Comparing Feature-Engineering and Feature-Learning Approaches for Multilingual Translationese Classification
This work addresses the task of translationese classification for computational linguistics, providing empirical evidence for translation universals, but it is incremental as it applies existing neural methods to a less-explored domain.
The paper tackled the problem of classifying translated vs. original texts by comparing traditional feature-engineering methods with neural feature-learning approaches, showing that neural architectures, especially BERT-based models, outperform others by over 20 accuracy points in both monolingual and multilingual settings.
Traditional hand-crafted linguistically-informed features have often been used for distinguishing between translated and original non-translated texts. By contrast, to date, neural architectures without manual feature engineering have been less explored for this task. In this work, we (i) compare the traditional feature-engineering-based approach to the feature-learning-based one and (ii) analyse the neural architectures in order to investigate how well the hand-crafted features explain the variance in the neural models' predictions. We use pre-trained neural word embeddings, as well as several end-to-end neural architectures in both monolingual and multilingual settings and compare them to feature-engineering-based SVM classifiers. We show that (i) neural architectures outperform other approaches by more than 20 accuracy points, with the BERT-based model performing the best in both the monolingual and multilingual settings; (ii) while many individual hand-crafted translationese features correlate with neural model predictions, feature importance analysis shows that the most important features for neural and classical architectures differ; and (iii) our multilingual experiments provide empirical evidence for translationese universals across languages.