Chapter 6.3 Was Proto Oceanic society matrilineal?

Proto Oceanic society is believed to have developed in the region of the Bismarck Archipelago in western Melanesia around 1500 BC.¹² Within a few hundred years daughter societies expanded eastward into the rest of Melanesia and into nuclear Micronesia, eventually reaching the remotest islands of Polynesia by 1000 AD (Kirch 1997).³ Lexical reconstruction has revealed a great deal about Proto Oceanic economy, technology and material culture [Ross, Pawley and Osmond 1998]⁴ but very little about aspects of social organisation. In particular there is considerable uncertainty about the type and even the existence of descent groups. My purpose is to suggest, through a combined analysis of historical linguistic, ethnographic and cross-cultural data, that POc society had unilineal, probably matrilineal descent groups and unilocal, probably matrilocal or matri-avunculocal residence rules.

Conjectures about descent groups in Proto Oceanic [read “early Oceanic”] society have a long history with contributions by anthropologists, linguists and archaeologists. Rivers (1914) assumed that early Melanesian society, like all early societies, was matrilineal on the grounds that only maternal kinship connections could be known for certain. He thought that matrilineal moiety systems developed early in Melanesian history as the result of a fusion between indigenous peoples and later arrivals, and he attributed shifts to patrilineal descent and succession to the influence of immigrants with an advanced culture. As M. Allen (1984:27) observed, Rivers’ argument was not really refuted, but simply “abandoned as a result of the general demise of evolutionary anthropology in favour of structural functionalism”. In a variation of Rivers’ scenario, Allen proposed that matrilineal descent groups were a precondition of social stratification in Melanesia, the argument being that that their restrictions on membership encouraged the formation of larger non-kinship political groups such as the graded societies of Vanuatu.

Several decades after Rivers, Murdock (1949) concluded from the application of an evolutionary algorithm that Proto Malayo-Polynesian and Proto Oceanic society were bilateral or “Hawaiian” in type and lacked descent groups of any kind. By Hawaiian, Murdock meant a cluster of traits including generational-Hawaiian kinship terminology, bilateral kindreds and bilocal extended families as well as the absence of unilineal descent groups. Goodenough (1955), noting that land tenure is generally associated with kinship groups in Malayo-Polynesian societies, amended Murdock’s conclusion by adding to it the feature of cognatic landholding descent groups. Linguistic evidence in support of Murdock’s reconstruction was provided by Milke (1958) who interpreted the basically generational-Hawaiian structure of POc kinship terminology, in which M = MZ = FZ and G = PGC as consistent with a bilateral form of social organisation.⁵ Evidence against the bilateral hypothesis was adduced by Blust (1980a) and Hage and Harary (1996) who showed that a strict application of Murdock’s algorithm leads to the conclusion that POc society was just as likely to have been “Iroquois” in type with either matrilineal or double descent.

In a recent work, Kirch (1997) proposed on general ethnographic grounds, that Lapita, i.e. Proto Oceanic, descent groups were cognatic in type:

Social anthropologists have long been aware that Oceanic peoples do not organize themselves onto strict unilinear descent groups but rather tend to have more flexible cognatic (sometimes called “ambilinear” or “non-unilinear”) systems of descent reckoning” (Kirch 1997:189-90)

While it is true that there are many cognatic societies in Oceania, they do not predominate in all regions and where they do exist, they may not be a continuation of POc society but, rather, a later development.

