Part four: Social category systems
p. 95-199
Full text
1In the previous chapter, I used ethnographic examples to show how kinship is articulated in an Aboriginal Australian society. However, in the Western Desert, as in many Australian societies, there is a further important layer in this domain: social categories. These social categories are compatible with kinship but must be distinguished from it since they lie at a different level of abstraction from the nature of a relationship. They constitute abstract or normative groups of people with identical features but, unlike lineages or clans, do not generate enduring corporations. They work as a superstructure above kinship and are a useful device for labelling people in generic ways, by making the identification of kin categories simpler, and by facilitating relationships and exchanges among different groups, ‘tribes’ or dialects. Well known to anthropologists, they constitute to some degree an Australian particularity. They are called ‘sections’, ‘subsections’, ‘semi-moieties’ and ‘moieties’.
2To fully understand these social categories and distinguish them from kinship, some general theoretical background is necessary. Kinship in general is an egocentrici recognition of relationships inside the web of relatedness built on filiation and marriage. ‘Egocentric’ means that whenever one needs to apply or understand terminology, behaviour, classification of relatives and so on, a person has to be defined as Ego, the speaker or point of reference. In contrast, social categories are sociocentricii: it is not necessary to have an individual’s point of view to establish and understand these categories since they are generally applicable, even though, as we shall see, they really become useful when, within these sociocentric groups or categories, we adopt an egocentric point of view. We need to spend some time discussing the notion of category since it will help us to better understand the social categories we are talking about here.
What is a category?
3A category is a collection of things that have some characteristics in common. In the natural world, for example, a set of trees could be seen as one category of plants, TREES, even though there are different types of trees. The word TREE thus becomes a category along with the category BUSH, GRASS, VEGETABLES or whatever types of plants a culture and language distinguish from other types. The first important point is that categories are made out of choices since there are many possible ways to group things. Even so-called ‘scientific’ categories are in fact arbitrary since there is no fundamentally objective reason to group together animal species, for example, in certain ways rather than in others. Rather than having REPTILES, BIRDS, MAMMALS etc., we could have categories such as ANIMALS THAT FLY (birds, flying foxes, bats, insects, etc.), ANIMALS THAT SWIM and so on.
4Categories are relative and depend on a culture and language. Remember, for example, the colour classification among the Hanunóo of the Philippines discussed earlier. Imagine there is a pack of black and white dogs and a pack of black and white cats. You could quite well organise your categories according to colour, and not distinguish dogs and cats. You could define the two categories as ‘black domestic animals’ and ‘white domestic animals’, for example. In either case, you eliminate certain criteria in favour of others. In the blackwhite classification you eliminate the fact that dogs bark and cats do not; in the dog-cat classification you eliminate the colour. Categories thus favour certain features or aspects over others. They are a simplification of the world.
5This leads us to the third important point. Categories may have subcategories, and subcategories within a category share common features but distinguish themselves from other subcategories through other features. I call this the rule of inheritance between categories: a subcategory inherits the fundamental features of its mother category. Both GRASS and TREES inherit the fact that they belong to plants and share features that every plant shares. However, TREES, as opposed to GRASS, have stems and branches. The structural principle of categories is hierarchical. The organising principles are inclusion, contrast and even opposition.
6Since categories are arbitrary and thus relative to particular cultural contexts and languages, the criteria or features which are inherited from category to subcategory and that contrast categories and subcategories are culturally meaningful. They are related to other social values and structures. If living beings are classed together according to colour, for example, we may expect ‘colour’ to have a specific cultural meaning and importance.
7Some argue against this proposition, claiming that there are objective criteria available to define categories that do not necessarily have a specific cultural meaning or value (but see Sahlins 1977 who contrasts this conclusion). If we return to the dog-cat/black-white example, it could objected that dogs never mate with cats and that a dog, whether white or black, will never produce a cat. We may indeed conclude that the world around us proposes a certain number of criteria that could potentially be applied in category-making, but again there is no absolute ‘objective’ reason why we should privilege mating over colour or reproduction over certain morphological aspects. These problems and questions apply to Australian Aboriginal social categories as well.
8Like ‘categories’ in general, a social categoryiii is a set of individuals who have something in common and who contrast with other people. There are social subcategories, and some of the features of each such subcategory are inherited from its mother category. Besides the characteristics indicated for categories in general, note that two other important features must be added to Australian social categories: firstly, they are dual or binary in that a social category never appears on its own but always with its counterpart(s) or fellow categories. Together, they constitute what is called a social category system. Secondly, a social category system is encompassing, that is every member of a group or society is born into one of the categories according to rules defined by the system. No exception is made to this inflexible rule.
9There are several types of social category systems in Australia: generational moieties, moieties, semi-moieties, sections and subsections. Semi-moieties are rare, limited to the Golf of Carpentaria and perhaps existing formerly in the southwest of Western Australia, and quite complex to understand so I will not describe the workings of this system. We now turn to the other systems.
Generational moieties: general principles
10Generational moieties divide society into two categories. Remember that generation and age are different concepts. The age of a person depends on date of birth, whereas the generation is the place of a person in a series of filiations. To understand this, I discuss a situation in which there are only very few people and follow them through their generations.
Figure 24: Age and generation

