It may be more difficult to extract fundamental utterance styles in real-life daily conversation than those in fictional utterances because the characteristics of utterance styles are exaggerated in fictional utterances. However, by utilizing a large-scale corpus of daily conversation, it is possible to identify the fundamental patterns of Japanese utterance styles. In this study, the NUCC was targeted and extraction of the characteristics of utterance styles was carried out using the statistical method of factor analysis. As a result, five factors (“Average style in NUCC,” “Avoid affirmation style,” “Frank teenager style,” “Dialect style,” and “Polite style”) were extracted quantitatively. Compared to fictional utterance styles, “Avoid affirmation style” is unique in real daily conversation. On the other hand, “Crude style” and “Hearsay style” do not appear. Although the similarities between the fictional corpus and the NUCC support the validity of the result, the factors were impacted by bias in the corpus. It would be desirable to utilize a speaker-balanced daily conversation corpus for a more precise analysis.
@InProceedings{MURAI18.1, author = {Hajime Murai}, title = {Factor Analysis of Japanese Daily Utterance Styles}, booktitle = {Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018)}, year = {2018}, month = {may}, date = {7-12}, location = {Miyazaki, Japan}, editor = {Hanae Koiso and Patrizia Paggio}, publisher = {European Language Resources Association (ELRA)}, address = {Paris, France}, isbn = {979-10-95546-16-0}, language = {english} }
Summary of the paper