Investigating Concept Definition and Skill Modeling for Cognitive Diagnosis in Language Learning

Language proficiency diagnosis is essential to extract fine-grained information about the linguistic knowledgestates and skill mastery levels of test takers based on their performance on language tests. Differentfrom comprehensive standardized tests, many language learning apps often revolve around word-levelquestions. Therefore, knowledge concepts and linguistic skills are hard to define, and diagnosis must bewell-designed. Traditional approaches are widely applied for modeling knowledge in science or mathematics,where skills or knowledge concepts are easy to associate with each item. However, only afew works focus on defining knowledge concepts and skills using linguistic characteristics for languageknowledge proficiency diagnosis. In addressing this, we propose a framework for language proficiencydiagnosis based on neural networks. Specifically, we propose a series of methods based on our frameworkthat uses different linguistic features to define skills and knowledge concepts in the context of thelanguage learning task. Experimental results on a real-world second-language learning dataset demonstratethe effectiveness and interpretability of our framework. We also provide empirical evidence withcomprehensive experiments and analysis to prove that our knowledge concept and skill definitions arereasonable and critical to the performance of our model.