A Corpus-Based Comparative Study of ChatGPT and DeepSeek in Chinese-English Translation

The purpose of this research is to compare the translation performance of two advanced AI-based machine translation systems (MT) - ChatGPT and DeepSeek - in Chinese into English translation, focusing on multidimensional linguistic features such as lexical, syntactic and discourse aspects. To achieve this research goal, the study poses the following questions: Which AI-based system exhibits better translation quality, ChatGPT or DeepSeek? What are the specific linguistic strengths and limitations of these systems? Data sources were original Chinese texts, translations generated by ChatGPT and DeepSeek, and human reference translations. Automated linguistic analyses were conducted using standard MT evaluation metrics (BLEU, METEOR, and BERTScore) and Coh-Metrix to assess lexical diversity, syntactic complexity and discourse cohesion. The results show that ChatGPT performs better than DeepSeek, with more accurate word choices, simpler language suited for readers, and better text flow, making translations easier to read and understand. This study highlights the strengths and weaknesses of AI translation and suggests further research on human-AI collaboration and its role in education.