The impact of generative AI on foreign language enjoyment: the roles of gender, English proficiency and usage duration among Chinese EFL learners

Abstract

Background

With the integration of Generative AI (GenAI) technology in language educational settings, the impact of GenAI tools has garnered increasing attention. Yet extant research has predominantly examined GenAI's cognitive impacts on learners' linguistic abilities, leaving the empirical landscape of learners' emotional experiences largely uncharted. This study, drawing on Positive Psychology, investigated the impact of GenAI tools on Chinese EFL students' foreign language enjoyment (FLE), as well as the roles of three factors—gender, English proficiency, and GenAI usage duration—on students' FLE.

Methods

A mixed-methods design was applied. Ninety-eight Chinese EFL university students completed the survey and 10 participated in the interview after the 12-week intervention. SPSS 24.0 was employed to perform descriptive statistics, independent-samples t-tests, and one-way ANOVA in sequence on the survey data, and the interview data were manually coded using thematic analysis.

Results

Findings indicated that participants reported medium level of FLE, and the FLE-teacher dimension was slightly higher than FLE-private and FLE-atmosphere. Among three influential factors examined, although neither gender nor English proficiency significantly influenced participants' FLE statistically, male participants exhibited a marginally higher level of FLE compared to their female counterparts. Noticeably, participants with relatively lower English proficiency levels tended to experience a slightly higher degree of FLE than those with higher English proficiency levels. However, this subtle intergroup difference did not reach statistical significance. Moreover, the duration of GenAI tool usage is shown to statistically impact the participants' FLE level, with six-month duration serving as a threshold at which participants' FLE undergoes a positive qualitative shift.

Conclusions

Participants generally perceived that GenAI tools fostered an enjoyable learning environment, but they also expressed concerns about potential over-reliance on GenAI technology and the possible erosion of their critical thinking abilities. The study recommends encouraging students to use GenAI tools for more than six months, enhancing female students' digital competence through learning tasks requiring cooperation between boys and girls, and designing more process-oriented tasks that require reflective thinking and sustained student engagement to strengthen critical thinking skills.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40359-025-03870-y.

Introduction

The impact of Generative Artificial Intelligence (GenAI) on English as a Foreign Language (EFL) learners has gained increasing attention due to its great effect on the language learning experience. GenAI tools can tailor educational resources to suit the individual needs of learners. This is particularly beneficial in EFL settings, where students often vary in proficiency and preferences, as it helps accommodate diverse learning styles [1].

Studies have proved that the usage of GenAI tools can help students improve their linguistic skills [2–4], as well as their non-linguistic abilities [5]. It has revealed that AI-assisted learning applications have an overall positive impact on English speaking ability [2]. Also, ChatGPT has advantages in assisting L2 writing, including "promoting learner autonomy, enhancing feedback literacy, and improving writing quality by providing immediate corrections and suggestions" [3]^p22. Moreover, students talking with the chatbot maintained a stable level of situational interest in the value dimension while doing extensive English reading [4]. Meanwhile, it is found that AI-based tools are helpful in developing students' information analysis skills and constructing arguments, and students rate their critical thinking skills as high or moderate [5]. Recently, some studies shifted their focus into GenAI tools' impact on language learners' emotional experience [6, 7]. One of them has shown that students found learning English more enjoyable in a less stressful digital space because of their higher willingness to communicate, reduced anxiety, and increased confidence in their communicative abilities [6]. However, given that the emergence of GenAI technology has been less than three years and relative studies are at initial stage, the impact of GenAI on language learners' emotions remains a vital area of study.

As GenAI offers unprecedented opportunities for language teaching and learning practice [1, 7], it is important to understand its influence on learners' emotions. Enjoyment, as the most common positive emotion experienced by L2 learners [8], has received more and more scholarly attentions [9, 10], so it is meaningful to discuss the impact of GenAI on language learners' enjoyment in GenAI-driven educational context. Concurrently, guided by the principle of Positive Psychology (PP), language educators are encouraged to create a positive emotion-arousal environment among learners in class [11]. In light of this, this study explores the impact of GenAI tools on Chinese EFL learners' foreign language enjoyment (FLE) and its influential factors by examining the intersection of PP and GenAI technology. The two research questions are: (1) What are the levels of FLE experienced by Chinese EFL learners when using GenAI tools? (2) How do gender, English proficiency, and the usage duration of GenAI tools influence Chinese EFL learners' FLE in GenAI-assisted educational context?

Literature review

FLE and influential factors

The concept of FLE was firstly introduced by Dewaele and MacIntyre in 2014 [9], and later they conceptualized it as “complex emotion, capturing interacting dimensions of challenge and perceived ability that reflect the human drive for success in the face of difficult tasks.” [12]^p126 Similarly, Botes et al. defined FLE as a positive emotion experienced by students in the process of language learning, especially when they go beyond their limitations to finish difficult learning tasks and their psychological needs are fulfilled in classrooms [13].

Ever since FLE was introduced, earlier studies have primarily focused on the development of scales of FLE in different educational and cultural context [9, 14]. Later, other studies are mainly about FLE’s relation to other individual variables. For example, prior studies have found that FLE is positively related to students’ language proficiency or academic achievement [15, 16]. Moreover, FLE has positive relation with learners’ L2 willingness to communicate [17], grit [18], engagement [19], and motivation [20]. Meanwhile, the influential factors of FLE, such as learners’ personality [21], classroom environment and emotional intelligence [22], as well as learners’ linguistic competence [23] have received sustained empirical scrutiny.

Virtually, FLE is shaped by a variety of internal and external factors that have the potential to either enhance or diminish the overall FLE. These factors can be broadly classified into three main categories: social, psychological, and pedagogical influences. With regard to social factors, the social context of language learning, particularly peer influence, plays a significant role in enhancing enjoyment. Collaborative learning environments, which foster interaction among learners, can foster a sense of belonging and mutual support, thereby rendering the learning experience more enjoyable [24]. Moreover, the quality of relationships with both peers and educators is crucial in enhancing enjoyment [25]. Secondly, psychological factors, including motivation and self-confidence [26], autonomy [27], and engagement [28] also play an important role in influencing FLE. Additionally, the teaching methodologies and the characteristics of teachers play a pivotal role in shaping FLE. Teacher immediacy, both verbal and non-verbal behaviors that enhance teacher-student interactions, can significantly elevate students' levels of learning enjoyment [29]. Also, "teachers' compliment, encouragement, motivation, and acknowledgement of learners' accomplishments, is a prevalent pedagogical method that has been linked to EFL learning enjoyment" [25]p2027. Moreover, factors such as age, gender, language proficiency personality traits, and individual interests contribute to the overall FLE [24].

