A Deep Learning-Based Blended Teaching Model for Enhancing English Proficiency in English Education

A Deep Learning-Based Blended Teaching Model for Enhancing English Proficiency in English Education

Feng Hongli

Foreign Language Teaching School, Ningxia Medical University, Ningxia, Yinchuan, China

DOI: https://dx.doi.org/10.47772/IJRISS.2024.803467S

Received: 15 December 2024; Accepted: 19 December 2024; Published: 18 January 2025

ABSTRACT

This research introduces an advanced blended instructional model for college-level English as second language  education, integrating traditional in-person teaching with cutting-edge online learning components powered by Conditional Random Field (CRF) techniques within a deep learning framework. As a hybrid paradigm, blended learning has gained prominence in higher education due to its ability to enhance student engagement and learning outcomes. The CRF-based model optimizes pedagogical strategies through dynamic, adaptive, and personalized learning experiences, addressing diverse cognitive profiles and preferences. It integrates various instructional modalities—classroom interactions, digital resources, and interactive activities—into a cohesive framework, with CRF algorithms modeling sequential dependencies crucial for language acquisition tasks such as syntactic parsing and part-of-speech tagging. These foundational tasks enable students to internalize linguistic structures, fostering proficiency in English. By leveraging advanced deep learning architectures like recurrent neural networks (RNNs) and transformer models alongside large-scale linguistic datasets, the model achieves significant gains in accuracy, generalization, and responsiveness to individual learner needs. Empirical results demonstrate a 16% improvement in overall English proficiency and a 27% enhancement in reading comprehension compared to traditional methods, underscoring the transformative potential of AI-driven methodologies in language education. This study not only advances theoretical insights into instructional design but also establishes a robust framework for optimizing higher education through the integration of deep learning techniques.

Keywords: Blended Teaching, Conditional Random Fields, Online Education, Deep Learning, English Proficiency

INTRODUCTION

In the dynamic and ever-evolving domain of higher education, English instruction continues to serve as a foundational pillar, providing students with the advanced communicative competencies necessary for both academic achievement and professional advancement. As the demands of a globalized, knowledge-based economy intensify, the role of English education at the collegiate level extends far beyond the mastery of basic linguistic skills [4]. It encompasses the cultivation of critical thinking, the refinement of academic writing, and the development of articulate verbal expression. College English instructors, through their integration of diverse literary works, advanced language proficiency exercises, and interactive pedagogical strategies, are critical in fostering students’ intellectual growth. By facilitating an environment conducive to deep intellectual engagement, these educators empower students to engage meaningfully with complex global issues, while equipping them with the analytical tools and communication skills required to navigate and contribute to an increasingly interconnected and multifaceted global society [21]. Thus, college English education is not merely a conduit for linguistic proficiency but a transformative process that shapes well-rounded individuals capable of intellectual innovation and cross-cultural communication.

The advent of online English instruction, propelled by rapid technological advancements and the global dissemination of knowledge, has fundamentally transformed the paradigm of language education. The extensive adoption of digital platforms has obliterated traditional geographical boundaries, creating unprecedented opportunities for both language acquisition and intercultural exchange in virtual spaces [7]. This shift to online learning environments offers unparalleled flexibility, enabling learners to chart individualized educational paths, supported by a rich array of multimedia resources tailored to diverse learning styles and preferences. Furthermore, online instruction significantly enhances inclusivity, engaging a broader spectrum of students, regardless of their backgrounds or proficiency levels, thereby fostering a collaborative and dynamic learning ecosystem  [23].

Through the integration of cutting-edge pedagogical approaches, real-time feedback mechanisms, and interactive language exercises, hybrid English education not only bolsters linguistic proficiency but also nurtures digital literacy, a crucial skill for success in today’s interconnected world. This mode of instruction provides students with the tools to engage critically with digital content, ensuring they are prepared for the challenges and opportunities presented by the globalized digital landscape. As technological innovation continues at an exponential rate, the future of online English instruction holds vast potential, offering increasingly effective, adaptable, and accessible pathways for language learning. In this evolving educational landscape, online English instruction will continue to redefine the contours of language education, paving the way for more innovative, inclusive, and transformative educational experiences on a global scale [25].

The implementation of a deep learning-enhanced blended college English teaching model represents a sophisticated synthesis of traditional pedagogical methodologies and advanced technological innovations, aimed at optimizing student engagement and improving learning outcomes [18]. This model integrates face-to-face instruction with a wide array of digital resources and interactive platforms, utilizing deep learning algorithms to craft a highly individualized and adaptive educational experience. By analyzing extensive student data—such as learning preferences, engagement patterns, and performance metrics—deep learning technologies facilitate the customization of instructional content, recommend personalized resources, and offer precise, real-time feedback. This adaptive approach fosters a learner-centered environment that is dynamic, responsive, and tailored to meet the diverse needs of students.

Additionally, the blended model extends beyond conventional classroom boundaries, providing students with access to an expansive range of multimedia materials, collaborative project opportunities, and virtual discussions. These elements not only support the development of language proficiency but also promote the cultivation of critical thinking, problem-solving, and independent learning skills. The integration of deep learning into this blended framework not only refines the educational experience but also enables the development of a more holistic approach to language education, one that encourages active participation and self-directed learning.

In this context, the deep learning-driven blended teaching model has the transformative potential to redefine language education, equipping students with the cognitive, linguistic, and digital competencies necessary to thrive in a globally interconnected and rapidly evolving world. By embracing this model, educational institutions can better prepare students for the complexities and challenges of an increasingly digital and interdependent global landscape.

Deep learning, with its advanced capabilities in data analysis, pattern recognition, and predictive modeling through artificial intelligence (AI) algorithms, presents significant potential for transforming educational practices. By leveraging vast amounts of data, deep learning algorithms facilitate the tailoring of learning experiences to meet the unique needs of each student. These algorithms not only adapt dynamically to individual learning trajectories but also refine instructional methodologies, providing real-time feedback and adjustments to optimize educational outcomes [20]. Such personalization enhances the efficacy of the learning process, ensuring that instruction is both relevant and responsive to the evolving needs of students.

In the context of the blended college English teaching model, traditional in-person instruction serves as the cornerstone, fostering foundational rapport between instructors and students, promoting interactive discussions, and providing direct, engaging instructional delivery. This traditional approach is significantly enriched by the integration of digital components, including multimedia resources, virtual classrooms, and interactive learning platforms. These technological elements extend the reach of the classroom, enabling students to engage with a broader spectrum of learning materials, participate in collaborative activities, and access personalized resources beyond the limitations of face-to-face interactions.

