语料库驱动的机器词典构建关键问题探讨

Critical issues on corpus-driven machine dictionarycreation

语料库驱动的语言研究试图最大限度地摆脱已有理论束缚,力求发现新的反映语言使用本质的事实。在这一理念指引下,我们重新审视和探讨了机器词典构建的几大关键问题(如:语言描述的核心、基本单位、释义模式等),并在此基础上提出一种新的机器词典构建理念,即:以意义为描述核心、以语言使用(或文本)为获取意义的本源、以集词汇、语法、语义、语用于一体的'扩展意义单位'为基本描述单位,采用列举与归纳相结合的释义架构和正则表达式的表示方法。

Corpus-driven linguistics tries to get rid of existing theories and to find out new realities which reflect the nature of language use.According to this idea,we try to re-examine several critical issues concerning machine dictionary creation:what is the core of language description,what is the fundamental unit,what is the appropriate model of explanation,etc.This paper proposes a new idea of machine dictionary creation which takes meaning as the core of description,language use(or text)as the essential source of meaning,and the extended unit of meaning as the fundamental unit of description,and makes use of enumerating and regular expressions to explain meaning.