Sociophonetic perspectives on stylistic diversity in speech research

Sociolinguistic data collection traditionally includes interviews, reading passages, and word lists (Labov 1972). Researchers have increasingly sought out elicitations tasks that have the benefits of read tasks (studying infrequently occurring variables; controlling for linguistic factors) while also eliciting styles more comparable to interview speech (see, e.g., Drager 2018). Examples include the Map Task (Brown et al. 1984) and the Diapix task (Baker & Hazan 2011). Other researchers have turned instead to elicitations tasks that maximize ecological validity, taking themselves out of the recording context altogether and training participants to collect their own field recordings (see, e.g., Podesva 2007; Sharma 2011). How comparable is the speech elicited from each of these tasks? Building on our previous results (Authors et al.015; Authors 2017) we consider three US English speakers’ vowel productions from interviews, reading passages, controlled ‘Lab Tasks’, and self-recordings. Our results suggest fewer differences across tasks than might be expected, suggesting that the interactional context may be more predictive of style shifting than the task, itself (Levon 2013).