11
The Dynamics of Swahili Culture
A Status-Centered View

As it has for centuries, the culture of the Swahili community continues to provide a basis for its members' social lives and acceptably effective procedures for pursuing their individual interests and personal needs. The sets of understandings that make this possible are changing, as they always have, with new elements, new organizations, and new distributions emerging at a rate that may be faster than in previous eras. But there can be no reasonable doubt that this culture works to serve those who variously share its elements. The aim of this book has been to illuminate the processes that make this possible, and a number of hypotheses have been proposed which attempt to do that.

For much of anthropology's history, the main approach to explaining culture's operation has been to invoke "shared beliefs and values." Once beliefs and values were identified as part of culture, the means whereby they actually affect those who share them received only slight attention. Since all group members share the same beliefs and values, the existence of the ones concerned with the behavior or institution at issue was often taken as a sufficient explanation.

There can be no serious question but that culture's effectiveness stems from its contents being shared. Even if all of its elements were shared by everyone, however, the processes involved in their operation deserve the sort of attention cognitive anthropologists have been giving their psychological aspects. And although the social processes also require attention, they have received only a limited amount.

The fact that sharing is less than universal for many or most cultural elements makes this an even more pressing matter. The ways in which under-

standings shared by only a small proportion of the community's members influence the behavior and lives of the whole community, including those who do not share them, has received little attention. The same is true of how understandings shared by everyone are effective, but the problem is more dramatic and obvious when our concern is with those cultural elements that may be unfamiliar to the majority.

There are a few understandings that seem to be shared by all, or almost all, group members. Those that are specific enough to guide behavior are usually of a broad and inflexible character of little use in dealing with the intricacies of life everyone faces every day. These sweeping and general understandings are of undoubted importance to both the individual and social life, but their usefulness in action depends on their being supplemented by a host of understandings whose sharing is often unevenly distributed among community members.

This type of distributed sharing is characteristic of most understandings. Some of the cultural elements most vital to the continuation of the community and to the adaptation of each of its members are shared only among a minority. This limited sharing is obviously characteristic of the procedural understandings that are the basis for such skills as compounding herbal medicines, the teaching of Arabic, and elaborate lovemaking, but it is also true of understandings concerning the specifics of performance and evaluation as a mother, a Muslim, or a debtor. Much remains to be learned about how this distributed sharing and its involvement in cultural organization operates, but there can be little doubt that just what is shared, by whom, and the social relations among the sharers and nonsharers are among the most vital aspects of cultural dynamics.

Clearly, much of the distribution of sharing is according to statuses of the sort just mentioned (see Schwartz 1978 and 1989:119–121, for a somewhat different view of the same phenomenon), but even within status-based categories, the sharing of understandings concerning the status itself can be distinctly limited. This has been shown not only for the Swahili (see chap. 4) and the four European groups to which they have been compared (Swartz 1982a ) but also for American college students (Holland 1985, 1987a ). Holland reports such sharing as concerns both the romantic and academic activities of the students she studied. For the Swahili, incomplete sharing was found in all three of the domains studied: nuclear family life and relations (chap. 5), intergenerational relations (chap. 6), and body functioning and illness (chap. 9).

It might be thought that the limited cultural sharing found in the Swahili community is an indication of the group's decline as a consequence of its cultural disintegration. Modern urban societies are famous for their fragmentary nature and the isolation of their nuclear families from one another and from other groupings (e.g., Bott 1971), and the Swahili are unquestionably urban

and in many respects "modern." Moreover, the community has experienced some notable changes in recent decades.

Despite these, and they include changes in community structure and the nature and frequency of collective action, the community remains the focus of its members' lives. Even without the broad-scale economic, political, and ritual activity that once characterized it, the community continues to serve as the matrix for its members' social relationships as well as the source of most of the means they use to satisfy their needs and, more broadly, pursue their personal goals.

Walking through Old Town in the evening, one encounters groups of Swahili young men standing in front of the houses and on the corners chatting among themselves. When their neighbors from other ethnic groups pass by, they greet them politely and even, sometimes, exchange a few pleasantries. But the "outsiders" do not stay to pass the evening. These groups of Swahili youth are just that; members of other ethnic groups do not belong even if they live next door. In fact, for community members of all ages, despite the constant association with people from other groups at work and in school, close relationships of any kind with outsiders are very rare.

In addition to almost all voluntary association being with other Swahili, the views and actions of fellow community members are taken most seriously in social, personal, and family matters. Those of outsiders have far less social and moral weight, however important they sometimes are economically and politically. The community, in sum, endures as the prepotent source of moral assessment, as well as of social life, for its members.

Invoking "Shared Beliefs and Values"—and Why Not

Perhaps the enduring value of the community is not surprising. The Swahili are a "traditional" community with a long, unbroken history, and one expects the culture of such a community to be effective. The temptation to explain this effectiveness by reference to "shared beliefs and values" is a powerful one. It can, after all, hardly be doubted that shared beliefs and values are the bedrock of life in this, as in every, community.

The fact that these beliefs and values, these understandings, are shared by some and not others complicates the explanation. But even if the issues focusing around sharing are set aside, major difficulties in understanding culture's effectiveness on the basis of calling attention to its contents remain. Thus, if all the understandings shared by community members were identified, the accomplishment would be only a very limited contribution to an explanation of the nature of the community and its members' behavior. In part, this is be-

cause, as Spiro has shown (most recently, in Spiro 1984), the sharing of culture does not always or necessarily entail behaving in accord with that culture.

In addition, then, to identifying the beliefs and values that people share, it is also necessary to address the vital issue of cultural conformity. If a large enough portion of the population is not influenced by the understandings shared among community members, it is hard to imagine how the group can be said to have an effective culture—or how it could exist.

Simply determining whether or not group members are guided by cultural elements is insufficient in assessing the contribution of the elements to the culture's effectiveness. This is because cultural elements that are not associated with the behavior they call for can nevertheless contribute to the effectiveness of the culture as a whole.

This last consideration has been discussed here (see chap. 6) as the part "tokens" play in social life. In short, people sometimes give indications of sharing beliefs or values with others even though those cultural elements do not actually guide their behavior. The indications themselves, however, encourage and smooth the course of interaction and the continuation of relationships, thereby contributing to culture's effectiveness despite the fact that they do not affect behavior in the way their contents seem to indicate.

"Tokens," as these not necessarily guiding understandings are called, are one aspect of processes that proceed on the basis of limited sharing. Another important issue, and one that has received virtually no attention, is how understandings affect group members who do not share them. Several processes by which this occurs have been described here. A brief review of them may be useful.

How Cultural Elements Affect Those Who Do Not Share Them: Statuses, Cultural Distribution, and Prediction

The importance of the processes whereby cultural elements affect those who do not share them is rooted in the fact that some, possibly many, of the understandings that serve to provide means for dealing with crucial problems are shared among only a relative few group members. Since it is true that despite this the culture serves the needs and interests of all group members a considerable proportion of the time, an examination of these processes is clearly vital to any adequate understanding of culture's functioning.

As Schwartz (1978) was first to point out, cultural elements are unevenly distributed even among those who are directly affected by them. Less than universal sharing of elements within a group is not necessarily a hindrance to the effectiveness of those elements or of the culture as a whole. In fact,

Wallace (1970) has argued persuasively that the simple—and undeniable—fact of incomplete sharing of culture's elements does not hinder the ability of culture to serve as the basis for social life and personal adaptation. Incomplete sharing, in fact, has quite positive functions.

Many a social subsystem simply would not "work" if all participants share common knowledge of the system. . . . [C]ognitive nonuniformity subserves two important functions: 1) it permits a more complex system to arise than most or any of its participants comprehend; 2) it liberates the participants—from the heavy burden of learning and knowing each other's motivations and cognitions (ibid., 35).

Wallace's well-known solution to the explanatory problems presented by the recognition of the incomplete sharing of culture's elements is what he calls "the organization of diversity." This guarantees orderly relationships "not by the sharing of uniformity, but by [the participants in the relationships] . . . capacity for mutual prediction" (ibid., 24). There are difficulties with Wallace's view of culture as "policy" as developed in his classic Culture and Personality , but there is a substantial basis for agreeing that mutual prediction plays a key role in culture's functioning.

Part of this mutual ability to predict the other's behavior is based in beliefs and values actually shared by those in interaction. However, even setting aside the fact that the universal sharing of cultural elements is not a necessary condition for their effective functioning, it is also true that when such sharing is present, it may not be sufficient for that functioning. Cultural sharing only affects what people do directly when the shared understandings are used as active guides for behavior. Sharing without sometimes conforming to some of what is shared makes some contributions to social life, but this only sets the stage, as tokens do, for the influence of understandings that are both shared and serve as effective guides. What is needed for many results is quite clear guidance, allowing people to accomplish their ends and to provide the limited predictability essential to social relations.

But the specific understandings that might provide such guidance are often inadequate. On the one hand, the understandings that are shared by all, or nearly all, community members are usually so coarse in their behavioral guidance as to preclude their effectiveness as a basis for predicting behavior or, even, accomplishing ends save in the most familiar circumstances. More directly to the point, these widely shared understandings are usually very broad and following them does not necessarily instill the needed confidence that behavior can be predicted, especially in the multiplex relationships that are crucial to individual and community life.

In chapter 8, it was noted that all interviewed community members mentioned a few of the same activities as shameful, and this suggested that they

all agreed that avoiding these behaviors was desirable or essential. Included among these prohibitions—and they all were that—was not being seen naked in public, not stealing, not using or selling alcohol, and not begging for food. These are undoubtedly important prohibitions and may add a significant, though sharply limited, element of predictability to relations among community members. However, they hardly qualify as a basis for making the predictions on which social life depends, especially not in the relations among kin, neighbors, and friends on which many aspects of community life depend.

The necessary predictability, however, can be based on the sharing, limited even in multiplex relationships, that seems to be characteristic of humans provided either that the relations are strictly limited in their scope or that they are mainly based on general expectations. In the first case, one needs few expectations to deal with bus conductors, and, in the second, those fundamental to relations with parents, spouses, and neighbors may be effective even though they are vague and general.

What is important about this is that culture's functioning as the basis for social relationships can depend on mainly specific understandings when only a few are involved, but in wider scope relations, the functioning can proceed on the basis of only a few, broad cultural elements. In either simplex or multiplex relations, then, limited sharing is quite sufficient for the relationships to proceed. Group members' belief in broad sharing may be useful, and may be provided by tokens, but in reality such unanimity of understanding is not called for.

