Acoustic Properties of Formality in Conversational Japanese

Abstract

This paper examines potential acoustic cues for level of formality in Japanese conversational speech using speech data gathered outside the laboratory, with the objective of using any significant cues to develop a model to predict level of formality in spoken Japanese. Based on previous work on the phonetic properties of formality in Japanese [1],[2] and other languages [3], and on a pilot study of informal geminate contractions in Japanese (section 2), the study examined the mean f0, articulation rate, and f0 range (the difference between the minimum and maximum f0 in an utterance) via direct examination of the data and a functional data analysis [4], [5].Analysis of the speech data shows significant relationships between all three variables and level of formality, and a binary logistic regression indicates that the variables have some potential as predictors of formality independent of lexical cues, although further refinement of any model will be necessary.In order to obtain appropriate conversational Japanese speech data to analyze, the decision was made to create a new small corpus. Although other corpora of spoken Japanese exist, notably the Corpus of Spontaneous Japanese [9] and the Chiba 3-way conversation corpus [10], they were judged to be

Capturing Formality in Speech Across Domains and Languages

“…However, our findings largely support previous work (e.g. [23,24]) that has pointed to the importance of variation in speech variables in encoding formality of spoken language.…”

“…In contrast to studies of speech transcripts, some work has measured acoustic-prosodic and linguistic features of formal and informal speech directly. Numerous authors, including [21,22,23], have found that speaking rate is an important indicator of formality across Korean, Elche Spanish, and Japanese respectively. [13]'s investigation of Dutch uncovered relationships between lower articulation rates and formal interactions, as well as the greater presence of interjections, filled gaps, laughter, and disfluencies in informal interactions.…”