Volume 91 1982 > Volume 91, No. 2 > Review article: the place of cognitive extensionism in the history of anthropological thought, by W. Shapiro, p 257-298
THE PLACE OF COGNITIVE EXTENSIONISM IN THE HISTORY OF ANTHROPOLOGICAL THOUGHT 1
SCHEFFLER, Harold W.: Australian Kin Classification. Cambridge Studies in Social Anthropology, Cambridge University Press, 1978. xx, 569 pp., figs, tables, map. Price £17.50.
“ . . . the occurrence of the most fundamental grammatical concepts in all languages must be considered as proof of the unity of fundamental psychological processes” (Boas 1911:67).
“We must remember that no matter how great an influence we may ascribe to environment, that influence can become active only by being exerted upon the mind; so that the characteristics of the mind must enter into the resultant forms . . .” (Boas 1962:176, originally published 1938).
Once upon a time, Robin Fox (1967:10) could argue: “Kinship is to anthropology what logic is to philosophy or the nude is to art: it is the basic discipline of the subject.” It is ironic that this was written as kinship studies were falling out of the limelight in anthropology, which they had held unchallenged since the publication of Morgan's Systems of Consanguinity and Affinity of the Human Family in 1871. For nearly a century they had been regarded as arcane but prestigious; and the materials from Aboriginal Australia held the high ground in both estimates: after all, Morgan, Durkheim, Freud, Malinowski, Radcliffe-Brown, Kroeber, Murdock, Lévi-Strauss, and other eminent theorists attached especial importance to Aboriginal social forms. However, times have changed. Since the middle of the 1960s kinship studies in general and the Aboriginal data more particularly have taken a far more limited place in our journals, and they are now widely regarded simply as arcane.
In this setting, the work under review would seem to be of only esoteric in- - 258 terest; and the author's commitment to genealogical notation, and to equivalence rules derived from structural semantics, will only scare away the uninitiated and highten this impression. But my intention here is to show that, appearances aside, Australian Kin Classification (hereafter AKC) has something to say about the nature of the human animal, as well as lesser issues in anthropological theory. I shall also assess it as a contribution to kinship studies.
One thing should be settled from the start. Unlike Morgan, Durkheim, Freud, and Lévi-Strauss, Scheffler does not attempt to parade the Aboriginal materials according to the sort of typological scheme wherein they are considered to mark a “stage” in an “evolutionary” process. This is akin to what Stephen Jay Gould (1977:56-62) has called the “ladder” view of biological evolution, in contrast to the “bushes” view that characterises selection theory. In “evolutionary” thought on human affairs, temporal imagery is employed to convey a political message; the probable extremes in this sense are the Marxists (the Evolutionary Left-Wing) and those whom Gould (ibid.: 237-42) calls the “pop ethologists”—e.g. Robert Ardrey and Desmond Morris (the Evolutionary Right-Wing). The most profound example of such “evolutionary” thinking is probably Freud's Totem and Taboo (in Brill 1938, originally published 1913), wherein the following equivalences are posited:
simple societies = prehistoric societies = child mentality = neurotic mentality = human nature, with the last element—human nature—being rather far to the Evolutionary Right.
For the most part, more recent anthropology has done two things to this bold paradigm. It has moved it much further to the Evolutionary Left: witness Leslie White's life-long redwashing, following Engels, of Morgan's racism (Harris 1968:246-49). 2 And it has focused on the first two elements of the Freudian equation, giving short shrift to the third and virtually ignoring the fourth. The “human nature” of the “neo-evolutionists” is that of B. F. Skinner's Walden Two, not William Golding's Lord of the Flies.
I shall return to these intellectual/political connections in my concluding remarks. 3 Scheffler, for his part, is more interested in the “bushes” of empirical inquiry:
. . . there is . . . much that is distinctive of . . . Australia, but there is much also that is not unique. In postulating a morally ordered universe and in attempting to comprehend it metaphorically, by analogy with the forms of their own social life, Australians demonstrate not how “primitive” they are but their intellectual kinship with the rest of mankind (p. 530).
This conclusion is not so much Durkheimian as Boasian. But in order to demonstrate this, as well as to assess other arguments put forwad in AKC, I shall have to undertake rather lengthy prolegomena.
2. Nuclear family theory
The expression “nuclear family” is polysemic: I shall use it here to refer to a co-resident grouping of a conjugal pair and its dependent offspring. The expression is also misleading: where such groupings exist, ethnographers usually take its - 259 “nuclear” quality for granted and do not attempt to demonstrate that it is the element out of which larger kin-groups are constructed on the ground, or are conceptualised. In this manner they beg the question as to whether “the nuclear family” is—to borrow from Lévi-Strauss (section 3)—“the elementary structure of kinship.”
Certain segments of Victorian social theory—most notably Morgan (1877)—claimed to find in the Aboriginal and other ethnographic materials evidence for the absence of the nuclear family. These claims were challenged by Malinowski (1913) for Aboriginal Australia, and by Lowie (1920:147-57), Malinowski (1962:132-50, originally published 1929), Murdock (1949:1-22) and others for the entire world. Murdock's numerical analysis was especially influential, with the result that, by the middle of the present century, its conclusion that “The nuclear family is a universal human social grouping” (ibid.:2) was generally accepted by social theorists. Parsons' attempt (1954) to account teleologically for certain (allegedly ubiquitous) properties of the nuclear family is perhaps the clearest signal of this acceptance (see also Fortes 1957:187).
Furthermore, it was generally maintained that other kin-groups are structural derivations of the nuclear family. The very expression “extended family” implies this position, as does Malinowski's declaration that “the clan always grows out of the family . . .” (1962:162, originally published 1930). 4 So too do the numerous statements in the ethnographic literature which imply that “the family,” “the lineage,” and/or “the clan” form a segmentary hierarchy. And just as sociocentric social forms were held to be “based upon” the nuclear family, so egocentric social forms—relationship terminologies—were said to be derived from individual connections “within” the nuclear family. This is the early version of the extensionist argument, as put forward by Malinowski, and Murdock (1949:132).
The evidence against the ubiquity of the nuclear family began to be marshalled almost immediately after Murdock's formulation. Spiro (1954) observed that in some Israeli kibbutizm parents do not live with their young children: instead, the children are reared in collective nurseries. Some societies in South India and West Africa have a rule of residence by which each spouse, after marriage, continues to reside in his/her natal abode, the man visiting the woman for sexual purposes (see e.g. Fortes 1949; Gough 1959). Segments of many modern Latin American societies have a remarkably high percentage of households in which no adult male is permanently resident (Adams 1960). This is true, too, of many black communities in the United States (Stack 1974).
But the coup de grace to nuclear family theory was delivered by Robin Fox, in his introductory text on kinship:
Take . . . polygyny, in which a man has several wives often housed in different huts, and sometimes in different parts of the country. This has been described as a “series of linked nuclear families with a father in common.” What sense does such an assertion make? The “facts” here are that several mother-child units exist, and that one male is responsible for them, circulating among them, as it were. In other cases, there is no institution of marriage at all, and the mother-child unit is not supported by the mother's mate or mates. . . . Sometimes the males spend all their time together, - 260 associating only briefly with the females. Sometimes the female has more than one mate, but none of these associates with her domestically. Sometimes sexual access to a female is limited to one male, but even so, he does not form a domestic unit with the female . . . Sometimes, even though the “one male, one female” pattern is regularised . . . this unit is lost in a larger unit from which it can only be separated artificially . . . The “universality” of the nuclear family can only be sustained by the loosest and broadest of definitions and the ignoring of “exceptions” (Fox 1967:38-9), emphases in original.) 5
I would add this. Nuclear family theory is a proposition, or a set of propositions, in which the elements of kinship are behavioural facts: it emphasises Behaviour, not Mind. Similarly, the kin-term extentionist argument derived from nuclear family theory is an argument which gives causal priority to Behaviour. It is surely no accident that Murdock's version of this argument springs from stimulus-generalisation theory in psychology (Murdock 1949:132); and that behaviourism was a major inspiration to both Malinowski and Murdock, the two most consistent exponents of nuclear family theory (see e.g. Malinowski 1939, Murdock 1945).
3. Alliance theory
What Schneider (1965b) and others have called “alliance theory” is, in fact, very much a mix; and if there is a “family resemblance” among the various alliance theories, between the remotest cousins it is all but invisible. I shall confine myself to Lévi-Strauss, the acknowledged leader of the school. By “alliance theory,” I shall refer only to Lévi-Strauss' publications on kinship.
Lévi-Strauss' first explicit break with nuclear family theory was his 1945 paper, “Structural Analysis in Linguistics and in Anthropology” (in French; translated and reprinted as Chapter 2 of Lévi-Strauss 1963). His argument is as follows (for “biological family” and “elementary family,” read “nuclear family”):
The primitive and irreducible character of the basic unit of kinship . . . is actually a direct result of the universal presence of an incest taboo. This is really saying that in human society a man must obtain a woman from another man who gives him a daughter or a sister (ibid.:46).
The idea . . . that the biological family constitutes the point of departure from which all societies elaborate their kinship system . . . has not been voiced solely by Radcliffe-Brown. There is scarcely an idea which would today elicit greater consensus. Nor is there one more dangerous . . . . Of course, the biological family is ubiquitous in human society . . . [But] in human society, kinship is allowed to establish and perpetuate itself only through specific forms of marriage. In other words, the relationships which Radcliffe-Brown calls “relationships of the first order” . . . depend upon . . . those which he considers secondary and derived. The essence of human kinship is to require the establishment of relations among what Radcliffe-Brown calls “elementary families.” Thus it is not the families (isolated terms) which are truly “elementary,” but, rather, the relations between those terms. No other interpretation can account for the universality of the incest taboo . . . (ibid.:50-51).- 261
I think the position Lévi-Strauss is attacking here is better attributed to Malinowski or Murdock than to Radcliffe-Brown (section 2); but it is not necessary to argue this here. So, too, with the facile assumption of the universality of “the” incest taboo—except for the hackneyed reminder that notions like “incest” have to do with sex and not, or not necessarily, with marriage. The crucial point is that, for Lévi-Strauss, the entailment of marriage by sex is part of an implicational chain which may be rendered as
sex = marriage = kin-group exogamy = exchange of women = alliance, wherein the = sign represents not “is equal to” but “entails.” Furthermore, the exchange of women is supposedly effected by their patrilateral kinsmen: “...in human society a man must obtain a woman from another man who gives him a daughter or a sister.”
