Par-delà le structuralisme?

This work is licensed under the Creative Commons | © Michael Lambek. ISSN 2049-1115 (Online). DOI: http://dx.doi.org/10.14318/hau6.3.006


Par-delà le structuralisme?

Response to Philippe Descola’s “Transformation transformed”

Michael LAMBEK, University of Toronto Scarborough


It is a great honor and pleasure to comment on Professor Descola’s paper.1 He is a very gracious person as well as a brilliant and erudite scholar. I am grateful that, unlike Lévi-Strauss’ reading of his dissertation, he let me see his text ahead of time. There is great charisma attached to both their names, and I have to say that what Descola says of his mentor is also true of himself, namely that he is “someone who [is] able to deal with arcane social facts by showing how they followed a complex logical pattern, and doing this with the clear and incisive mind of a great philosopher and the elegance and sensibility of a great writer” (p. 33).

Implicitly, then, Descola’s talk manifests succession between generations within a lineage of great scholars and also implicitly challenges any easy distinction between reproduction and transformation as ideal types. In other words, in elucidating how structuralism comprehends transformation, Descola’s talk enables us to ascertain the transformation of or within structuralism itself. The central ideas of structuralism are exemplified in the transformation from Saussure and Jakobson to Lévi-Strauss, and thence to Descola. I guess these intergenerational transformations could be analyzed as Goethean. I wonder, then, whether the relations between the ideas of those in the same generation as one another, namely between the models of Descola and Viveiros de Castro, are to be taken as Thompsonian?2 Likewise, [46]if we are to take a structuralist lesson in understanding what structuralism is, we can see it only in relations of difference to other models or theories of social phenomena, such as historical materialist or cybernetic ones. A key phrase in Descola’s talk offers a suggestion as to how to begin. “A structure,” he says, “is not a system” (p. 37).

Descola’s paper is deceptively short, with a large number of complex ideas packed in, and if I don’t exercise some discipline myself I will write as long in response. I find myself most in agreement with him when he depicts Lévi-Straussian structuralism and less so the further he himself departs from it.

Let me try to elaborate. Descola offers an extremely lucid exposition of structuralism. Since I have nothing to debate with him here—indeed how could I?—I simply remark on what I consider a blind spot of structuralism. This will provide an entrée to discuss Descola’s work.

I begin with the method. As Descola says, structuralism proposes that

no human phenomenon has a meaning in itself, that it only becomes relevant when it is contrasted with other phenomena of a similar kind – a system of marriage with other systems of marriage, a variant of a myth with other variants of a myth; so that the object of the inquiry is less the description of the phenomenon than the logic of its contrasts. (p. 37)

Descola adds the incisive remark of Jean Pouillon that the structuralist method consists of ordering a set of differences. So, structuralism not only begins with relations rather than discrete objects or meanings, but takes the primary relations to be those that differentiate. It proposes in the first instance a structure of differentiations not of positive relations. These differentiations are generally composed by means of binary, contrastive features.

But binary contrasts are not always evident. During my fieldwork among Malagasy speakers in Mayotte (Western Indian Ocean) I heard various words referring to members of a semantic domain or Wittgensteinian family of not generally visible and possibly nonmaterial beings I have called spirits. I was unable to discover clear binary contrastive features among all of them and hence to order them according to relations of difference. That was partly because the terms and concepts, coming from disparate historical and cultural traditions, were incommensurable to one another. By incommensurable, I mean, following Thomas Kuhn (1962), that they cannot be compared along a single yardstick or according to any neutral external measure. Translated with respect to structuralism, this means they cannot be ordered according to a set of binary differences. This was true not only between kinds of spirits but also between what I called the traditions of Islam, spirit possession, and cosmology in which they were located and that were all a part of life in Mayotte. Local practice entailed articulation of and among incommensurable forms of knowledge (Lambek 1993).

