Abstract
Artificial intelligence-driven chatbots have sparked significant changes in English education, which is challenging for pre-service teachers. To date, it remains unclear what are pre-service English teachers’ perceptions of and needs in integrating chatbots into English education. This study, therefore, explored Chinese pre-service English teachers’ (1) basic understanding of and attitudes toward chatbots, (2) perceptions of chatbots in English education, (3) perceptions of chatbots’ advantages and challenges and (4) needs in integrating chatbots into English teaching. From June to September 2023, 12 Chinese pre-service English teachers were purposefully sampled for semi-structured interviews, and thematic analysis was employed to identify key themes. The findings revealed generally positive attitudes toward chatbots, though teachers had limited understanding. Most teachers believed chatbots could improve language skills, but noted concerns like plagiarism. They emphasized human teachers’ irreplaceability in moral, emotional, and individualized teaching. Teachers also sought policy, training, and social support. These findings will not only help to better understand pre-service teachers’ views concerning integrating chatbots into English education internationally but also shed light on pre-service teacher education and chatbot development.
Keywords
Introduction
In recent years, rapid advancements in artificial intelligence (AI) technology have boosted the capability of chatbots and have accelerated their integration into education (Okonkwo & Ade-Ibijola, 2020; Smutny & Schreiberova, 2020). The Open University’s recent release of Innovating Pedagogy 2023 ranked the use of AI tools as the top 10 innovative methods, convincing the potential to trigger significant changes in educational practice in the next 5 to 10 years (Kukulska-Hulme et al., 2023). A typical example of applying AI in the educational area is the utilization of chatbots for teaching and learning. Chatbots are computer agents or programs that process the users’ input and give output by utilizing AI technology (Okonkwo & Ade-Ibijola, 2020). In English education field, chatbots can satisfy the students’ demands by offering personalized, interactive, and attractive learning experiences (Mohamed, 2023) and help enhance English listening (N. Y. Kim, 2018a), reading (Liu et al., 2022), writing (Guo & Wang, 2023), and speaking (Hwang et al., 2022; H. Yang et al., 2022) skills. Although quantities of chatbots’ advantages are mentioned, teachers lack the necessary competencies to apply chatbots in educational practice. Teachers are important mediums for integrating learning technologies (Saltan & Arslan, 2017). Notably, pre-service teachers are in a transitional stage and are at the stage of developing technological skills and knowledge (Valtonen et al., 2020). What they perceive and believe will directly influence their future teaching performance and practice (Istenic et al., 2021; T. C. Yang & Chen, 2023).
Although studies have noticed the English teachers’ perceptions of integrating chatbots into English teaching and learning (Bayram & Baturay, 2022; Nguyen, 2023), related studies on pre-service teachers’ perceptions of chatbots, and their needs in chatbots’ application in English education are lacking. Curriculum planners and training experts face challenges in developing and facilitating pre-service teachers with chatbot knowledge and skills. Since pre-service teachers are essential in teacher education, paying more attention to their perceptions can provide trajectories for integrating chatbots into English education.
Hence, this study aims to examine pre-service English teachers’ (1) basic understanding of and attitudes toward chatbots, (2) perceptions of chatbots in English education, (3) perceptions of chatbots’ advantages and challenges and (4) needs in integrating chatbots into English teaching. The findings can contribute to the international understanding of pre-service teachers’ views of and needs in using AI-powered tools to facilitate English education. Meanwhile, the exploration and analysis will provide implications for the design of pre-service training programs and therefore contribute to the enhancement of teachers’ digital literacy from a global view. It can also provide possible development directions for chatbots regarding the improvement of the functions.
Literature Review
What is an AI Chatbot
The advancement of AI has led to the proliferation of pre-trained Large Language Models (LLMs) represented by intelligent agents and their embeddedness in human beings’ lives (Guo & Wang, 2023). Chatbot, one of the most popular LLMs, is a system applied in various contexts including commerce, research, and education (C.-C. Lin et al., 2023; Okonkwo & Ade-Ibijola, 2020). An AI chatbot is a conversational or interactive software program that provides instant responses on given topics to users with the assistance of natural language processing (NLP) and machine learning or deep learning (Okonkwo & Ade-Ibijola, 2021; Zhai & Wibowo, 2022). AI chatbot has three major features: instant response, data support and multiple input and output (see Table 1). Hence, this personalized, on-demand and instant assistance will help individuals bridge the learning gaps, strengthen comprehension and acquire knowledge wherever and whenever they like (Chan & Lee, 2023). Large data support gives chatbots better capability to handle tasks, comprehend dialogs, and therefore can fulfill the requirements of users in a humorous and empathetic way (Zhai & Wibowo, 2022). The text-to-image or image-to-image chatbots can present the information in a more vivid way.
The Characteristics of Chatbots.
