Oysters, Smoked Salmon, and Stilton Cheese

• •11

Levi-Strauss is distinguished among the intel­lectuals of his own country as the leading ex­ponent of "Structuralism," a word which hascome to be used as if it denoted a whole newphilosophy of life on the analogy of "Marxism"or "Existentialism." What is this "Structuralism"all about?

The general argument runs something likethis. What we know about the extemal worldwe apprehend through OUT senses. The phe­nomena we perceive have the characteristics weattribute to them because of the way our sensesoperate and the way the human brain is de­signed to order and interpret the stimuli whichare fed into it. One very important feature ofthis ordering process is that we cut up the con­tinua of space and time with which we aresurrounded into segments, so that we are pre-

CLAUDE Lfvl-STRAUSS

disposed to think of the environment as consisting ofvast numbers of separate things belonging to namedclasses, and to think: of the passage of time as consistingof sequences of separate events. Correspondingly, when,as men, we construct artificial things (artifacts of allkinds), or devise ceremonials, or write histories of thepast, we imitate our apprehension of nature: theproducts of our culture are segmented and ordered inthe same way as we suppose the products of nature tobe segmented and ordered.

Let me give a very simple example of what I mean.The color spectrum, which runs from violet, throughblue, to green, to yellow, to red, is a continuum. Thereis no natural point at which green changes to yellow oryellow to red. Our mental recognition of color is a re­sponse to variations In the quality of the light input,notably to luminosity as between dark and light and towave length as between long and short. Wave lengthgets shorter as we move from infrared to ultraviolet,while temperature, as measured on a thermometer, getsless; luminosity is zero at either end of this spectrumand reaches a maximum In the middle-that is, in theyellow. l It is a discrimination of the human brain whichbreaks up this continuum Into segments so, that we feelthat blue, green, yellow, red, etc., are quite "different"colors. This ordering mechanism of the brain is such

• Physicists must forgive the archaic account of the rela­tion between color and thennal radiation. The practicaldescription of color difference is highly technical but, asan example, the ~reflectances" (luminosities) of the threestandard artists colOr! Emerald Green, Chrome Yellow, andCadmium Red, wilh wave lengths respectively 512, 581,and 600 millimicrons, are In the ratio 2: 3: I: A thennometerplaced in different parts of a spectrum derived from a whitelight source will register the greatest temperature rise in theInfrared and the least in the ultraviolet.

Oysters, Smoked Salmon, and Stilton Cheese

that anyone who is not color blind can readily be taughtto feel that green is the "opposite" of red in the sameway that black is the opposite of white. In our ownculture we have in fact been taught to make this dis­crimination, and because of this we find it appropriateto use red and green signals as if they correspondedto plus and minus. Actually we make a number ofoppositions of this kind in which red is contrasted notonly with green but also with other "colors," notablywhite, black, blue, and yellow. When we make pairedoppositions of this kind, red. is conSistently given thesame value; it is treated as a danger sign-hot taps.live electric wires. debit entries in account books. stopsigns on roads and railways. This is a pattern whichturns up in many other cultures besides our own andin these other cases there is often a quite explicit recog­nition that the "danger" of red derives from its "natural"association with blood.

Anyway, in our case, with traffic lights on both rail­ways and roads, green means go and red means stop.For many situations this is sufficient. However, if wewant. to devise a further signal with an intermediatemeaning-about to stop / about to go-we choose thecolor yellow. We do this because, in the sPe:etrum, itlies midway between green and red.

In this example the ordering of the colors green­yellow-red is the same as the ordering of the instructionsgo-cauHon-slop; the color system and the signal systemhave the same "structure," the one is a transformationof the other.

But notice how we have arrived at this transfor­mation:

a) The color spectrum exists in nature as a coUnuum.

CLAUDE LtVI-STRAUSS Oysters. Smoked Salmon. and Stilton Cheese

Figure I. Traffic-Signal Color Triangle

WAVE I..ENGTH

(movement)

This particular example has not, so far as I am awareever been used by Levi-Strauss, ~ut the structuralisthesis is that triangles of this kind, implying comparabletransformations of models of nature as apprehendedby human brains, have very general application, thoughIn the general case the possibilities are more compl!-cated.

