Variability and Consistency in the Idiosyncratic Selection of Fillers in Japanese Monologues: Gender Differences

Variability and Consistency in the Idiosyncratic Selection of Fillers in Japanese Monologues: Gender Differences Shunichi Ishihara The Australian National University Canberra, Australia shunichi.ishihara@anu.edu.au Abstract This study is a linguistic study on idiosyn- crasy using speaker classification tech- nique as an analytical tool. The goals of this study are to find out 1) to what extent Japanese filler words (e.g. um, you know in English) carry individual idiosyncratic information; 2) if there are any differences in the degree/nature of idiosyncrasy be- tween the sexes; and 3) what contributes to the identified gender differences, if there are any. Based purely on the individ- ual selection of fillers, we report in this study that 1) speaker discrimination per- formance was better in the male (ca. 85% accuracy) than the female (ca. 75% accu- racy) speakers by approximately 10%, and 2) the poorer performance of the female speakers was due to the larger within– speaker differences in the female speakers than the male speakers. That is, the selec- tion of fillers by female speakers is more variable, speech by speech, than that by male speakers, even under similar condi- tions (e.g. same type of audience and the same degree of formality). We also dis- cuss that the findings of the current study agree with the previously–reported differ- ences between the sexes in language use. 1 Introduction We intuitively know that different people talk/write differently, even when they try to convey the same message. We also know that people tend to use their individually selected preferred words despite the fact that in principle they can use any word at any time from the vocabulary built up over the course of their lives. This is due to the idiosyncratic choice of words, expressions and so forth. Every speaker of a given language has their own distinctive and individual version of the language—which is often referred to as idiolect (Halliday et al., 1964; Coulthard and Johnson, 2007). This idiolect manifests itself in various aspects of communication, such as the choice of words, expressions, or even grammar, morphology, semantics and discourse structure. The idiosyncratic nature of word selec- tion between speakers/writers has been studied in different fields. For example, it has been used to understand speaking styles of political leaders (Slatcher et al., 2004), to identify the authors of literary works (Thisted and Efron, 1987), to detect plagiarism (Woolls, 2003) and to enhance the performance of automatic speaker recognition (Doddington, 2001). In the domain of text (in contrast to speech), it has been demonstrated that word category usage is very stable across time and writing topics (Pennebaker and King, 1999). Besides the idiosyncrasies of individual speak- ers, men and women speak/write differently (Kop- pel et al., 2002). This has been well reported in various linguistic and non–linguistic aspects of speech (Lakoff, 1975; Coats, 1993). Particularly in terms of linguistic styles, it has been argued that women tend to be more stylistically flexible and varied than men in language use (Holmes, 1997; Holmes, 1998; Chambers, 1992). Thus, the cur- …

Content truncated to 500 words