What human kinship is primarily about: toward a critique of the new kinship studies

Abstract

The claims of the so-called ‘constructionist’ position in kinship studies are examined with reference to a recent article by Susan McKinnon. McKinnon's analysis is shown to be deeply flawed, primarily because she pays no attention to the phenomenon of focality, now widely established in cognitive science. Instead, she is trapped in unsupportable collectivist models of human kinship. It is argued that these models are part of a misguided critique of the Western European Enlightenment.

Introduction

My title and subtitle are anything but accidental. In 1972 David Schneider published ‘What is kinship all about?’ (Schneider 1972), which was followed a dozen years later by A critique of the study of kinship (Schneider 1984). His conclusions in both publications have been widely taken to mean that models of procreation¹ provide only one criterion for kin-reckoning throughout the world and are in no sense primary, as they are in the West.² This stance has been called ‘culturalist’ or ‘constructionist’ (‘social constructionist’, ‘cultural constructionist’) (Shapiro 2005a). The constructionist position has morphed into a highly self-conscious ‘new kinship studies’ (Carsten 2000: 3), which presents itself as part of a larger ‘deconstructionist’ movement in social theory (Yanagisako and Delaney 1995). Yet this larger movement is not without its critics. Nor are the claims of the new kinship studies in particular accepted by all (see e.g. Kuper 1999: 122–58; Patterson 2005; Shimizu 1991). But no one so far as I know has attempted a sustained analysis of this scholarship.³ I shall do so here, with reference to a particular article which, I hope to show elsewhere, is reasonably exemplary.

Prolegomena

But first I need briefly to consider a series of phenomena sometimes dubbed ‘prototype effects’. Thus we call Roman Catholic priest ‘father’, but we know intuitively that he is not as central a member of the ‘father’ class as one's genitor is. We might say that he is ‘like a father’, or fatherish, in that he is male, authoritative, and nurturant, that his position is likened to or modeled upon that of one's real father.⁴ We might also say that it is this latter who is the focal member of the ‘father’ class, and that membership in this class is extended to the priest. Such structuring of semantic space has been shown to apply quite widely in human cognition, and it has been suggested that it is in this matter that the mind constructs categories (e.g. D'Andrade 1995: 115–21; Kronenfeld 1996: 147–65; Lakoff 1987; Shapiro 2005a).

I need also to note, before proceeding, that my presentation assumes no previous acquaintance on the reader's part with kinship theory in anthropology.

Susan McKinnon on the nature of human kinship

A recent article by Susan McKinnon (2005b) exemplifies the constructionist position in kinship studies with remarkable clarity and expressly engages it adversarially with notions of kinship in evolutionary psychology. This engagement appears to be only the first in McKinnon's unfolding agenda, for she has more recently come out with a small book purporting to ‘deconstruct’ evolutionary psychology as a whole (McKinnon 2005a), and the ‘deconstruction’ enterprise figures heavily elsewhere in her scholarship (McKinnon 1995a, 2000, 2001; Franklin and McKinnon 2001). My concern here is mostly with her kinship piece. Here McKinnon argues that ‘the genetic calculus’ of evolutionary psychology disrespects the nuances of human kin-reckoning, as Sahlins (1976) argued three decades ago in his critique of sociobiology. One possible retort to this charge is that present-day Darwinian scholars are concerned not with kinship constructs but with kinship behaviour: thus there is a statistical tendency for maternal grandmothers to be the most investing of the four grandparents (Buss 2004: 237–40), for fathers to be far less likely physically and sexually to abuse their children than stepfathers their stepchildren (Daly and Wilson 1988: 86–91), and for kin to be far less likely to aggress lethally against each other than nonkin (1988: 17–35) – and all this is so quite apart from how particular kin relationships, and kinship in general, are conceptualised. But a more interesting answer is that evolutionary psychologists, who do not pretend to be specialists in the cross-cultural study of kinship, have managed to grasp the truth more profoundly than McKinnon, who does. In most of the rest of this article I shall document this surprising assertion, using McKinnon's own examples but also supplementing them for further and richer illustration.

A case in point is provided by the Wari Indians of southwestern Brazil. In Wari theory bodily substances can be shared in four ways: (1) through procreation, involving parents, children, and, less directly, other less immediate consanguines; (2) through wet-nursing; (3) through sexual intercourse between husband and wife, or between paramours; and (4) through killing enemies outside one's community, in which the killer and his victim are said to be in a father/son relationship (Conklin 2001a: 116–22; Vilaca 2000: 95). Now I shall guess that Susan McKinnon would seize upon (2), (3), and (4) especially and conclude that it is disrespectful of native views to stress (1) in analysing Wari notions of kinship, and that these notions are very different from what we have in the West. And she would be wrong on both counts. For we learn that, among the Wari, although ‘all consanguineal kin share some body substance … the most direct bodily connections are those … among parents and children …’ (Conklin 2001a: 118). Moreover, ‘[e]xchanges of … body fluids (through breast-feeding, sexual intercourse, and the killing of enemies) establish relationships that Wari recognize as being similar to, though weaker than, the consanguineal links that exist at birth’ (2001a: 118, emphasis added). Hence the parent/child relationship provides the quintessential kin-tie for the Wari, with other sorts of kin ties deemed to be, literally, substantially less.

