Enhancing English as a Foreign Language (EFL) Learners’ Writing

with ChatGPT: A University-Level Course Design

Yu-Ching Tseng1 and Yi-Hsuan Lin2
1
Department of English, Tamkang University, Taiwan
2
Department of English Language and Literature, Chinese Culture University, Taiwan
[email protected] (corresponding author)
[email protected] (corresponding author)
Abstract: This research explores the innovative integration of OpenAI’s GPT-3.5 within a university-level English as a Foreign
Language (EFL) writing course, illustrating a novel approach to academic instruction. The course follows the ADDIE
instructional design model, encompassing five systematic stages: analysis, design, development, implementation, and
evaluation. This model serves as the backbone of the course structure, ensuring a comprehensive educational experience.
The incorporation of the Technological Pedagogical Content Knowledge (TPACK) framework in this course facilitates the
effective integration of GPT-3.5 by enabling instructors to align advanced AI capabilities with appropriate pedagogical
strategies, thereby enhancing the learning experience. TPACK guides educators in applying GPT-3.5’s features in a manner
that is contextually relevant and pedagogically sound, ensuring the technology’s use complements the course content. The
findings from this research are significant. They reveal that GPT-3.5 addresses three fundamental challenges often
encountered in academic writing courses. Firstly, it enhances efficiency by providing immediate feedback and generating
content ideas, accelerating the writing process. Secondly, GPT-3.5 ensures cohesive organization within students’ writing,
guiding them to structure their thoughts more logically. Lastly, it serves as a reliable substitute for traditional peer reviewers,
offering critical and objective feedback that students can use to refine their drafts. As students engage with AI, they enter a
dynamic partnership. This collaboration with GPT-3.5 fosters critical thinking and empowers students to develop a distinctive
writing voice. Through this interaction, students are not merely passive recipients of knowledge but active participants in a
learning process that is augmented by cutting-edge technology. This study not only provides insight into the potential of AI-
augmented academic writing but also highlights GPT-3.5’s role in promoting writing proficiency. It demonstrates that the
application of AI in education can enhance the learning experience without compromising the individuality of student
expression.
Keywords: ChatGPT, AI-assisted writing, English as a foreign language (EFL), TPACK, ADDIE

1. Introduction and Problem Statement

The struggle between ChatGPT and EFL writing is an intriguing encounter that brings to light the tension between
traditional writing practices and the innovative capabilities of AI. Mastering writing skills poses challenges for
EFL students, including difficulties in applying essay organization principles (Han and Huang, 2017), limited peer
review support (Harutyunyan and Poveda, 2018; Li and Huan, 2022), and time constraints (Sogutlu and Veliaj-
Ostrosi, 2022). In this context, the idea of translating organizational knowledge into practical application, as
discussed by Lee (2002), becomes particularly pertinent. It reflects the complexities involved in adopting new
technologies like ChatGPT in educational settings, which necessitate guidance and structured practice. While
peer review remains an essential aspect of the writing process, the difficulty of finding available reviewers
outside of class is a significant obstacle (Ahmed, 2019). Additionally, students struggle with time management
due to distractions. To overcome these challenges, ChatGPT emerges as a valuable AI tool, providing real-time
assistance in organizing ideas, structuring essays, and offering instant feedback, maximizing writing quality
within a limited time.
While ChatGPT has advantages, such as availability, efficiency, instantaneous response, absence of human bias,
and multilingual capabilities, it is essential to exercise caution in its utilization. Concerns regarding
misinformation, limited contextual understanding, lack of empathy, and various ethical issues must be
considered (Garg et al., 2023). The widespread adoption of ChatGPT in educational settings has generated
interest in its potential applications across various disciplines. For example, studies in data visualization and
medical education have explored ChatGPT’s capabilities and limitations in enhancing learning experiences and
knowledge assessment (Chen et al., 2023; Gilson et al., 2023; Skalidis et al., 2023; Tanaka et al., 2023). However,
academic research on its effective integration into EFL curricula, particularly in writing settings, remains limited.
Therefore, the current study aims to optimize the utilization of ChatGPT, enhance students’ writing experiences,
and equip them with valuable skills to effectively incorporate AI technology into their writing process. The study
applies the ADDIE model (Aldoobie, 2015) under the TPACK (Koehler and Mishra, 2008) framework to create a

ISSN 1479-4403 0 ©The Authors

Reference this paper: , Tseng, Y and Lin, Y. 2024. Enhancing English as a Foreign Language (EFL) Learners’ Writing with
ChatGPT: A University-Level Course Design, The Electronic Journal of e-Learning, pp 00-00, available online at www.ejel.org
The Electronic Journal of e-Learning SI AI Preview 2024

writing course that incorporates GPT to strike a balance that empowers students to unlock the benefits of AI
while preserving their independence and autonomy as writers.

2. Background and Literature Review

The integration of technology in education has opened numerous opportunities to enhance the learning process,
transforming the way English as a foreign language (EFL) is taught and learned.
2.1 Challenges in EFL Writing
In EFL education, technology integration is reshaping language classrooms, yet challenges persist, particularly in
the domain of writing. Despite ongoing efforts, achieving significant improvements in students’ abilities to
organize content, develop ideas, and master grammatical structures remains notably challenging (Chen, 2022).
A pivotal aspect of addressing these challenges lies in recognizing the crucial role of feedback within the learning
process. Pedagogical research in the EFL context underscores the profound significance of feedback in engaging
students and fostering revision practices (Zhang and Hyland, 2022). Whether originating from instructors, peers
(Ravand and Rasekh, 2011; Ruegg, 2015; Baker, 2016; Zhang, 2020; Zhang and Cheng, 2021), or automated
systems (Xu, 2019; Zhang and Cai, 2019; Lin et al., 2020; Alharbi, 2022), such feedback substantially enhances
writing skills (Li et al., 2010). However, the availability of constructive peer critique often remains limited in
many learning environments, contributing to a notable gap in learning outcomes.
In today’s digital landscape, students frequently navigate an array of information streams and distractions while
writing. Traditional EFL writing classes frequently lack sufficient time and avenues for students to express their
challenges or engage in meaningful idea exchange. To effectively address these issues, AI-driven technologies
like ChatGPT emerge as innovative solutions, offering personalized and interactive writing assistance that can
bridge existing gaps and provide students with valuable support in their writing endeavors.
2.2 Technological Advancements in EFL Instruction
ChatGPT extends beyond mere suggestions to generate human-like responses to user prompts, enabling
dynamic conversations with users. This unique feature not only provides more personalized writing assistance
but also serves as a virtual peer reviewer, fostering an environment conducive to continuous improvement in
writing.
GPT models’ potential extends beyond language learning, with Chen et al. (2023) recognizing their
transformative role in data visualization and their potential contributions to computer science education. These
capabilities have also been assessed in various medical licensing exams, such as the United States Medical
Licensing Examination (Gilson et al., 2023), the European Exam in Core Cardiology (Skalidis et al., 2023), and the
National Medical Licensing Examination in Japan (Tanaka et al., 2023), underscoring their significance in medical
education and knowledge assessment. However, despite these advancements, understanding AI technologies’
impact on students’ autonomous learning in writing remains a challenge, as pointed out by Bouwmeester et al.
(2019), who highlight the variability of students’ autonomy based on the technological tools they use. While
technology-mediated learning tools like ChatGPT have the potential to enhance autonomous learning, further
investigation is required to determine the full extent of this impact.
As the role of technology in education continues to advance, both instructors and learners must adapt.
Instructors need to adapt their teaching practices both inside and outside the classroom (Goldberg, 2003;
Golonka et al., 2014), and learners must also adjust their self-learning processes to accommodate the use of
different technological tools (Cancino and Panes, 2021). The shift is particularly salient in EFL writing, where
technological advancement has reshaped the writing process and influenced cognitive strategy development
(Cancino and Panes, 2021). It’s therefore crucial for educators and learners to comprehensively utilize these
technological advancements to enhance the EFL writing experience.
2.3 An Integration Model for ChatGPT in EFL Writing Courses
EFL students encounter difficulties in producing high-quality academic English writing, often facing challenges
such as compromising grammatical structures needed for clarity, using incorrect or simpler vocabulary, and
sacrificing organizational coherence. Therefore, it is essential to investigate how technology-mediated writing
can aid students in producing high-quality academic writing (Xu et al., 2019; Ammade et al, 2020; Gayed et al.,
2022). Aspects such as linguistic complexity, grammar, and organizational structure are crucial in evaluating the
quality of English writing (Xu et al., 2019).

