Machine translation and legal terminology

Abstract

This chapter addresses machine translation (MT) with an eye to legal terminology. The translation of legal terms and phrasemes may be fraught with contextual complexities, and context has long been the Achilles’ heel of MT. Nevertheless, neural MT (NMT) and statistical MT (SMT) have made considerable progress in recent years, thanks to data-driven approaches making use of potentially related corpora to overcome contextual obstacles. Such approaches and the potential frozenness or repetitiveness of legal terms and phrases may allow MT to overcome some of these obstacles. This chapter reviews contextual complexities surrounding legal terminology, NMT and SMT architectures, and research on MT and legal translation to determine what might be expected from data-driven MT in this context.