Development and Evaluation of a Conversational Natural Language Processing Model for Technology-Enhanced Biology Learning in Secondary Schools

Development and Evaluation of a Conversational Natural Language Processing Model for Technology-Enhanced Biology Learning in Secondary Schools | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Development and Evaluation of a Conversational Natural Language Processing Model for Technology-Enhanced Biology Learning in Secondary Schools Khadijat Muhammad Awwal, Ismail Ibrahim Kuta, Abubakar Sadeeq Ahmad, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9301692/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 8 You are reading this latest preprint version Abstract Background The integration of digital technology into education has significantly transformed how students’ learning complex scientific subjects such as biology. Objective This study developed and evaluated a Conversational Natural Language Processing Model (CNLPM) designed to enhance secondary school biology learning in Niger State, Nigeria. Methology: The model was created via the ADDIE instructional design framework and evaluated via a quasi-experimental research design to determine its effects on students’ achievement, retention, and interest. A developmental research approach was adopted, combining the formative phases of analysis, design, development, implementation, and evaluation with an experimental validation stage. Six secondary schools were selected via a multistage sampling technique, yielding an intact sample of 261 senior secondary II biology students (146 males and 115 females). Three schools were assigned to the experimental group, which received instruction via the CNLPM, whereas the remaining three formed the control group and were taught via the conventional lecture method. Result The findings revealed that students exposed to the CNLPM significantly outperformed their counterparts in the control group (mean gains of 49.46 vs. 28.14). The model also contributed to improved retention and heightened interest in biology learning. These results indicate that conversational AI tools can meaningfully complement traditional instructional strategies. This study provides evidence supporting the adoption of technology-enhanced learning approaches to improve biology education in secondary schools. Conclusion Although the study was conducted in Niger State, Nigeria, the challenges addressed—limited access to interactive learning resources, variability in teacher expertise, and low student engagement—are common across many secondary education systems worldwide. Consequently, the findings offer transferable insights into the potential of conversational AI tools for enhancing science education globally, providing scalable strategies to improve learning outcomes in diverse educational contexts. Unique Contribution: This study makes a unique contribution by empirically demonstrating how an offline-capable conversational AI model, designed for resource-constrained secondary school contexts, can simultaneously enhance conceptual understanding, learner motivation, and equitable access to biology education, while directly advancing the goals of United Nations Sustainable Development Goal 4 (Quality Education) through inclusive and scalable AI-supported instruction Conversational NLP model technology-enhanced learning artificial intelligence secondary school educational technology biology Introduction Science and technology are pivotal drivers of national development, innovation, and global competitiveness. In the 21st century, scientific literacy and technological proficiency have become essential for building a knowledge-based societies. Science education equips students with foundational skills necessary for problem-solving, technological advancement, and sustainable development. However, in Nigeria, persistent declines in students’ performance in science subjects, particularly biology, have raised concerns among educators and policymakers. Research attributes these challenges to inadequate technology-enhanced learning tools, teacher-centered instructional approaches, low student motivation, and the inherently abstract nature of many biological concepts (Ayodele & Adebunmi, 2020 ). Biology, as a core science discipline, covers topics such as cell division, genetics, and ecology, which demand conceptual understanding, visualization, and abstract reasoning. Traditional lecture-based methods, widely employed in Nigerian secondary schools, often fail to provide interactive learning experiences that foster meaningful engagement (Abubakar et al., 2024 ). Consequently, students frequently perceive biology as difficult, resulting in low achievement and diminished interest. The West African Examinations Council (WAEC, 2022) reports that average performance in biology has consistently remained below 50% over the past decade, underscoring systemic instructional challenges. Recent advances in artificial intelligence (AI) present new opportunities to transform educational practice. In particular, conversational natural language processing models (CNLPMs) enable interactive, adaptive, and personalized learning experiences, supporting deeper student engagement than traditional teaching methods (Zhai, 2022 ). These AI systems simulate human-like dialogue, allowing learners to ask questions, receive immediate feedback, clarify misconceptions, and learn at their own pace. Such capabilities are particularly valuable in biology education, where understanding complex processes often requires individualized explanations and scaffolded support. Empirical evidence indicates that technology-enhanced learning tools can improve students’ achievement, retention, and motivation while helping to reduce gender disparities in STEM education (Carney, 2022 ; Wood, 2023 ). However, very few studies in sub-Saharan Africa have developed or evaluated AI-driven conversational models in secondary school contexts. Most existing AI tools are designed for Western educational settings, limiting their cultural and curricular relevance for Nigerian learners. This gap highlights the need for context-specific conversational models aligned with local curricula, linguistic patterns, and infrastructural realities. Nigeria’s educational environment presents additional challenges. While government initiatives encourage STEM education, many schools continue to face infrastructural deficits, including limited access to computers, unreliable internet connectivity, and inadequate teacher training in digital pedagogy (Victor & Ayoko, 2025 ). AI-driven tools that operate offline or require minimal bandwidth, such as locally developed CNLPMs, offer promising solutions for resource-constrained contexts. Gender-related factors further influence educational outcomes. Social norms, cultural expectations, and stereotypes often limit girls’ participation and confidence in STEM fields (Parker & Green, 2025; Collins & Shaw, 2025 ). International assessments, such as PISA, highlight persistent gender differences in literacy and numeracy performance (Robinson, 2025 ). Well-designed educational technologies, particularly those that provide personalized, judgment-free learning experiences, have the potential to reduce these disparities and promote inclusive STEM education (Garcia & Walker, 2023 ; Wood, 2023 ). In this context, the Conversational Natural Language Processing Model (CNLPM) developed in this study seeks to provide an interactive, learner-centered, technology-enhanced approach to teaching biology in Nigerian secondary schools. The model interprets student input, engages learners in natural-language dialogue, and offers step-by-step scaffolding for complex topics such as cell division and genetic inheritance. By supporting inquiry-based learning, personalization, and real-time feedback, the CNLPM addresses gaps in traditional instruction while aligning with global best practices in digital pedagogy. Statement of the Problem Niger State, Nigeria, remains an educationally disadvantaged region, characterized by limited access to digital infrastructure, insufficient technology-enhanced learning tools, and a shortage of teachers trained in modern pedagogies (Victor & Ayoko, 2025 ). Biology instruction is predominantly lecture-based, limiting student engagement and discouraging inquiry-based learning (Ikuesewo-Akinbami & Acheme, 2025 ). National assessment reports (WAEC, NECO, NABTEB, 2020–2024) consistently highlight weaknesses in students’ understanding of abstract biological concepts, attributing poor performance to outdated teaching methods, inflexible instructional strategies, and minimal use of interactive digital tools. Although interventions such as multimedia learning, computer-assisted instruction, virtual learning, and video conferencing have shown some improvements (Eze et al., 2023; Abubakar et al., 2024 ), persistent low performance in biology indicates a need for more interactive, adaptive, and student-centered instructional approaches. There is therefore a critical need to explore the potential of conversational NLP models as innovative tools to enhance biology learning in secondary schools. Aim and Objectives The aim of this study is to evaluate the effectiveness of a conversational natural language processing model in enhancing technology-supported biology learning in Nigerian secondary schools. Objectives : To design and develop a conversational NLP model tailored to the secondary school biology curriculum. To assess the effect of the model on students’ biology achievement compared with conventional teaching methods. To examine the impact of the model on students’ retention of biology concepts. To evaluate students’ interest in biology when taught using the model. To investigate gender differences in achievement when the model is implemented. Research Questions What strategies are employed in developing a conversational NLP model for technology-enhanced biology learning? How can a prototype of the conversational NLP model be designed and implemented for secondary school biology instruction? What is the effect of the model on students’ biology achievement compared with conventional teaching? How does the model influence students’ retention of biology concepts relative to traditional instruction? To what extent does gender affect students’ achievement in CNLPM-supported biology learning? Literature Review Technology-Enhanced Science Learning Science education, particularly biology, is widely recognized as cognitively demanding due to the abstract, microscopic, and dynamic nature of many core concepts, including cell division, genetic inheritance, and physiological processes. Learners frequently experience difficulty visualizing these processes and integrating new scientific information with prior knowledge, leading to surface learning and reliance on rote memorization rather than conceptual understanding. Contemporary learning theories emphasize that meaningful learning occurs when learners actively construct knowledge by connecting new information to existing cognitive structures, yet traditional teacher-centered instructional approaches often provide limited opportunities for such engagement (Gupte et al., 2021 ). In response to these challenges, a substantial body of research has explored technology-enhanced learning approaches in science education. Digital tools such as computer-assisted instruction, multimedia animations, simulations, and virtual laboratories have been shown to support visualization, reduce abstraction, and promote deeper conceptual understanding across primary, secondary, and tertiary education levels. Meta-analytical and empirical studies consistently report that technology-enhanced instruction improves retention, motivation, and learner engagement by enabling interactive exploration and immediate feedback (Yusuf, 2023 ; Baidoo-Anu & Ansah, 2023 ). These tools allow learners to manipulate variables, observe biological processes in real time, and test hypotheses, thereby supporting inquiry-based and experiential learning. Recent advances in artificial intelligence have extended the capabilities of educational technologies beyond static or linear interactions. Conversational natural language processing models (CNLPMs) represent a significant evolution in technology-enhanced learning by enabling dialogic, adaptive, and personalized interaction. Unlike conventional multimedia resources, conversational AI systems can interpret learner queries, diagnose misconceptions, and provide context-sensitive explanations through natural language dialogue. Research suggests that such systems support higher-order thinking and conceptual change by guiding learners through step-by-step reasoning and reflective questioning (Zhai, 2022 ). Empirical studies conducted in diverse contexts have demonstrated that dialog-based AI tools can improve learning outcomes in science subjects, particularly when designed to align with curricular goals and learners’ cognitive needs (Enebechi, 2021 ). Despite these advances, much of the existing research on conversational AI in education remains concentrated in higher education and high-income contexts. There is comparatively limited evidence on how conversational AI can support secondary school learners, especially in resource-constrained environments. This imbalance highlights the need for studies that examine how contextually adapted AI tools can enhance science learning at the secondary level and across diverse educational systems. Theoretical Framework This study is informed by three complementary theoretical perspectives—Cognitive Load Theory, Self-Directed Learning Theory, and the Learner-Centered Approach—which collectively provide a robust foundation for understanding how conversational AI can enhance biology learning. Cognitive Load Theory Cognitive Load Theory (CLT) posits that learning effectiveness depends on the efficient management of working memory resources. Instructional materials that are poorly structured or overly complex impose excessive extraneous cognitive load, thereby hindering learning and schema construction. Effective instructional design should minimize extraneous load, manage intrinsic load, and promote germane load to support deep learning (Razak et al., 2025 ). Research in multimedia learning demonstrates that integrated representations—such as synchronized text and visuals—facilitate comprehension by reducing split attention and supporting simultaneous processing of verbal and pictorial information (Zekeik et al., 2025 ). Conversely, redundant or overly detailed information can overwhelm learners’ working memory. Conversational AI systems address these challenges by presenting information incrementally, adapting explanations to learners’ responses, and using guided questioning to scaffold understanding. Through dialogic interaction, CNLPMs can reduce cognitive overload while supporting schema development in complex biology topics such as mitosis and meiosis. Self-Directed