www.ejel.org 0 ©The Authors

 Yu-Ching Tseng and Yi-Hsuan Lin

The integration of ChatGPT into EFL writing offers an innovative application of technology in English language
instruction. ChatGPT interacts with students’ writing, offering feedback on grammar, usage, mechanics, style,
organization, and content. Yet, despite ChatGPT’s revolutionary potential, the extent of its validity in the context
of EFL writing has not been sufficiently examined in prior studies. Future research is needed to ascertain the
true potential and impact of ChatGPT as an aid in EFL writing instruction.
To bridge this gap, this research paper employs the ADDIE model (Aldoobie, 2015) in conjunction with the TPACK
(Koehler and Mishra, 2008) framework, creating a systematic approach to integrate ChatGPT into a university-
level English writing course. The ADDIE model structures the process, facilitating the analysis of learning needs,
effective instructional design, strategy implementation, and evaluation of ChatGPT integration effectiveness
(Allen, 2017; Muruganantham, 2015). While ADDIE ensures consistency, its inflexibility in the digital age has
been noted (Bates, 2015). To address this, the study also incorporates the TPACK framework, which
accommodates evolving content and technologies, providing a strong foundation for designing effective
teaching and learning approaches. The TPACK framework ideally combines technological, pedagogical, and
content knowledge to guide the use of technology in writing instruction (Putri, 2019), ensuring that the
integration of AI technology is well planned and designed to meet the specific needs and challenges faced by
EFL students in their writing development, prioritizing learning first, followed by the technologies supporting
that learning (Schmidt et al., 2009). Accordingly, this study investigates two specific research questions:
Research Question 1. How can TPACK and ADDIE effectively integrate ChatGPT into EFL writing courses?
Research Question 2. What are students’ reflections on using ChatGPT in their writing process?

3. Method
This study employs a qualitative research methodology to explore the integration of the TPACK framework and
GPT-3.5 in an EFL writing course designed according to the ADDIE model. The qualitative nature of this study is
primarily focused on understanding students’ subjective experiences and perceptions regarding the use of AI in
their writing process. To collect data, the study utilizes two primary methods: analysis of students’ written works
and their reflective writings. These methods were chosen to gather rich, descriptive data about how students
interact with and perceive the integration of GPT-3.5 in their writing process. The course structure comprises
five sections that align with the ADDIE model: analyzing learning needs, designing instructions and prompts,
implementing ChatGPT strategies and techniques, evaluating the writing process, and assessing effectiveness
through student reflections.
3.1 Participants
This qualitative study is conducted within a university-level EFL writing course, English Composition III,
specifically targeting non-native English-major students at a prominent private university in Taiwan. Affiliated
with the Department of English at this university, the researcher has both designed and taught this course. The
study involved 15 participants (F=10, M=5) in academic year 2022-2023, all of whom were junior or senior
students majoring in English. All these participants were Taiwanese, non-native English speakers around the age
of 20-22. They were selected based on their academic progress and their current enrollment in the English
Composition III course. They had completed two years of prerequisite courses, English Composition I and II,
during their freshman and sophomore years, equipping them with a solid foundation in academic essay writing.
The participants willingly joined the research, giving consent through a signed form and understanding data
collection.
3.2 Procedure
The course design was framed under the Technological Pedagogical Content Knowledge (TPACK) framework (Fig.
1), which aligns with the Analysis, Design, Development, Implementation, and Evaluation (ADDIE) model. The
integration of these two models guided the design and implementation of a university-level English writing
course that incorporates GPT-3.5 as an assisting tool.

www.ejel.org 0 ISSN 1479-4403

The Electronic Journal of e-Learning SI AI Preview 2024

Technological
knowledge
(TK)

Technological Technological
Pedagogical Content
Knowledge Knowledge
(TPK) (TCK)
Technological
Pedagogical Content
Knowledge
(TPACK) Implement & Design & Analysis
Evaluation Development

Pedagogical
Content Content
Pedagogical Knowledge
knowledge knowledge
(PCK) (CK)
(PK)

Figure 1: The ADDIE-incorporating TPACK framework

The course design procedure begins with an understanding of three types of knowledge. These foundational
elements collectively form the basis of the Technological Pedagogical Content Knowledge (TPACK) framework.
• Technological Knowledge (TK): Comprehension of various technologies, including digital tools and
resources (Davis and Thompson, 2005). Within this course design, TK involves the use of GPT-3.5.
• Content Knowledge (CK): The instructor’s grasp of the subject matter to be learned or taught (Mishra
and Koehler, 2006). Here, it refers to the knowledge related to essay organization and writing skills.
• Pedagogical Knowledge (PK): Understanding methods and strategies for teaching and learning
(Shulman, 1986). In this course, PK includes integrating AI-assisted peer reviews and utilizing
interactive exercises with GPT-3.5 prompts to enhance students’ essay organization and writing skills.
3.3 Analysis Phase
The analysis phase examines students’ writing skills, needs, and areas for improvement in terms of CK and PK in
writing. The primary objective is to gain insights into the specific challenges students face during the writing
process. Through a combination of instructor evaluations and self-assessment, students’ strengths and
weaknesses are identified, enabling the effective integration of GPT-3.5 to enhance their learning experiences
in the most-needed areas. During this phase, the instructor reviews the basic components and principles of
organization in academic writing to ensure that all students have a solid foundation in composing well-
structured academic essays (Almelhi, 2021). The students engage in various writing exercises and assignments
without AI assistance. The instructor gains an understanding of the student’s writing abilities, which serves as a
basis for designing the subsequent stages of the course. The instructor assesses students’ needs (West, 1994;
Richards and Rodgers, 2001; Richards, 2013), identifying how GPT-3.5 can be integrated as a valuable tool to
enhance their writing skills.
3.4 Design/Development Phases
In the design and development phases, the course is structured to integrate GPT-3.5 into the students’ writing
process, incorporating technological pedagogical knowledge (TPK), technological content knowledge (TCK), and
pedagogical content knowledge (PCK). TPK involves the strategic integration of technology tools like GPT-3.5
into writing instruction, aligned with effective pedagogical approaches. TCK focuses on the instructor’s
understanding of how AI tools could enhance specific content knowledge related to writing skills, while PCK
guides the instructor in designing instructional strategies that effectively merge content and pedagogical
knowledge with technology integration. In the design phase, the instructor develops a framework outlining how
GPT-3.5 will be incorporated effectively into the course, selecting appropriate writing topics based on insights
from the analysis phase and ensuring alignment with students’ needs. The development phase involves creating
and refining course materials to facilitate GPT-3.5 integration, including instructions on formulating prompts,

www.ejel.org 0 ©The Authors

 Yu-Ching Tseng and Yi-Hsuan Lin

interacting with GPT-3.5 to refine prompts and maintain control over writing organization, and revising writing
based on GPT-3.5’s assistance.
3.5 Implementation Phase
During the implementation phase, the integrated knowledge of TPK, TCK, and PCK is applied to effectively
integrate AI tools like GPT-3.5 in the writing instruction. The instructor monitors students’ engagement with AI
integration and provides guidance to ensure a productive experience. Students actively engage with GPT-3.5 as
a writing tool, integrating AI assistance into their writing process by applying the instructions and techniques
acquired throughout the course to their writing practice. This phase involves utilizing GPT-3.5 to assist in
developing and improving the overall quality of their written works.
3.6 Evaluation Phase
In the evaluation phase, the study assesses the effect and impact of integrating TPK, TCK, and PCK to form TPACK
in the EFL writing course. This evaluation process encompasses both objective measures, such as improved
grammar and coherence in written work, and subjective measures, including students’ reflective writing.
Through student reflections, the instructor gains valuable insights into struggles and challenges faced during the
AI integration process, enabling them to gauge the effectiveness of AI integration and make informed
adjustments to optimize the learning process. In this phase, students engage in self-evaluation, reflecting on
their growth as writers and technology’s role in the writing process. As part of their final exam, students write a
reflective essay describing their development as writers while considering the impact of technology. This self-
reflective exercise deepens their understanding of AI tools like GPT-3.5 in writing education and raises
awareness of the potential risks associated with overreliance on AI during writing.

