“Signs and Customs : Lévi-Strauss, Practical Philosopher” Common Knowledge, 22:3, September 2016, pp. 415-430.

Symposium: Anthropological Philosophy

SIGNS AND CUSTOMS Lévi-­Strauss, Practical Philosopher

Patrice Maniglier

Translated by Matthew H. Evans

“Why did we obey? The question hardly occurred to us. We had formed the habit of deferring to our parents and teachers”: it was in this way that Henri Bergson introduced what remains one of the greatest works in the philosophy of the social sciences.1 It had not escaped Bergson’s attention that the social science disciplines, typically in Émile Durkheim’s work, had shown better than philosophy that the real problem posed by obligation, from both a theoretical and an ethical perspective, is that we are hardly aware of it. While philosophy approaches the problem of obligation as one of the legitimate conditions under which subjects can consciously submit themselves to the authority of others, the concept of society constructed by sociology figures obedience as, paradoxically, the more pervasive for its groundlessness. The very concept of society is meant to account for our unawareness of obligation as a problem: “We did not fully realize this, but behind our parents and our teachers we had an inkling of some enormous, or rather some shadowy, thing that exerted pressure on us through them. Later we would say it was 1.  Henri Bergson, The Two Sources of Morality and Religion, trans. Ruth Ashley Audra, William Horsfall Carter, and Cloudesley Brereton (London: Macmillan, 1935), 1.

Common Knowledge 22:3 DOI 10.1215/0961754X-3622260 © 2016 by Duke University Press

415

416 COMMON KNOWLEDGE

society.”2 The social is a stratum of causality irreducible to the causalities of individual and conscious action, which manifests itself in the absence of a question —  that of the legitimacy of authorities. Thus, the object of the social sciences should be to understand better why individuals do things that are imposed upon them for reasons that exceed their own understanding. Claude Lévi-­Strauss counted himself among the inheritors of the legacy of Durkheim, Marcel Mauss, and the French school of sociology. It is tempting to think of him as the third member of a triad — with Durkheim, Mauss, and Lévi-­Strauss each proposing his own response to the problem of obligation. If asked directly, Durkheim would have attributed the power of constraint to collective representations that differ in nature from individual ones. Mauss would have attributed obligation not to the nature of representations but to their participation in a system via the mechanism of exchange.3 As for Lévi-­Strauss, one would imagine him explaining the constraining character of social norms as a logical or, rather, a cognitive constraint. Yet it is easy to show — as Vincent Descombes does, drawing on Wittgenstein’s Philosophical Investigations — that any attempt to substantialize rules by understanding them as constraints ends in aporia.4 (Generalizing an objection often levied against Lévi-­Strauss, Descombes reproaches theoreticians of the social sciences at large for aspiring to “explain” rather than to “understand” behaviors.) To the question of why people do what they feel should be done, Lévi-­Strauss’s actual response was that people always tell us quite clearly why and that there is no reason not to believe them: “To whatever society he belongs, the individual is rarely capable of assigning a cause to this conformity: all he can say is that things have always been like this, and he does what people before him did.”5 The problem of obligation is thus displaced: it is not a problem of understanding why we do what we do but of understanding how we determine what has to be done. The question is how to ascertain what the common practice, habit, or custom really is, and doing so is not as simple as one might think. A collective behavior cannot be defined, as is often supposed, as an observable sequence of actions, common to several individuals maintaining stable relationships. The suggestion that we should look for the causes of observable behaviors at the “cognitive” or “symbolic” level is not the salient contribution that structuralism made to the social sciences. Rather, its contribution was to disclose a problem basic to linguistics, anthropology, history, and the other “sciences of

2. Bergson, Two Sources of Morality, 1. 3.  See Bruno Karsenti, L’homme total: Sociologie, anthropologie, et philosophie chez Marcel Mauss (Paris: Presses Universitaires de France, 1997).

4.  Vincent Descombes, Objects of All Sorts: A Philosophical Grammar, trans. Jeremy Harding and Lorna Scott-­Fox (Baltimore: Johns Hopkins University Press, 1986), 205. 5.  Claude Lévi-­Strauss, Structural Anthropology, vol. 1, trans. Claire Jacobson and Brooke Grundfest Schoepf (New York: Basic Books, 1963), 70.

6.  See Patrice Maniglier, La vie énigmatique des signes: Saussure et la naissance du structuralisme (Paris: Scheer, 2006). 7.  See Simon Bouquet, Introduction à la lecture de Saussure (Paris: Payot and Rivages, 1997); Johannes Fehr, Saussure entre linguistique et sémiologie (Paris: Presses Universitaires de France, 2000); Lucien Scubla, Lire Lévi-­Strauss: Le

