iDIDJ Australia Didgeridoo Cultural Hub

Local groups or bands in the classical social organisation included men and women of different clans. The focal male leaders of local groups centred their residence and group movements on lands of their own clan estates, for example, Thomson's notes and genealogies identify the focal men of the Yintjingga local group in 1928-1929 as several older Mbarrundayma clan men and their resident sister's sons of the Mumpithamu clan. (Rigsby & Chase 1998:199).

While not as circumscribed as Eastern Cape York, individual ranges in the Oenpelli area appear to have been comparatively limited and to have shown a high degree of overlap.

the Gunwinggu pattern of living was traditionally semi-nomadic -- 'semi-' because the region in which they moved was usually circumscribed. (Berndt & Berndt 1970:102-4)

Men did not keep only to their gunmugugur [clan] territory, but from our accounts they showed a preference for spending much of their time there and in immediately adjoining areas -- as a few of them still do. And it was in such a territorial constellation that other people would expect to find them. (ibid.:107)

The preference for geographically close marriage is self-evidently a central and essential component in the construction of geographically narrow ranges, which are the preferred target across much of northern Australia.

THE STRUCTURE OF ASYMMETRIES

At first glance, the various asymmetrical terminologies fall into three distinct groups. The Ngarinyin, Worora, and Wunambal terminologies focus on the clan and agnatic groupings. The Oenpelli Kunwinjku, Yir Yoront, and Yolngu terminologies focus on matrilateral vs patrilateral linkages. The Dyirbal, Wik and Kaanyju terminologies focus on comparative age differences. However, I will show that comparative age is also a central factor in the Oenpelli Kunwinjku, Yir Yoront, and Yolngu terminologies. The major distinction in asymmetric terminologies is therefore between the Ngarinyin, Worora, and Wunambal terminologies and the others.

Throughout most, if not all, of Australia, there was a strong preference for a woman's first husband to be significantly older, and generally at least the age of her mother. As we have seen the areas associated with Oenpelli Kunwinjku, Yir Yoront and Yolngu are all areas where marriage to genealogically close relatives is enjoined. In this situation, it is generally likely that a man's MBD will be younger than his FZD. Keen (1982:636-7) shows that a pattern of an average 40 year gap between father and child and an average 25 year gap between mother and child will result in the MBD being on average 15 years younger than a male EGO. The preference for MyBD among Thaayorre owners/speakers accords even more strongly with the preference for a significant age gap between husband and wife.

The same age-based preference for the MBD as spouse was found among Kariera owners/speakers. However, they differ from Yolngu and Cape York groups in also allowing the FZD as a legitimate spouse.

As the natives themselves put it to me, a man must look to his kaga to provide him with a wife by giving him one or more of his daughters. The relative who is most particularly his kaga,...,is his mother's brother, who may or may not be at the same time the husband of his father's sister. It is to this man that he looks first for a wife. If his own mother's brother has no daughter, or if she is already disposed of, he must apply to other persons who stand to him in the relation of kaga, to the husband of his father's sister for example. He may have to go much further afield and apply to some distant kaga, but this is only the case when there are available no nearer relatives. Thus we may say that the man who is pre-eminently kaga ... is his mother's brother; the woman who is pre-eminently toa is his own father's sister who should be the wife of the kaga; consequently the woman who is pre-eminently a man's nuba is the daughter of his own mother's brother, or failing this, of his own father's sister. It is this woman to whom he has the first right as a wife. (Radcliffe-Brown 1913:156)

The Oenpelli Kunwinjku first preference for the actual FZDDD also accords with this preference. Keen states that a FZDDD will be on average 35 years younger than a male EGO (Fig 1), based on the previously mentioned average 40 year gap between father and child and average 25 year gap between mother and child. From the viewpoint of gerontocratic marriage, an average difference of 35 years would appear to be desirable. It would allow a 45 year old man to claim a 10 year old wife. Keen (1982:637) states that 'A model of marriage between a man and his FZDDD in each generation shows that the age difference between such persons is reproduced at each generation.'

The use of the terms berlu 'FZ' and ngal-kurrng 'mWM' in the Oenpelli Kunwinjku asymmetry also appears to correlate with the age factors in marriages. Scheffler (1978:487-92) analyses the Oenpelli Kunwinjku terminology as basically a Kariera terminology, where the ngal-kurrng term distinguishes a particular sub-class of the overall berlu class. It is true that these two terms can be used in reference to the same individual. Thus among Oenpelli Kunwinjku, if a man marries kanjok, the daughter of a berlu, he will in most cases refer to his actual mother-in-law as ngal-kurrng (Berndt & Berndt 1970:100). Among Gundjeihmi owners/speakers any ngal-kurrng, including one's own mother-in-law, may alternatively be referred to as berlu (N. Evans pers.comm.). However, it should be noted that Gundjeihmi owners/speakers sometimes make use of a completely Kariera terminology, and that Gundjeihmi owners/speakers do not nowadays make use of the asymmetry which is standard among Oenpelli Kunwinjku owners/speakers.

