Digital literacy equity in AI-enhanced English learning: Access, participation, and implications for language education policy Authors: Ning Zuo, Chinaza Solomon Ironsi, Sarah Solomon Ironsi Published: 2026 ABSTRACT Despite the rapid integration of artificial intelligence (AI) into English language education, significant disparities persist in learners’ ability to benefit from AI-enhanced learning environments. Existing research has largely emphasized technological access, leaving insufficient empirical attention to the roles of digital literacy, participation, and learner agency in shaping equitable learning outcomes. This study addresses this gap by examining digital literacy equity in AI-enhanced English learning through a quantitative cross-sectional design. Quantitative data were collected from 220 tertiary-level English language learners using validated survey instruments and analyzed using descriptive statistics, regression, mediation, and group comparison analyses. The findings reveal that digital and AI literacy significantly predict learner participation and agency, which in turn mediate learning outcomes. Learners with lower literacy levels demonstrated reduced engagement and weaker outcomes despite comparable access to AI tools, highlighting the persistent digital divide. The study’s originality lies in empirically modeling participation and agency as co-mediating mechanisms linking literacy to learning outcomes, offering critical insights for equity-centered AI integration and language education policy.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-55014-w. Introduction The rapid integration of artificial intelligence (AI) into English language education is reshaping how learners access content, engage in language practices, and receive feedback. AI-enhanced English learning environments—such as adaptive learning platforms, automated writing evaluation systems, conversational chatbots, and technological tools—promise personalized instruction, immediate feedback, and scalable learning opportunities1–3. At the same time, these technologies demand increasingly sophisticated forms of digital literacy that extend beyond basic technical skills to include critical, ethical, and AI-specific competencies4,5. As a result, digital literacy has become a central determinant of who benefits from AI-mediated English learning and who remains marginalized. Digital literacy equity refers to the fair distribution of access, skills, participation opportunities, and learning outcomes associated with digital technologies6. In AI-enhanced English learning, equity concerns arise when learners have unequal access to AI tools, uneven capacity to use them effectively, or limited agency to shape how these tools mediate language learning7. While AI is often positioned as a democratizing force in education, research increasingly shows that it can reproduce or amplify existing inequalities linked to geographic location, institutional resources, and linguistic background8–10. These inequities are particularly consequential in English education, where language proficiency is closely tied to academic success, employability, and social mobility. Prior research showcases that AI tools can enhance English learning by providing individualized feedback, fostering learner autonomy, and increasing engagement11. Studies reported improved writing accuracy through automated feedback systems12, increased speaking practice via AI-driven chatbots13, and enhanced motivation through adaptive learning platforms14. Research also showed that generative AI supports idea development and language practice when learners possess adequate digital and critical literacy skills15. However, empirical evidence also highlights significant equity challenges. Studies found that learners from under-resourced institutions experienced limited access to AI tools due to infrastructure constraints and licensing costs16. Researchers reported that students with lower digital literacy struggled to interpret AI-generated feedback, leading to superficial engagement or overreliance on automated suggestions17. Teachers likewise faced difficulties integrating AI into pedagogy without sufficient training, often resulting in inconsistent or inequitable classroom practices18. Research on the digital divide further revealed that inequities operate at multiple levels. First-level divides concern access to devices, connectivity, and platforms; second-level divides involve differences in skills and usage; and third-level divides relate to unequal educational outcomes and benefits6,19. In AI-enhanced English learning, these divides intersect with language proficiency, cultural capital, and institutional policy, compounding disadvantages for already marginalized learners20. Despite growing interest in AI and language education, three major research gaps persist. First, existing studies rarely integrate digital literacy equity as a comprehensive analytical framework. Most research isolates either AI effectiveness or digital skills without examining how literacy disparities shape access and participation across contexts21. Second, limited attention has been given to participation and agency, particularly how learners negotiate power relations embedded in AI systems, such as algorithmic feedback norms and linguistic biases22. Third, there is a notable lack of policy-oriented analysis linking empirical findings to language education governance, teacher preparation, and curriculum design23. These gaps are significant because, without an equity-centered perspective, AI adoption risks reinforcing structural inequalities rather than mitigating them. Policymakers may implement AI initiatives without addressing the digital literacy conditions required for meaningful participation, leading to uneven outcomes and ethical concerns24. Addressing these gaps is therefore critical for developing inclusive, sustainable, and socially responsible AI-enhanced English education systems. The rationale for this study lies in the urgent need to align AI innovation with principles of educational equity and social justice. As educational systems increasingly institutionalize AI tools, understanding how digital literacy mediates access and participation becomes essential for informed decision-making. This study responds to calls for interdisciplinary research that bridges applied linguistics, educational technology, and policy studies25. By examining digital literacy equity in AI-enhanced English learning, the research provides evidence-based insights that can guide educators, institutions, and policymakers toward more inclusive practices. This study investigates digital literacy equity in AI-enhanced English learning, focusing on three interrelated dimensions: access, participation, and policy implications. The research contributes to the growing field of AI and language education by foregrounding equity as a central analytic lens rather than a secondary consideration. While existing studies often emphasize technological affordances or learning outcomes, fewer examine how digital literacy mediates learners’ meaningful engagement with AI systems. This research is essential because equitable participation in AI-enhanced English learning is not solely a technical issue but a structural and policy-driven concern that shapes who can benefit from educational innovation. This study makes three key contributions. First, it conceptualizes digital literacy equity as a multidimensional construct encompassing access, skills, participation, and outcomes in