In my example. the pattern was subject to two specialconstraints: tirst, it is a "fact of nature" that me se­quence of colors in the spectrum is green-yellow-redand not yenow-green-red or green·red-yellow, andsecond. there is the further fact of nature, which cer­tainly goes back to very early paleolithic times, thathuman beings have a tendency to make a direct asso­ciation between red as a color and blood as a substance,so that. if anyone of these three COtors is to be selectedto mean "stop-danger," it is much more likely to be redthan either yellow or green. On this account the corre-laUon between the members of the two triads are, in

/""RED

---- (stop)

YELLOW

(caution)

CREEN

(go)

SHORT •• •• LONG

(move) (don't move)

HICH

(change)

!LOW

(no change)

LUMINOSITY

(continuity )

b) The human brain interprets this continuum as ifit consisted of discontinuous segments.

c) The human brain searches for an appropriaterepresentation of a binary opposition plus/minusand selects green and red as a binary pair.

d) Having set up this polar opposition. the humanbrain is dissatisfied with me resulting discon­tinuity and searches for an intermediate position:not plus/not minus.

e) It then goes back to the original natural continuumand chooses yellow as the intermediate signalbecause the brain is able to perceive yellow as adiscontinuous intermediate segment lying betweengreen and recl.

f) Thus the final cultural product-the three-colortraffic signal-is a simplified imitation of a phe­nomenon of nature-the color spectrum-as al?'prehended by the human brain.

The essence of this whole argument may be exhibitedin a diagram (Figure I) which represents two super­imposed triangles. The corners of the first triangle arethe colors green, yellow, red, which are differentiatedalong two a~es: (I) short wave length/long wave lengthand (2) low luminosity/high luminosity. The comersof the second triangle are three instructions concerningmovement: go-continue in a state of movement;caution-prepare to change your state of movement;stop-continue in a state of non·movement. These mes­sages are again d:fferentiated along two axes: (I)movement/no movement and (2) change/no chang(l.By superimposing one schema on the other the colorsbecome signals for the underlying il)strUctions: thenatural structure of the color relations is the same asthe logical structure relating the three instructions.

CLAUDE Lfvl-STRAUSS 20

this case, more or less predetermined. The equivalences

Sred-yellow-green 1are given and we do not need to pay( STOP-CAUTION-COSattention to alternative possibilities offered by the rest

of tbe matrix.

STOP CAUTION co

"d yellow green _actual sequence

'" green yellowyellow "d green orneryellow green "d possible

green yellow red sequencesgre.n "d yellow

But in the general case, a structural analysis needsto start by setting out all the possible permutations andto proceed by examination of the empirical evidence ona comparative basis. Uvj·Strauss himself puts it this

way:

The method we adopt . . . consists of the following

operations;-(0 define the phenomenon under study as a relationbetween two or more terms, real or supposed;(ii) construct a table of possible permutations be­

tween these terms;(iii) take this table as the general object of analysiswhich, at this level only, can yield necessary connec­tions, the empirical phenomenon considered at the f>e..ginning being only one possible combination amongothers, the complete system of which must be con-structed beforehand. (Totemtsm (English translationof Le Totemisme aujourd'huiJ, p. 16)

As I have explained for the traffic-signal case, theultimate object of the exercise is to discover how rela-

Oysters, Smoked Salmon, and Stilton Cheese :u

tions which exist in nature (and are apprehended assuch by human brains) are used to generate culturalproducts which incorporate these same relations. ThtSpoint must not be misunderstood. Uvi-Strauss Is notan idealist in the style of Bishop Berkeley; he is notarguing that Nature has no existence other than in itsapprehension by human minds. Levi-Strauss' Nature iSa genuine reality "out there"; it is governed by naturallaws which are acceSSible, at least in part, to humanscientific investigation, but our capacity to apprehendthe nature of Nature is severely restricted by the natureof the apparatus through which we do the apprehend­ing. Levi-Strauss' thesis is that by noticing how w!apprehend nature. by observing the qualities of theclassifications which we use and the way we manipu­late the resulting categories, we shall be able to infercrucial facts about the mechanism of thinking. ~

After all, since human brains are themselves naturalobjects and since they are substantially the samethroughout the species Homo sapiens, we must supposethat when cultural products are generated in the way Ihave described the process must impart to them certainuniversal (natural) characteristics of the brain itselfThus. in investigating the elementary structures of cul­tural phenomena, we are also making discoveries aboutthe nature of man-facts which are true of you and meas well as of the naked savages of Central Braz.il. Uvi­Strauss puts it this way: "Anthropology affords me anintellectual satisfaction; it rejoins at one extreme thehistory of the world and at the other the history of my­self and it unveils the shared motivation of one andthe' other at the same moment." (TTistes Tropiques,p.62)

It is important to understand just what is being pro­posed. In a superficial sense the products of culture are

CLAUDE LtVI-STRAUSS 22

enormously varied, and when an anthropologist sets outto compare, let us say, the culture of the AustralianAborigines with that of the Eskimos or that of theEnglish he is first of all impressed. by the differences.Yet since all cultures .are the product of human brains,there must be. somewhere beneath the surface, featuresthat are common to all.

