Insights into EFL Students’ Perceptions of the ‘ChatGPT Essentials’ Training Course for Language Learning

Abstract

This paper introduces ‘ChatGPT essentials’, a pedagogically driven training course for pre-service English as a Foreign language (EFL) teachers at the College of Basic Education (CBE) in Kuwait. It responds to current ethical and academic integrity issues by empowering students to use ChatGPT both effectively and ethically. Prior to ChatGPT essentials training sessions, semi-structured interviews were conducted with twenty-five male undergraduate students in a Computer Assisted Language Learning (CALL) course to assess their familiarity with ChatGPT for language learning, followed by pedagogical and practical training sessions and a subsequent evaluation. The quantitative analysis indicates that the students generally valued the training on four levels (i.e., reaction, learning, behaviors, and results). Their perceptions and experiences have changed positively, indicating general positive attitudes towards using ChatGPT as a tool to develop their language skills. Qualitative data from post-training interviews and students’ reflective journals revealed that students valued the practical guidance on ethical usage and critical evaluation of ChatGPT practices, which enhanced their digital literacy skills and fostered responsible ChatGPT use. Such findings point to the benefits of implementing pedagogical training to enhance students’ ChatGPT usage.

1. Introduction

The popularity of ChatGPT has transformed education (Lim et al., 2023), in particular, the language learning field (Lo et al., 2024; Warschauer & Xu, 2024). The current CALL discourse surrounding ChatGPT and its impact on second/foreign language (SL/FL) learning has tended to focus on perceived opportunities and challenges for students and teachers (Kostka & Toncelli, 2023; Li et al., 2024; Zhu & Wang, 2025). Empirical research suggests that ChatGPT can promote learner’s autonomy and personalized learning (Agustini, 2023). It also can help in developing writing skills such as generating content and organization (Boudouaia et al., 2024; Ghafouri et al., 2024; Marzuki et al., 2023; Yan, 2023) and argumentative writing skills (Su et al., 2023). From a theoretical perspective, ChatGPT provides language learners with enhanced input and authentic human-computer interaction (Kohnke et al., 2023). It mediates the learning process and promotes self-regulated learning and motivation (Song & Song, 2023) by offering feedback related to language form and content (ElEbyary & Shabara, 2024; Guo & Wang, 2024; Jiang et al., 2023).

However, there also remain widespread concerns around its inability to offer the nuanced and context-specific guidance provided by human teachers (Kohnke et al., 2023), as well as: (1) students’ overreliance; (2) the provision of inaccurate information and feedback (Zirar, 2023); (3) academic integrity issues; (4) the risk of accusations of plagiarism and violation of academic policies (Baek et al., 2024); as well as causing (5) ‘cognitive offloading’, i.e., long-term detrimental consequences for students’ academic writing ability (Darwin et al., 2024).

This paper proposes the pedagogical training course ‘ChatGPT Essentials’ for enhancing students’ literacy skills, addressing concerns about ChatGPT usage. A systematic review of the literature revealed that the majority of studies either focus on exploring ChatGPT’s potential to enhance language learning skills or examining teachers’ and learners’ perceptions and experiences of using ChatGPT. Few studies explicitly address the need to train students to use ChatGPT, highlighting a critical gap in rigorously investigating methods of enhancing learners’ ChatGPT usage skills. An exception is Horn (2024), who studied Korean university students, demonstrating the effectiveness of integrating a well-designed training course for developing students’ literacy skills through a series of ChatGPT activities under teacher supervision. However, the present study differs by proposing a training model integrating pedagogical strategies aimed at fostering learners’ awareness of the inherent paradoxes associated with ChatGPT usage, thereby promoting a deeper sense of ethical responsibility.

Thus, the main aims of this paper are twofold: (1) to design a pedagogical training model that equips students with skills and strategies necessary to use ChatGPT to enhance students’ ChatGPT’s usage behaviors, and (2) to evaluate the effectiveness of this training by examining students’ perceptions of the training and the use of ChatGPT after attending the training sessions. The study involves exploring how the training influences students’ understanding of ChatGPT usage on four levels, namely: reactions, learning, behavior, and results, aligning with the widely used Kirkpatrick’s model for evaluating educational intervention.

2. Literature Review

2.1. Technology Acceptance Model (TAM)

The Technology Acceptance Model (TAM) proposed by Davis (1989) has been widely applied as a theoretical framework to understand users’ technology acceptance. Guided by social psychology theory, and particularly the Theory of Reasoned Action (TRA), TAM proposes that a user’s beliefs influence his/her motivation to participate. Davis identified two main key factors, namely Perceived Usefulness (PU) (i.e., the degree a user believes that using a specific technology will enhance their performance) and Perceived Ease of Use (PEOU), i.e., the degree to which a person believes that using technology will be free of effort. These two main external factors can influence users’ attitudes, behavioral intentions, and ultimately their actual use of technology.

In this paper, pedagogical training sessions were introduced to enhance students’ experiences with ChatGPT, which could increase both the perceived usefulness and the perceived ease of use, which are central constructs in the TAM model. In this paper, the pedagogical training refers to explicit instructions on how to use ChatGPT effectively and ethically. The training is designed to enhance language learning, reduce technical anxiety and academic concerns, and increase self-efficacy. Additionally, it aims to help students recognize ChatGPT’s potential to enhance their language learning skills. In the current study, TAM is referenced to provide theoretical background on factors that may influence students’ willingness to use ChatGPT for language learning rather than a primary framework for data analysis or interpretation of results.

