Sun-Young Oh
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Sun-Young Oh has a total of 26 co-authors. Here are the top co-authors by number of shared publications:
- Min-Chang Sung
Gyeongin National University of Education
2 shared publications - Soyeon Moon
Seoul National University Sun-Young Oh* Seoul National University
2 shared publications Chae-Young Ahn
1 shared publicationDong-Yeob Ham
1 shared publicationEun Kyoung Lee
1 shared publicationEun Young Bae
Department of Applied Linguistics, University of California, Los Angeles
1 shared publication
Latest Publications
- Open Access
Online processing fluency and explicit knowledge in data-driven learning: Rethinking grammatical difficulty
- Open Access
- 30 August 2026
- Published by Elsevier in SSRN Electronic Journal
Undergraduate L2 writers’ engagement with ChatGPT feedback: A prompt-by-prompt multiple-case study
Research on L2 writers’ engagement with ChatGPT feedback has grown rapidly, but less is known about how engagement unfolds across successive prompts or relates to writers’ prior experiences and educational contexts. This multiple-case study examined three Korean EFL undergraduates revising a persuasive essay with free versions of ChatGPT-4o and 3.5 in up to two 30-minute sessions, without a tutorial or task-specific prompt templates. Data included ChatGPT chatlogs, initial and final drafts, stimulated recall and semi-structured interviews, and screen and gaze recordings supporting recall. The three writers showed distinct patterns: reactive prompting accompanied by frustration, strategic attention to local language choices, and iterative work on discourse-level concerns. Similar revision behaviors, including acceptance of AI-generated revisions, reflected different forms and levels of cognitive and affective engagement. Motivation, prior ChatGPT use, prompting practices, writing beliefs, and Korean EFL educational experiences were associated with engagement in different ways. The findings show that feedback uptake alone is insufficient for identifying overreliance. Tracing prompt sequences alongside revision operations and writers’ accounts provides a fuller view of agency and decision-making in AI-assisted revision. Pedagogical implications include supporting students in formulating specific prompts and evaluating and selectively using AI feedback.
- 30 August 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessUndergraduate L2 writers’ engagement with ChatGPT feedback: A prompt-by-prompt multiple-case study
Research on L2 writers’ engagement with ChatGPT feedback has grown rapidly, but less is known about how engagement unfolds across successive prompts or relates to writers’ prior experiences and educational contexts. This multiple-case study examined three Korean EFL undergraduates revising a persuasive essay with free versions of ChatGPT-4o and 3.5 in up to two 30-minute sessions, without a tutorial or task-specific prompt templates. Data included ChatGPT chatlogs, initial and final drafts, stimulated recall and semi-structured interviews, and screen and gaze recordings supporting recall. The three writers showed distinct patterns: reactive prompting accompanied by frustration, strategic attention to local language choices, and iterative work on discourse-level concerns. Similar revision behaviors, including acceptance of AI-generated revisions, reflected different forms and levels of cognitive and affective engagement. Motivation, prior ChatGPT use, prompting practices, writing beliefs, and Korean EFL educational experiences were associated with engagement in different ways. The findings show that feedback uptake alone is insufficient for identifying overreliance. Tracing prompt sequences alongside revision operations and writers’ accounts provides a fuller view of agency and decision-making in AI-assisted revision. Pedagogical implications include supporting students in formulating specific prompts and evaluating and selectively using AI feedback.
