Decision Criteria for Scratch as a Bridge to Text-Based Programming: A Literature Synthesis

Block-based programming, exemplified by Scratch [1], [2], is the de facto entry environment for K-12 informatics in most jurisdictions including Vietnam under the 2018 General Education Curriculum [3]. The international literature on the block-to-text transition is split: some studies report block-based introductions producing motivational and short-run conceptual gains [4], [5], while others document durable Scratch-induced habits that persist into the text phase and impede transition [6], [7], [8]. The contradiction is not resolved by averaging effect sizes; it dissolves when the question is reframed from "is Scratch good or bad?" to "under which conditions does Scratch help, and under which does it hurt?". This paper synthesises the 2009–2022 literature on Scratch design intent, Scratch-induced habits, block-versus-text comparative studies, hybrid environments (Pencil Code, Hedy), transition scaffolding (Moors et al., PRIMM), notional-machine theory, and assessment literature, and extracts a decision-criteria framework. Seven decision criteria are specified — learner age and reading fluency, prior text-based exposure, target text language, instructor blockand-text dual familiarity, curriculum duration, classroom infrastructure, and parent-and-community Scratch-acceptance — each with definition, direction-of-effect on transition outcome, evidence anchors, threshold value, and failure mode if misapplied. The framework is engineered as a checklist that a curriculum designer can apply chapter-by-chapter rather than a global yes-or-no verdict. The framework is mapped onto two elixresearches textbook series: the From Scratch to Python (FS2P) series [9], which assumes Scratch entry and engineers an explicit transition, and the Python Pathway series, which skips Scratch entirely for a gifted-cohort target population. The synthesis is purposive, not exhaustive; validation comes from audit-loop review against the programme rubric, not from empirical trial. Three open questions are flagged for downstream Layer A research.