Efl students’ perceptions of using chatgpt as a language learning tool a case in the mekong delta

MINISTRY OF EDUCATION AND TRAININGCAN THO UNIVERSITY PHẠM LÊ THU DUNG EFL STUDENTS’ PERCEPTIONS OF USING CHATGPT AS A LANGUAGE LEARNING TOOL: A CASE IN THE MEKONG DELTA MASTER THESIS MAJ

MINISTRY OF EDUCATION AND TRAINING

CAN THO UNIVERSITY

PHẠM LÊ THU DUNG

EFL STUDENTS’ PERCEPTIONS OF USING CHATGPT AS A LANGUAGE LEARNING TOOL:

A CASE IN THE MEKONG DELTA

MASTER THESIS MAJOR: PRINCIPLE AND METHODS OF ENGLISH

LANGUAGE EDUCATION MAJOR CODE: 8140111

2024 MINISTRY OF EDUCATION AND TRAINING

CAN THO UNIVERSITY

PHẠM LÊ THU DUNG STUDENT CODE: M1622007

EFL STUDENTS’ PERCEPTIONS OF USING CHATGPT AS A LANGUAGE LEARNING TOOL:

A CASE IN THE MEKONG DELTA

MASTER THESIS MAJOR: PRINCIPLE AND METHODS OF ENGLISH

LANGUAGE EDUCATION MAJOR CODE: 8140111

SUPERVISOR 2: NGUYỄN HUỲNH TRANG (A/PROF DR.)

SUPERVISOR 1: NGUYỄN DUY KHANG (A/PROF DR.)

2024

ACKNOWLEDGEMENTS

I wish to extend my heartfelt thanks to the individuals mentioned below for theirsteadfast support and crucial contributions to the completion of this study Theirassistance was essential, and this thesis could not have been achieved without their help.First and foremost, I am profoundly grateful to my supervisor, Associate Professor Dr.Nguyen Huynh Trang, for her constructive feedback, unwavering support, patientguidance, and persistent encouragement throughout the research process Her invaluableadvice and suggestions were instrumental in shaping this work

I also extend my sincere thanks to all the instructors at the School of Foreign Languages,particularly Associate Professor Dr Nguyen Buu Huan, Can Tho University, for theirinsightful instruction throughout my master's program I am profoundly grateful to theparticipants who generously devoted their time and effort to assist me in completing thisresearch

Finally, I owe a special debt of gratitude to my parents and sister, whose constant supportand encouragement were my source of strength throughout this journey Their belief in

me made all the difference

ii

TÓM LƯỢC

Nghiên cứu này điều tra nhận thức của sinh viên ngôn ngữ Anh về việc sử dụngChatGPT như một công cụ hỗ trợ học ngôn ngữ Mục tiêu của nghiên cứu là đánh giánhận thức tổng thể của họ về việc sử dụng ChatGPT trong việc học tiếng Anh và tìmhiểu xem có sự khác biệt nào trong quan điểm giữa sinh viên nam và nữ hay không.Phương pháp nghiên cứu hỗn hợp đã được sử dụng, bao gồm bảng câu hỏi khảo sát 118sinh viên đại học tại Việt Nam và sau đó phỏng vấn bán cấu trúc với 10 sinh viên trong

số đó Kết quả cho thấy sinh viên nhìn chung đánh giá tích cực về ChatGPT, nhấn mạnh

sự dễ sử dụng và những lợi ích của nó trong việc cải thiện kỹ năng tiếng Anh (đặc biệt là

kỹ năng viết), tạo ý tưởng, cung cấp giải thích về cách sử dụng tiếng Anh, kiểm tra vàsửa lỗi bản nháp viết, tóm tắt văn bản, và nhiều hơn nữa Tuy nhiên, cũng có nhữngthách thức như nguy cơ ChatGPT đưa ra thông tin sai lệch, ngôn ngữ không tự nhiên,mối quan tâm đạo đức, và rủi ro phụ thuộc quá mức Không có sự khác biệt đáng kể nàođược tìm thấy giữa quan điểm của sinh viên nam và nữ Nghiên cứu cũng đưa ra đề xuấtcho sinh viên và giảng viên Đối với sinh viên, việc sử dụng ChatGPT có thể nâng cao

kỹ năng ngôn ngữ của họ, tuy nhiên, họ nên đánh giá một có sự cẩn thận đổi các câu trảlời mà ChatGPT đưa ra Đối với giáo viên, ChatGPT là một công cụ hỗ trợ quý giá, vàviệc phát triển chuyên môn được khuyến khích để tích hợp nó hiệu quả vào giảng dạytrên lớp

Từ khoá: ChatGPT, nhận thức của sinh viên, khác biệt giới tính, AI trong giáo dục.

iii

ABSTRACT

The study investigates English as a Foreign Language (EFL) students’ perceptions of

using ChatGPT as a language learning tool and whether there are any differencesbetween male and female students' perceptions A mixed-method approach wasemployed, combining a questionnaire administered to 118 university students in Vietnamand semi-structured interviews with 10 of these students The findings reveal thatstudents generally perceive ChatGPT positively, highlighting its ease of use and itsbenefits in enhancing English skills (especially writing), generating ideas, providingexplanations for English language uses, proofreading and error correction of writingdrafts summarizing texts, and more However, challenges such as the potential forincorrect information, occasional unnatural language, ethical concerns, and the risk ofover-reliance were also noted No significant differences were found between male andfemale students' perceptions The study concludes with implications for both studentsand teachers For students, using ChatGPT can enhance their language skills, thoughthey should critically evaluate its output For teachers, ChatGPT serves as a valuablesupplementary tool, and professional development is recommended to effectivelyintegrate it into classroom instruction.

Key words: ChatGPT, student perceptions, gender differences, AI in education.

iv

DECLARATION

The thesis entitled “EFL Students’ Perceptions of Using ChatgGPT as a LanguageLearning Tool: A Case in the Mekong Delta” is conducted under the supervision ofAssociate Professor Doctor Nguyen Huynh Trang from University of Economics Ho Chi

Minh City and Associate Professor Doctor Nguyen Duy Khang from Can Tho University

I declare that the information reported in this paper is entirely my own work, except where due references is made The present thesis has not been accepted for any other diploma

Can Tho, August 2024

Supervisor 2 M.A student researcher

Associate Professor Dr Nguyen Huynh Trang Pham Le Thu Dung Supervisor 1

Associate Professor Dr Nguyen Duy Khang

v

TABLE OF CONTENTS

ACKNOWLEDGEMENTS ii

TÓM LƯỢC iiii

ABSTRACT iv

DECLARATION v

TABLE OF CONTENTS vii

LIST OF TABLES ix

LIST OF ABBREVIATIONS x

CHAPTER ONE 1

1.1 Rationale for the study 1

1.2 Research aims 2

1.3 Research questions 2 1.5 Scope of the study 2 1.6 Organization of the thesis 3

CHAPTER TWO: LITERATURE REVIEW 4

2 1 Definition of the key constructs 4

2.1.1 Perceptions 4

2.1.2 Language learning tools 4

2.1.3 ChatGPT 5 2.2 AI-based chatbots in EFL language learning 5 2.3 ChatGPT’s usefulness in EFL language learning 6 2.4 ChatGPT and student’s motivation in learning English 8 2.5 Challenges and ethical considerations of ChatGPT in EFL language learning 9 2.6 The use of the Technology Acceptance Model (TAM) in education settings 11 2.7 Related studies 13

CHAPTER THREE: RESEARCH METHODOLOGY 15

3.1 Research design 15

3.2 Participants 16

3.2.1 Participants for the questionnaire 16

Table 3.1 Participants’ major 16

Table 3.2 Participants’ year of study 16

Table 3.3 Participants’ gender 17 3.2.2 Participants for the interviews 17 3.3 Research instruments 17 3.3.1 Questionnaires 17 Table 3.4 Summary of the second and third sections of questionnaires 19

vi Table 3.5 Cronbach’s alpha values 19

3.3.2 Semi-structured interviews 20

Table 3.6 Summary of the research instruments 21

3.4 Data analysis 21

3.4.1 Analysis of the questionnaire 21

Table 3.7 The application of five-point Likert scale in SPSS (Field, 2013) 21

3.4.2 Analysis of the semi-structured interviews 22 3.5 Procedures of the study 22 Table 3.8 Data collection procedures 22