Learning Theory Self-directed learning (SDL) emphasizes learners’ active role in diagnosing learning needs, setting goals, selecting strategies, and evaluating progress. SDL has been linked to deeper conceptual understanding, increased motivation, and improved metacognitive skills, all of which are essential for lifelong learning in rapidly evolving knowledge societies (Kim et al., 2021 ; Canaran, 2025 ). Conversational AI tools align closely with SDL principles by enabling learners to control the pace and direction of learning, ask questions without fear of judgment, and revisit explanations as needed. Dialog-based AI systems provide continuous formative feedback and encourage reflection, thereby fostering learner autonomy and responsibility. These features are particularly valuable in secondary education settings characterized by large class sizes and limited opportunities for individualized instruction. Learner-Centered Approach The learner-centered approach prioritizes active engagement, autonomy, and knowledge construction, contrasting with traditional teacher-centered pedagogies that emphasize information transmission. Learner-centered environments encourage inquiry, experimentation, and critical thinking, enabling students to take ownership of their learning (Riedel et al., 2020 ; Sadera et al., 2020 ). Conversational AI operationalizes learner-centered principles by adapting dialogue to individual learner responses, providing personalized scaffolding, and supporting inquiry-based exploration. Through interactive questioning and immediate feedback, CNLPMs facilitate meaningful engagement and promote deeper learning experiences. As such, conversational AI serves not merely as a technological tool but as a pedagogical agent that supports learner-centered instruction. Empirical Evidence on Conversational AI in Biology Education Empirical studies on technology-enhanced science instruction consistently demonstrate positive effects on academic achievement, motivation, and retention. AI-driven and dialog-based learning tools, in particular, have been shown to enhance conceptual understanding by supporting interactive and adaptive learning processes (Baidoo-Anu & Ansah, 2023 ). Studies conducted in secondary education contexts indicate that multimedia and interactive biology tutorials improve students’ understanding of abstract concepts such as human anatomy and cellular processes (Yusuf, 2023 ). Research specifically examining conversational AI in biology education remains limited but promising. Enebechi ( 2021 ) reported that chatbot-supported instruction significantly improved retention in cell division topics among secondary school students, while Zhai ( 2022 ) highlighted the role of conversational AI in promoting inquiry-based learning and conceptual change. However, most existing studies focus on tertiary education or controlled laboratory environments, underscoring the need for classroom-based research in secondary schools. Gender and Technology Use in Science Learning Research on gender differences in technology-enhanced learning presents mixed findings. Some studies report minimal or no gender-based differences in learning outcomes, suggesting that digital tools may help reduce traditional achievement gaps (Wood, 2023 ). Other studies, however, identify differences in confidence, engagement, and attitudes toward technology, often shaped by sociocultural norms and stereotypes associated with STEM fields (Essel et al., 2024 ; Collins & Shaw, 2025 ). AI-driven learning environments offer potential for promoting gender equity by providing personalized, non-judgmental support and reducing social pressures that may discourage participation. Inclusive conversational AI systems can adapt to learners’ needs and preferences, thereby supporting equitable engagement and learning outcomes (Garcia & Walker, 2023 ). Examining gender dynamics within AI-supported learning contexts therefore contributes to broader discussions on equity and inclusion in science education. Gaps in the Literature Despite growing interest in conversational AI for education, several critical gaps remain. First, there is a contextual gap, as many conversational AI models are developed for Western educational systems and may not align with curricula, linguistic patterns, or classroom realities in other regions. Second, an empirical gap exists due to the limited number of rigorous studies examining the effects of conversational AI on achievement, retention, and motivation among secondary school students, particularly in low- and middle-income countries. Third, an implementation gap persists, as few studies document the design, development, and classroom deployment of contextually relevant conversational AI models. This study addresses these gaps by developing a locally contextualized CNLPM for secondary school biology instruction and empirically examining its impact on students’ conceptual understanding, retention, motivation, and gender inclusivity. By situating the findings within global educational and theoretical frameworks, the study contributes to the expanding body of knowledge on how conversational AI can support equitable, effective, and context-sensitive science education worldwide. Research Design This study employed a developmental research design integrated with a quasi-experimental approach to both design and evaluate an instructional intervention. The developmental component focused on the systematic construction of a Conversational Natural Language Processing Model (CNLPM), ensuring that it was pedagogically grounded and aligned with secondary school biology curriculum standards. The quasi-experimental component evaluated the classroom effectiveness of the CNLPM by comparing learning outcomes with those of students taught using a conventional lecture-based instructional approach. The integration of developmental and quasi-experimental designs is appropriate for research that seeks to innovate instructional technologies while empirically validating their educational impact. Developmental research supports iterative refinement of educational tools through formative evaluation, while quasi-experimental validation allows effectiveness to be examined under authentic classroom conditions (McKenney & Reeves, 2025 ). However, consistent with quasi-experimental methodology, the absence of individual randomization necessitates cautious interpretation of causal inferences, a limitation that is acknowledged and addressed in the Discussion section. Population and Sampling The target population consisted of Senior Secondary II (SS II) students enrolled in biology courses in public secondary schools in Niger State, Nigeria. A multistage sampling strategy was employed. In the first stage, schools were purposively selected based on the availability of functional computer laboratories and relatively stable electricity and internet access to ensure the feasibility of delivering the digital intervention. While this selection criterion enabled effective implementation of the CNLPM, it inherently limits the generalizability of the findings to schools lacking such pre-existing infrastructure and therefore represents a specific subset of resource-constrained educational contexts. In the second stage, intact classes within the selected schools were randomly assigned to either the experimental or control groups. A total of 261 students (146 males and 115 females), aged between 15 and 18 years, participated in the study. Three schools constituted the experimental group, where instruction was delivered using the CNLPM, while three schools formed the control group and received conventional lecture-based instruction. The use of intact classes, while necessary to preserve natural classroom structures, introduces potential selection bias and threats to internal validity; these limitations are explicitly acknowledged in the interpretation of results. The biology topic Cell Division was selected due to its abstract and process-oriented nature, which requires visualization of dynamic microscopic processes such as mitosis and meiosis. These characteristics make the topic particularly suitable for evaluating a conversational Natural language processing model tool designed to support interactive, inquiry-based learning. Instrumentation Three instruments were used for data collection: Biology Achievement Test (BAT): A 40-item multiple-choice test designed to assess students’ conceptual understanding of cell division. Biology Retention Test (BRT): A parallel form of the BAT administered two weeks after instruction to measure knowledge retention. Biology Interest Inventory (BII): A 25-item, five-point Likert-scale instrument used to assess students’ interest and engagement in biology. Validation and Reliability Content validity was established through expert review by three specialists in educational technology, biology education, and guidance and counseling, who evaluated all items for clarity, relevance, and alignment with curriculum objectives. Reliability was determined through a pilot study involving 40 students from schools not included in the main study. Cronbach’s alpha coefficients indicated strong internal consistency: BAT (0.84), BRT (0.79), and BII (0.81), meeting acceptable reliability thresholds for educational research (Ahmed, 2024). Development of the Conversational NLP Model The CNLPM was developed using the ADDIE instructional design framework, encompassing analysis, design, development, implementation, and evaluation. Analysis During the analysis phase, learners’ cognitive needs, curriculum requirements, common misconceptions in cell division, and technological constraints were identified through curriculum review, teacher consultation, and preliminary classroom observations. These inputs informed both pedagogical and technical design decisions. Design The design phase focused on constructing dialogue flows, instructional prompts, and interaction pathways aligned with inquiry-based learning principles. The conversational structure was designed to support concept explanation, guided questioning, formative assessment, and immediate feedback. Development: CNLPM Architecture and Functionality The CNLPM was implemented using a hybrid rule-based NLP architecture, combining keyword matching, predefined dialogue trees, and conditional logic to manage learner–system interactions. This approach was selected to ensure transparency, reliability, and offline functionality within resource-constrained school environments. Biology content—including definitions, explanations, step-by-step processes, and conceptual summaries—was sourced from approved secondary school biology textbooks and curriculum guidelines and structured into modular learning units. Each unit corresponded to a specific subtopic (e.g., stages of mitosis, significance of meiosis) and was linked to a dedicated question bank. The question banks were designed to address varying cognitive levels and incorporated frequently observed student misconceptions identified during the analysis phase. Adaptive feedback was generated through conditional branching: learner responses triggered targeted explanations, corrective prompts, or follow-up questions depending on accuracy and response patterns. The conversational interface supported multiple interaction types, including: question-and-answer exchanges, stepwise concept explanations, guided inquiry sequences, formative self-check questions. Implementation The CNLPM was installed on school computer systems, and teachers and research assistants received training on system operation and classroom integration. The model was then deployed during regular biology lessons in the experimental schools. Evaluation Formative evaluation was conducted during pilot testing to refine dialog flows, feedback logic, and usability. Summative evaluation occurred during full implementation to assess the effectiveness of the CNLPM in improving achievement, retention, and interest. Ethical Considerations and Data Privacy Informed consent was secured from school authorities and parents/guardians and assent was obtained from participating students prior to data collection. All collected data were anonymized and stored securely to ensure privacy and confidentiality, in compliance with Secondary Education Board. Methodological Implications While the study design enabled authentic classroom implementation and practical evaluation, the use of purposive school selection and intact classes introduces limitations related to internal validity and generalizability. These constraints are acknowledged, and findings are interpreted as evidence of effectiveness within infrastructure-ready but still resource-limited school contexts, rather than all secondary schools. This transparency strengthens the methodological credibility of the study and aligns it with best practices in applied educational research. Data Analysis Data collected from the Biology Achievement Test (BAT), Biology Retention Test (BRT), and Biology Interest Inventory (BII) were analyzed using both descriptive and inferential statistical methods. Means and standard deviations were calculated to address the research questions, while Analysis of Variance (ANOVA) was employed in answering research question. ANOVA was first applied to assess baseline differences between the experimental and control groups at the pretest stage. The results indicated no significant differences, confirming that the groups were comparable in prior knowledge. Subsequently, ANOVA was used to compare posttest achievement scores, retention scores, and interest scores between groups. All statistical analyses were conducted using SPSS Version 21.0. Pre-test Equivalence of Groups Table 4.1 ANOVA summary of the pretest scores of the experimental and control groups Sources Sum of Squares df mean square f sig. Between group .089 1 .89 .05 0.94 Within groups 4629.113 267 17.338 Total 4629 .20 268 Table 4.1 presents the ANOVA results comparing the pre-test scores of students in the experimental and control groups. The analysis revealed no statistically significant difference between the pre-test scores of the experimental and control groups, F (1, 267) = 0.005, p = .94. This indicates that both groups were equivalent in Biology achievement prior to the intervention. Research Question 3: Effect of the CNLPM on Student Achievement Table 4.2 Means and standard deviations of the pretest and posttest achievement scores Group N Pretest Posttest Mean Gain X SD X SD CNLP Model 125 11.50 4.15 60.99 5.06 49.46 (experimental) Conventional teaching 136 11.46 4.17 39.60 3.94 28.14 Method (Control) The experimental group achieved a mean gain score of 49.46, whereas the control group recorded a mean gain of 28.14. This result indicates that students exposed to the CNLPM demonstrated greater improvement in biology achievement than their counterparts taught through conventional instruction. The large effect size implies that the conversational model significantly enhanced students’ conceptual understanding of cell division. The interactive nature of the CNLPM, combined with real-time feedback and personalized explanations, contributed to this improved learning outcome. 