The field of natural language processing (NLP) is growing rapidly with emergence of new state-of-the-art methods almost every year. As educators in the field, we are faced to make many decisions as to what to teach and how to do it. The fast-paced nature of NLP brings unique challenges for curriculum design, and the immense growth of the field has lead to not just core NLP courses, but also classes and seminars focused on subareas such as natural language understanding and social applications. What fundamental concepts should we be teaching future NLP researchers and practitioners? What kinds of paths through courses should we provide for students traversing an NLP-related degree? Our field is not limited to just computer science and linguistics. Students from social sciences, humanities, medicine, statistics and other areas are making NLP truly interdisciplinary. However, balancing the different goals and needs of students' from other disciplines also brings up the question of how we should tailor our courses for these wider audiences. In some cases, we must also scale our teaching for larger class sizes while preserving the accessibility of complex material. Finally, how do we teach responsible and ethical use of NLP? Given the availability of large-scale language data and accessible NLP tools, the irresponsible use of NLP is a growing problem. How do we teach reproducible and responsible practices? How do we provide students a solid understanding of NLP algorithms' impact on people and society so that they are ready to appropriately handle ethical and social issues such as misrepresentation, bias, and privacy?
We propose to hold a discussion-heavy workshop with participatory components on the topic of teachingNLP which aims to
- encourage a dialogue and collaboration between NLP educators and researchers inacademia and industry to address the questions above,
- provide a forum to share teaching and learningmethodologies and tools in NLP education, and
- create useful and accessible resources for teaching andlearning NLP.