Versión clásicaVersión móvil

Émergence et évolution de la parenté

 | 
Jean Lassègue

Tetradic Theory and the Origin of Human Kinship Systems

Nick Allen

Resumen

Une société ne peut perdurer dans le temps que si elle se soumet aux contraintes biologiques permettant son renouvellement. L’étude de la parenté a ceci de particulier qu’elle peut servir à rendre compte de l’origine des sociétés de petite taille : (i) parce que la parenté y possède une structure quasi logique dont on peut supposer qu’elle était présente dès l’origine et (ii) parce que le nombre des individus de la société est si restreint que les distinctions entre groupes selon la parenté sont immédiatement aussi des distinctions sociales. La question que l’on doit se poser est alors : « Quel est le système de parenté le plus simple logiquement qui permette cependant de reconnaître le caractère humain de la société ? »

On remarque que toutes les sociétés humaines essayent d’établir un lien de « recrutement » entre leurs nouveaux membres et certains prédécesseurs ou catégories de prédécesseurs en faisant usage du lien qui unit parents et enfants (lien caractérisé comme lien de parenté). Les modèles les plus simples doivent donc prendre en considération une relation horizontale (de type alliance) et une relation verticale (de type consanguinité) qui s’appuient sur une règle de prohibition de l’inceste définissant les partenaires sexuels permis ou interdits. Cette règle impose que tous les partenaires, permis ou interdits, soient discernables au moyen d’une terminologie. Pour prendre en compte les relations horizontale et verticale, la terminologie la plus simple doit donc permettre de distinguer linguistiquement dans la société des groupes d’individus des deux sexes entretenant entre eux des rapports d’alliance et de consanguinité. Quatre groupes sont nécessaires : les deux premiers (correspondant à la première génération) pratiquent l’alliance par le biais de la sexualité et produisent des enfants (correspondant à la seconde génération) qui font l’objet d’un recrutement dans les groupes différents de ceux de leurs parents des deux sexes. Il y a donc huit termes de parenté de base au moins. Les relations entre les deux ensembles de groupes sont concevables sous forme d’échange : les deux premiers groupes produisent les enfants qui appartiendront aux deux derniers qui eux-mêmes entretiennent ensuite des relations d’alliance qui reproduiront dans la descendance les appartenances aux groupes de leurs parents et ainsi de suite.

Si l’on émet une hypothèse sur l’origine historique de ce modèle impliquant quatre groupes (d’où son nom de « tétradique »), on remarque que les bandes de chasseurs-cueilleurs se retrouvent lors de rituels et que c’est là que se constituent les séparations en groupes, souvent sous forme de joutes partageant la société en deux. On peut imaginer un rituel possédant trois traits : (i) on y distingue au hasard deux groupes ; (ii) des actes sexuels dans chacun des deux groupes font partie du rituel (relation horizontale de type protomariage) et (iii) des recrutements d’enfants sont pratiqués (relation verticale de type proto-initiation) dans lesquels les nouveaux membres ne font pas partie des groupes auxquels leurs parents appartiennent. L’histoire de la parenté humaine a-t-elle commencé avec un système tétradique de ce genre ? Certains faits ethnographiques l’attestent : c’est en Australie que l’on trouve des systèmes de parenté proches de celui décrit dans la structure tétradique. Comme l’Australie a été peuplée il y a au moins 50 000 ans, on peut en conclure que c’est antérieurement que ce système s’est mis en place, peut-être dès la sortie hors d’Afrique.

Texto completo

1 The researcher curious about the origins of human society can either start with the animal kingdom, especially the great apes, and work forward in time, or start with anthropological materials and work backwards; for insofar as materials exist contemporary with the origins, they can be interpreted only by adopting one or other of these two perspectives. Without in the least rejecting the primatological perspective the present paper confines itself to the second or retrospective approach.

2The origins of society and the origins of kinship systems are not identical questions, and the relation between the two concepts will be discussed in more detail below. For the moment, suffice it to say, firstly, that in the context of small scale societies the two concepts are very closely linked, and secondly that, for anyone interested in extrapolating backwards in time, kinship systems offer a particularly attractive domain. This is because they lend themselves to the construction of formal and logical models which are less arbitrary in their assumptions and less remote from reality than would be the case with, say, economic, political or religious systems. The quasi-mathematical aspect of kinship, which has much less to do with statistics than with the logic of relations, renders it possible to answer the question “what is the simplest logically possible kinship system?” The same question can of course be posed about other societal systems, but there the chances of answering it with convincing reasons are smaller. Of course too, a gap exists between the propositions that X is the simplest kinship system of the general type one expects to find among humans, and that X is the original human kinship system. Nevertheless, until counterevidence or counter-arguments are advanced, the equation of simplest and earliest remains the most economical hypothesis.