Statuses: Bringing Culture to Bear on Everyday Concerns for Sharers and Nonsharers Alike

It has been a central tenet throughout this book that understandings, whether shared by many or few and whether specific or general in their reference, are useful as guides for behavior only as part of a cultural complex, "status," having three distinguishable sets of understandings. These three sets indicate who belongs in the category that is the basis of the status, under what conditions and what combinations with other statuses the one in question operates, and how those in the category can be expected to act and react in various situations and relationships. Through these three sets of understandings working together, culture's elements are brought to bear on individual's interests and problems as well as on social relations. Through their three sets of elements, statuses are a culturally constituted means whereby culture's elements are distributed among group members according to the situations that arise. Through this distribution statuses bring culture's components to bear as guides for behavior in actual circumstances and relationships, thereby mak-

ing culture not a heap of assorted understandings but a useful means of dealing with life.

Without the same emphasis on status as given here, Roberts (1964:438–452) made the general point concerning culture's ability to serve all of a group's members, although most of its elements are shared by only some of them. His view of culture is as an information economy involving storage, retrieval, and decision making. Roberts's approach has been materially elaborated as "distributed cognition" (Hutchins 1985), and recently there has been further development of the concept and an important empirical demonstration of its functioning (Cicourel 1988).

Evaluation as the Foundation of Social Life, Status as the Foundation of Evaluation

The distribution of understandings by statuses is both central to their effectiveness and unavoidable. It is only by understanding what people (and what sorts of people) do what under what circumstances that social life is possible and at least a minimum of one's needs and interests are attended to. Nor are statuses limited to their vital function as guides for social relationships and interactions. In fact, it is the evaluative functioning of statuses, more than any other single factor, that lies at the heart of culture's effectiveness. Hamlet's "There is nothing either good or bad, but thinking makes it so" would be a fuller, if less lyrical, characterization if it were clear that "thinking" is closely tied to the understandings in the statuses to whose categories we assign ourselves and to the understandings in the statuses whose occupants we evaluate and understand to be evaluating us.

The evaluations embedded in statuses are the bedrock of social life. As concerns actual behavior, interaction cannot proceed without those involved understanding each others' behavior as being within the limits set by the expectations of the statuses they assign to each other and themselves. It is in interaction that the basis for evaluation^[1] is manifested, and it is the statuses of both the evaluator and the evaluated (with all parties being both) which provide the standards for evaluation. Multiplex relationships depend more on inclusive evaluations of overall performance and less directly on particular behavior in specific contexts than simplex relationships do, but the evaluation process is central to both types of relations.

In the discussions of cultural models (chap. 7) and of shame (chap. 8), there was substantial evidence to show that the Swahili, like everyone else, are nonstop evaluators of everything they encounter, especially of the people (including themselves) they deal with. At the same time, everyone is very much alive to the fact that he or she is being evaluated by those same people.

Given the importance of the evaluations of those with whom one has ongoing relationships, there is a powerful impetus to act in ways understood as likely to win approval (or, at least, avoid disapproval) from partners in those relations. Putting this in terms of the status complex of understandings, those in interaction come to some agreement as to which statuses are salient and to meet, to some degree at least, what they take to be the expectations in those statuses.

In order for this to happen, the agreement must include what statuses each party occupies. This agreement emerges from and is confirmed by signs (actually symbols) of acceptable evaluations of the behavior that each participant gives the others. Such evaluation usually derives from at least partially shared understandings about what is acceptable and what is not. In other words, it is in social life that shared understandings (culture's elements) become "visible" as such through the behavior they guide and its acceptance. The acceptance of behavior as demonstrated by positive symbols from partners seems to be taken as acceptance of at least some of the understandings that guide that behavior.

The direct and immediate importance of social relationships to group members and the central part played by mutual evaluation in those relationships puts the understandings actors take as the base for evaluation at the center of culture's ability to operate. This is a point long emphasized by social theorists (e.g., Hallowell 1955:105–110, Durkheim 1961:52–55). Because of the central contribution of morality to cultural dynamics, I first proposed a definition of "culture" as "the sum of the morally forceful understandings acquired through learning and shared by the members of . . . [a] group" (Swartz and Jordan 1976:46). It is now clear, however, that this definition is seriously flawed in that it requires shared understandings to be "morally forceful" if they are to be included as part of a culture.^[2] This directs attention away (technically, it precludes it) from the empirical study of the presence or absence of moral force in shared understandings when such study is clearly worthwhile.

How Unshared Evaluative Understandings Serve to Affect Behavior

Moral or aesthetic judgments affect behavior, when they do, on the basis of two different sorts of understandings. One of these is "values," the evaluative understandings concerning what is good and bad, beautiful and ugly. Unlike these, the other sort is made up of descriptive understandings concerned with how people actually evaluate what one does and what the social consequences of behavior are likely to be.

These two sorts of understandings can, and sometimes do, affect the indi-

vidual in similar or identical ways. This happens when the individual holds evaluative understandings that would lead him to act in ways that he understands others to view as he does.

The two sorts of understandings, however, sometimes do not work together. This happens as the result of the actor believing that the evaluator holds evaluative understandings concerning the behavior at issue different from his. One way, and a common one, this occurs arises from the role relationship of the judged and the judge. Thus, understandings that hold that men having secret wives (see chap. 8) is a bad thing coexist in the culture with understandings saying it is satisfying and rather delightfully manly.^[3] Men who make such marriages understand many others to condemn them, but their own view is that such marriages (at least their marriage) are acceptable and, since they are permitted by Islam, quite free of sin or blame.

What seems important about the partial independence of the two sorts of understandings is that either can function independently of the other. It is obvious that people are affected by ideal understandings that they hold even if others do not share them. It may be less obvious that ideal understandings people do not share can also affect their behavior through the effect of descriptive understandings concerned with how others evaluate. So long as actors care about the evaluations of the likely evaluators, the evaluations the latter are understood as likely to make will affect what the actors do regardless of how accurate the understandings may be. Even should no one in the group personally understand secret wives and disregard of others' rights as intrinsically bad, the belief that they do may well affect what those who care about their views do.^[4]

Statuses as a Source of Morality When Understandings Differ

The present point is that for morals to affect behavior, it is not necessary that the understandings concerning the virtues of that behavior be shared by those affected. Some or all may share them, but the behavior could be the same for the sharers and the nonsharers provided that the latter are concerned about the evaluations of the former and respond to understandings that they believe some or all of the others hold.^[5] Much or all of this belief is the result of status-guided interaction where status assignment includes the attribution of expectations including what category members are believed to value.^[6]

Broader Consequences of Social Relationships

As important as social relationships are in themselves, their significance is by no means limited to participants' concern with the specific relationships

occurring at the moment. Many Swahili care little about the simplex relationships with grocery clerks, bus conductors, and tailors, but most of them do care about the results of those relationships, about what they get from participating in them. They care about the food they buy, the trips across town that they take, and the clothing that they wear.

To get these things, they must sustain relationships with the clerks, conductors, and tailors long enough to accomplish their ends, and this depends on meeting some of the expectations in the appropriate roles. One may care little for clerks' good opinions, but one turns over to them the money called for and refrains from too much quarrelsomeness to avoid trouble while getting the desired things they control. More than this, to some extent, one's interest in the food, trips, and clothing usually derives in some part from one's relationships with those who share meals, live in the places traveled to, and whose opinions about clothing matter. Finding a basis for getting along with occupants of statuses that matter little in themselves is, in fact, essential to getting along with occupants of statuses that matter a great deal.

"Delivery Systems" and Cultural Guidance for Life's Problems

The vital function performed by statuses in distributing culture's elements among actors and situations warrants a brief concluding examination. This distribution depends on the basic fact that in all interaction people are always categorized, by themselves and by others, according to identifying and salience understandings. These latter understandings provide what might be called "the delivery system" for bringing the expectations with which they are associated into the relationships where they guide behavior. Since situations result from an interaction between events and the statuses assigned and assumed by those involved, this process not only assigns particular cultural elements to specific individuals but does it with regard to the problems of the immediate situation.

The statuses people assign one another have expectations that work as the common standards in double contingency and determine to a considerable extent how they treat each other. Thus, a situation involving two men is one of patient care if the assigned and accepted role pair is practitioner-patient and is one of informal chatting if it is baraza member-baraza member. In this way, by agreeing on the complex of understandings that make up statuses—identifiers, expectations, and salience understandings—those in interaction jointly provide a culturally constituted means for the distribution of culture by mainly establishing the situation^[7] and by providing guidance for one another based on such sharing of understandings as may be present and effective and on each being willing to adapt to the other's apparent evaluations.

Nor is cultural distribution the only function statuses serve in addition to guiding social relationships. In order for culture's elements to have any utility, those who share them must be able to choose among them, rank them according to precedence, decide which call for or preclude which others, and group them according to similarity and difference. Such organization of culture's elements is a sine qua non of culture's effectiveness in guiding behavior. Cultural organization, like cultural distribution, cannot be understood without close attention to statuses and their operation, especially in guiding social relationships.

Limited Sharing within Statuses and the Sources of Statuses' Effectiveness: The Issues

The fact that culture is only very partially shared and that this is true within status categories as well as between them raises the question of how they can work effectively. A closely related question concerns the foundations for the differentiated cultural conformity without which such sharing as may exist has little significance either for individual survival or group continuation.

These two sorts of questions, concerning sharing and concerning conformity, are closely, in some instances inextricably, connected. The sharing issues are concerned with the processes involved in having at least the minimum interpersonal agreement necessary for statuses to operate. The conformity issues are concerned with the processes involved in expectations actually serving as guides to interaction. Together, the processes concerned with the issues involved in sharing and in conformity provide both the necessary and the sufficient conditions for interaction.

Both the reaching of agreement and the following of the expectations agreed on depend on social relations actually proceeding, since statuses function mainly in interaction rather than in the mind alone. Since this is so and since interaction only occurs when the participants believe they can predict each other's behavior, predictability is essential not only to guiding social relations but also to culture's general effectiveness as the basis for individual and group adaptation.

Conformity and Status Effectiveness: Universal Sharing and the Role of Tokens

A necessary step in statuses being generally effective is specific statuses being effective, with some of these contributing materially to the effectiveness

of others. In Old Town, and probably everywhere, among the statuses that contribute most broadly to the effectiveness of other statuses, there are inclusive categories that are rarely given names but might be referred to as "community member" or "decent person." These statuses, and their identifiers are historical and associational, involve a quite limited range of expectations, including those based on the few universally shared understandings about what must and must not be done (see above). Nevertheless, they play an important part in promoting the general effectiveness of many of the community's statuses.

