The increasing development of AI technologies has led to a greater acceptance of chatbot integration in educational environments, especially for language learning (Jeon, 2024; Koç and Savaş, 2025). Chatbots, or conversational agents, facilitate human-computer interaction using natural language, utilizing NLP technology (Das and Das, 2024). Research indicates that they can support language development across speaking, writing, listening, and reading, offering learners opportunities to practice and receive immediate responses at any time (Saeedi and Soltani, 2025). Lin and Mubarok (2021) found that students who engaged with chatbots showed higher levels of communication intensity and activity compared to those who did not interact with them. In a related study, Kim (2019) investigated Indigo, a voice chatbot, over a 16-week period with English language learners of varying proficiency levels. The study revealed that students interacting with the chatbot employed a wider range of negotiation strategies than peers in student-only groups, highlighting the potential of chatbots to diversify communicative tactics. Beyond speaking and negotiation, chatbots also contribute to writing and listening development by offering flexible, 24/7 practice opportunities that allow learners to identify weaknesses and work toward improvement (Winkler and Söllner, 2018). Importantly, they not only provide linguistic input but also emotional and motivational support, functioning as non-judgmental assistants that reduce anxiety and sustain engagement (Huang et al., 2022; Meng and Dai, 2021). Accordingly, these findings illustrate how chatbots extend their impact beyond linguistic gains to include affective benefits that are critical for sustained language learning.

Whereas chatbots offer important benefits for language learning and beyond, they also face notable challenges. Research shows that they struggle to convey empathy or foster emotional connection. Participants in Annamalai et al.'s (2023) study, for instance, reported a lack of emotional attachment during chatbot interactions. Concerns have also been raised about students' potential over-reliance on chatbots, which may hinder the development of critical thinking and problem-solving skills typically cultivated through human interaction (Darwin et al., 2024; Zhai et al., 2024). As Turkle (2016) cautions, digital tools can displace the depth of face-to-face communication, suggesting that the efficiency of chatbot use may come at the cost of empathy and intimacy in learning.

Overall, while chatbots have demonstrated clear potential for supporting both linguistic and affective aspects of language learning, their limitations highlight the importance of examining not only their instructional benefits but also their implications for learners' EI.

In today's digital world, the ability to manage emotions is increasingly recognized as essential for effective learning and communication. Audrin and Audrin (2023) emphasize that the proliferation of digital technologies requires individuals to cultivate emotional intelligence (EI) to navigate new forms of interaction and maintain wellbeing. Goleman's (1995) framework; self-awareness, self-regulation, motivation, empathy, and social skills, remains central for understanding how learners manage fear, anxiety, motivation, and interpersonal relationships in technology-mediated contexts.

Foreign language education illustrates the critical role of EI in learning outcomes. Language learning is highly emotion-driven, involving both negative and positive affective states. Negative emotions such as anxiety, demotivation, and boredom can hinder cognitive processes including memory, attention, and problem-solving (Mierzwa-Kamińska, 2025). In contrast, positive emotions such as enjoyment, resilience, and motivation strengthen persistence and engagement (Aydin and Tekin, 2023). Learners with higher EI are better equipped to regulate stress, sustain motivation, and adopt adaptive strategies in the face of these challenges (Li et al., 2021). Studies consistently show that EI supports interpersonal communication (Kakarla, 2025) and adaptability. EI has also been associated with performance in specific skills such as reading (Ebrahimi et al., 2018), writing (Ghanbari and Abdolrezapour, 2021), speaking (Afshar and Rahimi, 2016), and listening (Valizadeh and Alavinia, 2013). Beyond individual skills, EI predicts willingness to communicate (WTC), as emotional regulation and empathy reduce social anxiety and foster communicative confidence (Guo et al., 2024; Huang et al., 2025). EI also shapes the learning environment: teachers with higher EI tend to manage classrooms more effectively, build stronger relationships, and create supportive spaces that enhance students' emotional wellbeing (Brackett et al., 2011; Li et al., 2021). Learners with stronger EI likewise employ more effective strategies, including planning, monitoring, and evaluating their progress, which promotes persistence and academic success (Oxford, 2011).

