Levi-Strauss’ theory of kinship is built in some important respects
upon Radcliffe-Brown’sanalysisof Australian kinshipsystems,although
this is not asexplicitlystated in Levi-Strauss’writing as it might be.The
British Structural Functionalist’s influence is apparent from the fact
that Levi-Strauss adopts Radcliffe-Brown’s three forms of cross-cousin
marriage (bilateral, matrilateral and patrilateral) as his basic typology;
he re-analyses Australian kinship systems in the first of the ethnographic sections of his book The Elementary Structures of Kinship (first
published in France in 1949, second edition 1967, English translation 1969), and his theoretical position is frequently framed by
placing it in opposition to Radcliffe-Brown’s. Like Radcliffe-Brown,
L^vi-Strauss is interested in the life of social systems, not individuals,
and the needs of the individual aresubordinated to the alleged needs of
the system. One of the shortcomings of Radcliffe-Brown’s typological
approach noted by Leach was that there is no apparent limit to the
number of types and subtypes of society that can be devised (see
chapter 2). Levi-Strauss set out to demonstrate that there were logical
limits to the number of types among what he termed ‘elementary
locialsystems’.
Mauss’ theory that exchange perpetuates social relationships was
equally influential upon Levi-Strauss’ work (see chapter 4), but LeviStrauss argued that the structure created by exchange was itself determined by the structure of human thought, developing the ideas of
Hertz and van Gennep. Levi-Strauss’ work follows closely earlier
Dutch studies of kinship in Southeast Asia, especially Van Wouden’s
analysis of the practical consequencesof different types of cross-cousin
marriage and their representation in myth (Van Wouden 1968 [ 1935] ).
Structuralist theory explains the structure of society as the product of
ideas rather than the material conditions of existence.
Levi-Strauss pointed out that cross-cousin marriage is a phenomenon which occurs in many parts of the world: not just Australia (as
among the Aranda,‘Murngin’ and Kariera), but lowland South America
(the Yanomamo and others), Southeast Asia and India. Levi-Strauss
accepted that this could not have occurred through diffusion of the
custom from a common point of origin, and saw it as an expression of
universal patterns of human thought. Durkheim and Mauss had
attributed the origin of logical thought to the experience ofstructure in
the segments of a compound society. Levi-Strauss reversed this hypothesis, and argued that it was the structure of human cognition which
generated structure in social relationships. He argued that the
exchange of gifts and marriage partners were forms of communication
and should be treated like language, the best-studied medium of
human communication. While regarding the most basic structures of
cognition as universal, Levi-Strauss accepted Durkheim’s caution that
supposedly universal psychological mechanisms could not explain
human cultural diversity, and he interpreted the content of structural
thought as the property of specific cultural traditions, paralleling the
enormous diversity of languages.
Radcliffe-Brown’s theory of kinship was that relationships were
built outward from the nuclear family, increasing in scale as social systems achieved higher levels of complexity. Levi-Strauss followed
Saussure in arguing that kinship terms only gained meaningfrom their
place in a structural system, that is, in opposition to other kinship
terms, and not by extension from close relatives to more distant ones.
One of thesimplest might be the four-fold division created by the combination of patrilineal and generational moieties, and expressed
through the Australian four-section system (see Table 3.5). In this
structure, there are only four basic positions, since grandparents and
| S T R U C T U R A L I S M | 75 |
| Table 3.5 A four-section system created by combining generational and patrilineal | |
| moieties Patrilineal moiety A | Patrilineal moiety B |
Generation
A 1 Other }
A2 Own }
Father and fathers sister Mother and mother’s brother B 1
Ego and siblings Spouse and sibling-in-law B2
Mote:The table shows how ego’s closest relatives are distributed between the four
sections, A 1 , A2, B1 and B2.
grandchildren belong to ego’s generation. In the Kariera kinship
terminology, which is based on this four-fold division, the father’s
father and the mother’s mother’s brother are therefore called by the
same term because they occupy the same position (see chapter 2,
Figure 2.1 (a)). Parallel cousins occupy the same position as ego’s siblings (his brothers and sisters), and are therefore also called ‘brother’
and ‘sister’. Even the closest kinship relationships are thus determined
bythestructure of the system, and not extension (such as, from father’s
father to mother’s mother’s brother, or from sibling to parallel cousin).
Just as people are unconscious of the structure of their language, so
they are unconscious of the structure of their kinship system and
accept it implicitly (Levi-Strauss 1969:177).