It is evident that the factors influencing are numerous and the relationships among them are intricate. Due to the controllability of gender and the fact that participants' English proficiency levels can typically be scientifically categorized, empirical studies focusing on the impact of these two factors on FLE are commonly frequently examined [30–33]. Nevertheless, these aforementioned empirical studies concerning the impact of gender and English proficiency on FLE have predominantly been conducted within traditional language teaching environments. In contrast, research in the context of technology-enhanced language teaching, particularly in the emerging landscape of AI-assisted language instruction, remains remarkably scarce [7]. In addition, regarding research on the impact of AI-tool usage duration on users' emotions, a prior study posited that prolonged exposure to AI tools can enhance learners' perceptions of task value and enjoyment [34]. However, a previous study argued that the frequency of AI tool usage does not necessarily correlate with increased learning enjoyment [35]. Therefore, given the scarcity of relevant research and the inconsistencies in existing conclusions, within the context of GenAI-assisted English learning, investigating the effects of gender, English proficiency, and duration of GenAI tool use on FLE holds substantial research significance.

PP and FLE

PP emphasizes fostering learners' overall development through the promotion of positive experiences, the cultivation of positive traits, and the creation of positive environments [36]. MacIntyre and Gregersen were the pioneers to apply PP in language instruction and argued that learners’ positive emotions should be utilized to facilitate language learning [37]. However, from 2012 to 2015, PP had remained under-researched due to the dominant role of the cognitive perspective and faced criticisms, and the flourishing of PP research in applied linguistics began in 2016 due to two seminal books on PP in second language acquisition edited by MacIntyre et al., Gabrys-Barker and Gałajda [38]. Consequently, with PP movement in SLA, a number of studies have examined the positive emotions experienced by language learners [39, 40].

It is acknowledged that the positive emotions most frequently experienced by second language learners include enjoyment, gratitude, calmness, interest, anticipation, pride, inspiration, awe, and affection [41]. Among these positive emotions experienced by language learners, enjoyment is the most commonly researched in different contexts [9, 10]. Empirical evidence suggests that FLE plays a critical role in shaping the overall pleasant learning experience of language learners and improving learners’ linguistic competence, e.g., communicative skills [42, 43]. Nevertheless, by evaluating 36 empirical studies on FLE within the framework of PP, a review identified a salient research gap: the near-absence of investigations into FLE in AI-mediated language-learning environments [38]. Building on this foundation, the present study explores the intersection of PP and GenAI technology through a 12- week intervention to examine the impact of GenAI tools on Chinese EFL learners' FLE.

FLE in technology-assisted educational context from a PP perspective

In traditional classrooms, teachers can adopt various strategies to alleviate students' negative learning emotions, cultivate their positive emotions and traits, and create a classroom environment that positively stimulates emotional engagement [11, 44, 45]. Beyond traditional classrooms, PP also holds great potential in technology-assisted language education. In technology-driven contexts, emotions have also gained significant attention as a factor contributing to successful learning. It is found that learners experienced higher levels of positive emotions than negative emotions during online learning activities [46]. Based on a meta-analysis of 186 studies on emotions in technology-mediated learning environments, another study (2020) revealed a positive relationship between enjoyment and appraisals, achievement, and cognitive support [47]. Also, participants' enjoyment and excitement were positively correlated with their engagement in MOOCs [48]. Additionally, with the advancement of AI technology, language educators and researchers have shown particular interest in its role in shaping and enhancing language learning experiences [49]. For instance, a study highlighted the positive role of AI-speaking assistants in enhancing EFL students' FLE and willingness to communicate while mitigating foreign language anxiety [7]. Evidently, within the framework of PP, emotional factors are closely linked to the goals of improving students' overall well-being and fostering personal growth.

Clearly, although GenAI-driven online language instruction still falls within the broader category of online language-learning environments, it departs sharply from traditional online learning models. By dynamically generating fully customized materials and delivering precisely tailored support, it ushers in a new level of personalization for language learners. Therefore, understanding the impact of GenAI applications on the emotional dimensions of learners is therefore crucial. However, existing research on the impact of AI on language learners predominantly focuses on their linguistic abilities [2–4], while studies on its influence on learners' emotional experiences remain scarce. Most studies have concentrated on the effect of AI use in reducing learning anxiety [7, 50], with relatively little attention paid to emotions such as enjoyment, boredom, or complex emotions. Furthermore, despite the increasing attention given to PP, its application in Chinese foreign language contexts remains in its infancy [51]. There is a notable scarcity of intervention studies specifically targeting the effects of GenAI usage on language learners' emotions within the framework of PP.

Therefore, the present study will employ a quasi-experimental design to investigate the FLE level and three influencing factors (gender, English proficiency, and duration of GenAI use) among university students in China during the process of using GenAI to assist their English learning. The present study endeavors to supply empirical substantiation for the potential of PP in the context of GenAI-assisted foreign language education. Additionally, it aims to provide pragmatic recommendations for teachers and educators regarding the integration of GenAI technologies into language instruction.

Methodology

Research context and participants

The study was conducted at a typical public university in southeastern China. All first-year and second-year non-English majors were required to enroll in College English Courses I-IV, which consisted of two 45-min English reading and writing sessions per week, along with two additional 45-min English listening and speaking sessions every other week. These courses spanned approximately 16 weeks across four semesters and aimed to enhance students' comprehensive English skills, including listening, speaking, reading, writing, and translation, through a variety of activities.

The present experiment was conducted within the authentic teaching environment of the College English Course III. Prior to the start of the semester, the instructor carefully planned and designed a series of tasks that required students to utilize GenAI tools to assist their English learning throughout the semester. As it was conducted within a real classroom setting rather than in a purely experimental format, the research design is quasi-educational.

The participants in this study were 122 s-year students enrolled in College English Course III instructed by the first author during their third semester. They were selected through convenience sampling from three classes majored in financial management, logistics, and hotel management. The participants' scores from the most recent national standardized English proficiency test they took, namely China's National College English Test, Band 4 (hereinafter referred to as CET 4), were collected. Based on these scores, the participants were categorized into distinct groups according to their varying levels of English proficiency. Results show the participants' average score on their latest CET 4 was 408.33 out of 710, roughly equivalent to a TOEFL iBT score of 70 (Miami University Admission and Aid, n.d.), situating them within the CEFR B1 level (ETS, n.d.). However, six students failed to submit the questionnaire (see Appendix I) on time, and another seven students selected "No" in response to question 8 of the questionnaire (whether they experienced a certain degree of enjoyment while using GenAI to assist English learning). Additionally, four respondents provided uniform scores across all items on the 5-point Likert scale, indicating a lack of meaningful engagement with the survey items. Meanwhile, seven participants failed to provide their CET 4 grades. All these invalid data were subsequently removed from the analysis. As a result, the final experimental group consisted of 98 students (54 females and 44 males), with an average age of 20.17 years (SD = 0.75). Figure 1 presents the final sample size in the CONSORT-style flow diagram.

Fig. 1.

The CONSORT-style flow diagram of the sample size

Intervention device-GenAI tools

The intervention tools employed in this study consisted of widely available GenAI tools in Mainland China, such as Doubao, Wenxin Yiyan, Kimi, Xunfei Xinghuo, and Tengxun Zhiying, etc. These tools were utilized to support participants in improving their English skills both inside and outside the classroom by completing a variety of learning tasks assigned by the instructor or initiated by the learners themselves.

To further enrich the learning experience, participants were granted autonomy to select their preferred GenAI tools after being encouraged to explore several options based on convenience and availability. Additionally, learners were motivated to customize the features of the GenAI tools to align with their individual needs and preferences. For instance, when using Doubao for practicing English speaking, learners could choose the AI's gender and select between British or American English accents. This flexibility enabled learners to personalize their interactions with GenAI tools, fostering a more tailored and authentic language learning experience.

