Evaluation Of Long Sentence E-C Machine Translation In The Perspective Of Systemic Functional Grammar

With the ever-accelerated development of the global economy, traditional artificial translation hasn’t far satisfied the rapid growth of the needs of translation market. As the product of new era, machine translation has been impassionedly studied on a global scale by its merits of high efficiency and low cost. However, since machine is still not beyond the unique capacity of human in processing social knowledge, its quality of translated text still doesn’t make a breakthrough progress. Particularly, the issue of long sentence translation is still an unsolved problem in the field of machine translation(许建平1998).This thesis mainly studies the problems still existing in the E-C machine translation of long sentence, in which there are tow mainly creative points. At first, the author, through plenty of translation practice, makes a conclusion to different special sentence patterns in long sentence and their translated text, then realizes that the unsatisfactory translated text results from the wrongly manifestation of language metafunctions, which in turn attributes to that there are no counterpart of English sentence patterns in Chinese. Another innovative point is, based on systemic functional grammar, to assess the metaftunctions defaults of translated text and make a statistical analysis to them. By the way, some suggestions for improving long sentence MT are put forward.This thesis consists of six parts. The first one introduces the significance and objective of the study, research objects, research technique and thoughts exhaustively. Chapter one is literature review, in which the existing research achievements of long sentence translation and MT at home and abroad are concluded and assessed. Chapter two lays the theoretical foundation for the present study, including SFL and its influence on translation and MT studies, as well as the application of three language metafunctions in translation study. Chapter three concludes three kinds of long sentences from translation practice and their characteristics, step forwardly, sums up three obstacles in their machine translation. Chapter four chooses the long sentences from translation practice as examples, dissects their characters and metafunction defaults of their translated text. Then, statistics and discussion of the results are made. The last chapter is conclusion, some improvement suggestions of long sentence MT are proposed, besides, the major findings and limitations of the present study are concluded.