CHAPTER VII

Social Structures of Central and Eastern Brazil

DURING RECENT YEARS our attention has been focused on the institutions of certain tribes of central and eastern Brazil which had been classed as very primitive because of their low level of material culture. These tribes are characterized by highly complex social structures which include several systems of criss-crossing moieties, each with specific functions, clans, age grades, recreational or ceremonial associations, and other types of groups. The most striking examples are furnished by the Sherente, who have exogamous patrilineal moieties subdivided into clans; the Canella and the Bororo, with exogamous matrilineal moieties and other types of groups; and finally, the Apinayé, with non-exogamous matrilineal moieties. The most complex types, such as a double system of moieties subdivided into clans, and a triple system of moieties lacking clan subdivisions, are found among the Bororo and the Canella, respectively. (These tribes have been described by Colbacchini, Nimuendajú, and the present author, as well as earlier observers.)

The general tendency of observers and theorists has been to interpret these complex structures on the basis of dual organization, which seemed to represent the simplest form.1 This followed the lead of native informants, who focused their descriptions on the dual forms. I do not differ from my colleagues in this respect. Nevertheless, a long-standing doubt led me to postulate the residual character of dual structures in the area under consideration. As we shall see, this hypothesis later proved inadequate.

We propose to show here that the description of indigenous institutions given by field-workers, ourselves included, undoubtedly coincides with the natives’ image of their own society, but that this image amounts to a theory, or rather a transmutation, of reality, itself of an entirely different nature. Two important consequences stem from this observation, which until now had been applied only to the Apinayé: The dual organization of the societies of central and eastern Brazil is not only adventitious, but often illusory; and, above all, we are led to conceive of social structures as entities independent of men’s consciousness of them (although they in fact govern men’s existence), and thus as different from the image which men form of them as physical reality is different from our sensory perceptions of it and our hypotheses about it.

Our first example will be the Sherente, described by Nimuendajú. This tribe, which belongs to the central Ge linguistic family, is distributed in villages, each composed of two exogamous patrilineal moieties subdivided into four clans. Three of these clans are considered by the natives as the original Sherente clans; the fourth is attributed by legend to a foreign “captured” tribe. The eight clans, four in each moiety, are differentiated by ceremonial functions and privileges; but neither these clans, nor the two athletic teams, nor the four men’s clubs and the related women’s association, nor the six age grades function in the regulation of marriage, which depends exclusively upon the moiety system. We would expect, then, to find the usual corollaries of dual organization, namely, distinction between parallel-cousins and cross-cousins; merging of patrilateral and matrilateral cross-cousins; and preferential marriage between bilateral cross-cousins. This, however, is only rarely the case.

In another work whose conclusions we shall review briefly,2 we have distinguished three fundamental types of marriage exchange; these are expressed, respectively, by preferential bilateral cross-cousin marriage, marriage between sister’s son and brother’s daughter, and marriage between brother’s son and sister’s daughter. We have called the first type restricted exchange, implying the division of the group into two sections, or a multiple of two, while the term generalized exchange, which includes the two remaining types, refers to the fact that marriage can take place between an unspecified number of partners. The difference between matrilateral and patrilateral cross-cousin marriage arises from the fact that the former represents the richest and most complete form of marriage exchange, the partners finding themselves oriented once and for all in an open-ended global structure. Patrilateral cross-cousin marriage, on the contrary, is a “borderline” form of reciprocity, links groups only in pairs, and implies a total reversal of all the cycles with each succeeding generation. It follows that matrilateral marriage is normally accompanied by a kinship terminology which we have called “consecutive": Since the position of the descent groups in relation to one another is unchanging, their successive members tend to be merged under the same term, and differences of generation are ignored. Patrilateral marriage, on the other hand, is associated with an “alternating” terminology, which expresses the opposition of consecutive generations and the identification of alternating generations. A son marries in the direction opposite from his father—yet in the same direction as his father’s sister—and in the same direction as his father’s father—yet in the opposite direction from that of his father’s father’s sister. For daughters, the situation is exactly the reverse. A second result follows. In matrilateral marriage, we find two separate and distinct terms for two types of affinal relatives: “sisters’ husbands” and “wives’ brothers.” In patrilateral marriage, this dichotomy is transposed into the descent group itself, in order to distinguish first-degree collateral relatives according to sex. Brother and sister, who always follow opposite paths in marriage, are distinguished by what F. E. Williams, in Melanesia, described as “sex affiliation”; each receives a fraction of the status of the ascendant whose matrimonial destiny he or she follows or complements, that is, the son receives the status of his mother, and the daughter that of her father—or vice versa according to the situation.