Levi-Strauss classified the world’s kinship systems into three types:
elementary, intermediate and complex. In elementary systems, everyone known to a person stands in a definite kinship relationship to
them, even if they have no known genealogical link. There are precise
rules of marriage, which specify what type of relative a person must
marry.This is the type of kinship system found in Aboriginal Australia.
At the other extreme are the complex systems found in Europe. Here
only a fraction of the people known to anyone are regarded as their kin.
Marriage is regulated by a principle that close kin should not marry,
and people normally marry non-kin (the Sarakatsani, outlined in
chapter 2, exemplify this type of kinship system). Intermediate kinship
lystems are of the sort found among native North American and many
African peoples. Crow-Omaha systems are of this type (Figure 3.2).
Here, Levi-Strauss argued, the social universe is divided into a determinate number of lineages but these are not linked by a regular pattern
of marriage alliances. Ego can therefore only specify kinship relationihips with members of other lineages into which (s)he or their close
relatives happen to have married.
| Mother’s mother A= O | ‘Father’ ‘Father’s /*\ sister’ ^ A | ‘Grand father ‘Grand mother’ ; A = o | |
| Mother’s father | ‘Grand | ||
| mother’ ‘Grand | ‘Grand | ||
| ‘Father’s A sister’ V A ‘Father’ | ‘Grand father;a = O | Mother 0 | |
| ‘Father’ ‘Father’s 1 | |||
| sister’ ^ | mother’ father, A = O | Ego © | A |
In this example, descent is traced through women. Men marry women of other
lineages. The diagram shows the links which Ego, in the third generation, has with the
lineages of their father and mother’s father. In the father’s lineage, all women regardless
of generation are addressed by the same term as the father’s sister and all men born
into the group are addressed by the same term as the father. All women of the lineage
into which the mother’s father was born are addressed as ‘grandmother’ and their
hncKtanst« ‘oranHfathpr’
Figure 3.2 Example of a Crow-Omaha type kinship terminology
Levi-Strauss’ study of The Elementary Structures of Kinship is only
concerned with kinship systems of the first type. He argued that all
systems of this form can be classified according to three subtypes,
depending on which rule of cross-cousin marriage they follow.
Exchange, he argued, is the universal basis of kinship systems, and is
made possible by three properties of the human mind: to accept that
rules must be followed, to regard reciprocity as the simplest way of creating social relationships, and to consider that a gift, once given, binds
giver and receiver in a continuing social relationship. The structures
created by exchange depend on the type of marriage rule followed. In
making these claims, Levi-Strauss was clearly influenced by the work
of Malinowski and Mauss (exchange theory will be discussed further in
the following chapter). He follows Radcliffe-Brown, however, in arguing that the ultimate beneficiary of relationships created by exchange is
the social system, not the individual participants in it.
The Yanomamo case studyshowed how, wherever a society contains
exogamous, unilineal descent groups, cross-cousins will always belong
to a different group to ego. Once such lineages are linked by regular
marriage exchanges, parallel cousins will always belong to ego’s group.
Any social system which depends on such regular alliances can therefore specify cross-cousins as the ideal marriage partners and forbid
marriage with parallel cousins. Cross-cousins function as ‘markers’
who, even if they do not themselves become ego’s marriage partners,
signal the identity of the group into which he or she should marry.
Logically, as Radcliffe-Brown had already appreciated, there are only
three types of cross-cousin: patrilateral (the father’s sister’s child),
matrilateral (the mother’s brother’s child) and bilateral (where the
father’s sister’s child and mother’s brother’s child are one and the same
or, at least, occupy the same position in the structure of the kinship
system).
L£vi-Strauss’ most fascinating discovery was to realise that each type
of cross-cousin marriage produces its own structure of exchange. Van
Wouden had earlier made the same discovery and it is hard to believe
Levi-Strauss was unaware of his work (see Van Wouden 1968: v, vii,
xii). The insight can be illustrated with models which again assume
each line of descent has only one man and one woman in each generation. If the men of the two lines of descent exchange their sisters, the
mother’s brother marries the father’s sister. Ego’s marriage partner will
be at once his mother’s brother’s child and his father’s sister’s child.