But this is not all. The Wari have what has been called a ‘universal system of kin categorization’ (Barnard 1978) – which is to say that an individual applies kin terms to everyone with whom he or she associates. Does this mean that everyone is considered kin, though some are closer than others? Not at all. In fact the Wari distinguish lexically between ‘true kin’ and ‘those who are like kin but are not truly related’ (Conklin 2002: 215; see also Vilaca 2000: 94–5). Presumably the father/son relationship established between a killer and his victim is in the latter category, for the act of killing makes the two only ‘like real kin’ (Conklin 2001a: 121, 2001b; Vilaca 2002: 359) – i.e. they are likened to‘true kin’, in a manner comparable to my rendition of a Catholic priest as ‘father’. Similarly, when he takes a life a Wari man's abdomen is supposed to swell, and he is said to be in a state which is ‘like pregnancy’ but not one that ‘is pregnancy’ (Conklin 2001b: 161, emphasis in original).

These likenings stemming from killing may at first blush seem strange to Western minds, but a little reflection should dispel this appearance. For we have a considerable lexicon which renders sex as akin to violence, and vice versa. The polysemy of such expressions as to bang and to pound points in one direction, that of to cream and to whack in the other (Roscoe 1994; Shapiro 1995b). These renditions suggest parallels between triumphing over an opponent and, of all posited kin relationships, the tie linking husband and wife. But a father/son relationship, real or metaphorical, connected with violence is an element in the central creative act of both Judaism (Abraham's quasi-sacrifice of Isaac) and Christianity (the Crucifixion). And, like the Wari, both religions make use of birthing imagery in connection with hierarchical relationships between two males – specifically, Jonas' being regurgitated by the Great Fish, after which he becomes obedient to divine authority, to which Jesus' Emergence from the Tomb and similarly enhanced spiritual state is expressly compared in Matthew 12:38–41. Finally, the Roman Catholic version of Christianity goes one better than both its Protestant form and Judaism by rendering Jesus as maternal in much medieval and early modern symbolism (Bynum 1982) – much like the Wari warrior.

More prosaic are the other similarities between Wari and Western kinship notions. We too entertain the fiction – and we see it as such – that everyone is kin in expressions like ‘the brotherhood of man’ and ‘brothers and sisters in Christ’. We too have notions of wet-nursing as establishing a secondary sort of kinship (Fildes 1988). Many Protestant churches have wedding ceremonies in which bride and groom are said to be ‘one flesh’– this derived from Matthew 19:4–6:

And he answered and said unto them, Have ye not read, that he which made them at the beginning made them male and female? And said, For this cause shall a man leave father and mother, and shall cleave to his wife: and they twain shall be one flesh.

Wherefore they are no more twain, but one flesh.

Moreover, in her research on American marriage, Naomi Quinn found that ‘[o]ne of the features of the American model of marriage … is the general idea of sharedness, which is instantiated by different people in many different ways, including metaphors of merging such as “We were one person now”… or “It wasn't just the two of us anymore. We were a family”’ (email communication dated 25/4/06).

In a nutshell, Wari ideas about kinship, like Western notions of kinship, are grounded in native appreciations of procreation, and from this base they extend to other areas of experience. The claim of a West/Rest dichotomy, in this instance at least, is entirely without support.

But most of McKinnon's argument for such a dichotomy pertains not to notions of kinship in general but to kin classes, i.e. to those categories designated by what the old kinship studies called ‘kinship terminologies’. Her initial contention here is that, in the West, one's genetrix is the sole member of the mother class, whereas, among the Rest, the (only superficially comparable) class has multiple membership. This last assertion, I argue below, is misleading, but the first is plainly and simply false. I offer the following bits of auto-ethnography:

• 1
  I refer to the woman who (I am told) bore me and who (I know with certainty) nurtured me when I was young as my mother.
• 2
  This woman referred to another woman as my mother. I refer to this other woman as my grandmother, which designation I also use to refer to the woman referred to as my mother by the man I refer to as my father.
• 3
  When I was married my wife referred to a woman as my mother. I referred to her as my mother-in-law.
• 4
  Just before I reached puberty I was a member of a Cub Scout ‘pack’. I referred to the woman who superintended this group as my denmother.
• 5
  I regard English as my mother tongue and the United States of America as my mother country.
• 6
  Although I am not a Roman Catholic, I have Roman Catholic friends, each of whom has a godmother. And I know of some Roman Catholic women who are members of religious orders superintended by Mothers Superior.

This list is hardly exhaustive, but it should suffice to show that the ‘multiplicity of mothers’McKinnon (2005b: 109) finds in non-Western settings can be found in the West as well.