For example, ChatGPT (OpenAI, 2023a), the generative pre-trained transformer in the LLM field, has gained great popularity across the globe since its release. Compared with its previous versions, ChatGPT is now equipped with more powerful functions. It can provide human-like textual responses to a wide range of inquiries in a variety of languages (Ibrahim et al., 2023) and generate more truthful and less toxic prompts conversationally (Guo & Wang, 2023). So far, ChatGPT has rolled out voice and visual capability (OpenAI, 2023a). Becoming a versatile tool, ChatGPT has profound utility in numerous aspects, such as writing articles of all types (Taecharungroj, 2023), marketing and offering services (George & George, 2023), summarizing literature and improving papers (van Dis et al., 2023), etc. Furthermore, in 2023, GPT-4 appeared with higher accuracy, greater NLP ability, broader application scenarios and better human-bot interaction experience (OpenAI, 2023b), signifying the extensive potential of chatbots in the future.
AI Chatbots in English Teaching and Learning
Previous research shows that there are several technological affordances of AI chatbots in language teaching and learning such as timeliness, ease of use and personalization (W. Huang et al., 2022). Owing to these characteristics, chatbots are able to offer smooth and real-time interaction. This helps facilitating an authentic learning environment for language learners, and fosters synergies for interaction and communication, thereby assisting them in effectively addressing language communication and usage challenges (Z. Zhang & Hong, 2023). A review by Ji et al. (2023) provided possible roles that AI chatbots may serve in language education, including evaluator, resource evaluator, and need analyst, etc. In sum, chatbots have several utilities in English teaching and learning.
It is indicated that chatbots have been employed in English teaching and learning in three stages: (a) pre-class stage, (b) while-class stage, and (c) post-class stage. These are given in Table 2, along with representative usages and sample studies.
Chatbots in English Teaching and Learning.
Pre-Class Stage
In this stage, students and teachers are in need of abundant learning resources. Chatbots can search for information about unknown words and provide videos, pictures, memos, and songs to the learners (Hsu et al., 2021; J. Lin and Mubarok, 2021). In this scenario, chatbots serve as the resource provider. In addition, before the class, text analysis and student analysis are a must as they directly influence the teaching approach teachers adopt. Chatbots can be employed here in the following ways: simulation (W. Huang et al., 2022), estimating learners’ language proficiency level (Hassani et al., 2016) and making teachers well-prepared for the possible key and difficult points in the teaching process (Z. Zhang & Hong, 2023). Also, since chatbots can generate texts according to the prompts, it may help teachers write instructional designs as well (Z. Zhang & Hong, 2023).
While-Class Stage
During the class, the functions of chatbots can be exerted in more profound types. Studies demonstrated the possibility of integrating chatbots in real classes (Jia & Chen, 2008; N. Y. Kim, 2016). Firstly, since students get higher motivation and perform better if they learn in authentic contexts (Hwang et al., 2016), chatbots, equipped with the affordance of simulation, are able to help teachers by designing multiple scenarios, providing opportunities for learners to practice specific knowledge or words (Wang et al., 2017) and able to disguise as a specific interlocutor (e.g., waiter) to improve students’ expression (H. C. Yang & Zapata-Rivera, 2010). Another utilization is to promote teaching activities, such as searching for answers to teachers’ questions, making oral communications, etc. (Kohnke, 2022). While there is a time limitation of class, teachers are in urgent need of an assistant to solve frequently asked questions. Chatbots can liberate teachers from repeated questions, giving them more time to conduct teaching (Kohnke, 2022). Additionally, the conversational exchanges by chatbots provide students with more opportunities to express their points of view (H. Yang et al., 2022). Encouraging the expression of students not only helps to enhance students’ language capability but also liven up the class atmosphere.
Post-Class Stage
Owing to the timeliness and ease of use (W. Huang et al., 2022), chatbots can be applied in more after-class scenarios. It is in this stage that students can tightly interact with the AI chatbots. Student-AI interaction includes three types, namely cognitive, socio-emotional, and artifact-mediated interaction (J. Kim et al., 2022). For the cognitive part, when students are learning and reviewing individually, AI chatbots provide learners with explicit answers and personalized guidance (Huang et al., 2022; Mohamed, 2023). Chatbots can tailor the instructions based on the needs of learners and their language level (Jeon, 2021), which may offer an optimal learning experience for learners. For example, chatbots evaluate the students’ writing works and give improvement suggestions. Students may select whatever style and difficulty of language and grammar they like. For the socio-emotional part, studies proved the affordance of chatbots in increasing students’ motivation and interests (Liu et al., 2022; H. Yang et al., 2022) and reducing language anxiety when using chatbots (Fryer et al., 2019), enabling higher learning efficiency and better achievements. For the artifact-mediated interaction part, students can perceive the interfaces of AI chatbots and coordinate with them, thereby adapting to the presence of AI (J. Kim et al., 2022).
Teachers’ Perceptions of and Needs in Applying AI Chatbots in Education
According to Davis (1989), the successful adoption of technological innovations hinges on user acceptance. Two important determinants, namely perceived usefulness and perceived ease of use, weigh the most. The two determinants can be concluded as the cognitive process of users, or the users’ perceptions. Perception is a process in which organisms interpret and organize sensations to produce a meaningful experience of the world (Norman, 1977). Notably, how the individual interprets or perceives may be quite different from the reality and impact their action (Pickens, 2005). In the field of education, Choi et al. (2023) identified that teachers’ beliefs and perceptions are the decisive factors in maximizing the use of chatbots.
