This article has grown out of the frustration I felt a few years ago when I first tried to learn something about the subgroups of the Oceanic language family: there was no ready made up to date summary of what had been achieved in the field, nor of what was left to be done. One of the reasons that Oceanists may have felt reluctant to rush into print and produce an overview was perhaps the fact that, not surprisingly when there are almost five hundred languages in the Oceanic family, there still remains a fair degree of disagreement between scholars working in the field. Any field of academic enquiry, however, is in a constant state of flux. When old disagreements are settled, new ones are bound to emerge: a utopian state of agreement is never attained. Waiting for consensus is like waiting for Godot. The present article, then, is designed to present a snapshot of the state of the field as it is now. My purpose is not to try to resolve any of the disputes, but merely to give an indication of what is generally agreed upon, what disputes remain to be resolved, and to point the novice Oceanist in the direction of whatever literature s/he may need to read in order to find out more. Hopefully, the field will continue to make progress, and the 'use-by' date of this article will not be too far away.
The Oceanic languages are members of the Austronesian language family, a language family which, until the advent of European exploration and settlement of the 'New World', had spread out across a considerably larger proportion of the earth than had any other language family. Austronesian languages are spoken from Madagascar in the west to Easter Island in the east, and from Taiwan and Hawaii in the north to New Zealand in the south. Early European explorers marvelled at the fact that people who had obviously come from some common source had settled such a vast area of the globe, and the early students of Austronesian languages found a compelling motivation for their labours in the desire to learn what they could about Austronesian migration and settlement.
This paper will begin by sketching out the external relationships of the Oceanic family, and then examine the evidence for the existence of the group itself. The following section will cover the most contentious area of debate in the field: the dispersal of Proto-Oceanic (POC) and the existence or otherwise of large subgroups within the Oceanic family. The next section will briefly discuss some of the literature concerning the internal relationships of the smaller subgroups not already discussed. Next, I will make some conjectures about what the future may hold for the field. It is beyond the scope of the present article to discuss all of the material available on individual Oceanic languages in detail, but I will give some information on bibliographies and other more general material of historical interest at the end of the article.
Oceanic's closest relations to the west are thought to be the languages of the South Halmahera / West New Guinea (SHWNG) group (see Blust, 1978a for details). Blust (1984) and in other earlier papers, argued that beyond the Eastern Malayo-Polynesian (EMP) languages (Oceanic and SHWNG groups), was a Central Eastern Malayo-Polynesian (CEMP) group which consists of the EMP languages and those he dubs Central Malayo-Polynesian (CMP): the Austronesian languages of the Lesser Sunda Islands from Bimanese east, and of the southern and central Moluccas, including the Sula archipelago. Dyen (1978) had argued alternatively that these languages group more closely with those of western Indonesia, but now even the existence of a CMP group has been brought into question by Grimes (1991).
There has been much speculation about wider relationships beyond the confines of the Austronesian family. Many of these proposals (e.g links with Japanese, Indo-European, and Semitic languages) are not at all credible. Some more serious suggestions have been made, but little has appeared in print. The best candidates for language families having a close relationship with Austronesian ones are probably to be drawn from the Thai-Kadai and Austroasiatic groups, and perhaps with Miao-Yiao.
Amongst the innovations listed by Dempwolff were the loss of Proto-Austronesian (PAN) word-final consonants in POC, and a large array of phonemic mergers: according to Dempwolff PAN *e and *aw merge in POC, as do *b and *p; *g and *k; *t and *T; *n and *ñ; *r, *d and *D; and all of the PAN oral obstruents. According to Dempwolff, POC also allowed nasal clusters (*mp, *nt and *ngk) word-initially as well as word-medially (which was the only position they could occupy in PAN). The productive morphological process of nasalisation, according to Dempwolff, reconstructed for PAN, had disappeared in POC.