  • 1 For the history of the subject see F. Zimmermann, Enquête sur la parenté, Paris, PUF, 1993. For my (...)

3Within the arts and social sciences it is generally held that a problem of any complexity is best approached via its history; but this is much less true of the sciences, and kinship probably belongs closer to the scientific than to the artistic pole of the anthropological continuum. In any case, I doubt whether, in the present context, an attempt to introduce the views of previous writers would be helpful.1

Biological and social continuity

4An animal collectivity cannot endure unless the members who die are replaced by new members who are born. In the simplest models the new members are the offspring of the old members, rather than outsiders; thus the continuity of the group results from birth, copulation and death. These biological facts are the raw materials that humans have elaborated into kinship systems by establishing more or less explicit rules. The rules have been expressed and transmitted in language, and may be translated into practice with varying degrees of rigour, but the simplest kinship system is the one with the simplest rules.

5What sorts of rule does a kinship system need in order to be recognisably human? Probably the first such rule that comes to mind is one governing copulation – one prohibiting incest by dividing sexual partners into the classes of the forbidden and the permitted. For the purposes of the present paper we need not specify more closely the rules governing sexual partnership: the relation between biological factors (such as the prolonged helplessness of human infants) and pair bonding; between such bonding and the human institution of marriage; or the different forms of sexual partnership such as monogamy (serial or lifelong), plural marriage and concubinage. We need only say that a recognisably human society needs some sort of marriage rule.

6It may seem that the simplest society could function with a marriage rule alone. Indeed, with some effort, one can imagine a group of humans among whom the marriage of parents produces children, but these children have no socially recognised links with previous generations; the biological condition of child is associated with no filial role. New members of society would be simply that, or (a slightly less extreme case) they would be given some more specific social identity but on grounds having nothing to do with their parentage. It would be as if all children were foundlings. Such a group could indeed endure, but is it recognisable as a human society, let alone as the sort of human society one would expect in very early human history? The societies we know of have always made some attempt to link new members of society with particular predecessors or categories of predecessors, and have always done so by making some use of the biological link between parents and children. Therefore the rules of our simplest imaginable human society will need to cover not only “horizontal” relations (marriage) but also “vertical” relations, for which “recruitment” is a conveniently general term.

7To introduce the notion of kinship systems it is convenient to separate the horizontal and vertical rules, treating them as distinct dimensions (they are often alluded to under labels such as alliance and descent or affinity and consanguinity). However the two sorts of rule are not necessarily independent, and in the model on which we shall focus they form a single complex such that neither can be fully expressed without taking account of the other.

Kinship terminologies

8To carry the argument further, I move from rules to language. A distinction is needed straightaway between the “target” language used by participants in a kinship system and the metalanguage used by analysts to discuss the target language. The analysts themselves of course participate in their own native kinship system, and the everyday language they use in that capacity is most unlikely to constitute a satisfactory metalanguage. At the very least it needs to be supplemented by a certain number of technical terms and devices.

  • 2 Of course kinship terms may have functions other than classifying relatives, but this minimal funct (...)

9For a start we need precise ways of talking about how target languages lexicalise the domain of relatives. Every known language contains a kinship terminology, a set of single words distinguishing different types of relative, but the distinctions and assimilations operated by the terminology vary widely between languages across space and time.2 To analyse the variation we need symbols for the elementary or primary relations, and the following is one of the conventional notations.

10sex-neutral sex-specifying

11parents P F M

12siblings G B Z

13children C S D

14spouses E W H

  • 3 Following convention, and being male, I write with a consistent and explicit male bias. This explai (...)