11Richard and Henry are brothers. Both married at the same time. Richard married Esther and Henry married Elizabeth. Both had children. Richard and Esther had a girl called Janice. Henry and Elizabeth had a boy called Michael. Janice married when she was 20 years old and she had a girl called Emma. But Michael married only when he was 40 years old and he had a boy called George. Emma is thus at least 20 years older than George because her mother had her when she was 20. Therefore, if you take only age into account, Emma could be George’s mother. Emma and George however have something in common, some shared feature. Despite their age difference, they are both grandchildren of brothers: they are on the same generational level.
12Generational moieties are built around these generational levels, distributing all people of a society and beyond into two categories. Richard, Esther and their granddaughter Emma, as well as Henry, Elisabeth and their grandson George, are of the first generational moiety while Janice and Michael are in the second generational moiety. Generational moieties include members of alternate generational levels: a person is always in the same generational moiety as his brothers, sisters and cousins, his grandparents and his grandchildren. A person’s children and parents, on the other hand, are always in the other generational moiety. This is why they are sometimes called ‘alternate generational moieties’. We will examine this concept in some detail since the principles at work in generational moieties are also at work in section and subsection systems.
13We draw a square in which, say, all the people of a society are included. For the sake of making the demonstration easier, let us say that there is only one family in this society, composed by parents and their children.
14Inside the square, we draw a line separating the family into two halves. In the upper square M1 we assemble the parents, in the lower half M2 we regroup the children. We call M1 and M2 the two generational moieties because they divide the family (society) into two parts according to generational level membership.
Figure 25: Generational moieties: splitting into two groups

15We now decide on the following three rules:
Parents and children will never be in the same moiety.
There are only two moieties in the society.
All the people of a society have to be in one of these moieties.
16If these rules are followed, we must then ask: what happens with the children of M2, the children of the children? We cannot subdivide again since this would go against rule number two, and we cannot put them into M2 since this would contradict rule number one. The solution is of course to put the children of M2 back into M1 since this is the only way to follow all three of the rules. Hence, the children of M1 are M2 and the children of M2 are M1.
Figure 26: Generational moieties: parents and children

17Let us begin again. First, we have parents in M1 whose children are in M2. These children become parents themselves (horizontal arrow) and have children who must be in M1. These children become parents themselves and have children who will be in M2 and so on. If you follow this circle over a few generations, you see that a person is always in the same moiety as his grandparents and grandchildren while his or her parents, children, greatgrandchildren and great-grandparents will be in the opposite moiety. As we have seen when discussing terminologies, two important types of categories underpin Australian Aboriginal kinship systems: generations and gender. Generational moieties are organised around the former. Sections and subsections, as we shall see, add gender categories to the system.
Generational moieties in the Western Desert
18Without doubt, generational moieties are the most important category system in the Western Desert and many other Australian languages. They are usually named, but the recognition of such generational levels, whether named or not, is a structural feature of all Australian social category systems since they intrinsically underlie the kinship system. This characteristic was recognised by some of the earliest writers, among them Mathews (19031904:61; see Lawrence 1969 [1937]), who presumably coined the expression ‘alternating generations’ later adopted by Radcliffe-Brown (1930-31:443), albeit without acknowledgement. It is safe to hypothesise that generational moieties have been known in the Western Desert for much longer than other category systems, such as sections and subsections, because they are found throughout the desert and because their names are very similar, if not identical, from one group to another. This hypothesis is based on the general assumption that common words among dialects are older than words that are not shared, and thus point to the original language from which the dialects evolved.
19Western Desert people sometimes refer to persons of the same moiety as mobs, indicating a certain unity or identity. It is this unity or identity among people of the same generational level that may have led Elkin and others to misinterpret the Aluridja kinship system. From a sociological point of view, persons of the same generational moiety consider each other as brothers and sisters, as opposed to their fathers and mothers and their daughters and sons, who are in the opposite moiety. This principle is extrapolated to terminological usage. While a cross-cousin is always distinguished from a parallel cousin, these cross-cousins can also be considered as brothers and sisters and called by the same terms because they are co-generational.
20Alternate generational levels penetrate deeply into everyday social practice. Expectations of sharing in accordance with social rules and norms are stronger between persons of opposite moieties, where the relationship typically includes at least some restraint or even avoidance, than between persons of the same moiety. The unity or identity between persons of the same moiety, on the other hand, leads to behaviour that does not usually reflect relations of hierarchy or of opposition, but rather of reciprocity and identity. Laughren (1982:77) reports for the Warlpiri of Central Australia that ‘men’s sporting teams were formed according to this division’. Tonkinson (1991:75-76) notes for the Mardu that the alternate generational divisions are important in ceremonial activities as well, where ‘the two groups sit a short distance apart, and throughout the proceedings their members joust verbally with each other in loud and light-hearted fashion’. Stanton (1984:168) similarly explains for the Mount Margaret area that it ‘is the division of alternating generation levels which is of greatest significance in ritual activities’. As Ronald and Catherine Berndt formulated it,
Within a person’s own generation level are to be found, to some extent at least, ‘equals’: brothers and sisters, cross-cousins, age-mates and so on. The generation level above him includes those with some authority over him, directly or indirectly: father, mother, father’s sister, father’s sister’s husband, mother’s brother, mother’s brother’s wife, perhaps mother-in-law, father-in-law and so on. Deference, and in some cases avoidance, are relevant here (Berndt & Berndt 1992 [1964]: 87).
21In the Western Desert, these generational moieties are given two kinds of names: the first type is relativeiv, the second type is absolutev. Absolute names remain identical regardless of the position of the speaker. Relative names, on the other hand, depend on the speaker’s position. Among the Ngaatjatjarra, the names are as follows:
Relative terms for generational moieties | ||
Term | nganantarka | tjanamiltyan |
English gloss | us, we bone, we group | them, they flesh, they group |
Absolute terms for generational moieties | ||
Term | Tjintultukultul | Ngumpaluru |
English gloss | Sun side | Shade side |
22According to the absolute terminology, a person is born into one of the two moieties, for example Tjintultukultul (Sun side), and his or her membership of this moiety will never change. With the relative terminology, however, a person is always nganatarka (‘us’) for himself, but his or her parents consider themselves nganatarka as well. A person’s parents and children are in the tjanamiltyan moiety (‘them’), but they themselves will consider this person to be in tjanamiltyan from their point of view. In other words, relative terms work reciprocally: I call myself Y and the other person X, and the other person does the same thing. The tables and maps below show the distribution of absolute (Figure 27) and relative (Figure 28) terms across the Western Desert. They are reproduced from Dousset (2005).
Group / source | ABSOLUTE TERMS | |
Moiety name 1 (translation as in source) | Moiety name 2 (translation as in source) | |
Kukatja (Balgo) White 1981 | Maru | Libi (Nyilga) |
Mt Margaret Stanton 1984:169 | Ngumpulurutja (Shade) | Tjinturltawakulpa (Sun) |
Ngaanyatjarra Douglas 1977 | Tjirntulukultu (Sun side) | Ngumpalurru (Shade side) |
Ngaatjatjarra Dousset DATE?; Tindale 1963:39 ( according to Tindale the terms are identical in Warburton and in Laverton) | (Tjintultukultul / Tjindulakalanguru (Sun side, sections Karimarra and Purungu) | Ngumpaluru / Wiltjalanguru (Shade side, sections Tjarurru and Panaka) |
Pitjantjatjara (at Uluru) Harney 1960:63-4 | Djindarlagul | Wumbularu |
Pitjantjatjara (at Yalata) White 1981 | Biranba (White-light) | Maru (Black-dark) |
Wiluna Sackett 1978.Mainly Mandjildjara people | Djirndulu (Sun side) | Ngumbaluru (Shade side) |
Figure 27: Distribution of absolute terms of generational moieties in theWestern Desert