AI-enhanced English learning. Second, it synthesizes empirical evidence to reveal how disparities in digital literacy shape learner and teacher engagement with AI tools. Third, it translates these insights into policy-relevant implications, offering guidance for equitable AI integration in language education systems. By doing so, the study advances theoretical understanding and provides practical recommendations for addressing inequality in digitally mediated English education. Guided by these aims, the study addresses the following research questions: How does digital literacy influence equitable access to AI-enhanced English learning tools?In what ways do differences in digital literacy affect learner and teacher participation in AI-mediated English learning environments? How does digital literacy influence equitable access to AI-enhanced English learning tools? How does digital literacy influence equitable access to AI-enhanced English learning tools? In what ways do differences in digital literacy affect learner and teacher participation in AI-mediated English learning environments? In what ways do differences in digital literacy affect learner and teacher participation in AI-mediated English learning environments? Literature review Digital literacy in language education Digital literacy has emerged as a foundational competence in contemporary education, encompassing not only functional use of technology but also critical understanding of digital content, ethical considerations, and reflective engagement with digital tools4,5. In language education, digital literacy intersects with communicative competence, shaping how learners interpret, produce, and negotiate meaning through digital media. Some authors argued that digital literacy extends beyond technical proficiency to include evaluative judgments about media content5. Chiu et al. further proposed that AI literacy—a subset of digital literacy—requires learners and educators to understand not only how to operate AI tools but also how to interpret, critique, and responsibly leverage algorithmically generated content26. Studies have shown that inadequate digital literacy hampers learners’ capacity to benefit from technology-mediated learning environments. Zheng and Xiao27 reported that learners with stronger self-regulated learning strategies navigated digital learning platforms more effectively, supporting deeper engagement and better outcomes. Similarly, Gena and others highlighted that digital inequities compound linguistic disadvantage, particularly for learners outside urban or affluent educational settings28. These findings consistently underscore the role of digital literacy as a mediator of learning gains and equity. However, much of this research has focused on general digital skills rather than AI-specific literacies necessary for engaging with advanced language technologies (such as generative AI and adaptive feedback systems). This gap signals a need to conceptualize digital literacy in language learning as inclusive of AI competencies, not merely basic digital skills. AI-enhanced English learning AI-enhanced English learning encompasses a wide range of applications, from intelligent tutoring systems to automated writing evaluation and conversational agents. Researchers have documented positive impacts of AI on various language skills. Research found that automated writing evaluation systems improved ESL learners’ writing accuracy over time2. Other authors showed that AI-driven chatbots provided additional speaking practice, increasing learner confidence and exposure to authentic dialogue13. Some studies reported that generative AI tools helped learners to brainstorm ideas and practice writing fluency12,13. However, these benefits are not uniformly experienced. Crosthwaite observed that while automated feedback can accelerate revision cycles, learners often misinterpret feedback or rely on surface corrections without deep reflection29. Similarly, Celik emphasized that access to AI-enhanced tools is stratified; learners at well-resourced institutions benefit more than those in resource-constrained settings21. Contradictions emerge in how AI impacts learner autonomy. Some studies argue that AI scaffolds autonomy by offering self-paced support30, while others caution that overdependence on AI suggestions may undermine critical thinking29. These differences highlight that AI efficacy depends on user capability and context, not on intrinsic technological power. Despite evidence of positive effects, research remains fragmented. Studies often examine specific tools in controlled settings rather than holistic learning ecosystems. Few inquiries address how these technologies interface with broader educational structures, such as institutional policies, teacher roles, or curriculum requirements. This fragmentation suggests a research lacuna in systemic perspectives on AI-enhanced language learning that integrate technology, pedagogy, and educational context. Educational equity and the digital divide The concept of the digital divide has evolved from simple access disparities to multifaceted inequities involving skills and outcomes28. There is a layered model of digital divide: first-level divide (access to devices and connectivity), second-level divide (skills and usage), and third-level divide (outcomes and benefits)28. This framework is particularly instructive for examining AI-enhanced English learning because disparities at each level can block or distort equitable participation. Empirical research has validated this layered perspective. Hargittai32 demonstrated that access alone does not guarantee beneficial technology use; digital literacies mediate how individuals engage with technology. Warschauer and Matuchniak extended this argument to educational settings, showing that students with limited digital skills were less able to make use of online learning resources33. Ahmad and others6 reported that linguistic identities and sociocultural factors further intersect with digital divides, affecting participation in digitally mediated language learning. In AI contexts, equity concerns multiply. Technologies trained on the majority language varieties may inadequately support non-standard or multilingual learners, reinforcing linguistic hierarchies. Chiu and others emphasizes that algorithmic biases in language models can disadvantage learners whose language use falls outside dominant norms26. Khodabin and Arsalani similarly argued that AI in education must be evaluated not only for utility but also for ethical fairness and inclusiveness24. Although these studies foreground equity, few integrate digital divides with learner psychology or examine how affective factors (motivation, anxiety) intersect with technological inequities. For example, while there is evidence that learners with stronger digital literacy engage more confidently with AI tools, less is known about psychological barriers—such as technology anxiety or low academic self-efficacy—that may inhibit participation irrespective of access. This intersection remains underexplored in current literature. Participation and agency in AI-mediated learning Participation in AI-mediated learning is not merely a matter of having tools available; it also involves agency in selecting, interpreting, and applying AI output. Sociocultural theory posits that tools mediate cognition and learning34, suggesting that learner engagement with AI reflects broader social and cultural practices. In language education, agency includes negotiating meanings, making strategic