One can only wonder whether Lévi-Strauss took this last assertion seriously. Surely he must have been aware, even in 1945, of those “matrilineal” societies in which rights over a woman's disposal in marriage are vested in her mother's brother or some other elder kinsman of her matrilineal group—if not of those “patrilineal” societies, like the Nuer, in which matrilateral kin have secondary rights of this kind, signalled by their entitlement to a portion of the bridewealth (Evans-Pritchard 1940:17). All this is especially ironic because the assertion occurs in the context of an analysis of “the avunculate”. The analysis is glib in the extreme: a “plus” sign represents dyads which are “free and familiar,” a “minus” sign dyads “characterized by hostility, antagonism, or reserve...” (Lévi-Strauss 1963:44). This all-too-simple notational system, lifted from Radcliffe-Brown's student Warner (1931:180), is used to assert the proposition MB/ZS:B/Z::F/S:H/W (ibid.:42).
In fact, the whole scheme is ethnographic balderdash (Shapiro 1975a, 1976), and the asserted analogy has been disconfirmed cross-culturally (Ryder and Blackman 1970). Its continued employment in ethnographic analyses (e.g. Maddock 1970; Nsugbe 1974:84-85) reflects Lévi-Strauss' prestige, not the plausibility (much less the truth) of his propositions.
I take it this commentary is consistent with those of Leach (1970) and others, who worry about the sleight-of-hand quality in much of Lévi-Strauss' work. In any case, the implicational chain noted above appears to be the nucleus of alliance theory. Perhaps the foggiest link in this chain is the nature of the exogamous kin-group. In Lévi-Strauss' 1945 statement this is clearly the nuclear family—a stance taken also in his later piece on “the family” (Lévi-Strauss 1956). But in his magnum opus in kinship theory (Lévi Strauss 1949) the alliance unit is held to be a unilineal group of the clan/lineage sort: nonunilineal groupings “should not be considered here, because they have nothing to do with elementary structures” (Lévi-Strauss 1969:105).
All of which may be another instance of Lévi-Strauss' performing for a captive audience. 6 But there are problems if one takes these assertions in earnest. As indicated in the foregoing quotations, Lévi-Strauss (erroneously) believes the nuclear family to be universal. But he is surely aware that unilineal grouping is not. Even if it could be demonstrated—and Lévi-Strauss has never demonstrated this—that in societies where they exist, affinal links between unilineal groups are - 262 in some sense “elementary structures of kinship,” the question remains as to what their analogues are in societies where such groups are non-exogamous (Murphy and Kasdan 1959) or—as in most of the world—lacking altogether.
It is hard to avoid the conclusion that Lévi-Strauss' notion, or notions, about “the elementary structures of kinship” is nothing more than philosophic posturing. The elements of the alleged structure—be they nuclear families or unilineal groups—are absent in a large number of societies; and the proposition that the relations between or among the elements are logically prior to the elements themselves is a word-game propped up only by an established scholasticism (Gardner 1972).
Working on the assumption that alliance theory was intended to be modified or discarded when assessed against empirical materials, Hiatt (1967), Kuper (1970), and Scheffler (1970b, 1973:780-86) have confronted it with the data and found it grossly wanting. I have followed their lead with some suspicion and judged alliance theory to be mostly “a set of definitions masquerading as empirical propositions” (Shapiro 1979:99; see also Barnes 1971:165 et seq.,). But the real problem may be to understand why, in spite of everything, some otherwise intelligent and productive individuals continue to cling to it.
In the previous section I noted that nuclear family theory is a set of propositions about Behaviour. By contrast, Scheffler's extensionism is about Mind (section 4). Alliance theory is more of a mix and, as such, provides a bridge, however flawed, between the two: The Elementary Structures of Kinship is a book about societal integration, but it is also a book about ontogeny of thought (Barnes 1971:105). It leads right readily to Lévi-Strauss' far more interesting treatments of classification and mythology.
4. Cognitive extensionism
Cognitive extensionism was self-consciously entered into the anthropological arena by Floyd Lounsbury (Lounsbury 1964a, 1964b). It did not immediately present itself as a body of theory rival to nuclear family extensionism. On the contrary, especially in a paper published the following year, Lounsbury (1965) gives the decided impression of being a dedicated Malinowskian; and the opposition is rather the “social categories” analysis of Trobriand kinship by Leach (1958). Similarly, Scheffler's initial piece on cognitive extensionism (Scheffler 1966b) contains a critique of alliance theory, linked by scholastic affiliation to Leach's analysis, as well as the following assertion: “. . . although the nuclear family as a domestic unit may not be present in every society—it is present as such in better than ninety-nine per cent . . .” (ibid.:84). This was refuted by Robin Fox a year later (section 2). Although Scheffler remains properly critical of alliance theory (section 3), he and Lounsbury no longer maintain that the nuclear family is universal, or nearly so (Scheffler and Lounsbury 1971:62-63). Their position is now clearly distinct from that of Malinowski and Murdock.
Nor has cognitive extensionism anything to do with the process by which an individual, in the course or his/her maturation, acquires knowledge of kinship. Malinowski (see bibliography) seems always to have assumed that the sequence in which children learn to structure their social world corresponds to some arrange- - 263 ment of adult life. But Scheffler and Lounsbury (1971:61-62) have noted that
the order in which a child learns the several distinct meanings of a word cannot reasonably be used as evidence for the etymological relations between those meanings. Similarly, it cannot reasonably be regarded as evidence for the structural relations between them . . . . The child's conception of these distinctions may well differ from the adult's, but it is the adult's conception, which the child has to learn if he is to communicate satisfactorily with his fellows, that must be the focus of our attention when we study the content and structure of folk systems of classification . . . An adequate understanding of these is pre-requisite to an investigation of how they are learned.
The initial espoused aims of cognitive extensionism were not theoretical but descriptive. Its alleged goal was the formalisation of ethnographic materials (Keesing 1971b). As Lounsbury (1964a:351) put it:
. . . a “formal account” of a collection of empirical data has been given when there have been specified (1) a set of primitive elements, and (2) a set of rules for operating on these, such that by the application of the latter to the former, the elements of a “model” are generated; which model in turn comes satisfactorily close to being a facsimile . . . of the empirical data . . . A formal account is thus an apparatus for predicting back the data at hand . . .
More specifically, the task was “to predict accurately who gets called what in such and such a system of kinship terminology . . .” (ibid.:352).
Thus, Lounsbury aligned himself with the aims of “the new ethnography,” self-consciously set down by Ward Goodenough 13 years earlier:
The problem of rendering an ethnographic account . . . boils down, we feel, to trying to give the reader a basis for learning to operate in terms of the culture described in somewhat the same manner that a grammar would provide him with a basis for learning to speak a language. To seek to do this implies that a culture is as susceptible of rigorous analysis and description as is any language . . . Whenever possible, therefore, the ensuing account of Trukese social organization has sought to formulate rules and state the conditions governing their application in such a way that none of the available information contradicts them . . . (Goodenough 1951:10-11), emphases added).
But the paper in which Lounsbury first stated in detail his ethnographic goals was also a comparative study of Crow-Omaha systems of kin-classification, i.e. of typologically similar systems. AKC is also a comparative piece, though the basis for comparison is geographical and historical connection. Indeed, most of Lounsbury's and Scheffler's contributions have been comparative in one sense or the other, and Scheffler (1972d) is an attempt—admittedly preliminary—at applying cognitive extensionist theory to systems of kin-classification world-wide. All this coincides with, indeed, is part of, the movement of “the new ethnography” to a “new ethnology.” 7
For my part I have no doubt that cognitive extensionism had nomothetic ambitions from its inception. In any case, an examination of the pertinent literature suggests that it can be codified into the following set of propositions:- 264
It will be noted that this set of propositions posits no specific social group (like the nuclear family), or no specific set of on-the-ground social relations (like that between parent and child, or those between affines) as being either universal or structurally prior to other forms of social organisation. Unlike alliance theory, it does not posit that relations of consanguinity are somehow derived from those of affinity: rather, the two have no regular connection; and when they are associated, in “prescriptive” marriage, the derivational relationship is just the opposite (Scheffler and Lounsbury 1971:190-228). In short, the elementary structure of kinship in cognitive extensionism is not a set of behavioural facts but a set of cognitive postulates—specifically, that a man has an essential link with his child, and a woman with hers.
The latter postulate is more or less uncontroversial, but this cannot be said of the former. Alliance theory derives the status of father from that of mother's husband; and the alleged “ignorance of paternity” in some societies—most notably those of Aboriginal Australia and the Trobriand Islands—would seem to require this. Scheffler's remarkable analysis of this allegation for Australia will be considered in section 6. For the Trobriands he has noted that Malinowski's own material clearly indicates the positing of a “genitor” status (Scheffler 1970a:387-88); and this has been confirmed by recent ethnographic research (Montague 1971; Powell, cited by Goodenough 1970a:23; Weiner 1976:121-23). 9
The literature further suggests a distinction between “strong” and “weak” forms of cognitive extensionism. The “weak” form is concerned solely with certain universal, or quasi-universal, characteristics of systems of kin-classification: the genealogical grid, the distinction between primary and other meanings, and the cross-cultural similarities of the primary meanings and folk embryologies. By this tack cultural differences in equivalence rules, and in the secondary meanings which they generate, may be attributed to less nomothetic factors: hence a possible rapprochement between “psychological” and “sociological” interpretations of relationship terminologies (Keesing 1975:119). The “weak” form is easier to - 265 sustain (because it entails fewer propositions), is more in line with one established tradition in kinship studies (that represented most cogently by Murdock), and probably has more support in the cognitive extensionist literature. In referring to cognitive extensionism below, I mean (unless otherwise specified) it in this form.
