Exploring the Trade-off Between Model Performance and Explanation Plausibility of Text Classifiers Using Human Rationales | Synapse