NLP-based analysis of verb semantic cohesion in L2 writing: Interaction of genre and proficiency on verb networks

Abstract

Natural language processing (NLP) is a computer-based approach to linguistic analysis that applies algorithmic techniques to textual data, enabling the extraction of complex patterns and the examination of language structures. In second language (L2) writing research, NLP techniques have facilitated increasingly sophisticated analyses of semantic features, extending beyond traditional structural measures to capture deeper mechanisms of meaning construction. While previous studies have explored various dimensions of L2 writing through NLP-based approaches, less attention has been given to how genre influences semantic cohesion, particularly in verb usage. This study investigates genre effects on verb semantic cohesion in L2 writing, using an NLP-based semantic network analysis of narrative and argumentative essays written by higher- and lower-proficiency L2 learners. The semantic network analysis identifies how verbs function as semantic anchors within constructions, illuminating cohesion patterns that reflect writers’ conceptual organization. Results indicated that higher-proficiency writers demonstrated stronger semantic cohesion and more sophisticated verb usage in argumentative essays, characterized by denser and more interconnected verb networks. In contrast, their narrative essays showed less complex networks with concrete action verbs. Lower-proficiency writers exhibited less pronounced differentiation between genres, relying on a restricted range of verbs across both types of writing. These findings suggest that advanced L2 writing proficiency is marked by the ability to adapt verb usage to genre-specific demands, aligning with usage-based approaches to language learning. These findings are discussed in light of their potential implications for L2 writing instruction and assessment, emphasizing the importance of considering semantic cohesion and genre-appropriate verb usage.