But of course I need to add that not all mothers are for me equally ‘motherly’. In fact, there is for me a single central or – to use a term from the old kinship studies –focal member of this class, which is the woman described in (1). The remaining members are such only in a much ‘looser’ sense. Indeed, I might describe them as motherish, just as a Roman Catholic priest is ‘fatherish’: their membership in the ‘mother’ class stems from likening them to or modelling them on my real mother, insofar as they are (or can be construed as) female, nurturing, and, to varying degrees, authoritative. If I were asked simply who my mother is, I would nominate only the woman described in (1). And I believe my Roman Catholic friends would structure the ‘mother’ category in much the same way, as would people who have stepmothers. This might not be the case for people who are adopted, especially when very young, for the children of lesbian mothers, and for people brought into being via the new reproductive technologies, for whom the expression ‘real mother’ might have more ambiguous reference (see also Lakoff 1987: 74–84).

In my own field research among the Aboriginal Australian people of northeast Arnhem Land I also found ‘a multiplicity of mothers’. There one's genetrix is called ngarndi in some dialects, ngama in others – but so is her sister, nearly all other women who she calls ‘sister’, all of one's father's wives, most of the wives of men one's father calls ‘brother’, various other women, the estate and ritual objects associated with the ritual group of the genetrix, and other estates and ritual objects mythically linked to that group. But northeast Arnhem Landers distinguish between ‘full’ (dangang) and ‘partial’ (marrkangga) members of the ngama class. The genetrix is without question a member of the ‘full’ngama subclass. So are her sisters, though informants sometimes added, when so nominating them, something like ‘but she's not the one who bore me, the one from whose womb I emerged’. Which is to say that the membership of the genetrix's sisters in the ‘full’ngama subclass is subject to qualification, to hedging, as if it referred to a ‘grey area’ in people's structuring of their social worlds. Also inhabiting this area are the genetrix's ritual group ‘sisters’, as well as a woman the genetrix calls ‘sister’ whose own ritual group is not that of the genetrix but whose genetrix's ritual group is that of the genetrix of the genetrix of the informant. Another woman who the genetrix calls ‘sister’ may be said to be a ‘full’ member of the ngama class based on the consideration that her ritual group estate and its ritual objects are mythically linked to those of the genetrix, but such an assignment can be contested and, in any case, is likely to be successful only if the estates in question adjoin. All other women in the ngama class are members of the ‘partial’ngama subclass and are often described as ‘not really ngama’ or ‘ngama only by virtue of kin classification’.

Moreover, asked simply ‘Who is your ngama?’ an individual invariably nominates his/her genetrix. This of course jibes with my auto-ethnographic data. Also, northeast Arnhem Landers employ bodypart symbolism to represent kin class relationships, and a woman called ngama–any such woman, not just the genetrix – may be indicated by touching one's nipple – as if maternal succour provided the model for all such relationships. Consider too the following lexemes:

• ngamani= milk

• ngama'ngama'yun= to create (-yun being a verbalising suffix)

• nguy-ngamatirri= to love (nguy= heart, -tirri being a verbalising suffix)

Which is of course to say that notions of procreation and succour are conceptualised by northeast Arnhem Landers – as well as by Westerners – as quintessentially associated with the biological mother.⁵

The application of the kin term ngama to the ritual group estate of the genetrix and its ritual objects is logically dependent upon the identification of her as the focal member of the ngama class and is therefore a derived or secondary member of that class (cf. Scheffler and Lounsbury 1971: 60–1). The same point is made by reference to that estate as one's ‘milk country’. All this is remarkably comparable to my own sense that the United States of America is my ‘mother country’.

There is, finally, the consideration that ngama, in a language entirely unrelated to English, sounds much like ‘mama’. Murdock (1959) showed nearly a half century ago that this sort of phonological regularity occurs with more often than chance frequency in parental kin terms.⁶

McKinnon cites several examples of such ‘a multiplicity of mothers’, but she fails utterly to appreciate their semantic structure. Thus she notes that in systems of kin classification Lowie (1928) dubbed ‘generational’ all women of the parental generation are members of the ‘mother’ class (McKinnon 2005b: 110). This is fine, so far as it goes – but it does not go very far. As it happens, however, we have a recent detailed account of a ‘generational’ system by Richard Feinberg (2004). Writing on the residents of the Polynesian island of Anuta, Feinberg notes that, although the genetrix is merged with other female kin of her generation at a superficial level of classification, as is the genitor with the male kin of his, there is a special ‘parent’ term which is applied to both but which is not extended to others (2004: 68). Moreover, kin class reckoning depends upon parental kin class assignment: thus for example anyone who one genetrix calls ‘sister’ is called ‘mother’ (2004: 74–5) – which is to say that the kin class position of such a ‘mother’ is logically dependent upon, or derived from, that of the genetrix. Further still, these people make a gross distinction among the sphere of individuals to whom kin terms are applied: some are said to be ‘true’ members of their kin classes, while the membership of others is ‘outside’, or ‘a lie’ (2004: 81). There is some flexibility as to the membership of these gross subclasses, but this flexibility occurs along definite procreative lines, such that, for example, siblings are always ‘true’ kin but first cousins may or may not be (2004: 81–4). And, finally, asked to nominate members of various kin classes, people ‘usually answer as if the question were posed specifically about the genealogically closest relative in the designated category’ (2004: 82). In Feinberg's own words, ‘the basic model for assignment to kin classes is a genealogical one’ (2004: 80).⁷ The similarities to my own materials from northeast Arnhem Land – indeed, to the Wari data and my own venture into auto-ethnography – should be clear.

