By its very nature this kind of borrowing is difficult to identify. It usually shows up, however, in detailed comparative work as inconsistent sound correspondences and/or as dialect chaining. Thus, for example, in Papuan Tip (PT) languages in the Milne Bay Province of Papua New Guinea some reflexes of PPT ^* b, ^* d and ^* g occur voiced instead of being voiceless as expected (Ross 1988:198-202). These voiced reflexes are evidently the result of dialect borrowing from neighbouring languages.

Chaining occurs when a string of communities share sets of features with each other in an overlapping fashion so that community A shares some features with community B, which in turn shares some features with community C, and so on. The phenomenon was first described by Wurm and Laycock (1961-62) in discussing problems of classification in Papua New Guinea. Its existence introduces a certain arbitrariness in the definition of “language” and “dialect” in Melanesia and explains why linguists differ about the number of languages actually spoken therein. In his detailed study of the An languages of western Melanesia, Ross (1988:8) distinguishes between “chains” and “networks”, the former being used for “communalects that are typically spread along a coastline, each related most closely to its neighbour on either side”, and the latter for “communalects scattered over a land area or an archipelago, typically having neighbours on more than two sides and often sharing different innovations with several of these”.

Dialect borrowing therefore poses problems for the traditional family tree model of language diversification. To cope with it linguists are forced to adopt models that allow for diversification through contact-induced change, as Ross (1988) did in his exemplary study.

This is the most common kind of borrowing noted in Melanesia — at least it is the one most easily identified and most often reported. It is particularly common in NAn languages in contact with An ones and in Melanesian An languages in contact with Polynesian Outlier ones. In both these cases it is particular cultural items that seaboard or incoming Austronesians have that other resident communities seek and hence borrow vocabulary for. Thus, for example, the NAn Koita around Port Moresby have borrowed sailing and sea terminology (including terms for marine life) from their Austronesian neighbours, the Motu (Dutton 1994). The An languages near Polynesian Outlier languages have done very much the same thing, borrowing not only Polynesian terms for the sea and sailing but also words associated with kava drinking and artefacts made from coconut fronds (Clark 1994). This picture is repeated many times in the literature and is part of the folk wisdom of Oceanic linguists.^[12] By definition this kind of borrowing is one sided or at most unequal, although it is not always quite as straightforward as this suggests. Indeed it is very sensitive to socio-cultural realities and may be bound up with intimate borrowing as well. Thus, for example, Clark (1986) describes how the Polynesian Outlier languages of Emae and Mele-Fila in Vanuatu have borrowed intimately from neighbouring An languages which in turn only show effects of cultural borrowing.^[13] The reasons for this are apparently demographic — the Polynesians were (and still are) much fewer in number than their Melanesian neighbours and in seeking wives and in trading they would have had to, as Clark says,

deal with Melanesian speakers more often than not, whereas Melanesian speakers, on average, would have had only a minority of Polynesian contacts. Melanesian wives, in particular, marrying into Polynesian villages, bearing and rearing children, speaking a Melanesian-influenced second-language variety, would have accomplished both the physical and the linguistic assimilation of the immigrants (Clark 1986:341).

Such cases show that, even though the principles of cultural borrowing are clear, it is not possible to predict the effects in particular cases without knowing the social conditions operating. Consequently it is often not possible to distinguish between the effects of heavy cultural borrowing and intimate borrowing.

It has long been known that, structurally, the An languages of the Papua New Guinea area fall into two main types, those with a basic SOV (subject – object – verb) word order and those with a basic SVO word order.^[14] Practically all of the mainland An languages and those of south-eastern Papua and its associated offshore islands are of the first type while those of the second type are found mainly in New Britain and New Ireland. As well, the basic word order SOV coincides with that of most NAn languages. It has therefore been suggested that the structural difference between these two groups of An languages is to be attributed to contact with NAn languages some time in the past. Although it has not been possible to identify any particular NAn language or group of such languages to support this claim it seems highly likely that this difference was the result of intimate borrowing.

Apart from this case and a number of other well known cases of language “mixing” described briefly below (which derive from intimate borrowing between An and NAn languages),^[15] relatively few cases of this kind of borrowing have appeared in the literature until recently. One such case is Labu, an An language spoken near Lae, Papua New Guinea. This is a small language of 1700 speakers that is surrounded by speakers of two other An languages, Bukawa and Kela. For many years the genetic position of Labu vis-à -vis related languages in the same area has been ambiguous. Holzknecht (1994) shows, however, that Labu is a member of the Markham Family that has been heavily influenced by Bukawa and possibly also Kela (both closely related to one another and members of the North Huon Gulf Chain) in phonology, grammar and vocabulary. Although she does not say as much this case appears to be one of intimate borrowing between An languages. This is not only evident from the fact that all aspects of the structure have been influenced but also from the fact that the Labu have been socially in “more or less intense” contact with the Bukawa. Indeed, some of the early Labu “spent some time within Bukawa-speaking communities in a client (Labu)-patron (Bukawa) relationship” as refugees (Holzknecht 1994:371), and some children grew up bilingual in Labu and Bukawa. This is not regarded as a case of dialect borrowing, however, because, even though the languages are An, they are not sufficiently closely related or typologically similar to fit the definition given above.