The following section examines empirical research evidence related to CALL and presents the research findings about ChatGPT, focusing on its perceived usefulness in language learning, as well as several concerns related to its use.

2.2. ChatGPT and Language Learning: Empirical Research Findings

There has recently been increased research into the role of ChatGPT in language learning, including its benefits and drawbacks. A recurring theme concerns its potential to enhance language skills, particularly writing. Polakova and Ivenz (2024), Song and Song (2023), and Boudouaia et al. (2024) undertook a quantitative analysis that revealed substantial improvement in their participants’ academic writing performance. Both Polakova and Ivenz (2024) and Song and Song (2023) reported significant improvement in post-test results, specifically in: (1) organizational skills; (2) coherence, grammar, and vocabulary (Song & Song, 2023); (3) conciseness and inclusion of key information; and (4) the adept use of the passive voice (Polakova & Ivenz, 2024). Boudouaia et al. (2024) reported that these benefits significantly influenced students’ acceptance of the tool. Others such as Yan (2023) and Abdi et al. (2024) have provided qualitative insights concerning the effectiveness of using ChatGPT to support students’ writing skills. They found that using ChatGPT has helped students in generating ideas and outlining the topic. Others found that the use of ChatGPT improved students’ writing skills because of the provision of corrective feedback (Barrot, 2023; Yan, 2023).

Further research has confirmed students’ positive attitudes towards the use of ChatGPT (e.g., AbuSa’aleek & Alenizi, 2024; Alsalem, 2024; Allen & Mizumoto, 2024; Teng, 2024; Xiao & Zhi, 2023; Bao & Li, 2023; Baek et al., 2024; Slamet, 2024). For example, Slamet (2024) found that EFL Indonesian students appreciated ChatGPT’s ability to enhance engagement and motivation, alongside providing access to authentic language resources. Similar findings were also reported in Xiao and Zhi’s (2023) study involving Chinese EFL students, as well as Alsalem’s (2024) study of Saudi students. In both cases, students valued the use of ChatGPT for improving their pronunciation, colloquial language, and vocabulary, while its non-judgmental feedback helped alleviate speaking anxiety.

However, these studies also highlighted concerns related to ChatGPT’s inability to understand real interaction, along with threats to academic integrity, issues of overreliance, and other forms of ethical and academic misconduct (Al-Khreshehm, 2024; Baek et al., 2024). Alsalem (2024) and Hu and Škultéty (2024) both reported that students experienced frustration due to ChatGPT’s inability to provide rich and genuine interaction and its limited pronunciation, resulting in comprehension challenges. Abdi et al. (2024) reported students’ concerns about the discrepancies between questions and the answers provided by ChatGPT. Other ethical concerns include authentic authorship, academic integrity, and plagiarism (Creely, 2024; Yan, 2023). For example, Baek et al. (2024) found that some students feared institutional repercussions and accusations of plagiarism, along with inaccuracies, and skepticism relating to the safety of their data and ethics.

To address these concerns, previous studies have called for structured training to enhance students’ ability to use ChatGPT (e.g., Xiao & Zhi, 2023; Li et al., 2024; Teng, 2024; Warschauer et al., 2023). Horn (2024) argued that “ChatGPT is not yet equipped to teach on its own” (p. 4), and that educators have a role to play in enhancing students’ digital literacy and use of ChatGPT. Horn’s (2024) results indicated that training students has helped them to improve their appreciation of the affordances of ChatGPT for improving sentence construction and expanding their grammar, vocabulary, and practice of spoken English. However, some reservations remain regarding its effectiveness in improving English language skills. Additionally, the study reported ChatGPT’s benefits, such as enhancing students’ motivation in speaking activities, as well as in reading and writing, along with improving confidence and autonomy. Despite these promising findings, Horn’s (2024) training failed to raise students’ awareness of certain ethical and academic considerations. In another study by Warschauer et al. (2023), these academic considerations were addressed by proposing a five-part pedagogical framework that includes understanding, accessing, prompting, corroborating, and incorporating to help students and teachers integrate ChatGPT in their writing instruction. Warschauer et al. (2023) argued that the value of ChatGPT for language learners lies in navigating three main contradictions: the ‘imitation’, ‘the rich get richer’, and ‘with or without’ contradictions. To address these contradictions, Warschauer et al. (2023) argued that open discussion and critical engagement with AI tools are essential to enhance AI literacy.

In this paper, strong alignment is demonstrated with the perspectives of Warschauer et al. (2023) and Horn (2024), who emphasize the necessity of structured training to enhance ChatGPT literacy among language learners. Both studies underscore the value of pedagogically informed approaches for supporting students as they navigate the affordances and contradictions associated with AI tools in language learning contexts. Accordingly, a structured pedagogical training model called ‘ChatGPT essentials’ was developed, and its effectiveness was evaluated using Kirkpatrick’s model, as explained in Section 3.

The following research questions were proposed to evaluate the ‘ChatGPT Essentials’ training course:

• What impact, if any, does the ChatGPT Essentials training course have on learners’ general understanding of using ChatGPT?
• What are learners’ reported perceptions of ChatGPT after attending the ‘ChatGPT essentials’ training course?