Digital tools such as chatbots add another dimension to EI in education. Chatbots reshape communication by providing constant availability and immediate responses, reducing reliance on human support (Ashfaq et al., 2020; Qasem et al., 2023). While these affordances increase accessibility, they also risk diminishing the depth of human connection (Bhuvaneswari and Vijayakumar, 2021). To address this, developers increasingly design empathic and emotionally intelligent systems that can detect user moods and generate affectively appropriate responses (Følstad et al., 2018). Yet, research shows that users still prefer human interaction when empathy and intimacy are required, and often feel discomfort in chatbot communication (Adam et al., 2021; Rapp et al., 2021). This tension highlights both the promise and limitations of digital agents: while they can scaffold emotional regulation and support learning, they cannot fully replicate human-to-human emotional exchange.

Collectively, this body of research highlights the central importance of EI for both learners and teachers in a foreign language education, as well as the new dynamics introduced by digital tools such as chatbots. These findings highlight the need to consider not only the cognitive but also the emotional dimensions of technology-mediated learning.

Writing is one of the most emotionally demanding aspects of foreign language learning, often associated with anxiety, fear of failure, and reduced motivation (Cheng, 2002). Research has shown that learners with higher levels of EI are better able to regulate these emotions, persist in the writing process, and achieve more positive outcomes (Ghanbari and Abdolrezapour, 2021). In recent years, AI-assisted writing tools, particularly chatbots, have emerged as potential supports for these challenges (Wang, 2024). Chatbots offer safe and non-judgmental environments for the learners as the learners can take risks with language without embarrassment (Liu, 2024). Besides, Guo and Li (2024) conducted a mixed-methods study showing that self-made chatbots enhanced EFL students' writing motivation by increasing their confidence, clarifying goals, and fostering more positive attitudes toward writing. Besides motivational gains, recent evidence indicates that chatbots also foster learners' writing self-efficacy and self-regulated learning, as students report greater confidence, autonomy, and ability to manage their writing processes when supported by chatbot-assisted instruction (Apriani et al., 2024; Lee et al., 2025). Beyond affective benefits, chatbots also provide linguistic scaffolding by offering instant feedback (Labadze et al., 2023; Xiao, 2024). By providing immediate and non-judgmental feedback, chatbots can alleviate stress, foster adaptability, and scaffold learners' self-regulation (Ortega-Ochoa et al., 2024). In addition to surface-level corrections, chatbots have been found to guide learners in improving textual coherence, vocabulary richness, and stylistic appropriateness (Mills et al., 2025; Song and Song, 2023).

However, while chatbot-assisted writing has been considered highly beneficial, several challenges and limitations have been noted. Studies caution that excessive reliance on automated feedback may reduce opportunities for learners to exercise creativity, critical thinking, and independent judgment in the writing process. Dergaa et al. (2024) describe this risk in terms of AI-Chatbot Induced Cognitive Atrophy (AICICA), whereby disproportionate dependence on chatbots for cognitive tasks, such as generating ideas, problem-solving, or producing creative text, can lead to the underuse and eventual decline of essential cognitive abilities. Similarly, concerns about the originality of the texts have emerged, as some learners adopt chatbot-generated text without any modification or reflection (Laun et al., 2025). These issues highlight the importance of framing chatbots not as replacements for human instruction, but as supplementary tools that require careful pedagogical integration.

Taken together, while most existing studies have predominantly focused on the linguistic and motivational outcomes of chatbot-assisted writing (Lo et al., 2024), less attention has been given to the learners' emotional dimensions. For example, the study conducted by Mahapatra (2024) showed the potential of the chatbot-assisted writing in helping students overcome anxiety when requesting feedback. However, evidence on how learners experience broader affective states remains scarce. Since EI is closely linked to learners' ability to regulate anxiety, persist in writing tasks, and sustain motivation, examining EI in chatbot-assisted writing contexts offers a valuable perspective. Recent reviews explicitly recommend examining these emotional dimensions more systematically, as emotion plays a critical role in sustaining writing motivation and shaping learners' long-term attitudes toward L2 writing (Albadarin et al., 2024; Deep et al., 2025).