CHAPTER FOUR: FINDINGS 23

4.1 Findings from the questionnaire and interview 23

4.1.1 Multiple choice section 23

Table 4.1 The frequency of ChatGPT used by students in English language learning (%) 23

Table 4.2 Frequency of tasks using ChatGPT as a language learning tool 24

Table 4.3 The percentage of skill EFL students use ChatGPT to improve .26

4.1.2 5 point Likert-scale section 26

Table 4.4 Mean score of students’ perceptions of using ChatGPT as a language learning tool 26

Table 4.5 Mean scores of the clusters of students’ perceptions 27

Findings for the first research question: What are EFL students' perceptions of using Chatgpt as a language learning tool? 27

Table 4.6 Percentage of students’ perceived ease of use of ChatGPT as a language learning tool 27

Table 4.6 Percentage of students’ perceived usefulness of ChatGPT as a language learning tool 28

Table 4.7 Student's attitude toward using ChatGPT as a language learning tool .32

Table 4.8 Student's perceptions regarding the disadvantages using ChatGPT as a language learning tool .34

Students' suggestions to improve ChatGPT as a language learning tool .37

Findings for the second research question: Are there any differences in the perceptions of male and female EFL students of using ChatGPT as a language learning tool? 38

Table 4.9: Independent t-test for male and female in using ChatGPT as a language learning tool 38

4.3 Discussion 38

CHAPTER FIVE: CONCLUSIONS 42

5.1 Summary of the major findings 42

5.2 Implications 43

5.4 Limitations 44

5.5 Recommendations for further research 44

vii REFERENCES 45

APPENDIX A 50

APPENDIX B 54

APPENDIX C 59

APPENDIX D 60

APPENDIX E 80

APPENDIX F 99

LIST OF TABLES

Table 3.1 Participants’ major 16

Table 3.2 Participants’ year of study 16

17 Table 3.4 Summary of the second and third sections of questionnaires 19 Table 3.5 Cronbach’s

alpha values 19 Table 3.6 Summary

of the research instruments 21 Table 3.7 The

application of five-point Likert scale in SPSS 21 Table 3.8

Data collection procedures 22

Table 4.1 The frequency of ChatGPT used by students in English language learning (%)

23

Table 4.2 Frequency of tasks using ChatGPT as a language learning tool 24

Table 4.3 The percentage of skill EFL students use ChatGPT to improve. 26

Table 4.4 Mean score of students’ perceptions of using ChatGPT as a language learning

tool 26

Table 4.5 Mean scores of the clusters of students’ perceptions 27

Table 4.6 Percentage of students’ perceived ease of use of ChatGPT as a language

learning tool .27

Table 4.7 Student's attitude toward using ChatGPT as a language learning tool. 32

Table 4.8 Student's perceptions regarding the disadvantages using ChatGPT as a

language learning tool .34

Table 4.9: Independent t-test for male and female in using ChatGPT as a language

learning tool 38

ix

LIST OF ABBREVIATIONS

ChatGPT Chat Generative Pre-trained Transformer EFL

English as a foreign language

SPSS Statistical Package for the Social Sciences TAM

Technology Acceptance Model

AI Artificial Intelligence

CHAPTER ONE INTRODUCTION

This chapter addresses the background, the research aims, research questions, the

significance of this study, and the organization of the thesis.

1.1 Rationale for the study

In the current era of Industry 4.0, English plays a pivotal role as the lingua franca ofinternational communication and commerce, it is a gateway to information, innovation,and advancement The majority of online content, academic resources, and scientificpublications are available in English, making it an essential tool for those seekingknowledge and staying informed about global developments For students, learning

English is a strategic investment in their future It broadens their horizons, enhancestheir employability, and provides access to a vast pool of educational resources.

Presently, the technology that is garnering significant attention is artificial intelligence(AI) Artificial intelligence is "the ability of a machine to exhibit intelligent behavior,such as learning and problem-solving" (Newell & Simon, 1972, p 105) With thecapacity to learn, process information, and interact with humans, the integration of AIand natural language processing technologies into education has brought aboutinnovative tools that could revolutionize language learning experiences Among thesetechnologies, ChatGPT, an advanced language model created by OpenAI, is receivinggrowing attention ChatGPT has shown exceptional ability in understanding andgenerating human-like text, positioning it as a promising tool to assist EFL students intheir language learning journey

Ellis (1994) states that a language learning tool is any resource that can be utilized toassist language learners in acquiring a language ChatGPT can explain concepts, offersuggestions for written content, answer questions, and so on With its strong abilities,ChatGPT becomes a promising tool for EFL students to learn languages

However, as a technology emerging in late 2022, there have not been many studies onthe perceptions of EFL students regarding the use of ChatGPT for language learning,especially in the Mekong Delta region of Vietnam Additionally, there is limitedprevious quantitative research on ChatGPT Furthermore, current research has paidlimited attention to how male and female EFL students perceive ChatGPT as languagelearning tool This mixed-methods research aims to address these gaps by exploring theperceptions of EFL students towards ChatGPT, with a specific focus on the MekongDelta region of Vietnam, and investigating whether there are any gender-baseddifferences in these perceptions

1 What are EFL students' perceptions of using Chatgpt as a language learning tool?

2 Are there any differences in the perceptions of male and female EFL students of usingChatGPT as a language learning tool?

1.4 Significance of the study

The current study investigates EFL students' perceptions of using ChatGPT as alanguage learning tool in the Mekong Delta Understanding these perceptions is crucial

as it provides insights into the potential benefits and challenges of integrating AI-basedtools in language education

For students, this research highlights how ChatGPT can support their language learningprocess By identifying the specific ways in which ChatGPT aids language acquisition,the study offers valuable information on the tool’s effectiveness in enhancing students’language skills, fostering autonomy, and providing personalized learning experiences.For educators, the study’s findings offer a deeper understanding of how AI tools likeChatGPT can be incorporated into language teaching Insights into students’ preferencesand challenges with ChatGPT will enable teachers to develop more effective teachingstrategies, tailor their instructional methods, and better support students in their languagelearning journey Teachers can use these insights to guide students in maximizing thebenefits of ChatGPT, addressing any difficulties they encounter, and fostering a moreinteractive and supportive learning environment

Furthermore, this research contributes to the broader academic discourse on the use of

AI in education It serves as a reference for other researchers interested in exploring theintersection of AI and language learning, providing a foundation for future studies andfacilitating a deeper exploration of innovative educational technologies

1.5 Scope of the study

This study was conducted to investigate the perceptions of English majors at a university

in the Mekong Delta of Vietnam towards using ChatGPT as a language learning tool

2

1.6 Organization of the thesis

The thesis is structured into five chapters:

Chapter One: Introduction provides an overview of the background, purpose,

research questions, significance, and scope of the study

Chapter Two: Literature Review presents existing literature on perceptions,

language learning tools, and ChatGPT It includes a detailed examination of usingChatGPT for learning English, focusing on AI-based chatbots in EFL language learning,the benefits and challenges of ChatGPT, its impact on motivation, and relevant studies

Chapter Three: Research Methodology outlines the research design, data collection instruments, and data analysis procedures used in this study Chapter Four: Findings and Discussion presents and analyzes the study's results, offering a comprehensive

discussion of the findings

Chapter Five: Conclusions summarizes the major findings, draws conclusions,

discusses implications, acknowledges limitations, and provides recommendations forfuture research.

3

CHAPTER TWO LITERATURE REVIEW

This chapter reviews the literature relevant to using AI-based chatbots, especially

ChatGPT, as a language learning tool for EFL students and explains various aspects related to the topic of this research It starts with definitions of key terms such as

perceptions, language learning tools, and ChatGPT The next section focuses on the usefulness and challenges of using ChatGPT as a language learning tool Related

studies on this field are also included in the final section.

2 1 Definition of the key constructs

2.1.1 Perceptions

Efron (1969) states that perception is the primary way individuals engage cognitivelywith their surroundings The study of perception holds particular importance in both

philosophy and science, as all conceptual knowledge is based on or originates from thisfundamental form of awareness (Efron 1969) Gregory (1970) states that perceptions areregarded as hypotheses in which people attempt to infer something from what they seeand think Gregory highlights the importance of prior knowledge and experiences inshaping perceptions Meanwhile, Lindsay and Norman (1977) believe that perception isthe process of interpreting a circumstance into something that is important to one's pastexperiences.

In other words, perceptions refer to individuals' consciousness of their actions andbehaviors Therefore, it is crucial to examine students' viewpoints before encouragingthem to explore new things or implementing new things in the learning journey

In the context of this study, “perceptions” is defined as students' awareness of ChatGPT

as a language tool for learning English

2.1.2 Language learning tools

According to Warschauer (1996), language learning tools are "resources or technologiesused to facilitate language learning, ranging from traditional materials such as textbooksand language labs to contemporary computer-based programs and online platforms." Thepurpose of a language learning tool is preeminent to assist language learners in theireducational endeavors Having an effective and efficient tool for language learning cansignificantly enhance learners' engagement and enthusiasm acquisition, therebyexpediting the language process In the contemporary technological era, AI-basedchatbots, especially ChatGPT, stands out as a promising tool to support languagelearners Within the scope of this study, language learning includes the four skills oflistening, speaking, reading, and writing, as well as vocabulary and grammar, but does

4not encompass linguistics This is because ChatGPT's capabilities are not yet sufficientfor studying the linguistics of a language

2.1.3 ChatGPT

ChatGPT, developed by OpenAI, is an AI-based chatbot that mimics humanconversations using user input, as defined by Opara (2023) With ChatGPT's ability toengage learners in natural language interactions, it has the potential to enhancetraditional teaching methods This tool can offer a personalized and interactiveexperience that adapts to individual learning needs and preferences ChatGPT can createhuman-like dialogues by comprehending the context of a conversation and producingsuitable replies It utilizes a deep learning model known as GPT-3, which has beentrained on an extensive dataset of conversations (Deng & Lin, 2023) In other words,

ChatGPT is a computer program that can communicate and generate human-like text inresponse to a wide range of prompts and questions

2.2 AI-based chatbots in EFL language learning

Advancements in technology have paved the way for innovative educational tools,among which chatbots have emerged as a notable trend These digital conversationalagents are designed to engage with users, simulating human interactions in variouscontexts This section examines the impact and potential of chatbots in enhancinglanguage learning and students’ motivation.

Numerous studies have explored the beneficial effects of AI-assisted language learningtools on the language acquisition skills of English learners (Divekar et al, 2021; Bašić etal., 2023; Fitria, 2023) More specifically, AlKhayat (2017) investigated theeffectiveness of chatbots in improving English language proficiency It involved 22undergraduate students from an Arab university in Saudi Arabia, divided into threegroups The research focused on these students using a virtual agent for English practice,guided by situated learning and constructivist theories Survey results and teacherfeedback revealed that student proficiency levels and language skills play a vital role indetermining the success of chatbot integration in the classroom

Ebadi and Amini (2022) explored the attitudes of Iranian university students toward AIpowered mobile applications and analyzed how social presence and human-like featuresinfluence learner motivation through the use of chatbots 256 English as a foreign

language learners interacted with the chatbot "Computer Simulation in Educational

Communication (CSIEC)." Data collected included audio-recorded practices,

transcriptions, social presence scales, motivation measures, human-likeness ratings, and

a focus group interview The quantitative analysis, using Structural Equation Modeling,show that learner motivation is influenced by social presence and the human-like

characteristics of the tool Qualitative findings indicated that CSIEC's involvement

5boosted learners' motivation, enthusiasm, and confidence in learning English This studyoffers insights for the future use of chatbots as educational companions and may informpersonalized assessment and feedback strategies

Kohnke (2023) created a chatbot to support second-language learners at a highereducation institution in Hong Kong with their independent language studies Theresearch used a questionnaire with 128 participants and followed up with semi structuredinterviews with 12 participants to gain a detailed understanding of how learnersinteracted with the chatbot The results revealed that participants enjoyed theirinteractions with the chatbot, both in and out of the classroom, and believed it helpedimprove their English language skills This is especially significant in higher education,where students are expected to take responsibility for their own learning and engage inself-directed study This suggests that chatbots can be a valuable resource for offeringpersonalized support and bridging the gap between classroom learning and practicallanguage use

Zhang and colleagues (2023) developed a chatbot for independent, out-of-class training

on identifying logical fallacies in EFL argumentative writing 15 Chinese EFL studentsparticipated in the five-week training They used interviews and transcript analysis tounderstand how the training impacted participants' self-efficacy and EFL argumentativewriting skills To support their findings, pre-post questionnaires on writing self-efficacywere administered, and argumentative writings were evaluated using the Illinois CriticalThinking Essay Scoring Rubric The findings showed that the training improvedargumentative writing skills, although it might have slightly lowered self-efficacy.

In conclusion, the studies report that AI-assisted tools, particularly chatbots, have apositive impact on various aspects of English language learning, such as language,motivation, independent learning, and argumentative writing skills If used correctly andwith the right strategies, chatbots can have a significant positive influence on languagelearning However, it is the importance of balancing technological implementation withlearners' needs and psychological factors to ensure the most effective outcomes

2.3 ChatGPT’s usefulness in EFL language learning

Studies on the use of ChatGPT in English language learning for EFL students haveexplored both its effectiveness and potential as a supplementary tool in a classroomsetting Overall, these studies demonstrate that students generally find ChatGPTbeneficial for enhancing their language skills, especially in writing, but there are alsosome noted limitations

For example, Abd et al (2023) conducted a quantitative study with 181 students fromUniversiti Teknologi MARA in Malaysia, which found that the majority of studentsperceived ChatGPT as a useful tool for improving writing They highlighted the

6platform’s ability to provide personalized feedback, vocabulary suggestions, andgrammatical corrections, which not only improved their writing skills but also increasedtheir confidence in self-expression The findings underscore how ChatGPT can makeEnglish learning more enjoyable and engaging for students, reinforcing its value in theclassroom

Similarly, Bok and Cho (2023) examined how 71 Korean students in an academicwriting course used ChatGPT for paragraph revisions Students appreciated the instantfeedback on grammar, vocabulary, and paragraph coherence, citing the tool's availabilityanytime and anywhere as a significant advantage However, challenges such as the lack

of detailed explanations and unclear feedback were also reported The study suggestedthat ChatGPT could serve as a useful tool for writing revision but should becomplemented with teacher input to address its shortcomings This points to theimportance of blending AI and human feedback for a more balanced learning approach

In another study, Vera (2023) investigated the impact of ChatGPT on the Englishlearning process of Chilean university students In an experimental study with a controland experimental group, Vera found that students who used ChatGPT showed improved

language proficiency compared to those in traditional classroom settings The resultsemphasized the tool's ability to provide individualized, interactive learning opportunitiesthat enhance the learning experience However, Vera cautioned against viewingChatGPT as a replacement for traditional instruction, stressing that it should be seen as asupplementary tool that adds value through personalization and engagement.

Beyond student perspectives, educator views have also been explored Mohamed (2023)conducted interviews with 10 EFL faculty members at Northern Border University inSaudi Arabia to understand their opinions about ChatGPT’s role in language acquisition.While some faculty members acknowledged its value in providing prompt and accurateanswers, others expressed concern over its potential to weaken students' critical thinkingand research skills They also noted that AI could unintentionally reinforce biases orincorrect information Despite these concerns, the educators agreed that ChatGPT could

be a helpful complement to traditional teaching methods, but more research is needed tofully understand its impact

To address these mixed views, Tseng and Waschauer (2023) proposed a five-steppedagogical framework for educators, aimed at integrating ChatGPT into language

corroborating, and integrating—focuses on educating learners about how to responsiblyand ethically use ChatGPT while acknowledging its limitations It provides a structuredapproach to incorporating AI-generated content into students’ writing, ensuring theyunderstand how to appropriately use and cite AI tools in their work

7

In conclusion, while ChatGPT offers promising opportunities for personalized,interactive language learning, both students and educators have identified importantconsiderations Students appreciate the tool's ability to improve writing and provideconvenient feedback, though its limitations in detailed explanations and feedback clarityremain areas of concern Educators also recognize the potential benefits but are wary ofpossible drawbacks related to students' critical thinking skills Therefore, a careful,ethical integration of ChatGPT is recommended, ensuring it complements rather thanreplaces traditional teaching methods Frameworks such as that proposed by Tseng andWaschauer provide useful guidance for navigating this integration and leveragingChatGPT’s full potential in EFL contexts

2.4 ChatGPT and student’s motivation in learning English

Students can access ChatGPT anytime and anywhere, making learning English moreconvenient ChatGPT can provide quick and constructive feedback, offering interactiveand personalized tools to hone writing skills and enhance motivation (Jiang, 2022;Meunier et al., 2022; Yan, 2023) It means that using ChatGPT allows learners toparticipate in interactive tasks and activities that are designed specifically for theirlanguage level and interests This personalized approach can boost their sense of

accomplishment and keep them engaged in the learning process.