4.1.3 Retention Scores of Experimental and Control Groups Research Question 4: What is the difference between the retention scores of students taught Biology using CNLPM and those taught using the conventional method? Table 4.3 Mean and standard deviation of the posttest and retention test scores GROUP N Posttest Retention testMean Loss X SD X SD Conversational 125 60.99 5.06 60.145.21 0.85 NLP Model Conventional 136 39.60 3.94 39.113.82 0.49 teaching strategy As presented in Table 4.3 , the experimental group recorded a retention mean score of 60.14 (SD = 5.21), while the control group recorded 39.11 (SD = 3.82). Mean loss was higher in the experimental group (0.85) than in the control group (0.49), indicating stronger retention among students exposed to CNLPM. ANOVA results (Table 4.9) revealed a statistically significant difference in retention scores between the two groups, F (1, 267) = 1445.16, p < .001. Gender analysis Table 4.4 Mean Achievement Scores of Male and Female Students in the Experimental Group Group N Pretest Posttest Mean Gain \(\:\stackrel{-}{X}\) SD \(\:\stackrel{-}{X}\) SD Male 69 11.30 3.81 61.73 4.60 50.43 Female 56 11.75 4.56 60.08 5.49 48.33 Table 4.4 presents the mean achievement scores of male and female students in the experimental group. Both male and female students demonstrated substantial improvement in achievement following exposure to the CNLPM. Male students recorded a mean gain of 50.43, while female students recorded a mean gain of 48.33, resulting in a difference of 2.10 points in favour of male students. The observed difference in mean gain between male and female students was relatively small. Without further inferential statistical testing, this difference cannot be interpreted as statistically significant. Overall, the results indicate that both male and female students benefited comparably from interaction with the CNLPM. Findings This study found that the Conversational Natural Language Processing Model (CNLPM) positively influenced students’ achievement and retention in secondary school biology. As shown in Table 4.2 , students in the experimental group achieved higher posttest scores and larger mean gains than students in the control group. Retention results presented in Table 4.3 indicate that students taught using the CNLPM maintained higher performance levels over time, with minimal score reduction between posttest and retention test. Gender-based analysis (Table 4.4 ) revealed that both male and female students in the experimental group demonstrated notable learning gains following the intervention, with only minor differences in mean gain scores. Overall, the findings indicate that students exposed to the CNLPM outperformed those taught using the conventional teaching method in terms of achievement and retention. Discussion This study investigated the effects of a Conversational Natural Language Processing Model (CNLPM) on students’ achievement, retention, and learning equity in secondary school biology. The findings indicate that students exposed to the CNLPM demonstrated superior academic performance and stronger retention of biological concepts compared to their peers taught using the conventional lecture method. These results provide empirical support for the instructional value of AI-driven conversational learning tools in secondary science education. Effect of the CNLPM on Student Achievement The significantly higher posttest scores and mean gain recorded by students in the experimental group suggest that the CNLPM effectively enhanced students’ conceptual understanding of biology. This finding is consistent with previous research indicating that interactive and dialog-based instructional technologies promote deeper learning than passive instructional approaches (Graesser et al., 2005; Holmes et al., 2022 ). By allowing students to engage in continuous dialogue, ask questions, and receive immediate feedback, the CNLPM supported active cognitive processing, which is essential for meaningful learning. The conversational nature of the model likely encouraged self-explanation and reflection, processes known to strengthen conceptual understanding and reduce misconceptions (Chi et al., 2018 ). Unlike conventional lecture-based instruction, which often limits student participation, the CNLPM provided individualized scaffolding that approximates one-on-one tutoring—an instructional approach widely recognized for its effectiveness in improving student achievement (VanLehn, 2011 ). These features may explain the substantial learning gains observed among students in the experimental group. Effect of the CNLPM on Student Retention Findings related to retention revealed that students taught with the CNLPM maintained higher performance levels over time, with only minimal score reduction between the posttest and retention test. This suggests that the learning facilitated by the conversational model was not only effective in the short term but also durable. According to cognitive load theory, learning environments that reduce extraneous cognitive load and present information in structured, manageable formats are more likely to support long-term retention (Sweller et al., 2019 ). The CNLPM’s ability to simplify complex biological concepts and present explanations incrementally may have contributed to stronger schema formation. In addition, the dialogic structure of the model aligns with the spacing effect and retrieval practice principles, which emphasize repeated engagement with content to strengthen memory consolidation (Cepeda et al., 2006 ; Zepeda et al., 2024 ). Through repeated questioning and clarification during interaction, students were likely prompted to revisit key concepts, thereby reinforcing learning and reducing forgetting. Gender and Learning Outcomes The gender analysis revealed that both male and female students benefited comparably from the use of the CNLPM, with only minor differences in mean gain scores. This finding suggests that the conversational model functioned as an inclusive instructional tool that supported learning across gender groups. Prior research indicates that technology-enhanced learning environments, when designed to be interactive and learner-centered, can reduce traditional participation gaps and promote equitable learning opportunities (UNESCO, 2021 ; Zhai et al., 2023 ). The absence of substantial gender disparities may be attributed to the personalized and self-paced nature of the CNLPM, which allows learners to engage with content without social pressure or bias often present in traditional classroom settings. This aligns with studies suggesting that AI-driven personalization can support equity by adapting instruction to individual learners’ needs rather than demographic characteristics (Chen et al., 2022 ). Educational Significance of the Findings Beyond cognitive outcomes, the findings underscore the broader educational significance of conversational AI in science education. The observed improvements in achievement and retention suggest that AI-supported conversational models can address persistent instructional challenges, such as large class sizes, limited instructional time, and variability in teacher expertise—issues commonly reported in secondary schools in low- and middle-income contexts (Luckin et al., 2021 ; UNESCO, 2023 ). Educational Significance of the Findings Beyond cognitive outcomes, the findings underscore the broader educational significance of conversational AI in science education. The observed improvements in achievement and retention suggest that AI-supported conversational models can address persistent instructional challenges, such as large class sizes, limited instructional time, and variability in teacher expertise—issues commonly reported in secondary schools in low- and middle-income contexts (Luckin et al., 2021 ; UNESCO, 2023 ). Educational Significance of the Findings Beyond cognitive outcomes, the findings underscore the broader educational significance of conversational AI in science education. The observed improvements in achievement and retention suggest that AI-supported conversational models can address persistent instructional challenges, such as large class sizes, limited instructional time, and variability in teacher expertise—issues commonly reported in secondary schools in low- and middle-income contexts (Luckin et al., 2021 ; UNESCO, 2023 ). While the study was conducted within a specific context, its findings align with international research demonstrating the effectiveness of conversational AI and intelligent tutoring systems in enhancing learning outcomes across diverse educational settings (Kohnke et al., 2021 ; Zawacki-Richter et al., 2021 ; Crompton & Burke, 2023 ). However, consistent with best research practice, these findings should be interpreted as contextually grounded, with broader applicability requiring further empirical validation. Limitations and Future Research Despite its positive outcomes, this study has several limitations. The research was conducted in six secondary schools within Niger State, Nigeria, which may limit the generalizability of the findings. Additionally, the nine-week intervention period captured short-term learning effects but may not reflect long-term outcomes. The intervention focused on a single biology topic—cell division—which limits conclusions about broader curricular applicability. Technological challenges, including intermittent power supply, also affected implementation. Future studies should employ longitudinal designs, involve larger and more diverse samples, examine additional biology topics and STEM subjects, and explore ethical considerations such as data privacy, transparency, and algorithmic bias in greater depth. Conclusion and Policy Recommendations This study provides empirical evidence that integrating a Conversational Natural Language Processing Model (CNLPM) into secondary school biology instruction can enhance students’ achievement and retention. By offering interactive, dialog-based learning support, the CNLPM addressed instructional challenges commonly associated with conventional teaching methods. While the study was conducted in Niger State, Nigeria, the findings offer potential insights for similar educational contexts facing resource constraints and large class sizes. However, further research is required to validate the model’s effectiveness across diverse settings and levels of technological readiness. From a policy perspective, education authorities should consider pilot-based integration of AI-assisted instructional tools, supported by teacher training, infrastructure development, and ethical governance frameworks. When thoughtfully implemented, conversational natural language processing model has the potential to support inclusive, equitable, and high-quality science education in alignment with global educational goals such as Sustainable Development Goal 4 Declarations Ethical Approval This study was conducted in accordance with established ethical standards for educational research. Ethical approval was obtained from the relevant institutional research ethics committee prior to data collection. All procedures involving human participants complied with national and institutional guidelines for research ethics. Informed consent Informed consent was obtained from all participants involved in the study. For participants who were minors, informed consent was obtained from parents or legal guardians, and assent was obtained from the students themselves. Participation was voluntary, and participant were informed of their right to withdraw from the study at any time without penalty. Availability of Data and Materials The datasets generated and/or analyzed during the current study are available from the corresponding author upon request. Due to ethical considerations and institutional policies, access to raw student interaction data is restricted to protect participant privacy. Competing Interests The author declares that there are no competing interests, financial or non-financial, associated with this study. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Authors’ Contributions The author was solely responsible for the conceptualization, system design, data collection, model development, data analysis, interpretation of results, and manuscript preparation. The author read and approved the final manuscript. Acknowledgements The author gratefully acknowledges the support of participating schools, biology teachers, and students who contributed to the evaluation of the conversational learning system. Appreciation is also extended to academic colleagues who provided constructive feedback during the development of this study. AI Use Acknowledgment: Rubric AI was used only for grammar refinement and language clarity. The tool did not generate, modify, or influence research data, analysis, or conclusions. All the ideas, results, and interpretations are the author’s work. Consent to Publish Informed consent was obtained from all participants and from their parents or legal guardians for minors involved in this study. Consent was obtained for the publication of any potentially identifiable data. All data have been anonymized to protect participants’ privacy.” References Abubakar U, Falode AA, Ibrahim HA. Redefining student assessment in Nigerian tertiary institutions: The impact of AI technologies on academic performance and developing countermeasures. Adv Mob Learn Educational Res. 2024;4(2):1149–59. Ayodele AO, Adebunmi A. Early childhood development teachers’ perceptions on the use of technology in teaching young children. South Afr J Child Educ. 2020;10(1):1–10. Baidoo-Anu D, Ansah OL. Education in the era of generative artificial intelligence (AI): Understanding the potential benefits of ChatGPT in promoting teaching and learning. SSRN. 2023. https://doi.org/10.2139/ssrn.433748 . Canaran Ö. Navigating self-directed learning in the digital era. Exploring adult education through learning theory. IGI Global; 2025. pp. 163–90. Carney S. Reimagining our futures together: A new social contract for education. UNESCO; 2022. Cepeda NJ, Pashler H, Vul E, Wixted JT, Rohrer D. Distributed practice in verbal recall tasks: A review and quantitative synthesis. Psychol Bull. 2006;132(3):354–80. Chen L, Chen P, Lin Z. Artificial intelligence in education: A review. IEEE Access. 2022;10:75264–78. https://doi.org/10.1109/ACCESS.2022.3188796 . Chi MTH, Kang S, Yaghmourian DL. Why students learn more from dialogue than from monologue. Instr Sci. 2018;46(1):191–214. Collins M, Shaw A. The future of gender studies in education. Future Educ J. 2025;11(2):330–45. https://doi.org/10.2121/fuedu.2025.330 . Crompton H, Burke D. Artificial intelligence in education: A systematic review of research trends. Computers Education: Artif Intell. 