Editor

JLLT edited by Thomas Tinnefeld
Showing posts with label 81 Elmacı Er. Show all posts
Showing posts with label 81 Elmacı Er. Show all posts

Journal of Linguistics and Language Teaching

Volume 14 (2023) Issue 2, pp 167-185


Exploring the Role of AI-Based Conversational Agents for Collaborative Learning of English as a Foreign Language


Ebru Pınar Elmacı Er (Sakarya, Turkey) & Armin Weinberger (Saarbrücken, Germany)


Abstract

AI-based chat agents within educational settings present a promising pathway for bolstering collaborative learning (CL), especially in the sphere of English as a Foreign Language (EFL). This paper investigates the evolving role of chat agents based on Artificial Intelligence (AI) and Large Language Models (LLM), notably ChatGPT, in the landscape of CL specific to EFL. The primary aim is twofold: firstly, to discern the productive and unproductive applications of ChatGPT in a CL environment, and secondly, to investigate approaches for harnessing ChatGPT to amplify productive CL experiences in EFL contexts. We present a literature review addressing the main objective of this paper, i.e., the exploration of the productive and unproductive uses of ChatGPT for CL in the EFL context. The findings aim to provide insights into the capabilities and limitations of ChatGPT in fostering language skills, enhancing communication competence, and encouraging cooperative learning strategies among EFL learners. This present paper delineates approaches and possible pitfalls when employing AI chat agents in language learning, attempting to bridge the divide between AI technology and instructional design, aiming to facilitate a more informed and efficacious utilisation of AI within collaborative EFL learning environments.

Keywords: Artificial intelligence, conversational agents, EFL, collaborative learning, ChatGPT




1 The Gap between Productive and Unproductive Uses of AI in EFL Learning

Openly accessible Artificial Intelligence (AI) services have brought about a transformation in education including language education settings. AI-powered chat agents building on Large Language Models (LLMs), such as ChatGPT, have become popular tools in enhancing language learning, particularly in the realm of English as a Foreign Language (EFL). These AI systems, known for their interactive and adaptive features, provide distinguishing prospects for Collaborative Learning (CL). The pedagogical approach of CL refers to a variety of educational practices building on peer interaction (Dillenbourg et al. 2009).

CL is employed in formal education at all learning levels, from young learners to groups of university students working together to solve a problem, complete a task, or create a learning product (Laal & Ghodsi 2012). CL is often associated with higher achievement and greater productivity (Ansari & Khan 2020) and lends itself well to EFL instruction (Jiang & Zhang 2020, Kanno 2020). Computer-Supported Collaborative Learning (CSCL) allows learners to deliberate their individual viewpoints via text-based media or videoconferencing being guided by collaboration scripts (Weinberger et al. 2005), shared representations (Suthers 2003), or brought to reflect on dynamic visualisations of their interaction (Janssen & Bodemer 2013). The integration of AI and natural language processing technologies, particularly with the implementation of conversational agents, i.e. chatbots, building on services like Chat GPT, has shown great potential in advancing collaboration by providing adaptive guidance (Alam 2022, Namatherdhala et al. 2022). With the emergence of conversational agents based on LLMs, technology moves beyond offering itself as a platform for collaboration towards acting as a learning partner to help students construct arguments in interaction with their other (human) learning partners. Human-human interaction is accompanied by human-chatbot interaction. This may allow for designing learning environments that combine the benefits of interacting with peers with the potential that chatbots hold to understand natural language input and mobilise vast online resources to scaffold whenever needed. To date, there is a lack of comprehensive research on the experiences of EFL learners utilising ChatGPT as a CL tool.

The implementation of AI in educational settings, particularly in language learning, comes with its set of challenges. The productive use of AI-based chatbots in CL involves leveraging their capabilities to support and enhance human interaction rather than replacing it. These agents can serve as facilitators or conversation partners, providing linguistic input, feedback, and scaffolding tailored to the learners’ proficiency level (Huang et al. 2022). Unproductive uses might include over-reliance on AI for language production and abuse of AI help, which could impede the development of independent language skills or lead to the propagation of errors if the AI’s language model is not accurate or context-appropriate (Aleven, McLaren, Roll & Koeddinger, 2004, Nass & Brave 2005).

This paper aims to address this gap by exploring how to use ChatGPT in CL settings. Specifically, it aims to explore the dual facets of AI-based chat agents in the realm of collaborative EFL learning: identifying their productive and unproductive uses and devising instructional designs for their effective incorporation into language education. How can we improve students’ joint EFL learning experience leveraging the potential of AI while maintaining the essential human element of language learning? The findings from this state-of-the-art article will provide practical implications for teachers, curriculum developers, and educational technology designers to improve the quality and effectiveness of collaborative language learning experiences using ChatGPT as a valued pedagogical tool. 

Firstly, we will introduce collaborative language learning scenarios. Secondly, we will discuss the role of AI in education. Then we will address the research questions of risks and potentials of AI in language learning as well as develop a scenario of fruitful, AI-enhanced language learning, highlighting specific uses of AI services in EFL learning. By delineating these sections in the paper, we aim to offer a comprehensive understanding of the multifaceted relationship between AI technology and collaborative language learning, specifically in the EFL context.


