ABSTRACT
A fundamental goal of phonetic investigation is to build up an account of the systematic properties of speech performance. This, in turn, informs the development of rigorous phonetic taxonomies and the formulation of models of the learning, representation, and processing which underpin speech communication. The past 50 to 60 years of research (since the development of the spectrograph) has provided theoretical clarity in respect of many of these areas. However, there is a strong sense in which phonetic research has really only scratched the surface of what on closer inspection appear to be ever more complex
areas of human behaviour. So, for example, it is clear that the range of languages covered by quantitative phonetic research is heavily skewed to the relatively few languages spoken by the investigators themselves. Even within those languages which have received the bulk of the attention from phonetic research, there has tended to be relatively little quantitative research on the differences between varieties. Laboratory studies have typically been designed to focus on relatively small numbers of speakers producing material which is highly controlled. Furthermore, in order to test experimental hypotheses, they have tended to emphasise the statistically typical patterns of performance across groups of speakers, giving much less attention to the variability found within corpora of data. Finally, despite all the instrumental work that has been carried out over the years, there are still large tracts of our phonetic knowledge of languages which are cast exclusively within the terms of the auditory-segmental symbolic representations offered by the IPA (see Docherty & Foulkes, 2000, for further discussion).
