A Study of Text Classification Based on Machine Learning in Translation Studies
As is known,there is a close correlation between machine learning in translation studies and digital humanities.This paper proposes a five-steps method for text classification in translation studies,including distant reading of data,middle-distance reading of discriminative features,scrutinizing of randomly selected texts,identification of linguistic patterns and interpretation of linguistic patterns.With the five-steps method,it is found that the translation-oriented Chinese is characterized by less use of two noun phrase patterns—numeral+quantifier+noun and numeral+quantifier,which fall into the peripheral members in the respective categories concerned,and of which cognitive mechanism may be attributed to the translators'ignorance of the less prominent members of the semantic network.The case study reveals that text classification algorithm can improve the sketch of the integrity and objectivity of the translation-oriented language per se,while the close reading of data-driven text facilitates to pining on more fine-grained language patterns.This paper aims to shed some light on new methods for translation studies in the perspective of digital humanities.
digital humanitymachine learningtext classificationtranslation studiesfive-steps method
万方数据