McKinnon also claims that systems of kin classification that the old kinship studies usually called ‘Omaha’ posit ‘a multiplicity of mothers’ (McKinnon 2005b: 110), and once again the statement, though true, is misleading. In Omaha-type systems, the mother's patriline is singled out, such that the ‘mother’ and ‘mother's brother’ terms are applied to patrilineal descendants of the mother's brother: for example, both his daughter and his son's daughter are called ‘mother’. But even this rendition of things focuses on the mother (and her brother) – i.e. the position of other members of the ‘mother’ (and ‘mother's brother) classes is derived from those of these two close procreative kin. That this is so is underscored by the fact that the ‘mother’ term, when applied to other kin, is in such systems usually if not always accompanied by a lexical or other indicator of nonfocality, much like ‘godmother’ or ‘stepmother’ in English. Thus Aboriginal Australian people in a part of the Cape York Peninsula, just across the Gulf of Carpentaria from northeast Arnhem Land, refer to the daughter of the mother's brother as ‘little mother’ (McConvell and Alpher 2002: 163),⁸ whereas further west, just outside Arnhem Land itself, the same relative is rendered as ‘branch mother’– this in contrast to the genetrix, who is ‘trunk mother’ (2002: 171). In this latter example the trunk–branch opposition presumably suggests the base–derivative one. Among the Fox Indians of Illinois the mother's sister is called ‘little mother’, and it is this latter derived term, not the ‘mother’ term simpliciter, that is applied to the mother's brother's daughter and other women of the mother's patriline (Tax 1955: 252; see also Radcliffe-Brown 1941: 10). Especially remarkable here is Karl Heider's research on an Omaha-type system among the Grand Valley Dani of the Indonesian half of New Guinea. Much as I did in northeast Arnhem Land, and Feinberg seems to have done on Anuta, Heider (1978) asked his informants to name an individual to whom each Dani kin term is properly applied. Given the ‘mother’ term, nearly all informants nominated their genetrices. The only exceptions were a man whose mother had been killed early in his life, who nominated his mother's mother, who raised him; and another man, who first named his father's current wife, then his mother (Heider 1978: 238).

Similar considerations apply in those cases in which the mother's brother is called ‘mother’, put forward by McKinnon (2005b: 110) as further evidence for ‘a multiplicity of mothers’, now supposedly unconstrained even by gender. In all cases of which I am aware the mother's brother is more fully rendered as ‘male mother’. But the genetrix is not said to be a ‘female mother’, for this would be redundant – which is to say that the focal member of the class is female (see e.g. Kuper 1976; Middleton 2000; Shapiro 1981: 28). This is also true, for example, of the English class designated by the label ‘nurse’, with the associated rubric ‘male nurse’, which, I shall guess, is deemed politically incorrect these days.

McKinnon (2005b: 111) points out that Inuit naming practices sometimes skew kin term relationships so that, for example, a female infant who receives the name of her mother's mother is called ‘mother’ by her genetrix, thus presumably showing that ‘a multiplicity of mothers’ can exist not only independently of gender but, as well, of minimal age considerations. I have not been able to access the source she cites, but similar practices seem to be widespread in the Inuit area. Consider the following:

Personal names could sometimes complicate a genealogy … If a cousin was named after one's own father, for example, he could be referred to as …‘father’ rather than by the term for the role he actually filled. Obvious anomalies of this sort could usually be uncovered by a simple question about them. Responses would take the form, ‘He's not my [father] all right, he's my … cousin …’ but I call him [father]because he is named after my father. (Burch 1975: 68–9; emphases added)

In other words, role behaviour is based upon kin classification outside of naming; and the application of terms for close kin to others is logically dependent upon one's close kin relationships.

McKinnon (2005b: 111) invokes an article by Waltner (1996) dealing with Chinese materials as further evidence for ‘a multiplicity of mothers’. But Waltner's contribution has to do with legal and ritual notions surrounding the idea of ‘motherhood’, not with its semantic structure in the first place. Thus she cites a maxim that ‘a concubine has no children and a concubine's children have no mother’ (Waltner 1996: 72), which McKinnon quotes approvingly, without, apparently, realising that the proposition presupposes a ‘mother’ (and a ‘child’) category independent of and logically prior to itself. A far more detailed analysis of Chinese kinship by Feng (1948) shows that there is a handful of ‘nuclear terms’ subject to various modifiers. ‘Each nuclear term’, Feng 1948: 8) notes, ‘possesses a primary meaning and one or more secondary meanings’. Predictably, he gives the primary meaning of mu as ‘mother’ (1948: 9). ‘The primary meaning’, Feng goes on to say, ‘is assumed when the term is used independently’ (1948: 8) – which, it seems to me, is entirely comparable to my own self-report, already noted, that when I speak of ‘my mother’ I mean my genetrix and not my denmother, mother country, etc.⁹