2023;4:100113. https://doi.org/10.1016/j.caeai.2023.100113 . Enebechi RI. Effect of inquiry-based learning approach on senior secondary school students’ retention in biology. Br Int J Educ Social Sci. 2021;8(8):9–19. Essel HB, Opoku D, Amoako I. Gender differences in digital learning adoption among African students. Afr J Educ Technol. 2024;14(1):41–58. Eze AN, Ezenwafor JI, Onwusa SC. Effect of computer-based instruction on students’ achievement and retention of high- and low-achieving auto-mechanics technology students in technical colleges. Int J Sci Eng Res. 2020;11(8):1631–45. Garcia M, Walker J. The intersection of gender and educational technology. J Educational Technol. 2023;29(2):244–60. Graesser AC, Kohnke L, Lin Y. Conversational agents and learning: A review of educational dialogue systems. Educational Psychol Rev. 2022;34(3):1321–48. https://doi.org/10.1007/s10648-021 09633-8 . Gupte T, Watts FM, Schmidt-McCormack JA, Zaimi I, Gere AR, Shultz GV. Students’ meaningful learning experiences from participating in organic chemistry writing-to-learn activities. Chem Educ Res Pract. 2021;22(2):396–414. Holmes W, Bialik M, Fadel C. Artificial intelligence in education: Promises and implications for teaching and learning. Center for Curriculum Redesign; 2022. Ikuesewo-Akinbami A, Acheme RO. Evaluation of the role of radio in promoting Yoruba language by Adaba FM among traders in Oja-Oba market, Akure. Nasarawa J Multimedia Communication Stud. 2025;7(1):275–90. Kim D, Jung E, Yoon M, Chang Y, Park S, Kim D, Demir F. Exploring the structural relationships between course design factors, learner commitment, self-directed learning, and intentions for further learning in a self-paced MOOC. Comput Educ. 2021;166:104171. Kohnke L, Moorhouse BL, Zou D. Chatbots in education: A systematic review. Educ Inform Technol. 2021;26(4):4299–337. https://doi.org/10.1007/s10639-021-10418-8 . Luckin R, Holmes W, Griffiths M, Forcier LB. Intelligence unleashed: An argument for AI in education. Pearson Education; 2021. McKenney S, Reeves TC. (2025). Educational design research for relevant and robust scholarship. J Comput High Educ, 1–25. Parker SK, Ballard T, Billinghurst M, Collins C, Dollard M, Griffin MA, Walsh T. Quality work in the future: New directions via a coevolving sociotechnical systems perspective. Aust J Manage. 2025. https://doi.org/10.1177/03128962251331813 . Advance online publication. Razak RA, Alias NF, Idris AY, editors. Qualitative insights through applied cognitive task analysis. IGI Global; 2025. Riedel R, Vialle W, Pearson P, Oades LG. Quality learning and positive education practice: The student experience of learning in a school-wide approach to positive education. Int J Appl Psychol. 2020;5(1):53–75. Robinson L. Innovative practices to address gender disparities. Innov Educ. 2025;22(4):55–70. https://doi.org/10.7890/innoedu.2025.055 . Sadera JRN, Torres RYS, Rogayan DV. Challenges encountered by junior high school students in learning science: Basis for an action plan. Univers J Educational Res. 2020;8(12A):7405–14. Sweller J, Ayres P, Kalyuga S. Cognitive load theory. 2nd ed. Springer; 2019. UNESCO. Reimagining our futures together: A new social contract for education. UNESCO Publishing; 2021. UNESCO. Guidance on generative AI in education and research. UNESCO Publishing; 2023. VanLehn K. The relative effectiveness of human tutoring, intelligent tutoring systems, and other tutoring systems. Educational Psychol. 2011;46(4):197–221. Victor NCODE, Ayoko O. Information communication technology and academic staff job performance in tertiary institutions. Manuscript in preparation; 2025. West African Examinations Council. Chief examiners’ report for the West African Senior School Certificate Examination (WASSCE). WAEC; 2022. Wood K. Gender and equity in technology-enhanced classrooms: A meta-analysis. Comput Educ. 2023;194:104678. https://doi.org/10.1016/j.compedu.2023.104678 . Yusuf H. (2023). Effects of multimedia instruction on SS1 biology students’ academic achievement and retention in Bosso Local Government Area (Doctoral dissertation). Federal University of Technology, Minna. http://irepo.futminna.edu.ng Zawacki-Richter O, Marín VI, Bond M, Gouverneur F. (2021). Systematic review of research on artificial intelligence applications in higher education. International Journal of Educational Technology in Higher Education, 18(1), 1–27. https://doi.org/10.1186/s41239-021-00238-3 Zekeik H, Chahbi M, Sefian ML, Bakkali I. Augmented reality and virtual reality in education: A systematic narrative review on benefits, challenges, and applications. EURASIA J Math Sci Technol Educ. 2025;21(9):em2699. Zepeda CD, Een E, Butler AC. (2024). The mnemonic effects of retrieval practice. In Oxford Research] Encyclopedia of Education. Zepeda CD, et al. Spacing and retrieval practice effects on learning: A meta-analytic review. Educational Psychol Rev. 2024;36(1):1–29. Zhai X. ChatGPT and the future of education: Opportunities and challenges. Front Educ. 2022;7:1046789. https://doi.org/10.3389/feduc.2022.1046789 . Zhai X, Yin Y, Pellegrino JW, Haudek KC, Shi L. Applying machine learning in science education research: A systematic review. J Res Sci Teach. 2023;60(6):748–75. https://doi.org/10.1002/tea.21806 . Tables Tables 4.1 to 4.4 are available in the Supplementary Files section. Additional Declarations No competing interests reported. Supplementary Files APPENDIXC.pdf Tables.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 13 May, 2026 Reviewers agreed at journal 12 May, 2026 Reviewers agreed at journal 03 May, 2026 Reviewers invited by journal 22 Apr, 2026 Editor assigned by journal 22 Apr, 2026 Editor invited by journal 13 Apr, 2026 Submission checks completed at journal 11 Apr, 2026 First submitted to journal 11 Apr, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9301692","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":633767351,"identity":"b6d2fffe-3ce6-4852-8a12-5b9ef738fed5","order_by":0,"name":"Khadijat Muhammad Awwal","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA40lEQVRIiWNgGAWjYDACHh4gNmBjYGBvPgDkSsgQr4WH51gCSAsPkVpAtESOAZhPUIduz9mDH94U8MnbM+R8fnWjxoKHgf3w0Q34tJid7UuWnGPAZtjDcHabdc4xoMN40tJu4NVynsdAGugXxh7G3m3GOWxALRI8ZoS0GP8GarHvYeZ5ZpzzjxgtZ3vMQLYk9rDxMD/ObSNGy5kzZpZAvyT3nGEzY87tk+BhI+iXMznGN978OWbbPv/x48853+rk+NkPH8OrBQqOgQg2CTBJhHIQqAERzB+IVD0KRsEoGAUjDAAA1gdC4cCtiTYAAAAASUVORK5CYII=","orcid":"","institution":"Federal University of Technology","correspondingAuthor":true,"prefix":"","firstName":"Khadijat","middleName":"Muhammad","lastName":"Awwal","suffix":""},{"id":633767352,"identity":"01cdfe14-69dc-4c72-910a-8082c849b005","order_by":1,"name":"Ismail Ibrahim Kuta","email":"","orcid":"","institution":"Federal University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Ismail","middleName":"Ibrahim","lastName":"Kuta","suffix":""},{"id":633767353,"identity":"f3d5843d-d605-43ab-b7ee-7caa6aceef23","order_by":2,"name":"Abubakar Sadeeq Ahmad","email":"","orcid":"","institution":"Federal University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Abubakar","middleName":"Sadeeq","lastName":"Ahmad","suffix":""},{"id":633767354,"identity":"6ee0f3db-4e15-4e9e-b52d-cdcf51516a41","order_by":3,"name":"Abdulkareem Abubakar","email":"","orcid":"","institution":"Federal University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Abdulkareem","middleName":"","lastName":"Abubakar","suffix":""}],"badges":[],"createdAt":"2026-04-02 10:11:24","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9301692/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9301692/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108493322,"identity":"2747f436-6eff-4b56-97df-e1b76f1068fe","added_by":"auto","created_at":"2026-05-05 09:59:56","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":291495,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9301692/v1/b6499d4e-c881-48b3-961f-cbae24821b62.pdf"},{"id":108437996,"identity":"286946e4-bc38-4dac-9035-a0101e1f9304","added_by":"auto","created_at":"2026-05-04 16:05:24","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":35176055,"visible":true,"origin":"","legend":"","description":"","filename":"APPENDIXC.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9301692/v1/9b1e6c270924f93a8d8fff04.pdf"},{"id":108437995,"identity":"942eba0d-638d-44f8-8173-985696143043","added_by":"auto","created_at":"2026-05-04 16:05:24","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":15625,"visible":true,"origin":"","legend":"","description":"","filename":"Tables.docx","url":"https://assets-eu.researchsquare.com/files/rs-9301692/v1/46173f5860fa193f46162481.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"\u003cp\u003eDevelopment and Evaluation of a Conversational Natural Language Processing Model for Technology-Enhanced Biology Learning in Secondary Schools\u003c/p\u003e","fulltext":[{"header":"Introduction","content":"\u003cp\u003eScience and technology are pivotal drivers of national development, innovation, and global competitiveness. In the 21st century, scientific literacy and technological proficiency have become essential for building a knowledge-based societies. Science education equips students with foundational skills necessary for problem-solving, technological advancement, and sustainable development. However, in Nigeria, persistent declines in students\u0026rsquo; performance in science subjects, particularly biology, have raised concerns among educators and policymakers. Research attributes these challenges to inadequate technology-enhanced learning tools, teacher-centered instructional approaches, low student motivation, and the inherently abstract nature of many biological concepts (Ayodele \u0026amp; Adebunmi, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eBiology, as a core science discipline, covers topics such as cell division, genetics, and ecology, which demand conceptual understanding, visualization, and abstract reasoning. Traditional lecture-based methods, widely employed in Nigerian secondary schools, often fail to provide interactive learning experiences that foster meaningful engagement (Abubakar et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Consequently, students frequently perceive biology as difficult, resulting in low achievement and diminished interest. The West African Examinations Council (WAEC, 2022) reports that average performance in biology has consistently remained below 50% over the past decade, underscoring systemic instructional challenges.\u003c/p\u003e \u003cp\u003eRecent advances in artificial intelligence (AI) present new opportunities to transform educational practice. In particular, conversational natural language processing models (CNLPMs) enable interactive, adaptive, and personalized learning experiences, supporting deeper student engagement than traditional teaching methods (Zhai, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). These AI systems simulate human-like dialogue, allowing learners to ask questions, receive immediate feedback, clarify misconceptions, and learn at their own pace. Such capabilities are particularly valuable in biology education, where understanding complex processes often requires individualized explanations and scaffolded support.\u003c/p\u003e \u003cp\u003eEmpirical evidence indicates that technology-enhanced learning tools can improve students\u0026rsquo; achievement, retention, and motivation while helping to reduce gender disparities in STEM education (Carney, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Wood, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). However, very few studies in sub-Saharan Africa have developed or evaluated AI-driven conversational models in secondary school contexts. Most existing AI tools are designed for Western educational settings, limiting their cultural and curricular relevance for Nigerian learners. This gap highlights the need for context-specific conversational models aligned with local curricula, linguistic patterns, and infrastructural realities.\u003c/p\u003e \u003cp\u003eNigeria\u0026rsquo;s educational environment presents additional challenges. While government initiatives encourage STEM education, many schools continue to face infrastructural deficits, including limited access to computers, unreliable internet connectivity, and inadequate teacher training in digital pedagogy (Victor \u0026amp; Ayoko, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). AI-driven tools that operate offline or require minimal bandwidth, such as locally developed CNLPMs, offer promising solutions for resource-constrained contexts.\u003c/p\u003e \u003cp\u003eGender-related factors further influence educational outcomes. Social norms, cultural expectations, and stereotypes often limit girls\u0026rsquo; participation and confidence in STEM fields (Parker \u0026amp; Green, 2025; Collins \u0026amp; Shaw, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). International assessments, such as PISA, highlight persistent gender differences in literacy and numeracy performance (Robinson, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Well-designed educational technologies, particularly those that provide personalized, judgment-free learning experiences, have the potential to reduce these disparities and promote inclusive STEM education (Garcia \u0026amp; Walker, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Wood, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn this context, the Conversational Natural Language Processing Model (CNLPM) developed in this study seeks to provide an interactive, learner-centered, technology-enhanced approach to teaching biology in Nigerian secondary schools. The model interprets student input, engages learners in natural-language dialogue, and offers step-by-step scaffolding for complex topics such as cell division and genetic inheritance. By supporting inquiry-based learning, personalization, and real-time feedback, the CNLPM addresses gaps in traditional instruction while aligning with global best practices in digital pedagogy.\u003c/p\u003e\n\u003ch3\u003eStatement of the Problem\u003c/h3\u003e\n\u003cp\u003eNiger State, Nigeria, remains an educationally disadvantaged region, characterized by limited access to digital infrastructure, insufficient technology-enhanced learning tools, and a shortage of teachers trained in modern pedagogies (Victor \u0026amp; Ayoko, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Biology instruction is predominantly lecture-based, limiting student engagement and discouraging inquiry-based learning (Ikuesewo-Akinbami \u0026amp; Acheme, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). National assessment reports (WAEC, NECO, NABTEB, 2020\u0026ndash;2024) consistently highlight weaknesses in students\u0026rsquo; understanding of abstract biological concepts, attributing poor performance to outdated teaching methods, inflexible instructional strategies, and minimal use of interactive digital tools.