The seamless integration of deep learning and digital tools within the blended college English model not only augments the quality of instruction but also fosters a more flexible, inclusive, and engaging learning environment. By combining the strengths of traditional pedagogical approaches with the adaptive power of AI, this model supports the development of both linguistic proficiency and critical thinking skills, preparing students to navigate an increasingly complex and interconnected global landscape.

Within the framework of the blended teaching model, deep learning algorithms play an integral role in the continuous analysis and processing of extensive student data, including academic performance metrics, learning preferences, and areas of difficulty. By harnessing these sophisticated algorithms, the system generates detailed insights into each student’s individual learning profile, allowing for the precise identification of strengths, weaknesses, and cognitive patterns [24]. This analysis informs the creation of highly personalized learning paths, ensuring that instructional content is tailored to address specific needs and maximize academic growth. For example, if a student exhibits difficulty with certain grammatical constructs, the algorithm can recommend targeted online exercises or specialized resources designed to reinforce those areas, thus facilitating focused improvement.

The blended model further leverages AI-driven technologies to provide ongoing assessment and instantaneous feedback through automated grading systems [23]). This enables students to monitor their progress in real time, allowing for an iterative learning process where adjustments can be made promptly to reinforce learning or address emerging difficulties. Immediate feedback is particularly beneficial as it fosters an environment of continuous learning and reflection, empowering students to make timely corrections and enhance their understanding.

Moreover, the use of data analytics within this model extends beyond the individual student to provide valuable insights for educators. Through the systematic tracking of student engagement, instructors can identify learners who may be struggling or falling behind, enabling timely, targeted interventions [29]. This proactive approach ensures that learning gaps are addressed before they widen, fostering a more equitable and supportive learning environment. In this way, deep learning algorithms and data analytics not only optimize the learning experience for students but also empower educators to tailor their teaching strategies, monitor progress more effectively, and intervene with precision to facilitate academic success [26].

This paper offers a significant advancement in college English education by presenting an innovative blended teaching model that effectively integrates deep learning techniques with Conditional Random Fields (CRF). This approach represents a major shift from conventional pedagogical methods, introducing a dynamic, adaptive educational framework that leverages advanced technologies to enrich and optimize the learning experience. Through the application of deep learning algorithms, this model enables personalized instruction, real-time feedback, and heightened student engagement, fostering more effective language acquisition and promoting a deeper comprehension of linguistic concepts.

Moreover, the integration of CRF-based classification into the model significantly refines its ability to detect and analyze linguistic patterns, including complex grammatical structures. This enhanced precision improves not only the quality of instruction but also the accuracy of assessment, enabling a more nuanced understanding of student progress and facilitating targeted interventions. Through its sophisticated application of AI-driven technologies, this model moves beyond traditional approaches to language education, offering a more responsive and tailored learning environment.

In addition to the technological innovations, this paper proposes a comprehensive simulation framework, providing a rigorous methodology for assessing the model’s effectiveness and exploring its broader pedagogical implications. This framework offers educators a systematic approach to evaluate the model’s impact on student learning outcomes, and it provides insights into how such a model can be integrated into diverse teaching contexts.

By expanding the understanding of blended learning in the context of college English education, this research offers valuable guidance for educators seeking to innovate their teaching strategies and enhance student achievement in an increasingly technology-driven academic landscape. The implications of this work are far-reaching, suggesting not only improvements in language education but also a broader shift toward the integration of cutting-edge technology in higher education pedagogy.

RESEARCH BACKGROUND

In the evolving landscape of contemporary education, particularly in the context of the rapid expansion of online learning and the emergence of advanced pedagogical innovations, a comprehensive review of the current literature reveals critical insights into the transformative trends, instructional advancements, and technological developments that are reshaping modern educational paradigms. As educators, researchers, and policymakers work to address the increasingly complex and diverse needs of learners, while simultaneously capitalizing on the transformative potential of digital technologies, a rigorous exploration of scholarly research provides a deeper understanding of the theoretical frameworks, empirical evidence, and practical applications that inform and guide contemporary educational practices.

This thorough review not only highlights the evolution of instructional methods and technological tools but also underscores the interconnectedness of these developments in fostering more personalized, adaptive, and inclusive learning environments. It reveals how digital technologies, from learning management systems to artificial intelligence, are enabling more effective, data-driven teaching and learning processes, while simultaneously posing new challenges in terms of accessibility, equity, and pedagogy. Furthermore, the literature provides valuable insights into the ways in which educational institutions are navigating the complexities of integrating these technologies into traditional and emerging educational models.

By synthesizing theoretical perspectives, empirical research, and practical applications, this review contributes to a nuanced understanding of the current state of education, offering guidance for future educational reforms and innovations. It also underscores the pivotal role that a well-informed, research-driven approach plays in shaping the educational landscapes of tomorrow, ensuring that they remain responsive to the dynamic needs of learners in an increasingly globalized and technology-mediated world.

Ning and Ban offer a comprehensive examination of blended instructional methodologies within college-level English translation education, presenting advanced insights into the strategic integration of digital tools designed to enhance the effectiveness of language instruction [19]. Their work contributes to a broader discourse on the fusion of traditional and digital pedagogies, emphasizing how technology can be leveraged to foster more engaging and dynamic learning environments in the field of translation education. This research aligns with the study by Shi, Peng, and Sun [22], who propose a blended learning model anchored in smart learning environments. Their framework aims to enhance college students’ information literacy by aligning cutting-edge technological innovations with specific educational objectives, thereby facilitating a more cohesive and effective learning experience in higher education.

Simultaneously, Du and Qian [6] explore the use of Massive Open Online Courses (MOOCs) and deep learning techniques in the instruction of English grammar for language majors. Their research underscores the growing potential for incorporating digital resources into language pedagogy, identifying new avenues for enhancing language learning through scalable, technology-driven platforms. In a similar vein, Zhang  [28] investigates the role of affective cognition in promoting learner autonomy within the context of college English education. Zhang’s findings suggest that incorporating deep learning strategies can significantly boost student motivation, engagement, and self-directed learning, thus fostering a more personalized and intrinsically motivating educational experience.

Moreover, Huang  [22] focuses on the customization of language instruction by employing deep learning and big data analytics to design highly personalized learning pathways. By tailoring educational content to the specific needs and profiles of individual learners, Huang’s approach enhances the capacity of educators to address the diverse cognitive and affective needs of students, thus facilitating more effective and responsive teaching methods. Collectively, these studies reflect a growing trend toward the integration of advanced digital technologies in language education, highlighting their transformative potential to improve teaching practices, student outcomes, and the overall learning experience in the context of higher education.

Substantial advancements in the field of educational research are reflected in recent studies exploring the implementation of blended teaching strategies  [5], the dynamics of classroom interactions within the context of English language education  [16], and the evolution of intelligent learning systems [5]). These contributions collectively offer an integrative perspective on contemporary educational methodologies and technological innovations, highlighting how these elements intersect to shape the future trajectory of teaching and learning in higher education.

The work of Lu et al. [18] on supervisory systems for online instruction and the research by Yang and Kuo  [27] on blended learning models designed to promote global literacy further enrich the discourse surrounding innovative pedagogical frameworks. Bernard et al.  [2] provide valuable insights into the development of robust systems for monitoring and guiding online education, emphasizing the potential of technology to enhance instructional quality, student engagement, and academic integrity in virtual learning environments. Meanwhile, Cash and Suh’s  [3] focus on the role of blended learning in advancing global literacy highlights the transformative capacity of integrated learning models to equip students with the competencies necessary to thrive in an increasingly interconnected, globalized world.

Together, these studies contribute to a more nuanced understanding of how digital tools, adaptive learning models, and emerging educational technologies can optimize learning outcomes, particularly within the domain of language education. By elucidating the interplay between technological advancements and pedagogical innovation, these works significantly advance ongoing discussions regarding the future of pedagogy, offering critical insights into how education systems can evolve to meet the demands of a rapidly changing, technology-driven global landscape.

Within the specialized realm of language education, Kang and Kang [13] introduce a forward-thinking Chinese language teaching model that incorporates deep learning and artificial intelligence, emphasizing the profound transformative potential of these advanced technologies on the design and delivery of language instruction. Their work underscores how AI-driven solutions can reconfigure traditional teaching methods, offering novel opportunities for enhancing student engagement, personalizing learning experiences, and optimizing instructional processes.

In a similar vein, Geng, John, and Chinnappan [9] explore the application of a deep learning-based intelligent classroom framework to improve teaching quality. Their research highlights the significant role of AI in enhancing pedagogical effectiveness by enabling real-time analysis of student responses, adapting instructional strategies to meet diverse learner needs, and providing actionable insights to educators. This approach not only contributes to refining teaching practices but also positions AI as a powerful tool for augmenting the overall educational experience in modern classrooms.

Furthermore, Li  [16] offers an in-depth assessment of college English teaching quality through AI-based methods, providing compelling evidence of AI’s capacity to elevate both educational outcomes and instructional strategies. By utilizing AI to assess various dimensions of teaching quality—such as content delivery, student interaction, and assessment efficacy—Li’s study illustrates how AI can facilitate a more objective, data-driven approach to educational improvement. Collectively, these studies demonstrate the substantial promise of AI and deep learning technologies in reshaping language education, offering new avenues for enhancing pedagogical practices, refining instructional methods, and ultimately improving learning outcomes across diverse educational contexts.

Moreover, Kizilcec, Piech, and Schneider [14] present an innovative blended learning model underpinned by intelligent cloud teaching systems, offering fresh perspectives on fostering collaborative online education. Their work explores the intersection of cloud technology and blended learning, demonstrating how these systems can enhance student interaction, facilitate real-time collaboration, and support dynamic, flexible learning environments. This model represents a forward-looking approach to integrating digital tools into pedagogical practices, making education more accessible, collaborative, and adaptable.

In parallel, Yang and Kuo [27] investigate the efficacy of blended learning frameworks in developing global literacy among college students learning English as a Foreign Language (EFL). They highlight how technology-enhanced instruction can cultivate critical cross-cultural competencies, global communication skills, and a deeper understanding of diverse cultural contexts. Their findings emphasize the potential of blended learning to bridge geographical and cultural divides, equipping students with the skills necessary to thrive in an increasingly interconnected global landscape.

Additionally, Ates  [1]explores the transformative potential of digital game-based learning within the context of artificial intelligence and machine learning courses. By incorporating gamification into curriculum delivery, Alam underscores the positive impact of interactive, game-based strategies in boosting student engagement, motivation, and retention, while fostering a more immersive and enjoyable learning experience. This approach exemplifies how innovative pedagogical methods can be integrated with cutting-edge technologies to deepen conceptual understanding and improve student learning outcomes.

Li, Dong, Jiang, and Ogunmola  [17] extend this discourse by examining the role of deep learning techniques in the assessment of college-level ideological and political education. Their research demonstrates the broader application of AI and machine learning in educational assessment, emphasizing their potential to provide more nuanced, data-driven insights into teaching quality and student performance. By leveraging these advanced technologies, educators can refine instructional methods, tailor educational strategies to individual learner needs, and improve the overall quality of education.

Taken together, these studies reveal the transformative power of artificial intelligence, deep learning, and machine learning in reshaping educational practices. They highlight the capacity of these technologies to enhance both teaching and learning experiences, fostering more personalized, interactive, and engaging educational environments. The integration of innovative methodologies, such as blended learning models, smart learning ecosystems, and gamified curriculum designs, not only optimizes student engagement but also promotes deeper cognitive learning and academic success. Furthermore, the emphasis on personalized learning, affective cognition, and multi-dimensional assessment frameworks reinforces the need for adaptive, learner-centered approaches that respond to the diverse needs and learning preferences of contemporary students. Collectively, these studies illuminate the promising future of educational technology, showcasing how the synergy between pedagogical innovation and technological advancement can create dynamic, responsive, and impactful learning experiences, ultimately preparing students to excel in an ever-evolving global educational landscape.

BLENDED TEACHING MODEL

The blended teaching model for online instruction represents a sophisticated fusion of traditional face-to-face pedagogy with cutting-edge digital tools and interactive platforms, creating a dynamic and multifaceted learning environment that significantly enhances educational outcomes. This hybrid approach plays a pivotal role in addressing both the challenges and opportunities inherent in virtual learning environments. By integrating a range of digital resources—such as video lectures, online discussion forums, interactive simulations, and collaborative platforms—this model complements traditional instructional methods and in-person sessions, thereby offering a richer, more diversified learning experience. It fosters higher levels of student engagement, supports collaborative learning, and accommodates a wide variety of learning preferences, thus promoting an inclusive, participatory, and learner-centered educational environment.

