How EUR-lex parallel corpora improve the translation of general legal formulae in LELF

Abstract

Legal formulae are fixed expressions that can pose several translation challenges in view of system-related peculiarities, especially in international and multicultural settings such as the European Union. General bilingual dictionaries are often used by students and translators to cope with legilinguistic hurdles, although such tools do not always account for system-bound specificities. This may give rise to translation infelicities and/or partial translations. This paper addresses legal translation research and examines the usefulness of the EUR-Lex parallel corpora via the Sketch Engine platform to translate general legal formulae. More precisely, it focuses on neutral legal multi-words and compares corpus-driven data against entries obtained from three online general dictionaries (i.e., Cambridge, Collins, and WordReference). It assesses the translation of seven general legal formulae from Legal English as a Lingua Franca (LELF) into French and Italian. By comparing corpus-sourced equivalents with standard dictionary outputs, the study reveals that institutional parallel corpora generally outperform generic lexicographical tools. The findings also indicate that traditional dictionaries suggest decontextualised terms, whereas specialised parallel corpora offer contextually relevant terminology. Therefore, this investigation provides useful insights for the integration of corpus-driven data into professional legal translation practice and translator-training programmes.