iDIDJ Australia Didgeridoo Cultural Hub

Williams is sceptical about the applicability of the terms 'string', 'network', and 'web' to ba:purru names such as Gumatj, Dhalwangu, and Rirratjingu (1999: 134). In fact, I do not refer to groups such as these as networks or webs but as patrifilial groups or just 'groups', and I used 'string' consistently to refer to a number of connected groups. I describe networks of connection between places and groups, and discuss network images as metaphors of group cohesion and interrelations (Keen 1995: 515, 519, 520). Nevertheless if Dhalwangu or Gumatj extend their names in such a way, then it may indeed be appropriate to speak of Gumatj or Dhalwangu names as referring to 'strings' of groups, just like the names Manydjikay or Ma:tjarra.

Of course, it may be that Yolngu incorporate several varieties of cognitive schemes into their discourse about social relations, social entities and processes. Drawing on the work of a variety of researchers (including Williams, Morphy, Rudder, and Tamisari) I have posited images of connections to localities of bone, body, and limb; of root and branch, focus and extension, networks, and extendable strings. But no doubt Yolngu social-structural constructs embed other kinds of schema such as clusters; the point is that such schemes are involved in their constitution.

Yolngu people use constructs, Williams (1999: 126) argues, that may be more appropriately likened to the concepts 'descent group', 'lineage', and 'clan'; the latter does less violence to Yolngu concepts than 'strings of indeterminate length'. Responding to Williams's enquiries about the frequently used phrase 'same but little bit different', Dula Ngurruwutthun cautioned her that she should remember that Yolngu are 'all bundled together, just like a bundle of sticks' (1999: 126). This image, which Williams interprets as contrasting with network images, is a telling one. Certainly, the image implies a congerie of entities. To the Australian, English, French, or American reader it may evoke a picture of straight, neatly split pine, neatly tied into a bundle for firewood. But in Yolngu languages the word dharpa means a living or dead tree or woody shrub, the fallen or cut trunk or branches of a tree, firewood, and timber or lumber (Zorc 1986: 72). In northeast Arnhem Land the bundle of firewood from the forest c arried by a person over the shoulder, or in the back of a utility truck, is typically a heap of bent, twisted, and forked branches which interlock and tangle as such branches will. The image may well evoke a tangle of connections, as well as the distinctiveness of elements (and nowhere have I suggested that Yolngu do not recognize distinct groups).

Rudder (1997: 160) reproduces a drawing by Dangatanga which illustrates relations among five patrifilial groups of the Dhuwa moiety linked by ownership of the Morning Star ceremony. Each is represented by a yam plant with its own separate tuber, but the vines of the five plants are intertwined. Each plant is separate but together they 'bring fruit' as the result of affinal relations (of WM/WMB -- WDC/ZDC) among them. The drawing shows branches of the vines connecting adjacent and alternate groups. Terms such as 'network' and 'web' seem very appropriate here in describing relations among the groups as depicted by the drawing of these plants, and also, perhaps, the group itself.

Enduring identities

Sutton's assessment (1999) of my critique of the clan model occurs in the context of a useful investigation into the tenacity of Aboriginal 'country groups' over a lengthy period. He adduces evidence of very long-term associations of languages with regions, although there is also compelling evidence of migrations in recent millennia (1999: 36). He also emphasizes the genealogical, totemic, and geographical stability of Wik 'clans' and estates over some 160 years (1999: 38), and in the Daly River, for some 150 years (1999: 40-1). In a detailed study of the ethnographic record, Morton (1997) found some shifts in the boundaries of estates and ownership of country among Arrernte people, together with some stability, as has McKnight (1999: 82-3) in his study of Lardil land tenure.

Ba:purru identities have certainly persisted in northeast Arnhem Land and have appeared in anthropological and other records for more than sixty years. The stability of totemic identities in northeast Arnhem Land over several generations is attested by the written record dating back to the late 1920s, as well as genealogies collected by Donald Thomson in the 1930s. However, neither the relative stability of estates and their totemic affiliations nor the association of a particular lineage with a country over several generations is incompatible with my analysis of Yolngu patrifilial groups, nor do they support the use of the term 'clan' in Arnhem Land. The ba:purru names consistently associated with succeeding generations of agnates, with places, and with totemic entities, may be those of patrifilial groups of heterogeneous and indeed changing structure. What persists is their identity, differentiation, attachment to patrilineal succession, and, in broad terms at least, relations to locality.

