A catalogue record for this book is available from the British Library.
Printed in the United States on acid-free paper.
I. Abjuring a Definition and Other Matters
II. Holding a Definition in Abeyance and A Case for a Definition
III. Monothetic Definitions
IV. More on Monothetic Definitions
V. Multi-factorial Approaches: Family Resemblance and Polythesis
VI. A Prototype Approach
VII. Ethnocentrism and Distanciation
PREFACE TO THE PAPERBACK EDITION
In this Preface I set forth the central argument of the book, shorn for the most part of complexities encountered in the text. In doing so, I call the reader's attention to a matter otherwise insufficiently emphasized: that the idea of "family resemblances" can be applied productively not only to the category religion and to denominated families of religions, but also to elements that scholars variously attribute to religion and to religions (e.g., theism, soul concepts, rituals and ritualizations, etc.). Finally, I address an omission in the text that was noted by a reviewer.
Religion is a Western folk category that contemporary Western scholars have appropriated. As I put it elsewhere,
Western scholars who study religion develop some understanding of what is meant by religion in their society long before they become scholars. This observation is so unremarkable, so obvious and seemingly trite, that I would be embarrassed to voice it were it not important. But it is important. Long before European scholars of religion become scholars of religion, they have fairly well developed ideas of what to look for in searching the world for religions. In large measure, indeed, their scholarly efforts to define or characterize religion are efforts to refine and deepen the folk category that they began to use as children, and to foreground what they deem most salient or important about religion. ( Saler 1997: 28 )
For well over a hundred years, Western academics have labored at the task of refining and deepening the folk category, writing definition after definition and explication after explication. They have variously identified the essence of religion as the supposed fact of, or a special sensitivity to, or a belief in, or commerce with, the supernatural, the super-human, the spiritual, the sacred, the transcendent, the numinous, the wholly other, and the partially other (that is, the anthropomorphized). Or they have sought to locate religion's center of gravity in something special about people in their solitude, or people in their effervescent sociability, or people asserting self, or people projecting, or people otherwise engaging in therapy, or people symboling, or people being reflexive, and so on and so forth.
While these efforts have sometimes contributed to our understandings of the longings, hopes, ideas, expectations, and prac
tices of our fellow human beings, the task of identifying the essence or universal core of religion has largely been a failure, considering the lack of consensus among scholars.
Some scholars, indeed, cite the lack of definitional consensus as supporting their contention that the term "religion" does not point to some universal phenomenon, to some one "thing," that can be recognized throughout the world. Religion, they maintain, is a term with discrete Euro-American associations and accreted and multiple meanings in the Western world, and it is therefore difficult, and perhaps inadvisable, to apply it cross-culturally. Some even explicitly recommend dropping the term. But arguments similar to those made for abandoning the word religion have also been made for such terms as kinship, marriage, property, law, belief, tribe, peasant, and so on. Indeed, virtually all of the analytical terms used by Euro-American social scientists have distinct Western associations and multiple meanings. Were we to drop all of them, social scientists could well be rendered speechless. (I hope that no one cheers at the prospect! In fairness to social scientists, we ought to admit that sometimes they do say things that are informative and interesting!) A less radical solution, I argue in this book, is to consider an alternative approach that neither seeks to do away with terms that have served as convenient tools for talking about things that interest us nor unduly attempts to restrict the range of those terms by creating narrow definitions.
SOURCES OF THE SOLUTION PROFFERED
The approach to conceptualizing religion recommended in this book draws on the philosopher Ludwig Wittgenstein's discussions of "family resemblances." Further, it supplements and complements the idea of family resemblances with selected insights derived from prototype theory in the contemporary cognitive sciences. While I hold that the idea of family resemblances offers great potential advantages for conceptualizing religion, I do not think that a family resemblance approach by itself is adequate for scholarly purposes. Something more is required. Prototype theory supplies that something more.
The traditional theory of categorization endorsed by Western scholars holds that the members of a group comprehended by a category are mutually and equally members because they all share
some one feature, or some conjunction of features, in common. Such features are called "distinguishing features" or "defining features." In the traditional theory, they define the category. And they serve as standards for determining whether or not some candidate should be admitted to the group comprehended by the category. According to the traditional theory, for instance, all things called "game" must share something in common -at the very least, one "distinguishing feature" -if they are properly labeled by the category term.
Wittgenstein disagrees. While he acknowledges that the instantiations of a category might share one or more features in common, it is not necessary for them to do so in order to justify being mutually labeled by the category term. It is sufficient that they are linked together by overlapping similarities of various sorts, just as the members of a human family may show overlapping similarities -"family resemblances" -in facial features, body build, hair color, and so forth, without all the members of the family sharing the very same features.
