Revisiting Phonotactic Generalizations in Australian Languages

Australian languages are famous for their uniform phonological systems. Cross-linguistic surveys of (or including) Australian languages have reinforced this view of Australian inventories and phonotactics. Such uniformity is surprising and unusual given the phylogenetic diversity in the country (28 phylic families). Moreover, although Australianists have assumed that uniformity in phonemic inventory is coupled with unity in phonotactics, this has not been tested. Here we statistically test the generalizations current in the literature on Australian languages by deriving inventory information from lexical data (rather than grammatical descriptions). We utilize a comparative database of lexical items from predominantly Pama-Nyungan languages in order to test published generalizations about phoneme inventories, phonotactics, and other phenomena (such as root internal vowel harmony patterns). By using lexical materials to derive inventories and segment frequencies, we are able to assemble a nuanced picture of the diversity of systems present among the languages. Inventory studies confirm, to some degree, the impression of uniformity. However, phoneme frequencies vary substantially across the sample even among languages with similar inventory types. This work is of particular importance to phonological typologies of Australian languages, but it has implications for wider phonological theory as well. The survey used here is the largest comparative database of a single language family. Rarely do we have the opportunity to conduct a large-scale typological investigation of related languages in this way. We also make a contribution to the role of typology in Optimality Theory. A large-scale survey of markedness patterns (in related languages) allows us to study occurring and non-occurring grammars. Finally, we can investigate the predictions of competing theories