A Pilot Study 000901 - 010531 VINNOVA dnr 2001/02984

Substantial achievements regarding machine translation have been made on the international arena, commercially as well as from a research point of view. In spite of this, there is still no high quality machine translation system for Swedish. Thus in July 1999 the Swedish Ministry of Industry, Employment and Communications commissioned an investigation of the technical and economical preconditions for developing a Swedish-English basic module for machine translation. The investigation was primarily directed towards the translation of standardised industrial and governmental documents (Regeringsuppdrag N1999/7189/ITFOU). The task was carried out by NUTEK. As a result, a report was presented with an overview of the current situation and suggestions for further commitments (NUTEK/Näringsdepartementet 1999). The report is available at http://europa.eu.int/comm/translation/en/eyl/nutek.pdf). It is in Swedish. One line of action that is proposed in the NUTEK study is the investigation and further development of available research prototypes. This line is followed in the MATS project.

MATS aims at the investigation of MULTRA as the basis for a full-fledged, commercially valid machine translation system for Swedish. MULTRA is short for Multilingual Support for Translation and Writing (Sågvall Hein 1993, 1994, 1997). It is a research prototype primarily intended for high-quality translation of Swedish industrial documents within limited domains. Target languages, so far, are English and German. The development of Multra has been carried out at the Department of Linguistics at Uppsala University with support from HSFR (Swedish Research Council for the Humanities and Social Sciences) and NUTEK.

In the MATS project the process from prototype to product will be investigated and specified. Ways of coping with the reverse translation process, i.e. from English or German to Swedish will also be considered. In addition to the machine translation core system, the MULTRA translation engine, the resulting system will provide access to a translation memory. Support technologies for language checking (Almqvist & Sågvall Hein, 2000, forthcoming) and for the generation of translation dictionaries based on translated text (Tiedemann 2000a, 2000b) will also be part of the final system. The project also includes a market study concerning machine translation.

Project in Progress

• The MATS corpus
• The MATS link database
• Frequency counts:
  • Swedish [training] [evaluation] [complete] [checked]
  • English [training] [evaluation] [complete]

References

Almqvist, I. & Sågvall Hein, A., 2000.

A Language Checker of Controlled Language and its Integration in a Documentation and Translation Workflow. Translating and the Computer 22. Papers form the Aslib conference held on 16 & 17 November 2000.

NUTEK/Näringsdepartementet. 1999.

Om maskinöversättning. NUTEKs slutrapport av regeringsuppdrag N1999/7189/ITFOU 1999-09-22.

Sågvall Hein, A., 1997.

Language Control and Machine Translation. In: Proceedings of the 7th International Conference on Theoretical and Methodological Issues in Machine Translation. July 23-25, 1997. St. John's College, Santa Fe, New Mexico.

Sågvall Hein, A., 1994.

Preferences and Linguistic Choices in the Multra Machine Translation System. In: Eklund, R. (ed.), NODALIDA '93 Proceedings of '9:e Nordiska Datalingvistikdagarna', Stockholm 3-5 June 1993. Stockholm 1994.

Sågvall Hein, A., 1993.

On the Translation of Nominal Expressions in a Multilingual Unification-Based Setting. In: Hajicova, E. (ed.), Functional Approaches to Language Description. Proc. of a Conference in Prague, November 24-27 1992. Prague 1993.

Tiedemann, J., 2000a.

Extracting Phrasal Terms using Bitext. In Proceedings of the Workshop on Terminology Resources and Computation, held in conjunction with LREC 2000, Athens/Greece, May 2000.
[pdf, 167 kB] [ps, 146 kB] [gzipped ps, 60 kB]

Tiedemann, J., 2000b.

Word Alignment - Step by Step In Proceedings of the 12th Nordic Conference on Computational Linguisics, University of Trondheim/Norway 1999. .
[pdf, 442 kB] [ps, 683 kB] [gzipped ps, 208 kB]