| This Thesis is inspired by the boom of Artificial Intelligence and the alleged replacement of machine translation for human translators and interpreters.Two years have passed since the rollout of Neural Machine Translation(NMT).Where does NMT stand now in terms of Chinese-English translation?What kind of errors or mistakes NMT platforms made in the past and are there any improvements now?Do different types of texts or NMT platforms have different mistakes or errors?To answer these qLuestions,three types of text will be chosen and translated by Google Translate and Baidu Translate;and a set oferror-identifying criteria will be set in order to contrast and analyze changes of’ machine translation quality between 2017 and 2019.Then the author will conduct quantitative and qualitative analysis of translation results to evaluate NMT developments and identify typical problems of the latest NMT.Results show that NMT has coIme a long way during the past two years in terms of word and sentence correctness.fluency as well as readability.On the other hand,NMT platforms are still lackluster in translating rare words such as Chinese idioms,poetry,spoken words and words "with Chinese characteristics".Also,NMT still lacks the ability to read out hidden logic.translate complex sentences or combine text with context.In terms of text type,NMT platforms are proficient at translating professional text with certain patterns.good at government report and bad at spoken materials.Baidu NMT outperformed Google NMT in translating Chinese into English,excelling especially in processing rare Chinese words.In comparison,Google seemed to have lacked "learnig and training" in this regard,thus producing undesirable results.Besides,new problems such as repetitive translation or data mis-translation crop up in 2019.It’s proved in this thesis that NMT has made g eat strides between 2017 and 201 9.being able to assist human translators in a more efficient way.Nevertheless.seemingly insurmountable issues still exist in both NMT platforms,including issues related to cultural difference,context or communication.It’s suggested that human translators or interpreters open their eves to the latest developments of NMT and draw on the strength of machines to better adapt to the ever-changing AI age. |
