Evaluation of English-Chinese Translation Effects in Intelligent Translation Teaching with GPT Language Model

This paper addresses common issues in English-Chinese translation teaching, such as low translation quality and unnatural expression, by proposing a method to evaluate translation effects using the GPT (Generative Pre-trained Transformer) language model. It identifies bottlenecks in traditional translation methods, particularly the limitations of machine translation in fluency and semantic expression. The study employs an experimental comparison method, translating selected texts with various tools and comparing them to translations generated by GPT-4. Results indicate that GPT-4 outperforms traditional tools, achieving an average BLEU score of 86.5% for accuracy, with only one grammatical error on average, and a naturalness similarity score of 0.89. Overall, GPT-4 enhances students' understanding of complex syntax and authentic expressions, providing a scientific basis for high-quality translation teaching.