Although most Austronesianists have accepted Dempwolff's position on the existence of the Oceanic group, some modifications have been made to his original list of POC innovations. It is beyond the scope of this short article to go into all of the proposed modifications here: the interested reader is referred to Biggs (1965) (on generally adopted orthographic changes to Dempwolff's system), and to Blust (1978b), Capell (1971), Geraghty (1983), Grace (1969, 1976), Lynch (1978), Pawley (1972, 1973), and Ross (1988, Chapter 3, and 1989) for more substantive revisions.
There are always dangers in representing the the genetic affiliations of languages in terms of subgroups, particularly through the use of 'family trees'. Subgrouping is conceived of in terms of speech traditions splitting, while the conception can be false or misleading at best. Language differentiation may occur after innovations have spread across, for example, overlapping sections of a dialect chain. In other cases (e.g. Tokalau Fijian, see Geraghty, 1978) hitherto separate speech traditions may merge. I have eschewed the use of family trees in this paper, since I believe they often present a distorted view of what was really going on.
Another early proposal for subgrouping which should be mentioned is Dyen's (1965) large-scale lexicostatistical study of Austronesian languages. Dyen included data from around 100 Oceanic languages in his comparisons and showed a high level of lexical diversity in the Melanesian languages - so much, in fact, that the Oceanic languages were assigned to over 30 separate first level subgroups of Austronesian. One explanation which has been advanced for this high degree of lexical diversity among the Melanesian languages is the level of contact they have had with non-Austronesian languages in the region.
Although Dyen's conclusions have not been accepted by others working in the field, there is still one very striking parallel between Dyen's results, and those of Grace (and other subsequent commentators): all agree that there are a large number of well defined small groups of languages, but few agree about the existence of higher order subgroups.
Attempts at defining larger groupings have preoccupied a number of more recent studies. These efforts have generally concentrated on either the western or the eastern regions of Oceania. I will deal with each area separately.
Pawley also considered the position of Gilbertese (as a representative of the Micronesian languages) and Rotuman. These languages showed some characteristics symptomatic of Eastern Oceanic languages, but others which left him equivocal. Gilbertese and Rotuman were left unclassified. Although most of the detailed proposals for the large EO group have been overtaken by more recent classifications, Pawley (1972) was an important article since it was probably the first to present any detailed evidence for a putative large subgroup intermediate between OC and its smaller, better attested groupings.
Grace's (1976) review of Pawley's study was generally sympathetic, but he criticised Pawley for relying on a sole phonological innovation to support the EO group. In Pawley's (1977) paper he redefined the subgroup to exclude SES and renamed the rump 'Remote Oceanic'. He also proposed (Pawley, 1979) that Rotuman should be included as a branch of Central Pacific. Although more information concerning the Micronesian languages is now available there is still some equivocation about their position (see Pawley and R. Green 1984), and even now the position of the North and Central Vanuatu languages is not beyond question
Inclusion of more languages into putative groups intermediate between Oceanic and EO has been mooted, (see e.g. Lynch and Tryon 1985, who argued for the inclusion of South Vanuatu and New Caledonia languages in the group), but such proposals have been weakly supported and have gained little general acceptance. Since the evidence in support of any of these putative high-level subgroups is fairly meagre, the proposals have generally not been readily accepted, although little of the scepticism has been expressed in print.
The key to understanding the relationships between present-day languages probably lies in gaining a better understanding of the nature of 'proto-languages', or, to use a less-loaded term, the speech traditions from which the contemporary languages have developed. These issues have been more thoroughly addressed in the recent literature on the putative higher-level groupings of languages from the western part of Oceania. Although the details of 'subgrouping' are still sometimes contestable, a clearer picture of the range of situations which can lead to the differentiation of languages is emerging.
In a later paper, Pawley (1981), he spelled out his view of the prehistoric situation which had given rise to these difficulties in determining any large scale subgroups. Pawley saw POC as a long and rather complex dialect chain, for which we have evidence of '...various intersecting isoglosses, each reflecting an innovation which crosscuts the established subgrouping divisions but which does not itself define a larger subgroup because there are other isoglosses crosscutting it in turn' (Pawley 1981:280).
