Translator-led Evaluation Of Online Translation Tools

At such an age when translators are equipped with all kinds of online translation tools (OTTs) like online machine translation systems (OMTSs) and online computer-aided translation tools (OCTTs), the capability of using and evaluating them properly has been added to the "skill-sets" required of translators. Nevertheless, most of the previous researches on translation tools, especially on evaluating translation tools were guided by the faith of "technology first". In other words, they focused on the computational algorithms from the point of view of computer programmers, rather than the hands-on experiences from the point of view of translators.This thesis seeks to explore theoretically and practically on the methodology of conducting a translator-led OTT evaluation. It first argues on the necessity and feasibility of translator-led OTT evaluations via reviewing previous studies on OTTs and OTT evaluations home and abroad; then it goes on to conduct an evaluation of two frequently-used online machine translation systems (i.e. Google OMTS and Baidu OMTS) after a careful design of evaluation criteria and evaluation indicators, and selection of evaluation materials. An evaluation team (including four postgraduate who major in English language and literature in the direction of translation study and one clerk in a translation company with MTI degree) is also founded to operate the two OMTSs and give point(s) towards each indicator according to the criteria given. After that, the author collects and gives analysis on the data. Seen from the results, Baidu OMTS is in a leading position in operating experience, punctuation translation, and Chinese-to-English lexical and syntactic translation, while Google OMTS performs better in English-to-Chinese lexical and syntactic translation. Finally, four suggestions are put forward towards the Translator-led OTT evaluations:firstly, translators are supposed to pay more attention to OTTs and OTT evaluations; secondly, a comprehensive method should be applied in a translator-led OTT evaluations; thirdly, a careful design will lead to a more reliable result; fourthly, the evaluation should be conducted by a number of translators instead of only one translator.