Using ChatGPT for second language writing: Experiences and perceptions of

EFL learners in Thailand and Vietnam

Introduction
The study investigates how EFL learners in Thailand and Vietnam perceive and use ChatGPT as a tool
to improve their second-language (L2) writing skills. Writing in English is a significant challenge for
these learners due to various factors such as limited vocabulary, poor grammar knowledge, and
insufficient exposure to diverse writing styles. This is evidenced by consistently low writing scores in
international proficiency tests like TOEFL and IELTS.

To address these challenges, the study highlights the potential of generative AI technologies like
ChatGPT. Following widely accepted L2 writing frameworks, ChatGPT can aid in brainstorming ideas,
organizing content, providing personalized feedback, and scaffolding complex writing tasks. The
tool's features align with educational theories such as constructivism, social constructivism, and
cognitive load theory, as well as frameworks like Paul Nation's Four Strands. For example, ChatGPT
offers meaning-focused input through generating relevant ideas and model texts, supports meaning-
focused output by simplifying tasks, and enhances fluency by acting as a collaborative writing
partner.

The study also notes specific applications of ChatGPT across different stages of the writing process,
such as idea generation, editing drafts, and proofreading. Evidence from other regions, like Hong
Kong, Singapore, and the Philippines, illustrates its effectiveness in assisting students to refine
outlines, improve drafts, and correct errors in language, structure, and style. By investigating the
experiences of Thai and Vietnamese learners, this study provides insights into how ChatGPT can help
overcome L2 writing difficulties and suggests implications for curriculum design and teacher training
in EFL contexts.

2.3. Perceptions of English language learners about ChatGPT

The article examines perceptions of EFL learners from different contexts about ChatGPT and the
factors influencing its adoption for second-language (L2) writing. In China, students appreciate
ChatGPT for its support in brainstorming, feedback, proofreading, and critical thinking but worry
about plagiarism and learning dependency. South Korean learners value its error correction but
suggest improvements in feedback detail. Indian students highlight its potential as a formative
feedback tool, while Vietnamese students focus on its usability for feedback and information
searching but question source reliability. Indonesian learners find it helpful in translating ideas but
express concerns over accuracy and integrity.
Factors influencing ChatGPT adoption vary across contexts. The Technology Acceptance Model (TAM)
emphasizes perceived usefulness and ease of use. In China and Malaysia, elements like hedonic
motivation and system quality play key roles, while Thai and Vietnamese learners prioritize attitudes
and user-friendliness, respectively.
The study employs a mixed-methods approach, combining surveys and interviews for a nuanced
understanding of perceptions and practices. Conducted in Thailand and Vietnam, it compares
learners from a Thai university, high school, and a Vietnamese university, representing diverse
educational settings. The research aims to explore how ChatGPT's potential and challenges shape its
use in EFL writing instruction, offering insights for curriculum design and teaching strategies.
The study further explores the factors influencing participants' perceptions of ChatGPT for L2 writing.
These factors are categorized into school-related, teacher-related, and student-related elements. The
researchers emphasized the significance of these dimensions in shaping how learners perceive and
utilize ChatGPT during the writing process.

