Roger Levy

Timeline

1. 1996–2005

   Mathematics and Stanford linguistics

   Born in Philadelphia and raised in Tucson, he earns a BS in mathematics at the University of Arizona in 1996. Graduate study takes him to Stanford University, where he completes his linguistics PhD under Christopher Manning in 2005.

2. 2005–2006

   Edinburgh and UC San Diego

   After completing his doctorate in 2005, he spends a year as a postdoctoral fellow at the University of Edinburgh’s School of Informatics. He joins the Department of Linguistics at UC San Diego in 2006.

3. 2008

   A noisy-channel comprehension model

   In 2008, he publishes A Noisy-Channel Model of Human Sentence Comprehension under Uncertain Input at the Conference on Empirical Methods in Natural Language Processing. The work models human sentence understanding with uncertainty in the input.

4. 2010–2012

   CAREER and Sloan awards

   He receives an NSF Faculty Early Career Development Award and a UC San Diego Hellman Fellowship in 2010. In 2012, he receives an Alfred P. Sloan Research Fellowship while directing UC San Diego’s Computational Psycholinguistics Lab.

5. 2016

   Joining MIT brain sciences

   In 2016, he joins MIT’s Department of Brain and Cognitive Sciences. At MIT, he directs the Computational Psycholinguistics Laboratory and studies language processing and acquisition at the intersection of AI, linguistics and psychology.

6. 2018

   Testing grammar in neural networks

   With Ethan Wilcox, Takashi Morita and Richard Futrell, he coauthors What do RNN Language Models Learn about Filler–Gap Dependencies? in 2018. Controlled sentence tests show that recurrent neural networks can represent long-distance grammatical relationships and learn some constraints on where these relationships can occur.

7. 2023

   Measuring language model knowledge

   In 2023, he and Jennifer Hu compare prompting with direct probability measurements as methods for testing large language models’ linguistic knowledge. Their EMNLP paper finds that the models’ explicit judgments are less reliable than measurements taken directly from their internal representations.

8. 2024

   A Guggenheim Fellowship

   He receives a Guggenheim Fellowship from the John Simon Guggenheim Memorial Foundation in 2024. His research combines computational modeling of large datasets with experiments on human language processing to help develop algorithms that process human language.

9. 2024–2025

   Cognitive Science Society president

   In 2024–2025, he serves as president of the Cognitive Science Society.

10. Since 2025

    Chair of the MIT faculty

    In 2025, he begins a term as chair of the MIT faculty that runs through 2027. As of October 2026, he is a professor in MIT’s Department of Brain and Cognitive Sciences and directs its Computational Psycholinguistics Laboratory, combining computational models, experiments and language datasets to study how people understand and acquire language.