Author:

Keywords:

context-sensitive terminology management, translation environment tools

Abstract:

Purpose This paper wants to propose a system architecture for a context-sensitive terminology management system (TMS) in the domain of legal LSP. The TMS is expected to meet the needs of the translation services in the Belgian Federal Public Service of Justice in that our TMS will allow to search, (re)use and edit each term in its customized archetypical phraseological context. In a second part, we propose to link this TMS to EN 15038, the EU standard on Translation Services, which has been developed to streamline translation projects aiming to deliver more efficient multilingual LSP documentation. Method In a first step, automatic knowledge acquisition algorithms will be developed to mine context-sensitive terminological information from the online corpora of legal translations. In a second step, a feed-back loop between interface developers and expert users will be set up to optimize user access to the contextualized terminological knowledge base during the actual translation process. To meet these requirements, there will be an intense co-operation between terminologists, the developers of the knowledge extraction algorithms, database and software engineers, and expert users. Results The TMS is expected to display an interface offering the look-and-feel of a standard desktop CAT-tool. The actual application will run remotely on a server in a cloud-based architecture which will be available via an online interface. As a collaborative working environment, the tool will allow extensive and real-time interactive user management. Conclusion This paper analyses the challenges legal translators are facing in the Belgian Federal Public Service of Justice. At present, translation-oriented authoring and translation proper of legal documentation are being carried out on an ad-hoc basis. An R&D project has been set up to design a TMS which is expected to deliver correct and consistent legal terminology and collocations in both source and target languages.