From Google Translate to ChatGPT

Abstract

This chapter examines the quality of machine translated texts of various genres with a view to finding out to what extent these texts/genres have been accurately translated by neural machine translation systems and LLMs. The chapter also examines the potential use of these technologies in editing and revisions to enhance both quality and productivity. A theoretical and conceptual framework that is based on text-typology linguistic models and Mossop's parameters of revising and editing in translation is used to assess the quality of transactions produced by Google Translate (GT) as a neural-based machine translation system and ChatGPT as a large language model (LLM). In particular, the chapter investigates the GT and ChatGPT translations of argumentation within journalistic texts, exposition within both promotional and philosophical genres, and instrumental texts within the legal genre. This emphasis on diverse text types allows for a comprehensive evaluation of the translation performance of neural-based systems and LLMs across different communicative purposes and stylistic demands.

Ähnliche Arbeiten