Building on this need to better understand the interplay between artificial intelligence and emotional intelligence in language education, the present study narrows its focus to the influence of chatbots, gender, and technology use on EI. Specifically, we explore the effects of routine chatbot interactions on the emotional intelligence of pre-service English language teachers. Since technology is increasingly embedded in everyday life, its potential influence on the development and expression of EI warrants systematic investigation. In addition, gender-based differences are examined to provide a more nuanced understanding of how individual traits, such as EI and AI competence, may vary across demographic groups. Given the central role EI plays in effective teaching and classroom management, particularly in the emotionally charged context of foreign language education, this study is both timely and relevant. To this end, the study is guided by the following research questions:

By systematically exploring these research questions, this study aims to offer meaningful insights into incorporating AI tools such as chatbots within educational contexts, particularly in EFL instruction.

This study adopts a quantitative research design, chosen for its structured approach to explore the relationship between emotional intelligence and chatbot usage in EFL education (Creswell, 2011). It aims to fill a gap in the literature concerning the effects of chatbots on emotional intelligence within this educational context. The Turkish version of the Bar-On EQ-i, consisting of 88 items, with the final item serving as a self-check statement (“I responded to each question sincerely and accurately”), was employed to assess variations in participants' emotional intelligence (Acar, 2002). Ensuring the accuracy and validity of the findings, it is crucial to use scales translated into the participants‘ native language (Herdman et al., 1998). This approach ensures that the instruments accurately reflect the intended research objectives (Reichenheim and Moraes, 2007). The translated version of the Bar-On EQ-i was extensively reviewed and validated by field experts, and Acar (2002) reported strong internal consistency for the Turkish adaptation, with a Cronbach's α of 0.92. In the present study, the EQ-i also demonstrated acceptable reliability, with a Cronbach's α of 0.68 for the total scale.

Figure 1 presents the study design, showing the quasi-experimental pretest–posttest structure, participant group, intervention, measures, and data analysis procedures.

The participant cohort consisted of pre-service English language teachers enrolled at a state university in Istanbul, Türkiye. A total of 44 participants (see Table 1) were selected as the sample through convenience sampling, based on practical criteria such as ease of access, geographical proximity, availability at a specific time, or willingness to participate (Sedgwick, 2013). A sensitivity analysis conducted in G^*Power for a paired-samples t-test (α = 0.05, two-tailed, power = 0.80) indicated that with n = 44, the study was sufficiently powered to detect moderate effect sizes of dz ≥ 0.38. In order to demonstrate that they were voluntarily involved in the research project and that they comprehended the purpose of the research to confirm their voluntary participation in the research project and their understanding of its purpose, participants were asked to submit a written consent document. Furthermore, they were guaranteed that this research endeavor would not impact their academic records or career aspirations. In an effort to enhance ethical adherence, participants were properly briefed on the confidentiality of their data and their entitlement to resign from the study at any given time without facing any consequences.

The study is composed of three stages: pre-test, intervention, and post-test. The seven-step procedure was followed during the intervention which was illustrated in Figure 2. Step 1: participants completed the Bar-On EQ Inventory. Step 2: A tutorial introduced ChatGPT for creative writing. Steps 3–6: participants engaged in weekly supervised creative writing tasks, each aligned with an emotional intelligence (EI) subscale (Week 1 = intrapersonal, Week 2 = interpersonal, Week 3 = stress management, Week 4 = general mood). Step 7: participants completed the Bar-On EQ Inventory again at the end of the program.

