA/CA is serialized in new columns. + +--- + +## Gate boundaries (HARD — proposition only) + +- ❌ No prod CSV write, no DB write, no OWL regen, no `aif:` assertion emitted. +- ❌ No #674/#596/#666 merge, no régén launch. +- ✅ Proposition derived code=truth from the taxonomy CSV + AIF vocabulary confirmed by existing + usage (all proposed tokens native: `VerbalClassification_Inference`, `Deductive_Inference`, + `Logical_Conflict`, `ArbitraryVerbalClassification_Inference`, `VagueVerbalClassification_Inference`, + + the borrow-root's `OppositeConsequences_Conflict`, `SignFromOtherEvents_Conflict`). No fabrication + — 847/848/853 FAIL-LOUD with documented structural-ambiguity gap. +- ✅ Scheme-divergence from borrow-root (849/850 → Deductive, 854 → Vague) documented; cross-cluster + reuse (854 → PR-4) noted; third recurring ontology gap (ambiguity) flagged. + +Relates: dispatch `ih617l` (primary), #498 (chantier), PR-1 #699, PR-2 #701, PR-3 #703, PR-4 #705, +PR-6 #708, coverage-status #707, verification-audit #709, #133/#130 (existing OWL), #499 (inverse), +#677 (0 fabrication), #192, #458.