The “strong” form, by contrast, attributes even equivalence rules and secondary meanings to Mind and rejects sociological interpretations altogether. It is anathema to cultural materialists, Malinowskians, and others who think that human beings (except themselves) cannot think on their own (Shapiro 1977:44), 1979:80-82, 1981b). As it happens, this form of cognitive extensionism is supported by a rather large mass of data, including those having to do with some of the seminal debates in the history of kinship studies. I would even go so far as to speculate that Lounsbury's early work in this area was stimulated by a recognition of the bearing of these data on the debates and hence by the “strong” form of cognitive extensionism; and that both he and Scheffler continue to subscribe to this form, although both sometimes hedge in deference to contrary data and/or an expected hostile reaction from certain circles. Thus e.g. Lounsbury's Crow-Omaha paper was a devastating critique of the “unity-of-the-lineage” theory most closely associated with Radcliffe-Brown (1941). Lounsbury showed that the genealogical distribution of kin-terms in Crow-Omaha systems fit that theory only to a limited extent; yet he felt compelled to suggest in passing that the equivalence rules in such systems “express laws of succession . . .” (Lounsbury 1964a:383).
The “strong” form of cognitive extensionism has broad philosophical implications. So far as kinship studies are concerned, it has roots in the classic debate between Kroeber (1909) and Rivers (1914) on “psychological” versus “sociological” interpretations of relationship terminologies. Kroeber's position was weakened by careless language, especially his use of the hopelessly polysemic term “psychology”; whereas Rivers' was strengthened by its deterministic imagery, in contradistinction to the apparently capricious operation of Mind in Kroeber's scheme. David Schneider, by no means a cognitive extensionist (section 5), has given us a fresh appreciation of the Kroeberian position (Schneider 1968b); and the tendency of certain social forms to become elaborated without apparent outside stimulation is remarkable (see e.g. Maybury-Lewis 1979b; Shapiro 1971a). But the significant correlations between forms of kin-grouping and forms of kin-classification, and the demonstrable sensitivity of certain systems of kin-classification to outside influence (e.g. Eggan 1937; Murdock 1949), cannot be ignored either. Just how much the “strong” form of cognitive extensionism will have to be “weakened” to accommodate the data remains to be seen (Shapiro 1981b). 10
5. Neo-relativism in kinship theory
The anti-relativist implications of cognitive extensionism have not escaped its advocates (see especially Lounsbury 1969), or its opponents. Outstanding among the latter in the United States has been David Schneider, who maintains that the notion of a universal genealogical grid is an artificial imposition on the data; hence the “kinship” rubric “does not correspond to any cultural category known - 266 . . .” (1972:50, emphases in original). It will be noted that this contradicts Propositon (4) of cognitive extensionism (section 4). Further, Schneider holds that it is therefore an analytical fallacy “to study ‘kinship’ or religion or economics or politics, etc. as distinct cultural systems . . .” (ibid.:59). This contradicts Proposition (5) of cognitive extensionism.
Schneider's last statement rests largely, though not entirely, on his research on American culture (see especially Schneider 1968a, 1969), for which it has been substantially refuted by Scheffler (1976a). Otherwise, I find it difficult to take Schneider's position seriously as an empirical conclusion. On the contrary, he has encouraged a talented and prolific group of students at the University of Chicago to delve in particular into folk embryologies; and their findings on the whole give overwhelming support to Propositon (1) of cognitive extensionism, on which Propositions (2, (3), and (4) depend (see e.g. Basso 1973; Chock 1974; Montague 1971; Silverman 1971; Wagner 1967). 11 Which suggests that Schneider is primarily an ethnographer and a mentor of ethnographers; and that, his occasional theoretical statements aside, 12 the remarks quoted above are best viewed as warnings to field workers not to assume too much. Lounsbury and Scheffler, by contrast, are increasingly committed to comparison (sections 4 and 8).
Schneider's caution has paid dividends. Thus, another of his students, Lee Guemple (1969), has shown that the non-genealogical status ‘ritual sponsor’ 13 in parts of the Eskimoan Arctic is otherwise inseparable from genealogically-based statuses and must therefore be considered to occupy with them a single semantic domain. But Guemple (1972) goes too far: in stressing its “negotiability” or flexibility in kin-reckoning (as well as in plainly nonkinship institutions), he gives the impression that Arctic Eskimo society is something like a splinter group of Malinowski's Trobriand wheeler-dealers which has inexplicably moved northward. Fortes (1957) has tried to show that Trobriand society, at least, was not so “loosely-structured” as Malinowski was predisposed to see it. Those of us (like myself) who are not inclined to Fortes' sense of structure (section 7) may find Eskimo societies of especial interest because of this remarkable flexibility. But the ethnographic task, surely, is to show the bounds of this flexibility and to formalise, so far as we can, the strategies which give it shape (Keesing 1967, 1971b, 1972). A denial of this ethnographic order is tantamount to a denial of the order of Mind, and this makes fools of both us and the Eskimo (Shapiro 1982).
As it stands, Guemple's analysis of Arctic Eskimo kinship usages suggests the unsurprising conclusion that the “negotiability” he properly stresses occurs mostly in non-focal areas (compare Fjellman 1978). His account of adoptive parenthood and adoptive childhood (Guemple 1972:67-70) makes it abundantly clear that these relationships are modelled upon locally analogous biological relationships, as is generally the case (Goodenough 1970b). Similarly, the terminological transformation of affines into kin makes use of explicit equivalence rules (Guemple 1972:72). Finally, shifts from one kin-class to another because of an anomaly in age relative to Ego (ibid.) imply a focus against which the age anomaly is gauged (compare Kernan and Coult 1965; Rivière 1966; see also Scheffler and Lounsbury 1971:58-59).
None of this is to fault Guemple. On the contrary, like Schneider's other students, he has provided us with rich materials on the secondary and - 267 metaphorical significances of kinship terms and kinship constructs in a particular culture. But neither Schneider nor any of his students has refuted cognitive extensionism, or provided an alternative elementary structure of kinship. 14
In the Commonwealth the response to cognitive extensionism has been even less friendly. Here Rodney Needham has led the attack. In a remarkable instance of parallel invention (or parallel destruction), Needham (1971:4) has contradicted both Propositions (4) and (5) of cognitive extensionism through a rhetorical query:
What information is given . . . by the report that an institution has to do with “kinship?” Nothing, really, about social facts. For the label designates no distinct type of phenomena. . . . it cannot be inferred that . . . interest in kinship will be unconnected with economics or law . . .
Further, Needham (ibid.) has declared that the various cultural forms conventionally subsumed under the “kinship” rubric “have no intrinsic connexion with . . . the cultural idioms of procreation.” This contradicts Proposition (1) of cognitive extensionism.
Needham's anti-genealogical and anti-extensionist polemics antedate the advent of cognitive extensionism: indeed, the most famous (and most ironic) example is provided by Needham (1962), an attack upon a book co-authored by Schneider (Homans and Schneider 1955). But, as both Schneider (1972) and Scheffler (1973:783-84) have observed, Needham's numerous analyses of “prescriptive alliance systems” unwittingly presuppose genealogical constructs. This is also true of his students, who, as I have indicated elsewhere (Shapiro 1971b, 1975b), seem unquestioningly to adopt their mentor's scholastically inspired and empirically/flawed analytical procedures, and his dogmatic assertions of the priority of “category” over genealogy (see Maybury-Lewis 1979a for a recent example).
The limitations of Needham's approach are probably nowhere clearer than in his analyses of “lineal” relationship terminologies. Since 1966 he has correctly maintained that they are not invariably associated with unilineal descent groups (Needham 1966, 1967); yet he seems still to think that they are isomorphic with such groups. But this is usually not so: most such terminologies fail to distinguish maternal from paternal grandparents, a woman's children from a man's, and otherwise violate their allegedly “lineal” structure (Lounsbury 1964b:1079; Scheffler 1971a; Shapiro 1970a).
6. Scheffler's earlier confrontations with the Aboriginal materials
It is precisely at this point that Aboriginal relationship terminologies enter the clash; for they (more exactly, most of them) and they alone have the lineal structure posited in the models of Lévi-Strauss, Needham, and others. Paternal and maternal grandparents are distinguished. So too, reciprocally, are Ego's son's children and his/her daughter's children. So too, finally, are male Ego's children and female Ego's. The structurally basic “two-line” system may be represented as two father/son sequences of terms (with terms for females in parantheses), as follows (Shapiro 1979:47):
The more common “four-line” or “Aranda” system and other more complicated arrangements are essentially structural elaborations upon this scheme (Shapiro 1971a).
Probably its most remarkable characteristic is that a man and his wife (a woman and her husband) apply separate terms to their common offspring. It was this—presumably the contravention of Proposition (2) of cognitive extensionism—that initially led Scheffler (1969) to believe that cognitive extensionist theory does not apply to these Aboriginal systems. Within two years he had absorbed enough further data to present a preliminary argument that it is indeed applicable even in the absence of a generalised ‘child’ class (Scheffler 1971b); and in the years up to 1978 he substantially augmented this new position (Scheffler 1972a, 1972c, 1977a).
Now Proposition (2) of cognitive extensionism is logically dependent upon Proposition (1); and the latter plainly runs counter to the often-alleged “ignorance of paternity” in Aboriginal Australia. Scheffler (1973:750) suggests instead that Aboriginal folk embryologies do indeed posit a link between sexual intercourse and conception and therefore a “genitor” status; and that the more readily discussed (by male Aborigines) and therefore more widely reported (by male anthropologists) “spirit-finding” is held to occur not at conception but at quickening (see also Scheffler 1976b). There is a goodly amount of evidence for the first assertion (Shapiro 1979:10-12) but relatively little for the second. Yet as Scheffler (ibid.) notes, it is consistent with folk embryologies from other parts of the world. Future ethnographers working in Aboriginal Australia might well explore this hypothesis in situ. Those of us (myself included) who already have substantial data on the matter might consider dusting off our field notebooks.
The Aboriginal materials thus pose special problems and demand fresh analysis—not through “evolutionary” mysticism (section 1) but by someone able to link their historical uniqueness to a general theory of Humanity. Scheffler's credentials suggest that no one is better suited to this task.
7. The work under review
AKC is thus a full-dress display of the results of Scheffler's decade-long grapplings with the Aboriginal materials, inspired by Lounsbury's work during the preceding half-decade:
This book is an attempt to demonstrate that the categories by which the aboriginal peoples of Australia order their social lives are predominantly kin categories—a moot point in social anthropology—and, beyond this, to reveal the structures and the relations among the structures of Australian systems of kin classification (p. ix).