\u003c/p\u003e \u003cp\u003eAlthough interventions such as multimedia learning, computer-assisted instruction, virtual learning, and video conferencing have shown some improvements (Eze et al., 2023; Abubakar et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), persistent low performance in biology indicates a need for more interactive, adaptive, and student-centered instructional approaches. There is therefore a critical need to explore the potential of conversational NLP models as innovative tools to enhance biology learning in secondary schools.\u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eAim and Objectives\u003c/h2\u003e \u003cp\u003eThe aim of this study is to evaluate the effectiveness of a conversational natural language processing model in enhancing technology-supported biology learning in Nigerian secondary schools.\u003c/p\u003e \u003cp\u003e \u003cb\u003eObjectives\u003c/b\u003e:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eTo design and develop a conversational NLP model tailored to the secondary school biology curriculum.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eTo assess the effect of the model on students\u0026rsquo; biology achievement compared with conventional teaching methods.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eTo examine the impact of the model on students\u0026rsquo; retention of biology concepts.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eTo evaluate students\u0026rsquo; interest in biology when taught using the model.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eTo investigate gender differences in achievement when the model is implemented.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eResearch Questions\u003c/b\u003e \u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eWhat strategies are employed in developing a conversational NLP model for technology-enhanced biology learning?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eHow can a prototype of the conversational NLP model be designed and implemented for secondary school biology instruction?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eWhat is the effect of the model on students\u0026rsquo; biology achievement compared with conventional teaching?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eHow does the model influence students\u0026rsquo; retention of biology concepts relative to traditional instruction?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eTo what extent does gender affect students\u0026rsquo; achievement in CNLPM-supported biology learning?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Literature Review","content":"\u003cp\u003eTechnology-Enhanced Science Learning\u003c/p\u003e \u003cp\u003eScience education, particularly biology, is widely recognized as cognitively demanding due to the abstract, microscopic, and dynamic nature of many core concepts, including cell division, genetic inheritance, and physiological processes. Learners frequently experience difficulty visualizing these processes and integrating new scientific information with prior knowledge, leading to surface learning and reliance on rote memorization rather than conceptual understanding. Contemporary learning theories emphasize that meaningful learning occurs when learners actively construct knowledge by connecting new information to existing cognitive structures, yet traditional teacher-centered instructional approaches often provide limited opportunities for such engagement (Gupte et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn response to these challenges, a substantial body of research has explored technology-enhanced learning approaches in science education. Digital tools such as computer-assisted instruction, multimedia animations, simulations, and virtual laboratories have been shown to support visualization, reduce abstraction, and promote deeper conceptual understanding across primary, secondary, and tertiary education levels. Meta-analytical and empirical studies consistently report that technology-enhanced instruction improves retention, motivation, and learner engagement by enabling interactive exploration and immediate feedback (Yusuf, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Baidoo-Anu \u0026amp; Ansah, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). These tools allow learners to manipulate variables, observe biological processes in real time, and test hypotheses, thereby supporting inquiry-based and experiential learning.\u003c/p\u003e \u003cp\u003eRecent advances in artificial intelligence have extended the capabilities of educational technologies beyond static or linear interactions. Conversational natural language processing models (CNLPMs) represent a significant evolution in technology-enhanced learning by enabling dialogic, adaptive, and personalized interaction. Unlike conventional multimedia resources, conversational AI systems can interpret learner queries, diagnose misconceptions, and provide context-sensitive explanations through natural language dialogue. Research suggests that such systems support higher-order thinking and conceptual change by guiding learners through step-by-step reasoning and reflective questioning (Zhai, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Empirical studies conducted in diverse contexts have demonstrated that dialog-based AI tools can improve learning outcomes in science subjects, particularly when designed to align with curricular goals and learners\u0026rsquo; cognitive needs (Enebechi, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eDespite these advances, much of the existing research on conversational AI in education remains concentrated in higher education and high-income contexts. There is comparatively limited evidence on how conversational AI can support secondary school learners, especially in resource-constrained environments. This imbalance highlights the need for studies that examine how contextually adapted AI tools can enhance science learning at the secondary level and across diverse educational systems.\u003c/p\u003e\n\u003ch3\u003eTheoretical Framework\u003c/h3\u003e\n\u003cp\u003eThis study is informed by three complementary theoretical perspectives\u0026mdash;Cognitive Load Theory, Self-Directed Learning Theory, and the Learner-Centered Approach\u0026mdash;which collectively provide a robust foundation for understanding how conversational AI can enhance biology learning.\u003c/p\u003e\n\u003ch3\u003eCognitive Load Theory\u003c/h3\u003e\n\u003cp\u003eCognitive Load Theory (CLT) posits that learning effectiveness depends on the efficient management of working memory resources. Instructional materials that are poorly structured or overly complex impose excessive extraneous cognitive load, thereby hindering learning and schema construction. Effective instructional design should minimize extraneous load, manage intrinsic load, and promote germane load to support deep learning (Razak et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eResearch in multimedia learning demonstrates that integrated representations\u0026mdash;such as synchronized text and visuals\u0026mdash;facilitate comprehension by reducing split attention and supporting simultaneous processing of verbal and pictorial information (Zekeik et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Conversely, redundant or overly detailed information can overwhelm learners\u0026rsquo; working memory. Conversational AI systems address these challenges by presenting information incrementally, adapting explanations to learners\u0026rsquo; responses, and using guided questioning to scaffold understanding. Through dialogic interaction, CNLPMs can reduce cognitive overload while supporting schema development in complex biology topics such as mitosis and meiosis.\u003c/p\u003e\n\u003ch3\u003eSelf-Directed Learning Theory\u003c/h3\u003e\n\u003cp\u003eSelf-directed learning (SDL) emphasizes learners\u0026rsquo; active role in diagnosing learning needs, setting goals, selecting strategies, and evaluating progress. SDL has been linked to deeper conceptual understanding, increased motivation, and improved metacognitive skills, all of which are essential for lifelong learning in rapidly evolving knowledge societies (Kim et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Canaran, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eConversational AI tools align closely with SDL principles by enabling learners to control the pace and direction of learning, ask questions without fear of judgment, and revisit explanations as needed. Dialog-based AI systems provide continuous formative feedback and encourage reflection, thereby fostering learner autonomy and responsibility. These features are particularly valuable in secondary education settings characterized by large class sizes and limited opportunities for individualized instruction.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eLearner-Centered Approach\u003c/h2\u003e \u003cp\u003eThe learner-centered approach prioritizes active engagement, autonomy, and knowledge construction, contrasting with traditional teacher-centered pedagogies that emphasize information transmission. Learner-centered environments encourage inquiry, experimentation, and critical thinking, enabling students to take ownership of their learning (Riedel et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Sadera et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eConversational AI operationalizes learner-centered principles by adapting dialogue to individual learner responses, providing personalized scaffolding, and supporting inquiry-based exploration. Through interactive questioning and immediate feedback, CNLPMs facilitate meaningful engagement and promote deeper learning experiences. As such, conversational AI serves not merely as a technological tool but as a pedagogical agent that supports learner-centered instruction.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eEmpirical Evidence on Conversational AI in Biology Education\u003c/h3\u003e\n\u003cp\u003eEmpirical studies on technology-enhanced science instruction consistently demonstrate positive effects on academic achievement, motivation, and retention. AI-driven and dialog-based learning tools, in particular, have been shown to enhance conceptual understanding by supporting interactive and adaptive learning processes (Baidoo-Anu \u0026amp; Ansah, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Studies conducted in secondary education contexts indicate that multimedia and interactive biology tutorials improve students\u0026rsquo; understanding of abstract concepts such as human anatomy and cellular processes (Yusuf, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eResearch specifically examining conversational AI in biology education remains limited but promising. Enebechi (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) reported that chatbot-supported instruction significantly improved retention in cell division topics among secondary school students, while Zhai (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) highlighted the role of conversational AI in promoting inquiry-based learning and conceptual change. However, most existing studies focus on tertiary education or controlled laboratory environments, underscoring the need for classroom-based research in secondary schools.\u003c/p\u003e\n\u003ch3\u003eGender and Technology Use in Science Learning\u003c/h3\u003e\n\u003cp\u003eResearch on gender differences in technology-enhanced learning presents mixed findings. Some studies report minimal or no gender-based differences in learning outcomes, suggesting that digital tools may help reduce traditional achievement gaps (Wood, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Other studies, however, identify differences in confidence, engagement, and attitudes toward technology, often shaped by sociocultural norms and stereotypes associated with STEM fields (Essel et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Collins \u0026amp; Shaw, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAI-driven learning environments offer potential for promoting gender equity by providing personalized, non-judgmental support and reducing social pressures that may discourage participation. Inclusive conversational AI systems can adapt to learners\u0026rsquo; needs and preferences, thereby supporting equitable engagement and learning outcomes (Garcia \u0026amp; Walker, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Examining gender dynamics within AI-supported learning contexts therefore contributes to broader discussions on equity and inclusion in science education.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eGaps in the Literature\u003c/h2\u003e \u003cp\u003eDespite growing interest in conversational AI for education, several critical gaps remain. First, there is a contextual gap, as many conversational AI models are developed for Western educational systems and may not align with curricula, linguistic patterns, or classroom realities in other regions. Second, an empirical gap exists due to the limited number of rigorous studies examining the effects of conversational AI on achievement, retention, and motivation among secondary school students, particularly in low- and middle-income countries. Third, an implementation gap persists, as few studies document the design, development, and classroom deployment of contextually relevant conversational AI models.\u003c/p\u003e \u003cp\u003eThis study addresses these gaps by developing a locally contextualized CNLPM for secondary school biology instruction and empirically examining its impact on students\u0026rsquo; conceptual understanding, retention, motivation, and gender inclusivity. By situating the findings within global educational and theoretical frameworks, the study contributes to the expanding body of knowledge on how conversational AI can support equitable, effective, and context-sensitive science education worldwide.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eResearch Design\u003c/h2\u003e \u003cp\u003eThis study employed a developmental research design integrated with a quasi-experimental approach to both design and evaluate an instructional intervention. The developmental component focused on the systematic construction of a Conversational Natural Language Processing Model (CNLPM), ensuring that it was pedagogically grounded and aligned with secondary school biology curriculum standards. The quasi-experimental component evaluated the classroom effectiveness of the CNLPM by comparing learning outcomes with those of students taught using a conventional lecture-based instructional approach.