Technology's Influence on Translation Practices: Analyzing Machine Translation and Post-Editing Strategies in the Framework of COP Documentation

Abstract

This paper investigates the application of machine translation within the context of a specific, narrow topic: the Conference of the Parties (COP), relevant to climate change and international relations. This research is significant for my current academic focus and future career, motivated by curiosity about the potential utility of machine translation tools for the general public. The scope of this study is highly focused, as few existing studies compare multiple machine translation tools - most often only one or two are analyzed. Additionally, previous research predominantly relies on statistical and automated scoring methods using various metrics. Despite the widespread recognition of climate change, the specific relationship between COP terminology and its connection to climate issues and United Nations official language remains underexplored. The primary methodology employed is a qualitative assessment of machine translations generated by ChatGPT, Google Translate, and DeepL. These translations were analyzed by the author with the assistance of human translators, with evaluations based on four criteria: accuracy, readability, terminology consistency, and post-editing effort. Post-editing refers to the time and effort required to adapt machine-translated output to an acceptable standard, which may involve correcting errors or performing a comprehensive review, as conducted in this study. The findings indicate that DeepL and ChatGPT provide sufficiently accurate translations - neither perfect nor exceptional, but quite convenient and generally reliable. The translations are adequate but require additional editing time to reach a satisfactory quality for users. No machine translation tool is flawless or ideal. Given that current tools largely rely on statistical algorithms rather than advanced machine learning or AI, it is anticipated that future developments will improve translation quality. Further research is necessary to explore emerging tools, as the landscape of machine translation is rapidly evolving and remains underexplored.