پیوندهای مفید
اخبار و اعلانات
آمار
| تعداد نشریات | 44 |
| تعداد شمارهها | 1,882 |
| تعداد مقالات | 15,307 |
| تعداد مشاهده مقاله | 44,020,799 |
| تعداد دریافت فایل اصل مقاله | 17,809,011 |
Who Wrote This Email? Topical Structure as a Predictor of Native, Non-Native, and AI-Generated Writing | ||
| Applied Research on English Language | ||
| مقالات آماده انتشار، پذیرفته شده، انتشار آنلاین از تاریخ 03 شهریور 1405 | ||
| نوع مقاله: Research Article | ||
| شناسه دیجیتال (DOI): 10.22108/are.2026.149879.2808 | ||
| نویسنده | ||
| Sahar Zahed Alavi* | ||
| University of Bojnord, Bojnord, Iran | ||
| چکیده | ||
| The emergence of artificial intelligence (AI) has raised important questions about how AI-generated texts differ from human writing at the discourse level. While previous research has primarily focused on lexical and syntactic features, little attention has been paid to topical progression as a discourse marker of organization. Following the framework of systemic functional linguistics (Halliday, 2009), this study investigated whether native, non-native, and AI-generated English business emails differ systematically in their topical progression patterns and whether these topical progressions can predict text origin. Four topical progression patterns (i.e., Parallel, Sequential, Extended Parallel, and Extended Sequential) were analyzed using Poisson generalized linear models controlling for text length, followed by multinomial logistic regression for text classification. The results revealed selective rather than uniform differences across the three text types. Native texts demonstrated significantly greater use of Extended Parallel topical progression, whereas AI-generated texts exhibited higher frequencies of Extended Sequential progression. Non-native texts differed from native texts primarily through a lower frequency of Parallel topical progression, while Sequential progression showed no significant differences across the groups. The multinomial model achieved an overall classification accuracy of 85.6%, indicating that topical progression frequencies provided strong discriminatory power. Thus, it could be an interpretable linguistic framework for distinguishing AI-generated, native, and non-native writing. | ||
| کلیدواژهها | ||
| Topical Structure Analysis؛ Business Emails؛ Generative AI؛ Authorship Prediction | ||
|
آمار تعداد مشاهده مقاله: 18 |
||