417 Anthropological Philosophy •

Maniglier

culture” — a problem pertaining to the nature of their data. Structuralism begins from a negative but profoundly instructive observation that the units of cultural practice (whether matters of speech, ritual, myth, or custom) are not given in any observable way. The universe of cultural practices is defined by an essential and moreover dual variability: the variability in ways of speaking one and the same language, as well as the variability between different languages. We congratulated ourselves for denouncing structuralism as a new Eleaticism, while truly it had done its best to show us that this variability is not contingent but lies firmly within our modes of determining the units that, together, comprise our cultural practices.6 I would like to show here that the Lévi-­Strauss of the Mythologiques series, in redefining structural analysis through the concept of transformation, curiously stumbled back upon the problem that had originally been Saussure’s, namely, the identity of the sign. But I would like as well to bring to light how the construction of a new method, the structural method — given its operative concepts of “difference” and “system” — had to work necessarily through the formulation of a new philosophical problem: the particular kind of identity or unity that is uniquely constituted through the correlation of differences. It is a problem that we might well call ontological. There has been recently a renewal of interest in reading those authors — Ferdinand de Saussure, Roman Jakobson, Nikolai Trubetzkoy, and Lévi-­Strauss — who had been identified, for a time, with structuralism, though only on condition that they be segregated from the structuralist movement, which then can be relegated to the status of a passing fad.7 What seems to me necessary, however, is the reverse procedure: to reclaim these authors for the structuralist movement and to do so from the vantage point of a plainly philosophical problem that intersected with the “cultural sciences.” The coherence, which by all accounts was uncertain and contestable, of what became known as structuralism must be sought in the internal imbalances and overemphatic claims that the discovery of a philosophical problem within their own methodology imposes on “theories of practice.” Semiology is less the name for a new theoretical realm than it is a designation for the recurrence of this philosophical problem within a diverse group of disciplines. There is not much to gain by opposing a “good” ( purely methodological) structuralism against a “bad” structuralism that engages in philosophical speculation. Only by coming to understand its articulation of both methodological projects and speculative con-

déploiement d’une intuition (Paris: Odile Jacob, 1998); and Patrick Sériot, Structure and the Whole: East, West, and Non-­Darwinian Biology in the Origins of Structural Linguistics, trans. Amy Jacobs-­Colas (Boston: Walter de Gruyter, 2014).

418 COMMON KNOWLEDGE

structions will we come to discover the real problems that structuralism poses. It is a great merit of structuralism to have shown, in passing, that philosophy need not be condemned forever to track backward toward first principles (in indifference, very often, to whatever the philosopher is seeking to ground) and that philosophy could become a necessary recourse in the process of constituting any new intellectual discipline.

Saussure and Philosophy: The Identity of the Sign as an Ontological Problem We are only now beginning to see that we know nothing, or almost nothing, when it comes to Saussure. The discrepancy is so great between the Saussure depicted in the study guides to structuralism and the Saussure of the critical editions and scholarly commentaries that it is easy to become discouraged about reopening his dossier. A coherent interpretation of Saussure’s endeavor is nevertheless possible, as long as we are clear that he was not concerned as much with meaning as with signs. His ambition was never to establish a general theory of signification. Wittgenstein was right to say that we can do all kinds of things with signs, among which what we call “signifying” or “communicating” is only one game among others. But Saussure’s question is simpler and more immediate: we have difficulty talking about the manipulation of signs because their identification poses formidable problems. “Every time I utter the word Messieurs,” Saussure writes, “I renew its material being: it is a new act of phonation and a new psychological act. The link between the two uses of the same word is not based upon material identity, nor upon exact similarity of meaning, but upon facts the linguist must discover, if he is to come anywhere near to revealing the true nature of linguistic units.”8 What must be understood is that, with each reiteration, everything changes: it is not a matter of simple variation, since the preservation of even a minimal “kernel” becomes impossible, thus making it impossible as well to separate the accidental from the essential in the sign. If the sign is real, then it is an unobservable reality — unobservable, in the sense that it is unmeasurable, experimentally indemonstrable. The history of linguistics after Saussure’s day would demonstrate the futility of any contrivance put in place to measure the empirical domain of the sign. In 1943, during the first of the “Six Lectures on Sound and Meaning” that he gave at the New School for Social Research in New York — lectures at which Claude Lévi-­Strauss was a very attentive auditor —  Jakobson explained that a radiographic film, shot in the 1930s, of the vocal apparatus in action demonstrated for the first time that the observable units of the articulatory gestures did not match the units of phonological perception.9

8.  Ferdinand de Saussure, Course in General Linguistics, trans. Roy Harris (1916; repr., London: Duckworth, 1983), 128 – 29.

9.  Roman Jakobson, Six leçons sur le son et sur le sens (Paris: Minuit, 1976).

419 Anthropological Philosophy •

Maniglier

But Saussure, long before this experimental validation of his viewpoint, had posited that linguistics could not have an object like those of any other empirical science, and for a reason both simple and vertiginous. To distinguish and classify the sound bread, for example (in the sentence, whose sounds likewise would need to be distinguished and classified, “I would like some bread”), it is not enough to record the distinctive empirical particularities of the sound. A signification must be associated with it, or rather we must be able to associate one or more audible differences with one or more differences on an entirely different plane, the precise nature of which matters little, at least not at the outset. (It could be the visual plane or the psychological, for example — we might call this plane semantic.) What matters is the concomitance or correlation of two differences or series of differences. Saussure does not simply claim that the sign is defined by the set of its distinctive features; he moreover shows that, for there to be a difference between the sounds bread and dread, there must be another difference of an entirely distinct order. In the linguistic sense, therefore, a structure is not, nor did any authentic structuralist ever claim it to be, a formal system of rules — a syntax, in other words. A structure is, rather, a system constructed by the reciprocal definition of two systems of differences. A structure, in the logical sense of a formal system, may well be what Noam Chomsky disclosed in his Syntactic Structures (1957), but of course Chomsky was never a structuralist, if by structuralism we mean the expansive and problematic movement that passed from linguistics to anthropology and then from anthropology to the culture at large. The centrality of the concepts of “difference” and “biplanarity” is too often underestimated by those thinkers whose concept of structure bears little correspondence to any that emerges from the actual practice of structural analysis, since generally they prefer to look anywhere than to structuralist authors themselves for a definition of the term.10 Around the “question of the sign,” scientific knowledge and philosophical reflection come together. In a handwritten note on Germanic legends, cited more than once by Lévi-­Strauss, Saussure writes: It is true that in going more deeply into things, we notice in this field, as in the related field of linguistics, that all incongruities of thought spring from inadequate reflection on the nature or identity, or the characteristics of identity, when we are dealing with a non-­existent entity like 10.  The philosophers closest to the structural method are not those who, like Gilles-­G aston Granger (in Formal Thought and the Sciences of Man [Dordecht: Reidel, 1983]), make the strongest claims to be so. Authors such as Gilles Deleuze, Jacques Derrida, Michel Foucault, and Louis Althusser understood perfectly well that “difference” was the key issue for structuralism. See, in particular, Deleuze, “How Do We Recognize Structuralism?” in