Group/ source | Own moiety as in source | Other moiety as in source |
Antakarinja Tindale 1965:48 | Nganandarka | Jianamiltjan |
Kokatha (SA) Tindale (1965:45) | Nganantarka | Tarbula / Tanamiltjan |
Kukatja (Balgo) White (1981, following pers. comm. Peile) | Maringara | Yinyarungura |
Mardu (Jigalong) Tonkinson (1991) | Marirra | Yinara |
Mt Margaret Stanton (1984:168-9) | Nganatarka | Tarputa |
Ngaatjatjarra Gould (1969:106) | Nganatarka | Inyurpa / Tarputa |
Ngalea (Nullarbor Plain) Tindale (1965:31) | Nganandaraka | Tjanamiltjan (Tarbudu) |
Pindiini / Wonggai Tindale (1965:70) | Nganatarka | Tanaamiltjan |
Pintupi Myers (1986) | Nganarnitja | Yinyurrpa |
Pitjantjatjara Goddard (1992), Elkin (1938-40:213) and White (1981) for Yalata | Nganantarka | Tjanamilytjan |
Tjalkadjara (north of Laverton) Tindale (1965:91) | Nganataruka | Tjanamiltjan |
Yankunytjatjara Goddard (1985) | Nganantarka | Tjanamilytjan |
Figure 28: Distribution of relative terms of generational moieties in the Western Desert

Patrimoieties and matrimoieties
23When anthropologists talk of ‘moieties’ and not ‘generational moieties’, they understand something different from what we have discussed above. We have seen that ‘generational moieties’ divide society into two horizontal categories. They are horizontal because they cut across families and groups to unite people at the same genealogical level. ‘Moieties’, however, are understood to divide society into vertical categories, grouping lines of people following descent and filiation. There are two types of such moieties: patrimoieties and matrimoieties. The former is found in societies where patrifiliation is of importance, the latter where matrifiliation is dominant. We have seen that generational moieties have a close relationship with kinship terminology and with behaviour towards relatives. We will now see that patrimoieties and matrimoieties have a close relationship with principles of descent and of inheritance.
24The working principle of moieties is that belonging to a moiety is inherited from generation to generation via descent. In the case of patrimoieties, children belong to the moiety of their father but only the father’s sons transmit the patrimoiety to their own children. In the case of matrimoieties, children belong to the moiety of their mother but only the mother’s daughters transmit the matrimoiety to their own children. Another important distinction between generational moieties and other moieties is that the former are endogamous while the latter are exogamous. This simply means that a person has to find a spouse within his or her own generational moiety since one cannot marry a person who is classified as ones mother, father, aunt, uncle, son or daughter. In patrimoieties or matrimoieties, people of the same moiety are considered to share a substance or essence that is inherited from generation to generation. In other words, people of the same moiety are alike, and they cannot marry another alike person. Members of a patrimoiety or a matrimoiety must therefore marry a member of the other patrimoiety or matrimoiety.
Figure 29: Patrimoieties in a genealogy

25We have two patrimoieties: the red moiety and the yellow moiety. In a patrimoiety system, moiety membership is inherited through the male line. Members of a patrimoiety choose a spouse from the other patrimoiety. Thus, members of the red moiety marry people from the yellow moiety and vice versa. We begin reading the figure with man Number 1. He is considered to be the oldest known male ancestor of the red moiety and as such is the ancestor of the red moiety. He has married a woman from the yellow moiety. This woman, like other wives and husbands, is not represented in the figure to avoid too much complexity. You need to imagine each person’s spouse. Because inheritance here goes through the males’ line, the children (2 and 3) of man 1 are of the red moiety. Let us follow the boys’ line: number 2 has a son number 4, who has a son himself, number 8. They all belong to the red moiety, of course. Number 2 also has a daughter, number 5. This daughter will have to marry a man from the yellow moiety. Her children, 10 and 11, will inherit their membership from their father, who is of the yellow moiety.
26We now move back up to the sister of number 2, who is number 3. She is of the red moiety, but her husband belongs to the yellow moiety. Again, since belonging is inherited through the males’ line, the children of number 3 are of the yellow moiety. Understandably, the children of number 6 are of the yellow moiety. The children of number 7, on the other hand, since she will have to marry a person from the red moiety, will be of the red moiety as well.
27The figure below represents the same principles as they operate in a matrimoiety system. We could have changed the figure so as to have a woman as the ancestor of the moiety. However, to facilitate comparison between patrimoieties and matrimoieties, I have taken the same genealogical structure and simply recoloured people according to their membership of a matrimoiety.
Figure 30: Matrimoieties in a genealogy

