Abstract
We analyzed longitudinal survey data on language change of honorifics use at home in Okazaki city of Aichi Prefecture, Japan. The Okazaki survey was carried out three times over 55 years, in 1953, 1972, and 2008, by National Institute for Japanese Language and Linguistics (NINJAL) and The Institute of Statistical Mathematics (ISM). A number of studies have demonstrated that a process of the language change follows a kind of S shape curve. Yokoyama, Asahi, and Sanada (2008) proposed a logistic regression model to predict language changes, representing as a multi-variate S shape curve. The model assumes two major variables; the birth year of the participants and the year of survey. We made our predictions of the 2008 data, based on logistic regression analysis of the 1953 and 1972 data. The result showed the differences between the prediction by the model and the observed data in middle or old (or high) age class. We discussed what factors determine the differences between the prediction and the observed data.
