A Gestural Account of Consonantal Length Contrasts: Articulatory Evidence from Japanese Bilabial, Alveolar, Palatal, and Dorsal Mimetic Geminates

We present articulatory data regarding how Tokyo Japanese speakers produce consonantal length contrasts. Six speakers produced fifteen mimetic items across bilabial, alveolar, palatal, and dorsal places as singletons and geminates. Data was recorded using Electromagnetic Articulography. Analyses focused on intra-gestural timing, kinematics, and inter-gestural timing properties. Geminate production shows a modestly longer closing phase but a much longer plateau phase duration, with no difference in release phases. Dynamic Time Warping analyses also localize temporal expansion to the plateau. Kinematically, geminates exhibit greater constriction amplitude, reduced peak velocity, and lower stiffness. Inter-gesturally, vowel–to–target and vowel–to–vowel lags increase, leading to increased acoustic duration of preceding vowels, while vowel–to–closure lags remain stable. Linear Discriminant Analyses reveal that temporal cues outperform spatial ones in classification. Based on these findings, we propose possible phonological accounts, including a unified gestural account: a local prosodic temporal modulation ($\mu_{T}$) prolongs activation, yielding plateau extension and emergent spatial strengthening, parsimoniously explaining gemination across articulators.