Critical NLP

This is a special interest group led by me and a handful of graduate students and postdocs at the University of Toronto. The idea is to shift NLP beyond the dominant heuristic practice by grounding it in rigorous theoretical and interdisciplinary discourse. Our aim is to foster dialogue shaped by diverse fields, encouraging critical reflection, shared learning, and the development of more socially grounded perspectives and methods in NLP. For more information, refer to our website.

Objectives

Instead of functioning only as a reading group cycling through papers periodically, this group will host a range of focused activities on defending substantive claims, interrogating taken-for-granted practices, and exploring conceptual resources from beyond NLP. These sessions aim to expand our epistemic boundaries and develop more theoretically grounded ways of doing NLP.

Modes of Discussion

Currently, we are supporting three modes of engagement:

  • Bring a Positive Claim: Participants come prepared with a claim drawn from NLP research or practice that they find meaningful and are willing to defend. Each claim is supported by 2-4 papers and, optionally, a set of known critiques the presenter has chosen to engage rather than avoid. The group's role is to interrogate the claim, assumptions, and boundaries of its validity.
  • Practices on Trial: Participants identify a method, metric, or framing in NLP that tends to go unquestioned, and treated as standard, neutral, or simply the way things are done. The discussion moves beyond asking whether the practice works, and instead examines it as a social and epistemic phenomenon: why it emerged, what assumptions it encodes, whose interests it serves, and what it makes harder to see. The goal is interpretive rather than evaluative.
  • Epistemic Shift: Participants bring a concept, method, or framework from outside NLP and make the case for its relevance. The discussion unpacks what the idea actually means in its home discipline, traces where it may already be quietly shaping NLP without being named, and explores what it would look like to apply it more deliberately. The goal is to import dimensions from other disciplines that could reorient how NLP approaches core problems, expose blind spots, reframe questions, or suggest novel methodologies.

Participation

We currently meet biweekly or monthly at the UofT DGP Lab or online. All UofT students and affiliates are welcome to join. If you are interested, feel free to reach out by email. For those outside UofT, stay tuned: we are working on organizing open events that bring together experts from diverse disciplines to discuss shared themes across fields, including contextual reasoning, cultural sensitivity, and related topics.