A Study On The Relation Between Deverbalization And AI Translation Quality

Is it feasible to apply translation or interpretation theories and strategies to AI translation via deverbalization,one core interpretation theory? Aiming to address this research question and to provide a brand new vision for scholars and scientists alike dedicating themselves to the improvement of the quality of AI translation,this study chose three different texts from the fields of politics,culture and technology respectively encompassing 727,760 and 774 Chinese character,and compared the translations done by two human interpreters,Baidu Translation and Google Translation against the benchmark of de-verbalization.Through detailed analyses,this study attempted to address two research questions: one is whether the quality of translated text is positively correlated to the quantity of deverbalization occurring in translation;the other is that whether deverbalization can be an effective way to improve translation quality of AI translation.In this study the parameters for deverbalization and translation quality assessment are set.Supplement,restructuring,replacement and omission are the four parameters of deverbalization;the four parameters of translation quality assessment are grammatical mistakes,meaning deviation,redundancy and information missing.By calculating and comparing the quantity of deverbalization and the mistakes occurring in each translation,this study found that translation quality bears prominently positive relationship to the quantity of deverbalization in translation.Pearson Correlation,which shows that the number of errors in a translation is negatively related to the participation of deverbalization at 64%,reinforces the conclusion.This study also found that by adopting deverbalization,nearly half of the mistakes in translation can be corrected.So this study arrived at the conclusion that deverbalization can potentially serve as an effective way to improve the quality of AI translation.