his Desert Islands and Other Texts, 1953 – 1974, ed. David Lapoujade, trans. Michael Taormina (Cambridge, MA: MIT Press, 2004). See also Manfred Frank, What Is Neostructuralism?, trans. Sabine Wilke, Martin Schwab, and Richard John Gray (1984; repr., Minneapolis: University of Minnesota Press, 1989). The notion of “poststructuralism” is misleading, since it embraces all of the most faithfully structuralist thinkers.

420 COMMON KNOWLEDGE

the word, or the mythic person, or a letter of the alphabet, which are merely different forms of the SIGN, in the philosophical sense.11

It follows, then, that a philosophical clarification is needed in order to get past the impasses in which linguistic thought tends to gets trapped: a clarification that leads to the recognition that the problems posed by linguistics — problems as methodological as they are philosophical — are common to an array of other phenomena. Semiology, therefore, is not defined by the hypothesis that its objects share a common function, that of signification, but instead by the revelation that the nature of those objects — their “mode of existence,” to use an expression coined by Gilbert Simondon and recently generalized by Bruno Latour — is such that they cannot be repeated without variation.12 Thus, Saussure’s note continues: The graphic individual and, by the same token, the semiological individual will not, unlike the organic individual, have a means of proving that it has remained the same because it is built in its entirety upon a free association. As we see, the incapacity to maintain a certain identity cannot ultimately be accounted for by the effects of time — t hat is the remarkable error of those interested in signs — rather it is deposed in advance in the very constitution of the entity that we pamper and observe like an organism, when it is really only the ghost that emerges from the fleeting combination of two or three ideas. It’s all a matter of definition. . . . We must come to see, one piece at a time, that it is on this fundamental nature of these entities that mythography generally works its reason.13

Saussure’s problem was not how to retrieve the true linguistic identities behind the various ways of speaking but, rather, how to achieve an “epistemology of comparative grammar.”14 What comparative grammar does, when it shows that languages as seemingly different as Latin, Gothic, and Sanskrit were originally “one and the same,” is to demonstrate that, in trying to speak one language, we end up speaking another. “French does not come from Latin,” Saussure declared in his inaugural conference at the University of Geneva, “it is Latin.”15 11.  Cited in Claude Lévi-­Strauss, The Origin of Table Manners, vol. 3 of Mythologiques, trans. John Weightman and Doreen Weightman (Chicago: University of Chicago Press, 1978), 315. 12.  Bruno Latour, Enquête sur les modes d’existence (Paris: La Découverte, 2012). 13.  Ferdinand de Saussure, “La légende de Siegfried et l’histoire Burgonde ( présentation et édition de Béatrice Turpin),” trans. M. H. Evans, in Saussure, ed. Simon Bouquet (Paris: Éditions de l’Herne, 2003), at 387; translator’s version.

14.  See Simon Bouquet, Introduction à la lecture de Saussure (Paris: Payot and Rivages, 1997); Johannes Fehr, Saussure entre linguistique et sémiologie (Paris: Presses Universitaires de France, 2000); Arild Utaker, La philosophie du langage: Une archéologie saussurienne (Paris: Presses Universitaires de France, 2002); and Patrice Maniglier, La vie énigmatique des signes: Saussure et la naissance du structuralisme (Paris: Léo Scheer, 2006). 15.  Ferdinand de Saussure, Writings in General Linguistics, ed. Simon Bouquet and Rudolph Engler, trans. Carol Sanders and Matthew Pires (New York: Oxford University Press, 2006), 101.

Beyond Causes and Reasons: The Identity of What Is Done It is the relationship of this dual problem to that of obligation that motivated Lévi-­Strauss’s extension of the structural method into the domain of the social sciences. Just as Saussure has it that “we say man and dog because our predecessors said man and dog,”19 so Lévi-­Strauss holds that the subject is right to say that “he does what people before him did.”20 If this response to the question “why?” seemed “perfectly veracious” to Lévi-­Strauss, that is because it shows how the question is not as much about the causes of an action as about its definition. Likewise for Saussure, the problem of learning a foreign language is less a matter of comprehending sentences than one of perceiving them. And just as, for Saussure, the key problem of linguistics is not one of meaning but one of semiosis, for Lévi-­Strauss the real problem that the social sciences face is that of demarcat16.  For a more thorough rereading of Saussure informed by this hypothesis, see Maniglier, La vie énigmatique des signes. 17. Saussure, Writings in General Linguistics, 154.