Ross (1988) tried to address these kinds of problems in detail, and began his study by isolating four 'clusters' of languages in the Western Melanesian region. These 'clusters', according to Ross, owe their existence to a genetic relationship, but not necessarily by virtue of being descendents of a common proto-language according to the Stammbaum model. The clusters isolated by Ross are an Admiralties cluster, a North New Guinea group, a Meso-Melanesian cluster, and a Papuan Tip group. The Admiralties cluster and the Papuan Tip group had been recognised by earlier writers (by Blust (1978b) and by, for example, Pawley (1975) respectively), but the other two groups subsume a number of smaller groups proposed by earlier writers. The North New Guinea cluster extends along the northern coast of New Guinea and through a large area of south-west New Britain, also taking in the languages of the Huon Gulf. It includes the SW New Britain, Astrolabe Bay, Manam-Schouten and Sepik groups from Grace's (1955) classification. Meso-Melanesian includes the Austronesian languages of the north and north-east of New Britain, and of the island chain which stretches from New Ireland through Bougainville, Choiseul to New Georgia and Santa Isabel, about half-way down the Solomon Islands. This group includes Grace's (1955) New Georgia, Choiseul, Buka-Bougainville and New Ireland-North New Guinea groups. The Papuan Tip languages are the Austronesian languages of the Papuan coast and nearby islands: the D'Entrecasteaux, Trobriands, Woodlark and Missima islands. I will return to the Admiralties languages in section 5, since they do not figure in Ross' larger Western Oceanic group.
It is in the treatment of the other groups that Ross' work is most innovative: Ross proposes a Western Oceanic group which subsumes all of the other three groups but which is not to be seen as involving a proto-language in the commonly accepted sense. He maintains that speakers of languages outside the Western Oceanic group migrated from the area in which POC was spoken, and that the languages of the Western Oceanic group evolved by a process of dialect differentiation from that point on. The three remaining clusters (North New Guinea, Papuan Tip, and Meso-Melanesian) spread out from the POC homeland later, but maintained linkages of various sorts, sometimes across groups.
The argument for Western Oceanic rests on five morphosyntactic developments: there appear to be no phonological developments shared exclusively by the whole the group (although POC *d and *dr have merged here, as they have in Micronesia, SES, and parts of North and Central Vanuatu, and POC *r and *R have merged in a very large number of Western Oceanic languages).
Papuan Tip is the most clearly defined subgroup within Ross' Western Oceanic: its essential unity has long been recognised in works such as Capell (1943), Grace (1955), and Pawley (1975). Pawley (1975) provided many of the arguments used by Ross to demonstrate the group's homogeneity. All the Papuan Tip languages show mergers between POC *r and *R, *d and *dr, *s and *c, and *n and *ñ (except in POC *ñamuk 'mosquito'). There were splits in POC *p and *k. POC *q also merges with the lenis reflex of *k. There were also a number of morphosytactic innovations (Ross, 1988:208ff).
The North New Guinea cluster is defined on the whole by losses rather than innovations, and hence the group is probably not as secure as the Papuan Tip group. While POC distinguished between a true object noun phrase and an 'incorporated object', which was treated as part of the verb, the distinction does not appear to be made in any of the North New Guinea languages. While POC appeared to have three categories of preposition-like morpheme, the North New Guinea languages only have one. In the North New Guinea languages, the POC transitive suffix *-i has been lost.
Ross' third cluster making up the Western Oceanic group, the Meso-Melanesian group have all, according to Ross (1988:265) undergone three phonemic mergers (of POC *r and *R, *d and *dr, *s and *c), while POC *p- split into PMM *p- and PMM *v-, and POC *-p- became PMM *-v-. A number of morphosyntactic innovations are also mentioned by Ross (1988:271ff), but the specifics of these are too complicated to be entered into here.