In a linguistic approach to the problem, Pawley (1981) reconstructed a POc term *kainaŋa as meaning ‘descent group’ and, following Goodenough (1955) as ‘landholding descent group’ [but see §4.2.2.6]. However, Chowning (1991) objected that *kainaŋa may be attributable only to a lower-order branch of the Oceanic family tree and she argued that even if *kainaŋa, or an equivalent term *qapusa, could be attributed to POc,

we would have no clue as to what kind of descent group or category might be represented. For the former, cognate terms designate a patrilineal group in Tikopia [a Polynesian outlier in Melanesia], a matrilineal one in Truk [in Micronesia] and a cognatic one in Maori [in Polynesia]. Furthermore, the kinship terms reconstructed so far for POc do not solve the problem. If POc had unilinear descent, we would expect either that a term for cross-cousin was reconstructable, or that many more of the societies would be like Truk and the Trobriands in having kinship systems (Crow, in these cases) that group cross-cousins with other kin types. I would also expect a reconstructable term for FaSi [father’s sister] unless, as in Kove, she was called by the same term as MoBro [mother’s brother], but this does not seem to happen in many Melanesian societies (Chowning 1991:70).

The problem of descent in POc society is not as insoluble as it might appear. POc kinship terminology (Milke 1958), like that of Proto Polynesian (Marck 1996) and Proto Malayo-Polynesian (Blust 1980a) terminology, was bifurcate merging for males in the first ascending (+1) generation: F = FB ≠ MB (POc *tama ‘F, FB’, *matuqa ‘MB’. [See ch. 2 of this volume for a detailed reconstruction of POc kinship terms.] It is uncertain for all three of these protolanguages whether +1 terminology for females was also bifurcate merging (M = MZ ≠ FZ) or generational, as Milke (1958) thought, but as far as inferences about descent are concerned it does not matter at all. Nor does it matter that the POc terminology lacked terms for cross-cousins (PssGC ≠ PosGC) or that Crow type terminologies (FZ = FZD, F = FZS) are not more common in ethnographically known Oceanic societies. Generalising from an earlier study by Murdock (1947), the presence of a term for MB alone is sufficient to establish that descent in Proto Oceanic, Proto Polynesian and Proto Malayo-Polynesian society was almost certainly unilineal.⁶ As shown in Table 3.1, bifurcate merging terminologies for +1 males in non-unilineal societies are almost always found together with unilineal descent groups (patrilineal, matrilineal or both) 85 percent of the time.⁷ The rare cases of bifurcate merging terminologies in non-unilineal societies are best interpreted as survivals from earlier unilinear states, under the assumption that changes in kinship terminology usually follow changes in descent rules (Lowie 1948; Murdock 1949; Fox 1967). As shown in Table 3.2, bifurcate merging terminology is also associated with unilocal residence—91 percent of the time.⁸ Proto Oceanic society was either patrilineal and patrilocal, or matrilineal and matrilocal or matri-avunculocal. Using the same cross- cousin sample as for Tables 3.1 and 3.2, 81 percent of the matrilineal societies are either matrilocal or avunculocal.

The case of a separate term for FZ is interesting and deserves a comment in view of its uncertain existence in POc, PPn and PMP kinship terminologies (Pawley 1981; Marck 1996; comments on Blust 1980a [and §2.5.1.2.6 of this volume]).⁹ In Oceanic the presence of a term for FZ almost always implies the presence of a term for MB. This is displayed in Table 3.3 which is based on the terminologies reported in Murdock (1970) for all the Oceanic-speaking societies in the World Ethnographic Atlas (WEA) (Murdock 1967).¹⁰ The implicational relationship between these two terms is an expression of a marking rule in which the presence of the marked term (FZ) implies the presence of the unmarked term (MB) but not necessarily conversely (Greenberg 1980).¹¹ In most Oceanic as in most Polynesian and Malayo-Polynesian kinship terminologies (Hage 1996, 1998a; Hage and Harary 1996; Blust 1980a) ‘male’ is the unmarked term. Diachronically interpreted, if POc, PPn and PMP terminologies had a separate term for FZ it was lost before the term for MB.

