Macalester College DigitalCommons@Macalester College Linguistics Honors Projects Linguistics Department 2011 Vowel Duration and Perceptions of the Gay Accent Laura Panfili Macalester College Follow this and additional works at:http://digitalcommons.macalester.edu/ling_honors Recommended Citation Panfili, Laura, "Vowel Duration and Perceptions of the Gay Accent" (2011).Linguistics Honors Projects.Paper 7. http://digitalcommons.macalester.edu/ling_honors/7 This Honors Project is brought to you for free and open access by the Linguistics Department at DigitalCommons@Macalester College. It has been accepted for inclusion in Linguistics Honors Projects by an authorized administrator of DigitalCommons@Macalester College. For more information, please [email protected]. Honors Project Macalester College Spring 201 1 Title: Vowel Duration and Perceptions of the Gay Accent Author: Laura Panfili Vowel Duration and the Gay Accent 1 Vowel Duration and Perceptions of the Gay Accent Laura Panfili May 201 1 Honors Project in Linguistics Macalester College Vowel Duration and the Gay Accent 2 Abstract This study aimed to examine a potential linguistic cue that signals a speaker's sexual orientation. I examined the relationship between vowel duration and perceived sexual orientation for male speakers of American English. Speakers recorded a passage that was heard by nalve listeners and ranked according to perceived sexual orientation. There was no significant difference in vowel duration between men perceived to sound gay and men perceived to sound straight. However, the gay-sounding men produced their diphthongs with more variance in duration than did the straight-sounding men. Vowel Duration and the Gay Accent 3 Vowel Duration and Perceptions of the Gay Accent Encoded in our speech is far more than just the information we speak. What we say and the way we say it are both reflections of who we are, or who we want people to think we are. Our language conveys various aspects of our identity, such as our socioeconomic status, race, and where we grew up (e.g., Labov 1972 and Trudgill 1974). This study examines one particular aspect of identity as it relates to speech - sexual orientation. Sexual orientation and speech is a fairly new topic within sociolinguistics; its study is sometimes referred to as "queer linguistics" or "lavender linguistics" (Munson 201 1). Previous linguistic studies have shown that a certain way of speaking is associated with the gay community that listeners - both from inside and outside the gay community - can recognize as sounding gay (e.g., Gaudio 1994; Pierrehumbert, Bent, Munson, Bradlow & Bailey 2004). This is not to say that all gay people speak with the gay accent; it is an accent likely adopted to discretely identify oneself as a member of the gay community. As with other features of our speech, the gay accent can be downplayed or emphasized depending on the circumstances. There have been a handful of recent studies regarding the male gay accent, but there have been fewer studies regarding the gay accent amongst lesbians. The limited research done on the female counterpart of the accent has shown that while it does exist, it is significantly different from the male accent and less easily identified by listeners (e.g., Moonwornon-Baird 1997); for these reasons, the present study focuses only on the male gay accent. As not all gay males exhibit the gay accent, a linguistic study that groups all gay males into the same category would not be an accurate examination of the gay accent. For Vowel Duration and the Gay Accent 4 this reason, research aimed at studying the gay accent studies accents that are perceived to sound gay. Previous studies have determined which voices are perceived to be gay- sounding or straight-sounding by having nalve listeners rank how gay or straight- sounding they perceive the voices to be, generally on a five or seven point scale (e.g., Gaudio, 1994; Rogers, Jacobs & Smyth, 2003). Voices are then analyzed once they are grouped into the categories of perceived orientation provided by the listeners. However, the body of linguistic cues that leads listeners to perceive speech as sounding gay or straight is not fully understood. The following is a review of studies of such cues. Gaudio (1 994) examined the relationship between pitch and perceived gayness. Eight men - four gay and four straight - read two passages, one technical and one dramatic. Thirteen nayve listeners then heard the recordings and indicated their perceptions of four aspects of the speaker, including gaylstraight. His study showed that average pitch (average FO) was correlated with neither perceived sexual orientation nor actual sexual orientation. However, his data suggest that male voices that use more pitch variation and a larger pitch range were perceived to sound both gayer and more feminine. As female speech tends to have both of these linguistic features, speaking with more pitch variation and a wider pitch range may be a way of showing the gender nonconformity often seen amongst gay males (Rieger, Linsenmeier, Gygax, Garcia & Bailey, 2010). Rogers et al's (2003) findings confirm the results of Gaudio (1 994) with a larger sample. They worked from a data bank of 25 male voices reading passages in various tones that were ranked on various continuums, including gaylstraight and masculinelfeminine. Rogers et a1 (2003) also found no correlation between perceived Vowel Duration and the Gay Accent 5 sexual orientation and FO. They did find that listeners thought it acceptable to list voices as both gay-sounding and masculine-sounding; sounding gay did not always correspond to sounding feminine; Gaudio (1994) found the opposite, that "straight" and "masculine" corresponded and "gay" and "effeminate" corresponded. The changing stereotypes across the decade between these studies could explain this difference. Munson, Jefferson and McDonald (2006a) looked at fricative identification and perceived sexual orientation. They created a synthetic Is/ to /I/ continuum, from which 40 listeners heard a subset. Listeners rated their perception of the speaker's sexual orientation. The study did not find that perceived sexual orientation had any relationship with fricative