linguistlist.org>
Editor for this issue: Renee Galvis <reneelinguistlist.org>
cs.cmu.edu>ESSLLI-2002 Workshop on Recent Advances in Speech Translation Systems August 12-16, 2002 Trento, IT A workshop held as part of the 14th European Summer School in Logic, Language and Information ESSLLI-2002 Trento, Italy August 5-16, 2002 ** REVISED CALL FOR PAPERS ** ** NOTE EXTENDED SUBMISSION DEADLINE: APRIL 15, 2002 ** ORGANIZERS: Alon Lavie and Lori Levin (Carnegie Mellon University) Fabio Pianesi (ITC-irst) DESCRIPTION: Speech Translation research has made significant strides over the last decade, with several large scale research efforts (C-STAR, Verbmobil, SLT, NESPOLE! and others) significantly advancing the state-of-the-art. A wide variety of different approaches to MT has been pursued in the various research efforts, and in some cases these have been combined in multi-engine approaches. Nevertheless, current speech translation technology is still far away from broad commercial application, with a basic tradeoff between quality of translation and domain coverage. Some recent research has focused on issues of robustness and domain-portability and on enhancing the communication abilities using multi-modal interaction. The purpose of this workshop will be to present the current state-of-the-art of speech translation research and explore the current promising trends and developments. This workshop is intended to complement the ACL-02 workshop on Speech-to-Speech Translation, which will be held in conjunction with ACL-02 in Philadelphia about one month earlier. Our intention is to have deeper and more focussed presentations and discussions on several identified key topics and issues in current speech translation research. In support of this goal, the workshop program will be organized around daily theme topics, taking advantage of the ESSLLI workshop format of five daily sessions of 90 minutes each. Each daily theme will consist of one or two long (30 minute) presentations of research and/or position papers which explore the theme, followed by extensive time for discussion of the main issues related to the theme. Groups and researchers are encouraged to submit distinct papers to both workshops. Please note that the submission deadline to this workshop has been extended to April 15th, in order not to conflict with the deadline of the ACL-02 workshop. Some possible theme topics include: - Architechture and design considerations for ST systems - New approaches to ST systems and their components - Domain and language portability issues for ST systems - Improving communication robustness - Robust Speech Recognition for ST applications (i.e, dealing with noise, bandwidth and platform issues) - Integration of ST with alternative modalities for cross-lingual communication - Evaluation of Speech Translation - specific problems and approaches. - Moving from prototypes to real-world systems and applications (i.e, issues related to translation quality, user interfaces, speech translation on small devices, etc.) SUBMISSION: We invite both research and position paper submissions from all researchers in the area of speech translation and related topics. Submissions will be Electronic, in either postscript, pdf or MS word formats. Submissions should not exceed 10 (A4 or letter) pages, typeset in 10-12 point, with at least 2.5 cm / 1 inch margins. All submissions will be reviewed by an international program committee. The accepted papers will be made available in a summer school reader. A joint volume of expanded versions of a selection of papers from both the ACL-02 and ESSLLI-02 workshops is being planned. A joint editorial board will be established after the workshops to select candidate papers from those presented at the two workshops and to consider possible publishing venues. Submissions should be sent by Monday, April 15, 2002 to the following email address: alavieMail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue
let.kun.nl>CALL FOR PAPERS PIONIER-Workshop on Variation in form versus variation in meaning Location: University of Nijmegen, The Netherlands Date: 11-12 July 2002 Organizers: Marco Haverkort, Helen de Hoop, Maurits van den Noort Invited speakers: - Arto Anttila (New York) & Vivienne Fong (Singapore) - Manfred Krifka (Berlin) - Tom Roeper (Amherst) - Matthias Schlesewsky (Potsdam) At the end of the conference there will be a forum discussion. Invited forum participants: - Peter Ackema (Utrecht) - Marian Klamer (Leiden) - Wietske Vonk (Nijmegen) - Henk Zeevat (Amsterdam) Apart from one-to-one mappings between forms and meanings, languages display one-to-many mappings (ambiguity), many-to-one mappings (optionality) and weighted versions of ambiguity and optionality (that account for preferred readings and preferred expressions, respectively). The competition between different types of constraints, may lead to various kinds of outcomes, including optionality and ambiguity. Blutner (2000) and Zeevat (2000) develop an account of these phenomena within a bidirectional optimization theory. An alternative view is offered within the framework of Anttila and Fong (2000) who argue that grammars are partial rather than total orders of constraints, which can be interpreted quantitatively. One very robust generalization in language that pops up in one form or another in all linguistic disciplines and at all levels of representation, is that marked forms typically have concomitant marked meanings (e.g., Bresnan, to appear). Levinson (2000) formulates this generalization as follows: "What is said in an abnormal way indicates an abnormal situation, or marked messages indicate marked situations" (Levinson 2000: 136). The pragmatic generalization that marked forms tend to be used for marked interpretations, is accounted for by Blutner's theory of bidirectional optimization. The structures that compete in one direction of optimization are constrained by the outcomes of the other direction and vice versa. A potential problem for an approach such as Blutner's may be the occurrence of true optionality and true ambiguity in language. For example, De Hoop (2000, to appear) extensively argues that the relative word order between an adverb and a direct object is truly optional in Dutch for several classes of noun phrases, among which those introduced by a definite article (that is, there exist two forms). Within the appropriate contexts, both orders can be used to refer to an anaphoric definite that is preceded by a linguistic antecedent, and both orders can be used to refer to a non-anaphoric definite (that is, there exist two meanings). Because definites scramble more often than not and get an anaphoric interpretation more often than not, we may predict, on the basis of Blutner's approach, that the unscrambled form is the marked one which is used for the marked meaning (the non-anaphoric meaning in the case of definites), whereas the unmarked form (the scrambled order) is used for the unmarked (anaphoric) reading. Although this tendency does exist, there are many exceptions (anaphoric definites that don't scramble, as well as non-anaphoric definites that do), as observed in De Hoop (to appear). The main conclusion of the discussion above is that two different word orders do not always serve to disambiguate meaning. The generalization that a difference in forms co-occurs with a difference in meanings, based on general economy principles in communication, must be weakened. Different forms are not necessarily restricted to different interpretations. For this workshop we would like to put forward the following hypotheses: a. When there is no optionality in form (or optionality decreases), there may be ambiguity in meaning (or ambiguity increases). b. When there is no ambiguity in meaning (or ambiguity decreases), there may be optionality in form (or optionality increases). The former hypothesis is quite generally accepted, we think, while the second hypothesis (the natural counterpart of the first one) may seem quite controversial. We cordially welcome contributions from theoretical, descriptive and psycholinguistic perspectives that add new insights to the above discussion. There will be room for about 10 contributed papers. Send your abstract of 1-2 pages (preferably a PDF-file or a Word-file) by e-mail to: M.v.d.NoortMail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue