Three Styles in the Study of Kinship

J.A. Barnes

• Publisher: Taylor & Francis
• Year: 2013
• Print ISBN: 9780415330084
• eBook ISBN: 9781136535000
• Edition: 1
• Topic: Social Sciences
• Subtopic: Anthropology
• Index: Social Sciences

The study of kinship is a fundamental part of the study and the practice of social anthropology. This volume examines the work of three distinguished anthropologists that bear on kinship and determines what theoretical models are implicit in their writings and assesses to what extent their claims have been validated. The anthropologists studied are from France, the UK and USA: Claude Levi-Strauss, Meyer Fortes and G.P. Murdock.
First published in 1971.

I Safety in numbers

In scientific anthropology, it would seem,
there is safety in numbers.

Murdock (1940a: 369)

I Safety in numbers

I Introduction

George Peter Murdock has had a friajor influence on theoretical studies in kinship and social organization. The publication of his book Social structure marked the establishment of a distinctive trend in comparative anthropological inquiry. His interest in comparative studies based on information about a large number of societies from all parts of the world led him to initiate the Cross-Cultural Survey, later to grow into the Human Relations Area Files. He founded the journal Ethnology to provide an outlet for publications in this field for articles which ’specifically incorporate or relate to some body of substantive data⁵ (Murdock et al. 1962a: 2). His ‘World ethnographic sample⁵ (1957a) has been used by many other scholars for a great variety of investigations. Its replacement, the Ethnographic atlas, which appeared in instalments in Ethnology over many years before being published separately, continues as a sampling frame for general use. We can have no hesitation in identifying a distinct school or sub-branch within social anthropology, characterized by its own method of cross-cultural analysis. Two collections of papers, Readings in cross-cultural methodology (Moore 1961) and Cross-cultural approaches: readings in comparative research (Ford 1967), provide an ostensive definition of the school and indicate its range of interests.

Yet although quantitative world-wide cross-cultural studies ‘have been appearing of late at a geometrically increasing rate’ (Murdock 1967: 3), the fundamental assumptions common to these studies do not command unqualified professional support. Now that professional activities as a whole, and not merely cross-cultural studies, are expanding rapidly and new bandwagons threaten to create an indigestible intellectual traffic jam, it is no surprise that a line of inquiry marked out some thirty years ago should have many

competitors. More surprisingly, those who have followed other lines have for the most part been content to ignore the cross-cultural method and to develop their own techniques without reference to it. A few writers have stated briefly their unequivocal mistrust of the method, but usually without examining Murdock’s arguments and assumptions in detail. For some reason or other, most of the sustained discussion of cross-cultural method has been about blemishes and limitations in the practical application of the method rather than about fundamental principles. Criticism has been directed more at the way ethnographic data should be selected and coded for analysis than at the type of analysis performed. Yet there are many social anthropologists, in the United States as well as in France and Britain, who have no enthusiasm for the cross-cultural endeavours of Murdock and those who have followed him; the quantitative aspect of these studies has met with particular disapproval. A striking example of this lack of enthusiasm was the absence from Britain for many years of any copy of the Human Relations Area Files; this cannot have been due solely to shortage of funds. Silence among the critics cannot be explained by uncertainty about the stated aims and premisses of cross-cultural research, for Murdock has set out the assumptions underlying his inquiries, as he sees them, with great gusto and forcefulness. The technique of inquiry he has developed has its roots in the work of one of the founders of anthropology, E. B. Tylor. It is one instance of what Kòbben (1952: 131, 137-138) calls the holo-geistic method, whose practitioners seek to ‘identify associated variables that transcend the vagaries of historical contact and local conditions’ It aims at nomothetic, rather than idio-graphic or genetic, explanations and its statistical procedures are similar to those used very widely in cognate disciplines such as psychology and sociology, and in the natural sciences. The intellectual credentials of the school thus seem to be impeccably traditional and scientific. The results of applying the techniques of cross-cultural inquiry now form a substantial part of the contemporary literature in social anthropology. We cannot merely ignore them because, for example, we happen to find the structuralist dialectic more exciting, or the ethological approach more firmly based on verifiable fact. If we think that quantitative cross-cultural studies as now carried out are along the wrong lines, we should give our reasons. This is what I try to do in this chapter. I concentrate my attention on Mur-dock’s principal theoretical work, Social structure (1949a), and on the strenuous efforts he has made since that book was published to improve his sample of societies and to meet other criticisms.