identification in male voices. Gay men in Pierrehumbert et al's 2004 study produced vowels using a more expanded vowel space than straight men. Their study involved self-identified gay and straight men and women reading a set of sentences that were played for naYve listeners. The listeners rated the recordings using a seven point scale, ranging from "sounds totally straight" to "sounds totally gayllesbian." As their listener judgments were generally quite accurate, they did not regroup speakers by perceived sexual orientation and instead left them grouped by actual sexual orientation. The gay men and lesbian women had more dispersed vowel spaces than their heterosexual counterparts. Straight women had a more expanded vowel space than straight men and tended to have more precise vowels. The vowel space expansion among gay men may be a female speech feature they have adopted as a way of showing gender nonconformity. This explanation, however, does not explain the same phenomenon among lesbian women, though it is perhaps just the result of the backing of lo/ and /u/. Munson, McDonald, DeBoe & White 2006b, discussed Vowel Duration and the Gay Accent 6 below, found an expanded vowel space to be associated with the perception of sexual orientation, though an expanded vowel space was not associated with actual sexual orientation. Munson et a1 (2006b) examined vowel space and Is1 skewness in gay and lesbian speech. Forty-four speakers recorded a list of single words; these data showed that there is no significant correlation between vowel space and sexual orientation, but that there is significant correlation between self-reported sexual orientation and Is/ skewness. Gay men had more negatively skewed Is/ sounds, though the difference was not significant for lesbian women; this Is1 skewness is what is often referred to as the stereotypical "gay lisp." Munson et a1 (2006b) also included a perception experiment in which 40 listeners heard the word lists used in the previous part of the experiment. They ranked, using a five-point scale, their perception of various qualities of the speaker, one of which was gayness or straightness. Predictors of a more gay-sounding rating for women were lower F 1 and F2 values and a more contracted vowel space, contradicting Pierrehurnbert et a1 (2004). Predictors of a more gay-sounding rating for men were higher F1 and F2 values and a more negatively skewed Is/. The current study examines another aspect of vowels and perceived sexual orientation - vowel duration. Based on previous findings that vowel placement (height and backness) cues listeners to make a judgment about the speaker's sexual orientation, as well as my own observations, I hypothesized that vowel duration would also cue a judgment of sexual orientation. The hypothesis driving this study was that male gay- sounding speech includes longer vowels than male straight-sounding speech. I also Vowel Duration and the Gay Accent 7 hypothesized that this effect might be greater in diphthongs than in monophthongs, based on Pierrehumbert et al's (2004) finding that men speaking with the perceived gay accent articulate more clearly than others; longer diphthongs would emphasize the presence of two vowels in one syllable. In order to test this hypothesis, I followed the methodology of previous studies by recording self-identifying gay men and self-identifying straight men. Listeners then heard these recordings and ranked on a seven-point scale how gay or straight they thought the voice sounded. Based on these rankings, voices consistently judged to sound very gay or very straight were selected for vowel duration measurement. Methods Experiment 1 - Production The goal of Experiment 1 was to record speakers as experimental stimuli and data to test my hypothesis that vowel duration is longer in speech perceived to sound gay than in speech perceived to sound straight. 1.1 Participants Forty men participated in Experiment 1. Twenty were self-identifying gay men and 20 were self-identifying straight men. Within each of those groups, half were freshmen and sophomores (henceforth underclassmen) and half were juniors and seniors (henceforth upperclassmen). All were native speakers of American English and current students at a small liberal arts college in the upper Midwest. They were recruited through word of mouth. 1.2 Speech Materials The reading, a passage from Wikipedia about the history of pizza, is a neutral article that does not invite a particularly emotional or dramatic reading (see Appendix A Vowel Duration and the Gay Accent 8 for the speech materials). It was adapted to include at least two tokens of all the vowels and diphthongs of American English, with the exception of 131, which many Americans the age of the participants do not produce. The recordings were made using a unidirectional microphone sending information directly to PCQuirer running a PC computer. 1.3P rocedure Speakers recorded a short passage in a sound attenuated booth in the Linguistics Laboratory over various sessions in October and November of 20 10. Speakers were told that they were participating in a study entitled "Sexual Orientation and Communication." They were presented with the passage, which was typed in 12 point Times New Roman font and double-spaced, in the Linguistics Laboratory of the college. Participants were asked to read through it silently to verify that they were familiar with all the target words, though the target words were not specified to participants. All participants reported familiarity with the target words. They were instructed to record the passage in a normal speaking voice at a normal volume. Small errors in reading were accepted, but speakers were allowed to re-read the passage if they made any major errors, which occurred twice. Each recording took approximately ninety seconds. Following the recording, speakers filled out a survey indicating their gender and sexual orientation, native language, age and hometown (see Appendix B). They were asked to indicate on a scale from one to seven how gay or straight they think their voice sounds. An open-ended question asked speakers to reflect on whether they change their voice to sound gayer or straighter, and under what circumstances.