Web-based learning (WBL), particularly web-based collaborative learning, offers advantages over traditional learning by overcoming time and place constraints and enabling peer interaction. While research highlights the value of WBL and the influence of students' beliefs on their learning practices, existing studies often lack a qualitative, longitudinal perspective. This study addresses this gap by employing a qualitative case study approach to explore the dynamic relationship between university EFL student readers' beliefs about web-based collaborative learning and their actual practices over a four-month period. The research aims to provide a nuanced understanding of how these beliefs evolve and impact students' engagement with WBL, contributing to better pedagogical practices in this area.

The literature review establishes the theoretical framework by examining learners' beliefs as a factor-sensitive construct interacting with practice. Students' beliefs, understood as evaluative stances toward learning activities, are shaped by past and ongoing experiences and can be positive, negative, or mixed. Positive beliefs generally drive practices, while negative beliefs constrain them. However, this relationship is dynamic and influenced by contextual factors. The review then examines existing research on WBL, noting a predominance of quantitative studies focusing on correlations between beliefs and learning outcomes at a single point in time. Qualitative studies are often limited in data sources or time span. This study aims to address these limitations by employing a comprehensive qualitative approach with multiple data sources and a longer timeframe.

This study adopted a qualitative case study approach, focusing on an EFL reading course with 20 female university students in China. The course incorporated web-based collaborative learning using a domestically developed platform similar to Google Docs. Students engaged in two types of activities: posing questions related to their pre-class reading and helping each other answer these questions. Data collection involved three rounds of semi-structured interviews (conducted in Chinese), two reflective essays from each student, and researcher field notes on students' online activities. Data analysis used inductive and deductive thematic analysis, focusing on linguistic resources expressing students' evaluative stances (emotions, judgments, and assessments) and their openness to other perspectives. Data sources were analyzed iteratively to identify themes and answer the research questions.

The findings are presented chronologically across four phases: **Phase 1 (Early Phase):** Initially, students held generally neutral beliefs about using the web platform and negative beliefs about the reading activities. These beliefs did not interact significantly. Positive beliefs about platform operation were offset by neutral or negative beliefs about its pedagogical value, and the difficulty of the reading activities and the transition from prior high-stakes test-focused learning led to negative beliefs about their ability to handle the demands. Practices were primarily driven by the compulsory nature of the tasks, with limited voluntary participation. **Phase 2 (Changing Beliefs):** Students' beliefs about the platform’s pedagogical value became increasingly positive due to positive experiences with peer learning and the platform’s affordances, while their beliefs about reading activities showed increasing positivity, albeit still mixed due to ongoing challenges. The positive beliefs about the platform began to moderate their beliefs about reading activities. Practices reflected this shift, with increased mutual assistance and deeper engagement with the reading texts. **Phase 3 (Dynamic Factors):** The positive beliefs were mitigated by technological issues (temporary, recurrent, and recoverable) and the emergence of peer pressure among slower-paced students (chronic). For the fast-paced students, positive beliefs about reading were mitigated by difficulties with certain texts. These mitigated beliefs were mutually bolstered, with the importance of reading driving perseverance despite technology issues. Practices showed a mismatch with beliefs: negative emotions affected engagement, while commitment to the course also contributed to continued engagement. **Phase 4 (Stabilized Beliefs):** Students' beliefs became stable and predominantly positive, with negative beliefs coexisting but remaining inert. This was due to the maturation of their understanding of learning challenges, which created a framework for accepting and overcoming difficulties. Practices became regular and consistent, reflecting the predominantly positive and stable beliefs.

The study reveals the dynamic nature of EFL students’ beliefs about web-based collaborative learning and the complex interplay of factors influencing their beliefs and practices. The findings challenge the static view provided by quantitative studies and offer a more nuanced understanding of the relationship between beliefs and practices, emphasizing the importance of contextual factors. The study's longitudinal qualitative approach allows for a detailed exploration of how beliefs change and interact over time, highlighting the influence of both positive and negative experiences.

This study contributes to a deeper understanding of the dynamic relationship between EFL students’ beliefs and their practices in web-based collaborative learning. The findings highlight the importance of addressing both technological aspects and the challenges of critical literacy in designing effective WBL environments. Future research could investigate the impact of different web-based activities and explore the generalizability of these findings to other contexts and learner populations.

The study's limitations include its focus on a specific group of EFL learners at one university, the limited time frame (one semester), and the focus on specific web-based collaborative learning activities. Future research could address these limitations by investigating diverse learner populations and contexts and examining a broader range of WBL affordances.