The very nature of Australian social institutions and of their comparability to the institutions of other peoples is still very much in dispute. For anyone who believes, as I do, that kinship systems are a constant feature of human societies, the challenge of the Australian data still remains (p. xi).- 269
The dispute arises from considerations noted in the preceding section; and also from Radcliffe-Brown's earlier efforts, adopted by Elkin (1938) and others, to represent Aboriginal relationship terminologies as conjugally-linked “patrilines.” 15
Scheffler proceeds to consider Aboriginal folk embryology, arriving at the conclusions noted in section 6. He then provides a critical review of the history of the dispute just alluded to, tracing it to its origins in Victorian social theory (compare Scheffler 1972c).
The bulk of the book consists of cognitive extensionist analyses of particular schemes of kin-classification—based upon the published sources, unpublished manuscripts lodged at the Australian Institute of Aboriginal Studies and elsewhere, and Scheffler's own penetrating if brief field research among various Aboriginal groups. The first generation of students produced by Radcliffe-Brown's Sydney regime have been remarkably hesitant in publishing their findings: we are indebted to Scheffler for interviewing some of the senior Sydney-siders still with us and delving into dusty manuscripts, thus rescuing much useful information from oblivion. Kinship buffs in particular will be interested in his re-analyses of the more controversial materials: Kariera, Aranda, Dieri, Ngarinyin, “Murngin,” and others. 16
These analyses point to the following conclusions:
. . . much of the diversity among Australian systems of kin-classification may be accounted for as the product of various combinations of a fairly small stock of structural elements. These are, principally, half-a-dozen or so dimensions of conceptual opposition variously combined to yield a somewhat larger number of principal kinclasses and subclasses and a few rules of genealogical structural equivalence. Because there is relatively little variation at the level of principal classes . . ., the structurally most distinctive differences among the systems are differences in their respective sets of equivalence rules (p. 418).
This being so, and also because the determination of kin-class foci is both logically prior to, and of greater theoretical importance than, the equivalence rules (sections 4 and 8), I shall first consider Scheffler's treatment of the former—especially his contention that the focal referents of Aboriginal kin-classes are genealogically close kin. The following, regarding the Aranda, is exemplary:
According to [the missionary Carl] Strehlow . . ., the postfix larra . . . is one of two that may be combined with a basic [relationship] term to indicate a “class” relative rather than a “blood” relative of the designated category. The other is knara, which Strehlow glosses as ‘big’. A kata knara, for example, is an older brother or classificatory ‘brother’ of one's ‘own’ or ‘proper’ father (i.e., one's kata atja); a kata larra is a younger brother or classificatory ‘brother’ of one's father; and similarly for most other Aranda [kin-terms]. These forms signify the relative age of the designated kinsman in relation to the closest kintype denotatum (or denotata) of the basic term, which is marked as the atja, ‘own’ or ‘proper’, designatum of the term. From all this it should be clear that the semantic function of atja is to specify - 270 the focal or logically primary designatum of a term, for it is in relation to this kintype that the terminological statuses of other kintypes as knara or larra are determined (p. 25).
We may therefore conclude that kata in Aranda has “father” for its primary meaning, and that its other meanings are secondary. But note that there are no comparable data to indicate a distinction between brothers and ‘brothers’: Scheffler seems rather to take such a distinction for granted, a consideration to which I shall very soon return.
Moreover, AKC presents little further analysis in support of Proposition (2) of cognitive extensionism. Part of the blame for this deficiency must surely be rooted in the data Scheffler has at his disposal, collected as it has been by ethnographers unsophisticated in semantic analysis, and/or with an unquestioning commitment to an extensionist or anti-extensionist stance. But these ethnographic gaps seem to bother Scheffler hardly at all: as with the alleged Aranda distinction between brothers and ‘brothers’, most of the book simply carries on from the assumption that Proposition (2) must necessarily be true.
Further evidence for this assertion comes from Scheffler's selective treatment of the north-east Arnhem Land materials. He cites (p. 540) a personal communication from Nicolas Peterson, according to whom each “Murngin” relationship term may be modified by the adjective yuwalk (‘true’) or by the suffix—'mirringu, so as to designate “the genealogically closest kin of that category . . .; ” When Scheffler informed me of Peterson's findings, well before the manuscript of AKC was completed, I wrote to him that they did not tally with mine. My own research indicated that, although yuwalk does indeed separate “the genealogically closest kin” from non-kin who happen to be of the same kin-class, it does not separate the former from kin of that class who are genealogically more distant; 17 and that—mirringu is a stylistic suffix which signals that the lexeme in question is a kin-term but does not separate genealogically close kin from anyone else (Shapiro 1981a:31 et seq.). Thus, for example, the expression yuwalk bapa applies to Ego's genitor but also to his FB, FFBS, FMZS, and others; and anyone Ego calls bapa—not just his genitor—may also be referred to as bapa'mirringu.
I do not wish to insist that Peterson is wrong and that I am right: indeed, it is conceivable that we may both be right (or both wrong); for, although the “Murngin” groups among whom we did most of our work are closely related historically and continue to interact with each other, their local forms of social classification are decidedly different (Shapiro 1977:41). Whatever the case, my main point is that Scheffler's outright dismissal of my findings on these matters is yet another indication that, for him, Proposition (2) of the cognitive extensionist position is in danger of transgressing the bounds wherein it can be questioned.
For my part I suspect that Proposition (2) is, in the main, correct; but I insist that it be empirically demonstrated and not assumed. I shall attempt to effect such a demonstration in the case of the north-east Arnhem Land lexeme bapa by appealing to the sort of data which Scheffler has not normally taken into account. The following considerations strike me as germane:
I would not, however, argue that, because the kin-class bapa is associated with a certain sociocentric division which is defined, so far as Ego is concerned, by reference to his genitor, bapa therefore has “genitor” as its primary meaning. Scheffler has relied heavily upon arguments of this sort. Thus, he notes that the anti-extensionist
theory maintains, in effect, that the Aranda expression kata [‘father’] is used in two different though related senses; the primary sense is “man of my father's subsection” and the derivative sense is “my father” . . . [This position] is illogical because the category designated by the term in its alleged primary sense . . . not only includes ego's father but is defined by reference to ego's father; this definition treats ego's father as the relative by reference to whom membership in the broader category is reckoned . . . (p. 24, emphasis in original).- 272
But surely here the anti-extensionist position requires only that the Aranda somehow posit the status of genitor, and that they allow this status a certain salience in the reckoning of a larger class of persons which includes it. I cannot see how these considerations (as opposed to those noted earlier) are sufficient to establish that kata primarily means “father.”
Nor can one assume in a work of this sort—Scheffler repeatedly makes the assumption—that, given two or more referents of a relationship term, the most senior referent is focal. This is indeed generally true of systems of kin-classification, as Scheffler points out (pp. 98-99), a consideration which may properly be used in the elucidation of particular cases. But the subsequent employment of these cases in support of the cognitive extensionist position gives this position a tautological ring and deprives it of some of its empirical status (compare Scheffler 1967:171-73).
Scheffler's analyses of the relations among kin-classes whose focal members have, presumably, been ascertained are far more numerous, and far less questionable. Consider the following, based on Scheffler's own field work among the Arabana of South Australia:
. . . one of my informants said that, as between bilya ‘father's mother’ and tandi ‘mother's father’, “bilya is the main word, the big word; tandi is like inside bilya.” As I understand it, this means that tandi designates a special subclass of the cross-grandparent class, whose designation is bilya; when this special subclass is removed from the general class this leaves a residual subclass whose members may be designated as bilya (p. 252).
Scheffler's appreciation of these and less exegetically-based relationships between sub- and superclasses (or “principal classes”) is one of the strongest points of AKC (see especially pp. 460 et seq.). Of particular interest for the history of kinship theory is his persuasive argument that the so-called “section” systems of Aboriginal Australia are but one form in which egocentric superclasses may be realised. But even here Scheffler's arguments may be questioned. I would urge the following:
1. In arguments concerning class inclusion, such as the one just quoted, Scheffler places great weight on the “syntactic” relations among overt (i.e. labelled) kin-classes, especially—by his own admission (p. 418)—those of reciprocity. Thus, if it can be shown empirically that kin-class X contains subclasses X1, X2, and X3, and if it can be similarly demonstrated that X and Y are reciprocal classes, it is taken to follow logically that Y is analogously divided into Y1, Y2, and Y3: no independent evidence for this latter subdivision is thought to be needed. Scheffler has elsewhere (Scheffler 1977b) argued this point in some detail—again on the basis of logical necessity. But the logic appealed to is a Western logic; and the degree of its application elsewhere constitutes an empirical finding of very considerable import, not a matter to be assumed. 19
2. In arguing for the existence of superclasses, Scheffler places some (though considerably less) weight on the substantial literature on Aboriginal bodypart symbolism, wherein lexically distinct classes can be represented by the same body part. But this tactic must be used with caution. Thus, in portions of south-central Arnhem Land, the ‘MF’ and ‘FM’ classes are lexically distinct, both from each other and from all other lexically designated kin-classes; but these two classes - 273 alone are associated with the chin, a consideration which suggests a superclass relationship for which I could present considerable further evidence from my field notes. In the north-east, however, the same bodypart association holds for these two classes, but they are linked by suffixes to two other classes. There is compelling evidence that these last two are simply derivations (subclasses) of the first pair. Yet the bodypart symbolism appears to be askew: the derived classes are associated not with the chin but with the knee, and they share this association with a number of other classes (e.g. ‘WM’, ‘WMB’) to which neither Scheffler nor I would wish to argue they are related by class inclusion (Shapiro 1977, 1981a:32, 94).
This apparently mysterious state of affairs is brought to order by a single consideration: all the “knee” classes—and only those classes—have as their members individuals belonging to the matrilines of a man's actual or potential wives. They are thus members of a sort of secondary superclass—one based upon their common affinal connotations subsequent to their derivation from primary superclasses (Shapiro 1968, 1981a:93-96). Their logical status is identical to that of a more familiar class of objects designatable as “rusty nails”: rather than use such a nail to help affix a photograph to the wall of my den (as I might well have done when I could consider it simply as a “nail”), I am more likely to toss it (and other “rusty” things) into my rubbish bin.