Now, I can learn from Descola’s work that perhaps what these beings have in common is that they are all premised on having a similar interiority but different physicality from humans. And yet, if I look more closely, some emphasize similarities of physicality, and for some the interiority is more or less knowable. The relations are a matter of degree rather than mutual exclusion and the properties and criteria are generally left vague. (It’s doubtless analogism.) To ask too [47]many questions, to challenge the ambiguity, to insist on finding an order and create a typology or paradigm, would be an ethnocentric and destructive act on my part.3 I think that structuralism rests on commensuration. It is brilliant in showing commensurability where at first sight it is not evident, but it cannot readily address contexts where commensurability simply is not there. Incommensurability marks the limit of what structuralism can say. Thus, for Kuhn, it marks the limit of one paradigm and the impossibility of clearly establishing its relationship to a successive or adjacent one. Put another way, incommensurable paradigms cannot be derived as transformations of one another. There is a break between them. Incommensurable does not mean incomparable or untranslatable, but it implies that translation is not exact or self-evident and always leaves a residue.

Lévi-Strauss has taught us that one of the activities of mind lies in making things commensurable to one another, structuring them, or, if the entities are not experienced prior to the thought, creating structures such that they will produce relations and secondarily the beings that are then in relation, and then enacting the kinds of transformations among them that Lévi-Strauss and Descola demonstrate. This has been described, with various levels of abstraction and theoretical consequence, as la pensée sauvage, generative grammar, structuring structures, and even rationalization in Weber’s sense, underlying everything from art through sentences to social hierarchy and bureaucracy.

Nevertheless, if we “give sense to the world by detecting in it salient features that can be organized in contrastive sets” (p. 35), we also reach limits and have to make sense of those limits. These limits or aporia are more widespread than the sort of anomalies Mary Douglas (1966) talked about. One of the limits might be the very incommensurability between the world and language. By “the world” here I include what Descola lists as “animals, plants, physical processes, artifacts, images, and other forms of beings” (p. 35). You will perhaps notice the word that is not uttered here (see below).

I admit I had to look up the definition of gnoseology, which, as you all know, is the philosophy of knowledge and cognition. Descola refers to “the assumption that there exists a physical continuity between, as [Lévi-Strauss] says, ‘the states of subjectivity and the properties of the cosmos’.” I take this as a credo of monism. However, I am suspect of monism taken tout court and I do think that Lévi-Strauss’ monism was always balanced by his repeated adversion to dualism. At the least, a tension between monism and dualism is characteristic of his thought and, I think, of Descola’s and of human existence more generally. (My own predilection is for a monism of the both–and sort, i.e., a monism that somehow encompasses dualism rather than a monism of the either–or sort that excludes dualism. Yet making that distinction itself operates dualistically . . . It is a paradox human thought cannot escape.)

Descola suggests that the relations between language and world are not incommensurable, that there is “a vast array of meaningful differences between qualities and beings that can be systematically organized according to these differences” [48](p. 35). This organization is at first view a mental and human one, but he thinks that is merely a surface manifestation of something deeper and more widespread. I am enough of a naturalist to agree that human cognition is a part of nature, but that does not convince me that the products of mind are always fully commensurable with the world to which it attends. I am uneasy conflating structural order with ecological system, or language with world. (Remember, a structure is not a system.) It is not that they are inaccessible to each other, and not that they cannot be compared or related, only that they are incommensurable. There would be a lot more agreement if it were otherwise.

I do think a tree falling in the forest makes a sound when there is no human being or even no animal to hear it, but I do not know in what sense the relations between trees and forest are part of a structure of commensurable differences until human thought and language make them so. I am also inclined to think, along with Wittgenstein, that if trees or forests were to speak, we would not be able to understand them.

