Examining English language learners’ longitudinal development of syntactic complexity across five CEFR levels with a robust measurement design: A mixed-methods approach

Abstract This study investigated 104 English language learners’ longitudinal development of syntactic complexity across the A1 to C1 levels on the Common European Framework of Reference for Languages (CEFR) scale. Using a derivative corpus of the EFCAMDAT, the study analyzed changes in judiciously selected measures—mean length of clause (MLC), coordinate phrases per clause (CP/C), dependent clauses per clause (DC/C), and complex nominals per clause (CN/C)—via negative binomial regression. The study also qualitatively analyzed essays whose syntactic complexity measures were not well predicted by the regression models. Results suggested that MLC divided the CEFR levels into two bands (i.e. A1/A2/B1 and B2/C1), while DC/C and CN/C distinguished adjacent CEFR levels. CP/C exhibited non-linear growth, potentially reflecting its construct ambiguity. Learners’ unanticipated performance was characterized as the absence of dependent clauses at the B2 to C1 levels. Two case studies indicated that such unanticipated performance was potentially caused by learners’ ongoing transition from subordination to phrasal complexity, different communicative needs demanded by different tasks, and the computational tool’s limited capability to analyze learner language.