28No detailed discussion is necessary here since by now you should understand the working principles of moieties. Let me just mention a few steps: the children of number 1 are yellow because number 1’s wife is yellow and she is the one transmitting membership to the couple’s children; number 4 is red again since number 2’s wife is red; number 8 is yellow again since his father, number 4, married a yellow woman and so on.
29To conclude these short explanations on moieties, let us recall some of the rules that structure them:
Moieties divide society into two halves.
Membership is genealogically inherited. In the case of patrimoieties, membership isinherited from ones father; in the case of matrimoieties, membership is inherited from ones mother.
Moieties are always exogamous. Marrying someone of the same moiety as oneself isconsidered incestuous.
30Patrimoieties and matrimoieties are well represented in Australia. Matrimoieties are found in the Perth area, in Arnhem Land and in South Australia. Patrimoieties are found in Arnhem Land, in Cape York, in southern West Australia, in the Kimberleys and so on. Many groups combine patrilineal and matrilineal features. The Dieri people, for example, have matrilineal and patrilineal units called madu and bindara. A person has a special relationship with the patrilineal totem of his or her mother and the matrilineal totem of his or her father (Berndt and Berndt 1992:53). When discussing sections and subsections, which are widespread in Aboriginal Australia, we shall see that they are a combination of the principles and rules present in generational moieties, patrimoieties and matrimoieties at the same time.
31These moieties are inherent in Australian social organisation as a structural feature. But they also have concrete applications. Moieties are not well distributed in the Western Desert, with the exception of the Pintupi at the central eastern edge of the desert who have most probably inherited them from their Warlpiri neighbours quite recently (see Meggitt 1986) and the Mardu at the north-western edge of the desert. The Pintupi, according to Myers (1986), use two unnamed patrimoieties. Although unnamed they distribute people according to role and responsibility. For every site in space, there are owners (yarriki or kirta) and managers (kurtungurlu). Owners have primary rights while managers have secondary rights to sites in space. Managers have the duty to clear the grounds and paint the dancers, who are the owners, before ritual activity on a particular site. Managers and owners are of opposite patrimoieties, and they have a shared responsibility over the land. Members of both moieties have rights over these sites and over rituals, but they have different responsibilities during the rituals concerning a particular site in space. Similarly, according to Tonkinson (1991:75-77), the Mardu recognise two patrimoieties which play different roles in some rituals and that have influences in camping arrangements. They are egocentrically termed as marndiyarra (‘my own patrimoiety’) and yarigirra (‘the other patrimoiety’).
Sections: a formal presentation
32The last type of social organization we are going to discuss here are sections and subsections, as anthropologists name them, but often referred to in English as skin names by Aboriginal people themselves. I shall use the word sections and subsections because these better represent what they are really about. Indeed, you can think of a section as a part of a cake. In the case of sections, the cake has been cut into four equally sized pieces. In the case of subsections, it has been cut into eight equally sized pieces. Putting the pieces back together reconstructs the entire cake. We will have a detailed look at sections first. Subsections can be considered as a variety, or as a refinement, of sections subdividing each piece into two according to a few rules.
33Sections and subsections are widespread in Australia and are used nowadays in an increasing number of languages. In the Western Desert, sections have spread widely as well within the last few hundred years. Western Desert people call them skin name in English, but more appropriately yinni (also meaning ‘name’), miri (also meaning ‘skin’, ‘skin colour’) and yara (also meaning ‘symbol’ or ‘metaphor’).
34Like generational moieties and moieties, sections are social categories: every person belongs to a specific section and all the people in the same section have something in common. You will easily understand sections if you understood generational moieties as they are a subdivision of them, as well as patrimoieties and matrimoieties since they are a combination of these: if one divides a society into two patrimoieties and two matrimoieties at the same time, the result are four pieces or sections. One valid hypothesis is to think that section systems emerged in areas where neighbouring groups had patrimoieties and matrimoieties and where these groups started exchanging and intermarrying.
35Let us have a look again at what we explained for generational moieties. All the people of a community, dialect or tribe are represented in one large square. This square is divided into two generational moieties M1 and M2, where M1 are the parents and children of M2 and M2 the parents and children of M1. So far, this reflects exactly what we have seen for generational moieties.
Figure 31: Generational moieties again

36We have seen that generational level is one of the central characteristics of Australian Aboriginal kinship systems. When we discussed terminologies as well as patrimoieties and matrimoieties, we saw that the other central characteristic is gender: bifurcation and merging which is gender-based in terminologies, and membership to a moiety with respect to transmission by gender. We now need to combine generation and gender to construct the section system.
37Inside each generational moiety, we begin by distinguishing fathers from mothers. Following the basic rule of incest prohibition, people who marry each other should not be identical; they should be distinguished. But what happens with their children? Should they be distinguished according to gender? In principle they could be, but let us recall that these children, who are in M2, will themselves become parents of M1. This means that if we want to distinguish fathers from mothers in M2 just as we did in M1, then we need to group brothers and sisters and distinguish them from their spouses.
Figure 32: Generational moieties: distinguishing fathers from mothers

38Now, since brothers and sisters are considered identical (in our case they are in M2a) and if we want to make this applicable to all elements of the system, then we need to add to the mother’s part (M1a) her brother, and to the father’s part (M1b) his sister. The result is that M2b, the son’s wife and the daughter’s husband are themselves brothers and sisters, which is something we would have expected, having understood the Dravidian terminology and the direct exchange system that underpins marriage in such a system. We have now constructed a section system.
Figure 33: The section system