Probably the most important contribution made by Ross for Oceanic studies was that he turned his back on the Stammbaum model where it was clearly inapplicable, and that he described language change as occuring in terms of other models where they were appropriate. Perhaps this can best be illustrated by discussing the prehistoric situation which Ross sees as having given rise to the current linguistic situation in Oceania. He says that the POC language community probably inhabited an area no greater 'than the flat triangle whose apex is formed by the French Islands (where the Bali-Vitu communalects are now spoken) and whose base stretches from the islands of the Vitiaz Strait in the west along the north coast of New Britain to Lolobao Island (today the home of the Meramera) in the east' (Ross 1988:386). Proto-Western Oceanic would be the putative ancestor of the Oceanic languages now spoken along the northern coast of New Guinea, the Huon Gulf and the Papuan coast and nearby islands (the D'Entrecasteaux, Trobriands, Woodlark and Misima islands), as well as the languages of New Britain and the area stretching from New Ireland through Bougainville, Choiseul, New Georgia and Santa Isabel in the Solomons. This ancestral language could also be thought of as late POC since it was the language spoken by the people who stayed at home after the first migrant groups had left the area. This language, however, would have been in the process of disintegration, and the result of this would be the variious intersecting isoglosses noted by Pawley and discussed above.
The following section deals with the internal relationships between some of the smaller language groups which have not yet been discussed. For more details on the internal relationships of the groups not dealt with below, the interested reader is referred to the literature cited above.
The Sarmi Coast: Although Grace (1971) discusses aspects of the phonological history of the Oceanic languages west of the border between Papua New Guinea and Irian Jaya, he did not attempt to show explicitly that these languages form a subgroup of Oceanic. However, he did make some phonological observations which suggest that they may indeed form a subgroup (they have all undergone a merger of POC *R, *l, *d, and possibly *r, for example). Ross did not examine these languages in detail in his study either, and nor has anyone else to the best of my knowledge. The question of their closest relationships, if any, remains open.
North-West Solomons: Although these languages form part of Ross' Meso-Melanesian cluster, they are discussed again here, because their inclusion in the larger group may still be controversial, and because they have been discussed by a number of other writers. Tryon and Hackman (1983) for instance, established three groups (Choiseul, New Georgia, and Ysabel) which they proposed grouped together into a larger grouping of Western Solomons languages. Amongst the innovations they proposed for the group was the replacement of the first person singular disjunctive pronoun *(n)au with *(a)rau. They had not looked at the languages of Bougainville in their study, and Ross (1986) pointed out that this innovation was also shared by the Bougainville languages and redefined the group to include them. Ross (1988) discusses the internal relationships and the earlier commentators in more detail than I have space for here.
The South-East Solomons: It is commonly accepted that there are two major groups within the area: Guadalcanal-Nggelic (GN) and Cristobal-Malaitan (CM). Pawley (1972) spelt out the detailed evidence for these two groups. Tryon and Hackman (1983) provide a major survey of all the languages of the Solomon Islands and give details of the subgrouping within both the GN and CM groups. Their position on the CM group (that it splits into a Malaitan group and a Cristobal group) has been overtaken by the carefully argued view propounded by Lichtenberk (1988): that there are three major sub-divisions within the CM subgroup: Central and North Malaitan, South Malaitan-Cristobal, and an isolate; Longgu. As I have already said, the wider relationships of SES remain open to debate.
North and Central Vanuatu: Extensive surveys of all the languages of Vanuatu by Tryon (1976) and of the languages of south Malekula by Charpentier (1982) now provide a firm platform for comparative work to be built on. Clark (1985) describes in some detail the interrelationships within the North and Central Vanuatu subgroup, painting a subtle picture of the 'groups, chains, clusters and waves' which define their complex affiliations. Clark proposes a primary split between a northern group (consisting of all non-Polynesian languages from the Banks and Torres islands Efate, including Eastern Santo and the Malekula interior), and a central group containing all the other non-Polynesian languages north of Erromango. The major division in the northern group runs between Santo and Malekula, and between the Raga language (at the northern end of Pentecost island) and the rest of Pentecost. In the central group there is a primary division between Epi-Efate and the rest. For more details, refer to Clark (1985).
