Introduction

Smart mobile learning (SML), which integrates artificial intelligence (AI) and mobile learning, provides users with a personalised, smart, and mobile learning environment. SML enhances communication and collaboration and enables the delivery of personalised learning experiences (Yu et al., 2019). With the advancement of smart technologies, smart electronic devices have become widely accepted for learning purposes. This online learning, facilitated by cutting-edge AI technology, signifies the dawn of smart educational approaches (Lyapina et al., 2019). Furthermore, SML has made distance mobile learning feasible (e.g., Basoglu and Akdemir, 2010; Nah et al., 2008), with the goal of offering personalised learning experiences anywhere and anytime (Bajaj and Sharma, 2018; Yu et al., 2019).

SML incorporates a variety of technologies to cater to diverse user needs and facilitate the learning process (Abdel-Basset et al., 2018). SML applications such as Liulishuo employ advanced technologies such as speech recognition, natural language processing, machine learning, and recommendation systems to optimise mobile learning experiences. Similar SML applications include Duolingo, Rosetta Stone, and HelloTalk, among others (De la Vall and Araya, 2023; Karakaya and Bozkurt, 2022; Rukiati et al., 2023).

Liulishuo incorporates AI through various techniques and algorithms that facilitate intelligent features and functionalities. First, Liulishuo utilises natural language processing (NLP) algorithms to analyse and comprehend learners’ spoken input. Second, Liulishuo harnesses AI-based speech recognition technology to evaluate learners’ pronunciation and fluency. Third, Liulishuo uses machine learning algorithms to tailor learning to cater to the specific needs and development of each individual. Last, Liulishuo leverages recommendation systems to suggest personalised content and exercises based on each learner’s performance and preferences (Liulishuo.com, 2023). Consequently, the most notable difference between Liulishuo and a system that does not employ AI is that of intelligence level. The AI capabilities of Liulishuo enable it to function as an intelligent tutor, providing guidance, explanations, and suggestions that are tailored to each learner’s specific needs. It can pinpoint areas in which a learner may encounter difficulties and provide targeted exercises or additional resources to address these challenges (Bai and Stede, 2022).

NLP technology empowers the system to provide precise feedback, correct pronunciation, and offer personalised language learning suggestions (e.g., Evanini and Zechner, 2020). AI-based speech recognition technology contrasts the learner’s speech patterns with the accents of native speakers and provides real-time feedback to enhance pronunciation and intonation (e.g., Gong et al., 2022). Machine learning algorithms monitor each learner’s personalised performance, identify strengths and weaknesses, and adjust the content and exercises accordingly. This ensures that learners are consistently challenged at an appropriate level, which promotes progress and engagement (e.g., Pedro et al., 2019). Finally, recommendation systems delve deeper into learners’ interaction patterns, engagement levels, and learning pace, providing an advanced level of personalisation that not only enhances the learning experience but also optimises the learners’ time (Chen et al., 2021). As a result, SML applications such as Liulishuo have evolved into smart learning solutions.

The advancement of AI technologies has enabled SML applications such as Liulishuo to gain widespread popularity in the learner market (Kurni et al., 2023; Yu, 2022). Supported by advanced intelligent features and functionalities, Liulishuo is progressively reaching maturity. Currently, more than 216.5 million users have registered with Liulishuo to learn English in a personalised way (Liulishuo.com, 2023). However, faced with competitive pressure from many other English learning apps, Liulishuo must retain its current users. Researchers have indicated that it is important to retain users by studying their continuance intention (Bhattacherjee, 2001). At the individual level of users, the continuous usage of SML is also crucial to improving their English level.