When we apply these definitions to the Sherente, we immediately perceive certain anomalies. Neither the kinship terminology nor the marriage rules coincide with the requirements of a dual system or a system of restricted exchange. Rather, they contradict one another, each pattern being associated with one of the two fundamental types of generalized exchange. Thus the kinship vocabulary offers several examples of consecutive terms, as, for instance:

father’s sister’s son = sister’s son
wife’s brother’s son = wife’s brother
father’s sister’s husband = sister’s husband = daughter’s husband

The two types of cross-cousins are also distinguished. However, marriage (for male Ego) is permitted only with the patrilateral cousin and is prohibited with the matrilateral cousin, which should imply an alternating terminology, and not a consecutive one—as is precisely the case. At the same time, several terminological identifications of individuals belonging to different moieties (mother and mother’s sister’s daughter; brother, sister, and mother’s brother’s children; father’s sister’s children and brother’s children; etc.) suggest that this moiety division does not represent the most essential aspect of the social structure. Thus, even a superficial examination of the kinship terminology and marriage rules leads to the following observations: Neither the terminology nor the rules of marriage coincide with an exogamous dual organization. The terminology, on the one hand, and the marriage rules, on the other, belong to two mutually exclusive patterns, both of which are incompatible with dual organization.

On the other hand we find indices of matrilateral marriage which contradict the patrilateral pattern, the only one for which we have evidence. These are: (1) plural union—a form of polygyny usually associated with matrilateral marriage and matrilineal descent, although in this case the descent is actually patrilineal; (2) the presence of two reciprocal terms among affinal kin, aimapli and izakmu, which leads us to believe that affines maintain a unidimensional relationship with one another, that is, that they are sisters’ husbands or wives’ brothers, but not both at the same time; (3) finally, and above all, there is the role of the bride’s maternal uncle, which is unusual for a moiety system.

Dual organization is characterized by reciprocal services between moieties which are, at the same time, associated and opposed. This reciprocity is expressed in the set of special relationships between a nephew and his maternal uncle, who belong to different moieties regardless of type of descent. But among the Sherente, these relations, restricted in their classic form to the special narkwa bond, seem to be transposed to the husband or bridegroom, on the one hand, and to the bride’s maternal uncle, on the other. Let us examine this point further.

The bride’s maternal uncle performs the following functions: He organizes and carries out the abduction of the bridegroom as a preliminary to the marriage; he takes in his niece in the event of a divorce and protects her against her husband; if the niece’s husband dies, he forces her brother-in-law to marry her; together with her husband, he avenges his niece if she is raped. In other words, he is his niece’s protector with, and if necessary against, her husband. If, however, the moiety system had a truly functional value, the bride’s maternal uncle would be a classificatory “father” of the bridegroom, rendering his role as abductor (and as protector of the wife of one of his “sons,” thus hostile to the latter) absolutely incomprehensible. There must, therefore, always be at least three distinct descent groups—Ego’s group, Ego’s wife’s group, and the group of Ego’s wife’s mother—and this is incompatible with a pure moiety system.

On the other hand, members of the same moiety often reciprocate services. At the occasion of female name-giving, ceremonial exchanges take place between the alternate moiety to that of the girls and their maternal uncles who belong to the officiants’ moiety. The boys’ initiation is performed by their paternal uncles who belong to the same moiety; at the giving of the name Wakedi to two boys (a privilege reserved to the women’s association), the maternal uncles of the boys accumulate game that is then taken by the women of the opposite moiety, which is therefore the moiety of the uncles as well. In short, everything happens as though there were a dual organization, but in reverse. Or, more accurately, the role of the moieties is lost. Instead of moieties exchanging services, the services are exchanged within the same moiety, on the occasion of a special activity held by the other moiety. Three partners, therefore, are always involved instead of two.

Given these conditions, it is significant to discover, at the level of the associations, a formal structure which corresponds exactly to a law of generalized exchange. The four men’s societies are organized in a circuit. When a man changes his association he must do it in a prescribed and immutable order. This order is the same as the one governing the transfer of feminine names, which is a privilege of the men’s societies. Finally, this order


krara → krieriekmū → akemhā → annōrowa →(krara)


is the same, although inverted, as that of the mythical origin of the societies and of the transfer, from one society to another, of the obligation to celebrate the Padi rite.

