During its 58th annual meeting, the Association for Computational Linguistics (ACL) awarded its 2020 Lifetime Achievement Award to Professor Bonnie Webber.
Prof. Webber is an emeritus professor of intelligent systems in the Institute for Language, Cognition and Computation (ILCC) in the School of Informatics, University of Edinburgh. She received her PhD from Harvard in 1978 and taught at the University of Pennsylvania for 20 years before joining the University of Edinburgh in 1998. She retired in 2016, though she continues to supervise student research.
Prof. Webber is a Fellow of the Association for Computational Linguistics (ACL) (for significant contributions to discourse structure and discourse-based interpretation), the Association for Advancement of Artificial Intelligence (AAAI), and the Royal Society of Edinburgh (RSE).
Prof. Webber was introduced to Natural Language Processing by Bill Woods, who both taught at Harvard and led the Natural Language Processing group at Bolt Beranek & Newman (Cambridge MA), where he pioneered both the Lunar Sciences Natural Language System (LUNAR) for NASA, and BBN's first speech understanding system.
It was while working on Question-Answering dialogues for LUNAR that Webber began to address the kinds of anaphoric expressions that were rife in such dialogues -- and which are critical to understanding what question is being asked. Instead of treating this as a reference resolution problem, Webber asked the reverse question: What entities does a text make available for subsequent anaphoric reference? This became the subject of her 1978 PhD dissertation, "A Formal Approach to Discourse Anaphora". Discourse is now an important area within our field, and this early useful work established Prof. Webber as one of its founders.
She is perhaps best known for developing the Penn Discourse TreeBank (PDTB), together with Professor Aravind Joshi and other collaborators at the University of Pennsylvania. (Professor Joshi was himself awarded the first ACL Lifetime Achievement Award back in 2002.) The PDTB was inspired by work done by Joshi and his colleagues on Lexicalized Tree-Adjoining Grammar (L-TAG). The PDTB takes a similar lexicalized approach to discourse relations, grounding them in either explicit words or phrases, or simply in adjacency, just as in L-TAG. The PDTB remains the largest corpus manually annotated for discourse phenomena. Its lexical focus has meant that judgments can be traced to specific lexical material, enabling annotation to be more consistent. Not least because it avoids being tied to a particular discourse theory, it is a major resource in computational linguistics; e.g., it has made empirical work on discourse parsing possible and there is now an important discourse parsing subcommunity that organizes shared tasks and workshops. The PDTB was also the inspiration for similar efforts in many other languages: Arabic, Chinese, Czech, Hindi, Italian, Polish, Portuguese and Turkish.
Hinrich Schütze, ACL President