7 July 2022 in Natural Language Engineering
Although Korean language education is experiencing rapid growth in recent years and several studies have investigated automated writing evaluation (AWE) systems, AWE for Korean L2 writing still remains unexplored. Therefore, this study aims to develop and validate a state-of-the-art neural model AWE system which can be widely used for Korean language teaching and learning. Based on a Korean learner corpus, the proposed AWE is developed using natural language processing techniques such as part-of-speech tagging, syntactic parsing, and statistical language modeling to engineer linguistic features and a pre-trained neural language model. This study attempted to determine how neural network models use different linguistic features to improve AWE performance. Experimental results of the proposed AWE tool showed that the neural AWE system achieves high reliability for unseen test data from the corpus, which implies metrics used in the AWE system can help differentiate different proficiency levels and predict holistic scores. Furthermore, the results confirmed that the proposed linguistic features–syntactic complexity, quantitative complexity, and fluency–offer benefits that complement neural automated writing evaluation.
26 September 2024 in Education and Information Technologies
1 October 2026 in Classroom Discourse
1. In spoken interaction, participants frequently encounter problems in hearing, understanding, or producing talk. Such troubles are typically managed through repair, a fundamental mechanism for ma...
25 March 2025 in Applied Linguistics Review
This study investigates the role of isomorphism in sentence processing by Korean heritage speakers in the United States. Employing acceptability judgement and self-paced reading tasks, we examine how Korean heritage speakers comprehend two Korean clausal constructions expressing transitivity: active transitive and suffixal passive. Results reveal a nuanced interplay between isomorphism, construction-specific cues, and task demands in sentence processing. While Korean heritage speakers’ acceptability ratings are largely comparable to those of monolingual Korean speakers, they spend more time reading critical regions and less time reading spill-over region than monolingual Korean speakers. Notably, there are no significant reading-time differences (i) across all critical and spill-over regions within identical grammaticality conditions and (ii) in the verb and post-verb regions across grammaticality pairs within the same construction and canonicity conditions. General proficiency in Korean does not substantially affect heritage speakers’ performance across the tasks. These findings advance our understanding of heritage language (processing) and sentence-processing architectures, also highlighting the unique linguistic experiences and challenges encountered by heritage language speakers.