- Open Access
- 27 April 2026
- Published by Elsevier in SSRN Electronic Journal
Divergent cohesion profiles: Comparing L2 proficiency levels and ChatGPT-mediated revisions in EFL argumentative writing
The emergence of ChatGPT has elevated interest in comparing the linguistic features of AI-generated texts with human-authored ones to inform writing pedagogy. One feature that remains underexplored is cohesion, which functions as a basic linguistic resource for textual connections in L2 writing. Using a mixed-methods approach, this study combines TAACO–based quantitative analyses with qualitative examinations of students’ excerpts. It examines how cohesion features in 90 Korean EFL college learners’ argumentative essays are reshaped by ChatGPT revisions and how these changes differ from the patterns observed across proficiency levels. The results revealed significant divergence of cohesion profiles in ChatGPT-revised texts from EFL learners’ writing, with distinct effects across proficiency levels. While the learners’ cohesion progressed from explicitness to implicitness in meaning connections, ChatGPT revisions favored explicit cohesive use with relatively high levels of lexical diversity throughout the text. Notably, ChatGPT-driven cohesion shifts demonstrated potential risks of attenuating writers’ original rhetorical styles and cultural nuances. Extending Warschauer et al.’s (2023) framework of ”affordances and contradictions,” this study reframes ChatGPT as a mediational artifact that affords opportunities for textual enhancement while introducing inherent tensions in L2 writers’ voice and discursive identities. Pedagogical implications for writing instruction and assessment in AI-mediated contexts are discussed.
- 27 April 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessDivergent cohesion profiles: Comparing L2 proficiency levels and ChatGPT-mediated revisions in EFL argumentative writing
The emergence of ChatGPT has elevated interest in comparing the linguistic features of AI-generated texts with human-authored ones to inform writing pedagogy. One feature that remains underexplored is cohesion, which functions as a basic linguistic resource for textual connections in L2 writing. Using a mixed-methods approach, this study combines TAACO–based quantitative analyses with qualitative examinations of students’ excerpts. It examines how cohesion features in 90 Korean EFL college learners’ argumentative essays are reshaped by ChatGPT revisions and how these changes differ from the patterns observed across proficiency levels. The results revealed significant divergence of cohesion profiles in ChatGPT-revised texts from EFL learners’ writing, with distinct effects across proficiency levels. While the learners’ cohesion progressed from explicitness to implicitness in meaning connections, ChatGPT revisions favored explicit cohesive use with relatively high levels of lexical diversity throughout the text. Notably, ChatGPT-driven cohesion shifts demonstrated potential risks of attenuating writers’ original rhetorical styles and cultural nuances. Extending Warschauer et al.’s (2023) framework of ”affordances and contradictions,” this study reframes ChatGPT as a mediational artifact that affords opportunities for textual enhancement while introducing inherent tensions in L2 writers’ voice and discursive identities. Pedagogical implications for writing instruction and assessment in AI-mediated contexts are discussed.
- 31 January 2026
- Published by Korean Association for the Study of English Language and Linguistics (KASELL) in Korea Journal of English Language and Linguistics
- Vol. 26, 580-605
Developing and Validating a Curriculum-Based Rating Scale for Korean Middle School EFL Writing: An Argument-Based Validation Approach
- 31 January 2026
- Published by Korean Association for the Study of English Language and Linguistics (KASELL) in Korea Journal of English Language and Linguistics
- Vol. 26, 580-605
Developing and Validating a Curriculum-Based Rating Scale for Korean Middle School EFL Writing: An Argument-Based Validation Approach
- 31 January 2026
- Published by Korean Association for the Study of English Language and Linguistics (KASELL) in Korea Journal of English Language and Linguistics
- Vol. 26, 546-563
Grade-level classification of the basic vocabulary in the 2022 revised English curriculum: A corpus-based analysis of English textbooks
- 31 January 2026
- Published by Korean Association for the Study of English Language and Linguistics (KASELL) in Korea Journal of English Language and Linguistics
- Vol. 26, 546-563
Grade-level classification of the basic vocabulary in the 2022 revised English curriculum: A corpus-based analysis of English textbooks
- Open Access
- 1 January 2026
- Published by National Foreign Language Resource Center in Language Learning & Technology
- Vol. 30 (1) , 1-25
Agency to autonomy in mediated data-driven learning: A longitudinal study
Drawing on Sociocultural Theory (SCT), this longitudinal case study examines how secondary EFL learners exercise agency, engage, and develop autonomy within digitally mediated data-driven learning (DDL). It began with a 16-month compilation of a local learner corpus, followed by 7 months of four pedagogical phases—paper-based, hands-on, customized, and self-directed—progressing from teacher-guided instruction to autonomous engagement. Multimodal data reveal that learner autonomy in DDL goes beyond technical mastery of corpus tools; it unfolds as a socially, emotionally, and contextually mediated process shaped by identity, affect, and strategic growth. Suho, the Planner, progressed through structured, scaffolded support, while Jimin, the Seeker, exercised exploratory agency through multimodal engagement aligned with personal interests. Theoretically, the study extends SCT by demonstrating reciprocal mediation in digital learning environments and highlighting the interdependence between agency, identity, and affect. Pedagogically, findings underscore the central role of teacher expertise—particularly adaptive technological, pedagogical, and content knowledge—in making DDL accessible and sustaining learner engagement. Methodologically, the study highlights the value of longitudinal, multimodal, and ecologically situated approaches for capturing the process of agency, engagement, and autonomy.
