Cross-Cultural Comparison of Argument Structures Among English Learners

This interdisciplinary study analyzes 6,085 learner argumentative essays across 16 language backgrounds using argument mining. Automated scoring of organization and argument strength is used to assess learners’ argument proficiency. The extraction of Argumentative Discourse Units (ADUs) (such as claims and premises) enables a comprehensive examination of argument structures at the sentence (ADUs), paragraph (ADU flows), and full-text (argumentative communication style) levels. Cross-cultural comparisons, based on language family and cultural context, reveal three major findings: (1) Learners often exhibit appropriate argument structure despite cultural differences, but reasoning is insufficient frequently. (2) High- and low-context cultures share the same top-3 ADU flows. While premises appear more frequently than claims in both groups, more notably in low-context cultures, high-context essays contain more non-argumentative units. (3) Over 87.2% of the essays reflect a direct argumentative style by placing claims at the beginning of the texts, especially in low-context cultures. Yet, only about 40% of the essays offer adequate supporting premises, with high-context cultures more often providing well-supported claims. In short, computational argumentation (argument mining along with organization and argument strength scoring) enhances language education by reducing manual annotation and enabling large-scale analysis of argumentative texts, providing insights into how culturally-diverse learners construct arguments.