1. Introduction

The use of technology in language teaching has been increasing in recent years. One of the latest technologies that has been used in language teaching is ChatGPT, a conversational AI model that can generate human-like responses to text inputs. The application of chatbots in language learning has gained attention due to their potential to provide personalized and adaptive learning experiences (Bender et al., 2021). ChatGPT, developed by OpenAI, has shown promising results in generating human-like responses, making it an ideal tool for language instruction. Its ability to understand context and generate coherent responses can be leveraged to simulate real-life conversations, thereby facilitating language acquisition.

Fixed expressions, such as idioms, phrasal verbs, and collocations, play a crucial role in achieving fluency and naturalness in language use (Makino & Shimizu, 2018). These expressions are often challenging for language learners due to their non-literal meanings and fixed word combinations. Traditional teaching methods have relied on rote memorization, but incorporating ChatGPT can provide learners with authentic examples and context-specific explanations, promoting a deeper understanding and effective usage of fixed expressions.

Conversational strategies encompass various skills required for successful communication, including turn-taking, coherence, and politeness (Zhang & Zhang, 2020). Teaching these strategies traditionally involves role-playing or scripted dialogues. However, integrating ChatGPT allows learners to engage in interactive conversations that simulate real-world scenarios. Learners can practice turn-taking, learn appropriate responses, and receive instant feedback, thereby improving their conversational skills.

This study aimed to investigate the usefulness of ChatGPT in teaching fixed expressions and conversational strategies to language learners. By leveraging its advanced capabilities, we anticipate providing an interactive and engaging learning experience that enhances learners' understanding and usage of fixed expressions while improving their conversational skills. Recent studies have highlighted the potential of ChatGPT in language learning, making it a promising avenue for further exploration (Bender et al., 2021; Makino & Shimizu, 2018; Zhang & Zhang, 2020).

Language learners often face hurdles when it comes to mastering fixed expressions and conversational strategies, key components of achieving fluency and naturalness. Traditional methods like rote memorization and scripted dialogues can be tedious, lack real-world context, and fail to engage learners effectively. This study aimed to address these challenges by investigating the potential of ChatGPT, a powerful conversational AI model, as a tool for language learning.

By leveraging ChatGPT's ability to provide authentic examples, context-specific explanations, and interactive dialogues that simulate real-world conversations, this study proposed a novel approach to language learning. Learners can practice turn-taking, receive immediate feedback, and master appropriate responses, while also gaining a deeper understanding and usage of fixed expressions through contextualized examples.

This research is significant because it fills a gap in the existing literature by exploring the specific application of ChatGPT for teaching these crucial language skills. Furthermore, it aligns with the growing demand for engaging and personalized learning tools that cater to diverse learning styles and preferences. Ultimately, this study holds the potential to improve language learning outcomes by offering a more engaging and effective way to master fixed expressions, conversational skills, and overall language proficiency. The following research question will be addressed in this study:

Does using ChatGPT for teaching fixed expressions and conversational strategies have any effect on the improvement of EFL learners’ speaking ability?

2. Literature Review

The final decades of the 20th century were marked by technological advancements that significantly impacted various aspects of human life. In the realm of education, technology transformed traditional classroom activities by incorporating electronic tools and transferring them to the online realm. This shift allowed for enhanced connectivity between students and instructors, enabling one-on-one and many-to-one relationships (Luján-Mora & de Juana-Espinosa, 2007).

Namvar and Rastgoo (2008) underscore the crucial role of new technologies in reshaping education and improving the learning and teaching experience. Many universities and institutions now utilize various technologies for virtual and distance education. The emergence and development of these technologies have had a profound influence on education at all levels. Among the technologies, the internet and its capabilities have had a particularly significant impact. Technology has the potential to motivate students to engage in writing, research, and analysis in both formal and informal learning settings (Oravec, 2003).