A substantial part of the effectiveness of these statuses derives from the expectations of others understood as characteristic of those categorized in them. Importantly, these include highly visible behaviors, mostly of the sort that are carried out (or, more often, avoided) in public. To be well thought of by "decent people/community members" is understood to entail a broad range of behaviors. If one fails to meet expectations for obvious, publicly performed activities, one is likely to be negatively evaluated for the whole range. Conversely, and importantly, if one meets the expectations for public behavior, one is at least eligible for positive evaluation in far more significant domains.

The consequences of the community member status extend beyond the restricted, direct guidance they provide for group members' overt behavior. These derive from the fact that behavior guided by them serves to reassure community members that those around them are like them, that the base for the predictability that is indispensable to social life is present. Even between community members who have substantial differences in their understandings of emotionally charged issues, there are exchanges of tokens that give them reassuring indications that they share beliefs and values allowing continuation of their relationships with one another.

In chapter 6, the tokens involved in generational relations were seen to contribute to their continuation despite serious differences in understandings concerned with the foundations for such relations. The tokens in the form of statements in a particular context gave each of the age groups a basis for believing that those in the other group were willing to take their sensibilities into account by suggesting sharing in areas where, at least as concerns the guidance of much behavior, it does not exist. The assertions members of each age group make about the sources of intergenerational strife provide tokens for the members of the other grouping.

"Tokens" are functionally defined. Any symbolic exchange in which the participants directly or indirectly indicate to one another that there is substantial sharing of understandings between them involves an exchange of tokens. These tokens may also be guides for behavior beyond the exchange in which they occur, but they need not be. The symbols that serve as tokens are not all expressed in speech. By never appearing naked and, probably more impor-

tant in daily life, by appearing in the kind of clothes community members regularly wear,^[8] Swahili proclaim their membership in the group.

By dressing as other group members do and by wearing clothing slightly different from that of the members of any other group, group members give one another tokens suggesting that there are important similarities uniting them and differentiating them from people outside the community. They show that they accept the identifying symbols of being a "community member" and thereby can be taken as predictable and safe for interpersonal relations. Indicating interest in and agreement about important issues reinforces and continues this.

If it can be accepted that interaction is vital for social relationships, that the functioning of social relationships is vital for statuses, and that statuses are at the heart of culture's general effectiveness, then tokens can be seen to play a significant part in cultural dynamics through their contribution to the interaction. Tokens can be based in understandings that are also guides, but even when that is not so, they contribute to social life through promoting what might be called a social and psychological "medium" that is conducive to interaction.

When Tokens are Not Guides But Have Manifest Results

Working from the hypothesis that tokens play an important part in cultural dynamics by helping to provide the medium for the social interaction essential to culture's functioning, a further hypothesis emerges. This is that tokens differ among themselves in their contribution to cultural dynamics, with some having distinct dangers to those dynamics built into them.

It seems likely that tokens that are not guides serve best in encouraging a productive social climate if they leave no "footprints," that is, if they do not interfere with the effectiveness of the complex of understandings in the relationships they promote. When tokens are of the sort found in the study of generational differences, the fact that they are not guides has limited consequences.

The younger Swahili can continue to act as they wish regarding clothing, demeanor, and such, despite saying, in what appears to be their "community member" status, that their own actions are both "bad" and a result of their personal shortcomings. In their statuses as young people, however, they are not bound by their public assertions of self-blame. Similarly, the older people are speaking as community members when they blame themselves for the misbehavior they report, but when they act as parents or as fellow mature men or women, they are not hampered in denouncing the youth and their behavior. The relative isolation of members of different generations from each other

outside the home renders token use less likely to cause difficulties than it would if there were more open and frequent association.

Such insulation from the consequences of asserting views in one status which are less than consistent with those taken in another, however, may not be present or effective in all situations and roles. Tuzin (1976:177) reports that among the Ilahita Arapesh, there is a widespread avowal of the virtue and commonality of fraternal co-residence but that, in fact, brothers often do not live with one another. He notes that normative statements are of a different order from descriptions of experience and that "while they [i.e., the statements of norms] appear to refer directly to experience they are actually of a different logical order, one which, up to a point, cannot be contradicted . . . by the statistics of 'real' behavior" (ibid., 197).

However, he goes on to note that continual and frequent violation of the norm, expressed as a "token" in my sense, leads to an undermining of the effectiveness of the understandings as useful guides.

These many exceptions [to the "rule" and asserted practice of fraternal co-residence], taken collectively and over the long term, pose a serious and increasing threat not only to the particular norm but to the wider set of values to which it contributes. [If group members come to perceive the invalidity of the statement that brothers live together, this] . . . would expose the prescription as starkly, massively [emphasis in original] unworkable and, by pernicious implication, invalid. Here, then, is the psychological raw material of revolutionary and counter-revolutionary ardor, of the unutterable bleakness of cultural collapse (ibid., 198).

What Tuzin's discussion of the Ilahita suggests is that tokens that are not also guides work best when they have least direct bearing on aspects of social reality manifestly at odds with the tokens' content and unavoidably perceptible by the tokens' users. If the tokens can indicate deeply important sharing between users and still offer little possibility that their referred to understandings will be seen as false, they would seem likely to be both highly effective and free from the sort of danger to shared understandings generally that Tuzin notes.^[9]

The exchange of tokens serves to increase the likelihood that partners in interaction anticipate acceptable behavior from one another and, therefore, that they are willing to interact. As Tuzin's work suggests, however, tokens can produce a "crisis of faith" in the ability of those in contact with one another to predict each others' behavior. This happens if the tokens' references are to observable conditions that are manifestly contradictory to what the tokens assert.

Even when they work most effectively, tokens that are not also guides have distinct limitations on their contribution to individual satisfaction and group continuation. They can serve to make social life smoother and to encourage

participation in it, but unless they are also guides, they contribute nothing to the "business" of life. Cooking meals, treating the sick, earning a living, and the like depend on social relationships, to be sure, but these relationships need to be more than just social links. The statuses of those involved must contain substantial expectations concerning cooking and who shares the meals they prepare, how to treat illness, and the rest of the concerns addressed in the relationships. More than this, the expectations must at least sometimes be met.

Some understandings must be guides some of the time or there can be no continuing social life or any basis for individual adaptation. Despite the importance of tokens, the Swahili community has not endured for centuries on the basis of a culture limited to sources of mutual reassurance that are no more than that. The necessities of individual and social life depend on statuses whose expectations contain understandings that, if followed, lead to some sort of satisfaction. Such statuses are effective only if there is at least minimal sharing of some of their constituent understandings and at least some conformity to these.

Divergent Understandings and Double Contingency

The minimal sharing just mentioned is all that is required for social relationships. Some time ago, Wallace (1970:32–34) argued on the basis of a gedanken experiment that group members need not share even one cognitive map. He argued that useful and productive relationships can be and are carried out on the basis of what he called "equivalence structure." As noted earlier, Wallace views predictability rather than sharing as the basis for social life, with the needed predictability arising from "equivalent mutual expectations . . . [that] may be termed an implicit contract [emphasis in original], in the general sense of the word contract" (ibid., 35).

Although Wallace does not say so specifically, it would appear that any actor's behavior can be affected by what fellow participants in an interaction do and that their behavior, in turn, is affected by what the first actor does in response to their act (ibid., 27–29). This is similar to Talcott Parsons's (1964:36–43) "double contingency" (see also chap. 7, above), which involves the mutual adjustment of actors to one another's behavior. That is, in any interaction, what the second actor does is contingent on what the first actor did and the first's response to the second's response takes account of what the latter response was.

There can be little doubt that double contingency is a fundamental element in social life. Through smoothing the course of particular interactions, it contributes to the continuation of relationships. The process is one characterized

by continual self-correction, with each participant reassessing his own and his companions' statuses according to clues received in interaction.

The primary task of double contingency, as far as interaction beginning and continuing is concerned, is to achieve agreement on mutual status assignment. People may not begin an interaction with shared understandings about their own and their partner's category memberships, but they will respond to each other's cues until they do or they will cease trying to interact. Even if they share—or come to share—identifying understandings about their own and the other's category and its salience relative to other categories, they must also come to agree on the expectations the other will manifest in response to what the first does. They will respond to each other's cues and reach agreement, or the interaction will stop when it is hardly begun.

The Necessity for "Common Standards" and the Question of Sharing

Double contingency is much influenced by the immediate responses of those in direct relations with one another, but their responses are not completely ad hoc. Parsons (1964:37) is quite explicit in maintaining that an element of shared "orientation" ("understandings" is close to this in this context) is essential to the operation of this basic process: "The orientation of one actor to the contingent action of another inherently involves evaluative orientation, because the element of contingency implies the relevance of a [shared] system of alternatives . . . the particular acts of evaluation on both sides should be oriented to common standards."

In order for people to be concerned about how others respond to them and about how they should act themselves, they have to have some commitment to the relationship. It is also essential that they have some ability to predict the partner's response, and it is hard to see how this can be based in anything other than some sort of understanding of how the partner will choose among alternatives. This last is an understanding of the evaluations of different courses of action the partner is likely to make. Insofar as the understandings of alternative evaluations held by the different participants are similar, their interactions are probably likelier to lead to results they will find acceptable or desirable. Since, however, evidence supports much of Wallace's skepticism about the existence of sharing of understandings, it seems worthwhile to examine processes whereby what Parsons calls "common standards" are brought into action.

As has been made abundantly clear here, an "all by all" model of cultural sharing is quite unjustified by the facts. If acceptable, it would account for Parsons's "common standards," but the fact of generally incomplete sharing precludes its invocation. At the same time, a "none by none" view is clearly

untenable. Some sharing among at least some group members is always present. Every study devoted to the topic has shown that sharing is limited, even sharply limited, but all of them, from Roberts's study of the Navaho at the beginning of the 1950s to Holland's study of university students at the end of the 1980s, show the presence of some sharing. The question is not whether there is sharing, then, but what is shared and by whom?

Universal Sharing and the Importance of Relationships: "I Know You!"

Among the Swahili, universally shared procedural understandings are mainly limited to such behaviors as not appearing naked in public. There is, however, universal, or nearly universal, sharing of another order of understanding. As seen in chapter 4, members of the Swahili community are quite uniform in their attachment to and concern about certain relationships. Individuals differ to some extent about which particular relationships are more important to them than others. They also differ in the intensity of their commitment to the relationships they find most important. Despite these differences, everyone has some relationships in which his or her behavior points to their being understood as having deep emotional and/or social significance.

In every case I know about, all such relationships are based on kinship, marriage, and neighborhood. These are the relationships to which people devote most of their energy and attention and the evaluations of partners in them are generally the most weighty and lasting in effect. The broad importance for culture's dynamics of the expectations that are the bases for evaluation in these relationships can hardly be overemphasized. It is a central hypothesis of this study that these expectations in multiplex relationships are fundamental not only to those relationships but to the processes that make culture an effective base for individual and community life.

