The Actres Project: Using corpora to assess English-Spanish translation

Abstract

Assessing translation quality is generally seen as a difficult and elusive task because of the inadequacy of the tools available. The aim of this paper is to demonstrate the usefulness of a corpus-based contrastive methodology developed at the University of León (Spain) for identifying instances of translationese. The ACTRES project functional framework draws on the work by Bondarko (1991) and Chesterman (1998) and has been designed for translation-oriented cross-linguistic analysis (Rabadán et al. 2004). The long-term study focuses on those semantic areas that are typically problematic for our language pair (modality, quantification, modification, aspectuality, etc). The contrast features a two-step procedure: contrasting typical ways of expressing similar meanings in English and Spanish, and spotting differences between original Spanish and translated Spanish. First, empirical data are extracted from two monolingual ‘comparable’ corpora -The Bank of English and the Corpus de Referencia del Español Actual (CREA); secondly, these results are compared with data from a custom-made translation corpus containing English original texts and their corresponding Spanish translations. The three sets of results provide different types of useful information- i) the resources available (or absence of) in each of the languages to express a given meaning and their relative centrality, ii) the solutions favored by translators to bridge the crosslinguistic disparities and/or gaps and iii) the erroneous or non-existent uses and structures transferred from the source language into the target language. Translation practice, translator training and translation quality assessment (TQA) are the main areas that can benefit from this type of research.