3. Methodology

3.1. Research Design

An evaluation research design was employed to assess the effectiveness of the “ChatGPT Essentials” training course for EFL students using both qualitative and quantitative data collection methods. This approach is well-suited to capture the nuanced perceptions and experience of EFL students who participated in the training sessions. Furthermore, this approach allows for a holistic understanding of how the training influenced students’ attitudes and the potential use of ChatGPT for language learning. By focusing on students’ lived experiences and subjective interpretations, this design yields insights that can inform future improvements to the training program and guide similar interventions in other contexts.

3.2. Setting, Participants, and Research Procedures

The participants consisted of twenty-five male undergraduate EFL students aged nineteen to twenty-four, who were enrolled in the CALL module for twelve weeks within an English language teacher education program in the College of Basic Education (CBE) in Kuwait. All participants were preparing to become English language teachers, and the module is an elective course in their degree pathway.

Participants were advanced English language speakers, as admission to the teacher education program requires a high level of English proficiency, typically demonstrated by prior standardized admission tests. Students had completed advanced general English language courses, applied linguistics, and literature courses before enrolling in the CALL module. The primary motivation for learning English among participants was professional, as all students were aiming to become future English language teachers in schools. Their engagement with the CALL module was directly related to their anticipated teaching careers, where they would be expected to integrate technology into language instruction. The CALL module is open to students from multiple years of study who have completed prerequisite courses such as advanced English writing and applied linguistics courses. This flexible enrollment policy allows students to select the module according to their study plan. As a result, the course cohort included students at various stages of their undergraduate program, from second to fourth year. 16% of the participants were in their second year, 50% in their third year, and 34% in their final year of the teachers’ education program. The background interviews revealed that all participants (n = 25) were familiar with ChatGPT as an AI tool. Fifteen of the twenty-five had previously used it for general purposes, with eight having used it as a tool for learning English. The remainder of participants (n = 10) reported that they had not made use of it previously. Those reporting using ChatGPT (n = 15) stated their daily use (n = 4), while the rest indicated that they only used it weekly (n = 6) or monthly (n = 5) when they needed to accomplish their assignments or find answers to questions. This demonstrated that all participants had some level of familiarity with ChatGPT.

The study was conducted over a three-month period, beginning in February 2024 and ending in April 2024. In week one, the students received a detailed explanation of the study, including that their participation was voluntary, and their final grade would not be impacted if they decided not to participate1 (Appendix A). While the study’s general aims were outlined to ensure transparency, care was taken not to disclose research questions or expected outcomes. In addition, during this week, semi-structured background interviews were conducted to explore students’ familiarity with ChatGPT, as well as their attitudes towards its use, both generally and for learning English (Appendix B).

In week two, the researcher conducted phase one (Appendix C), which consisted of two sessions each lasting for one hour, after students’ normal class time, to introduce the ‘ChatGPT Essentials’ training course to develop their pedagogical understanding of ChatGPT, including its four paradoxes (Lim et al., 2023). In week three, the researcher implemented phase two, modeling interactions with ChatGPT using an LCD projector and demonstrating activities including (1) practicing real-life scenarios with ChatGPT; (2) writing with an AI partner; and (3) checking ChatGPT’s answers. Then, an evaluation form was distributed for students to evaluate phases 1 and 2. Phase 3 ‘Practicing ChatGPT’ took place between weeks four and ten, with students asked to complete assignments related to the CALL course, utilizing the assistance of ChatGPT. They were also asked to write a reflective journal entry after each phase. The researcher conducted semi-structured interviews with the students during weeks eleven and twelve, along with post-training questionnaires to evaluate the effectiveness of the training course (Appendix D and Appendix E). The timeline for the research procedures is illustrated in

Table 1. Study procedure timeline.

Week		Session	Description and Objectives
1	Reflection	Orientation session	Distributing information sheets and consent forms Background semi-structured interviews
2		Phase 1 Comprehension & understanding	Session 1 (1 h) - Paradox 1&2 Session 2 (1 h) - Paradox 3&4
3		Phase 2 Modeling	Demonstrating some activities with ChatGPT - Practice real-life scenarios with ChatGPT - Writing with an AI partner - Checking ChatGPT’s Information Evaluation Round 1: Distributing the training evaluation questionnaire to measure reaction and learning.
4–10		Phase 3 Practicing ChatGPT	Engaging in some activities with ChatGPT - Practice real-life scenarios with ChatGPT - Writing with an AI partner - Checking ChatGPT’s Information Using ChatGPT as part of the CALL module - Students work on four different assignments using ChatGPT as an assisted tool
11–12		Wrap-up session	Evaluation Round 2: Distributing the training evaluation questionnaire to measure behavior and results Semi-structured interviews with students

.

As shown in Table 1, the evaluation process was intentionally divided into two distinct evaluation rounds to align with established educational evaluation practices and to capture different dimensions of the training. In week 3, immediately after the training session (Round 1), an evaluation was conducted to assess students’ comprehension and understanding of the materials. This immediate assessment is critical for determining whether the instructional objectives were met and if students had acquired the foundational knowledge and skills intended for the training. Round 2 was conducted after students had the opportunity to actively engage with ChatGPT through practical activities in week 11. This subsequent evaluation allowed for the assessment of students’ ability to apply what they had learned in authentic contexts, reflecting higher levels of evaluation. By separating the evaluations, the study will be able to distinguish between initial comprehension and actual transfer of learning to practice, providing a more comprehensive and nuanced understanding of the training’s effectiveness. The following section introduces the design of the ‘ChatGPT Essentials’ training course.