2   Collaborative Learning

CL is an instructional approach that stresses the importance of social interaction in learning. Learners work together to solve problems, complete tasks, or create projects to co-construct knowledge (Laal & Ghodsi 2012). Research has demonstrated that group work not only boosts student motivation (Tran 2019) and academic achievement but also improves communication, critical thinking, and analytical abilities (Chandra 2015). Moreover, it fosters better classroom behaviour and peer relationships while increasing student engagement in the learning process. Additionally, group work promotes learning and empowers students to take ownership of their education (Koç 2018). Johnson &  Johnson (2020) outline five criteria for collaboration in the classroom: 

1. Interdependence: Team members recognize their shared responsibilities and rely on each other to achieve their common goal;

2. Individual accountability: Each member’s performance is evaluated individually;

3. Face-to-face promotive interaction: Team members provide positive feedback to encourage one another.

4. Group processing: Members reflect on the process used to complete a task and discuss ways to enhance it;

5. Social skills: Effective CL necessitates the proficient use of interactional skills, like questioning, giving feedback and reaching consensus.

CL has special applications in the foreign language classroom (Momtaz & Garner 2010). Collaborative EFL learners are exposed to language input and ample opportunities to practise their speaking and listening skills. When learners engage with each other during such interactions, they share a focus on using the target language while also noticing challenges in language use and receiving feedback from their peers (Long 1996).

Learning a foreign language builds on adequate materials and tasks, including language use. Ideally, this includes opportunities for learners to engage with native speakers. One effective approach is for both, teacher and students, to utilise the target language as much as possible (Eurydice 2012). By doing so, learners are given opportunities to interact with each other and their teacher in the target language, which greatly aids in developing listening and speaking skills. Moreover, such interaction enables learners to recognize and correct their mistakes while also learning from the errors made by others. Therefore, the need for interaction becomes apparent in the process of acquiring relevant input in foreign language classrooms. According to Long’s Interaction Hypothesis (1996), when learners interact, they receive feedback. This means that individuals get feedback on their statements, which helps them focus on linguistic structures in a specific context. The Output Hypothesis, developed by Swain (1995), suggests that learners gain knowledge of their interlanguage and become aware of the differences between the structures of the target language and their current level of skill. Interlanguage here refers to the dynamic and evolving linguistic system that is the transitional state or the evolving language system that learners create as they move from their native language toward proficiency in the target language.

CL surpasses teacher-led dialogue by fostering psychologically safe eye-to-eye interaction among peers, where knowledge is collectively constructed and shared, as opposed to hierarchical interaction with teachers conveying and assessing knowledge. CL thus empowers students as active contributors, cultivating an environment where learners can practise speaking a foreign language and productively learn from their mistakes. Through collaborative endeavours, students not only exchange ideas but also develop vital social skills, including communication, teamwork, and conflict resolution, essential for navigating real-world scenarios. 

Collaboratively learning a foreign language can build on different scenarios, such as acting out scenarios or role-playing exercises, solving problems together as a team, engaging in discussions on topics, participating in debates to explore differing perspectives, and delivering presentations to share information or ideas. Accordingly, Koç (2018) found that, with CL presenting a variety of scenarios, EFL teachers showed a preference for CL activities over solitary learning activities. 

A key question in designing CL scenarios is whether learners, when left to their own devices in groups, can engage in interactions that are productive for learning. Some studies on collaborative language learning observe beneficial learner behaviour and provide support for the benefits of collaborative language learning per se, e.g., engaging in dialogues fosters oral language skills (Bao 2020). Here, beginners learning Chinese within the Danish EFL context practising oral language skills were found to achieve high language proficiency through collaborative dialogues as they provided opportunities for participation and information sharing. Additionally, learners were found to offer mutual support for higher-level functions that are typically challenging to perform individually, thereby indicating an increase in language competence. Mutual support among peers seems to be particularly beneficial in the language acquisition journey (Zeng & Takatsuka 2009). It seems that engaging in learning within environments proves to be both an efficient and effective approach for enhancing foreign language learners’ language and communication abilities due to mutual support, mutual feedback, and opportunities for practising the target language across contexts. 

Multiple factors can make or break CL, however, and need to be considered when designing pair and group activities such as the learners' proficiency level, interests, and preferred learning styles. Providing instructions and ongoing support adapted to these factors throughout the activity seems advisable. Therefore, implementing CL for language learning can pose challenges for teachers, and there is some agreement that teachers can greatly benefit from receiving proper prior training to design and facilitate CL activities (Avci & Adıgüzel 2017, Bao 2020, Zeng & Tatsuka 2009). 

To sum up, CL is a teaching approach that can greatly benefit EFL learners. CL of EFL builds on specific scenarios and requires instructional support for learners to practise speaking and develop listening skills. 


3   The Evolution of Education: From Traditional to Digital

In the past, interactions between teachers and students took place mostly face-to-face in classroom settings. Lessons from textbooks and teacher-centred pedagogies predominated, with few opportunities for students to actively participate in learning. Similar to other types of training, teaching EFL focused on language norms, rote memorisation, and grammar drills. The advent of computers and the internet at the end of the 20th century brought about a transformation in education through the introduction of digital learning tools. This marked the beginning of the shift from traditional teaching methods to digital learning environments. Interactive multimedia, online resources, and computer-based language learning programs began to supplement traditional materials. AI-driven technologies promise to further advance the personalization and interactivity of such digital learning environments.

A paradigm change of AI in education has resulted from AI's capacity to process enormous volumes of data, discern patterns, and adjust to individual learning needs. In EFL instruction, by automating assessment and creation of talk and, thus, enabling real-time language practice through virtual chat, AI-powered systems can offer tailored learning paths, immediate feedback, and interactive scenarios of language use.