In a deliberate and systematic endeavor to integrate the principles of PP, we carefully designed the intervention to foster a positive and empowering language-learning environment. The GenAI tools served as a key component in this intervention, providing consistently supportive and encouraging feedback during interactions with learners. These positive reinforcements were strategically incorporated to align with the core tenets of PP, which emphasize the cultivation of positive emotions and the enhancement of psychological well-being.

Instruments

Foreign Language Enjoyment Scale (FLES)

The questionnaire (See Appendix I) is divided into two parts. Part I aims to collect profile information, including gender, age, and use duration of GenAI tools, etc. Part II is a revised version of Chinese Version of the Foreign Language Enjoyment Scale (CFLES) [14], consisting of 11 items of a five-point Likert scale (ranging from "completely disagree"1 to "completely agree"5).

The CFLES is a 5-point Likert scale adapted by Li et al. [14] from the Foreign Language Enjoyment Scale (the FLES) originally developed by Dewaele & MacIntyre [9]. With 11 items, the CFLES is tailored for Chinese EFL learners and comprises three sub-scales: FLE-Private (items 1, 2, 3, 4, 6), FLE-Teacher (items 7, 8, 9), and FLE-Atmosphere (items 5, 10, 11). For the study purpose, original items were contextualized to reflect GenAI-assisted English learning context. For instance, the original item 3, "I learned a lot of interesting things,"was modified to, "While using GenAI tools to assist my English learning, I learned a lot of interesting things."In this study, the modified scale demonstrated excellent overall reliability (Cronbach's α = 0 0.91) and structural validity (KMO = 0.86, p = 0.00 < 0.05).

Semi-structured interview

The interview was conducted in Chinese, guided by an interview outline (See Appendix II) comprising 15 open-ended questions.1 The questions were structured around eight primary thematic areas: (1) Background information of the participants (questions 1–4); (2) Overall level of experienced FLE (questions 5–6); (3) Influence of English proficiency on FLE (questions 7–8); (4) Impact of gender on FLE (questions 9–10); (5) Effect of GenAI usage duration on FLE (questions 11–12); (6) Other influential factors (question 13); (7) The most enjoyable scenario while using GenAI to assist English learning (question 14); and (8) Suggestions for improving GenAI-mediated FLE (question 15). As depicted in Fig. 1 above, to ensure the balanced distribution of factors including gender, English proficiency, and duration of GenAI tool usage, 15 out of 98 survey respondents were invited for interviews, with 12 accepted and subsequently completing the interview voluntarily. After reviewing the recordings, we retained the ten most informative transcripts, excluding two brief interviews with female participants to maintain gender parity. The detailed profiles of the interview participants are presented in Table 1.

Table 1.

The profiles of the interviewees

Interviewees	Age	Gender	GenAI tools used	CET 4 passed or not	Use duration of Gen AI/single session
S1	21	female	Wenxin Yiyan, Xunfei Xinghuo, Zhipu Qingyan, Kimi	Yes	6–12 months/30-60min
S2	21	female	Wenxin Yiyan, Xunfei Xinghuo	No	3–6 months/20 min
S3	22	female	Wenxin Yiyan, Xunfei Xinghuo, Doubao, Kimi	Yes	6–12 months/45 min
S4	20	female	Kimi, Doubao, Tengxun Zhiying	No	3–6 months/10 min
S5	20	female	Wenxin Yiyan, Kimi, Suno	Yes	1–3 months/30 min
S6	20	male	Wenxin Yiyan, Kimi, Tongyi Qianwen	No	6–12 months/30–45 min
S7	21	male	Wenxin Yiyan, Xunfei Xinghuo, Zhipu Qingyan, Kimi	No	6–12 months/20–30 min
S8	21	male	Wenxin Yiyan, Xunfei Xinghuo, Doubao, Suno	Yes	3–6 months/10 min
S9	20	male	Wenxin Yiyan, Doubao, Tengxun Zhiying	No	3–6 months/over 30 min
S10	20	male	Wenxin Yiyan, Xunfei Xinghuo, Kimi	Yes	3–6 months/20–30 min

S1 refers to the first interviewee, with the rest following a similar pattern

As is presented in Table 1, among the 10 interviewees, there was an equal gender distribution, with five males and five females. The sample was also evenly divided into two groups based on English proficiency, with five individuals who passed CET 4 and another five who failed. The average age of the interviewees was 20.6 years. The GenAI tools employed by the 10 participants to aid in their English learning were highly diverse. Each individual utilized between two and five distinct types of GenAI tools. These tools include the latest mainstream GenAI tools available in Chinese mainland (eg., Kimi,Wenxin Yiyan, Zhipu Qingyan, Doubao) as well as an internationally recognized GenAI product (Suno). In terms of total usage duration of GenAI tools, five participants had used them for a duration of 3–6 months, four for over 6 months, and one for less than 3 months. Regarding the usage duration of GenAI-assisted English learning per session, five participants engaged in it for more than 30 min, four for 20–30 min, and one for approximately 10 min. Overall, the 10 participants exhibit a high degree of diversity and representativeness across multiple dimensions, including gender, English proficiency, age, types of GenAI tools, their experience with GenAI tools, and duration of single-use sessions.

Procedures and data collection

The study took place between late September 2024 and late-December 2024. All 122 participants from three classes were taught by the first author. As scheduled, the application of GenAI tools was an essential part of the semester's English teaching and learning activities. Firstly, to encourage active engagement with GenAI tools in supporting English learning, during Week 1, the teacher provided an introductory tutorial lecture on the feature of GenAI tools, their basic functions, ways to find suitable GenAI tools and design effective prompts. In addition, all participants were informed and written consent to participate was obtained from all of the participants in the study. From Week 2 to Week 3, students are strongly encouraged to utilize various GenAI tools as extensively as possible. This period is designed to enable students to become familiar with these GenAI tools and to explore the potential contexts in which English learning can be enhanced through these tools.

Afterwards, from Week 4 to Week 15, participants were required to complete intervention tasks aligned with the textbook content every other week, submitting their assignments via a campus-based teaching platform.2 For each biweekly intervention task, students were provided with a detailed record sheet outlining the steps for task completion and specific guidance on how to use GenAI tools effectively (e.g., the record sheet for task 1 is shown in Appendix III). Over the 12-week intervention period, students completed a total of six GenAI-assisted tasks outside the classroom. To reinforce the importance of these tasks and enhance motivation, students were informed that their performance in these tasks would account for 30% of the students' regular course grades for formative assessment, and exemplary assignments from 2–3 students were demonstrated in class every other week. This approach ensured that all participants took the GenAI-assisted English learning tasks seriously, and the integrity and reliability of the subsequent questionnaire survey and interview data would be guaranteed.

In Week 16, all participants were invited to complete the post-intervention survey through the same campus-based teaching platform mentioned earlier, thereby furnishing the quantitative data. Semi-structured interviews were then conducted in Weeks 17–18 to collect the qualitative data.3 The systematic procedure is shown in Fig. 2.