Ali and colleagues (2023) examined how ChatGPT affects motivation for learningEnglish while taking into account a number of other elements that affect studentinvolvement 80 teachers and students who used ChatGPT in its initial phase in early

2023 were the subjects of the study, which used a quantitative research design to gatherdata A non-probability sampling method was used to choose the sample through anonline survey The findings show that ChatGPT has an advantageous impact onencouraging students to improve their reading and writing abilities The fact thatstudents' questions were answered and feedback was provided has motivated them intheir English learning process

Similarly, Song and Song (2023) investigated how ChatGPT influences the writing skillsand motivation of Chinese EFL students The researchers employed a mixed methodsapproach, integrating quantitative and qualitative techniques The quantitative phasefeatured a pre-test and post-test design to evaluate writing skills Fifty EFL students,with similar proficiency levels, were randomly assigned to either an experimental groupusing AI-assisted instruction via ChatGPT or a control group receiving traditionalinstruction Their writing samples were scored using established rubrics Simultaneously,semi-structured interviews were conducted with some participants to explore theirmotivation and experiences with AI-assisted learning The findings reveal significantimprovements in both writing skills and motivation among students who participated inAI-assisted instruction

8Similarly, Yıldız (2023) investigated how incorporating ChatGPT-generated dialoguesinto language teaching materials affects language learners' motivation The quasi

experimental study involved second-year university students from various departmentsand consisted of 10 sessions, each lasting 15 minutes The experimental group usedChatGPT for post-lesson activities, while the control group followed the traditionalcurriculum Motivational changes were measured using the 'Motivational Strategies forLearning Questionnaire' (Pintrinch & De Groot, 1990) along with pre- and post-tests,with data analyzed using SPSS The results showed significant differences in motivationsubcategories (such as self-regulation, intrinsic values, and test anxiety) across differentmajors ChatGPT's feedback, personalized support, and flexible guidance were found tosignificantly enhance students' motivation and improve language learning outcomes.However, Woo (2023) presented a slightly different perspective by focusing on HongKong secondary students' perceptions of learning to write with ChatGP In theworkshop-format study, twenty-one students were taught prompt engineering skills andcompleted a 500-word writing task with the help of ChatGPT Data were gathered usingpre-workshop and post-workshop questionnaires, as well as think-aloud protocols duringthe writing task The results indicate that there was no significant change in students'motivation levels before and after the workshop, despite a slight increase in average

motivation scores Students experienced a high cognitive load, particularly with promptengineering, yet they reported overall high satisfaction with the workshop Thesefindings indicate that while ChatGPT can be a useful tool, it might not independentlyincrease student motivation and can be cognitively challenging to use effectively.

In conclusion, there are diverse perspectives on ChatGPT's impact on language learningmotivation While several studies highlight its potential to enhance motivation throughinteractive and personalized feedback, there are still studies that show ChatGPT does nottruly motivate students Therefore, it is important to conduct careful research beforeimplementing this tool and to tailor its use to each learning environment to maximize itspotential

2.5 Challenges and ethical considerations of ChatGPT in EFL language learning

ChatGPT has emerged as a powerful tool for language learning, offering engaginginteractions and personalized practice opportunities for English language learners (EFLlearners) However, alongside its potential benefits, concerns have also arisen regardingits potential impact on learning and student autonomy

Hostetter and colleagues (2023) conducted a study to examine whether artificialintelligence (AI) could generate written content that remained indistinguishable Thestudy involved 83 students and 82 faculty members The researchers gave theparticipants four writing samples: three from college students and one from ChatGPT

3-an AI chatbot They discover that neither faculty nor students could identify

AI-9generated text with any degree of certainty The results show that effectivelydifferentiating AI-generated text from human-authored work is a substantial issue forboth students and instructors Participants also voiced ethical worries about AI's place ineducation, and professors reported increased concern about how it would affect teachingmethods The ability of faculty members to identify AI-generated content wassignificantly increased by prior ChatGPT-3 experience and expertise in responsestructure analysis, indicating that familiarity and domain-specific analytical abilities canimprove identification accuracy

Mohammadkarimi (2023) investigated how teachers perceive academic dishonesty inEFL students' papers in the setting of AI 67 teachers participated in the study and sharedtheir opinions via questionnaires and interviews The results show that teachers havediffering opinions about the advantages of AI technologies for students Some teacherssee advantages, while others worry about how it would affect academic integrity.Teachers all agreed that AI had a detrimental impact on students' commitment toacademic honesty because they see it as encouraging dishonesty and impeding skilldevelopment The study underscores the necessity of addressing ethical consequencesand highlights the importance of teachers in spotting AI-generated assignments.Activities for addressing problems, tools for detecting plagiarism, and the incorporation

of AI into instruction are some of the strategies mentioned Although some teachers

noted difficulties in spotting academic dishonesty connected to AI, the study emphasizesthe significance of thorough training and support for teachers to properly use AI whileretaining academic integrity.

Hieu and colleagues (2023) investigated how Vietnamese EFL students from twouniversities perceive the difficulties of using ChatGPT in writing activities Theresearchers used qualitative methods, conducting semi-structured interviews with 20students The study highlights several potential drawbacks to using ChatGPT in EFLwriting exercises First, there is a concern that students might become overly reliant onthe tool, hindering their development of independent writing skills This aligns withbroader discussions about technology dependence in education Second, the study raises

a red flag about the accuracy and appropriateness of content generated by ChatGPT.EFL learners, especially, might be vulnerable to misleading or grammatically incorrectinformation Integrating fact-checking and critical evaluation skills alongside ChatGPTuse could be a valuable solution Finally, the research suggests that ChatGPT might limitopportunities for creative expression Creativity is a fundamental aspect of languagelearning, as it empowers us to express ourselves in unique and meaningful ways.Incorporating prompts or exercises that encourage students to go beyond basicfunctionalities of ChatGPT could address this limitation

10Karataş and colleagues (2024) conducted a qualitative case study involving 13preparatory class students at the School of Foreign Languages in a Turkish university.Over four weeks, the researcher, acting as a language teacher, introduced ChatGPT tothe students through various learning experiences The qualitative data collected frominterviews were analyzed using thematic analysis The findings reveal that, whileChatGPT-assisted language learning offers advantages such as improving students'writing, grammar, and vocabulary acquisition, and enhancing motivation, it also hasnotable disadvantages These include over-reliance on the tool, negative impacts oncritical thinking and learner autonomy, the generation of incorrect information, andethical concerns related to privacy and academic integrity It suggests that teachersshould play a crucial role in guiding students to use AI responsibly, helping themdevelop critical thinking skills, and maintaining the integrity of the learning process Bydoing so, students can benefit from the technological advancements while avoidingpotential pitfalls

In conclusion, while ChatGPT has the potential to enhance learning throughpersonalized feedback and support, it also presents significant challenges that must becarefully managed Challenges such as difficulty in distinguishing AI-generated contentfrom human-authored work, concerns about academic dishonesty, over-reliance ontechnology, and potential impacts on critical thinking and creativity are prominent Toaddress these challenges, it is crucial to develop strategies that balance the advantages of

ChatGPT with the need for maintaining academic integrity, fostering independentlearning, and encouraging critical engagement

2.6 The use of the Technology Acceptance Model (TAM) in education settings

Since this study uses the Technology Acceptance Model (TAM) as the framework for itsquestionnaire, it is essential to review how TAM has been applied in previous research.TAM, developed by Davis (1989), is widely recognized for examining the acceptanceand use of new technologies by focusing on perceived usefulness (PU) and perceivedease of use (PEOU) These factors are seen as critical in shaping users' attitudes,behavioral intentions, and actual usage of technology In recent years, TAM has beenadapted to explore user perceptions of artificial intelligence tools like ChatGPT,particularly in educational settings This section highlights several studies that haveutilized TAM to investigate factors influencing user acceptance of ChatGPT in variouscontexts