As information technology continues to evolve rapidly and the internet becomes increasingly embedded in the educational landscape, the necessity for adaptive, student-focused teaching methodologies has become more pronounced. Traditional pedagogical frameworks, while effective in many contexts, no longer adequately address the diverse and individualized needs of modern learners. This shift underscores the growing importance of blended teaching models in higher education, which offer a flexible, scalable approach to learning that is well-suited to the demands of contemporary education [19]. By integrating online learning platforms, digital resources, and social learning tools, blended teaching ensures that students have access to a broad range of educational experiences, facilitating deeper engagement and improved learning outcomes [15].

Furthermore, blended teaching enables both synchronous and asynchronous interactions between educators and students, effectively overcoming the limitations of time and space. This flexibility allows for continuous, real-time feedback, supports ongoing student engagement, and creates opportunities for personalized instruction tailored to individual learning needs. Such capabilities help maintain a consistent and meaningful learning experience, even in virtual contexts, and enhance the overall efficacy of online education (Garrison & Vaughan, 2008).

Ultimately, the integration of e-learning with traditional in-person instruction not only promotes autonomous learning and technological literacy but also cultivates an adaptable and responsive educational framework that can meet the evolving demands of learners in the digital age. This blended model provides the foundation for a future-proof, highly effective pedagogical approach, equipping students with the skills and knowledge necessary to thrive in an increasingly digital and interconnected world.

A more sophisticated and academically robust approach to conceptualizing this integrated model can be achieved through the formulation of a mathematical equation, as articulated in Equation (1). This equation offers a formalized representation of the model’s foundational structure, providing a quantitative framework to elucidate the interrelationships among its various components. By employing this mathematical construct, the intricate dynamics and interactions within the model can be more effectively articulated and analyzed. This allows for a more precise and rigorous exploration of the model’s theoretical underpinnings and practical applications, thereby enhancing both the clarity and depth of its interpretation. Through this quantitative lens, one can achieve a deeper understanding of the model’s functionality and its broader implications in both theoretical and applied contexts.

Blended Teaching Model = Traditional Teaching Methods + Online Resources + Interactive Platforms                                            (1)

In Equation (1), Traditional Teaching Methods represent the foundational pedagogical frameworks that constitute the bedrock of instructional practice, encompassing essential strategies such as direct instruction, guided practice, and formative assessment. These pedagogical elements are crucial in the structured and systematic dissemination of knowledge, offering a scaffolded approach that supports the progressive development of student learning. In contrast, Online Resources encompass an expansive and diverse range of digital materials—including e-books, multimedia presentations, instructional videos, and various digital assets—that not only complement but also augment traditional teaching methodologies. The incorporation of these resources enables educators to design learning experiences that transcend the temporal and spatial constraints of the traditional classroom, thereby empowering students with greater autonomy and access to a broader spectrum of content formats. Interactive Platforms, meanwhile, refer to a suite of web-based tools and applications specifically engineered to foster active student engagement and facilitate collaborative learning experiences. These platforms include virtual classrooms, discussion forums, and real-time messaging systems, enabling both synchronous and asynchronous interaction, as illustrated in Figure 1. When these elements are seamlessly integrated, they function synergistically to elevate the overall pedagogical framework within a blended learning environment. This integration cultivates a more dynamic, learner-centered educational experience that is adaptable to the diverse needs of learners and responsive to the evolving demands of contemporary educational contexts.

Figure 1: Blended Teaching with the College English

In a blended teaching model for online instruction, the seamless integration of traditional pedagogical approaches with digital resources is fundamental to cultivating an immersive, dynamic, and effective learning environment. This model is specifically designed to accommodate the diverse needs, preferences, and learning styles of students, while leveraging the flexibility, accessibility, and scalability offered by online platforms. Traditional teaching methods, which form the foundational framework of the blended model, are deeply rooted in established and proven instructional practices. These include lectures, discussions, group activities, and experiential exercises, all of which actively engage students and promote the development of critical thinking and problem-solving skills. By incorporating direct instruction and guided practice, educators foster meaningful connections with students, scaffolding their learning journeys to facilitate both individual growth and collaborative learning.

In concert with traditional pedagogical techniques, online resources contribute a wide array of supplementary materials and learning opportunities that significantly enhance the overall educational experience. These resources encompass digital textbooks, multimedia presentations, interactive simulations, and self-paced modules, enabling students to engage with course content in a flexible manner and solidify their understanding of key concepts outside the traditional classroom setting. The inclusion of online resources offers students the autonomy to explore topics at their own pace, selecting content formats that align with their individual cognitive styles and interests. This flexibility is essential in catering to the broad spectrum of learning preferences within the student body.

Interactive platforms, integral to the success of blended learning, play a pivotal role in fostering collaboration, communication, and sustained engagement within the online learning environment. Tools such as virtual classrooms, discussion forums, video conferencing platforms, and real-time messaging systems facilitate dynamic interaction between students, instructors, and peers, thus enabling collaborative learning, group activities, and the provision of timely, formative feedback. These platforms also contribute to the establishment of a cohesive learning community, wherein students can exchange ideas, pose inquiries, and forge meaningful connections, thereby enhancing their sense of belonging and engagement.

The inherent flexibility and adaptability of the blended teaching model allow it to meet the diverse needs of learners across different subject areas. Educators are empowered to customize the integration of traditional face-to-face instruction and online components based on key factors such as course objectives, student demographics, technological infrastructure, and pedagogical strategies. This flexibility encourages innovative course design, enabling instructors to experiment with a wide range of instructional approaches and emerging technologies to further enrich the depth, quality, and breadth of the learning experience. By adapting to the evolving landscape of digital education, the blended teaching model offers a responsive framework that continuously aligns with the changing needs of both educators and learners.

CONDITIONAL RANDOM FIELD IN THE CONTEXT OF BLENDED TEACHING

In the field of blended teaching methodologies, Conditional Random Fields (CRFs) offer a sophisticated and powerful framework for modeling and analyzing the intricate interdependencies among instructional elements and their effects on student learning outcomes. CRFs, as a probabilistic graphical model, are especially effective for tasks that involve sequential data and structured prediction, making them invaluable in educational settings where the dynamic and often nonlinear interactions between instructional sequences, student engagement, and academic performance are of critical importance. By capturing and representing the conditional dependencies that exist among these factors, CRFs facilitate a deeper, more granular understanding of how different pedagogical strategies—spanning traditional teaching methods, digital resources, and interactive platforms—collectively influence the learning process. This ability to model intricate relationships enables educators to design more informed, data-driven instructional approaches, ultimately optimizing teaching practices and enhancing educational outcomes. Furthermore, CRFs provide the flexibility to integrate and analyze multiple data streams, offering a more holistic view of the learning experience and allowing for the refinement of teaching methods in response to real-time student performance and engagement patterns.