Despite these examples of overlapping reference, there are other factors which suggest that ngal-kurrng should not be viewed as a subclass term with berlu as a superclass term. Firstly, among Oenpelli Kunwinjku owners/speakers, behaviour towards a berlu/ngal-kurrng mother of a kanjok spouse is not the same as behaviour towards the ngal-kurrng mother of a kakkali spouse.

De facto gagali [kakkali] and nagurng [na-kurrng]-ngalgurng [ngal-kurrng] are expected to behave substantially as if their bonds were 'real' -- that is, as if they were based on genealogical ties. A man has the usual obligations toward his actual mother-in-law, his wife's mother, including partial avoidance, but what is uncertain is how far he and she are obliged or entitled to use the special gungurng kun-kurrng] vocabulary in speaking together. It is the only proper medium of conversation in this affinal relationship, but ideally it is restricted to 'real' nagurng-ngalgurng, where the genealogical connection is traceable or implied (Berndt & Berndt 1970:100)

A similar difference in behaviours is found among Gidjingali owners/speakers.

I mentioned that there were special restrictions on social intercourse between a man and his WM or potential WMs. The one exception was when a person married his FZD. The relationship between a man and his FZ was familiar and affectionate, and it did not change if he married her daughter. He continued to address his WM as baba [FZ]. (Hiatt 1965:72)

For Oenpelli Kunwinjku owners/speakers, the first preference ngal-kurrng, who appears in the asymmetry, cannot be alternatively referred to as berlu. Rather, this first preference mother-in-law is the daughter of a woman called berlu. As such, she is, by the general logic of kinship terminologies, a cousin rather than an aunt. And indeed, this is exactly how she is referred to in Kriol. If a Kunwinjku speaker is asked for the Kunwinjku equivalent of the Kriol term cousin, then the first response is na-/ngal-kurrng and not kanjok. This response pattern is in fact found throughout the western Top End and eastern Kimberleys, regardless of whether the language has a symmetrical or an asymmetrical terminology. If a language has a specific term for mother-in-law, and you ask for the equivalent of the Kriol term cousin, the first response will be the term for mother-in-law, and not a term which you would otherwise understand to mean 'cousin'.

The term kanjok is translated as cousin if a Kunwinjku-Kriol translation is requested. The term ngal-kurrng can be specifically distinguished in Kriol as the poison cousin. Again, this pattern is found throughout the western Top End and eastern Kimberleys. However, most analyses of kinship terminologies would predict that the mother-in-law should be known as the poison auntie rather than the poison cousin. This predicted Kriol term appears in Wik Mungkan where the mother-in-law is known as poison aunt, and the father-in-law as poison uncle (Kilbam et al. 1986:382, 385).

The evidence therefore suggests the prototypical mother-in-law for Oenpelli Kunwinjku speakers is a type of cousin, and not a type of aunt. Further, it appears that this view of the mother-in-law has some mote generally applicable basis. If we return to consider the general conditions of age relations in Aboriginal Australia, then this more general basis becomes evident. In Aboriginal Australia, the prototypical aunt is the actual FZ, who will belong to an age cohort senior to EGO. The prototypical mother-in-law, on the other hand, very commonly belongs to the same age cohort as a male EGO. Thus prototypical 'aunts' and 'mothers-in-law' are commonly women at two very different stages of life. When marriage practices have a first preference for actual kin, as they do among the Oenpelli Kunwinjkn, it is even more likely that the first preference mother-in-law will be of the same generation as a male EGO (Fig 1).

In considering the significance of this prototypical age difference, it is necessary to examine the general semantics of English kinterms, as this is central to Kriol translations of Aboriginal language kin terms. English kinterms are usually defined in terms of reproductive relationships to EGO, as are kinterms generally: i.e. 'aunt' is the sister of my father or mother, etc. There has been much debate as to whether this is the correct or appropriate way of defining kinterms, either in English or more generally. I do not address this debate, but merely note that reproductive relationships have been central to analyses of the semantics of kinterms, whether correctly or otherwise.

There is, however, another general aspect of the meaning of English kinterms, which has not been a central focus of attention; that of comparative age relationships. Kin terms in English have age implicatures, as illustrated by the following sentences.

She's my cousin, but she's (much) older than me.

? She's my cousin, but she's the same age as me.

She's my cousin, but she's (much) younger than me.

? She's my aunt, but she's (much) older than me.

She's my aunt, but she's the same age as me.