The uses and gratifications (U&G) theory (Katz et al., 1973) serves as a valuable tool for examining sustained user behaviour towards SML. This theory has been effectively applied to diverse media based on computer-mediated communication and has been adapted for research on various emerging platforms (e.g., Khan, 2017; Liu et al., 2016). It has also been applied to mobile learning based on mobile devices (Aburub and Alnawas, 2019; Chang et al., 2021; Shukla, 2021). Nevertheless, current U&G research on mobile learning primarily focuses on the categorical level of gratification and fails to introduce specific constructs that encapsulate this type of gratification. Moreover, there is a noticeable gap in the literature concerning SML. As the future direction of mobile learning, SML, infused with AI technology, should have distinct constructs that reflect user gratification with this intelligent mobile learning modality. Unfortunately, new gratification for the intelligent dimension of SML has yet to be identified.

Consequently, this study focuses on the predictive influence of specific constructs under the gratification type on continuance intention towards SML. Additionally, this research endeavours to propose new gratification for the intelligent dimension of SML. Ultimately, this study seeks to empirically uncover the hierarchy of importance among various gratifications, each with specific constructs, that impact continuance intention towards SML, and partial least-squares structural equation modelling (PLS-SEM) is employed as the analytical method.

This study is expected to offer the following potential contributions. First, this study will demonstrate that various factors compose the gratification associated with SML usage and reveal the hierarchy of these factors in predicting sustained usage. Second, by introducing unique new gratifications for the intelligent dimension of SML, this study will augment U&G theory, making it more relevant and applicable to research in the realm of smart communication technologies. Third, U&G theory can provide a complete understanding of users’ post-acceptance behaviour by explaining different types of learner gratification after usage. This study will clearly relate U&G theory to the post-acceptance research of the information system (IS) represented by SML. As a result, this study will promote the generalisability of the theory in the context of SML and therefore is an indispensable step in developing U&G theory (Alvesson and Kärreman, 2007; Johns, 2006). This study can thus help explain the causes of SML’s continuous usage from the perspective of U&G and help individual users understand and improve their behaviour in their post-acceptance stage, which is beneficial for users to better manage their learning.

Theoretical framework

Uses and gratifications (U&G) theory

While the initial decision to embrace an SML system is a crucial starting point, the capacity of the SML system to maintain users over an extended period hinges on its continued usage in the post-acceptance phase. Accordingly, U&G theory is employed in this study to assess the gratifications of users in their post-acceptance engagement with SML systems.

The research on uses and gratifications originated in the 1930s. After decades of development by many researchers, Katz et al. (1973) proposed U&G theory. This theory explores sociopsychological needs, which can help explain individual users’ medium usage (Xu et al., 2012). It has been effectively applied to a variety of computer-mediated platforms relying on communication technology, including the internet (Stafford et al., 2004), smartphones (Joo and Sang, 2013), social networking services (SNSs) (Liu et al., 2016), online videos (Khan, 2017), internet-based games (Li et al., 2015; Wu et al., 2010), instant messaging (Gan and Li, 2017), and mobile learning facilitated by mobile devices (Aburub and Alnawas, 2019; Chang et al., 2021; Shukla, 2021). Given that SML facilitates access to computer-mediated learning, applying U&G theory to this context is pertinent because it relates to the internet, new media, and IS.

U&G theory focuses on several key areas: (1) psychological and social origins that lead to (2) user needs; these needs, in turn, cultivate (3) user expectations of (4) the medium or computer-mediated communication systems. These expectations then give rise to (5) varying patterns of medium usage, which yield (6) the gratification of these needs and may result in (7) other outcomes, potentially including unintended consequences (Katz et al., 1973; Palmgreen, 1984; Rubin, 1994).

With the development of information and communication technology, researchers have new interpretations of U&G theory. Finn (1997) suggested that the integration of media and computer-mediated communication technology has changed the exposure patterns of users. Moreover, Ruggiero (2000) indicated that the interactive nature of the internet dramatically enhances the core of the U&G concept in terms of active usage; a computer-mediated medium also enhances the effectiveness of U&G theory. As a result, U&G theory can be applied to the study of telecommunications media effectively and efficiently (e.g., Ruggiero, 2000).