Yilmaz et al (2023) employed TAM in their investigation of university students’perceptions of ChatGPT, focusing on variations across gender, grade level, major, andprior experience with AI tools Their quantitative study of 239 students from the Scienceand Mathematics Education Program in Kazakhstan found that participants generallyheld positive perceptions of ChatGPT The study highlighted significant differences in

11perceptions based on gender, with female students rating the "Perceived ease of use" ofChatGPT lower than their male counterparts Interestingly, no significant differenceswere found across grade levels, but major and prior experience with AI played roles inshaping perceptions, particularly in the "Perceived social influence" dimension Thisstudy's findings provide valuable insights into the nuanced ways different demographicfactors can influence users' acceptance of ChatGPT, especially through the lenses ofperceived ease of use and social influence, both key elements in TAM

Ma et al (2024) expanded the application of TAM by surveying 784 ChatGPT users inChina, providing a large-scale analysis of how users' perceived usefulness (PU) and ease

of use (PEOU) influenced their behavioral intentions and actual use behavior Theirfindings confirmed the interrelationships between these core TAM constructs, showingthat behavioral intention acted as a mediator between both PU and PEOU and actual usebehavior Additionally, they found that behavioral intention had a stronger impact on usebehavior among female and urban users compared to males and rural users, respectively.This study underscores the relevance of TAM in understanding technology adoptionacross different user demographics, suggesting that individual and contextual factorssuch as gender, location, and education level can amplify the pathways from perceivedusefulness and ease of use to actual usage

Saif et al (2024) applied TAM in a more specialized educational context, examininghow stress and anxiety among students contribute to their adoption of ChatGPT as atechnology-assisted solution The study surveyed 156 final-year management science

students in Pakistan and used Structural Equation Modeling (SEM) to analyze therelationship between students' academic stress, their adoption of ChatGPT, and theirperceptions of the tool’s usefulness and ease of use The study revealed that stressinduced anxiety drove students to adopt ChatGPT for completing tasks, with PEOU and

PU playing pivotal roles in reducing stress and shaping positive attitudes toward thetool This study highlights the psychological dimensions in the adoption of ChatGPT,suggesting that TAM can be effectively utilized to explore not only technological factorsbut also behavioral concerns, such as anxiety and stress, that may motivate technologyuse in academic contexts

Overall, these studies illustrate the versatility and robustness of TAM in examining theadoption of AI tools like ChatGPT By incorporating variables such as gender,educational background, and psychological factors, researchers have extended thetraditional TAM framework to capture a broader range of influences on user behavior.These findings underscore the importance of considering both technological andindividual factors when studying the acceptance of new educational technologies

12

2.7 Related studies

In recent years, the fusion of AI and Natural Language Processing (NLP) technologieswith language education has marked a significant shift in EFL learning approaches Theintroduction of AI-driven conversational agents, notably ChatGPT, has sparked growinginterest as a potential solution to address the multifaceted challenges faced by EFLlearners This section investigates the existing body of literature, focusing on studies thatinvestigate the perceptions and experiences of both EFL students and teachers regardingthe incorporation of ChatGPT as a language learning tool

Li and colleagues (2023) used the narrative inquiry approach to capture the unique,colorful experiences of three graduate students in Applied Linguistics who usedChatGPT for a two-month period Three different data sources are gathered: writtenreflections, focus group interviews, weekly notes, and final narratives Their study offersimportant new information about graduate students' perceptions of the advantages anddifficulties of using ChatGPT in academia, including scaffolding for term clarification,teaching examples, cognitive engagement, and confidence building, as well as worriesabout fake information, a lack of citations, and inappropriate ChatGPT use by students.Shaikh and colleagues (2023) investigated the use of ChatGPT in formal Englishlanguage learning with 10 participants from diverse backgrounds, including differentlevels of English proficiency, educational levels, and nationalities Using a standardquestionnaire, the study collected feedback on ChatGPT’s effectiveness for tasks such asconversation, writing, grammar, and vocabulary The quantitative analysis of responsessuggests that ChatGPT is a promising tool for formal English language learning

However, the study acknowledges limitations, such as potential biases from participants'familiarity with technology and the need for further research into the tool's long-termusability and effectiveness.

Xiao and Zhi (2023) explored students' experiences with ChatGPT and their views on itsrole in language learning through a qualitative study involving five students from aleading international university in China, majoring in Marketing, Translation, and DataScience Their findings suggest that ChatGPT can be a valuable learning aid and assistwith language tasks The students demonstrate critical evaluation skills in assessingChatGPT's output and adjusting prompts to enhance learning, mitigating potentialacademic integrity issues While the study indicates that ChatGPT can provideimmediate feedback and personalized learning, its small sample size of five studentsmay limit the generalizability of the results

In Vietnam context, there is a study by Ngo (2023) explored university students'perceptions of using ChatGPT as a learning tool The research examined both theadvantages and disadvantages students encountered To gather data, researchersdistributed an online survey (questionnaire) to 200 students and conducted in-depth

13interviews with 30 The findings revealed a generally positive student sentiment towardsusing ChatGPT for learning Students highlighted benefits like time savings, access todiverse information, personalized guidance, and brainstorming assistance for writing.However, the study also identified challenges, such as difficulty evaluating sourcecredibility, proper citation formatting, and accurate use of language nuances like idioms.Addressing these issues through educational interventions or integrating fact-checkingand citation tools within ChatGPT itself could significantly improve its effectiveness inacademic settings

The studies reveal the complex connection between ChatGPT and education Studentsexpress concerns about accuracy and data security even though they are aware of thepossible benefits of ChatGPT for language learning and improved learning abilities.Collaboration among stakeholders is advised to guarantee ethical AI inclusion Teachershold a variety of opinions; although some see ChatGPT as advantageous, others areconcerned about its ethical and privacy risks These various viewpoints show theimportance of guidelines and moral considerations when implementing AI in education Despite the growing interest in utilizing artificial intelligence language models likeChatGPT in language learning environments, there remains a gap in understanding EFLstudents' perceptions of its effectiveness and potential challenges, especially in theMekong Delta Besides, existing research is limited by small sample sizes or the depth

of study, often relying solely on questionnaires or interviews This study aims to addressthis gap by employing both questionnaires and semi-structured interviews to gain acomprehensive understanding of EFL students’ perceptions of ChatGPT as a learningtool

CHAPTER THREE RESEARCH METHODOLOGY

This chapter begins with a brief discussion of the research design used in this study Next, participants, research instruments, and data collection procedure and data analysis are discussed.

3.1 Research design

This study was designed as descriptive research with a mixed method employed to collect thedata The three common approaches to conducting research are quantitative, qualitative,and mixed methods

Creswell and Creswell (2017) describe quantitative research as a means of testingobjective theories by examining the relationship among variables Quantitative researchaims to gather numerical data to explain, forecast, or control phenomena of interest(Brown, 2010) The outcome of quantitative research typically manifests as numbers,presented in tables, graphs, or other statistical forms Common data-collectioninstruments in quantitative research include questionnaires, which are widely used inconstruction research (Smith, 2015)

On the other hand, qualitative research aims to build a detailed and nuancedunderstanding of phenomena within their natural settings (Brown, 2010) Denzin andLincoln (2011) note that qualitative research provides insight into participants' livedexperiences and allows researchers to view the world from their perspectives The goal

is to gain a deeper understanding of social or human issues by examining them fromvarious viewpoints (Denzin & Lincoln, 2011) This approach uses diverse datacollection methods, including observations and textual or visual analysis, with

interviews being one of the most frequently utilized methods (Smith, 2015).

