Scientific Development Research

The rapid development of neural machine translation (NMT) technology has significantly improved translation accuracy and efficiency, leading to its widespread application across various specialized fields. However, due to limitations in corpus data, specialized knowledge, and contextual understanding, machine translation still encounters numerous challenges in translating aviation texts, with lexical and syntactic issues being particularly prominent. This study focuses on aviation texts and aims to identify common error types in machine translation. By applying pose-editing, it seeks to effectively enhance the professionalism, readability, and overall quality of the output. It is found that terminological mistranslations and contextual mismatches at the lexical level are likely to cause conceptual confusion and even factual errors. At the syntactic level, rigid adherence to the source structure and semantic ambiguity result in disorganized logic and awkward expression, undermining the readability of the translation and the accuracy of information delivery. Post-editing can effectively identify and correct errors at both the lexical and syntactic levels, thereby improving the accuracy and readability of the translation. It is hoped that this study will provide valuable reference for post-editing similar texts and promote the further exploration and practice of machine translation post-editing in the aviation field.

Aviation texts; Machine translation; Types of errors; Post-Editing