This study aims firstly to develop and describe three types of voice quality measurement: acoustic; perceptual ratings made by expert judges, and naive attitudinal ratings. Its second aim is to test the usefulness of the three measures indiscriminating selected voice types, and to show the strengths of the relationships among the measures. Thus, some indication of the way the voice types function in Australian society can be given. Thirdly, the study develops a taxonomy of the functions of voice quality, providing a theoretical basis and direction for the empirical work in this area.
The first measure involved expert perceptual ratings of voice quality. Five voice types were measured: breathy, creaky, nasal, tense and whispery voice. Six male and six female speakers representing three broad age groups each produced examples of these five voice types. All speakers were Australian born. Two judges rated the voices on each of the five voice qualities in terms of the degree of auditory presence.
The second study was a replication of previous research using the long term average spectrum, which averages the amplitude or intensity spectrum of the voice across a selected frequency range for a continuous stretch of speech. The specific acoustic measure used was the alpha parameter. Whispery voice was shown to be significantly different from all other voice types, and breathy voice significantly different from nasal voice. This study replicated earlier research in showing the alpha parameter to be a weak discriminatory tool in voice quality research.
Following the replication study, a different acoustic measure based on the long term average spectrum was developed. The new measure reduces and normalises the spectral information in the frequency range 0-2 KHz. Multidimensional scaling and discriminant analysis showed this measure to be a more powerful tool than the alpha parameter in discriminating the five voice types. Forty-five of the sixty voices were accurately reclassified into the irrespective voice type groups. Most of the remaining voices were shown to be either poor or very complex examples.
Naive attitudinal ratings of the voices were obtained using a matched guise design. Forty Australian and forty American subjects rated one half of the voices on scales representing the underlying dimensions of status and solidarity. Another forty Australian subjects then rated the other half of the voices in the same way. The major results indicated that male tense voice is seen as very high in status, while greater solidarity is perceived in female breathy voice. Nasal voice obtained very low status ratings but somewhat higher solidarity ratings. Whispery voice was rated higher in solidarity when used by older age speakers, and finally, male creaky voice received higher status ratings than did female creaky voice. Subjects were also asked to nominate an age range for each voice. The one overriding result was that all five voice types raised the perceived age from that of the speakers' usual voices, but that creaky voice was perceived as oldest.
The strengths of the relationships between pairs of measures were then examined using multiple regression. These analyses showed that degree of auditory presence of breathy, creaky, tense and whispery voice is systematically reflected in the acoustic measure developed. …………………