The document describes a study that used a cognitive diagnostic model to examine the grammatical knowledge of remedial Japanese university students. 548 students took a 55-item grammar test before and after remedial or non-remedial instruction. Item analysis showed 53 items were effective diagnostically, and the test was highly reliable. While the remedial group improved their score by 6 points, the non-remedial group's score dropped by 6 points. Analysis of mastery probabilities indicated the remedial group best mastered passive voice, while the non-remedial group's mastery of comparatives did not decrease significantly.