Translation choices compared: Investigating variation in a learner translation corpus

This paper focuses on the concept of variation in translation through the analysis of different target versions of the same source text. The study draws upon a small subset of an English-to-Italian learner translation corpus. By observing the translation choices made by several trainees translating three source texts, belonging to different genres and characterised by different degrees of specialisation, the study seeks to identify which elements (e.g. lexical items, syntactic structures) are most open – and, conversely, less susceptible – to variation, as well as how different translation solutions relate to – or even impact on – text meaning. A twofold methodological approach is adopted for that purpose, which combines n-gram extraction with the manual analysis of multi-parallel concordances. The analysis shows limited full invariance in learner translations, but it also points to the existence of preferred translation solutions and variations which can be placed on a cline of acceptability, and which need better characterisation.