As a result, the SML system meets the U&G theory assumptions: (1) The users are active. (2) Users who choose a medium/computer-mediated communication system are goal-oriented and purposeful with expectations. (3) Users know their needs and expectations to use a specific medium/computer-mediated communication system. After usage, the user’s needs are either satisfied or not satisfied. If their needs are satisfied, their continuance usage will be strengthened (Katz et al., 1973).

This study seeks to empirically assess whether users’ expectations of SML are being gratified. Given its focus, U&G theory functions as an appropriate approach for studying the post-acceptance of SML. The U&G framework is capable of examining user satisfaction in the post-acceptance stage, providing insight into whether users feel gratified with an information technology following its acceptance. After all, a central research question posed by U&G theory is as follows: what gratifications of corresponding expectations do users experience after using a specific medium? (e.g., Ruggiero, 2000). As such, U&G theory offers a logically coherent paradigm, seamlessly allowing for the exploration of users’ continuance intention in the post-acceptance phase (e.g., Luo and Remus, 2014).

In forecasting the usage of information technology, research predominantly based on technology diffusion theories, such as the technology acceptance model (TAM), often neglects the emotional components that emerge from users’ personal and social realms. In contrast, U&G theory fills this gap by addressing the emotional factors that users contemplate after employing technology, hence serving as a supplement to the limitations of technology diffusion theory in delineating a user’s post-acceptance usage. This stems from the significance of considering emotional elements, which are intrinsic to users’ personal and social facets, during the post-acceptance phase of technology usage. Additionally, the notion of satisfaction in U&G theory transcends the simple fulfilment of a need. It includes a pleasurable element, essentially indicating that gratification is an amalgamation of satisfaction and happiness. Following our previous discussion, U&G theory thereby emerges as a fitting framework for investigating continuance intention during the post-acceptance phase. Dissecting different forms of user satisfaction post-usage can provide a comprehensive understanding of the user’s post-acceptance gratification. This holistic perspective, which includes acknowledging emotional factors and expanding the understanding of satisfaction, underscores the advantages of the U&G approach in examining post-acceptance phenomena.

Research construct and hypothesis formulation

Uses and gratifications typologies

Prior research leveraging U&G theory has classified user gratification in the post-acceptance stage across various representative media platforms. Table 1 presents a typology of uses and gratifications corresponding to representative new (computer-mediated) media.

Table 1 Typologies of uses and gratifications pertaining to representative new (computer-mediated) media.
Full size table

Research constructs

In this study, five types of gratification are identified and derived from a range of theories, including the incentive theory of motivation, flow theory, diffusion of innovation theory, self-determination theory, and learning theory. Following a comprehensive review of the literature on these theories, it is concluded in this study that several formative factors underpin these five gratifications. Specifically, technology gratification is composed of factors such as perceived intelligence (Bartneck et al., 2009) and convenience (Ko et al., 2005). Hedonic gratification is constructed from factors such as perceived enjoyment (Ryan and Deci, 2000) and concentration (Koufaris, 2002). Users’ social gratification is epitomised by the factor of status (Venkatesh and Brown, 2001), while achievement (Wu et al., 2010) embodies a factor associated with utilitarian gratification. Finally, education (Stafford et al., 2004) emerges as a factor that aligns with users’ content gratification. The research constructs that are employed in this study are outlined in Table 2.

Table 2 Research constructs.
Full size table

Hypothesis formulation

Continuance intention is a predominant anticipated behavioural outcome in the field of IS post-adoption research (Bhattacherjee, 2001). In this study, continuance intention is defined as the probability that a user will continue to use SML over an extended period.

The literature suggests that there is a positive correlation between technology gratification and continuance intention (Gan and Li, 2017; Gulvady, 2009). In this study, intelligence is used to describe the ability of Liulishuo’s SML system to continually acquire and store knowledge via self-learning processes (Ritter et al., 2011). This inherent smartness and intelligence, exhibited by Liulishuo’s SML system, is something that users actively assess and appreciate (Lee et al., 2007). In terms of the technical benchmarks that users consider, Liulishuo’s SML system should fulfil users’ technology gratification expectations, particularly in perceived intelligence. Prior research has empirically substantiated that perceived intelligence can exert a positive influence on users’ behavioural intentions concerning Internet of Things (IoT) systems (Dong et al., 2017) and personal intelligent agents (Moussawi et al., 2023). Thus,

H1. Perceived intelligence is positively associated with continuance intention towards SML.