- 1 January 2026
- Published by National Foreign Language Resource Center in Language Learning & Technology
- Vol. 30 (1) , 1-25
Open AccessAgency to autonomy in mediated data-driven learning: A longitudinal study
Drawing on Sociocultural Theory (SCT), this longitudinal case study examines how secondary EFL learners exercise agency, engage, and develop autonomy within digitally mediated data-driven learning (DDL). It began with a 16-month compilation of a local learner corpus, followed by 7 months of four pedagogical phases—paper-based, hands-on, customized, and self-directed—progressing from teacher-guided instruction to autonomous engagement. Multimodal data reveal that learner autonomy in DDL goes beyond technical mastery of corpus tools; it unfolds as a socially, emotionally, and contextually mediated process shaped by identity, affect, and strategic growth. Suho, the Planner, progressed through structured, scaffolded support, while Jimin, the Seeker, exercised exploratory agency through multimodal engagement aligned with personal interests. Theoretically, the study extends SCT by demonstrating reciprocal mediation in digital learning environments and highlighting the interdependence between agency, identity, and affect. Pedagogically, findings underscore the central role of teacher expertise—particularly adaptive technological, pedagogical, and content knowledge—in making DDL accessible and sustaining learner engagement. Methodologically, the study highlights the value of longitudinal, multimodal, and ecologically situated approaches for capturing the process of agency, engagement, and autonomy.
- 1 August 2025
- Published by John Benjamins Publishing Company
Production and understanding of change‑of‑state tokens in English talk‑in‑interaction among L1 and L2 speakers
This study investigates the production and understanding of change-of-state tokens like oh and ah in English conversations among one L1 speaker and three L2 speakers. Using conversation analysis, it examines how these speakers do interactional work to achieve shared understanding through these tokens. The L1 speaker produces oh to signal changes of state (Heritage 1984), which L2 speakers respond to appropriately. On the other hand, the L2 speakers often use ah, apparently influenced by its usage in their native language (Korean), which the L1 speaker also understands as indicating a change of state. While these discourse particles help achieve intersubjectivity, the study identifies moments of potential misalignment in orientations, as L2 speakers sometimes use ah in ways the L1 speaker does not fully grasp. The findings highlight the importance of the situated adaptability in multilingual interactions and the mutual responsibility to embrace interactional variations between L1 and L2 speakers.
- 1 August 2025
- Published by John Benjamins Publishing Company
Production and understanding of change‑of‑state tokens in English talk‑in‑interaction among L1 and L2 speakers
This study investigates the production and understanding of change-of-state tokens like oh and ah in English conversations among one L1 speaker and three L2 speakers. Using conversation analysis, it examines how these speakers do interactional work to achieve shared understanding through these tokens. The L1 speaker produces oh to signal changes of state (Heritage 1984), which L2 speakers respond to appropriately. On the other hand, the L2 speakers often use ah, apparently influenced by its usage in their native language (Korean), which the L1 speaker also understands as indicating a change of state. While these discourse particles help achieve intersubjectivity, the study identifies moments of potential misalignment in orientations, as L2 speakers sometimes use ah in ways the L1 speaker does not fully grasp. The findings highlight the importance of the situated adaptability in multilingual interactions and the mutual responsibility to embrace interactional variations between L1 and L2 speakers.