基于会话语料库的日语性别角色语言自动提取方法论  

Automatic Extraction Methodology of Japanese Gender Role Language Based on Conversational Corpus

作　　者：孙昊 

机构地区：[1]日本同志社大学文化情报系,日本京都6100394

出　　处：《东北亚外语研究》2018年第1期48-54,共7页Foreign Language Research in Northeast Asia

摘　　要：本文对基于语料库挖掘的日语性别角色语言提取方法加以概述。该方法首先从日语口语语料库中抽取异性会话的文本,其次统计在异性的会话参与者中各种词汇的使用频度,最后引入特征选择方法之一的fisher精确检验的思想来抽取异性的会话中使用差异性较大的词汇。本研究将这种方法应用于日语自然会话语料库来提取性别角色语言。本研究的结果表明,除了已经被广泛接受的人称代词和终助词以外,日语中的「うん/(嗯)」「ええ/(嗯)」「ああ/(啊)」「ねー/ねえ/(表示确认)」「なあ/(啊)」「さ/(吧)」「よー/(哟)」「どー/(怎样)」「という/(表示说话,思考的内容)」等词汇可以被视为日语的性别角色语言。This article describes a corpus-based approach to Japanese gender role extraction.This method first extracts the text of the conversation between opposite gender from the Japanese oral corpus.Second,it counts the opposite gender conversation participants’frequency of use of various vocabularies.Finally,the idea of Fisher’s exact test,one of the feature selection methods,is introduced to extract the more significant vocabulary from conversations using the opposite gender.This study applies this method to Japanese natural conversation corpus to extract gender role language.The results of this study show that besides the widely accepted personal pronouns and finalizers,“un”,“ee”,“aa”,“nee”,“naa”,“sa”,”you”,“dou”,“toiu”,can be treated as Japanese gender role language.

关 键 词：角色语言 语料库挖掘 卡方检验 fisher精确检验1 

分 类 号：H0[语言文字—语言学]