Convenience refers to the ease with which users can obtain what they need by using a system (Chou et al., 2004). Convenience value, as defined by Larivière et al. (2013), refers to the value derived from accomplishing a task in a quick, efficient, and effortless manner. In fact, this convenience value is recognised as a principal driving force behind internet usage (Flanagin and Metzger, 2001). In service-oriented research, various dimensions of convenience have been identified and deliberated in the literature (e.g., Berry et al., 2002). Convenience has been empirically validated as a significant motivation for mobile media usage (Leung and Wei, 2000), YouTube utilisation (Haridakis and Hanson, 2009), the use of tourism mobile apps (Xu et al., 2019), and the utilisation of a learning management system (Bansah and Darko, 2022). Thus,

H2. Convenience is positively associated with continuance intention towards SML.

Prior research conducted on internet-based games (Li et al., 2015) and general portals (van der Heijden, 2003) has demonstrated that perceived enjoyment is a predictive factor for continuance intention. Perceived enjoyment has been acknowledged as a primary determinant of media usage (Ryan and Deci, 2000). Empirical evidence from previous studies indicates that when a usage process is perceived as enjoyable, it strengthens users’ intentions to continue using the system (Gallego et al., 2016). This viewpoint has been confirmed in research on online accommodation booking platforms (So et al., 2021). Thus,

H3. Perceived enjoyment positively influences continuance intention towards SML.

Users who are deeply engaged in their activities tend to exhibit a heightened concentration level (Koufaris, 2002). This high level of concentration, directed towards media usage, often culminates in what is referred to as a flow state (Sherry, 2010). Past research findings posit that this intense concentration can notably enhance the usage of e-learning systems (Lee, 2010). Thus,

H4. Concentration positively influences continuance intention towards SML.

Users derive a sense of social gratification from the enhanced status associated with the use of Liulishuo’s SML system. Drawing upon the literature on the diffusion of innovations, it is posited that the pursuit of higher social standing is a fundamental motivator for IS usage (Rogers, 1995). This perspective also applies to online technologies, where the allure of elevated status continues to drive user engagement (Yan et al., 2021). Thus,

H5. Status positively influences continuance intention towards SML.

Achievement, in the realm of using an e-learning system, is embodied by users earning learning points, completing missions, levelling up, and competing with others (Ryan and Deci, 2000). As users engage with Liulishuo, they incrementally elevate their proficiency levels. Past studies have empirically validated the idea that this sense of achievement serves as a motivating factor that can predict users’ intention to continuing using a system (Wu et al., 2010). This has been confirmed in recent research on smartphone apps (Mi et al., 2021). Thus,

H6. Achievement positively influences continuance intention towards SML.

Users’ content gratification, within the context of Liulishuo’s SML system, is signified by the educational benefit they accrue. This form of gratification pertains to the valuable information relayed by the media, which fulfils users’ expectations (Cutler and Danowski, 1980). Underpinned by a user-friendly social mechanism and the contributions of nearly 100 elites in English culture and education, Liulishuo expanded from merely oral learning to encompass a broad spectrum of English culture learning. Previous empirical studies have confirmed that content gratification can foster users’ engagement with online media (Stafford et al., 2004) and enhance online learning (Li and Liu, 2023). The education acquired by users through Liulishuo, as an SML system, can enhance their intention to continue using the platform. Thus,

H7. Education is positively associated with continuance intention towards SML.

Figure 1 shows the research model.

Fig. 1: Research model.
figure 1

Description: The figure depicts the research model.