Repeated over several generations, a closed alliance develops between
two lines of descent, of the type exemplified by the Yanomamo (Figure
3.3 (a)). One alternative is to open up the pattern of alliance by following a rule that, while the men of one group give their sisters to the men
of a second one, they themselves receive their wives from a third.This is
the pattern generated by following a rule that men must marry their
mother’s brother’s daughters, but not their father’s sister’s daughters
(Figure 3.3 (b) ), which Van Wouden called asymmetric connubium
(Van Wouden 1968: 86-7). This system can be extended indefinitely,
creating a chain of allied groups which only terminates when the chain
closes in upon itself to become a circle. For this reason, Levi-Strauss
referred to bilateral cross-cousin marriage as restricted exchange, and
matrilateral exchange as generalised. Both the Kariera and the Aranda
have restricted exchange but the so-called Murngin (Yolngu ) of northern Australia, whom Radcliffe-Brown chose to exemplify another
of his types of Aboriginal society, practise matrilateral cross-cousin
marriage.
The third possibility is for a series of lineages each to transfer the
members of one sex in marriage one way in one generation, as in
matrilateral exchange, but to reverse the direction of exchange in the
following generation, as if to repay the debts incurred by the receipt of
marriage partners in the previous generation.This is what happens if a
A=0 0 =A A 0 =A 0 =A O A 0 =A 0=A O
A =0 OA A 0 =A OA O A=0 A=0 A =0
A=0 OA A 0 =A 0 =A 0 A 0= A 0=A 0
| (a) Bilateral cross-cousin marriage | (b) Matrilateral cross-cousin marriage | (c) Patrilateral cross-cousin marriage |
Figure 3.3 The structures created by the three types of cross-cousin marriage
S T R U C T U R A L I S M 79
rule of patrilateral cross-cousin marriage is followed (Figure 3.3 (c)).
Van Wouden anticipated this would have a disruptive effect on social
relations, while Levi-Strauss regarded it as tantamount to a sudden loss
of confidence in thestability of the system. Both matrilateral and patrilateral exchange create patterns of debt and credit linking a chain of
lineages, but in the case of matrilateral exchange, each lineage has to
accept the cost of being permanently indebted to the group who give
them their wives, even though it remains permanently in credit with
the group to whom it is giving itssisters. Even in bilateral exchange, the
two lines remain in a balanced state of debt and credit over successive
generations. In the case of patrilateral exchange, the system constantly
‘returns to a point of inertia’ ( Levi-Strauss 1969: 444) from which
social relationships must once more be re-initiated. It is a system with
inherent internal contradictions, which will either succumb to these
contradictions and collapse, or transform itself into a more stable condition (Levi-Strauss 1963: 311). Levi-Strauss’ theory appears to be
supported by the fact that patrilateral cross-cousin marriage is in fact
the rarest of the three types.
To what extent do the participants in such a system need to be aware
of its structural consequences? Like Malinowski, Levi-Strauss considered that there was no more need for them to be aware of the total
structure than there was for native speakers of a language to be capable
of consciously articulating its grammatical rules. They do need to
recognise the obligations incumbent on them by virtue of their position in that system. Levi-Strauss admits that matrilateral cross-cousin
marriage seems a ‘risky venture’ from ‘an individual and psychological
viewpoint’, because each man gives his sister to one group, but relies on
the goodwill of another to receive a wife (Levi-Strauss 1969: 451). He
quotes two proverbs from a Sumatran people who practise matrilateral
cross-cousin marriage. One, explaining the prohibition on marrying a
patrilateral cross-cousin, asks ‘how is it possible that water can flow up
to its source?’ The other, justifying confidence that anyone who has
given away his sister in marriage will receive a wife, states ‘the leach
rolls toward the open wound’ (Levi-Strauss 1969: 449)
Criticisms of Levi-Strauss’ theory of cross-cousin marriage
TWo principal lines of criticism have been directed by British and
American anthropologists at Levi-Strauss’ theory.The first is generally
known as the ‘preference or prescription’ debate. Critics ask what
proportion of marriages have to follow the rule of cross-cousin marriage for Levi-Strauss’ predictions about the structural consequences
of cross-cousin marriage to be realised. The Asante told Fortes that
they considered the ideal marriage to be with a bilateral cross-cousin,
but Fortes found that only 8 per cent of marriages actually accorded
with this ideal and alliances between different lineages were rarely
perpetuated (Fortes 1950: 279). Chagnon, on the other hand, found
that 70 per cent of Yanomamo marriages took place between members
of already-allied lineages, and many of them were with first cousins
(Chagnon 1968: 73). It is clear that, while the Yanomamo exemplify
Levi-Strauss’ predictions concerning the type of structure generated
by bilateral cross-cousin marriage, the Asante do not. Another
version of the criticism asks how tightly the pool of marriageable
women has to be defined to generate Levi-Straussian structures. Even
the Yanomamo allow marriage with second cousins, or more distant
relatives, providing they are members of the same lineage as egos
cross-cousins. Suppose the whole of a small-scale society were simply
divided into three categories, such that (from the individual’s perspective) one-third were members of his own category, one-third were
potential wife givers and one-third potential wife receivers? The
Purum of Burma, a small communityof four villages containing a total
of ninety households, appear to have organised matrilateral crosscousin marriage somewhat in this way (Wilder 1971). Unfortunately
for the Purum, their villages came under attack by the Japanese during
the Second World War and their marriage system was not available for
reinvestigation at the time of the debate stimulated by Levi-Strauss’
theory.
