A comparative study of phonetic sex-specific differences across languages
Extensive reviews of phonetic and phonological investigations into sex-related differences reveal a mottled history. The investigations suffer from methodological and theoretical deficits: the most serious being the misrepresentation of the interaction between variables, a lack of homogeneous data and its misinterpretation, and the widespread neglect of women's speech. Existing phonetic databases are shown to be inadequate and poorly-controlled, admitting too many unwanted variables. A very tightly-controlled database, constructed for this research, contains data for eighty female and male speakers of two accents of British English. This contribution is regarded as important per se. Digital acoustic analysis of the data permits quantification of the phonetic divergence shown by the sexes in British English. Previous attempts to normalize the acoustic effects of speaker-sex on vowels have been largely unsuccessful. Here, the application of an innovative auditory normalization procedure reflects how perceptual normalization may be achieved. It further demonstrates that male/female phonetic differences remain after normalisation, which cannot be accounted for by anatomy, but are accountable by social-role conditioning (i.e. learned). These differences are statistically significant. Speaker-sex and gender are thus shown to interact at the phonetic level. Extending this technique to five other languages/dialects corroborates the central hypothesis that the degree to which the sexes diverge phonetically will vary from speech-community to speech-community. Exploration of the possibility that contoids will reveal similar systematicity shows this to be unlikely across languages. The examination of suprasegmental sex-associated differences, however, merits further pursuit. Implications of these experimental findings are discussed for 'inter alia' speech technology, language-planning and medical aids. Using sex-linked differential voice quality as a springboard, it is suggested that sex-appropriate norms are required in speech pathology. The need for socio- phonetics to be recognized as an important new discipline is thus underlined.