15A relation of course usually links two things – two poles – which are not necessarily commutable. In the case of relatives, the two poles are the individual (ego) and the person whom he has as relative (alter), and one has to conceptualise the relation as starting from ego. Consequently, since we read from left to right, it makes sense to read the symbol F (for father) as if it has immediately on its left an invisible symbol for ego. Usually this virtual ego need not be written, but it becomes relevant in certain contexts: for instance, the relation of male ego to his father can usefully be distinguished from that of female ego to hers by writing mF ≠ fF.3 Moreover, when we move from the primary relatives, those indicated by a single capital letter, to remoter ones (secondary, tertiary, etc.), the additional links are shown to the right, i. e. further from ego – whatever the grammar of the genitive in the analyst’s native language.

16Using this notation, we can make some elementary empirical observations about kinship terminologies, which will give us an idea of the direction in which simplification can be sought. When seen in world-wide perspective, French, like other European terminologies, is not very typical. Thus, where French oncle assimilates the two sorts of parent’s brother (it makes the equation FB = MB), most languages discriminate them (FB ≠ MB). Very commonly too, languages assimilate same-sex siblings (ssG), so that F = FB, M = MZ, and they do not confine this to cases where the sibling link comes at the end of a chain. Thus one often finds B = FBS = MZS, Z = FBD = MZD, and so on. Expressed in sex-neutral symbols this amounts to G = PssGC: the terms for siblings also cover parallel cousins. However, the coverage is not limited to first cousins, since the G within the formula PssGC can itself be replaced by PssGC, and the replacement can be repeated as often as one likes. The terms covering this class of relatives can then be rendered in the metalanguage as “classificatory siblings”, and if the assimilation is carried through consistently, the terminology as a whole can also be called classificatory.

17Among such classificatory terminologies a good number, represented in all continents apart from Europe, have a separate term or terms assimilating paternal and maternal cross-cousins (FZD = MBD, or in short, PosGC, where os stands for opposite-sex), and in that case the society sometimes has a positive marriage rule, prescribing that every ego should marry a cross-cousin. Because the terminology is classificatory, the prescription does not necessarily concern a first cross-cousin. The logical and empirical concomitants of such kinship systems are described in text-books, and are difficult to grasp without the aid of genealogical diagrams. But let us move on.

  • 4 I recommend drawing the diagram in the following order: the four brother-sister or sister-brother p (...)

18Occasionally, and particularly in terminologies which accord with the simplest cross-cousin marriage diagrams, one also finds a tendency towards the assimilation of alternate genealogical levels. In attested terminologies such assimilation seems always to be partial or patchy, but it would be logically simpler if it were total and consistent. The result is again best grasped with the aid of a genealogical diagram. Let us emphasise that we have now left the domain of what has been reported by ethnographers or historical linguists.4

Fig. 1. Genealogical diagram for focal tetradic society.

Fig. 1. Genealogical diagram for focal tetradic society.

19Although the diagram uses standard genealogical conventions, it needs to be read quite differently from a family tree, where a particular triangle represents a single named individual with a datable life-span. Clearly the triangle here represents something altogether more abstract and more general. In part this is because we are dealing with a classificatory kinship system, where same-sex siblings are assimilated. Thus each triangle represents a male together with his brothers (as well as other relatives), and the same applies to the filled-in triangle representing ego. But since the triangle in quadrant 1 represents not only ego’s father but also the latter’s brothers, FBS too is covered by ego’s triangle. One simply reads up the ordinary filiation line, then down it again.

  • 5 The answer is the ego triangle.

20Since we are accustomed to diagrams in which one generation is followed by another as the eye moves down the page, the cycling filiation line round the outside may initially be disconcerting. Let us concentrate first, not on interpreting its essence, but simply on how to read it. Followed from ego downwards and outwards, it leads round to ego’s children, who are represented by the triangle and circle in quadrant 1: one might supply the line with an arrow pointing in this direction, to indicate that the eye is moving in the direction in which time advances. However, the line can just as well be read in the opposite direction: trying to follow the filiation upwards from ego’s father and backwards in time, one is carried round the outside and back to ego’s triangle, which thus also covers ego’s FF. Whether one traces it upwards or downwards, the male line simply shuttles back and forth between the triangles in quadrants 0 and 1. It is easy to see that the circle in2 covers not only ego’s wife (the relation resulting from marriage), but also his FZD and MBD (a relation that exists even before his marriage). It may be less obvious to the eye, but one will find that ego’s father, like all the other males, also has as wife a PosGD. As an exercise, to increase familiarity with the notation, it may be worth working out which symbol covers FMBDSSWDH (answer in footnote).5 The point of the exercise is to emphasise that, provided the marriage and recruitment rules are obeyed, no genealogical chain, however long, can carry one outside the diagram; all possible relatives are accommodated within it.