choices, and co-constructing learning with peers and educators. Empirical research indicates that agency is shaped by learners’ competencies and by instructional design. Fryer and Carpenter35 found that learners who actively set conversational goals with chatbots derived more benefit than those who used tools passively. Research also found that teachers who co-designed AI tasks with learners fostered richer engagement than those who adopted off-the-shelf AI activities36. These findings illustrate that pedagogical mediation matters as much as technological presence. However, gaps remain concerning how power relations embedded in AI systems affect participation. AI feedback often reflects norms encoded in training data, privileging certain linguistic expressions and genres. Some studies cautioned that such embedded biases can constrain learner agency by normalizing certain language forms while marginalizing others. These issues point to a need for research examining how AI-driven norms shape learner identities and participation, especially among linguistically diverse populations. Policy perspectives on AI and digital literacy Policy discussions on AI in education are gathering momentum internationally. OECD37 articulated opportunities and risks associated with AI, emphasizing equity and ethical use. Research provided guidance on generative AI in education, underscoring the need for digital literacy development and safeguards against misuse38. However, policy frameworks often lack specificity regarding language education contexts or equity implementation strategies. Studies on policy implementation reveal tensions between innovation and regulation. There are suggestions that policy discourse often emphasizes AI’s potential while underplaying equity and teacher support. This imbalance risks accelerating adoption without sufficient frameworks for professional development, curricular integration, and evaluation. Consequently, there is limited evidence on how policies translate into practice in language classrooms, especially concerning equitable access and participation. Synthesis and research gaps Across the literature, three themes emerge: (1) digital literacy is central to meaningful technology use; (2) AI can support language development, but its benefits are uneven; and (3) equity concerns pervade access, participation, and outcomes. While existing studies provide foundational insights, they also have limitations. Research often isolates technological efficacy from sociocultural and policy contexts, overlooks psychological mediators of participation, and rarely integrates equity concerns into a comprehensive framework. Specifically, the gaps include a lack of AI-specific digital literacy frameworks integrated with language learning research. There are also limited systemic perspectives that connect technology, pedagogy, and policy. Exegesis from broader literature suggests insufficient attention to the psychological dimensions of participation, such as motivation and anxiety. Aside from these psychological variables, there is also scarce empirical evidence on how equity considerations inform policy and practice in AI-enhanced language education. Addressing these gaps is essential to ensure that AI does not merely augment educational technologies but does so in ways that expand equitable access and enable meaningful participation for diverse learners. Theoretical and conceptual framework This study is grounded in an interdisciplinary theoretical framework that integrates sociocultural theory, digital literacy theory, and critical perspectives on artificial intelligence (AI) in education to examine digital literacy equity in AI-enhanced English learning. Together, these perspectives provide a robust lens for understanding how access, participation, and learning outcomes are socially mediated, technologically structured, and institutionally governed. Sociocultural theory posits that learning occurs through mediated social interaction, where tools and symbols shape cognitive development34. In AI-enhanced English learning, AI tools function as mediational artifacts that influence how learners engage with language, feedback, and interaction. These tools do not operate neutrally; rather, their pedagogical impact depends on learners’ prior knowledge, digital competencies, and the instructional contexts in which they are embedded. This perspective foregrounds participation and agency as central to equitable learning experiences. Digital literacy theory extends this view by emphasizing that effective participation in digital learning environments requires layered competencies, including technical, cognitive, critical, and ethical dimensions4,5. Recent scholarship has highlighted AI literacy as an essential extension of digital literacy, encompassing the ability to interpret al.gorithmic output, recognize bias, and make informed decisions when using AI tools4. From an equity standpoint, disparities in digital and AI literacy constrain learners’ ability to benefit from AI-enhanced English learning, reinforcing second- and third-level digital divides6,19. Critical AI in education perspectives further inform this framework by interrogating power, bias, and governance in AI systems. Researchers argue that AI tools have linguistic norms and epistemic values that may marginalize non-dominant language varieties and learner identities22,25. These perspectives stress the importance of policy, ethics, and institutional accountability in shaping how AI is deployed in educational settings. Without deliberate equity-oriented design and regulation, AI risks reproducing existing structural inequalities rather than mitigating them. Building on these theories, this study proposes a conceptual framework that positions digital literacy equity as a foundational condition shaping engagement with AI-enhanced English learning environments. Digital literacy equity is conceptualized as encompassing equitable access to AI tools, differentiated skill development, and AI-specific literacy competencies. These factors shape learners’ capacity to engage meaningfully with AI-mediated pedagogical practices. Within the framework, AI-enhanced English learning functions as the instructional space where AI tools interact with pedagogical design. Learner participation and agency mediate this interaction, reflecting the degree to which individuals actively interpret, adapt, and apply AI-generated input. Participation is understood as both behavioral engagement and psychological investment, while agency refers to learners’ capacity to make informed, autonomous decisions in AI-supported learning processes. Policy and institutional context operate as an overarching structural influence that shapes all components of the framework. Policies related to infrastructure investment, teacher professional development, curriculum design, and ethical governance condition the distribution of digital literacy resources and the implementation of AI tools. These contextual factors ultimately affect equitable learning outcomes, defined as fair opportunities for language development and meaningful educational benefits. This conceptual framework enables systematic examination of how digital literacy equity influences access and participation in AI-enhanced English learning and how policy interventions can address inequities at multiple levels. Methodology Research design This study adopts a quantitative cross-sectional research design to examine relationships among digital literacy, AI literacy, participation, learner