Contribution
A reflective keynote, not a research article. What it adds is a distinction: Lévi-Strauss used “transformation” in two incompatible ways, which Descola names after Goethe and D’Arcy Thompson, and only one of them fits myth. It also states plainly a consequence often left implicit: in the analysis of myth the transformation group is not given by the material but cut by the analyst. And it places Descola’s own ontological fourfold as a structural group of the Goethean kind.
Key insight
Structural analysis does not describe a thing; it orders a set of variants so that each is a transformation of the others. Whether the ordering is discovered or constructed depends on the regime: developing an initial form generates its variants by a rule, while deforming between given forms in myth leaves the path to the analyst’s knowledge and imagination.
Assumptions
- That method and doctrine in Lévi-Strauss can be separated.
- That the domain is “superstructures” (kinship, classification, myth, food norms, art), sets of phenomena organized systematically.
- That variants of a set have no privileged member: “each variant is a variant of the other variants and not of any of them in particular”.
Key results
The lecture’s theses, argued from Lévi-Strauss’s practice and Descola’s own work:
- Method. Elements are characterized by position, not intrinsic properties; the analyst brings to light, in a table of permutations, “the whole set of possible combinations”; the resulting model “provides the syntax of the transformations which allows one to move from one variant to another within a class of phenomena”.
- “A structure is not a system.” A structure requires “invariant relations which allow moving from one set to another by the means of a transformation”.
- Goethean regime (Les structures élémentaires): reciprocity, in its minimal form as restricted exchange between two moieties, develops into generalized exchange (Australian four- and eight-section systems), its combination with bride purchase (the Kachin), and mixtures in China and India. Lévi-Strauss, against Radcliffe-Brown, takes as initial the structure that develops the principle most widely.
- Thompsonian regime (myth): a mythical transformation is the group formed by variants of a myth that keep its structure, “including by inverting it”, and by myths of neighbouring societies that borrow mythemes, inverting their motifs or swapping their functions. Unlike organisms, whose differences rest on the discontinuities of the genetic code, the group of myths is “a virtual continuum upon which the analyst performs arbitrary cuts” (citing L’homme nu, 603–4). D’Arcy Thompson needs stable forms at both ends; myth analysis has none, and mythical transformation introduces discontinuity by contrast, while anatomical transformation introduces continuity by deformation.
- The ontological group. With interiority and physicality each similar or different between a subject and an unspecified other: similar interiority, different physicality is animism; different interiority, similar physicality is naturalism; shared elements of both within a class distinct from other classes is totemism; both distinct but with small enough differences for correspondences is analogism.
- Symmetrization. Generalizing a local concept (mana, hau), a local epistemic stance (Dumont, Strathern) or a native doctrine (Tempels, Viveiros de Castro) all fall short; the combinatorial matrix symmetrizes better because its totality is never given in advance and is reconfigured as elements are added, though it requires worldwide data.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Structural analysis can be practised without Lévi-Strauss’s doctrine of mind and sign | moderate | argument; Descola’s own practice |
| C2 | Lévi-Strauss’s kinship analysis is Goethean, though he credited D’Arcy Thompson | moderate | reading of Les structures élémentaires, with Lévi-Strauss’s own statements cited |
| C3 | In myth analysis the transformation group is cut by the analyst, not given by the material | moderate | Lévi-Strauss’s own admission (L’homme nu, 603–4), cited |
| C4 | The four ontologies form a structural group with no privileged variant | weak (as stated here) | summary of Beyond Nature and Culture, not argued in the lecture |
| C5 | Combinatorial comparison is the fairest form of symmetrization | weak | argument by contrast with other strategies, whose shortcomings are “too long to develop here” |
Concepts
- transformation: an ordered passage from one variant to another within a group, preserving invariant relations.
- structure / system: a structure is defined by invariants across sets that allow transformation between them; a system is a set of related elements without that.
- Goethean regime: transformation as development of an initial blueprint (the Urform, the Urpflanze) into its variants.
- Thompsonian regime: transformation as continuous deformation in a coordinate space between forms already given.
- symmetrization: placing the observer’s and the observed’s cultural features on an equal footing.
- aliud: an undetermined other, not yet specified by a relation (after Descombes).
Connections
The lecture reads Lévi-Strauss: Les structures élémentaires de la parenté, Anthropologie structurale (LIT-775, unread), Du miel aux cendres and L’homme nu, with Petitot’s genealogy of structuralism’s morphological sources. Its Goethean regime is, by my connection rather than Descola’s, the one Propp’s Morphology announces in its Goethe epigraphs (LIT-828); Descola does not mention Propp. Its account of meaning by contrast is Saussure’s (LIT-774), which it names as the semiological dimension of Lévi-Strauss’s approach.
Bearing on the record
- On Lévi-Straussian transformation as a method. The lecture supports reading structural analysis as a theory of transformations between variants, with invariants, rather than as a static inventory of oppositions. It also carries a caveat that matters for any use of it as a model of how variants are actually related: in myth, by Lévi-Strauss’s own account, which variants count as transforms of which, and the path between them, are the analyst’s construction. Descola says this; he does not discuss computation or formal modelling.
- THEORY-159 (value fixed by relations among coexisting terms) is Saussure’s; Descola’s “no human phenomenon has a meaning in itself” is the anthropological generalization, asserted, not argued, here.
- No THEORY is filed: the lecture argues a method and reports a reading; it establishes no finding of its own.
- No instruction for machine-learning practice; nothing for the anthology.
Note of 2026-10-16: THEORY-432 now cites LIT-822 as a source.
Limitations
- A keynote with a personal register; it summarizes rather than shows.
- The Goethe/Thompson distinction is Descola’s interpretation; Lévi-Strauss credited Thompson, and Goethe only “in passing”.
- The ontological fourfold is stated, not defended; the defence is in Beyond Nature and Culture.
- Nothing is said about how the two regimes would be told apart in a given analysis, beyond the presence or absence of an initial form.
Open questions
- If the transformation group of myths is cut by the analyst, what would make one ordering of variants better than another? Descola names no criterion.
- Can the Thompsonian regime be made non-arbitrary for myth by fixing stable end forms, as D’Arcy Thompson’s method requires?