Previous studies investigated teachers’ perceptions of AI in education. T. C. Yang and Chen (2023) pointed out that whether integrating chatbots into instruction or not was strongly related to teachers’ perceptions and intentions. In the realm of English teaching, studies mainly highlighted the perceptions of the specific chatbots in English teaching. Chuah and Kabilan (2021) and Ulla et al. (2023) investigated English teachers’ perspectives on using ChatGPT in English teaching, indicating both the recognition and concerns of in-service teachers. However, only a limited number of studies related to pre-service English teachers have been conducted. Among them, J. Yang (2022) examined the English pre-service primary teachers’ perceptions of the potential benefits of employing AI chatbots, concluding that chatbots have high affordance in English language classes and more training is needed to compromise pre-service teachers’ lack of English education competency.
Compared with perceptions, needs are more practical and concrete. Needs refers to the basic substance (food, etc.) that is physiologically required by the body, or some social and personal factors as well as things derived from complex learning (knowledge, skills, achievements, prestige, etc.; Reber, 1985). The analysis of needs was first proposed by Michael West in the 1920s (West, 1994) and later formed the concept and was adopted in English special purpose (ESP) research (Richterich, 1980). There are four main models for needs analysis: (1) Target Situation analysis (TSA) by Munby (1981) or the Communication Needs Processor (CNP), which focuses on the learners’ communication needs in future career or learning scenarios. (2) Present Situation Analysis (PSA) by Allwright (1982) or the Deficiency Analysis, which focuses on the gap between learners’ present knowledge and the language needs of the targeted situation. (3) The model by Hutchinson and Waters (1987) put learners at the center and analyzed both target needs and learning needs. (4) Dudley-Evans and St John (1998) adopted means analysis and ecological approach and gave seven folds of needs analysis model to facilitate curriculum development.
Conducting the needs analysis in the language teaching and learning field is greatly beneficial to all facets including curriculum design, syllabus construction, class opening, curriculum evaluation, etc. (Chen, 2009). As integrating cutting-edge AI chatbots into teaching brings great challenges to teachers, it is necessary to analyze the needs of future English teachers so as to help them improve relevant knowledge and competencies. Yet the studies are rare, and thus there is an urgent need to identify them. For the current study, the second model Present Situation Analysis would be adopted.
Pre-Service Teacher Education in the AI Era
Teachers play a vital role in the application of new educational technology in the curriculum. Nevertheless, previous studies indicated that various factors (i.e., learning dimensions, job replacement, sociotechnical blindness, and the configuration of AI systems) will affect the AI anxiety of teacher candidates (Hopcan et al., 2023), which will accordingly exert a negative impact on chatbot applications in English class. Thus, it is necessary to promote AI-supported pre-service teacher education in many forms as they will be at the forefront of education in the future (C. Zhang et al., 2023).
Previous literature noted the significance of developing the training program as it is helpful for pre-service teachers to develop a more sophisticated conception of digital literacy (Lim, 2023), and improve innovation and collaboration. Notably, the designers of the program should customize the program to meet diverse needs, interests, and concerns (Hopcan et al., 2023). Research demonstrated technological and pedagogical content knowledge (TPACK) framework may be suitable to better improve pre-service teachers’ technical and ethical skills to master the AI-based tools with high pedagogical and technological affordances (Celik, 2023; Rets et al., 2023).
TPACK is a framework used to understand teacher knowledge with the integration of technology and seven knowledge domains: Content Knowledge (CK), Pedagogical Knowledge (PK), Technological Knowledge (TK), Pedagogical Content Knowledge (PCK), Technological Content Knowledge (TCK), Technological Pedagogical Knowledge (TPK), and Technological Pedagogical Content Knowledge (TPACK; Mishra & Koehler, 2006). It is useful for teachers to think, analyze and evaluate their necessary integrated knowledge to integrate technology into teaching (Baran et al., 2011). For pre-service teachers’ development in TPACK, Koh and Divaharan (2011) revealed the situation that pre-service teachers mainly developed TK (Technological Knowledge) and TPK (Technological Pedagogical Knowledge), but more emphasis should be laid on TCK (Technological Content Knowledge) and TPACK.
Additionally, as pre-service teachers have less requisite knowledge and mental representation, open educational resources (OERs) should be offered during and after education to facilitate their self-regulated learning (L. Huang et al., 2022).
