this paper, and for indicating some points that needed clarification. We hope we have provided a better explanation of these

This paper, accompanied by peer group commentary and author’s response, is a discussion paper concerning the state of the art in Machine Translation. The current orthodoxy is first summarized, then criticized. A number of research projects based on the standard architecture are discussed: they involve the use of Artificial Intelligence techniques, advanced linguistic theories, and sublanguage. Alternative approaches discussed are systems which develop or update their grammars semi-automatically, dialogue MT, and corpus-based MT including example-based and statistical approaches.

Ariane-G5 is an integrated environment initially designed to facilitate the development of multilingual MT systems for revisors (MT-R), where output quality is obtained by using the heuristic programming facilities of its 5 rule-based languages to specialize the lingware components to the sublanguage at hand. It can support many MT architectures and linguistic methodologies and accepts whole paragraphs or pages as units of translation rather than separate sentences. For MT-R, B.Vauquois' mutltilevel transfer approach has given excellent results on a large number of MT-R mockups and prototypes, as well as two large scale operational systems. As both the computer tools and the linguistic methodology are not embodiments of a particular theory, they are quite easy to adapt to new problems. In the last few years, they have actually been revised and further developed in the framework of new research on high quality MT for monolingual authors (MT-A), relying on a disambiguation dialogue with the author (DBMT), following an all-paths analysis. The evolution of Ariane-G5/LIDIA and associated linguistic methodologies is now motivated by two projects of different aims and requirements. The UNL project of personal multilingual high-quality communication over the Internet requires the construction of a large lexical database from which coherent dictionaries for MT and for interactive disambiguation will be generated. For the C-STAR Speech Translation (ST) project, it becomes necessary to start from phonetic lattices output by a speech recognizer, and to transmit some linguistically useful memory from one dialogue turn to the next. Keywords Machine Translation for Revisors, MT-R, Dialogue-Based Machine Translation, DBMT, Speech Translation, Machine Interpretation, Ariane-G5 MT shel...

This paper evaluates the di#erent tasks carried out in the translation of pronominal anaphora in a machine translation (MT) system. The MT interlingua approach named AGIR (Anaphora Generation with an Interlingua Representation) improves upon other proposals presented to date because it is able to translate intersentential anaphors, detect co-reference chains, and translate Spanish zero pronouns into English---issues hardly considered by other systems. The paper presents the resolution and evaluation of these anaphora problems in AGIR with the use of di#erent kinds of knowledge (lexical, morphological, syntactic, and semantic). The translation of English and Spanish anaphoric third-person personal pronouns (including Spanish zero pronouns) into the target language has been evaluated on unrestricted corpora. We have obtained a precision of 80.4% and 84.8% in the translation of Spanish and English pronouns, respectively. Although we have only studied the Spanish and English languages, our approach can be easily extended to other languages such as Portuguese, Italian, or Japanese.

Computational neurolinguistic$ (CN) is an approach to computational linguistics which includes neurally-motivated constraints in the design of models of natural language processing. Furthermore, the knowledge representations in- cluded in such models must be supported with documented behaviorial evince, normal and patho- logical.

emaine dernitre, les importations de boWns d'abattage out chutd de 62% en regard de la semaine prdcddente, totulisant 334 bouvillor;; et 50 taures. In te ms of its general structure, our translation model may be viewext as being composed of three abstract relatious: (i) the soui;e analysis/synthesis relation: anasyut. s(T_S, SurfSyu_.S SenL_S) which defines a set of well-formed triples, where T. $ is a source language text, SnrfSyL.S and and Sem_S are respeclively a sm'faee syutactic su'ucturc and a semautic stmcmm for this text, both beiug source language dependent; (it) the target analysis/synthesis relation: anasynt_t(T_T, SurfSyn._T, Sem_T) which is the analogue of auasynt__s tbr the target hmguage; a,d (iii) file transfhr relation: tr(Sem_.S, Sero_T) which defiues a set of couples, where Sem__S aud Sem_T ace respectively source and target semantic structures that are cousidered to be trauslationally equMdcnt The anasynt._s and anasynt t relations are formally and computatiou