Kinship, theology and deep grammar

One of the most salient paradoxes in the study of kinship systems is their sheer analytical complexity, from the point of view of an external observer, and simultaneously the ease by which those very same systems are assimilated by the natives themselves. Whereas no special training, costly rituals, harsh indoctrination seems to be needed to master the intricacies of kinship terminologies and marriage rules for those who are born into them, the situation for the anthropologist seems to be the very opposite. What could be the reason for this apparent inconsistency, simplicity for the native and complexity for the stranger? My hunch is that the paradoxes of kinship run very deeply into the nature of what human kinship is all about. Far from solving (or dissolving) this paradoxical nature, the purpose of this text is to contribute to its clarification by means of a comparison between kinship, language and religion.

Let us start with the comparison between kinship and religion. In the scientific study of religion a crucial distinction should be drawn between popular religion and what we might call erudite or systematic theologies. Whereas there is no such a thing as a ‘religious system’ in what concerns popular religion, the systematic character of theological knowledge is all too apparent (I’d like to thank Benson Saler for drawing my attention to this important point). In popular religion we normally have a rather fragmentary and quite often inconsistent or incoherent set of beliefs and practices. Very little explicit or formal instruction is needed to elicit this set of practices and beliefs, for they manage to colonise human minds through an inferential process instead of explicit semantic indoctrination (Boyer 1994; Sperber & Wilson 1996). Popular religion cannot be said to form a ‘cultural system’ in any way. Neither is it ‘systematic’ nor ‘cultural’ stricto sensu. Even though some exposure to a particular cultural environment is certainly needed for any form of popular religiosity to emerge, this is very far from the explicit and costly cultural indoctrination needed to assimilate fully fledged cultural systems such as science and, needless to say, theology.

Now would kinship systems fall closer to the ‘theology’ or to the ‘popular religion’ pole? Kinship systems are certainly ‘systematic’ as the name implies. They do not consist of a bunch of fragmentary beliefs and practices but they clearly possess an internal logic. So at first sight they look very much like theologies rather than popular religions. But then here is the paradox: it takes long years of harsh study in theological faculties or seminaries to assimilate all the theological knowledge of a particular religion, whereas nothing of that is needed for the assimilation of a kinship system, I insist, for those who are born into it. In other words, despite being clearly systematic they do not seem to be very ‘cultural’. Would that mean that kinship systems are in any way innate? Or, rather that we possess some sort of innate deep kinship grammar that enables humans to learn their respective kinship systems, in spite of the poverty of the associated cultural stimulus? Here is where the comparison with language is instructive.

Consider the case of the grammatical rules of human languages. At first sight, we are confronted with a similar situation: despite the outstanding complexity of those rules, children learn their native languages with no special difficulty. Of course, ever since Chomsky we know that this paradox as far as language is concerned is more apparent than real. Children do not have major problems in learning their native languages precisely because they do not have to learn those complex grammatical rules for they have them already hard-wired into their brains. The problem lays not in learning those rules but in learning them consciously, since we already possess them as unconscious knowledge. Now could the very same thing apply to kinship systems? Do we have a kinship grammar in the same way as we have a language deep grammar? Thus, it would be the transformation of such grammar into a set of explicit rules what would make the study of kinship systems so hopelessly complex to the foreign anthropologist, but not its unconscious assimilation by the native. In other words, the anthropologist’s role would be equivalent to that of the linguist, who approaches the study of a natural language by means of sophisticated formal methods acquired through years of intense training, whereas those born into a particular kinship system would correspond to ‘mere’ native speakers of that natural language.

