If there is one point that stands out above all others in this survey it is that there has been much more contact between the different types of languages in Melanesia than most scholars have been prepared to acknowledge in the past. As a result, we must be prepared to accept that the linguistic and archaeological prehistories of the area are that much more complicated than hitherto suspected or acknowledged.

Scholars interested in the classification and origin of An languages have long known about and been fascinated by the diversity aspect of Melanesian languages. While those languages share what Grace called “a characteristic uniformity” (or “tendency to sameness”) in some respects, they also display a “remarkable diversity” in others (Grace 1968:67). However, those same scholars have often argued at cross purposes about the underlying reasons for this because they have generally failed to recognize that there is not just one kind of diversity, but several.

The first kind of diversity to be recognized is that which is referred to in the opening paragraph to this paper, notably, the sheer number of languages concentrated in a small area. Actually it is not merely the absolute number that is the interesting and significant thing, but, as Pawley (1981) points out, the number of languages per island group. Thus, when compared with Polynesia, for example, Melanesia has many languages per island group while Polynesia usually has only one. Why is this so? Pawley’s answer was that Melanesian diversity was not “brought about by mechanisms of a radically different kind from those which operated in Polynesia” (Pawley 1981:273). Rather, the sequence of diversification was much the same in both areas, the only difference being that “the cycle of linguistic diversification” had more time to apply in Melanesia than in Polynesia (Pawley 1981:298). While he saw various socio-economic and political factors as important in this speciation process, Pawley did not see contact with NAn communities as a “necessary ingredient in the recipe for Melanesian linguistic diversity” (Pawley 1981:274-275). He did acknowledge, however, that the presence of such populations may have hastened the process. As he himself noted on a later occasion (Pawley 1990), this type of diversity is not much of a puzzle. It can be accounted for by the sorts of mechanisms outlined in his paper.

Two other kinds of diversity are more interesting. They have to do with variation between An languages in Melanesia. In the first kind the focus is on the way some languages differ markedly from other An languages and reconstructed proto-languages in their sound systems and grammars. The second type has to do with the distribution of cognates across languages, some languages having relatively few. Those that are most divergent in these senses are often referred to as “aberrant” or “problematic” (Grace 1990), and those that have changed least as “conservative” or “exemplary”. Between these two are degrees of aberrancy so that languages can be ranged on a scale (although no one has actually tried to do this) from least to most aberrant (Grace 1990). The Markham River languages in Papua New Guinea, New Caledonian languages and those of south Tanna in Vanuatu in particular have reputations of being the most aberrant. Indeed, although An languages in Melanesia are classified as An, that is, as descendants of Proto-Oceanic, they vary so widely in structure and vocabulary that linguists are still debating how they should be classified internally (see Pawley and Ross, this volume).

Scholars have long puzzled over the reasons for these two kinds of diversity and have sought to explain them in varying ways. The first to do so was S.H. Ray (1926), who suggested that the present forms of An languages in Melanesia resulted when incoming An speakers from Indonesia settled in Melanesia and came into contact with resident NAn populations. Later, Capell (who was actually a student of Ray’s) carried this idea forward in various publications, giving rise to what is known as the “pidginization hypothesis”. Both Ray and Capell pointed to the small proportion of vocabulary in Melanesian An languages that could be related to Indonesian vocabulary.^[19] Capell (1943) further claimed that a study of this vocabulary shows that the languages of southeast Papua, for example, derived their particular features from having been in contact with a number of “regional” NAn languages, certain features of which could be reconstructed from that same vocabulary.

For a long time this pidginization hypothesis was rejected by later linguists as based on false assumptions and bad methodology.^[20] Grace (1962), in summing up the controversy, challenged supporters of it to find a Melanesian language in which the non-An elements could be attributed to a specific NAn language. Thurston (1982) took up this challenge and provides evidence from Lusi in West New Britain that he says is “capable of resolving the argument in favour of Capell and Ray” (Thurston 1982:2). His examination of both lexical and grammatical evidence leads him to conclude, as already pointed out, that Lusi is a creole language which has developed from a pidginized An language through contact with the NAn language Anêm. Thurston is so convinced by what he finds in Lusi and other languages of the same area that he believes that pidginization (in his sense) was the major factor in the diversification of Mn languages (1987, 1994).

Meanwhile, Lynch (1981) also took up the challenge implicit in Pawley’s 1981 paper referred to above and argued that the many-languages -per-island-group type of diversity as found in Melanesia was not solely a reflection of different time depths and particular socio-economic and political conditions. He felt that contact between An and NAn languages was also an important factor. In support of his claim he appealed to the specific mixed-language-type case studies of Magori, Maisin, Lusi and Reefs-Santa Cruz that have been mentioned above.

Since then Grace, responding to more recent detailed work by a number of younger scholars including those already mentioned, has acknowledged that there is much that is left unexplained by the traditional views of language diversification and change based on the family tree model. He has called for a reassessment of “all the facts” (1985:3).

Part of the problem in the past has been, of course, that scholars have attempted to provide a unified “solution” to what has often been called “the Melanesian problem” (Grace 1968), without distinguishing between the different types of diversity described above and without attempting to find separate “solutions” for each kind. What such scholars failed to acknowledge was that there are many factors besides time depth and language contact that have been involved in producing the kinds of diversity found in Melanesia today. Attitudes to language, and what Don Laycock used to call “conscious human monkeying with language”, are two such factors. Laycock describes how in one dialect of Buin, a NAn language of south Bougainville, gender distinctions in grammar are the exact reverse of those in other dialects — what is male in other dialects is female in this dialect (1982:35). Melanesians thus appear to foster linguistic diversity purposefully because they see linguistic differences as important badges of group identity. It is, as it were, a Melanesian choice to promote diversity (Laycock 1982:34).

One thing is certain, and that is that diversity must be functional in some way; otherwise, as Grace points out, it is maintained at too great a social cost — why learn to be different when being the same would be less taxing? Nor can diversity be explained solely by migration, as that would require too many independent moves (Grace 1975; Thurston 1987:94ff.). At the same time, Melanesian societies have placed a high value on multilingualism. The public display of knowledge of other languages has long been noted as an important means of gaining prestige in Melanesia (Salisbury 1962; Sankoff 1977).

Another factor that needs to be taken into account in attempting to explain linguistic diversity in Melanesia is word tabooing. This is the social action of forbidding the use of a word that is associated in some way with a member of the community who has just died. It is a common feature of many Melanesian societies and leads to “unnatural” replacement of vocabulary (Simons 1982; Holzknecht 1988).