Indeed, in the case of some categories ("game," for instance), some instantiations (e.g., solitaire) will not share any discrete features with other instantiations (e.g., football). In such cases, however, there are typically still other instantiations of the category with which they overlap. Let us suppose that our category has three subsets of instances, A, B, and C. A and C share no features in common. Members of both, however, share some features with members of the subset B. Instances of B, therefore, serve to link instances of A and C, just as the central links in a chain connect the peripheral links on either end. Such linkages, the philosopher J.R. Bambrough suggests (see Chapter Five), provide an "objective" justification for the category and its instantiations.
I recommend that we think of "religion" in a similar way. The various instantiations of the category, popularly called "religions," need not all share some one feature, or some specific conjunction of features, in common.
We can deal with religion in terms of a pool of elements that we deem typical of religion, without supposing that any one element is necessary for the existence of a religion. These elements or "typicality features" (see the reference to Jackendoff in Chapter Seven) mutually pertain to our general model. But while all of them are formally predicated of "religion," not all of them will be found in all "religions." Thus, for instance, we tend to suppose that theism is typical of religion, and we predicate it of our general model. Yet while many religions encountered in the world do indeed accord central
importance to "gods," some (e.g., Canonical Theravada Buddhism) do not. It is not necessary for all complexes of phenomena that we call religions to share the very same elements, even though such elements are typical of what we usually mean by religion.
In the text, I make it clear that this logic is also likely to apply to the different instances of some denominated religion. Thus, for instance, the doctrine of the dual nature of Christ -the doctrine that Jesus Christ is both fully human and fully divine -specifically applies to the Christian family of confessions. But not all Christians accept it. Monophysite Christians, for instance, reject it, maintaining that Christ has only one nature, the divine, which incorporates or assumes human attributes. This helps make the point that "Christianity" is not some monolithic religion. Rather, it is a family of religions, linked together by family likenesses. Although we can and should predicate the doctrine of the dual nature of Christ to our scholarly model of "Christianity," we cannot legitimately insist that it is found in all Christianities.
In the text, I suggest that this logic also applies to the elements or typicality features that we predicate of our general model of "religion" (see, for instance, my treatment of E.B. Tylor's discussion of the category "spiritual beings" in Chapter Three). In retrospect, however, I now think that I fail to emphasize the point adequately. So let me make amends in this Preface.
If we accept the applicability of the idea of family resemblances to what some authors call "religion in general" and to denominated families of religions, we should also deal with the likelihood that that idea will apply at an even more fundamental level. That is, the elements or typicality features that we predicate of our scholarly model of religion ("religion in general") are themselves likely to be organized by family resemblances. Numbers of anthropologists either recognize this or verge on doing so. Take, for example, "sacrifice," which many scholars associate with religion. Maurice Bloch ( 1992: 25 ) writes that "I believe it is right to stress the great variety that exists among the various examples of 'sacrifice' as they have been described in the anthropological literature..." Yes, it is right to stress this "great variety." The problem, however, is how best to deal with it.
In the argument advanced in this book, insights derived from prototype theory suggest how we might deal productively with "great variety." And they help defuse a common complaint voiced
about a family resemblance approach to religion: the complaint that, in the absence of a set of necessary features that distinguish religion from all else, a huge and bewildering array of phenomena can be assigned to the category, and this renders the category virtually useless as a scholarly analytical tool. That complaint calls in effect for sure borders erected by stipulating necessary features or conditions. Prototype theory, in contrast, induces us to celebrate central tendencies and peripheries rather than necessities and borders.
Prototype theory is not monolithic. Prototype theorists disagree among themselves on a number of issues. Prototype theory, nevertheless, can be narrowly described as attempts -attempts among which there are discernable family resemblances -to account for "prototype effects."
Prototype effects are differences in the effective judgments that people render about how well various instances of a category exemplify that category. Many persons, for instance, may judge robins and sparrows to be clearer or "better" examples of "bird" than penguins, or apples and oranges to be clearer or "better" examples of "fruit" than olives. These judgments may be explicitly given. Or they may be suggested by such things as the order in which people list examples of a category, or the differential alacrity that individuals manifest in citing examples or in responding to examples voiced by others. The clearest or "best" examples in judgments rendered are called "the prototypes" of their categories, or "the most prototypical exemplars."
Research accomplished by a diversity of persons suggests that prototype effects are widely encountered. The classical theory of categorization cannot account for that finding. That theory, it may be recalled, holds that all of the members of a group comprehended by a category are equally members by virtue of sharing some set of necessary features.