The quantitative data from Table 1. highlights that learners found ChatGPT most beneficial during
the pre-writing stage, particularly for generating ideas and gathering necessary information.
However, lower ratings for post-writing tasks, such as reviewing and proofreading, reveal limitations.
For instance, while participants appreciated ChatGPT's capacity to generate ideas (4.17, Rank 1), they
expressed concerns regarding its feedback accuracy in the post-writing phase.
Qualitative data provides further insights. Students found ChatGPT effective in supporting
brainstorming and organizing ideas. For example, Thai students mentioned its utility in gathering
information and offering diverse perspectives. However, some participants noted that tools like
Grammarly were more precise in identifying errors during the revision phase.
This distinction between pre- and post-writing effectiveness reflects a nuanced perception: while
ChatGPT excels in idea generation and textual support, its automated feedback lacks the clarity some
learners prefer for error identification. This suggests that its application in writing could be
complementary rather than comprehensive, particularly in post-writing tasks. The document
examines the perceptions and usage of ChatGPT in second-language (L2) writing among learners,
combining quantitative and qualitative findings. Table 2.outlines the factors influencing these
perceptions, showing an overall mean score of 3.59 (Agree). Personal factors scored the highest, with
items such as the ease of integrating ChatGPT into writing tasks and its usefulness in enhancing skills
receiving mean scores of 4.18, 4.06, and 4.03, respectively. In contrast, institutional factors like
unclear guidelines (3.16), insufficient school facilities (3.02), and lack of adequate training (2.87) had
the lowest scores. This indicates that while learners find ChatGPT beneficial for personal skill
development, institutional challenges limit its effective application.
The qualitative data aligns with these findings. Minh from Vietnam emphasized how ChatGPT
provides comprehensive explanations, helping to understand various perspectives. Mali from
Thailand shared that it aids in grammar, spelling, and punctuation improvement. Similarly, Ngan from
Vietnam highlighted how ChatGPT reduces stress in the writing process. However, barriers such as
insufficient school support were also noted. Nam from Vietnam mentioned the lack of clear
guidelines for responsible ChatGPT use, Preecha from Thailand pointed out inadequate access to
resources, and Suda from Thailand expressed a need for more training on its effective use.
Table 3. presents the results of a one-way ANOVA, showing statistically significant differences in
perceptions among three groups: Vietnamese university students, Thai university students, and Thai
high school students. Vietnamese university students' perceptions significantly differed from the two
Thai groups, but there was no significant difference between the Thai university and high school
groups.
The document also discusses how learners utilize ChatGPT during different stages of the writing
process. In the pre-writing stage, learners use it to brainstorm, gather perspectives, and refine
outlines. For instance, Kanya from Thailand used ChatGPT to gain detailed insights and improve her
arguments, while Pim expanded her ideas with its suggestions. During the writing stage, ChatGPT is
employed to check linguistic accuracy, clarify vague concepts, and enrich the content. Mali from
Thailand used it to ensure coherence in her work, and Dao simplified complex terminology with its
help. In the post-writing stage, learners rely on ChatGPT to check grammar, style, and adherence to
academic conventions. Ngan from Vietnam used it to refine grammar and punctuation, and Somchai
from Thailand noted how it helped elevate vocabulary to a more sophisticated level. These findings
underscore ChatGPT’s value in the writing process while highlighting the need for better institutional
support to maximize its potential.
Conclusion

The findings of this study reveal that participants held positive perceptions of ChatGPT,
recognizing its value in enhancing their writing skills, subskills, and overall writing process.
ChatGPT was viewed as an effective tool for generating meaningful and polished written
outputs, with individual-level factors such as relevance, usefulness, accessibility,
engagement, and ease of use being the most influential. When comparing perceptions
across the three groups of EFL learners, statistically significant differences were observed,
particularly between participants from Vietnam and those from Thailand, underscoring the
influence of contextual factors such as culture and available resources on learners’ attitudes
toward ChatGPT.

Participants utilized ChatGPT extensively throughout various stages of the writing process,
including brainstorming, organizing ideas, refining outlines, and editing drafts for
appropriateness and accuracy. Beyond serving as a learning assistant, ChatGPT was
perceived as an active collaborator, reflecting the evolving concept of human-machine
collaboration in educational practices. These findings align with and provide evidence for
educational theories and frameworks, such as Paul Nation’s Four Strands, which advocate
for the integration of tools like ChatGPT into language learning.

This study highlights the need for deliberate integration of ChatGPT into L2 writing
instruction, ensuring alignment with established learning theories while addressing
individual, teacher, and institutional factors. Learner perceptions, existing practices, and
contextual elements should be considered when designing instructional strategies to create
an optimal learning environment where students can fully leverage ChatGPT for their writing
development. To support such integration, comprehensive training for teachers and
learners, alongside policies and resources at institutional and systemic levels, is essential.
These measures would enable sustainable and effective implementation of ChatGPT-
enhanced instruction, turning positive perceptions into tangible learning outcomes.

While this study contributes valuable insights to the growing body of literature on the use of
ChatGPT in L2 writing, it does have limitations in scope and methodology. Future research
could expand to include more diverse EFL contexts across the ASEAN region or different
educational levels. Additionally, further studies could examine the factors influencing learner
engagement with ChatGPT or its direct impact on L2 writing performance.