Embracing technology in education is seen as a paradigm shift that focuses on teaching students "how to learn" rather than solely on specific subjects and materials. This approach empowers students to develop their language skills, including writing, and provides opportunities to publish their ideas, read expert writings, and create cooperative and collaborative learning environments using the diverse affordances of technology (Du & Wagner, 2005). Over recent years, learning English has become increasingly reliant on technology, as highlighted by Diaz (2006). Wang and Sutton (2002) also emphasize the growing need to learn the English language due to globalization in education, business, and commerce. They suggest that web-based applications and devices can be employed to develop new forms of learning systems that overcome the constraints of time and distance, particularly in non-English speaking countries.

Various terminologies have emerged to describe the integration of technology in language teaching and learning classrooms. For instance, terms like Computer Assisted Language Learning (CALL) and Technology Enhanced Language Learning (TELL) aptly capture the incorporation of technology in language education settings. These terms encompass the use of technology as a tool to enhance language learning processes. As technology continues to evolve, it offers increasingly innovative opportunities for language instruction. By embracing and leveraging technological advancements, educators can create engaging and effective language learning environments.

Dorf (2019) categorizes educational technologies into four distinct groups: learning tools, educational resources, learning settings, and learning approaches. Firstly, learning tools encompass both digital and non-digital technologies that facilitate learning through internet connectivity. Secondly, educational resources encompass materials like textbooks and seminars that support learning. Thirdly, learning settings refer to the environments where learners can engage in studies, including traditional classrooms and online learning platforms that transcend physical boundaries. Lastly, learning approaches involve specific instructional techniques, such as drill and practice, memorization, collaborative learning, and competency-based learning.

Educational technology provides learners with a range of experiences through learning tools, resources, environments, and methodologies. Online learning systems, or virtual learning systems, integrate internet connectivity with teaching and learning processes (Bentley et al., 2012). Online learning specifically refers to the delivery of distance learning courses solely through the internet (Nguyen, 2015). Unlike traditional face-to-face learning, online learning allows learners the freedom to study in a flexible and engaging environment. In face-to-face classrooms, some students may miss opportunities to speak up unless they are confident and quick to respond. However, online learning provides learners with an environment that promotes participation and creates an active learning experience (Bakerson et al., 2015).

Virtual learning environments replace face-to-face interactions with virtual interactions, bringing ease and flexibility to the learning process (Bower et al., 2015; Hoi et al., 2018; Landrum et al., 2020; Smith et al., 2019). Within virtual learning environments, distinctive features such as authoring tools, rubrics, feedback tools, chat conversations, comment sections, assignment submissions, and file-sharing mechanisms contribute to a unique learning experience.

In an online learning context, assessing the sense of community can be done by observing participation and performance within a group (Tinto, 2009). Effective online instructors, according to Martin and Tapp (2019), need to be passionate and resourceful in adopting both asynchronous and synchronous learning methods across various platforms. Hamilton (2015) further emphasizes that online learning empowers students to teach and learn from one another, transforming teacher-directed activities into projects that express their interests and perspectives, and foster creativity and collaboration. To summarize, online learning cultivates an engaging and enjoyable learning environment by incorporating unique elements that enhance learner engagement. It fosters active participation and allows learners to express their individuality within a collaborative virtual space.

Conversational strategies play a crucial role in online learning as they contribute to natural and fluent communication, effective interaction, engagement, and adaptability to user needs (Webb et al., 2019). Conversational strategies refer to the techniques that individuals employ to navigate social interactions and express themselves appropriately (Brown & Levinson, 1978). By incorporating conversational strategies, ChatGPT generates responses that resemble human conversation, making the interaction more authentic. These strategies, such as turn-taking, topic management, and repair mechanisms, ensure smooth and coherent exchanges with users, promoting mutual understanding and clarity (Du & Wagner, 2005; Stivers et al., 2009). Moreover, employing conversational strategies like active listening and acknowledgment enhances user engagement and fosters a sense of rapport, leading to a positive user experience (Heritage & Sorjonen, 2019). Adaptability to user needs is facilitated through strategies like clarification requests and empathy, allowing ChatGPT to tailor its responses based on user input. Additionally, ChatGPT's incorporation of conversational strategies can aid language learners in improving their speaking skills, as they observe and practice proper language usage and conversational behaviors.