21Because ego can reach alter via an indefinite, indeed infinite number of genealogical paths, from one point of view the domain of relatives that the model organises is not the same as the domain of relatives as it exists in contemporary France. But from another point of view, this difference disappears. Ego’s relatives are those with whom he is linked, by a relation or series of relations of siblingship, filiation and marriage, and the rules of marriage and recruitment that create the multiple paths to a single alter make no difference to the definition.

22It was kinship terminologies that led us to the diagram, and we can now return to them. We have already mentioned kinship systems that prescribe crosscousin marriage and the genealogical diagrams that relate to such systems. In the simplest cases, a single symbol in the diagram corresponds to a single kinship term. Applying the same principle to our eight-symbol diagram gives us a theoretical eight-term terminology covering all possible relatives – past, present or future. The question arises whether yet further simplification is possible. One can easily postulate a single base lexeme per quadrant and distinguish the sexes by either two suffixes or a single one for the sex that is “marked” (in the linguists’ sense); but I am uncertain whether a human society could operate with four sexneutral kinship terms, entrusting the determination of sex, in those contexts where it was relevant, to other methods, linguistic or extralinguistic.

The diagram

23Let us call the diagram the “focal tetradic diagram” – tetradic because it consists of four quadrants, focal because, out of the various four-quadrant genealogical diagrams one can draw, it is on this one that, for various reasons, I focus the discussion (the most obvious variant has each quadrant consisting of a husband-wife pair, rather than a brother-sister pair). We are now in a better position to see what it is that the diagram represents. It is not enough to say that the symbols it contains represent classificatory relatives, so that, say, the circle in three represents a classificatory mother. The statement is true, but more fundamentally and more abstractly, what the circle represents is one component in a structure. Let us say that a structure is a totality whose components are linked by relations which remain constant when the content of the components changes (whether over time, or because of a shift of context). Over time, for ego, the individuals covered by the « classificatory mother » circle change as old ones die off and young ones are born, but the relation with the other components remains unchanged. However, we must also take account of change of context. Relatives are people, kinship terms are words – entries in a dictionary; but when we shift from the one domain to the other, the diagram loses none of its relevance. The circle now represents one word among the (let us say) eight that make up the hypothetical terminology, and the diagram shows its semantic relations to the others. One could also say, putting the emphasis solely on meanings, that the circle represents one slot within the semantic field of relatives.

24What is to be said about the rules that give order to this field? Firstly, the rules eliminate marriage with any primary relative. Ego cannot marry his mother in quadrant3, his daughter in 1, or his sister in 0 – or indeed their classificatory equivalents; he can only marry within quadrant2. Secondly, we need to consider the nature of the rules. We introduced the marriage rule by means of the empirical observation that societies quite often require ego to marry a classificatory PosGD, and we then introduced the more or less non-empirical rule prescribing systematic assimilation of alternate genealogical levels. Applied to the male ego of our focal model, this recruitment rule entails his assimilation to FF and SS, not to mention remoter relatives, such as those in the male line from more distant even-numbered generations; in sex-neutral language the assimilation is to ssPssP and to ssCssC. Applied to ego’s wife, the rule entails her assimilation to ego’s FM and DD (among others), and the model allows ego to marry these relatives, either in the minimal or in the classificatory sense (note that, being secondary relatives, FM and DD are closer than PosGD, who is tertiary). In other words the tetradic marriage rule cannot be properly stated without taking account of the recruitment rule, and the converse is also true: the category to which ego is assimilated includes not only FF but also, because of the marriage rule, MMB (osPssPosG), represented by the same triangle. The marriage and recruitment rules are interwoven., or more precisely, they are the horizontal and vertical implications of the unitary structure. Moreover, although it seems natural to introduce the structure as the product of rules, one can also reverse the perspective: the structure generates the rules, which are merely its expression within a jural idiom.

25So far we have been talking of a kinship system envisaged as composed of ego and his relatives. However, it is common enough in small-scale societies, with total populations numbered in hundreds, for every member of society to be regarded as a relative. For simplicity of exposition let us assume that the society is demographically isolated. If the domain of relatives is co-extensive with society, the discriminations between relatives become the divisions of society, and the sets of relatives assimilated to form the various kinship categories become the groups that compose society. The metalanguage moves away from egocentric analytical terms to a different set which can be called sociocentric.