Dealing with anomalies

In adhering to a clan model, anthropologists have recently tried to deal in a variety of ways with cases that do not fit. One way is simply to ignore the difficulties, but there have been more creative strategies.

In the first, the unified clan is a target or ideal to which people aspire and which they sometimes achieve. The modal form of the Wik clan, according to Sutton (1999: 30), consisted of one or more 'patrilines' (lineages) whose members claimed the same land as primary country, and who had the same patrilineal totems and totemic names:

In other words, 'clan' may be more neatly applicable in some cases than in others; not all clansmen are clansmen in relation to all the country claimed by any one of them, and not all territories have a 'clan' in possession of primary and unique rights over them. But it is fair to say that the clan as a patrilineal land-holding totemic unit with a unique country is the target towards which the flux of reality is continually pushed, and forms the model into which people attempt intellectually to compress the often somewhat ragged facts (Sutton 1978: 59-60, cited in Sutton 1999: 30).

The ultimate Wik target was to belong to a descent-based group whose claims had been recognized by the relevant jural public 'since time out of mind' (1999: 30). The application of this theory to northeast Arnhem Land requires evidence that a certain structure is indeed a shared ideal or the target. Presumably the structure of complex groups such as Djambarrpuyngu, and dual groups such as Djapu, would be described with reference to the ideal. The drawing of patri-groups as yam plants mentioned earlier does imply a conceptual equivalence of the patrifilial groups so represented, but we cannot infer from this equivalence that the groups have a uniform organization of a kind implied by the term 'clan'.

A second strategy is to regard the anomalous cases as stages in a process in which clans merge and split. Morphy (1997) has elaborated a model of clan dynamics, building on the work of Peterson (1983) (see also Keen 1978: 97; Peterson, Keen & Sansom 1977). According to Peterson's (1983) model, a clan's land and the foraging range of the related band (a residence group consisting of some members of two or more clans) will tend to maintain a more-or-less constant size. Clans, however, expand and contract according to their demographic fortunes. As a clan grows, more of its male members live on other clans' estates. Where a clan is dying out, members of certain other clans will have the opportunity to merge all or part of its estate with their own. The merging of two foraging ranges, however, entails costs, so that in the long term two bands will re-emerge, reinforcing divisions in the enlarged clan, perhaps through differences in matrilineal ties, which then splits. The range and estate would become congruent once more.

In Morphy's model (1997) different segments of a clan, with distinct matrifilial links and related marriage rights, may grow in size independently. When conflict reaches a certain level the clan may split and create new units. Rapidly growing clans (or segments) may constitute a threat to another clan by blocking men's access to women as wives, or through physical violence, resulting in its territory being taken over. In this way Morphy explains variations in clan structure as resulting from a Fortesian process of clan growth, internal conflict, division, extinction, and takeover, articulated with the politics of marriage. The account is a reconstruction of past practices, for over the last two decades especially, the mobility of younger people has transformed marriage practices by enabling some men and women to evade the system of bestowal to find spouses outside Arnhem Land. As mentioned, Morphy (1991: 50) also reports the merger of patrifilial identity as a specific strategy to simplify land tenure for a land case.

These models depend on the existence of a modal clan which is the site for internal division, fission and fusion, and, in Peterson's model, the beginning and end state of the trajectories of transformation. Given the variety of forms of Yolngu patrifilial groups, processes of growth, decline, and succession are likely to have been more complex than the models suggest. Because of the ways in which matrilateral and agnatic connections cross-cut, and because of the variety of grounds for succession claims, processes of division and takeover also have a greater complexity. For example, there can be rival contestants from different claimant groups, on the ground that the target country is that of their mother's mother's group (Williams 1986: 53), and on other grounds such as spirit conception.

In order convincingly to account for variation in the structure of Yolngu patrifilial groups, it would be necessary to locate all (or most) of the known variants as stages of transformation, and to show that documented processes of change are consistent with the models, taking into account global process of social change such as changes in marriage practices. It is hard to see how the complexities of Djambarrpuyngu structure might be accounted for in this way.