3.1 AI for Language Education

AI is a multidisciplinary field that aims to create machines capable of simulating human intelligence processes (Sumakul et al. 2022). One of the core components of AI is Machine Learning, a subset that allows computers to learn from data and improve their performance over time. At its core, AI seeks to enable machines to perform tasks that typically require human intelligence (Jian, 2022). This entails problem-solving, making decisions, comprehending spoken language, spotting patterns, and adapting to novel circumstances. AI systems are designed to replicate human cognitive processes while also exploring problem-solving strategies.

Machine learning is a branch of AI that focuses on creating algorithms and models that allow computers to get better at a particular task by learning from data (Jordon & Mitchell 2015). Instead of being explicitly programmed to perform a task, machine-learning algorithms learn from examples and data patterns. As more data is fed into these algorithms, they become more accurate and proficient at making predictions or decisions. LLMs, like GPT-3 (Generative Pre-trained Transformer 3), are advanced, 'generative' AI systems designed to understand, generate, and process human language. These models are built upon transformer neural network architectures, capable of processing vast amounts of text data to learn patterns, relationships, and nuances within language. Transformers are a specific type of neural network architecture explicitly developed to process sequential data, like text, more effectively. They excel in capturing long-range dependencies within the data by processing the information in parallel.

In conclusion, AI concepts and machine learning are foundational to the development of intelligent systems that can enhance EFL instruction and digital learning environments (Kohnke et al. 2023). The ability of machine learning algorithms to analyse language patterns, adapt to individual learners, and provide personalised feedback underscores their potential to advance language education and create more effective and engaging learning experiences.


3.2 Enhancing Language Acquisition through AI

The integration of AI into EFL teaching has immense potential to facilitate language learning by tailoring the learning experience to individual needs. (Shazly 2021, Sumakul et al. 2022). An et al. (2023) describe adaptive learning as a personalized method of teaching that ensures that students are engaged with material that is appropriate for their level of knowledge, preferred learning pace, and learning styles. AI systems track students' progress and dynamically change the type and level of activities to maximize engagement and retention. This is done through data analysis and pattern recognition. The automated evaluation of students’ essays is a great example of AI’s potential to be used in language learning contexts. Automated evaluation refers to the process of assessing or grading tasks using computer algorithms and software instead of relying solely on human judgement. In an experimental study, Wang (2022) found that learners held high expectations for the automated evaluation tools, and the effectiveness of intelligent scoring was higher than teacher-made scoring. 

Furthermore, AI-driven solutions effectively address the challenge of catering to diverse learning needs (Chen et al. 2020). By analysing individual strengths and weaknesses, AI-powered platforms can identify areas where additional support is required, ensuring that learners receive targeted resources and exercises. Language learning becomes an inclusive endeavour, as AI accommodates various learning preferences, linguistic backgrounds, and paces of progress. The application of AI potentially minimizes the risk of learners feeling overwhelmed or left behind, contributing to a supportive and inclusive learning environment. Ultimately, the synergy of adaptive learning, assessment, and personalized support could be designed through AI support, resulting in a dynamic and learner-centred EFL experience, where each student's unique journey is nurtured, leading to more effective and comprehensive language acquisition. 

In recent years, the emergence of language models has brought about a transformation due to our interactions with AI. Advanced language models comprise sophisticated AI programmes made to comprehend and produce text or speech that sounds like human-generated text or speech. These models have a high degree of fluency and coherence when processing and producing natural language text. Such models include, among others, GPT-3, GPT-4, and BERT. They are helpful for a variety of tasks, including content creation, translation, summarising, and responding to inquiries. They are adaptable and can be included in many programmes, including chatbots. It has been shown how chatbots, like ChatGPT, can be used for research and learning using language models. ChatGPT, a language production model, can create text in several genres (e.g. short essay format, itemised list, poem), depending on its training and the input it receives (Oravec 2023). OpenAI created a sizable language model for it. It is a strong tool that may be applied to many different tasks, including cooperative learning.

Chatbots like ChatGPT can provide immediate assistance to students who need help with their assignments, e.g., by connecting students to helpful resources or providing potential solutions to difficult issues, which makes them a quick and efficient way to receive knowledge (Wang et al. 2023). They might be able to provide students with immediate support if they are struggling academically. They can provide tailored comments on writing projects, assistance with navigational and technological issues, and language translation services (Wang et al. 2023). While chatbots cannot replace a professor's knowledge and advice, they can be used by students as a supplemental resource while they wait to speak with their professor in person during office hours (Wang et al. 2023). However, generative AI chatbots can also be misused, and have complicated the many issues with academic fraud that educational institutions already face. (Oravec 2023).

The effectiveness or ineffectiveness of ChatGPT use in group learning tasks primarily depends on how it is included in the teaching procedure and the users' intents. The influence of ChatGPT on CL is dual; the next section examines both its beneficial and detrimental effects.


4   The Study

Despite the growing integration of AI tools in educational contexts, there remains a substantial gap in understanding their specific roles and impacts within CL environments, particularly in EFL settings. CL, with its emphasis on social interaction and collective knowledge construction, presents unique challenges and opportunities for AI implementation. The nuances of language acquisition in such settings, when mediated by AI technology, are yet to be fully explored and understood. This gap in the literature underscores the critical need for empirical research aimed at unravelling the complexities of AI-assisted CL in EFL contexts. The purpose of this study, therefore, is to delve into this underexplored area, examining the dual aspects of AI-based chat agents: their potential to facilitate and the risks they might pose in collaborative EFL learning environments. This investigation is guided by two primary research questions:

1. Which risks and potentials have been identified for AI tools such as ChatGPT for CL?

As ChatGPT is openly accessible and capable of taking over knowledge tasks, we identify risks of supplementing central cognitive processes as well as opportunities for enhancing such processes.