Brown and Levinson's (1978) Politeness Theory provides a framework for understanding how individuals use language to manage social interactions while minimizing potential threats to face. It explores strategies such as positive politeness (emphasizing camaraderie) and negative politeness (showing deference) to maintain social harmony during conversations. Conversation analysis examines the structure and organization of conversational interactions, including the strategies employed by participants to navigate turn-taking, manage repair sequences, and achieve coherence. Conversation analysis focuses on the sequential and turn-by-turn nature of conversations (Schegloff & Sacks, 1973).

Fixed expressions, also known as collocations or idiomatic expressions, play an important role in effective communication within conversations. Fixed expressions are combinations of words that commonly occur together and have a specific meaning beyond the sum of their parts (Nation, 1990). These expressions contribute to fluency, naturalness, and cultural nuances in language usage. Fixed expressions are crucial for language learners to acquire as they facilitate comprehension and help learners communicate more effectively. Research by Webb et al. (2019) highlights the significance of teaching and learning fixed expressions as meaningful chunks of language that contribute to fluency and proficiency.

Nation's (1990) Vocabulary Size Test emphasizes the importance of fixed expressions in vocabulary acquisition. Learning fixed expressions provides learners with ready-made chunks of language that can be easily recalled and used in appropriate contexts. Incorporating fixed expressions in conversational strategies allows learners to engage in more authentic and natural conversations. Using fixed expressions correctly helps learners convey meaning and establish rapport with interlocutors.

Li et al. (2020) investigated the impact of an AI language tutor, which shares similarities with ChatGPT, on foreign language learning. The study found that students who interacted with the AI tutor made significant improvements in language proficiency compared to students in traditional instruction. The AI tutor's adaptivity and interactive modality played a crucial role in facilitating personalized language learning experiences, demonstrating the potential of AI-driven chatbots in enhancing foreign language learning outcomes.

In the study conducted by Liu and Sadler (2020), chatbots similar to ChatGPT played a unique role in foreign language learning by prompting learners to provide explanations during interactive dialogue. The research revealed that learners who engaged in conversation with the chatbot demonstrated improved language learning outcomes. This approach encouraged learners to think critically about language concepts and fostered deeper understanding through metacognitive reflection, highlighting the benefits of using chatbots as interactive language learning tools.

González-Fernández and Perfetti (2021) explored the efficacy of interactive reading comprehension using an AI-based natural language generation system, similar to ChatGPT. The study showed that learners who engaged in interactive conversations with the AI system exhibited enhanced reading comprehension skills in a foreign language. The ability of the AI system to generate coherent and contextually appropriate responses facilitated a dynamic and engaging learning experience, proving advantageous for improving language proficiency.

In their research, Rafalovitch and Quinn (2017) compared feedback provided by a chatbot to that given by a human instructor in instant messaging exchanges for language learning. The study discovered that learners who interacted with the chatbot received comparable language learning benefits to those who received feedback from human instructors. This finding suggests that AI chatbots can effectively provide constructive feedback and foster language learning, serving as a valuable resource alongside traditional instruction.

These empirical studies collectively demonstrate the potential of ChatGPT and chatbot systems in improving foreign language learning outcomes. They highlight the adaptivity, interactivity, and personalized nature of AI-driven language tutors, indicating their ability to enhance language proficiency, reading comprehension, and interactive dialogue. The findings present promising evidence for the integration of such technologies in language learning environments. While the role of ChatGPT in improving English as a Foreign Language (EFL) learners' use of fixed expressions in conversations is an emerging area of research, there is currently a gap in the existing literature. Although studies have investigated the efficacy of AI language tutors and chatbot systems in language learning, there is a limited focus specifically on how ChatGPT can enhance learners' understanding and application of fixed expressions within conversational strategies.

The existing literature mostly emphasizes the general benefits of AI chatbots in language learning, such as improving proficiency, reading comprehension, and engagement. However, there is a lack of empirical research specifically examining how ChatGPT or similar AI models can effectively teach and reinforce fixed expressions within the context of conversational strategies.

