Discourse complexity measures as indices of L2 writing proficiency: Insights from propositional connection and relation

Complexity is a crucial feature of L2 writing development. While substantial measures have been well-established as indices of L2 writing proficiency, they are primarily focused on the lexical and syntactic complexity of learner texts. Discourse complexity (DC), especially concerning connections and relations between propositions, remains underexplored despite its importance in discourse organization in L2 writing. This study investigates DC measures for L2 written production in both the absolute and relative/cognitive dimensions. A Rhetorical Structure Theory (RST) dependency treebank of 216 argumentative essays was created to investigate the discourse structures in learner texts across three proficiency levels in the Common European Framework of Reference for Languages (CEFR). Our analysis suggests that more proficient L2 writers tend to produce more logically complex rhetorical relations, longer rhetorical length, and stronger propositional connectedness in content development and organization. Further, we show that overall propositional connectedness and mean discourse distance are both able to distinguish between texts at the CEFR A2, B1, and B2+ levels, and can be useful indicators of L2 writing proficiency. These findings not only provide novel insights into L2 complexity at the discourse level, but also carry pedagogical implications for the teaching of coherence and rhetorical organization in L2 writing classrooms.

© 2025 Elsevier Inc.