Fig. 2.

The experimental procedure of the study

Data analysis

The quantitative data of the survey were analyzed using SPSS 24.0. First, descriptive statistics were generated, and the mean scores of participants' FLE were calculated to assess their overall enjoyment levels, addressing the first research question. To answer the second research question, firstly, independent samples t-tests were subsequently performed to compare enjoyment levels across groups based on gender and English proficiency. A one-way ANOVA was subsequently performed to examine differences in enjoyment levels across groups defined by the duration of GenAI tool usage. In addition, the interview data were analyzed by two coders using thematic analysis, according to the six steps (1. familiarizing yourself with your data; 2. generating initial codes; 3. searching for themes; 4. reviewing themes; 5. defining and naming themes; 6. producing the report.) proposed by a prior study [52] to explore their perceptions of the effects of GenAI tools on FLE and its key influential factors in GenAI-driven context. The Krippendorf's Alpha coefficient was calculated to be 0.87, which indicates a satisfactory level of agreement between the two coders [53]. Thus, the results derived from the interviews served to corroborate and enrich the interpretation of certain findings obtained from the quantitative data of the questionnaire, specifically in relation to research questions one and two.

Results

Overall FLE level

To gain an overall understanding of the participants' experienced enjoyment while engaging in English learning assisted by GenAI, we first conducted a descriptive analysis of the scale data pertaining to enjoyment in the second section of the questionnaire. Table 2 presents the participants' overall self-assessment of FLE and their evaluations across three sub-dimensions.

Table 2.

Descriptive data of overall FLE and the three sub-dimensions

	Mean	Min	Max	SD
Overall FLE	3.90	2.73	5.00	.50
FLE-private	3.85	2.20	5.00	.52
FLE-teacher	4.09	2.67	5.00	.60
FLE-atmosphere	3.80	1.67	5.00	.63

N = 98

Overall, the participants reported a moderate level of FLE, with a mean score of 3.90 out of 5. However, notable differences were observed among the three sub-dimensions. Enjoyment related to personal factors (FLE-private) and environmental factors (FLE-atmosphere) was comparable, both scoring moderately, with mean scores of 3.85 and 3.80 respectively, whereas enjoyment linked to teacher-related factors (FLE-teacher) was marginally higher (M = 4.09).

Meanwhile, according to the interview data, all 10 interviewees claimed that they derived enjoyment from utilizing GenAI tools to facilitate their English learning process. For instance, S1 remarked, “Yes, using generative AI tools to assist with English learning brings me a sense of enjoyment”. S6 concurred, noting that “Yes, because it (the AI tool) provides me with excellent answers most of the time.” Specifically, seven interviewees (S2, S3, S5, S6, S7, S9, S10) indicated that they experienced a moderate level of enjoyment, whereas three interviewees (S1, S4, S8) reported a high level of enjoyment. This finding aligns with the survey results, in which participants rated their FLE at a moderate level.

Key influential factors

Gender

In this study, the 98 participants were first divided into two groups based on gender (male and female). Descriptive statistical analyses were conducted to examine the overall FLE and its three sub-dimensions for each group. Subsequently, an independent samples t-test was performed to compare the FLE levels between male (n = 44) and female (n = 54) participants to identify potential gender differences.

As shown in Table 3, the mean scores indicated that the male group's overall FLE and the three sub-dimensions were ranging from 3.86 to 4.08. Similarly, the female group's overall FLE and the three sub-dimensions were ranging from 3.75 to 4.10. The results indicate that both male and female participants derived a high degree of enjoyment from using GenAI tools in their English learning process, with moderate levels of enjoyment. Moreover, independent-samples t-tests revealed no significant gender differences in overall FLE (p = 0.69 > 0.05), the FLE-private dimension (p = 0.69 > 0.05), the FLE-teacher dimension (p = 0.81 > 0.05), or the FLE-atmosphere dimension (p = 0.39 > 0.05).

Table 3.

Gender differences in overall FLE and its three sub-dimensions

	Groups	Mean	SD	95% Confidence Interval of the Difference
				Lower	Upper	t	df	p (Sig.)
Overall FLE	male	3.93	.63	-.17	.26	.40	69.02	.69
	female	3.89	.39
FLE-private	male	3.88	.64	-.18	.27	.40	68.78	.69
	female	3.83	.40
FLE-teacher	male	4.08	.71	-.27	.22	-.24	96	.81
	female	4.10	.51
FLE-atmosphere	male	3.86	.68	-.15	.37	.86	96	.39
	female	3.75	.59

male (N = 44), female (N = 54)

However, according to the interview, most interviewees admitted that gender might exert a certain degree of influence on their FLE. And this gender difference largely derived from the differences between male and female students' habits of using GenAI tools to assist English learning. It is suggested "females may place greater emphasis on details and the accuracy of language use, and are more inclined to accept detailed explanations and examples provided by AI" (S1). In contrast, "Boys may be more inclined to use GenAI to assist the completion of their homework, and there is a high probability that they will not use AI to assist them in learning English except homework, whereas girls may be more inclined to use AI chat to help them improve their English” (S7). In addition, girls may pay more attention to "the emotional support of the generated content" (S8), or "AI's verbal emotions, including AI's voice and intonation" (S10), and boys may care more about "generating content that is logical and expresses rational thought" (S8).

Clearly, females tend to engage in meticulous language practice through GenAI, while males focus more on efficiency and task completion. Consequently, girls are more "cautious and diligent users”, and males are "more casual and adaptive users" (S1). This difference in preferences directly affects the enjoyment derived from using GenAI. In other words, females—who are typically more attuned to detail and accuracy—may become frustrated when GenAI feedback lacks precision. Conversely, males, who tend to prioritize efficiency and logical coherence, are likely to be dissatisfied if the tool's responses are slow or logically inconsistent. However, some interviewees suggested that the influence of gender on enjoyment is relatively minor and that the emotional experiences are more likely to be shaped by differences in "learning habits and objectives" (S6), "individual personality" (S7), and "varying levels of English proficiency" (S4, S5).

English proficiency

Additionally, participants were divided into two groups based on their scores in the CET 4 exam (the full score is 710): the passing group (Group One) and the non-passing group (Group Two). The results of the independent samples t-test revealed a significant difference (p = 0.000 < 0.05) in CET 4 national English examination scores between Group One (n = 43, mean = 499.53, SD = 27.05) and Group Two (n = 55, mean = 379.00, SD = 36.20). Descriptive statistical analyses were conducted to assess the overall FLE and its three sub-dimensions for both groups. An independent samples t-test was then performed to compare differences in FLE levels between participants with different English proficiency.

As shown in Table 4, the mean scores indicate that the first group's overall FLE and the three sub-dimensions were ranging from 3.72 to 4.05. Similarly, the second group's overall FLE and the three sub-dimensions were ranging from 3.87 to 4.12. This indicates that participants with varying levels of English proficiency both experienced a moderate degree of enjoyment. Moreover, independent-samples t-tests revealed no statistically significant differences between the two groups in terms of overall FLE (p = 0.30 > 0.05), the FLE-private dimension (p = 0.29 > 0.05), the FLE-teacher dimension (p = 0.59 > 0.05), or the FLE-atmosphere dimension (p = 0.26 > 0.05).

