The claim
A155 closed on a thesis: “A trope is a candidate communicative invariant: a recognizable relational pattern that persists across many realizations, while its surface content and even its rhetorical function may change.” And refined it: “Rather than treating a trope as something that must be perfectly invariant, we can characterize it as a family of observational profiles connected by admissible transformations, with a distinguishable core of preserved relations.”
So a trope is not one profile but a family of them (CASE-038). The admissible transformations are A155’s vocabulary: recasting, inversion, subversion, deconstruction and parody. What makes them admissible is that they keep the core relations apart from the rest (CLAIM-077), and transformations and the invariants they keep define each other (CLAIM-113). Because parody keeps the structure and changes the function, a trope’s instances can pull apart the two kinds of preservation that a translation runs together: of a relational pattern, and of what the pattern is for. A178’s Study B: “Distinguish preservation of the trope’s recognizable relational pattern from preservation of its rhetorical or pragmatic function.” A191 §8: “These interventions separate narrative structure from narrative function.”
The test A155 §5 proposes: “The central test is whether D_trope(W, W′) remains low even if lexical, visual, or surface-narrative similarity becomes small.” For each source example it generates a surface-altered version, a relation-altered version, an explicit subversion and a cross-modal realization, and compares an embedding baseline with transport (TERM-008) measures “on held-out trope families”. “The crucial result would be whether the latter distinguish same trope, different surface from same surface, different trope.” The distortion is read component by component (TERM-043): structural preservation in one component, pragmatic in another. The parody case is the Villon–Henley one (CASE-010).
It answers QUESTION-002 in part, for one kind of material: what survives a recast is the core of relations, and whether the function survives is a separate question.
What it does not say
- It does not say a trope has a definition. A155 §1: “This is not yet a formal definition of the trope”.
- It does not say a trope classifier would test the theory. A155 §3: “The simplest experiment would train a classifier … But that would miss much of what makes this case attractive.”
- It does not say the analysis is new in kind. Propp’s functions (THEORY-167: “differ in who performs the functions and how”) and Lévi-Strauss’s transformations (CLAIM-019) are the same move on folktales and myths. A155 names neither.
- It does not say the catalogue’s labels can validate it. A trope’s description is written to fit its examples, so agreement with the labels is editorial agreement (CLAIM-145, QUESTION-005). A155 §5’s own remedy, “a small, independently validated set of tropes” and evaluation “on held-out trope families”, would meet that objection if it were carried out.
Note of 2026-10-29: what holds it back
The result its defeat condition names is missing. No study in the record compares relational measures against surface or embedding similarity as predictors of human judgements of trope identity on recasts. A155’s test with held-out trope families, which A178 kept as Study B, has never been run. The other route is the owner’s ruling that the characterization is held as a definition, with the test bearing only on its use.