Heterogeneity

Why would one expect a uniform constitution, or even modal type, among Yolngu groups? That Yolngu share a clan system is an assumption built into the concept of a regional culture or system. The clan model itself presupposes a type of organization across the region, such that the application of terms like 'clan' and 'localized sub-group' is governed by recognizable criteria. Williams (1986: 18), for example, writes of the 'Yolngu system of land tenure'. In her response to the critique of the clan model, however, Williams (1999: 130) emphasizes heterogeneity, suggesting that my critique of the clan model reflects research on the western fringe of the Yolngu 'cultural bloc' whereas her own work was focused on the most northeasterly area. This is rather ironic given my recent (1994) emphasis on heterogeneity across the region.

The implication of Williams's point is that the critique of the clan model may apply to groups at Milingimbi (in the northwest) but not Yirrkala (in the northeast). In fact, many groups represented at Milingimbi moved there from further to the east after the establishment of the mission. Moreover, the research that led me to rethink the concept of clan was a brief but intensive study of a dispute over the ownership of land, carried out in 1987 at Yirrkala, Gapuwiyak, and Galiwin'ku, in the north and northeast of the region. In any case, the critique is just as applicable to the analysis of easterly and northeasterly groups, for several of the anomalies thrown up by the clan model concern those groups (e.g. Gumatj, Djapu, and Djambarrpuyngu of Dhambaliya). The critique allows for variation in the constitution of patrifilial groups; indeed, the insistence on the constitutive role of tropes makes it possible to examine variation in the construction of group-identity in Yolngu discourse and practice without bein g hamstrung by the constraints of the clan concept.

I suggest that the variation in group structure has to do in part with the nature of Yolngu ancestral law and politics, as well as processes of social change during the twentieth century. While people more-or-less agree about principles and related beliefs, there is no overarching political or legal authority to impose uniform dogma, rules, and arrangements. (The state has no jurisdiction in such matters, except indirectly through the Land Rights Act and Native Tide Act.) People of a local network inherit and work out their own particular way of doing things, subject to their relations with neighbours, and to demographic and other contingencies.

Translation

More general points about analysis and translation raised in the original article (Keen 1995) have not been taken up in the debate. This drew on cultural anthropology and the philosophy of language to explore in some detail the way in which (some) Yolngu constitute their world. The article makes suggestions about how the anthropological meta-language draws on and modifies imagery embedded in the meaning and use of its concepts and models, leading to difficulties of description and translation. It seems to me to be potentially a fruitful exercise to attempt to modify or move away from terms in our customary anthropological meta-language, in order to explore further how Yolngu constitute their social world, necessarily in their own terms, and to find resources in our own languages to match those terms.

Problems of translation arise in Williams's use of the expression 'a Yolngu concept of corporateness'. Drawing on Stoljar (1973), she understands certain developments in England and northeast Arnhem Land as parallel but distinct histories of corporateness. The classic incidents of formal incorporation which had developed in England by the twentieth century, with appropriate alterations, characterize Yolngu land-owning groups (Williams 1986: 96; 1999: 131; and see above). Williams does not merely draw an analogy between concepts of corporation in English law and Yolngu constructs, but treats 'corporateness' as a Yolngu concept. However, the application of the word 'corporate' to Yolngu concepts, beliefs, rules, and practices is an act of translation; not in this case by substituting an English word or expression for a Yolngu word or expression, but in the use of an English word, derived from legal and social theory as well as practical use, to label complex Yolngu constructs. Not only are the connotations of terms inappropriate here, but incidents such as the right to a common seal, the capacity to make by-laws and to sue and be sued are completely alien, reflecting a quite different system of law. For some purposes at least, Williams (1983: 105) prefers to follow Stanner's (1956) advice and construe Yolngu practices in terms of Western legal concepts.