2. What are productive, exemplary scenarios of collaborative EFL learning?

With ChatGPT harnessing AI for any teacher and instructional designer as well as for students, we identify and explore good practices of instructional design building on or implementing ChatGPT.


4.1 Methods

To address the research questions, an extensive literature review was conducted. The review encompassed scholarly articles, books, conference proceedings, and relevant online resources. Keywords and search terms related to "AI-based chat agents," "collaborative learning," and "English as a Foreign Language (EFL)" were employed across major academic databases. Criteria for inclusion involved relevance to AI in CL environments, specifically focusing on ChatGPT or similar conversational AI technologies utilized within EFL contexts. Publications were assessed for their exploration of productive and unproductive uses of AI in CL scenarios and strategies or frameworks related to enhancing EFL teaching through AI-based chat agents.


4.2 Findings

The investigation into the role of AI-based chat agents within the context of CL in EFL led to comprehensive insights into their multifaceted contributions and limitations. This section presents the synthesised outcomes derived from an extensive literature review conducted to address the above-mentioned two pivotal research questions.


4.2.1 Productive Uses of ChatGPT in CL

The following productive uses of ChatGPT in CL are documented in the literature:

Generating ideas and brainstorming solutions:

Idea generating and brainstorming are common components of CL tasks. ChatGPT can help us in this process by providing a variety of viewpoints, suggesting innovative ideas, and encouraging creativity (Sok & Heng 2023). It can be a useful tool for students to investigate various perspectives and broaden their thinking. For instance, students can collaborate to ask ChatGPT questions about a subject, and then analyse and discuss the answers.

Providing feedback and support:

ChatGPT can be used to give students feedback and support on their school- or university-related assignments (Dai et al. 2023). Asking ChatGPT questions allows students to get feedback on their assignments. By identifying their learning weaknesses, students can be helped to improve their overall learning ability.

Promoting critical thinking:

ChatGPT fosters the development of critical thinking by prompting students to interrogate and examine materials. By leveraging ChatGPT, students can identify and evaluate multiple perspectives after having done collaborative research on a topic. 

ChatGPT as a study partner:

By means of fostering discussions, ChatGPT supports collective learning. It can be useful as a study partner. Through ChatGPT, absent human partners may allow students to quiz one another on course content.  

Real-time tutoring assistance:

ChatGPT can offer learners rapid feedback and direction, assisting them in overcoming challenges and making concepts clear. It can serve as a substitute tutor, clarifying concepts, and responding to queries to improve learning.

Language practice:

Language practice is frequently part of CL, particularly when studying subjects like writing or foreign languages. ChatGPT can give students the chance to develop their language abilities through discussions, written comments, and general language proficiency improvement (Su et al. 2023).

Instant Access to Information:

Access to an array of educational materials is among ChatGPT's strongest benefits when students are working together to learn. Helping students locate information on a topic, ChatGPT can also answer questions and alleviate worries when used within a virtual classroom setting or CL environment. With this access to knowledge, the learning process can be accelerated and valuable hours can be recovered.

Personalised support:

ChatGPT may be tailored to provide students with individualised support. By offering individualised explanations and resources, it can adjust to different learning styles and rates. This flexibility can encourage students to participate in group learning activities and help them to better understand complex topics.


4.2.2 Unproductive Uses of ChatGPT in CL

While chatbots like ChatGPT can be valuable resources, it is important to recognize that chatbots cannot replace the expertise and guidance of human educators but may be seen as complementary tools (Wang et al. 2023). In addition, some potential uses of ChatGPT risk becoming unproductive distractions rather than enhancing the learning process. Oravec (2023) claims that AI-powered chatbots have given academic cheating new dimensions because they make it simple for students to access potent systems for creating content that may be passed off as their own in assignments or exams.

Relying on ChatGPT exclusively, without engaging any fellow students, could undermine the social and communicative benefits of collaborative work. Also, asking it to directly complete assignments would circumvent the latter’s educational purpose and not truly assess students’ intellectual understanding. Students may also be tempted to cite its responses without quoting any source, which would raise academic integrity issues. Hence, Using ChatGPT may substitute for human interactions and independent work, and over-dependence on the system could hinder the development of self-directed learning skills. Some major unproductive uses of ChatGPT in CL are listed and described below:


Using ChatGPT to cheat or plagiarize: 

The ease of access to information through ChatGPT can inadvertently lead to plagiarism. ChatGPT can be used to generate text that is similar to human-written text. This ability could be used to cheat on exams or to plagiarise work (Cotton et al. 2023, Oravec 2023, Pinochet et al. 2023, Zhu et al. 2023). Students may copy and paste responses generated by the chatbot without fully understanding the content. This would not only undermine the educational process but also raise ethical concerns.

Using ChatGPT to avoid taking responsibility for learning:

ChatGPT can be used to provide students with answers to their questions. Students can relinquish responsibility and agency for their own learning (Oravec 2023), aiming to complete assignments rather than develop an understanding of the material.

Overreliance on ChatGPT:

A potential pitfall is over-reliance on ChatGPT, which can inhibit learners' critical thinking and problem-solving skills. Relying too heavily on the model's responses without engaging in independent thought may lead to a passive learning experience.

Lack of human interaction:

CL emphasises human interaction and social engagement. Excessive reliance on ChatGPT may diminish the importance of face-to-face or peer-to-peer interactions, which are crucial for developing communication and interpersonal skills. If students primarily interact with AI-driven chatbots instead of their peers, they may miss out on the interpersonal skills and social connections that are integral to CL.