To sum up, previous studies show that the perceptions of pre-service teachers will exert a direct impact on their future teaching performance and practice (Istenic et al., 2021; T. C. Yang & Chen, 2023). Although research indicates various applications of chatbots in English teaching and learning, showing the immeasurable potential in English education, few studies explore how pre-service English teachers perceive chatbots and their functions in English education, where they acquire such knowledge, and what are the challenges and needs in their integrating chatbots into English education. Focusing on teachers’ perceptions of and needs in their pre-service stage will help them to handle the challenges and obstacles in their practicing instruction (Sorge et al., 2019), at the same time, enhancing their technological literacy in their teaching (Ye et al., 2021). Hence, this exploratory study aims to investigate Chinese pre-service English teachers’ perceptions of and needs in integrating chatbots into English teaching and learning. The following two research questions are set for the study:
Methods
Participants
The participants were sampled from a provincial normal university in Zhejiang province, China to eliminate the influence of geographical variables. Altogether, 12 Chinese pre-service English teachers were recruited purposefully controlling their gender, grade level, and teaching experience. The demographic profile of participants is summarized in Table 3.
Demographic Profile of the Participating Pre-Service English Teachers.
Data Collection
In the present study, semi-structured interviews were adopted to explore pre-service English teachers’ perceptions of and needs in integrating chatbots into English education (Weiss, 1994). The interview questions were developed based on literature review and pre-service English teachers’ real practice. The questions were primarily open-ended and primarily focused on seven topics (See Appendix 1) to ensure the flexibility (Merriam, 2009). An academic colleague who was not involved in the research was invited as an external auditor to review the questions and provided comments to enhance validity.
The interviews were conducted from June to September 2023 via videoconferencing to eliminate the limitation of time and space. To make sure the participants express themselves freely, the interviews were conducted in their mother tongue, Chinese. The pilot study was conducted with one participant to check the design of the interview guidelines, the online facilities and the interview time. Based on the result of the pilot study, we refined the description of some questions to ensure that the participants’ understanding is aligned with the targeted meaning.
Each formal interview lasted for approximately 50 minutes. To establish a positive relationship with the participants, the researchers began each session by introducing themselves and expressing their gratitude for the interviewees’ participation as suggested by Denzin and Lincoln (2005). Before the interview, the interviewees’ informed consent was obtained and they were informed of their basic rights. To ensure the confidentiality of the participants’ identities, pseudonyms were used throughout the reporting of research findings, as recommended by Miles and Huberman (1994). The interviews were recorded so the responses could be transcribed and analyzed.
Data Analysis
In this study, the process of analyzing interview data adhered to the fundamental steps involved in qualitative research data processing. To begin, we converted the audio data obtained from the interviews into written texts. We then assigned names to each of these text documents and read the transcripts to highlight and label the sections of interest. Furthermore, we read each interview text thoroughly, refining the text and clarifying the detailed information based on the interview questions. This process helped us identify the primary themes that emerged from the data. Additionally, we generalized the detailed original data and transformed it into aggregate coding. This step involved summarizing and grouping similar information into broader categories to facilitate analysis. Finally, we reviewed the research questions and literature, refined and integrated the codes, developed the spindle code, and began preliminary writing.
Through this process, six categories were generated to clarify research questions, including: basic understanding, experience and attitudes; impact of chatbots on English teaching and learning; advantages and challenges of integrating chatbots; sources of chatbot knowledge and skills; external needs and future advancement needs of chatbots. And a final set of 17 themes were identified. For example, the first category “Basic understanding, experience and attitudes” include two themes: “Pre-service English teachers’ understanding and experience (Theme 1)” and “Pre-service English teachers’ attitudes toward chatbot application and human-bot coordination (Theme 2)”. Sub-themes were also generated to further illustrate the points. For instance, for Theme 1, there are four sub-themes: “Chatbots’ popularity”, “Chatbots’ diversity”, “Pre-service English teachers’ understanding”, and “Pre-service English teachers’ frequency of chatbot use” (See the summary of emergent themes attached as the Supplemental Information). To ensure rigor, member checking with several participants were conducted to ensure the accuracy and relevance of the themes to their experiences and perceptions.
Findings and Discussion
Perceptions Regarding Chatbots in English Teaching and Learning
Basic Understanding, Experience and Attitudes
Unsurprisingly, chatbots have gained great popularity among the participants. All the participants acknowledged the popularity of chatbots and had heard of chatbots.
