Previous research Travis (1981) –first study to evaluate homosexual voice perception • Homosexual participants had higher fundamental frequency modes in English • Worked from an abnormal perspective – the homosexual participants should be “corrected” in clinics Gaudio (1994) • Goal: isolate the specific variables that indicate homosexual speech in English • Variable analyzed: variety of intonation • 13 participants listened to 4 heterosexual men and 4 homosexual men • Variable does not indicate the sexuality of participants – constructed from the total segment 2 Previous research Levon (2006; 2007) -More pertinent to present study • One homosexual speaker rated by listeners • Variable: /s/ realization (duration and range of tone) in English • Digitally manipulated variants (one manipulation for each segment) by percentage (shortened sibilants by 17%) • Did not find statistically significant effects Mack (2010, 2011) • First study of perceived sexual orientation in Spanish • The participants produced the following /s/ variants: [s], [h], [Ø] instead of digital manipulation • Methodology did not control for competing factors that could influence rankings 5 The study: objectives 1) To analyze the linguistic variables of sheísmo and /s/ retention, variants associated with feminine speech in Buenos Aires Spanish 2) To eventually expand the binary variable of [+/- heterosexual] to study human sexuality and masculinity as a continuum 6 Variables under analysis “ll” realization and Pre-consonantal, word-internal (V__C) ex. casco the realization of palatal “ll” as in calles Fontanella de Weinberg (1978) change in progress [ʃ] [dʒ] led by women Wolf and Jiménez (1979) [ʃ] = women [ ʃ ] replaces [dʒ] Chang (2008) [dʒ] before [s] “Since /s/ realization is so highly correlated with social factors from a production standpoint, future studies are needed to assess the impact of /s/ production on perception of social identities” (Mack 2010: 54). [h] [Ø] variation in /s/ production by sex: Buenos Aires Fontanella de Weinberg (1973b) cross-dialectical comparison Terrell (1981) Havana (Dohotaru 2004) /s/ variation by sex women = [s] prestige variant, men = [h] [Ø] vernacular variant [ʃ] 1975 after /s/ realization 7 The study Research Questions: 1) How will the heterosexual participants, as identified via an extensive background questionnaire, pattern in the production of the palatal “ll” and /s/ ? 2) How will the listeners identify the speech of the heterosexual men along the scale of masculinity? 3) Will L1 Spanish listeners pattern similarly to L2 Spanish listeners in the classification of masculinity of the speech samples? 8 Methodology: Speakers Participants: 5 native speakers of Buenos Aires Spanish -self-identify as heterosexuals age range: 26-52 -recruited through personal contacts of investigator based on self-identification 1) Background questionnaire – demographic and sexual activity information 2) Recording– Read the following text out loud: Mi perro y yo andamos por las calles cerca del barrio y nuestros vecinos nos llaman por nuestros nombres. Prefieren que vayamos por la vereda. Aunque ya tenía su pelota, mi perro se cayó cuando atrapó otra pelota que encontró. Le pregunté a mi vecino, “Hola, ¿qué haces? ¿Dónde vive el dueño de la pelota?” y respondió él, “No sé. ¿Dónde la encontraste?” Written to elicit phonetic variables under analysis Recorded on computer microphone and sent to investigator from Argentina 9 Questionnaire Questions about participants’ sex lives • • • • Self-identified sexual orientation Gender identity Sexual experience (physical) Sexual fantasies (content, preferences) Calculation of Kinsey number for each participant Homosexual behavior/ Heterosexual behavior X 6 = K (K=Kinsey number) The Kinsey Scale 0 Exclusively heterosexual 1 2 3 4 Bisexual 10 5 6 Exclusively homosexual Manipulation Manipulation of each variable utilizing PRAAT (Boersma & Weenink 2014) to produce perception task 1) One example of each /s/ variant was located in each speaker 2) Replaced /s/ realization within each speaker with the three variants 3) Each of the three variants was spliced into the same word (respondió) Classification of /s/ realization [s] [h] -Controlled for all other factors -Manipulation of /s/ will be discussed today 11 [Ø] Variant