39If we read this figure in a different way, I as Ego am in M2a, so my brothers and sisters must also be in M2a. My spouse and cross-cousins will be in M2b, my mother will be in M1a and my father in M1b. Since my spouse’s parents are classificatory mother’s brothers and father’s sisters, my father-in-law belongs in the same category as my mother, M1a, and my mother-in-law is in the same category as my father, M1b.
40However, as was the case with generational moieties, the figure and the section system itself work without any specific Ego, because they are sociocentric. Indeed, M2 are also parents of M1. We can thus reread the figure in the following way:
Any M1a person marries an M1b person, and any M2a person marries a M2b person.
Any child of an M1a mother will be M2a, and any child of an M2a mother will be M1a.
Any child of a M1b mother will be M2b, and any child of an M2b mother will be M1b.
Any child of an M1a father will be M2b, and any child of an M2b father will be M1a.
Any child of an M1a father will be M2a, and any child of an M2a father will be M1b.
41The system is so perfect that it is also compatible with patrimoieties and matrimoieties, as I had already foreshadowed. Using patrimoieties and matrimoieties to number and represent sections makes them even clearer. If we ignore the M1 and M2 labelling, which we drew from generational moieties, we can reframe the section system using patrimoieties and matrimoieties without changing its internal structure. Let us say there are two matrimoieties called A and B and two patrimoieties called 1 and 2. Each individual is a member of one matrimoiety as well as one patrimoiety. For example, a person is A1, meaning that he or she belongs to the matrimoiety A and to the patrimoiety 1. Remember that people cannot marry within their moieties. This means that a person A must marry a person B and that a person 1 must marry a person 2. However, since every individual is now a combination of letters and numbers (A1, A2 etc.) because every person is member of both a patrimoiety and a matrimoiety, people need to be exogamous to both moiety types. In other words, an individual A1 must marry an individual B2: change from A to B and from 1 to 2.
42Also, we now know that one inherits membership of the matrimoiety from ones mother, and membership of the patrimoiety from ones father. Children of A are A, of B are B, of 1 are 1 and of 2 are 2. Again, if we combine this, since each person has a letter and a number, we can see that children of an A1 woman, who is married to a B2 man, are A2: they inherit the letter from the mother and the number from the father.
Figure 34: Matrimoieties in section systems and patrimoieties

43Each individual is thus at the intersection of two overlapping categories: a matrimoiety and a patrimoiety. Each spouse of each individual also has the combination of the two contrasting categories. Each individual has inherited one category from the mother and one from the father. The diagonal lines represent the relationship of filiation between fathers and their children; the vertical lines represent the relationship of filiation between mothers and their children. What is important to understand is that these matrimoieties and patrimoieties may actually not be named or recognized by languages and groups, but the principle that underpins them is implicit in the section system.
44The usual figurative representation of these relationships in section systems is symbolised below. The equal sign between sections mean that they intermarry. I have also added on the right side of the figure the section names used by the Ngaatjatjarra to illustrate that they are not merely abstract constructs but have real counterparts (the father-child diagonal arrow has been omitted, as is usually the case in these figures).
Figure 35: Formal and actual representation of the Ngaatjatjarra’s section system.

Sections in the Western Desert
45As is the case with generational moieties, sections have names, which are frequently used where this system is in place. Among some tribes, such as the Warlpiri or the Arrernte of Central Australia, section names are used systematically to call or name a person (Glowczewski 1991). Below is a table with the section names used by some dialectal groups of the Western Desert. The table is to be read as follows: Column 1 is the mother of 2 or the father of 4. Column 2 is the mother of 1 or the father of 3. Column 3 is the mother of 4 or the father of 2. Column 4 is the mother of 3 or the father of 1. Note that Pintupi people use subsections today. Their section names in the table were recorded in 1932 by Fry (1934), before the Pintupi adopted the subsection system.
1 | 2 | 3 | 4 | |
Ngaatjatjarra | Karimarra or Milangka | Tjarurru | Purungu | Panaka or Yiparka |
Ngaanyatjarra | Karimarra or Milangka | Tjarurru | Purungu | Panaka or Yiparka |
Ooldea | Garimara or Milanga | Dararu | Burong | Banaga or Ibaga |
Pintupi | Purukula | Tararu | Purunga | Panaka or Yiparrka |
Mardu | Garimara | Milanga | Banaga | Burungu |
46
Misinterpretations of sections and subsections?
Sections and subsections have been labelled ‘marriage classes’ by former anthropologists such as Lévi-Strauss (1967) because they obviously conform to (some of) the marriage rules at work. Today, scholars regard this conception of sections as erroneous. Sections are not marriage classes and thus cannot be taken as the sole element determining marriage partners; nor do they embed and express marriage rules, but they are a useful general guides to expected behaviour towards kin. Those who are potential spouses will be in the appropriate section. A woman or girl always has the same section as her mother’s mother, and a man or boy always has the same section as his father’s father. This has sometimes been labelled indirect filiation or indirect descent. Remember that filiation is the relation a person has with his or her mother and father and descent is the sum of these filiations over generations. We need to be careful with the notion of indirect filiation or descent in Australia since filiation and descent are associated with the transmission of rights and obligations. This may not always be the case in Australia and is certainly not the case with sections. Thus section systems should not be interpreted as governing, organising or structuring social elements such as marriage, filiation and descent. Von Brandenstein (1970, 1972, 1978) suggests that sections reflect an emic and universal conception of classifying and transmitting humours or temperaments and that rules between sectionsin fact reflect universal genetic-like rules. In fact, though, sections are simply a metaphoric overlay on the kinship system. They are a higher-level abstraction over the kinship terminology. In this context, it is notable that the proper name for the section system among the Ngaatjatjarra is yara, meaning ‘symbol’ or ‘metaphor’.
47Most Western Desert groups, in contrast to Warlpiri or Arrernte people and their subsection system, do not use section names in daily interaction. However, people use them to narrow down kinship relationships in cases where the application of the relational triangle, which we discussed earlier, is difficult or impossible. In other words, the section system is not used within a community, where everyone knows each other, but in contexts in which individuals from different communities, or even different languages, meet. Sections are a ready and easy reference guide for kinship classification, as Fry (1933: 267) explained. The applicability of the section system in contexts where ‘foreigners’ meet is one reason for its rapid diffusion throughout the desert.
48What now follows is an example with a difficult genealogical relationship, which shows how the section system is applicable in such situations to narrow down actual kinship relationships. We will use the Ngaatjatjarra section terms, illustrated in Figure 35 above. We use abbreviations in the figure to make reading easier.
Figure 36: Example of the application of the section system to a genealogical grid