Full size image

Research method and data

Measurement development

In this study, a quantitative research methodology was implemented, and survey data were utilised. The participants involved in this study were users who had engaged with Liulishuo. All the measurement scales were adapted from previous literature, included perceived intelligence (Bartneck et al., 2009), convenience (Ko et al., 2005), perceived enjoyment (Moon and Kim, 2001; van der Heijden, 2003), concentration (Koufaris, 2002), status (Venkatesh and Brown, 2001), achievement (Wu et al., 2010), education (Stafford et al., 2004), and continuance intention (Bhattacherjee, 2001), and were tailored to fit the SML context. All measurement items were evaluated using a seven-point Likert scale. Table 3 provides a detailed view of the measurement items.

Table 3 Measurement items.
Full size table

The survey items were initially in English. The translation process of the survey items in this study from English to Chinese was meticulously carried out. The specific steps and techniques employed included forward translation, back translation, comparison and revision, and pilot testing and finalisation. These established translation methods were used to ensure linguistic and conceptual equivalence of the survey items (e.g., Smith et al., 2022).

Data collection methods

A pilot test of 107 samples was conducted prior to the formal study and was not included in the main survey. Preliminary evidence underscores the measurement scale’s reliability and validity.

A conditional random sampling procedure was implemented for our online questionnaire survey, which was hosted nationwide by the Baidu sample service in China. Given the study requirements, only respondents who confirmed prior use of Liulishuo via their response to the first question, “Have you used Liulishuo before?”, were permitted to proceed with the rest of the questionnaire. Since the platform studied, Liulishuo, primarily caters to a younger demographic, the nature of our sampling method inherently resulted in a sample that predominantly consisted of relatively young individuals, including college students and working professionals, which is reflective of the demographic profile of Liulishuo’s user base (Liulishuo.com, 2023). Participation in the survey was voluntary, with measures put in place to guarantee the authenticity and reliability of the responses collected. The data collection process spanned approximately 15 days, and after careful validation of responses, a total of 495 participants’ valid responses were deemed suitable and included in the subsequent analysis.

Data analysis and results

PLS-SEM can be utilised to assess the relationships among independent, dependent, mediating, and moderating variables, making it better suited for the data analysis in this study. When compared to covariance-based structural equation modelling (CB-SEM), PLS-SEM demonstrates fewer identification problems and displays greater robustness (e.g., Hair et al., 2011). Given that the research model involves a considerable number of variables and exhibits complexity, PLS-SEM can more effectively validate the relationships among the variables in the research model (Chin et al., 2003).

Descriptive statistics

Among 495 valid participants, 51.9% were female, and 48.1% were male; a total of 95.8% of the participants had a college degree or above. In addition, most of the participants (98%) were between 19 and 45 years old. The descriptive statistics align well with Liulishuo’s target audience. Earlier data analysis reports indicate that Liulishuo, being an English learning application for adults, predominantly caters to a demographic of college students and working professionals aged roughly between 19 and 45 (Liulishuo.com, 2023). Further academic studies suggest that this user base has a higher propensity to leverage technology for educational pursuits (Ding, 2019). Figure 2 illustrates the demographics of the participants. Figure 3 presents the visualisation of the Likert scale data.

Fig. 2: Demographic characteristics of participants.
figure 2

Description: The figure illustrates the demographics of the participants.

Full size image
Fig. 3: Visualisation of the Likert scale data.
figure 3

Description: The figure presents the visualisation of the Likert scale data.

Full size image

Measurement model

Chin (2010) suggested that in the first step of model evaluation, the measurement scale applied in each construct should be confirmed to be reliable and valid through analysis of the measurement model. Therefore, utilising the SmartPLS 3.0 version, the measurement model was initially evaluated in this study before the structural model was tested.

To assess the psychometric properties of the constructs, composite reliability (CR), Cronbach’s alpha, and average variance extracted (AVE) were calculated in this study, as shown in Table 4 (e.g., Gao et al., 2022; Nunnally and Bernstein, 1994). The results reveal that the composite reliability (CR) values exceed 0.800, the Cronbach’s alpha coefficient values surpass 0.700, and the average variance extracted (AVE) values exceed 0.500. As a result, the constructs of this study demonstrate good internal consistency reliability, as shown in Table 4.