18. Saussure, Course in General Linguistics, 16. 19. Saussure, Course in General Linguistics, 108. 20.  Lévi-­Strauss, Structural Anthropology, 70.

421 Anthropological Philosophy •

Maniglier

It was his discovery that the repetition of linguistic signs engenders their transformation that led Saussure to show, retroactively, that synchronic identity itself is not guaranteed. The form that his editors gave to his Cours de linguistique générale obscured one of the richest aspects of his thought: the idea that linguistic variability stems from the sign’s own mode of determination. His concept of a “system of linguistic values” was meant to model the logic of what is destined for variation.16 Saussure encouraged philosophers to take the full measure of the positive discoveries of comparative grammar: “The crucial response of the study of language concerning the theory of signs, and the eternally new horizon it will open up . . . , will be to impart to that theory a whole new aspect of the sign, the fact that the sign can only begin to be truly known when it is understood that it is something not only transmissible, but intrinsically destined to be transmitted, modifiable.”17 This philosophical generalization enables both a redefinition of comparative method and its extension into other domains: the study of folklore, myths, legends, customs, and so forth. Saussure thus enables us to take intrinsic variability as a property defining the object of the “cultural sciences.” Our doing so, moreover, would parallel his definition of semiology as “a science that studies the life of signs within society” — the study, in other words, of how signs transform themselves as long as they are in “circulation.”18 Saussure’s elaboration of what would become the structural method thus poses and resolves a pair of related problems. The first concerns the determination (rather than interpretation) of signs — the determination of what it is that gets said. The second concerns how the identity of what it is that gets said is such that its repetition brings about, whether over the long or the short term, its variation.

422 COMMON KNOWLEDGE

ing — within the continuity designated in The Savage Mind as “praxis” (a word no less adequate for being old-­fashioned) — the boundaries of an action, a practice, or that which is commonly done. In the end, “that is our way” is the only valid answer. In effect, doing anything — getting married, looking at one’s watch, writing a philosophical article, or even committing suicide — is something only insofar as it actualizes a cultural identity, a way of doing what is done. Similarly, each way of saying Messieurs is a way of realizing a sign and would not even be perceptible if we could not recognize in it the actualizing of a latent possibility. The identity of what it is that gets done is thus as problematic as the identity of the sign. It is, moreover, a fundamental question for the social sciences, as even a quick rereading of Durkheim’s Le suicide will attest. In his introduction to that foundational text, Durkheim looks for an objective working definition of suicide, dismissing the “ordinary” meaning, which he treats as a vulgar usage. An objective definition is requisite for the implementation of the statistical approach that Durkheim defends. The point is to address each individual suicide as an occurrence of a generic event, entirely disregarding the qualitative diversity that the agents themselves would be inclined, necessarily, to recognize (the way that each suicide was carried out, any motivations alluded to, the personality of the actor, and so on), in order just to count them. Even my own suicide is only one suicide, which is to say yet another suicide. It is with an eye to reducing the acts to so many instantiations of a category, distinguishing one from the other in no other than numerical terms, that Durkheim constructs an objective definition (in the format of the formula suicide = x). Approaching actions as mute facts, susceptible of being recorded, nevertheless raises a number of difficulties that many of Durkheim’s readers soon encountered.21 In the end, Durkheim himself was obliged to acknowledge some intentional or subjective factors: refusing to define suicide in accordance with conscious motivations, he defined it instead as committing an action that one knows will lead to death. Whether it is a case of the noble soldier or of the hapless man who throws himself in front of a train, the act is committed “en connaissance de cause.”22 But the problem remains intact: how do we know that the subject knew? And is the definition of death not equally vulgar and “ordinary” as that of suicide? Is there no risk, in other words, of the new definition becoming no less equivocal than the old? Second, the new definition seems to contradict the statistical data at Durkheim’s disposal. From an ethnomethodological perspective, authors such as Jack Douglas and J. Maxwell Atkinson or Harvey Sacks have insisted that the statistics do not represent the number of times that individuals have committed suicide but the number of times that deaths have been classed as suicides by those

21.  The problem was raised as early as 1930 in Maurice Halbwachs, Les causes du suicide (Paris: Félix Alcan, 1930), 339 – 50.

22.  Émile Durkheim, Le suicide: Étude de sociologie (1897; repr., Paris: Presses Universitaires de France, 1930), 3, 5.

23.  See Bernard Conein, “Le suicide est-­il une chose sociale? Catégorisation, classification et connaissance sociale,” Raisons Pratiques 11 (2000): 87 – 106. 24.  See Pierre Bourdieu, Distinction: A Social Critique of the Judgement of Taste, trans. Richard Nice (London: Routledge, 1984), 481 – 86; and Bourdieu, Practical Reason: On the Theory of Action, trans. Randall Johnson (Stanford, CA: Stanford University Press, 1998), 31. 25.  This simplification of Durkheim is the basis of the criticism against him leveled by Atkinson, Douglas, and Sacks and then reprised by Conein. See Conein, “Le suicide est-­il une chose sociale?”; John Maxwell Atkinson, Discovering Suicide: Studies in the Social Organization of Sudden Death (London: Palgrave Macmillan, 1983); Jack