agency, and learning outcomes in AI-enhanced English learning. A correlational approach was employed to identify predictive and mediating relationships among variables. This design is appropriate because it enables the systematic examination of structural relationships among key constructs associated with digital literacy equity. Research context The study was conducted across three tertiary institutions located in China, in an English language learning context where AI-enhanced tools, such as automated writing feedback systems, adaptive learning platforms, and generative AI applications, were actively used in instructional practice. These contexts were selected because they represented sites where AI adoption was expanding rapidly and where issues of digital literacy equity were particularly salient. Participants and sampling Participants included English language learners (ELLs). Learners were enrolled in English courses that incorporate AI-enhanced learning tools. Stratified purposive sampling was employed to ensure representation across varying levels of digital access and literacy, which is central to the study’s equity focus. The sample includes approximately 220 learners, representing diverse levels of English proficiency and prior digital learning experience. It was required that experienced teachers who were conversant in the use and application of AI instruction assist in implementing the study. Demographic data collected include age, gender, and educational level, language background, access to digital devices, internet connectivity, and prior exposure to AI-based learning tools. Participants ranged in age from 18 to 29 years (M = 21.8), with approximately 56% female and 44% male learners. Based on self-reported measures, participants demonstrated heterogeneous levels of access to AI tools and digital literacy. The study employed purposive sampling to select institutions and participants actively engaged in AI-enhanced English learning. This technique was appropriate because it ensured that participants possessed direct experience with the phenomena under investigation, thereby increasing the relevance and richness of the data. Within selected institutions, stratified purposive sampling was used to capture variation across proficiency levels and access conditions, supporting equity-focused analysis. It was ensured that τhe survey distribution was restricted through institutional access, including course enrollment lists and controlled links. This ensured that only eligible participants could respond. Ethical considerations Ethical approval was obtained from the University’s ethics committee prior to data collection. It was confirmed and approved that the research adheres to and will be conducted in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments, or comparable ethical standards. Human ethics and consent to participate All participants provided written informed consent and were informed of their right to withdraw at any time without penalty. Data was anonymized using coded identifiers, and no personally identifiable information is reported. Given the involvement of AI tools, particular attention was paid to data privacy, platform transparency, and ethical use of AI-generated content, in line with international guidelines. All digital data was securely stored on password-protected systems accessible only to the research team. Data collection procedures Quantitative data were collected through an online survey administered to learners. The survey includes a Digital literacy and AI literacy scale adapted from validated instruments39,40, self-reported measures of participation and engagement in AI-enhanced English learning, and perceptions of institutional and policy support41. These scales are reported to be a valid and reliable instrument that can measure constructs like digital and Artificial intelligence literacy for learners and levels of participation and engagement while using AI in English language instruction39,40, and41. The survey used a five-point Likert scale and was piloted prior to administration to ensure validity and reliability. To assess access to AI tools, 5 items were included that measured learners’ availability and the quality of access to AI-enhanced learning resources. Example item: “I have reliable access to AI-based tools for learning English.” Digital literacy had 6 items and assessed general digital competencies, including navigation, evaluation, and effective use of digital tools. Example item: “I can effectively use online platforms to support my English learning.” For AI literacy, 6 items were included to measure learners’ ability to understand, interpret, and critically evaluate AI-generated content. Example item: “I can judge whether AI-generated feedback is accurate and useful.” Participation had 5 items and captured the extent of active engagement with AI tools during English learning. Example item: “I actively use AI tools when completing English learning tasks.” Learner agency also comprised 5 items that assessed learners’ sense of autonomy and control when using AI tools. Example item: “I decide how to use AI tools to support my English learning.” The learning outcomes comprised 5 items and measured perceived improvements in English language skills and learning effectiveness. Example item: “Using AI tools has improved my English learning outcomes.” Face validation of the instrument was conducted to ensure that the items were properly worded. Afterward, it was administered to 43 randomly selected participants to determine how reliably the instrument measures the constructs it purports to measure. A Cronbach’s alpha score of 0.88 was obtained, indicating that the instrument is reliable for measuring the study’s constructs. Data analysis procedures Quantitative data analysis was conducted in four stages: Descriptive statistics and correlation analyses were used to summarize distributions and examine bivariate relationships. Multiple regression analysis tested the predictive effects of digital literacy and AI literacy on learner participation. Mediation analysis was conducted to examine whether participation and learner agency mediated the relationship between digital/AI literacy and learning outcomes, following established mediation procedures. Group-comparison analyses (independent-samples t-tests) compared learners with high versus low digital literacy and access using median splits to identify equity-based differences in participation and outcomes. Findings This section presents the quantitative findings based on the dataset (N = 220). Results are organized according to the research objectives, focusing on (1) access to AI-enhanced English learning, (2) the role of digital and AI literacy in participation and agency, and (3) equity-based differences in learning outcomes. Table 1Descriptive statistics of key variables (N = 220).VariableMeanSDAccess to AI tools3.4951.121Digital literacy3.6231.081AI literacy3.5911.117Participation3.1910.902Learner agency3.0680.979Learning outcomes2.7910.947 Table 1Descriptive statistics of key variables (N = 220).VariableMeanSDAccess to AI tools3.4951.121Digital literacy3.6231.081AI literacy3.5911.117Participation3.1910.902Learner agency3.0680.979Learning outcomes2.7910.947 Descriptive statistics of key variables (N = 220). Table 1 presents descriptive statistics of key variables. Table 1 shows that learners reported moderate access to AI tools (M = 3.50), indicating that AI technologies are generally available across the sampled contexts. Both digital literacy (M = 3.62) and