<![CDATA[Curriculum changes in Indonesia often raise questions about the continuity of thematic learning in Madrasah Ibtidaiyah (MI) and elementary schools (SD). Changes in curriculum structure do not necessarily mean that the underlying principles of learning are abandoned. This study aims to trace the development of the principle of integration from thematic learning to Deep Learning and to formulate a framework of integrated learning that is relevant to current developments in MI/SD education. This study employed a literature review and policy document analysis by examining relevant scholarly articles, theoretical works, and educational policy documents concerning integrated learning, thematic learning, curriculum development, and Deep Learning. The analysis was conducted through thematic mapping, comparison of forms of integration across curriculum developments, policy document analysis, and conceptual synthesis. The findings show that the principle of integration remains an important part of learning in MI/SD, although its form of implementation has changed over time. In the 2013 Curriculum, integration was realized through thematic learning, with themes serving as the main organizing element, although implementation still showed forms of formal-administrative integration. The Merdeka Curriculum provides greater flexibility through learning outcomes, contexts, IPAS, and learning experiences. Deep Learning further strengthens connections among concepts, meaningful learning experiences, and holistic student development. Based on these findings, this study proposes a substantive-holistic integration framework consisting of three interrelated pillars: epistemological, pedagogical, and assessment. This framework provides a basis for understanding the continuity of the principle of integration while helping teachers and prospective teachers design more meaningful and integrated learning experiences that are responsive to the development of MI/SD students. Keywords: substantive-holistic integration; conceptual reconstruction; thematic learning; Deep Learning; MI/SD]]>
Copyrights © 2026