McKinnon's final example of ‘a multiplicity of mothers’ is from her own fieldwork in the Tanimbar Islands of eastern Indonesia. Here men of differently ranked groups sometimes have a relationship designated by an expression which McKinnon renders as ‘elder-younger brothers who treat each other well’ (2005b: 111; see also McKinnon 1991: 100). And she goes on to say (McKinnon 2005b: 111), taking the perspective of a child of one of these men: ‘Because these father's brothers are also one's father’– i.e. are terminologically equated with one's father –‘the wives of these men are therefore one's mothers’– i.e. are terminologically equated with one's mother. But this very statement shows that the status of these other ‘mothers’ (and ‘fathers’) is logically dependent upon their relationship to one's father, who, I shall guess, occupies his focal position through the application of a procreative model. That such a model is not foreign to the Tanimbarese is indicated by another expression, rendered by McKinnon (1991: 117) as ‘true elder-younger same-sex siblings’. This expression sometimes has a wider application, but, McKinnon tells us, ‘it can also be interpreted more narrowly to include only those who have been born of the same mother and father’ (1991: 117). Which is to say that the wider application signals a grey area in semantic space – one presumably like the designation of the mother's sister as ‘mother’ in northeast Arnhem Land.

But this is not at all McKinnon's analysis. She insists instead that ‘[t]he fact of “treating one another well” says it all: the relationship is created and maintained by acts of nurturance and solicitude that constitute the very definition of kinship’ (McKinnon 2005b: 111). Rhetorical considerations aside, ‘the fact of treating one another well’ not only does not say it all; it does not tell us in the first place what native criteria are used in the Tanimbar Islands to determine who gets treated how –‘well’ or otherwise. Similarly, ‘acts of nurturance and solicitude’, far from providing ‘the very definition of kinship’ in this locale, are logically consequent upon a definition of kinship and kin classes derived from other bases. It is relatively clear from McKinnon's data, though obscured by her analysis, that these bases are procreative among the Tanimbarese – as indeed they are probably everywhere else.

McKinnon's contention that adoption defies procreative models (2005b: 112–13), also indebted to Sahlins (1976), has comparable ethnographic and analytical flaws. She cites the Inuit area and Polynesia as regions in which adoption is especially common. But in both areas they are special ‘adopted child’ and/or ‘adopting parent’ terms, made up of the focal ‘child’ and ‘parent’ terms accompanied by a suffix or other linguistic marker (see e.g. Burch 1975: 46; Damas 1972: 43; Guemple 1972: 68; Hooper 1970: 56; Howard et al. 1970: 43). These latter terms are thus derivates of the former, just as, say, godmother is derivative of mother. Behaviourally, moreover, procreative kinship is especially salient in both areas, as Joan Silk has shown in several important contributions (Silk 1980, 1987a, 1987b). Here is her summary of the situation:

The patterns of … adoption in Oceania and the Arctic are strikingly similar. First, in each of these societies, natural parents who give up primary responsibility for raising their children typically delegate care of their offspring to close consanguineal kin. Second, natural parents are uniformly reluctant to give up their children to others permanently, and often express regret at the necessity of doing so. Third, parental investment is not necessarily terminated when adoption … arrangements have been completed. Even after children have left their households, natural parents may maintain contact with them, continue to contribute some resources to their care, and retain their rights to retrieve their offspring if they are mistreated. Fourth, natural parents are often very selective in their choice of … adoptive parents; they typically prefer adults who can offer their children better economic prospects than they can themselves. Finally, there is some evidence of asymmetries in the care of natural and adoptive children, as adopted … children may be required to work harder, may be disciplined more forcefully, or allocated fewer familial resources than natural children. (Silk 1987b: 46)

Towards the end of the ‘multiplicity of mothers’ section of her article, McKinnon remarks that

many people do make the distinction between ‘real’ and other forms of kinship, although who counts as real kin in any particular culture is not always – or even often – defined genetically. Even allowing for such a distinction, however, it is clear that the patterns of nurturance, altruism, and allocation of resources follow from specific cultural classifications of kin relations and cultural understandings of appropriate kin behavior that are never simply reflections of genetic relations … (2005b: 113)

My argument so far has been that, on the contrary, ‘real’ kin/others distinctions are usually if not always defined by local notions of genetic connection. McKinnon's further assertion of a correspondence between kin class and behavioural class – so that, for example, anyone called ‘mother’ in a particular community is supposed to be treated in much the same way as one's genetrix – is the stuff of introductory anthropology textbooks and communitarian fantasy, but, so far as I am aware, it has never been demonstrated even for a single case. But there are several counter-demonstrations (e.g. Goodenough 1951: 111–19; Kronenfeld 1975; Shapiro 1997: 204–7). Keesing (1969) makes the crucial point that the behavioural norms informants present to anthropologists pertain to focal members of kin classes – which is to say that ‘patterns of … altruism’ follow less‘from specific cultural classifications of kin relations’ and more from ‘genetic relation’– the exact antithesis of McKinnon's assertion (see also Peterson 1997; Shapiro 2005b). And this in turn is still more evidence that, when informants talk about kin categories, they have in mind close procreative kin.

