Development and Validation of Academic Vocabulary Complexity Indicators for Meaning-Centered Approaches in Automated Writing Assessment.

This study designs meaning-based lexical complexity indicators focused on the usage patterns of academic vocabulary in argumentative texts written by native Korean speakers and analyzes the relationship between these indicators and writing assessment elements. The research utilized 1,231 argumentative texts on the same topic from the ‘2023 National Writing Ability Diagnostic System Development Corpus’ by the National Institute of Korean Language. Three indicators were designed and applied: Academic Vocabulary Diversity (SDiv), Level-Weighted Index (SWeight), and Usage Rate (SRate). Quantitative analysis revealed that the level-weighted index and usage rate showed significant positive correlations with assessment scores, with SWeight and SRate demonstrating positive effects on content scores in regression analysis. Qualitative analysis of functional type usage patterns revealed distinct differences between high-scoring and low-scoring groups in type distribution, connection structure, and development patterns. The high-scoring group utilized various types (analysis, justification, causality) in a balanced manner with complex connection structures, while the low-scoring group exhibited linear thinking patterns with excessive dependence on conceptualization. These findings suggest that academic vocabulary-based complexity indicators can contribute to a paradigm shift from form-centered to meaning-centered writing assessment.