linguistlist.org>
Editor for this issue: Renee Galvis <reneelinguistlist.org>
microsoft.com>--- CALL FOR PARTICIPATION --- --- ONLINE REGISTRATION NOW AVAILABLE! --- The Association for Machine Translation in the Americas AMTA-2002 Conference Location: Tiburon, California Dates: October 8-12, 2002 The Association for Machine Translation in the Americas (AMTA) is pleased to announce its fifth biennial conference, planned for October 8-12, 2002, in Tiburon (near San Francisco), California. Online registration is now available on the conference web site: http://www.amtaweb.org/AMTA2002/ Register at a discounted rate until August 11, 2002! A preliminary program, providing the schedule for tutorials, workshops, exhibits, accepted papers, panels, and invited speakers for the conference, is also now posted on the conference web site. We look forward to seeing you in Tiburon! CONFERENCE THEME: From Research to Real Users Ever since the showdown between Empiricists and Rationalists a decade ago at TMI-92, MT researchers have hotly pursued promising paradigms for MT, including data-driven approaches (e.g., statistical, example-based) and hybrids that integrate these with more traditional rule-based components. During the same period, commercial MT systems with standard transfer architectures have evolved along a parallel and almost unrelated track, increasing their coverage (primarily through manual update of their lexicons, we assume) and achieving much broader acceptance and usage, principally through the medium of the Internet. Web page translators have become commonplace; a number of online translation services have appeared, including in their offerings both raw and post-edited MT; and large corporations have been turning increasingly to MT to address the exigencies of global communication. Still, the output of the transfer-based systems employed in this expansion represents but a small drop in the ever-growing translation marketplace bucket. Now, 10 years later, we wonder if this mounting variety of MT users is any better off, and if the promise of the research technologies is being realized to any measurable degree. In this regard, we pose the following questions: Why aren't any current commercially available MT systems primarily data-driven? Do any commercially available systems integrate (or plan to integrate) data-driven components? Do data-driven systems have significant performance or quality issues? Can such systems really provide better quality to users, or is their main advantage one of fast, facilitated customization? If any new MT technology could provide such benefits (somewhat higher quality, or facilitated customization), would that be the key to more widespread use of MT, or are there yet other more relevant unresolved issues, such as system integration? If better quality, customization, or system integration aren't the answer, then what is it that users really need from MT in order for it to be more useful to them? INVITED SPEAKERS We are pleased to announce that invited speakers for the conference will include Yorick Wilks and Ken Church, both notable participants at TMI-92, and Jaap van der Meer, former CEO of ALPNET. We anticipate that the speakers will provide a sharp and stimulating focus on the theme of the conference. CONFERENCE ORGANIZERS Elliott Macklovitch, General Chair Stephen D. Richardson, Program Chair Violetta Cavalli-Sforza, Local Arrangements Chair Bob Frederking, Workshops and Tutorials Laurie Gerber, Exhibits Coordinator PROGRAM COMMITTEE Arendse Bernth (IBM Research) Christian Boitet (GETA, CLIPS, IMAG) Ralf Brown (LTI, CMU) Robert Cain (Foreign Broadcast Information Service) Michael Carl (RALI) Bill Dolan (Microsoft Research) Laurie Gerber (Language Technology Broker) Stephen Helmreich (CRL, NMSU) Eduard Hovy (ISI, USC) Pierre Isabelle (XRCE) Christine Kamprath (Caterpillar) Elliott Macklovitch (RALI) Bente Maegaard (CST) Michael McCord (IBM Research) Robert C. Moore (Microsoft Research) Hermann Ney (RWTH Aachen) Sergei Nirenburg (CRL, NMSU) Franz Och (RWTH Aachen) Joseph Pentheroudakis (Microsoft Research) Jessie Pinkham (Microsoft Research) Fred Popowich (Gavagai Technology Inc.) Florence Reeder (MITRE) Harold Somers (UMIST) Keh-Yih Su (Behavior Design Corp.) Eiichiro Sumita (ATR) Hans Uszkoreit (DFKI) Lucy Vanderwende (Microsoft Research) Hideo Watanabe (TRL, IBM) Andy Way (Dublin City Univ.) Eric Wehrli (Univ. of Geneva) John White (Northrop Grumman IT) Jin Yang (SYSTRAN) Ming Zhou (Microsoft Research)Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue
cis.uni-muenchen.de>Second Call for Papers WORKSHOP "ONTOLOGICAL KNOWLEDGE AND LINGUISTIC CODING" Munich, February 26-28, 2003 at the 25th annual meeting of the German Linguistics Society (Deutsche Gesellschaft fuer Sprachwissenschaft) - -------------------------------------------------- The relation between nonlinguistic knowledge and linguistic competence is still a most controversially discussed issue. This workshop will focus on a less often addressed aspect of this domain: Given that every notion or concept of the human cognitive system that is amenable to linguistic coding can be assigned a position in the ontological hierarchies we have analytical knowledge of, what is the relation between this systematic position and the structure of the linguistic signs that encode it? Whereas phonetic realisations are largely arbitrary, this does obviously not hold for semiotic complexity. Cutting up the sign inventory of a language according to an `expense' or `weight' hierarchy (roughly grammatical morpheme < lexical root < derived stem < composed stem < phrase), it seems that the encoded concepts tend to build a corresponding hierarchy of basicness or elementarity: Grammatical morphemes preferably encode very elementary categories, whereas concepts which require at least a phrase to be encoded are in general highly derived. The factual conditions are, of course, highly complex, one reason being the combinatorial possibilities of the encoding means. Accordingly, the governing principles are as yet known only in rough outline. Existing findings need to be worked out, cf. Giv�n's (1990) `Less predictable information will be given more coding material', the preference for root-encoding of basic-level categories (Rosch 1978), or what could be called Talmy's (2000) law, which states that roots of motion verbs tend to co-encode, alongside with the superconcept of motion itself, exactly one additional factor such as path, manner, figure etc. Universal preferences are equally interesting as typological differences and among the corresponding cross-linguistic investigations, cross-modal studies comparing sound and sign languages are especially revealing and welcome. A focal point will be on eventity ontologies, in particular those for motion eventities and communicative acts. As these complex themes suggest an interdisciplinary approach, not only linguists are hereby addressed, but also computer scientists, psychologists and philosophers. REFERENCES: Givon, Talmy (1990): Syntax. vol II. Amsterdam, Philadelphia: Benjamins. Talmy, Leonard (2000): Toward a Cognitive Semantics. Vol.1: Concept Structuring Systems. Cambridge, MA, London: MIT Press. Chapter 1: Lexicalization Patterns, 21-146. Rosch, Eleanor (1978): Principles of Categorization. In: Rosch, Eleanor/ Barbara B. Lloyd (eds.) Cognition and Categorization, Hillsdale, NJ: Lawrence Erlbaum, 27-48. A more detailed description of the workshop can be found at http://www.cis.uni-muenchen.de/~andrea/DGfS2003/ Invited speakers: - --------- Nicola Guarino (National Research Council Padua, Italy) Leonard Talmy (SUNY at Buffalo, USA) We invite abstracts for a 30 minute presentation (including discussion). Abstracts should not exceed one page (including references) and should be sent electronically in one of the common formats (preferably ASCII) to andreaMail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue