This website is a free e-resource, from the book entitled 'Music, Lapita, and the Problem of Polynesian Origins' by Mervyn McLean. The complete text may be read either online or in paginated form as a downloadable pdf here.


Chapter 5


During the past several decades, linguists have worked closely with archaeologists in Oceania to provide context for the dating of Lapita sites, the probable origin of Lapita peoples, and for working out the nature of Lapita society.


Dempwolff and the comparative method

The comparative method is a technique used by linguists to examine systematic sound changes in languages in order to infer relationships among them. Thus, to provide a simple example, the Tahitian song type pehe, is "reflected" as pe'e in Rarotonga but pese in Samoa. This is shown not to be a chance resemblance when it is found that other words containing an "h" in Tahitian, also have and "s" in Samoan, and a glottal in Rarotongan. But to trace the origin of the term it is not sufficient to judge from only three examples of it. In this case, the word is found in numerous other Polynesian languages as well, but not in Tonga or Niue, possibly excluding them from the ancestral language common to most of Polynesia, and suggesting an origin a step closer to the present.  


The Austronesian subgroups

The comparative technique was used most famously in the 1920s and 30s by the German linguist Otto Dempwolff whose contribution to Oceanic linguistic was profound.

Already known when Dempwolff began his work was the over-arching existence of a vast language family, now known as Austronesian, containing nearly all the languages of Island Melanesia, Eastern Micronesia, and Polynesia, together with others in SE Asia and as far afield as Madagascar. Dempwolff's great achievement was to prove the existence within Austronesian of a large subgroup of languages now called Oceanic, made up of all of Near and Remote Oceania, and subject to investigation and refinement by linguists ever since it was first proposed by Dempwolff.

Successive subgroups from oldest to youngest within the larger Austronesian family with which the present book is mostly concerned are:

Proto Austronesian

Proto Malayo Polynesian

Proto Oceanic

Proto Eastern Oceanic 

Proto Central Pacific

Proto Polynesian

Within Oceanic, a minimum of three subgroups, Admiralties, Western Oceanic, and Eastern Oceanic has until recently been recognised for purposes of reconstructing proto Oceanic terms.


Eastern Oceanic Subgroup

For the best part of 40 years since a detailed discussion of the Eastern Oceanic subgroup by Pawley (1972), this term has been standard in linguistic tree diagrams as a label for all of the languages spoken in what is now known as Remote Oceania. Although evidently still in use as late as 2004 by the Australian National University Oceanic Lexicon Project (ANU 2004), it has since been discarded, and is now replaced by separate high order subgroups for most of its former components (Pawley 2007, Fig. 3). This leaves no steps at all relevant to Polynesian between Oceanic at the beginning of the tree and the Central Pacific subgroup at the end of it which contains Fijian, Rotuman, and Polynesian. As Lapita potters demonstrably did not traverse this distance in a single hop, and would have spoken related dialects on their way, it seems justifiable to retain the older designation of Eastern Oceanic as an umbrella term in the present book, especially as it cannot be avoided when quoting from earlier literature, but with the proviso that it may have been no more than a transitory dialect chain in most of the areas where it was once spoken.

Western Oceanic is a complex of loosely related dialects that developed after initial Lapita colonists had left the area (Bowden 1993); and Admiralties is of special importance because of indications that it shared a period of development with St Matthias (ANU 2004), and the presence in these places of Lapita sites of Manus and Mussau (Specht 2007:Table 2).

Also of obvious importance is the later history of the rump group of the Western Oceanic dialect chain as this extended its range beyond the Bismarck Archipelago. This broke up into three subgroups named Meso Melanesian, reaching as far as the NW Solomons; Northern New Guinea, along the north coast of New Guinea; and Papuan Tip, in the Northern, Milne Bay, and Central provinces of Papua New Guinea (Pawley 2007:22). These are frequently referred to in cognate sets cited in later chapters of the present book, abbreviated MM, NNG, and PT respectively. Other abbreviations in the cognate sets refer to subgroups of Eastern Oceanic occupying the remainder of the Solomon Islands, Vanuatu, New Caledonia and the Loyalty Islands, and Fiji. For a full list of subgroup abbreiviations see Appendix 4. 

The languages of Micronesia fall into three groups. On the southern fringe there are two Polynesian Outliers: Nukuoro and Kapingamarangi which, in common with other Outliers, have languages that are Samoic in origin and were settled by back-migration out of Western Polynesia. In Western Micronesia, the Mariana Islands, Yap, and Palau have language affinities with

Malayo Polynesian languages of the Philippines and Indonesia, and are thought to have been settled directly from these places. All of the other languages of both Western and Eastern Micronesia form a single large language family called Nuclear Micronesian.

It is probably true to say that no one really knows where Eastern Micronesians came from, with both the Solomon Islands and Vanuatu nominated in the past, and no consensus reached on either, despite the revision of subgroups referred to above. One may speculate that this issue will not be resolved until difficulties with the Lapita hypothesis have been settled and the exact relationship of Micronesia with Polynesia has been determined.

Although speakers of Austronesian languages form a majority in Oceania they are not quite exclusive to it. In New Guinea they are outnumbered by non-Austronesians, with speakers collectively known as Papuan occupying most of the western half of New Guinea as well as the Highlands of Papua New Guinea, and are present in enclaves as far east as the Solomon Islands. They have been in New Guinea for millennia longer than the Austronesian peoples and have interacted with them in important ways. 


Quantitative Methods

Lexicostatistics is a method of language comparison devised in the 1950s by the linguist Morris Swadesh. It makes use of a list of 200 word meanings, such as father, mother, ear, eye, sit, stand etc., for which terms are believed to exist in all languages. For each pair of languages under scrutiny, a count is made of the number of words for which terms of similar form are found, such as father in English and vater in German. The percentage of such hits provides a measure of cognacy between the two languages. Glottochronology is a later application of Swadesh lists which has the more ambitious objective of calculating the time lapse between pairs of related languages. Both methods have been controversial since their inception because of assumptions that do not always hold true, and results that may be at odds with other approaches. For such reasons the method has now largely been abandoned in favour of more rigorous dates that are now available from radiocarbon dating of archaeological sites. 


Reconstructing the past

The current state of knowledge has been reached by linguists essentially through a process of moving from the known to the unknown, both geographically and through time, working back from the most familiar and best-studied languages, which by and large were Polynesian, and adding others as knowledge of interrelationships increased. Emphasis to begin with was mostly on Proto Polynesian, with a homeland somewhere in Western Polynesia, with interest shifting progressively back through the tree of language affiliations and linguistic subgroups, first to Proto Oceanic, where the homeland was probably in the Bismarck Archipelago, and most recently to Proto Austronesian and an origin most likely in Taiwan. In all of this effort it has been reconstruction of proto terms that has proved crucial, as will be seen, especially, in chapters 6, 11, 12, and 13. 

At each stage in the process, the object has been to find shared innovations whose origin could be pinpointed to particular levels of the family tree. Thus, in the domain of music, did the speakers of Proto Polynesian possess the Eastern Polynesian word for skin drum, and if so did it mean the same thing, or was it a general term that meant something different to which the new meaning was later assigned? As will be seen, the last possibility has turned out to be the case, and similar transformations have occurred for other terms, illustrating just one of the pitfalls of the reconstruction process. 

A step further back still the difficulties multiply when an effort is made to determine the cultural inventory of the Lapita potters. Thanks to a mammoth Oceanic Lexicon Project at the Australian National University we now know a great deal about the material culture of the potters as well as their food plants and much else, with more to come. But the musical and other evidence in the present book again suggests unresolved problems with some of the subgroups that are crucial to the Lapita hypothesis.


Limitations of subgrouping

It has to be accepted that all of the languages of the Austronesian language family spoken throughout Island Melanesia, Polynesia, and Eastern Micronesia belong ultimately to the single subgroup of Oceanic, whose initial location was somewhere in the region of the Bismarck Archipelago. The internal relationships of the Oceanic languages, however, are another matter. The huge number and diversity of Melanesian languages and cultures compared with the smaller number and equally striking homogeneity of Polynesian languages and cultures has been the subject of much debate and speculation since Otto Dempwolff first advanced his famous Oceanic linguistic hypothesis in the 1930s (see Pawley 1981 for a discussion). It is a debate that has become closely associated with the problem of Polynesian origins insofar as this relates to members of the hypothetical Eastern Oceanic subgroup which emerged from the breakup of Oceanic. Multiple interactions undoubtedly occurred, both between Papuan prior occupants of the area and Austronesian languages, and among Austronesian languages themselves, all contributing to the present diversity. Representing all of these languages as members of a single branching "tree", however, has probably unduly contributed to acceptance of the standard Lapita hypothesis, by providing a false sense of progressive stops or stations along the path taken by the putative "fast train" or even "slow boat" proposed for the Lapita dispersal.

Pawley (1981:274) has suggested that the linguistic homogeneity now characteristic of Polynesia represents a situation that must once have been the case in the homeland of the Oceanic subgroup. As will be seen, the same is true of music. Pawley and other linguists, however, have repeatedly warned that the branching tree diagrams conventionally used for the representation of subgroups are not necessarily representative of migrations of peoples, a matter known to linguists for more than 40 years , when a prescient warning was issued by the linguist Bruce Biggs:

It should be emphasized that linguistic subgrouping is concerned with internal relationships of languages in a language family. Inferences as to migrations, first settlements, homelands, cultural affiliation and so on should be drawn from such data with caution, and a full awareness of the limited application of linguistic conclusions to such problems (Biggs 1972:143-4).   

Biggs (1972:146) was of the opinion that such inferences "would be justified only if we had prior knowledge of the homeland of each linguistic subgroup in advance, and knew for sure that each island had been settled only once." But no such certainty exists.

It is unfortunate that the tree diagrams commonly used by linguists to represent linguistic subgroups invariably carry implications of the kind warned against by Biggs. The matter is put into perspective by Lynch et al. (2002:92-4) who eschew tree diagrams and represent subgroups in prose for this very reason.

Although the tree format implies successive splits and breakups of proto languages from a common ancestor, this is necessarily true only of trees compiled using subgrouping data now referred to as "innovation defined". An alternative, called "innovation linked", which refers to dialect chains, carries no such implication, and all but SE Solomonic of the Central Eastern Oceanic subgroups leading to Proto Polynesian are innovation linked (Lynch et al. 2002:119), complicating interpretation of the internal subgroups. This explains

why no homeland can be identified for the Eastern Oceanic and Central Pacific subgroups from which Proto Polynesian is shown in the linguistic tree to derive. Importantly also, it explains conflicts which have emerged between the linguistic subgroups and attempts to match them with archaeological and other evidence. 

>>> Chapter 6. Music