Core Concepts
This paper proposes an integrated system that combines automated writing evaluation (AWE) and grammatical error correction (GEC) to provide language learners with comprehensive feedback and assessment, facilitating their writing skill development.
Abstract
The paper presents an integrated system that combines automated writing evaluation (AWE) and grammatical error correction (GEC) to enhance the language learning process for second language (L2) learners.
The system enables language learners to simulate examination situations, where they can submit their written essays, and the system will provide them with both assessment of the writing quality and corrective feedback on grammatical errors. This integrated approach aims to bridge the gap between AWE and GEC, offering a more comprehensive solution compared to standalone systems.
The AWE component utilizes neural models to provide holistic scores and rubric-based evaluations, drawing on the ASAP and ASAP++ datasets. The GEC component employs sequence-to-sequence and edit-based neural models, leveraging various pre-trained language models such as BERT and BART, to detect and correct grammatical errors.
The authors discuss the potential of the system to facilitate language education, alleviate the burden on instructors, and empower learners to "write it right" through simulated examination scenarios. They also outline plans to expand the system to support multilingual inputs and incorporate instructor feedback into the workflow.
The paper highlights the importance of integrating AWE and GEC to provide comprehensive and efficient feedback, which can significantly improve the language learning experience for L2 learners.
Stats
The paper does not provide specific numerical data or statistics. However, it mentions that the current GEC model achieves a 65.29 F0.5 score on the BEA 2019 test set.
Quotes
"By leveraging the power of natural language processing (NLP) and machine learning algorithms, AWE and GEC systems have been developed separately to provide language learners with automated corrective feedback and more accurate and unbiased scoring that would otherwise be subject to examiners."
"Our approach empowers language learners to 'write it right' by enabling simulated examination situations."