More elaborate cases are those described briefly by Clark (1994), who discusses a complex situation obtaining amongst the Polynesian Outlier languages. He notes that none of the Polynesian languages is “pure” in the sense of having undergone no influence but the influence is scaled — it is greatest on those languages geographically closest to non-Polynesian languages and least in the most remote or isolated communities. Two languages, Mele-Fila and Emae, show unmistakable signs of intimate borrowing from non-Polynesian languages nearby. Two others, Rennellese and Fagauvea, show signs of cultural and intimate borrowing at work together. In other cases it is not clear if the evidence is indicative of intimate borrowing or the influence is a secondary effect of cultural borrowing on a large scale. The picture is further complicated by the fact that (i) at least in one case, Anuta, multiple settlements have occurred from Polynesia on the one island, (ii) in other cases there has been contact with neighbouring Polynesian communities that have a common settlement history. On the other hand, none of the non-Polynesian languages has undergone anything like the borrowing seen in the more affected cases of Polynesian languages. In general they have only borrowed culturally relevant vocabulary of a kind already noted above. The explanation for the observed borrowing pattern is demographic — the Polynesian communities are all small even by Melanesian standards.

These are special cases of intimate borrowing where the languages have undergone so much influence from some other language or languages that there has been considerable debate about their original status. They include Magori, Maisin and the Reef-Santa Cruz languages, and in a slightly different category, Lusi.

Magori is an An language of the southeast coast of Papua New Guinea. At the time of European contact it was only spoken in two small villages in the Bailebo River valley almost opposite Mailu Island where the principal village of the NAn Mailu (or Magi) is situated. When first recorded it was thought that Magori was part of the Mailu language, if apparently somewhat aberrant. Subsequent investigation showed, however, that it was originally an An language from which the Mailu had initially borrowed. Later, it appears that something happened to put the Magori in a position where at first contact they were living in fear of the Mailu and were learning their language. In the process they were borrowing back words that the Mailu had originally borrowed from them in a different form (Dutton 1982). This borrowing has been very heavy but so far it does not seem to have affected the grammar and phonology very much. However, Magori and remnants of three other closely related An languages, Ouma, Yoba and Bina, have become more alike as a result of mutual loaning and borrowing with other neighbouring An languages. Lexically, they now appear to be co-ordinate between eastern and western groups of An languages along the south coast of Papua (Dutton 1976, 1982).

Maisin is spoken in two areas corresponding to two dialect divisions in the Northern Province of Papua New Guinea — one in the coastal area of Collingwood Bay and one in several villages in the swamps between the mouths of the Musa and Bariji Rivers in Dyke Ackland Bay (Dutton 1971). For many years this language had been an enigma, some scholars classifying it as originally An and others as originally NAn. In 1977, however, Lynch showed fairly conclusively that although it had a vocabulary of mixed origins its grammar was basically An. It is now clear that it originally belonged to the Are-Tauopota Chain of Papuan Tip languages (Ross 1988). Its present “mixed” status arises from the fact that it was evidently in close contact with some (as yet unidentified) NAn language or languages in which the speakers were bilingual and as a result has been heavily influenced by it or them.

The Reefs-Santa Cruz languages spoken in the islands of the same name in the eastern Solomon Islands are similar to Maisin in that investigators have disagreed about their original status. Wurm (1978), for example, argued that they were originally NAn and Lincoln (1978) argued that they were An. This controversy has not been satisfactorily resolved to date.

Lusi is an An language spoken in the West New Britain Province of Papua New Guinea. According to Thurston (1982), Lusi is a creole language (that is, a pidgin language that has become the native language of a group of speakers) which has developed from a pidginized An language through contact with the NAn language Anêm (Thurston 1982:71). Thurston’s use of the word “pidginized” is aimed at emphasizing the rapid nature of change that was involved in this transformation and the changes the language has undergone.^[16] It was an exaggerated position adopted to draw attention to the normally accepted view that language change is gradual. In reality it is no different from related languages in the area. While it has undergone changes not undergone by others, most probably as a result of contact with Anêm, it is not so different for it to be put in a special category.