摘要
本文基于英文语料库研究论著前言语篇,采用自下而上的语步分析法,利用"语篇瓦片叠压"(TextTiling)技术和多特征多维度分析法,考察语篇内部的语步结构、语篇功能及词汇语法表现。研究发现,英文语料库研究论著前言平均每篇包含7个左右语步,其中约36%的前言包含4—5个语步。后续的多维分析发现,多个语步体现出前言体裁的信息表达功能,主要用于陈述研究话题、写作背景等,还有少部分语步呈现出学术论证的特点,而在靠结尾部分多出现显性劝说型表述,旨在推销所介绍的书籍。其中,显性劝说虽然篇幅不大,但最具统计学上的显著性,可见它是前言语篇的特色功能。本实验表明,整合"语篇瓦片叠压"及多维分析的无监督体裁分析法,相对于通过经典的人工识别语步进行语言特征分析的做法,不但可行,而且更为客观高效,因此具有广泛的应用前景。
This paper experiments with the bottom-up move analysis of English corpus linguistics book prefaces combining TextTiling and Biberian Multi-dimensional methods.The within-text move structures,their discourse functions,and lexico-grammatical are examined.The results show that the prefaces are composed of seven moves on average,and 36%of all texts consist of four to five moves.The follow-up multi-dimensional analysis reveals that several moves are characterized by a functional focus of information packaging signifying research topics and background.A couple of moves are concerned with learned exposition.The concluding moves exhibit overt expression of persuasion serving promoting purposes.This last discourse function of the prefaces seems to identify itself as a characteristic of the genre in light the highest statistical significance among the three functions,despite of its limited proportion of all moves.The paper concludes that comparing with the canonical human identification of discourse moves and linguistic analysis,the unsupervised TextTiling and Biberian multidimensional analysis combined approach is not only viable but more efficient and less subjective,which promises wider application in other genres.
出处
《语料库语言学》
2020年第1期18-31,113,共15页
Corpus Linguistics