3. Although I agree with Scheffler that “section” systems are structurally (and presumably historically) derived from relationship terminologies, I must insist that these systems, once derived, can and do carry on sui generis. This argument is presented at length in Shapiro (1977). I would call attention here to Scheffler's utterly incorrect acceptance of the received view, perpetrated by Radcliffe-Brown in over-reaction to Victorian social theory, that it is always kin-classes and never “sections” that effectively determine marriage in Aboriginal societies, and to the overwhelming contrary evidence (Shapiro 1977, 1979:67-71). Scheffler's argument is that such sociocentric divisions have only an illustory significance in affinal politics “because in general the women whom a man may claim as wives by right of kinship belong to the section or subsection formally paired ‘by marriage’ with his own” (p. 470). The evidential basis for this position is astonishingly slender and in no wise points in the direction Scheffler imagines (Shapiro 1981a:143-45).
These last two criticisms are relatively trivial. The first, by contrast, suggests the most radical flaw of AKC, particularly when coupled with some of Scheffler's arguments, noted above, on focality: a decided tendency towards circularity, concealed beneath a mammoth quantity of empirical materials and analysis and even a cautionary word (pp. 402-03). I shall return to this matter in my concluding remarks.
I wish to turn now to Scheffler's treatment of the equivalence rules allegedly operative in Aboriginal systems of kin-classification. It will be recalled that Lounsbury's pioneer contribuion to cognitive-extensionist theory was on the structure of Crow-Omaha systems (section 4). Now, in most such systems, relationship terms are not used especially widely: 20 hence the problem of dealing with more remote kin-types is not particularly salient. The same appears to hold for most of the other relationship terminologies heretofore analysed by cognitive ex- - 274 tensionists. Aboriginal societies, by contrast, are famous for their application of kin-terms to everyone. Hence any analysis of an Aboriginal system of kin-classification that claims “to predict accurately who gets called what” would presumably have to deal with kin-types considerably more remote than a collateral remove or two.
But AKC does nothing of this sort. Collateral kin-types are given especially short shrift, Scheffler's analyses terminating at no more than two collateral removes (e.g. MMBD, FMBSD). This is really not his fault; for if Scheffler's informants, and those of the other ethnographers whose data he uses, resembled mine in their proclivity for genealogical chains (as I must assume they did), then he is lucky to have got even this far. I shall expand upon this assertion by recalling some of my own experiences in eliciting genealogies in north-east Arnhem Land.
As intimated above, my first inquiries into this topic were made through the kin-terms themselves, a tactic which leads rather readily to tautology. After a time, however, I learned to express the elements I had in mind by a set of more specialised lexical items, which circumvented the lexicon of kinship terms 21 and which I present here in the form of glosses:
‘bearer’ (genetrix) ‘one who was borne by...’ (a woman's child)
‘finder’ (genitor) ‘one who was found by...’ (a man's child)
With the help of certain other lexemes, I could then create more complex constructions, some of which were readily grasped by informants, e.g.:
‘the bearer of your finder’ (an informant's FM); and
‘others, besides X, who were found by Y and borne by Z’ (X's full siblings).
But some of my constructions were grasped less readily, or not at all. In particular I noticed a decided tendency among my informants to balk when one or more kinsman was posited as linking the one initially designated to the one I had in mind, e.g.
‘the bearer of the finder of your finder’ (an informant's FFM),
or, referring to even first-degree collateral kin and using a kin-term for the sibling link,
‘the ones who were borne by the sister of your finder’ (an informant's FZ's children).
I did not lose information because of difficulties of this second kind, because informants were entirely comfortable with my recasting the kin-type in question as ‘the ones who were borne by your aunt’ (i.e. so long as they could consider the linking kin by relationship term), and also because, in conventional Notes and Queries fashion, I usually obtained the personal names of an informant's FZs before querying him about their children. But I do think it remarkable (and worth repeating) that, in my experience, Aboriginal Australians easily decode the messages ‘aunt's children’ and ‘X's children’ but not the message ‘father's sister's children’. Comparable data from other parts of the continent (or elsewhere, for that matter) would be of considerable interest.
By contrast, difficulties of the first sort—‘the bearer of the finder of your finder’—presented substantial problems in compiling genealogies. Informants were unable to decode messages of this sort, though, as with FZ's children, they could supply the appropriate relationship term once I did likewise for the linking kinsman (FF). But only some informants knew their lineal kin of the third ascen- - 275 ding generation by personal name, very few commanded this information for those of the fourth ascending generation, and none at all for “higher” generations.
I must add a rider to this. By “personal name” I mean one or more of an individual's secular names. Both men and women also have ritual names, and men (though not, I think, women) are supposed to know these for lineal agnates as remote as the FFFF. Each man also has a ‘head name’—the ritual name of his FF suffixed by a morpheme meaning “towards”—which is known by other men and applied to him in ceremoniai contexts. ‘Head names’ can thus be used to separate male Ego's (or X's) FBSs from his other collateral kin.
Similarly, because nearly all informants knew the names of their grandparents, both FBSs and certain other close collateral kin—specifically, parents' siblings and their descendants—were relatively easy to record on genealogies. Beyond this point, however, difficulties began to occur. Those informants who did not know their great-grandparents by name were often unsure whether alleged siblings of their grandparents were their actual siblings. Many could not even nominate (say) a FMZ or a MMB. The situation could occasionally be rectified by consulting an older or otherwise more knowledgeable man, but this was obviously not always possible. At further removes—great-great-grandparents and their descendants other than great-grandparents, etc.—information of the kind required by the conventional “genealogical method” was skimpy in the extreme.
In dealing with such relatively remote kin-types and even with close collateral kin, informants were generally more comfortable operating through the relationship terminology: it made little or no personal or social difference to them whether (say) an alleged brother of the MM was in fact a MMB or a more remote ‘brother’ of the MM—though, for reasons detailed below and elsewhere (Shapiro 1981a:54-55)—it was of considerably more concern whether or not the alleged brother in question was a clansman of the MM. In any case he is called ‘MMB’, and his daughters (informants seemed never to tire of pointing this out) are potential mothers-in-law (Shapiro 1969, 1970b, 1981a:46).
But even this last fact holds only if the daughters are members of the ‘WM’ kin-class, and this in turn can be guaranteed only if their mother is a member of one of the two ‘WMM’ kin-classes. If she is a member of another kin-class—(say) ‘mother’—the daughters may still be called ‘WM’, by patrifiliation (‘following the father's road’); but, unless they are of the MM's clan, they are more likely to be called ‘sister’, by matrifiliation (‘following the mother's road’) (Shapiro 1981a:34 et seq.).
Now none of these further considerations—clan membership, relations of patri- and matrifiliation between kin-classes—however “nongenealogical” they may appear from a Notes and Queries standpoint, fails to resort to genealogical constructs of some sort: Scheffler has repeatedly and properly made this point (e.g. Scheffler 1972b:376; Scheffler and Lounsbury 1971:142). It is the “sort” that is remarkable, and which suggests that, apart from relationships of parentation, what one “needs to know in order to predict accurately who gets called what” in north-east Arnhem Land has little to do with equivalence rules. This “little” shrinks in handling even close collateral kin-types. It fades into insignificance when called upon to deal with more complex genealogical chains, - 276 which may be meaningful to Scheffler but are not, so far as I can ascertain, to Aboriginal Australians.
I say “may be,” because of the original goals of cognitive extensionism (section 4). I have already noted Scheffler's remarkably limited concern with such chains in AKC. He has, moreover, relatively recently appreciated “relative product” analyses of the kind just illustrated (Scheffler and Lounsbury 1971:140-50). Hence his continued concern with genealogies is something of a mystery.
Two considerations suggest a solution to the puzzle. One is that Scheffler apparently now believes that “relative product” statements volunteered by informants can only be interpreted as emic vindications of equivalence rules. The best evidence Scheffler offers for this is from his own field work among the Baniata of the Solomon Islands (Scheffler 1972b; but see also Scheffler and Lounsbury 1971:144-47), and it consists of comments of the following sort:
“The X (‘sister’) of my Y (‘mother’) I call Y also.”
“The children of this latter Y I call Z (‘sibling’).”
But these are not necessarily statements of kin-class extension, nor do they even presuppose that an informant can forge a genealogical link between himself and the person or persons in question. On the face of the evidence presented, they could simply be statements of the relationship between kin-terms.
The second consideration is more interesting. I noted in section 4 that cognitive extensionism has from its inception had both ethnographic and ethnological aims. Now if one's goal is comparison, there must be a “data language” in terms of which disparate elements may be compared; and in the history of kinship studies this has usually, though not always (section 5), been provided by genealogies. Clearly Scheffler thinks—and I agree with him—that there are good reasons why this should be so: I shall take this up again in the next section. I wish here to point out only that, all this being as it is, it is necessary to confine oneself to limited genealogical constructions, since (even if societies such as those of Aboriginal Australia applied kin-terms on the basis of detailed genealogical reckoning—which they do not!) the unrestricted application of relationship terms is a relatively rare phenomenon. In most societies—not just those of or derived from Western Europe—such terms are used only within a comparatively limited social circle. It may turn out, then, that the primary importance of equivalence rules will lie in their cross-cultural applicability to relatively restricted genealogical chains (compare D'Andrade 1970:111; Kronenfeld 1980).
Be this as it may, the most distressing aspect of AKC is Scheffler's dogged determination, following and even surpassing Fortes (1969:102-21), to reduce much of the rest of Aboriginal social organisation to egocentric kinship. Thus, in his argument, noted above, that “section” systems are one kind of realisation of kinship superclasses, he rejects the idea, propounded by several anthropologists, that such systems have an underlying moiety structure. 22 The grounds for this rejection seem to be the following:
1. The moiety systems which allegedly underlie “section” systems are often absent when the latter are present.
2. Even when both moiety and “section” systems are present, the former are sometimes defined relationally—e.g. ‘our side’/‘their side’—not by proper name.- 277
3. Relationally defined moiety systems are sometimes merely pairs of aggregates of kin-classes, upon which classes they are therefore structurally dependent (see Shapiro 1979:45-48 for illustration).
It is curious that Scheffler is prepared to grant a sort of higher analytical status to moiety systems with proper names, such as the Duwa/Yirritja division in much of Arnhem Land (Shapiro 1977), than to those that are relationally defined: I shall return to this below. For north-east Arnhem Land I have shown that the distribution of kin-terms for any Ego is such as to result in an own moiety/other moiety division of the population, and that the content of this division is Ego-independent (Shapiro 1970a:381). Here, then, a relationally defined moiety scheme exists alongside one with proper names, and the two have the same structure. Elsewhere, apparently (this has yet to be demonstrated), a similar ordering of the terminology exists independently—i.e. without a moiety system with proper names. Still elsewhere, the same structure exists in ‘our side’/‘their side’ form. Even if we concede Scheffler's argument that this structure is sometimes achieved through kin-classification or “section” systems, is it still not remarkable that, in a series of geographically contiguous and historically related societies, it is repeatedly if variously achieved?
