Linear functions relating the horizontal position of the tongue back (TBx) at the consonant release (dependent variable) to TBx at the middle of the vowel (right) and F2 at the consonant release to F2 at the middle of the vowel (left) for each of the consonants examines. Each line fits 90 tokens from six vowels for the EMMA subject.
Analysis of Intercepts into their constituents: slope and average of F2 or TBx at C and V according to Eq. (2) for EMMA data. Frontness increases vertically.
Analysis of Intercepts into their constituents: slope and average of F2 or TBx at C and V according to Eq. (2) for XRMB data. b/p refers to the labial stop, d/t refers to the alveolar stop, and g/k refers to the velar stop. Frontness increases vertically.
Trajectories of TBx for consonants from EMMA data: For all consonants, the first articulatory sample in the time series corresponds to the acoustic frame in which the formant transition for the consonant begins. Trajectories in black are for the back vowel context, and trajectories in gray are for front vowel context. Frontness increases vertically.
Comparison of statistics of CV: TBx relation at the initial point of the formant transition and at the point of tightest closure in the consonant for EMMA data. Frontness increases vertically.
Spatial effects in coarticulation: Means and standard deviations of change in TBx between initiation of C formant transition and middle of V for each consonant in EMMA data, separated into back vowels (black) and front vowels (white).
EMMA velar data split by front and back vowels. Upper panel: Average TBx position at the beginning of F2 onset (c). Lower panel: TBx-based slopes.
Words used from the Wisconsin XRMB database. Each word is followed by the particular CV included in the analysis.
Linear parameters for EMMA data.
Linear parameters for XRMB data. The upper number in each cell contains the average across all subjects, for regressions done within each subject. The lower number in each cell contains the statistics for the regressions performed on the data pooled across subjects.
Mixed effects GLM results for each place of articulation vs. Labial: effect sizes (Dif), 95% confidence intervals (LB, UB), and MCMC derived p-values.
Normalized coarticulation resistance.
Article metrics loading...
Full text loading...