My view, that anthropological descriptions 'necessarily' incorporate certain metaphors, is not defensible in Williams's belief (1999: 132). Some anthropologists, she writes, have been at pains to avoid certain inappropriate connotations of terms such as 'corporate' and 'boundary' by defining them in such a way as to exclude them (Williams 1999: 136), as I have also pointed out (Keen 1995: 506, 519). But the specific definition of terms does not free one from all their non-stated implications. Whatever the qualifications, William's scheme divides Yolngu patrifilial groups into 'clans' and 'localized sub-groups'. The inescapable implication is that the Yolngu of whom she writes arrange themselves into tokens of a recognizable type of organization. The tropes implicit in this scheme include those which help constitute a taxonomy, especially 'levels' of equivalent elements in which elements at a lower level are incorporated into an element at a higher level (hence the term 'sub'-group). The tropes are necessaril y implied, unless you assert with Humpty Dumpty, 'When I mean a word ... it means just what I choose it to mean -- neither more nor less' (Carroll 1962: 114).

Attempts at a closer translation pose problems for comparison and generalization. One response to the original paper was that comparison would be inhibited without the use of generalizing concepts such as 'clan'. However, James J. Fox reacted differently to a seminar version of the 1995 article. Finding parallels as well as differences between the Yolngu cultural constructs described in the paper and those of Rotinese with whom he has worked, Fox remarked, 'Now we can talk to each other!' In his view, the analysis fostered comparison. Indeed, the use of terms such as 'clan' may actually obscure and distort particularities and differences by subsuming local practice within globalizing terms. If the clan concept distorts Yolngu practice, then comparisons based on it will also be misleading.

In the end, resolution of the issue, if resolution is possible, depends on evidence. In order to support the clan model and to refute the critique it would be necessary to provide evidence for some or all of the following: (a) the clan model does not in fact generate anomalies; (b) Yolngu clans are tokens of a type of organization, namely clans; (c) Yolngu subscribe to an ideal of clan structure to which existing forms tend to conform; (d) processual models in which clans are the modal state account for all or most of the variation in group structure. A more fruitful way forward would be to investigate the constitution of a number of patrifilial groups in detail (including discourses and strategies of identity, ancestral connections and rights in sacra, associated social action and relations, social and demographic processes, adaptations to contingencies, and so on). My hunch is that a degree of heterogeneity will become apparent that has so far been hidden by the imposition of the clan model, and that canno t be accounted for by uniform processes across the region.

NOTES

I thank Nicolas Peterson and Peter Sutton for their comments on an earlier draft of this paper, and anonymous referees for suggestions.

(1.) While the details vary, I think that it is fair to say that the work of Morphy (1977; 1984; 1991), Peterson (1972), Schebeck (1968), Shapiro (1981), Williams (1986), and myself (Keen 1978; 1982; 1988) provided varieties of the general model outlined above. The Berndts proposed a somewhat different account in which each patrilineal local group was formed at the intersection of cross-cutting 'group' (mala) and 'language' (matha) identities (Berndt 1965; 1976).

(2.) I am grateful to Peter Sutton for raising this point.

(3.) In support of the use of the term 'clan', Williams (1999: 126) cites Ganambarr (1994: 235-6), who contrasts links through women 'binding us and the tribes in this area all together in a complex way' with 'our father's line where we are a single group'. Williams (1999: 136) makes much of this expression of unity of identity through the father, and writes that the account 'surely constitutes a most convincing case for the use of the term clan to refer to this entity in Yolngu society'. But it does not constitute a case for using the term 'clan' at all, for the question hangs on the constitution of patrifilial groups, their similarities and differences, and relations among them, not the unity of a particular 'father's line'. Affirmations of unity are common in Yolngu discourse. Interestingly, in her explanation of the expression yothu-yindi ('child-mother') Ganambarr uses the words 'clan' to refer to Datiwuy, her own patrifilial group, 'tribe' for Gumatj, her mother's group, and 'tribes' and 'clans' interc hangeably to refer to what are called in Yolngu languages ba:purru, matha, and mala.

(4.) In an earlier publication Williams (1983: 98-9) makes the same remarks about clans in Aboriginal Australia generally.

(5.) Williams (1999: 135) does make some remarks about hierarchy in the sense of social rank within patrifilial groups, apparently in response to my suggestions about taxonamic hierarchies.

(6.) Williams (1999: 132-3) takes me to task for misrepresenting her when I wrote that she 'emphasizes the corporate characteristics of matha, defined by the ownership of land'. However, I expanded this remark in a footnote. I do not think that I seriously distorted her views, given the scope of the article.

