Previous researchers have conducted some studies on Chinese university students’ acceptance and use of MELR. In a study that examined Chinese undergraduates’ perceptions of mobile English learning, Zou and Yan (2014) found that 78% of the students had a positive attitude toward using MELR. In addition, Zhang and Pérez-Paredes (2019) reported that passing exams and improving test scores were the most important reasons for students to use MELR. This is in line with Xue’s (2014) finding that using mobile devices in English learning was helpful to improve students’ test scores. As for the most frequently used MELR, vocabulary and translation applications were far more popular than other listening, speaking, reading, and writing resources (Lai et al., 2022). This is consistent with Steel’s (2012) finding that vocabulary acquisition was the most popular learning activity among Chinese language learners when it comes to MELR. In order to explore the effectiveness of MELR on English language learning, Yin (2013) suggested that Wechat – a social media platform, was an effective way to prepare students for CET-4 and CET-6 test through enhancing vocabulary, grammar, etc. Ruan and Ma (2014) also demonstrated that a designed mobile APP could facilitate students’ grammar learning.

Numerous research studies have been carried out to investigate the factors that influenced students’ self-directed use of MELR. García Botero et al. (2018) extended the UTAUT model to explore the factors that affected students’ behavioral intentions toward using mobile devices for English learning. Their findings indicated that social influence, performance expectancy, and facilitating conditions impacted students’ attitudes to mobile English learning, and behavioral intentions had effects on actual use of mobile English learning. Similarly, Hoi (2020) surveyed higher education learners in Vietnam to understand their adoption of mobile-assisted language learning (MALL). They found that attitude and performance expectancy had significant effects on learners’ behavioral intentions toward using mobile-assisted language learning, while facilitating conditions exerted no direct influence on learners’ MALL acceptance. In terms of English vocabulary learning, Guo and Li (2022) suggested that performance expectancy, effort expectancy, price value, facilitation condition, and habit had a significant positive impact on the use of English vocabulary APP learning by college students. Using a different model (TAM), Kim and Lee (2016) investigated Korean students’ adoption of mobile-assisted language learning and potential factors affecting their MALL use. Results indicated that perceived ease of use, perceived usefulness, perceived enjoyment, and content reliability significantly predicted students’ acceptance of mobile English learning. However, few studies have been conducted to explore factors influencing Chinese students’ use of mobile devices, especially in higher education context. To fill in this gap, this study aims to examine the factors that affect Chinese university students’ self-directed use of mobile English learning resources applying the UTAUT model.

The Unified Theory of Acceptance and Use of Technology (UTAUT) was developed by Venkatesh et al. (2003), which is also the most recent model to explain the users’ technology acceptance process. The UTAUT model is applicable in studies exploring technology acceptance among college students (Kumar and Bervell, 2019), which has also been one of the most popular models for mobile learning (Venkataraman and Ramasamy, 2018). The UTAUT has four core constructs, including performance expectancy, effort expectancy, social influence, and facilitating conditions. Their definitions are summarized as follows (Venkatesh et al., 2003):

Performance Expectancy (PE): This refers to “the degree to which an individual believes that using the system will help him or her to attain gains in job performance.”

As shown in the UTAUT model (Figure 1), performance expectancy, effort expectancy, and social influences have an impact on behavioral intention while facilitating conditions have an impact on use behavior. In addition, gender, age, experience, and voluntariness have a moderating effect on behavioral intention.

The objective of this study is to investigate the factors that affect behavioral intention toward adopting MELR among Chinese undergraduate EFL learners. In this model, Mobile English learning resources (MELR) refer to mobile applications, English podcasts, online courses, e-books, and websites delivered via mobile devices. Based on the UTAUT, the author proposed a research model (Figure 2) and made some modifications. First, the proposed model has added perceived playfulness (PP) as an independent construct. Perceived playfulness has been defined as “level of a user’s concentration, curiosity and enjoyment in the interaction with the technology environment” (Moon and Kim, 2001), which has been found to predict intention to adopt mobile learning (Wang et al., 2009). Second, this study only investigated the main hypotheses while dropping the effects of the moderators. Third, this study eliminated the construct of use behavior and only measured behavioral intention. Thus, this study has formulated and tested the following five hypotheses:

