Abstract
The purpose of this study is to investigate the articulatory patterns of consonants and vowels in Taiwan Mandarin from the perspective of dynamical system. In the dynamical system framework, we need two pieces of information to characterize phonological system, namely the gestural targets for consonants and vowels and the interactions between gestures. The first goal is static, in which the static articulatory targets for consonants and vowels in Taiwan Mandarin were thoroughly described by using Electromagnetic Articulometry (EMA). Further, by comparing consonants and vowels with the same place of articulation, some manner effects on articulatory targets were revealed. The second goal is dynamic, in which dynamic CV transition and coarticulation were analyzed. Qualitatively, time functions of articulator movements were presented as a visualization of the CV coarticulation. Quantitatively, the degree of the observed coarticulation were numerically measured. Specifically, the second part of this study is to explore the coarticulatory patterns in Taiwan Mandarin, by means of direct visualization of CV coarticulation as well as numerical analyses on the degree of coarticulation. For the numerical analyses, a systematic scheme of the quantification of CV coarticulation is proposed. The coarticulatory patterns in Taiwan Mandarin revealed in this study is generally consistent with those in other languages described in the literature, and support the statement in the degree of articulatory constraint (DAC) model that a sound segment has greater degrees of coarticulatory aggressiveness and resistance at articulators that are more involved in the production of that sound segment (e.g., the tongue tip for /t/). Furthermore, the CV coarticulation results revealed that V-to-C effect is dominant over C-to-V effect in Taiwan Mandarin, which can be attributed to C-V gesture in-phase relation. By examining the dynamic patterns of CV coarticulation, this study presents that V-to-C coarticulation is mostly caused by gesture overlapping, which is not the case for C-to-V coarticulation.