Another surprise awaits us when we turn to the myth. The myth actually presents the associations as age grades, created in a succession from youngest to oldest. For mask-making, however, the four associations are grouped in pairs linked by reciprocal services, as though they formed moieties, and these pairs consist of age grades which are not consecutive but alternate, as though each moiety were composed of two marriage classes in a system of generalized exchange. (See Figure 4.) We find the same order in the rules of aikmā—the commemoration of the deaths of illustrious men.

i_Image1

FIGURE 4

The following outline sketches the main features of the preceding discussion:


1. There are no rigid barriers between exogamous moieties, associations, and age grades. The associations function as marriage classes. They fulfill the requirements of the marriage rules and kinship terminology better than the moieties do. On the level of myth, the associations appear as age grades, and in ceremonial life they are grouped within a theoretical moiety system. Only the clans appear extraneous and seemingly indifferent to this organic whole. Everything functions as if the moieties, associations, and age grades were awkward and fragmentary expressions of an underlying reality.

2. The only possible historical evolution that could account for these contradictory characteristics would be:

The Bororo head the list of our other examples, which we shall sketch more briefly. First, we must note the remarkable symmetry between Sherente and Bororo social organization. Both tribes have circular villages divided into exogamous moieties, each with four clans and a central men’s house. This parallelism goes even further, despite the opposition of terms that is due to the patrilineal or matrilineal character of the two societies. Thus the Bororo men’s house is open to married men and that of the Sherente is reserved for bachelors; it is the scene of sexual license among the Bororo, while chastity is imperative in the Sherente men’s house; Bororo bachelors drag in girls or women with whom they then have extra-conjugal sexual relations, whereas the Sherente girls enter only to capture husbands. A comparison between the two tribes is therefore certainly justified.

Recent studies have provided new information concerning social organization and kinship. For the latter, the rich documents published by Father Albisetti show that although the dichotomy between “cross” and “parallel” relatives exists (as we should expect in a system of exogamous moieties), it does not coincide with the moiety division but, rather, cross-cuts it, since identical terms occur in both moieties. We shall limit ourselves to a few striking examples. Ego equates brother’s children and sister’s children, although they belong to different moieties. Although in the grandchildren’s generation we find the expected dichotomy between “sons and daughters” (terms theoretically limited to grandchildren of the moiety opposite Ego’s own) on the other hand and “sons-in-law” and “daughters-in-law” (terms theoretically restricted to grandchildren of Ego’s moiety) on the other, the actual distribution of these terms does not correspond to the moiety division.

We know that in other tribes—for example, the Miwok of California—such anomalies indicate the presence of groupings different from, and more important than, the moieties. Furthermore, in the Bororo system, we note certain striking terminological equivalents, such as:

mother’s brother’s son’s son is called: daughter’s husband, grandson; father’s sister’s daughter’s daughter is called: wife’s mother, grandmother;

and especially:

mother’s mother’s brother’s son and mother’s mother’s mother’s brother’s son’s son are called: son.

These equivalents immediately bring to mind kinship structures of the Bank-Ambrym-Easter Island type. The similarities are corroborated by the possibility of marriage with the mother’s brother’s daughter’s daughter in both cases.3

Regarding social organization, Father Albisetti specifies that each matrilineal moiety always consists of four clans and that there is preferential marriage not only between certain clans but between certain sections of these clans. According to him, each clan is actually divided into three matrilineal sections: Upper, Middle, and Lower. Given two clans linked by preferential marriage, unions can take place only between Upper and Upper, Middle and Middle, and Lower and Lower section members. If this description were correct (and the observations of the Salesian Fathers have always been trustworthy), we see that the classic picture of Bororo institutions would collapse. Whatever the marriage preferences linking certain clans, the clans themselves would lose all functional value (as we have already observed for the Sherente), and thus Bororo society would be reduced to three endogamous groups—Upper, Middle, and Lower—each divided into two exogamous sections. As there are no kinship relationships between the three principal groups, these would really constitute three sub-societies (Figure 5).

i_Image1

FIGURE 5

Since the kinship terminology seems explicable only in terms of three theoretical descent groups, ultimately split into six—wife’s father, mother, daughter’s husband—and linked by a system of generalized exchange, we are led to postulate an original triadic system transformed by the addition of a dual system, as among the Sherente.

To regard the Bororo as an endogamous society is so startling that we should hesitate even to consider this possibility had not an analogous conclusion already been drawn for the Apinayé by three different authors working independently with documents collected by Nimuendajú.