South Vanuatu: Lynch has been the major contributor to our understanding of the southern Vanuatu languages. Lynch 1978a examines the relationships between the languages of south Vanuatu (the languages spoken on Erromanga, Tanna and Aneityum. These languages have often been seen as somewhat aberrant. Although all of them are clearly set apart from the languages of North and Central Vanuatu, they are also quite different from each other, particularly lexically. Lynch demonstrates that the languages do all share some phonological innovations, however, and proposes that there is a primary split between an Erromangan group, a Tanna group, and Aneityum. Lynch (1983) examines the languages of Erromanga in more detail.
New Caledonia: The languages of New Caledonia and the Loyalty Islands have been relatively well described for a long time now, ever since the pioneering work of Leenhardt (1930) and (1946) was published. Until recently, however, comparative work on the languages of the region has not proceedeed very far. Grace, in a series of articles published over the last twenty or so years (see Grace, 1971, 1981, and in press) has pointed to some of the sociolinguistic circumstances that have made this work difficult. Haudricourt (1971) gives a fairly comprehensive overview of the area. Geraghty (1989) suggests that the languages of New Caledonia and the Loyalty Islands form a subgroup he calls 'Southern Oceanic', and that this group has its closest external relation in the South Vanuatu group. According to Geraghty, Southern Oceanic splits into a Loyalties and a New Caledonia group.
'Nuclear' Micronesian: The 'Nuclear' Micronesian languages are all those Micronesian languages which are members of the Oceanic subgroup. Some Micronesians (e.g. the people of Chamorro) presumably arrived in the area after migration from the Philippines area rather than from the Oceanic region as did the speakers of the nuclear languages. Bender (1971) and Bender and Wang (1985) detail the relationships of the subgroup. Jackson (1983) is the most comprehensive survey of the interrelationships between Micronesian languages.
Central Pacific: The fact that all the Polynesian languages form a closely related group has been clear since the days of the first European explorers in the region. One officer serving under Captain Cook reported his experiences on first arriving in Hawaii (over 3,000 km from Tahiti) thus: 'what more than all surprised us, was our catching the sound of Otaheite [Tahiti, J.B.] words in their speech; and on asking them for hogs, breadfruit, and yams in that dialect, we found we were understood' Beaglehole (1967:263). The geographic region of Polynesia can be thought of as a triangle, with New Zealand, Hawaii, and Easter Island at its points. Some of the languages, however, (the 'Polynesian outliers') are spoken outside this area, in Micronesia, Vanuatu, the Solomon Islands, and New Caledonia; their speakers presumably arriving in their current homes after late westward return migrations from within the triangle. Pawley (1966) set out the essentials of Polynesian subgrouping: a first split between Tongan and Niuean (the Tongic languages) from the 'Nuclear Polynesian' languages (all the rest); Nuclear Polynesian has a primary split between a western 'Samoic-Outlier' branch containg Samoan, Tokelauan, etc. and the outlier languages, and an Eastern branch which contains Maori, Rarotongan, Hawaiian, Tahitian, Easter Island and other languages.
After some initial doubt about Grace's (1959) proposals, it is now generally accepted that Rotuman is also closely related to the Polynesian languages (see Pawley 1979, Geraghty 1986). Fijian is also a member of the Central Pacific subgroup which includes Rotuman and Polynesian. Fijian, however, has a long and complicated history, whereby once very well-differentiated dialects have converged and become more alike. The details of this are too complicated to be discussed here: Geraghty (1983) explores the matter in detail. Geraghty (1986) examines the sound system of Proto-Central-Pacific.
George Grace, in a perceptive series of articles, (Grace, 1981, 1985, forthcoming) has pointed to the difficulties encountered by the classical comparative method in regions where multilingualism is as prevalent as it is in many parts of Oceania. The comparative method makes presumptions that changes in a linguistic tradition are changes in individual languages rather than changes in groups of languages. In reality, it may be that common changes spread across several languages simultaneously. In Oceania, it is often the case that people from neighbouring villages are all speakers of the languages of the neighbouring villages as well as that of their own. Participating in a sprachbund of this kind, change can spread throughout an entire area, across language boundaries. Some of the higher level subgroups may possibly be subgroups which share not a common proto-language, but perhaps a common 'proto-sprachbund', and different 'proto-languages'.