AI literacy (M = 3.59) are relatively high, suggesting that learners possess foundational competencies for engaging with AI tools. However, participation (M = 3.19) and learner agency (M = 3.07) are only moderate, indicating variability in how actively learners engage with AI-enhanced learning environments. Notably, learning outcomes are comparatively lower (M = 2.79), falling below the midpoint of the scale. This suggests that the availability of AI tools and literacy skills does not automatically translate into strong perceived learning gains, highlighting a potential gap between access and effective use. Table 2Correlation matrix of study variables.VariableAccessDigital literacyAI literacyParticipationAgencyLearning outcomesAccess1.000− 0.033− 0.049− 0.053− 0.093− 0.031Digital literacy− 0.0331.0000.0800.6230.7060.592AI literacy− 0.0490.0801.0000.3250.3350.325Participation− 0.0530.6230.3251.0000.7460.694Agency− 0.0930.7060.3350.7461.0000.725Learning outcomes− 0.0310.5920.3250.6940.7251.000 Table 2Correlation matrix of study variables.VariableAccessDigital literacyAI literacyParticipationAgencyLearning outcomesAccess1.000− 0.033− 0.049− 0.053− 0.093− 0.031Digital literacy− 0.0331.0000.0800.6230.7060.592AI literacy− 0.0490.0801.0000.3250.3350.325Participation− 0.0530.6230.3251.0000.7460.694Agency− 0.0930.7060.3350.7461.0000.725Learning outcomes− 0.0310.5920.3250.6940.7251.000 Correlation matrix of study variables. Table 2 reports correlations among key variables. The correlation analysis reveals several important patterns. First, digital literacy shows strong positive relationships with participation (r = 0.623), agency (r = 0.706), and learning outcomes (r = 0.592). This indicates that learners with higher digital literacy are more likely to engage with AI tools and achieve better learning outcomes actively. Second, participation and agency exhibit very strong correlations with learning outcomes (r = 0.694 and r = 0.725, respectively). This suggests that active engagement and learner control are central mechanisms that make AI-enhanced learning effective. In contrast, access to AI tools shows negligible and slightly negative correlations with all variables, including participation (r = − 0.053) and learning outcomes (r = − 0.031). This indicates that mere access does not significantly influence engagement or outcomes, reinforcing the distinction between access and meaningful use. AI literacy shows moderate relationships with participation (r = 0.325) and learning outcomes (r = 0.325), suggesting that while it contributes to engagement, its effect is weaker than that of general digital literacy. Table 3Regression results predicting participation.PredictorβIntercept1.12Digital literacy0.32AI literacy0.28 Table 3Regression results predicting participation.PredictorβIntercept1.12Digital literacy0.32AI literacy0.28 Regression results predicting participation. A regression analysis examined whether digital literacy and AI literacy predicted learner participation. This is presented in Table 3. Both digital literacy (β = 0.32) and AI literacy (β = 0.28) emerged as positive predictors of participation, with digital literacy exerting a slightly stronger influence. These results indicate that learners’ ability to use and critically interpret AI tools significantly shapes their level of engagement, reinforcing the importance of literacy equity. Both digital literacy and AI literacy significantly predicted learner participation, with digital literacy exerting a slightly stronger influence. Figures 1 and 2 present the relationship between participation and digital literacy/AI literacy. Fig. 1Participation in digital literacy. Participation in digital literacy. Fig. 2Participation in AI literacy. Participation in AI literacy. Figures 1 and 2 illustrate relationships between participation and (a) digital literacy and (b) AI literacy. Figure 1 shows a positive trend between digital literacy and participation, with higher literacy associated with greater engagement. Figure 2 shows a similar pattern for AI literacy, though with wider dispersion, reflecting uneven confidence in AI use. Together, these figures visually confirm that literacy disparities correspond to unequal participation, even when access is relatively widespread. Table 4Mediation Path Coefficients.PathβDigital literacy → participation0.623Digital literacy → agency0.706Participation → learning outcomes0.694Agency → learning outcomes0.725 Table 4Mediation Path Coefficients.PathβDigital literacy → participation0.623Digital literacy → agency0.706Participation → learning outcomes0.694Agency → learning outcomes0.725 Mediation Path Coefficients. Table 4 presents mediation path coefficients. The mediation analysis shows that digital literacy strongly predicts both participation and learner agency, which in turn significantly predict learning outcomes. Participation (β = 0.694) and agency (β = 0.725) both demonstrate strong effects on learning outcomes, indicating that engagement and autonomy are key pathways through which digital literacy influences learning. The strength of these relationships suggests that the impact of digital literacy on learning outcomes is largely indirect, operating through these mediating variables. Table 5Mean differences by digital literacy group.Outcome variableHigh digital literacyLow digital literacyParticipation3.6522.686Learner agency3.6522.429Learning outcomes3.2262.314 Table 5Mean differences by digital literacy group.Outcome variableHigh digital literacyLow digital literacyParticipation3.6522.686Learner agency3.6522.429Learning outcomes3.2262.314 Mean differences by digital literacy group. Table 5 presents mean differences by digital literacy group. Group comparison analysis reveals substantial disparities between learners with high and low digital literacy. Learners in the high digital literacy group report significantly higher participation (3.65 vs. 2.69) and greater learner agency (3.65 vs. 2.43). Similarly, learning outcomes are notably higher for high-literacy learners (3.23 vs. 2.31). The largest difference is observed in agency, indicating that digital literacy strongly influences learners’ sense of control and autonomy in AI-enhanced learning environments. These findings demonstrate that digital literacy functions as a critical equity factor, shaping not only engagement but also the quality of learning experiences. Discussion This study examined digital literacy equity in AI-enhanced English learning by quantitatively analyzing relationships among access, digital and AI literacy, learner participation, agency, and learning outcomes. Grounded in a multi-level digital divide framework and learner agency theory, the findings provide empirical support for the proposition that equity in AI-enhanced language education is not determined by access alone but is structurally mediated by learners’ capacities to participate and exercise agency. Digital literacy as a structural enabler of equity The findings demonstrate that digital literacy and AI literacy are significant predictors of participation, confirming that literacy functions as a foundational condition for equitable engagement in AI-enhanced English learning. While learners reported relatively high access to AI tools, variability in literacy levels translated into unequal participation and outcomes. This supports the argument that second-level digital divides—differences in skills and competencies—are more issues than access alone1,19. These results align with prior research showing that learners with stronger digital