3.3. ‘ChatGPT Essentials’ Training Course

‘ChatGPT Essentials’ is a pedagogical training course comprising explicit instructions on how to use ChatGPT effectively and ethically. It builds upon existing frameworks emphasizing functional skills (i.e., understanding, accessing, promoting, and corroborating) (Warschauer et al., 2023). It consists of three main training phases: (1) Comprehension and Understanding; (2) Modeling ChatGPT; and (3) Practicing ChatGPT (Figure 1). Additionally, the students were encouraged to engage in iterative reflection (see Appendix C). Paradoxes, i.e., meaning statements that seem contradictory or absurd but may in fact be true or plausible (Cowie, 2023) and often involve two opposing ideas that exist simultaneously, were central to ‘ChatGPT training’ sessions. The use of paradoxes as a pedagogical framework in training courses like “ChatGPT essential” is grounded in previous similar research (Lim et al., 2023; Warschauer et al., 2023), which highlighted the complex and sometimes contradictory affordances of AI tools in education and particularly in language education. By explicitly foregrounding these paradoxes in the training, students can engage with contradictions that stimulate critical thinking, creativity, and problem-solving. Paradox-centered training was designed to move beyond purely technical training, encouraging students to think critically about the opportunities and limitations of ChatGPT. In this way, the training interventions align with current pedagogical recommendations that cultivate a ‘paradox mindset’, preparing students to understand and navigate ChatGPT’s complexities and tensions in educational practices.

Phase 1 ‘Comprehension and Understanding’, introduces the four paradoxes of ChatGPT, as identified by Lim et al. (2023), raising students’ awareness of various ethical issues and academic misconduct behaviors related to the use of ChatGPT. These paradoxes are: firstly, ChatGPT is a ‘friend’ yet a ‘foe’; secondly, ChatGPT is ‘capable’ yet ‘dependent’; thirdly, ChatGPT is ‘accessible’ yet ‘restrictive’; and fourthly, ‘ChatGPT becomes popular even when banned’. Phase 2 consists of modeling the use of ChatGPT by engaging in various activities. The researcher practices the following activities in front of students: Practicing real-life scenarios with ChatGPT; writing with an AI partner; and checking ChatGPT’s answers. Each of these activities has a pedagogical purpose, as outlined below:

○

Practicing real-life scenarios with ChatGPT: To enhance students’ ability to communicate using their English language skills and develop confidence in conversing with ChatGPT.

○

Writing with an AI partner: To distinguish between copying and recycling from ChatGPT and to demonstrate that ChatGPT is a tool that can be used during pre-and post-writing stages to minimize overreliance.

○

Critical thinking and digital literacy: To show students that ChatGPT can generate inaccurate information and teach them to verify the accuracy of its responses by cross-checking with reliable sources and recognizing its limitations.

In the final phase, students practice ChatGPT independently to assist them with their assignments. During all phases, students are encouraged to reflect on the use of ChatGPT based on Gibbs’ (1988) model (as cited in Grosbois & Sarre, 2016), that is, to describe their experience and feelings, as well as to analyze and evaluate ChatGPT and draw conclusions. They are also encouraged to talk about their action plan for dealing with ChatGPT in the future or potentially beneficial changes.

3.4. Data Collection Instruments

The researcher started by undertaking semi-structured background interviews with the students (Appendix B) to explore their level of familiarity with ChatGPT and their perceptions about its effectiveness as a tool for learning the English language. A formal pre-training evaluation was not conducted because it was anticipated that some students might not have had prior exposure to ChatGPT before the training. In light of this potential, a baseline test might have produced data about their initial ChatGPT’s familiarity that was unrepresentative and inconsistent. Therefore, the focus was placed on post-training evaluation, after students had experienced ChatGPT’s activities, which probably offered more insightful data about the training effects. The post-training semi-structured interviews, alongside a questionnaire, were developed using Kirkpatrick’s (1994) model of evaluation (Appendix D and Appendix E). This consisted of four main levels: (1) Reaction (six statements to gauge students’ immediate responses to the training); (2) Learning (nine statements to measure the knowledge and skills acquired by attending the training session); (3) Behaviors (seven statements to assess students’ application of ChatGPT in real situations); and (4) Results (six statements assessing the impact of ChatGPT). The Cronbach’s alpha coefficient achieved a value of 0.89, reflecting the consistency of all the elements of the questionnaire. Finally, each student submitted three reflective journals, i.e., a total of seventy-five, and post-training face-to-face semi-structured interviews were conducted with students. Each interview lasted for 15–20 min and was audio-recorded.

3.5. Analysis

To answer research question 1, the researcher performed quantitative descriptive statistical analysis for the training evaluation, using IBM SPSS 25. To present the findings, the four levels of evaluation proposed by Kirkpatrick were considered, namely reaction, learning, behavior, and result. For triangulation purposes, qualitative evidence from the interviews was integrated to enhance the validity of the interpretation. To answer research question 2, thematic analysis was performed for the background and post-training semi-structured interview data and students’ reflective journals (Braun & Clarke, 2006). The analysis was data-driven, commencing with familiarization, in which the researcher read over the interview data multiple times to obtain an overall idea of the content. This was followed by the development of preliminary codes to group words and phrases related to the research focus, to create different themes. Two independent coders (i.e., the researcher and research assistant) reviewed all the data and generated codes and broader themes. They then met to discuss and negotiate the findings. To enhance the coding reliability, the data coding of the transcripts was also checked by a third coder for blind coding of 35% of the randomly selected teachers’ extracts. Agreement on the codes and themes was assigned to 82% of students’ data extracts.