4. Results
RQ1: Developing an AI-Integrating Writing Course
Throughout the course, the researcher employs the ADDIE model as a systematic framework to integrate TK,
CK, and PK into TPACK (Fig. 2).

ADDIE Course Plan TPACK in writing instruction

Analysis Phase The essay writing process involves teaching essay The process involves assessing students'
organization principles, analyzing examples, needs and determining the optimal use of
brainstorming, drafting, peer review, instructor technology (GPT-3.5) to enhance their
feedback, and revisions, culminating in a well- learning experiences within the course.
crafted essay that reflects significant effort.

Design & The design and development phase integrates The process integrates TK with PK and
Development GPT-3.5 into the writing curriculum, offering CK to teach students how to develop
Phase scaffolding techniques for the "self-introduction" prompts and use them step-by-step to
and "revision" topics. enhance the organization of their writing.

Implement Phase During the Implement phase, students utilize The successful implementation of
GPT-3.5 as an assisting tool in the writing TPACK empowers students to achieve
process, combining their knowledge of essay improved writing skills through effective
organization principles with AI support to draft, utilization of technology.
revise, and refine their essays.

Evaluation Phase The evaluation phase involves In the evaluation phase, TPACK
assessing the impact of AI integration informs the development of
integration on students' writing AI-enhanced writing curricula for future
performance and their reflections courses.
on the learning experience.

Figure 2: ADDIE model and TPACK writing instruction

4.1 A Need for Revision
During the analysis phase, students’ academic writing requires extensive revisions. Traditional essay writing
methods, such as structured outlines, handwritten drafts, and standard peer reviews, while valuable, come with
limitations such as time constraints and a lack of personalized attention. Ideally, students would receive
feedback from knowledgeable reviewers, but such resources are often scarce. ChatGPT offers a promising
solution to these challenges. It provides immediate feedback and constructive suggestions, enabling students to
refine their work continuously without time restrictions. It serves as a virtual reviewer, offering personalized

www.ejel.org 0 ISSN 1479-4403

The Electronic Journal of e-Learning SI AI Preview 2024

guidance and effectively increasing the available reviewing resources. This integration augments traditional
writing methods and empowers students to improve their writing.
Without the support of AI technology, the students’ essays are examined by the two researchers before and
after the review process. Figure 3 presents the improvements made in terms of grammar, coherence, and clarity
following the review. However, some lingering issues in the revised versions are highlighted in red, possibly
requiring additional revision.

V1 (Before Review) V2 (After Review) Advancements

For the purpose of living a better life, For the purpose of living a better life, most V2 improves grammar and clarity
most parents will ask their children to parents will ask their children to concentrate on compared to V1 by using a more concise
concentrate on their studies whatever their studies wherever they live. and grammatically correct sentence,
they inhabit in which area. stating that parents encourage their
children to concentrate on studies
regardless of where they live, aiming for
a better life.

First, both Mac and Windows offer First, they offer visually appealing appearances. V2 significantly enhances grammar,
visually appealing designs. However, Mac's users are always attracted by its sleek clarity, and coherence by using concise
Mac is known for its sleek and appearances. It is known for its one-piece language, providing specific details
consistent appearance. On the other alloy computer body and the minimalist about Mac's sleek appearance, and
hand, Windows provides a more design. On the other hand, Windows provides presenting a better flow of ideas
customizable and versatile interface. several kinds of visually appearances. compared to Version 1, which had some
Customers can find their favorite one in the repetition and less descriptive language.
numerous Windows computers.

First, If You constantly buy products First, shopping becomes an addiction if you V2 demonstrates improved grammar,
you don't need, the purchase is constantly buy products you don't need, and clarity, and coherence with its well-
typically followed by negative the purchase is typically followed by negative structured and concise explanation,
feelings such as guilt or regret, feelings such as guilt or regret. The two main presenting shopping as an addiction
advertisement and marketing reasons for this problem are the producer's when constantly buying unnecessary
strategies are pushing the new marketing strategies and the consumers’ products, followed by negative
products to cause this problem. It never-ending desire to collect new products. emotions like guilt or regret, primarily
will let us immerse in those Exaggerated advertising methods can make influenced by producers' marketing
platforms. what's more, many people consumers buy impulse purchases and buy a lot strategies and consumers' desire for
will buy something new that doesn't of unnecessary items. Shopping addiction will new items.
useful, it is a very to gratify our result in wasting money on merchandise
vanity. Shopping becomes an which we don't need.
addiction, impulse shopping will
result in wasting money on the
merchandise which we didn't really
need.

Figure 3: Example of student’s writing before and after review

4.2 A Scaffolding Approach to Developing Prompts
The design and development phase employs a scaffolding strategy to enhance students’ writing skills through
the focused tasks of self-introduction (autobiography) and revision. The autobiography task stresses both
organization and authenticity. It encourages students to structure their narratives logically while expressing their
unique identities and experiences. This task serves a practical purpose, preparing students for real-world
situations where they need to introduce themselves in a compelling and organized manner. Meanwhile, the
revision task utilizes GPT-3.5 as a substitute for traditional peer reviewers. It releases teachers and peers from
the reviewing role and capitalizes on the AI’s capacity to provide immediate feedback, allowing students to
revise and refine their work autonomously. Both tasks help students improve writing quality and efficiency,
providing a balanced approach to mastering different facets of the writing process. The two tasks break down
the writing process into manageable steps.
4.3 Self-Introduction
The objective of the task is to develop students’ skills in utilizing GPT for organization while preserving their
individuality and originality. In the task, students begin by providing a list of their advantages and strengths.
Using prompts, students direct GPT-3.5 to follow their planned organization, facilitating a step-by-step
transformation of their points into well-structured essays (Fig. 4).

www.ejel.org 0 ©The Authors

 Yu-Ching Tseng and Yi-Hsuan Lin

Figure 4: A scaffolding approach for self-introduction

• The instructor begins by providing instruction on the basic structure of a self-introduction, offering
students a solid foundation for their writing.
• Building upon this knowledge, students proceed to list their advantages, identifying their unique
strengths and attributes (Fig. 5).

Figure 5: Student’s advantage list

• Students communicate these key points to GPT-3.5, ensuring that the AI model is informed about the
purpose of the autobiography and the intended content for each paragraph (Fig. 6).

www.ejel.org 0 ISSN 1479-4403

The Electronic Journal of e-Learning SI AI Preview 2024

I am applying for a master's program of (PROGRAM NAME) in (UNIVERSITY NAME),

This is a list of my advantages; write a seif-introduction for me: (PROVIDING THE LIST).
In the first paragraph, state the reason why I want to be a teacher and summarize my strength
with a strong hook sentence. In the second paragraph, summarize my past experiences in academic
learning. In the third paragraph, focus on my research ability. In the fourth paragraph, focus on my
teaching ability. In the fifth paragraph, show my potential to be an excellent student in
the program. Finally, close the self-introduction with a strong statement.

Figure 6: Prompt for the first draft

• Students revise the details and derive the first draft of their self-introduction. This step allows them
to fine-tune their writing, ensuring that their ideas are well-articulated and coherent.
• Students employ the strategic use of the STAR (situation, task, action, and result) method (Cook, 2009)
for reorganizing first drafts and the SWOT (strengths, weaknesses, opportunities, and threats)
technique (Misbah and Mahboob, 2017) for crafting final paragraphs, enhancing the quality and
impact of their self-introductions (Fig. 7). The integration of GPT-3.5 further enhances the
effectiveness of these techniques because GPT-3.5 can understand the prompts, analyze the original
text, and follow the instruction to accurately generate text based on these methods, leading to more
impactful self-introductions.

Use the STAR method to describe my experiences in each

paragraph of my self-introduction. Don't make a list; write it
into a cohesive piece. Mark (S, T, A, R) after every single
sentence.

Use SWOT to rewrite the final paragraph. Don't make a list.

Write a cohesive piece.

Figure 7: Prompts for using STAR and SWOT

The following examples show the contrasting organization of paragraphs using the STAR method (Fig. 8).
Applying this method ensures a systematic and engaging presentation of the applicant’s situation, task, actions,
and results, effectively highlighting strengths and capturing readers’ attention.