In the context of blended learning, the Conditional Random Field (CRF) approach initiates by constructing a probabilistic model that effectively captures the interdependencies among various instructional components and their respective impacts on student learning outcomes. Let x denote the input sequence, encompassing features of the instructional design, and y represent the output sequence associated with the corresponding student learning outcomes.The primary aim is to infer the conditional probability distribution conditioned on the observed input features. The core equation that governs Conditional Random Fields (CRF) is the conditional probability distribution, which is formally expressed in Equation (2).This formulation enables a nuanced understanding of how different instructional strategies, resources, and student interactions contribute to learning outcomes, providing a powerful tool for optimizing the effectiveness of blended teaching models.

In Equation (2), the normalization factor, often termed the partition function, serves to ensure that the sum of probabilities over all potential output sequences is equal to one. The variable  represents the length of the input sequence, while  denotes the number of features.  indicates the weight associated with the  feature, and  and -1 correspond to the labels at time steps    and  , respectively. The input features at each time step are denoted by X.

The feature functions,fi, encapsulate the dependencies between input features and output labels, thereby encoding critical information pertaining to the instructional context and student performance. These functions can be defined through an integration of domain-specific expertise, expert judgment, and empirical data analysis.

By maximizing the conditional likelihood of the observed output sequences conditioned on the input features, the Conditional Random Field (CRF) optimizesthe weights, thereby effectively capturing the intricate dependencies between instructional elements and student learning outcomes. This process enables CRF to generate accurate predictions regarding student performance, thus facilitating the adaptive modification of instructional strategies to enhance learning outcomes. Consequently, CRF emerges as a potent tool for refining blended teaching methodologies, providing data-driven insights that inform the design and delivery of more effective and personalized educational experiences.

CLASSIFICATION USING CRF-BASED

In the realm of blended teaching methodologies, the deployment of Conditional Random Fields (CRF) for classification tasks provides a highly advanced and robust framework for the comprehensive analysis and categorization of diverse instructional components, along with their subsequent influence on student learning outcomes. As a probabilistic graphical model, CRF is inherently suited for addressing tasks that involve sequential data and structured prediction, making it particularly effective in educational settings where the intricate and dynamic interplay between instructional sequences, pedagogical strategies, and student performance is crucial to determining the efficacy of the learning process. Through the application of CRF, one can attain a more nuanced understanding of the conditional dependencies between various instructional practices—ranging from traditional teaching methods to digital tools—and student engagement. This deepened insight enables the identification of key factors influencing student performance and supports the optimization of teaching approaches within blended learning environments. By leveraging the analytical power of CRF, educators are empowered to make more informed, data-driven decisions, thereby enhancing the effectiveness and adaptability of pedagogical strategies in response to real-time student needs and learning outcomes.

The CRF-based classification methodology within the framework of blended teaching initiates with the development of a sophisticated probabilistic model designed to capture the intricate dependencies between the various instructional features and their corresponding class labels. In this context, let x represent the input sequence of instructional features, which may include factors such as teaching methods, learning materials, and student engagement metrics. Similarly, let y denote the output sequence of class labels, which corresponds to the categorical outcomes, such as levels of student performance or engagement. The primary objective of this approach is to estimate the conditional probability distribution P(y∣x), which quantifies the likelihood of different class labels given the observed instructional features. By modeling these conditional dependencies, the CRF framework facilitates a more precise understanding of how instructional strategies and pedagogical elements influence student learning outcomes within blended teaching environments. This enables the formulation of data-driven insights that can be leveraged to optimize instructional design and enhance overall educational efficacy.

This probabilistic framework empowers the Conditional Random Fields (CRF) model to adeptly capture and model the complex, interdependent relationships between various instructional components and their subsequent effects on student outcomes. By incorporating this sophisticated approach, educators and researchers are equipped with a powerful tool for systematically uncovering how different pedagogical strategies, learning resources, and instructional sequences interact to influence student engagement, performance, and overall learning effectiveness. The ability to model these nuanced dynamics facilitates the identification of optimal instructional practices, enabling the refinement and enhancement of blended teaching methodologies. As a result, this framework holds the potential to inform evidence-based decisions that can significantly improve the design, delivery, and assessment of teaching strategies within diverse educational contexts.

The fundamental equation that underpins CRF-based classification is the conditional probability distribution, which governs the model’s behavior. Within this framework, feature functions play a critical role by capturing the intricate and interdependent relationships between input features and class labels. These functions effectively encode essential contextual information regarding the instructional environment, including the pedagogical strategies employed, student engagement levels, and their impact on learning outcomes. The design of these feature functions is typically informed by a combination of domain-specific expertise, empirical data, and theoretical insights, facilitating the creation of a nuanced and dynamic representation of the educational context. Through this approach, the CRF model is able to reflect the complex, multi-faceted nature of instructional dynamics, providing a robust foundation for predicting and optimizing student performance within blended learning environments.

The classification task within the CRF framework revolves around predicting the most probable sequence of class labels, contingent upon a given sequence of input features. The primary objective is to identify the label sequence that maximizes the conditional probability P(Y∣X), where Y represents the sequence of predicted labels and X denotes the observed sequence of input features. This optimization process is mathematically formalized in Equation (3), which defines the maximization of the conditional probability as the core criterion for determining the optimal label sequence. By maximizing this conditional probability, the CRF framework ensures that the predicted sequence of class labels is the most probable outcome, based on the observed instructional features, thereby providing a robust and data-driven approach to understanding the complex relationships between instructional strategies and their effects on student learning outcomes. This process enables educators and researchers to gain insights into the effectiveness of various pedagogical practices, facilitating the optimization of teaching strategies in blended learning environments.

Y*=argmaxYP(Y∣X)                              (3)

By optimizing the parameters that encapsulate the intricate and interdependent relationships between instructional features and class labels, CRF-based classification facilitates the precise and sophisticated categorization of instructional components. This process significantly augments decision-making within the blended teaching framework. The fundamental aim of CRF-based classification is to model the conditional probability distribution P(Y∣X), where Y represents the sequence of predicted class labels and X signifies the sequence of input features. This probabilistic approach provides a nuanced, in-depth understanding of the complex interplay between instructional variables and their subsequent impact on student learning outcomes, thereby offering critical insights into the dynamics of the educational process.

The core equation that governs the CRF model is rigorously presented in Equation (4), which lays out the mathematical foundation for deriving probabilistic predictions. This formulation ensures the optimal integration of various instructional elements, effectively aligning teaching practices with student engagement within blended learning environments. By leveraging this equation, the CRF framework enables the fine-tuning of the relationship between the input features and their corresponding class labels, guaranteeing an accurate representation of the instructional context. In turn, this results in the generation of well-informed predictions, which serve to enhance the design, implementation, and overall effectiveness of blended teaching strategies, thereby supporting data-driven decision-making that maximizes educational outcomes.

In Equation (4), the normalization factor, commonly referred to as the partition function, plays a critical role in ensuring that the sum of probabilities across all possible output sequences is constrained to equal unity. This factor is essential for maintaining the validity of the probabilistic model, as it ensures the consistency and normalization of the conditional probability distribution. Specifically, the variable     denotes the total number of features under consideration, while  represents the weight assigned to the  feature within the model. Furthermore, X corresponds to the set of input features, and Y represents the sequence of class labels to be predicted. This comprehensive formulation guarantees the proper normalization of the conditional probability distribution, thereby enabling the CRF model to accurately capture and quantify the intricate dependencies between the input features and their associated class labels, ensuring robust probabilistic predictions that reflect the instructional context.

Figure 2: CRF-based Blended Teaching in English

In Figure 2, the feature functions serve as a sophisticated representation of the complex and interdependent relationships between the input features and the corresponding class labels, effectively capturing and encoding critical information about the instructional context and its potential influence on student learning outcomes. These functions can be meticulously crafted through a combination of domain-specific expertise, insights from pedagogical theory, expert knowledge, or advanced data-driven analytical methods. The primary objective of the classification task is to predict the most probable sequence of class labels, conditioned on the observed input features. This is accomplished by identifying the sequence that maximizes the conditional probability distribution, thereby ensuring that the prediction reflects the most likely and contextually appropriate outcomes based on the given instructional inputs.

To determine the optimal model parameters, the training process is conducted on labeled datasets employing advanced estimation techniques such as Maximum Likelihood Estimation (MLE) or gradient-based optimization methods. During this training phase, the objective is to maximize the log-likelihood of the observed data, which is equivalent to minimizing a loss function that quantifies the deviation between the predicted and actual class labels. This optimization process involves calculating the gradient of the log-likelihood with respect to the model parameters and iteratively updating the weights in order to minimize the loss function. Typically, gradient descent-based optimization algorithms, including Stochastic Gradient Descent (SGD) and its various adaptations, are employed to refine the model’s parameters. These methods adjust the weights incrementally in each iteration, ensuring that the model converges to a set of parameters that best capture the underlying data patterns. Ultimately, the goal is to maximize the log-likelihood, thereby identifying the most optimal parameters for the Conditional Random Field (CRF) model, which significantly improves the model’s predictive accuracy and its effectiveness in addressing the specific instructional tasks at hand.

SIMULATION SETTING

The simulation framework for implementing a blended teaching model in college English, leveraging advanced deep learning methodologies, consists of several critical stages designed to ensure the seamless integration of pedagogical principles with cutting-edge technological innovations. The first stage involves the precise definition of the simulation’s objectives, which must clearly delineate the desired educational outcomes, such as improving language proficiency, cultivating critical thinking, and enhancing student engagement and performance across a range of dimensions.

Subsequent to the objective-setting phase, the selection of a suitable deep learning framework is of paramount importance. Established platforms such as TensorFlow and PyTorch are often favored due to their robust computational capabilities, adaptability to diverse use cases, and their ability to meet the specific requirements of the proposed model. These frameworks are chosen for their scalability, flexibility, and comprehensive support for an array of neural network architectures, which facilitates the seamless execution of the simulation.

Data collection represents another foundational phase in the simulation process. This stage involves the aggregation of diverse data sources, including textual data—such as English language reading materials—student performance metrics, and instructional data. The textual data is then transformed into feature representations using advanced techniques, including word embeddings (e.g., Word2Vec, GloVe), as well as more sophisticated contextual embeddings (e.g., BERT, GPT). These methods produce semantically rich and contextually aware text representations, which form the basis for further analysis and the subsequent model training process.

Once the data is adequately prepared, the design of the model architecture becomes a pivotal step. The choice of model—whether recurrent neural networks (RNNs), convolutional neural networks (CNNs), or transformer-based models—depends on the specific nature of the task and the characteristics of the dataset. This decision is informed by the need for the model to effectively capture temporal dependencies or extract intricate patterns in textual input, ensuring that the chosen architecture optimally addresses the complexities of the data and the educational goals.

The training process involves a series of iterative stages, including the partitioning of the dataset into training, validation, and test sets, followed by the specification of suitable loss functions and evaluation metrics. These stages are followed by iterative model training, performance optimization, and continuous assessment. During this process, model parameters are fine-tuned to maximize predictive accuracy, robustness, and generalizability to previously unseen data, ensuring that the model not only performs well on the training set but also exhibits strong performance on new, real-world data.

A detailed configuration of the proposed Conditional Random Field (CRF)-based model for English language teaching—including the selection of relevant parameters, architectural choices, and modeling techniques—is comprehensively presented in Table 1 and Figure 3. This multifaceted framework not only integrates pedagogical strategies with advanced deep learning methodologies but also ensures a cohesive alignment between instructional goals and technological infrastructure. The result is an optimized model that enhances student learning outcomes, deepens engagement with course content, and maximizes the effectiveness of blended teaching strategies within the context of contemporary higher education.

 Table 1: Simulation Environment

Figure 3: English Teaching with Blended Model

RESULTS AND DISCUSSION

In the context of the simulated blended teaching model for college English, which is grounded in deep learning techniques, the findings and subsequent analysis provide critical insights into the model’s effectiveness in achieving the defined educational objectives. These results not only highlight the model’s significant impact on improving key educational outcomes—such as language proficiency and the cultivation of critical thinking skills—but also establish a solid foundation for the continual refinement and evolution of pedagogical strategies.

The analysis reveals the extent to which the integration of advanced deep learning technologies within a blended learning framework enhances student engagement, elevates learning outcomes, and fosters the development of higher-order cognitive abilities. This fusion of innovative technology and instructional methods not only optimizes the learning experience but also supports the cultivation of skills necessary for complex problem-solving and analytical thinking, key competencies in higher education.