Table 4 Discriminant validity.
Full size table

Discriminant validity was assessed in two steps. The first step was to load all items into the corresponding latent variable. In this study, it was found that all items exhibited loadings greater than 0.700. At the same time, the factor loading of an item in this study onto its associated construct surpassed that of any other nonconstruct item (e.g., Venkatesh et al., 2012). Thus, the measurement model exhibits strong convergent validity. Table 5 outlines the PLS loadings and cross-loadings. In the second step, Fornell and Larcker’s (1987) criteria were used in this study to assess whether the square root of any construct’s AVE in the research model was larger than the correlation between this construct and any other construct. If the squared correlations between the constructs do not exceed 0.800 in the correlation matrix, the multicollinearity issue can be disregarded without affecting the model’s estimation (Hair et al., 2006; Provenzano et al., 2020). Table 4 shows that the analysis results conform to Fornell and Larcker’s (1987) criteria. This suggests that the measurement model of this study exhibits good discriminant validity. Put simply, the five dimensions of the gratification construct and continuance intention are identifiable and distinguishable.

Table 5 PLS loadings and cross-loadings.
Full size table

Structural model

Tenenhaus et al. (2005) recommended the goodness of fit (GoF) measure (0 < GoF < 1) as a global fit index for PLS-SEM modelling. The GoF value of this model is calculated to be 0.64, exceeding the benchmark value of 0.36 for GoFlarge (e.g., Croasdell et al., 2011; Wetzels et al., 2009), suggesting that the model aptly fits the data. Consequently, the model demonstrates 63.1% of the variance for continuance intention.

Table 6 shows the measurement of the path relationships between the constructs in this model and the assessment of their significance levels. The findings from the data analysis validate all the hypotheses, except H4. These findings reveal that continuance intention is influenced by intelligence (β = 0.171, P < 0.01), convenience (β = 0.125, P < 0.05), perceived enjoyment (β = 0.120, P < 0.05), status (β = 0.147, P < 0.05), achievement (β = 0.131, P < 0.05), and education (β = 0.150, P < 0.05). However, concentration does not significantly influence continuance intention. As a result, the structural model analysis results empirically indicate the sequence of the importance of different gratifications affecting continuance intention towards SML.

Table 6 Hypothesis testing.
Full size table

Figure 4 presents the structural model.

Fig. 4: Structural model.
figure 4

Description: This figure depicts the structural model.

Full size image

Discussion and conclusions

Leveraging the U&G framework, this study reveals five primary categories of user gratifications influencing SML continuance intention, each of which embody an array of factors. Each type of gratification is dissected in order of its significance. The empirical results, which explore the relationship between each gratification type and continuance intention, shed light on their alignment or contradiction with the literature. Additionally, explanations are provided to rationalise the obtained findings.

Discussion

This study, focusing on Liulishuo as a typical research object of SML, employs empirical research to establish that five categories of gratification can effectively predict users’ intention to continue using Liulishuo. According to the order of importance, the first category is technology gratification, which is represented by intelligence. The second category is content gratification, which is represented by education. The third category is social gratification, which is represented by status. The fourth category is utilitarian gratification, which is represented by achievement. The fifth category is technology gratification, which is represented by convenience. The sixth category is hedonic gratification, which is represented by perceived enjoyment.

Hypothesis H1 investigates the influence of perceived intelligence on users’ continuance intention towards SML. The results demonstrate that intelligence, as a technology gratification, emerges as the most influential factor in predicting the continuous usage of Liulishuo. This finding aligns with the research findings of Stafford et al. (2004) on the internet, those of Liu et al. (2016) on SNSs, those of Gan and Li (2017) on instant messaging, and those of Moussawi et al. (2023) on personal intelligent agents. A higher degree of intelligence implies that users can avail themselves of more astute learning guidance through Liulishuo, thus enhancing their learning efficiency, comparable to having a personal human tutor. For instance, Liulishuo evaluates users’ performance and precisely identifies areas that need improvement, which can effectively and efficiently elevate the users’ English proficiency. This gratification derived from the intelligent features of Liulishuo bolsters users’ intention to continue its use.