Figure 8: Contrast with/without STAR

www.ejel.org 0 ©The Authors

 Yu-Ching Tseng and Yi-Hsuan Lin

The two concluding paragraphs below (Fig. 9) exemplify how the SWOT method shapes the structure.
Implementing this method offers a comprehensive assessment of strengths, weaknesses, opportunities, and
threats, leading to a more balanced and insightful portrayal of the applicant’s qualifications.

Figure 9: Contrast with/without SWOT

In the “self-introduction” assignment, students progressively fine-tune their ideas into coherently structured
essays, utilizing GPT-3.5 to follow their outlined organization. They adopt both STAR and SWOT techniques to
restructure their initial drafts and craft compelling conclusions. These systematic methods not only improve
their organizational skills but also enhance the persuasive power of their writing.
4.4 Revision
The “revision” task is designed to bolster students’ skills in employing GPT as a peer reviewer, simultaneously
enhancing their capacity for self-evaluation. The task aims to equip students with a deeper understanding of
their writing problems, fostering their ability to identify these issues independently.
In the “revision” task, the scaffolding approach helps students transition from an existing article to outlining
essential points for improvement, allowing them to focus on specific aspects of their writing.

Figure 10: A scaffolding approach to revision

www.ejel.org 0 ISSN 1479-4403

The Electronic Journal of e-Learning SI AI Preview 2024

The proofreading process employs systematic step-by-step scaffolding (Fig. 10), ensuring a thorough review of
the manuscript. The step-by-step scaffolding ensures that each stage of the proofreading process builds upon
the previous one. By following this structured approach, the proofreader can effectively address grammatical
errors, enhance clarity and coherence, and deliver a well-structured and logically presented composition:
• As shown in Figure 11, students establish GPT-3.5’s role as a proofreader for non-native English
compositions, involving the review of manuscripts submitted by Taiwanese college students majoring
in English.
• The original text is carefully examined to identify grammatical errors. This step focuses specifically on
grammatical errors, avoiding distractions that may otherwise arise during the process. The grammar
check serves as the foundation for improving the manuscript’s clarity, laying the groundwork for the
following steps.
• Each paragraph is scrutinized to ensure a clear and coherent structure. Checking the overall structure
ensures the manuscript is well organized. Examining the presence of clear topic sentences and well-
supported arguments helps to establish a coherent and logical flow of ideas.
• The application of general-to-specific organization further strengthens the manuscript’s coherence.
Applying general-to-specific organization allows the main points to be presented logically and
cohesively.
• Crafting a closing statement offers a concise summary of the main points discussed throughout the
article.

Figure 11: Prompts for revision

It is crucial to note that a table is created in each step to list all the identified issues along with their specific
locations in the text, providing a clear overview for students to address each error systematically (Fig. 12). This
table presentation makes it easier to pinpoint areas requiring improvement, serving as a valuable tool for
students to gain a deeper understanding of their writing issues. Visually displaying errors and their specific
locations equips students with a clearer picture of the weak spots in their writing. Thus, the table presentation

www.ejel.org 0 ©The Authors

 Yu-Ching Tseng and Yi-Hsuan Lin

not only aids in correcting current mistakes but also contributes to the development of students’ self-
assessment abilities, fostering a deeper understanding and awareness of their writing strengths and
weaknesses. The table below, generated by GPT-3.5, exemplifies this approach:

Figure 12: An example generated by ChatGPT

With GPT-3.5 acting as a virtual proofreader in the revising process, students can carefully examine grammatical
errors, improve paragraph coherence, and enhance overall organization. This innovative approach reduces the
need for extensive peer review and the instructor’s traditional role in the proofreading process.
4.5 Integration of AI into the Writing Process
During the implementation phase, students engage with GPT-3.5 in two tasks. In the “self-introduction” task,
they employ GPT-3.5 to enhance essay structure. Students provide specific prompts, guiding the AI to craft
autobiographies with distinct elements, such as engaging hooks and compelling conclusions (Fig. 13). This skill
enables students to retain their autonomy while utilizing AI’s assistance, ensuring the generated content reflects
their individual voices and preferences.

Write an autobiography with six paragraphs based on the experiences I had. The autobiography needs
to be natural which means that others can’t tell it’s written by AI. Start with an attraction hook in
the first paragraph. It should include my academic score, my research experiences, and my teaching
experiences. The second paragraph focuses on my academic learning. How it benefits my future
learning in XXX. The third paragraph focuses on my research experiences. The fourth paragraph
focuses on my teaching experiences. The fifth paragraph focuses on describing my personality which
includes self-disciplines, leadership, and being willing to help others. Show my strong potential of being
a graduate student in XXX. Finally, close the autobiography impressively.

Figure 13: Student’s prompt 1

www.ejel.org 0 ISSN 1479-4403

The Electronic Journal of e-Learning SI AI Preview 2024

As the process unfolds, students refine each paragraph through follow-up prompts (Fig. 14). Through these
prompts, they demonstrate critical thinking and authorship, directing GPT-3.5 to produce an autobiography that
genuinely captures their unique experiences.

Revise the second paragraph. Add what I have learned in the courses and how I
maintain my good grades.
Revise the third paragraph. Add what I have learned in the research. Connect it with
my expected teaching strategy “CALL.”
Revise the fourth paragraph. Add what I have learned in the teaching experiences.
How I apply what I have learned in academic learning and the research I did in my
teaching.

Figure 14: Student’s prompt 2

Students use GPT-3.5 as a tool to refine and optimize their writing. For instance, they might initially instruct GPT-
3.5 to rewrite a self-introduction using specific techniques. If the generated content doesn’t meet expectations,
they refine their prompts for a more satisfactory output. The iterative process fosters effective communication
with the AI. Figure 15 illustrates their intention to communicate more effectively with GPT-3.5 to receive the
expected result on the first attempt, reflecting their adaptive approach to utilizing AI technology in the writing
process.

PLEASE Rewrite the self-introduction. Use STAR to

strengthen the act and result in each paragraph. Use
SWOT to close the article. Make it less than 2000 words.

I don't want a list, I want something like "Tamkang

Times.... (Situation) Hone writing skills.... (Task)
Wrote 100 news... (Action) Won Prizes... (Result)"

If I want you to do these at first, what signal you could

understand?

Figure 15: Student’s prompt 3

Throughout the task, students evaluate and potentially disagree with GPT-3.5’s outputs, emphasizing the
importance of critical thinking and independent decision-making in AI-assisted writing. The examples below (Fig.
16) highlight the collaborative nature of the human-AI interaction in the writing process, where students actively
engage with GPT-3.5 to refine their compositions and uphold their unique voices.
“In addition, I had the privilege of participating in a Graduate
Speech program, which sharpened my public speaking and
Nobody would use "in conclusion" in the first paragraph. presentation skills.” This paragraph is incorrect.
“Lastly, I have experience in competing for results in a college
degree program.” This sentence is wrong.

In the last paragraph, don't use "to conclude'' at the beginning and
make it sound more natural. Add more benefits that the company In fourth paragraph, I don’t fluent in Spanish, can your revise
will get if they give me the position. that?

It is a little bit off the topic. That activity is a two-day activity

and it is minor. However, when I face an urgent situation, I Can you replace “proud” in the first paragraph?
can resolve it with a clam attitude.

Make it sound literal Can you be simpler?

The fifth paragraph needs an example: I had helped Understanding others’ needs should be written
visually impaired students to go to their classrooms. with summer club experience.