Relationship Terms and Cultural Models

One of these processes depends on the cultural models expressed in the standard terms used to characterize social behavior, including, especially, behavior in the multiplex relationships themselves. These evaluative terms serve to promote the common standards essential not only to those relationships but, more generally, to interaction within the community as a whole. The terms do this by bringing these standards to bear on individual behavior in a direct and explicit way, with particular relevance to what the individual himself views as the most important relations.

Lakoff and Johnson's influential work (1980) has shown a similar role for

metaphor, but here the process focuses on the direct use of words in ordinary speech to characterize in a value-laden way how people behave, especially how they behave in social relationships. Through their relatively uniform use when applied to different individuals, these characterizations provide an impetus to uniform standards in similar social relationships.

None of the terms used for these characterizations is neutral. Some are approving, though most Swahili terms concerned with behavior in relations indicate disapproval. But in either case, their applicability to acts and types of behavior is quite unambiguous. The terms as used thus openly and explicitly display for all those exposed to them understandings of what is desirable, acceptable, good, and beautiful, or, more often, the opposite. They apply particularly and directly to relationships with others, including those from the social groups (family, neighborhood, and community) that matter most to most people. They thereby serve as models of the parts of the culture applicable in various key relationships for occupants of particular statuses and, by extension, to other relationships.

The approval—more commonly, disapproval—involved in the terms harnesses, so to speak, the concern people have for the evaluations of those who matter to them and makes them amenable to the pressures for conformity to the common standards they imply. These terms, in other words, serve as cultural models that not only display shared understandings but do so in a way that is conducive to their actually affecting behavior.

Keesing (1987:374), views a "cultural model" as constructed from elements of the same general sort as the relationship terms of interest here:

What . . . makes them models? Presumably, it is their paradigmatic, world-proposing nature. These cultural constructions of the everyday world do not consist of disconnected bits of cultural wisdom, expressed in precepts, parables, proverbs, or pragmatic, probabilistic operating strategies, but of world-proposing . . . models embodied or expressed in these bits.

As chapter 7 makes clear, the relationship terms used in Swahili evaluations concerned with characterizing how people act in social relationships are not "disconnected bits" but unite particular characterizations into overall, integrated models. These models find their most direct and behaviorally effective expression when they are applied to individuals by their partners in multiplex relationships through characterizing them, their general behavior, or their behavior in some limited context according to emotion- and value-rich terms.

These terms differ in their scope, but they always involve evaluation of the performance of their subjects in some of their statuses. Sometimes the statuses to which the terms apply are only situational (e.g., apply only to hosts or guests), and sometimes they are broad, such as "Muslim," "man,"

"woman," or "community member." There are terms that are appropriate only for occupants of particular statuses, but there are others that can be used for everyone. The meanings of some terms are the same regardless of the statuses of those to whom they apply or who uses them, but others vary according to the statuses of the referent and/or the user. Some are uniform in their implied judgments, but many relationship terms entail different judgments according to the statuses of the user and the person referred to.

The existence and, especially, the use of quite standard terms that evoke the understandings of what is and what is not proper behavior in particular statuses direct the universal concern for evaluations into active interpersonal and psychological forces promoting the "common standards" Parsons rightly places at the base of social life. The judgments involved in applying the relationship terms to any community member supplies all those who are aware of the judgments, not just those to whom they are applied, with a compelling opportunity to become aware of common standards for behavior or, more often, to be reminded of the costs or benefits flouting or following them bring. The fact that the standards evoked by the terms are applied to or about a specific person carrying out particular behavior gives the standards a clarity otherwise hard to attain. That the terms clearly entail approval or disapproval is often obvious by the manner of their use and application, and this approval or disapproval is often unambiguously associated with strong positive or negative emotions.

In the same way, the judgments made by those about whom one cares are obviously more compelling than those made by others about whom one cares little, but there is often a transitivity involved. If A cares for B and his judgments and C also cares for B, the possibility that A and C will have common standards is much increased, despite their indifference to or ignorance about one another, by their common concern about B. This calls for A and C each understanding B's judgments of him as being based on the same set of understandings. The likelihood of this is increased when B's judgments of both are expressed in the same relationship terms. The use of the terms in characterizing particular actions of specific individuals, not just their existence, much reduces their ambiguity and is a key base for their part in encouraging common standards.

Similarly, if A's relations with B are very important to A, B's views about A's behavior in the relationship A has with C are likelier to be effective than otherwise. This insertion of a third person's standards in a relationship not involving that third person offers the prospect of a standardization of understandings in such relationships when their participants have important relationships with similar "third persons." If the views of the "middle man" (B) are those often expressed in the ordinary relationship terms, with the common specificity these involve, the likelihood that this standardization will be broader and more generally effective is increased.

The terms that Swahili use to characterize participation in social relationships provide exposure to a constant source of clearly implied—sometimes explicitly stated—understandings about how people should and, more often in actual use, should not behave. They are less often of the "X is a bad person" sort than of the stated or implied "X's having done what he did in his dealings with Y is an mshenzi [uncivilized person]," with the connection to X's behavior, his relationship to Y, and the condemnation as an "mshenzi" being quite clear.

Because of this clarity, the terms often make clear the standards that are used in positive and in negative judgments of easily identified behavior in particular relations.^[10] The fact that the terms apply to and are not infrequently made by partners in multiplex relationships enhances their ability to affect behavior.

The Importance of Positive Assessments: A Universal Lesson and Ubiquitous Source of Cultural Conformity

It may well be that the particular "shalts" and "shalt nots" taught in childhood socialization are not the most important lessons of this crucial period as concerns culture's effectiveness. They are unquestionably important, but what is learned varies from family to family and person to person. This variability is far less characteristic of the teaching of two quite unspecific things.

The first is that the judgments passed on one depend on others' assessments of one's behavior. The second is that positive assessments are far more desirable than negative and that this is truer as the judges matter more. If group members learn these two things, and they do in every society as an essential part of becoming human, a solid basis is laid for participation in social life that is independent of extensive and uniform sharing of the specific understandings that guide behavior, including behavior in social relationships.

Learning that evaluation is constant and that positive evaluation is profoundly desirable has a number of important consequences for culture's effectiveness. One of these is making the cultural models displayed through the use of common terms characterizing behavior an effective force for cultural conformity. These models are not the only source of the common standards essential to interaction, but they use the universal concern with evaluation to encourage the operation of a reasonably uniform foundation for interaction. They do this regardless of differences among individuals concerning the virtues or vices of specific behaviors by depending less on particular beliefs and values for their strength than on everyone's awareness of being constantly judged by people whose judgments matter.

The judgments may or may not involve understandings that the judge, the judged, and those aware of the particular judgment all share in the sense that they had similar knowledge of them beforehand or in the sense of all equally accepting them as just, true, or desirable. What they do is provide schemata concerning what is and is not desirable in social behavior as that is understood by some group members. These are presented in a concise but value- and emotion-laden way through relationship terms whose use in evaluation is unmistakable.

Cultural Conformity: Bases for Shame and Guilt

The effectiveness of this process, of course, depends in part on people's concern with others' evaluations. Such concern is not equally present in everyone or equally common in different communities. An oversimplified but useful division is according to the nature of the goals sought. One sort of schema aims primarily at supplying the individual with the positive evaluations of his fellows. A related, but different, schema puts its focus mainly on attaining intrinsically rewarding goals, with others' approval of them or of the means for attaining them secondary.

The difference between the exact behavioral guidance provided by each of the schema would not be obvious. One might seek goal A because one understood that others admired the quest, or, alternatively, one might seek A because one wanted it for itself. In either case, the directly observable behavior is A being sought. Differences would appear in how the individual felt about gaining or not gaining what the schemata called for and the nature of the pleasure or pain experienced.

If painful feelings associated with others' negative judgments are the result, it would appear that what was in operation is what Obeyesekere calls "shame" (1981:131). This is the meaning of the Swahili term aibu, a word often heard in Old Town.

A concern with the correspondence between one's actions and one's own evaluative understandings independent of others' judgments is a different sort of process, one that is nearer to what is sometimes called "guilt" (e.g., Piers and Singer 1971 [1953]:26–27). Although it occurs among the Swahili, it is less commonly heard about there and, possibly, less common in occurrence (see Swartz 1988).

Aibu does not depend on the actor evaluating himself or his behavior in a negative way but on his belief that others do. The Swahili are by no means unusual in experiencing feelings of disgrace, dishonor, and dysphoria as a consequence of believing others do or would disapprove of what they do.

They may, however, be different from the members of some other groups in the extent to which a very substantial proportion of them dwell on the possibility.

Conformity to Communitywide and Status-Specific Understandings

Two different sources of shame can be distinguished among the Mombasa Swahili (see chap. 8). One sort comes from the judgments of what I called "sanctioners," with every community member being a sanctioner as concerns the behavior of some others under some circumstances. The sanctioners' judgments are, and are taken by community members to be, dependent on the category in which the person judged is classified for purposes of judgment and, also, the category that is salient for the sanctioner when the judgment is made. In the status of brother-in-law, for example, a man's judgment of another's adultery is strongly negative, but in the status of crony, the judgment is quite different even though it could be of the same acts by the same man were he not married to the sanctioner's sister.

The importance of status-specific cultural conformity will be obvious, and the emotional consequences of the sort of judgment sanctioners make is an important support for just such differentiated conformity. It is worth noting that such judgments are not limited to communities like the Swahili where shame is a frequent and active focus of attention. Although a pervasive concern with shame may make others' judgments more poignant for more of the population, it is not at all a necessary part of their influencing behavior. If this is so, it implies that even without the extra emotional impetus of shame, the judgments that people attribute to others are universally important in encouraging differentiated conformity even if substantial cultural heterogeneity is present.

People are not, of course, equally concerned about all judgments of themselves. One of the aspects of the social world being divided into culturally significant categories is that assignment to one or another establishes the judgment of a potential judge as being more and less important as concerns the assessment of what an individual does as a member of one or more of his categories. This "choice" of sanctioners is central to the shame process, but it is not limited to shame. Even in instances when the judgments carry little emotional charge, their prospective social consequences can make them a force for conformity to the judge's expectations when the judged understands them as significant. The one judged, then, plays a main part in the potency of a judgment by the status he assigns the judge. Moreover, since the judge's expectations are what determine his judgment, the status he assigns himself

and is accorded by the judged vis-à-vis the behavior at issue is crucial. There is, thus, a dual dependence on status assignment at the heart of the socially, culturally, and psychologically vital assessment process.