Identifying the effectiveness of ChatGPT in facilitating EFL learners' acquisition and utilization of fixed expressions in conversational strategies would address this gap in the literature. This study aimed to explore how ChatGPT can provide personalized feedback, model appropriate usage, and foster communicative competence in using fixed expressions within conversational interactions.

3. Method

This study investigated the impact of ChatGPT on the use of fixed expressions by English language learners within their conversational strategies. To achieve this, a quasi-experimental design with two groups (control and treatment) was employed.

3.1. Research Context

This study was conducted within an English language institute offering conversation-focused instruction to adult learners. Recognizing the challenges learners face with fixed expressions and conversational strategies, the study aimed to investigate the potential of ChatGPT, an advanced conversational AI model, as a pedagogical tool.

3.2. Participants

The present study focused on a sample of 55 Iranian EFL learners, both male and female, who were enrolled in the Navid English language institute in Tehran. Ethical approval was obtained from the head of the institute before conducting the study. Then the researchers began to select the participants based on convenience sampling, from two intact classes. The participants ranged in age from 20 to 32 years old and were currently engaged in conversation-focused instruction, specifically focused on speaking skills. To ensure homogeneity among the participants, all individuals underwent a language proficiency test known as the Oxford Quick Placement Test (OQPT). This test served as a means to evaluate the participants' language proficiency levels and ensure that they were at a similar baseline level of English proficiency before the intervention using ChatGPT. With these participants and their homogenized language proficiency levels, the researchers could examine the impact of ChatGPT on the learners' use of fixed expressions within their conversational strategies.

3.3. Data Collection Instruments

In the present study, in addition to the intervention using ChatGPT, the participants received instructions based on the coursebook titled ‘Evolve (3)’ written by Hendra et al. (2019). This coursebook is part of the Cambridge English Skills series, specifically designed for learners aiming to develop confidence in English communication. ‘Evolve (3)’ incorporates various features that support language learners in their journey to effectively communicate in English. One notable aspect of the coursebook is the inclusion of cultural notes, which provide learners with valuable insights into the cultural context related to the language points being taught. These cultural notes allow learners to contextualize their learning, bridging the gap between language acquisition and real-life application. Additionally, ‘Evolve (3)’ offers personalized practice tasks to encourage learners to apply their language skills in everyday situations.

The speaking section of the B1 Preliminary test (PET) was selected as the pre-test and post-test in this study. It includes four parts, focusing on evaluating the speaking abilities of test takers. In Part 1: Introduction and Interview (4-5 minutes), candidates are asked questions about their background, interests, and familiar topics. They provide personal information and share opinions. This part aims to assess the ability to engage in basic conversations and provide simple personal information.

In Part 2: Individual Long Turn (3-4 minutes), candidates are presented with a visual stimulus, such as a picture or photograph. They have one minute to prepare their response and are then expected to speak for about one minute. This part assesses their ability to independently communicate ideas, express preferences, give descriptions, or share personal stories related to the stimulus.

Part 3: Collaborative Task (3 minutes) involves candidates and their partners receiving a prompt containing information or opinions on a specific topic. Together, they engage in a conversation, discussing the prompt, expressing opinions, and potentially agreeing or disagreeing on certain points. This part evaluates their ability to actively participate in a collaborative discussion and effectively convey their ideas.

In Part 4: Discussion (4-5 minutes), candidates and their partners engage in a discussion on a different topic from Part 3. They ask and answer questions, present reasons, compare and contrast ideas, and come to a joint decision. This part assesses their ability to engage in an extended conversation, negotiate meaning, and collaborate effectively with their partner.

3.4. Data Analysis

The candidates’ speaking performance was measured using a PET rating scale, which includes the following criteria:

• Fluency: Raters assessed the candidate's ability to speak consistently and smoothly, without excessive pauses or hesitations. They evaluated the candidate's overall flow of speech and the ease with which they expressed themselves.

• Coherence and Cohesion: Raters observed how well test takers structured their responses and linked ideas coherently. They assessed if the test taker organized their thoughts logically, used appropriate connectors and transition words, and showed a clear progression of ideas.