Table 4.

English proficiency differences in overall FLE and its three sub-dimensions

	Groups	Mean	SD	95% Confidence Interval of the Difference
				Lower	Upper	t	df	p (Sig.)
Overall FLE	1	3.84	.47	-.31	.10	−1.05	96	.30
	2	3.95	.53
FLE-private	1	3.79	.51	-.32	.10	−1.05	96	.29
	2	3.90	.52
FLE-teacher	1	4.05	.63	-.31	.18	-.54	96	.59
	2	4.12	.59
FLE-atmosphere	1	3.72	.58	-.40	.11	−1.13	96	.26
	2	3.87	.67

Group 1 (N = 43), Group 2 (N = 55)

Additionally, the interview results revealed that nine out of ten interviewees acknowledged that their relatively limited English proficiency had a negative impact on their FLE. Only one male participant with relatively high English proficiency contended that his English proficiency did not adversely affect his enjoyment of utilizing GenAI tools for English learning. Various aspects of English proficiency, such as insufficient vocabulary, a weak grammatical foundation, poor oral expression, and limited writing ability, were found to influence the sense of enjoyment to varying extents. For instance, "A lack of adequate vocabulary can hinder comprehension of the content generated by the AI, thereby disrupting the fluency of the learning process" (S1). Also, "A poor grasp of grammar can impede understanding of the complex sentence structure analyses provided by the AI, consequently diminishing the sense of enjoyment” (S1). "In writing, due to an inadequate foundation in grammar, it is challenging to determine whether the grammatical suggestions provided by AI are accurate. This uncertainty leads to a sense of unease, which in turn diminishes the enjoyment of using AI" (S3). Similarly, "The revision suggestions provided by AI for compositions may be difficult to fully assimilate due to limited English proficiency" (S7).With regard to oral English, S6 argued that"when I attempted to practice speaking with AI last week, my non-standard pronunciation led to recognition errors by the AI, which generated irrelevant responses and left me feeling frustrated" (S6).

Notably, there exists a dynamic relationship between English proficiency and the enjoyment derived from AI use. On one hand, initially, users may experience frustration due to inadequate abilities. Insufficiencies in vocabulary, grammar, speaking, and writing skills can significantly diminish the enjoyment derived from the use of AI. The primary manifestations of this issue include comprehension barriers, communication breakdowns, and low learning efficiency. Therefore, advancing one’s English proficiency is pivotal for leveraging GenAI more effectively, which in turn heightens both enjoyment and a sense of accomplishment. As S1 observed, “Once my English improved to the level where I could interact fluently with the AI, the experience became far more rewarding and enjoyable.”

Use duration of GenAI tools

Moreover, the participants were categorized into three groups based on their duration of GenAI tool usage: Group One (up to 3 months, n = 29), Group Two (between 3 and 6 months, n = 37), and Group Three (6 months and more, n = 32). Moreover, the results of question 7 in the first section of the questionnaire, which pertains to the background survey of the participants, revealed that the majority of participants used GenAI tools to assist with English learning on a daily or weekly basis. The specific distribution is shown in Fig. 3.

Fig. 3.

Participants' frequency of GenAI usage for assisting English learning

To explore the difference among the three groups, a one-way ANOVA was performed. As shown in Table 5, the duration of use is a key factor influencing participants' overall FLE as well as its three sub-dimensions. Generally, the difference is significant, but it is noteworthy that the differences between the groups varied. Concerning overall FLE levels, no significant difference was observed between Group One and Group Two (p = 0.10 > 0.05). However, statistically significant differences were found between Group One and Group Three (p = 0.00 < 0.05), as well as between Group Two and Group Three (p = 0.01 < 0.05).

Table 5.

Use duration of GenAI differences in overall FLE and its three sub-dimensions

	Group One	Group Two	Group Three	F	p(Sig.)	LSD
Overall FLE	3.66 ± 0.52	3.86 ± 0.37	4.17 ± 0.52	8.93	.00	c > b > a
FLE-private	3.67 ± 0.50	3.81 ± 0.37	4.07 ± 0.61	5.13	.01	c > b > a
FLE-teacher	3.77 ± 0.62	4.14 ± 0.61	4.33 ± 0.46	7.68	.00	c > b > a
FLE-atmosphere	3.56 ± 0.62	3.68 ± 0.51	4.16 ± 0.63	9.50	.00	c > b > a

Group One (N = 29), Group Two (N = 37), Group Three (N = 32); a = Group One, b = Group Two, c = Group Three

A similar pattern was observed in the sub-dimension of private factors, where no significant difference in FLE levels was found between Group One and Group Two (p = 0.25 > 0.05). However, significant differences were identified between Group One and Group Three (p = 0.00 < 0.05), as well as between Group Two and Group Three (p = 0.03 < 0.05). In contrast, within the teacher-related sub-dimension, no significant difference in FLE levels was observed between Group Two and Group Three (p = 0.15 > 0.05). However, significant differences were found between Group One and Group Two (p = 0.01 < 0.05), as well as between Group One and Group Three (p = 0.00 < 0.05). Additionally, the results for the atmosphere factors sub-dimension were similar to those observed for the private factors sub-dimension and overall FLE. Specifically, no significant difference in FLE levels was found between Group One and Group Two (p = 0.44 > 0.05). However, significant differences were observed between Group One and Group Three (p = 0.00 < 0.05), as well as between Group Two and Group Three (p = 0.00 < 0.05).

Equally importantly, the interview result echoed that the duration of GenAI usage has a significant impact on FLE. Most interviewees' emotional experience of using GenAI can be roughly divided into three stages, the initial fresh stage, the mid-adjustment stage and the late bottleneck or breakthrough stage. At the onset of utilization, "the novelty of the technology tends to evoke a strong sense of enjoyment" (S2). However, in the second phase, "after several months of use, the novelty tends to wane" (S4), worse still, "the sense of enjoyment may subsequently decline in the absence of noticeable progress in learning" (S1). In the third phase, the users may "get used to using the GenAI tools"(S2) or "feel AI tools are only tools" (S4), even "as use deepens, enjoyment may decline due to content repetition or learning bottlenecks" (S7). Luckily, in the phase of bottleneck, enjoyment can be reinvigorated by "continuously adjusting learning strategies and discovering new sources of enjoyment" (S1) or "adjusting learning objectives" (S7). Meanwhile, the interviewees reported that the duration of a single session of GenAI-assisted English learning typically over 30 min (see Table 1). During extended periods of use, issues such as waning attention and repetitive content may arise, which can negatively impact the sense of enjoyment. To address these challenges, they employed various strategies, including"taking appropriate breaks" (S10), "switching learning tasks" (S6), "changing GenAI tools" (S8) and "adjusting the learning objectives" (S7) or "regenerating content" (S5).

Additional influential factors

According to the interview, in addition to English proficiency, gender, and duration of GenAI usage, several other factors may influence the enjoyment when using GenAI for English learning. These factors include "the effectiveness of teacher guidance" (S1, S6), "the level of engagement in peer interaction" (S3, S6), "the appropriateness of task difficulty and amount" (S3, S9, S10), "the interest in learning content" (S6, S7), "the diversity of learning methods" (S6, S9), "the peers' perceptions of GenAI tools" (S5, S8), "the tranquility and comfort of the learning environment" (S2), and "the atmosphere of peers' using GenAI tools" (S4,S5, S7).