Inaccurate or misleading information:

While ChatGPT is a powerful language model, it is not infallible. It may occasionally provide inaccurate or misleading information (Pinochet et al. 2023), especially if the data it was trained on is flawed or incomplete (Oravec 2023). Learners must exercise caution and verify the information obtained from ChatGPT to ensure its reliability. Relying solely on the chatbot for information without critical evaluation can lead to the dissemination of misinformation among learners.

Reduced creativity:

Although ChatGPT can contribute to idea generation, over-reliance on the model's suggestions can stifle learner creativity. It is essential to strike a balance between leveraging ChatGPT's input and fostering independent thinking to encourage originality and innovation (Zhu et al. 2023).

Using ChatGPT in group learning scenarios can enhance the educational experience. It can provide helpful support, promote engagement, and progress individualised learning when used properly. Overconfidence, plagiarism, social isolation, and misinformation are a few examples of potential threats that could lead to unsatisfactory outcomes (Pinochet et al. 2023). To fully deploy ChatGPT in CL, teachers and students must strike a balance between using AI as a tool for assistance and maintaining the essential components of independent thought and human contact in the learning process (Cotton et al. 2023).

Some strategies can greatly facilitate the implementation of ChatGPT within CL environments: Establishing clear guidelines and specifying expectations for ChatGPT utilisation lays the foundation for its seamless integration into the learning process. By delineating specific expectations regarding its usage, educators can ensure alignment with the learning objectives while also maintaining focus and direction in its implementation. Moreover, students often require comprehensive training on leveraging ChatGPT effectively. Equipping learners with the necessary skills and strategies will ensure that they are able to use this tool as a facilitator rather than a stand-alone solution and can empower learners to realise the potential of ChatGPT.

Ongoing monitoring of student engagement with ChatGPT remains essential. This monitoring serves as a means of ensuring its appropriate and productive use within the CL framework. It allows educators to intervene, guide, and redirect usage as needed. Encouraging students to reflect on their use of ChatGPT can lead to stimulating discussions about the benefits and challenges of using this tool, prompting critical thinking and self-assessment, facilitating learners to evaluate their experiences, and fostering a deeper understanding of how ChatGPT contributes to their CL development.

By combining these measures – establishing clear usage expectations, providing adequate training, active monitoring, and fostering reflective discussions – a holistic framework for integrating ChatGPT into CL environments emerges to develop a proactive and reflective learning culture among students.


4.2.3 Productive Collaborative, Exemplary Scenarios of EFL Learning

Learning EFL often seems to be an individual pursuit, where students toil independently to assimilate vocabulary and grammar concepts (Zkan & Kesen 2008). Nevertheless, research suggests that for EFL students, CL might be more effective than that (Avci & Adiguzel 2017). This approach involves sharing knowledge, engaging in discussions, and providing mutual support while acquiring new language skills. In the context of EFL classes, chatbot agents emerge as valuable tools to facilitate CL (Alemdag 2023). A chatbot represents an AI system capable of conversing with individuals using everyday language and offering guidance and responses based on preset rules or machine learning algorithms (Vaccino-Salvadore 2023). Similarly, these chatbots can serve to offer feedback on student assignments, address inquiries, and ignite discussions within the respective teaching and learning setting (Klmová & Ibna Seraj 2023). As Alemdag (2023) highlights, prior research has already illustrated the advantages of using chatbots to create a realistic practice environment for students in foreign language dialogues.

Designing CL situations can be done in a variety of ways using chat agents like ChatGPT. Here are several scenarios:

Pair or group discussions:

Chatbots can be used to facilitate pair or group discussions in language learning. For example, a chatbot could be used to pose a question to students and then collect their responses. The chatbot could then summarize the responses and provide feedback (Alemdag 2023). 

Drama and role-play: 

Chatbots can be used to facilitate drama and role-play activities. They can simulate character interactions, provide scripted dialogues aligned with the scenario's theme, offer feedback and corrections, cover various language contexts, incorporate cultural nuances, encourage improvisation, expand vocabulary, and allow for role reversal, helping students improve their language skills, cultural awareness, and conversational abilities dynamically and engagingly.

Practise English Speaking Skills:

An AI-based chat agent can also be used to practise English speaking skills as a conversation partner (Hsu et al. 2021). In their study, Hsu et al. employed a chatbot that allowed EFL learners to practise English-speaking skills in Taiwan through interactive conversations.

Information exchange activities: 

Chatbots can be used to support information exchange activities. For example, a chatbot could be used to create a jigsaw activity in which students have to share information to complete a task in an EFL context.     

An example for a comprehensive scenario that combines the various uses of chat agents like ChatGPT for CL outlined above, is provided in the following subsection.


4.3 Integrated CL Scenario: The Language Immersion Week

At the outset of the Language Immersion Week, students engage in a multifaceted language learning experience facilitated by ChatGPT that is connected to a text-to-speech generator, hence providing both written and spoken language. Respectively, learners’ talk is being automatically recognized and transcribed for ChatGPT to process. Throughout the week-long program, ChatGPT serves as an omnipresent language companion, orchestrating various activities to enhance students' language proficiency, cultural awareness, and collaborative skills in an EFL context.


Activity 1: Pair and Group Discussions 

In the first half of the day, ChatGPT initiates pair or group discussions by posing thought-provoking questions related to cultural norms, societal values, or current affairs. For instance, the topic of the Language Immersion Week could be “Social Media’s Influence on our Lives”. Students interact within designated groups, responding to prompts and engaging in dialogue. ChatGPT, acting as a mediator, collects responses, offers feedback, and synthesizes group discussions.