She's my aunt, but she's (much) younger than me.

There is an implication that cousins are in the same age cohort as EGO, and it is very marked to assert this as being contrary to expectations. Similarly, there is an implication that aunts are of the age cohort immediately senior to EGO, and again it is very marked to assert this as being contrary to expectations.

I argue that Kriol speakers are aware of both the 'reproductive relationship' and 'age cohort' aspects of English kinterms, and that their translations of Aboriginal language kin terms may give prominence to either factor. We may illustrate this by considering the full range of Kriol usages of the term cousin. We have seen that Kriol speakers use the term cousin for certain types of in-laws, and for cross cousins. Kriol speakers also use cousin in compound terms to describe parallel cousins: cousin-brother and cousin-sister. Kriol speakers are generally aware that English speakers use the terms, brother, sister, and cousin, to distinguish lineal from collateral kin, and not parallel from cross kin. The use of cousin to refer to cross-cousins, and in the compound terms to refer to parallel cousins, depends on the 'reproductive relationship' meaning. Cousins, cousin-brothers, and cousin-sisters may be of very different age-cohorts to EGO.

On the other hand, I argue that the use of the term cousin to translate specifically in-law kinterms, focally the mother-in-law & son-in-law dyad, depends on the 'same age cohort' meaning of this term, and not on the 'reproductive relationship' meaning. Given the universal preference for the construction of affinal relationships with cross rather than parallel kin, it is to be expected that in-laws will be cousins, rather than brothers or sisters.

While a man and his mother-in-law may often belong to the same age cohort, the kind of relationship that holds between a man and his mother-in-law is marked by obligation and deference on his part, which are paralleled by the kinds of deference and obligation he owes to older men and women. There are grounds, therefore, for viewing mothers-in-law as both 'aunts' and 'cousins'. I would suggest that they are oft en viewed as both, simultaneously and irresolubly.

There are two factors which favour the general terminological classification of mothers-in-law with aunts, as cross-kin of the parental generation level. One is the salience of the avoidance behaviours, which class the mother-in-law with senior generation kin. The other factor is the emphasis on the orthodox transitive applicability of kin terms. If the mother-in-law is terminologically classified with aunts, then the general transitive application of kin terms is unproblematic. If the mother-in-law is classified in EGO's own generation, then the transitive application of kin terms is problematic, as previously discussed for Oenpelli Kunwinjku. Similar problems arise with the Dyirbal terminologies, which also classify in-laws as belonging either to EGO's own generation or to the grandparental generation (Dixon 1989).

It is in the context of these factors that the common element mukul in the Yolngu terms mukul ba:pa 'FZ' and mnukul rumaru 'wM, MBW' is probably to be understood, even though the relationships that hold between a male EGO and women called by these two kin terms are very different.

CORRELATIONS BETWEEN TERMINOLOGICAL SYSTEMS AND MARRIAGE PRACTICES.

There are a number of correlations evident between terminological systems and marriage practices, but few of these are absolute.

a. Asymmetric terminologies are only found in areas where geographically close marriage is enjoined, and where marriage to genealogically close kin is permitted. However, there are areas where geographically close marriage is enjoined and where marriage to genealogically close kin is permitted which are associated with symmetric terminologies (e.g. Gidjingali).

b. In the small number of cases where marriage practices enjoin first preference marriage to genealogically close kin, then the terminologies are mostly asymmetric (Oenpelli Kunwinjku, Yir Yoront, Yolngu). However, the marriage practices of Kariera owners appear to have enjoined first preference marriage to a genealogically close relative and the Kariera terminology is symmetric.

There is a direct causal connection between the prescription or permission of genealogically close marriage, as is the case with all the asymmetrical terminologies, and the attainment of geographically close marriage. The prescription or permission of genealogically close marriage permits a pattern of marriage between the same neighbouring groups to be repeated at each generation level. We may compare this with situations where geographically close marriage is prescribed, but genealogically close marriage is proscribed, as among owners/speakers of Jawoyn. In this situation, it is not possible to repeat marriages between the same groups at each generation level. It is possible to alternate marriage preferences by generation. So, for example, for a male EGO, the preferred spouse could be a member of his FM's group, but members of his M's group would be proscribed spouses.

However, it is most unlikely that an alternating system of this type would lead to the construction of ranges, on average as narrow as those found in systems permitting genealogically close marriage. The general practicalities of finding a spouse entail that people must on average marry over a wider geographical range in areas where some groups of geographically close people are proscribed on the basis of genealogical closeness, than in areas where nobody is proscribed for this reason.

