Translation Quality Assessment of Legal Contracts Translated by Google Translate and ChatGPT

Legal contracts are among the most difficult documents to translate because contracts are known for their linguistic and stylistic intricacies. Nowadays, people rely mainly on automated translation systems such as Google Translate and ChatGPT as translators to render different types of texts. This study aims to evaluate the quality of the translation of legal contracts rendered by Google Translate (GT) and ChatGPT using House's (2015) Model for Translation Quality Assessment (TQA). The study uses a descriptive qualitative approach to investigate the accuracy of Google Translate and ChatGPT when rendering legal contracts. This study focuses on analysing the performance of both systems in translating contracts with emphasis on accuracy, through the identification of errors, precisely covert and overt ones, as classified by House's Model. The Model is employed to assess translation quality on multiple levels, including linguistic, stylistic, and contextual accuracy. An authentic contract written in Arabic is translated into English by both Google and ChatGPT, and errors were identified and classified according to House's Model (TQA). The results of this study demonstrate that both systems exhibit shortcomings when they render a legal contract at different levels, including: the linguistic, contextual, and stylistic levels. These shortcomings affect the quality of both translations negatively because both systems exhibit lexical inaccuracies and structural inconsistencies, which result in ambiguity, lack of domain-specific knowledge, inconsistent jargon, addition of redundant information, failure to preserve the formal and binding tone of the original text, and a need for human oversight.