The evidence for matrilineal rather than patrilineal descent in Proto Oceanic society is distributional, linguistic and historical in nature. As Rivers (1914) and M. Allen (1984) emphasised, matrilineal descent groups are found extensively in many areas of Melanesia including

the Huon Gulf, New Britain [east of the Willaumez Peninsula] and New Ireland [in the Bismarck Archipelago], the Massim Archipelago [apart from the northern d’Entrecasteaux], Bougainville, parts of the Solomons and much of north and central Vanuatu (M. Allen 1984:26). [Brackets are Hage’s]

Double descent groups are also found in Melanesia, e.g. Bunlap [Sa, south Pentecost] (Tattevin 1928), Vanua Levu [Fiji] (Quain 1948) and also in Micronesia, e.g. Pingelap (Damas 1979) and in Polynesia in Pukapuka (Beaglehole and Beaglehole 1938). If Murdock (1940, 1949) was right, one of the origins of double descent is the intrusion of patrilineal institutions into a strongly integrated matrilineal system. In a survey of double descent systems Murdock found that exogamy is everywhere associated with both matrilineal and patrilineal groups, but that inheritance and succession are almost always patrilineal. Residence is always patrilocal and political organisation is patrilineal.

In Table 3.4, all the Oceanic-speaking societies in the WEA are classified by type of descent group. In Polynesia, the last region of Oceanic settlement, descent groups are almost entirely cognatic; in nuclear Micronesia they are almost entirely matrilineal; and in Island Melanesia,¹² the homeland of Proto Oceanic society, they are predominantly matrilineal or formerly matrilineal, as attested by systems of double descent.

With a larger sample of societies, the rows in Table 3.4 could specify major subgroups of the Oceanic language family. On the basis of available data in the WEA, matrilineal and double descent are not confined to societies in any one subgroup. It must be noted that anthropologists are sometimes uncertain about the existence of double descent, Thus Hogbin changed his mind about double descent in Wogeo, referring in 1970 to matrilineal moieties and agnatic residential groups. Schneider (1984) changed his earlier view of Yap (Schneider 1961) when he, most unfortunately, came to reject the entire metalanguage of kinship analysis.

Historical linguistic evidence for matrilineal descent in Proto Oceanic society is provided by Blust (1986-87) whose argument may be summarised. In the Austronesian languages terms for ‘orphan’ are in some cases monomorphemic as in Bontok so ‘be an orphan’ but in other cases descriptive as Malay anak piatu (anak ‘child, piatu ‘desolate, orphaned’). In some Oceanic languages the descriptive term refers to one or both parents. In a sample of 12 Oceanic languages, Blust found that in eight languages there is an agreement between the parent mentioned in the descriptive term and the rule of descent. For example, in Ere ‘orphan’ is timan pwi ‘father none’ and descent is patrilineal; in Woleaian ‘orphan’ is sile-mas and descent is matrilineal; in Samoan ‘orphan’ is matua-oti ‘parents dead’ and descent is ambilineal (cognatic). This correlation suggests that in Austronesian societies an orphan is defined in relation to his/her descent group rather than his/her parents. The four exceptions are Kwara’ae, Lau, ’Are’are and Sa’a, all of which are spoken in patrilineal societies in Malaita in the southeast Solomons. In all four languages there is a pair of terms reflecting Proto Malaita-Cristobal *tina mate ‘mother dead; and *tina mauri ‘mother living’ which refer to both parents, or to the father alone, and to persons of high and low status.

It is noteworthy that a paired term reflecting earlier *tina mauri (“mother living”) is found widely in the Southeast Solomons (but not in Micronesia). Compare LAU (C.E. Fox 1974) inamae/inomae ‘orphan, relatives dead, poor and unprotected’, inamauri/inomauri ‘parents alive, prosperous, important’, inamouri ‘eldest son of a living chief’; (Catherine Tyhurst, pers. comm.) inomae ‘eldest son of a deceased man’, faa-inomae ‘to bereave’.—‘ARE’ARE (Geerts 1970) inamae ‘orphan’, inamauri ‘a very big chief, a person of very great importance’.—SA’A (Ivens 1929[a]) inemae ‘be an orphan, be bereft of parents, an orphan’, inemauri ‘be a chief, have a due succession of chiefs, a chief’.—AROSI (C.E. Fox 1970) inemae ‘child whose mother is dead, orphan’, inemauri ‘child whose mother is living’ (Blust 1986-87:220).