Moreover, the interviewees described several particularly enjoyable scenarios or events (see Table 6) they experienced while utilizing GenAI to assist English learning. The repetitive depiction of enjoyable scenes is achieved through the utilization of GenAI tools to refine their English writing drafts. (S1, S2, S7, S9). Additionally, an interviewee (S3) found enjoyment in interesting example sentences provided by GenAI tools that facilitated a deeper understanding of grammatical structures. Another notable event was when GenAI tools facilitated the completion of group projects, which subsequently earned them praise from the teacher (S6). It is noteworthy that interactive, interesting, and group-cooperative learning scenarios are the most enjoyable for students in the context of GenAI-assisted English learning. The learning tasks that elicit a greater sense of enjoyment exhibit multimodal characteristics. For instance, the English writing task in this study incorporates the text-to-speech function of GenAI tools. The picture-generation task involves the text-to-diagram modality. The English speaking tutoring and song generation task utilizes the text-to-audio function of GenAI tools. Meanwhile, the digital human generation task employs the text-to-video function.

Table 6.

The most enjoyable scenarios

Question

Key points of response from interviewees

Would you please describe a specific event or incident that you particularly enjoyed while using GenAI to assist your English learning?

collaborative projects and creativity (lyrics writing, musics composition and digital human news broadcast video generation): "AI generated very beautiful English songs aligned with the main idea of the text provided by our group…"(S4) interactive language learning (English speaking tutoring and cultural exploration): "GenAI acted as a very good English speaking tutor when we talked about heroes, and it encouraged me to speak more and gave some examples in China and other countries, …"(S7) most enjoyable learning tasks: writing-assisting tasks (S1, S2, S7, S9); translating-assisting tasks (S2,); grammar and sentence analysis task (S3); music generation task (S4); digital human broadcasting video generation task (S5); English debates-assisting task (S6); role-players and situational simulation dialogue (S7); using AI to recalls or elicits knowledge that one has learned (S8); picture-generation task (S10)

Last but not least, the interviewees offered several recommendations to enhance the sense of enjoyment when utilizing GenAI tools. In terms of the functionality of GenAI tools, the following two suggestions were made: Integrate personalized learning path recommendations (S1,S9) and enhance the capability for content personalization generation (S8); Optimize GenAI functions to provide more in-depth and reliable responses (S3), retain context from previous interactions with users (S4), and improve speech recognition capabilities for GenAI tools. Regarding the design of teaching tasks by educators, the interviewees proposed: (1) Design AI-based collaborative group learning tasks, such as role-playing, group discussions, and debates (S1, S3, S6, S7); (2) Create more interactive and engaging tasks in diverse formats (S4, S5, S7, S10); (3) Develop tasks that integrate the use of GenAI with textbook topics (S7).

Discussion

The present study seeks to explore the impact of GenAI tools on Chinese EFL learners' FLE, the key influential factors, and learners' perceptions of GenAI usage. The following sections discuss the findings in relation to the two research questions.

Overall FLE level

The first research question aims to examine the participants' overall levels of FLE. The findings, derived both from the questionnaire and the interview, reveal that the usage of GenAI tools had a positive impact on language learners' emotional engagement. This finding aligns with the results of a prior study which reported a significant increase in FLE among participants who utilized GenAI tools to support their English-speaking practice [7].

Notably, this study also found that participants' perceived FLE on the teacher factor dimension was significantly higher than that on both the personal and atmosphere factor dimensions. On the one hand, it suggests that the teacher's guidance to participants in this experiment, including the introduction of the GenAI tool's functions and the design of learning tasks incorporating the GenAI, were highly recognized by students. Consistent with the Control-value Theory within the framework of PP, the perceived controllability of the learning process or outcome by learners significantly influences their emotional experiences [54]. In the current study, participants were granted the flexibility to choose their favorite GenAI tools and interact with them at any appropriate time and place. This arrangement effectively enhanced their sense of autonomy and control over the learning process. On the other hand, it also confirms the finding that teacher enthusiasm can enhance students' foreign language learning enjoyment and thus promote the positive development of their academic mood, as found in a related study [55].

Additionally, while interviewees were describing the most interesting scenarios in their GenAI-assisted English learning, they mostly mentioned that they experienced the highest level of enjoyment when GenAI tools helped them with creative learning tasks (e.g., English lyrics writing, English song generation, and digital human news broadcasting, etc.), which echoes the notion that FLE was found to occur when an individual achieved something, pushed his or her limits, experienced something novel, challenging or made positive changes [56].

Key influential factors of FLE

The second research question delves into the three pivotal factors—gender, English proficiency, and the duration of utilizing GenAI—that impact the participants' FLE levels within the context of GenAI-assisted English learning.

Gender

In this study, both male and female participants reported a moderate level of FLE. Notably, the FLE level of male participants was marginally higher than that of the female group. Nevertheless, the difference between the two groups did not reach statistical significance. This finding stands in contrast to previous research [30–32], which consistently reported that female learners tend to exhibit higher levels of enjoyment but also experience a greater degree of anxiety in comparison to their male counterparts. This discrepancy may stem from the distinct learning contexts in which the studies were conducted. The current study was situated within a GenAI-mediated learning environment, whereas the previous studies were carried out in traditional language learning settings.

It is noteworthy that the interview results indicate that both male and female participants generally perceive gender as a potential factor influencing their enjoyment of using GenAI tools to support their English learning. This perception is attributed to differences in personality, study habits, and English proficiency levels. However, the questionnaire results reveal that the gender difference is not statistically significant. This seemingly contradictory result is actually understandable. On one hand, participants sometimes deviate from their own perceptions and the actual situation, on the other hand, the perceived gender differences among the participants did not achieve statistical significance when evaluated on a larger scale. Moreover, this seemingly paradoxical outcome actually highlights a significant nexus between gender and the adoption of technology. An earlier study posited that women, particularly in developing countries, are often situated on the disadvantaged side of the digital divide [57]. Furthermore, another study revealed that within the realm of higher education, women may have a preference for student—centered pedagogical methods, yet they tend to report a lower sense of competence in utilizing digital tools [58]. It is likely that the existence of the aforementioned gender-based digital divide is the reason why, in this study, male participants demonstrated a relatively higher level of understanding and acceptance of GenAI tools. Consequently, they exhibited a slightly higher level of enjoyment than females. However, this gender difference is not yet significant. Of course, the specific reasons for this gender difference still await further in-depth investigation.

English proficiency

Interestingly, questionnaire results indicated that although there was no significant difference between the two groups with different English proficiency, and the group with slightly lower English proficiency experienced a marginally higher level of FLE during the process of using GenAI tools to assist English learning. It is also intriguing to observe that nine out of ten interviewees reported that their limited English proficiency had a negative impact on their enjoyment of using GenAI tools for English learning.