This study consisted of 206 Chinese undergraduate students with English education majors at Yangzhou University in Mainland China. Specifically, 21 participants (10.2%) were male, and 185 (89.8%) were female. Participants were distributed almost equally across four different levels. 46 participants (22.3%) were Year one; 58 (28.2%) were Year two; 52 (25.2%) were Year three; 50 (24.3%) were Year four. This study used online survey called Sojump to collect data. An anonymous link was administered via social media platforms such as QQ and WeChat. A total of 247 questionnaires were returned and 206 were considered valid for further analysis.

This study used a quantitative questionnaire to collect data, which consisted of two parts. The first part collected demographic information, including gender, age, and grade. The second part involved 18 UTAUT-based items to measure six constructs, namely performance expectancy, effort expectancy, social influence, facilitating conditions, perceived playfulness, and behavioral intention. All items were adapted from previous studies (Wang et al., 2009; Chung et al., 2015; Fagan, 2019; Hoi, 2020; Nikolopoulou et al., 2020). The items were scored on a five-point Likert scale, ranging from strongly disagree to strongly agree. The items and their sources are listed in the Appendix. The questionnaire was translated from English to Chinese so that participants can better understand the items.

In this study, Structural Equation Modeling (SEM) with Amos 26 tool was used to analyze the data. This study followed the two-step approach recommended by Anderson and Gerbing (1988).

The first step was to assess the validity and reliability of the measurement model. In this step, factor loadings, composite reliability (CR), average variance extracted (AVE), Cronbach’s alpha, and inter-construct correlations were calculated to describe the validity and reliability of each construct (Fornell and Larcker, 1981).

The second step was to analyze the structural model to test the research hypotheses. To achieve a good model fit, Chi-square to its degree of freedom (CMIN/df) should be lower than 3 (Schumacker and Lomax, 2010). Goodness-of-fit index (GFI) and adjusted goodness-of-fit index (AGFI) should be greater than 0.8 (Forza and Filippini, 1998). Comparative fit index (CFI) and Tucker-Lewis index (TLI) should be above 0.9 (Kline, 2005). Root mean square error of approximation (RMSEA) and standardized root mean square residual (SRMR) should be lower than 0.08 (Steiger, 2007).

Confirmatory factor analysis (CFA) was conducted to examine the convergent and discriminant validity of the measurement model. Convergent validity was measured by factor loadings, composite reliability (CR), and average variance extracted (AVE). The recommended values for factor loadings, CR, and AVE were 0.6, 0.7, and 0.5, respectively (Hair, 2006; Awang, 2015). As shown in Table 1, all factor loadings were above 0.6. All CRs were higher than 0.7 and all AVEs were above 0.5, indicating that all constructs had appropriate convergent validity. In addition, all Cronbach’s Alpha values were larger than 0.70, indicating that all constructs had acceptable internal consistency.

Discriminant validity would be regarded as appropriate if the square root of the AVE of each construct was greater than its correlation coefficients with other constructs (Gefen et al., 2000). As presented in Table 2, for each construct, the square root of AVE was greater than the correlation coefficients with other constructs, demonstrating good discriminant validity.

The second step was to test the structural model. The model fit was examined by several key goodness-of-fit indices, which were summarized in Table 3. Based on the results, the research model showed a good fit to the data.

As shown in Figure 3, the model explained 76% of the variance in behavioral intention, which was higher than the threshold of 70% recommended by Venkatesh et al. (2003). Table 4 also listed the path coefficients and their significance level for each hypothesis. Among all the five hypotheses, four hypotheses were supported. No significant relationship was found between social influence and behavioral intention (β = −0.021, p > 0.05). Thus, Hypothesis 3 was not supported. In addition, facilitating conditions (β = 0.383, p < 0.001) exerted the greatest influence on behavioral intention, followed by effort expectancy (β = 0.253, p < 0.01), perceived playfulness (β = 0.214, p < 0.01), and performance expectancy (β = 0.205, p < 0.05). Therefore, Hypothesis 1, Hypothesis 2, Hypothesis 4, and Hypothesis 5 were supported.