Hypothesis H2 explores the impact of convenience on users’ continuance intention towards SML. The findings reveal that convenience, as a technology gratification factor, significantly influences a user’s sustained usage of Liulishuo. This insight aligns with the studies of Ko et al. (2005) on the internet, Liu et al. (2016) on SNSs, Xu et al. (2019) on tourism mobile applications, and Bansah and Darko (2022) on learning management systems. The ease and comfort provided by Liulishuo while learning English meet users’ need for convenience, thereby strengthening their desire for continued engagement.

Hypothesis H3 assesses the impact of perceived enjoyment on users’ continuance intention towards SML. The results demonstrate that perceived enjoyment, as a hedonic gratification, can effectively predict the continuous usage of Liulishuo users. This confirms previous research results on home personal computers (Venkatesh and Brown, 2001), internet-based games (Li et al., 2015), online videos (Khan, 2017), and online accommodation booking platforms (So et al., 2021). Users’ enjoyment and pleasant experiences when learning English using Liulishuo can meet their demand for enjoyment, thus strengthening their intention of continuous usage.

Hypothesis H4 investigates the impact of concentration on users’ continuance intention towards SML. Surprisingly, the results demonstrate that concentration does not affect users’ intention to continue using Liulishuo. This indicates that users consider concentration to be less critical when using Liulishuo.

Hypothesis H5 explores the influence of status on users’ continuance intention towards SML. The results suggest that status as a social gratification reliably predicts the intention to continue using Liulishuo. This confirms previous research on personal computers (Venkatesh and Brown, 2001), online videos (Khan, 2017), and online technologies (Yan et al., 2021). This demonstrates that users of Liulishuo place significant importance on gaining status through Liulishuo’s social systems.

Hypothesis H6 evaluates the effect of achievement on users’ intention to continue using SML. The results indicate that achievement, viewed as a form of utilitarian gratification, significantly impacts the sustained use of Liulishuo. This aligns with earlier research on internet-based games (Li et al., 2015) and smartphone apps (Mi et al., 2021). Thus, this study highlights that users attach significant importance to the utilitarian purpose of using Liulishuo, such as achieving specific goals.

Hypothesis H7 investigates the influence of education on users’ continuance intention towards SML. The findings confirm that education, as a content gratification, can effectively predict the ongoing usage of Liulishuo. This aligns with previous studies on the internet (Stafford et al., 2004) and online learning platforms (Li and Liu, 2023). These findings emphasise that users prioritise the educational content provided by Liulishuo in their continued usage.

Conclusions

This study contributes to understanding users’ gratification stage concerning SML by identifying specific gratifications. Doing so expands the scope and generalisability of U&G theory to the SML context. By examining the relationships among various types of gratifications and continuance intention, this study enhances our understanding of users’ gratification stage in SML. This insight is instrumental for continually advancing users’ language abilities. This study reveals that technology gratification, represented by intelligence, is the most influential factor in effectively predicting the sustained usage of SML. Furthermore, the empirical results highlight that users’ expectations of the SML system are of greater significance than their basic intrinsic needs.

Theoretical contributions

While prior research using uses and gratifications (U&G) theory to investigate IS-based media has primarily concentrated on predicting the role of hedonic, utilitarian, and social gratifications in fostering continued usage, there has been a noticeable gap in exploring the relationship between technology gratification, especially concerning the intelligence inherent in AI-based information systems, and sustained use. This study, however, establishes that intelligence, functioning as a form of technology gratification, is the most potent predictor of continuous SML usage. Therefore, this study makes a pioneering contribution by introducing intelligence as a novel factor of technology gratification. In doing so, it contributes to the expansion of U&G theory, rendering it more applicable to SML-related research. The research also affirms that the technology gratification of convenience positively impacts the continuous usage of SML, further bolstering the integral role of technology gratification in the context of sustained SML usage.