It should be made explicit that those involved in this process are not necessarily limited to the actual participants in the role in which the judged behavior occurred. Judgments on behavior are often made on one's performance in a status by people having no direct connection to that status. Thus, one's mother may judge one's performance as an employee, and her judgment may be important to one's behavior because of the character of the multiplex relationship with her.^[11] A neighbor's judgment of the same behavior, however, might be far less influential despite there being a multiplex relationship with him as well. It might be, however, that the neighbor's assessment is nevertheless more influential than that of a customer who is nothing more than that since the simplex relationship, customer-employee, is likely to be influential in only a limited way.

By definition, multiplex relationships cross a number of life's domains, and their participants are the same individuals over long periods of time. The statuses that are the basis for participation in these relationships provide expectations that may serve as bases for evaluations of performances by fellow participants in situations and interactions where the judge may not be directly involved or, even, present. Simplex relations, by contrast, may be highly important to role performance, but their importance is limited to the role in which they exist.

Since people are aware of the "outside" judgments involving multiplex relationships and wish to be positively evaluated by their partners in these relationships, various strategies based on shared understandings are used to increase the likelihood of such evaluation. In the Swahili community, none of these is more striking than control of information.

Even the rather extreme secrecy of the Swahili, however, does not prevent almost every act from being subject to the judgments of sanctioners. More than that, in the Swahili community (and possibly elsewhere), there is an additional and much more limited set of judges, the "arbiters."

These are men of substantial prestige who are distinctive in that they are said to tisha, frighten, their fellow community members. Their judgments are not, as the sanctioners are, relative to the status of the judged. Every community is judged in a similar way, with little attention to differences in gender and age and none to differences in wealth, learning, or anything else. The standards they employ are universal: what is unacceptable for one is unacceptable for all.

Interestingly, however, they make no explicit judgments, nor do they act differently toward those who are supposed to have won their displeasure. Young men—young women may be spoken of similarly, but I never encountered it—are praised for "fearing the faces of the mature people" (kucha uso

wa wazima ),^[12] and this means that the young men behave in accord with universally applicable standards as though they feared the judgments of high-prestige men, the arbiters. The arbiters make no public—or, as far as I know, private—statements about who is naughty and who is nice. Nor are they explicit about which behaviors are judged as one or the other.

The arbiters are, in fact, a sort of evaluative Rorschach. The understanding that they would disapprove of something is what is important since they do not actually disapprove, or approve either, of anything in ways that can be perceived. Not only do they refrain from making explicit judgments but they also continue to behave toward the judged with the same dignified and restrained politeness they show everyone else.

They function in at least one respect as an internalized set of standards does: the judgments are no less real for having no external signs. From the perspective of an observer, doing something to avoid displeasing the arbiters is indistinguishable from doing something because conscience dictates it be done. As with internal prompting, the person involved "just knows" what the arbiters approve and disapprove. In neither case is there a social process others can see.

The only thing that can be seen and that connects the arbiters to approval or disapproval of specific behavior is what can be inferred from their own behavior. What they openly do is almost always viewed as acceptable for everyone since they are the avatars of good behavior.

The arbiters have no Swahili name, but this unmarked category is as much a status as are those occupied by the sanctioners. The arbiters, unlike the sanctioners, have but a single status, and their imagined judgments are virtually uniform regardless of who is judged. Still, both sorts of judges are effective only if their statuses are. Only when men tisha (inspire fear in) others are their judgments taken with the seriousness arbiters' judgments are, and inspiring fear is a main expectation of those in the arbiter status. Similarly, a brother-in-law's disapproval of one having a "secret wife" is in accord with the expectations of his status vis-à-vis one's own as sister's husband and, therefore, is likely to be taken seriously so long as the brother-in-law demonstrates the general interest in his sister's welfare called for by his status's expectations. In effect, then, the ability of judgments to affect behavior depends on the judgments of the judges, whether arbiters or sanctioners.

Statuses and Cultural Conformity

According to the argument to this point, a number of important inducements to cultural conformity operate differentially based on the status of the individual and that of those judging his behavior. "Sanctioners," and everyone is one, may make quite different judgments of the same act depending on

the statuses of the judged and the status salient for them when they make the judgments.

Even the recondite "judgments" of the arbiters can be understood only in the light of the complex of understandings that form the arbiter status and the mainly undifferentiated status, "community member," assumed by those who take themselves as judged. Here the judgments of the same act are understood as uniform provided only that the judged is a community member.

This judgmental process with its two different sorts of judges provides support for the differentiated conformity called for by the differences among the expectations in different statuses and their roles. It also supports a broad general conformity to a relatively few universal expectations applying to all community members. "Sanctioners" are surely present in every society, and every human is one, but "arbiters" may not be ubiquitous. It is likely that universally applicable standards work in a similar way even if they are not seen as embodied in the judgments of high-prestige individuals.

Judging and Sharing: Conformity Independent of Consensus

From the point of view of cultural dynamics, it is important to note that one need not share the understanding that is the basis for judgment to be judged by it or agree with it to be affected by it. Since evaluations are relative to the statuses of both the actor and the assessor, the shared understandings brought to bear are not from a widely shared and generally applicable inventory of evaluational rules. Instead, the expectations in the roles that unite the statuses of the judge and the judged provide the basis for the assessment.

Since every status has as many roles as its occupants have relations with different categories of others, the expectations of a given status may be quite different according to the status of the individual with whom there is a relationship. The husband has expectations in the husband-wife's brother role that are quite different from those in the husband-crony role, and the judgments will differ accordingly.

The diversity of standards presents a potential difficulty for the evaluated and, since everyone is constantly evaluated, for all the members of the group. To put it melodramatically, the differentiated conformity called for by the fact that everyone has many statuses and an even larger number of roles offers the potential for painful individual conflict about the existence of potentially contradictory evaluations. Social difficulties also threaten in that individuals may be reluctant to participate in some kinds of relationships due to uncertainty about which standards will be applied to them.

If the numerous pressures for conformity to different, sometimes contradictory, expectations all work simultaneously, the individual's situation could become impossible. The fact that people not participating in a relationship, as well as those directly involved, assess behavior in it makes this even more difficult. Virtually everything a person does could be approved by some with whom there are important relationships and, simultaneously, disapproved by others with whom relations are equally vital.

To complete the melodrama, the social and individual dangers inherent in there being a variety of standards by which conflicting judgments may be made for the same behavior are not greater than the threat posed by a single standard being uniformly followed by all. If none of the differing pressures for conformity to different, sometimes contradictory expectations are effective in producing a differentiated conformity, there is the prospect of status distinctions failing and, with them, the performance based in different expectations that is indispensable to social life and individual satisfaction.^[13]

Since Swahili individuals continue to function and the community perdures, there are unquestionably ameliorating factors at work which allow differential conformity and, at the same time, reduce the incidence of contradictory pressures on people. One of these is the limitation on the array of statuses likely to be occupied by those who know about a given behavior.

This reduction in contradictory expectations and judgments, no doubt, comes about in a number of ways in different societies, but in the Swahili community, it is in some part the result of the quite sharp separations in the personnel and the location of activities involved in different domains. This serves to limit the range of statuses occupied by both judges and the judged in each domain and, thereby, limits the variety of expectations bearing on particular behaviors.

The business of earning money is mainly separated from family relations, juniors are separated from seniors, and, for most of the day and the great majority of activities, men are separated from women. This, of course, reduces the likelihood that the same behavior will be subject to judgment according to different standards by reducing the range of statuses whose occupants are aware of the activities.

An even more pervasive source of reduction in the difficulties arising from conflicting judgments derives not from social structure but directly from culture, namely, the widely shared understandings that the restriction of information about one's activities and those of people with whom one is closely connected is, pari pasu, always to be preferred to the broad dissemination of information. Insofar as this understanding guides behavior, it promotes uniformity in judgments by reducing the number of judges and reduces conflicting judgments by allowing the judged to choose who will have the information needed.

Limiting Cultural Diversity with Cultural Models

A closely related problem to that of multiple evaluators is that of multiple standards. Anything that encourages those involved in a relationship or interaction to apply similar understandings to participation and evaluation will lessen diversity of judgment, if only "locally," that is, to the individuals or situation at hand. The processes that promote "local" homogenization include double contingency and, a key source of its effectiveness, the use of emotionally loaded mnemonics such as those found in the use of relationship terms. These two processes are more focused and current than socialization and enculturation on which they depend, to some extent, for their effectiveness. More immediately and directly important to the effectiveness of double contingency and the display of cultural models is that those judged care about the relationship in which these processes occur.

Judgments are likelier to be effective when they are made by partners in multiplex relationships. Since these relationships are broad in scope (but by no means all-inclusive) by definition, their effectiveness in strengthening evaluations applies rather broadly and, since membership in them overlaps, making broad networks of relations, is conducive to some uniformity within the community.

The uniformity promoted in this way is not a uniform commitment to a single set of understandings. Mothers do not expect sons to perform in the mother-son role as they expect them to perform in the sister-brother role, much less the student-teacher role, but the mother's assessment of all of these roles is often influential. The woman who is mother, that is, may introduce or maintain some uniformity in effective understandings by her assessments of those who are in family roles with her and also in other roles that concern her. The "uniformity" they promote is, of course, differentiated according to the statuses and roles in which she judges her family members.

Note that since the son who is evaluated by his mother in the employer-employee relationship is reacting in some part to his mother's evaluation and by his actions is affecting the employer, the employer's behavior is affected by his worker's mother's standards whether the employer shares them or not. This sort of process does not guarantee uniformity throughout the group, but it is surely a force in that direction.

The pressures for uniformity based in evaluations are always through roles, not statuses as wholes. A mother's behavior is evaluated differently by sons, by daughters, by teachers, and by other mothers, with uniformity promoted less as regards the mother status as a whole than as regards its component roles. This is probably a considerable part of the basis for the finding here and, possibly, in Holland's study (1987a ) that status occupants do not share

more of the understandings concerned with their statuses with fellow status members than they do with nonmembers.

Mothers evaluate one another mainly in the mother-mother role and have only a limited basis for assessing, or directly influencing, one another in the other roles of the mother status since they are rarely present when these roles are played. Nuclear family members, however, not only interact with the mother in their particular roles but also influence her in her other roles involving fellow family members. Thus, the "local" homogenization in the family leading to more sharing among family members as concerns the mother status than among people who are all mothers but operate in different families is quite in accord with the processes hypothesized here.

The cultural conformity that is essential for social life and for individual adaptation is a highly differentiated one. Actors must follow at least some of the understandings that constitute the expectations in the various statuses they occupy in response to the requirements of their current situation and its relationships. Their conformity is not to a single set of expectations but, rather, must be attuned to the behavior of others in the particular situation and must shift as statuses and their roles change from situation to situation and relationship to relationship.