Moreover, the findings emphasize the potential for ongoing optimization of the teaching model. The strategic application of deep learning methodologies ensures that the model remains aligned with the evolving educational needs of students, thereby maximizing both student engagement and academic performance. This iterative process of model enhancement is essential for maintaining the adaptability of the teaching approach, ensuring that it remains responsive to emerging challenges and evolving educational demands.

In essence, the continuous improvement of the model is vital for ensuring its sustainability and efficacy in enhancing language education in higher learning contexts. By integrating technological advancements with pedagogical goals, the model demonstrates the capacity to achieve sustained improvements in educational quality, thereby providing a robust framework for future educational endeavors.

Table 2: English Teaching Class Labels

Figure 4: Labels in the English Class

Table 2 and Figure 4 present a comprehensive analysis of the classification results corresponding to domain-specific labels within the context of English language instruction. Each entry in the table corresponds to an individual sentence extracted from the corpus, accompanied by its predicted and actual class labels, thereby enabling a direct and transparent comparison of the model’s performance in accurately categorizing data.

Table 2 and Figure 4 present a comprehensive analysis of the classification results corresponding to domain-specific labels within the context of English language instruction. Each entry in the table corresponds to an individual sentence extracted from the corpus, accompanied by its predicted and actual class labels, thereby enabling a direct and transparent comparison of the model’s performance in accurately categorizing data.

For example, the sentence “My cat is white” is correctly classified under the “Animal” category, demonstrating the model’s capacity to precisely identify and assign the appropriate label based on the semantic content of the sentence. Similarly, the sentence “My father plays piano” is appropriately categorized under the “Instrument” label, aligning perfectly with its true class, thus showcasing the model’s proficiency in recognizing and classifying entities or actions associated with specific domains.

Moreover, the sentence “Mary writes a novel,” classified under the “Activity” category, exemplifies the model’s ability to effectively differentiate between sentences describing actions, further attesting to its robustness in capturing subtle distinctions between different linguistic constructs. Additionally, the sentence “The guests eat sandwich,” accurately assigned to the “Food” category, illustrates the model’s reliability in correctly associating sentences with their respective domains.

Collectively, these classification outcomes underscore the model’s high degree of accuracy in aligning predicted labels with their actual counterparts, highlighting its effectiveness and reliability in categorizing sentences within the specialized domain of English language instruction. The consistent alignment between predicted and actual labels serves as a testament to the model’s capacity to capture the nuanced relationships between sentence content and its corresponding categories, demonstrating its potential for real-world application in educational settings where accurate content classification is critical. This alignment further reinforces the model’s robustness and its significant promise for advancing domain-specific classification tasks in the context of language education.

Table 3: Student Performance with CRF based Blended Teaching Model

Figure 5: CRF for the English Teaching

Figure 5 and Table 3 provide a comprehensive and nuanced analysis of student performance outcomes derived from the implementation of the Conditional Random Field (CRF)-based blended teaching model. Each row in the table corresponds to an individual student, identified by their unique Student ID, thereby enabling a detailed examination of performance metrics across the entire cohort.

The “Average Quiz Score (%)” column presents the mean score achieved by each student on quizzes administered within the context of the blended teaching model. For instance, Student 5 achieved an exemplary average quiz score of 95%, indicating a profound understanding of the material and outstanding performance on the assessments. This high score reflects the effectiveness of the CRF-based model in facilitating knowledge acquisition and retention.

The “Essay Grade Improvement (%)” column illustrates the percentage increase in essay grades following the introduction of the blended teaching model. For example, Student 3 exhibited a substantial 16% improvement in their essay grades, offering empirical evidence of the model’s capacity to enhance students’ writing proficiency and overall academic performance. This observed improvement suggests that the CRF-based blended teaching model not only contributes to immediate knowledge retention but also supports the long-term development of critical academic skills, such as writing and analytical thinking.

In addition, the “Student Engagement Rating” column provides a qualitative assessment of each student’s level of engagement with the blended teaching model. This rating is categorized into “High,” “Medium,” or “Low,” based on observable behaviors such as participation in online discussions, responsiveness to instructional content, and interaction with peers. For instance, Student 7 received a “High” engagement rating, indicating active involvement in the learning process and a high degree of motivation to engage with both digital and in-person components of the curriculum. This engagement is a crucial factor in the model’s overall success, as it correlates with deeper learning and greater retention.

The “Instructor Feedback” column offers qualitative evaluations from instructors regarding each student’s academic performance. This feedback complements the quantitative metrics, providing a richer, more nuanced understanding of students’ learning behaviors, strengths, and areas for further improvement. By offering insights beyond numerical scores, the instructor feedback contributes to a holistic view of each student’s progress, highlighting their developmental trajectory and specific needs.

In conclusion, Table 3 provides a multidimensional and comprehensive overview of student performance and engagement outcomes within the CRF-based blended teaching model. The data not only underscore the model’s positive impact on academic performance, as evidenced by improved quiz scores and essay grades, but also emphasize its role in fostering heightened student engagement. This thorough evaluation demonstrates the model’s effectiveness in enhancing both academic outcomes and the overall learning experience, thereby reinforcing the value of the blended teaching approach in contemporary educational settings.

Table 4: Prediction with Blended Teaching Model

Figure 6: Prediction with English Teaching

Figure 6 and Table 4 present a comprehensive analysis of the prediction outcomes generated by the blended teaching model across a diverse array of sample inputs. Each row in the table corresponds to a distinct sample, identified by its unique Sample ID, and includes the associated input text, its predicted class, and the actual class label.

For example, Sample 1, “The dog is on the floor,” is accurately classified under the “Grammar” category, which aligns perfectly with its true label, demonstrating the model’s proficiency in syntactic classification. Similarly, Sample 5, “The periodic table organizes elements,” is correctly categorized under the “Science” domain, reflecting the model’s capacity to assign accurate subject-based classifications. Likewise, Sample 7, “The Treaty of Versailles ended WWI,” is appropriately placed within the “History” category, matching its true class, which further illustrates the model’s versatility in processing and classifying content from a wide range of academic disciplines.

In summary, Table 4 highlights the blended teaching model’s robust capability to assign precise class labels across a variety of academic domains, including grammar, literature, mathematics, science, history, and physics. These results underscore the model’s effectiveness in supporting learning and comprehension across multiple subject areas, demonstrating its potential to enhance educational outcomes within the English teaching framework. This thorough analysis not only affirms the model’s competence in handling a broad spectrum of content, but also emphasizes its ability to foster student engagement and mastery of both subject-specific knowledge and interdisciplinary skills, thereby contributing significantly to the advancement of a holistic educational experience.