competencies engage more effectively with technology-mediated learning environments17,14. Similarly, studies on AI-enhanced language learning suggest that the benefits of tools such as automated feedback and generative AI depend heavily on users’ ability to interpret and apply outputs meaningfully17,15. The present study extends this body of work by empirically demonstrating that AI literacy, distinct from general digital literacy, plays an independent and significant role in shaping participation. However, the findings also diverge from techno-optimistic perspectives that frame AI as inherently democratizing8,22. While prior literature often suggests that AI can reduce inequities through personalization, the current results indicate that personalization does not compensate for insufficient literacy. Learners with lower literacy levels were less able to benefit from AI tools despite having comparable access, highlighting that equity cannot be achieved through technological provision alone. From a theoretical standpoint, these findings reinforce digital literacy theory by showing that literacy is not merely an individual skill set but a structural enabler that conditions access to learning opportunities. Within the study’s framework, digital literacy equity operates as the entry point in the pathway linking access to outcomes. This insight is significant because it shifts the focus of equity interventions from infrastructure provision to capacity building. In terms of practice and policy, this finding underscores the need for integrating explicit digital and AI literacy instruction into English language curricula. Instructional design should move beyond tool usage to include critical interpretation of AI-generated content, error recognition, and strategic application. At the policy level, investments in AI technologies must be accompanied by systematic efforts to develop learners’ literacy competencies, ensuring that access translates into meaningful participation. Participation as a primary engagement mechanism Participation emerged as the strongest mediator between digital literacy and learning outcomes, indicating that active engagement with AI tools is the primary pathway through which literacy advantages are realized. Learners with higher literacy levels demonstrated significantly greater participation, and this engagement was strongly associated with improved learning outcomes. This finding aligns with sociocultural theory, which posits that learning occurs through active interaction with mediational tools26. AI systems, as mediational artifacts, require users to engage actively to produce learning gains. The results are also consistent with research showing that engagement with digital platforms enhances learning outcomes when learners can navigate and use features effectively7,27. At the same time, the findings extend existing literature by quantitatively demonstrating that participation is not merely a byproduct of access or motivation, but a structurally conditioned behavior shaped by literacy. This challenges studies that assume engagement will naturally follow from exposure to AI tools13. Instead, the results suggest that participation must be intentionally scaffolded, particularly for learners with lower literacy levels. Interestingly, the magnitude of participation differences between high- and low-literacy learners was larger than the differences in learning outcomes. This suggests that while some learners may achieve acceptable outcomes with minimal engagement—possibly through passive use of AI tools—the depth and quality of learning are likely compromised. This nuance adds to the literature by highlighting that participation is not only a quantitative measure but also indicative of meaningful engagement. From an instructional perspective, this finding has direct implications for classroom practice. Educators should design AI-enhanced activities that require active interaction, such as iterative writing with AI feedback, guided chatbot conversations, and reflective evaluation of AI outputs. Simply providing access to AI tools without structured engagement risks reinforcing inequities. At the policy level, participation should be recognized as a key indicator of equity in AI-enhanced learning environments. Monitoring engagement patterns can help identify underserved learners and inform targeted interventions, ensuring that all students can participate meaningfully in AI-mediated learning processes. Agency as a mediated construct The study finds that learner agency is significantly influenced by digital literacy and functions as a mediator between participation and learning outcomes. However, agency exerts a slightly weaker effect than participation, suggesting that agency is a secondary, though still critical, mechanism in the learning process. This finding aligns with sociocultural and self-regulated learning theories, which conceptualize agency as the capacity to make informed, autonomous decisions within mediated environments6,7. The results support prior research indicating that learners who can critically interpret and strategically use digital tools exhibit greater autonomy and achieve better outcomes28,32. At the same time, the findings complicate existing assumptions that position agency as the primary driver of success in AI-enhanced learning26,33. Instead, the results suggest that agency is contingent on participation, which, in turn, is enabled by digital literacy. In other words, learners must first engage with AI tools before they can meaningfully control their learning processes. This layered relationship represents a key contribution of the study. By modeling participation and agency as co-mediators, the research advances understanding of how these constructs interact within AI-mediated environments. It demonstrates that agency is not an isolated trait but a relational, context-dependent construct shaped by both individual competencies and technological affordances. From a practical perspective, fostering agency requires more than encouraging autonomy; it requires equipping learners with the skills to evaluate AI outputs critically, make informed choices, and adapt strategies34. Instructional design should incorporate reflective activities, such as comparing AI-generated and self-produced texts, evaluating the quality of feedback, and making revision decisions. At the policy level, agency development should be embedded in curriculum standards for digital and AI literacy. Educational systems must move beyond procedural training to support critical and ethical engagement with AI, ensuring that learners are not passive consumers but active participants in AI-mediated learning. Equity implications in AI-environments Regarding equity implications in AI environments, the findings provide compelling evidence that AI-enhanced learning environments can both enable and constrain educational equity, depending on how digital literacy is distributed and supported. Despite relatively uniform access to AI tools, significant disparities were observed in participation, agency, and learning outcomes between high- and low-literacy learners. This confirms the persistence of second- and third-level digital divides3,35,36. These results align with critical perspectives on AI in education, which caution that technological systems can reproduce existing inequalities if underlying structural conditions are not addressed [33; 37, 39]. The study further supports research indicating that algorithmic systems may embed biases and require critical literacy for