Assessing voice vs. writing quality

Abstract Using a mixed-methods design, the study aimed to distinguish voice features from the criteria considered in most L2 writing rating scales (e. g., topic development, organization, and quality of language use). Accordingly, a sample of 200 argumentative essays written in English was scored for voice, writing quality, and the critical features of academic writing such as text length, lexical sophistication, and grammatical complexity (see Cumming et al., 2005 ). Multiple regression and structural equation modeling (SEM) were conducted to verify which of and how the critical elements of academic writing interacted to determine the scores for voice and writing quality. Think-aloud protocols (TAPs) of raters scoring voice and writing quality were also compared. Voice was found an indication of genre competence, rated independently of language use and mostly based on the structure of arguments and the quality with which writers employed certain textual voice elements. In assessing writing quality, however, textual voice elements were dismissed, and the writing quality scores mainly reflected the quality of language use rather than the argument structure. The study provides implications for second language writing assessment and instruction, as well as for future research to be conducted.