I do not think one can purify anthropology of its anthropocentrism, not only because by definition it is the study of humankind, but also because I take a Gadamerian hermeneutic position that we cannot fully escape our own prejudices. Just as there is no ethnographic study that is entirely purged of its ethnocentrism, or, let us say more charitably, purified of the subject position of the ethnographer (albeit that may have been one of Lévi-Strauss’ goals), so there can be no human assumptions that are not at some level anthropocentric. Descola himself speaks of “the affordances which the world offers to the specifically human dispositions” (p. 40). Anthropocentrism is not equivalent to Eurocentrism; animals are not the equivalent of people in other societies and domestication is not equivalent to colonization. More deeply put, without some level of anthropocentrism we would not be able to understand ourselves. Moreover, from what noncircular position could we confirm with authority that we had surpassed or transcended anthropocentrism?

My position is not one of pure relativism; we can always broaden our horizons. Moreover, I acknowledge the importance of ecology and models that locate human beings within the ecosystem rather than outside or above it. Countless philosophers have also reminded us of the hubris of human reason. I also agree (with some caveats) with the remark of Lévi-Strauss that “the right of the environment, which everyone talks about, is the right of the environment in regard to man, and not the right of man in regard to the environment.”4

I have not read Kohn’s book (2013) and cannot—or, rather, should not—comment on it, but I don’t think it is original to turn to iconic and indexical signs. I quite agree that these processes move us beyond the “language-like properties of social life” (p. 35), but I would like to maintain the distinction between human language and communication. Gregory Bateson (1972) spoke of the communication of information, and defined information (structurally?) as the difference that makes a difference. It occurs at all levels, not only within ecosystems broadly conceived, but also at the infracorporeal and even the cellular level. The point about the symbol, for authors like Suzanne Langer (1942) and Leslie White (1949), was how it took human language and thought to a different level. The location of culture [49]in language and symbols was precisely to say it was neither internal nor material, neither mind nor body. Indexicality was described in the 1960s work on ritualization among animals and Roy Rappaport (1968) showed the cybernetic connections between ritual events of humans and the size of both human and pig populations in Papua New Guinea. Human language and social practice evidently depend on indexical and iconic functions as well as symbolic ones, and not only with respect to parole; the Mythologiques are full of iconic and indexical relations. Here we come to the relation between structuralism and semiotics, on which there are people in this conversation far more qualified than me to comment, but I would draw attention first to the concept of the interpretant and second to the question of whether indexical and iconic signs in nature always find themselves in structural relations, in sets of ordered differences, in the same way that symbolic signs are purported to do.

I think that the urge to evade anthropocentrism can quickly slip into the projections of anthropomorphism, which are themselves displaced and mystified forms of anthropocentrism. (I am speaking here of certain Western thinkers, not Descola and not of the members of societies characterized by animism.) It is one thing to complement structuralism with ecology—which I applaud—and quite another to start talking about forests as being or having selves. We slip here into heady Nietzschean territory of the fluid transformation and ambiguity between the literal and the metaphorical.

Communication is not the same as reasoning. I mean no disrespect to forests when I say I don’t think they think. (They are, of course, welcome to respond.) But the fact that we alone rely on symbols does not make us superior. I do love forests, and especially to lose myself within them. But that is also to say that I have never learned fully how to know the forest from the trees and I am not so sure that forests and trees know how to do so either.

I take agency as including the possibility to do otherwise and to act knowingly. I accept that the tiger deploys strategy when he hunts me. But do glaciers exhibit agency when they melt? Melting ice has causal effects—on humans and also on polar bears and sea levels, and sinking or rising land masses, and so forth, and these have feedback loops. But I think it melts the concept of agency to extend it to melting. And in any case, Lévi-Strauss was uninterested in agency; what humans and other beings have in common was to be found by dissolving man, not by reconstituting other beings in man’s likeness.

How we think about tigers or glaciers takes us directly to Descola’s four ontologies (2013). Thinking about shrinking forests in the Amazon or Madagascar, the question is how to articulate the knowledge from environmental science with what we know about human action and society and also with social justice to formulate a sound political ecology or ecological politics. Taking Descola’s four ontologies seriously, is the question how to recontextualize naturalism from the perspective of animism or totemism, how to articulate insights and practices taken from each of them, or how to formulate an animist politics?