\u003c/p\u003e \u003cp\u003eThe integration of developmental and quasi-experimental designs is appropriate for research that seeks to innovate instructional technologies while empirically validating their educational impact. Developmental research supports iterative refinement of educational tools through formative evaluation, while quasi-experimental validation allows effectiveness to be examined under authentic classroom conditions (McKenney \u0026amp; Reeves, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). However, consistent with quasi-experimental methodology, the absence of individual randomization necessitates cautious interpretation of causal inferences, a limitation that is acknowledged and addressed in the Discussion section.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003ePopulation and Sampling\u003c/h2\u003e \u003cp\u003eThe target population consisted of Senior Secondary II (SS II) students enrolled in biology courses in public secondary schools in Niger State, Nigeria. A multistage sampling strategy was employed. In the first stage, schools were purposively selected based on the availability of functional computer laboratories and relatively stable electricity and internet access to ensure the feasibility of delivering the digital intervention. While this selection criterion enabled effective implementation of the CNLPM, it inherently limits the generalizability of the findings to schools lacking such pre-existing infrastructure and therefore represents a specific subset of resource-constrained educational contexts.\u003c/p\u003e \u003cp\u003eIn the second stage, intact classes within the selected schools were randomly assigned to either the experimental or control groups. A total of 261 students (146 males and 115 females), aged between 15 and 18 years, participated in the study. Three schools constituted the experimental group, where instruction was delivered using the CNLPM, while three schools formed the control group and received conventional lecture-based instruction. The use of intact classes, while necessary to preserve natural classroom structures, introduces potential selection bias and threats to internal validity; these limitations are explicitly acknowledged in the interpretation of results.\u003c/p\u003e \u003cp\u003eThe biology topic Cell Division was selected due to its abstract and process-oriented nature, which requires visualization of dynamic microscopic processes such as mitosis and meiosis. These characteristics make the topic particularly suitable for evaluating a conversational Natural language processing model tool designed to support interactive, inquiry-based learning.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eInstrumentation\u003c/h2\u003e \u003cp\u003eThree instruments were used for data collection:\u003c/p\u003e \u003cp\u003eBiology Achievement Test (BAT): A 40-item multiple-choice test designed to assess students\u0026rsquo; conceptual understanding of cell division.\u003c/p\u003e \u003cp\u003eBiology Retention Test (BRT): A parallel form of the BAT administered two weeks after instruction to measure knowledge retention.\u003c/p\u003e \u003cp\u003eBiology Interest Inventory (BII): A 25-item, five-point Likert-scale instrument used to assess students\u0026rsquo; interest and engagement in biology.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eValidation and Reliability\u003c/h2\u003e \u003cp\u003eContent validity was established through expert review by three specialists in educational technology, biology education, and guidance and counseling, who evaluated all items for clarity, relevance, and alignment with curriculum objectives.\u003c/p\u003e \u003cp\u003eReliability was determined through a pilot study involving 40 students from schools not included in the main study. Cronbach\u0026rsquo;s alpha coefficients indicated strong internal consistency: BAT (0.84), BRT (0.79), and BII (0.81), meeting acceptable reliability thresholds for educational research (Ahmed, 2024).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eDevelopment of the Conversational NLP Model\u003c/h2\u003e \u003cp\u003eThe CNLPM was developed using the ADDIE instructional design framework, encompassing analysis, design, development, implementation, and evaluation.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eAnalysis\u003c/h2\u003e \u003cp\u003eDuring the analysis phase, learners\u0026rsquo; cognitive needs, curriculum requirements, common misconceptions in cell division, and technological constraints were identified through curriculum review, teacher consultation, and preliminary classroom observations. These inputs informed both pedagogical and technical design decisions.\u003c/p\u003e \u003cp\u003eDesign\u003c/p\u003e \u003cp\u003eThe design phase focused on constructing dialogue flows, instructional prompts, and interaction pathways aligned with inquiry-based learning principles. The conversational structure was designed to support concept explanation, guided questioning, formative assessment, and immediate feedback.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eDevelopment: CNLPM Architecture and Functionality\u003c/h2\u003e \u003cp\u003eThe CNLPM was implemented using a hybrid rule-based NLP architecture, combining keyword matching, predefined dialogue trees, and conditional logic to manage learner\u0026ndash;system interactions. This approach was selected to ensure transparency, reliability, and offline functionality within resource-constrained school environments.\u003c/p\u003e \u003cp\u003e Biology content\u0026mdash;including definitions, explanations, step-by-step processes, and conceptual summaries\u0026mdash;was sourced from approved secondary school biology textbooks and curriculum guidelines and structured into modular learning units. Each unit corresponded to a specific subtopic (e.g., stages of mitosis, significance of meiosis) and was linked to a dedicated question bank.\u003c/p\u003e \u003cp\u003eThe question banks were designed to address varying cognitive levels and incorporated frequently observed student misconceptions identified during the analysis phase. Adaptive feedback was generated through conditional branching: learner responses triggered targeted explanations, corrective prompts, or follow-up questions depending on accuracy and response patterns.\u003c/p\u003e \u003cp\u003eThe conversational interface supported multiple interaction types, including:\u003c/p\u003e \u003cp\u003equestion-and-answer exchanges, stepwise concept explanations, guided inquiry sequences, formative self-check questions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eImplementation\u003c/h2\u003e \u003cp\u003eThe CNLPM was installed on school computer systems, and teachers and research assistants received training on system operation and classroom integration. The model was then deployed during regular biology lessons in the experimental schools.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eEvaluation\u003c/h2\u003e \u003cp\u003eFormative evaluation was conducted during pilot testing to refine dialog flows, feedback logic, and usability. Summative evaluation occurred during full implementation to assess the effectiveness of the CNLPM in improving achievement, retention, and interest.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eEthical Considerations and Data Privacy\u003c/h2\u003e \u003cp\u003e \u003cstrong\u003eInformed consent\u003c/strong\u003e \u003cp\u003ewas secured from school authorities and parents/guardians and assent was obtained from participating students prior to data collection. All collected data were anonymized and stored securely to ensure privacy and confidentiality, in compliance with Secondary Education Board.\u003c/p\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eMethodological Implications\u003c/h2\u003e \u003cp\u003eWhile the study design enabled authentic classroom implementation and practical evaluation, the use of purposive school selection and intact classes introduces limitations related to internal validity and generalizability. These constraints are acknowledged, and findings are interpreted as evidence of effectiveness within infrastructure-ready but still resource-limited school contexts, rather than all secondary schools. This transparency strengthens the methodological credibility of the study and aligns it with best practices in applied educational research.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section2\"\u003e \u003ch2\u003eData Analysis\u003c/h2\u003e \u003cp\u003eData collected from the Biology Achievement Test (BAT), Biology Retention Test (BRT), and Biology Interest Inventory (BII) were analyzed using both descriptive and inferential statistical methods. Means and standard deviations were calculated to address the research questions, while Analysis of Variance (ANOVA) was employed in answering research question.\u003c/p\u003e \u003cp\u003eANOVA was first applied to assess baseline differences between the experimental and control groups at the pretest stage. The results indicated no significant differences, confirming that the groups were comparable in prior knowledge. Subsequently, ANOVA was used to compare posttest achievement scores, retention scores, and interest scores between groups. All statistical analyses were conducted using SPSS Version 21.0.\u003c/p\u003e \u003cp\u003ePre-test Equivalence of Groups\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4.1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eANOVA summary of the pretest scores of the experimental and control groups\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"1\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSources Sum of Squares df mean square f sig.\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBetween group .089 1 .89 .05 0.94\u003c/p\u003e \u003cp\u003eWithin groups 4629.113 267 17.338\u003c/p\u003e \u003cp\u003eTotal 4629 .20 268\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e4.1\u003c/span\u003e presents the ANOVA results comparing the pre-test scores of students in the experimental and control groups. The analysis revealed no statistically significant difference between the pre-test scores of the experimental and control groups, F (1, 267)\u0026thinsp;=\u0026thinsp;0.005, p = .94. This indicates that both groups were equivalent in Biology achievement prior to the intervention.\u003c/p\u003e \u003cp\u003eResearch Question 3: Effect of the CNLPM on Student Achievement\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4.2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eMeans and standard deviations of the pretest and posttest achievement scores\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"1\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGroup N Pretest Posttest Mean\u003c/p\u003e \u003cp\u003eGain\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eX SD X SD\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCNLP Model 125 11.50 4.15 60.99 5.06 49.46\u003c/p\u003e \u003cp\u003e(experimental)\u003c/p\u003e \u003cp\u003eConventional teaching 136 11.46 4.17 39.60 3.94 28.14\u003c/p\u003e \u003cp\u003eMethod (Control)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe experimental group achieved a mean gain score of 49.46, whereas the control group recorded a mean gain of 28.14. This result indicates that students exposed to the CNLPM demonstrated greater improvement in biology achievement than their counterparts taught through conventional instruction. The large effect size implies that the conversational model significantly enhanced students\u0026rsquo; conceptual understanding of cell division. The interactive nature of the CNLPM, combined with real-time feedback and personalized explanations, contributed to this improved learning outcome.\u003c/p\u003e \u003cp\u003e \u003cb\u003e4.1.3 Retention Scores of Experimental and Control Groups\u003c/b\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003eResearch Question 4:\u003c/h2\u003e \u003cp\u003eWhat is the difference between the retention scores of students taught Biology using CNLPM and those taught using the conventional method?\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4.3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eMean and standard deviation of the posttest and retention test scores\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"1\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGROUP N Posttest Retention testMean Loss\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eX SD X SD\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eConversational 125 60.99 5.06 60.145.21 0.85\u003c/p\u003e \u003cp\u003eNLP Model\u003c/p\u003e \u003cp\u003eConventional 136 39.60 3.94 39.113.82 0.49\u003c/p\u003e \u003cp\u003eteaching strategy\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eAs presented in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e4.3\u003c/span\u003e, the experimental group recorded a retention mean score of 60.14 (SD\u0026thinsp;=\u0026thinsp;5.21), while the control group recorded 39.11 (SD\u0026thinsp;=\u0026thinsp;3.82). Mean loss was higher in the experimental group (0.85) than in the control group (0.49), indicating stronger retention among students exposed to CNLPM.\u003c/p\u003e \u003cp\u003eANOVA results (Table\u0026nbsp;4.9) revealed a statistically significant difference in retention scores between the two groups, F (1, 267)\u0026thinsp;=\u0026thinsp;1445.16, p \u0026lt; .001.\u003c/p\u003e \u003cp\u003eGender \u003cb\u003eanalysis\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4.4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eMean Achievement Scores of Male and Female Students in the Experimental Group\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"1\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGroup N Pretest Posttest Mean Gain\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\stackrel{-}{X}\\)\u003c/span\u003e\u003c/span\u003eSD \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\stackrel{-}{X}\\)\u003c/span\u003e\u003c/span\u003e SD\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale 69 11.30 3.81 61.73 4.60 50.43\u003c/p\u003e \u003cp\u003eFemale 56 11.75 4.56 60.08 5.49 48.33\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4.4\u003c/span\u003e presents the mean achievement scores of male and female students in the experimental group.\u003c/p\u003e \u003cp\u003eBoth male and female students demonstrated substantial improvement in achievement following exposure to the CNLPM. Male students recorded a mean gain of 50.43, while female students recorded a mean gain of 48.33, resulting in a difference of 2.10 points in favour of male students.\u003c/p\u003e \u003cp\u003eThe observed difference in mean gain between male and female students was relatively small. Without further inferential statistical testing, this difference cannot be interpreted as statistically significant. Overall, the results indicate that both male and female students benefited comparably from interaction with the CNLPM.\u003c/p\u003e\u003c/div\u003e"},{"header":"Findings","content":" \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003cp\u003eThis study found that the Conversational Natural Language Processing Model (CNLPM) positively influenced students\u0026rsquo; achievement and retention in secondary school biology.