26The focal tetradic diagram now becomes a representation of the social structure of a hypothetical focal tetradic society. The quadrants can now be called by the standard technical term “sections”, and we can also give a name to the components of society that consist of pairs of sections. The half of society that consists of the two sections of a given parity (even or odd) is a generation moiety. Thus ego’s generation moiety is made up of sections 0 + 2, 1 +3 constituting its counterpart. For male ego, the 0 + 1 pair of sections constitutes a patrimoiety, leaving the2 +3 pair to form another one; and 0 + 3 constitutes a matrimoiety, as does 2 + 1. We are here concerned as analysts with the shift from egocentric to sociocentric description of the tetradic structure, and it is immaterial whether the moieties are named by members of the hypothetical society. On the other hand, it is important to see that the matrimoieties are every bit as real as the patrimoieties. If the latter are more obvious to the eye, this is an artefact of the explicit male bias in the drawing of the diagram.

27To appreciate this, suppose for a moment that we reverse the conventional values of triangle and circle, so that triangle signifies female (this amounts to just the same as replacing triangles by circles and vice versa). The diagram will now have a female bias: female ego’s mother and children will be in quadrant 1, her father in quadrant3, and the matrimoieties will now appear more salient than patrimoieties. But the structure – the set of relations between elements – remains unaffected by the changed interpretation of the symbols. The structure is in fact symmetrical between the sexes, but the conventional notation makes it hard to devise a single diagram that shows this. Indeed, one can easily become confused when comparing the viewpoints of male and female ego as regards relatives in the odd level. Quadrants (parts of a diagram) and sections (parts of a society) are not the same thing. Thus I and my sister, who are in the same section, share a father, but we locate him and his section in different positions on the diagram. This is because to me he is ssP, so in quadrant 1, but to her he is osP, so quadrant3. One can envisage the sections (enduring groups of people sharing an unchanging identity such as a name or totem) changing from one side of the diagram to the other as one changes the sex of ego.

28Relations between the moieties can be neatly expressed in terms of exchange. Ego’s generation moiety gives the children produced in the wombs of its members to constitute the other generation moiety, a fact that is represented by the outside cycling filiation lines; and the prestation is reciprocated, as is shown by the ordinary filiation lines. Ego’s patrimoiety gives its sisters and daughters in marriage to the other patrimoiety, and receives the corresponding counterprestation, as is shown by the four marriage lines. If we swap the reading of triangles and circles the exchange is of brothers and sons between matrimoieties. In sex-neutral language ego’s ssP recruitment moiety gives osG and osC to the other recruitment moiety in exchange for their equivalents.

  • 6 One can also envisage the structure being applied to various other domains: within the kinship doma (...)

29It is clear by now that an isomorphism exists between (a) the vocabulary, rules and behaviour in the domain of relatives and (b) the structure of society as a whole, and that this isomorphism arises because in all cases we are dealing with realisations of a single underlying structure.6 We can now see more clearly the essence of the cycling filiation lines. They not only enable ego’s generation moiety to make a return to the other for the new lives it receives, but also make it possible for the system to function perfectly, given a large enough population. An egocentric generation is a very different sort of thing from a sociocentric generation moiety: in an ordinary kinship system ego’s parent’s generation eventually dies off, whereas in the tetradic society ego’s parental generation moiety cannot die off, being constantly replenished by the children of ego’s generation moiety. Thus they in turn constantly replenish ego’s generation moiety, and the risk of ego failing to find an appropriate spouse simply does not arise. The standard genealogical diagrams used to illustrate non-tetradic « elementary structures of kinship » are based on generations and, essential though they are for many analytical purposes, they cannot be perfectly realised by a demographically realistic population. From this point of view, tetradic structures of kinship are the only truly elementary ones.

Logical simplicity and historical priority

  • 7 I leave the definition of simplicity informal and implicit.