Figure 16: Student’s prompt 4

www.ejel.org 0 ©The Authors

 Yu-Ching Tseng and Yi-Hsuan Lin

In the revision task, students skillfully utilize GPT-3.5 as a tool to optimize their writing. By crafting personalized
prompts that effectively guide the AI, they instigate considerable improvements in their compositions. This is
manifested in heightened coherence, enriched vocabulary usage, and more refined sentence structures (Fig.
17). The process not only results in enhanced text quality but also augments their own understanding and
mastery of effective writing techniques.
V1 (Without GPT) V2 (With GPT) Advancements
The number of teachers is one of the A notable disparity between urban and rural Grammar:
obvious gaps between urban and schools is evident in the number of teachers In V1, the phrase "only needs to teach Chinese
rural schools. In big cities, every available. Urban schools benefit from a well" is grammatically correct but could be more
teacher in the school can perform sufficient number of specialized educators, with concise and specific. In V2, the sentence is
their own duties. The Chinese teacher each teacher focusing on their specific subject rephrased to include "each teacher focusing on
only needs to teach Chinese well and area. In contrast, rural schools often face the their specific subject area," providing a clear and
does not have to worry about other challenge of one teacher handling multiple concise description of specialized educators in
subjects. However, it is very common subjects due to limited resources. urban schools.
that one teacher wears more than one Additionally, high turnover rates and the Clarity:
hat in the country schools. For constant adaptation to new substitute teachers In V1, the phrase "one teacher wears more than
instance, an English teacher teaches further hinder the learning experience for rural one hat" is somewhat vague and could be clearer.
not only English but also civics students. In V2, the sentence is rephrased to specifically
because there are no other teachers. state that one teacher handles multiple subjects in
Another situation is that even though rural schools due to limited resources, providing a
there are substitute teachers for other clearer understanding of the situation.
subjects, there is high teacher Coherence:
turnover rate, and students have to In V1, the sentence includes multiple points
constantly adapt to new substitute without clear connections, making it less
teachers. Either way, both can coherent. In V2, the sentence is rephrased to
adversely affect students learning. present the related ideas in a more cohesive
manner, emphasizing how high teacher turnover
rates and constant adaptation to new substitute
teachers impact the learning experience for rural
students.

User continuity is one of the places Another area where Mac surpasses Windows is Grammar:
where Mac beats Windows. Apple user continuity. Mac provides seamless In V1, there is a grammar error in the phrase
has the whole Apple family (Apple integration across its devices, including Apple "all they need to do is airdropping," which
Watch, Iphone, Mac) as support and Watch, iPhone, and Mac, through built-in should be "all they need to do is use AirDrop."
all they need to do is airdropping apps like AirDrop. This built-in integration In V2, the sentence is rephrased with proper
which is a built-in App that you can system enhances productivity and grammar, using "through built-in apps like
exchange data between different convenience, allowing users to effortlessly AirDrop" to describe the seamless integration
Apple devices. On the other hand, exchange data between their Apple devices. across Mac devices.
when Windows users wish to In contrast, Windows users need to download Clarity:
exchange data from other devices additional applications to achieve similar In V1, the statement "Mac users may leverage
with their computers, Windows users functionality, resulting in a less streamlined the built-in integration system" lacks
need to download extra applications. user experience. Mac users can leverage this specificity in describing the advantages of the
In short, Mac users may leverage the built-in integration system to boost job integration system. In V2, the sentence is
built-in integration system to boost productivity and make their lives easier. rephrased to explicitly state that "This built-in
job productivity and make their life Therefore, when it comes to user continuity, integration system enhances productivity and
easier. Thus, when it comes to user Macs should be the preferred choice without convenience," providing a clearer explanation
continuity, Macs should be your only any hesitation. of the benefits for users.
choice without any considerations. Coherence:
In V1, the transition word "Thus" does not
create a strong logical connection between the
previous and following sentences. In V2, the
transition word "Therefore" is used, which
creates a more coherent flow and reinforces the
conclusion that Macs should be the preferred
choice for user continuity.

Figure 17: Comparison of students’ writing before and after applying ChatGPT
The successful integration of GPT-3.5 into the writing process exemplifies the effective combination of TK, PK,
and CK within the context of TPACK. In the implementation phase, students skillfully employ TK by crafting
personalized prompts and refining instructions to direct GPT-3.5 in generating desired content. Their adept use
of PK is evident in their communication with the AI, seeking clarifications and adjustments to maintain their
autonomy as writers and effectively follow the instructor’s step-by-step scaffolding approach. As students
integrate GPT-3.5 into their writing process, they apply their pedagogical knowledge to understand and
implement the instructor’s guidance, which includes crafting personalized prompts, refining instructions, and
seeking clarifications for effective communication with the AI. Additionally, their CK is showcased as they
structure written work and ensure the intended focus for coherent compositions. This harmonious integration
of TK, PK, and CK results in significant improvements in their writing, demonstrating the successful application
of TPACK principles to enhance writing instructions.
RQ2: Exploring Students’ Complex Perceptions of AI in Writing
In the evaluation phase, students reflect on their experiences with integrating GPT-3.5 into their writing process
through an introspective essay. Students also delve into their emotions, expressing their reservations about AI’s
efficiency. This introspective process empowers students to become thoughtful writers, fostering a harmonious
relationship between human creativity and AI assistance in the writing process.
4.6 GPT Makes Writing Efficient
In the evaluation phase, students recognized GPT-3.5’s efficiency, noting its role in time-saving and simplifying
the writing process. They reported that GPT-3.5’s rapid generation of information expedited their research of
points, and its capabilities for refining compositions and structuring pieces saved them considerable time.

www.ejel.org 0 ISSN 1479-4403

The Electronic Journal of e-Learning SI AI Preview 2024

The most impressive part of ChatGPT is that it can give us a lot of information in a short time. From
Taiwan to the whole country, all things can be found in it.
Technology has influenced my writing in several ways. ChatGPT has provided me with a wide range of
information. Researching and finding many data has become much easier … with just a few clicks.
The students’ reflections confirmed that GPT-3.5’s time-saving efficiency was critical to their improved writing
process. This newfound efficiency not only sped up their writing but also empowered them to maintain a high
quality of work within the time constraints. While this newfound efficiency accelerated their writing, it was
crucial to maintain a balance between using AI tools and developing their own writing abilities. The course was
designed not to create dependency on AI but to use it as a supplement to enhance students’ writing skills and
critical thinking.
At the beginning of this semester, I was a student who wrote an article very slowly. … After trying this
tool, I found that my writing efficiency improved. I can just input my points and ask for recombination.
Also, after finishing my essays, I can feed my writing drafts to ask for some suggestions of reversion.
Throughout the course, I have witnessed significant development in my writing abilities. … I now spend
only 30 minutes on an essay compared to the 2-3 hours I used to spend before. These tools have
provided creative suggestions and fresh ideas, enabling me to develop my writing more efficiently and
achieve better results.
At the beginning of this course, I spent countless hours on homework and writing. However, as the course
progressed, I realized the impact that technology brings to the writing process. The accessibility of
information, editing, and the ability to connect information have all expanded my perspectives and
my writing ability.
The reflections consistently highlight enhanced efficiency as GPT-3.5 accelerates writing, optimizes research,
and streamlines editing, revolutionizing the writing experience. This transformative AI efficiency becomes
invaluable, maximizing writing quality within limited time frames.
4.7 GPT Makes Organization Easier
In the evaluation phase, students’ reflective essays reveal how GPT-3.5 serves as a valuable aid in organizing
their essays, ensuring coherence and logical flow throughout the text. The integration of GPT-3.5 revolutionizes
their writing approach.
Students stated how GPT-3.5 assisted them in generating well-structured essays by providing accurate and
relevant responses to their prompts, thus improving the overall quality of their writing.
Take my own experiences as an example. I used ChatGPT to help me write my personal statement. I
believe that what ChatGPT wrote is better than what I wrote because the prompts I gave it are
accurate.
Traditionally, people have to think about everything when writing. For instance, wording, grammar,
structure, etc. However, with the help of ChatGPT, people can produce an essay simply by giving
instructions.
Beyond its immediate benefits, students found that AI’s ability to handle the mechanical aspects of writing freed
them to focus on developing cogent arguments and thoughtful organization. The use of AI effectively shifted
their focus from the granular task of writing to the higher-level task of reasoning and structuring. This delegation
of labor to GPT-3.5, a proficient tool for grammar and structure, allowed students to fine-tune their arguments,
fostering their critical thinking skills.
It is no more necessary to create an article word by word on one’s own. Instead, people should pay
attention to the organization, which might be one of the points that humans could perform better
than Al.
Technology lets us pay more attention to our thinking logic. …What those digital tools are really good
at is the structure, formulaic writing performance. We don’t need to worry about if the essays we wrote
by using online tools have huge grammatical errors. We can use them to make double confirmation on
structure and grammar in our writing works.