423 Anthropological Philosophy •

Maniglier

tasked with doing so. The real sociological problem, then, is the institutional or cognitive procedures that inform the agents responsible for classing certain deaths as suicides.23 Pierre Bourdieu, for his part, held that statistical method presupposes the nature of what it is that is in question — namely, belonging or not belonging to a determinable class — and made a sociological object of “classification struggle” (“lutte des classements”).24 Thus, if Durkheimian method in itself entails a philosophical problem, it is not, as some philosophers complacently have it, because of a pretense to explain (rather than understand) human actions by their objective causes but, rather, because Durkheim wants to define those actions objectively, as though the event itself bore observable markers that would permit its relegation to a generic category. Yet it would be incorrect to say that, with this definition of suicide, Durk­ heim purports to identify a homogeneous “social thing” or that evidence of a relatively constant suicide rate is indicative of a unique social cause.25 On the contrary, Durkheim endeavors to dissolve his own category by showing that the statistical rate conceals a qualitative diversity among suicides. In studying statistical correlations, we can shed light on this diversity only by focusing on the dependence (covariation) or independence of statistical series: “One result now stands out prominently from our investigation: namely, that there are . . . various forms of suicide” : “egoistic,” “altruistic,” and “anomic” types.26 Durkheim’s ambition is less to explain why individuals commit suicide than to construct a scientific means of classifying suicides, and the approach by way of causality is only one path to this taxonomic end. In effect, what classification demands is a “reverse method” that favors an “etiological classification” over a “morphological” one, a refusal, in other words, to class suicides on the basis of qualitative dissimilarities between the internal properties of each act and a decision to retain only the statistical correlations between circumstances external to the act itself (the agent’s belonging to such and such a profession, age range, region, and so forth).27 From this point of view, the relatively arbitrary character of the indicator is ultimately

Daniell Douglas, Social Meanings of Suicide (Princeton, NJ: Princeton University Press, 1967); and Harvey Sacks, “Sociological Description,” Berkeley Journal of Sociology 8 (1963): 1 – 16. 26. Durkheim, Le suicide, 312. 27.  On this point, see Durkheim, Le suicide, 139 – 43: “Malheureusement, une classification des suicides raisonnables d’après leurs formes ou caractères morphologiques est impraticable. . . . Mais nous pourrons arriver à notre but par une autre voie. Il suffira de renverser l’ordre de nos recherches. En effet, il ne peut y avoir des types différents de suicides qu’autant que les causes dont ils dépendent sont elles-­mêmes différentes.”

424 COMMON KNOWLEDGE

of little importance, as long as it is homogeneous, since the aim is to illuminate variations among suicide rates in relation to other parameters and, through those correlations, to show their relation to entirely heterogeneous statistical tendencies. What matters is less the absolute volume of suicides than their distribution relative to various factors involved.28 On the one hand, this method allows us no more than to “assume the diversity of types without being able to identify them. It may prove their existence and number but not their special characteristics.”29 Yet Durkheim, on the other hand, purports to retrieve, at least to an extent, a qualitative diversity by means of the quantitative distribution of suicides.30 We thus are enabled to show that some suicides (altruistic suicides, for example) are themselves only variants of other acts (homicide, in this instance), and thus we can extract, from the apparent diversity of everyday activities, the social virtualities (“tendencies”) that are actualized variously in relation to context — in relation, that is, to the entities that comprise the social realm. The “objective” definition, far from presupposing an essence of suicide, is a methodological condition that enables us to approach human phenomena en masse and to have them function as variables. We thus go from the indeterminate to the general, from quantity to quality, number to nature, cause to effects, explanation to comprehension. The problem posed by Durkheim’s approach, therefore, less involves his hypostatizing a statistical category into a social one than his supposing that empirical or observable properties of suicide enable, at least initially, the identification of the act, on the basis of objective criteria, with a mass of other acts. The structural method, as formulated in Lévi-­Strauss’s Mythologiques, has as its origin the radical questioning of just this possibility. It is to this end that Lévi-­Strauss, in The Origin of Table Manners, opposed structural and historical methodologies: The difficulties start with the defining of the facts. At no point does the historical method raise the question of what constitutes a fact in folklore. Or, to be more exact, it accepts as factual any element that the observer’s subjective appreciation of the ostensible content of the narrative takes to be such. Little or no attempt is made to effect a reduction that would show how two or more themes, superficially different from each other, stand in a transformational relationship to each other, with the result that the status of scientific fact is not attributable to each particular theme or to such and such themes but rather to the schema which gives rise to them, although this schema itself remains latent.31

28.  On the pertinence and limits of Durkheim’s critics, see Christian Baudelot and Roger Establet, Durkheim et le suicide (Paris: Presses Universitaires de France, 1984), 46 – 75.

29. Durkheim, Le suicide, 142. 30. Durkheim, Le suicide, bk. 2, chap. 6. 31.  Lévi-­Strauss, Origin of Table Manners, 228.