In any case, the middle part of Descola’s talk takes me back to the debates that characterized my graduate student days, debates between structuralists and materialists and the ways that the more powerful thinkers, such as Sahlins, along one track, and Rappaport, along a very different track, attempted to transcend them, [50]via dialectics and cybernetics, respectively. Perhaps the discussions today are more sophisticated or have taken a radically new turn, but I think they illustrate another lesson from Lévi-Strauss, namely that any mythological system—and here I include a tradition like anthropology—is built by adding layers around an intractable core, call that core mind and matter, language and world, stability and change, structure and history, or even culture and nature—to mention the unmentionable word.5 There are central oppositions around which human thought works itself and which it cannot give up. The core oppositions, those significant and complex enough to generate so much talk and argument, but also art, are likely to be incommensurable ones, not those easily placed in binary contrastive relations of difference.

As for seeking the unconscious activity of mind in cultural constructions or by means of cognitive psychology, I do not think these projects need be mutually exclusive. Had cognitive psychology been as sophisticated in Lévi-Strauss’ day, he might have made use of it. But not all cognitive psychology is as sophisticated as it thinks itself, and without serious ethnographic work, like that of Rita Astuti (forthcoming), it remains highly susceptible to ethnocentrism or ontological bias and at the mercy of its techniques and methods. And no amount of cognitive psychology or MRI machines could compensate for Lévi-Strauss’ fascination with myth or for our fascination with what he has to say about and by means of myth, even if it does not provide the royal road into the unconscious, or even if that road is better conceived in psychoanalytic than cognitivist terms.

I return to transformation. Descola is extremely insightful in pointing to the centrality of transformation for Lévi-Straussian structuralism and in showing how it takes two distinct forms, with respect to the work on kinship and myth, respectively. Here I have to admit my ignorance, not having read either Goethe or D’Arcy Thompson on the subject—so I hope you will forgive any misunderstanding that follows.6

One question is whether the two models of transformation are meant to show how we may best organize the diversity we perceive through empirical investigation or to model how diversity was and continues to be produced.

Another question is how the two models of transformation address the relations of structure to history. Is restricted exchange prior to other forms of marriage exchange only logically or also temporally?

The Goethean model is one of the unfolding of possibilities derived from an original form. While these possibilities are described as logical, there is also a hint that there might be a temporal or linear order to them such that one manifestation must be historically prior to another. Are they reversible in the sense that Lévi-Strauss describes marital exchange itself? Or perhaps irreversible in the sense in which Lévi-Strauss can remark in Tristes tropiques ([1955] 1973: 150), “I cherish the reflection . . . of an era when the human species was in proportion to the world it occupied”? The principle of reversibility is more straightforward in the D’Arcy [51]Thompson model, where there is no original form. In the Mythologiques there is only the continuous shuffling of the mythemes, which are equivalent to one another. There is no growth in complexity. Indeed, it is the severity and abstinence of this model in the face of our hegemonic ideologies of change and growth that is so wonderful. The problem is that it leaves intellectual successors to the model with nothing to do but further reshuffling.

Finally, Descola offers us a guide to his magisterial book. This is a brilliant and stimulating model, and its capacity to illuminate is made evident by the wealth of ethnography and philosophy it encompasses. With the waning of primitive worlds to investigate, we witness replacement not only with the questionable romance with animals but also with the significant resurgence of armchair anthropology. In Descola’s book the Goethean core is not a principle or a golden bough but a primary opposition and the set of relations it establishes along two intersecting axes. In my commentary in a previous issue of HAU (Lambek 2014) I called it “the elementary structures of being.” It is constituted by what Descola in his essay calls “the awareness of a duality of planes between material processes . . . and mental states” (p. 40). Note the dualism here, which is actually a doubled dualism since each plane is set on an axis of same and different. Yet Descola does not want to identify his planes with Cartesian body and mind or with anthropological nature and culture. This is because he sees the latter two oppositions as manifestations of only one of the four ontologies his more abstract original pair can generate. I read the par-delà of his title to indicate “beyond” in the sense of moving forward from, enlarging or surpassing, rather than abandoning nature and culture. However, whereas the same and different distinction is both binary and commensurable and also understood as a continuum, a matter of degree of likeness, I suggest the material/mental distinction is an incommensurable one and hence that the two axes are not equivalent to one another. I admit that I am not sure what the consequences of my suggestion are, but they might preempt the following questions.