Activity 2: Drama and Role-Play

Based on the syntheses of the morning discussions, ChatGPT creates scripts for drama and role-play in the afternoons. In the case of the social media topic, such a roleplay could be enacted on a social media platform. ChatGPT creates immersive scenarios aligned with cultural contexts, providing scripted dialogues and prompts for students to continue the role-play following their own script. Students immerse themselves in the role-playing activities, first following the script, but then being prompted to continue their role without a script, leveraging ChatGPT's guidance to navigate conversations, expand vocabulary, and improve conversational fluency through simulated real-life situations. Within the role-play, ChatGPT also assumes the role of a dynamic conversation partner providing nuanced cultural insights, and feedback on the topic of choice.


Activity 3: English Speaking Practice

During dedicated conversation sessions, ChatGPT pairs up with individual students as a conversation partner. In the previous activities, ChatGPT has profiled learners’ skills in speaking a foreign language and has prepared personalised speech training. Using conversational strategies derived from Hsu’s et al. (2021) study, ChatGPT engages students in interactive discussions that are tailored to the individual student and refine their English language skills. Students practise articulation, pronunciation, and spontaneous conversation, receiving instant feedback and guidance from ChatGPT. This could culminate in a talk on the topic of social media recorded on video, ready to be posted on social media.


Activity 4: Information Exchange

In the final session, ChatGPT orchestrates an information-sharing puzzle activity. It assigns students specific roles or topics within groups and encourages collaborative information sharing to complete a collective task, such as creating a critical social media message. Students interact within the ChatGPT-facilitated platform, pooling their knowledge and language skills to complete the task within the EFL context. Then they present their results to the community of learners. ChatGPT facilitates this presentation by connecting to a picture or video AI service to create artistic slides that underline the claims of the group of learners in the background.

Throughout Language Immersion Week, ChatGPT's multifunctional capabilities synergistically merge, fostering a holistic learning environment. It not only augments language proficiency but also cultivates cultural sensitivity, communication skills, and CL strategies among students in the domain of EFL.


5   Discussion

The synthesis of the literature has highlighted the multifaceted nature of ChatGPT's integration into CL environments. The findings corroborate the assertions made by Oravec (2023) that AI-based chat agents, such as ChatGPT, hold the potential to facilitate CL by providing immediate feedback and fostering interaction among learners. However, caution must be exercised as certain unproductive uses, as highlighted by Zhu et al. (2023), reveal instances where reliance solely on AI agents could hinder the development of critical thinking and human-to-human interaction in the learning process.

The nuances discerned from the literature underscore the importance of a balanced approach, where ChatGPT serves as an augmentative tool rather than a substitute for traditional instruction. Strategies proposed by Zhu et al. (2023) emphasise the need for educators to scaffold interactions, encouraging meaningful engagement while leveraging ChatGPT's strengths in providing instant guidance.

In the domain of EFL, the utilisation of ChatGPT presents a promising way to bolster language acquisition. The findings resonate with studies by Kohnke & Lee (2023), who suggest that AI-based chat agents can facilitate authentic language practice, enabling learners to engage in real-time conversations to hone their linguistic skills. Furthermore, the adaptability of ChatGPT, as discussed by Vaccino-Salvadore (2023), allows for personalised learning experiences, catering to individual proficiency levels and learning styles.

Future developments in the field of EFL are poised to be shaped by the incorporation of advanced conversational agents powered by LLM, which will function as collaborative partners in the EFL learning process. These conversational agents, usually referred to as 'chatbots', offer immense potential for enhancing the language learning experience and providing personalised support to EFL learners. The adoption of AI technology in language education has become a focal point in recent scholarly inquiries (Yang et al. 2022). AI chatbots can function as conversation partners in EFL speaking classes, offering learners valuable opportunities to refine their speaking skills (Yang et al. 2022). Through real-time voice conversations, they provide a more effective method of improving EFL speaking skills than text-based interactions ever could (Kohnke et al. 2023). Integrating AI chatbots into the EFL classroom promises to foster enhanced interaction and verbal communication, promoting learner-centred methods (Shazly 2021).

AI-powered chatbots can cultivate a supportive and non-critical environment, enabling EFL learners to practise speaking without apprehension of criticism, as highlighted in Shazly's research (2021). These chatbots offer immediate feedback and guidance, a factor noted by Shazly (2021) as contributing to reduced anxiety levels and enhanced speaking performance among learners. Moreover, employing AI chatbots in synchronous voice-based chat modes has proven effective in elevating speaking proficiency among Chinese EFL learners (Chew & Chen 2021). The integration of AI chatbots into EFL speaking classes opens up opportunities for learners to engage in real-time communication and interaction, ultimately leading to enhanced speaking fluency, as corroborated by Chew & Chen's study (2021).

In the context of the collaborative EFL classroom, the concept of agentic engagement, which signifies active involvement and participation in the learning process, is pivotal (Almusharraf & Bailey 2021). AI chatbots can be intentionally designed to promote agentic engagement by offering interactive and captivating learning experiences, as underscored by the research of Almusharraf & Bailey (2021). Moreover, AI chatbots can contribute to nurturing a collaborative language learning orientation by motivating learners to collaborate and interact with the instructor in class, which aligns with the findings of Almusharraf & Bailey (2021). The integration of AI chatbots into EFL instruction can foster a more academically conducive and amicable learning environment, ultimately promoting CL among peers. 

However, challenges in contextual understanding and cultural nuances pose limitations to the efficacy of ChatGPT in EFL contexts. Integrating cultural sensitivity and contextual relevance into AI-driven interactions emerges as a critical aspect to ensure meaningful language acquisition experiences for diverse learner groups (Kohnke et al. 2013).