www.ejel.org 0 ©The Authors

 Yu-Ching Tseng and Yi-Hsuan Lin

Throughout their reflective essays, students highlighted the transformation in their writing process, where GPT-
3.5 and other digital writing tools acted as indispensable partners in ensuring coherence and logical flow in their
compositions. By taking on the organizational aspects, these AI tools allowed students to explore the depth of
their ideas and strengthen their argumentation, leading to enhanced writing skills and more effective
communication.
4.8 GPT Replaces Peer Reviewers
In the evaluation phase, the use of GPT-3.5 as an effective tool for immediate feedback and revision became a
recurring theme in students’ reflective essays. The AI not only detected errors but also facilitated a
comprehensive understanding of their writing problems. By delivering a clear overview of errors and
suggestions, GPT-3.5 enabled students to gain a deeper insight into their areas for improvement.
In the early stages, word processors and spelling and grammar checkers provided immediate feedback,
helping me refine my work. These digital tools not only corrected errors but also highlighted areas
where I could improve clarity, coherence, and conciseness. Additionally, access to online resources
allowed me to delve deeper into research, broadening my understanding and enriching the depth of my
writing.
Al helped me improve my writing skills because it can detect my problems in the essay, for example,
my coherence problems, or gives better word usage. It can also review my paragraph and tell me the
problem I had and the ways to revise it … and you can see its difference with the essay you wrote so
that you can understand the problems in your grammar. It can list all the errors for you with one click.
Students also mentioned how the integration of GPT-3.5 had reshaped their editing process, making it more
collaborative and efficient. This transformation underscored GPT-3.5’s role in providing critical writing support
and guidance, especially for students without immediate access to extensive writing resources.
With the advent of digital tools, the revision and editing process has become more efficient and effective.
Features like track changes and comments allow for collaborative editing, enabling me to receive
feedback from peers and instructors seamlessly.
AI can serve as an accessible learning tool, offering writing guidance and resources to individuals who
may not have immediate access to extensive writing support. It can provide real-time assistance and
feedback, allowing students to independently improve their writing skills, especially in cases where
immediate outside resources may not be readily available.
Throughout their reflections, students emphasized how GPT-3.5’s language checks and organizational support
enhanced the writing process, providing valuable feedback and suggestions similar to peer reviewers.
4.9 Perplexity toward GPT
In reflective essays, students highlighted their growth as thoughtful writers and the complexities they faced
when integrating AI into their writing process. They recognized both the efficiency of AI and the importance of
maintaining their authorial identity. While a few students’ reflections have been highlighted for their illustrative
value, these perspectives are representative of broader sentiments expressed by the group.
One student recounted using ChatGPT for a presentation, yet the AI fell short in crafting a satisfactory
conclusion. The student realized the need for human input and the importance of critical thinking.
I use ChatGPT to prepare my presentation. … I also ask the ChatGPT to come to a conclusion for me.
Unfortunately, on the presentation day, the conclusion is not as good as the summary and analysis. …
ChatGPT is good, but never totally rely on it. You are the master of it. I still need to think critically on
my own. That’s the importance of critical thinking ability for everyone.
Another student nostalgically looked back on the past when writing was done using a dictionary and pen before
the advent of technology. The nostalgic sentiment underscored the emotional depth and personal connection
inherent in traditional writing, acknowledging the trade-offs between traditional and technologically assisted
methods.
Look back at the past, I appreciated that I was born in a changing era. When I first started writing, the
only thing I had was a dictionary. No ChatGPT, no Joplin, no Google translation but my pen. I still miss
the feeling of writing on a tight new white paper without any limits. It was the place where I could share
my feelings and stories and basically, another me in another chapter.

www.ejel.org 0 ISSN 1479-4403

The Electronic Journal of e-Learning SI AI Preview 2024

In navigating the balance between AI’s convenience and maintaining their individual voice, students grappled
with ethical considerations. One student questioned whether it is reasonable to consider AI-generated
sentences as their own writing. This internal debate demonstrated their sensitivity to issues of originality and
artistic expression. Their thoughtful consideration of this ethical dilemma showcased their growth as
conscientious writers willing to navigate the complexities of using AI technology while maintaining their
authentic integrity as writers.
Nonetheless, the question still remains at the bottom of my heart. However good the sentences AI
generates, is it reasonable to just mark them as the users’ writings? As a writer, I highly disagree with
this. But the fact that AI is an effective reversion tool is doubtless.
These individual reflections are part of a larger group of student experiences. The collective insights indicate
that while students recognized the practical advantages of AI, they also remained critically aware of the
importance of maintaining their authorial identity and the need for thoughtful engagement with technology.
Throughout their reflections, students experienced a transformation in their perceptions of AI’s role in writing.
They recognized the advantages of AI in content generation and revision while simultaneously acknowledging
the importance of their own critical thinking and creativity. Their complex understanding of AI’s potential and
limitations showcases the multifaceted nature of their experiences, demonstrating a mature and critical
relationship with the technology and contributing to the ongoing discourse on the integration of AI in the writing
process.

5. Discussion
5.1 GPT in Writing: Efficiency, Structure, Peer Review
The incorporation of GPT-3.5 into writing practices has emerged as a transformative strategy for addressing
several issues outlined in the introduction. Firstly, by quickly generating content and creative suggestions, GPT-
3.5 reduced the time allocated for research and content generation. Secondly, GPT-3.5 provided precise and
relevant responses to prompts and aided in structural organization. GPT-3.5 ensured consistency and logical
progression in students’ compositions, which is exactly what they need at this stage of learning to write (Xu et
al., 2019). Furthermore, students recognized GPT-3.5’s potential as a dependable replacement for peer
reviewers, acknowledging the crucial role peer reviewers play in achieving good writing (Harutyunyan and
Poveda, 2018). The AI’s thorough language evaluations and feedback provided students with critical insights into
their writing, allowing them to independently identify areas for refinement (Gayed et al., 2022).
5.2 GTP, ADDIE, and TPACK in Writing Enhancement
The integration of GPT-3.5 within the writing framework demonstrated the practical application of both the
ADDIE model and the TPACK framework.
5.2.1 Analysis/design: Laying foundation
During the analysis stage, the course designers recognized the need to establish a strong grasp of essay structure
before introducing AI-generated content. Therefore, fundamental principles of organization and coherence
were reinforced to provide students with a solid basis for proficient writing.
In the design stage, a scaffolded approach was developed to guide students through diverse writing tasks. The
autobiography task aimed to improve students’ self-introduction skills by highlighting their strengths, employing
a structured method to communicate with GPT-3.5, and facilitating the creation of well-structured essays. The
STAR and SWOT techniques further enhanced the self-introductions. The revision task utilized a methodical
proofreading process with GPT-3.5 to address grammar, paragraph structure, coherence, and concise closing
statements, resulting in a refined outcome. Tables were employed for efficient assessment, streamlining the
revision process.
Through a step-by-step systematic approach, students were conscientiously guided to interact with AI
technology in their writing process (Ammade et al., 2020). Each decision made, from listing their advantages to
refining their compositions, was done thoughtfully and with purpose. This approach instilled in them a
heightened sense of consciousness and critical thinking, enabling them to utilize AI’s capabilities while retaining
full control, cultivating a new generation of mindful and skillful writers (Yot-Domínguez, 2017).

