Discourse Analysis of Argumentative Essays of English Learners Based on CEFR Level

Proceedings of Recent Advances in Natural Language Processing, pages 468–474 Varna, Sep 4–6, 2023 https://doi.org/10.26615/978-954-452-092-2_052 468 Discourse Analysis of Argumentative Essays of English Learners based on CEFR Level Blaise Hanel and Leila Kosseim Computational Linguistics at Concordia (CLaC) Laboratory Department of Computer Science and Software Engineering Concordia University, Montr´eal, Qu´ebec, Canada blaise.hanel@concordia.ca, leila.kosseim@concordia.ca Abstract In this paper, we investigate the relationship between the use of discourse relations and the CEFR-level of argumentative English learner essays. Using both the Rhetorical Structure Theory (RST) and the Penn Discourse Tree- Bank (PDTB) frameworks, we analyze es- says from The International Corpus Network of Asian Learners (ICNALE), and the Corpus and Repository of Writing (CROW). Results show that the use of the RST relations of EX- PLANATION and BACKGROUND, as well as the first-level PDTB sense of CONTINGENCY, are influenced by the English proficiency level of the writer. 1 Introduction In a world where over 7,000 languages are used, much research has focused on improving meth- ods to teach and learn natural languages. In par- ticular, the field of Natural Language Processing (NLP) has a long history of developing tools to as- sist language learners and reduce learning barriers. Previous works on surface linguistic features and language learning, such as Webber (2009), Bac- hand et al. (2014), and Abdalla et al. (2018) have shown significant difference in discourse usage across textual genre and simplicity level. To our knowledge, very few studies have investigated the relationship between discourse structures and lan- guage learning. Corpus research on the use of discourse struc- tures among different CEFR levels can provide valuable insights into how well language learners are able to organize and convey their ideas in writ- ten or spoken language. Such an analysis can also identify common patterns of language use that are particularly challenging for learners at different CEFR levels, leading to the development of more effective teaching materials and strategies that tar- get learners’ specific needs (Aoyama, 2022), while simultaneously reducing the workload of human graders (Mieskes and Pad´o, 2018). Findings can also inform the development of more reliable as- sessment tools that accurately measure learners’ proficiency in the use of discourse structures. Ac- curate assessment is essential for learners to iden- tify their strengths and weaknesses and make in- formed decisions about their language learning goals and strategies. In this paper, we investigate the usage of dis- course relations using the Rhetorical Structure Theory (Mann and Thompson, 1988) and the Penn Discourse TreeBank (Prasad et al., 2008) frameworks to discover trends in their usage in ar- gumentative English learners across various profi- ciency levels. Results show that the RST relations of EXPLANATION and BACKGROUND are statis- tically used more often by writers with a lower CEFR language level, and the use of the PDTB relation of CONTINGENCY decreases as CEFR level increases. 2 Background 2.1 Discourse Analysis Frameworks In order to analyze the discourse structure of texts computationally, two main frameworks have been developed: Rhetorical Structure Theory (RST), proposed by Mann and Thompson …

Content truncated to 500 words