ABSTRACT This paper investigates whether nonbinary speakers’ imitation of extended voice onset time (VOT) in word‐initial English /p, t, k/ is impacted by whether they believe they are listening to a nonbinary or binary model speaker. Forty‐five nonbinary American English speakers participated in an online VOT shadowing task, and the results find that nonbinary speakers diverge to a significantly greater degree away from a binary‐identified speaker than a nonbinary‐identified speaker or a speaker whose gender is not identified. Social information can impact the degree to which one speaker phonetically converges with another speaker, and nonbinary speakers have been shown to alter their speech due to their social environment, specifically when there is a threat of being misgendered. These patterns suggest that spaces that are not identified as nonbinary‐inclusive present a social threat for nonbinary speakers, thus motivating them to shift their linguistic productions depending on social context.
Jack Rechsteiner (Mon,) studied this question.