Mapping the Intellectual and Conceptual Structure of Early Childhood Science Education: Epistemic Continuity, Thematic Evolution, and Emerging Research Fronts

Abstract

This study maps the development, intellectual roots, and conceptual transformation of early childhood science education between 1990 and July of 2026 through performance analysis and science mapping of 1147 English-language research articles indexed in the Scopus Social Sciences subject area. Analyses conducted in R with bibliometrix/Biblioshiny examined performance indicators, collaboration, direct citation paths, co-citation patterns, and author-keyword structure. Publication output increased sharply after 2019: 635 articles (55.4%) appeared between 2019 and July 2026, including 537 (46.8%) in the completed years 2019–2025 and 98 in the partial 2026 period. Research in Science Education and the International Journal of Science Education were the most productive sources, while Fleer ranked first in both publication count and local citations. Internationally co-authored articles represented 12.57% of the corpus according to bibliometrix’s Multiple Country Publications ratio. The historiograph retained a connected sequence of influential works. Under the baseline settings, the displayed co-citation network was partitioned into four Louvain groups concerned with cultural-historical theory, empirical science pedagogy, nature of science and standards, and cognitive development. Modularity was low (Q = 0.13), and the number of groups was parameter-sensitive, while author keywords indicate that inquiry, play, and teacher pedagogy remain prominent alongside the more recent visibility of STEM, computational thinking, and computer science education. Together, these patterns suggest continuity in established research traditions alongside an expanding thematic vocabulary. They also motivate a hypothesis about core–periphery organization, although no core–periphery model was fitted. An expanded query (n = 2739) reproduced the post-2019 rise in publication counts and was not used for the primary maps. The findings are limited to the retrieved corpus and do not estimate the prevalence of these themes across the wider STEM, robotics, coding, computational thinking, or artificial intelligence studies.