classification: [s] 12 Variant classification: [h] 13 Variant classification: [Ø] 14 Methodology: Listeners Participants: 13 sequential Spanish-English bilinguals (no simultaneous) L1 Spanish/ L2 English L1 English/ L2 Spanish 6 7 3F/3M 5F/2M Spain, Mexico, Peru, Argentina Spain, Mexico Number of participants Gender Countries of origin/study A methodological shortcoming that will be discussed -all advanced graduate students & Spanish instructors 15 Methodology: Listeners 1) Questionnaire - personal information, experience with linguistics, phonology 2) Rating – forced choice socio-perceptual identification task on Quia 5 practice + 15 speech samples (3 tokens x 5 speakers) -fixed order, no sequential repetition of speaker or variant “‘¿Dónde vive el dueño de la pelota?’” y respondió él… How would you classify the speaker? Masculine More or less masculine Not masculine, more effeminate -no other visual stimulation -statistics run with SPSS (IBM Corp) 16 Results: The scale of masculinity Calculation of Kinsey number Distribution of participants along Kinsey scale Names have been changed to protect participant privacy 17 Results: Production of variables /s/ Production scale by participant in respondió /s/ Production scale by participant in nuestros 18 Results: Production of variables “ll” Production scale by participant in calles 19 Results: Matched-guise task Perceived masculinity by L1 and L2 Spanish speakers 20 Results: Matched-guise task Perceived masculinity by speaker 21 Results: Matched-guise task Perceived masculinity by /s/ variant 22 Results: Matched-guise task Listener found to be significant (not surprising) and needs to be controlled for Mixed model controls for repeated measure (listener) 1) Generalized estimating equation -controlled for listener as random effect 2) Ordinal logistic regression (like logistic but for 3 DV) -accounted for tripartite dependent variable (ranking 1-3) Variant presented significant (Wald Chi-Square 9.796, df=2, Sig=.007) [Ø] (Wald-Chi-Square 4.121, df=1, Sig=0.042) [h] (Wald-Chi-Square 0.36, df=1, Sig=.849) [s] 23 No statistical difference between aspiration and realization - statistical difference between [+/- production] variable Discussion 1) How will the heterosexual participants, as identified via an extensive background questionnaire, pattern in the production of the palatal “ll” and /s/ ? • Production varies for both variables but center around [h] and [Ø] on spectrum 2) How will the listeners identify the speech of the heterosexual men along the scale of masculinity? • Variable under analysis, /s/, predict rankings of masculinity • No statistical difference between aspiration and realization – statistical difference between [+/- production] variable • Different from Levon (2006; 2007) who did not find support • Speaker was significant 3) Will L1 Spanish listeners pattern similarly to L2 Spanish listeners in the classification of masculinity of the speech samples? • Significant differences between L1 and L2 Spanish speakers 24 Conclusion and future work Future work: Additional speakers, expand spectrum of sexuality to include selfidentifying bisexual and homosexual L1 and L2 Spanish from/learned in Argentina In high-contact metropolises such as BA, second dialect contact is unavoidable Conclusion: Difficult to isolate variables that affect perception of masculinity More matched-guise studies needed to examine L1 and L2 differences How do learners acquire horizontal variation? Additional studies into languages other than English to examine possible cross-linguistic patterns 25 ¡Gracias! Thank you to Sound Change in Interacting Human Systems Workshop, especially Drs. Keith Johnson, Andrew Garrett, & Larry Hyman The Department of Spanish and Portuguese at Indiana University, especially Silvina Bongiovanni, Megan Solon, and Danielle Diadone Dr. María Eugenia Pérez Ibáñez, Conservatorio Municipal de Música de Falla, Buenos Aires Indiana University Statistical Consulting Center Dr. Laura Colantoni , University of Toronto Contact information: Meg Cychosz: mcychosz@indiana.edu 26 Masculinity rating by listener gender 3 2 1 Marco Felipe Luis Javier Carlos Rating distribution by female and male listeners 1 2 3 Female listeners N N Male listeners Marco Felipe Luis Javier Carlos Marco Felipe Luis Javier Carlos