Janet Pierrehumbert

Janet Pierrehumbert is a linguist and language modeller whose work uses computational methods to study how people learn words and how language changes across communities. She earned a BA in linguistics at Harvard in 1975 and a PhD in linguistics at MIT in 1980, developed a model of English intonation for synthetic speech, and worked in Linguistics and Artificial Intelligence Research at AT&T Bell Laboratories from 1982 to 1989. She joined Northwestern’s linguistics faculty in 1989, became professor in 1993, and moved to Oxford as Professor of Language Modelling in 2015. Her work includes 2021 embeddings that model word meaning across time and social space, a 2025 study of analogy in large language models, and a 2026 analysis of climate discourse using large language models. As of October 2026, she is Professor of Language Modelling in Oxford’s e-Research Centre and Senior Research Fellow at Trinity College; she was elected to the US National Academy of Sciences in 2019 and received the ISCA Medal for contributions to phonology in 2020.

Now at: Trinity College, University of Oxford
Badges: Bridge
Loyal

Timeline

1975–1980

Harvard and MIT degrees

She earns a BA in linguistics at Harvard in 1975 and completes a linguistics PhD at MIT in 1980.
1980–1989

English intonation and Bell Labs

Her 1980 PhD dissertation develops a model of English intonation that generates pitch contours in synthetic speech. From 1982 to 1989, she works as a Member of Technical Staff in Linguistics and Artificial Intelligence Research at AT&T Bell Laboratories.
1989–2015

Northwestern and Oxford

She joins Northwestern’s linguistics faculty in 1989, becomes professor in 1993, and serves as department chair and director of graduate studies before moving to Oxford in May 2015.
2019

National Academy election

She is elected a member of the US National Academy of Sciences in 2019.
2020

ISCA Medal for phonology

She receives the 2020 ISCA Medal for contributions to phonology.
2021

Dynamic word embeddings

With Valentin Hofmann and Hinrich Schütze, she introduces dynamic contextualized word embeddings in 2021. The model represents words as a function of linguistic and extralinguistic context, and models time and social space jointly.
2025–2026

Word analogy and climate discourse

She coauthors a 2025 study comparing human judgments with GPT-J on English adjective-to-noun formation. It finds that large language models generalize language patterns by analogy from stored examples. In 2026, she coauthors a computational analysis of climate discourse in financial news using large language models.
As of October 2026

Oxford professor and Trinity fellow

As of October 2026, she is Professor of Language Modelling in Oxford’s e-Research Centre and Department of Engineering Science, and Senior Research Fellow at Trinity College. Her research uses machine learning and computational linguistics to study word formation and language learning, and her group applies natural language processing methods to language patterns in human communities.