摘要
•Parasagittal tongue sensor positions alone unreliably measure lateralization.•Lateralization is reliably quantified using EMA with 3D palate shape data.•Tongue-side lowering relative to the palate distinguishes /l/ from /t/ and /n/.•Taiwanese Mandarin /l/ does not show greater lateral asymmetry than /t/ or /n/.•Lateral tongue bracing before /l/ release is vowel-dependent, occurring only in /i/.
This study uses electromagnetic articulography to examine the articulation of three coronal sounds in Taiwanese Mandarin: /l/, /t/, and /n/. Two methods are employed to analyze their lateral kinematics. The first compares the positions of parasagittal sensors with those of midsagittal sensors, quantifying lateralization based on their relative positions, with laterals expected to be produced with lowered sides of the tongue. The second compares the positions of the parasagittal sensors against the position of an estimated three-dimensional palate surface model of the speaker. Results from the first method indicated that lateral consonants do not necessarily involve lowering of the sides of the tongue relative to the midsagittal position. Results from the second method indicated that a significantly larger opening is measured between the parasagittal sensors and the palate for the lateral /l/ when compared to /t/ and /n/. Together, the results indicate that including the position of the palate is critical when quantifying lateralization. Coarticulatory effects of the following vowel were also more clearly interpreted from the second method, with the high-front vowel /i/ triggering lateral tongue bracing prior to the release of /l/. These findings demonstrate that incorporating palate morphology is essential for accurately capturing lateral articulation and associated coarticulatory patterns.