In subsequent argument McKinnon deals with ‘the systems of kinship known as unilineal, in which descent is traced either through the male line to constitute patrilineal groups, or through the female line to constitute matrilineal groups. Either way, such a delineation of groups will always entail that some genetic kin will be in other groups while some more distant genetic kin will be in one's own group’ (2005b: 113–14). Her debt, yet again, to Sahlins (1976) is duly recorded, but her presentation needs badly to be repaired, to wit:

First, all this about ‘tracing of descent’ is mostly another example of textbook ‘wisdom’, in reality confined largely to what have been called ‘segmentary lineage systems’ in the Muslim Middle East and parts of Africa. In Aboriginal Australia, by contrast, detailed genealogical reckoning is absent and ‘the tracing of descent’ is replaced by a This World/Other World distinction in which the right hand member of the opposition is assigned ontological, moral, and temporal priority (Shapiro 1979: 13–14). Much the same holds for the so-called ‘descent groups’ of Aboriginal North America (Tooker 1971) and present-day or recent Amazonia (Murphy 1979). Although the matter badly needs attention for other areas, this latter pattern is probably more common than the former.

Second, what we know about segmentation in segmentary lineage systems – that it follows genealogical lines – is entirely consistent with genetic logic. Segmentation in other kinds of ‘descent groups’ has been much less studied, but, for northeast Arnhem Land at least, there is strong indication that the same logic is at work, though concealed by a genealogically minimising ideology (Keen 1995).

Third, whatever descent or descent-like constructs exist in a community, kinship is nearly everywhere reckoned bilaterally, and, as I have suggested, in ways that are largely compatible with genetic notions. Whatever the intentions, McKinnon's wording suggests, quite wrongly, that effective kin reckoning in ‘unilineal’ populations is at odds with these notions.

Fourth, it also suggests, again wrongly, that in such populations these groupings are especially salient, whereas in fact their importance in everyday life varies very considerably ethnographically. When they are nonlocalised, as they often are, they can be said to be ‘groups’ only conceptually. Even when localised, they are never the sole basis for social action (see esp. Keesing 1971; Kuper 1982).

This last point needs to be pursued in view of McKinnon's assertion that ‘[i]n most societies around the world, marriage is governed, in the first instance, by systematic relationships between groups …’ (2005b: 122). As evidence for this assertion she offers a sketchy presentation, from a secondary source, of the four and eight ‘section’ systems of Aboriginal Australia. But the ‘sections’ are in no sociological sense ‘groups’. They are categories in terms of which marital and ritual obligations are sometimes expressed (Shapiro 1979: 70–2), somewhat comparable to Western astrological classes and other conceptual schemes (1979: 72–4).

During the 1950s and early 1960s a very considerable literature grew out of Levi-Strauss' attempt to see presumed marital relationships between patrilineal groups in Aboriginal Australia¹⁰ as constituting ‘elementary structures’ of sociality (Levi-Strauss 1969[1949]: 146–220). The ‘origins myth’ character of Levi-Strauss' scheme makes it suspect as an empirical exercise, as I have argued elsewhere (Shapiro 1998; see also Conkey 1991). But even as ethnographic analysis, we now know that it is hopelessly flawed. An important consideration is that these groups are not localised, as Radcliffe-Brown (1931: 4) had assumed, absent any real evidence, they were (Shapiro 1973). This in itself suggests their unimportance in the politics of marriage. But Hiatt's work in north-central Arnhem Land (Hiatt 1965: 38–44) went further and showed that primary rights to bestow an Aboriginal girl are vested in individuals – not groups – and that these individuals are usually not even members of the girl's patrilineal group. Such groups, he argued, are significant only in the distribution of ritual rights. My own subsequent research further east in Arnhem Land supported Hiatt's argument in every particular (see esp. Shapiro 1981).

McKinnon's own field materials point in a not entirely distinct direction. Tanimbarese patrilineal groups are localised and there are both enduring and ephemeral marital relationships among them, but, as in Aboriginal Australia, individual marriages are arranged by close kin of the bride and groom. Moreover, primary obligations to give and receive bridewealth fall on particular kin and not groups. A man has marital rights to the daughter of his mother's brother – not, apparently, because she is a member of a particular group but by virtue of his kinship position per se (McKinnon 1991: 134–62, 199–258; 1995b).

So, even where unilineal groups exist, they are not necessarily the effective units in arranging marriages; indeed, as Scheffler (1973: 784–6) has argued, they are rarely if ever so. Moreover, any ‘group’ rendition of Third and Fourth World marriage is even more plainly untenable in the absence of anything resembling ‘unilineal’ reckoning, as in much of Amazonia and South Asia. In these areas what is called ‘cross-cousin marriage’ is practised – i.e. marriage between a man and either his mother's brother's daughter or his father's sister's daughter. Now in populations with either patrilineal or matrilineal groups both women are members of units other than a man's own. But in most of these two regions no such groups exist. Hence McKinnon's claim that cross-cousin marriage ‘depends … upon a distinction between one's own group and others’ (2005b: 123) is invalid even without recourse to the actual politics of marriage (see e.g. Gardner 1972; Gregor 1977; Kaplan 1975; Kensinger 1995; Yalman 1962).