49A man, indicated as Ego in this figure, encounters a girl (far right). They do not know how they are connected in genealogical terms, and you can see that the genealogy is quite complex: she is Ego’s sister’s husband’s father’s female cross-cousin’s husband’s sister’s daughter. Also, imagine they do not know a common third person to whom they could apply the relational triangle. Ego, however, immediately knows that the girl is a sister, a parallel cousin, a mother’s mother or a daughter’s daughter, a grandmother or a granddaughter. He knows that she is not a mother-in-law, which would have been problematic because he would have had to avoid her; he also knows that she is not a crosscousin and thus a potential partner. Ego knows all this because he is of the Karimarra section and, meeting the girl, he asks what section she is in. The girl answers that she is Karimarra as well. The figure above retraces the section name for each genealogical position so that you can follow how this can be verified using the section schemas of Figures 34 and 35, as well as the schema above the genealogy of Figure 36. Remember that vertical lines connect mothers to their children, that the equal signs means marriage and that sections that stand diagonally connect the father with his children. Ego is A1 and thus his sister is as well. A1 marries B2. The mother of B2 is a B1, and her husband is A2. The mother of A2 is A1 and so on. Had the girl said she was in the Purungu section, he would have known that she was a cross-cousin. Had she said Tjarurru, he would have known she was a mother or a sister’s daughter. Had she said Panaka, he would have known she was a father’s sister.
Subsection systems
50Subsection systems are not widespread in the Western Desert, being limited to its eastern edges among the Pintupi and the Luritja people. However, it is strongly represented among Central Australian groups such as the Warlpiri and the Arrernte, as well as further north towards Darwin. While based on rules identical to those of the section systems, the subsection system adds a further distinction between first cross-cousinsvi and second cross-cousinsvii.
Figure 37: First and some second cross-cousins (1st cc designates first cross-cousins; 2nd cc designates second cross-cousins)

51Basically, with first cross-cousins the genealogical link is in the first generation above the cousins, while with second cross-cousins the link is two generations above. Figure 37 shows that, predictably, first cross-cousins are your mother’s brother’s children and your father’s sister’s children. Second cross-cousins are your mother’s mother’s brother’s daughter’s children (MMBDS, MMBDD) and your father’s father’s sister’s son’s children (FFZSS, FFZSD). It is important to understand that the figure represents only actual cross-cousins but that this principle is extrapolated to include classificatory cross-cousins as well. Indeed, any first parallel cousin of my second cross-cousin is a first cross-cousin for me and any second parallel cousin of my second cross-cousin is also a second cross-cousin for me, and so on. This reckoning becomes highly complex after a few steps, so subsections are useful in such cases.
52The rules of subsection allocation are such that the subdivision of sections into subsections produces a single integrated system rather than two parallel section systems. Rules of subsection allocation are somewhat different and certainly more complicated than is the case with the section system. However, it is not absolutely necessary to fully understand the functioning of subsection systems in order to apply these principles in the field, and even Aboriginal people themselves at times have difficulties explaining exactly what is happening. What you need to remember is how the different subsections are connected to one another. Relationships then flow automatically.
53The basic idea is as follows: because first and second cross-cousins are distinguished, their mothers and fathers have to be distinguished too. The mother of a second crosscousin is your mother’s first parallel cousin’s brother’s child. Therefore, what in a section system is your mother’s and mother’s brother section will, in a subsection system, be divided into your mother, mother’s sister and mother’s brother as a first half, and as a second half, your mother’s first parallel cousin and her brother. The children of the first half are your first cross-cousins while the children of the second half are your second cross-cousins. In a formal subsection system (theoretically, at least, because in reality there are often some variants of the following rule), you may only marry persons of your second cross-cousins category but no first cross-cousins. In reality, though, while second cross-cousins are the preferred choice, first cross-cousins are an acceptable alternative.
Figure 38: Distribution of relatives within the eight subsections

54Just as was the case with sections and moieties, subsections have names, and the way of representing subsection systems follows the ideas defined for the section system: the equal sign links intermarrying subsections and vertical arrows link a mother’s subsection to her children’s subsection. The names of subsections are additionally distinguished by gender. For each subsection there is a masculine and a feminine name. For example, in the case of the Pintupi, Luritja and Warlpiri subsection system, Tjakamarra and Nakamarra are two names for the same subsection, with Tjakamarra designating a male and Nakamarra a female. In reality then a subsection system has sixteen rather than eight names.
55Because the division of a section system does not produce two independent section systems, but a complete new system, the link between a mother and her child is no longer reciprocal. This means that the arrow does not go back from the daughter to the mother but goes first through other subsections before coming back in the fourth generation to the starting point.
Cycles in subsection systems
Subsection systems have two matricycles of four generations each and four patricycles of two generations each. You can easily verify this by drawing a little genealogy, giving Ego any subsection name as a starting point, then using the schema of Figure 39 below. In the following steps, you attribute to each person the corresponding subsection using the same schema. You will see that it takes two generations to find a person in direct line to Ego with the same subsection as Ego’s father, and four generations to find Ego’s mother.
Figure 39: Subsection names of the Pintupi and Luritja people

