Senior Research Fellow

Janet Pierrehumbert

  • I am Professor of Language Modelling in the Oxford e-Research Centre and the Department of Engineering Science.

  • I have an interdisciplinary background in linguistics, mathematics, electrical engineering and computer science.

  • I am currently working on the relationship between the dynamics of language and the structure of linguistic systems.

  • I studied at Harvard and MIT and then worked in the Linguistics and AI Research Department of AT&T Bell Laboratories and the Linguistics faculty at Northwestern University before coming to Oxford in 2015.

Image
Janet Pierrehumbert

Research

My research focuses on the relationship between the dynamics of language — in acquisition, processing, or historical change — and the structure of linguistic systems. I am interested in creating robust and beneficial language technology, and also in formalizing and testing theories in the language sciences. These goals go hand-in-hand. I use classical algorithms, deep learning models, and experimental data to address them.

I am a Fellow of the American Academy of Arts and Sciences, the Linguistic Society of America, and the Cognitive Science Society. I was elected as a member of the National Academy of Sciences in 2019,  and of the Academia Europaea in 2024. 

Selected Publications

Lin, F. La Malfa, E., Hofmann, V.,  Yang, E. M, Cohn, A., and Pierrehumbert, J.B. Graph-enhanced Large Language Models in Asynchronous Plan Reasoning, International Congress on Machine Learning (ICML) 2024. (2024).

Hofmann, V., Pierrehumbert, J.B. and Schuetze, H., ‘Dynamic Contextualized Word Embeddings’,  Proceedings of the Association for Computational Linguistics 2021. (2021)

Hofmann, V., Pierrehumbert, J.B. and Schuetze, H., ‘Superbizarre is not Superb: Derivational Morphology Improves BERT’s Understandings of Complex Words’  Proceedings of the Association for Computational Linguistics 2021. (2021)

Röttger, P,  Vidgen, B, Nguyen, D, Waseem, Z, Margetts, H, and Pierrehumbert, J.B. ‘Hatecheck: Functional Tests for hate speech detection models’, Proceedings of the Association for Computational Linguistics 2021. (2021)

Péter Rácz, Clay Beckner, Jennifer B Hay, Janet B Pierrehumbert. ‘Morphological Convergence as On-Line Lexical Analogy’. Language 96(4), 735-770. (2020). 

Todd, S., Pierrehumbert, J.B., and Hay, J.B., ‘Word frequency effects in sound change as a consequence of perceptual asymmetries: An exemplar-based model’, Cognition 185 (2019), 1-20.

Needle, J. and Pierrehumbert, J.B., ‘Gendered Associations of English Morphology’, in Harrington, Pouplier, and Reinisch (eds), special issue on Abstraction, Diversity and Speech Dynamics (2018), Laboratory Phonology

Beckner, C., Pierrehumbert, J.B. and Hay, J.B., ‘The emergence of linguistic structure in an on-line iterated learning task’, Journal of Language Evolution (2017)

Pierrehumbert, J.B., ‘Phonological Representation: Beyond abstract versus episodic’, Annual Review of Linguistics (2016) 2, 33-52 

Hay, J.B., Pierrehumbert, J.B., Walker, A. J. and LaShell, P., ‘Tracking word frequency effects through 130 years of sound change’, Cognition (2015) 139, 83-91

Daland, R. and J. B. Pierrehumbert, ‘Learning diphone-based segmentation’Cognitive Science (2011) 35(1), 119-155

E.G. Altmann, J.B. Pierrehumbert, and A.E. Motter, ‘Beyond word frequency: Bursts, lulls, and scaling in the temporal distributions of words’, PLoS One (2009) 4(11), e7678

Pierrehumbert, J.B., ‘Word-specific phonetics’Laboratory Phonology VII (2002) Mouton de Gruyter, Berlin, 101-139

Pierrehumbert, J.B. and J. Hirschberg, ‘The Meaning of Intonational contours in the Interpretation of Discourse’, in P. Cohen, J. Morgan, and M. Pollack, (eds), Intentions in Communication (MIT Press, Cambridge MA, 1990) 271-311.

Professor Pierrehumbert
janet.pierrehumbert@trinity.ox.ac.uk

How do language dynamics – in acquisition, processing or historical change – relate to the structure of linguistic systems? How do social and cognitive factors interact in shaping human languages? I explore these questions using experiments, statistical analyses of large corpora, and computational modelling, including new neural network approaches to learning and prediction.