The other correlations cannot be explained in such directly causal terms, partly because the available materials are too limited to allow for the full testing of hypotheses. However, it is possible to propose some lines of enquiry. One factor which appears to be of importance is a central tension in Australian marriage practices. In systems such as those found in Australia which positively specify potential spouses, there is always a tension between the size of the pool of potential spouses available to each individual man, and the number of men competing for each individual woman. The wider the class of women designated as potential spouses within a particular terminology, then the greater the pool of men legitimately competing to marry each individual woman within that class. The more narrowly the class of potential spouses is constructed, then the fewer the men who can legitimately compete for each woman.

Given that polygynous marriage was a target of attainment for men, it appears reasonable to posit that most men would desire both to specify as wide a class of legitimate potential spouses for themselves as possible and to restrict the legitimate competition from other men for each individual spouse. However, these two desires cannot both be satisfied within Australian terminological systems. It is therefore unsurprising to find that there is variation across Australian terminologies in the size of class of potential spouses that they delineate.

It appears that the terminologies delineating the widest range of potential spouses are those of certain Western Desert peoples, and other peoples of western South Australia. These terminologies are notable for the fact that they do not distinguish parallel from cross kin. This means that there is a very wide class of EGO's own generation relatives who are potential first preference spouses, provided that they are 'distant' (Elkin 1940). It appears that this very wide potential class, not distinguishing distant cross from parallel kin, extends to owners/speakers of other Western Desert varieties which do allow for a cross vs parallel distinction (Scheffler 1978:98).

Of the widely attested types, Kariera terminologies delineate the broadest category of potential, first preference spouses, as they allow for marriage to all cousins, with the exception that close cousins are terminologically distinguished as non-marriageable in many systems. Aranda terminologies delineate a more limited category of first preference spouses, as they divide the preferred second cousin spouse from first cousins. However, for most owners/speakers of languages with Aranda terminologies, while first cousins are not the first preference spouse, they are second preference spouses (Berndt & Berndt 1985:51, (6) Maddock 198 1:65).

The exact degree of 'secondness' and how this is marked, varies somewhat. For Gidjingali owners/speakers, marriage to a first cousin arragudja is equally respectable to marriage to a second cousin mangga (Hiatt 1965:7 1). The terminology of Bunuba is Aranda, and in generic discussions of kinship terminology, Bunuba owners/speakers distinguish the first cousin gurndayi from the second cousin manggayi. However the usage of these terms does not in many cases reflect concerns of genealogy, but rather those of distance. Distant first cousins may be termed manggayi, and are marriageable. Conversely, some second cousins may be termed gurndayi (Rumsey 1982:161).

All the asymmetrical terminologies are like Aranda terminologies, in that in comparison to a Kariera terminology, they delineate a more limited class of first preference spouses. However, the marriage practices of Yolngu and Kunwinjku Oenpelli owners differ from those of owners of Aranda terminologies in that they involve a much clearer differentiation between first preference and other kinds of spouses. Marriages to 'cousins' not designated by the 'spouse' terms are not second preference marriages of the same degree of legitimacy as marriage to an arragudja among the Gidjingali. We have already seen that this is the case with the Yolngu terminologies. The Oenpelli Kunwinjku do allow a second preference within the broad kanjok 'cousin' class. However, Berndt & Berndt provide evidence that this is definitely a second preference marriage.

Otherwise, minor departures from the ideal marriage type do not attract much notice in the ordinary way. They are most likely to come to the surface in arguments and quarrels. A husband and wife in such circumstances have a ready-made grievance that either of them can use, even after years of marriage. They can accuse each other of being only ganjuig [kanjok] and not real gagali [kakkali], adding (for instance), 'Those [named] are my gagali -- I should be married to them, not to you!' 'My mother didn't give me to you [or vice versa]; you're not the right husband [or wife] for me!' (Berndt & Berndt 1970:101).

We may compare the Gidjingali, Oenpelli Kunwinjku, and Yolngu terminologies:

Gidjingali: First preference marriage to certain mangga (2nd cousin). Second preference marriage to other mangga and to arragudja (1st cousin) of equal respectability to first preference.

Oenpelli Kunwinjku: First preference marriage to kakkali (FZDDD). Second preference marriage to kanjok (cousin), not of equal status to first preference marriage.

Yolngu: First preference marriage to galay (matrilateral cousin). Marriage to a distant dhuway (patrilateral cousin) apparently possible, but only one of a number of low preference marriage possibilities -- not a secondary marriage preference.

One vexed question here is to what extent 'marriage practices' and the meanings of particular kin terms are to be distinguished from one another. Thus is arragudja simply 'first cousin' or is it 'first cousin and perfectly acceptable marriage partner'. Similarly is kanjok simply 'cousin' or is it 'cousin and possible but not preferred marriage partner'? I do not attempt to resolve this question, but simply note the Oenpelli Kunwinjku and Yolngu terminologies are associated with a much clearer differentiation of the class of first preference spouses from other types of spouses than is the case with Aranda terminologies.