56Figure 39 has to be read as follows. Say, for example, that you are a Nakamarra woman. You would marry a Tjapaltjarri man. Your children would be Nungurrayi and Tjungurrayi, and they would marry Nangala women and Tjangala men. Their children would be Nampitjinpa/Tjampitjinpa if their mother is Nungurrayi and Napaltjarri/Tjapaltjarri if their mother is Nangala.
57People who live in areas where section and subsection systems co-exist, such as in the north-east and central eastern part of the Western Desert, use both systems and terminologies. One section simply corresponds and is equivalent to two subsections. Below is a table summarising these correspondences between the eastern section system (Ngaatjatjarra and Ngaanyatjarra) and the Pintupi and Luritja subsection system. As you can see, there are sometimes even linguistic similarities between the names, such as Karimarra and Tjakamarra or Nakamarra, testifying to a common historical origin.
Sections | Subsections (male) | Subsections (female) |
Karimarra, Milangka | Tjakamarra Tjampitjinpa | Nakamarra Nampitjinpa |
Tjarurru | Tjungurrayi | Nungurrayi |
Tjapanangka | Napanangka | |
Purungu | Tjapaltjarri Tjapangarti | Napaltjarri Napangarti |
Panaka, Yiparrka | Tjupurrula Tjangala | Napurrula Nangala |
58A few concluding remarks about subsections are necessary. I have already stressed that sections are little used within a community where members know one another. Sections are not precise enough to distinguish kinship categories sufficiently to make them an everyday tool for behaviour. The situation is somewhat different with subsections, where eight categories become sixteen when gender-marked names are counted. Because of the subsection system’s greater capacity for precision in the classification of relatives, it is used frequently for terms of address and reference, as well as to replace personal names, even among members of the same community, in particular among Warlpiri, Arrernte, Luritja and Pintupi people.
Where did sections originate?
59The diffusion of section systems into the eastern Western Desert region is relatively recent and reached completion in the 1930s or 1940s. It superimposed itself onto a social organisation that up until then had been dominated by generational moieties only. According to Tindale (1988 [1972]: 254), the Pitjantjatjara only became familiar with the section system in 1933 and even today, although Pitjantjatjara people know how to use section systems, they are certainly less familiar with them than their immediate neighbours to the west, the Ngaatjatjarra and Ngaanyatjarra. Berndt and Berndt (1992 [1964]: 47) report that sections diffused through the southern part of the Western Desert in the 1940s.
60A notable characteristic is that the names of the sections (and subsections) are not identical among the groups possessing this form of social categories. So, although the structural principles, the way the system works, is known and has diffused over thousands of kilometres, the terminology associated with it has not always been adopted and, in some cases, it has been considerably modified. What this means is that just because you know the names and relationships between names for a section system found in one group, you cannot automatically apply it to another group. You first need to check if the section names are organised in the same way and, if not, you must discover the equivalence rules between the section names used by different groups. It may well be that, as a result of the system’s diffusion, the Karimarra section in group A became Purungu in group B and vice versa.
61I have undertaken an extensive study of these routes of diffusion and of the transformations that have occurred in the section system from one group to another (Dousset 2005). It is a complex study, so I confine this discussion to a few summary remarks and conclusions.
62The section system seems to have followed two major routes of diffusion into the Western Desert. One route began in the Pilbara where the section names Karimarra, Panaka, Paljeri, Purungu were used and, to the south of it, Milangka. The other route was from the south-west of Western Australia, where Tjarurru and Yiparrka possibly originated, into the Western Desert. The south-west is also the possible origin of the western section system (McConvell 1990, 1996), a system that must have moved northwards initially and then in a north-easterly direction into the desert at a later stage.
63The north-western or Pilbara-Kimberley system followed two diffusion routes, one parallel to the coast and southwards along the desert, the other south-eastwards to the Yulparija and the Pintupi peoples. The south-western system went straight into the desert along with elements of the north-western system. Before penetrating the Western Desert itself, the north-western or Kimberley-Pilbara terms moved along the coast to a point in the south where they encountered new terms or another system. This meeting may have enforced the system’s diffusional capacity and given it new strength, probably because of migratory movements and cultural affinities. The section system then moved into the desert.
Figure 40: General view of the overlap of trade routes and cultural diffusion in the Pilbara, Kimberley and Western Desert regions

