Is there any evidence of rhoticity in historical Australian English?

Australian English is traditionally regarded as having been non-rhotic throughout its history, but a recent study by Trudgill and Gordon (2006) has found rhoticity levels of 1% to 20% in audio recordings of six Australian men born near the end of the 19th century, suggesting that Australian English was once a rhotic dialect. The present study re-examines the three most rhotic speakers in the Trudgill and Gordon dataset, as well as archival recordings from an additional seven speakers, to further investigate the presence of rhoticity in Australian English around the turn of the 20th century. Approximately 30 minutes of audio interview data from each speaker was examined impressionistically for the presence of non-prevocalic /r/. Other postvocalic instances of /r/ were also identified in order to explore the relationship between non-prevocalic /r/, linking /r/ and intrusive /r/. On re-examination, the three subjects from Trudgill and Gordon (2006) were found to exhibit little or no rhoticity according to our criteria. Analysis of the additional seven speakers further weakens the argument for rhoticity in Australia at the end of the nineteenth century. Some speakers do, however, have traces of rhoticity, and only one shows consistent use of linking /r/. In this paper, we will explore the phonetic conditioning environments for the vestigial rhoticity in historical Australian English and will discuss some implications for phonological theory.