Uncovering the dimensionality of authorial voice in the integrated writing task: A confirmatory factor analysis study

Research has long noted the importance of establishing an authorial voice as a hallmark of academic writing. However, despite purported connections between academic writing and integrated writing tasks (Gebril & Plakans, 2014; Huang & Hung, 2013), little attention has been paid to examine the role and effect of authorial voice on writing quality in a test task that involves use of source texts. The primary intent of this study is to shed light on the internal structure of authorial voice in an integrated reading-to-write task by way of Confirmatory Factor Analysis (CFA). The participants of this study included 315 Chinese college-level EFL students who wrote for a task addressing the topic of “eating for professional situations”. Based on an analytic rubric developed by the researchers with indicators informed by Hyland's (2005) interactional framework of authorial voice, the collected essays were rated on a scale of 1–5 based on the quality as it pertains to each of the indicators under the rubric. The results of a model comparison approach to CFA suggested that the current data are best characterized as a unidimensional construct; it represents one single factor. Furthermore, the Structural Equation Modeling (SEM) analysis indicated verbatim source use is inversely associated with quality of authorial voice, corroborating our hypothesis that improper source copying negatively influences the rater's perception of a test-taker's authorial voice. Practical implications regarding rubric use and instruction are discussed.