We know that the Apinayé moities are non-exogamous and that marriage is regulated by the division of the group into four kiyé, as follows: a man A marries a woman B, a man B marries a woman C, a man C marries a woman D, etc. Since boys belong to the kiyé of their fathers and girls to that of their mothers, the apparent division into four exogamous groups masks a real division into four endogamous groups: men of A and women of B, who are related; men of B and women of C, also related; men of C and women of D; men of D and women of A. The men and women grouped into the same kiyé, on the other hand, are not related at all. This is exactly the situation we have described among the Bororo, based on information currently available, except that the latter would have only three endogamous groups instead of four. Certain clues suggest the same type of groups among the Tapirapé. Under these conditions we may ask ourselves if the Apinayé marriage rule that prohibits cousin marriage and the endogamous privileges of certain Bororo clans (whose members may contract marriages, although they belong to the same moiety) do not aim, by antithetical means, to counteract the division of the group, either by incestuous exceptions or by marriages contrary to the rules, which the remoteness of kinship ties makes it difficult to distinguish.

Unfortunately, gaps and obscurities in Nimuendajú’s work on the Eastern Timbira do not allow us to carry the analysis to this point. At any rate, we can be certain that here again we are in the presence of the same elements of a complex common to the entire culture area. The Timbira have a systematically consecutive terminology in which:

father’s sister’s son = father
father’s sister’s daughter = father’s sister
mother’s brother’s son = brother’s son
daughter’s daughter = sister’s daughter

And the prohibition of cross-cousin marriage (as among the Apinayé) despite the presence of exogamous moieties; the role of the bride’s maternal uncle as the protector of his niece against her husband, a situation already encountered among the Sherente; the rotating cycle of age grades, analogous to that of the Sherente associations and the Apinayé marriage classes; and, finally, the regrouping of alternate pairs of age grades in athletic contests, like that of the Sherente associations in ceremonies—all this leads us to assume that the problems raised would be quite similar.

Three conclusions emerge from this schematic presentation:

1. The study of social organization among the populations of central and eastern Brazil must be thoroughly re-examined in the field—first, because the actual functioning of these societies is quite different from its superficial appearance, which is all that has been observed until now; and second, and more important, because this study must be carried out on a comparative basis. Undoubtedly the Bororo, the Canella, the Apinayé, and the Sherente have, each in their own way, created real institutions which are strikingly similar to one another and, at the same time, simpler than their explicit formulation. Furthermore, the various types of groupings found in these societies—specifically, three forms of dual organization, clans, sub-clans, age grades, associations, etc.—do not represent, as they do in Australia, so many functional groups. They are, rather, a series of expressions, each partial and incomplete, of the same underlying structure, which they reproduce in several copies without ever completely exhausting its reality.

2. Field-workers must learn to consider their research from two different perspectives. They are always in danger of confusing the natives’ theories about their social organization (and the superficial form given to these institutions to make them consistent with theory) with the actual functioning of the society. Between the two there may be as great a difference as that between the physics of Epicurus or Descartes, for example, and the knowledge derived from contemporary physics. The sociological representations of the natives are not merely a part or a reflection of their social organization. The natives may, just as in more advanced societies, be unaware of certain elements of it, or contradict it completely.

3. We have seen that, in this respect, the native representations of central and eastern Brazil, as well as the institutional language in which these are expressed, constitute an effort to regard as basic a type of structure (moieties or exogamous classes) whose true role is quite secondary, if not totally illusory.

Behind the dualism and the apparent symmetry of the social structure we perceive a more fundamental organization which is asymmetrical and triadic;4 the requirements of a dualist formulation lead to insuperable difficulties in the harmonious functioning of the organization.

Why do societies affected by a high degree of endogamy so urgently need to mystify themselves and see themselves as governed by exogamous institutions, classical in form, of whose existence they have no direct knowledge? This problem (to which we have elsewhere sought a solution) belongs to general anthropology. Raising it in a technical discussion and with respect to a limited geographical area at least shows the contemporary trend of anthropological research and demonstrates that henceforth in the social sciences, theory and research are indissolubly linked.

NOTES

1. By 1940, however, Lowie had cautioned against drawing false analogies to the Australian systems.

2. See C. Lévi-Strauss, Les Structures élémentaires de la parenté (Paris: 1949).

3. Among the Bororo, however, marriage remains possible with the mother’s brother’s daughter, which indicates that we must not push the comparison too far.

4. This triadic organization had already been pointed out by A. Métraux among the Aweikoma, but it was disputed because it would have been "unique to Brazil." (For the authors cited in this argument, see the bibliography at the end of the book.)