(7.) Cases described or indicated in the literature include Dultji Warramiri and Mirtamirta Warramiri (Shapiro 1981: 121-2), Djapu (Morphy 1991: 48), and the Djambarrpuyngu groups (Rudder 1997: 168). Ganalbingu and Dhalwangu groups may also have this structure (Nicolas Peterson pers. com.; Peter Toner pers. com.).

REFERENCES

Barnes, R.H. 1980. The meaning of corporation in Eastern Indonesia. In The meaning of marriage payments (ed.) J.L. Comaroff, 93-124. London: Academic Press.

Berndt, R.M. 1965. Marriage and the family in North-eastern Arnhem Land. In Comparative family systems (ed.) M.F Nimkoff, 77-104. Boston: Houghton Muffin.

----- 1976. Love songs of Arnhem Land. Chicago: University of Chicago Press.

----- & C.H. Berndt 1977. The world of the first Australians. Sydney: Lansdowne Press.

Birdsell, J. 1953 Some environmental and cultural factors influencing the structuring of Australian Aboriginal populations. American Naturalist 87, 171-207.

Carroll, L. 1962 (1873). Through the looking glass and what Alice found there. London: Macmillan.

Evans-Pritchard, E.E. 1940. The Nuer. Oxford: Clarendon Press.

Ganambarr, M. 1994. Da:tiwuy. In Macquarie Aboriginal words (ed.) N. Thieburger & W. McGregor, 234-65. North Ryde: Macquarie Library Pty Ltd, Macquarie University.

Groger-Wurm, H.M. 1973. Australian Aboriginal bark paintings and their mythological interpretation 1: Eastern Arnhem Land. Canberra: Australian Institute of Aboriginal Studies.

Keen, I. 1978. One ceremony, one song: an economy of religious knowledge among the Yolngu of North-east Arnhem Land. Ph.D. thesis, Australian National University.

----- 1982. How some Murngin men marry ten wives: the marital implications of matrilateral cross-cousin structures. Man 17, 620-42.

----- 1994. Knowledge and secrecy in an Aboriginal religion: Yolngu of Northeast Arnhem Land. Oxford: Clarendon Press.

----- 1995. Metaphor and the metalanguage: 'groups' in northeast Arnhem Land. American Ethnologist. 22, 502-27.

----- 1997. Western Desert v the rest. In Australian Aborigines and contemporary social enquiry: essays to honour L.R. Hiatt (eds) F. Merlan, J. Morton & A. Rumsey, 65-93. Canberra: Aboriginal Studies Press.

Kraus, W. 1998. Contestable identities: tribal structure in the Moroccan High Atlas. Journal of the Royal Anthropological Institute (N.S.) 4, 1-22.

Lakoff, G. & M. Johnson 1980. Metaphors we live by. Chicago: University of Chicago Press.

McKnight, D. 1999 People, countries, and the Rainbow Serpent: systems of class classification among the Lardil of Mornington Island. New York: Oxford University Press.

Morphy, H. 1977. Too many meanings: an analysis of the artistic system of the Yolngu people of North-east Arnhem Land. Ph.D. thesis, Australian National University.

----- 1984. Journey to the Crocodile's Nest. Canberra: Australian Institute of Aboriginal Studies.

----- 1991. Ancestral connections: art and an Aboriginal system of knowledge. Chicago: Chicago University Press.

----- 1997. Death, exchange and the reproduction of Yolngu society. In Australian Aborigines and contemporary social enquiry: essays to honour L.R. Hiatt (eds) F. Merlan, J. Morton & A. Rumsey, 123-50. Canberra: Aboriginal Studies Press.

Morton, J. 1997. Arrernte (Aranda) land tenure: an evaluation of the Strehlow model (Strehlow Research Centre, Occasional Papers 1). Alice Springs: Strehlow Research Centre.

Myers, ER. 1986. Pintupi country, Pintupi self: sentiment, place, and politics among Western Desert Aborigines. Washington: Smithsonian Institution Press.

Peterson, N. 1972. Totemism yesterday: sentiment and local organisation among the Australian Aborigines. Man (N.S.) 7,12-32.

----- 1983. Rights, residence and process in Australian territorial organisation. In Aborigines, land and land rights (eds) N. Peterson & M. Langton, 134-45. Canberra: Australian Institute of Aboriginal Studies.