effective use22,10. However, the findings diverge from narratives that suggest AI inherently promotes inclusion by providing scalable, personalized learning opportunities24,23,38. The evidence indicates that without targeted support, AI may actually amplify inequities, as more literate learners are better positioned to leverage its benefits. The novelty of this study lies in its integrated empirical modeling of equity as a mediated process, linking digital literacy, participation, agency, and outcomes within a single framework. Unlike prior research that treats these variables independently, this study demonstrates how they interact dynamically to produce unequal learning experiences. This contribution advances both theoretical and empirical understandings of equity in AI-enhanced education. In relation to the theoretical framework, the findings validate the integration of sociocultural theory, digital literacy theory, and critical AI perspectives. AI tools function as mediational artifacts (sociocultural theory), whose effective use depends on layered competencies (digital literacy theory), and whose impacts are shaped by structural and ethical considerations (critical AI perspectives). Together, these frameworks explain why equity must be addressed at multiple levels—individual, pedagogical, and institutional. The reliance on self-reported data limits the ability to capture actual behavioral engagement with AI tools. Nonetheless, the findings challenge deterministic views of AI as inherently equitable and instead highlight the importance of human capacity, pedagogical design, and policy support in shaping inclusive learning futures. Conclusion This study investigated digital literacy equity in AI-enhanced English learning by examining how access, digital literacy, AI literacy, learner participation, and agency interact to shape learning outcomes. Drawing on a multi-level digital divide framework and learner agency theory, the study provides empirical evidence that equitable outcomes in AI-supported language education depend less on access to technology alone and more on learners’ capacity to engage meaningfully with and control AI-mediated learning processes. The quantitative findings demonstrate that digital literacy and AI literacy are significant predictors of learner participation and agency in AI-enhanced English learning environments. Learners with higher literacy levels reported greater engagement with AI tools and stronger perceptions of control over their learning, which in turn translated into improved learning outcomes. Mediation analyses further revealed that the effects of digital literacy on learning outcomes are largely indirect, operating through participation and learner agency. These findings confirm that literacy functions as a structural enabler, transforming technological access into educational benefit. Group comparison analyses reinforce the equity dimension of the study by showing significant disparities between learners with high and low digital literacy. Despite shared access to AI tools, learners with lower literacy levels consistently demonstrated reduced participation, weaker agency, and poorer learning outcomes. This pattern highlights the persistence of second- and third-level digital divides within AI-enhanced learning contexts, where inequalities are reproduced through differences in skills and engagement rather than solely through infrastructure. Collectively, the findings validate the study’s theoretical framework, which conceptualizes digital literacy equity as a processual pathway linking access, literacy, participation, agency, and learning outcomes. The study advances existing research by empirically modeling participation and agency as co-mediating mechanisms, offering a more nuanced understanding of how equity operates in AI-enhanced English education. By moving beyond access-centered perspectives, the study underscores the need to reconceptualize equity as a dynamic interaction between technological systems and learner capacities. This study contributes robust quantitative evidence that AI-enhanced English learning can both enable and constrain equity, depending on how digital and AI literacy are distributed and supported. Without intentional efforts to develop learners’ competencies and agency, AI integration risks amplifying existing inequalities rather than mitigating them. The findings thus call for a shift in research, practice, and policy toward equity-centered AI literacy development in language education. Implications of the study The findings of this study carry important implications for theory, practice, and educational policy in AI-enhanced English language learning. From a theoretical perspective, the study strengthens digital divide and learner agency frameworks by empirically demonstrating that participation and agency function as key mediating mechanisms between literacy and learning outcomes. This suggests that future models of AI-enhanced learning should conceptualize equity not as a static condition but as an evolving process shaped by learners’ engagement and control within AI-mediated environments. The findings encourage scholars to more explicitly integrate psychological constructs, such as agency, into digital equity research. The group-comparison results show that learners with lower digital literacy experience had significantly lower participation and outcomes. To address this, instruction should include scaffolded AI integration, such as guided AI prompts like pre-written input templates and step-by-step instructions for interpreting feedback. Teacher can model effective AI use for their students. For example, teachers can demonstrate, “Here is how to question AI feedback instead of accepting it directly.“This reduces inequities by making AI tools accessible to learners with varying levels of digital competence. Pedagogically, the results highlight the need to embed digital and AI literacy instruction within English language curricula. Simply providing AI tools is insufficient; learners must be supported in understanding how these tools work, how to critically evaluate their outputs, and how to integrate them into language learning tasks strategically. Educators should design AI-enhanced activities that scaffold participation and promote learner agency, particularly for students with lower literacy levels, to prevent disengagement and marginalization. For example, in AI-assisted writing activities, learners can be required to generate an initial draft independently, use AI tools like Grammarly to revise the text, and critically evaluate AI-generated feedback by identifying accurate suggestions and incorrect or misleading outputs. Teachers can implement structured prompts such as: “Which AI suggestions improved your writing, and which did not? Why?” This approach ensures that learners do not passively accept AI outputs but develop critical engagement skills, which are essential for equitable participation. Since participation emerged as a primary predictor of learning outcomes, AI-enhanced activities should require active, iterative engagement rather than a one-time use of the tool. For instance, teachers can design multi-stage AI-supported tasks, such as iterative essay development (Draft → AI feedback → revision → peer review → final submission). Teachers can design AI-mediated speaking practice where students interact with chatbots to simulate real-life conversations. These tasks can include classroom practices followed by reflection tasks evaluating fluency and accuracy. This study