3.6. Researcher Reflexivity

As a researcher with a background in CALL and applied linguistics, I recognize that my expertise and positive orientation toward implementing technology and AI tools in the language learning process may have influenced both the design of the study and the interpretation of the findings. My familiarity and experience with ChatGPT specifically, and my belief in its potential benefits, could have shaped the selection of the research questions, the design of the training, and its evaluation. Throughout the research process, I remained attentive to these influences by practicing reflexivity and critically questioning my assumptions to ensure that my professional perspectives did not unduly bias the analysis or reporting of the results.

4. Findings

The findings section is divided into two main parts: the first presents an evaluation of the ‘ChatGPT Essentials’ training course, and the second focuses on examining students’ perceptions of ChatGPT after attending the training course.

4.1. Evaluating the Effectiveness of the ‘ChatGPT Essentials’ Training Course

4.1.1. Reaction and Learning Levels

The data for these two levels were collected immediately following phases 1 and 2, to elicit students’ general attitudes towards the training sessions (i.e., reaction) and to measure the knowledge and skills they had acquired (i.e., learning).

Table 2. Evaluating reaction and learning levels.

Levels of Evaluation	Mean	Std. Deviation
Reaction	4.12	0.299
Learning	4.35	0.292

shows that the students tended to find the training program engaging, relevant, and satisfactory. In addition, a high mean of 4.35 suggests that students strongly agreed that their skills and knowledge have improved.

As shown below in

Table 3. Reaction level.

	Strongly Disagree		Disagree		Neutral		Agree		Strongly Agree
Reaction	F	%	F	%	F	%	F	%	F	%	Mean	Std. Deviation	Items sorted by means
The ChatGPT essentials training was well organized	0	0	0	0	1	4.0	20	80.0	4	16.0	4.12	0.440	3
The ChatGPT essentials training activities were engaging	0	0	0	0	0	0	22	88.0	3	12.0	4.12	0.332	4
The training facility was conducive to learning	0	0	0	0	2	8.0	20	80.0	3	12.0	4.04	0.455	5
The length of the training was appropriate	0	0	1	4.0	0	0	21	84.0	3	12.0	4.04	0.539	6
I had sufficient opportunities to ask questions and receive helpful answers.	0	0	0	0	1	4.0	17	68.0	7	28.0	4.24	0.523	1
The handouts provided helped me to meet all my training needs	0	0	0	0	2	8.0	17	68.0	6	24.0	4.16	0.554	2

, the detailed descriptive statistical analysis for the reaction level indicated that students highly valued ‘ChatGPT essentials’ training sessions. Opportunities to ask and discuss unclear ideas about its usage, along with the practicality of the handouts, were scored the highest on this level. While the training facilities and the length were rated slightly lower, they still received strong positive evaluations. The absence of negative responses and low variability further reinforces the overall students’ satisfaction with the training on the reaction level.

To get an in-depth insight into the learning level,

Table 4. Learning level.

	Strongly Disagree		Disagree		Neutral		Agree		Strongly Agree
Learning	F	%	F	%	F	%	F	%	F	%	Mean	Std. Deviation	Items sorted by means
After the training, I understand the different uses of ChatGPT and can apply them appropriately	0	0	0	0	1	4.0	15	60.0	9	36.0	4.32	0.557	4
I can identify the limitation of using ChatGPT and know to deal with them	0	0	1	4.0	2	8.0	14	56.0	8	32.0	4.16	0.746	8
ChatGPT essentials training has helped me to effectively write prompts to generate desired texts from ChatGPT	0	0	0	0	1	4.0	17	68.0	7	28.0	4.24	0.523	6
I am confident in my ability to evaluate the quality of text produced by ChatGPT	0	0	0	0	1	4.0	17	68.0	7	28.0	4.24	0.523	7
Through the training, I learned ethical considerations and academic misconduct behaviors related to using ChatGPT	0	0	0	0	0	0	17	68.0	8	32.0	4.32	0.476	5
After the training session, I can distinguish between ChatGPT copying and recycling behaviours	0	0	0	0	0	0	13	52.0	12	48.0	4.48	0.510	3
I can use ChatGPT to enhance my English language skills	0	0	0	0	2	8.0	8	32.0	15	60.0	4.52	0.653	2
I can use ChatGPT to solve any problems related to my studies	0	0	0	0	7	28.0	7	28.0	11	44.0	4.16	0.850	9
ChatGPT essentials training course has helped to change my attitudes towards using ChatGPT	0	0	0	0	1	4.0	5	20.0	19	76.0	4.72	0.542	1

below presents a detailed descriptive statistical analysis. All students overwhelmingly reported positive learning outcomes from attending ‘ChatGPT Essentials’ training, specifically regarding changes in their attitudes towards using ChatGPT. The statement “The training course has helped to change my attitudes towards using ChatGPT” had the highest mean (4.72), with 76% strongly agreeing and 20% agreeing. Furthermore, students reported high confidence in ChatGPT usage (mean = 4.32), using ChatGPT to improve their English language skills (mean = 4.48), evaluating ChatGPT’s output (mean = 4.32), and distinguishing between copying and recycling behaviors (mean 4.32). The ability to write effective prompts and deal with ChatGPT’s limitations was also rated positively (means = 4.24 and 4.16, respectively). Additionally, the training was highly effective in building confidence in the practical and ethical use of ChatGPT (mean = 4.32).

