Developing a reading proficiency scale for Chinese as a second language: a confirmatory factor analysis approach

Abstract

There is a long history of operationalizing theoretical constructs of language proficiency by developing language proficiency scales. However, existing scales are not directly applicable in teaching and assessing Chinese as a second language (CSL) reading in US higher education contexts. Teachers need guidance regarding what component reading skills/knowledge should be taught and measured. This study developed a CSL Reading Proficiency Scale aligning to Koda’s (2005) theoretical framework that models L2 reading as a three-factor construct. The scale development began with modeling the construct of CSL reading ability on the basis of literature review. Drawing on this process, an operational descriptive scheme and a descriptor pool were established and subjected to expert review of content validity. Descriptors that survived expert review were then compiled into a Likert-scale questionnaire. This instrument was tested through teachers’ evaluations (N = 179) of whether CSL learners ‘can do’ what were denoted by descriptors. Confirmatory factor analysis exhibits solid goodness of fit between descriptors and the hypothesized three-factor model. The total score and the three subscale scores show high reliability. Evidence of convergent validity, discriminant validity, and concurrent validity indicates that the Scale measures CSL reading ability as intended. 摘要 长期以来, 外语教学和测试领域通过制定语言水平量表来描述语言水平的理论构念。然而, 现有的量表在应用于美国高校汉语作为第二语言 (CSL) 阅读教学和测试中有一些局限性。对于具体应该教和考哪些阅读子技能和子知识, 教师们需要更多的有效指导。本研究基于Koda (2005) 的三因素第二语言阅读模型, 开发了一个新的 CSL 阅读水平量表。量表的开发始于根据已有的研究来描述 CSL 阅读能力的构念。在此基础上, 作者建立了一个描述语体系和描述语语料库。语料库中 57 条通过了专家内容效度评审的描述语被编制成一套李克特量表问卷。教师使用这套问卷对 179 位 CSL学习者的阅读能力进行了评估。验证性因子分析结果显示数据与假设的三因素模型之间具有良好的拟合度。总分和三个子量表的分数都显示出很高的信度。 此外, 该量表具有很好的收敛效度、区分效度和共时效度。