In the October 2004 issue of the Journal of the Acoustical Society of America, there's an article by J.B. Pierrehumbert, T. Bent, B. Munson, A.R. Bradlow and J.M. Bailey, entitled "The Influence of Sexual Orientation on Vowel Production" (.pdf).
The abstract:
Vowel production in gay, lesbian, bisexual (GLB), and heterosexual speakers was examined. Differences in the acoustic characteristics of vowels were found as a function of sexual orientation. Lesbian and bisexual women produced less fronted /u/ and /ɑ/ than heterosexual women. Gay men produced a more expanded vowel space than heterosexual men. However, the vowels of GLB speakers were not generally shifted toward vowel patterns typical of the opposite sex. These results are inconsistent with the conjecture that innate biological factors have a broadly feminizing influence on the speech of gay men and a broadly masculinizing influence on the speech of lesbian/bisexual women. They are consistent with the idea that innate biological factors influence GLB speech patterns indirectly by causing selective adoption of certain speech patterns characteristic of the opposite sex.
They don't say anything about the distribution of ages and other characteristics in the 103 "Chicago-area self-identified GLB and heterosexual women and men" they studied ("26 self-identified heterosexual men, 29 self-identified gay men, 16 self-identified heterosexual women, 16 self-identified lesbian women, and 16 self-identified bisexual women"). However, these people were "participating in a broad-based social psychology study of sexual orientation", so presumably that information will come out eventually.
The interesting thing was that listeners were able to get some information about speakers' sexual orientation from neutral laboratory-setting readings of phonetically-balanced reference sentences like "It's easy to tell the depth of a well". The self-identified straight male speakers were given an average rating of 3.2 on a scale of 7 (1=totally straight, 7=totally gay), while the (s-i) gay men were given an average rating of average of 4.6. Among the women, the (s-i) straight female speakers had an average rating of 3.2, while the (s-i) lesbian and bisexual women averaged 4.3.
The paper considers and rejects the hypothesis that the effects are due to an overall scaling of vocal-tract resonances. Adult human males are about 8% larger on average than adult females (in linear dimensions like vocal tract length), and adult male larynxes sit lower in the neck than adult female larynxes do, leading to a overall difference of about 15% in average vocal tract resonance frequences (with a differential effect on front and back vowels). It's possible for people to control overall vocal tract length to some extent -- by retracting and protruding the lips, or by raising and lowering the larynx -- but this is was not what happened in their recordings. Of course, it's logically possible that there might be average anatomical differences as well, but again, this is not what they found.
Instead, the authors postulate that their speakers are using particular socially-evaluated ways of talking to signal aspects of their identities. One specific example they cite is the idea that LB females' further-back /u/ vowels might be connected to an earlier sociolinguistic study which found that "a back variant of /u/ was associated with membership in a group known for its 'tough' stance". The paper hints, but doesn't quite assert, that the (more diffuse) vowel differences between (self-identified) male straight and gay speakers might reflect a difference in precision of articulation, which might have other correlates as well, for example the differences in aspiration that have sometimes been found associated with sexual orientation differences among men. (An overall difference in precision of articulation between males and females is well established, as this paper mentions, and would provide a basis for development of greater precision as a norm or stereotype for gay males.)
It's important to note that the study doesn't (try to) demonstrate, and therefore doesn't claim, that the vowel differences they found are responsible for the perceptual effects that they found.
It's also worth noting, I think, that the study doesn't try to distinguish between empirical norms and stereotypes, nor to look at context effects, either in the actions of the speakers or the reactions of the listeners. Of course, that would require a different sort of experiment. Still, some clues could be gleaned from looking at this experiment's data in other ways. For example, they show us the (mean) values of vowel formants as a function of the speakers' sex and (self-identified) orientation --but they don't show us the data broken out in terms of the listener-identified orientation. Were there particular speakers (of whatever self-identified orientation) that were particularly strongly identified by listeners as straight or gay? If so, what were their formant measurements like? And so on...
Compare (my recent discussion of) the Foulkes et al. study of the perception of children's sex from short spoken stimuli. That experiment was the opposite of this one: Foulkes et al. looked at the relationship between phonetic measurements and listeners' judgments, whereas Pierrehumbert et al. look at the relationship between phonetic measurements and speakers' identities.
In both cases, in my opinion, it would be nice to see more of the data. Journals generally have rather strict constraints on page counts, but in this day and age, there's no reason not to put most of the raw data on the web. In the case of studies like these, there may be some reasons not to publish the speech recordings (due to confidentiality and informed consent issues), but there's no reason not to publish all the raw acoustic-phonetic measurements along with the associated sociosexual metadata, and all the raw listener judgments.
Posted by Mark Liberman at October 19, 2004 11:07 AM