Towards methodologically more rigorous corpus-based translation studies

Abstract

This paper argues in favour of a more rigorous methodology for corpusbased translation studies. According to our proposal, research papers in the field should be minimally required to (i) provide a meticulous overview of the corpus materials used and of the exact procedures for selecting, annotating and sifting the data; (ii) comment on any specific problems encountered during data selection and annotation, including explicit and motivated statements as to the solutions being adopted; (iii) include elaborate testing for statistical significance as a complement of, not in opposition to, thorough qualitative analysis. This approach, we suggest, not only offers a way around many theoretical and methodological problems that have been noted in the recent literature (e.g., House 2008; Becher 2010; Bernardini and Ferraresi 2011), it also facilitates more rigorous replication and reinterpretation of previous work, potentially leading to a re-assessment of some popular but unproven assumptions such as the notion that linguistic features in translations are independent of source language or genre. By thus moving forward the empirical cycle of testing and re-testing of hypotheses, the methodology advocated here encourages collaborative research, and leads ultimately to more scientific progress.