Speech Prosody 2002
This paper discusses some of the problems regarding the unnaturalness of speech data currently used in research on oral Japanese politeness and proposes improved techniques. Two experiments were carried out. First, to extract natural unscripted utterances, within a specific vocabulary and context, the experimental design of the Map Task was employed with the population controlled for social status. Second, to elicit a perception of the level of politeness of spoken Japanese, a rating experiment of formality was conducted. Raters scored the degree of formality of lexically similar but non-identical tokens without any manipulation. Magnitude Estimation (ME) was employed to re- flect the perceived formality. Results from the production side showed that raising the fundamental frequency (F0) or changing speech rate are not always correlated with increasing formality. In the perception experiments, listeners did not show good agreement in their judgement of formality for most stimuli, though some trends were observed in stimuli. Although the speakers produced the utterances with appropriate honorifics, neither F0 nor speech rate seem to be dominant factors for judging formality. Therefore there is a need to explore the interaction between lexical cues and acoustic cues for conveying formality.
Bibliographic reference. Ito, Mika (2002): "Japanese politeness and suprasegmentals - a study based on natural speech materials", In SP-2002, 415-418.