In a proof-of-concept experiment on using natural language processing (NLP) to provide effective student feedback, 949 participants from the UK (paid $1 each) completed a learning activity on identifying poison ivy. Participants first received one of two randomly assigned explanations to induce prior knowledge. They then took a six-question multiple-choice pre-test. Afterward, they answered four similar questions requiring a short written explanation of their reasoning. Based on these written responses, participants received one of two targeted feedback paragraphs. Finally, participants completed a post-test identical to the pre-test but with randomized question order. Researchers evaluated the NLP feedback's effectiveness by measuring learning gains (post-test minus pre-test scores).