Most discussions of marriage preferences do not detail the degree of legitimacy of second and lower preference marriages. This includes the available materials on marriage practices of other asymmetric terminologies. Consequently, it is not possible to determine whether the marriage practices of owners of asymmetric terminologies may be distinguished systematically from those of owners of Aranda terminologies in involving a much clearer differentiation of the class of first preference spouses.

From a pan-Australian perspective, the available materials suggest that there is an association between the extent of ranges, the size of the class of potential spouses and the size of the pool of men competing over each individual marriage. The largest class of potential spouses and thereby the largest set of competing men is found in areas of the Western Desert where the construction of geographically wide ranges is mandated. The smallest class of potential spouses and thereby the smallest set of competing men appears to be that found in areas associated with Yolngu varieties where the construction of geographically narrow ranges is mandated.

However, as we have seen no direct causal connection can be drawn between range size and size of the spouse class. Rather the connection appears to be one involving the appearance vs non-appearance of options. The possibility of a narrowly delimited spouse class arises only in areas which have a preference for narrow ranges which in turn are found only in the well-resourced parts of Australia. These options are absent in poorly resourced parts of Australia, for which only comparatively wider ranges and wider spouse classes are reported. Comparatively wider ranges and comparatively wider spouse classes are also options in the well-resourced areas of Australia, and in a certain sense appear to be the default options.

The reasons why the options of comparatively narrower ranges and spouse classes are adopted in particular areas remain to be investigated. Keen (1982) provides a detailed comparative study of the differences between Gidjingali and Yolngu marriage patterns, focussing on the high rates of polygyny among Yolngu in comparison with Gidjingali. Keen identifies a number of factors as contributing to this difference, including the wider allowance for 'straight' marriage among Gidjingali. This allows for a wider range of potential claimants in a woman's first marriage (Keen ibid.:631, 638). This suggests a study of comparative polygyny rates would be of interest.

KIN TERMINOLOGIES AND REGIONAL IDENTITIES

We have thus far been considering marriage practices and kinship terminologies only from the perspective of each individual marriage. However, marriage practices must also be considered from a group perspective. It is necessary to investigate whether the common possession of a particular type of marriage practice and/or kinship terminology correlates with the common possession of other practices and systems of nomenclature. If there is such a correlation, then the common possession of a particular type of marriage practice and/or kinship terminology can be viewed as partly constitutive of a regional identity.

This is the case in the Oenpelli area. The particular asymmetrical pattern found in the Oenpelli Kunwinjku terminology appears to have been characteristic of other kinship terminologies in the area extending north and west towards the coast from Oenpelli (Amurdak, Gaagudju, Garik, Iwadja, Marrgu). There are two other cultural patterns, whose long term geographical distribution correlates strongly with the range of the skewing. One is the Ubarr set of ceremonial rituals, which is autochthonous to this area (Berndt & Bemdt 1970:124). The other is a system of phratry names (Berndt & Berndt 1970:61--6), which linguistic evidence argues to be of some antiquity across much the same region as the skewing. Phratry terms in the various languages of the region are set out as follows.

Phratry 1: Amurdak warri-wudjali, Gaagudju [PHI]-yarrmangiiru (Masc), njing-garrmangiiru (Fem), Garik man-djirra-wudjali, Kunwinjku yarri-burrik

Phratry 2: Amurdak warri-arnkurrk, Gaagudju [PHI]-yarrangaalbu (Masc), njing-garrangaalbu (Fem), Garik man-djirra-wuli, Kunwinjku yarri-karnkurrk

Phratry 3: Amurdak warri-arninj, Gaagudju [PHI]-yarraadjawa (Masc), njing-garraadjawa (Fem), Garik man-balnggidj, Kunwinjku yarri-yarninj

Phratry 4: Amurdak warr-ugarr, Gaagudju [PHI]-yarrabarnaadjinggi, 0-yarrabarnaadju (Masc), njing-garrabarnaadjinggi, njing-garrabarnaadju (Fem), Garik yarri-wurrgan, Kunwinjku yarri-wurrgan

Phratry 5: Amurdak warri-marrangadj, Gaagudju djimburruwoodju (Masc & Fem), Kunwinjku djoned

A number of the phratry names are evidently related to one another, either in part or in whole, both within and between languages. However, the relationships are irregular, both within and between languages. The commonalities cannot therefore reflect recent borrowings, as these would show only regularities. The one exception is the Garik term for Phratry 4, which is evidently a recent borrowing from Kunwinjku. Sources from early contact list only the first three phratry names for the Cobourg Peninsula area, where Garik is found (Earl 1842:240--1, Spencer 1914:46).