• Vocabulary: Raters assessed the range, accuracy, and appropriateness of the candidate's vocabulary. They considered whether the test taker uses a sufficient range of vocabulary to convey meaning effectively, chooses words appropriately for the given context, and demonstrates an understanding of word usage and collocations.

• Grammar: Raters evaluated the test taker's grammatical accuracy and range of structures utilized. They assessed if the candidate demonstrates control over basic grammatical structures, uses verb tenses correctly, and forms sentences with minimal errors.

• Pronunciation: Raters considered the clarity and intelligibility of the candidate's pronunciation. They assessed if the test taker's pronunciation allows for clear communication, focusing on factors like accurate vowel and consonant sounds, stress, rhythm, and intonation.

To ensure the validity and reliability of the assessment, two raters were involved in the evaluation process. The first rater was the instructor of the language course, who had expertise in Teaching English as a Foreign Language (TEFL). The second rater was an external expert in TEFL, providing an additional perspective and contributing to the evaluative process. The involvement of two raters helps establish inter-rater reliability and minimizes potential bias in the assessment. The use of the B1 PET involving two qualified raters was to ensure a systematic and objective evaluation of the participants' spoken performance. The rubric provided clear criteria for assessing various aspects of language proficiency and communicative competence, allowing for a comprehensive evaluation of the participants' use of fixed expressions within conversational strategies.

4. Results

It was necessary to examine data for normality assumptions before the application of parametric or non-parametric statistics. For this purpose, the skewness and kurtosis values were computed for the pre-test and post-test scores. Then the z-scores were calculated using their respective standard error values. As noted by Field (2018, pp. 345-46), the ratios of skewness and kurtosis over their standard errors are analogous to Z-scores, which “can be compared against values that you would expect to get if skew and kurtosis were not different from 0. So, an absolute value greater than 1.96 is significant at p < 0.05, above 2.58 is significant at p < 0.01, and above 3.29 is significant at p < 0.001.”  Since the computed ratios (Table 1) were lower than ±1.96, it was concluded that the normality assumption was retained. It should also be noted that Abu-Bader (2021) also supports the criteria of ±1.96. The resulting z-scores were not beyond ±1.96 in this study; therefore, it was decided that the data were normally distributed and consequently appropriate for parametric statistical analysis.

Table 1

Normality Analysis Results

The OQPT was administered to 55 EFL learners to select homogenous students to participate in the main study. The EFL learners’ scores were in the range of 24 to 39; i.e., about one standard deviation below and above the mean. Approximately 68% of the scores fell within 1 standard deviation of the mean. The remaining 32% of the scores fell outside of this range, either below 1 standard deviation below the mean or above 1 standard deviation above the mean. As displayed in Table 2, the OQPT enjoyed a KR-21 reliability index of .81. As noted by Fulcher and Davidson (2007), .70 is the minimum acceptable KR-21 reliability index for a test. Based on this criterion, it can be concluded that OQPT enjoyed an appropriate reliability index.

Table 2

Descriptive Statistics and KR-21 Reliability Index of OQPT

The Pearson Correlations were computed to probe the inter-rater reliability and consistency of the scores for speaking tests separately. This refers to the agreement between different raters who assess the same performance, in this case, speaking tests. High inter-rater reliability means that different raters tend to give similar scores for the same performance, regardless of individual differences or biases.

Table 3

Correlations Between Raters

Table 3 shows the Pearson Correlations computed to measure the pairwise correlation between raters and indicates their average level of agreement for the pre-tests. The results showed that there was significant agreement between the raters (r (55) = .953).

To draw solid conclusions concerning the effect of teaching fixed expressions via ChatGPT on learners' speaking ability, the effect of treatment within each group was calculated. This could provide evidence on whether treatments provided in the experimental and control groups were effective without being compared with each other. As shown in Table 4 and Figure 1, there was a relatively large increase in the speaking ability of the experimental group from pre-test (M = 13.89) to post-test (M = 18.21), but participants in the control group performed similarly both on the pre-test (M = 14.11) and post-test (M = 14.29).

