世新大學圖書館

Source: Dissertation Abstracts International, Volume: 65-06, Section: A, page: 2184.

Adviser: John H. Esling.

Thesis (Ph.D.)--University of Victoria (Canada), 2004.

The voices of heroes and villains in Japanese animation (anime ) are thought to represent the vocal stereotypes of good and bad characters in Japanese culture. In this study, phonetic properties of the voices of heroes and villains in anime were examined.

In Chapter 1, four hypotheses about the auditory, acoustic, and perceptual characteristics of the voices of heroes and villains were formulated based on previous research on vocal stereotypes and vocal cues to personality and emotion. After a preliminary study using the voices of heroes and villains from four TV series, 20 anime series and movies were selected for the main analysis (Chapter 2). In Chapter 3, the auditory analysis of the voices of 88 anime characters was performed, using a modified version of Laver's descriptive framework for voice quality (Laver, 1994, 2000). Based on this analysis, in which epilaryngeal settings (i.e., laryngeal sphinctering vs. pharyngeal expansion) played a significant role, four voice types were identified to categorize the voices of heroes and villains. Following the auditory analysis, a series of acoustic analyses, namely pitch, vowel formant, and spectrographic analyses, were performed, and the relationship between these acoustic measures and the character roles and voice types was examined (Chapter 4). In Chapter 5, in order to investigate whether the identified auditory characteristics contribute to people's perception of good and bad characters, Japanese laypersons' perceptions of selected speech samples were examined in an experimental setting where 32 participants listened to content-masked speech excerpts of the 27 selected target speakers and rated their impressions of age, gender, physical characteristics, personality traits, emotional states, and vocal characteristics. Quantitative and qualitative analyses were performed in order to examine the relationship between auditory correlates and the participants' trait attributions. Lastly, the results from the three aforementioned components of the present study (i.e., the auditory and acoustic analyses and the perceptual experiment) were compared statistically by calculating correlations among the three, using correlation analyses, factor analysis, and cluster analysis

The findings of this study are as follows (see Chapter 7). The present study was able to identify the auditorily critical vocal components that differentiate good and bad characters, namely epilaryngeal states. Whereas the majority of the heroes' voices exhibited an absence of pharyngeal constriction and the presence of breathy voice, the majority of villains' voices exhibited non-neutral epilaryngeal states (i.e., moderate to extreme laryngeal sphinctering or pharyngeal expansion). The perceptual experiment that contrasted epilaryngeal states in anime voices was successful in confirming the effects of these settings on laypersons' perceptions. Participants attributed unfavorable physical traits, personality traits, emotional states, and vocal characteristics to speakers who exhibited non-neutral epilaryngeal states regardless of the roles they played in the original cartoons. The acoustic analysis results were less clear-cut in differentiating voices of heroes from those of villains. Mean F0 and F0 range did not differ very much between the two roles; as for vowel formants, only F2 was found to be consistently lower in villains than in heroes, which was attributed to pharyngeal expansion and, in the case of females, pharyngeal constriction as well. The results of the correlation analyses suggest that the auditory analysis results accounted for more of the variance found in the perceptual experiment than the acoustic measures. (Abstract shortened by UMI.)

School code: 0244.