Due to the difficulty of creating high-quality labelled training data for different languages, the low-resource problem is crucial yet challenging for automated essay scoring (AES). However, little attention has been paid to addressing this challenge. In this paper, we propose a novel zero-shot cross-lingual scoring method from the perspectives of pretrained multilingual representation and writing quality alignment to score essays in unseen languages. Specifically, we adopt multilingual pretrained language models as the encoder backbone to deeply and comprehensively represent multilingual essays. Motivated by the fact that the scoring knowledge for evaluating writing quality is comparable across different languages, we introduce an innovative strategy for aligning essays in a language-independent manner. The proposed strategy aims to capture shared knowledge from diverse languages, thereby enhancing the representation of essays written in unseen languages with respect to their quality. We include essay datasets in six languages (Czech, German, English, Spanish, Italian and Portuguese) to establish extensive experiments, and the results demonstrate that our method achieves state-of-the-art cross-lingual scoring performance.

Data augmentation aims at expanding training data with clean text using noising schemes to improve the performance of grammatical error correction (GEC). In practice, there are a great number of real error patterns in the manually annotated training data. We argue that these real error patterns can be introduced into clean text to effectively generate more real and high quality synthetic data, which is not fully explored by previous studies. Moreover, we also find that linguistic knowledge can be incorporated into data augmentation for generating more representative and more diverse synthetic data. In this paper, we propose a novel data augmentation method that fully considers the real error patterns and the linguistic knowledge for the GEC task. We conduct extensive experiments on public data sets and the experimental results show that our method outperforms several strong baselines with far less external unlabeled clean text data, highlighting its extraordinary effectiveness in the GEC task that lacks large-scale labeled training data.