In the period since the break-up of POc perhaps 3,500 years ago, many of the descendants of POc have undergone radical transformations while others have changed much less. The most conservative languages, lexically, have probably been some members of the Meso-Melanesian Cluster, especially Bali-Vitu and the Willaumez chain, all the languages located in the eastern half of the main Solomon Island chain from eastern Santa Isabel and Guadalcanal to Santa Ana, parts of central and northern Vanuatu, all the Fijian languages and some Polynesian languages (especially Tongan). Most of these languages retain reflexes of 35-45 per cent of some 250 Proto-Malayo-Polynesian basic vocabulary items reconstructed by Blust (1981, 1993). At the other extreme are languages which retain only a small percentage — in the most extreme cases, fewer than 10 per cent — of the reconstructed etyma. Most languages are somewhere between these extremes. Among Oceanic languages there is a rough — by no means exact — correlation between lexical and grammatical conservatism and a much lesser correlation between lexical and phonological conservatism. With a handful of exceptions all the lexically most innovative languages are found in the area of the New Guinea mainland and New Britain.

Linguists have for several generations been vexed by these highly innovative (sometimes called “aberrant”) languages. Among the explanations that have been offered to account for their high degree of vocabulary replacement are: (i) imperfect acquisition of an Oceanic language by a non-Austronesian (Papuan) speech community, (ii) long-term bilingualism between Oceanic and Papuan neighbours, sustained by trade and intermarriage, (iii) very small size of speech communities, as the result of migration, political structures, etc., (iv) social pressures to develop a distinct language from one’s neighbours, (v) taboos on using words coinciding with the names of chiefs or of the dead, (vi) changes of physical environment following migration, (vii) cultural changes generated internally, (viii) cultural changes generated by contact, and (ix) phonological change creating problematic (especially ambiguous) word forms.

By far the most controversial of these proposals, because it conflicts with the hypothesis of founder settlement, is (i), the so-called “pidginization” hypothesis, elaborated by Ray (1926) and Capell (1943). The current consensus is that few, if any, Oceanic languages show evidence of the imperfect learning and heavy substrate residue which would indicate their takeover by a community whose previous language was Papuan. To explain most cases we need to look to a combination of the other factors.

In parts of Melanesia at least as far east as the central Solomons there has been sustained interaction between Oceanic and Papuan language communities, associated with trade and intermarriage. The resultant bilingualism has in a number of cases caused a radical restructuring, especially in grammar, of Oceanic languages. Lynch (1981) argues that the subject-object-verb order and postpositions of the Papuan Tip group are attributable to Papuan influence on Proto-Papuan Tip. One Papuan Tip language, Maisin, has undergone so much Papuanization that earlier commentators could not decide whether it was Papuan or Austronesian (Ross forthcoming). Dutton (1982) has analysed the borrowing and reborrowing of vocabulary in Magori, also a Papuan Tip language. The Bel languages of the North New Guinea group show significant Papuanization in their grammatical structures (Ross 1987, forthcoming), and Lincoln (1976) describes the contrast between Papuanized Piva and un-Papuanized Banoni, two otherwise closely related languages of Bougainville. However, while such sustained language contact and bilingualism has certainly promoted linguistic change, it seems rarely to have led to the adoption of Austronesian languages by Papuan speakers which Ray and Capell argued for. Where the latter did occur, it probably wrought changes in phonology rather than grammatical structure (Ross 1994a).

It should surprise no-one that a change of physical or cultural environment led some Oceanic language communities to lose many POc words and change the meanings of others. Sometimes it is possible to work out the approximate period in the history of these languages when the terms in question were lost. For example, although New Zealand Maori no longer has terms for a great many plants and animals of the Pacific tropical environment (Biggs 1994), we can be sure that such terms were present in the immediate common ancestor of Maori, Rarotongan, Tahitian and Tuamotuan, because the latter retain Proto-Polynesian terms for many such items, and we can infer that the losses occurred only after Maori separated from these other languages. In the same way, even though in the central Pacific the megapode or brush turkey is now found only on one remote island (Niuafo’ou, between Tonga and Samoa), we know that speakers of Proto-Polynesian retained the POc (and Proto-Eastern Malayo-Polynesian) word for the megapode and that the term was lost in daughter languages only after the break-up of Proto-Polynesian. This inference can be made because the Niuafo’ou language (a Polynesian language spoken on Niuafo’ou Island between Tonga and Samoa) retains the POc term, *malau. It is supported by the recent discovery of megapode bones in archaeological sites in Fiji. In certain Oceanic languages spoken inland on large islands in Melanesia the loss of terms for the maritime environment can be shown to be fairly recent, because the languages in question are closely related to coastal languages which retain many POc maritime-related terms.