This, in itself, is no new idea. A much older genera­tion of anthropologists. notably Adolf Bastian (1826­

1905) in Germany and Frazer in England held thatbecause all men belong to one species there must bepsychological universals (Elementargedanken) whichshould manifest themselves in the occurrence of similarcustoms amopg peoples "who had reacbed the samestage of evolutionary development" all over the world.Frazer and his contemporaries assiduously compiledimmense catalogues of "similar" customs which weredesigned to exhibit this evolutionary principle. This isnot wbat the structuralists are up to. The recurrenceof a detail of custom in two different parts of the mapis not a matter to whicb Uvi-5ttauss attaches any par-ticular importance. In his view. the universals of humanculture exist only at the level of structure, never at thelevel of manifest tact. We may usefully compare thepatterning of the relations which links together sets ofbuman behaviors, but we shall not learn anything ifwe simply compare single cultural items as isolates. Inthe ttaffic-signal case, it is the contrast between thecolors and the switching from one color to another thatprovides the information; each color has relevance onlyin relation to the others.

These very general ideas are a development of argu­ments originally developed by the Prague scbool ofsttuctural Unguists but particularly by Roman Jakobson

Oysters, Smoked Salmon, and Stilton Cheese 23

(18g6- ), who has resided in the United Statessince Ig42 and who was an academic colleague of Uvi­Strauss at the New School for Social Research in NewYork at the end of World War II. The influence onUvi·Strauss of Jakobsen's style of phonemic analysiS,which derives in turn from much earlier work of Saus­sure, has been very marked. Levi-Strauss repeatedlymakes an assumption that other modes of cultural ex­pression, such as kinship systems and folk taxonomies.are organized like human language. This culture/lan­guage analogy has been developed out of Jakobson'sdistinctive feature theory, but Levi·Strauss has not ex­ploited the additional insights which might have beenderived from Chomsky's thinking about generativegrammars. Incidentally, Chomsky himself has expresslydeclared that Levi-Strauss' use of linguistic analogies isunjustified, though he agrees that Jakobson's argumentmust constitute a basic part of any general linguistictheory, including his own.!· a

It is interesting to see how Uvi-5trauss sets aboutderiving his cultural generalizations from his linguistic

~ See Noam Chomsky, Current Issues in Linguistic Theory(The Hague, 1964). p. 67.J In the view of many professional linguists the publicationof Noam Chomsk.y's Syntactic Stroctures (1957) had a sig­nificance for IInguisticli comparable to that of Einstein'searly papers on relativity theory for physics, and it hassometimes been argued. to Levi-5trauss' dl.scredit, that Inrelying on a Jakobson-style linguistics, he is foUowing amodel that is no longer viable. Two points need to bemade on the other side. First, even If Chomsky's work is anadvance on that of Jakobsen, It does not invalidate thegenuine merits of the latter; second, the characteristics ofChomsky's linguistics, which are subsumed under the titlesgenerative and transformational grammars, have manypolniS in common with the generative and transformationalrules for myth analysis which Levi-Strauss developed on

CLAUDE LfvI-STRAUSS 24 Oysters, Sm.olted Salmon. and Stilton Cheese

base. His discussion of the "culinary triangle" providesa case in point. This is one of the major themes whichpersist throughout the four published volumes ofMytfwlogiques, but it has also been the subject of anindependent article, which I will summarize here.~

Levi-Strauss begins with a brief reference to J akob­son's thesis in the following terms:

In all the languages of the world the complex systemsof oppositions between the phonemes are no morethan a multidirectional elaboration of a more simplesystem which is common to all, namely the contrastbetween consonant and vowel, which through theworking of a double opposition between compact anddiffuse, acute and grave, generates on the one handwhat we may call the "vocalic triangle":

and on the other the "consonant triangle":

his own quite independently. But on the other side again,"The idea of a mathematical investigation of languagestructures, to which Levi·Strauss occasionally alludes, be·comes meaningful only when one considers Tules withinfinite generative capacity." (Chomsky, p. 66) Levi-Strausshas been concemed to demonstrate only that varieties ofcultural forms, as they are actually recorded, are transfor­mations of one another. Chomsky has tackled the morefundamental problem of seeking to fonnulate grammaticalrules that will discriminate between transformations whichmake acceptable sense and those which do not. Why canwe say: "The cat sat on the mat," but not "The mat saton the cat"?• Claude LeVi-Strauss, "Le Triangle culinalre," L'Arc (Aix.en-Provence), No. 26 (J965), pp. J9-29. English version inNew Society (London). December 22, J 966. pp. 937-40.

k

/~. ,Most readers are likely to find such a pronouncementsomewhat baffling, so I will give a rather more extendedversion of the original doctrine.

