An open-source shallow-transfer machine translation engine for the Romance languages of Spain

EAMT 2005 Conference Proceedings 79 An Open-Source Shallow-Transfer Machine Translation Engine for the Romance Languages of Spain Antonio M. Corbí-Bellot1, Mikel L. Forcada1, Sergio Ortiz-Rojas1, Juan Antonio Pérez- Ortiz1, Gema Ramírez-Sánchez1, Felipe Sánchez-Martínez1, Iñaki Alegria2, Aingeru Mayor2, Kepa Sarasola2 1Transducens group, Departament de Llenguatges i Sistemes Informàtics Universitat d'Alacant, E-03071 Alacant 2IXA Taldea, Informatika Fakultatea, Euskal Herriko Unibertsitatea, E-20071 Donostia acorbi@dlsi.ua.es, mlf@ua.es, sortiz@dlsi.ua.es, japerez@dlsi.ua.es, ge- ma@internostrum.com, fsanchez@dlsi.ua.es, acpalloi@si.ehu.es, jibmamaa@si.ehu.es, ksara- sola@si.ehu.es Abstract. We present the current status of development of an open-source shallow-transfer machine translation engine for the Romance languages of Spain (the main ones being Span- ish, Catalan and Galician) as part of a larger government-funded project which includes non-Romance languages such as Basque and involving both universities and linguistic technology companies. The machine translation architecture uses finite-state transducers for lexical processing, hidden Markov models for part-of-speech tagging, and finite-state based chunking for structural transfer, and is largely based upon that of systems already devel- oped by the Transducens group at the Universitat d'Alacant, such as interNOSTRUM (Spanish—Catalan) and Traductor Universia (Spanish—Portuguese). The possible scope of the project, however, is wider, since it will be possible to use the resulting machine transla- tion system with new pairs of languages; to that end, the project also aims at proposing standard formats to encode the linguistic data needed. This paper briefly describes the ma- chine translation engine, the formats it uses for linguistic data, and the compilers that con- vert these data into an efficient format used by the engine. 1. Introduction This paper presents the current status of deve- lopment and the main motivations of an open- source shallow-transfer machine translation (MT) engine for the Romance languages of Spain (the main ones being Spanish (es), Catalan (ca) and Galician1 (gl)) as part of a larger government- funded project which will also include MT en- gines for non-Romance languages such as 1Most scholars consider Galician and Portuguese (pt) the same language; however, the official or- thography of Galician is very different from the ones used for European and Brazilian Portuguese. There- fore, while grammatical resources will be rather re- usable, lexical resources will not easily be. Basque (eu) and involving four universities and three linguistic technology enterprises.2 The shallow-transfer architecture will also be suit- able for other pairs of closely related languages which are not Romance, for example, Czech— Slovak, Danish—Swedish, etc. The multilingual nature of Spain is recog- nized, to a varying extent, in laws and regula- tions corresponding to the various levels of go- 2TALP (Universitat Politècnica de Catalunya), SLI (Universidade de Vigo), Transducens (Universitat d’Ala- cant), IXA (Euskal Herriko Unibertsitatea), ima- xin|software (Santiago de Compostela), Elhuyar Fundazioa (Usurbil), and Eleka Ingeniaritza Linguis- tikoa (Usurbil, coordinator). Corbí-Bellot et al. 80 EAMT 2005 Conference Proceedings vernment (the Constitution of Spain and the Statutes of Autonomy granted to Aragon, the Balearic Islands, Catalonia and Valencia (ca), Galicia (gl), and Navarre and the Basque Count- ry (eu)). On the one hand, demand by many citizens in these territories make private com- panies …

Content truncated to 500 words