| Summary: | 碩士 === 國立高雄第一科技大學 === 應用英語系口筆譯碩士班 === 103 === This study explores how to control the source Chinese text for Google Translate to produce more accurate Thai translation, so Thai people in Taiwan can learn about Taiwan’s culture and society by reading the Thai machine translation. Drawing on the relevant theories of machine translation (MT), controlled language writing and the linguistic features of Chinese, Thai and English as the theoretical framework, this research expects to identify the norms that should be followed in controlled Chinese (CC) writing for effective Thai MT application and how the norms are different from those used for Indo-European MT application. Ten web texts are adapted in CC for Thai MT application, including three cultural texts, two texts on tourist attractions, three web texts on Taiwanese folktales and two texts on Taiwan’s food and beverages. In addition, a questionnaire was administered on twenty Thai native speakers to learn the Thai respondents’ reception of Thai MT outputs of both texts in uncontrolled and controlled Chinese respectively.
The findings shows that the similarities between the CC norms for the MTs in Thai and in Indo-European languages include: 1) the use of vernacular Chinese, not classical Chinese, 2) the use of a paraphrase or an explanation for four-character words and fixed phrases, 3) the use of the SVO structure, and 4) the use of cultural terms in target languages. In contrast, there are some different norms, including: 1) more use of disyllabic words for Thai MT and less strict rules on the number of syllables for MTs in Indo-European languages in the lexical area, 2) more use of active voice for Thai MT and more use of passive voice for MTs in Indo-European languages in the syntactic area, and 3) less use of indefinite article and more use of measure words for Thai MT and more use of indefinite article and less use of measure words for MTs in Indo-European languages in the grammatical area. With regard to the results of the questionnaire, Thai respondents’ understanding of the Thai MT output of the uncontrolled Chinese text was 37.75 % on average while that of CC texts rose up to 89.75% on average. The statistical results show the effectiveness of CC use for Thai MT application.
In conclusion, this research suggests that CC norms can be designed to be suited to different target languages for MT application, so the accuracy and understandability of the MT output can be highly improved. In light of the benefits of CC writing, we recommend that the similar research model be applied to different controlled languages for effective MT application in different source and target language pairs in future studies.
|
|---|