When asked to list some usual chatbots in their daily lives, approximately all the participants considered ChatGPT as the first example. Also, some participants listed the voice assistants on mobile phones such as Siri (the voice assistant on iPhone), Little Yi (the voice assistant on Huawei), etc. The process of enumeration indicates the diversity of chatbots, showing that chatbots have several types in different fields. For instance, P12 commented: First, there’s ChatGPT; it has different versions like 3.0, 3.5, and 4.0. Also, Siri is an Apple mobile phone with a chatbot. Similarly, there may be domestic chatbots, like Baidu developed one called ERNIE Bot, Little Yi and so on. AI chatbots are now on our mobile phones, for example, some apps can also carry AI chatbots, for example, WeChat public subscription accounts seem to have an AI oral chat study room, like, I do not remember, anyway. There are also some oral English chatbots on some apps. You can try it. (P12)
Then we investigated the participants’ basic understanding of chatbots. The participants used specific words to describe functions of chatbots and exemplified their perspectives. P4 commented: Take ChatGPT as an example. I know ChatGPT can help collect information. First, you raise a question to it, and then it will help you collect various resources online, and then help you organize. Therefore, I think the function of chatbots is to collect data first, and then organize and arrange these data to form a logical answer, and then present it. (P4)
Though the participants had some basic understanding of chatbots, there exist misunderstandings about chatbots that are not aligned with reality. The misunderstandings mainly include incomplete understanding and delayed understanding. Firstly, the incomplete understanding happened most frequently. Almost all the participants had a limited knowledge of chatbots as they just used general words to describe the nature and functions of chatbots, which is also found by research of Belda-Medina and Calvo-Ferrer (2022). A few participants could come up with one conclusive feature while others could not. P8 made a brief introduction and illustrated one characteristic from his point of view: First of all, it is a language model trained with the algorithm, which can generate a natural language text and communicate according to the information of related problems provided by me. I think one of its biggest characteristics is that it uses a lot of data training, including textbooks on the Internet. But it has a disadvantage, that is, it is easier to identify, and sometimes the answer is not right. (P8)
However, most participants just listed a lot of application scenarios without being able to express clearly the nature and working mechanism of chatbots. P3 is a typical example: Like asking the chatbot to come up with some exercises for students. Does that count? Then for example, sometimes I want to be lazy, so I make chatbots to help me write an introduction to a certain topic, and some of its expressions are great, so they can be used in some works. And as I mentioned earlier, I used chatbots to practice speaking, and they would talk to me and I could practice speaking. (P3)
Parallel to the incomplete understanding, the participants had a delayed understanding of chatbots that do not conform to reality owing to the disconnection with the forefront of technological development. For instance, since ChatGPT has developed an empathetic tone and human-like voice as well as an artificial image in the new version, some participants’ conceptions still stayed in the primitive stage. P3 & P6 commented: Chatbots may not convey the emotions as people, and it can be seen from the lines. The words are not so special and rich, that is, not as close as a real person when communicating with you. (P3) In terms of spoken language, I think it is possible to design some images for chatbots, and it is best that users can design these images themselves. In this way, whether we are speaking or listening, we will have a stronger sense of presence, and it feels like we are really chatting with people. (P6)
Additionally, some participants indicated the low efficiency of chatbots and expressed the chatbots’ ill performance negatively from their perspectives, which is aligned with the findings of Limna et al. (2023). As P9 commented: For example, when I chat with it in oral communication, it may not be able to grasp the key point I want to express at once, and the focus will be one-sided. And it may not be able to accurately identify the central meaning of what I am saying. (P9)
In addition, we examined the users’ frequency of chatbot use. The results indicated that the frequency varied according to their grade levels. For example, P8 was in Grade Four and was experienced in teaching while P9 was in lower grade without enough teaching experience. There may exist a correlation between the pre-service development stage and the frequency of chatbot use.
If the lesson preparation is also counted, I use it more frequently. I can use chatbots to roughly understand some ideas of the course design process, but if it is said in the teaching process, it may not be used because of the constraints. (P8) Relatively few, because I have not yet reached the stage of junior or senior year, and I am only a freshman, so I only use chatbots for learning, such as ChatGPT, and I will ask more questions for homework completion. (P9)
The quote from P8 demonstrates that the pre-service teacher has begun to use chatbots to prepare for lessons. Evidence could also be found in P11.
I have only used ChatGPT once so far when I was an intern some time ago, and I wanted it to generate a writing sample for students. This is the purpose for which I use it. (P11)
Although only used ChatGPT for once, P11 used it for the purpose of teaching, which again shows the pre-service teachers’ awareness of using chatbots to enhance teaching.
Another important finding is the pre-service English teachers’ attitudes toward chatbots. Firstly, participants conveyed strong support for the integration of chatbots into English teaching and learning. Almost all the participants (N = 11) directly expressed a supportive attitude and were in urgent need of further action. This result aligns with the study by Tlili et al. (2023), who examined the public discourse in social media about ChatGPT and got a generally positive comment.
Second, as we asked if chatbots can be a substitute for human teachers, all the participants gave a straightforward negation that is in accordance with the study by Edwards and Cheok (2018). They listed some traits of human teachers, such as moral education, teaching methods, and emotional support to certify human teachers’ irreplaceability. Some typical responses are as follows.
Moral Education: Human teachers also bear the responsibility for the moral education of human beings. Chatbot has no educational function, and it is difficult to guide students with the correct values. (P2)
Teaching methods: The role of human teachers is not only to pass on knowledge, but more importantly, to enlighten students on how to solve problems, and to solve some psychological problems that students may have in the learning process, but the chatbot can only tell students the answer. (P3)
Emotional support: One of the biggest differences between chatbots and human teachers is emotional output. Chatbots are unable to empathize with people or detect students’ psychological changes through their micro-expressions. (P4)
Finally, in terms of human-bot coordination, all the participants acknowledged the necessity of human-bot coordination, and some of them posed feasible application scenarios (Edwards & Cheok, 2018; Pérez et al., 2020) and needs.
