Extracting Terms and Terminological Collocations from the ELAN Slovene–English Parallel Corpus

Extracting terms and terminological collocations from the ELAN Slovene- English parallel corpus Špela Vintar Dept. of Translation and Interpreting Faculty of Arts University of Ljubljana spela.vintar@guest.arnes.si Abstract In many scientific, technological or political fields terminology and the production of up- to-date reference works is lagging behind, which causes problems to translators and results in inconsistent translations. Experience gained in various projects involving parallel corpora show that automatic extraction of terms and terminological collocations is an achievable goal, however methods and techniques may vary according to the language pair selected. The paper describes how a methodology for multi-word term extraction and bilingual conceptual mapping was developed for a Slovene-English corpus. We used word-to-word alignment to extract a bilingual glossary of single-word terms, and for multi-word terms two methods were tested and compared. The statistical method is broadly applicable but gives results of very limited use, while the method of syntactic patterns extracts highly useful terminological phrases, however only from a tagged corpus. A vision of further development is given and how these methods might be incorporated into existing translation tools. 1 Introduction Through the past decade Central and Eastern Europe has been undergoing a process of rapid economic, cultural and ideological transition, which invariably affects all aspects of communication, particularly translation. In Slovenia, new international contacts, foreign investments as well as preparations for full membership in the EU have caused an enormous increase in the volume of translations produced and required. The translation of EU-legislation alone is an overwhelming task for such a small translation market as over 160,000 pages are due within the next two years, and other fields such as localization and web technologies are facing similar demands for translations, mostly from or into English. In order to accomplish these tasks with maximum efficiency and quality, two aspects should be particularly stressed. The implementation of state-of-the-art translation technologies such as translation memories and terminology management systems can prove very useful both in supporting translators' work and in providing reusable resources for the future, however a thorough analysis of the text and translation environment is essential before these tools are introduced. The second aspect is related to the first one and involves terminology work in general, which tends to be even less well-organized under such circumstances. Continuous overload in translation work and massive production of translated texts inevitably means that terminologists cannot keep up with the linguistic, technological and terminological developments within the field, and terminographical reference works are compiled – if at all – with a backlog that sometimes completely annihilates their usefulness. Lack of reliable terminological resources in turn results in even greater inconsistency. A possible solution for these problems is the creation of a domain-specific parallel corpus, a kind of archive of past translations and their originals, which serves as reference as well as terminology source for further translations within the domain. A parallel corpus is however not equivalent to a terminology database, and the way leading from former to the latter – if done manually – …

Content truncated to 500 words