And so what appears to be a porcupine in one region may be semiotically what appears, in a different region, to be a grebe. Identity, in social or cultural sciences, must be structural, which also means that it will be discernible only by way of a comparative methodology. It is from this perspective that Lévi-­Strauss claims a Saussurian filiation for himself, while citing the manuscript of Saussure’s, quoted earlier, on the identity of the sign as a philosophical problem.33 It follows that, if anthropology participates in a more general semiology (given that “all the phenomena of interest to social anthropology can actually be characterized as signs”), its participation stems not from those phenomena having a communicative function of the kind that linguistic phenomena possess but, rather, from their nature posing the same irreducibly philosophical and methodological problems as linguistic phenomena do.34 To be sure, linguistic phenomena, like anthropological ones, can be determined, individuated, singularized, only as variations and not in accordance with shared observable properties. It would not be productive, therefore, to criticize, as Descombes does, the so-­called structuralist conception of signification.35 For the real stake of structuralism consists in its having relativized the question of meaning, its having shown that the “cultural sciences” can be regarded as having a specific unity because they face, whether or not they are aware of it, the same primary methodological problem — that of how to define their data. What still needs to be explained is why the method put in place by structuralism to resolve this problem is one with the structuralists’ definition of their own subject matter as the “sign.” We must explain, more precisely, why the expansion of the structural method from linguistics into social anthropology had to pass through the famous dictum of Lévi-­Strauss that we must look not for a “sociological theory of symbolism” but for a “symbolic origin of society.”36 32.  Lévi-­Strauss, Origin of Table Manners, 200. 33.  Lévi-­Strauss, Origin of Table Manners, 315. 34.  Claude Lévi-­Strauss, Structural Anthropology, vol. 2, trans. Monique Layton (New York: Basic Books, 1976), 2:11.

425 Maniglier

•

The task I have undertaken is quite different: it consists in proving that myths which are not alike, or in which the similarities seem at first sight to be accidental, can nevertheless display an identical structure and belong to the same group of transformations. It is not, then, a question of listing common features, but of demonstrating that, in spite of their differences, and perhaps even because of them, myths which seem at first to present no similarities, proceed according to the same principles and originate from a single group of operations.32

Anthropological Philosophy

The structural method, on the other hand, grounds neither the identity of nor the differences between narratives on their similarities, because those similarities are apparent only to the ethnographer and hence are subjective:

35.  Vincent Descombes, Grammaire d’objets en tous genres (1983; repr., Paris, Minuit, 1996), n. 179. 36.  Claude Lévi-­Strauss, Introduction to the Work of Marcel Mauss, trans. Felicity Baker (1958; repr., London: Routledge, 1987), 21.

426 COMMON KNOWLEDGE

Practical Life and Semiological Life The central problem for the structural method is that we cannot presuppose the nature of the features of any given behavior that will be pertinent to its definition. The structural method contradicts that of Durkheim almost term for term. Durkheim begins with a preliminary and abstract definition of suicide, so as to reestablish a qualitative diversity by examining relations between quantitative series, whereas Lévi-­Strauss starts with a particular myth and analyzes the manner in which the qualitative variations that it exhibits in relation to other versions are concurrent with one another. Lévi-­Strauss’s interest is in the parallel transformation of narrative “motifs,” for a change “never occurs alone but always in correlation with other changes.”37 These correlated transformations bring to light a “system of compatibilities and incompatibilities” that enables us to define each motif not by its substantial quality but by the distribution of oppositions that it actualizes.38 It is of no importance whether feature A is positive or negative, as long as you can show that, when it is positive, feature B is negative and, inversely, that, when it is negative, feature B is positive. Just as for Saussure distinctive phonetic features can be considered pertinent only insofar as they are associated with distinctive semantic features, it is the correlation between distinctive features that, for Lévi-­Strauss, enables us to deduce those features pertinent for the actors themselves. Each version, it follows, can be redefined according to its relative position within a system of transformations, in relation to those terms for which it can be substituted. Or, more precisely, a system of transformations can be defined only within a system of systems of transformations. It is in this way that Lévi-­Strauss was driven, step by step, to expand and develop structural analysis so that it could encompass more and more myths of the Americas — all in order to understand the single Bororo myth with which the first volume of Mythologiques opens.39 In pursuit of this quarry, particular contents (this mythical narrative or that kinship formula) are progressively reduced to algebraic values, definable in formal terms according to their position within a structure, without requiring us to presuppose the nature of a variable (suicide = x) under which different contents would be subsumed. The variable is captured progressively, in step with its qualitative variations, by studying the correlations between those variations that bring to light the individual contents as variants of one another. We are not moving from the indeterminate to the diverse but from the diverse to the structural.

37.  Claude Lév i-­S t rauss, L’homme nu, vol. 4 of Mythologiques (Paris: Plon, 1971), 604. 38.  Claude Lévi-­Strauss, Anthropologie structurale deux (1973; repr., Paris: Plon, 1996), 162. 39.  For the grounds and consequences of this approach, which Lévi-­Strauss laid out in his introduction to The Raw and the Cooked, see Patrice Maniglier, “L’humanisme inter-

minable de Claude Lévi-­Strauss,” Les Temps modernes 609 (2000): 21 – 241. A further elaboration of the “transformation group” concept is found in Patrice Maniglier, Le Vocabulaire de Lévi-­Strauss (Paris: Ellipses, 2002), and Maniglier, “L’être du signe: Linguistique et philosophie dans le projet sémiologique de Ferdinand de Saussure” (PhD diss., Université de Paris-­Nanterre, 2002), www.theses .fr/2002PA100121.

