Vowels in checked syllables are almost short and abrupt in Chinese Mandarin dialects. This paper proposed a parametric statistical modeling for the analysis of these vowels.The parameters were described empirically wi...Vowels in checked syllables are almost short and abrupt in Chinese Mandarin dialects. This paper proposed a parametric statistical modeling for the analysis of these vowels.The parameters were described empirically with the method of probability distribution. The effects of the multiparameter classifications were estimated by the binary and multinomial logistic models, and the method of support vector machine was adopted to carry out the pattern recognition of the classifications. The results show that the parameters of duration, DCT_（1-3）are significant to the external classification, and the parameters of DCT_（0-3） have significant main effects on the internal classification. The automatic testing shows that the hit rate of the external pattern recognition is only 43.6% but the rate is 96.1% of the internal. It indicates that the abrupt feature should be additive rather than distinctive.展开更多
基金supported by the Innovation Project of Phonetics and Grammar of Chinese Dialects from Chinese Academy of Social Sciencesthe Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institution(PAPD)
文摘Vowels in checked syllables are almost short and abrupt in Chinese Mandarin dialects. This paper proposed a parametric statistical modeling for the analysis of these vowels.The parameters were described empirically with the method of probability distribution. The effects of the multiparameter classifications were estimated by the binary and multinomial logistic models, and the method of support vector machine was adopted to carry out the pattern recognition of the classifications. The results show that the parameters of duration, DCT_（1-3）are significant to the external classification, and the parameters of DCT_（0-3） have significant main effects on the internal classification. The automatic testing shows that the hit rate of the external pattern recognition is only 43.6% but the rate is 96.1% of the internal. It indicates that the abrupt feature should be additive rather than distinctive.
