Although we do not have records to show how often this occurred in the past — Melanesian societies kept no written records and were apparently not sufficiently concerned about the demise of languages to keep oral traditions about them for any significant length of time — there were sufficient numbers of cases of languages in the last stages of obsolescence at the time of European contact for us to presume that it was a reasonably regularly occurring event prehistorically. It would appear that the conditions favouring language shift in Melanesia are similar to those noted elsewhere — the absorption of remnants of larger groups seeking refuge from intertribal warfare and natural disasters (such as droughts, floods and epidemics) or small numbers of people pursuing perceived economic or cultural advantages. In any case, both An and NAn languages were involved (Dutton and Mühlhäusler 1991).

It is hard to find cases in this category because the potentially instructive ones either have not been fully described or do not live up to expectations on further investigation. For example, even though we have reported cases of dual lingualism involving An and NAn language speakers, in the best described one the An languages involved, Banoni and Piva, do show low levels of cultural borrowing from neighbouring Papuan languages (Lincoln 1976:97-99).