AI-Powered Map Reveals How Multimedia Learning Research Exploded Into a New Era

Multimedia learning, the science of how people learn when words and pictures work together on a screen, has quietly become one of the most consequential research fields of the digital age. Every instructional video, interactive simulation, and online course draws, whether its designers know it or not, on decades of research into how visual and verbal information combine in the human mind. Now, a sweeping new analysis has charted the entire landscape of that field, and its findings reveal a discipline that has not only grown explosively but has fundamentally transformed what it studies, how it studies it, and where it is heading next.

Hasan Çoruk and Fatih Erdoğdu of Zonguldak Bülent Ecevit University in Turkey set out to solve a persistent problem in the literature: while individual studies of multimedia learning abound, no one had systematically traced how the field’s themes have shifted over a quarter century. Their solution was to combine two powerful analytical approaches. The first, bibliometrics, applies quantitative techniques to publication data, counting outputs, tracking citations, and mapping networks of scholarly influence. The second, topic modeling, uses machine learning to automatically detect hidden thematic structures in thousands of documents at once. Together, these methods allowed the researchers to see not just how much the field has published, but what it has actually been talking about.

The scale of the undertaking was formidable. The researchers drew on 7,412 publications indexed in the Scopus database between 2000 and 2024, a corpus spanning everything from cognitive psychology experiments to medical education interventions. Scopus, widely regarded as one of the most comprehensive multidisciplinary bibliographic databases available, provided the broad coverage necessary to capture a field that spills across education, psychology, computer science, and the health sciences. For the topic modeling component, the team employed Latent Dirichlet Allocation, or LDA, a probabilistic technique that treats each document as a mixture of topics and each topic as a distribution of words. By running this algorithm across the full corpus, the researchers could identify coherent thematic clusters that no single keyword search could reveal.

The temporal trends alone tell a striking story. Publication activity in multimedia learning increased significantly after 2018, a surge that coincides with, and was likely accelerated by, the global shift toward digital education during the COVID-19 pandemic. When schools and universities worldwide were forced into emergency remote teaching almost overnight, the theoretical and practical questions that multimedia learning researchers had been investigating for decades suddenly became urgent problems for millions of teachers. The analysis suggests this period did more than inflate publication counts; it pushed the field into new interdisciplinary domains, as researchers from medicine, nursing, chemistry, and vocational education began applying multimedia principles to their own specialized contexts.

The LDA analysis distilled the field’s enormous thematic diversity into four dominant clusters, and the picture they paint is genuinely revealing. The most popular research area, titled Design and Development of Interactive Teaching Tools, has remained the field’s backbone even as its relative popularity has declined over time. This cluster encompasses the practical craft of building the applications, modules, and platforms through which multimedia instruction actually reaches learners, from interactive language learning systems to educational video games and virtual laboratory environments. Its enduring dominance reflects a field that has always been intimately connected to the technologies it studies.

The second cluster, Interactive Video Applications Model for Teaching and Learning, stands out for a different reason: adaptability. According to the analysis, this theme has proven remarkably capable of evolving beyond basic information presentation to accommodate various user needs. Video, once a passive one-way medium, has been transformed by interactivity into something far richer, incorporating features such as pause controls, embedded tasks, and adaptive pacing. Recent studies in this tradition examine questions with direct practical stakes, such as whether task instructions and the mere availability of a pause button change how effectively people learn from educational videos. The theme’s evolution mirrors the broader technological maturation of online video itself.

The third cluster, Cognitive Effects of Visual and Text-Based Instructional Design, represents the field’s theoretical heart. Research in this area categorizes studies of the cognitive processes that underlie learning from multimedia, drawing on foundational frameworks that emerged in the late twentieth century. These include cognitive load theory, which describes the limits of working memory during problem solving and instruction, and dual coding approaches, which propose that verbal and visual information are processed through partly distinct channels. Richard Mayer’s cognitive theory of multimedia learning, first articulated in the 1990s and still actively refined today, synthesized these ideas into testable design principles, such as the observation that people learn better when corresponding words and pictures are presented together rather than separately. The persistence of this cluster demonstrates that questions about how the mind handles combined visual and verbal input remain as scientifically alive as ever.

The fourth and smallest cluster may be the most strategically significant. The Role of Educational Interventions in Medical Care and Education occupies limited space in the overall literature, yet the analysis identifies it as disproportionately important because it facilitates interdisciplinary multimedia integration in specific domains such as health, life skills, and vocational competencies. The supporting evidence is tangible: multimedia education programs have been tested in randomized controlled studies with patients recovering from surgery, used to improve knowledge and self-efficacy around disease prevention among pregnant women, and embedded in simulation-based training for cardiopulmonary resuscitation among healthcare professionals. In these contexts, multimedia is not merely a convenience; it can affect clinical outcomes, professional competence, and ultimately patient safety, which gives even a small body of research considerable real-world leverage.

The methodological marriage of bibliometrics and topic modeling deserves attention in its own right. Conventional bibliometric analyses excel at quantifying growth and mapping collaborations, but they often struggle to capture thematic substance, relying instead on author-assigned keywords that can be inconsistent and incomplete. Topic modeling addresses this weakness by reading the actual text of thousands of abstracts and titles, allowing themes to emerge from the data rather than from pre-existing classification schemes. The approach has been increasingly adopted across educational research, appearing in recent analyses of technology use in mathematics education and of multimodal teaching trends more broadly. By applying it to multimedia learning specifically, Çoruk and Erdoğdu have provided the field with something it previously lacked: a data-driven map of its own intellectual territory, showing which themes anchor the field, which are rising, and which are fading.

For researchers, the implications are clear. The field’s post-2018 expansion into interdisciplinary territory suggests that the most fertile ground may lie at the boundaries, where multimedia principles meet domain-specific challenges in medicine, vocational training, and beyond. For practitioners, the map offers a reminder that effective digital instruction rests on a deep empirical foundation, and that design choices about video interactivity, visual layout, and cognitive load are grounded in decades of experimental evidence rather than intuition. As artificial intelligence, virtual reality, and adaptive systems continue to reshape how educational content is created and delivered, this comprehensive analysis provides both a record of how multimedia learning research became what it is today and a scientific baseline for anticipating where it will go next. The study, published in Current Psychology, involved no external funding, and the authors report no competing interests, with both researchers contributing equally to the work.

Subject of Research: Bibliometric and topic model analysis of multimedia learning research trends from 2000 to 2024

Article Title: Mapping multimedia learning research: a bibliometric and topic model analysis

Article References: Çoruk, H., & Erdoğdu, F. (2026). Mapping multimedia learning research: a bibliometric and topic model analysis. Current Psychology, 45(18), Article 1505. https://doi.org/10.1007/s12144-026-10051-6

Image Credits: AI Generated

DOI: 10.1007/s12144-026-10051-6

Keywords: multimedia learning, bibliometrics, topic modeling, LDA, cognitive load theory, educational technology, interactive video, instructional design, medical education, Scopus, research trends, Current Psychology