Discourse dependency distance minimization in agricultural academic abstracts: Evidence from the AgriDTB Corpus

This study investigates how second language (L2) writers organize discourse in academic texts to facilitate cognitive efficiency from the perspective of Dependency Distance Minimization (DDM). While previous studies have widely focused on syntactic-level structures, less is known about how writers manage the organization of clauses and propositions at the discourse level, particularly in the context of scientific writing. The study uses a self-built corpus of agricultural research article abstracts to investigate the way L2 writers maintain text organization in their writing. The findings of the study reveal that related discourse elements tend to be placed closer than syntactic units, suggesting a tendency toward minimizing dependency distance at the discourse level. Statistical analysis further shows that the distribution of discourse dependency distance follows Zipf's law. These results also emphasize the importance of discourse-level organization in L2 academic writing and provide useful insights for teaching English for Specific Purposes (ESP), particularly in science and technology contexts. Moreover, the study highlights that discourse dependency patterns show systematic tendencies in discourse parsing.