Speaker and material selection for the Australian national database of spoken language*

Abstract

The Australian National Database of Spoken Language (ANDOSL) was collected to provide spoken speech data for the research community in Australia. It was intended that the data be representative of the major speech varieties in Australia, and that the collection would have sufficient coverage to be adequate to the needs of several disciplines, such as speech scientists, linguists, TESOL and TEFL teachers, engineers, computer scientists and speech pathologists. The data on which we report here is the foundation upon which it is hoped that a very large database will be established. As this data discussed in this paper was to be the foundation material and would probably be the largest single input to the collection, it was important to collect an adequate and representative basic core of material which could underlie a variety of the speech research in Australia. So, the current material is a collection of the appropriate segmental distribution for the three major dialects of Australian English currently described as Broad, General and Educated/Cultivated Australian. In addition, it was possible to commence collection of a limited sample of the English of Australians who were born overseas and speak English with an accent.

Three-year-olds' production of Australian English phonemic vowel length as a function of prosodic context

“…For instance, duration could be used as the primary cue to distinguish (1) long vs short vowels, (2) voiced vs voiceless fricatives, (3) voiced vs voiceless postvocalic consonants, (4) phrase-final vs non-final syllables, (5) stressed vs unstressed vowels, (6) the presence vs absence of focus/ accent. In AusE, Fletcher and McVeigh (1993) examined the segmental durations of a male speaker from the SHLRC-ANDOSL database (Vonwiller et al, 1995), confirming the durational contrast between the /AE+/ -/AE/ vowel pair. In line with Klatt (1976), they also identified final lengthening and accentuation as factors contributing to segmental and syllable duration.…”

Phonetic Archaeology and 50 Years of Change to Australian English /iː/

“…The results may possibly reflect a minor change in progress. Cox and Palethorpe (2001), in a separate study of speakers from the ANDOSL database (Vonwiller et al 1995), did show a change for fronting of /iː/ (that is, higher F2) as a trend for females, but the extent of onglide was not measured. Contrary to this, Mannell (2008) found no evidence of change in either production or perception in the degree of /iː/ for males or females over the period 1990-2007.…”

“…Broader /iː/ displays a greater extent of articulator movement (indicative of diphthongization) and an increased duration of onglide. Harrington et al (1997) in an apparent-time analysis of the ANDOSL database (Vonwiller et al 1995), indicate that for males recorded in 1995, the oldest Broad speakers appeared to have the most extensive onglides and the youngest Cultivated speakers the least onglided /iː/. In Harrington et al (1997) the onglide was described with reference to the degree of delay to the vowel target which was identified as the point at which the second formant (F2) reached its peak.…”

Vowel transcription systems: An Australian perspective

“…His aim was to present a phonologically adequate but phonetically accurate representation of vowel productions. More recently, Harrington et al (1997) have proposed similar revisions to the traditional phonemic transcription system for SAusE vowels (the HCE system) following an acoustic analysis of vowels in isolated words produced by 55 male and 64 female speakers with an age spread of 18 to over 46 years contained in the Australian National Database of Spoken Language (ANDOSL) (Vonwiller, Rogers Cleirigh & Lewis, 1995). 4 A comparison between the symbols sets of Mitchell (1946) and Harrington et al (1997) can be found in Table I.…”