PREAMBLE
Behavioral differences between men and women have generally been attributed either to natural, and therefore, essential, differences in biology, physiology, genetics or to cultural, and therefore non-essential impositions, the fortuitous demands and/or accidents of a social system and the dialectics of history and/or of the human mind. By this logic, the observable phenomena that distinguish men from women must be biological or social, natural or cultural. The terminology of discourse is inexorably bifid: we talk of biological sex, on the one hand, and social role, on the other. Although some confound the two kinds of factors,1 most explanations of sex differences may be lumped into one or other of these polarized classes.
For many purposes this formula is both convenient and appropriate. But where the objective is an understanding of the whole human creature, the binary habit is a twofold impediment. First, if we insist on a nature versus culture dichotomy, we cannot deal with the fact that (even) that which is to be called nature is a construct of culture; that the classification of a human being as male or female is done (normally, if not always) at birth by other human beings according to their superficial reading of quite superficial body signs; and that this classification (normally, if not always) determines the subsequent socialization and, to some considerable extent, the behavior of that human being. Second, perhaps consequently, we lack or have lacked a terminology common to both biological and sociological approaches. We have used the same terms to refer to quite different phenomena and have confused each other horribly.
Some solution to the problem of terminology was both prerequisite and consequent to discussion in this symposium. All of us were soon persuaded that a minimum of three levels of sex difference had to be recognized: sex or sex class,2 gender behavior, and sex role. Thus:
1. Human beings are slotted into sex class male or sex class female and the “folk” classification (nearly always) tallies with the “scientific” distinction between male and female sex.
2. Gender or gender behavior is the social correlate of sex class: being in or having been slotted in a particular sex class, one learns to behave accordingly, to be masculine or feminine according to the norms of a particular culture.
3. Within each culture certain things are done, certain functions performed by men, others by women, and these sex-linked tasks give rise to sex roles.
We still did not know the relative weight of these three sets of factors, but at least we could begin to discuss it.
This is all to the good. Improved communication between the sciences of man is no mean achievement. But returning to my own corner of the problem and the effort to explain degrees of variation in the significance of sex or sex differences within, between, and across cultures, I find I am no further ahead. Certainly, where I am using the materials of others and largely where I am using my own, it remains extremely hard to decide whether the sex differences observed and/or reported are functions of sex class, gender behavior, or sex role. It may be that terminology is the least of our difficulties.
INTRODUCTION
In the explanation of sex difference, social anthropologists have been extraordinarily careful to keep their distance from biologists. The reasons for their doing so are by now common-place: the founding fathers and mothers were not only struck by the significance of cultural factors, they were also constrained as a profession to distinguish themselves from biologists and anxious, for good moral reasons, to correct the errors of vulgar evolutionism.3 The effect has been to lean heavily on the concept of social role and so to demonstrate that the business of being a woman—relating to other women, to men, to the economic and social system—depends largely if not only on “the position of women,” on the way in which “the female role” is defined in each society.
The weight put here on the female side is not added to fit the topic to this symposium: it is striking that sex differences or variations in the balance between men and women are never analyzed from the perspective of “the male role” or “the position of men”; and although all role theorists list sex roles somewhere in their compendia, “the position of men” or “the male role” is not normally mentioned at all. Banton, for example, asks us to consider “What would be the social consequence of allowing an individual to occupy both the female role and the priestly role?” (1965:32–33), but refers later to the probable reaction to a “man filling what is considered to be a woman’s job (e.g. nursing)” (Banton 1965:51). To write the second statement in the form of the first makes a very different point: viz. “What would be the social consequences of allowing an individual to occupy both the male role and the nursing role?”
The difference in gloss is not simply that “the role of female tends to be less independent of other roles than that of male” (Banton 1965:34), nor that social scientists, being notoriously androcentric, can more readily stereotype and so objectify “the female role.” The point to be made is that two peculiar difficulties inhere in the analysis of sex roles. First, while role is an explicitly “cultural” concept, even when applied to pre-cultural/proto-cultural species (Benedict 1969), the conceptualization of sex roles involves inexplicit assumptions about the “essential” attributes of each sex class. Second, the concept of “female role” is not equal and opposite to the apparently corrresponding concept of “male role.” In combination, it is as though membership in the female sex class constitutes a role, but membership in the male sex class does not (cf. Sullerot 1971:7).
The object of this exercise is to examine the effect of this unstated hypothesis on the analysis of sex difference. I shall begin by posing the reverse, i.e., sex, specifically female sex, does not constitute a role, so that it may be tested in a number of ethnographic situations. I shall try to distinguish the logic of situations in which sex is (only) perceived to be a role from those in which it (actually) is a role. I shall work towards the conclusion that there is no analytic meaning in such concepts as “the position of women” or “the female role” is society X or society Y, since the inevitable presence of the sex attribute tells us nothing at all about its general social relevance. The phrases, like the attribute to which they refer, must be narrowly qualified by context (exactly when and how is sex class socially relevant); and by perception (in whose is it relevant?).
