ESSLLI-98 Workshop on MACHINE TRANSLATION August 24 - 28, 1998 A workshop held as part of the 10th European Summer School in Logic, Language and Information (ESSLLI-98) August 17 - 28, 1998, Saarbruecken, Germany ** FIRST CALL FOR PAPERS ** ORGANIZER: Frank Van Eynde (K.U. Leuven) Web site: http://www.dcs.warwick.ac.uk/~esslli98/workshops.html BACKGROUND: The Programme of ESSLLI-98 (Saarbruecken, August 17-28, 1998) features a workshop on Machine Translation, to be held during the second week of the School. As in the case of the other ESSLLI workshops its main aim is to provide a forum for advanced Ph.D. students and other researchers to present and discuss their work. Machine Translation is, of course, a rather large topic. In a sense, there is not a single branch of computational or formal linguistics which is not directly or indirectly relevant to it. In order to arrive at a reasonably coherent programme for the workshop it is therefore proposed to focus on a limited number of topics. 1. Formal Theories of Translation There are many theories of translation (esp. in literary criticism), but few of the theories are sufficiently explicit and formalized to provide a useful frame of reference for MT. A notable exception is the one presented in M.T. Rosetta. Compositional Translation (Kluwer, 1994), recently reviewed in JoLLI. Contributions on this theme can take the form of replies to M.T. Rosetta or of presentations of an alternative formal theory of translation. 2. Computational Semantics and Machine Translation The importation of methods and analyses from formal and computational semantics has proved useful for modeling the translation of such notoriously difficult expressions as determiners, pronouns, negation and tense/aspect markers. Of potential interest for MT is also the recent attention in computational semantics for the modeling of reasoning on the basis of underspecified representations. The contextual disambiguation of word senses, for instance, could be seen as a form of reasoning with underspecified semantic representations. Presentations on this theme can take the form of new contributions in one of these areas. 3. Competence-based vs. performance-based models for MT As in other fields of natural language processing, there has been a growing interest in the use of probabilistic techniques during the last decade, leading to a shift from rule- or constraint-based models to corpus-based or example-based models. Contributions are invited which report on the use of probabilistic techniques in translation systems (notice the emphasis on MT, rather than on NLP in general). 4. Machine Translation of Spoken Language Till a few years ago, virtually all MT efforts concerned the translation of written language, but the recent rise of interest (and funding) in speech processing has changed this: most of the currently started MT efforts explicitly aim at the translation of spoken language. Contributions on this theme should report on work in this field, preferably on those aspects which are in the intersection of speech processing and MT (rather in their union). 5. Translator's tools There is a school of thought in MT which dismisses the efforts to arrive at Fully Automatic High Quality Translation (FAHQT) as misguided and wasteful. What should be aimed at instead is the development of tools for human translators, such as smart text editors, on-line access to multilingual dictionaries and term banks, automatic recognition of multi-word units, and the like. A recent survey of this work can be found in the special issue of Machine Translation on New Tools for Human Translators (Volume 12, nos. 1-2, 1997). Contributions on this topic should report on original work in this field. For all five of the topics, but especially for the last three, contributors are encouraged to include a demo, either as part of the presentation, or as an extra, at the end of the session. WORKSHOP FORMAT: There will be 5 sessions of 90 minutes, each containing three slots of 30 minutes. Some of these slots will be assigned to invited speakers, but the large majority will be assigned on the basis of submitted proposals. SUBMISSIONS: All researchers in the area, but especially Ph.D. students and young researchers, are encouraged to submit a proposal. Proposals should include 1. Name, affiliation, address, e-mail of the submitter(s) 2. An indication of which of the 5 themes will be addressed 3. Two-page abstract of a paper 4. If applicable, requirements for the demo They should be sent to Frank Van Eynde Centrum voor Computerlinguistiek - K.U.Leuven Maria-Theresiastraat 21, B - 3000 Leuven, Belgium frank.vaneynde@ccl.kuleuven.ac.be fralau@iol.it fax : +32/16/325098 REGISTRATION: Workshop contributors will be required to register for ESSLLI-98, but they will be eligible for a reduced registration fee. IMPORTANT DATES: Feb 15, 98: Deadline for submissions Apr 15, 98: Notification of acceptance May 15, 98: Deadline for final copy Aug 24, 98: Start of workshop FURTHER INFORMATION: To obtain further information about ESSLLI-98 please visit the ESSLLI-98 home page at http://www.coli.uni-sb.de/esslli