Facilitating conditions were the most powerful determinant of behavioral intention to adopt MELR. This finding was in accordance with the results of previous studies on mobile English learning (Mekhzoumi et al., 2018; Ahmed et al., 2021; Guo and Li, 2022) and other studies on mobile learning in general (Al-Adwan et al., 2018; Shameem and Sanjeetha, 2021), which consolidates the importance of this construct in the UTAUT model. When students perceive that necessary resources and technical support are available to them, they will have stronger intentions to use MELR for English learning. However, this finding was contrary to Hoi, 2020 study in Vietnam, where facilitating conditions were shown to have no direct effect on the adoption of mobile assisted language learning. One possible cause could be that students involved in this study might have more abundant MELR resources and better technical support, such as high-speed internet access. This would encourage Chinese undergraduate students to adopt mobile devices for English language learning.

Effort expectancy was the second most important predictor of behavioral intention toward using MELR. This is consistent with other findings in mobile English learning context (Mekhzoumi et al., 2018; Guo and Li, 2022) and in mobile learning in general (Nassuora, 2013; Fagan, 2019). This finding implies that students who perceive MELR easy to use will have more positive intentions to adopt MELR for English learning. This result was also demonstrated by Kim and Lee, 2016 research. Using a different model (TAM), they found that effort expectancy was an important factor that affected the usage of mobile assisted language learning among Korean students. Therefore, when developing mobile English learning applications, designers should take this factor into consideration, making applications as user-friendly as possible.

Perceived playfulness was also a significant indicator of MELR adoption. This result aligned with Doan’s (2018) research on mobile English learning and other studies on mobile learning in general (Jawad and Hassan, 2015; Karimi, 2016). As expected, when MELR are perceived as pleasurable and enjoyable, students will be more likely to intend to use mobile devices for English learning. A potential reason for this result is that mobile English learning in China mainly occurs in informal settings outside class (Chen, 2013). Consequently, students will expect mobile English learning platforms to be interesting to use. Thus, these platforms should strive to create enjoy-and-learn environment so as to engage students.

Finally, performance expectancy had a positive influence on students’ behavioral intention toward adopting MELR. This result coincided with other studies on the effect of performance expectancy on behavioral intention (Fagan, 2019; Nikolopoulou et al., 2020). Some studies also suggested that performance expectancy was the most important predictor of students’ intention to adopt mobile English learning (Mekhzoumi et al., 2018; Hoi, 2020). However, this study found that performance expectancy exerted the least influence on behavioral intention compared to other constructs. There are two possible explanations for such difference. First, Chinese students mainly regard MELR as supplementary English language learning material in that they adhere to traditional textbooks (Zhang and Pérez-Paredes, 2019). This will reduce their expectation of improving academic performance through MELR. Another reason involves the limitations of MELR. Small screen size, typing difficulties, and poor application design further undermine students’ desire to improve language performance using MELR.

Notably, social influence did not have a significant effect on behavioral intention to adopt MELR. This is in line with the findings of Arain et al. (2019), Fagan (2019), and Abbad (2021), and contrary to the findings of Ameri et al. (2020) and Ahmed et al. (2021). A possible explanation is that students examined in this study belong to digital natives who grow up in digital environment. When it comes to MELR, they are more comfortable with self-directed learning and thus are less influenced by instructors or peers.

This study has some limitations. First, this study had a relatively limited sample size with only 206 students at Yangzhou University. Thus, future research can involve more students and be conducted in other universities in order to generalize the results. Second, it only analyzed the core constructs of the UTAUT model and did not include the effects of potential moderators (age, gender, experience, and voluntariness) on the main relationships. Future studies can include these moderators to have a better understanding of students’ behavior intention. Finally, this study only investigated behavioral intention with a self-reported measure, which may change over time due to new knowledge and experience. In order to gain more accurate results, it is helpful to use a longitudinal study to measure students’ actual use of MELR.