According to Johnson and Onwuegbuzie (2004), mixed methods research involvesgathering, analyzing, and combining both quantitative and qualitative data Thisapproach provides researchers with a comprehensive understanding by leveraging thestrengths of both methods and addressing their individual limitations In this study, amixed-method design was used to explore students' perceptions of ChatGPT as a tool forlearning English Quantitative data from questionnaires were collected to gain insightsinto students' perceptions, while qualitative data from semi-structured interviews wereused to gain a deeper understanding of their viewpoints (Johnson & Onwuegbuzie,2004)

15

3.2 Participants

3.2.1 Participants for the questionnaire

118 EFL students (General program) participated in this survey This included studentsfrom both the English Language and English Teacher Education majors, both of whichare part of the Department of Foreign Languages This is a convenience sampling

Convenience sampling is a non-probability sampling technique where participants areselected because they are readily available and easy to access (Fraenkel et al., 2012).This study chose convenience sampling because it is cost-effective, requiring minimalinvestment of time and resources in participant recruitment Additionally, it is time

efficient, allowing for the quick collection of data by targeting readily available

individuals Lastly, convenience sampling is feasible in situations where obtaining arandom sample of the target population is challenging or impractical

Among the 118 EFL students who participated in this study , 65 students were from theEnglish Studies program, accounting for 55.08% of the total participants 33 studentswere from the English Translation and Interpretation program, accounting for 27.97% ofthe participants (the two programs belong to the English Studies major) The EnglishTeacher Education major was represented by 20 students, accounting for 16.95% of theparticipants

In terms of cohorts, 10 students belonged to Cohort 46, accounting 8.47% of the total

70 students belong to Cohort 47, accounting for 59.32% of the participants, and 38students belong to Cohort 48, accounting for 32.2%

In terms of genders, 51 were male, accounting for 43.22% of the participants, while 67were female, accounting for 56.85% of the participants

Table 3.1 Participants’ major

Major Frequency Percent Valid Percent

Cumulative Percent

English Studies major 65 55.08 55.08 55.08

English Translation and

Interpretation 33 59.32 27.97 83.05 EnglishTeacher

Education 20 16.95 16.95 100.0 Total 123100.0 100.0

Table 3.2 Participants’ year of study

Course Frequency Percent Valid Percent Cumulative Percent

16

Valid

46 10 8.47 8.47 8.47 47 70 59.32 59.32 67.8

48 38 32.2 32.2 100.0 Total 118 100.0 100.0

Table 3.3 Participants’ gender

Gender Frequency Percent Valid Percent Cumulative Percent

Male 51 43.22 43.22 43.22Valid

Female 67 56.85 56.85 100.0 Total 118 100.0

100.0

3.2.2 Participants for the interviews

Of the total 118 students, ten students voluntarily participated in the semi-structuredinterviews when invited They are students from the English Language Studies andEnglish Teacher Education programs in the Faculty of Foreign Languages participated inthe survey before

3.3 Research instruments

3.3.1 Questionnaires

3.3.1.1 Rationale of the questionnaires

Questionnaires are a popular choice for researchers due to their ability to efficientlygather data from a large number of participants (Creswell, 2014) This method isparticularly well-suited for collecting information on perceptions and attitudes, as itallows respondents to express their viewpoints in a structured format (Babbie, 2010) Inthis study, questionnaires were chosen as the primary data collection tool to gain insightsinto participants' perceptions of using ChatGPT as an English language learning tool Thequestionnaire draws on existing studies by Ngo (2023), Zhang (2023), and TheTechnology Acceptance Model (TAM) framework

Proposed by Davis (1989), TAM is a well-established theoretical model forunderstanding how users accept and adopt new technologies According to theTechnology Acceptance Model (TAM), users' intention to adopt a technology isinfluenced by two key constructs: perceived usefulness (PU) and perceived ease of use

(PEOU) Perceived ease of use refers to the extent to which users believe that using thetechnology will be effortless, while perceived usefulness reflects their belief that thetechnology will enhance their performance or productivity Additionally, this studyincorporates a third construct: attitude toward use This construct represents users'overall feelings and evaluations about the technology, shaped by their perceptions of its

17usefulness and ease of use TAM offers a structured method for analyzing users'perceptions, and attitudes toward technology adoption, making it a suitable frameworkfor examining EFL students' perceptions of using ChatGPT for language learning

While TAM provides a solid foundation for understanding technology acceptance, itmay not fully capture all aspects relevant to the context of using ChatGPT for languagelearning Therefore, the questionnaire adapts and improves upon TAM by incorporatingadditional sections to address specific needs and concerns of EFL students, that is thefrequency and types of tasks they use ChatGPT for and their perceptions of thechallenges encountered when using the tool

Overall, first, after the items on frequency and usage, the main part of the questionnairefocuses on perceptions, including a section where participants evaluate how easily theycan navigate and use ChatGPT for language learning

Then, the questionnaire assesses the perceived usefulness of ChatGPT as a languagelearning tool Participants are asked to rate the extent to which they believe ChatGPTcan help improve their English language skills on a 5-point Likert scale Participants areasked to identify the benefits they believe they gain from using ChatGPT, such asimproved vocabulary, grammar, and conversational skills This section aligns with thecore concept of perceived usefulness in TAM

Next is the attitudes Participants are asked to reflect on their general impressions andsatisfaction with the technology, which helps provide a fuller picture of their willingness

to integrate it into their learning routines

Last is a section on perceived challenges Participants are asked to identify any obstaclesthey anticipate encountering while using ChatGPT for language learning, such asconcerns about receiving incorrect answers or becoming overly reliant on the tool

3.3.1.2 Questionnaire design

The questionnaire with 32 items consists of three main sections: participants’ personalinformation (Gender, cohort, and major), students' frequency and preferences of usingChatGPT in English learning, students’ perceptions of using ChatGPT in Englishlearning (see Appendix A)

The second section, with 3 items, using a multiple-choice approach, helps understandthe frequency of using ChatGPT for learning, which tasks are more frequently used withChatGPT, and which English language skill of students is primarily enhanced

The third section with 29 items using five-point Likert-scale (1=Strongly disagree, 2=Disagree, 3= Neutral, 4= Agree, 5= Strongly agree) was categorized into four clusters,namely perceived ease of use, perceived usefulness, attitude toward using, perceptionsregarding the disadvantages Below is the summary of the questionnaire.

18

Table 3.4 Summary of the second and third sections of questionnaires

Cluster Item

Frequency and usage (multiple choice) (1), (2), (3)

Perceived ease of use (4), (5), (6), (7)

Perceived usefulness (8), (9), (10), (11), (12), (13), (14), (15), (16), (17)

Attitude toward using (18), (19), (20), (21), (22), (23) (24)Perceptions regarding the disadvantages (25), (26), (27), (28), (29), (30), (31), (32)

3.3.1.3 Piloting the questionnaires

Before the research began, the questionnaires were first written in English and thentranslated into Vietnamese to simplify the response process for participants A pilot testwas conducted online using Google Forms with 35 students The Google Form link wassent to English Language and English Teaching students via Messenger and Zalo Theirresponses were analyzed using Cronbach's alpha (α) test to ensure reliability Cronbach'salpha, also referred to as coefficient alpha, serves as the predominant measure of internalconsistency (reliability), indicating the degree to which a set of items are interrelated.Notably, Cronbach's alpha values typically range from 0 to 1, with higher coefficientsindicating greater correlation among questionnaire items, reflecting a shared underlyingconstruct The questionnaire data obtained in this study underwent reliability testingaccording to the aforementioned principles, yielding a Cronbach's alpha value indicative

of good scale reliability (α = 0.819) (see Appendix E) As a result, the reliability of thequestionnaire used in this study was verified The specific criteria for Cronbach's alphaare detailed below:

Table 3.5 Cronbach’s alpha values

Cronbach’s alpha values Reliability levels

.9 ≤ α < 1 Excellent.8 ≤ α < 9 Good.7 ≤ α < 8 Acceptable.6 ≤ α < 7 Questionable

α < 6 Unacceptable

In summary, Table 3.5 indicates that the acceptable range for Cronbach's Alpha value isgenerally between 0.7 and 0.9, with values above 0.9 sometimes indicating redundancyamong items If the Cronbach's alpha falls within this range, the questionnaire is

considered reliable However, if the Cronbach's alpha is below 7, it suggests issues withthe questionnaire, rendering it unusable.

