Bringing Sociolectal Variation to the Extremes: Investigating the role of gradience in perceptual judgments of sexuality from pitch accents

Abstract

The effects of belonging to certain dialectal and sociolectal groups on human speech has been the topic of much research since the conception of the field of linguistics. For the speech of gay men, there has been a plethora of studies on different acoustic-phonetic markers of queerness in speech. However, only one research project (White 2022) has begun to analyze whether there are underlying differences in the intonational systems of gay men as compared to their straight counterparts. That project showed that listeners considered bitonal ToBI pitch accents as significantly more gay sounding when produced by a gay speaker than monotonal pitch accents. The goal for this thesis is to recapitulate the findings of White 2022 in order to affirm the positive correlation between L+H* pitch accents and the indexing of gay identity in speech. Furthermore, both H* and L+H* pitch accents were manipulated to test whether increasing the extremity of an intonational token would yield more extreme judgments of that token. The overarching aim of the project is to use the answers to those research questions to provide insight into the nature of intonational variation for gay male speakers of American English.

Using an open-guise perception study, responses were collected from participants from a wide range of demographics judging the perceived gayness level of a single gay male speaker, with experimental design conditions being pitch accent type and amount of F0 manipulation. Linear regression models revealed that, as in White 2022, L+H* tokens were rated as sounding significantly more gay than H* tokens, across manipulation steps. For the manipulated stimuli, more phonetically extreme instances of L+H* pitch accents were correspondingly rated as sounding more gay, with positive effects rising gradiently with manipulation step number. However, compared to the effect of pitch accent type, this finding was less statistically significant. Interestingly, gayness ratings of H* also rose gradiently with manipulation step, with a discrete jump in ratings between Steps 2 and 3 of manipulation. This finding suggests that H* and L+H*, while still categorically different for gay speakers from the perspective of listeners, exist on a realizational spectrum (as found by others, e.g., Ladd & Schepman 2003) where more extreme tokens of H* begin to look more like—and thus be interpreted as—L+H* pitch accents. This thesis investigates the consequences of looking at intonation through a variationist lens and reveals fruitful areas for future research as linguists continue to investigate intonational variation across dialects and sociolects.