When individuals or families select spouses, some types of person are ruled out or disapproved of, while others are thought particularly appropriate, either because a preference exists in favour of such marriages, or because the persons concerned are already related in a particular, prescribed way. Marriage preferences are found in all societies, but prescriptions are confined to those exhibiting what Levi-Strauss called "elementary structures of kinship (Levi-Strauss 1969). Because many elementary structures involve "cross-cousin marriage, Homans and "Schneider (1955) assumed that Levi-Strauss was trying to explain the widespread preference for marriage among first cross-cousins, but "Needham (1962) argued that this was a misunderstanding. Elementary structures are not just unusually strong preferences for marrying specific close relatives, but global systems of classification, whereby prescriptive relationship terminologies divide up a person’s entire kinship universe into marriageable and non-marriageable categories. The marriageable category may include cross-cousins, but is not limited to them.
In this technical context, therefore, both terms have meanings differing significantly from their dictionary definitions (Needham 1973: 177). With regard to the three aspects of kinship — classification, rules, and behaviour — ‘preference’ seems to imply the exercise of choice at the level of individual behaviour, while ‘prescriptions’ ought to be requirements or rules. In fact, however, ethnographers normally take marriage preferences to be customs with some degree of jural force, whereas prescriptions are now usually seen as classificatory phenomena.
South Indian Maravars, for example, are able to classify all their caste fellows — potentially hundreds of thousands of people – according to gender, relative age, generation, and whether their relationship with ego is parallel or cross, as shown in Figure 5 (Good 1981). Only one such category of relative is marriageable. Men marry women classed as kolundiyal, junior cross-relatives of their own generation, while women marry attans, senior cross-relatives of their own generation. The prescribed categories include relatives on both the patrilateral and matrilateral sides, but Maravars also state a preference for men to marry their father’s sister’s daughter (FZD); i.e. for patrilateral first cross-cousin marriage. This asymmetric preference for one particular close relative therefore distinguishes more precisely among the broad categories of relatives classed as marriageable. However, whereas all Maravar marriages accord with the prescription, for reasons explained below, their stated asymmetric preference is scarcely reflected in their behaviour, since mother’s brother’s daughter (MBD) marriage is just as common as FZD marriage.
In the second edition of his topic Levi-Strauss tried to downplay the distinction between prescription and preference, which he first equated to that between model and reality (1969: xxxiii), then to the difference between mechanical and statistical models (1969: xxxv). In these mutually inconsistent statements, categories, rules, and behaviour are thoroughly muddled up. Even Needham partly conflated categorical and jural phenomena in his notion of prescriptive alliance (1973: 176), since alliance is a jural phenomenon rather than a matter of classification. However, the two basic types of alliance structure, symmetric and asymmetric exchange, are indeed often associated with prescriptive terminologies (Maybury-Lewis 1965: 224). The terminologies are usually differently structured in the two cases (though not necessarily, since categories and rules vary independently), because the former implies direct reciprocity and the latter a distinction between spouse-givers and spouse-takers.
The diagnostic test for distinguishing between preferences and prescriptions is to look at what happens when behaviour does not conform to them (Good 1981). When people marry, their choice of partner conforms to or conflicts with a set of preferential rules having customary or judicial force. (It is better not to speak of ‘obeying’ or ‘disobeying’, since rules serve more to criticize or justify behaviour after the event than to coerce people to behave in certain ways; cf. Bourdieu 1977.) Violating such rules generates social disapproval or even formal punishment, whose severity depends on how seriously the rule concerned is taken; certain marriages may simply be thought inappropriate to one’s status, while others may even be criminal offences. In either case there is no undoing what has taken place; as with any other ‘crime’, the breach of rules may be forgiven, but cannot be wished away. Prescription, by contrast, is characterized by the existence of an entire category of ‘marriageable relatives’, so if you marry a previously unrelated person, then of course — since s/he is, ipso facto, marriageable – your spouse and in-laws are thenceforth categorized as if s/he had belonged to the marriageable category all along. Crucially, the same thing happens when people marry relatives from a non-marriageable category; that is, their spouses — whom they have married and who are therefore again ipso facto marriageable — are reclassified, and so are all their immediate relatives. Every marriage in a prescriptive system thus conforms to that prescription by definition, at least in retrospect. The fact that such redefinitions are infrequent in practice does not mean that a prescription is merely an extremely strong rule. Rather, to violate a prescription is to violate a system of clas-sificatory meaning; it is ‘hard to think’, and for many people actually unthinkable.
In short, then, marriage preferences are jural rules or customs saying that certain types of relative must, should, or could marry, though the sanctions backing them up vary greatly in strength, and the punishments for ignoring them may be correspondingly strong or weak. Marriage prescriptions, on the other hand, arise where the entire kinship universe is divided up into marriageable and non-marriageable categories; and where the terminological relationships between all actual couples and their respective families are defined on the premise that — by definition — people who marry must stand in marriageable categories with respect to one another.