I wish to make it clear that I am not appealing to a universal dualism in human thought of the sort posited by Lévi-Strauss (1966b and elsewhere) but to a specific dualistic classification of society found widely in (though not confined to) Aboriginal Australia. A full appreciation of this structure has yet to be determined, though it probably transcends named moiety systems, “section” systems, and perhaps even systems of kin-classification (Brandenstein 1970; Shapiro 1979:72-74).
In any case, Scheffler's most profound capitulation to Fortesian analysis is to be found in the final chapter of AKC, which consists primarily of a re-analysis of the materials on the Walbiri of Central Australia, especially Mervyn Meggitt's outstanding monograph (1962) and his more recent Morgan Centennial paper (Meggitt 1972). The latter in particular is guaranteed to draw fire from cognitive extensionists. Consider:
. . . if we confine ourselves to analysis of the dyadic relationships expressed in [the kinship] system . . ., there is no way of accounting for the other manifestly significant features of Walbiri society . . . moieties, subsections, and lines of descent or filiation do not emerge contingently from . . . a simple addition of particular, similar dyadic relationships . . . Rather, the broader categories that are . . . deduced from fundmental metaphysical postulates of the Walbiri . . . are . . . superimposed on . . . the kinship matrix . . . (ibid.:82).
One might think that Scheffler's critique would be founded on the argument that the “fundamental metaphysical postulates of the Walbiri” are themselves derived from kinship concepts; that “moieties, subsections, and lines of descent or filiation” are indeed contingent upon “dyadic relationships.” But this is only partly the case. Thus, the “subsection” (eight-“section”) system is dismissed almost in passing—not so much because of its likely structural derivation from the relationship terminology (see above) as because of Meggitt's comments (1962:169) on its functional unimportance.- 278
In the same vein, what Meggitt calls “patrimoieties,” “matrimoieties,” and “alternate generation-levels” are held by Scheffler to be nothing “more than aggregates of kin classes” (p. 507). In keeping with his remarkably limited sense of dual organisation, Scheffler also notes that these divisions “have no proper names. They are identified and distinguished only by reciprocal expressions . . .” (ibid.). The status of the “alternate generation-levels” is denigrated because these categories “are of no jural significance” (p. 513), that of the patrimoieties because they “exist, to the extent that they may be said to exist at all, only in ritual contexts” (p. 515). This reliance upon functional considerations is remarkable, and I shall return to it.
From a purely ethnographic standpoint I am prepared to concede Scheffler's argument regarding Walbiri matrilines and the occasions for their action, e.g. marital bestowal and mortuary obligations: “The affairs in question focus on particular individuals, so the matrilines are not fixed sets but only the close uterine kin of those individuals . . .” (p. 511). But the matrikin mentioned by Meggitt (1962:192 et seq.) form a matrilineal and not simply a matrilateral set, and they appear to have in common certain symbolic qualities. This is similar to the situation in north-east Arnhem Land. But elsewhere in Arnhem Land comparable qualities and functions are ascribed to sociocentric rather than egocentric matrilineal divisions, i.e. to matriclans (Shapiro 1981a:89-105). If, then, the distinction between sociocentric and egocentric matriliny is not a salient one for Arnhem Landers—indeed, not for Aboriginal Australians generally (Shapiro 1979:27-30)—, why should it be so for their analysts?
This argument is patently similar to the one I made above for moiety systems, and it presents legitimate typological problems. 23 If, following Goodenough (1970a) and Keesing (1966), our discourse is ethnographic in the strictest sense, our typology should be such as accurately to reflect locally meaningful units; and there is no indication that the Walbiri recognise sociocentric matriliny and distinguish it from egocentric matriliny, which they do recognise. Hence, in dealing solely with the Walbiri materials, one is justified in insisting on this distinction. At the opposite pole of analysis, the same distinction probably needs to be made: thus, it appears unlikely that nomothetic theories on the development and disintegration of matrilineal corporations (e.g. Aberle 1961; Gough 1961a) will apply also to egocentric matriliny—i.e. that the latter, when considered simply as “matriliny” would lend itself to what philosophers of science call a “natural typology” (see Hempel 1965:146 et seq.).
Between the two extremes, however, is the more limited comparison of spatially and temporally related materials, which presumably reflects local historical processes. In the cases at hand, the (superficial) differences that exist presumably reflect different degrees of elaboration upon the Aboriginal premises that females are secular and not ritual beings; that certain activities, especially marital bestowal, are secular and not religious; and that the people responsible for them should therefore be related through females (Shapiro 1979:28-29). All this being so, the distinction between sociocentric and egocentric matriliny is not particularly salient for our appreciation of Aboriginal thought.
We have seen that, according to Scheffler, Walbiri subsections, moieties, and matrilines have the sort of secondary ontological status that Fortesian analysis - 279 gives to everything save corporations (see Shapiro 1981a:25-27, 102-03; Wagner 1974). But what of Walbiri patriclans? 24 Fortes (1969:117-18) is prepared to admit them into the hallowed grounds reserved for descent groups, but Scheffler is not: this on the basis of certain considerations, listed below more or less in the order in which they occur in AKC:
Perhaps the most remarkable conclusion from all this is that two of these considerations—the fourth and the sixth—are functional rather than structural: they have to do not with the cultural construction of the patriclans but with their relative unimportance in Walbiri life. In the same vein is Scheffler's dismissal, noted above, of the Walbiri “subsection” system on functional grounds; of patrimoieties as existing “to the extent that they may be said to exist at all, only in ritual contexts”; and of the “alternate generation-levels” as having “no jural significance.” One might even mention the quotes that Scheffler finds it necessary to use in connection with clan ownership; as if incorporeal property were a hand-me-down version of corporeal property. The tie with a mode of analysis designed primarily to glorify the idea of a kinship corporation could hardly be more obvious. Nor could it be more depressing, coming as it is here from a man who, once upon a time, was in the vanguard of the attempt to liberate the notion of “descent” from that of “the corporation” (Scheffler 1964, 1965, 1966a; see also Schneider 1967).
Even so, it is not at all clear that “descent-ordered” societies are so very different. Higher-order segments in such societies tend to be corporate only with respect to ritual, leaving the control of subsistence property to lower-order segments (Befu and Plotnicov 1962). Fortes (1945:147) notes for the Tallensi that “a woman is a minor in jural and ritual matters,” and closely comparable statements can be made for the Mae Enga (Meggitt 1964), at least some portions of traditional Chinese society (Freedman 1958:85), probably all Islamicised lineage systems, and even some of the societies in the “matrilineal belt” of Central Africa (Richards 1950).
Turning to structural and cultural considerations, I fail to see that membership in Tallensi, Enga, Chinese, Islamic, or Central Bantu lineages is based upon - 280 anything other than filiation, as is the case with Walbiri patriclans (Scheffler's first consideration; compare Goodenough 1970a:46). The most striking structural characteristic of groups of the former kind is that, unlike those of the latter, they are organised according to a genealogical segmentary hierarchy—this presumably related to their larger size, the greater pressure they therefore place upon subsistence resources, and their more obvious capacity for corporate control and collective action (Fortes 1953; Sahlins 1961). 26 I have no particular objection to their being called “descent groups” for these reasons, indeed, as I have intimated above, this last consideration has been the effective basis for the application of the expression. It should be pointed out, however, that this usage makes the “connexion” between descent grouping and corporate characteristics a matter of definition instead of a matter of fact; and that “descent theory” is thus primarily concerned with the assortment of ethnographic specimens (see note 23) and not with the development of empirical propositions.
But Scheffler, in fine Fortesian style (see especially Goody 1961), will simply not let the matter drop. Just as unnamed moiety systems are analysed out of existence, so Walbiri patriclans, apparently lacking both generic and specific names (Scheffler's second and third considerations), are somehow less real than Tallensi lineages, and their ethereal quality is not altered by the consideration that they may be otherwise identified. These alternative modes of identification, noted above, have to do with the relation of Walbiri clansmen to Dreamtime beings, and to one another, and with Scheffler's remaining considerations—his fifth, seventh, and eighth—to which I now turn.
That the Walbiri posit that Dreamtime beings are “still-existent” is beyond question. But Scheffler's conclusion from this that they cannot therefore also be ancestral is entirely foreign to the metaphysics encoded in Walbiri ritual (Munn 1964). Furthermore, it seems to suggest that the distinction is absolute even in “descent-ordered” societies: if so, what are we to make of the congeries of institutions frequently dubbed “ancestor cults”? Finally, though Western theology is not my game, it occurs to me that the position of God in Judaism, and that of the Trinity in Christianity, have at least this much in common with the notion of the Dreamtime in Walbiri thought. 27
Which leads right readily to a further set of conclusions. In Judaeo-Christian contexts it is sometimes said that all human beings are “children of God.” Similarly, Christians sometimes refer to themselves collectively as “brothers in Christ” and address one another by sibling terms. In most Christian Churches the entailed “filial” relationship is established through rituals in which individuals are said to be “reborn” or “born again.” Most of us will concede that these usages pertain to a domain of metaphorical kinship, one which is modelled upon a kinship domain simpliciter but contains at least partly different elements and a different logic: my son would not be said to be a “grandchild of God” but, like, me, a “child of God.”
Walbiri initiation is rather closely comparable to Christian baptismal rituals: each youth is “reborn” as the “son”—more precisely, the ‘man's son’ (section 6)—of a Dreamtime being. There are cultural and sociological differences, of which perhaps the most obvious are that the number of ritual groups among the Walbiri is necessarily greater than one, whereas the number exceeds one only by - 281 historical contingency in Christianity (which has experienced repeated attempts to restore it to a single such group); and that Walbiri initiation excludes females. But in the Walbiri case, too, the kinship idiom is metaphorical: a man's father is said to be a ‘man's son’ of their common Dreamtime being, not its ‘brother’.