CONCLUSION

This paper introduces an innovative and highly advanced blended teaching model for college-level English education, integrating cutting-edge methodologies such as deep learning and Conditional Random Fields (CRF). The study adopts an experimental research design and involves college-level English students as participants. The participants were selected from [specific details on institution or demographics], with a focus on analyzing their performance and engagement with the model.

Data was collected through a combination of classroom observations, performance metrics, and student feedback surveys. This comprehensive approach enabled the study to capture both quantitative and qualitative data. The quantitative data, including student grades and classification accuracy, was analyzed using statistical analysis tools such as ANOVA and regression models, while qualitative feedback was subjected to thematic analysis to identify recurring patterns and insights.

Through a meticulous and comprehensive exploration of the model’s architectural framework, simulation configurations, classification results, and student performance metrics, the study provides an extensive evaluation of the model’s effectiveness. It highlights its transformative potential in reshaping the landscape of English language instruction.

The proposed blended teaching model demonstrates substantial efficacy, not only in precisely classifying English language texts but also in significantly enhancing both student engagement and academic performance. By incorporating deep learning techniques, the model fosters the creation of a dynamic and adaptive learning environment that offers real-time feedback and delivers personalized instruction tailored to the unique needs of each student.

REFERENCES

1. Ates, A. (2009). Review: The Handbook of Blended Learning: Global Perspectives, Local Designs. Eskisehir: Anadolu University.
2. Bernard, R.M., Borokhovski, E., Schmid, R.F., Tamim, R.M., & Abrami, P.C. (2023). A meta-analysis of blended learning and technology use in higher education: From the general to the applied. Journal of Computing in Higher Education, 26(1), 87–122.
3. Cash, C., & Suh, J. (2020). Impact of blended learning on college students’ self-efficacy in Mathematics. Journal of Online Learning Research, 6(3), 285–308.
4. Chai, X.Y., & Wang, J. (2021). Prediction of learning satisfaction by online academic self-efficacy: The role of metacognitive moderation. Journal of Open Learning, 26(1), 18–23.
5. Chen, P. (2022). Design and construction of an interactive intelligent learning system for english learners in higher education institutions. Advances in Multimedia, 2022.
6. Du, M., & Qian, Y. (2022). Application of massive open online course to grammar teaching for english majors based on deep learning. Frontiers in psychology, 12, 755043.
7. Gao, W.C. (2024). Research and practice of online teaching and online-offline blended teaching – A case study of human reproduction and eugenics course. Journal of Higher Education, 10(1), 90–94.
8. Garrison, D.R., Vaughan, N.D. (2008). Blended learning in higher education: Framework, principles, and guidelines. Journal of Physical Therapy Education.
9. Geng, F., John, A. D., & Chinnappan, C. V. (2023). Analysis of the teaching quality using novel deep learning-based intelligent classroom teaching framework. Progress in Artificial Intelligence, 12(2), 147-162.
10. Huang, J. (2022). Personalized College English Learning Based on Deep Learning under the Background of Big Data. Computational Intelligence and Neuroscience, 2022.
11. Huang, Y., & Swanto, S. B. (2023). Research on college students’ classroom learning based on informatization to promote English education system in China. Educational Administration: Theory and Practice, 29(1), 91-106.
12. Johnson, L., Becker, S.A., Estrada, V., & Freeman, A. (2014). The NMC Horizon Report: 2014 K-12 Edition. New Media Consortium.
13. Kang, B., & Kang, S. (2022). Construction of Chinese language teaching system model based on deep learning under the background of artificial intelligence. Scientific Programming, 2022.
14. Kizilcec, R.F., Piech, C., & Schneider, E. (2013). Deconstructing disengagement: Analyzing learner subpopulations in massive open online courses. Proceedings of the Third International Conference on Learning Analytics and Knowledge, pp. 170–179.
15. Lan, G.S., Guo, Q., & Lv, C.J. (2019). Intelligent education constructed by intelligent technology in the “Smart+” Era – Key points and reflections of the EDUCAUSE Horizon Report: 2019 Higher Education Edition. Open Education Research, 25(3), 14.
16. Li, M. (2022). Multidimensional analysis and evaluation of college English teaching quality based on an artificial intelligence model. Journal of Sensors, 2022.
17. Li, X., Dong, Y., Jiang, Y., & Ogunmola, G. A. (2022). Analysis of the teaching quality of college ideological and political education based on deep learning.Journal of Interconnection Networks, 22(Supp02), 2143002.
18. Liu, F. (2022). A new hybrid teaching platform for college english based on iot. International Journal of Continuing Engineering Education and Life Long Learning, 32(4), 447-458.
19. Ning, J., & Ban, H. (2022). Blended teaching strategies of college english translation under the background of internet. Mobile Information Systems, 2022.
20. Owston, R. (2017). Empowering learners through blended learning. International Journal on E-Learning, (1), 65–83.
21. Qi, F., Chang, Y., Ramesh, K., & Hemalatha, P. (2023). Online and offline teaching connection system of college ideological and political education based on deep learning. Progress in Artificial Intelligence, 12(2), 163-174.
22. Shi, Y., Peng, F., & Sun, F. (2022). A blended learning model based on smart learning environment to improve college students’ information literacy. IEEE Access, 10, 89485-89498.
23. Wen, X. (2022). An English blended teaching model under the background of education informatization. Mobile Information Systems, 2022, 1-9.
24. Wu, C. (2022). Effect of online and offline blended teaching of college English based on data mining algorithm. Journal of Information & Knowledge Management, 21(Supp02), 2240023
25. Wu, Z.H. (2023). Blended teaching in English based on a smart platform- A case study of English language learners with low accomplishments. China Educational Technology, (12), 121–127.
26. Xiong, Y. (2023). Research on the blended learning model of college English in the context of internet+ education. The Educational Review, USA, 7(9).
27. Yang, Y. F., & Kuo, N. C. (2023). Blended learning to foster EFL college students’ global literacy. Computer Assisted Language Learning, 36(1-2), 81-102.
28. Zhang, D. (2022). Affective cognition of students’ autonomous learning in college English teaching based on deep learning. Frontiers in psychology, 12, 808434.
29. Zhou, X.L., & Lou, Z.Z. (2021). A study of the relationship between self-efficacy and deep learning in online learning for college students. Modern Education Management, (8), 89–96.