But if there is a deep grammar for kinship systems – a kinship module – the questions arises as to its evolutionary origins. This is different from the evolutionary origins of language deep grammar, for it could be cogently argued that the ability to learn and use a language clearly increased the inclusive fitness of our hominin ancestors (Pinker 1994) – whatever might have sparked the initial mutations in their preverbal forebears, a controversial issue in itself but which can be set aside in the present discussion. Notice that kin selection is of no help here – and this is a remarkable fact that needs to be emphasised. Kin selection could have been operative in language evolution by increasing the reproductive fitness of a mutant female who, thanks to her enhanced verbal skills, could improve the communication with her offspring, and vice versa. But the same argument cannot be applied to kinship, for biological and cultural kinship are different, as anthropologists have been tirelessly emphasising for eons, since the very beginning of the discipline of social anthropology. But the question is what should we make of such difference? One way of addressing this issue is exemplified by Dunbar’s reassessment of Hughes seminal sociobiological approach to the evolution of human kinship (Dunbar 2011; Hughes 1988). There is a lack of correspondence between biological and social (cultural) kinship, Dunbar concedes, but this lack of correspondence is not random. ‘If there is any degree of consistency between social and biological kinship, no matter how small, then, form an evolutionary point of view, investing in one’s social kin will have the consequence, on the long-term average, of investing in one’s biological kin. Evolution is a statistical process, not a deterministic one, and a very great deal of statistical slop can often be tolerated’ (Dunbar 2011: 137; see Hughes 1988: 130-131 for a parallel argument).

Let us pause for a moment: investing in one’s social kin has the consequence, on the long-term average, of investing on one’s biological kin. But how could that be? Take the case of so-called classificatory kinship terminologies, which is the type of kinship terminology that we normally find in tribal societies. In a bifurcate merging system I use the term ‘mother’ to refer to my biological mother and to all her parallel female cousins. But whereas I share fifty percent of my genome with my biological mother, my degree of genetic relatedness with, say, my mother’s fourth cousin is practically equivalent to what I share with any other member of the human species! And yet my kinship system makes me call the two women by the same term. I fail to see how any ‘statistical slop’ could give a selective advantage to any hominin who called ‘mother’ his mother’s fourth parallel cousin, specially taking into account that, as Dunbar himself recognises, after a few generations relatedness becomes ‘smeared out’ across the whole population (2011: 142). No matter how many people or how much they invest on their mother’s fourth cousin, their investment will not revert to their biological mother in the long run. It does not seem to me that statistical reiteration could make these kinship terminologies adaptive in any way. So we are back to the paradox mentioned above: kinship systems seem to be ‘cultural’ despite the fact that they are not assimilated through any specific process of cultural indoctrination.

An apparent way out of this dilemma would be to postulate that there is no domain-specific kinship module in the human mind but that kinship systems are merely the product of the activation of domain-general cognitive principles, that is, those also prevalent in other systems of classification. This was the thesis put forward by Kemp and Regier (2012), according to which a kinship terminology is a system of classification resulting from the trade-off between two principles: simplicity and informativeness. The simpler a particular system of kinship categories is the less informative it will be (the more different kinship types will be subsumed under the same category), and conversely, the more informative (the more categories we have or the closer they are to actual biogenetic relationships) the more complex it will be. In all probability, kinship systems do actually conform to this model of classificatory logic. Kemp and Regier compare kinship systems with systems of colour classification precisely to illustrate how the same domain-general principles apply to both. Hence kinship terminologies would be just an instance of the general capacity of human minds to classify things, be these colours or human beings. But this domain-general approach misses a very important characteristic of human kinship and the comparison with colour classification turns out to be particularly informative in this sense.

Whereas we all see colours, even though we might not classify them in the same way, nobody can see his or her kin! Again, kin selection plays no role here, for even if humans might have some innate device to detect individuals with whom they share a substantial part of their genome, their kin in biological terms, these very often do not coincide (too often to be mere exceptions to the rule) with the members of their kinship systems, their kin in cultural terms, as we have seen with the case of classificatory kinship terminologies. In other words, the members of a kinship system are created by the very system they belong to, something which cannot be true of a system of colour classification. A very similar point has been recently made by Dwight Read (2012: 30-31): ‘The constructed system of kinship is not an elaboration upon, nor emergent from, a system of biological relations. Instead, it is the essence of what we mean by a cultural idea-system—a constructed (in the sense that it is not modeled on what is extrinsic to us) conceptual system with an internal logic that ensures common understanding and meaning for that system of concepts by those who share it in common through enculturation’.

