Research Questions and a Proposal for the Future Governance of Translation Data

Abstract

This article seeks to assess the impact of data-driven methods of machine translation (MT), not just on translators, but more broadly on industry and society. We consider translation as a shared resource that is the result of producers and consumers who share the overall objective of providing sustainable access to reliable multilingual information. This view of translation as a shared knowledge resource is based on the reliance of machine learning-based translation techniques on the pooling and leverage of parallel text in the form of existing translations. These techniques build on the well-established practice of leveraging such resources as translation memories. We use the institutional analysis and development (IAD) framework to assist with analysis of the dynamic situations emerging through the use of MT, and consider why changes, such a shared ownership model, would ultimately benefit all stakeholders. We finally suggest some research questions that might help to assess sustainability for translation as a resource, an industry, and an occupation, such as 'what would happen if we left translation to machines?', 'why do translators not act collectively', and 'could principal and end customer buy-in be heightened by quantifying the threat to sustainability of access to translations?'.