การศึกษาเปรียบเทียบการใช้กลวิธีลดทอนความหนักแน่นและอคติของเอไอ ในบทความวิจัยด้านการสอนภาษาอังกฤษของชาวไทยและเจ้าของภาษา
Year: 2026
Type: article
Abstract: The purpose of this study was to investigate the rhetorical complexities of hedging within the Results and Discussion sections of English Language Teaching (ELT) research articles. Employing a qualitative comparative design, a corpus of 20 peer-reviewed articles (10 authored by Thai NNES and 10 native speakers) published between 2019 and 2024 was compiled. Data were extracted using AntConc software and manually coded according to Hyland’s (1998) typology of hedging, followed by a thematic analysis. The results revealed a significant pragmatic divide: Thai authors relied heavily on modal verbs (98 instances) for a reader-oriented approach, while native authors preferred limiting conditions (54 instances) to precisely bound their claims. Furthermore, the repetitive hedging patterns of Thai authors, characterized by lower lexical diversity, were observed to align with linguistic features known to trigger false positives in AI detection tools. These findings underscore the urgent need for pedagogical interventions to expand NNES scholars’ rhetorical repertoires and highlight the necessity for linguistically equitable AI detection methodologies.
Institution: Rajamangala University of Technology
Language: Thai
Subfield: Linguistics and Language
Field: Social Sciences
Domain: Social Sciences
Open Access status: diamond
OpenAlex ID: W7220598059