Jakobson claims that young children gain control ofthe basic vowels and consonants so as to generate mean­ingful noise patterns in a standardized sequence.~ Thechild first develops the basic vowel/consonant opposi­tion by discriminating a contrast in loudness:

Vowel (V) / Consonant (C)(high-energy noise) (low-energy noise)

(loud-compact) (soft-diffuse)

The undifferentiated consonant (C) is then split bydiscriminating pitch-a low-frequency (grave) com­ponent ("p") and a high-frequency (acute) component("t"). The high-energy (compact) velar stop consonant("k") then complements the undifferentiated high­energy (compact) vowel ("a") while the low-energy(diffuse) consonants ("p," "t") are complemented bycorresponding low-energy (diffuse) vowels ("u"-grave,''f'-acute).

The whole argument may be represented by a doubletriangle of consonants and vowels (Figure 2) dis­criminated as compact/diffuse, and grave/acute.

But let me go back to the "culinary triangle." Afterhis initial brief reference to the lingUistic prototype,

~ See R. Jakobsen and M. Halle, Fundamentals of Language(New York, 1956), pp. 38 ff.

CLAUDE LtVt-STRAUSS .8 Oysters, Smoked Salmon, and Stilton Cheese

PITCH

CULTURE "---•• NATVRE

CIlAVE ••- ~. ACUTE

(low frequency) (high frequency)

a (k)

/~Figure 3. The Culinary Triangle (Primary Form)

TaANSFORMED COOKED aOTTEN

(elabore)(marque)

RAW

/\!

NORMAL

(non elabore)(non marque)

STATE OF MATERIAL

(degree ofelaboration)

i (t)u (p) _

IDIPFUSE

COMPACT

LOUDNESS

(noise energy)

Figure 2. JalloblOTl', Primary Vowel-COTIIOTIant Triangle

Levi-Strauss observes that just as there is no humansociety which lacks a spoken language so also there is nohuman society which does not, in one way or another,process some of its food supply by cooking. But cookedfood may be thought of as fresh raw food which hasbeen transformed (elabon!) by cultural means, whereasrotten food is fresh raw food which has been trans­formed by natural means. Thus, juSt as Jakobson'svowel-consonant triangles represent the binary opposi­tions compact/diffuse and grave/acute which have be­come internalized into the child's computet-like mentalprocesses, so also we can construct a culinary triangleto represent the binary oppositions normal/transformedand culture/nature, which are (by implication) in-ternalized into the etdos of human culture everywhere.'1

ft For Ihis use of the term eidos see Gregory Balcson, Navel!(New York, 1936), p. 220. In Bateson's language eidosrefers to M a slandardizatlon of the cognitive aspects of Ihepersonal1ty of Individuals,"

It is not a necessary part of Levi-Strauss' argumentthat raw (unprocessed) food must lie midway betweenthe natural and the cultural, though it is, of course, afact that most unprocessed human foodstuffs fall intothe category "'domesticated plants and animals,'" i.e.,they are both cultural and natural.

Finally Levi-Strauss completes his exercise in intel­lectual gymnastics by claiming that the principal modesof cooking form another structured set which is theconverse of the first:

(a) Roastil1g is a process in which the meat isbrought into direct contact with the agent ofconversion (fire) without the mediation of anycultural apparatus or of air Or of water; theprocess is only partial-roast meat is only partlycooked.

(b) Boiling is a process which reduces the raw foodto a decomposed state similar to natural rotting,but it requires the mediation of both water anda receptable-an object of culture.

Figure 4. The Culinary Triangle (Developed Form)

CLAUDE LfvJ-STRAUSS 2.8

(c) SmolUng is a process of slow but complete cook­ing; it is accomplished without the mediationof any cultural apparatus. but with the media­tion of air.