Division of labor: Chatbots are more responsible for more mechanical work, such as finding some data, and humans may be more responsible for some experiential or creative work. (P1)
Difference in order: Chatbots can first display the most essential things or the most important things, and if human teachers observe that students have problems, they can timely adjust, further explain or supplement. (P4)
Needs: I will learn more about chatbot theory. I also hope to have the intervention of school forces to carry out more training. In addition, I think our students and parents, they need to have a certain understanding of human-bot collaboration. (P11)
Impact of Chatbots on English Teaching and Learning
To ensure the feasibility of integrating chatbots into English teaching and learning, we should clarify the pre-service teachers’ perceptions of whether chatbots exert impact on multiple English skills. Specifically, we focus on chatbots’ impact on listening, speaking, reading, writing, vocabulary, grammar and comprehensive English skills. In this category, we asked 12 participants about the pros and cons from their perspectives of using chatbots as an aid for English teaching and learning. Table 4 illustrates their perspectives on the supportive points and concerning points of different English skills. To make it clearer, we selected some excerpted quotes that are on behalf of similar views.
Pre-Service English Teachers’ Perceptions of Impact of Chatbots on English Teaching and Learning in Seven Skills.
As displayed in the table, the findings highlight the potential impact of chatbots application in seven aspects. For English listening and speaking, participants (N = 8) believed that chatbots can exert a positive impact on their conversational skills, which is also supported by N. Y. Kim (2018a) and Muhammad et al. (2020). Especially speaking, they noted chatbots can reduce the negative impact of shyness (Chien et al., 2022). For reading skills, previous studies highlighted the utility of enhancing students’ reading interest (Liu et al., 2022), whereas the participants of the present study mainly focused on the functions of selecting and providing reading material as well as comprehension. For English writing, 10 participants mentioned the function of providing resources in the pre-writing stage (Guo et al., 2022) and polishing the articles in the post-writing stage (Guo & Wang, 2023). Concerns on dependence (N = 3) and logic restraint (N = 2; Mohamed, 2023) drew our attention, while the previous study demonstrated the possibility of training students’ logic by chatbots (R. Zhang et al., 2023). Participants offered several utilities for enhancing students’ vocabulary (N. Y. Kim, 2018b) and grammar skills (N. Y. Kim, 2019). However, they questioned if chatbots can present the knowledge interestingly.
To conclude, supportive points include accurate feedback, instant response, personalized instruction (Ait Baha et al., 2023; Kuhail et al., 2023), and a profound knowledge base. Nevertheless, the concerning points lie in the possibility of inappropriate feedback, defects in pronunciation, student dependence, limitations of thoughts and lack of interest. The participants have different supportive perceptions and concerns about each skill.
Advantages and Challenges of Using Chatbots in English Teaching and Learning
Furthermore, to ensure the smooth integration of chatbots into English teaching and learning, exploring the advantages and challenges from the users’ perspective is of great significance. Table 5 illustrates their perceptions.
Pre-Service English Teachers’ Perceptions of Chatbots’ Advantages and Challenges in English Teaching and Learning.
The advantages include providing experience, providing accurate language, updating mindset, collecting material, reducing cost, providing instant response (Pérez et al., 2020), improving confidence (Song et al., 2022), and promoting interaction (Ait Baha et al., 2023). They mentioned the chatbots’ usefulness in pre-, while-, and post-class stages from several perspectives (Gonda & Chu, 2019). However, there are apparent disadvantages mentioned by the participants, including values concerns, cultural gap, unsatisfactory results, public opinion, humanistic care, hard to access, educational inequality, privacy and security (Sharkey, 2016), reducing job opportunities (Reich-Stiebert & Eyssel, 2016), and plagiarism and academic integrity (Okonkwo & Ade-Ibijola, 2021).
Since the participants offered lots of pros and cons of chatbots in English teaching and learning, the approaches to solving the concerns are valuable to explore. They gave possible solutions to the concerns. P1 admitted the necessity to get acquainted with the chatbots (Cheng et al., 2020), and P7 advised to strengthen the supervision during the process. Some participants also emphasized the standard use of chatbots as suggested by Mendoza et al. (2022). Exemplary quotes are as follows: First you have to contact the chatbot; you must not avoid it, that is, you have to use it, and then you know where its weaknesses are, you know where its strengths are. (P1) Students should learn some rules, such as how to use the robot, and how often. For teachers, how they use chatbots to assist English teaching every day should be regulated. (P11)
Sources of Chatbot Knowledge and Skills
To probe deeper, we further investigated how participants gained the relevant knowledge and skills. Among the 12 participants, the Internet was mentioned 11 times, which is the primary source of acquiring chatbot knowledge. Also, some participants (N = 3) claimed that they acquired the knowledge by chatting with friends and classmates. A few participants (N = 2) noted the usage of journals and academic papers in spreading relevant knowledge. Furthermore, we examined how pre-service English teachers obtained skills in operating chatbots. The results are similar to the previous question. Half of the participants mentioned the use of the Internet. The rest is searching on social software (N = 3), attending lectures (N = 2), asking friends (N = 2), and reading relevant books (N = 2).
