A Look inside the Pedagogy of Natural Language Processing

Natural Language Processing is truly an interdisciplinary field, combining elements of Linguistics, Computer Science, Statistics, and more. In the classroom, NLP educators choose which aspects to teach based on multiple constraints such as class length, student experience, recent advancements, program focus, and even personal interest. As a result, two NLP courses can look very different in terms of their content — despite teaching the same field. NLP courses often serve as students first introduction to the field and so the content can have a significant impact on the path these future scholars take and what is judged to be important in the future. In essence, these courses influence the shape of what’s to come in NLP!

Stanford student Lucy Li and I are conducting a series of interviews with faculty on how they teach natural language processing. We’d like this series to initiate dialogue on how NLP is taught across different institutions, and show the challenges and successes that people have experienced while planning curriculum and engaging students. In each interview, we talk with people at different stages of their career on how they have made the design decisions to construct their NLP courses, with an eye towards insight into how we as a field might find common ground. Though academia emphasizes researching and making intellectual advances in the field, we wanted to examine another important role it plays: educating the next generation of NLP practitioners and researchers.

Links to current interviews:

• Dan Jurafsky (Stanford, Linguistics)
• Yejin Choi (University of Washington, Computer Science)
• Emily M. Bender (University of Washington, Linguistics)
• Jason Eisner (Johns Hopkins University, Computer Science)