Les Structures élémentaires de la parenté, by Claude Lévi-Strauss, has acquired the stature of a classic in social anthropology. In the present monograph I wish to examine its scientific claims to this reputation.
Lévi-Strauss’s work is considered a remarkable feat of analysis not only by social anthropologists, but also by a surprising number of writers in other fields. To cite the opinion of one of the most prominent modern psychologists, for Piaget the anthropological structuralism of Lévi-Strauss ‘presents an exemplary character and constitutes the most striking model – neither functional nor genetic nor historical, but deductive – that has been employed in an empirical human science’ (Piaget 1968: 90). For Boudon, similarly, Lévi-Strauss’s treatment of marriage rules constitutes an ‘exact’, ‘general’, ‘falsifiable’ theory (1968: 203), and the ‘scientific importance’ of Lévi-Strauss’s ‘discoveries’ in anthropology needs no further demonstration (1968: 10).
For Lévi-Strauss himself, Les Structures élémentaires de la parenté is ‘an introduction to a general theory of kinship systems’. It is presented as an example in method, and indeed as the last word as far as the explanation of kinship systems is concerned: ‘even if some aspect of the problem treated in the work were developed no new idea would need to be introduced’ (Lévi-Strauss 1949: xi).1
I
Les Structures élémentaires de la parenté was first published in 1949. The book constituted, according to Lévi-Strauss, a stage in his effort to produce ‘a sort of inventory of mental constraints, an attempt to reduce the arbitrary to an order, to discover a necessity immanent in an illusion of liberty’. He chose kinship, he later explained, because it was a domain which might seem at first to be characterized by ‘its incoherent and contingent character’, and he tried to demonstrate that it was possible to reduce this domain to a ‘very small number of significant propositions’ (1963a: 630).
For Lévi-Strauss, ‘kinship and its related notions are at the same time prior and exterior to the biological relations to which increasingly we tend to reduce them’. Kinship relations ‘offer immediately to the mind a framework of logical classification; this framework, once it is conceived, is used in order to situate, by reference to these pre-established categories, individuals with whom anyone happens to be, or not to be, empirically related’ (1955: 106). He is against the classification of kinship terminologies into ‘descriptive’ and ‘classificatory’ systems; ‘descriptive’ being the term applied to systems in which the terms designate particular relations between an individual and his parents, ‘classificatory’ those systems that group certain types of relatives and others, very distant or simply theoretical, in large classes. For him, ‘every kinship system is at the same time descriptive and classificatory’ (1955: 106).
In these views, Lévi-Strauss follows a long tradition starting perhaps with Durkheim by the end of the nineteenth century.
Durkheim criticized Kohler’s Zur Urgeschichte der Ehe in 1897, because it implied that ‘[kinship] nomenclatures express links of blood’ (1897: 313). He explained that if a term, in certain society, is applied to designate the actual mother, but also a series of other women, this is not because all these women are thought of as being in the same relationship of consanguinity towards the person who applies the term. It is because all those women belong to the same group, the group where the father had married, and the term thus explains the particular relationship of an individual towards the whole group designated by that term. Consanguinity is therefore a very different thing from kinship (parenté). ‘Con- sanguinity is only the sufficient condition of kinship’; kinship is essentially a series of juridical and moral obligations, and these obligations are neither regulated nor classified as consanguineal relationships (1897: 316).
Lang, also, saw quite clearly that ‘our ideas of sister, brother, father, mother, and so on, have nothing to do . . . with the native terms which include, indeed, but do not denote these relationships as understood by us’ (1903: 101).
Lévi-Strauss’s position coincides also with Kroeber’s and Hocart’s. Kroeber, when discussing the fallacious differentiation between ‘classificatory’ and ‘descriptive’ terms (1909), and Hocart, when talking about the ‘disastrous’ effects of the theory of the kinship ‘extensions’ (1937), both tried to establish the principles upon which relationship terminologies are built. Hocart shows that the Fijians do not need to think in terms of ‘family trees’ in order to classify a person. All they need, he says, is to place that person according to line [‘side’] and generation. He proposes, then, that the specification of a native term should be done by reference to the principles involved in the classificatory system, and not by genealogy.
