Australian Languages: Classification and the Comparative Method (review)

This book comprises papers presented at the Workshop on Reconstruction and Subgrouping in Australian Languages which formed part of the 15th International Conference on Historical Linguistics held in Melbourne in August 2001. In ‘Introduction: Subgrouping methodology in historical linguistics’, the editors state that ‘this book arose out of our desire to test, as rigorously as possible, the traditional ideas of subgrouping and the comparative method in the Australian context’ (1). They note that a subgroup is a set of languages more closely related to one another than to others in the family by virtue of shared innovations. They also note that subgrouping can be represented by the family tree model where splits are neat and discrete, but it needs to be adapted where diffusion has obscured clean splits (9).

Harold Koch provides ‘A methodological history of Australian linguistic classification’ in which he gives a thorough overview of the various classifications of Australian languages. Of particular relevance to the papers in this volume is his account of the recent views of R. M. W. Dixon (48–60). Dixon has doubts about how successfully the comparative method can be applied in Australia where, in his view, languages have converged after a long period of equilibrium (Dixon 2001, 2002:20–44), and he trenchantly attacks the validity of the widely accepted Pama-Nyungan family (Dixon 2002:xx, 44–54). In the most widely used classification of Australian languages compiled by Geoffrey O’Grady, Kenneth Hale, and Stephen Wurm (1966; see also O’Grady, Voegelin, & Voegelin 1966), the languages of the Australian mainland are classified into twenty-nine families with one family, Pama-Nyungan, covering all of the mainland except for the northern part of Western Australia and the Top End of the Northern Territory. The basis for the classification was lexicostatistical, but Blake 1988 shows that certain pronoun forms such as ngali ‘we two’ are found only in Pama-Nyungan, as are the ergative case alternation -lu/-ngku and the analogous locative -la/-ngka. Moreover, Evans 1988 shows that initial apicals merge with initial laminals in Pama-Nyungan (t → t^y, n → n^y). Dixon points out that lexicostatistics cannot provide evidence for genetic classification, and he claims the figures do not support Pama-Nyungan anyway (2002:44ff.). He dismisses the evidence of pronouns like ngali with the claim that they are likely to have been spread via diffusion and the evidence of distinctive case alternations on the grounds that they have a restricted areal distribution within Pama-Nyungan. He also shows that the area reflecting the apical-laminal merger does not coincide exactly with Pama-Nyungan (2002:51).

In ‘Pama-Nyungan as a genetic entity’, Luisa Miceli points out that Pama-Nyungan needs to be established as a family rather than a subgroup ‘because there is no known larger entity of which these languages could form a branch’ (61). This echoes a now widely held view that it will be difficult to reconstruct proto-Australian convincingly: although all of the languages of the Australian mainland look as if they are related since there are a number of widespread roots such as bu- ‘to hit’, ka- ‘to carry’, and na-/nha-/nya- ‘to see’, in general, cognates are scarce, particularly among the non-Pama-Nyungan languages of the north. It is worth keeping in mind that humans entered the Sahul landmass—comprising Australia (including Tasmania) and New Guinea—about 50,000 years ago, and the roots widespread on the Australian mainland are not found in Tasmania (cut off from the mainland 14,000 years ago) or New Guinea (cut off 7,000 years ago).

In ‘The coherence and distinctiveness of the Pama-Nyungan language family within the Australian linguistic phylum’, Geoff O’Grady and the late Ken Hale argue strongly against Dixon’s alleged denial that the comparative method can be applied to Australian languages and say that...