30Since tetradic kinship systems are so remote from our own experience, they may not be simple to grasp at first reading, but even so it should be intuitively clear how and why they are logically simple.7 The simplicity is partly a matter of the small number of classes that suffice to classify the whole domain of relatives/society without ambiguity or overlap, but perhaps one should emphasise even more the fact that a single structure shapes a multiplicity of contexts that we are accustomed to think of as separate. Our relatives, scattered here and there across the map; the kinship terms with which we identify them as relatives (somewhat complicated by the decline of marriage, family breakdown and new reproductive technology); the laws regarding incest; the results of marriage, especially the immediate appearance of a new set of affinal relatives, who previously were simply not relatives; the statistical tendency towards class endogamy; social structure seen as consisting of vaguely or arbitrarily defined socio-economic classes – some such picture presents an obvious contrast with the tidy unitary structure of tetradic society. Could one argue that this tidiness is due to the level of abstraction at which the analysis has chosen to operate, that the argument is circular since of course if one simplifies by abstracting the end point will be simple? This would be to miss the point that a system in which the domain of relatives and the structure of society are isomorphic really is simpler than one in which separate descriptions are needed for each.

31It is natural to ask whether the process of simplification can be carried further, and indeed some have speculated that human society began when two hordes met and started exchanging women. But it is not possible to reduce a tetradic society to a dual organisation while retaining the prohibition on marriage with primary relatives as an implication of the rules of social structure. Whatever arrangement one tries, it will always be necessary to add a rule formulated in non-classificatory language to eliminate one or other primary relative. This new rule, so contrary to the classificatory style governing the rest of the system, nullifies, indeed reverses the apparent gain in simplicity achieved by reducing four sections to two moieties. A double dichotomy is the furthest one can go in simplifying the domain of relatives/society, if one hopes to remain within what is humanly plausible.

32One possible use for the tetradic model is as a pedagogic device. To understand how the model operates helps one think clearly about kinship systems in general, much as understanding non-Euclidean geometry or imaginary numbers can help one think about space or arithmetic. To approximate to a system such as we meet in the real world, the abstract model has naturally to be enriched with all sorts of features (spatial organisation, legal fictions and rituals such as adoption or blood-brotherhood, step-relatives, marital choice, local beliefs about reproduction, kinship and gender…), and to appreciate this is to perceive the gap that has to exist between models and reality. Similarly, to conceptualise the transformations that are needed to bring the model into line with attested systems could facilitate thinking about historical transformations of kinship systems. There is nothing wrong with this, but my focus here is on the world historical question with which the paper started. Did human kinship systems start out as tetradic? Can the idea that simplest equals earliest be supported by empirical evidence as well as by logic?

33Potentially the question could lead to a great deal of discussion and indeed research, and I offer here only a few brief remarks. (1) It is a textbook commonplace that small-scale tribal societies are pervaded by kinship, and this condition is met by tetradic society. (2) It has often been held that understanding the origins of incest rules would take us a long way towards understanding the origins of human society, and rules of this sort are embodied in the tetradic structure. (3) The individual features that enter into tetradic society are all ethnographically attested – notably bilateral cross-cousin marriage, terminological equations between alternate genealogical levels, societies with four sections and generation moieties; it is only their combination in minimal form that is an invention. (4) Other forms of assimilation between alternate generations are not uncommon (for instance reincarnation of grandparents in grandchildren), and can be interpreted as continuing the assimilation originally expressed in the sections and in the terminology. (5) Such historical evidence as we have suggests that kinship systems tend to lose features that are present in the tetradic model and gain ones that are alien to it, rather than changing in the opposite direction.

  • 8 The “cognatic” equation FZ = MZ, as in tante, raises different issues. Relatively uncommon across t (...)

34Quite a strong argument emerges if one thinks about the motivations and mechanisms of semantic change in terminologies. The tetradic terminology contains three different types of equation: alternate generation equations (associated with the vertical dimension of kinship), horizontal equations resulting from the marriage rule, and classificatory equations. Since all of these occur outside the model, critics opposed to the idea of an original tetradic terminology need to be able to offer explanations of how they arose. For instance, they need to argue that languages which at one moment distinguish the real mother from classificatory mothers (so that M ≠ MZ) become so obsessed with the idea of ssG equivalence that they lose the distinction. Being based on the visible event of parturition, the distinction must always be relevant to some speakers in some contexts, and its retention would not be incompatible with the use of the ssG equivalence principle elsewhere in the terminology. Nevertheless, the critic must argue, one or other of the two separate words (presumably the one for MZ) becomes obsolete and disappears, so that the language now has M = MZ. Equally implausible changes are needed to explain, for instance, the replacement of W ≠ MBD or MB ≠ ZS with the corresponding equations. Such problems do not arise if the equations are present from the start.8

35The only obvious alternative to a tetradic starting point for humanity is to suppose that the earliest kinship terminology covered only primary relatives and then gradually extended outwards to secondary ones, tertiary ones and so on. At its simplest this process of lexical innovation would produce a terminology without equations, which would then have to be introduced by the problematic processes mentioned in the last paragraph. Alternatively, one has to suppose that the process of extension was guided by a vision of marriage and recruitment rules, but then the problem is to explain how this vision originated in the absence of language in which to express it. Moreover, it is far from clear how the vision could guide early phases of the expansion, say from primary to secondary relations.