What Lévi-­Strauss calls a structure is not a schema of relations among terms reduced to abstract variables (variables indifferent, in other words, to the contents that might substantialize the structure). Rather, a structure is for him the “transformation group” within which each one of the contents is determinable as a variant. Thus, if contents are said to be “structured,” that is not because they have been subjected, as though from outside, to the imposition of an abstract form but, rather, because contents can be defined only in relation to one another.43 A structure is not an ensemble of rules that can be isolated from their applications; it is a field of virtualities on which each element emerges as a variant, as each actualizes a relative possibility. It is in this respect that the practical units of structural analysis (mythical narratives, ritual acts, vestimentary customs, and even technologies) can be regarded as signs, conforming in this way to the definition given by C. S. Peirce and reprised by Lévi-­Strauss: “That which replaces something for someone” — or, in other words, that which is equivalent to another sign by way of several systematically interrelated transformations.44 It is an essential property of the sign to be potentially other. Hence, if a tool such as a stone axe is a sign, as Lévi-­Strauss said it was in his inaugural lecture at the Collège de France, it is only insofar as, “[in] a given context, and for the observer capable of understanding its use, it stands for the different implement which another society would use for the same purpose.”45 40.  Lévi-­Strauss, Structural Anthropology, 1:316n62. 41.  See Claude Lévi-­Strauss, The Naked Man, vol. 4 of Mythologiques, trans. John Weightman and Doreen Weightman (1973; repr., New York: Harper and Row, 1981), 676, where he writes that the structural method has a “symmetrical and inverted relationship to statistical analysis: it tends to replace quantitative precision by qualitative precision, but in either case precision is only possible as an aim because both forms of analysis have at their disposal a multiplicity of cases which display the same tendency to organize themselves spontaneously in space and in time.” 42.  Lévi-­Strauss, Structural Anthropology, 2:116.

427 Anthropological Philosophy •

Contrary to formalism, structuralism refuses to set the concrete against the abstract and to recognize a privileged value in the latter. Form is defined by opposition to material other than itself. But structure has no distinct content; it is content itself, apprehended in a logical organization conceived as a property of the real.42

Maniglier

Thus, Lévi-­Strauss insists that the structural method is “Galilean,” as it aims “to uncover the law of variation, in contradistinction to the ‘Aristotelian’ outlook, which is concerned primarily with inductive correlations.”40 The “logic of qualities” within which structuralist method is placed distinguishes it from statistical method as much as from the method of formalism:41

43.  See Jocelyn Benoist, “Structures, causes, et raisons: Sur le pouvoir causal de la structure,” Archives de Philosophie 1 (2003): 73 – 88, where he argues that “form for the structuralists is the contrary of essences” because it is not “tied to the nature of such and such a content, but to their indifference.” Benoist’s analysis would gain much from marking the distinction between form in Lévi-­Strauss’s sense and in the sense of “grammatical rules,” since Lévi-­ Straussian form is, one might say, made of contents. 44.  Lévi-­Strauss, Structural Anthropology, 2:11. 45.  Lévi-­Strauss, Structural Anthropology, 2:12.

428 COMMON KNOWLEDGE

A behavior is not “symbolic” because we have accorded it the a priori function of bearing a “signification” (however you understand that function) but, rather, because the behavior is defined solely by that for which it is substitutable. (Consequently, as I have indicated, it can be apprehended only by means of comparative methodology.) When Lévi-­Strauss defined anthropology as a theory of the universal laws of the human mind — and identified those laws in turn with the “symbolic function” — his intention was not to illuminate the “mental constraints” that, for example, compel subjects to reciprocate a gift that they have been given ( pace Descombes). He was looking for what it is that enables subjects to construct the “incorporeal” units or identities (in Saussure’s sense) — to become sensitive to nonobservable units in accordance with strictly differential parameters. Symbolic thought is, above all, a way of organizing sensible reality by bringing to the surface entities that correspond to no substantial invariance. To do so is possible because reversing the values of parameters (high/low, raw/cooked, and so on) is sufficient to produce the same sign. But how does this concept of the sign open onto the construction of a theory of practice? An action (the narrating of a myth, for instance) can be grasped in its particularity only by the actors themselves, insofar as the action is a realization or instantiation of a purely differential entity that could have been actualized differently, given that the agents inhabit a symbolic system conceived as a field of life possibilities. (In this example, the symbolic system would be the “virtual mythological system” within which the given version of the narrative is determinable.) A different absolute value could be assigned to the oppositions structuring the narrative without changing anything as to their correlations. An action can take place only within the context of a practice, understood as a virtual system of customs. Hence, to act is always to do what is usually done — that is, to realize a common practice. There is no other way of determining a unit of practice, which is not an empirically verifiable action. A custom is not an observable behavioral sequence (an empirical type) that gets repeated by several individuals and whose occurrences (tokens) can be counted. Rather, a custom is a purely differential virtuality that is actualized in diverse ways and can be defined only in distinction from other customs. So if, as Lévi-­Strauss has it, we continue doing what is done, the reason is not that we are obedient to norms that influence us unconsciously but, rather, that an action is always a performance — not in the Chomskian sense of an application of rules that together define a competence but in the musical sense of realizing what could be realized otherwise. Even suicide is a performance (it is a common practice), and the problem that an anthropology of suicide faces is not how to explain why an agent performed an act whose nature, without much examination, we believe that we understand. The problem, on the contrary, is how — based on correlative variations that bring to light the features relevant in a given context — to reconstruct the system of practices that define the act.

