A database of voice onset time (VOT)* values from 122 languages and 37 language families. Values were collected in a survey of approximately 350 theses, articles, grammars, and manuscripts, in which 164 sources contained stop-specific VOT means. VOT data points with shared place and voice were averaged within each study, resulting in a total of 1079 VOT means. These are grouped by a shared laryngeal feature value or gesture, as specified in the primary source.
Data collected by Eleanor Chodroff (Northwestern), Alessandra Golden (JHU), and Colin Wilson (JHU). Accompanying manuscripts: Chodroff, E., Golden, A., and Wilson, C. (in prep). Covariation of voice onset time: a universal aspect of phonetic realization. A large portion of this dataset also appears in Chodroff, E. (2017). Structured variation in obstruent production and perception. PhD Thesis. Johns Hopkins University.
*Voice onset time (VOT) is a commonly measured phonetic dimension of stop consonants (e.g., /p t k b d g/). It refers to the duration from stop release to the onset of vocal fold vibration, which can occur either before or after the release. For languages that have a laryngeal contrast (e.g., /p t k/ vs /b d g/), VOT is typically one of the primary correlates: /p t k/ have longer VOT values than /b d g/. There are also known differences across place of articulation within a laryngeal series. Dorsal stops (e.g., /k g/) have longer VOT values than corresponding labial stops (/p b/); the VOT of coronal stops tends to fall intermediate to those of dorsals and labials. As you can see in the figure, the differences across place of articulation are highly regular across languages. The strength of the correlations suggest that this difference may be reducible to a single dimension of variability. We argue that the correlations arise from a uniformity constraint requiring (ideally) uniform phonetic realization of the laryngeal feature value for all specified stops.