It is clear, therefore, from this position, that relationship terms denote social categories and not ‘degrees’ of kinship. The kind of status that a set of relationship terms denotes is peculiar to the society in which it is applied (cf. Needham 1960d). The comparability of a set of relationship terms with another can only be seen by reference to the classificatory principles involved in them.
To follow this position consistently one should think of relationship terms as social categories. In this respect, Lévi-Strauss’s use of the term ‘degree’ throughout Les Structures élémentaires de la parenté introduces some confusion. He talks about ‘prohibited degrees’ when referring to the category of people that an individual is not allowed to marry (cf. 1949: 137, 367, 370; 1965: 17), and he actually identifies the concept of ‘degree’ with that of ‘relationship’ in paragraphs such as this: ‘la notion du degré de parenté, c’est-à-dire la notion du rapport’ (1949: 549). On the other hand, in a paragraph such as ‘the ancient Chinese conceived of kinship not as a series of degrees but as a hierarchy of categories’ (1949: 388), one is bound to think that the other systems he analyses or mentions are composed of ‘a series of degrees’.
II
Whether kinship systems are composed of ‘degrees’ or ‘categories’, Lévi-Strauss’s main idea is that they are always the expression of some sort of exchange between groups, which is, therefore, the origin of the different rules of marriage. In this, Lévi-Strauss follows Mauss’s idea that exogamy is an exchange of women between clans (1920). For Lévi-Strauss,
Exchange – and consequently the rule of exogamy which expresses it – has in itself a social value. It provides the means of binding men together, and of superimposing upon the natural links of kinship the henceforth artificial links . . . of alliance governed by rule (1949: 595).
The exchange appears to us under different forms, says Lévi-Strauss. It can be ‘direct’ or ‘indirect’, ‘continuous’ or ‘discontinuous’; it sometimes appears as a cash or short-term transaction and at other times as a long-term transaction; sometimes it is ‘explicit’ and sometimes ‘implicit’; at some times ‘closed’ and at others ‘open’; at some times it is ‘concrete’ and at others it is ‘symbolic’. But always it is ‘the fundamental and common basis of all modalities of the institution of marriage’ (1949: 592-3).
From all these dichotomies that characterize ‘exchange’, Lévi-Strauss uses ‘closed/open’ as the criterion to distinguish between the two main types of kinship structure. The exchange is closed ‘when marriage must satisfy a special rule of alliance between marriage classes or a special rule of observance of preferential degrees’, and it is open ‘when the rule of exogamy is merely a collection of negative stipulations, which, beyond the prohibited degrees leaves a free choice’ (1949: 592). From the preface to the first edition of the book under consideration, it is clear that ‘elementary structures’ are based on ‘closed exchange’, and Lévi-Strauss’s definition of these structures is based on the formal character of their terminologies: ‘the nomenclature permits the immediate determination of the circle of kin and that of affines’. Elementary structures, then, ‘prescribe marriage with a certain type of relative . . . while defining all members of the society as relatives, divide them into two categories, viz., possible spouses and prohibited spouses’ (1949: ix).
‘Complex structures’ do not possess this kind of terminology; they are based on ‘open exchange’, and thus do not involve a positive determination of the spouse. But they ‘can be explained as the result of the development or combination of elementary structures’ (1949: 576).
III
Although the idea that all kinship systems express ‘exchange’, and the inference that all of them derive from ‘closed’ systems of exchange, are both controversial, the attempt to examine and define all the ‘closed’ systems of social classification is in itself interesting.
