Legal idioms and artificial intelligence: translation strategies in the face of complexity

Abstract

    The research problem lies in the existence of questions about the ability of machine translation and artificial intelligence to produce legal texts characterized by accuracy and stylistic robustness compared to human translation, especially in light of the increasing reliance on these technologies in the legal field. This study aims to offer a comparative viewpoint on the legal translations rendered by three types of agents: postgraduate translation students, Google Translate, and AI-driven MT engines. Through an analysis applied to a sample of chosen legal sentences, the quality of every translation has been rated by criteria drawn from Peter Newmark’s (1988 ) two categories of translation – communicative and semantic. Quality assessment consisted of an analysis of linguistic and legal correctness, stylistic uniformity and clarity. The study hypothesizes that AI-powered translation engines will outperform Google Translate and approach the quality of human translation in terms of stylistic coherence and legal accuracy, provided there is human supervision during the translation process. The findings of the study reinforce the promise of artificial intelligence in legal translation, particularly when there is human oversight, as well as the need for translation theory in evaluating translation quality in particular areas of specialization.