Socially stratified phonetic variation and perceived identity in Puerto Rican Spanish

This dissertation examines the interaction between phonetic variation and perceptions of speaker identity in Puerto Rican Spanish. Using an interdisciplinary approach, three experiments were designed and carried out: (1) an descriptive study of stereotypes about sexual orientation and male speech, (2) an observational study examining the relationship between acoustic parameters and perceived sexual orientation, perceived height, perceived social class, and perceived age, and (3) an implicit-processing experiment examining the influence of social stereotypes on memory for voices.;The study was carried out in the San Juan, Puerto Rico, metropolitan area and included ninety-six participants. Results of the first experiment indicate that there is considerable uniformity in notions of speech variation associated with the gay male speech stereotype for the participants in the study, and that the most cited stereotypical markers of sexual orientation are related to stereotypical notions of gender. However, a majority of the respondents explicitly stated that although they realize a stereotype exists, they do not believe there is necessarily a correspondence between stereotypes of gay men's speech and real life production. Results of the second experiment show that listeners do evaluate speakers' voices differently in terms of perceived sexual orientation, and that perceptions of sexual orientation are most strongly predicted by one acoustic measure of vowel quality (the second resonant frequency of the vowel /e/, which relates to tongue position in the anterior-posterior dimension). An examination of the relationship between perceptions of sexual orientation and perceptions of height, age, and social class revealed that perceptions of height were correlated with perceived sexual orientation. The third experiment showed that listeners responded more quickly to speakers previously rated as more gay sounding than they did to speakers rated as more straight sounding, and the slowest mean responses were for the deleted variant. Most significantly, a d-prime analysis showed the strongest signal detection in the case of the sibilant ([s]) when produced by stereotypically gayer sounding speakers. The results suggest a relationship between /s/ variation and listener perceptions of sexual orientation as well as a possible effect of perceived sexual orientation on speech processing. Taken together, these results underscore the need for methods that measure both conscious and subconscious effects of stereotypes in speech production and perception.