Abstract

In this thesis we center the design and development of natural language technology around humans. We are motivated from two angles, roughly summarized as: (i) who are the users of these systems, and what do they want?, and (ii) how can we use our knowledge of human language processing and acquisition? We argue that a human-centered approach to NLP is essential to help us understand model behavior and capabilities, identify where and how modeling can be improved, and make sure models are in line with users’ needs. Each of the chapters in this thesis is driven by one or more of these aspects.