The data underwent analysis using the SPSS 22.0 software package. For continuous data, descriptive statistics were provided, including the mean and standard deviation. The normal distribution of the data was assessed using the Shapiro-Wilk test. To compare the means of two independent samples, the independent-samples t-test was employed, while the paired-samples t-test was utilized for comparing the averages of two dependent samples. In addition to significance testing, effect sizes were calculated using Cohen's d for paired comparisons to provide information on the magnitude of observed differences. The relationship between variables was examined using the Pearson correlation test, and effect sizes were also reported using r. Because multiple comparisons were conducted across the five EI subscales, the Holm–Bonferroni procedure (Holm, 1979) was applied to control the familywise error rate at α = 0.05. A statistical significance level of 0.05 was set for all analyses.

The paired sample t-test was applied to examine whether there was a difference in the mean scores applied at different times by the same people. Table 2 shows descriptive statistics on pre-test and post-test results obtained from emotional intelligence, intrapersonal, interpersonal, adaptability, stress management, and general mood scales. Shapiro–Wilk tests indicated that both pre-test (W = 0.969, p = 0.283) and post-test (W = 0.950, p = 0.055) EQ-i total scores did not significantly deviate from normality, supporting the use of parametric analyses. A paired-samples t-test on the EQ-i total scores revealed no significant change from pretest (M = 285.27, SD = 17.07) to posttest (M = 286.00, SD = 17.22), t₍₄₃₎ = −0.20, p = 0.839, 95% CI (−7.91, 6.45). The effect size was negligible (Cohen's d = 0.03), suggesting minimal overall change in emotional intelligence levels.

When Table 3 was examined, it was seen that there was an increase of 0.2809 points in the mean adaptability score. At the uncorrected level, this increase reached statistical significance (p = 0.034, d = −0.22). However, after applying the Holm–Bonferroni correction, the effect did not remain significant and should therefore be interpreted as a trend rather than a robust finding. All other subscales and overall EI showed negligible changes (|d| <0.10).

When the Table 4 was examined, it was seen that the mean scores of reality testing, adaptability and stress management during the intervention differed according to the gender variable (p < 0.05). While the mean reality testing and adaptability scores of males were statistically significantly higher than females, the mean stress management scores of females were higher than males. In terms of intrapersonal, interpersonal, and general mood, male and female participants showed similar tendencies.

Based on the statistics provided in the Table 5, the relationships between various variables were examined. There was no significant relationship found between technology interest level and emotional intelligence (r = −0.097, p > 0.05). Similarly, no significant associations were observed between technology interest level and interpersonal or intrapersonal emotional intelligence (r = -, p > 0.05). However, a significant negative relationship was identified between technology interest level and interpersonal emotional intelligence focused on social responsibility (r = −0.304, p < 0.05).

Likewise, no significant correlation was found between technology skill level and emotional intelligence (r = −0.216, p > 0.05), nor between technology skill level and interpersonal or intrapersonal emotional intelligence (r = -, p > 0.05). There was also no significant relationship between technology skill level and adaptability (r = 0.003, p > 0.05). However, a notable negative correlation was found between technology skill level and stress tolerance (r = −0.420, p < 0.05).

Regarding artificial intelligence level, a significant negative correlation was observed with emotional intelligence (r = −0.408, p < 0.01) and with intrapersonal relationships (r = −0.521, p < 0.01). Conversely, there were no significant relationships found between artificial intelligence level and interpersonal relationships, adaptability, stress management, or general mood (p > 0.05).

