How to correctly understand natural language query and catch the useful information from it is a long-standing research question in the chatbot field. However, in many cases only a small set of words are sufficient for closed-domain tasks. In this paper, we propose Minimal Dictionary Language (MDL): a simple language architecture which allows users to express their needs through certain basic syntax. For the pilot study, we selected Enron email open dataset[2] for our experimental use and apply MDL as the language for email retrieval task. We also built a front-end search interface by Python Flask and Vue.js for test and evaluation purposes.