46.  It is worth noting, along these lines, that many who have attempted suicide “deny having wanted to die, either affirming that they had envisioned something else, or seeming to afford no significance to what happened to them” (Geneviève Morel, introduction to Clinique du suicide, ed. Morel [Toulouse: Erès, 2004] 13). 47.  The work of Descola illustrates well how the theory of practice belongs to structural anthropology. See, for example, Philippe Descola, La Nature domestique: Symbolisme et praxis dans l’écologie des Achuar (Paris: Éditions de la Maison des Sciences de l’Homme, 1986), 136 – 68. 48.  Here I have in mind Leon Vandermesch’s essay “Suicide in China” (in Morel, Clinique du suicide, 53-63) and Maurice Pinguet’s admirable book Voluntary Death in

429 Anthropological Philosophy •

Maniglier

If we are in need of a theory of practice, and if acquiring it demands an innovation in method, that is because, as Christ said on the cross, “they know not what they do.”46 Their unawareness may be no reason to forgive them or even to resent them, but it is a reason to practice structural anthropology.47 By its methods we discover that, while for Durkheim’s contemporaries the act of a soldier who consents to die is identical to that of a woman who consents to give birth, for us today the act of a hopeless man who gives up on his own life is identical to that of a driver who forgets that others’ lives are in his hands. These cases are identical not because they resemble one another but because they are opposed. Moreover, terms that appear to be identical may reveal themselves, from a structural point of view, to differ significantly. The act of suicide in China is not of the same nature as it is in France, insofar as the oppositions between life and death, femininity and masculinity (and so forth) are not distributed in the same ways.48 An anthropological inquiry would likely account for permanent features as well as statistical variations, without having recourse to the psychological explanations that Durkheim, and many other sociologists after him, eventually found themselves compelled to accept. We would do well, therefore, to recall the contradictions that Wittgenstein, in the Philosophical Investigations, exposed as arising from belief that rules of behavior are distinct from their so-­called applications. Starting from this insight, he concluded that to obey rules is only to be caught up in practices and that practices are necessarily collective.49 It must be admitted that, while Wittgenstein’s tendency is to think of a custom in terms of accustomization to a given practice, his evocation of the practical is not a response to a well-­formulated philosophical problem but, rather, is the philosophical exposure of a bogus philosophical problem (the problem of formulating a general definition of language). On the other hand, what Lévi-­Strauss’s extension of the structural method to all cultural data demonstrates is, first, that a practice is neither the consistent utilization of

Japan, trans. Rosemary Morris (Cambridge, MA: Polity, 1993). 49.  See Ludwig Wittgenstein, Philosophical Investigations, 4th ed., ed. P. M. S. Hacker and Joachim Schulte, trans. G. E. M. Anscombe, Hacker, and Schulte (1953; repr., Oxford: Blackwell, 2009), §202: “That’s why ‘following a rule’ is a practice. And to think one is following a rule is not to follow a rule. And that’s why it’s not possible to follow a rule ‘privately’; otherwise, thinking one was following a rule would be the same thing as following it” (88). Again: “A person only goes by a signpost only in so far as there is an established usage, a custom” (86). More generally: “To follow a rule, to make a report, to give an order, to play a game of chess [one might add: to commit suicide], are customs (usages, institutions)” (87).

4 30 COMMON KNOWLEDGE

a material thing nor the repetition of a behavior and, second, that any attempt to theorize our own practices must reconstruct the philosophical concepts of identity, unity, virtuality, and so on. What defines a practice is not as much there being agents to perform it (even if we suppose, in agreement with Bourdieu, that agents too are structurally defined in relation to each other) as there being things to do. The intractability of a theory of practice is wrapped up in the mode of being proper to it. A custom is of necessity collective not because its identity might be reduced to a generic concept but, rather, because its identity is negotiated in the copresence of diverse variants. Another advantage of the structural approach is its enabling us to see that the identity of a custom cannot be fixed by the representation that subjects make of it. Its identity, having no interiority, is determined, outside of the subject, by other possible customs. It is sufficient that they change for the value, which is to say the definition, of a usage to change. This simple yet very complex intuition, according to which we cannot presuppose the nature of a practice by relying on resemblances that we are inclined to notice within it, is at the heart of any structuralist approach. Was it not this intuition that led Foucault to argue, in Discipline and Punish, that the prison as a punitive procedure changed in nature as soon as it was aligned with an ensemble of “disciplines” that included the school, the army, and the hospital? It is one and the same thing that makes us seem to follow rules and drives us ceaselessly to change them. Symbolic practices bear some resemblance to the game of croquet that Alice finds herself playing in Wonderland, with its pink flamingos by way of mallets and its hedgehogs by way of balls, sticking their heads up unexpectedly as she attempts to strike them and as, meanwhile, other players seize the moment to advance. Structuralism has posed an inescapably theoretical and speculative question: what is the nature of what is done if, in trying to do it, we end up doing something completely different? That question has bound together two other intractable problems: how to construct a theory of practice and how to construct an ontology of customs and a logic of their variation. One may argue, then, that the deployment of structural method was already in tune with the “renversement du platonisme” for which Foucault, Deleuze, and Derrida would come to argue. The strength of Lévi-­Strauss is his having recognized the necessity of the passage through philosophy (“to a certain extent, we cannot help ourselves from doing philosophy,” he wrote) at the same time as he refused to make a telos of it (“philosophical reflection is a means, not an end).”50 The philosophy of structuralism, then, is in every respect a practical philosophy.

50.  Claude Lévi-­Strauss, “Philosophie et anthropologie,” Cahiers de philosophie 1 (1966): 54; translator’s version.