One distinctive feature of the cross-cultural movement, if we may call it that, is that it has acquired not only a common set of intellectual aims and research techniques but also specialized bibliographic institutions and several key published documents. The Cross-Cultural Survey was established at Yale University in 1937 under Murdock’s leadership as part of the Institute of Human Relations. Extracts of published and unpublished ethnographic material on selected societies were classified according to a scheme set out in the Outline of cultural materials (Murdock et aL 1938, subsequently revised). Material in foreign languages was translated into English. During World War II several handbooks were produced with the help of the Survey. In 1949 the Human Relations Area Files were developed from the Survey. Whereas the Survey is confined to Yale, the Files were established to allow the extracted ethnographic material to be distributed to other universities (Murdock et aL 1950: xii-xiv). Both the Survey and the Files were designed to facilitate the formulation and testing of cross-cultural generalizations using quantitative methods. The societies included in the Files were chosen so as to form a fair sample of all known cultures (Murdock 1940a: 369). Later, as the number of societies increased, the objective shifted slightly; the societies in the Files are now seen as forming a collection from which a satisfactory sample can be drawn with minimum effort. Naroll (1968: 254) comments that recently societies thought to be of strategic interest to the United States government tend to have been selected disproportionately. The goal of the Survey is, or was, to cover ‘a representative ten per cent sample of all the cultures known to history, sociology, and ethnography’ (Murdock 1949a: viii). When writing Social structure, Murdock used a sample of 250 societies, 85 of them taken from those covered by the Survey at that time.

While the Survey and the Files may be seen essentially as bibliographic aids, a commitment to quantitative cross-cultural inquiry has also led Murdock to construct a series of standard samples of cultures and/or societies. His Outline of world cultures (1954a) establishes a list of all known cultures and suggests a suitable sample. His ‘World ethnographic sample’ (1957a) contains 565 cultures whose main characteristics are indicated in succinct coded statements. The sample has been used by many other ethnologists as a basis for their own inquiries. A revised version appeared in 1961 (Murdock 1961; cf. Kòbben 1967: 9). Publication of the ‘Ethnographic atlas’ began in 1962. With the twenty-first instalment the Atlas reached a total of well over 1,100 societies. Finally, Murdock has constructed a standard world sample of 200 cultures. A new organization, the Cross-Cultural Cumulative Center (CCCC) will use this sample to re-test correlations found earlier and'to intercorrelate the findings of different studies and thus raise the rate of scientific accumulation from an arithmetic to a geometric level’ (Murdock 1968b: 306). Other institutions have followed Mur-dock’s lead, and a Permanent Ethnographic Probability Sample is being established at Northwestern University (Naroll 1968: 254).

Although the principal stimulus to develop the Survey, the Files, the Atlas, and so on has been the requirements of the cross-cultural method, Murdock has claimed from the start that these research tools can have other uses. The Survey, he writes, ’should prove useful in nearly every type of research which anthropologists and other social scientists have hitherto pursued’ (Murdock 1940a: 363). Some anthropologists who criticize the cross-cultural method are nevertheless ready to support the Atlas and similar documents as providing them with handy cues, directing attention to new portions of the ethnographic corpus that may merit closer scrutiny. The Files may be seen as a convenient set of indexed extracts from a huge body of scattered literature, and the Atlas provides an even more succinct key to the contents of ethnographic monographs* It is obvious that, as the amount of ethnographic writing continues to increase, we need more effective ways of finding our way around the literature; and it may well be that the Atlas adequately earns its keep as an index alone. However, the use of the Atlas as a pointer back to the literature is quite distinct from its use as a lead forward to statistical cross-cultural inquiries, and it is with the latter that we are here mainly concerned.

In the Files we have a relatively expensive and elaborate tool for library research. Murdock has been the driving force behind their development and he has been the obvious person to announce the achievements and possibilities of this undertaking. Typically, he has tried to assess quantitatively the efficacy of the Files as an aid to research. He states that with their help one of his articles (1950b) was written in a total elapsed time of twenty-five hours (1950c: 720; 1953: 485), whereas without their aid he would have needed at least twenty-five days. Similarly Udy (1964: 169) reports that he was able to extract all the information he needed ten times faster by using the Files than by reading through the source monographs themselves.

It is therefore not surprising that to many observers Murdock has become identified with a set of ethnographic data organized in distinctive fashion in the Files, as well as with a theory of functional relations between cultural items and a statistical technique for establishing these relations. Thus Nadel is led to note, rather peevishly, his suspicion that Tor Murdock, nothing anthropological is scientific unless it is (a) based on the Human Relations Area Files and/or (b) contains some acknowledgement of Clark L. Hull’s learning theory⁵ (Nadel 1955: 346). A more accurate assessment is made by Leach when he says that, although Murdock may be generally associated with a particular style of cross-cultural comparison, the volume of his collected papers ‘is a valuable reminder that Six-Gun Pete has had other aces up his sleeve’ (Leach 1966: 1518). Similarly, the vigorous diversity of methods, range, and ethnographic content of the articles appearing in Ethnology, and the even wider range of interests shown by his pupils in the festschrift presented to their teacher (Goodenough 1964; cf. Fox 1966), give convincing evidence that Nadel was wrong.