This finding deviates from the conclusion of a prior study, which argues that there is a positive correlation between higher language proficiency and increased enjoyment [33]. The reason for this divergent conclusion may be attributed to the fact that although GenAI-assisted English learning is a form of online learning, this interactive and inquiry-based online learning approach may be more engaging than traditional online learning methods. In conventional online learning environments, learners with higher language proficiency may benefit more due to their stronger learning motivation and clearer learning objectives, thus experiencing more enjoyment. However, in the context of GenAI-assisted English learning, learners with lower English proficiency may more readily experience the enjoyment. This may be because with the help of GenAI tools, those students with lower English proficiency can finish learning tasks more easily and efficiently. This phenomenon was corroborated in the interviews with some of the participants. For instance, a student (S3) mentioned that

"my spoken English is not good, and I often mispronounce words, so I feel shy to speak English with others. However, when using AI tools, it can understand what I am talking about, which I find quite fascinating.”.

Noticeably, the interview data suggest that gender and English proficiency may influence FLE, yet the quantitative results show no significant effects. This discrepancy is common in mixed-methods research. Actually, the value of mixed-methods research lies in integration and cross-validation, rather than requiring consistent outcomes. Discrepancies in findings may themselves reveal deeper research questions [59]. This inconsistency may be attributed to three distinct levels: methodological issues, psychological and cognitive factors, and contextual and tool-related influences. First and foremost, methodologically, survey-based quantitative instruments typically measured “overall enjoyment” through one—or at most a handful—of broad constructs. A single-item indicator (e.g., Item 6. Learning English with GenAI tools is fun.) on a five-point Likert scale) lacks the granularity required to chart the nuanced, moment-to-moment affective changes that characterize authentic learning episodes. Conversely, semi-structured interviews allowed students to describe their emotions in specific contexts. A prototypical excerpt from a relatively high-proficiency male learner (S10) illustrates this affordance:

“the multimodal, interactive experience with AI enhances the enjoyment of using AI to support English learning, but I felt bored when AI responses became repetitive.”

This integration of positive and negative emotions might result in a "non-significant" aggregate score on the survey.

Secondly, according to Attribution Theory, individuals naturally seek causes for their success, failure, enjoyment, or anxiety [60]. Gender and English proficiency are readily available, stable, internal attribution labels that students can easily identify and articulate. Indeed, because factors influencing enjoyment were explicitly mentioned, participants may have been primed to attribute their affective responses to precisely those variables. During the interviews, learners across the proficiency spectrum—both high- and low-achieving—uniformly attributed dips in enjoyment to perceived linguistic deficits, most commonly citing “limited vocabulary” as the decisive constraint (S1, S7). In other words, regardless of their varying levels of English proficiency, interviewees expressed a common belief that their relatively lower English proficiency hinders their ability to fully comprehend GenAI-generated content or engage effectively in interactions, thereby diminishing their enjoyment of GenAI-assisted English learning. This shared sentiment likely explains why the questionnaire results reveal no significant differences in enjoyment levels across different proficiency groups. This phenomenon may be attributed to the fact that neither group of students possessed a sufficiently high level of English proficiency to mitigate these challenges. Moreover, it is important to note that, although the majority of interviewees acknowledged that gender can shape enjoyment in GenAI-assisted English learning, they emphasized that gender influences the sources or modalities of enjoyment rather than its overall intensity. This distinction is clearly captured in statements such as:

“Female students may place greater emphasis on oral practice; consequently, if AI tools provide accurate error correction and offer a diverse range of expressions, they are likely to find the learning experience more enjoyable. In contrast, male students may be more inclined to construct complex conceptual frameworks of English knowledge. When AI delivers in-depth grammatical analyses and academically enriching content, their learning experience tends to be more positive and engaging” (S3).

However, the survey did not include any specific items that directly addressed the influence of gender or English proficiency on students’ enjoyment of using GenAI tools to support English learning. As a result, students may not have been psychologically primed to attribute their enjoyment to these factors.

Thirdly, regarding contextual and tool-usage factors, students of different English proficiency levels and genders might use GenAI tools for different learning purposes. High-proficiency learners use AI for creative learning tasks, for instance, writing reflections on English news reports (S1) or using cinematic plot scenes to describe subordinate clauses (S3), drawing enjoyment from intellectual challenge; their less-proficient peers rely on the same tool for vocabulary learning (S4) and getting answers to exercises (S9), experiencing satisfaction chiefly through task completion. The antecedents and affective intensity of these two forms of enjoyment differ qualitatively, yet the survey treated them as interchangeable. Interviews effectively uncovered this variance in experience arising from "the same tool, different applications."

Therefore, the findings are not contradictory but rather form a more complete picture: At the group level, gender and English proficiency fail to emerge as significant predictors of the enjoyment students report when using GenAI tools for language learning—an encouraging outcome that allows educators to recommend GenAI tools to a diverse student body with the expectation that most can derive positive experiences, regardless of gender or proficiency. At the individual level, gender and English proficiency shape individuals’ unique usage experience profoundly. When offering guidance, educators must take these individual backgrounds into account to provide personalized support. Certainly, discrepancies between qualitative and quantitative findings constitute a critical agenda for future inquiry.

Usage duration of GenAI tools

Based on the questionnaire results, the duration of GenAI usage is the sole factor that significantly affects the variation in the participants' FLE levels. Meanwhile, drawing from the previously discussed interview results, the emotional experience of participants in GenAI-assisted English learning can be broadly categorized into three distinct phases: the initial novelty phase, the mid-term adjustment phase, and the late-stage bottleneck or breakthrough phase. When correlated with the questionnaire results, it is evident that the participants' enjoyment levels generally intensifies with prolonged use. In the early stages of engagement, participants exhibited a pronounced sense of novelty. Although some users experienced a temporary dip in enjoyment during the mid-term phase, the overall level of enjoyment remained high in both the second and third stages, following adjustments to learning methods or goals. Notably, a qualitative shift in the users' experience occurred after six months of consistent use of the AI tool.

A prior study noted that users who view ChatGPT positively and are satisfied with it are more likely to continue using the tool in the future [61]. In contrast, the present study offers a novel perspective, suggesting that extended use of GenAI can lead to a significant enhancement in FLE. This increase in FLE may, in turn, have a positive impact on users' satisfaction and their intention to continue utilizing the technology.

The significant impact of GenAI usage duration on the level of FLE among participants may be attributed to the increased variety of GenAI tools employed and the deeper understanding of their functionalities as usage duration extends. As shown in Table 1 of this study, participants who used GenAI for over six months had tried a significantly larger number of AI tools, typically ranging from three to five. Meanwhile, participants who used GenAI for longer periods engaged in a wider range of scenarios for English learning, thereby developing greater confidence in the effectiveness of GenAI—assisted learning, which in turn enhanced their FLE levels. As one interviewee (S10) stated,

"The longer I engage with GenAI tools, the more proficient I become in its application. I uncover additional features and capabilities that were previously unknown to me. The multimodal and interactive nature of the AI experience significantly enhances the pleasure derived from using AI tools to support my English learning journey."