Certainly, once again we seem to get closer to the ‘theology’ pole than to the ‘popular religion’ pole. But notice how peculiar is this process of enculturation, how different it is from the enculturation of theology or any other standard cultural system. This is a quasi-natural process of enculturation in which individuals learn a system of cultural categories, their kinship system, with the same ease with which they learn their natural languages, but without any kinship-acquisition device. Everything looks as if kinship systems were the ‘missing link’ in human biocultural evolution: the half-biological half-cultural connection that would enable humans to transcend the limits on social life imposed by natural selection, i.e. the limits on cooperation imposed by reciprocal altruism and kin selection, but without the need for a fully-fledged cultural system – such as that of moral gods (Norenzayan 2013) – with its costly rituals and harsh indoctrination. Perhaps the old Lévi-Straussian dictum that human kinship embodies the transition from nature to culture was not so far fetched after all.


Boyer, Pascal. 1994. The Naturalness of Religious Ideas. A Cognitive Theory of Religion. Berkeley: University of California Press.

Dunbar, Robin. 2011. ‘Kinship in Biological Perspective’. In N.J. Allen, H. Collon, R. Dunbar and W. James (eds.) Early Human Kinship. From Sex to Social Reproduction. Oxford: Blackwell Publishing, 131-150.

Kemp, Charles and Terry Regier. 2012. ‘Kinship Categories Across Languages Reflect General Communicative Principles’. Science 336: 1049-1054.

Hughes, Austin L. 1988. Evolution and Human Kinship. New York: Oxford University Press.

Norenzayan, Ara. 2013. Big Gods. How Religion Transformed Cooperation and Conflict. Princeton: Princeton University Press.

Pinker, Stephen. 1994. The Language Instinct. New York: HarperCollins.

Read, Dwight W. 2012. How Culture Makes Us Human. Primate Social Evolution and the Formation of Human Societies. Walnut Creek: Left Coast Press.

Sperber, Dan and Deirdre Wilson. 1996. Relevance: Communication and Cognition. Oxford: Blackwell Publishing.



  • comment-avatar
    Olivier Morin 13 May 2014 (14:06)

    Thanks for a quite thought-provoking post. Let me quickly react to your claim that biological kinship is invisible to us, and that, as a consequence, kinship systems as we conceive of them have absolutely nothing to do with it. Though common in social anthropology, this view seems much too strong to me.

    We all agree (I think) that kinship terminologies do not do a very good job of tracking biological kinship — or, to put it differently, they would do a terrible job of it if that was their main function; but they have many other functions — legal, social, political — up to and including the creation of patently fictitious kinship ties for the purposes of alliance-building. It is tempting to conclude (as is often done) to the complete irrelevance of biology to the study of human kinship. This conclusion, however, comes a little too quickly.

    It forgets that there is much more to kinship cognition besides kinship terminologies. We have many ways of tracking these relations besides naming them. Just like color terminologies do not exhaustively describe color cognition (and often give a very misleading picture of it, suggesting that some peope lack certain concepts or certain distinctions that they do master) the study of kinship terminologies gives an impoverished picture of our cognitive capacities in this area.

    It forgets that some kinship ties, at least, are represented and tracked with a high degree of accuracy both in humans and in other primates. Critiques of human behavioral ecology, in the wake of Marshall Sahlins, have relished pointing out the exceptions and disregarding the rule, but we humans are indeed very good at knowing who our fathers, mothers and grandmothers — better, indeed, than our primate forebears. The error rate is most often of the kind that kin selection can forgive. (For evidence, let me refer to Bernard Chapais’ recent synthesis, Primeval Kinship, OUP, 2008.) Kinship is visible, because kinship-related behaviors are. The fact that kinship vocabularies may often fail to reflect this is not a good argument against this view (cf. my first point).

    I applaud this refreshing revival of Lévi-Strauss’s ambition — to redefine kinship as the nexus of nature and culture; yet, to start on the right footing, it might be useful not to dismiss biological kinship outright, as so many previous inquiries have done, thus losing sight of the “nature” part of the problem.

