Examining the Methods Employed to Analyze the Spoken Discourse of Japanese Learners of English: Characteristics of Discourse Analytic Measures

为了分析以英语为第二语言的口语转录语篇，语篇分析方法通常被用来具体评估流利性、准确性、句法复杂性和词汇复杂性。尽管它们在二语习得研究中经常被使用，但这些措施的特点以及哪些措施应该在特定的语境中使用还没有被研究过。本研究旨在从多个角度探讨语篇分析方法的特点。研究结果显示了三个主要发现。首先，能够持续检测学习者熟练程度变化或差异的指标是“每分钟令牌的数量”。其次，每一项测量都与其他每一项测量以某种程度上可预测的方式相关联，并且每一项测量都评估了演讲表现的一部分。第三，当任务涉及自我介绍和图片描述时，相同的衡量标准可以在不同任务中以类似的方式被解释。

For analyzing spoken transcribed discourses in English as a second language, discourse analytic measures have often been employed to specifically assess fluency, accuracy, syntactic complexity, and lexical complexity. Despite their frequent use in second language acquisition research, the characteristics of these measures and also which measures should be used in a certain context have not been examined. The current study aims to explore the characteristics of the discourse analytic measures from multiple perspectives. The results indicated three main findings. First, a measure that can consistently detect a change or a difference in a learner's proficiency is "the number of tokens per minute." Second, each measure is related to every other measure to some degree in a predictable manner, and each measure assesses a part of the speaking performance. Third, the same measure can be interpreted in a similar manner across tasks when the tasks involve self-introduction and picture description.