The substantial geographical co-incidence of these three, unrelated patterns suggests a greater density of contact between people within the area than in other directions, at least for some periods of time. The common possession of the kinship skewing both partly constitutes this regionalised density of contact, and partly indicates the potential for re-newing the regionalised density of contact. It should be noted however, that there is no autochthonous name, in any of the languages of the area, referring to this regionalised density of contact. Consequently, it is an inchoate regional identity in comparison to identities reified by a name, such as 'upriver' people or equivalent named regional identities.

In other areas of Australia, particular types of kinship terminologies and/or particular patterns of marriage practices correlate with regional identities. One factor of considerable importance in determining regional identities is that of ecologically based oppositions. Sutton (1978:115) provides evidence for a coastal vs inland distinction in regional marriage patterns among Wik owners/speakers. This coastal vs inland distinction is also reported for marriage patterns in north-east and north-central Arnhemland (White et al. 1990:178). It seems likely that investigation of other oppositions suggested by Sutton (1990:75) of river vs hinterland, and plains vs hills, would also establish distinctions in marriage patterns.

There are also correlations with regional identities determined by language names. Merlan indicates that there was a regional aspect to marriage among Jawoyn language owners.

It would appear that marriage among people who considered themselves to be affiliated with the same identity [the Jawoyn regional land-language identity] of this kind was previously more common than it is now. Genealogical documentation in preparation for the Katherine Land Claim of all those people considered Jawoyn showed a much greater proportion of Jawoyn married to partners also recognised as Jawoyn in earlier generations, as compared with more recent marriages. (Merlan 1998:120).

The range of possible correlations between kinship terminologies and marriage practices, and other kinds of regionalised practices and named identities remains to be fully explored. The precise motivations for the various patterns of regionalisation also remain to be fully explored (see Keen 1997, Peterson 1976 for discussions of regional identities).

CONCLUSIONS

The kinship terminologies and marriage practices of Oenpelli Kunwinjku speakers/owners are of interest on a number of grounds. The asymmetry in the Oenpelli Kunwinjku terminologies is of a kind otherwise unreported. The form of the asymmetry overtly raises issues about the kinship status of mothers-in-law, as belonging to EGO's own generation and/or to the parental generation, which are otherwise only overtly raised by the Dyirbal terminologies. Covertly, this issue appears to be one of more general significance, as suggested by translations of the Kriol terms 'cousin' and 'poison cousin' with specifically 'in-law' kin terms across the western Top End and eastern Kimberleys.

This in turn raises issues for consideration with respect to the super- and sub-classing of kin terms and their significance. As Rumsey (1981) demonstrates, super- and sub-classing are contextually dependent, and not reducible to a single pattern. It appears that context can be extended to include generational and behavioural factors. While mothers-in-law may class with aunts for some purposes, they do not necessarily do so for all purposes.

While the particular structure of the asymmetry is unique, it classes broadly with the other, more well known, asymmetries found in kinship terminologies of Australian languages. The structure of the asymmetry revolves around the considerable discrepancy between the ages of a girl and her husband in her first marriage, as do the other asymmetries, apart from the 'clan' type asymmetry.

The Oenpelli Kunwinjku asymmetry also classes broadly with all of the other asymmetries in that its geographical associations do not appear to be entirely random. Asymmetrical terminologies appear only in areas where genealogically close marriage is permitted or prescribed. Prescription or permission of genealogically close marriage is in turn found only in areas where there is a preference for the construction of geographically narrow ranges, an option which is itself restricted to the better resourced areas of Australia. A preference for geographically narrow ranges is reported for many of the better resourced areas of Australia, and permission of genealogically close marriage appears to be not uncommon in these areas. Both prescription of genealogically close marriage and asymmetric terminologies are rare, even in the better resourced areas.

It is not possible to provide any definitive account as to why asymmetric terminologies might be adopted in particular areas because of limitations in the available materials. As we have seen Oenpelli Kunwinjku and Yolngu terminologies are associated with a clearer distinction between the class of first preference spouses and other spouses than is the case with other terminologies. Further research is required to determine whether this is true of other asymmetric terminologies. One effect of more clearly distinguishing the class of first preference spouses is to narrow the class of legitimate spouses, and to thereby narrow the pool of men legitimately competing for each individual woman. The effects of this remain to be fully investigated.

The full extent of the correlations between kinship terminologies, marriage practices and regional identities also remains to be explored. We have seen that common possession of the Oenpelli Kunwinjku skewing is one of the factors constituting a regional identity, though in this particular case a somewhat inchoate identity, as it is un-named. Kinship terminologies and marriage practices appear to co-incide with other regional practices or named identities, such as ecological zones or land-language associations. The particular factors relevant in each correlation require further research.

