Third International Workshop on Models and Analysis of Vocal Emissions for Biomedical Applications (MAVEBA 2003)

Florence, Italy
December 10-12, 2003

Analysis and Evaluation of Nasalized [g] Consonant in Continuous Japanese

Hisao Kuwabara

Teikyo University of Science & Technology, Uenohara, Kitatsuru-gun, Yamanashi, Japan

Nasalized velar consonant [g] in continuous Japanese is often observed in some dialect and is said to decrease in frequency year by year. This paper deals with acoustic and perceptual analysis of this phenomenon. Test materials used in this experiment are read version of Japanese short sentences by NHK's (Japan Broadcasting Corporation) professional announcers. Each sentence includes at least one [g] consonant that would likely be pronounced as nasalized. An evaluation test reveals that less than 60% of nasalization has been found to occur for [g] consonants for which 100% nasalization bad been observed decades ago. Acoustic analysis for nasalized and non-nasalized [g] sounds has been performed mainly through waveform parameters. It has been found that power ratio between consonant and vowel is the most effective parameter for distinguishing nasals from non-nasals. But it is highly speaker dependent.

Index Terms. Nasals, velar, perception, waveforms

Full Paper (reprinted with permission from Firenze University Press)

Bibliographic reference.  Kuwabara, Hisao (2003): "Analysis and evaluation of nasalized [g] consonant in continuous Japanese", In MAVEBA-2003, 273-276.