The DX+ conceptual framework for AI-supported inclusive block programming in higher education
Abstract
Learning to program remains disproportionately difficult for students with dyslexia and dyslexia-related reading difficulties, for whom the text-dense, syntax-sensitive nature of conventional integrated development environments amplifies an already heavy decoding and working-memory load. This article proposes DX+ , a conceptual design framework - proposed but not yet implemented - for an AI-supported visual block-programming environment that lets learners assemble algorithms from syntax-valid visual blocks and automatically transpile them into Python, C++, and Java. The contribution is the principled integration of three strands that have largely been pursued in isolation-block-based programming, Universal Design for Learning (UDL), and code-oriented large language models (LLMs)-into a single four-layer architecture comprising an inclusive presentation layer, a visual block-programming layer, a block-to-source transpilation engine, and an adaptive AI tutoring module designed around a code model intended for parameter-efficient fine-tuning. We articulate the design problem, position DX+ against existing tools - including hybrid block-text editors and other multi-language block platforms - through a comparative analysis, and present its anticipated theoretical and practical implications as design hypotheses for future empirical validation. The work addresses a clear gap in localised, dyslexia-oriented programming tools and offers a coherent, theoretically grounded model for inclusive information-technology (IT) education.