I shall first try to make clear the ways in which I am using the notions of situation and role, and shall then deal with some ethnographic material in four sections in such a way as to demonstrate if not to exaggerate the peculiar epistemology of sex difference. It will be obvious that the purview of this paper is the social differentiation of male and female. Given the range of contributors to this volume I have neither the temerity nor the need to treat biological and ethological perspectives as such. The conclusion, however, attempts to relate the various “kinds” of human sex difference to each other, and to problems raised in the previous sections. The effort is very tentative, but anomalies in the data demand that it be made.
The contents are:
1. The logic of sex class situations
2. The problem of sex roles
3. The over-sexed peasant
4. The asexual ethnographer
5. One ethnographer × 3 presentations of self × n situations
6. The North American perversion: gender behavior idealized
7. Conclusion
1. THE LOGIC OF SEX CLASS SITUATIONS
Models for the analysis of situation are not wanting in the social sciences. For my purposes the most useful are probably those of Goffman (1959, 1971) and Nadel (1957). Gellner’s (1973a) paper underlines the enormous difficulties inherent in circumscribing (or, as he argues it, refusing to circumscribe) the context in which a concept is meaningful. The same difficulties inhere in attempts to define situation: it is hard to know where to stop. Indeed, without explicit criteria of relevance, the tendency is to define a situation in wholly subjective or psychological terms: the situation is only what the subject perceives. Jarvie’s recent exploration of the notion is very explicit. He itemizes the nonsubjective dimensions of situation and castigates holistic and psychologistic approaches for ignoring their effect, and denying the systematic importance of unintended consequences (Jarvie 1972:3–36; see also Popper 1959, passim; 1972: 78).
It is not a coincidence that scholars in economic anthropology are least prone to this fallacy: they are necessarily concerned with real phenomena, with the perception of those phenomena, and with the consequences of the perception. Firth’s various statements of the tasks of the subdiscipline constitute both a plea and a prescription for situational analysis and imply a logic by which the situation can be bounded (Firth 1939:1–31, 352–365; 1951). His latest statement might also be a credo: “There is a structure at all levels—in the phenomena, as in the perceptions which order them and in the concepts which interpret their logical relationships; and it is presumptuous to assign to one level more ‘reality’ than to another” (Firth 1972:38).
I interpret this to mean that the phenomenon, the actor’s perception of the phenomenon, and the observer’s perception of the phenomenon are each integral parts of a situation, and that they act and react on each other in circular—perhaps spiral—not linear sequence. In a discussion of sex differences neither the social nor the biological “side” can claim precedence over the other. Gender is clearly more than a mental construct of the actor and/or the observer, but, in terms of the logic of the situation, it is not more important than those constructs. And, in the time-honored functional formula, a variation in one part constitutes a variation of the whole thing.
2. THE PROBLEM OF SEX ROLES
Concerning role generally, Banton’s formula is conveniently minimal: Role is “a set of norms and expectations applied to the incumbent of a particular position” (Banton 1965:29). Being relatively unspecific, this version allows me the inferences I want. These are that: roles are performed in and constitute part of situations rather than whole societies (Nadel 1957); there may be variations in and conflicting definitions of appropriate performance (certainly most readably described by Goffman 1959); roles and role systems may be classified and differentiated according to a number of different criteria (as Banton himself attempted).
Having said that, it is necessary to take note of the fact that the concepts of status and role are peculiarly subject to conflicting definitions. Since “they are widely used as basic concepts in social analysis, and they are used in attempts to solve an enormously wide variety of sociological problems, this need not surprise us at all. The usefulness of a particular version in one analytic context cannot be expected to carry over into all others (Wallman 1974a). If there are many versions it is because there are many problems.
But such proliferation has two important corollaries. Corollary one: not all the differences are equally useful. Some have been articulated into approaches as distinct from each other as the legalistic (which emphasizes the rights and obligations that comprise a role) and the dramatic (which emphasizes individual performance of a socially scripted part). Others smack of entirely unhelpful pedantry, i.e., the question of how many roles make a status, for example, might better be asked of pre-Renaissance theologians than of empirical social scientists. Corollary two: it is appallingly easy to get one’s methodological wires crossed, to confuse the parts and the purposes of one role model with the parts and purposes of another.
Against this background, the striking fact that most role theorists have extraordinary trouble with sex roles might be attributed to ordinary methodological confusion. This is entirely feasi-ble: all sciences need better theories. But for our present purposes, a better explanation is suggested when we look at the particular kind of trouble sex roles give. Let me try to demonstrate it very briefly.
Nadel (1957) classes sex roles as “non-relational” and “ascribed.” Non-relational means that their performance requires no “other.” Not only can you play them by yourself, but they (apparently) require no “other’s” perception/interpretation/evaluation of performance. Ascribed roles are “assigned to individuals without reference to their innate differences or abilities” (Banton 1965:29). A sex role then does not have to be performed, and it has nothing to do with innate differences. But if it does not have to be performed, it entails no particular rights or obligations; being non-relational, it entails no opposite number, no audience, no interaction. It is not then a role, socially scripted, it is an attribute, socially passive. But an attribute is (usually?) an innate difference, and sex roles, being ascribed, (should) bear no reference to innate differences. The methodological confusion is compounded.
Banton (1965:33) tries to clear it by differentiating roles along a scale from “basic” to “independent” that “compares the extent to which particular roles can be played independently of other roles.” On this scale, sex roles are the most basic and are graphically placed ve...