A Comparative Assessment On GNMT Translation And Human Translation From A Manipulation Perspective

In recent years,with the development of artificial intelligence and deep learning technologies,there has been a breakthrough in neural machine translation research.Particularly,Google developed a new generation of neural translation(GNMT)system in September 2016,which uses the most advanced deep learning model available today and a long short-term memory(LSTM)network to overcome the inherent weaknesses of the traditional machine translation model.GNMT has dramatically improved the quality of machine translation.However,sometimes it still makes mistakes when it comes to Chinese-English translation.This thesis attempts to compare the quality of human translation and GNMT translation from the perspective of Manipulation Theory.The author selects the response of China's Foreign Minister,Wang Yi,to the questions on the Korean Peninsula issue in the press conferences of China's Foreign Ministry during the NPC and CPPCC annual sessions from 2017 to 2018 as the source language text.The thesis compares the interpreter Sun Ning's interpretation with that of GNMT and conducts comparative analysis from the perspectives of words,sentences and discourse.The differences of translation results are interpreted from the perspective of Manipulation Theory in this thesis,which aims to help translators and interpreters recognize and strengthen the advantage of human translation on the basis of getting a better understanding of the limitations of GNMT at the current stage.It can also provide some insight at the linguistic level for R&D engineers to improve the quality of machine translation.