ACKNOWLEDGEMENTS

This paper is based primarily on many hours of discussion with Peggy Balmana (Mangub -- now deceased). She gave patient and considered answers to often repetitive questions, and I would like to acknowledge the centrality of her contribution here. I would like to thank Narelle and Stephen Etherington for providing information on the use of terminologies and marriage practices across the wider Oenpelli community. The comments of two anonymous referees were also very helpful. The fieldwork on which this paper is based was supported by grants from Parks Australia and the Australian Institute of Aboriginal and Torres Strait Islander Studies. The Gagudju Association also provided material assistance.

NOTES

(1.) Bininj Gunwok is the name for the overall technically defined language which includes the language varieties known as Kunwinjku. Bininj Gunwok includes a number of other named language varieties.

(2.) This paper focuses on first preference marriages as these are central to terminologies. Given that a woman's first husband was significantly older than her, most women were likely to have more than one husband in the course of their life. The patterning of second and subsequent marriages can be quite different from that of first marriages. Any study of overall marriage practices would necessarily encompass the practices accompanying these marriages. However, they lie beyond the scope of this paper.

(3.) S. Etherington (pers.comm.)

(4.) This information was confirmed by S. Etherington.

(5.) The preceding question mark is used to indicate sentences which are odd without being completely unacceptable.

(6.) The Berndts present this point in terms of subsections.

REFERENCES

ALPHER, B. 1991. Yir-Yoront Lexicon: Sketch and Dictionary of an Australian Language. Berlin: Mouton tie Gruyter

BERNDT, R. and C. BERNDT 1970. Man, Land and Myth in North Australia. Sydney: Ure Smith 1985. The World of the First Australians. Adelaide: Rigby

BLUNDELL, V. and R. LAYTON 1978. Marriage, Myth and Models of Exchange in the West Kimberleys. Mankind 11:231--45

BULMER, J. 1994. Victorian Aborigines: John Bulmer's Recollections 1855--1908. A. Campbell and R. Vanderwal (eds), Melbourne: Museum of Victoria.

CHASE, A. and P. SUTTON 1987. Australian Aborigines in a Rich Environment. In W. Edwards (ed.), Traditional Aboriginal Society: A Reader. pp. 68--95. Melbourne: Macmillan

DIXON, R.M.W. 1980. The Languages of Australia. Cambridge: Cambridge University Press 1989. The Dyirbal kinship system. Oceania 59: 245--68

EARL, G.1842. On the Aboriginal Tribes of the Northern Coast of Australia. Journal of the Royal Geographical Society 16:239--51

ELKIN, A.1940. Kinship in South Australia. Oceania 10:198--234

ELKIN, A., R. BERNDT and C. BERNDT 1951. Social organisation of Arnhemland: 1 western Arnhemland. Oceania 21:253--301

FALKENBERG. A and J. FALKENBERG 1981. The Affinal Relationship System: A New Approach to Kinship and Marriage among the Australian Aborigines at Part Keats. Oslo-Bergen: Tromso Universitetsforlaget

FISON. L. and A. HOWITT 1880. Kamilaroi and Kurnai. Melbourne: George Robertson

HEATH, J. 1980. Dhuwal (Arnhem Land) Texts on Kinship and Other Subjects, with Grammatical Sketch and Dictionary. Sydney: Oceania Linguistic Monograph 23

HIATT, L. 1965. Kinship and Conflict: A Study of an Aboriginal Community iii Northern Arnhemland. Canberra: Australian National University Press

1966. The Lost Horde. Oceania 37:81--92

1984. Your Mother-in-law is Poison. Man 19:183--98

KEEN, I. 1980. Alligator Rivers Stage II Land Claim. Darwin: Northern Land Council

1982. How Some Murngin Men Get Ten Wives: The Marital Implications of Matrilateral Cross-cousin Structures. Man (NS) 17: 620--42

1988. Twenty-five Years of Aboriginal Kinship Studies. In R. Berndt and R. Tonkinson (eds), Social Anthropology and Australian Aboriginal Studies: A Contemporary Overview, pp. 79--123. Canberra: Aboriginal Studies Press

1994. Knowledge and Secrecy in an Aboriginal Religion. Oxford: Clarendon

1997. A Continent of Foragers: Aboriginal Australia as a 'Regional System'. In P. McConvell and N. Evans (eds), Archaeology and Linguistics: Aboriginal Australia in Global Perspective, pp. 261--73. Melbourne: Oxford University Press

KILHAM, C., M. PAMULKAN, J. POOTCHEMUNKA, T. WOLMBY. 1986. Dictionary and Sourcebook of the Wik Mungkan Language.