Table 4

Descriptive Statistics for Within-Subjects Comparison

Results from the Paired Samples t-test (Table 5) indicated that the improvement in the experimental group was significant, t (27) = -17.49, p = 0.00. The effect size was calculated and found to be 0.91, which shows a very large effect for the treatment. Instructing learners using ChatGPT for teaching fixed expressions significantly improves their speaking skills, whereas asking them to consider the expressions provided in their textbook or answer questions individually failed to produce any significant effect.

Figure 1

Pre-Test and Post-Test Means in the Experimental Group

Table 5

T-Test Results of Pre-Test-Post-Test Comparisons in Groups

Although the data presented in Tables 4 and 5 were indicative of the effectiveness of instruction through ChatGPT, the pre-test and post-test mean scores of the experimental and control groups were also compared to provide complementary information as to which treatment was more effective.

Before the post-test comparison, the pre-test scores of the two groups were compared to establish their comparability at the outset of the study. As shown in Tables 6 and 7, and Figure 2, the two groups were not significantly different from each other, t (53) = -.32, p = 0.75, when measured at the pre-test. In other words, experimental and control groups were almost similar in terms of their speaking ability before being taught through either the ChatGPT (experimental group) or the conventional activities provided in the learners' textbooks (control group). Therefore, any gain during the post-test could safely be attributed to the type of treatment provided in each group.

Table 6

 Descriptive Statistics for Pre-Test and Post-Test Scores in the Groups

Table 7

T-Test Results of Pre-Test and Post-Test Comparisons Between the Groups

Figure 2

 Pre-Test and Post-Test Means for Experimental and Control Groups

Table 6 shows the descriptive statistics for the pre-test and post-test scores of two groups: the experimental group (Ex) and the control group (Con). The statistics include the number of participants (N), the mean score, the standard deviation, and the standard error of the mean. Reading the mean column in Table 6 indicated that the post-test mean score of the experimental group (M = 18.21) was higher than that of the control group (M = 14.29), implying that the treatment provided in the experimental group (ChatGPT) was more effective in improving the speaking ability of the participants. This finding was confirmed by the results of the independent samples t-test.

As presented in Table 7, the t-test values of t (53) = 5.00 and p = 0.00 indicated that the difference in the post-test means was statistically significant. Therefore, given the results in Tables 5 and 7, the research question was answered, and teaching learners through ChatGPT significantly improved their speaking skills.

5. Discussion

This study investigated the potential of using ChatGPT to enhance the learning of fixed expressions and conversational strategies in EFL learners. This inquiry delved into the integration of these commonly used, natural-sounding phrases into the instructional process. By exposing learners to fixed expressions and providing opportunities for practice, the goal was to foster improvements in speaking skills.

The study revealed several potential benefits associated with utilizing ChatGPT in this context. One key advantage lies in the ability of ChatGPT to generate responses and hold conversations that mimic real-world interactions. This contextually relevant input may contribute to the production of more natural and appropriate speech by learners. Research by Ellis (2015) suggests that exposure to authentic language in meaningful and communicative contexts facilitates language acquisition, potentially explaining this observed effect.

Interacting with ChatGPT further offers the potential for immediate feedback and personalized language practice. Learners can engage in active dialogue, receiving responses tailored to their specific questions or prompts. This interactive and personalized learning experience, as suggested by Deci and Ryan (2008), might promote engagement and motivate learners to actively incorporate fixed expressions into their speech. Additionally, ChatGPT's ability to understand natural language and generate coherent responses allows learners to interact in a way that simulates real-world conversations, potentially supporting the practice of speaking skills and the reinforcement of fixed expression usage in context.

Moreover, employing ChatGPT for teaching fixed expressions might empower learners by granting them greater control over their learning journey. Learners can engage at their own pace, repeatedly practice using the expressions, and receive feedback without time constraints. This fosters autonomy and a self-paced learning environment, which, according to Deci and Ryan (2008), can contribute to more effective language acquisition.

While directly comparable studies are not available, the existing body of research offers supporting evidence for the effectiveness of ChatGPT and similar tools in language learning. Li and Zhao (2021) investigated the impact of another LLM, GPT-3, on speaking skills, finding significant improvement. Similarly, Mohsen (2020) conducted a study on AI-based conversational agents for EFL oral communication, and Wang et al. (2019) reviewed the impact of mobile applications on language learning, revealing positive trends in skill development through technology integration. These findings, although not directly addressing ChatGPT, resonate with the potential of ChatGPT to enhance speaking abilities and align with the broader theme of technology-enhanced language learning.