The sort of judgments sanctioners make provide impetus for just such conformity through the judgments' constant dual-status dependence—on that of the judged and on that of the judge. It is crucial to this process that people do not respond equally to all judgments by all judges. Because of the greater commitment to them, the judgments by those with whom the judged share multiplex relationships are usually more significant across a broader range of domains than judgments by those in simplex relationships. Participation in multiplex relationships, in other words, helps organize the various understandings available for guidance in a particular situation. Those likelier to guide behavior so that a positive evaluation in multiplex relationships results have an attraction in the Swahili community that understandings in simplex relationships do not have.^[14]

Multiplex Relationships, Conformity, and General Expectations

Individuals differ as to which of their multiplex relationships are the source of more influential judgments and which less, but it is always from partners in some such relationships that people draw the judgments that affect their behavior most. In the Swahili community, there are shared understandings that encourage this focusing more on certain multiplex relationships than on others.

A Swahili cultural model, expressed in proverbs as well as relationship terms, emphasizes the importance of permanent relationships (mainly of kinship) as opposed to more transitory ones. To the extent that attachment is associated with sensitivity to judgment, this promotes the effectiveness of judgments made by those in such relationships.

The judgments in all relationships, whether simplex or multiplex, always involve the expectations in the statuses of those in the relationships. Expectations, however, are not always different in substance from the other types of understandings that make up statuses. It has been noted several times that the three types of understandings that make up statuses are analytically distinguishable but not necessarily different in fact.

Often, identifying understandings overlap with expectations so that what people do influences how they are categorized. Both kinds of understandings work together to indicate the salience of category membership in different situations and, when several are appropriate at once, to indicate its proper strength in combinations with other statuses. It is what people do that can be seen and assessed, however, and therefore it is expectations that are the understandings most directly tied to evaluation.

But expectations are never divorced from identifiers and salience understandings. Metaphorically, the expectations are the tool's edge, but the haft is composed of identifiers and salience understandings. The three distinguishable components of statuses, however, are subject to joint influence. By having direct influence on expectations, that is, by calling for some rather than others, the social pressures in judgments, double contingency, and cultural models also affect the other parts of statuses through their indications of which are appropriate.

Some expectations are quite specific. A bus passenger is understood to give the conductor money, and in return the conductor is understood to allow the passenger to ride. The ability of the parties to such a simplex relationship to determine whether or not their expectations have been met is to a considerable extent dependent on little more than direct observation and very limited interpretation.

This is not so as concerns many or most of the expectations in multiplex relationships which are of the sort called "general expectations" here. These are broad in scope and only loosely identified with specific behaviors. Although multiplex relationships include specific expectations, the sort that are most vital to them are not met through specific behaviors but rather by a general interpretation of a whole range of actions.

The broad behavioral scope of these general expectations gives them a kind of flexibility in influencing behavior that is particularly important to culture's effectiveness. Because of this, as will be seen in a moment, the importance of participants in multiplex relationships meeting their partners' expectations

goes beyond the relationships themselves to affect how the community as a whole operates.

In an important sense, general expectations differ from specific expectations with respect to the amount of interpretation of behavior they call for. When Swahili say that in relations with those closest to you, your satisfaction or lack of it is due less to what your partners do than it is their nia (purpose or intentions), they are referring to just what is meant by "general expectations."

General Expectations, Cultural Sharing, and the Scope of Multiplex Relationships

Whether or not A's behavior meets B's general expectations is a matter involving A's interpretation of events quite as much as the events themselves. In the grocery clerk-customer relationship, the clerk does not usually need to make abstruse interpretations to decide whether or not the customer has handed over the payment asked for. In the Swahili husband-wife relationship, however, it is far more difficult to establish whether, for example, the husband has "shown love" to the wife even though doing so is frequently mentioned as an expectation in this relationship. Because of their dependence on general expectations, multiplex relationships have two notable qualities.

First, the relationships can function despite participants not sharing many specific understandings, including some of those concerned with the relationship itself. Some sort of balance of satisfied as opposed to violated expectations is probably essential to the maintenance of social relations. This balance, however, is more likely a psychological than a quantitative one. Many failures to keep the bathroom floor dry can be outweighed by a single manifestation of what is understood as concern and love. To the extent this is true, family relationships, and multiplex relationships generally, function in some part, at least, through their participants' general, empirically broad, expectations of one another which lead them to interpret quite a wide range of behavior as in accord with their most heavily weighted expectations.^[15]

Second, the relationships have an indefinitely broad scope, since just what each participant can, should, and might do for the other is not, by their nature, specified in the general expectations. In the specific expectations most characteristic of simplex relationships, the limits of commitment and responsibility are usually rather clearly understood as part of sharing the understandings that constitute those expectations.

In multiplex relationships, there are no such sources of limitation. A person whose nia toward you is "good" is one who can be called on for a very broad range of things, and if your nia is similar, you can be similarly called

on. Each of you may understand issues and circumstances that cannot or should not be dealt with within the relationship, and some of these limiting understandings may be shared, but the boundaries are very wide.^[16]

Sharing Identifying Understandings

What seems most consistently and broadly shared by participants in multiplex relationships are the understandings that identify participants to one another. Such sharing of identifiers is characteristic of all relationships, especially multiplex relationships. The sharing of the identifying understandings may not be quite as complete as community members sometimes indicate it is, since people sometimes do identify as "friends" those who privately identify themselves as only acquaintances or who even understand themselves to be enemies. But the sharing is usually general, so that almost everyone is quite sure who their partners in multiplex relationships^[17] are. Given the importance of general expectations in multiplex relationships, the sharing of identifiers (which sometimes include these expectations) for these relationships entails a sharing of such expectations as well.

As we will now see, these expectations play a key role in cultural dynamics beyond their direct service in multiplex relationships. Working together with the ability to identify those in multiplex relationships and the indefinite boundaries of expectations in those relationships, these inclusive expectations are an essential basis for transmitting, so to speak, the effects of understandings to those who do not share—or even know about—them.

It may well be that there are a considerable variety of processes by which this takes place. Two of them were recognized among the Swahili. In one, understandings are "imported" from one relationship to another. In the other, limitations in similar expectations in all but one in a set of connected multiplex relationships give a highly distinctive character to the one relationship, husband-wife, that does not have those limitations.

Importation in Swahili Medical Treatment: It isn't What You Know

An instance of the transmitting of understandings and their effects through what might be called importation was seen in the examination of Swahili understandings about the body and illness and how people choose among types of medical care (see chap. 9). This choice was not commonly made on the basis of intrinsically organized schemata involving selecting a medical practitioner on the basis of understandings about the practitioner's views of illness being in accord with the patient's. Most patients were found to have few understandings of any kind about either the treatment of illness or differences among types of practitioners.

Rather, the choices were made according to the advice of people, usually kin or neighbors, of whom the patients had general expectations to the effect that the adviser would help the patient and could be trusted. These expectations made the advice worthy of following. When it was followed, it led to the patient being affected, in the choice made and the treatment received, by understandings about medical care held by the adviser or often, at a further remove, only by the adviser's adviser or her adviser.

What is particularly striking about the advice is that it is usually accepted. The patient goes, or allows himself or herself to be taken, to the kind of medical practitioner the adviser recommends. When this happens, the nonsharers are affected by the medical understandings that are, according to the definition of "culture" used here, part of the community's culture despite the patient's ignorance of them. If two or more members of the community who maintain some sort of an active relationship with one another share an understanding, it is part of the culture of the group. The fact that there are individuals who do not share the element does not bar it from being part of the culture. Swahili culture would consist of nothing save understandings such as that one must not go naked in public if only universally shared items were included.

Nor is this only an arid definitional matter. If culture's ability to promote individual life and social relationships is to be explained, and if a substantial proportion of culture's contents is less than universally shared, it is obviously essential to examine closely whether understandings shared by only a few affect others who do not share them. When this does happen, and it is surely quite a common phenomenon, the processes whereby it happens call for description and analysis. The Swahili medical care study is an attempt to do just that for one domain.

The understandings that lie behind medical care affect those who receive that care just as the understandings behind the generation of electricity affect people who read by electric light even if they are innocent of understandings about coils and magnetic fields. Medical care, unlike electricity in modern cities, is not "just there," that is, so much a part of life that it requires effort to avoid.

Everyone I talked to knew that Mombasa has a variety of different medical practitioners, and everyone was quite clear about knowing that different people consulted various of them when ill. But understanding that there are practitioners who can be consulted about your illness does almost nothing to lessen the effects of that illness. Only choosing one and accepting the treatment offered may do that. Having no clear understandings about how illness arises and none about how different medical practitioners deal with illness, the basis for this choice is not obvious. It is constantly made, of course, but the basis for it cannot, for most, be understandings concerning the body and its treatment since they do not have these.

For most people, the choice is made on the basis of advice from partners

in multiplex relationships. Their advice is taken seriously enough to be the immediate basis for action as a consequence of the general expectations in those relationships. These relationships "transmit medical understandings" in the sense that they expose patients to activity based on them (what practitioners do when the patients come to them) whether the patients themselves have any familiarity with those understandings or not.

Repeatedly, informants said that they sought the treatment of a practitioner recommended to them not because of what they directly knew about the practitioner or the theory of body functioning and illness he or she followed but because of their relationship with the adviser who recommended the practitioner. When asked why they followed the advice given them, the patients usually said that the adviser "cared about" them (and so would give "good advice") or that the patients "trusted" the adviser or, less often, that the adviser "knew about" the illness in question or medical matters generally. Thus, for the overwhelming majority of the community who are ignorant of medical understandings, these understandings affect their lives through the agency of quite different understandings; namely, the expectations in their relations with others when these others were kin, friends, or neighbors.

There is little new or enlightening in observing that medical knowledge is limited in its distribution and that medical practitioners share understandings (with other practitioners and, to a limited extent, with a relatively few "dedicated amateurs") that most group members do not share. What is worth noting is that the medical understandings affect people because those people have expectations of others, including others as free of medical understandings as they themselves are, which lead them to accept the treatment of the medical experts.

Those in the relationships who give advice are only sometimes more knowledgeable about body functioning and medical care than are the patients they advise. When they are not, they get information from a third person about what should be done. This third person may share a substantial body of medical understandings with practitioners, but it appears that the "chains" of social relationships between the medical practitioner who ultimately treats the patient and the patient can sometimes be rather long. In each link of this chain, the connection between adviser and advised is more often through shared general expectations than through shared understandings about the body and illness.