MADDOCK, K. 1981. The Australian Aborigines: A Portrait of Their Society. (2nd ed.). Melbourne: Penguin Books

MCCONNEL, U. 1934. Wikmunkan and Allied Tribes of Cape York Peninsula. Oceania 4: 310--367

1940. Social Organisation of the Tribes of Cape York Peninsula. Oceania 10:434--55

MERLAN, F. 1988. Gender in Aboriginal Social Life: A Review. In R. Berndt and R. Tonkinson (eds), Social Anthropology and Australian Aboriginal Studies: A Contemporary Overview, pp. 17--76. Canberra: Aboriginal Studies Press

1990. Jawoyn Relationship Terms: Interactional Dimensions of Australian Kin Classification. Anthropological Linguistics 31:227--63

1997. The Mother-in-law Taboo: Avoidance and Obligation in Australian Aboriginal Society. In F. Merlan, J. Morton and A. Rumsey (eds), Scholar and Sceptic : Australian Aboriginal Studies in Honour of LR Hiatt, pp. 95--122. Canberra: Aboriginal Studies Press

1998. Caging the Rainbow. Places, Politics, and Aborigines in a North Australian Town. Honolulu: University of Hawi'i Press

MYERS, F. 1986. Pintupi Country, Pintupi Self: Sentiment, Place and Politics among Western Desert Aborigines. Canberra and Washington D.C.: Australian Institute of Aboriginal Studies and Smithsonian Institution.

PETERSON, N. 1971. The Structure of Two Aboriginal Ecosystems. Ph.D. thesis, University of Sydney

1976. The Natural and Cultural Areas of Aboriginal Australia. In N. Peterson (ed.), Tribes and Boundaries in Australia, pp. 50-71. Canberra: AlAS

RADCLIFFE-BROWN, A. 1913. Three tribes of Western Australia. Journal of the Royal Anthropological Institute 43:143--94

RIGSBY, B. and A. CHASE. 1998. The Sandbeach People and Dugong Hunters of Eastern Cape York Peninsula: Property in Land and Sea Country. In N. Peterson and B. Rigsby (eds), Customary Marine Tenure in Australia, pp. 192--218. Oceania Monograph 48 Sydney: University of Sydney Press

RUMSEY, A. 1981. Kinship and Context Among the Ngarinyin. Oceania 51:181--92

1982. Gun-gunma: An Australian Aboriginal Avoidance Language and its Social Functions. In J. Heath, F. Merlan and A. Rumsey (eds).. Languages of Kinship in Aboriginal Australia, pp. 160--181. Oceania Linguistic Monograph 24

SCHEFFLER, H. 1978. Australian Kin Classification. Cambridge: Cambridge University Press

SHAPIRO, W. 1981. Miwuyt Marriage: The Cultural Anthropology of Affinity in Northeast Arnhem Land. Philadelphia: Institute for the Study of Human Issues.

SHARP, L. 1934. The Social Organisation of the Yir Yoront Tribe: Kinship and the Family. Oceania 4:404--31

SPENCER, W. 1914. Native Tribes of the Northern Territory of Australia. London: Macmillan

STANNER, W. 1965. Aboriginal Territorial Organisation: Estate, Range, Domain and Regime. Oceania 36: 1--26

SUTTON, P. 1978. Wik: Aboriginal Society, Territory and Language at Cape Keerweer Cape York Peninsula, Australia. Ph.D. thesis, University of Queensland

1990. The Pulsating Heart: Large Scale Cultural and Demographic Processes in Aboriginal Australia.. In B. Meehan and N. White (eds), Hunter-Gatherer Demography: Past and Present, pp. 71-80. Oceania Monograph 39, University of Sydney

TAYLOR, J. 1976. Mapping Techniques and the Reconstruction of Aspects of Traditional Aboriginal Culture. Australian Institute of Aboriginal Studies Newsletter ns 5:34-43

THOMSON, D. 1934. The Dugong Hunters of Cape York. Journal of the Royal Anthropological Institute 64:453-537

1972. Kinship and Behaviour in North Queensland: A Preliminary Account of Kinship and Social Organisation on Cape York Peninsula. Australian Aboriginal Studies 51. Canberra: Australian Institute of Aboriginal Studies.

WARNER, W. 1930. Morphology and Functions of the Australian Murngin Type of Kinship I. American Anthropologist 32:207-56

1933. Kinship Morphology of Forty-one Australian Tribes. American Anthropologist 35:63-86

1958. A Black Civilization. (2nd ed.). New York: Harper and Row

WHITE, N., B. MEEHAN, L. HIATT and R. JONES. 1990. Demography of Contemporary Hunter-gatherers: Lessons from Central Arnhemland. In B. Meehan and N. White (eds), Hunter-gatherer Demography: Past and Present, pp. 171-85 Oceania Monograph 39, University of Sydney