This study examines the effects of chatbot integration on emotional intelligence components (intrapersonal skills, interpersonal skills, adaptability, stress management, and general mood) among pre-service English language teachers. The results indicate that participants consistently demonstrated stronger interpersonal skills than intrapersonal skills, as reflected by the highest scores in the interpersonal dimension and the lowest scores in the intrapersonal dimension, both before and after the intervention. This trend is consistent with foundational models of emotional intelligence, which conceptualize EI as a multidimensional construct encompassing both interpersonal and intrapersonal competencies (Goleman, 1995; Bar-On, 1997). However, recent studies emphasize that AI-mediated chatbots may particularly enhance the socially oriented dimensions of EI. For instance, Herath's (2025) review of 59 studies on emotionally intelligent chatbots in mental health demonstrates that such systems can reduce anxiety, foster emotional self-awareness, and promote supportive social interactions by simulating empathy. At the same time, the review draws attention to unresolved challenges, including cultural bias in emotion recognition and risks of over-dependence on AI, which highlight the need for cautious and ethically grounded integration in educational contexts. Similarly, Lu and Guo (2025) provide empirical evidence that emotionally intelligent chatbots can alleviate loneliness and social anxiety while enhancing social self-efficacy and psychological resilience among college students. Their study with 120 participants using the Replika chatbot over a 5-month intervention showed significant improvements in emotional resilience and social skills compared to a control group. These findings highlight the potential of emotionally intelligent chatbots not only to support social connection but also to strengthen adaptive emotional competencies in educational contexts, making them highly relevant for pre-service teacher training. Building on this, recent empirical research also supports the positive association between chatbot use and EI in educational contexts. Mosleh et al. (2024), in a cross-sectional study of 529 undergraduates in the UAE, found a statistically significant positive correlation between chatbot utilization and total EI scores. Students who interacted more frequently with chatbots demonstrated higher levels of emotional awareness, sociability, and adaptability, suggesting that consistent engagement with AI-mediated tools may contribute to EI growth in learning environments. These findings highlight not only the potential of chatbots to reduce anxiety and foster social connection, but also their capacity to strengthen core EI competencies when embedded into academic contexts.

Interpersonal skills, such as social awareness, empathy, and relationship management, are often more visible and frequently practiced in social contexts, potentially explaining their higher scores (Mayer and Salovey, 1997). In contrast, intrapersonal skills, which involve self-awareness and self-regulation, require deeper self-reflection and may be more challenging to develop without targeted interventions (Bar-On, 2006). The consistency of this trend across pre-test and post-test results suggests that the intervention may have impacted all areas equally without significantly altering the relationship between these dimensions. This emphasizes the need for future AI-assisted writing programs to integrate strategies that specifically enhance intrapersonal competencies, such as mindfulness training, reflective journaling, and emotional regulation exercises, to ensure a more balanced EI profile (Petrides and Furnham, 2001). Such strategies could complement the socially oriented gains facilitated by chatbot interaction, responding directly to the caution raised in recent reviews regarding over-reliance on AI for social-emotional support. Empirical evidence from Naseer et al. (2025) supports this concern, showing that among 500 educators and health professionals, frequent and extended use of AI was significantly linked to cognitive overload, reduced attention span, and heightened emotional stress. Their findings indicate that while AI can enhance efficiency and emotional engagement, unchecked dependence may compromise well-being and decision-making, reinforcing the importance of balanced and reflective integration in educational contexts.

When Table 3 was examined, it was observed that there was an increase of 0.2809 points in the mean adaptability score. Although this result reached statistical significance at the uncorrected level (p = 0.034, d = −0.22), it did not remain significant after Holm–Bonferroni adjustment and should therefore be interpreted as a trend rather than a robust effect. Nonetheless, the direction of change is consistent with prior research. For example, Dewi et al. (2025) reported that AI-driven tools improved adaptability in educational research, and Hessari et al. (2024) demonstrated that generative AI tools enhanced employee adaptability while reducing work overload. These studies suggest that adaptability is one of the more malleable components of emotional intelligence, and the trend observed in this study aligns with the possibility that structured AI-mediated interventions may support its development.