  • comment-avatar
    Chris Kavanagh 14 May 2014 (08:32)

    Thanks for the interesting post! I certainly agree that the fluency with which even young children can navigate, what appear to an outsider, as bewildering kinship taxonomies is striking but I remain unconvinced that this provides evidence of some mental kinship module which is entirely divorced from biological kinship.
    An article just published by Warren Shapiro titled “On the alleged disjunction between cultural and darwininan understandings of human kinship” in the Journal of Cognition and Culture seems directly relevant to the points raised here. In the article Shapiro presents quite a compelling case that the supposed lack of relevance of biological kinship to kinship terminology has been exaggerated and fails to take account of the ability to distinguish between concepts which can be referred to with the same or similar constructs but remain semantically distinct. For example, a stranger approaching you in the street to ask for money and referring to you as “brother” can still presumably distinguish you from their actual biological kin. And while that example might seem a bit flippant I think the central point remains important and that we need to be very careful to distinguish between explicit kinship systems and the ability for people to recognise, for example, their biological mother and who else was born from said mother.

    Incidentally, although its by no means my area of expertise, I believe that it is inacurate to present Universal Grammar, in the sense of having innate linguistic knowledge, as if it is established fact since as far as I am aware the theory remains highly controversial and has been subject to harsh critiques in recent years (see for instance, Evans & Levenstein, 2009).

  • comment-avatar
    Dwight Read 21 May 2014 (08:19)

    As I discuss in (Read 2012), the evolutionary pathway from the non-human primates to Homo sapiens involves a major change in social systems from the systems of face-to-face interaction that non-human primate societies are dependent upon to the relation based social systems of human societies. The driving factor for this qualitative transition stems (among other factors) from the increase in social complexity introduced by the phylogenetic trend in the non-human primates towards increased individualization of behavior, including the formation of coalitions. This trend also involves another trend towards behavior specified more with respect to behavioral capacities, rather than specific behaviors. It is no coincidence that in the great apes we find individualistic behavior comparable to what we find in humans coupled with the beginnings of traditions – patterns of behavior transmitted phenotypically rather than genotypically. What the chimpanzees and other great apes lack, though, is the means to incorporate individualistic behavior into their social milieu without reverting to small social units as a way to deal with the social complexity introduced through individualistic behavior and coalition formation. Chimpanzees live in communities of 50 – 100 or so individuals, but within this community females are largely solitaire in their behavior (except, obviously, for mating and raising of offspring) and males only form small, unstable social units composed of 6 or fewer males.

    What our hominin ancestors were eventually able to cognize, due to increase in mental capacity (including an increase in the size of working memory [Read and van der Leeuw 2008]), was the idea that a relation of relation is again a relation. The macaques apparently cognize (in some sense) a mother relation; that is, they distinguish female/offspring dyads according to the difference between the ‘mothering’ behavior of a female towards her offspring and the behavior she may have towards the offspring of other females (Dasser 1988). At some point in hominin evolution leading toward Homo sapiens, our hominin ancestors had the cognitive capacity to be able to recursively construct a new relation from an already known relation through something like Theory of Mind: if I recognize female A as mother, then I can imagine that female A recognizes some female B as mother, and so I have a relation to female B, namely the ‘mother of mother’ relation.

    This construction of a new relation from an already known relation provides the link taking us from biological foundations to the beginnings of a constructed (hence cultural) system of kin relations we refer to as genealogical relations. From the recursive construction of new relations based on recognition of a mother relation and, reciprocally, a child relation (and, eventually, a father relation), another revolutionary change took place with the shift from system of genealogical tracing from one person to another to the symbolic, terminological system that enables us to specify a kin relation between two persons directly. Thus English speakers can say that Sam is the grandfather of Sue without reference to any intervening persons. Common to all kinship terminologies is the ability not only to specify kin relations in this manner, but the ability to compute kinship relations from the terminology itself. As numerous ethnographers have reported, when person A meets unknown person B, all they need to determine their kinship relation is a person C for whom each has a known kinship relation. For English speakers, if A meets B for the first time and A and B determine, through their conversation, that A refers to person C as uncle and person B refers to person C as father (hence C refers to person B as son), then A know that to refer to B as cousin from that information alone. As Laurent Dousset (2008) has noted for the Ngaatjatjarra of Australia, this kind of calculation by A and B suffices for A to work out his or her kinship relation to every person in B’s group since each of those persons has a kinship relation to B expressible through a kin term that B uses to refer to that person.