The interviews corroborated these quantitative findings and revealed that the participants valued the ‘ChatGPT’s essentials’ training course because of its ability to enhance confidence and improve their ethical practices. For example, the following S8 and S16 have declared that:

“After attending the training sessions, I realized that there were some unethical practices that I was doing, such as copying and pasting from ChatGPT. Now, after the training, I feel more confident to use it ethically.”

(S8)

“Indeed, the training presented ChatGPT differently, I thought that using ChatGPT is a way of cheating; however, after attending the paradoxes sessions, I understood that it depends on how it is used.”

(S16)

Moreover, the training allowed students to enhance their understanding of some academic misconduct practices, such as copying and pasting. As S5 mentioned:

“I was very concerned about my use of ChatGPT, because, if you remember, before the training, I was worried about cheating my professor, now I knew that ChatGPT is similar to any other assisted tool if we you use it properly, avoiding copying and pasting, it is going to be a great source and tutor.”

(S5)

The previous two levels, namely reaction and learning levels, presented students’ immediate reactions to the training and their self-reported learning outcomes, corresponding to the first two levels of Kirkpatrick’s model. In the following section, findings are presented for the other deeper levels of evaluation, namely behavior and results, which can provide a more comprehensive understanding of how the training influenced students’ actual behavioral changes and tangible outcomes over time.

4.1.2. Behavior and Results

The behaviors and results were evaluated after the completion of students’ assignments in Phase 3. The analysis examined whether students’ behaviors changed positively or negatively towards using ChatGPT.

Table 5. Evaluating behaviors and result levels.

Levels of Evaluation	Mean	Std. Deviation
Behavior	4.39	0.594
Result	4.06	0.567

demonstrates that the training exerted a positive influence, with a mean of 4.39, indicating that the students applied their acquired skills and knowledge into practical actions and transformed their habits. Additionally, a mean of 4.06 indicates tangible outcomes, i.e., improvements in students’ productivity and language skills.

Table 6. Behavior level.

	Strongly Disagree		Disagree		Neutral		Agree		Strongly Agree
Behaviour	F	%	F	%	F	%	F	%	F	%	Mean	Std. Deviation	Items sorted by means
I have used ChatGPT to assist me in my studies	0	0	1	4.0	4	16.0	11	44.0	9	36.0	4.12	0.833	7
I have applied the techniques learned in the training to improve my ChatGPT prompts	0	0	0	0	4	16.0	11	44.0	10	40.0	4.24	0.723	5
After attending the training sessions, I started to use ChatGPT ethically in a proper way	0	0	1	4.0	0	0	10	40.0	14	56.0	4.48	0.714	4
I have explored new ways to use ChatGPT to improve English language that I did not know before the training	0	0	0	0	0	0	11	44.0	14	56.0	4.56	0.507	1
I have used ChatGPT affordances to improve my language skills	0	0	0	0	2	8.0	7	28.0	16	64.0	4.56	0.651	2
I have used ChatGPT to enrich my knowledge related to different subjects	0	0	0	0	5	20.0	9	36.0	11	44.0	4.24	0.779	6
I became a user of ChatGPT after engaging in ChatGPT essentials training sessions	0	0	2	8.0	0	0	5	20.0	18	72.0	4.56	0.870	3

presents a detailed descriptive statistical analysis for behavior level, showing that all behavioral items received a high mean score (4.12–4.56), indicating strong agreement that students not only used ChatGPT but also applied specific strategies and ethical considerations learned from the training. Student declared that the training allowed them to explore new ways of using ChatGPT to improve their English language skills (mean = 4.56), use ChatGPT ethically in a proper way (mean = 4.48), and apply techniques learned to improve prompts (mean = 4.24).

Similarly, the result level in

Table 7. Result level.

	Strongly Disagree		Disagree		Neutral		Agree		Strongly Agree
Results	F	%	F	%	F	%	F	%	F	%	Mean	Std. Deviation	Items sorted by means
Learning how to use ChatGPT has improved my productivity	0	0	0	0	5	20.0	9	36.0	11	44.0	4.24	0.779	1
Learning how to use ChatGPT has improved the quality of my language skills.	0	0	0	0	2	8.0	15	60.0	8	32.0	4.24	0.597	2
Learning how to use ChatGPT has saved my time	0	0	0	0	9	36.0	10	40.0	6	24.0	3.88	0.781	5
Using ChatGPT has helped to learn new things	0	0	1	4.0	5	20.0	9	36.0	10	40.0	4.12	0.881	4
Using ChatGPT has made learning more enjoyable	0	0	2	8.0	10	40.0	6	24.0	7	28.0	3.72	0.980	6
Using ChatGPT has enhanced my knowledge in different topics	0	0	0	0	6	24.0	9	36.0	10	40.0	4.16	0.800	3

below indicates an overall positive agreement among students that engaging in ‘ChatGPT essential training’ has resulted in an improvement in their ChatGPT usage. Learning how to use it effectively has improved students’ productivity (mean= 4.24), the quality of their language skills (mean = 4.24), saved their time (mean = 3.88), and enhanced their knowledge about different topics (mean = 4.16).