Now far from denying this metaphorical quality, Scheffler repeatedly asserts it. Thus:
Drawing on their theories of how human beings reproduce, some Australian cultures postulate an analogous process as part of their theories of the relationships between human and Dreamtime beings. In this process the Dreamtime being whose “essence” has become incorporated in an individual is likened to the individual's ‘father’ in the special sense that he is, as it were, the individual's “spiritual genitor”—not the genitor of his or her . . . body, but the genitor of certain other aspects of his or her being. Conversely, the individual is said to be the [‘man's] child’ of the Dreamtime being, in the special sense that he is, as it were, the Dreamtime being's “spiritual off-spring.” Thus, when used to express a relationship between a Dreamtime being and one of his human representatives, the expressions ‘father’ and [‘man's] child’ are used in . . . metaphorical senses. They refer not to genealogical relationships established through the human reproductive process but to similar relationships established through a similar process (pp. 526-27, emphasis added). 28
Presumably, then, we could relabel the relationship between a Dreamtime being and one of its human incarnations as “metaphorical filiation.” But the generative powers of such a being are posited to have operated during and since the Dreamtime, including not only the creation of certain natural features but, as well, the spiritual generation of an individual's human ancestors, known and posited (Munn 1970, 1973). Is this being not therefore a metaphorical ancestor? Is not “the continuity of the Dreaming” therefore tantamount to metaphorical agnation? And are not Walbiri patriclans—and those in other Aboriginal societies—therefore metaphorical descent groups?
This is not intended as an exercise in lepidopterology. My point is that once we recognise the existence of metaphorical kinship, we should be prepared also to recognise its elaboration. The “totemic clan” systems of Aboriginal Australia and elsewhere (Shapiro 1981a:26-27) probably represent a set of structurally similar realisations of this elaborative tendency. Metaphorical agnation in Christianity, and Christendom as a metaphorical descent group, are further realisations of this tendency. This mode of analysis strikes me as especially promising (see also note 28), and I hope elsewhere to pursue it.
Even in the truncated form presented here, its debt to Scheffler is obvious. But Scheffler himself draws very different conclusions from the data. According to him, Walbiri patriclans are not descent groups but are more like kindreds; the ego or propositus is the still-existent Dreamtime being. Unlike most other cultural categories described in the ethnographic literature as “kindreds,” these Walbiri categories are mutually exclusive; . . . no individual can be a member óf more than one. Because of this difference, some anthropologists may want to avoid describing them as kindreds . . . If a general designation is - 282 necessary, “patrifilial kin groups” seems adequate (p. 522; for references to partly similar arguments, see Shapiro 1981a:106).
This allows Scheffler to extend his argument, such that Aboriginal totemic clans are assigned a status similar to “section” and moiety systems and matrilines: “even the ‘patrilineal clan’ is a kind of kin class” (p. 523).
For my part, I have no particular objection to calling Aboriginal patriclans “patrifilial kin groups”—so long as it is recognised that, so far as recruitment is concerned, a similar label might be attached to (say) Tallensi descent groups (see above). But I think it plain that the more important issue for Scheffler has to do with whether such divisions are—to use Goodenough's distinction (1961)—ancestor-oriented or ego-oriented, as Scheffler contends (see also Scheffler 1973:756-65). The following considerations seem to me to be relevant:
Scheffler's attempt to render Aboriginal social life primarily as egocentric kinship may fairly be compared to Maddock (1974), in which Aboriginal society is portrayed as a set of corporations linked by affinity and ritual ties. Both works oversimplify. AKC comes closer to the mark than The Australian Aborigines: A Portrait of Their Society, largely because Scheffler is able to operate more independently of Fortes than Maddock is of Lévi-Strauss. Maddock's book is part of the cult of a metaphorical and still-existent ancestor, whose creative powers the empirical data merely illustrate. Scheffler's has more profound links with the scientific enterprise.- 283
It is for this reason that I have chosen to give AKC what will doubtlessly be construed as an essentially negative review. I am disturbed by Scheffler's commitment to the sterility of Fortesian analysis and still more by the circularity with which the tenets of cognitive extensionism are applied to the Aboriginal data in this book. It is not that these tenets are not solidly grounded: on the contrary, it is precisely because cognitive extensionism is well on its way towards revealing the true elementary structures of kinship that we must be especially alert to anything that would lessen its status as an empirical theory.
For my part, too, I should like to see more concern by cognitive extensionists—and by the ethnographers upon whose data they rely—with native genealogical constructs: what Conklin (1964) has called “ethnogenealogies” (see also Barnes 1967). Apart from his Baniata and Siriono analyses (section 7), Scheffler has accepted without question the traditional methods of gathering the data on which cognitive extensionist theory depends. It is worth pointing out in this connection that genealogies obtained in Notes and Queries fashion are not analogous to the texts obtained by linguists. The latter are direct samples of the data to be analysed. The former are samples of data arranged according to a grid predetermined by the analyst and would therefore seem to be a degree removed from the data sui generis.
Yet, when all is said and done, the success of cognitive extensionism suggests rather strongly that the kinship constructs of our informants are not so different from our own, certainly not as different as neo-relativists like Needham and Schneider (section 5) would have us believe. For all its ethnographic limitations, “the genealogical method” is not simply a convenient way of getting and comparing data on systems of kin-classification: it seems to be reasonably close to the manner in which the human mind—whether it be that of the trained ethnographer, the Western layman, or the Aboriginal Australian—manages much of these data.
Such cross-cultural uniformities in cognition are by no means confined to systems of kin-classification. The seminal study of colour classification by Berlin and Kay (1969) is clearly the most famous example from other domains, and several commentators have noted the parallels between the two lines of inquiry (e.g. Burling 1970:45-54; Leech 1974:232-62). Especially remarkable is the regularity in the focal membership of classes in both domains. The corresponding variation in non-focal membership may be due to local historical or sociological causes (section 4); and it provides a superficial relativism which anti-cognitivists like Marvin Harris (1968:592-96) have chosen to emphasise. 30 Similar conclusions have already been shown to be applicable to still other domains of classification, wherein cross-cultural research has only recently taken hold (Witkowski and Brown 1978).
The research by Joseph Greenberg and his associates on language universals—recently summarised by Greenberg (1975)—provides a link between such nomothetic research on denotative meaning and other aspects of cognition. So too does the recent spurt of interest in the cross-cultural study of connotation and metaphor (e.g. D'Andrade and Egan 1974; Osgood et al. 1975). Wallace's work on “revitalization movements” is nothing less than an attempt to lay bare the - 284 cognitive basis of religious groupings everywhere (Wallace 1956, 1970:188-99). Theories of transformational grammar have been subjected to massive psycholinguistic testing (Greene 1972:107-88), and anthropologists have become increasingly aware of the necessity of confronting Lévi-Strauss' theories of classification and mythology with further empirical materials (e.g. Hammel 1972; Hiatt 1969).
This enumeration of research on regularities in cognition is merely illustrative and by no means exhaustive. Even so, at least three sets of observations may profitably be made in connection with it.
First, such research emphasises endogeny—the ability of human beings to create structures which are largely independent of external circumstances (Shapiro 1981b). Contrast this with the behaviourist view of the human organism as a set of “contingencies of re-inforcement” and “contingencies of survival” (Skinner 1974). Harris (1964, 1979:15) is much taken with behaviourism though its inability to account for structure-dependent behaviour has been apparent for several decades (see e.g. Chomsky 1959; Lashley 1951). So far as the elementary structure of kinship posited by cognitive extensionism is concerned, the universality of the mother/child dyad might seem to be amenable to reinforcement theory though I daresay it is better handled as something akin to what the ethologists call “imprinting” (Sluckin 1972). 31 In any case, it could be argued that reinforcement theory is applicable to the father/child dyad in Western European societies. But I should hate to be in the shoes of the Skinnerian ethnographer trying to deal with this dyad in (say) certain segments of traditional Nayar society, where paternity is accorded jural status in the virtual absence of behaviour between a man and his children (see section 4 and Gough 1961b:363-64).
Second, although all this suggests a marked innatist component in human affairs, cognitive extensionism and related areas of theory and research have proceeded almost entirely independently of selection theory in biology. This is mainly a matter only of academic background and scholarly priorities: most of those who have pursued nomothetic inquiry into human cognition are tough-minded empiricists, more concerned with exposing regularities that continue to stun relativists than with accounting for these regularities by appeal to one or another established interpretive scheme. But at least two further considerations have, or may have, played relatively minor roles:
1. The postulate of continuity, in selection theory, between humanity and other animals is in the main not supported by cognitivist research (see e.g. Chomsky 1972:66-71; Miller et al. 1960:141-42). Admittedly, some of the gap has been bridged by recent research on communication with the great apes (reviewed by Hill 1978); but most of it remains. Thus, e.g. even if we allow the mother/child dyad in mammalian societies to be considered anlagen of the forms of that dyad posited in human folk embryologies, there still appears to be no rudiments of a father/child relationship in nonhuman societies. This “missing link” has been recognised for some time: precisely four decades separate the virtually identical conclusions of Lowie (1933) and Barnes (1973) on the matter. Attempts to locate such a “link”—either in the primatological data or in the more recent spate of anthropological interest in the social carnivores—have been highly speculative and/or based on flawed appreciation of the data at hand (e.g. Fox 1975; Mackey 1976).- 285
2. In their efforts at closing such gaps, ethologists—“pop” and otherwise—have relied massively on the metaphor of typological resemblance as historical connection (section 1): hence the “links” just alluded to are purely formal and are at best only suggestive of phylogenetic relationship. One can sympathise with the perplexity that prehistorians must feel when they try to read behaviour (and still more, cognition) from the fossil record; and it can be assumed, petitio principii, that father/child behaviour and ideologies of paternal connection must have arisen at some point in hominid phylogeny. But the attention lavished upon the savannah baboon during the decade of the 1960s is now presumably only a chapter in the history of anthropology—not a reason to take seriously a particular origin myth of human sociality. 32
Finally, although modern cognitive science comes from a variety of academic backgrounds, and although, even within anthropology, it remains an essentially convergent development, it none the less bears a “family resemblance” to the cognitivism of Franz Boas and his early students. The point seems to be fairly widely recognised (e.g Stocking 1974:19; Tyler 1969:2). But the portrait of “the Boas school” as a homogenised pack of relativists, smeared ad nauseam by Leslie White and his students (e.g. White 1963; White and Dillingham 1973:53-54), is far more generally accepted. Marvin Harris, who has inherited the Throne of Cultural Materialism from White and perpetuated the Doctrine of Boas as the Anti-Marx, has accepted it (Harris 1968:250 et seq.). 33 So, too, has ethological anthropology (see e.g. Berghe 1975:9; Freeman 1970). There is indeed support for this interpretation in the Boasian literature. But Boas and his students were far too complex a group for caricature. It seems fair to suggest that the disentanglement of relativist and innatist components in the Boasian tradition remains a problem for future research (Lounsbury 1969, Shapiro 1980, Stocking 1974).
