The variation among the subject-honorific constructions from the Edo period to the Showa period : A Bayesian approach using multinomial mixed-effects logistic regressions

エド　キ　カラ　ショウワ　キ　ニ　カケテ　ノ　ソンケイ　ゴ　ノ　バリエーション　タコウ　コンゴウ　コウカ　ロジスティック　カイキ　ニ　ヨル　ベイズ　トウケイ　ガク　テキ　アプローチ

ヤマダ, アキタカ

Yamada, Akitaka

Graduate School of Language and Culture, Osaka University

オオサカダイガク　ダイガクイン　ゲンゴブンカ　ケンキュウカ

多項混合効果ロジスティック回帰分析

尊敬語

変異

通時的変化

Bayesian statistics

multinomial mixed-effects logistic regression

subject honorifics

variation

diachronic development

This paper is a quantitative multifactorial study of the Japanese subjecthonorific constructions used from the Edo period to the Showa period. Subjecthonorific expressions are found in genealogically-unrelated languages, but unlike other languages, Japanese possesses several different competing constructions to encode the speaker's respect for the referent, and thus the examination of this peculiar system is expected to develop our understanding of subject honorifics, and the language variation in general. Using the data taken from the Corpus of Historical Japanese (CHJ), this study compares three subject-honorific constructions o…ni nar, …nasar, and o…nasar by conducting a Bayesian multinomial mixed-effects logistic regression. The findings are as follows: First, the imperative shows the largest effeet size indicating the difficulty of the o…ni nar construction's taking this sentence mood. Second, verbs show their own selectional tendencies. Finally, the selection of as ubject-honorific construction is also subject to the genre; o. ni nar is abandunt in textbooks of standard Japanese.