Stop Allophony and Voicing in Australian Aboriginal Languages  Phonetics

Justin Lo

Yale University

Many Australian Aboriginal languages have a single phonological stop series, although several have been reported to distinguish at least two stop series.  The nature of the stop contrasts in various Australian languages has been variously described as that of voicing (Austin 1988), fortis/lenis (Butcher 2004), length (Evans & Merlan 2004), or even a combination of factors like “stress, meter and vowel length” (Evans 1996).  Few studies, however, address the nature of allophonic stop variation that is also present throughout the Australian languages, and even fewer attempt to compare the stops from languages across the Australian continent.  The goal of my work is to address these important gaps in the literature and refine our understanding of the phonetics of Australian languages in general.

Hamilton (1996:54-55) contains many preliminary observations about stop allophony.  He observes that the canonical stop in Australian languages is typically not voiced, although there are several languages reported to have the voiced stop as canonical.  Still, these same languages can realize their stops with a “considerable degree of variation in voicing, degree of stricture, and tension” which fall into two patterns: (1) free variation between allophones, and (2) contextually conditioned allophony. 

Hamilton’s account of allophony has some gaps in its logic.  For instance, if the different stop allophones really were in free variation, statements about the “canonical stop” would be meaningless, for in principle it would then become an arbitrary choice which allophone to choose as “canon”.  Hamilton’s pattern of contextually conditioned allophony is a more promising explanation.  I propose that the phonetic and phonological context strongly predicts the allophonic form that a stop may take.  

In order to test this hypothesis, I segmented several sound files from four representative Australian languages with one or two stop series– Bardi, Kayardild, Warlpiri, and Yan-Nhangu– into individual words and using Praat software  observed the phonetic and phonological context in which the stops of the language occur, such as preceding consonant (if the stop is part of a cluster) and syllable position.  Then I took several acoustic measurements, including VOT (measuring the voicing of the stop), VTT (measuring the persistence of voicing into the stop), and the length of the stop. 

Preliminary data that I have already gathered appear to support the idea that stops are contextually conditioned. For example, in Warlpiri there appears to be an expected correlation of  place and stop voicing following a nasal segment as predicted by Hayes & Steriade 2004.  Furthermore the degree of persistent voicing in otherwise “voiceless” stops in my data seems to support the model of passive devoicing (Anderson & Maddieson 1994).  These results call into question the usefulness of the voicing contrast model with respect to Australian languages.


Make a Free Website with Yola.