Other influential factors

The interview results revealed that, in addition to the three primary factors under investigation in this study—gender, English proficiency, and usage duration of GenAI tools—several other factors also influence the participants' enjoyment of utilizing GenAI tools to assist with English learning. These additional factors can be categorized into internal and external factors. Internal factors include, for example, the students' interest in English learning and their learning strategies. External factors includes teacher guidance, peer interaction, and the environment in which the GenAI tools are used. These factors collectively impact the participants' FLE levels to varying extents.

This finding essentially echoes a prior research which found that FLE was positively linked to multiple learner-internal variables (self-perceived FL proficiency, relative standing among peers, attitudes towards the FL, the teacher, and FL-related culture) and teacher-related variables (the use of FL in class, enthusiasm, predictability, and friendliness) [29]. Moreover, another study argued that the multifaceted factors, including psychological, social, and pedagogical factors interact to shape learners' emotional responses and engagement levels [62]. In this sense, the current study also confirms the fact that personal factors, teacher factors, and environmental factors are three critical dimensions that impact the enjoyment of learning, regardless of whether it is in a traditional language teaching environment or within the context of GenAI-assisted English teaching.

Concerning suggestions for improving enjoyment, the interviewees' perceptions are also in line with the recommendations proposed by a previous study [56], focusing on collaborative learning, teachers' effective guidance, and teaching design optimization, emphasizing the need for tailored instructional strategies that prioritize enjoyment in the language learning experience. Leveraging GenAI tools can promote a personalized learning experience and offer a more supportive and effective learning context to substantially benefit learners [63].

Last but not least, with regard to participants' general attitudes toward GenAI tools, the finding of this study aligns with the results of several prior studies [64–66], which all highlight that students from diverse educational backgrounds express varying perceptions of GenAI tools. These studies emphasize the benefits of GenAI, including its potential to enhance language acquisition, improve instructional design, and foster a productive learning environment. However, they also underscore the challenges associated with its use, such as skepticism towards AI-generated feedback, concerns over excessive dependence on technology, and the potential erosion of critical thinking skills.

Conclusion and implications

In this study, we examined the impact of GenAI tools on Chinese EFL learners' FLE, and its key influential factors, informed by the principles of PP. The findings revealed that the application of GenAI tools resulted in moderate FLE levels. Among the three key factors examined, gender and English proficiency did not significantly influence FLE, whereas the duration of GenAI tool usage was identified as the primary determinant. Qualitative data from interviews further highlighted that GenAI-facilitated interactions fostered an enjoyable and supportive learning environment, consistent with the principles of PP. These findings suggest that GenAI tools can create positive learning experiences by promoting engagement and enjoyment.

Despite these valuable insights, the study has several limitations. Firstly, the final sample size of 98 participants is relatively small, and all were drawn from a single university. This restricts the generalizability of the findings. Future research should employ larger-scale surveys across diverse educational contexts to enhance the representativeness and generalizability of the results. Secondly, the current study focused primarily on FLE through questionnaires and interviews. Future investigations should adopt multi-perspective approaches to provide a comprehensive assessment of GenAI tools' impact on other emotions, such as anxiety and boredom, as well as on participants' linguistic competence. Thirdly, there are discrepancies between the interview results and the questionnaire findings. For instance, the majority of interviewees assert that their English proficiency and gender influence their enjoyment of using GenAI tools for English learning. However, the statistical analysis of the questionnaire data did not reveal any significant impact of these two factors. The underlying reasons for this inconsistency warrant further empirical exploration. Perhaps future research endeavors should encompass participants of both genders with a more extensive spectrum of English proficiency levels to provide a more comprehensive understanding of the factors influencing the use of GenAI tools in English learning. Fourthly, this study did not restrict the types of GenAI tools employed by the participants, nor did it undertake a comparative analysis of the enjoyment experienced by participants using different GenAI tools. Future research could explore the GenAI tools themselves as variables to provide a more nuanced understanding of their differential impacts. Fifthly, the tasks varied in nature, and the differing amounts of time students allocated to each task may have influenced participants’ perceived enjoyment, potentially affecting the study outcomes. Future research could delve deeper into examining the effects of learning task design and intervention dosage. Finally, although this study identified six months as a critical turning point for changes in FLE, longitudinal studies spanning one to two years are necessary to comprehensively examine the long-term emotional responses of learners using GenAI tools.

Notwithstanding these limitations, this study has significant implications for EFL education. Firstly, it highlights the potential of GenAI tools to shape positive emotions among foreign language learners. Secondly, given the relatively lower FLE among female students, instructors might consider assigning cooperative learning tasks that pair girls and boys to support more effective use of GenAI tools and optimize learning outcomes for female students. Thirdly, the finding that learners with lower English proficiency experienced higher FLE suggests that teachers can design appropriate GenAI-facilitated tasks to help these students transform their enjoyment into better learning outcomes. Fourthly, the positive impact of GenAI tool usage duration underscores the need for teachers to create more opportunities for students to engage with these tools both inside and outside the classroom, encouraging sustained use beyond six months. As learners engage with these GenAI tools, they can often develop a deeper appreciation for the learning process, foster increased motivation and enhance learning enjoyment [34]. Finally, to address concerns regarding over-dependence on GenAI tools and potential negative impacts on critical thinking, teachers should design more process-oriented tasks that require greater reflective thinking and sustained student engagement. Overall, this study underscores the potential of GenAI tools to enhance the emotional and educational experiences of EFL learners while highlighting areas for further research and practical application.

Abbreviations

GenAI

Generative Artificial Intelligence

EFL

English as a Foreign Language

FLE

foreign language enjoyment

PP

Positive Psychology

FLES

The Foreign Language Enjoyment Scale

CFLES

Chinese version of the Foreign Language Enjoyment Scale

Authors’ contributions

X.Y. conceptualized the study, designed the methodology, administered the project, and wrote the original draft. J.R. is contributed to the analysis of the interview data and revision of the manuscript. W.M. contributed to the conceptual development, supervised the project, and revised the manuscript through critical review and editing. All authors reviewed and approved the final manuscript.

Funding

This paper is funded by the Chongqing Municipal Education Commission Research Project of Humanities and Social Science in 2025 titled "A Study on the Innovative Teaching Pathways of College English Empowered by GenAI" (No. 25SKGH155); 2024 "University-Level Joint Science and Education Project (Humanities) General Project of Minjiang University: Generative AI Technology Aiding the Practice and Research of College English Teaching" (No. MJKJ24024); Minjiang University 2024 Education and Teaching Research and Reform Project: "Digital Human Teachers and ChatGPT Co-assisting Digital Transformation of the English Language Curriculum Research and Practice) (No. MJUJG2024A011).

Data availability

The survey and interview data are available on Figshare at DOI: 10.6084/m9.figshare. 30836675.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the ethical standards of the institutional and/or national research committee and adhered to the principles outlined in the 1964 Helsinki Declaration and its subsequent amendments or comparable ethical standards. Approval for this study was granted by the Specialized Committee for Scientific Ethics and Academic Integrity of the Academic Committee, Sichuan International Studies University (Approval No. 202400005). Prior to participation, informed written consent was obtained from every participant. All participants first received detailed information about the study’s purpose, procedures, potential risks and benefits, confidentiality safeguards, and their right to withdraw at any time without penalty.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.