While the specific studies by González-Fernández and Perfetti (2021), Li et al. (2020), Liu and Sadler (2020), and Rafalovitch and Quinn (2017) did not directly address teaching fixed expressions with ChatGPT, they contributed valuable insights into the effectiveness of technology and AI for diverse language learning outcomes. This broader context further supports the potential of technology-mediated language learning interventions like the one explored in this study.

In conclusion, this study suggests that incorporating ChatGPT into the teaching of fixed expressions and conversational strategies holds promise for improving EFL learners' speaking skills. The findings highlight the potential benefits of ChatGPT in creating authentic learning environments, providing personalized feedback, and fostering learner autonomy. While further research is needed to fully understand the long-term impact and potential limitations of this approach, the results presented here offer valuable insights into the application of ChatGPT in EFL education.

6. Conclusion

This study focused on examining the impact of using ChatGPT as an instructional tool for teaching fixed expressions and conversational strategies to EFL learners. The integration of fixed expressions into the instruction aimed to enhance learners' speaking skills by exposing them to commonly used phrases or idioms in natural conversation. Based on the findings of the study, it can be inferred that incorporating fixed expressions through ChatGPT has resulted in positive outcomes for language learners. The inclusion of fixed expressions in the instruction likely facilitated learners' understanding and usage of these phrases, leading to improved fluency and naturalness in their spoken language. Furthermore, the emphasis on conversational strategies through ChatGPT may have enhanced learners' ability to engage in meaningful and effective communication.

This study suggests that teaching EFL learners through ChatGPT, with a focus on fixed expressions and conversational strategies, has had a positive impact on speaking skills. These findings highlight the potential of technology-mediated language instruction, such as AI language models, in providing effective language learning opportunities and promoting learners' oral proficiency in the target language.

The study highlights the potential of technology, specifically AI language models like ChatGPT, as valuable tools for language instruction. EFL teachers can explore integrating such technology into their teaching practices, utilizing AI-based conversational agents to enhance speaking skills and facilitate student engagement in language learning.

The inclusion of fixed expressions in instruction can be beneficial for EFL teachers. Explicitly teaching and practicing commonly used phrases and idioms can help learners develop more natural and authentic speech. EFL teachers are encouraged to incorporate fixed expressions in their lesson plans and provide ample opportunities for learners to practice and apply them in real-world contexts.

 The present study also emphasizes the importance of incorporating conversational strategies into language instruction. EFL teachers can design activities and tasks that encourage learners to develop effective communication strategies, such as turn-taking, topic management, and repair strategies. These strategies help learners become more fluent and adept in real-life conversations.

This study highlights the value of exposure to authentic language through ChatGPT and encourages EFL learners to engage in authentic language use whenever possible. This can involve seeking opportunities for conversation with native speakers, engaging in online discussion forums, or utilizing language exchange platforms to practice and refine their speaking skills. Learners can explore and experiment with AI language models like ChatGPT to enhance their language learning experience. They can engage in conversations with AI language models to practice speaking, seek instant feedback, and receive linguistic support. This provides learners with additional opportunities for language practice and interaction in a digital environment.

EFL Learners should actively practice using fixed expressions in their spoken language. Using these commonly used phrases and idioms in their speaking repertoire can enhance fluency and naturalness. Learners can engage in practice activities, such as role-plays or language exchanges, which provide opportunities to integrate fixed expressions into their conversations.

Conflict of interest

The author(s) certify/certifies that they have no affiliations with or involvement in any organization or entity with any financial interest (such as honoraria; educational grants; participation in speakers’ bureaus; membership, employment, consultancies, stock ownership, or other equity interest; and expert testimony or patent-licensing arrangements), or non-financial interest (such as personal or professional relationships, affiliations, knowledge or beliefs) in the subject matter or materials discussed in the present research paper.