摘要
评价意义是语篇语义层的人际意义资源,评价话语分析是以评价理论为分析框架对文本中的评价意义模式进行研究。由于评价意义与词汇语法形式之间没有一对一的体现关系,现有的评价话语分析一般需要在标注精密度和文本规模之间进行取舍。数据科学的介入有助于解决该困境。基于数据科学的基本原则阐释了话语分析中的数据处理流程,提出在人工精密语义标注的基础上使用数据处理语言辅助进行模式识别,从而提高研究的系统性和可重复性。以使用R语言辅助冲突话语中会话者评价风格识别为例,探讨数据科学在话语分析研究中的应用价值。该方法不仅可用于评价话语分析,也可应用于其他理论的话语分析,是后定性研究在话语分析领域的一次尝试。
Evaluative meaning is an interpersonal meaning potential on the discourse semantics stratum.Appraisal-based discourse analysis is to identify patterns in evaluative meaning-making based on the appraisal framework. However, since there is not always a definite set of lexicogrammatical realisations for a certain evaluative meaning, researchers need to manually annotate the data to ensure the accuracy, which renders processing large scale datasets difficult. A data analysis workflow normally used by data scientist is modified to suit the needs of appraisal-based discourse analysis. This workflow relies on fine-grained manual annotations as inputs and uses data science techniques for data analysis and identifying evaluative meaning-making patterns to improve research accuracy and reproducibility. A case study on the identification of evaluative meaning-making patterns in adversarial interactions is conducted to illustrate the workflow implemented with R, a programming language. The workflow is not limited to data analysis in appraisal-based discourse analysis, but can be used in discourse analysis in general to add to post-qualitative research methodologies.
作者
许庆欣
陈居强
XU Qing-xin;CHEN Ju-qiang
出处
《天津外国语大学学报》
2020年第6期24-38,155,156,共17页
Journal of Tianjin Foreign Studies University
关键词
评价理论
数据科学
话语分析
评价意义
后定性研究
appraisal theory
data science
discourse analysis
evaluative meaning
post-qualitative research
