The Linguist is a languages magazine for professional linguists, translators, interpreters, language professionals, language teachers, trainers, students and academics with articles on translation, interpreting, business, government, technology
Issue link: https://thelinguist.uberflip.com/i/135206
© EUROPEAN UNION, 2013 FEATURES SUPPORT NETWORK A translator at the European Commission uses translation memory software to search for passages of text that have been translated previously a document arrives for translation, Euramis searches for any relevant segments and creates a dedicated memory for that document alone. Once the translation is completed, the new translation units are uploaded to the Euramis database, ready to be retrieved in future. This was a great leap forward, but translators could hardly be expected to create a new memory manually for every document, import data, and then export data to the central memory. So a series of macros were devised, the main ones enable translators to: • Create a memory by loading the document to be translated into Word, and then clicking 'Create TWB Memory'. The next stage is simply to begin translating. • Export translation units automatically and upload them to Euramis by clicking on 'Cleanup', once the translation has been revised and corrected. This also strips the document of all TWB codes. Word 2003 with TWB toolbar and custom tools The database is searchable through concordance, but it did not supplant other databases. The metasearch engine Quest was therefore developed to allow searching, with a single click, through a whole series of Vol/52 No/3 2013 resources, including IATE, EUR-Lex and Euramis, but also a further 24 databases from other EU institutions and countries. This idea of finding reference material through a single click was taken further with the Docfinder tool, which, like Quest, got its own toolbar. Rather than searching for text, it resolves references in EU legislation and opens the relevant document in the browser, including a bilingual split-screen view when desired. Docfinder results with bilingual split-screen These tools are being updated with DGT's switch to Windows 7 and Trados Studio, but they are mature and useful tools that translators rely on in their day-to-day work. There is one young upstart that has not yet become a habitual part of the workflow, and that is machine translation. DGT is building a machine translation system, dubbed MT@EC. It was initially opened up to DGT translators, but is ultimately intended for use by the other Directorates-General, internal use and some public-facing services as well, though not as a directly accessible MT portal. The project is being built using statistical MT technology: the open-source MOSES engine, fed with the millions of segments stored in Euramis, which cover the 23 official languages of the EU. DGT translators are taking an active part in the project. Those with an interest in machine translation got a taster of a first generation of engines from English into their languages in early 2011. These volunteers assessed whether the translations provided by the system were useful for their work in a 'maturity check', completed in May 2011. As was to be expected, different language pairs gave different results. Translators for 10 of the languages (working from English) felt it was sufficiently useful to have machine translation integrated into the existing workflows by having it automatically included in the translation memory created using the TWB macros. Since then, and with the advent of a new generation of engines, seven more languages have joined, with the Dutch, English, Estonian, Finnish, German and Slovak still reluctant, for various reasons. Translators have been very helpful in pointing out flaws in the system, including character corruption, extraneous spaces and inconsistent spelling due to spelling reforms (Portuguese and Maltese), and a third generation of engines fixing these problems should be available to them soon. So the tools available to DGT translators run the gamut from administrative tools to reference tools to machine translation, all built into a specific workflow. We believe it makes their job easier and their work better. Hopefully the new kid on the block – machine translation – will soon become as standard and helpful a tool as the rest. * This series of articles will end with a detailed look at IATE, a database of EU terminology that is freely available to the public. JUNE/JULY The Linguist 27