64The diffusion routes of the section system have been summarised here to show the dynamics of Aboriginal kinship and culture in general. Languages and dialects had in the past, and still have today, the capacity to adapt to new contexts, new organisational models and new kinship terms and categories. Kinship and social organisation are dynamic domains because they are among the vital enablers of human interaction. If the section system diffused over such large areas, it means that groups and languages are not isolated from one another but in constant interaction and exchange. Cultural diffusion with the section systems as an object can be linked to the existence of wider exchange systems, comprising three major constituents: trade routes, mythological tracks of the Tjukurrpa (Dreaming) and migrational patterns. While this exchange system criss-crosses and covers all areas of the Western Desert, some branches are clearly dominant, being the major arteries for the flow of goods and ideas, if not people. These branches lead to important entry points at the desert’s edges which are, clockwise, the region south-west of Kalgoorlie, Anna Plains and surrounding areas in the Pilbara, the area of Balgo south of the Kimberleys, the Rawlinson-Musgrave axis in the east and Ooldea in the south-east. An additional portal of entry with a Kintore-Papunya axis must have operated from the 1930s onwards at least, following Pintupi people’s eastwards drift and their adoption of subsections. These portals can, with reasonable probability, be considered as constituting the principal entry points of ideas, goods and elements of social organisation, including sections. Coincidentally, they are situated along some of the limits determined by Peterson’s (1976) drainage basin model.
65This historical dimension has implications for our understanding of the ‘Western Desert cultural bloc’, a question I began to address in Part 1 of this book. What was a debatable concept when Berndt (1959) proposed it and has since then been taken for granted by Western Desert specialists has now benefited from additional evidence: the cultural bloc is a dynamic ensemble of groups with local identities but situated within a larger entity, which is not solely an anthropological artefact but is also ethnographically proved and historically grounded. Western Desert culture emerges from the dialectic between local identity and global similarity or embeddedness, a dialectic that simultaneously promotes vast networks and exchange systems and facilitates the incorporation of goods and ideas travelling along these systems into local social structures and cultures.
Summary of Part Four and concluding remarks
66In this chapter we have considered another important aspect of the domain of kinship in Aboriginal Australia, social categories, which, along with categories in general, reflect a world view, a way of classification, as seen from a particular cultural perspective. Social categories divide society into an even number of units (two, four or eight), each with certain characteristics that generate the principle of social classification. In Aboriginal Australia, all the various types of social category systems are based on identical factors that are or are not deployed in any given category system: gender and generation, which when combined produce a third factor, ‘crossness’, which is the result of bifurcate merging.
67Generational moieties, the first social category system described, divide society into two halves. The first half, or moiety, includes a person and his or her close or classificatory brothers, sisters, cousins, grandparents and grandchildren with their brothers, sisters and cousins. The other moiety includes a person’s parents and children, along with all their close or distant brothers, sisters and cousins. Generational moieties organise elements of everyday and religious life; more than this, though, they are the key to understanding Western Desert kinship.
68Moieties, in particular patrimoieties and matrimoieties, are particularly important for our understanding of the other system types, sections and subsections, because they latently or overtly structure these systems. Moieties, which divide society into two halves, are organised around a principle of descent. Membership of a moiety is determined either by the mother’s (matrimoiety) or father’s (patrimoiety) membership.
69Where social category systems exist, every person is born into a section or a subsection, just as every person is in a generational moiety. Relations between sections or subsections are organised around rules of marriage (cross-cousin marriage) and of filiation (both, matrilineal and patrilineal). A person’s membership derives from the memberships of his or her parents. Subsections, which divide society into eight categories, or sixteen if one considers the gender difference indicated in each subsection name, are frequently used within communities to address and refer to other individuals and are even used as personal names. This is far less the case with sections. There are possibly two reasons for this situation: the first is that sections are not precise enough to delimit actual kinship relationships between two individuals. There are about 20 categories in the kinship terminology, which are narrowed down into four categories only in the context of sections. This is problematic for an individual since kinship categories are a guide to appropriate behaviour between people. The second reason is the positive consequence of the first: if sections are not prominent within a community, yet are nevertheless used and appreciated, it must be because their utility becomes clear when different communities meet. In these cases, where visitors or people who rarely meet encounter one another, sections are a good enough approximation of the general type of expected behaviour between individuals. Sections can be considered a vehicle, a means for inter-community and inter-language communication. The analysis of the diffusion of the section system has confirmed these assertions: section systems are, in the first place, a tool for interaction between neighbouring and distant groups and indicate the existence of extensive social networks.
Bibliography
DOI are automaticaly added to bibliographic references by Bilbo, OpenEdition’s bibliographic annotation tool. These bibliographic references can be downloaded in APA, Chicago or MLA formats.
Format
- APA
- Chicago
- MLA
Further reading
Some general books on social organisation
BARTH, Fredrik (ed.) 1969. Ethnic groups and boundaries: The social organization of culture difference. Oslo: Scandinavian University Books.
BOHANNAN, Paul and John MIDDLETON (eds) 1968. Kinship and Social Organization. Garden City, New York: The Natural History Press.
10.1086/212357:FIRTH, Raymond 1971. Elements of Social Organization. London: Tavistock [1951].
LOWIE, Robert H. 1950. Social Organization. New York: Holt, Rinchart and Winston.
Some books on Australian social organization
BRANDENSTEIN, Carl G. von 1982. Names and Substance of the Australian Subsection System. Chicago and London: The University of Chicago Press.
BURRIDGE, Kenelm O. L. 1973. Encountering Aborigines: A Case Study, Anthropology and the Australian Aboriginal. New York: Pergamon Press.
FISON, Lorimer and Alfred W. HOWITT 1991. Kamilaroi and Kurnai. Canberra (Melbourne): AIATSIS (Robertson) [1880].
KEEN, Ian 2004. Aboriginal Economy and Society. Melbourne, Oxford, New York: Oxford University Press. MADDOCK, Kenneth 1972. The Australian Aborigines. A Portrait of their Society. Ringwood - Victoria: Penguin Books.
SHAPIRO, Warren 1979. Social Organization in Aboriginal Australia. New York: St. Martin’s Press
TURNER, David H. 1980. Australian Aboriginal Social Organisation. Canberra, Atlantic Highlands: AIAS, Humanities Press.
Endnotes
i A term or relationship that indicates an individual’s point of view (see Sociocentric).
ii A term or relationship seen from a social or sociological point of view (see Egocentric).
iii A way of classifying people in some form or another into distinct groups. The expression is usually applied to Australian Aboriginal sections, subsections, moieties and generational moieties. Each of these category systems divides society into an even number of complementary classes based on kinship relationships.
iv A term or name that depends and varies on the speaker. Examples are relative generational moiety names: moiety A calls itself X and calls the other moiety Y, and moiety B calls itself X and the other moiety Y (also see Absolute terms).
v Terms and relationships that do not depend on the speaker and that work like names. Absolute names are in use in moieties or sections where each moiety or section has its specific and invariable name (also see Relative terms).
vi This is a synonym for cross-cousins but is distinguished from second cross-cousins.
vii A type of cross-cousin relationship in which Ego and his/her crosscousin are two generations removed; e.g., a MMBDD is a second cross-cousin.
Only the text can be used under the OpenEdition Books License license. Other elements (illustrations, attached files) are “All rights reserved”, unless otherwise stated.
Australian Aboriginal Kinship
An introductory handbook with particular emphasis on the Western Desert
Laurent Dousset
2011