Cultural Organization and General Expectations

The schemata that commonly serve Swahili as bases for getting medical treatment have components that are extrinsically connected to one another. People use general expectations about advisers as a connection between the

understandings that they are ill and need treatment and those that hold treatment is available and can be obtained from a specific source. Since "cultural organization" refers to the relations among shared understandings, in the extrinsically based schema, the general expectations take over the part played by the intrinsic relations (of the sort used by chess players in deciding on moves) in organizing the elements in that schema.

Nor is organization through statuses involved only in extrinsically based schemata. Some organization derives directly from understandings about understandings (Swahili proverbs sometimes express these) wherein doing or being one thing is said to be better or more important than being or doing something else. A substantial part of establishing relations among understandings, however, is due to the functioning of statuses and their constituent roles in determining what is appropriate, what is more and less important, and, as already seen, what is likelier to be approved by whom. This depends mainly on the significance of evaluation in the relationship, so understandings whose guidance leads to behavior that is positively evaluated by partners in multiplex relationships thereby have a quality affecting their desirability and choice other understandings do not have.

A rather extreme example of the choice among understandings immediately guiding behavior can be seen in an aspect of the relationship between Swahili mothers and daughters. A woman cooking in her kitchen, for example, can sometimes be heard to direct strong insults to her own daughters.^[18] Women tell me that they would not use such insults on their peers or even the daughters of their peers but that with their own daughters such behavior is acceptable and they are not despised by anyone, including the daughter, for using them (Swartz 1990b ). The mother-daughter role involves, inter alia, the understandings that the mother status occupant is dominant, that mother's abuse of a daughter "means nothing," and that daughters accept their mothers' behavior, even if it is harsh, as "instruction."

What goes together with what and what does not, that is, one of the relations among guiding understandings, is established according to statuses and the expectations they have in various roles. Without organization to indicate which understandings can and cannot be used together, what comes first or last, and what is more salient than what, the elements of culture are useless in guiding behavior. The main base for this organization is the expectations in the statuses, and this is as true for general expectations as for specific ones. However, general expectations have effects that specific expectations do not.

"Patterns" or Common Element Organizations

An example of the organizing effects of general expectations beyond what is involved in schemata can be seen as concerns the type of cultural organiza-

tion often referred to as "patterns." This organization is one in which a common element is present in the complexes of understandings associated with a number of different domains. The common understanding provides an element of similarity in the different domains in the way exemplified by Benedict (1934) in her distinction between Apollonian and Dionysian cultures. This similarity may be one of style, substance, or both, but its presence presumably serves to promote cultural conformity through harnessing habituation and the sense of rightness that comes with familiarity.

As shown in chapter 9, at least one instance of just this sort of cultural organization is found in Swahili culture. Understandings concerned with the usefulness or desirability of "balance" are a common element in several different domains. Thus, in the domain of body functioning, Galenic medicine is based on understandings about the close relationship between elemental balance and health; in social relationships, the balance involved in maintaining status differences and meeting expectations is valued; and in character, the highly important "balance" calls for adherence to the differential proprieties in different situations.

A "Pattern" and How It Can Be Effective Given Limited Sharing

The understanding that balance is desirable provides an organization based on the presence of a common element in the diverse sets of understandings that severally concern their different domains. In all of them, the desirability and benefit of including the "proper" proportions of different things is noted and also the cost of not doing so. The simple existence of the same sort of understanding in quite different areas of life may make it likelier that the understanding will be influential in each of them. To the extent that the understanding is a morally charged one, as balance in social relations and character is, it becomes even more effective and able to influence or guide behavior in a variety of areas.

The difficulty with these common element organizations, "patterns," having a substantial effect on general cultural conformity is that their influence depends on there being a good deal of sharing of the understandings involved. On the face of it, the balance notion in body functioning can only be made compelling for those who share it and also share the balance understandings concerning relationships and character. Since most people know nothing of Galenic medicine, including its emphasis on balance, the acceptance of this type of medicine would, it might seem, hardly be increased by the fact that those same people are committed to balance in social relations and in character.

There is a substantial basis, however, for the pattern being effective in promoting conformity despite the widespread lack of sharing of its component

understandings. This basis is in the fact that somewhere in the chain of advisers who serve to supply the "missing link" between the understanding that one is ill and the understanding that there is help in dealing with illness, there is usually someone who knows about balance in medicine as well as in social relations and personal character. By inclining the experts or serious amateurs who know about Galenic balance to recommend Galenic practitioners rather than other kinds, the organization affects some of those who do not share the balance understandings in some domains.

It may be of some interest to suggest that not only do understandings affect those who do not share them but organizations can affect those who are innocent of their components. At least one process by which the latter occurs is the same as the one that brings about the former. This is, of course, that people accept advice from each other, with an important source of the acceptance being the general expectations in the multiplex relationships within which it is given.

Transmission by Simplex Relationships and Its Limits

Clearly, advice is also given and accepted in simplex relationships. The practitioner-patient relationship is an example of this. Here acceptance, when it occurs, is usually on the basis of the specific expectation that the practitioner knows about illness and can make useful recommendations. In order for the patient to continue to accept the practitioner's advice, however, the patient must see acceptable results, that is, the specific expectations in the relationship call for particular things to happen within a relatively limited time span.

The general expectations vital to multiplex relationships do not call for highly specific returns, and the time spans involved in exchanging whatever is exchanged may be quite long. A mother who sends a son or daughter to a practitioner whose treatment does not help as soon as the patient thinks it should is, nevertheless, quite likely to be asked for advice again. The patient is likelier to abandon treatment from the practitioner than advice from the parent. The "pay-off" in multiplex relationships is quite different from that in simplex relationships, making the former far more resilient in most cases.

Moreover, the general expectations in multiplex relationships provide a far broader scope for the relationships than is found in simplex relationships. Parents or neighbors are consulted because they are expected to be committed to one's interests as much or more than because of the breadth of their command of understandings.

Simplex relationships do serve to transmit unshared understandings in the way seen for Swahili medical care, but their "reach" is limited and their "cost" is high. Simplex relationships, by definition, begin and end in a single domain, and their specific expectations are all within that domain. One does

not have a relationship with a medical practitioner, if the only role is practitioner-patient, outside the domain of medicine. This relationship can be part of a broad transmission of the effects of understandings to those who do not share them by those who take themselves to have benefited but only if the links in the chain of transmission that follow it are the general expectations in multiplex relationships.

Generally, simplex relationships cannot serve to make cultural organizations effective in the way multiplex relationships can and do. A practitioner may study and practice Galenic medicine because of understandings in other domains which make the balance understandings in that scheme attractive to him, but his ability to bring people to accept the kind of medicine he practices depends more on the actual results they believe he achieves than on his advocacy of the scheme.

In a multiplex relationship, the partner's enthusiasm, which may be based in her being struck by the inherent "rightness" of the balance, or some other, understanding does matter in that her advice is followed not so much because of the qualifications she has for giving the advice but more because of her understood commitment. If a mother wants a child to consult an herbal doctor, a central reason for following the advice is that she gave it. General expectations do not call for specific results, nor are they based in understood command of specific understandings. Their broad and vital part in cultural dynamics is mainly a consequence of these two facts.

The effects of general expectations in the operation of patterns and in the transmission of the results of unshared cultural elements suggest that these broad expectations have effects beyond the relationships in which they occur. In fact, through their central part in multiplex relationships and the broad consequences of these relationships for the social life of the community, they play an important part in the operation of social structure as a whole. These broad expectations not only bind together spouses, parents and children, neighbors, and many others who "mean" a great deal to one another but also play a central part in making the whole culture effective. Their contribution to the community's whole social structure is particularly critical.

Social Structure as an Independent Influence on Behavior

"Social structure" here refers to the statuses in a community's culture and their connections, direct and indirect, by mutually involving references in their expectations and salience understandings. Since the statuses making up social structure are each composed of shared understandings and since these statuses are joined by the mutual references of one of these sorts of component understandings (i.e., expectations), it is a culturally constituted system. As

will be seen, however, despite its being composed wholly, but not quite solely, of culture, social structure has effects that are independent of culture.

A social structural perspective focuses attention on how the statuses that guide relationship affect one another and, therefore, the nature of the guidance available within the community seen as a system, sensu stricto, of relationships. A social structural point of view directs attention to the connections among understandings rather than directly to the understandings themselves.

As an example of the results of a social structural perspective, consider recent changes in the statuses of Swahili women. These changes in the expectations in the statuses daughter, sister, and wife as well as woman lessen the difficulty women have in being in statuses, especially "employee," whose expectations call for spending a good deal of time outside the home. The expectations in these "outside" statuses affect the statuses women (or, perhaps, "respectable women"), daughters, sisters, wives, mothers, and neighbors and the wide variety of roles involving these statuses. These changes in women's statuses and roles, in turn, affect the statuses connected with them and their component roles, thus altering the sets of understandings that affect a wide range of the community's relationships.

The Social Structural Importance of Multiplex Relationships and Their General Expectations

Every status affects all the others directly or indirectly connected to it in role relationships through the presence of the mutual references in the roles' expectations. This is true whether the relationships are simplex or multiplex. The examination of how medical care is obtained illustrated how multiplex relationships such as mother-child affect simplex relationships such as practitioner-patient by leading people to participate in them, but it should not be thought that all influence flows from multiplex to simplex relationships. The influence goes the other way as well. So, for example, the expectations in the statuses in such simplex relations as employer-employee doubtless affect expectations in the statuses of a variety of multiplex relations within the family and neighborhood.

Given this caveat, it is nevertheless true that the statuses in multiplex relationships are particularly important to a community's social structure. Because their general expectations are flexible and inclusive and because, by definition, the relationships involve a number of different domains, it is multiplex relationships that most often bring the effects or products of a wide range of the group's culture to community members who may not share the understandings they rest on. No one in any group has even indirect access to

all the cultural elements available in all of the statuses in the community's social structure, but such access is not necessary to be affected by them. Relations with others who do have access to these understandings, whether direct or indirect, may serve to transmit their effects provided the relations have expectations promoting such transmission.

The hypothesis that has been advanced here is that it is multiplex relations with their general, rather than specific, expectations which do this. The transmission may be through the sort of advising that was seen in Swahili medical care, or it may be through one participant in a multiplex relationship bringing the effects of his or her participation in a quite different relationship to bear on the fellow participant.

