How reconstructable is Proto Trans New Guinea? Problems, progress, prospects

This paper asks how far it is possible to make reliable reconstructions of Proto Trans New Guinea (pTNG), the putative ancestor of more than 400 Papuan languages of New Guinea. In order to make such an assessment we need to answer a number of more specific questions, including: (1) Are there trustworthy criteria for determining membership in TNG? (2) Can a clear-cut determination be made for every language? (3) Can we determine whether a particular reconstructed etymon should be attributed to pTNG or to a later stage? That is to say, is the high order subgrouping of TNG well established? (4) Are the available cognate sets sufficient in number and quality, and widely enough distributed, to arrive at reliable reconstructions of pTNG phonology, lexicon and morphology? TNG is a valid family in that there is a large set of languages that share features clearly indicating common origin. However, the precise membership and high-order subgrouping of the family remain uncertain and the lexical residues of common origin to which the Comparative Method can be applied are sparse. Furthermore, New Guinea is an area where there has been a great deal of bilingualism, leading to lexical and structural borrowing. Consideration of such factors led several commentators in the 1970s and 80s to conclude that trustworthy reconstructions of pTNG are unattainable using topdown methods, i.e. by searching for systematic agreements between languages belonging to distantly related subgroups. Contrary to this view, I argue that using a top-down strategy has been essential to achieving initial breakthroughs in reconstructing pTNG. But it is clear that gaining a more complete and finergrained understanding of the history of the family will require the integration of top-down with bottom-up reconstructive work.