These findings were corroborated by the post-training interviews and the reflective journals data, confirming the positive impact of ‘ChatGPT essentials’ training sessions on students’ behaviors. The majority of students declared that after attending ‘ChatGPT Essentials’, they became confident in using ChatGPT, and their behavior has changed. Analysis of the interviews revealed three main themes: Development in ethical practices (n = 23 students), an increase in self-confidence (n = 18 students), and development in ChatGPT usage (n = 22 students). For example, S14 declared that his behavior has improved as a result of attending the ChatGPT training sessions:

“I completely changed the way I use ChatGPT. Before attending the training course, I was copying and pasting things without thinking about any negative impact on my academic performance. After the training, I realized that I should use it wisely, because over-reliance may affect me as a writer.”

(S14)

Similarly, S4 (who had been concerned about unethical practices) highlighted changes in his attitudes towards using ChatGPT and declared that after the training, his way of interacting with ChatGPT has changed.

“For me, the training sessions were very helpful I assessed myself and realized that I was using it wrongly, I’ve now changed the way I write prompts to ChatGPT and changed when and how to use it while working on my assignments.”

(S4)

An additional student explained that the training allowed him to see the value of using ChatGPT and reduced the fear of committing academic misconduct behavior “I think the training assisted me to be a better user of ChatGPT, I now feel ready to use it without having any fears of committing plagiarism” (S10). Two additional students respectively stated that their ideas about ChatGPT completely changed after the training sessions and they became more confident about their skills “ I think after the training sessions, my skills have developed, I became more competent to use ChatGPT independently” (S23), “ I can now consult ChatGPT whenever I want to because I had the confidence to use it ethically and professionally” (S18).

4.2. Students’ Perceptions About Using ChatGPT After the Training

Generally speaking, all students valued the training, as it helped them to understand how to use ChatGPT effectively, as illustrated in the previous section. Figure 2 below shows that students had positive perceptions towards using ChatGPT due to its affordances to develop their own language skills, including communication, grammar and vocabulary, and writing. For example, S9 noticed an increased ability to communicate with ChatGPT, which helped to develop his English communication skills as he stated: “After the training, I learned how to write proper prompts for ChatGPT and how to communicate with ChatGPT, it is like interacting with a real person, my communication skills have developed”. Another student reported improvements in his grammatical and vocabulary knowledge: “I have developed my language skills, especially vocabulary and grammar. When I interact with ChatGPT, I always ask it to support me with synonyms and correct me whenever I make a grammatical mistake, it is super effective”. All students declared that the training allowed them to experience ChatGPT, which had significantly developed their writing skills. For example:

“I noticed that the training allowed me to use ChatGPT and explore its benefits for developing my writing skills. I would definitely continue using ChatGPT because I noticed great improvement as a result of the immediate feedback I received from ChatGPT on the style of my writing and my grammatical mistakes”.

(S23)

Despite these positive attitudes, few students remained apprehensive as to why, how, and when it should be used, along with the outcomes of such interactions (n = 2). For example, S11 stated that he still did not feel confident about ChatGPT and had no intention of using it in the future:

“Although I learned a lot from the ChatGPT essentials training, I do not plan to use ChatGPT in the future because for me it is a waste of time (…) If I want to find reliable information, I’d prefer Google Scholar. I don’t believe that interacting with AI is realistic for me, especially if I want to develop my speaking skills. To be honest, I am not interested in AI technologies at all.”

(S11)

S21 reported that he remained uneasy about sharing private data with ChatGPT, he stated that:

“While interacting on ChatGPT, I noticed that my chats are encrypted, but it is unclear where this data is stored and processed long-term. Also, I tried to enter my own written text in ChatGPT and realized that it was saved and stored. That’s why I feel that there are alternative AI tools that can be safer than ChatGPT.”

(S21)

5. Discussion

This study has proposed a ‘ChatGPT Essentials’ training course and explored the impact of such pedagogical training on students’ understanding of ChatGPT usage, including their behaviors and perceptions, resulting in the following two themes.

5.1. Transformation of Understanding and Attitudes

The findings of this study support the arguments of Warschauer et al. (2023) and Horn (2024) concerning the importance of teaching students AI literacy. Findings revealed that conducting a pedagogical training had a positive influence on students’ understanding of ChatGPT and consequently improved their practices. We agree with Warschauer et al. (2023) that ChatGPT is not simply a cultural tool to assist students (i.e., accomplishing assignments), but rather transforms the whole writing process. To assist students coping with this transformation, ‘ChatGPT Essentials’ was developed and implemented. In line with Horn’s (2024) findings, the current study indicated that the training had positively shifted students’ attitudes towards using ChatGPT. Similar to previous research findings (Horn, 2024; Xiao & Zhi, 2023; Teng, 2024), many students had misconceptions about ChatGPT prior to the training, with some being skeptical of its potential to enhance their language skills. Issues included unethical practices (i.e., copying and pasting from ChatGPT and recycling content) (Lim et al., 2023) and misconceptions of AI’s role, with students viewing ChatGPT as a replacement for putting in the effort, rather than a tool for transforming behaviors and thinking. This attitude may stem from a lack of AI literacy or ethical awareness. Similar to Horn’s (2024) findings, the results indicated that implementing ‘ChatGPT essentials’ has not only helped students recognize the value of ChatGPT as a language development tool but also simultaneously altered their attitudes, practices, and behaviors.