    This implies that the evolutionary pathway from non-human primate social systems to human social systems begins, as it must, at the level of biological kin, but ends, through the symbolic kinship terminology systems that enable the computation of kin relations by reference to the logic of the kinship terminology, at a constructed, cultural level. The terminology, as I have shown for wide range of terminologies (Read 2013), not only seems to be logical, but is logically coherent — as it must be for this kind of computation to be the basis (as it is) of working out kin relationships. Strikingly, we can now account for the great division of terminologies into what Morgan referred to as classificatory terminologies (that is, terminologies that do not consistently distinguish lineal from collateral genealogical relations) and descriptive terminologies (those that consistently make that distinction) by which of two possible ways that sibling relations are conceptualized. One way, familiar to English speakers, is through sibling constructed as ‘child of parent’ (where this refers to the kin term product of the terms child and parent, not to genealogical relations) and the other in which sibling is defined by common parentage. The latter implies that sibling is a primary, not a constructed, kin relation; the former that sibling is a constructed kin term. The latter provides the logic that gives rise to the classificatory terminologies; the former the logic that gives rise to the descriptive terminologies.

    What makes all kinship systems easy to learn from the child’s perspective is that all terminologies are based on the same generative logic, but differ at the level of the implementation of that logic, such as whether sibling is a primary kin term or a derived kin term. What makes the classificatory terminologies difficult for English speakers is that the mapping of the terms for a classificatory terminology onto genealogical relations, makes no sense from the perspective of a descriptive terminology, which has a structure that roughly parallels that of a genealogical structure. English speakers try to make the comparison between the terminologies through genealogy, which is confusing, rather than through the structural logic of the two kinds of terminologies, which clarifies the differences between the two kinds of terminologies.

    What kinship systems may share with languages is the capacity of the brain to work out structural logic, either in the form of a non-consciously understood grammar in the case of languages, or in the form of the computational logic (which is much like the logic of an addition table or a multiplication table for numbers) for kinship terminologies. Whether this is better understood as a ‘module’ or simply as part of the general capacity of the brain I leave to others more knowledgeable than I about these matters.

    In sum, I quite agree with Salazar’s comment: “Everything looks as if kinship systems were the ‘missing link’ in human biocultural evolution: the half-biological half-cultural connection that would enable humans to transcend the limits on social life imposed by natural selection, i.e. the limits on cooperation imposed by reciprocal altruism and kin selection, but without the need for a fully-fledged cultural system…” Kinship does not require a ‘fully-fledged cultural system’, for with the evolution of kinship systems we have the evolutionary origin of a cultural system.

    Dasser, V. 1988. ‘Mapping Social Concepts in Monkeys’. In: R.W. Byrne & A.Whiten (eds) Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans. New York: Oxford University Press, 85–93.

    Dousset, Lawrence. 2008. ‘The ‘Global’ Versus the ‘Local’: Cognitive Processes of Kin Determination in Aboriginal Australia’. Oceania 78:260-279.

    Read, Dwight W. 2012. How Culture Makes Us Human. Primate Social Evolution and the Formation of Human Societies. Walnut Creek: Left Coast Press.

    Read, D. 2013. ‘A New Approach to Forming a Typology of Kinship Terminology Systems: From Morgan and Murdock to the Present’. Structure and Dynamics 6 (1).

    Read, D., and S.E. van der Leeuw. 2008. ‘Biology is Only Part of the Story …’. Philosophical Transactions of the Royal Society B 363:1959-1968.