Welcome to the home page of the Computation, Language, Intelligence, and Grounding (CompLING) Laboratory at the University of Waterloo! We are a group of researchers working towards understanding human language with the help of computational methods. We are part of the natural language processing group in the David. R. Cheriton School of Computer Science at the University of Waterloo.

Research Foci

• Grounded language learning: How do humans acquire language so effectively and efficiently? One promising conjecture is that we learn language from both language itself and rich grounding signals (such as vision, sound, and touch) around us. Along this line, we use computational models to test hypotheses from developmental psychology and cognitive science and bring new insights to the fields.
• Cross-lingual language processing: Cross-lingual grounding is another important aspect of language grounding. We are interested in investigating how multiple languages interact in humans and machines, how the mechanisms differ, and how we can use these insights to build better cross-lingual language processing systems.
• Understanding neural language processing models: We analyze deep neural networks, particularly those trained in grounded settings, including multimodal and cross-lingual models. We are interested in answering questions such as: What can/do these models learn? Can we manipulate part of the model to achieve some desirable effect? What training and inference objectives can make the model more robust and generalizable?
See more discussion on what grounding means to us here on pages 15--18.

Funding Support

We are grateful for the extramural support we have received for our work from the Natural Sciences and Engineering Research Council of Canada (NSERC), the Vector Institute, and Cohere. If you would like to help support our work, please let us know!