www.ejel.org 0 ©The Authors

 Yu-Ching Tseng and Yi-Hsuan Lin

5.2.2 Development/implementation: encouraging authenticity

During the development stage, course materials were carefully crafted to encourage authenticity. The selection
of the autobiography writing topic prompted students to draw from personal experiences, fostering originality
and discouraging plagiarism. By promoting personal narratives, the course nurtured a strong sense of
authorship, underlining the significance of individual creativity in their work.
In the implementation stage, students adeptly incorporated GPT-3.5 into their writing process by creating
personalized prompts to guide the AI. Improvements were evident through enhanced coherence, enriched
vocabulary, and refined sentence structures. Throughout their interaction with GPT-3.5, students skillfully
preserved their unique authorial voice and ownership, ensuring the AI-generated content harmonized with their
viewpoints and preferences.
As students interacted with GPT-3.5, there was a noticeable effort to retain their unique authorial voice. This
was visible as they guided the AI’s input to align with their personal perspectives and preferences, thereby
indicating their control over the final work. This behavior could be seen as a subtle shift towards becoming more
reflective writers who can utilize advanced technology without compromising personal insight or creativity.
5.2.3 Evaluation: Cultivating conscientious writers
The evaluation phase provided intriguing insights into the effectiveness of the systematic integration of GPT-
3.5. Notably, it showcased the symbiosis between the ADDIE and TPACK models, which was apparent in the
students’ ability to effectively utilize GPT-3.5 to enhance their writing. This successful implementation
demonstrated that the fusion of ADDIE’s rigorous instructional design process with TPACK’s intricate
understanding of the intersection between technology, pedagogy, and content can offer a structured yet flexible
approach to AI integration in education (Ammade et al., 2020). More specifically, the students’ ability to craft
personalized prompts and refine the AI’s outputs illustrated their technological and pedagogical proficiency, a
cornerstone of TPACK. This shows that students were not just passively using technology but actively and
thoughtfully engaging with it to better their learning experience and outcome. Simultaneously, their efforts to
retain their authorial writing style underscored the critical reflection inherent in the ADDIE process. Students
were critically reflecting on their work, an action promoted by the ADDIE model, and this reflection aided in the
development of their authorial voice and overall writing skills. Thus, the findings shed light on a promising
strategy for facilitating conscientious authorship and critical thinking, intertwining technological advancements
with traditional pedagogical approaches in writing education.
5.2.4 Fostering authorship and critical thinking
The students in this study articulated feelings of complexity and internal debate surrounding the utilization of
GPT-3.5 in writing. This reaction reflected their growing cognizance of ethical considerations and the importance
of authorship conscience. Their thoughtful consideration of whether AI-generated sentences truly constituted
their own writing exemplified their sensitivity to issues of originality and artistic expression.
The course facilitated the development of conscientious writers by encouraging students to navigate these
complexities with care and critical thinking. The course’s integration of AI fostered critical thinking among
students by presenting a complex task: blending their unique voices with the AI’s suggestions while preserving
authorship. They were prompted to critically assess the AI’s inputs and their work, fostering a deeper
understanding of their narratives. Simultaneously, the course nurtured conscientious writers, as students
grappled with AI’s implications in terms of originality and artistic integrity. By negotiating this intricate task,
students developed a balanced approach to AI, upholding their authorship and demonstrating thoughtful use of
technology, an essential skill for the modern writer.
5.3 Assessing AI-Assisted Writing: Balancing Technology and Human Input
In response to queries regarding the assumptions about AI-generated content and its comparison to human
texts, it is important to clarify that this study does not posit AI-generated results as superior or perfect compared
to student texts. Rather, the study views AI as a tool that complements and enhances students’ writing process.
• AI versus Human Texts: The study recognizes that while AI, such as GPT-3.5, can offer valuable
assistance in terms of generating ideas, structuring content, and providing language suggestions, it
does not replace the complex nature of human thought and expression. The goal is not to create a
dependency on AI but to utilize it as a supplement that aids in the development of students’ writing
skills.

www.ejel.org 0 ISSN 1479-4403

The Electronic Journal of e-Learning SI AI Preview 2024

• Instructor’s Role in Testing AI Results: As part of the course, the instructor plays a crucial role in
evaluating the AI-assisted work submitted by students. This evaluation is not solely focused on the
quality of the AI-generated content but also on how effectively students integrate this content into
their writing while maintaining their unique voice and critical thinking. The instructor assesses
whether the students are using AI as a tool for improvement, ensuring that the final output reflects
their understanding and creativity.
• Encouraging Critical Engagement with AI: The course encourages students to critically engage with
AI outputs. Students are taught to scrutinize the AI-generated content, evaluate its relevance and
accuracy, and modify it to suit their purpose and style. This approach fosters a balanced relationship
with technology, where students learn to employ AI’s strengths while being mindful of its limitations.
In sum, this study emphasizes a balanced approach to AI in writing education, advocating for its use as an aid in
the creative and critical process rather than as a substitute for human ability. The objective is to train writers
who can blend technology with their own unique insights and skills, thus advancing their writing in the context
of evolving digital tools.

6. Conclusion
In utilizing ChatGPT as an instructional tool for writing, the study’s findings underscore the positive impact of
the combined TPACK and ADDIE approach on students’ writing abilities. This research aligns with the findings of
Ammade et al. (2020), Gayed et al. (2022), and Xu et al. (2019), specifically emphasizing how the incorporation
of technology can assist students in generating better academic writing. Students exhibited significant
improvement in their writing quality, validating the effective integration of ChatGPT into writing instruction
under the thoughtful guidance of the TPACK framework and the structured approach of the ADDIE model.
For the wider EFL teacher community, these findings suggest that AI tools like ChatGPT can be effectively
integrated into writing instruction, provided they are used within a structured and pedagogically sound
framework. This can lead to improvements in students’ writing quality and foster a more engaging and
interactive learning environment. However, while GPT-3.5’s integration into the writing process offers various
benefits, this study acknowledges its limitations. First, there is a potential for further research to explore its
personalized application in generating diverse writing types, accommodating individual writing styles and
requirements. Additionally, conducting in-depth qualitative studies, such as interviews or think-aloud protocols,
could provide valuable insights into students’ experiences with AI integration, potentially revealing underlying
reasons for their writing challenges. Furthermore, it’s important to note the limitation posed by this study’s
small sample size. While the qualitative data from the 15 participants provides valuable initial insights, a larger
sample would allow for more generalized conclusions and a broader understanding of AI’s impact on writing
education. Finally, quantitative assessments measuring time savings and comparing AI-assisted writing to
traditional methods could provide concrete insights into the practical advantages of AI in the writing process.
Future research plans include expanding the sample size and exploring ChatGPT’s application in diverse writing
contexts to accommodate individual styles and needs. Additionally, employing quantitative methods and in-
depth qualitative approaches like interviews will provide a more holistic understanding of AI’s role in writing
education. Regarding validity and reliability, the study employed rigorous methodological approaches within its
scope, as the collected data were scored and validated by two researchers who also teach academic writing.
Future experiments will aim to enhance both the validity and reliability of the findings through varied and more
extensive data collection methods.
Despite these constraints, this study provides crucial insights into the potential of AI integration to enhance
writing education and foster critical thinking skills, paving the way for comprehensive future investigations.

Acknowledgements
This work was partially supported by the Ministry of Education (MOE) in Taiwan [PED1121084].

References
Ahmed, P.H., 2019. Major writing challenges experienced by EFL learners in Soran University. Journal of University of
Human Development, 5(3), pp.120-26.
Aldoobie, N., 2015. ADDIE model. American International Journal of Contemporary Research, 5(6), pp.68-72.
Alharbi, M.A., 2019. Exploring the potential of Google Doc in facilitating innovative teaching and learning practices in an
EFL writing course. Innovation in Language Learning and Teaching, 14(3), pp.227-42.
https://doi.org/10.1080/17501229.2019.1572157.