It is manifested in the findings that the Internet is the primary source that the participants acquired chatbot knowledge and how-to skills, which clearly demonstrates the urgent need of formal education of AI and AI-based tools, and of designing and developing training programs for pre-service English teachers to enhance their digital literacy. Additionally, a wealth of relevant material shall be uploaded to the web or social platforms so that pre-service teachers will have more access to the resources concerning chatbot knowledge and usage.
Needs Regarding Integrating Chatbots Into English Teaching and Learning
Apart from the perceptions of the pre-service English teachers in various facets, we further investigated the needs regarding integrating chatbots into English teaching and learning. The following aspects were highlighted by the participants.
External Needs
Participants proposed some external needs to promote the chatbot application in English teaching and learning. Specifically, the needs mainly focus on the following parts: policy support, equipment support, financial support, practice support, training support, and social consciousness support. Table 6 illustrates the needs and the representative quotes.
Pre-Service English Teachers’ External Needs.
It is noticeable that the participants are in the most urgent need of policy support (N = 7) and training support (N = 7). Calls for policy support from the government gained high frequency as some participants conveyed that only the government has the power to promote the application domestically. Additionally, the participants also mentioned the necessity to offer equipment support (N = 4) from both the research and development organization and the schools, and financial support (N = 4). Some participants noted they needed the practice support (N = 2) and the social consciousness support (N = 3). More sample applications should be offered and the parents’ approval should be ensured.
Policy support in AI-assisted education was the most-frequent mentioned aspect by the participants. However, their statements did not involve much detail, Schiff (2022) noted that the national education agencies should enact the policy from three aspects, including developing more AI experts, preparing the workforce for AI and arousing public AI literacy, especially the students and teachers. Endris et al. (2024) proposed that the policy regarding AI-assisted education should be contextual, consultative, dynamic, implementable and measurable and gave the seven steps for policy development. Hence, it is advisable for policymakers to consider the perspectives of teachers and the recommendations of researchers when making decisions about integrating AI into education.
From an individual perspective, several participants pointed out the importance of attending some training courses, which is also mentioned by Chuah and Kabilan (2021). We conducted further research to determine their specific training program needs, particularly regarding the content that should be covered during any training sessions. The results mainly include teacher ethics (N = 4), human-bot cooperation (N = 1), scenario training (N = 7), emergency processing (N = 1), and chatbot selection (N = 1). Table 7 shows the need for training content.
Pre-Service English Teachers’ Needs for Training Content.
It is noted that more than half of the participants said they needed some scenario training. To be specific, the participants mainly focused on how to use chatbots to facilitate class preparation, how to conduct class activities with chatbots, and how to evaluate students by using chatbots.
When teachers use chatbots to teach listening, speaking, viewing, reading and writing, as well as cultivate students’ comprehensive quality, I think teachers should be trained in different skill types. (P2) I think teachers need some training in the process of guiding students to give feedback after class. (P4) For example, how should the teacher guide the student to use the chatbot to hand in the homework and correct the homework, and then how to ask the chatbot questions. How to make chatbots help teachers correct the homework and so on. (P5)
The findings are aligned with the statements by Novella-García and Cloquell-Lozano (2021). The training programs should include basic instrumental practice so that teachers are able to handle technology. In this process, they can reach a better level of digital competency (Fernández-Batanero et al., 2020). If we examine closer the participants’ needs within the TPACK framework, they were mostly in need of training in TPK, PCK, and CK. For example, human-bot cooperation primarily involves the integration of TK and PK, that is, knowing how to effectively use chatbots to collaborate with humans and improve the teaching and learning experience. And scenario training is connected with PCK because it encompasses designing and implementing strategies for situational learning that enhance student engagement and comprehension, often grounded in the specific subject matter. It is very likely that they may also need a systematic training of the whole TPACK framework as their responses were very limited, which reflected their lack of general knowledge of the framework.
In addition, the training programs should also focus on the ethical and critical dimensions and therefore, teachers can solve the social, psychological, or educational problems in the classroom (Torres-Hernández & Gallego-Arrufat, 2022).
Future Advancement Needs of Chatbots
Moreover, the participants envision the future improvement of chatbots from several perspectives, namely improving language processing and information retrieval abilities (N = 2), ameliorating oral output (N = 4), enhancing humanistic care (N = 2), promoting cross-platform support (N = 1), integrating VR technology (N = 2), and developing diverse and specialized chatbots (N = 4). Table 8 illustrates the results.
Pre-Service English Teachers’ Needs of Chatbots’ Future Development.
Specifically, some participants suggested the technological improvement of chatbots to offer more reasoning and handle more complex questions, which is in line with the findings of Chuah and Kabilan (2021). Regarding the awkward pronunciation of chatbots, participants also envision a more fluent and human-like voice. It is recommended that chatbots incorporate more humanistic care (Belda-Medina & Calvo-Ferrer, 2022) and integrate VR technology with more accessible platforms. Additionally, developing diverse and specific chatbots was also mentioned since different chatbots have different utilities (Pérez et al., 2020).