\u003c/p\u003e \u003cp\u003eAs shown in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e4.2\u003c/span\u003e, students in the experimental group achieved higher posttest scores and larger mean gains than students in the control group. Retention results presented in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e4.3\u003c/span\u003e indicate that students taught using the CNLPM maintained higher performance levels over time, with minimal score reduction between posttest and retention test.\u003c/p\u003e \u003cp\u003eGender-based analysis (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4.4\u003c/span\u003e) revealed that both male and female students in the experimental group demonstrated notable learning gains following the intervention, with only minor differences in mean gain scores.\u003c/p\u003e \u003cp\u003eOverall, the findings indicate that students exposed to the CNLPM outperformed those taught using the conventional teaching method in terms of achievement and retention.\u003c/p\u003e \u003c/div\u003e "},{"header":"Discussion","content":"\u003cp\u003eThis study investigated the effects of a Conversational Natural Language Processing Model (CNLPM) on students\u0026rsquo; achievement, retention, and learning equity in secondary school biology. The findings indicate that students exposed to the CNLPM demonstrated superior academic performance and stronger retention of biological concepts compared to their peers taught using the conventional lecture method. These results provide empirical support for the instructional value of AI-driven conversational learning tools in secondary science education.\u003c/p\u003e \u003cp\u003eEffect of the CNLPM on Student Achievement\u003c/p\u003e \u003cp\u003eThe significantly higher posttest scores and mean gain recorded by students in the experimental group suggest that the CNLPM effectively enhanced students\u0026rsquo; conceptual understanding of biology. This finding is consistent with previous research indicating that interactive and dialog-based instructional technologies promote deeper learning than passive instructional approaches (Graesser et al., 2005; Holmes et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). By allowing students to engage in continuous dialogue, ask questions, and receive immediate feedback, the CNLPM supported active cognitive processing, which is essential for meaningful learning.\u003c/p\u003e \u003cp\u003eThe conversational nature of the model likely encouraged self-explanation and reflection, processes known to strengthen conceptual understanding and reduce misconceptions (Chi et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Unlike conventional lecture-based instruction, which often limits student participation, the CNLPM provided individualized scaffolding that approximates one-on-one tutoring\u0026mdash;an instructional approach widely recognized for its effectiveness in improving student achievement (VanLehn, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). These features may explain the substantial learning gains observed among students in the experimental group.\u003c/p\u003e \u003cp\u003eEffect of the CNLPM on Student Retention\u003c/p\u003e \u003cp\u003eFindings related to retention revealed that students taught with the CNLPM maintained higher performance levels over time, with only minimal score reduction between the posttest and retention test. This suggests that the learning facilitated by the conversational model was not only effective in the short term but also durable. According to cognitive load theory, learning environments that reduce extraneous cognitive load and present information in structured, manageable formats are more likely to support long-term retention (Sweller et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). The CNLPM\u0026rsquo;s ability to simplify complex biological concepts and present explanations incrementally may have contributed to stronger schema formation.\u003c/p\u003e \u003cp\u003eIn addition, the dialogic structure of the model aligns with the spacing effect and retrieval practice principles, which emphasize repeated engagement with content to strengthen memory consolidation (Cepeda et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2006\u003c/span\u003e; Zepeda et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Through repeated questioning and clarification during interaction, students were likely prompted to revisit key concepts, thereby reinforcing learning and reducing forgetting.\u003c/p\u003e \u003cdiv id=\"Sec27\" class=\"Section2\"\u003e \u003ch2\u003eGender and Learning Outcomes\u003c/h2\u003e \u003cp\u003eThe gender analysis revealed that both male and female students benefited comparably from the use of the CNLPM, with only minor differences in mean gain scores. This finding suggests that the conversational model functioned as an inclusive instructional tool that supported learning across gender groups. Prior research indicates that technology-enhanced learning environments, when designed to be interactive and learner-centered, can reduce traditional participation gaps and promote equitable learning opportunities (UNESCO, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Zhai et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe absence of substantial gender disparities may be attributed to the personalized and self-paced nature of the CNLPM, which allows learners to engage with content without social pressure or bias often present in traditional classroom settings. This aligns with studies suggesting that AI-driven personalization can support equity by adapting instruction to individual learners\u0026rsquo; needs rather than demographic characteristics (Chen et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec28\" class=\"Section2\"\u003e \u003ch2\u003eEducational Significance of the Findings\u003c/h2\u003e \u003cp\u003eBeyond cognitive outcomes, the findings underscore the broader educational significance of conversational AI in science education. The observed improvements in achievement and retention suggest that AI-supported conversational models can address persistent instructional challenges, such as large class sizes, limited instructional time, and variability in teacher expertise\u0026mdash;issues commonly reported in secondary schools in low- and middle-income contexts (Luckin et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; UNESCO, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec29\" class=\"Section2\"\u003e \u003ch2\u003eEducational Significance of the Findings\u003c/h2\u003e \u003cp\u003eBeyond cognitive outcomes, the findings underscore the broader educational significance of conversational AI in science education. The observed improvements in achievement and retention suggest that AI-supported conversational models can address persistent instructional challenges, such as large class sizes, limited instructional time, and variability in teacher expertise\u0026mdash;issues commonly reported in secondary schools in low- and middle-income contexts (Luckin et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; UNESCO, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eEducational Significance of the Findings\u003c/p\u003e \u003cp\u003eBeyond cognitive outcomes, the findings underscore the broader educational significance of conversational AI in science education. The observed improvements in achievement and retention suggest that AI-supported conversational models can address persistent instructional challenges, such as large class sizes, limited instructional time, and variability in teacher expertise\u0026mdash;issues commonly reported in secondary schools in low- and middle-income contexts (Luckin et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; UNESCO, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWhile the study was conducted within a specific context, its findings align with international research demonstrating the effectiveness of conversational AI and intelligent tutoring systems in enhancing learning outcomes across diverse educational settings (Kohnke et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Zawacki-Richter et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Crompton \u0026amp; Burke, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). However, consistent with best research practice, these findings should be interpreted as contextually grounded, with broader applicability requiring further empirical validation.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eLimitations and Future Research\u003c/h3\u003e\n\u003cp\u003eDespite its positive outcomes, this study has several limitations. The research was conducted in six secondary schools within Niger State, Nigeria, which may limit the generalizability of the findings. Additionally, the nine-week intervention period captured short-term learning effects but may not reflect long-term outcomes.\u003c/p\u003e \u003cp\u003eThe intervention focused on a single biology topic\u0026mdash;cell division\u0026mdash;which limits conclusions about broader curricular applicability. Technological challenges, including intermittent power supply, also affected implementation.\u003c/p\u003e \u003cp\u003eFuture studies should employ longitudinal designs, involve larger and more diverse samples, examine additional biology topics and STEM subjects, and explore ethical considerations such as data privacy, transparency, and algorithmic bias in greater depth.\u003c/p\u003e "},{"header":"Conclusion and Policy Recommendations","content":"\u003cdiv id=\"Sec31\" class=\"Section2\"\u003e \u003cp\u003eThis study provides empirical evidence that integrating a Conversational Natural Language Processing Model (CNLPM) into secondary school biology instruction can enhance students\u0026rsquo; achievement and retention. By offering interactive, dialog-based learning support, the CNLPM addressed instructional challenges commonly associated with conventional teaching methods.\u003c/p\u003e \u003cp\u003eWhile the study was conducted in Niger State, Nigeria, the findings offer potential insights for similar educational contexts facing resource constraints and large class sizes. However, further research is required to validate the model\u0026rsquo;s effectiveness across diverse settings and levels of technological readiness.\u003c/p\u003e \u003cp\u003eFrom a policy perspective, education authorities should consider pilot-based integration of AI-assisted instructional tools, supported by teacher training, infrastructure development, and ethical governance frameworks. When thoughtfully implemented, conversational natural language processing model has the potential to support inclusive, equitable, and high-quality science education in alignment with global educational goals such as Sustainable Development Goal 4\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthical Approval\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was conducted in accordance with established ethical standards for educational research. Ethical approval was obtained from the relevant institutional research ethics committee prior to data collection. All procedures involving human participants complied with national and institutional guidelines for research ethics.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInformed consent\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eInformed consent was obtained from all participants involved in the study. For participants who were minors, informed consent was obtained from parents or legal guardians, and assent was obtained from the students themselves. Participation was voluntary, and participant were informed of their right to withdraw from the study at any time without penalty.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of Data and Materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated and/or analyzed during the current study are available from the corresponding author upon request. Due to ethical considerations and institutional policies, access to raw student interaction data is restricted to protect participant privacy.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe author declares that there are no competing interests, financial or non-financial, associated with this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe author was solely responsible for the conceptualization, system design, data collection, model development, data analysis, interpretation of results, and manuscript preparation. The author read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe author \u0026nbsp;gratefully acknowledges the support of participating schools, biology teachers, and students who contributed to the evaluation of the conversational learning system. Appreciation is also extended to academic colleagues who provided constructive feedback during the development of this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAI Use Acknowledgment:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRubric AI was used only for grammar refinement and language clarity. The tool did not generate, modify, or influence research data, analysis, or conclusions. All the ideas, results, and interpretations are the author\u0026rsquo;s work.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to Publish\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eInformed consent was obtained from all participants and from their parents or legal guardians for minors involved in this study. Consent was obtained for the publication of any potentially identifiable data. All data have been anonymized to protect participants\u0026rsquo; privacy.\u0026rdquo;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAbubakar U, Falode AA, Ibrahim HA. Redefining student assessment in Nigerian tertiary institutions: The impact of AI technologies on academic performance and developing countermeasures. Adv Mob Learn Educational Res. 2024;4(2):1149\u0026ndash;59.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAyodele AO, Adebunmi A. Early childhood development teachers\u0026rsquo; perceptions on the use of technology in teaching young children. South Afr J Child Educ. 2020;10(1):1\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBaidoo-Anu D, Ansah OL. Education in the era of generative artificial intelligence (AI): Understanding the potential benefits of ChatGPT in promoting teaching and learning. SSRN. 2023. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2139/ssrn.433748\u003c/span\u003e\u003cspan address=\"10.2139/ssrn.433748\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCanaran \u0026Ouml;. Navigating self-directed learning in the digital era. Exploring adult education through learning theory. IGI Global; 2025. pp. 163\u0026ndash;90.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCarney S. Reimagining our futures together: A new social contract for education. UNESCO; 2022.