Contextual sociophonetic prediction beyond the gender binary

Abstract

Previous research has demonstrated social information affects listeners' linguistic decision-making. Strand and Johnson (1996) showed that imputed gender shifts listeners' sibilant category boundaries. Further research has shown sibilant identity influences listeners' binary categorization of gender, suggesting social-linguistic bidirectionality (Bouavichith et al., 2019). This study extends this body of literature by investigating how sibilant categorization changes when acoustically masculinized speech is framed within differing social contexts. Participants completed a lexical decision task, where each word consisted of a synthesized sibilant onset and a naturalistic rime. In block 1, rimes were minimally manipulated; listeners were told the speaker identified as female during the recording. In block 2, rimes were masculinized; this manipulation was contextualized in condition 1 as the speaker's gender transition and in condition 2 as digital manipulation. If sensitive to the use of phonetic variation to convey social meaning, condition 1 listeners would be more likely than condition 2 listeners to adopt a categorization strategy in block 2 consistent with hearing a male-sounding voice (i.e., more likely to categorize ambiguous sibilants as /s/). As expected, condition 1 participants were more likely to categorize a sibilant as /s/ in block 2, while condition 2 participants did not differ across blocks.