www.ejel.org 0 ©The Authors

 Yu-Ching Tseng and Yi-Hsuan Lin

Allen, M., 2017. Designing online asynchronous information literacy instruction using the ADDIE model. In T. Maddison and
M. Kumaran, eds. Distributed learning. Witney, UK: Chandos Publishing. pp.69-91. https://doi.org/10.1016/B978-0-
08-100598-9.00004-0.
Almelhi, A.M., 2021. Effectiveness of the ADDIE model within an e-learning environment in developing creative writing in
EFL Students. English Language Teaching, 14, pp.1-20. http://doi.org/10.5539/elt.v14n2p20.
Ammade, S., Mahmud, M., Jabu, B., and Tahmir S., 2020. TPACK model based instruction in teaching writing: an analysis on
TPACK literacy. International Journal of Language Education, 4(2), p.129. https://doi.org/10.26858/ijole.v4i2.12441.
Baker, K.M., 2016. Peer review as a strategy for improving students’ writing process. Active Learning in Higher Education,
17(3), pp.179-92.
Bates, T., 2015. Teaching in a digital age. Vancouver, BC: Tony Bates Associates Ltd.
Bouwmeester, R.A., de Kleijn, R.A., van den Berg, I.E., ten Cate, O.T.J., van Rijen, H.V., and Westerveld, H.E., 2019. Flipping
the medical classroom: effect on workload, interactivity, motivation, and retention of knowledge. Computers and
Education, 139, pp.118-28. http://doi.org/10.1016/j.compedu.2019.05.002.
Cancino, M., and Panes, J., 2021. The impact of Google Translate on L2 writing quality measures: evidence from Chilean EFL
high school learners. System, 98, 102464. https://doi.org/10.1016/j.system.2021.102464.
Chen, A.H., 2022. The effects of writing strategy instruction on EFL learners’ writing development. English Language
Teaching, 15(3), 29. https://doi.org/10.5539/elt.v15n3p29.
Chen, Z., Zhang, C., Wang, Q., Troidl, J., Warchol, S., Beyer, J., Gehlenborg, N., and Pfister, H., 2023. Beyond generating
code: Evaluating GPT on a data visualization course. Preprint arXiv:2306.02914.
https://doi.org/10.48550/arXiv.2306.02914.
Cook, S., 2009. Coaching for high performance: How to develop exceptional results through coaching. Ely, UK: IT
Governance Ltd.
Davis, N., & Thompson, A., 2005. The evaluation of technology-related professional development, Part 2. In C. Crawford et
al., eds. Proceedings of Society for Information Technology and Teacher Education International Conference 2005.
Chesapeake, VA: AACE. pp.825-30.
Garg, R.K., Urs, V.L., Agrawal, A.A., Chaudhary, S.K., Paliwal, V., and Kar, S.K., 2023. Exploring the role of Chat GPT in patient
care (diagnosis and Treatment) and medical research: a systematic review. medRxiv, 2023.06.13.23291311.
https://doi.org/10.1101/2023.06.13.23291311.
Gayed, J.M., Carlon, M.K.J., Oriola, A.M., and Cross, J.S., 2022. Exploring an AI-based writing assistant’s impact on English
language learners. Computers and Education: Artificial Intelligence, 3, Article 100055.
https://doi.org/10.1016/j.caeai.2022.100055.
Gilson, A., Safranek, C., Huang, T., Socrates, V., Chi, L., Taylor, R., and Chartash, D., 2023. How does ChatGPT perform on
the United States Medical Licensing Examination? The implications of large language models for medical education
and knowledge assessment. JMIR Medical Education, 9, e45312. https://doi.org/10.2196/45312.
Goldberg, A., Russell, M., and Cook, A., 2003. The effect of computers on student writing: a meta-analysis of studies from
1992 to 2002. The Journal of Technology, Learning and Assessment, 2(1), pp.1-20. Available at:
<https://ejournals.bc.edu/index.php/jtla/article/view/1661>.
Golonka, E.M., Bowles, A.R., Frank, V.M., Richardson, D.L., and Freynik, S., 2014. Technologies for foreign language
learning: a review of technology types and their effectiveness. Computer Assisted Language Learning, 27, pp.70-105.
https://doi.org/10.1080/09588221.2012.700315.
Han, T., and Huang, J., 2017. Examining the impact of scoring methods on the Institutional EFL Writing Assessment: A
Turkish perspective. PASAA: Journal of Language Teaching and Learning in Thailand, 53, pp.112-47.
Harutyunyan, L., and Poveda, M.F., 2018. Students’ perception of peer review in an EFL classroom. English Language
Teaching, 11(4), pp.138-51. https://doi.org/10.5539/elt.v9n9p36.
Koehler, M.J., and Mishra, P., 2008. Introducing TPCK. In AACTE Committee on Innovation and Technology, ed. The
handbook of technological pedagogical content knowledge (TPCK) for educators. Mahwah, NJ: Lawrence Erlbaum
Associates. pp.3-29.
Lee, I., 2002. Teaching coherence to ESL students: a classroom inquiry. Journal of Second Language Writing, 11(2), pp.135-
59. https://doi.org/10.1016/S1060-3743(02)00065-6.
Li, J., and Huang, J., 2022. The impact of essay organization and overall quality on the holistic scoring of EFL writing:
perspectives from classroom English teachers and national writing raters. Assessing Writing, 51, Article 100604.
https://doi.org/https://doi.org/10.1016/j.asw.2021.100604.
Li, L., Liu, X and Steckelberg, A., 2010. Assessor or assessee: how student learning improves giving and receiving feedback.
British Journal of Educational Technology, 41(3), pp.525-36.
Lin, C.A., Lin, Y.L., and Tsai, P.S., 2020. Assessing foreign language narrative writing through automated writing evaluation:
a case for the web-based Pigai system. In Serpil Meri Yilan, ed. ICT-based assessment, methods, and programs in
tertiary education. Hershey, PA: IGI Global. pp.100-119. http://doi.org/10.4018/978-1-7998-3062-7.ch006.
Misbah, S., and Mahboob, U., 2017. Strengths, weaknesses, opportunities, and threats analysis of integrating the World
Health Organization patient safety curriculum into undergraduate medical education in Pakistan: a qualitative case
study. Journal of Educational Evaluation for Health Professions, 14, p.35.
Muruganantham, G., 2015. Developing of e-content package by using ADDIE model. International Journal of Applied
Research, 1(3), pp.52-54.

www.ejel.org 0 ISSN 1479-4403

The Electronic Journal of e-Learning SI AI Preview 2024

Putri, S.E., 2019. The implementation of TPACK in teaching writing recount text in a senior highschool level. RETAIN
Journal, 7(2), pp.156-63.
Ravand, H., and Rasekh, A.E., 2011. Feedback in ESL writing: toward an interactional approach. Journal of Language
Teaching and Research, 2(5), pp.1136-45. https://doi.org/10.4304/jltr.2.5.1136-1145.
Richards, J.C., and Rodgers, T.S., 2001. Approaches and methods in language teaching. Cambridge University Press.
Richards, J.C., 2013. Curriculum approaches in language teaching: forward, central, and backward design. RELC Journal,
44(1), pp.5-33.
Ruegg, R., 2015. The relative effects of peer and teacher feedback on improvement in EFL students’ writing ability.
Linguistics and Education, 29, pp.73-82. https://doi.org/10.1016/j.linged.2014.12.001.
Schmidt, D.A., Baran, E., Thompson, A.D., Mishra, P., Koehler, M.J., and Shin, M.J., 2009. Technological pedagogical content
knowledge (TPACK): the development and validation of an assessment instrument for pre-service teachers. Journal of
Research on Technology in Education, 42(2), pp.123-49.
Shulman, L.S., 1986. Those who understand: knowledge growth in teaching. Educational Researcher, 15(2), pp.4-14.
Skalidis, I., Cagnina, A., Luangphiphat, W., Mahendiran, T., Muller, O., Abbe, E., and Fournier, S., 2023. ChatGPT takes on
the European Exam in Core Cardiology: an artificial intelligence success story? European Heart Journal—Digital
Health, 4(3), pp.279-81. https://doi.org/10.1093/ehjdh/ztad029.
Sogutlu, E., and Veliaj-Ostrosi, M., 2022. EEL learners’ challenges in essay writing: the case of a non-public high school in
Albania. Journal of Positive School Psychology, 6(4), pp.3958-81.
Tanaka, Y., Nakata, T., Aiga, K., Etani, T., Muramatsu, R., Katagiri, S., Nomura, A., 2023. Performance of generative
pretrained transformer on the National Medical Licensing Examination in Japan. Preprint from medRxiv.
http://doi10.1101/2023.04.17.23288603.
West, R., 1994. Needs analysis in language teaching. Language Teaching, 27(1), pp.1-19.
Xu, Z., Banerjee, M., Ramirez, G., Zhu, G., and Wijekumar, K., 2019. The effectiveness of educational technology
applications on adult English language learners’ writing quality: a meta-analysis. Computer Assisted Language
Learning, 32, pp.132-62. https://doi.org/10.1080/09588221.2018.1501069.
Yot-Domínguez, C., and Marcelo, C., 2017. University students’ self-regulated learning using digital technologies.
International Journal of Educational Technology in Higher Education, 14(1), pp.38. https://doi.org/10.1186/s41239-
017-0076-8.
Zhang, L.J., and Cheng, X.L., 2021. Examining the effects of comprehensive written corrective feedback on L2 EAP students’
performance: a mixed-methods study. Journal of English for Academic Purposes, 54(101043), pp.1-15.
https://doi.org/10.1016/j.jeap.2021.101043.
Zhang, W., and Cai, W., 2019. Research on English writing feedback based on online automatic evaluation and reform.
Education Research, 2019(3), pp.102-103.
Zhang, Z.V., 2020. Engaging with automated writing evaluation (AWE) feedback on L2 writing: student perceptions and
revisions. Assessing Writing, 43, 100439. https://doi.org/10.1016/j.asw.2019.100439.
Zhang, Z.V., and Hyland, K., 2022. Fostering student engagement with feedback: an integrated approach. Assessing
Writing, 51, 100586. https://doi.org/10.1016/j.asw.2021.10058.

www.ejel.org 0 ©The Authors