There is in social anthropology another long tradition that differentiates ‘closed’ or ‘prescriptive’ systems, involving a positive determination of the spouse, from ‘non-prescriptive’ systems (Fison, Howitt, Kroeber, Hodson, Lowie, the Leiden school). Lévi-Strauss’s intention when writing Les Structures élémentaires de la parenté seemed to be to continue this tradition. It was thus understood by Needham, among others, who reinforced this idea by assimilating Lévi-Strauss’s ‘elementary structures’ to ‘prescriptive systems’ (Needham 1962b; cf. Josselin de Jong 1952). The line of demarcation between prescriptive and non-prescriptive systems was traced by Needham, following Lévi-Strauss’s own definition of ‘elementary’ and ‘complex’ structures, by reference to the formal traits of their relationship terminologies. This interpretation of Lévi-Strauss’s definitions not only conferred an internal logic on his typology, but it also separated the notions of ‘prescription’ and ‘preference’ as analytically different. ‘Prescription’ referred to the positive determination of the spouse which articulated a prescriptive terminology; ‘preference’ did not refer to a particular type of terminology defining a system. In fact, ‘preference’ refers to a trait definable in any possible system whether ‘elementary’ or ‘complex’. It refers either to explicit rules of marriage with a certain category of people or to the actual higher frequency of marriage to a certain category of spouse. In neither case is it definable by a formal trait of the system, and as it can be discerned in any system it does not provide the same sort of typological criterion that ‘elementary/complex’ was meant to supply. On the other hand, in a prescriptive system the ‘preferred’ category of spouse, defined either by an explicit rule or by statistical frequency, can coincide or not with the category prescribed by the terminology.
Thus Lévi-Strauss’s use of both terms when referring to the prescribed category of spouse was certainly unfortunate. But if ‘preferred’ was translated as ‘prescribed’ whenever Lévi-Strauss refers to the prescribed category, the typology he offered was still coherent.
But for Lévi-Strauss the use of ‘prescription’ and ‘preference’ as synonyms is far from unfortunate. Both terms refer, as he affirms in 1965 and reaffirms in 1967 in the preface to the second edition of his book, to the same reality inasmuch as they refer to the models that represent this reality. ‘Even a preferential system is prescriptive at the level of the model, while even a prescriptive system cannot but be preferential at the level of the reality.’ His book, on the other hand, is ‘concerned exclusively with models and not with empirical realities’ (1967: 58 n. 20).
IV
The idea of the present monograph arose as the result of the effort to understand Lévi-Strauss’s propositions in the light of his own reappraisal of ‘elementary structures’ since 1965. The main reason for writing it was to see whether or not Lévi-Strauss’s proposals about elementary structures deserved to be included in the study of the particular kind of ideological systems represented by prescriptive terminologies.
While writing this monograph, and in response to the published articles that constitute preliminary versions of some of the chapters included here, I have encountered two kinds of objection that I should like now to answer. The first kind questions the usefulness of criticizing an author, particularly when one considers that his work is not an example to be followed. But if one pays critical attention to the writings of an author, it is because one is concerned with the subject treated by that author. If, moreover, one disagrees with views which are almost universally praised as being of outstanding theoretical importance, then a critique becomes a duty. In any case, the present monograph is by no means only a critique; its argument rests centrally upon a series of original analyses (cf. chapters 2, 4, 5, 6). There is in this respect an international and increasing trend towards the publication of very general assess- ments of Lévi-Strauss’s theories, taking his empirical examples at their face value. The task seems to be, on the contrary, and as I hope I show here, to test Lévi-Strauss’s assertions and propositions by intensive re-analysis.
The other class of objections expresses in various ways the defence of Lévi-Strauss made by Mary Douglas: ‘I do not think it is fair to such an ebullient writer to take him literally’ (1967: 50). The richness of his thought is held to lie in the very ambiguity of his style. Now, if Lévi-Strauss’s intention were to write poetry, this view might have force; but his express intention was to propound a general theory of kinship systems. If, in this case, the final product of his thought is ambiguous – and Leach even finds that ‘some passages of Lévi-Strauss when translated into English seem almost meaningless’ (1967b: xvi-xvii) – how can anybody be sure of grasping correctly the ideas that he wishes to communicate?
If it were further argued in Lévi-Strauss’s favour that to bring together the ideas on kinship expressed by Durkheim, Mauss, Kroeber, Hocart, Lowie, and others, in a solidary compendium on kinship systems, is in itself valuable, I should agree. But the value of the product of this effort, and the value of the ‘small number of significant propositions’ that Lévi-Strauss claims to have achieved, still remains to be determined. It also remains to be considered whether or not in the field of so-called kinship systems any ‘new idea’ can be introduced.
1 Page references to Lévi-Strauss’s monograph are to the original French edition of 1949, or to the second French edition of 1967 when referring to the preface to this edition. The passages cited may readily be located in the English edition (1969) by means of the page-concordance (Korn and Needham 1969).