Nevertheless, the incorporation of AI-driven tools in the domain of EFL may encounter obstacles. The willingness of teachers to embrace AI-driven applications can be influenced by an array of factors, encompassing their perceptions, attitudes, and personal beliefs. Addressing these factors and providing teachers with the necessary support and training can facilitate the seamless integration of AI-based applications into EFL classrooms. Moreover, there is a pressing need to deal with the ethical considerations and pedagogical ramifications associated with AI in the EFL landscape. Subsequent research may concentrate on probing the educational and ethical implications of AI in the EFL context while also scrutinising the potential advantages and obstacles linked to the utilisation of AI-driven applications in EFL instruction.


6   Conclusion

The exploration of the role of AI-based chat agents, particularly ChatGPT, in CL within the EFL context elucidates both opportunities and challenges. The synthesised insights underscore the need for a nuanced approach that harnesses the strengths of AI agents while addressing their limitations. By embracing ChatGPT as a facilitator rather than a replacement for human interaction, educators can leverage its capabilities to provide timely feedback, foster collaborative engagement, and personalise learning experiences in EFL settings. The findings highlight the importance of pedagogical strategies that combine AI technology with human intervention to create enriched learning environments. This study contributes to the discourse surrounding AI in education by delineating pathways for maximising the productive use of ChatGPT in collaborative EFL learning while advocating for continued research and pedagogical innovation to leverage its potential.



References 

Alam, A. (2022). Employing adaptive learning and intelligent tutoring robots for virtual classrooms and smart campuses: reforming education in the age of artificial intelligence. In: Advanced Computing and Intelligent Technologies: Proceedings of ICACIT 2022 (pp. 395-406). Singapore: Springer Nature Singapore. (https://doi.org/10.1007/978-981-19-2980-9_32; 05-09-2023).

Alemdag, E. (2023). The effect of chatbots on learning: a meta-analysis of empirical research. In: Journal of Research on Technology in Education, 1-23. (https://doi.org/10.1080/15391523.2023.2255698; 04-09-2023).

Almusharraf, N. (2021). Online engagement during Covid‐19: role of agency on collaborative learning orientation and learning expectations. In: Journal of Computer Assisted Learning 37 (5), 1285-1295. (https://doi.org/10.1111/jcal.12569; 20-09-2023).

An, X. et al. (2023). Modeling English teachers’ behavioral intention to use artificial intelligence in middle schools. In: Education and Information Technologies, 28 (5), 5187-5208.

Ansari, J. A. N. & N. A. Khan (2020). Exploring the role of social media in collaborative learning the new domain of learning. In: Smart Learning Environments 7 (1), 1-16. (https://doi.org/10.1186/s40561-020-00118-7; 21-09-2023).

Avci, H., & T. Adiguzel (2017). A case study on mobile-blended collaborative learning in an English as a foreign language (EFL) context. In: The International Review of Research in Open and Distributed Learning 18 (7). (https://doi.org/10.19173/irrodl.v18i7.3261; 21-09-2023).

Bao, R. (2020). Collaborative dialogue between complete beginners of Chinese as a foreign language: Implications it has for Chinese language teaching and learning. In: The Language Learning Journal 48 (4), 414-426. (https://doi.org/10.1080/09571736.2017.1422136; 21-09-2023).

Chandra, R. (2015). Collaborative learning for educational achievement. In: IOSR Journal of Research & Method in Education (IOSR-JRME) 5 (3), 2320-7388.

Chen, L. et al. (2020). Artificial intelligence in education: A review. In: IEEE Access 8, 75264-75278.

Chew, S. & Y. Chen (2021). Speaking performance and anxiety levels of chinese efl learners in face-to-face and synchronous voice-based chat. In: Journal of Language and Education 7 (3), 43-57. (https://doi.org/10.17323/jle.2021.11878; 05-09-2023). 

Cotton, D. R.et al. (2023). Chatting and cheating: Ensuring academic integrity in the era of ChatGPT. In: Innovations in Education and Teaching International, 1-12.

Dai, W.et al. (2023, July). Can large language models provide feedback to students? A case study on ChatGPT. In: 2023 IEEE International Conference on Advanced Learning Technologies (ICALT),  IEEE, 323-325.

Dillenbourg, P. et al. (2009). The Evolution of Research on Computer-Supported Collaborative Learning. In: Balacheff, N. (Eds). Technology-Enhanced Learning. Springer, Dordrecht. (https://doi.org/10.1007/978-1-4020-9827-7_1; 05-09-2023). 

Eurydice. (2012). Key data on teaching languages at school in Europe. Education, Audiovisual and Culture Executive Agency.

Hsu, M. H. (2021). Proposing a task-oriented chatbot system for EFL learners speaking practice. In: Interactive Learning Environments, 1-12. (https://doi.org/10.1080/10494820.2021.1960864; 05-09-2023).

Huang W, et al. (2022). Chatbots for language learning – are they really useful? A systematic review of chatbot-supported language learning. In: Journal of Computer Assisted Learning, 38 (1): 237-257.

Jiang, D., & L. J. Zhang (2020). Collaborating with ‘familiar’strangers in mobile-assisted environments: The effect of socializing activities on learning EFL writing. In: Computers & Education 150, 103841. (https://doi.org/10.1016/j.compedu.2020.103841; 25-09-2023).

Jiang, R. (2022). How does artificial intelligence empower EFL teaching and learning nowadays? A review on artificial intelligence in the EFL context. In: Frontiers in Psychology 13, 1049401.

Johnson, D. W. et al. (2000). Cooperative learning methods: A meta-analysis. (https://www.researchgate.net/publication/220040324_Cooperative_learning_methods_A_meta-analysis; 01.12.2023).

