This project supports students’ writing of science explanations. One main goal is to develop novel, automated natural language processing (NLP) technology to provide middle school students and teachers with real-time feedback on their science essays, to help them reflect on the quality of their explanations, and to foster their ability to write scientific explanations grounded in scientific evidence. A science unit on roller coasters is presented to students in a web-based environment that includes a roller coaster simulation to conduct experiments. Students respond in writing to open-ended questions about the physics of roller coasters, and write two short essays. Essay feedback is provided in the form of a checklist of main ideas. We also apply NLP methods to assess students’ answers to science questions to facilitate researchers' analysis of students' overall performance in writing about science.