To predict syntactic complexity in second/foreign language writing, some studies have advocated the use of T-unit and clausal subordination measures while others have argued for the use of phrase-based measures. This study seeks to identify syntactic features that can be regarded as discriminators among different levels of writing quality. For this purpose, a corpus of argumentative essays by EFL learners was compiled and then the essays were rated and placed into three groups of high-rated, mid-rated, and low-rated essays. The corpus was then coded and analysed for both phrasal and clausal features. The phrasal features were manually coded based on the development scheme hypothesized by Biber, Gray and Poonpon (2011) for academic writing, and the clausal features were analysed using the online L2 Syntactic Complexity Analyzer developed by Lu (2010). A separate ANOVA test was used to compare the three groups of essays for each of the phrasal and clausal features. The findings of the current study demonstrated that subordination and dependent clauses were not good indicators of different writing qualities in our corpus. Also, the pattern of noun phrase complexity predicted by Biber et al. (2011) was not observed across argumentative essays from three different levels of writing quality.

Keywords