Thus, as to means. roasting and smoking are naturalprocesses whereas boiling is a cultural process, but. asto end-products, smoked food belongs to culture butroast and boiled food to nature.

Levi-Strauss summarizes his whole argument in thediagram shown in Figure 4.

In his original article, "Le Triangle culinaire," Levi­Strauss qualifies the generality of this schema by notingthat our own system, which distinguishes grilling fromroasting, and steaming from boiling. and adds a cate-­gory frying (which is a fonn of boiling in which oil issubstituted for water), requires a much more compli­cated model-and at this point some English-speakingreaders might begin to suspect that the whole argumentwas an elaborate academic joke. But exactly the samediagram (Figure 4) appears on page 406 of Mytholtrgiques III (lg68) accompanied by the same text, so wemust try to take the matter seriously. This is ratherdifficult. Levi-Strauss has not adhered to his own rulesof procedure as specified. above (page 20), and the

(-)air

(+)smoked

COOICED

RAW

roast(-)water

(+)boiled

R<rrTRD

Oysters. Smoked Salmon, and Stilton Cheese 29

. whole operation suggests a game of acrostics in whichappropriate words have been slipped into the vacantslots of a prearranged verbal matrix. Elsewhere Levi­Strauss has claimed that "behind all sense there is anon-s~nse~7 but perhaps the best that one could claimfor this fandangle is that behind the nonsense there isa sense, even if it is not the sense of ordinary conver­sation.

What Levi·Strauss.is getting at is this. Animals just'eat food, and food is anything which is available whichtheir instincts place in the category "edible." But humanbeings, once they have been weaned from the mother'sbreast, have no such instincts. It is the conventions ofsociety which decree what is food and what is not foodand what kinds cif food shall be eaten on what occasions.And since the occasions are social occasions there mustbe some kind of patterned homology between relation­ships between kinds of food on the one hand and rela·tionships between social occasions on the other.

Moreover, when we look into the facts, the categorieswhich are treated as significant kinds of food becomeinteresting in themselves. The diet of any particularhuman population is dependent upon the availabilityof resources and, at the level of actual items of food­stuff (bread, meat, cheese and so on), there is verylittle overlap between the shopping list of an Englishhousewife and the inventory of comestibles availableto an Amazonian Indian. But the English housewifeand the Amazonian Indian alike break up the unitarycategory "food" into a number of subcategOries, "foodA," "food B," "food C," etc., each of which is treated ina different way. But, at this level, the categories A, B,

r Claude Uvi-Strauss, "Reponses A quelques questions,"Esprit (Parls), November 1963, pp. 6:18-53.

CLAUDE LtVI-STRAUSS 3° Oysters, Smoked Salmon, and Stilton Cheese

C, etc., turn out to be remarkably alike everywhere.They are, in fact, categories of the kind which appearin Figure 4, and the significant thing about such cate­gories is that they are accorded very different levels ofsocial prestige. I do not mean only that the differentcomponents of the feast can always be fitted into ourprearranged slots-oysters (raw), smoked salmon(smoked), lobster soup (boiled), saddle .of mutton(roast), souffle (cooked), SUlton cheese (rotted)-butrather that foods of these different general classes beara standardized relationship to each other. For example,according to our conventions, whenever the menu in­cludes a dish of roast meat it will be accorded pride ofplace.in the middle; steamed and boiled foods, on theother hand, are consIdered especially suitable for in­valids and children. Why should this be? Why shouldwe tend to think of boiled fowl as a homely dish but ofroast chicken as a party dish?

All sorts of rational1z.aUons can be devised to fit anyparticular case-for example that boiling fowls arecheaper than roasters, or that boiled food is "more di­gestible" (what is the evidence for this?), but all suchexplanations begin to look rather thin once it is realiz.edthat other peoples, with very different cultures from ourown, sort out their foodstuffs in very similar ways andapply status distinctions of comparable sort. Some foodsare appropriate only to men, others only to women;some foods are forbidden to children; some can only beeaten on ceremonial occasions. The resulting patlern isnot always the same, but il is certainly very far fromrandom: Levi-Strauss has even claimed that the highstatus which attaches to roasting as against boiling is auniversal cultural characteristic, so that boiled food ishighly regarded only in relatively democratic types ofsociety. "Boiling provides a means of complete conser-

vation of the meat and its juices, whereas roasting isaccompanied by destruction and loss. Thus one denoteseconomy; the other prodigality; the latler is aristocratic,the former plebeian:' ("Le Triangle culinaire," p. 23)