Cognitive extensionism, operating within that tradition, has revealed remarkable regularities in the data. AKC, despite its flaws, goes a long way towards bringing the Aboriginal materials and kinship studies back into the mainstream of anthropological theory. For this reason it is a seminal book.
- 286 Page of endnotes
- 287 Page of endnotes- 288
A Soldier's View of Empire
THE REMINISCENCES OF JAMES BODELL
* * A NEW ZEALAND BOOK
The extraordinary story of an English Soldier who became MAYOR OF TAURANGA!!!
“This extraordinary book makes most interesting reading, Professor Sinclair has done a fine job of editing.”
*BODLEY HEAD $19.95*
**AVAILABLE FROM ALL GOOD BOOKSHOPS**
1 Thanks are due to the Faculty Academic Study Program of Rutgers University for providing me with the free time to develop this paper—the first to be published in a series which I have called “Innatism and Endogeny in Boasian Anthropology” (Shapiro 1980). I wish also to express my gratitude to the Rutgers Research Council for providing research funds.
2 Compare e.g. White (1949), in all his Kremlinophilic glory, with White (1948), a grudging recognition of Morgan's racism. Something quite comparable seems to have been done with Freud's innatism (Sulloway 1979).
3 I am especially indebted to Nisbet (1969) in developing this argument. See also Murdock's penetrating comments on evolutionary theory (Murdock 1959), and the revealing diagram in Goldenweiser (1922:22).
4 Fortes (1957:175) has pointed out the absurdity of this position, as well as its controversion in Malinowski's own writings.
5 These remarks, especially those pertaining to polygyny, should cause us to re-examine Malinowski's contention (see above) that the nuclear family is ubiquitous in Aboriginal Australia. My materials on patterns of association in north-east Arnhem Land, only some of which have been published (Shapiro 1973, 1981a:72), do not unequivocally support Malinowski. I hope eventually to deal with this in detail. I should note here Fox's—and Adams' (1960)—argument that what is universal in human social organisation is the mother/child dyad. This is a vast improvement over nuclear family theory; but it is controverted by the kibbutz data, noted above. More to the point, so far as its being the elementary structure of kinship is concerned, it is not structurally prior to most other kinship groups. Thus, Fox (1975) finds it necessary to posit a structurally independent mating relationship, from which the position of mother's husband (not that of father), is, presumably, historically (not structurally) derived. But there is another alternative (section 4), which is structurally simpler, more consistent with the data, and which does not rely on “evolutionary” conjecture.
6 This interpretation is supported by Lévi-Strauss' return, in the final chapter of The Elementary Structures of Kinship, to the proposition that “the biological family” is the unit, or at least a unit, of exchange. See also Lévi-Strauss (1966a:17 et seq.).
7 Further examples of this shift are provided by Goodenough (1970a), Witkowski and Brown (1978), and in section 8. The distinction between ethnography and ethnology is derived from Goodenough (1956).
8 The use of a single term for one's children, regardless of sex, is common in systems of kin-classification. Most Aboriginal relationship terminologies present special analytical problems in this area (section 6).
9 I know of two recent reports (Monberg 1975; Tonkinson 1978), pertaining to ethnographically very disparate societies, in which the “ignorance of paternity” bogus is raised afresh, but I shall have to deal with them elsewhere. Suffice it to say here that, although I suspect—as do Lounsbury and Scheffler—that “recognition of paternity” is universal, even its quasi-universality (if this were the case) resolves problems that cannot be handled by nuclear family theory or alliance theory. The same applies in regard to the other universals, or quasi-universals, in cognitive extensionist theory. See the notions of “near universals” and “statistical universals” developed by Greenberg and his associates (1966:xx) in connection with the study of languages.
10 Kroeber later modified his position in deference to the data. His distinction in the sphere of social organisation between “basic patterns” (arising from Behaviour) and “secondary patterns” (originating in Mind) is the single most important inspiration of this paper (see Kroeber 1952:210-18, originally published in 1939).
11 Scheffler, too, was Schneider's student, though by 1966 the influence upon him of Lounsbury, his senior colleague at Yale, had become much more substantial.
12 The most extensive of these is Schneider (1976).
13 In keeping with the conventions of “the new ethnography,” and revealing of its American origins, I use single quotes to signal glosses—approximate translations of foreign expressions. Double quotes are otherwise used in American fashion. I am indebted to The Journal of the Polynesian Society for allowing this departure from Commonwealth practice.
14 See Keesing (1974:83-88) and Marshall (1977) for recognitions, from very different perspectives, of this theoretical poverty amid ethnographic riches. For critiques of cognitive extensionism by those less equivocally associated with “the new ethnography,” see D'Andrade (1970:110 et seq.) and Tyler (1966).
15 Compare e.g. Radcliffe-Brown (1931) with Radcliffe-Brown (1951), the latter a far more sophisticated piece of structural-semantic analysis. One of the bonus payoffs of AKC is Scheffler's fresh appreciation of Radcliffe-Brown's “structural principles,” heretofore consigned to oblivion.
16 In the main I shall not comment here on Scheffler's analysis of the north-east Arnhem Land (“Murngin”) data. I saw it in manuscript and offered some data and critical commentary, which Scheffler kindly acknowledges (p. 287). In turn, I sent him portions of the manuscript of Shapiro (1981a)—though too late to be taken into account in AKC. I would add only that, where the two works conflict, I see no reason to withdraw or modify any of my conclusions.
17 My expressions “nonkin” and “kin . . . who are genealogically more distant” are deliberate. Scheffler, by contrast, assumes that anyone to whom a kin-term is applied is ipso facto a kinsman. I shall return to this below.
18 This analysis and its conclusion should be taken as superseding Shapiro (1979:56-57).
19 Thus, e.g. Dole (1969) has shown that a very large number of relationship terminologies do not follow what Tax (1955:19) calls “the rule of uniform descent.”
20 I pass this off impressionistically: the proposition remains to be rigorously demonstrated. I do know of exceptions, both in Australia and elsewhere, wherein relationship terms are applied society-wide.
21 The evidence for this circumvention is presented in Shapiro (1981a:16, 31-32, 87). It should be recognised that terms for kin are not necessarily kinship terms (Schneider 1965a). Nor does the use of kinship terms presuppose a relation of kinship—a point to which I shall return in the following section (see also note 17).
22 Readers unfamiliar with this argument should see Shapiro (1979:59-61, 64).
23 Compare the moot issues properly dismissed by Leach (1961:2 et seq.) as “butterfly-collecting,” wherein typology is treated as an end in itself.
24 In the literatuire on the Walbiri these units are variously called “patrilines,” “patrilodges,” and “patrilineages.” But since they are (unsurprisingly) very similar to the units most often termed “patriclans” in the rest of Aboriginal Australia, and since Scheffler's argument is intended to have this wider application, I have chosen this last term.
25 Compare north-east Arnhem Land baparru, noted above.
26 But compare Keesing (1971a), who draws attention to the limitations of describing social action, even in such societies, in terms of “descent.”
27 Although I have the highest regard for Munn's analysis of Walbiri temporal notions cited above, I am unconvinced that these are radically different from ours. The foregoing paragraph is in part intended as support for my position, which I would buttress by the following. A north-east Arnhem Land informant (whose notion of the Dreamtime can, I presume, be regarded as reasonably close to that of the Walbiri) once observed in my presence that the Dreamtime existed ‘an exceedingly long time ago’. There are no doubt metaphysical schemes which presuppose a “nonlinear” sense of time, as opposed to the “linear” sense that dominates Western secular thought. But to suppose, with Leach (1961:126), that individuals are culturally bound to one or the other scheme is to embrace an unwarranted relativism. The data of this and the following paragraph suggest that individuals readily move from one kind of scheme to the other.
28 This analysis is part of a larger one in which Scheffler attempts to link Aboriginal initiation with “spirit-finding” (section 6), the latter “just a special case of a more widespread theory about how human beings come to instantiate Dreamtime beings . . .” (p. 526). I have added above certain aspects of Judaeo-Christian theology for comparative purposes. I hope elsewhere to pursue this productive line of inquiry.
29 A modification of this line of inquiry has recently been pursued by Turner (1974, 1977, 1980). But I shall have to deal with this elsewhere.
30 In all fairness to Harris, it should be pointed out that the Berlin and Kay study had not been published when he wrote his famous critique of cognitive anthropology; and that “the new ethnography” had barely begun to move to a “new ethnology.” But Lounsbury's early papers on cognitive extensionism were already available. Harris' comments on the Crow-Omaha paper indicate that he fails completely to grasp its anti-relativist implications. Worse still, and despite its adherence to an all-but-obsolete behaviourism, The Rise of Anthropological Theory enjoys a celebrity which is probably unparalleled in American anthropology. I fail to see how anyone who has read Paul Kay's devastating retort (Kay 1970) to Harris on “the new ethnography” can take at least that part of Harris' book seriously.
31 On this matter, see the massive array of data and theory marshalled by Bowlby (1969, 1973).
32 For critiques of the baboon model of protohominid society, see Crook (1973) and Pilbeam (1977). Consider also the following observation: “Primate behavior is quite variable and presents nothing like a unified model from which human behavior may be considered to have emerged” (Fried 1967:40). Perceptions of this sort bear an obvious logical and temporal relationship to the paradigm shift from the savannah baboon (and, to a lesser extent, other nonhuman primates) to the social carnivores (see especially Schaller and Lowther 1969). It is worth noting, too, the similarities, in both folk and scientific circles in the West, between such creatures which, somehow, are both contemporary and ancestral, and Aboriginal Dreamtime beings (section 7).
33 The facetious tone of this sentence is inspired, in very different ways, by Lowie (1946) and Harris (1979:280-81).