Success through Error: Using Error Analysis of ChatGPT Output in English as a Foreign Language Learner Writing Instruction
Saved in:
| Title: | Success through Error: Using Error Analysis of ChatGPT Output in English as a Foreign Language Learner Writing Instruction |
|---|---|
| Language: | English |
| Authors: | Arifi Wak (ORCID |
| Source: | International Society for Technology, Education, and Science. 2024. |
| Availability: | International Society for Technology, Education, and Science. 944 Maysey Drive, San Antonio, TX 78227. Tel: 515-294-1075; Fax: 515-294-1003; email: istesoffice@gmail.com; Web site: http://www.istes.org |
| Peer Reviewed: | Y |
| Page Count: | 13 |
| Publication Date: | 2024 |
| Document Type: | Reports - Research Speeches/Meeting Papers |
| Education Level: | Higher Education Postsecondary Education |
| Descriptors: | Artificial Intelligence, Computer Software, Second Language Learning, Second Language Instruction, English (Second Language), Word Frequency, Technology Integration, Writing Instruction, Error Patterns, Computational Linguistics, Arabic, Native Language, Standards, Language Tests, Language Proficiency, Writing Evaluation, Educational Benefits, Error Analysis (Language), Contrastive Linguistics, College Freshmen, Foreign Countries, English for Academic Purposes |
| Geographic Terms: | Saudi Arabia |
| Assessment and Survey Identifiers: | International English Language Testing System |
| Abstract: | Questions exist as to whether AI tools, such as ChatGPT, can aid learning. This study examined whether in-class exercises involving error detection in text generated by ChatGPT can aid students' foreign language writing. Participants were Arabic-English speakers who were classified as ranging from modest to competent English users according to IELTS standards. There were two conditions, each involving four sections of an English communication course, distributed across consecutive semesters and taught by the same instructor. In the treatment condition (n = 101), students performed error detection exercises using text generated by ChatGPT. In the control condition (n = 112), students performed the same exercises without ChatGPT. Then, they wrote a descriptive essay. It was hypothesized that if ChatGPT exercises enhanced students' confidence as writers, they would write longer essays. If ChatGPT exercises enhanced attention to detail, the exercises would increase writing quality (e.g., greater use of low-frequency and abstract words, and greater reliance on complex syntactic structures). In this study, students exposed to ChatGPT exercises used more low-frequency words and wrote more sentences but of shorter length than control condition students. The number of words did not differ, suggesting that ChatGPT exercises enhanced writers' attention to the shorter sentence length of the English language and its vast vocabulary. The results of this study suggest that ChatGPT exercises yield minor benefits for improving the writing of second-language speakers. Yet, extensive and broader exercises may yield more substantial benefits. [For the complete proceedings, see ED672804.] |
| Abstractor: | As Provided |
| Entry Date: | 2025 |
| Accession Number: | ED673080 |
| Database: | ERIC |
| Abstract: | Questions exist as to whether AI tools, such as ChatGPT, can aid learning. This study examined whether in-class exercises involving error detection in text generated by ChatGPT can aid students' foreign language writing. Participants were Arabic-English speakers who were classified as ranging from modest to competent English users according to IELTS standards. There were two conditions, each involving four sections of an English communication course, distributed across consecutive semesters and taught by the same instructor. In the treatment condition (n = 101), students performed error detection exercises using text generated by ChatGPT. In the control condition (n = 112), students performed the same exercises without ChatGPT. Then, they wrote a descriptive essay. It was hypothesized that if ChatGPT exercises enhanced students' confidence as writers, they would write longer essays. If ChatGPT exercises enhanced attention to detail, the exercises would increase writing quality (e.g., greater use of low-frequency and abstract words, and greater reliance on complex syntactic structures). In this study, students exposed to ChatGPT exercises used more low-frequency words and wrote more sentences but of shorter length than control condition students. The number of words did not differ, suggesting that ChatGPT exercises enhanced writers' attention to the shorter sentence length of the English language and its vast vocabulary. The results of this study suggest that ChatGPT exercises yield minor benefits for improving the writing of second-language speakers. Yet, extensive and broader exercises may yield more substantial benefits. [For the complete proceedings, see ED672804.] |
|---|