36In any case an expansion hypothesis is unnecessary. Since kinship is usually thought of in terms of relatives and their lexical classification, that was how we started on the path that led us to the tetradic model. But other paths are possible. Someone who is already familiar with the idea of a four-section system can reach the same end-point simply by attaching to the system an isomorphic or congruent egocentric nomenclature; and one can introduce four-section systems by talking of cross-cutting moieties and their rules of marriage and recruitment without reference to cross-cousins and the like. Approached in this way, the kinship system becomes an implication of the social structure, one among the ways in which the structure is manifested. If we can propose an origin for the social structure, we do not need a separate origin story for the terminology.

Origin of tetradic model

37Hunter-gatherer societies spend much of their time living in small bands with a couple of dozen members, but from time to time they assemble, typically for ritual purposes. We can thus contrast phases of dispersal and concentration. It is surely more likely that a whole group will divide itself up – give itself a structure – when it is gathered together in one place than when it is dispersed, but one would like to go further and try to model the processes that led up to cross-cutting generation and recruitment moieties. This will inevitably be speculative, but a clue is perhaps provided by the association of demographic concentration with ritual.

38Rituals and similar activities quite often require the division of participants or congregation (dancers, singers, actors, contestants…) into two or more sides or teams, so the problem lies not so much in explaining holistic divisions of society, but in seeing how these divisions could be associated with rules of marriage and recruitment. Since the aim is to explain the origin of the rules, we have to make the effort to imagine that they are not already in existence.

39Suppose then that the society splits into two teams, selected at random, with an identity and membership that endures from one period of concentration to the next. Suppose too that, as part of the ritual, couples within each team pair off and have intercourse. Let us say that this relationship is socially recognised, even though it is confined to the ritual context. We might even think of it as protomarriage, and of the ritual as constituting or including a sort of proto-wedding. But so far team A does not necessarily contain the parents and children of team B: we now have somehow to incorporate the vertical dimension of kinship. So, in due course the time comes to initiate new members of society (whether or not they were conceived in the proto-marriages does not matter), and the collective initiation also takes place during a period of concentration. But the newcomers are initiated not into the team of the mother who bore them but into the other team (prototype of a recruitment rule). In this way the initial random allocation of individuals to teams gives way, in the ritual context, to child-exchanging generation moieties.

  • 9 Logically, the allocation could as well be to the other half of the appropriate team, the half that (...)

40Suppose that the ritual now becomes more complicated and the two teams are each bisected, randomly, but in such a way that each individual within an established couple belongs to a different half-team, newcomers to the team being allocated randomly to one or other half. This random element can again in due course be phased out, the new participant being allotted to the half-team of his or her mother’s mother.9 I talk of mother’s mother rather than father’s father because the identity of an individual’s mother is always clear while that of the father need not be (probably, in this case, will not be). However, it is unnecessary to decide whether matrimoieties or patrimoieties came first: the cross-cutting of generation moieties with one form of recruitment moiety necessarily generates the other form.

41So far, nothing has been said about marriage or recruitment during the dispersed phase, and there is no need to assume that what happens in the concentrated phase is immediately transferred into ordinary life when the assembly breaks up into bands. The bands might for generations follow whatever more or less pre-social arrangements they had inherited from the past. During this transitional period the dispersed and concentrated phases would contrast as nature to culture, and the innovations developed in the context of ritual would only gradually spread to take over tribal life as a whole. Perhaps one can think of the aesthetic elegance of the tetradic structure as playing some part in its development and spread. Might there also have been socially sanctioned selection pressures operating against non-conformists and hence in favour of those who by virtue of their genes were most inclined to grasp the new rules of ritual and kinship and most willing to follow them?