Conclusion
Theoretical and Practical Implications
In the present study, we elaborated on the participating pre-service English teachers’ perceptions of and needs in integrating chatbots into English teaching and learning. The findings corroborate the previous studies that revealed the potential of LLM-based tools, especially chatbots, to enhance the efficiency of English teaching and learning and to blaze a trail of human-bot coordination. The study, therefore, steps into a point for discussion and suggestions to better facilitate chatbot-powered English teaching and learning.
To sum up, from the perspective of perceptions, the findings show that participants have an incomplete and delayed understanding of chatbots, and their knowledge and how-to skills about chatbots mainly come from the Internet. Their application frequency varies with their grade level. In addition, the participants held supportive attitudes toward chatbots but strongly believed that they could not replace human teachers. They also envisioned the possibility of human-bot coordination in future English teaching and learning. From the perspective of needs, the study revealed six external needs: policy support, equipment support, financial support, practice support, training support, and social consciousness support. Meanwhile, they looked forward to the future development of chatbots from six aspects.
On the basis of the findings of the study, there emerged two important implications. Firstly, it is necessary to raise awareness and enhance knowledge of chatbots among pre-service English teachers. As the present research revealed that the participants obtained most of the information about chatbots from the Internet, which might lead to their misunderstandings, the formal training is in urgent need. Training programs specifically focused on chatbots shall be provided to help pre-service English teachers to better understand their nature, functions, applications, and limitations within the TPACK framework. In particular, the following shall be highlighted: TK, such as knowledge about diverse chatbots and their working principles; TCK, such as knowledge about chatbot selection for teaching oral English; and TPACK, such as using chatbots to design a project-based learning project where students research a topic on English festival, create a PowerPoint presentation in English, and then present it to the class using digital presentation tools. In addition, teacher ethics, human-bot cooperation, teaching scenarios, emergency processing shall be important contents in the program. This can help teachers integrate chatbots into their teaching practice more effectively. In addition to providing training for pre-service English teachers, other external support, specifically, policy support, equipment support, financial support, practice support, and social consciousness support shall be strengthened. This will provide them with necessary resources, which can boost their confidence and comfort in using chatbots in English teaching and learning.
Secondly, to keep up with the rapid development of technology in the era of AI, it is essential to continuously improve and innovate chatbots to meet the evolving needs of English teaching and learning. In particular, advancements including improving language processing and information retrieval abilities, ameliorating oral output, enhancing humanistic care, promoting cross-platform support, integrating VR technology, and developing diverse and specialized chatbots shall be anticipated. Therefore, it is crucial to continue to invest in research and development in chatbot technology to improve its functionality and applications in English teaching and learning.
Limitations and Further Directions
Although this study explored the perceptions of and needs in pre-service English teachers regarding the integration of chatbots into English teaching and learning, which could contribute to the understanding of pre-service English teachers’ perceived knowledge of AI-based tools, and could shed light on pre-service teacher education and integration of chatbots into English education, there are several limitations which should also be addressed. Firstly, the findings were produced by qualitative analysis using in-depth interviews with a small number of participants. Future studies could adopt mixed methods with larger samples to better elaborate their perceptions and needs. Secondly, our study investigated pre-service English teachers’ perceptions and needs cross-sectionally. As chatbot technology advances with such an unbelievable speed, a longitudinal study that tracks their cognitive change would be valuable to provide a more complete insight into the field.
Supplemental Material
sj-docx-1-sgo-10.1177_21582440251321853 – Supplemental material for To Chat or Not: Pre-Service English Teachers’ Perceptions of and Needs in Chatbot’s Educational Application
Supplemental material, sj-docx-1-sgo-10.1177_21582440251321853 for To Chat or Not: Pre-Service English Teachers’ Perceptions of and Needs in Chatbot’s Educational Application by Yilong Su, Meina Luo and Chenyin Zhong in SAGE Open
Footnotes
Appendix 1
Author Contributions
Yilong Su contributed to the literature review, collected the data, performed the analysis, and drafted the manuscript. Meina Luo conceived the main conceptual ideas, supervised the research process and contributed to the material preparation, data collection, and the interpretation of findings. Chenyin Zhong contributed to data analysis and the interpretation of findings. All authors discussed the results and commented on the manuscript. All authors read and approved the final manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research is supported by funding from Zhejiang Province Education Science Planning Project (2022SCG376). It is also supported by Tsinghua University 2023 Visiting Scholar Program.
Ethics Approval
Although there are no mandatory ethical checks required by the university, the authors have voluntarily adhered to strict ethical guidelines and practices throughout the entire research process.
Consent to Participant
This study received informed consent from all participants.
Consent for Publication
The authors accept responsibility for releasing this material to be published.
Code Availability
N/A
Data Availability Statement
The summary of emergent themes is attached as the Supplemental Information.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