Hence not a single one of McKinnon's objections to the handling of kinship in evolutionary psychology is supported by the evidence. The most that can be said is that the genealogical skewing of unilineal descent and Omaha-type kin classification is not predictable from Darwinian principles. But these forms of skewing exist in populations which otherwise act and classify in ways that accord remarkably well with these principles. The distinction between ‘a genetic … calculus’ and ‘a system of social classification’ (McKinnon 1995b: 123), which McKinnon thinks is crucial to her argument, is in fact entirely meaningless.

Conclusions

From all this I think a number of conclusions can be drawn about the constructionist approach in kinship, at least as advocated by Susan McKinnon.

First, this approach – despite its claim to analyse non-Western notions of kinship ‘in indigenous terms’ (Carsten 1997: 292) – is in fact remarkably disrespectful of the principles by which people around the world classify their kinship universes. This stems partly from an astonishing ignorance of focality theory. This latter is a scholarly shortcoming of a very high order – not only because of its long history in kinship studies in particular (see e.g. Malinowski 1929; Scheffler and Lounsbury 1971), but, as well, because similar results, as I noted earlier, have been demonstrated in other areas of classification.

Second, the other main factor that distances many of the new kinship scholars from their own ethnographic materials is a commitment to Marxist theory, especially the hopelessly antiquated fantasies of Engels (1972[1884]) on the origin and development of the family. This works in concert with their ignorance of focality theory to produce a grossly distorted view of Third and Fourth World sociality. Hence McKinnon's concoction of group motherhood, of collective childcare through this and adoption, of the pervasiveness of kinship in human relations, and of the salience of descent groups– though this last probably owes less to Marxism than to other Victorian theories (Kuper 1988). In any case, the fact is that kinship in our species is nothing if not individual, because the bonding that we undergo, especially as children, is socially selective (Flanagan 1999: 40–2). Attempts to collectivise it – whether in Fourth World universal systems of kin categorisation (Shapiro 2005b) or Western communes (Brumann 2003) – have at best a very limited success.

Third, although the new kinship scholars present themselves as comparativists, the time-honoured project of earnest cultural comparison is at best tangential to their main project, which is the belittling of the West, especially Western science and what is sometimes called ‘the traditional family’. These are of course familiar targets for feminists and Marxists (see e.g. Gross and Levitt 1994: 107–48; Tobias 1997: 214–20). Especially pertinent here is McKinnon's recurrent use of the language and tactics of ‘deconstruction’– not to illuminate the social conditions which encourage particular forms of ideology or scholarship but to denigrate these forms. Thus she writes contemptuously of ‘the middle class ideal’ of motherhood (McKinnon 2005b: 112). She asserts that ‘[e]volutionary psychologists presuppose a restrictive understanding of kinship … that is a reflection of Western upper-class concerns’ (2005b: 117). She refers to their causal imageries¹¹ as ‘stories’ (McKinnon 2005a: 7) and ‘myths’ (2005a: 2) and concludes that ‘their science is ultimately a complete fiction’ (2005a: 4). But this is far too grave a conclusion. Whatever biographic or social considerations underlie the work of evolutionary psychologists, they have produced a series of testable propositions about human behaviour. Some of these propositions may need to be modified or even discarded (see esp. Buller 2005), but this is true of any scientific enterprise. A far more cogent case can be made that it is McKinnon's social environment as an academic, with its doctrinaire Marxism and feminism, that is projected onto ethnographic materials to which, as I believe I have demonstrated, it is entirely foreign. The result of this projective process is a Manichean anthropology based on the concoction of an Individualist West versus a Collectivist Rest.¹²

Fourth, and related to all three of my previous conclusions, there is a salient link between the ignorance of focality theory and the remarkable hostility to the traditional family in the new kinship studies. For if, as I have argued, close procreative kin are probably everywhere distinguished, the suggestion is that these kin participate in special relationships that are very nearly universal and not, pace Marxism, the dispensable product of a particular socioeconomic regime. This is of course just a restatement of conclusions reached by Malinowski (1913), Lowie (1920: 147–85), and others nearly a century ago, but apparently the lesson needs to be relearned.

Fifth, the new kinship scholars view human affairs as part of an extrasomatic process which has little if anything to do with Homo sapiens as a biological species. They thus draw upon a long tradition of ‘biophobia’ in social theory (Daly and Wilson 1988). But their view of biology is antediluvian: they equate biological causation with the reflex arc, whereby only one outcome is predetermined, and they show no awareness of contingency sensitivity in biological systems (e.g. Oyama 1985; Pinker 2002; Shapiro 2008). Hence McKinnon (2005b: 127) claims that evolutionary psychology insists upon ‘universal forms of behavior’– quite unaware, apparently, that all its propositions rely on probability calculi. And she declares that ‘the mind is a flexible and creative tool capable of creating diverse cultural forms’ (2005b: 127) – a proposition she advances to counter evolutionary psychology but which is in fact assumed by all Darwinian students of human affairs with whose work I am familiar. What this latter group of scholars seems to share is a concern with establishing the limits of this flexibility – more particularly, with how certain elements may or may not be combined in the generation of cultural forms. This in turn invites us to reconsider the supposed antagonism between structure and freedom. One could cite Chomsky here, or Levi-Strauss, but, at least in professional anthropology, the apical ancestor is A.L. Kroeber's remarkable 1909 article, ‘Classificatory systems of relationship’ (Kroeber 1909). What Kroeber argued here is that the human mind is capable of isolating and combining certain elements, such as gender and collaterality, so as to generate a fairly large but finite number of systems of kin classification, and, more, that these systems are mostly independent of institutional influence. There is more than a passing resemblance between this piece of prescient brilliance and the arguments presented here, as well as a very considerable corpus of literature in both cognitive science and Darwinian anthropology (Shapiro 2008). This being so, it is fallacious to present ‘deconstructionism’ as a freedom-promoting alternative to ‘biological determinism’.