The results of this study indicate a moderate and statistically significant difference in emotional intelligence (EI) based on gender. Accordingly, female participants demonstrated greater competence in stress management compared to their male counterparts, consistent with the findings of a study conducted by Gentry et al. (2007). The study revealed that females prefer to employ more effective coping techniques for stress management than males. Similarly, a longitudinal study by Ptacek et al. (1992) investigated gender differences in coping strategies among undergraduate students. In their research, 186 male and female participants provided daily reports over 21 consecutive days. The findings revealed that female participants utilized a broader range of coping strategies compared to men. Furthermore, Graves et al. (2021) found that females are more likely to employ emotion-focused coping strategies. In a similar vein, Gupta et al. (2019) conducted a study to explore coping styles among male and female young adults. While male participants scored detached and avoidance coping styles, having a tendency to ignore stressors, female participants scored higher on emotional coping style with an inclination to manage stress better compared to males. Another study by Watson et al. (2011) indicated that women may have better stress management skills compared to men. However, contrasting findings have emerged in the literature, some studies (Graves et al., 2021; O'Rourke et al., 2022; Thoits, 1995) revealed that no gender differences were found in coping strategies in response to stressors. According to the findings of the study conducted by Thoits (1995) female and male participants employed similar coping patterns during childhood and adolescence. Moreover in a recent study conducted by Vidic (2024), no gender differences were observed in stress management skills. Apart from these studies, some studies also found non-consistent results with the current study. To illustrate, Hampel and Petermann (2005) also examined potential gender differences in coping styles with stressors, and according to its results female adolescents were found to use maladaptive coping strategies more frequently than male adolescents. Likewise, Al-Bahrani et al. (2013) aimed to investigate how gender influences coping styles and found similar results. According to their findings, female adolescents were significantly more likely to use maladaptive coping styles than male adolescents. This study highlights a moderate and statistically significant difference in emotional intelligence (EI) and stress management, favoring female participants. Supporting evidence from studies such as Gentry et al. (2007), Ptacek et al. (1992), and Gupta et al. (2019) suggests that females employ broader and more effective coping strategies, including emotion-focused approaches. However, contrasting findings from Thoits (1995) indicate no gender differences in coping, while studies by Hampel and Petermann (2005) and Al-Bahrani et al. (2013) suggest females may use maladaptive coping styles more frequently in adolescence. These mixed results highlight the complexity of gender differences in stress management, warranting further exploration.

The findings of this study also revealed that male participants exhibit greater adaptability compared to their female counterparts. The literature has also highlighted similar findings on adaptability among female and male participants. Reed and Reedman (2020) explored how gender influences resilience, adaptability and reactivity, and according to the findings, males demonstrated higher adaptability compared to their female counterparts. Besides, Zamarripa (2020) has shown that in the COVID-19 environment, women were found to be more stressed and less adaptive compared to men. Besides, several studies also support the notion that males demonstrate higher adaptability than females (Al-Bahrani et al., 2013; Hampel and Petermann, 2005).

The study revealed a moderately negative relationship between technology use and interpersonal emotional intelligence, suggesting that a greater interest in technology is associated with lower social intelligence. This finding highlights the potential trade-offs of increased technology use, as excessive time spent on computers and digital devices may reduce opportunities for personal communication and engagement in social activities. This aligns with Caplan's (2005) research, which found that individuals with limited self-presentational skills often prefer online social interaction over face-to-face communication. This preference may stem from the perceived anonymity and reduced social risk of online platforms, which can make them more appealing for individuals who struggle with in-person interactions. Besides, a recent study by Shanmugasundaram and Tamilarasu (2023) reported that excessive use of digital devices and social media is associated with diminished social skills and fewer face-to-face interactions, which is aligned with the findings of Benvenuti et al. (2023). In their study, the impact of technology use on adolescents' behavioral, social, physical, and cognitive development was investigated and they found that the pervasive use of smartphones and tablets among adolescents impacts their social behaviors, physical health, and cognitive development negatively.

