Mitchell P. Marcus
Also known as Mitch Marcus
Mitchell P. Marcus researches language parsing and the acquisition of linguistic structure. Marcus earned his MIT doctorate in 1978 and joined Penn in 1987 after working at Bell Laboratories. He created and led the Penn Treebank project through the mid-1990s. With Beatrice Santorini and Mary Ann Marcinkiewicz, he published its 1993 account. He also contributed to OntoNotes and later unsupervised morphology research. The 2018 shared model uses reliable word-form patterns to improve segmentation, with results tested in English, Turkish and Finnish.
- Works
- 145
- Citations
- 14,472
- h-index
- 27
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
1978–1987
From the MIT AI laboratory to Penn
Marcus received his PhD from MIT’s Artificial Intelligence Laboratory in 1978. He worked at AT&T Bell Laboratories before joining Penn in 1987. His research concerns statistical language processing and learning linguistic structure.
-
1993
An annotated corpus for language research
With Beatrice Santorini and Mary Ann Marcinkiewicz, Marcus published Building a Large Annotated Corpus of English: The Penn Treebank. He created and led the project through the mid-1990s. Its training corpus supported improved natural-language parsing.
-
2007
A shared semantic representation
Marcus coauthored the 2007 OntoNotes paper on a unified relational semantic representation, with Pradhan, Hovy, Palmer, Ramshaw and Weischedel. He was a project co-principal investigator.
-
2018
Learning word structure without labels
With Hongzhi Xu, Charles Yang and Lyle Ungar, Marcus studied unsupervised morphological segmentation. Their model finds statistically reliable sets of word endings, feeds them back into learning and prunes unreliable relations. The paper reports improved segmentation on English, Turkish and Finnish Morpho-Challenge data. These results apply to the tested languages and data.
Sources and credits
Photo: www.cis.upenn.edu; creator not stated. License: Not stated by source. Source: cis.upenn.edu. Cropped to a square and resized. Ask us to remove or replace this photo.