The cognitive advantages to retaining a restricted counting system (without exponentiation) even as a more complicated one is being developed are not immediately obvious, but follow from the information about upcoming complexity that is implicit in the use of distinct numerals. Kanum, a language from the south of New Guinea, where “systems with limited extent” are widely reported, has base-6 counting systems with full use of exponentiation in one system, and no possibility of extension in another. The evidence suggests the more complex systems were internally motivated, yet the simpler systems have not been abandoned.
Donohue (2005a) argues that the SVO order of most southern Austronesian languages found between mainland Southeast Asia and New Guinea is due to contact with non-Austronesian languages. I offer a number of other correlations between word order features and geographic area, establishing that the well-discussed division between “eastern” and “western” (or “Papuan” and “Austronesian”) languages in the Indonesian archipelago is not a crisp one, but is one that should be essentially maintained. Despite the fact that the division, traditionally based on the position of the genitive, generally matches the western boundary of “Central Malayo-Polynesian” (Blust 1993), a better explanation for the break is shown to be influence from languages with a typology matching that found in western New Guinea. At the same time, the much less discussed break between the northern and the southern Austronesian languages is established, mapping the contrast between the northern, “Philippinetype” Austronesian languages and their southern neighbours, despite the lack of any well-accepted genetic boundary between these two areas, implying substratal influence similar to that which characterises the eastern Austronesian languages.