In part, the scope of articles appearing in Ethnology under Murdock’s editorship is explained by the division he draws between ethnographic accounts and comparative studies; we shall have more to say on this later. However, this is only part of the explanation, for he has always held that the cross-cultural method is not the only way to arrive at propositions that are valid transculturally. He has often expressed his approval of the inquiries conducted by Mead in Samoa into the biological and cultural causes of adolescent stress, and by Holmberg among the Siriono into sex anxiety in a society with chronically uncertain food supply, for these investigations were made in field situations where the appropriate variables occurred naturally in the combinations desired. If experimentation with human beings was possible, these are situations one might well construct artificially (Mead 1928; Holmberg 1950; Murdock 1950a: 573; 1951b: 1; 1954b: 27; 1957b: 252; 1966: 97). But since, like astronomers, we cannot experiment, we have to rely mainly on the other method distinctive of anthropology, that of subjecting hypotheses to quantitative comparative tests. Although most of his book Social structure is aimed at ’scientific results of universal application’ (1957b: 249), Chapter 8 and Appendix A, where he discusses the evolution of social organization, deal with historical (or prehistorical) reconstruction, though some of the ethnographic evidence educed is expressed quantitatively. Elsewhere, as in his book Africa, he has pursued primary historical interests with minimal reliance on quantified ethnographic data (1959a: 40), and one of his many complaints against British social anthropologists is their alleged complete lack of interest in history (1951a: 468).

Murdock’s training during World War I and immediately thereafter was decidedly unorthodox for a recruit to the anthropological profession in the United States at that time (1965a: 356-358). His eclectic background in social science and history, and his early association with sociology at Yale, show themselves in the battery of disciplines on which he draws for the explanation of social phenomena. Whereas many of his more orthodox contemporaries have tried to establish a distinctive cultural level of discourse, freed from the taint of reductionism, Murdock tries to manufacture from carefully selected portions of cognate disciplines a comprehensive theory of human behaviour and of culture (cf. Lévi-Strauss 1963a: 306). His theory impinges on the activities of individuals as well as groups, and relates to instrumental behaviour as well as to the construction and transmission of symbols. In keeping with this broad frame of reference Murdock looks to the common characteristics of social behaviour at all times and all places, rather than in some delimited temporal, geographical, or ecological context. The quartette of disciplines – sociology, historical anthropology, behaviouristic psychology, and psychoanalysis -specifically invoked by Murdock in Social structure (1949a: xi-xvii; cf. 1949b) could well be applied to an analysis of the social behaviour of a single community or tribe. Although he has published the findings of his own fieldwork in British Columbia, Oregon, and Truk (1934a, 1936, 1938, 1948, !958?, 1965b; Murdock and Goodenough 1947), it is clear that analysis at this level is not Murdock’s main interest. Indeed, another of his criticisms directed against his British colleagues is their concentration on the intensive study of a few societies (1951a: 467). His preferred universe of discourse is the whole of human social life, whether contemporary or extinct, tropical or arctic, tribal or industrial. He is not concerned with social life among other animals, for the culture of those animals with a pronounced social life, such as termites, is inherited and, unlike human culture, does not have to be inculcated (1940a: 364-365). Although learning principles are essentially the same for man and most mammals, man’s ability to use language makes his form of life distinctive. All human cultures, because they are carried by men, have a great deal in common. These common aspects make possible generalizations that are true for all humanity or all cultures.

This reference to all human cultures explains why Murdock is perennially preoccupied with constructing an adequate ethnographic sample. Only in this way can he be sure that his propositions are genuinely expressive of a character of human activity as a whole, and not true merely of human beings in Africa, or during the Middle Ages, or at a subsistence level of economy. We might argue that, despite their limited reference, propositions that are true for only one segment of human experience, whether this is delimited temporally, spatially, or by cultural and social criteria, are nevertheless of scientific value and worth the effort needed to establish them. But for Murdock these propositions are definitely of lesser value; hence his appeal to sociologists, political scientists, and others to find in the pages of Ethnology data from a sample of societies ‘extensive enough to be truly representative of the total range of variation in human behaviour, a more extensive field than is provided by the materials on European and American culture to which their attention has traditionally been limited’ (Murdock et al. 1962a: 1-2).