What is the ontological status of Descola’s model of ontologies? Is it distinct from them? Do the ontologies form distinct worlds? If we think wild gorillas or pet cats share something with us, are we still in the ontology of naturalism or have we stepped into animism or totemism? In what sense is posthumanism also postnaturalism? Also, doesn’t analogism continue to hold everywhere, as, I think, Lévi-Strauss thought? Might Descola entertain the idea of ontological pluralism—or would that be an oxymoron?

In any case, Descola counters all this with the magnificent “basic principle of structural analysis which holds that each variant is a variant of the other variants and not of any of them in particular which would be privileged” (p. 41). And he concludes with the brilliantly enunciated manifesto on symmetrization that reminds us why structuralism continues so profoundly to matter.


Astuti, Rita. Forthcoming. Cognition in the field.

Bateson, Gregory. 1972. Steps to an ecology of mind. Chicago: University of Chicago Press.[52]

Descola, Philippe. 2013. Beyond nature and culture. Translated by Janet Lloyd. Chicago: University of Chicago Press.

Douglas, Mary. 1966. Purity and danger. London: Routledge.

Kohn, Eduardo. 2013. How forests think: Toward an anthropology beyond the human. Berkeley: University of California Press.

Kuhn, Thomas S. 1962. The structure of scientific revolutions. Chicago: University of Chicago Press.

Lambek, Michael. 1993. Knowledge and practice in Mayotte. Toronto: University of Toronto Press.

———. 2014. “The elementary structures of being (human): Comment on Philippe Descola’s Beyond nature and culture.” Hau: Journal of Ethnographic Theory 4 (2): 245–251.

Langer, Susanne. 1942. Philosophy in a new key. Cambridge, MA: Harvard University Press.

Lévi-Strauss, Claude. (1955) 1973. Tristes tropiques. Translated by John and Doreen Weightman. London: Jonathan Cape.

Rappaport, Roy. 1968. Pigs for the ancestors. New Haven, CT: Yale University Press.

Schatz, Adam. 2011. “Jottings, scraps and doodles,” review of Claude Lévi-Strauss: The poet in the laboratory by Patrick Wilcken, London Review of Books 33 (21): 3–7.

White, Leslie. 1949. The science of culture. New York: Grove.


Michael LAMBEK holds a Canada Research Chair in Anthropology at the University of Toronto Scarborough. He is currently a fellow at the Wissenschaftskolleg zu Berlin (2016–17). He has written a number of books, from Human spirits (Cambridge University Press, 1981) through The ethical condition (University of Chicago Press, 2015) and has been a frequent contributor to HAU.

Michael Lambek
Department of Anthropology
University of Toronto Scarborough
1265 Military Trail
Scarborough, Ontario
M1C 1A4


1. The invitation came from Alejandro Paz and the commentary was delivered well before the author became interim coeditor of HAU. I am grateful to Amira Mittermaier for very helpful comments.

2. On Goethean and Thompsonian, see Descola’s essay.

3. I don’t mean to suggest that people in Mayotte don’t ask questions themselves. (For full discussion, see Lambek 1993.)

4. As cited by Adam Schatz (2011). Source and page numbers in Lévi-Strauss not given.

5. At the Collège de France, Descola does hold the Chair in Anthropology of Nature.

6. However, I have heard Goethe cited by contemporary alchemists and practitioners of complementary medicine in Switzerland, practices that in turn exemplify very neatly the analogism of the pensée sauvage.