The latter can involve a sort of chain reaction effect. For purposes of illustration, imagine that A is in a multiplex relationship with B, B is in any relationship with C, and C in any with D. D imposes expectations on C deriving from relationships D has about which C need have no understanding. The effect of this on C leads him to impose expectations on B who, in turn, knows nothing of their provenance but is affected by them and imposes them on A. A knows nothing of what led B to impose the expectations he did, but A is affected by them nevertheless. Since the A-B relationship is multiplex, the limits to the kind and extent of the influence that B brings into the relationship as a consequence of his relations with C are quite wide. Thus, the imposition of expectations by D affects the relationship between A and B even though neither of them need be aware of the relationship between C and D for this to happen. (See Swartz and Jordan 1976:86–98, for a fuller discussion of this process and an example.)

The point is that through advising and through the sort of chain reaction of expectations, social relationships spread the effects of components of the group's culture so that they affect individuals even if they do not share those components. Simplex relationships can and do operate in these processes as well as multiplex relationships, but the former, unlike the latter, are probably limited to the domains in which they mainly operate, so that, for example, economic relations can only transmit the effects of mainly economic understandings.

This limitation is to be expected since it is the expectations in social relationships that serve to transmit cultural influence to those who may not share the particular understandings that affect them as concerns the matter at issue. Given the central part played by expectations, it must be remembered that it is the identifying and salience understandings that make the expectations effective. These latter status components are the culturally constituted means for promoting the culture's groupwide effectiveness regardless of less than complete cultural sharing.

Multiplex relationships and their general expectations are a particularly important part of this because of their scope and their relative immunity to

the weakening of relations that can result from the failures to meet specific expectations that are inevitable and generally more harmful to simplex relationships. There are failures to meet specific expectations in multiplex relationships, of course, but the broader and more diffuse foundation of these relationships often diminishes the harm these do to the relationships and makes them a more enduring part of the community's social structure.

Social Structure as an Independent Influence on Behavior

The interconnected set of statuses that forms the group's social structure is crucial to the functioning of the community's culture through the operation of the components of the statuses, but that same social structure also serves to influence behavior independently of the culture that is its base and most of its substance. I say "most" of its substance because community members need not have understandings about the connections among statuses for those structural connections to affect behavior. A father can be affected in his status as employee by his boss's expectations, some of which stem, for example, from the boss's spouse relationship. The father may impose expectations on his children as a consequence of the expectations imposed on him as an employee even though neither the father nor the children know the boss is married, much less how his relations with his wife affect what he expects from his employees.

The connections among relationships derived from the mutual reference and interdependence of expectations have an influence on behavior that stems from, but is independent of, shared understandings themselves. Parsons is the locus classicus of the view that social structure has an influence on behavior that is independent from culture (1964 [1951]:6, 17–21 passim), and this view is important to the scheme developed here.

In an earlier work, I used data from Pitt-Rivers's (1961) study of an Andalusian village to illustrate one of the processes by which this operates. The Spanish villagers were shown to respond to expectations in one of their social relationships despite having no understandings about the usefulness or desirability of the expected behavior itself. Some of the expected behavior, in fact, was contrary to shared understandings about how people should behave (Swartz and Jordan 1976:93–98).

The expected behavior was forthcoming solely because of the importance those involved attached to maintaining the relationships in which the behavior was expected. This importance, in turn, was a consequence not of the value or utility of the relationship in itself but of its connections to other relationships (ibid., 89–92). In a sense, the influence of the one relationship can be, and frequently is, derived not from understandings about its intrinsic merit

or worth but from its connection with other relationships that are understood as valuable in themselves.

Culture and "Cultural Products"

In the Spanish case, influence comes not from people's shared understandings about the behavior at issue but from their commitment to the social relations themselves. Thus, the social structure affects behavior independently from culture. Nor is this the only way in which that occurs. To appreciate this source of influence in its proper context, it is useful to look at social structure as a product of culture much as tools or buildings are products of culture.^[19]

People are affected by the tools they use, the buildings they occupy, the food they eat, and the clothes they wear without respect to their sharing the understandings that produced the tools, buildings, and clothing. People are usually led to make use of cultural products such as tools they do not know how to make or repair through their participation in social relationships. Frequently, they actively seek ends they clearly recognize by means they are thoroughly unfamiliar with, as most Swahili patients do in striving to become well. This sort of intentional striving, however, is by no means an essential part of all the processes whereby cultural elements affect those who do not share them.

Wives' Power De Nihilo: Social Structure's Effects Independent from Culture

Chapter 10 showed that almost every understanding concerning Swahili spouse relations holds that wives are subordinate to their husbands and must accept their decisions. Almost everyone I talked to regardless of sex said that women are subject to the authority of men because it is part of God's plan as revealed in the Koran. In fact, those men who do exercise this divinely commanded authority over their wives in ways that deny them the expensive ceremonies and finery most want appear to be entirely successful in doing so. These men are not negatively evaluated by men or women.

And yet many women actually do spend very considerable sums of money in ways their husbands say they disapprove. The women's expenditures, the men say, are the single most important factor in the Swahili being less prosperous than the Arabs and Indians among whom they live in Old Town.

Without repeating the argument and evidence, it is enough to say that the central issue is why men choose not to exercise their very real power when dealing with their wives' wishes to use money in ways the men do not approve

and from which, in fact, the men derive little direct benefit. The answer to this, the evidence indicates, is to be found in social relations—not just those between husbands and wives but the various relationships involving men and involving women.

Examination of men's relationships shows that intimacy and emotional support are not readily available to them in any relationship other than that with the wife. Expectations in the broad "man" status make emotional warmth and support difficult to give and to receive in almost every relationship involving men save that between spouses. For women, however, many relationships have expectations that encourage warmth and support, so that for them the spouse relationship is only one source among many.

The men's emotional dependence on the spouse relationship together with the women's relative independence gives the wife a source of power she need not admit, or even recognize, in order to use. Her unhappiness resulting from not getting what she wants is transformed, more or less without her willing it, into her withdrawing emotional support from the husband who frustrated her wishes. For most husbands, this leads to serious efforts to avoid wifely unhappiness. The women need not admit, even to themselves, that their husbands are emotionally dependent on them. All they have to do is act as they feel—and doing this is an expectation of the status "woman"—to increase the likelihood that their husbands will give them what they want. For present purposes, what is important about this source of wives' power is that it demonstrates a means whereby a community's social structure affects community members in ways independent of, in fact, contrary to, the directly concerned elements of culture.

In considering Swahili medical care, it was seen that understandings can and do affect community members, including those who do not share them. Here we see behavior affected in ways that do not depend on anyone sharing understandings that produce the effect seen and without anyone needing to acknowledge, or know about, the basis for what is happening. There is no understanding, certainly not one people admit, holding women have the ability to use their husbands' money as the women wish regardless of their husbands' views. Yet they do.

Social Structure's Effect: Blocking and Channeling

Men say they believe and are aware that sanctioners act as though they believe that it is improper, impolitic, or shameful to be emotionally expressive and to accept emotional support. In their relations with all of those with whom the relationship is in any way open to observation, including those with mother, sister, and other female kin, such behavior is negatively evalu-

ated by those involved, including the men themselves. This is not true of women's relationships, most of which include expectations of emotional expressiveness.

The unique presence, for men, of emotionality in the spouse relationship needs to be seen in light of the fact that this relationship is carried on entirely in the privacy of the home with no one but family members ever seeing spouses together. It is not so much that the expectations in the spouse relationship openly and explicitly admit male dependence as it is that the privacy in which much of this relationship takes place makes its expression possible. Further, the positive evaluation of "love" between spouses encourages giving and accepting support and warmth. This support for husbands is not specifically called for and need not be recognized by either husbands or wives, but it is available in the spouse relationship, in large part as a consequence of love and privacy. A good deal of evidence has been cited in support of the hypothesis that men do, in fact, derive emotional support from their relationship with their wives even if they never talk about it or admit its presence.

As seen, from a cultural point of view, the shared understandings concerning the spouse relationship accord men complete control, so that the wives' power to get their husbands to give them money is de nihilo, or so it seems. In fact, the social structure of the Swahili community—or, more exactly, the unique character of the spouse relationship within the structure—is a key resource for wives in dealing with their husbands.

This is the key fact here, and it is the consequence of the differences in the whole set of statuses and roles focusing on husbands as contrasted with the set focusing on wives. This difference is the main basis for wives' power, which is to say that their power is directly attributable to the social structure, not, as we saw, the elements of culture that, in fact, give all power to the husband.

The statuses that make up the social structure are all culture and nothing but culture. The structure itself, however, is not just the sum of its cultural parts. It also involves the relations among these parts, including the effects relations have on one another because of the connections between them. Some of these connections are not the result of understandings in any community member's mind but of the effects relationships have on one another through the influence of their expectations "spilling over" into one another. Some such mutual effect of relations on one another is understood by participants, who know, for example, some or all of the ways an employer-employee relationship can affect a parent-child relationship. But the effect of one relationship on another can come about without such understandings through the meeting of the expectations in one or more relationships affecting the expectations in some other relationship.

This is just what is seen in the spouse relationship. The expectations restricting emotionality in all of men's other multiplex relationships give the

spouse relationship an importance for them it would not otherwise have and that, because of the different expectations in their multiplex relationships, it does not have for wives. Since social structure is a cultural product, culture's part is hardly a distant one. But its effects come from the way it channels behavior through expectations that encourage it here and block it there. There is no understanding to the effect that wives should spend substantial sums on weddings and bracelets even if husbands object. Quite the opposite. What there is, is a complex series of statuses and their roles that give wives alternatives for emotional gratification but give none to husbands.

Culture contains statuses that guide relationships and indicate what these relationships contain and do not contain. These have consequences for other relationships and the occupants of the statuses in those relationships. Wives' power is one of these consequences.

Hakuna Refu Lisilo Ncha : Nothing is So Long that it has No End

There can be no question that the culture of the ancient Swahili community works, even though it has been seen that most of its elements are shared only among various sized parts of the population rather than by all. In asking how this culture works, then, a basic question has been how cultural elements affect those who do not share them.

The main answer to this has been through statuses, with particular importance attributed to the statuses in multiplex relationships and the general expectations that are vital to them. These bundles of understandings are taken to bring culture's guidance to specific issues and situations and, at the same time, to serve as a central part of a variety of processes that make culture's components, or the result of their use, available to community members, including some who do not share them.

If the perspective derived from this study can be summarized in a single statement, it is that culture is a sort of ouroborus, the Greek snake with its tail in its mouth. Social relationships cannot operate without the culturally constituted statuses that guide them, but culture cannot operate without the social relationships that distribute its components and are the main force in organizing them.

Culture is not ineffable; it is a natural phenomenon, though a very complex one, that can be increasingly well understood as more useful concepts and theories are developed. This book has been an attempt to contribute to such concepts and theories and to test them on a body of ethnographic data collected from some of the world's most gracious people.