Wider questions

42This speculative little story about the origin of the tetradic structure is less important in my argument than (a) the ritual context and (b) the derivation of the classificatory terminology from the social structure. As regards (b), one can remain agnostic about the early existence of terms for primary relatives, or for some of them, especially mother, provided it is clear that these terms did not serve as a nucleus from which the terminology gradually expanded. However, there is still a deeper question, namely whether a tetradic society needs to possess a full spoken language at all. To operate the system in a minimal manner, ego only needs to know two things: into what section he can marry, and in what section his children belong. But section membership could be signalled by nonlinguistic means such as haircut, body painting or ornamentation, even perhaps, within the ritual context, by posture or position in space (position on the terrain of dance?). Thus in theory at least, language is not essential. On the other hand one might well doubt whether the organisation of teams and half-teams could really occur without something resembling spoken language.

43This leads to the problem of how, in the early history of the human species, innovations in the sphere of kinship and social structure relate to innovations of other sorts, and hence to absolute dates. Since kinship systems typologically close to the focal tetradic model are well attested in aboriginal Australia, the model surely developed before the human settlement of that continent some 50 000 years ago. Furthermore, since it is so difficult to imagine a fully human society that lacks a kinship system, I suppose that the model was already familiar to those of our ancestors who dispersed from Africa twice as long ago as that. Whether one can go further back still, I do not know. Tetradic theory may hold clues or suggestions to those approaching the problems of human origins from other perspectives. Perhaps for instance, some of those using a Darwinian framework put too much emphasis on the imperatives of subsistence and reproduction, too little on the genesis of ritual and social organisation. As regards language, perhaps the functions served by that human innovation during the concentrated phase may have differed from those of the dispersed phase, and may have been equally or even more important. In any case, no satisfactory account of human origins can ignore the domain of kinship and social structure, and no satisfactory account of this domain can ignore the tetradic framework, within which alone these two aspects of society can be perfectly congruent.

Notas

1 For the history of the subject see F. Zimmermann, Enquête sur la parenté, Paris, PUF, 1993. For my own earlier writings see “The prehistory of Dravidian-type terminologies”, in M. Godelier, T. R. Trautmann and F. E. Tjon Sie Fat (eds.), Transformations of Kinship, Washington, Smithsonian Institution, 1998, p. 314-331, and chapters3-4, in Categories and Classifications: Maussian Reflections on the Social, Oxford, Berghahn,2000.

2 Of course kinship terms may have functions other than classifying relatives, but this minimal function will suffice here. Let us also simplify by assuming that a language has only a single set of terms.

3 Following convention, and being male, I write with a consistent and explicit male bias. This explains the ordering E W H earlier in the paragraph: male ego’s W is given priority over female ego’s H.

4 I recommend drawing the diagram in the following order: the four brother-sister or sister-brother pairs, the ordinary filiation lines, the marriage lines, the cycling filiation lines; finally fill in ego and add the quadrant numbers.

5 The answer is the ego triangle.

6 One can also envisage the structure being applied to various other domains: within the kinship domain to the patterning of attitudes towards different categories of relative (familiarity, avoidance, respect and joking), and outside it to the construction of a cosmology. In some languages (aboriginal Australia) the social structure even interacts with the grammar, affecting the choice of pronoun.

7 I leave the definition of simplicity informal and implicit.

8 The “cognatic” equation FZ = MZ, as in tante, raises different issues. Relatively uncommon across the world, it is not a classificatory equation in the sense we have been using the word. In particular, the equation is not just between the shortest pair of formulae from an indefinitely large set of ever-increasing genealogical remoteness. If the significance of the difference between the father’s and mother’s side of the family declines for any reason, the replacement of the distinction by the equation is not puzzling.

9 Logically, the allocation could as well be to the other half of the appropriate team, the half that contains not MM but FM. This will result in a tetradic structure of a non-focal type, a topic which will not be examined here.

Índice de ilustraciones

Título Fig. 1. Genealogical diagram for focal tetradic society.
URL http://books.openedition.org/editionsulm/docannexe/image/9170/img-1.jpg
Archivo image/jpeg, 10k

Autor

School of Anthropology, Université d’Oxford.

Salvo indicación contraria, el texto y otros elementos (ilustraciones, archivos adicionales importados) se puede utilizar bajo licencia OpenEdition Books License.

Comprar

Volumen papel

decitre.fr
Buscar en OpenEdition Search

Se le redirigirá a OpenEdition Search