It is also possible that linguistics on its own will be unable to deal with all of the problems. Perhaps I should leave the last word on this matter here to Andrew Pawley (1978:145-146):
I doubt if linguistic research carried out in isolation from other disciplines will be sufficient to clarify the development and relationships of the Oceanic languages. Linguists surely need also to study the social, economic and geographic variables affecting the development of Oceanic speech communities. For example, in order to define subgroup boundaries and to evaluate reconstructive hypotheses one needs to know the extent and nature of borrowing between the languages concerned. In some cases this knowledge can be gained pretty well from the linguistic evidence alone. But generally a much better assessment can be made if there has been thorough ethnographic and archaeological documentation of the region in question, along with careful study of the way in which linguistic elements tend to move along trade routes and marriage-exchange networks. (Pawley, 1978:145-146).If what Grace and Pawley say does reflect the reality of the situation, then we will still need a more sophisticated understanding of the sociolinguistic scenarios involved before many of the outstanding issues are resolved.
Hollyman (1960) is the only published source I am aware of which provides a comprehensive listing of Oceanic languages, and it is still of some use in spite of its antiquity. Lincoln (1977) may be harder to track down and it does not provide alternative names for languages, but it is more current. Perhaps the best readily accessible reference to all of the Austronesian languages would be the atlas of Pacific languages edited by Wurm and Hattori (1981, 1983). Tryon (1993), although not yet seen by the writer, promises to become the standard first point of call for all interested in matters of comparative Austronesian linguistics.
Kleineberger (1957) is a comprehensive bibliography of sources on Oceanic languages, but of course it includes nothing written after 1957. Some of the works already mentioned in the text of this paper have quite extensive bibliographies, particularly Ross (1988), and Pawley (1972). Bowden (1992) has a large bibliography, and the surveys discussed below also contain useful bibliographies.
Sebeok, ed. (1971) contains survey material on all of the Oceanic languages, while Tryon and Hackman (1983) survey the languages of the Solomon Islands. Tryon (1976) surveys the whole of Vanuatu and Charpentier (1983) looks in detail at south Malakula. Wurm, ed. (1976) contains a wealth of material on the Oceanic languages of New Guinea.
This article has not addressed questions of historical syntax. Some useful sources are Pawley (1973) on Proto-Oceanic syntax generally, Clark (1976) on Proto-Polynesian syntax, Lichtenberk (1985) on the expression of possession, Clark (1973) on transitivity in Eastern Oceanic, and Lichtenberk (1986) on changes in syntactic categories.
Another area of potential interest to readers not mentioned in this article has been the question of cultural reconstruction. Pawley and R. Green (1985) is an extensive review of the linguistic evidence pertaining to the Proto-Oceanic language community. Pawley (1982) looks at the question of leadership in Proto-Oceanic society and Walter (1989) examines fishing strategies. Within the smaller subgroups, Pawley and K. Green (1971) discuss the evidence for the Polynesian homeland, and Marck (1975) discusses the origins of the Oceanic speaking Micronesians. A small explosion of interest in matters of cultural reconstruction is reflected in a number of forthcoming papers: Pawley and Pawley (forthcoming) examines Austronesian terms for canoe parts and seafaring, Pawley and Ross (forthcoming) examines Austronesian cultural reconstruction in general and provides a fairly detailed overview of Austronesian subgrouping, and Pawley and Ross, eds (forthcoming) will provide a wealth of material on these matters, some of it written purely from the perspective of linguistics, and some from other disciplines.
Many proto-language word lists are secreted away in computer databases in various academic institutions; I will confine my discussion to published sources. The most important general source is probably Wurm and Wilson (1975), an English finder list of a large number of Proto-Oceanic and Proto-Austronesian vocabulary. There have also been a number of reconstruction lists published in Oceanic Linguistics and the University of Hawaii Working Papers in Linguistics: Blust (1972), Grace (1969), Milke (1968). Walsh and Biggs (1966) is a listing of some Proto-Polynesian vocabulary.