Designing Controlled Chinese Rules for MT Pre-Editing of Product Description Text

Pre-Editing Based On Controlled Language

Controlled language (CL) refers to “subset of natural languages whose grammars and dictionaries have been restricted in order to reduce or eliminate both ambiguity and complexity”(ISO, 2017). Controlled language, commonly controlled English, has been mainly applied to technical documents including technical guides and maintenance manuals in technology industries such as aviation since the 1970s, with the aim of improving the readability and translatability of technical texts. Since 2003, the research on CL has turned to controlled translation, i.e., the combination of CL and MT. CL can effectively eliminate the ambiguity of the source language content, which is the biggest challenge for MT, greatly improve the accuracy of MT and reduce post-translation editing workload (Yuan, 2003). Studies on CL have focused on translation between European languages (such as English, German and French), and the results show a generally positive impact of CL rules on the MT output (Bernth & Gdaniec, 2001; Reuther, 2003; Marzouk, 2021). Marzouk’s recent study (2021) indicates that neural MT offers a promising solution that no longer requires CL rules for improving the MT output in the context of technical texts. But we wonder if the idea of CL can be introduced to the MT of non-technical texts, and we aim to answer that question in this paper.