A Comparative Content Analysis of GPT Translation vs Human Translation: The Case of Journalistic Political Texts

Abstract

Translation is crucial for global communication, enabling the exchange of ideas, culture, and knowledge in different fields such as journalism. Recent advancements in artificial intelligence (AI) have revolutionized translation processes, posing challenges to traditional human translators. This study aims to compare and explore the effectiveness of translation between human translators and ChatGPT, focusing on cultural, syntactical, semantic, and linguistic dimensions. Qualitative research method, specifically comparative content analysis, is employed to address two key research questions which are: how does ChatGPT's translation differ from human translation in terms of accuracy and conformity to the source text? And To what extent can ChatGPT handle vocabulary, cultural nuances, semantics, and syntax compared to human translation and what are the reasons behind the committed mistakes? The study findings highlight distinct strengths of human translators in capturing subtle cultural nuances and emotional resonances, managing idiomatic expressions, metaphors, and semantic distinctions with clarity and accuracy, in journalistic political writing. In contrast, ChatGPT's translation errors often stem from its reliance on statistical correlations rather than nuanced contextual understanding and cultural sensitivity. They are predominantly unintentional due to algorithmic limitations rather than intentional biases. Future research should explore hybrid approaches combining AI capabilities with human oversight to enhance translation quality. Developing AI models that integrate cultural and contextual awareness could mitigate current limitations and the need for collaborative efforts between AI developers and human translators to improve AI's ability to handle complex linguistic and cultural nuances effectively.