Translators and EU parallel texts: Help or trap? Exploring terminological differences between EU and national legal documents

Recent technological advances have determined an increase in the use of parallel corpus data among translators and for a number of translation-related applications, such as automatic bilingual terminology extraction and statistical machine translation (e.g. Zanettin 2002, Bowker 2011, Kohen 2005). Parallel corpora – i.e. collections of texts in one language aligned to their translations in another language, mostly known by translators in the form of Translation Memories (Bowker and Barlow 2008) or of websites offering access to parallel texts (e.g. linguee.com, eur-lex.europa.eu) – are becoming all the more successful among both professional and trainee translators in that they allow to observe translation strategies and solutions adopted by previous translators, and apparently offer ready-made solutions to translators’ doubts. Freely available, reliable parallel data is however rather scarce, and almost limited – as far as the Italian language is concerned – to texts produced by the European Union. Even though translation quality should not an issue, translators – especially unexperienced ones – risk losing track of the fact that the language/terminology used within European institutions is a special language, specifically developed to reflect the supra-national nature of the EU (e.g. Fischer 2010, Cosmai 2000), which cannot be used within national contexts without interpretive, even legal, risks. This is particularly evident with respect with terminology belonging to social science fields such as law and economics, where European drafters need to use – or create – terms which are not linked to already existing national concepts in order to avoid confusion between national and EU concepts and enhance translation (EU 2013). In order to investigate the extent of this phenomenon, we conducted a small-scale corpus-based comparison of the language of EU and Italian national institutions in the domain of banking supervision and banking regulations. The study was based on a monolingual comparable corpus containing relevant official texts produced by national and EU institutions, which was analysed using a combination of automatic term extraction and hands-on corpus work. The study shows that even when the same terms (especially simple, single-word terms) are found in both subcorpora, there might be considerable differences in terms of frequency and of the creation of complex terms (or MultiWord Expressions). The paper aims to shed light on such differences and on possible reasons for the variation (i.e. non-standardisation) that can be observed even at the intracultural level, arguing that awareness of such differences is of paramount importance for translators and translation quality.