The second critique concerns what Leach termed the ‘structural
consequences’ of cross-cousin marriage. Will the structures generated
by a particular marriage rule be the same, even where it is embedded in
otherwise very different social systems (Leach 1961b)? Leach pointed
out that the Katchin of Burma, whom he had studied, were hilldwelling, dry-rice cultivators who used matrilateral cross-cousin
marriage as a means of creating closed alliances between aristocrats
and similar alliances among small groups of commoners. Each alliance
might embrace no more than three lineages. Certain women were
given in marriage between aristocrats and commoners to create links
of dependence. Leach argued that the alliances were unstable.
Commoners sought to maintain equality, while aristocrats attempted
S T R U C T U R A L I S M 8 1
to extract tribute, in the way that neighbouring princes did in the
wet-rice cultivating valleys, but without their access to agricultural
surpluses (Leach 1954).The‘Murngin’ of Australia,on the other hand,
wereegalitarian hunter-gatherers, whose use of thesame marriage rule
created chains of relationships among numerous clans. Leach’s study
suggests the material conditions of existence may have a radical effect
on the way that cognitive structures are expressed.
Just how the Murngin system really worked became the subject of
intense debate.Lloyd Warner, an American student of Raddiffe-Brown,
had carried out extensive fieldwork among the Murngin. In his book A
Black Civilisation (Warner 1958 [1937] ), Warner had published the
Murngin kinship terminology (Figure 3.4) which clearly identified
seven lines of descent. Many anthropologists assumed each line on
Warner’s chart corresponded to a distinct clan, or descent group, on
the ground.Since Warner had also written that the Murngin had patrilineal moieties, some assumed that Warner must have missed an eighth
line of descent, otherwise the two outer lines whom, it was inferred,
closed the circle of the alliance, would be guilty of intra-moiety marriage. Lawrence and Murdock published an interpretation along these
lines (1949) which was ridiculed by Radcliffe-Brown as demanding a
greater complexityof kinship relationships than even native Australians,
masters of kinship though they are, could handle (Radcliffe-Brown
1951)! It is clear from Warner’s own work that the Murngin do not
create such tidy marriage alliances (Figure 3.5). Yolngu marriages in
fact create a network of alliances which sometimes turn back on themselves, so that two local groups appear to be engaging in bilateral
exchange. Leach realised this was because someclans consisted of more
than one exogamous lineage (Leach 1961b: 70; compare Warner 1958:
26 and Morphy 1978: 217).
When the second edition of The Elementary Structures of Kinship
was published in 1967, Levi-Strauss responded to earlier criticisms by
contending that, even though in practice matrilateral cross-cousin
marriage might involve no more local groups than were needed in the
Aranda system, the potential of the two types was always different: the
Aranda system would always tend to close in upon itself, whereas the
Murngin system would always tend to open out into longer chains of
alliance. His argument has received support from Keen’s comparison
of marriage practices among the Yolngu (Murngin) and their neighbours the Gidjingali, who have an Aranda-type system (Keen 1982).
A ©=A © “A
Natchiwaiker
Mari Nati MarLkmo Due Kutara Dumunger
A O’ A w- =A
Gawel Marelker Gawel Bapa Waku Gurrong Waku
© =A
Natchiwalker
Mari Galle Ego Due Kutara Dumunger
A OA @ =A ‘A © =A OA ©-A
Gawel Marelker Gawel Gatu Waku Gurrong Waku
| A Natchi | Galle |
| – | Mari |
| waiker |
| Due | Kutara |
| Maraitcha | Dumunger |
Members of e^o’s moiety
Figure 3.4 ‘Murngin’ (Yolngu) kinship terminology
S T R U C T U R A L I S M
A
/\ Yiritja dans
O Dun clans
Figure 3.5 Actual marriage exchanges between ‘Murngin (Yolngu) clans