This study has demonstrated the effectiveness of the ‘ChatGPT essentials’ training, with students reporting positive attitudes towards using ChatGPT. This contrasts with previous studies (Baek et al., 2024; Creely, 2024), in which students continue to express concerns. The training sessions encouraged students to view ChatGPT as a tool to transform their thinking, behaviors, and productivity. Our findings are consistent with those of other studies (Bao & Li, 2023; Xiao & Zhi, 2023; Yan, 2023) in that the students valued the role of ChatGPT in enhancing their linguistic abilities, including grammar and vocabulary, as well as written and communication skills. This study also uniquely reported other benefits, including the positive impact on self-confidence, ethical practices, and development in ChatGPT usage. The participants’ intention to use ChatGPT in the future can be attributed to implementing the ‘ChatGPT essentials’ training course, which: (1) reduced anxiety by raising students’ attention to the proper ethical considerations associated with ChatGPT and (2) provided opportunities for students to experience the benefits of ChatGPT.

5.2. Conservatism and Reluctance

However, even after the training, a small number of students remained skeptical about ChatGPT. This accords with previous studies (Teng, 2024; Alsalem, 2024; Hu & Škultéty, 2024). Such ongoing reluctance may stem from multiple factors, including (1) fear of change; (2) a preference for traditional teaching methods; (3) distrust of AI technologies; and (4) prior AI background knowledge. Similar to Marzuki et al. (2023) and Teng (2024), some students felt that ChatGPT could impair their creativity and writing ability. It is worth noting that the majority of students completely changed their views and became users of ChatGPT. This indicates that students tend to be resistant to accepting any change lacking clear guidance, confirming the importance of training to enhance AI literacy (Horn, 2024; Warschauer et al., 2023).

6. Concluding Remarks

6.1. Theoretical Contributions

A key point to emphasize is that this study contributes to the growing body of CALL literature. Firstly, it represents a new way of exploring learners’ intention to adopt ChatGPT after exposing them to a pedagogical training, allowing them to develop a new understanding of the use of ChatGPT. Secondly, as far as we are aware, this is the first study that proposes a comprehensive pedagogical training model to develop learners’ ethical and pedagogical awareness of ChatGPT. Thirdly, this study is among the first to evaluate the proposed training course from learners’ perspectives. Finally, this study offers a novel theoretical contribution to the TAM model, proposing that the ‘ChatGPT essentials’ training course is a significant external factor that may not only correlate positively with ease of use but also raise students’ awareness of perceived usefulness.

6.2. Practical Contributions

At a broader level, this study suggests that higher education and policymakers need to develop a shared vision for implementing ChatGPT, using a pedagogical training to help students understand ways of using it both ethically and effectively as a regular tool to aid their language learning process, rather than a means for cheating, i.e., copy and paste. We feel it is vital for both teachers and students to accept the integration of ChatGPT, agreeing with Baek et al.’s (2024) argument that: “a dichotomizing approach of either using AI or completely banning it in higher education institutions is unrealistic, given ChatGPT’s widespread adoption” (p. 6).

This study has shown that the proposed training course ‘ChatGPT Essentials’ has a positive impact on shifting students’ attitudes towards ChatGPT and understanding its ethical use. Nevertheless, it is important to note that no two classrooms are the same, and it is vital for teachers to remain flexible, including when it comes to activities or adding further phases to comply with their specific needs. The current study emphasizes the essential role of the teacher in guiding and developing students’ understanding and awareness. It proposes a balanced approach that focuses on using ChatGPT as a tool to assist teachers and students in accomplishing classroom activities, rather than to replace human educators. One of the most important implications of this paper is teaching students the pedagogical principles of ChatGPT, so as to promote understanding of its technical, ethical and pedagogical aspects.

6.3. Limitations and Future Research

This study has a number of limitations and future research recommendations. Firstly, the dual role of the researcher (i.e., being also the teacher) may have inadvertently introduced response bias, as students may have felt inhibited in providing favorable feedback to meet their teachers’ perceived expectations. This may also have prompted students to overstate their satisfaction with the ‘ChatGPT essentials’ training course to please the teacher/researcher. While triangulation and coding reliability procedures were employed to enhance the credibility of the findings, these measures cannot fully eliminate the risk of bias stemming from the researcher-instructor role. Future research could mitigate this risk by involving independent data collectors to reduce potential bias. Secondly, there is also a possibility of researcher bias, with the researcher’s close involvement in designing and implementing the whole research process having a potential influence on the interpretation of the findings. To minimize this bias, several strategies were followed, including triangulation, second coders for the data, and peer reviews. Thirdly, there is a lack of gender diversity among participants, and only a small number were included. All students were male, due to the gender-segregated structure of the institution where the study was conducted. As this restricts the generalizability of the results, this study recommends that future research should involve a larger number of students in mixed-gender settings to capture a broader range of views. An additional limitation is the use of self-reported data in the form of reflective journals and interviews, which are subject to biases such as social desirability and recall errors. These limitations can affect the reliability of the findings, even when a triangulation procedure is used. Finally, this study lacked formal pre-training evaluation forms, due to the researcher’s anticipation that the students would have little (or no) prior exposure to ChatGPT, instead employing background interviews to gain a more nuanced understanding of students’ familiarity, perceptions, and behaviors. While this approach provided a good exploratory starting point, it limits the ability to quantitatively measure the exact degree of change before and after the training. Therefore, this study recommends that future research combine both methods of assessment to systematically measure any changes.