----- I. Keen & B. Sansom 1977. Succession to land: primary and secondary rights to Aboriginal estates. In Hansard of the Joint Select Committee on Aboriginal Land Rights in the Northern Territory, 1002-14. Canberra: Government Printer.

Rudder, J. 1997. Yolngu cosmology: an unchanging cosmos incorporating a rapidly changing world? Ph.D. thesis, Australian National University.

Schebeck, B. 1968. Dialect and social groupings in north east Arnhem Land. Typescript.

Australian Institute of Aboriginal and Torres Strait Islander Studies, Canberra.

Scheffler, H.W. 1978. Australian kin classification. Cambridge: Cambridge University Press.

Schneider, D.M. 1984. A critique of the study of kinship. Ann Arbor: University of Michigan Press.

Shapiro, W. 1981. Miwuyt marriage: the cultural anthropology of affinity in Northeast Arnhem Land. Philadelphia: Institute for the Study of Human Issues.

Stanner, W.E.H. 1956. The Dreaming. In Australian signpost (ed.) T.A.G. Hungerford, 51-65. Melbourne: Cheshire.

Stoljar, S.J. 1973. Groups and entities: an inquiry into corporate theory. Canberra: Australian National University Press.

Sutton, P. 1978. Wik: Aboriginal society, territory and language at Cape Keerweer, Cape York Peninsula, Australia. Ph.D. thesis, University of Queensland.

----- 1999. The system as it was straining to become: fluidity, stability, and Aboriginal country groups. In Connections in native title: genealogies, kinship and groups (CAEPR Research Monograph (eds) J.D. Finlayson, B. Rigsby & H.J. Bek 13), 13-57. Canberra: Centre for Aboriginal Economic Policy Research, Australian National University.

Tindale, N. 1974. Aboriginal tribes of Australia. Berkeley: University of California Press.

Turner, S.P. 1979. Translating ritual beliefi. Philosophy of the Social Sciences 9, 401-23.

Wagner, R. 1974. Are there social groups in the New Guinea Highlands? In Frontiers of anthropology (ed.) M. Leaf, 95-122. New York: Van Nostrand.

Warner, W.L. 1937. A black civilization: a social study of an Australian tribe. New York: Harper and Brothers.

Williams, N.M. 1983. Yolngu concepts of land ownership. In Aborigines, land and land rights (eds) N. Peterson & M. Langton, 94-109. Canberra: Australian Institute of Aboriginal Studies.

----- 1986. The Yolngu and their land: a system of land tenure and the fight for its recognition. Canberra: Australian Institute of Aboriginal Studies.

----- 1999. The relationship of genealogical reckoning and group formation: Yolngu examples. In Connections in native title: genealogies, kinship and groups (eds) J.D. Finlayson, B. Rigsby & H.J. Bek, (CAEPR Research Monograph 13), 125-39. Canberra: Centre for Aboriginal Economic Policy Research, Australian National University.

Zorc, S. 1986. Yolngu-matha dictionary. Batchelor, N.T.: School of Australian Linguistics.

Un fatras de batons: le debat sur les clans Yolngu

Resume

Une critique du modele de clan tel qu'il a ete applique dans l'ethnographie du nord-est de L'Arnhem Land a provoque a son tour un certain nombre de responses critiques. Cet article defend la critique originale et prend a parti les questions soulevees dans les reponses. Selon cette critique, un manque de congruence entre certains elements du modele de clan et les conceptions des Yolngu se rapportant aux notions d'identite, de pays et des ancetres aurait produit des anomalies. Cette assertion fut corroboree par un compte rendu des metaphores et autres tropes trouvees a la fois dans les conceptions Yolngu et dans les concepts et les modeles anthropologiques. Cet article discure les points cles en question: la cloture des terres et les limites des groupes, l'etablissement des genealogies, l'homologie des segments, les niveaux taxonomiques, l'organisation collective, les ensembles et les reseaux sociaux, la durabilite des identites patrifiales et la capacite des modeles evolutifs de prendre compte des anomalies. L a portee de ce debat s'etend non seulement a des modeles semblables dans d'autres regions mais aussi a la nature de la description, de la traduction et de la generalisation en anthropologic.