19

3.3.1.4 Delivering the questionnaires

After the piloting and testing procedures for the questionnaire's reliability were

completed, the questionnaire was adjusted and ready to be delivered to 118 participants

In the third week of the study, permission was requested via email from the professor tosurvey the students in their classes With the professor's permission, the class was

attended, and students were asked for their consent to participate in the survey If theyagreed, they completed the survey immediately in the classroom

3.3.2 Semi-structured interviews

3.3.2.1 Rationale for using semi-structured interviews

Interviews are a good method for obtaining detailed insights into a subject’s experiencesand perceptions (Merriam, 2009) Creswell (2014) notes that interviews enable anuanced understanding of how people perceive and interpret their surroundings Amongvarious interview types, the semi-structured format is particularly flexible Fraenkel andWallen (2019) describe it as a guided dialogue where the interviewer uses preparedquestions but can also adjust the conversation to explore new topics in depth Thisformat is advantageous due to its cost-effectiveness and its ability to probe the "how"and "why" behind participants' experiences (Flick, 2014) Consequently, a semistructured interview will be used to explore participants' perceptions on using ChatGPT

as a tool for learning English

3.3.2.2 Interview design

The purpose of the interview was to further understand how EFL students perceive usingChatGPT to learn English, which was in line with the study's main goal It employedfive open-ended questions to explore how students use ChatGPT, whether it motivatesthem to learn English, the perceived advantages and disadvantages of the tool, and

suggestions for improvement as a language learning tool (see Appendix C)

3.3.2.3 Piloting the semi-structured interviews

Interview questions were written in Vietnamese before the actual interview and pilotedwith two students through Google Meet The aim of the pilot interview was to make thequestions clearer and easier to follow for the participants

3.3.2.4 Conducting semi-structured interview

Ten students who participated in the questionnaire were invited for interviews in thisstudy based on their voluntary participation All the interview questions were asked byVietnamese Each interview lasted about 30 minutes The interview responses wererecorded by Google Meeting

Table 3.6 Summary of the research instruments

Instruments Questionnaires Semi-structure interviews

Functions

Elicitation format

To collect data about

students’ perceptions ofusing ChatGPT as alanguage learning tool

3 items using multiple choiceapproach

32 items using 5 point-Likertscale with 4 clusters

To explore in-depth

students’ perceptions ofusing ChatGPT as alanguage learning tool

5 open-ended questionsregarding their perceptionsand their suggestions forChatGPT

Response format Written Spoken

3.4 Data analysis

3.4.1 Analysis of the questionnaire

Quantitative data analysis was conducted using IBM SPSS Statistics version 20 (SPSS).Descriptive statistics were employed to assess students' perceptions of using ChatGPT inlearning language

Following data collection, responses from all questionnaires were entered into SPSS.The percentage will be calculated for the 3 items in the Frequency and Usage cluster thatuse the multiple-choice approach The five-point Likert scale (1=strongly disagree,2=disagree, 3=neutral, 4=agree, 5=strongly agree) with 29 items converted intonumerical values for efficient data entry and analysis Additional variables includinggender, major, and course enrollment were also collected and uploaded into SPSS forfurther analysis

Table 3.7 The application of five-point Likert scale in SPSS (Field, 2013) Value

Range Label Mean of scale 1 0.01 1.80 Strongly disagree Very low 2 1.81

-2.60 Disagree Low 3 2.61 - 3.40 Neutral Average 4 3.41 - 4.20 Agree High 5 4.21

- 5.00 Strongly agree Very high

When examining EFL students' perceptions of using ChatGPT as a language learningtool through a 5-point Likert scale, it's essential to understand the nuanced meaningsbehind each range Those who strongly disagree (0.01 - 1.00) likely hold firm opposition

to the idea, expressing skepticism about ChatGPT's effectiveness in language learning.Respondents falling within the disagree range (1.01 - 2.00) may have doubts or

21reservations, indicating a level of uncertainty about ChatGPT's suitability for language

acquisition In the neutral range (2.01 - 3.00), students neither favor nor opposeChatGPT, possibly reflecting a need for more information or experience On the otherhand, those who agree (3.01 - 4.00) recognize ChatGPT's potential benefits in languagelearning but may not be entirely convinced Finally, respondents in the strongly agreerange (4.01 - 5.00) express unwavering support for ChatGPT as a valuable tool forenhancing language skills, indicating a high level of confidence in its efficacy.

3.4.2 Analysis of the semi-structured interviews

Thematic analysis is a qualitative research approach used to identify, analyze, and reportpatterns or themes within data collected from semi-structured interviews This methodinvolves systematically organizing and interpreting interview transcripts to revealunderlying meanings and insights According to Braun and Clarke (2006), thematicanalysis includes several iterative steps: becoming familiar with the data, generatinginitial codes, searching for themes, reviewing themes, defining and naming themes, andproducing the final report

Thematic analysis is notably flexible and adaptable, making it suitable for variousresearch contexts and theoretical frameworks It enables researchers to delve deeply intocomplex phenomena, capturing rich and detailed perspectives from participants.Furthermore, it allows for the integration of diverse viewpoints, which enhances thecredibility and reliability of the research findings

3.5 Procedures of the study

All the data were gathered in the second semester of the academic year 2023-2024

Table 3.8 Data collection procedures

Weeks Research activities

Week 1 - 2 Conducting and piloting questionnaires, interviews

Week 3 - 4 Email the professors to ask for their permission to survey the participants

Delivering the questionnaires through paper to the participants

Week 5 - 6 Synthesizing the data from the questionnaires Arrange the

schedule for the interviewWeek 7 - 9 Interviewing the participants

Week 10 - 11 Synthesizing the data from observations, the interviews

Week 12 - 15 Analyzing the data from questionnaires, the observations, and interviews

Writing the report and completing the thesis

22

CHAPTER FOUR FINDINGS

This chapter reports the findings from the questionnaire and the interview The

following section examines participants' perceptions using ChatGPT as a language

learning tool.

4.1 Findings from the questionnaire and interview

4.1.1 Multiple choice section

A Descriptive Statistics Test was used to examine this multiple choice section about

student’s frequency and usage of ChatGPT as a English learning tool Item 1 is about the

frequency of using ChatGPT for learning English Table 4.1 shows the result of the

analysis

Table 4.1 The frequency of ChatGPT used by students in English language learning

(%)

Frequency of using ChatGPT

Frequency Percent Valid Percent

Cumulative Percent

Valid

Always 29 24.6 24.6 24.6 Usually 72 61.061.0 85.6 Sometimes 17 14.4 14.4 100.0Never 0 0 0 0 Total 118 100.0 100.0

The findings reveal a high prevalence of ChatGPT use among EFL students in the study

A significant majority (85.6%) reported using ChatGPT regularly for English language

learning, with nearly a quarter (24.6%) indicating it has become an integral part of their

routine This suggests strong acceptance and potential for this tool as a learning aid The

remaining students (17%) sometimes use ChatGPT, suggesting it complements their

learning approach but might not be used daily Interestingly, no students reported never

using ChatGPT, suggesting a strong awareness of the tool

In item 2, because each respondent could select multiple options for the question

regarding their usage of ChatGPT, the analysis of usage was based on the frequency of

individual tasks chosen rather than percentages Among the six tasks provided in the

questionnaire, three tasks were most frequently utilized: generating ideas, providing

23explanations for English language use, and

proofreading and error correction of writing

drafts Table 4.2 presents the results of the

analysis

Table 4.2 Frequency of tasks using ChatGPT as

a language learning tool

Generat ing ideas

Translating Giving

explanations

Pr

Missing 24 61 58 Learning new

words

N 64 64

The findings from both the questionnaire and interviews reveal a clear preference amongstudents for using ChatGPT to support specific tasks in their English language learning.Most frequently selected tasks, according to the questionnaire, the top three tasksstudents use ChatGPT for are: (1) generating ideas, (2) proofreading and errorcorrection, and (3) providing explanations for English language use Generating ideaswas the most selected task (104 times), with students recognizing the challenge ofarticulating ideas in a new language and valuing ChatGPT’s assistance in providingoutlines for writing and presentations Proofreading was selected 88 times, highlightingChatGPT’s effectiveness in detecting spelling errors, grammar mistakes, and wordusage Finally, providing explanations was selected 70 times, as students often soughtclarification on language rules and grammar

The interviews echoed these preferences, but in a different order Students mostfrequently used ChatGPT for proofreading, followed by paraphrasing, and generatingideas Ninety percent of the interviewees emphasized the tool’s utility in catchinggrammar mistakes and offering better word choices, with several students viewing it asmore than just a proofreading tool—an essential resource for improving writing skills.For instance, Student 4 shared, "I take it a step further by giving ChatGPT poorly writtensamples, which helps me spot common mistakes and get better at writing." In the case ofgenerating ideas, students in the interviews appreciated the structure ChatGPT offered,though they emphasized the need to refine and personalize its suggestions to meet theirindividual needs

Writing is the most utilized skill from the questionnaire (presented more detail in table4.3) 83% of students reported using ChatGPT predominantly for writing tasks Thisoverwhelming preference indicates that students rely heavily on the tool for their writtenEnglish work ChatGPT’s inability to provide listening and speaking functionalitieslikely contributes to the low percentages for these skills, with listening selected by only3.4% and speaking by 9.3% of students Reading was selected by 16.9%, where somestudents used ChatGPT to generate reading passages and assess their comprehensionthrough self-created quizzes

24

In the interviews, writing also emerged as the most common skill for which studentsused ChatGPT However, the interviews provided additional insights into how studentsuse ChatGPT to improve their writing As mentioned earlier, proofreading andparaphrasing were highlighted as essential tasks For example, Student 2 noted, "Itcatches grammatical errors I might miss and offers alternative phrasings that improve

my writing flow," reflecting a deep reliance on ChatGPT’s ability to polish written

drafts In paraphrasing, while students valued the clarity ChatGPT offered, they alsoexpressed concerns about occasional robotic or unnatural suggestions, as seen in Student3’s comment: "Sometimes, it suggests great alternatives, but other times, its suggestionssound robotic or unnatural."