This study revealed that content gratification, represented by education, has the second most significant positive effect on the continuous usage of SML. Additionally, this study confirmed that hedonic gratification, represented by perceived enjoyment; social gratification, represented by status; and utilitarian gratification, represented by achievement, can effectively predict continuous usage. The results demonstrate that various factors contribute to gratification with SML usage and indicate the order of importance of these factors in predicting continuous usage.

The findings of this research are theoretically valuable, as they deepen our understanding of the driving factors behind SML usage from the perspective of uses and gratifications (U&G) theory. The findings can provide a basis for further theoretical exploration into how users can optimise their learning management for increased efficiency. These insights not only contribute to the enhancement of English proficiency but also potentially extend the theoretical framework to other online learning sectors. By elucidating the primary gratifications leading to continued SML use, this research builds upon, integrates, and expands existing theories, thus offering a more comprehensive theoretical basis for the study of online learning engagement and commitment.

U&G theory offers a comprehensive understanding of users’ post-acceptance behaviour by explaining various types of user gratification after usage. This study effectively relates U&G theory to the post-acceptance research of IS, represented by SML. This connection is also theoretically sound, as U&G theory addresses users’ gratifications with expectations after using a specific medium (e.g., Ruggiero, 2000).

This study has effectively expanded the range of U&G theory, originally rooted in communication research, by applying it to the domain of SML systems. These systems, as conveyors of learning information, can be classified broadly as computer-mediated mediums. Consequently, the theoretical justification for this extension is sound and brings a fresh perspective to the analysis of online learning systems, which have traditionally been examined from an IS viewpoint. By merging the theoretical strengths of both the management IS field and the communication field, this study offers an enriched, interdisciplinary perspective on users’ post-acceptance behaviour.

Practical implications

The findings offer practical guidance for retaining existing SML users in China by addressing their specific needs for gratification. By doing so, it could bolster user engagement and, in turn, foster customer loyalty. As insights from this study draw attention to particular user gratifications, they significantly inform practitioners in their design of future mobile learning products. Aligning these offerings with user needs and expectations ensures a more engaging and beneficial learning experience. Furthermore, the strategic optimisation of the design elements of SML systems, informed by these user gratifications, can significantly enhance learning efficiency.

The findings emphasise the crucial role of technological gratifications, notably intelligence and convenience, in promoting the continuous usage of SML systems. For practitioners in the SML field, these findings suggest that enhancing the intellectual capabilities of these platforms is crucial, making them more akin to smart, personalised tutors. Additionally, the element of convenience should not be overlooked. An easy-to-navigate interface, coupled with the flexibility of mobile access, offers users the advantage of learning anytime and anywhere.

In addition, this research emphasises the importance of content gratification, particularly in terms of educational value, as a critical factor in driving continuous usage. This serves as a significant insight for SML practitioners, suggesting that they should focus on enriching the educational content provided by these platforms to better meet user needs.

Moreover, hedonic gratification, illustrated by perceived enjoyment; social gratification, depicted by status; and utilitarian gratification, represented by achievement, have been identified in this study as essential factors that motivate SML users’ continued engagement during the gratification stage. These findings suggest that SML practitioners should focus on improving platform features and functions that enhance perceived enjoyment, elevate user status, and facilitate achievement. This could involve introducing more interactive and entertaining learning activities, developing social features that allow users to share progress and compete with peers, and establishing clear goals and reward mechanisms to boost users’ sense of achievement.

Finally, the insights garnered from this study could guide individual SML users in understanding and managing their own learning processes more effectively, leading to substantial improvements in their language proficiency.

Limitations and future research directions

This study also has certain limitations, offering avenues for improvement in future studies. The samples are geographically limited to China. On the one hand, the cultural context of the study might introduce a degree of cultural bias. On the other hand, because the subjects hail from a developing country, the findings may encompass a certain level of economic distortion. The findings can be more generalised if future studies use samples from other Western developed countries for comparison.