Now, while many categories are associated with prototype effects, prototype theorists do not hold that all categories enjoy the same structure. There are significant differences in the organizational principles hypothesized for different categories, and prototype effects must be accounted for in different ways, depending on the organizational structures of their respective categories.
I advocate that "religion" be conceptualized as a "graded" category, on the model of "tall person" or "rich person." Some tall persons are taller than others and some rich persons are richer than others. And while various individuals or public agencies may suggest guidelines, there are, insofar as I know, no sure, sharp,
As the scholarly literature on religion indicates, there are no sure, sharp, and universally accepted criteria for marking off religion from not-religion. (Some scholars, indeed, have proposed such labels as "quasi-religions" or "semi-religions" to indicate that various complexes of phenomena resemble religions in some ways, but not sufficiently in other respects to justify the unqualified label "religion.") Further, some religions exhibit more of the typicality features that we associate with our general model of religion than do others, and perhaps greater elaboration of these features than is the case elsewhere. Some religions, in a manner of speaking, are "more religious" than others.
Religion, then, in my approach, is a graded category the instantiations of which are linked by family resemblances.
I further advocate that we acknowledge explicitly that for most Western scholars the clearest examples of the category religion, the most prototypical exemplars of it, are those families of religions that we call "Judaism," "Christianity," and "Islam." Those families exhibit the greatest clusterings of typicality features that we associate with religion. And, of course, those are the religions that most of us, as children, first identified with the term religion.
Fairly early in life we learned to set religion off from other things, though imperfectly. We learned, moreover, that just as religion may help unite persons who participate in the same confession, it can also divide them off from others.
As scholars, we seek to refine and deepen such understandings in systematic ways. We use our understandings, indeed, to search the world for other religions. We identify various verbal assertions and other behaviors in non-Western societies as "religious" by analogy to what we find in the West.
Now, while our most prototypical religions are important cognitive reference points, useful for purposes of orientation, illustration, and comparison, they do not immediately supply or fully disclose our general scholarly model of religion. That model is developed out of all that we have learned about religions, Western and non-Western.
Our general model is, of course, biased by our experiences in our own society. By recognizing such bias, however, we are in a better position to correct for it in our own attending to the world. We have some clear cases of what we mean by religion, and then candidates that are increasingly less clear and more problematic. But contrary to the charge that a family resemblance approach is
promiscuously inclusive, we find that as our typicality features diminish, there are fewer reasons to term peripheral candidates "religions." Our category, to repeat what was said earlier, rests on central tendencies, and centrality implies distance and periphery.
Peripheral cases are themselves extremely interesting. To the extent that they manifest elements or features that we associate with religion in our clearest or most central cases, they provide us with opportunities to study such elements or features as they may occur outside of the purview of what we unhesitatingly label religion. This, indeed, suggests that to some extent we can transcend "religion" while attending to a religious dimension in human life.
In the final pages of this book, I recommend that anthropologists selectively borrow non-Western categories and experiment with them "for probing and describing the cultures of people who do not employ them, just as we now use religion as a category for probing and describing the cultures of people who have no word and category for religion." In a review published in 1994 (p. 179 ), the anthropologist Brian Morris registers surprise that in entering my suggestion I fail to note that anthropologists have long engaged in inter-cultural borrowing by using terms such as "mana," "taboo," and "totem." Morris is justified in expressing surprise. In conceding that, I take the opportunity to clarify my position.
Mana, taboo, and totem are examples of what I would hope to avoid if my proposals were enacted fully. Those terms were inadequately explored in their original cultural settings, and they have been applied dubiously elsewhere. Franz Steiner ( 1999) gives us reasons to conclude that Victorian anthropologists, and numbers of their successors, oversimplified what may be involved in "taboo." Roger Keesing ( 1984) indicates that established anthropological notions of "mana," going back to the writings of R.H. Codrington ( 1891), are confused and not to the point in certain ethnographic applications. And the term "totem," borrowed from the Ojibwa and imposed on various other populations, is hardly a unitary phenomenon, and totem as an "ism," Claude Lévi-Strauss ( 1962) argues, is an illusion fashioned by anthropologists.
Non-Western folk categories that might be borrowed as tools for cross-cultural analysis should be studied as thoroughly as possible in their own cultural settings. One ought to explore in detail, for instance, the uses to which the category terms are put among the populations that employ them. We might very well find that those
terms are polysemous, that the categories to which they pertain are organized by family resemblances, an d that members of local populations judge some instantiations of those categories to be clearer exemplars than others. Indeed, the prescriptions given in this book for conceptualizing religion might well apply to many other folk categories, among both Western and non-Western populations. While the pages that follow have much to say about "religion," I use religion as a case study for exploring how we might best conceptualize analytical categories in our efforts to study the human condition. Benson Saler, 1999