A Study On The Movement Of Tongue In The Vowel System Of Shanghai Dialect

It is an essential way to research the phonetics from the perspective of physiological. With the rapid development of science and technology, equipments used by experimental phonetics researchers have been updated incessantly in order to obtain the articulator data more effectively. Nowadays, electromagnetic articulograph has been more and more used to obtain data of articulator in the mouth for the real-time, directly, and non-prejudicial advantages. In this paper, we researched tongue movement by the electromagnetic instrument NDI Wave. The purpose is to find rules of tongue movement of Shanghai Dialect’s finals depending on using the clustering algorithm.First, the paper introduced the related basic knowledge, such as dynamic time warping, cluster analysis, principal component analysis, etc. Then according to the unequal time lengths of the experimental data, the paper propose a improved cluster method based on dynamic time warping(DTW), and the method can compute the unequal data directly. Compared with traditional way, this improved algorithm no longer need to find a sample as a template, which is used to warp the other samples’ time length, so the cluster result will not influenced by different templates.Second, in the experiment stage, the paper explained how to effectively used the limited number of sensors, and proposed a way that three sensor(left and right mastoid, upper incisor) used to constitute a reference plane in order to remove the interference to tongue movement from head rigid motion in experiments, and all features are determined on the basis of this reference plane.Third, after a series of data preprocessing, we used the improved algorithm to get cluster result. By the aid of phonetics professional knowledge, we determined the number of cluster is six.Finally, through the observing and comparing the results and the simplified trajectory, we found the similarities between components and features, and finals in same class also have similarities. We proposed the possibility that the data from tongue tip and tongue back may distinguish Shanghai Dialect’s finals better.