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCepeda NJ, Pashler H, Vul E, Wixted JT, Rohrer D. Distributed practice in verbal recall tasks: A review and quantitative synthesis. Psychol Bull. 2006;132(3):354\u0026ndash;80.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen L, Chen P, Lin Z. Artificial intelligence in education: A review. IEEE Access. 2022;10:75264\u0026ndash;78. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ACCESS.2022.3188796\u003c/span\u003e\u003cspan address=\"10.1109/ACCESS.2022.3188796\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChi MTH, Kang S, Yaghmourian DL. Why students learn more from dialogue than from monologue. Instr Sci. 2018;46(1):191\u0026ndash;214.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCollins M, Shaw A. The future of gender studies in education. Future Educ J. 2025;11(2):330\u0026ndash;45. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2121/fuedu.2025.330\u003c/span\u003e\u003cspan address=\"10.2121/fuedu.2025.330\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCrompton H, Burke D. Artificial intelligence in education: A systematic review of research trends. Computers Education: Artif Intell. 2023;4:100113. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2023.100113\u003c/span\u003e\u003cspan address=\"10.1016/j.caeai.2023.100113\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEnebechi RI. Effect of inquiry-based learning approach on senior secondary school students\u0026rsquo; retention in biology. Br Int J Educ Social Sci. 2021;8(8):9\u0026ndash;19.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEssel HB, Opoku D, Amoako I. Gender differences in digital learning adoption among African students. Afr J Educ Technol. 2024;14(1):41\u0026ndash;58.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEze AN, Ezenwafor JI, Onwusa SC. Effect of computer-based instruction on students\u0026rsquo; achievement and retention of high- and low-achieving auto-mechanics technology students in technical colleges. Int J Sci Eng Res. 2020;11(8):1631\u0026ndash;45.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGarcia M, Walker J. The intersection of gender and educational technology. J Educational Technol. 2023;29(2):244\u0026ndash;60.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGraesser AC, Kohnke L, Lin Y. Conversational agents and learning: A review of educational dialogue systems. Educational Psychol Rev. 2022;34(3):1321\u0026ndash;48. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10648-021 09633-8\u003c/span\u003e\u003cspan address=\"10.1007/s10648-021 09633-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGupte T, Watts FM, Schmidt-McCormack JA, Zaimi I, Gere AR, Shultz GV. Students\u0026rsquo; meaningful learning experiences from participating in organic chemistry writing-to-learn activities. Chem Educ Res Pract. 2021;22(2):396\u0026ndash;414.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHolmes W, Bialik M, Fadel C. Artificial intelligence in education: Promises and implications for teaching and learning. Center for Curriculum Redesign; 2022.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIkuesewo-Akinbami A, Acheme RO. Evaluation of the role of radio in promoting Yoruba language by Adaba FM among traders in Oja-Oba market, Akure. Nasarawa J Multimedia Communication Stud. 2025;7(1):275\u0026ndash;90.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim D, Jung E, Yoon M, Chang Y, Park S, Kim D, Demir F. Exploring the structural relationships between course design factors, learner commitment, self-directed learning, and intentions for further learning in a self-paced MOOC. Comput Educ. 2021;166:104171.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKohnke L, Moorhouse BL, Zou D. Chatbots in education: A systematic review. Educ Inform Technol. 2021;26(4):4299\u0026ndash;337. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10639-021-10418-8\u003c/span\u003e\u003cspan address=\"10.1007/s10639-021-10418-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLuckin R, Holmes W, Griffiths M, Forcier LB. Intelligence unleashed: An argument for AI in education. Pearson Education; 2021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcKenney S, Reeves TC. (2025). Educational design research for relevant and robust scholarship. J Comput High Educ, 1\u0026ndash;25.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eParker SK, Ballard T, Billinghurst M, Collins C, Dollard M, Griffin MA, Walsh T. Quality work in the future: New directions via a coevolving sociotechnical systems perspective. Aust J Manage. 2025. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/03128962251331813\u003c/span\u003e\u003cspan address=\"10.1177/03128962251331813\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Advance online publication.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRazak RA, Alias NF, Idris AY, editors. Qualitative insights through applied cognitive task analysis. IGI Global; 2025.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRiedel R, Vialle W, Pearson P, Oades LG. Quality learning and positive education practice: The student experience of learning in a school-wide approach to positive education. Int J Appl Psychol. 2020;5(1):53\u0026ndash;75.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRobinson L. Innovative practices to address gender disparities. Innov Educ. 2025;22(4):55\u0026ndash;70. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.7890/innoedu.2025.055\u003c/span\u003e\u003cspan address=\"10.7890/innoedu.2025.055\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSadera JRN, Torres RYS, Rogayan DV. Challenges encountered by junior high school students in learning science: Basis for an action plan. Univers J Educational Res. 2020;8(12A):7405\u0026ndash;14.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSweller J, Ayres P, Kalyuga S. Cognitive load theory. 2nd ed. Springer; 2019.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUNESCO. Reimagining our futures together: A new social contract for education. UNESCO Publishing; 2021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUNESCO. Guidance on generative AI in education and research. UNESCO Publishing; 2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVanLehn K. The relative effectiveness of human tutoring, intelligent tutoring systems, and other tutoring systems. Educational Psychol. 2011;46(4):197\u0026ndash;221.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVictor NCODE, Ayoko O. Information communication technology and academic staff job performance in tertiary institutions. Manuscript in preparation; 2025.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWest African Examinations Council. Chief examiners\u0026rsquo; report for the West African Senior School Certificate Examination (WASSCE). WAEC; 2022.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWood K. Gender and equity in technology-enhanced classrooms: A meta-analysis. Comput Educ. 2023;194:104678. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.compedu.2023.104678\u003c/span\u003e\u003cspan address=\"10.1016/j.compedu.2023.104678\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYusuf H. (2023). \u003cem\u003eEffects of multimedia instruction on SS1 biology students\u0026rsquo; academic achievement and retention in Bosso Local Government Area\u003c/em\u003e (Doctoral dissertation). Federal University of Technology, Minna. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://irepo.futminna.edu.ng\u003c/span\u003e\u003cspan address=\"http://irepo.futminna.edu.ng\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZawacki-Richter O, Mar\u0026iacute;n VI, Bond M, Gouverneur F. (2021). Systematic review of research on artificial intelligence applications in higher education. International Journal of Educational Technology in Higher Education, 18(1), 1\u0026ndash;27. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s41239-021-00238-3\u003c/span\u003e\u003cspan address=\"10.1186/s41239-021-00238-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZekeik H, Chahbi M, Sefian ML, Bakkali I. Augmented reality and virtual reality in education: A systematic narrative review on benefits, challenges, and applications. EURASIA J Math Sci Technol Educ. 2025;21(9):em2699.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZepeda CD, Een E, Butler AC. (2024). The mnemonic effects of retrieval practice. In \u003cem\u003eOxford Research] Encyclopedia of Education.\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZepeda CD, et al. Spacing and retrieval practice effects on learning: A meta-analytic review. Educational Psychol Rev. 2024;36(1):1\u0026ndash;29.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhai X. ChatGPT and the future of education: Opportunities and challenges. Front Educ. 2022;7:1046789. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/feduc.2022.1046789\u003c/span\u003e\u003cspan address=\"10.3389/feduc.2022.1046789\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhai X, Yin Y, Pellegrino JW, Haudek KC, Shi L. Applying machine learning in science education research: A systematic review. J Res Sci Teach. 2023;60(6):748\u0026ndash;75. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/tea.21806\u003c/span\u003e\u003cspan address=\"10.1002/tea.21806\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTables 4.1 to 4.4 are available in the Supplementary Files section.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"discover-artificial-intelligence","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"diai","sideBox":"Learn more about [Discover Artificial Intelligence](https://www.springer.com/44163)","snPcode":"","submissionUrl":"","title":"Discover Artificial Intelligence","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Conversational NLP model, technology-enhanced learning, artificial intelligence, secondary school, educational technology, biology","lastPublishedDoi":"10.21203/rs.3.rs-9301692/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9301692/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eThe integration of digital technology into education has significantly transformed how students\u0026rsquo; learning complex scientific subjects such as biology.\u003c/p\u003e\u003ch2\u003eObjective\u003c/h2\u003e \u003cp\u003eThis study developed and evaluated a Conversational Natural Language Processing Model (CNLPM) designed to enhance secondary school biology learning in Niger State, Nigeria.\u003c/p\u003e\u003ch2\u003eMethology:\u003c/h2\u003e \u003cp\u003eThe model was created via the ADDIE instructional design framework and evaluated via a quasi-experimental research design to determine its effects on students\u0026rsquo; achievement, retention, and interest. A developmental research approach was adopted, combining the formative phases of analysis, design, development, implementation, and evaluation with an experimental validation stage. Six secondary schools were selected via a multistage sampling technique, yielding an intact sample of 261 senior secondary II biology students (146 males and 115 females). Three schools were assigned to the experimental group, which received instruction via the CNLPM, whereas the remaining three formed the control group and were taught via the conventional lecture method.\u003c/p\u003e\u003ch2\u003eResult\u003c/h2\u003e \u003cp\u003eThe findings revealed that students exposed to the CNLPM significantly outperformed their counterparts in the control group (mean gains of 49.46 vs. 28.14). The model also contributed to improved retention and heightened interest in biology learning. These results indicate that conversational AI tools can meaningfully complement traditional instructional strategies. This study provides evidence supporting the adoption of technology-enhanced learning approaches to improve biology education in secondary schools.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eAlthough the study was conducted in Niger State, Nigeria, the challenges addressed\u0026mdash;limited access to interactive learning resources, variability in teacher expertise, and low student engagement\u0026mdash;are common across many secondary education systems worldwide. Consequently, the findings offer transferable insights into the potential of conversational AI tools for enhancing science education globally, providing scalable strategies to improve learning outcomes in diverse educational contexts.\u003c/p\u003e\u003ch2\u003eUnique Contribution:\u003c/h2\u003e \u003cp\u003eThis study makes a unique contribution by empirically demonstrating how an offline-capable conversational AI model, designed for resource-constrained secondary school contexts, can simultaneously enhance conceptual understanding, learner motivation, and equitable access to biology education, while directly advancing the goals of United Nations Sustainable Development Goal 4 (Quality Education) through inclusive and scalable AI-supported instruction\u003c/p\u003e","manuscriptTitle":"Development and Evaluation of a Conversational Natural Language Processing Model for Technology-Enhanced Biology Learning in Secondary Schools","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-04 16:05:13","doi":"10.21203/rs.3.rs-9301692/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-13T04:40:35+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"162442869080406078904587476479492775478","date":"2026-05-12T07:27:51+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"218176257310510850210211882167513369879","date":"2026-05-04T01:01:41+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-22T23:39:12+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-22T23:34:25+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-04-13T12:19:41+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-12T03:36:00+00:00","index":"","fulltext":""},{"type":"submitted","content":"Discover Artificial Intelligence","date":"2026-04-12T03:31:47+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"discover-artificial-intelligence","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"diai","sideBox":"Learn more about [Discover Artificial Intelligence](https://www.springer.com/44163)","snPcode":"","submissionUrl":"","title":"Discover Artificial Intelligence","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"dae65221-1142-4bf7-80a7-e7e2a479d3d4","owner":[],"postedDate":"May 4th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-13T04:40:35+00:00","index":31,"fulltext":""},{"type":"reviewerAgreed","content":"162442869080406078904587476479492775478","date":"2026-05-12T07:27:51+00:00","index":30,"fulltext":""},{"type":"reviewerAgreed","content":"218176257310510850210211882167513369879","date":"2026-05-04T01:01:41+00:00","index":27,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-04T16:05:13+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-04 16:05:13","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9301692","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9301692","identity":"rs-9301692","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}