Furthermore, there is a notable and statistically significant negative difference between technology and stress management. Consequently, those with a higher level of interest in technology are less likely to effectively manage stress. Those who use technology excessively may suffer from incompetency with stress management (Moroney et al., 2023). In exploring this dynamic, Qi and Yang (2024) conducted a study and found that higher digital resilience is connected with lower technological stress. Similarly, Suharti and Susanto (2014) suggested that a higher level of technology usage is associated with lower stress management. However, contrasting findings have emerged in the literature (Jerath et al., 2023; Ladakis et al., 2024). Research conducted by Jerath et al. showed that smartwatches provide promising avenues for stress management through continuous HRV monitoring and personalized feedback interventions. In a similar vein, (Ladakis et al. (2024) highlight the significant potential of virtual reality systems, often incorporating guided imagery and bio-sensing technology, to reduce stress in an immersive environment. On the other hand, (Eisen et al. (2008) compared computer-based and traditional face-to-face stress-management interventions and found that while both were effective in reducing stress immediately, there were no significant differences between them in terms of longer-term effectiveness, suggesting that the benefits of technology for stress management may depend on how it is implemented and used over time. These findings illustrate the complexity of the relationship between technology and stress management, with results depending on the specific technological tool, its application, and the individual's resilience and usage patterns.

A moderate negative correlation was observed between technology interest and interpersonal emotional intelligence, suggesting that higher technology usage might detract from social skills, as supported by Caplan (2005). There is also a negative association between technology interest and effective stress management, indicating challenges in stress handling among those highly engaged with technology.

A statistically significant moderate negative correlation was observed between artificial intelligence competence and overall emotional intelligence, suggesting that higher AI proficiency may be associated with lower emotional intelligence. One possible explanation is that as learners devote more attention to technical interaction with AI systems, they may engage less frequently in socially and emotionally rich exchanges with peers or instructors. Alternatively, individuals who are more confident in using AI may rely less on interpersonal support, which could also be reflected in lower EI scores. Our study aligns with the findings of (Annamalai et al. (2023), which explored chatbots using the framework of self-determination theory and discovered that participants experienced a lack of emotional connection during their interactions with chatbots. However, contrasting evidence exists. For instance, (Mosleh et al. (2024) reported that individuals who regularly engaged with chatbots demonstrated higher levels of emotional intelligence, particularly in dimensions of emotional regulation and adaptability. Their study suggests that chatbots may provide a safe, judgment-free environment that fosters self-reflection and the practice of emotion management strategies, thereby contributing positively to certain aspects of EI. In addition, shyness arises when the proficiency of a language learner in the desired language is impacted by the individual they are interacting with, as explained by Nunan (1991). During such circumstances, students frequently encounter feelings of worry or apprehension when engaging with other individuals. In order to mitigate these adverse sentiments among pupils, chatbots can be utilized. Hsu (2022) conducted a study where participants utilized the NeuroSky Mindwave to collect immediate data on their levels of attention. When interacting with an AI chatbot, their degree of mindfulness is noticeably increased, suggesting a condition of enhanced relaxation.

Despite this, AI has positively impacted adaptability in second language writing, aiding students in adapting to feedback and learning environments, as shown in studies by Bardzell and Bardzell (2013) and Lin and Chang (2020). These studies highlight productivity and skill improvement through AI and chatbot interaction.

There was a statistically significant moderate negative correlation between artificial intelligence competence and interpersonal emotional intelligence, which may suggest that greater reliance on AI is associated with reduced confidence in interpersonal contexts. With regard to adaptability, the findings indicated a small improvement, though the effect size was modest and should be interpreted cautiously. Nevertheless, this tendency aligns with previous research. AI may improve students' ability to adapt to L2 writing by delivering customized feedback, enabling adaptable learning experiences, replicating authentic language usage, providing error analysis and correction recommendations, and enhancing accessibility and flexibility in writing practice. For instance, Bardzell and Bardzell (2013) reported that productivity was the most commonly cited reason for using chatbots, highlighting their value in providing efficient support and information. Building on this functional role, Lin and Chang (2020) demonstrated how a chatbot (Chatbot DD) could also enhance the learning process more directly. In their study, students reported that the chatbot helped them identify weaknesses in their writing and improve as reviewers, with many perceiving noticeable gains in their writing skills. Together, these findings suggest that chatbots not only serve as productivity tools but can also foster adaptability by enabling customized feedback, error correction, authentic language practice, and flexible learning opportunities.