, An odd line o{ thought, certainly, yet if we acceptLevi-Strauss' unexpected frame of reference, such com­ments are not nearly so arbitrary as they may appear.In that we are men, we are all a part of nature; in thatwe are human beings, we are all a part of culture. Oursurvival 3S men depends on our ingestion of food(~hich is a part of nature); our survival as humanbeings depends upon our use of social categories whichare derived from cultural classifications imposed onelements of nature. The social use of categories of f<KX!is thus homologous with the social use of categories ofcolor in the tr3ffic-signal case (page 19). But food isan eSpecially appropriate "mediator" because, when weeat, we do establish, in a literal sense, a direct identitybetween ourselves (culture) and our food (nature).Cooking is thus universally a means by which natureis transformed into culture, and categories of cookingare alw;ays peculiarly appropriate for use as symbols ofsocial differentiation.

In another context, in which Uvi-8trauss is concernedto debunk the anthropologic.al mystique that has clus­tered around the concept of totemism, he has criticizedthe functionalist thesis that totemic species are givensocial value because they are of economic value. On thecontrary, says Levi-Strauss, it is the species themselvesconsidered simply as categories that are socially valu­able: totemic species are "goods to think with~ (bonnesIi penser) rather than "goods to eat" (bonnes Ii manger)The culinary tri3ngle is the other side of the same argu­ment. Foodstuffs, as" such, are of course "goods to eat";but this alone does not explain the complications which

CLAUDE LtvI·STRAUSS 32

we jnject into the classification of food; food species,like totemic species, are "goods to think with."s (Cf.pages 40-42.)

This is an unfamiliar style of discourse, and it hasto be admitted that here, as elsewhere in Uvi-Strauss'writings, there is an element of verbal sleight of handwhich invites caution rather than enthusi~m. All thesame, the reader should not imagine that the "culinarytriangle" is ju~t an elegant jeu d'esprit by a master ofthe unexpected analogy. Levi-Strauss has by now mar­shaled a great deal of evidence to show that the proc­esses of food preparation and the categories of foodwith which they are associated are everywhere elabo­rately structured and that there are universal principlesunderlying these structures. Moreover, the method ofanalysis, however bizarre it may appear, has wide appli­cation. The culinary triangle first appeared in print onlyin 1965, but triangles of comparable type occur in manyearlier parts of the Levi-Straussian corpus.

In the 1945 paper which is the foundation work forall his subsequent structural anthropology. "L'Analysestructurale en lInguistique et en anthropologie,"$ thecorners of the triangle are MUTUALITY, RIGHTS, OBLIGA­

TIONS, while the binary oppositions appear to be ex­change/no exchange and receivers/givers. In Les

~ Several critics have rebuked me for mistranslation, but Infact I cile Uvi-Strauss' own words to avoid this imputation.IJter~lly, bonne. d penser means "good 10 think," bonne.d manger "good to eat." But "good to think" Is not English,and the adjectival plural of the French is untranslatable.It seems to me thai here, as so often, lbl-Strauss isplaylng a verbal game. TOlemic species are categories ofthings, and it does in fact convey the meaning better 10refer to them as "goods" than my critics would allow.• An English translation of this paper appears as Chapter :2

of Structural Anthropology (New York, 1963), the EngUsllversion of Anthropologie structurale (Parts, 1958).

Oysters, Smoked Salmon, and StiltC?7l Cheese 33

StrucruTes eUmentaires de la paTente: (1949), the tri­angle becomes BILATERAL MARRIAGE,. PATRILATERAL

CROSS.cOUSIN MARRIAGE, MATRILATERAL CROSS-cOUSIN

MARRIAGE, and the oppositions are symmetty/asyro­metry, alternation/repetition. "La Geste d'Asdiwal"(1960) includes a highly complicated triangle whichcombines geographical and food calegory parameters insuch a way that vegetable food is opposed to animalfood, sea to land, East to West, and definition to lackof definition. 1G TItis is not just a game. Uvi-Strauss isendeavoring to establish the rucliments of a semanticalgebra. If cultural behavior is capable of conveyingInformation then the code in which cultural messageslJIf: expressed must have an algebraic structure. It ipossible that Levi-Strauss is making larger claims forthe importance of this algehra than is justified by theevidence. but there is more to it than a tricksier's gameof tic-tac-toe. Let us go back to the beginning.

10 An English translation, 'lhe Story of Asdiwal," may befound in E. R. Leach, ed., The Structural Study of Mythand Totemf.sm (London, 1967), pp. 1-48.