Kanno, M. (2020). Maintaining and enhancing students' collaborative learning in a Japanese EFL higher education context. In: Journal of Education, Innovation, and Communication 91-106. (https://doi.org/10.34097/jeicom_sp_june2020_4; 21-10-2023).

Koç, E. (2018). Exploring collaborative learning with a focus on group activities in EFL Classrooms. In: İnönü Üniversitesi Eğitim Fakültesi Dergisi, 582–597. (https://doi.org/10.17679/inuefd.385741; 21-10-2023).

Kohnke, Lucas et al. (2023). ChatGPT for Language Teaching and Learning. In: RELC Journal 54 (2). (https://doi.org/10.1177/00336882231162; 01.12.2023).

Laal, M., & S. M. Ghodsi (2012). Benefits of collaborative learning. In: Procedia-social and Behavioral Sciences 31, 486-490. (https://doi.org/10.1016/j.sbspro.2011.12.091; 20-10-2023).

Long, M. H. (1996). The role of linguistic environment in second language acquisition. In: Ritchie, W. & T. K. Bhatia (Eds.): Handbook models for education. Learning and Individual Differences 103: 102274.

Momtaz, E. & M. Garner (2010). Does collaborative learning improve EFL students’ reading comprehension? In: Journal of Linguistics and Language Teaching 1 (1), 15-35.

Namatherdhala, B. et al. (2022). A comprehensive overview of artificial intelligence tends in education. In: International Research Journal of Modernization in Engineering Technology and Science 4 (7). (https://www.researchgate.net/publication/361912952_A_COMPREHENSIVE_OVERVIEW_OF_ARTIFICIAL_INTELLIGENCE_TENDS_IN_EDUCATION; 01.12.2023)

Nass, C. I. & S. Brave (2005). Wired for speech: How voice activates and advances the human-computer relationship. Cambridge: MIT press, p. 9.

Oravec, J. A. (2023). Artificial Intelligence Implications for Academic Cheating: Expanding the Dimensions of Responsible Human-AI Collaboration with ChatGPT. In: Journal of Interactive Learning Research 34 (2), 213-237.

Pinochet, L. H. C. et al. (2023). Collaborative Work Alternatives with ChatGPT Based on Evaluation Criteria for its Use in Higher Education: Application of the PROMETHEE-SAPEVO-M1 Method. In: Procedia Computer Science 221, 177-184.

Sok, S. & K. Heng (2023). ChatGPT for education and research: A review of benefits and risks. Available at SSRN 4378735.

Su, Y. et al. (2023). Collaborating with ChatGPT in argumentative writing classrooms. In: Assessing Writing 57, 100752.

Sumakul, D. T. Y. et al. (2022). Artificial intelligence in EFL classrooms: Friend or foe? In: LEARN Journal: Language Education and Acquisition Research Network 15 (1), 232-256.

Suthers, D. D. (2003). Representational Guidance for Collaborative Inquiry. In: Andriessen, J. et al. (eds) Arguing to Learn. Computer-Supported Collaborative Learning, vol 1. Springer, Dordrecht. (https://doi.org/10.1007/978-94-017-0781-7_2; 10-08-2023).

Swain, M. (1995). Three functions of output in second language learning. In: G. Cook, & B. Seidlhofer (Eds.): Principle and practice in applied linguistics: Studies in honour of H. G. Widdowson. Oxford: Oxford University Press, 125-144.

Tran, V. D. (2019). Does Cooperative Learning Increase Students' Motivation in Learning? In: International Journal of Higher Education 8 (5), 12-20.

Uribe, D., et al. (2003). The effect of computer-mediated collaborative learning on solving III-defined problems. Educational Technology Research and Development 51 (1), 5-19 (https://doi.org/10.1007/BF02504514; 15-08-2023).

Vaccino-Salvadore, S. (2023). Exploring the Ethical Dimensions of Using ChatGPT in Language Learning and Beyond. In: Languages 8 (3), 191.

Wang, T. (2023). Exploring the Potential Impact of Artificial Intelligence (AI) on International Students in Higher Education: Generative AI, Chatbots, Analytics, and International Student Success. In: Applied Sciences 13 (11), 6716.

Wang, Z. (2022). Computer-assisted EFL writing and evaluations based on artificial intelligence: a case from a college reading and writing course. In: Library Hi Tech 40 (1), 80-97.

Weinberger, A. (2005). Epistemic and social scripts in computer-supported collaborative learning. In: Instructional Science 33, 1-30. (https://doi.org/10.1007/s11251-004-2322-4; 15-08-2023).

Yang, H. (2022). Implementation of an ai chatbot as an English conversation partner in EFL speaking classes. Recall 34 (3), 327-343. (https://doi.org/10.1017/s0958344022000039; 10-09-2023).

Zeng, G. & S. Takatsuka (2009). Text-based peer–peer collaborative dialogue in a computer-mediated learning environment in the EFL context. System 37 (3), 434–446. (https://doi.org/10.1016/j.system.2009.01.003; 18-09-2023).

Zhu, G. et al. (2023). Embrace Opportunities and Face Challenges: Using ChatGPT in Undergraduate Students' Collaborative Interdisciplinary Learning. In: arXiv preprint arXiv:2305.18616.





Authors:

Ebru Pınar Elmacı Er

Lecturer

School of Foreign Languages

University of Applied Sciences

Sakarya

Turkey

Email: pelmacier@subu.edu.tr



Dr Armin Weinberger

Full Professor

Department of Educational Technology

University of Saarland

Email: a.weinberger@edutech.uni-saarland.de