Automated Essay Scoring and Language Certification: Assessing Generalizability, Agreement and Validity for French
Researchers have developed an enhanced argument-based validation (ABV) framework for automated essay scoring (AES) systems, focusing on French language tests. This improved framework includes fairness analysis, linguistic feature correlations, prediction error evaluation, and model-human rater agreement. The study applied this framework to compare eight different model architectures using a large corpus of French essays, aiming to provide a more comprehensive understanding of AES model capabilities and limitations. AI
IMPACT Provides a more robust evaluation methodology for AI systems used in high-stakes language assessments.