anticipates that such designs ensure that AI use becomes a process of engagement rather than a shortcut to answers. The findings show that learner agency significantly influences outcomes. To foster agency, teachers should design activities that allow learners to make decisions about AI use rather than follow instructions. For example, students can be encouraged to engage in AI/ human feedback comparison tasks. Here, students can compare AI-generated corrections with teacher feedback and then decide which is more accurate and justify their choice. They can mentor students on when to use AI and explain their reasoning. These activities encourage learners to control AI tools, rather than be controlled by them. From a policy perspective, the study underscores the need for equity-oriented AI integration strategies. Institutional and national policies should prioritize professional development for learner-focused AI literacy programs and inclusive design standards that account for learner readiness variability. Policymakers should move beyond infrastructure-focused initiatives and invest in capacity-building measures that ensure all learners can benefit from AI-enhanced education. There is a need to include provision for teacher AI training programs in our institutions. Finally, the study’s findings caution against techno-deterministic assumptions that AI inherently democratizes learning. Instead, they call for intentional, evidence-based approaches that align AI deployment with equity goals. By foregrounding literacy, participation, and agency, this study provides a foundation for more just and effective AI-enhanced English language education systems. Limitations This sought to elicit information on how digital literacy influences equitable access to AI-enhanced English learning tools and how differences in digital literacy affect learner participation in AI-mediated English learning environments. On this premise, this study relied solely on quantitative measures rather than experimental approaches. This limitation can be addressed by encouraging future studies to incorporate experimental approaches, such as classroom-based observations and task-based analyses of AI-mediated language-learning activities. Equally, such research designs can be integrated with qualitative approaches, such as interviews, to provide deeper, more contextually grounded insights into the same or similar issues. Appendix SURVEY Instructions: Please indicate your level of agreement with each statement using the following scale: 1 – Strongly Disagree. 2 – Disgree. 3 – Neutral. 4 – Agree. 5 – Strongly Agree. ………………………………………………………………………………………………………. PART 1. Age:……………………………………….Gender: ………………………………….Educational Level: ……………………….Type of device used most frequently: Smartphone / Laptop / Tablet / Other. Internet reliability: Very poor → Very strong (5-point scale). Access to paid AI tools: Yes / No. Age:………………………………………. Age:………………………………………. Gender: …………………………………. Gender: …………………………………. Educational Level: ………………………. Educational Level: ………………………. Type of device used most frequently: Smartphone / Laptop / Tablet / Other. Type of device used most frequently: Smartphone / Laptop / Tablet / Other. Smartphone / Laptop / Tablet / Other. Smartphone / Laptop / Tablet / Other. Internet reliability: Very poor → Very strong (5-point scale). Internet reliability: Very poor → Very strong (5-point scale). Very poor → Very strong (5-point scale). Very poor → Very strong (5-point scale). Access to paid AI tools: Yes / No. Access to paid AI tools: Yes / No. ………………………………………………………………………………………………………. PART 2. Items54321I have reliable access to AI-based tools for learning English.I can easily access AI tools when I need them for my studies.My internet connection allows me to use AI tools without interruption.I have access to appropriate devices, such as a laptop and a smartphone, to use AI tools.My institution provides access to AI-enhanced learning platforms.I can effectively use online platforms to support my English learning.I am confident in navigating digital tools and applications.I can evaluate the quality of online learning resources.I can solve basic technical problems when using digital tools.I can use digital tools to complete language learning tasks independently.I understand how to use different digital tools for different learning purposes.I understand how AI tools generate responses or feedback.I can judge whether AI-generated feedback is accurate.I can critically evaluate suggestions from AI tools.I am aware that AI tools can sometimes make errors.I know how to adjust my use of AI tools to improve learning outcomes.I understand the limitations of AI tools in language learning.I actively use AI tools when completing English learning tasks.I frequently engage with AI tools during my studies.I use AI tools to practice different language skills, such as writing and speaking.I interact regularly with AI tools for feedback or support.I explore different ways of using AI tools in my learning.I decide how to use AI tools to support my English learning.I feel in control of my learning when using AI tools.I choose when and how to rely on AI tools.I adapt AI tools to suit my learning needs.I critically reflect on how AI tools influence my learning.Using AI tools has improved my English learning outcomes.AI tools help me better understand English concepts.I am confident in my English skills when using AI tools.AI tools help me complete learning tasks more effectively.AI tools have positively impacted my English learning experience. Items54321I have reliable access to AI-based tools for learning English.I can easily access AI tools when I need them for my studies.My internet connection allows me to use AI tools without interruption.I have access to appropriate devices, such as a laptop and a smartphone, to use AI tools.My institution provides access to AI-enhanced learning platforms.I can effectively use online platforms to support my English learning.I am confident in navigating digital tools and applications.I can evaluate the quality of online learning resources.I can solve basic technical problems when using digital tools.I can use digital tools to complete language learning tasks independently.I understand how to use different digital tools for different learning purposes.I understand how AI tools generate responses or feedback.I can judge whether AI-generated feedback is accurate.I can critically evaluate suggestions from AI tools.I am aware that AI tools can sometimes make errors.I know how to adjust my use of AI tools to improve learning outcomes.I understand the limitations of AI tools in language learning.I actively use AI tools when completing English learning tasks.I frequently engage with AI tools during my studies.I use AI tools to practice different language skills, such as writing and speaking.I interact regularly with AI tools for feedback or support.I explore different ways of using AI tools in my learning.I decide how to use AI tools to support my English learning.I feel in control of my learning when using AI tools.I choose when and how to rely on AI tools.I adapt AI tools to suit my learning needs.I critically reflect on how AI tools influence my learning.Using AI tools has improved my English learning outcomes.AI tools help me better understand English concepts.I am confident in my English skills when using AI tools.AI tools help me complete learning tasks more effectively.AI tools have positively impacted my English learning experience. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1