In generating ideas, it showed that students often turn to ChatGPT for generating ideas

In the questionnaire, generating ideas was the most frequently selected task, and studentsvalued ChatGPT’s ability to provide outlines for writing, speaking, and presentations.Similarly, in the interviews, students described ChatGPT as a helpful tool forbrainstorming and structuring their work Student 6 mentioned, "When it comes togenerating ideas, I find ChatGPT really helpful it provides me with outlines and ideas

on what to say."

However, the interviews added depth to this finding While students appreciatedChatGPT’s initial suggestions, many emphasized that these ideas often requiredrefinement to align with their personal style or academic needs Student 3 commented, "Iprefer to begin with my own thoughts and then use ChatGPT to polish them it helps toenhance the clarity and impact of my concepts." This demonstrates that while ChatGPTplays a significant role in the brainstorming process, students still recognize theimportance of personal input and creativity

Paraphrasing was selected by 61 students in the questionnaire, placing it in the middlerange of tasks In the interviews, eight students used ChatGPT for paraphrasing, but theiropinions were more varied While some found ChatGPT effective for generatingalternatives and improving sentence clarity, others felt that its suggestions wereoccasionally too robotic Student 9 saw the paraphrasing feature as a learningopportunity, stating, "ChatGPT can explain its reasoning behind paraphrases, offering avaluable learning opportunity." Despite some concerns about the unnaturalness of certainparaphrases, students acknowledged the tool’s usefulness in restructuring sentences andenhancing clarity, particularly when they struggled to express ideas clearly

Proofreading and Mistake Correction is the second most selected task in the

questionnaire and the most commonly reported task in the interviews The interviewsprovided a richer understanding of how students use this feature Students praised

ChatGPT’s ability to catch errors they often overlooked and appreciated its suggestions

25for improving word choice and writing flow Student 2 remarked, "It catchesgrammatical errors I might miss and offers alternative phrasings that improve mywriting flow." Several students even used ChatGPT to analyze poorly written samples,using the tool as a learning resource to spot common mistakes and improve their writingskills

This demonstrates that students view ChatGPT not only as a tool for error correction butalso as an active contributor to their development as writers Its role extends beyondsurface-level corrections to providing feedback that enhances students’ overall writing

style and confidence in their abilities.

In Providing Explanations and Learning New Words, In the questionnaire, 70 studentsselected ChatGPT’s ability to explain English language usage, while 64 selected it forlearning new words Although not explored extensively in the interviews, this highlightsChatGPT’s role as a resource for language clarification, helping students grasp difficultconcepts or unfamiliar grammar rules Students frequently rely on ChatGPT to provideexplanations for their English learning queries, which aligns with its function as anaccessible, on-demand resource

Table 4.3 The percentage of skill EFL students use ChatGPT to improve.

Skill of English Frequency Percent Valid Percent Cumulative Percent

4.1.2 5 point Likert-scale section

A Descriptive Statistics Test was used to examine students’ perceptions of using

ChatGPT as a language learning tool Table 4.1 shows the result of this test

Table 4.4 Mean score of students’ perceptions of using ChatGPT as a language

learning tool

N Min Max M SDStudents’ perceptions 118 1 5 4.12 0.95

Table 4.1 shows that the mean score of students’ perceptions is higher than scale 3 in the

five-degree scale (M=4.12, SD=.95), indicating that students’ perceptions of using

ChatGPT as a language learning tool were at a high level

26

A Descriptive Statistics Test was conducted to measure the mean scores of four

clusters, namely perceived ease of use, perceived usefulness, attitude toward using,

perceptions regarding the disadvantages Table 4.4 shows the result

Table 4.5 Mean scores of the clusters of students’ perceptions

Clusters N Min Max Mean SD Perceived ease of use 118 1 5 4.3 8 Perceived

usefulness 118 1 5 4.01 87 Attitude toward using 118 1 5 3.88 86 Perceptions regarding the

disadvantages 118 1 5 3.69 1The findings of the study reveal that EFL students generally hold a positive perception

of ChatGPT as a language learning tool Firstly, with a mean score of 4.3, the perceivedease of use suggests that students find ChatGPT relatively straightforward to navigate,

indicating a user-friendly interface or intuitive functionalities This ease of use is crucialfor ensuring that students can effectively utilize the tool without many difficulties.Additionally, the mean score of 4.01 for perceived usefulness indicates that studentsview ChatGPT as highly beneficial for language learning purposes They recognize itsvalue in providing instant feedback, generating natural language responses, and offeringpersonalized learning experiences tailored to their individual needs Despite a slightlylower mean score of 3.96 for attitude toward using, falling within the "high" category, itstill signifies a generally positive disposition towards incorporating ChatGPT into theirlanguage learning activities Lastly, the mean score of 4.23 for perceptions regarding thedisadvantages suggests that while students acknowledge the advantages of ChatGPT,they are also cognizant of its limitations.

Findings for the first research question: What are EFL students' perceptions of

using Chatgpt as a language learning tool?

Perceived ease of use

Specific analysis of students’ perceived ease of use of ChatGPT as a language learningtool from the questionaire is presented in the following table

Table 4.6 Percentage of students’ perceived ease of use of ChatGPT as a language

learning tool

Percentage Items Statements Strongly Disagree

Disgaree Neutral Agree Strongly AgreeM SD

6ChatGPT understands my requests regardless of

whether I use phrases,

0 5.2 12.7 40.7 41.5 4.19 85

Total average0.43 5.08 10.18 37.08 47.23 4.26 0.87The data suggests that participants generally find ChatGPT to be user-friendly and easy

to navigate, with a significant majority strongly agreeing or agreeing with this statement

(85.5%, M=4.29) The level of agreement regarding the ease of giving requests to ChatGPT is slightly higher, with 90.7% strongly agreeing or agreeing (M=4.37) This

indicates that while the interface is user-friendly, some users may still encounterdifficulties in effectively communicating with ChatGPT, as 5.9% and 5.1% disagree withthese two items, respectively

Moreover, participants seem to have varying perceptions regarding ChatGPT's ability to

understand requests While a majority (77.9%, M=4.18) agree that it understands

requests regardless of format, there is a notable proportion of participants who areneutral or disagree with this statement (21.1%)

Regarding the clarity and ease of understanding of ChatGPT's responses, the majority ofparticipants agree that ChatGPT's responses are clear and easy to understand (82.2%,

M=4.19) However, there remains a small percentage of participants who are neutral or

disagree with this statement (17.9%), suggesting some room for improvement in thisaspect

Perceived usefulness

Table 4.6 Percentage of students’ perceived usefulness of ChatGPT as a language

learning tool

Percentage Items Statements Strongly Disagree

Disgaree Neutral Agree Strongly AgreeM SD

8ChatGPT helps me generate ideas and improve the1.7 2.5 14.4 39.8 41.5 4.17 89

28outlines of my English

writing assignments.

9ChatGPT can analyze my writing texts and

suggest

language components where improvements can be

made for better unity, coherence

and language use.

materials, making it easier

and faster to grasp the main components

(participants, methods, findings, etc.) of the research.

12ChatGPT can offer effective reading strategies

to help me self-study easily (when preparing for exams,

conducting research for assignments, or simply enhancing my