Blust interprets the discrepancy between the proto-term for ‘orphan’, and its reflexes as evidence for a shift from matrilineal to patrilineal descent in Malaita and by implication in the “wider Oceanic context”. Under this interpretation the correlation between lexical glosses of terms for ‘orphan’ and rules of descent in his sample of Oceanic-speaking societies is “exceptionless”.

Specific historical evidence in favour of Rivers’ matrilineal hypothesis was adduced by Lane (1961). Using kinship data collected by different observers over a period of nearly 100 years, Lane described the breakdown of matrilineal social organisation in two Vanuatu societies, Mota in the Banks Islands and Barabet in Pentecost. Matrilineal clan and moiety organisation was weakened or disappeared, residence (in Barabet) became bilocal and Crow kinship terminologies [cf. §2.3.3], which are usually associated with matrilineal descent, gave way to modified generational Hawaiian terminologies [cf. §2.3.1]. Lane attributes these changes to the effects of sudden and extensive depopulation which fatally weakens more rigid lineage systems and moves them in the direction of more flexible bilateral systems. Generalising this result, Lane, following Dole (1967), concluded that the bilateral organisation of Polynesian, especially eastern Polynesian, societies developed in response to the difficulties faced by small unstable populations settling widely separated islands.

Kinship systems are generally conservative in nature: they change slowly, sometimes glacially, under normal (non-acculturative) circumstances, and they are more resistant to diffusion than other cultural traits (Dyen and Aberle 1974; Eggan 1955; Murdock 1949). If Proto Oceanic society was, in fact, matrilineal then one might expect a continuation of matrilineal traits in its descendants. The converse would be true if Proto Oceanic society was patrilineal. Data for testing this hypothesis come from a recent study by Burton et al. (1996).

Burton and his colleagues, building on the work of Murdock (1967), have shown that nine world regions can be distinctively characterised by two dimensions of social structure: gender and descent. These two dimensions score 63 different traits of social structure.¹³ The first dimension contrasts matricentric and patricentric traits of social organisation and kinship terminology.

In order of strength from highly to weakly positive,

[m]atricentric social organization traits include matrilocal or uxorilocal residence, monogamy, and the absence of marriage exchange.¹⁴ Hence, matricentric societies tend to organize kinship groups around women through matrilocal or uxorilocal residence or through matrilineal kinship groups (Burton et al. 1996:93)

In order of strength,

[p]atricentric social organization traits include nomadic or seminomadic settlement patterns, clan communities, localized or dispersed patrilineal groups, patrilocal residence, polygyny, and bridewealth payments. Hence, patricentic societies tend to organize kin groups around men through patrilocal residence, patrilineal descent or polygyny… Strongly matricentric kinship terminologies include generational aunt terms, bifurcate merging aunt terms and Crow cousin terms. The former two terminologies classify mother and mother’s sister together, as one would expect of societies that keep related women together after marriage, and Crow cousin terms are well known to be associated with matrilineal descent. Strongly patricentric kinship terminologies include bifurcate collateral aunt terms and Omaha cousin terms. Bifurcate collateral terminologies assign separate terms to mother and mother’s sister, as one would expect of societies that separate women after marriage, and Omaha terms are well known to be associated with patrilineal descent.

The second dimension…contrasts unilineal and bilateral traits… Unilineal social organization traits include clan communities, dispersed or localized patrilineal groups, dispersed matrilineal groups, patrilocal residence, nonsororal polygyny, cousin marriage, patrilocal residence, and bridewealth payments. Bilateral social organization traits include bilateral kin groups, ego-centered kindreds, virilocal residence, bilocal residence, monogamy, and prohibition of cousin marriages.