In recent decades we have been invited to choose between a vision of anthropology as a science and one of it as an art form – or, more specifically, as a branch of hermeneutics. The new kinship studies suggest that a more vital choice nowadays is between anthropology as a child of Enlightenment scepticism – and the consequent requirement to demonstrate the truth of a proposition – on the one hand, and, on the other, anthropology as a branch of collectivist dogma.

Footnotes

• 1

  I prefer expressions like ‘models of procreation’ or ‘procreation models’ (Yeatman 1983) to ‘genealogical’ (etc.) because the latter suggest the sort of extended genealogies found in the Bible and ‘the tracing of descent’ of introductory anthropology textbooks. In fact, as I note later, such genealogies are relatively restricted ethnographic phenomena, whereas models of procreation are probably universal.

• 2

  Actually, Schneider sometimes went further, insisting that the procreative model is a construct of kinship studies, and that (other) native English-speakers employ a model in which procreative relationships are neither sufficient nor necessary (Schneider 1968, 1972: 49–56, 1984: 92 et seq.). I argue in the body of this paper that the former assertion is false. On the latter, see Scheffler (1976).

• 3

  A quarter century ago I tried to do so (Shapiro 1982), but this was before the politicisation of kinship studies. For a more recent effort on my part, see Shapiro (1995a: 201–88).

• 4

  This is metaphorical usage. In semantic research a distinction is often made among primary, secondary, and metaphorical members of classes, but I find it unnecessary to do this here.

• 5

  There are comparable – and complementary – symbolic expressions of paternal creation and nurturance (see Shapiro 1981: 16–20, 2005b: 51).

• 6

  Most of the foregoing analysis of the semantics of ngama has appeared elsewhere (see Shapiro 1981: 87–92, 2005b: 51).

• 7

  In an earlier publication Feinberg (1981) provides examples which require modification, but not abandonment, of these conclusions.

• 8

  It might be argued that ‘little’ is simply the opposite of ‘big’ and does not signify nonfocal status. But this is contrary to what we know of systems of kin classification in general (Scheffler 1987: 214–16) and, indeed, of human categorisation at large, wherein focality is usually associated with indications of superior size or age. Vide English ‘How tall are you?’ and ‘How old are you?’ (Kronenfeld 1996: 95–7). ‘How young are you?’ is derived and an obvious sop to the elderly.

• 9

  I have confirmed this analysis with two of my own Chinese-speaking informants – Louisa Schein, Associate Professor of Anthropology at Rutgers University, and Ching-I Tu, Professor of Asian Languages at the same institution. Professor Schein has carried out ethnographic fieldwork in parts of southwest China. Professor Tu is a native Mandarin speaker. He translates sheng mu, the birth mother, who both McKinnon and Waltner treat as just another member of the mu class, as ‘prototypical mother’.

• 10

  For the sake of simplicity, I refer to Aboriginal Australian ritual groups as ‘patrilineal’, because the father/child link is probably the single most important principle of recruitment to these groups. But the expression misleads – this for two reasons. First, within such groups there is only a very limited ‘tracing of descent’, group unity being based on other principles (see above). Second, this link is probably everywhere on the continent, most extraordinarily in the Western Desert (see esp. Myers 1986: 129–30), supplemented by other principles. This is also true of Tanimbarese ‘houses’ (see below) – though for men the father/child link is the most favoured principle of recruitment (McKinnon 1991: 84–106, 1995b). For this reason I refer to these units as well as ‘patrilineal’.

• 11

  I borrow the useful expression ‘causal imagery’ from Stinchcombe (1968), who uses it without explicit definition.

• 12

  This is not to argue that Marxism has no insights into particular historical or ethnographic situations. I wish only to highlight here its apocalyptic quality, something widely appreciated in certain scholarly circles (see esp. Campion 1994: 425–53). Nor do I have any quarrel with the idea that women should have equal access with men to highly-placed jobs and other life opportunities. My quarrel here is with the blatant anti-family stance of much feminist writing.

Acknowledgement

This essay is dedicated to the memory of my father, Charlie Shapiro, from whom I seem to have inherited, by Darwinian and perhaps other means, an utter inability to put up with falsity and pretence. I want also to thank Herb Damsky, Tom Gregor, Adam Kuper, Tom Parides, and Mel Spiro for their encouraging remarks on earlier drafts of this article, though I need to add that responsibility for it rests with me alone.