In this chapter we are not concerned with the whole range of Murdock’s writings as an anthropologist, nor with the qualities that his work has in common with that of his colleagues. Our scrutiny is directed to the distinctive features of the cross-cultural method, as exemplified in the work of Murdock and those associated with him. Hence I cannot follow his example, and construct a sample of cross-cultural ethnologists or of articles and books using the cross-cultural method, which could be analysed quantitatively. Instead of picking at random, I try to exercise judgement and discretion to bring to light the structure of thinking underlying the cross-cultural method and to test it for logical consistency. First, I describe briefly the theories used by Murdock, what data provide grist for the cross-cultural mill, and how these are put into standard form. Then I discuss a crucial analytical step, the use of synchronic data to infer the existence of diachronic processes. The next section deals with the need for statistical techniques and is followed by a discussion of the encoding procedure characteristic of the cross-cultural method. Lastly I discuss the concept which, for me, constitutes the Achilles⁵ heel of the method, the sampling unit. The chapter ends with a brief assessment of the value of Murdock’s work in the recent development of social and cultural anthropology.

2 Data and disciplines

Murdock begins his book Social structure with the sentence: ‘This volume represents a synthesis of five distinct products of social science – one research technique and four systems of theory’ (Murdock 1949a: vii). In this chapter I am concerned more with Murdock’s ‘research technique’ than with the systems of theory that he employs, for it is essentially the research technique that marks off the writings of his school of cross-cultural inquiry from other contemporary publications in anthropology as well as from the products of other modes of comparative investigation. The research technique is distinctive, whereas the four systems of theory are widely employed in social science, even if Murdock’s selective amalgam of the four is idiosyncratic. Despite Murdock’s tendency to identify comparative studies in anthropology with that particular variety of comparative inquiry which he has developed, there have in fact been many studies using other techniques to assess evidence from a large number of diverse societies and cultures (cf. Lewis 1956). Similarly, many anthropologists have used statistical methods for purposes quite distinct from those envisaged by Murdock (cf. Kluckhohn 1939; Mitchell, J. C, 1967). In this chapter we shall confine our attention to one particular application of statistical methods, and for brevity I use the label ‘cross-cultural’ in this restricted sense. As already mentioned, the cross-cultural method was not developed from scratch by Murdock, but had its origins in the work of Tylor (1889), Steinmetz (1896; cited in Naroll, 1962: 5-6) and others (cf. Kòbben 1952). Murdock has, however, written extensively about the assumptions and aims of the method, and our attention will be concentrated on what he has to say.

Yet before we can come to grips with the research technique specific to the cross-cultural method, we must look briefly at the four theories, at how they are used and at the ethnographic data they are called upon to explain. Only after we have discussed theories, data, the verbal arguments linking them, and the model of society which emerges from this linkage, can we begin to consider to what extent the quantitative research technique provides adequate tests of the theories and of the various testable propositions derived from the theories. In this section we shall therefore introduce a few definitions, sketch the range of information about the real world which Murdock handles, and classify the kinds of explanation he seeks to establish.

It will help our discussion of Murdock’s cross-cultural method to follow a terminology based on that used by Textor (1967: 20-21). By cultural ‘characteristic’ I mean some aspect of culture such as ‘mode of marriage’, ’settlement pattern’, ‘mean size of local communities’, or ‘linguistic affiliation’. Characteristics typically appear as the names of columns in Murdock’s ‘World ethnographic sample’ and in the Atlas. By ‘attribute’ I mean the value taken by a cultural characteristic in a given society. Thus ‘marriage with bride-service’, ‘migratory or nomadic bands’, ‘communities with fewer than fifty persons’, and ‘Khoisan language stock’ are typical attributes, and are the values that happen to be taken by the four characteristics mentioned in the culture of the Kung Bushmen of the Kalahari, the first society listed in the Atlas. Murdock himself uses a variety of terms – trait, element, item, etc. – for our ‘characteristic’ and ‘attribute’ (cf. Murdock 1932: 204-205). The essence of the cross-cultural method is the establishment of statistical associations between pairs of attributes.

The number of attributes associated with a given characteristic must be at least two (‘present’ and ‘absent’). In the Atlas, most characteristics have more, with a maximum of 128 attributes for ‘linguistic affiliation’. The ‘World ethnographic sample’ provides information on thirty characteristics. Some of this information was presumably employed in Social structure, but the information on inheritance, extr...

Table of contents

• Cover
• Title page
• Copyright page
• Dedication
• Contents
• Acknowledgements
• Preface
• 1 Safety in numbers
• 2 Real models
• 3 Irreducible principles
• Postscript
• References
• Index