Applying the method of correspondence analysis (Greenacre 1984) these two dimensions define a space in which societies can be located by their social structural traits. Halves of this space are matricentric versus patricentric and unilineal versus bilateral, while quadrants are matricentric and bilateral versus matricentric and unilineal, and so on. Two of the nine world regions identified by Burton et al. are the Southeast Asia and the Insular Pacific (Micronesia and Polynesia) region, which is matricentric, and the Australia, New Guinea and Melanesia region, which is unilineal. Since the authors give the rating on each dimension for every society in their world-wide sample, it is possible to define a region strictly by language group. We will consider only the Oceanic-speaking societies in Polynesia, Nuclear Micronesia and Melanesia, cross-cutting the two worlds just mentioned.

If Proto Oceanic society was strongly matrilineal, one would expect to find a continuation of matricentric traits in its descendants, including those that later developed different descent rules. Support for this hypothesis is given in Figure 3.1. With two exceptions (Seniang [Sinesip] and Lau, Fiji) all the Oceanic-speaking societies in the WEA are matricentric whatever their rule of descent—cognatic, patrilineal, matrilineal or double descent.

We conclude, on the basis of historical-linguistic, ethnographic and cross-cultural evidence, that descent in Proto Oceanic society was matrilineal and residence matrilocal or matri-avunculocal. This result would account for certain unusual ethnographic facts, such as the presence of Crow kinship terminology and patrilineal descent in Seniang [Sinesip] (a case in which changes in kinship terminology, as usual, lag behind changes in descent), marriage with the MBW (Rivers 1914) (a practice commonly found in matrilineal societies of generalised exchange; Lévi-Strauss 1969, Hage 1998b), and double descent (the undermining of a strongly integrated matrilineal system by patrilineal institutions).

It may be, as Lane (1961) and Dole (1967) have argued, that demographic and geographic constraints—small unstable populations and great distances—led to the development of more flexible cognatic social organisation in the settlement of Polynesia. It may also be the case that relative isolation after settlement led to the weakening or disappearance of matrilineal descent in some Oceanic societies. As Lévi-Strauss (1984:183) observed with respect to Micronesia, in those societies that have lived in relative isolation, “we find…a retreat from matrilineal institutions: left to themselves these institutions, by reason of their well-known instability, have a tendency to evolve spontaneously towards other forms.” Damas (1979) makes this point with respect to isolated atolls in the Carolines—Pingelap, Mokil and Ngatik—where matrilineal descent was left with few functions other than exogamy:

I would argue that the comparative vitality of matrilineal emphasis in the Yap and Truk districts is closely related to regular reinforcement of those ties through the process of clientship, trade relations, and (in the case of Yap) a system of tribute which operated largely within a matrilineal context. By contrast, relative isolation of the atolls and islands in the eastern Carolines appears to have promoted conditions which serve to weaken matriliny (Damas 1979:192).

One should also mention here the two isolated outliers, Enewetok and Ujelang, in the solidly matrilineal Marshall Islands which evolved patrilineal descent groups (moieties) (Hage and Harary 1996). According to Ann Chowning (pers. comm.) the majority of Oceanic-speaking societies in the New Guinea mainland are patrilineal rather than matrilineal. In this case, rugged terrain rather than geographical distance was a barrier to social communication which led to the weakening or disappearance of matrilineal institutions.

The hypothesis of matrilineal Proto Oceanic social organisation should have interesting culture historical implications. Rivers, as mentioned, attributed moiety systems, which are common in Melanesia, to the fusion of indigenous and immigrant peoples. As evidence he noted the presence of inter-moiety hostility, native traditions of separate origins and the attribution of physical and mental differences to members of different moieties. As Blust (1981c) has observed, there is no linguistic evidence for the “fusion hypothesis”, i.e., no significant differences in vocabulary, phonology or grammar between members of different moieties in Oceanic societies. An alternative, more plausible explanation, which applies to moiety systems in general, derives from a theory of Murdock’s (1940, 1949). Typically, when a matrilineal society fissions, the new community, in order to preserve the proximity of males to their natal groups, consists of two exogamous lineages. Eventually these two groups may develop onto a moiety system with all its attendant symbolism. Could such a process account for the prevalence of moiety systems in Melanesia?¹⁵

Notes⇫