Developing Computational Thinking: A Multi-Modal Cognitive Progression Architecture

Introductory computing education in primary schools (Kindergarten through Grade 5, Ages 6–12) frequently suffers from an acute pedagogical dilemma: early learners either engage in unconstrained block-based tinkering (Scratch) that reinforces dysfunctional algorithmic habits (bricolage, spaghetti concurrency, and global state fragmentation), or are prematurely subjected to authentic textual syntax (Python) that exceeds their orthographic reading-fluency and working-memory thresholds. This monograph presents a unified, multi-modal cognitive progression architecture for K–5 computing education that bridges code-free computational thinking to authentic textual programming fluency. Synthesizing findings across eighteen foundational pedagogical sub-studies in the Elix Researches Program, we formalize a three-stage developmental ladder anchored in Bruner's Enactive-Iconic-Symbolic representation theory, Cognitive Load Theory, Notional-Machine formation, and the PRIMM instructional model:(1) Stage I (Grades 1–3, Ages 6–9, Band B0) deploys code-free (phi-code) unplugged activities, including an 18-exercise tactile battery and the physical "Cup-and-Strip" memory model, establishing deterministic program-counter and state-mutation intuitions with zero digital hardware;(2) Stage II (Grades 4–5, Ages 8–11, Band B1) institutes disciplined block-based programming governed by three anti-bricolage invariants (Single Dispatcher, State Centralization, Mandatory Custom Blocks) paired with a concurrent dual-pane Scratch-to-Python Abstract Syntax Tree (AST) mutual-translation bridge and a 7-criterion quantitative transition framework; and(3) Stage III (Grade 5+, Ages 10–12, Band B2) achieves authentic textual algorithmic fluency via the Luggage-Tag variable model, bounded iteration precedence (for before while), an explicit 4-step "Bug Hunter" debugging protocol, 4-tier Parsons puzzle scaffolded fading, and a minimalist zero-install browser sandbox progression. We establish developmental readiness floors based on Chall's reading stages (>= 90 WPM expository floor) and keyboard motor ergonomics, formulate a 28-misconception taxonomy cross-examined with seven Austroasiatic/Vietnamese linguistic interference hypotheses (H1–H7), articulate four delivery modalities (M1–M4) tailored for low-resource schools without home computers, and demonstrate direct curricular operationalization across a 104-lesson modular framework aligned with Vietnamese (GDPT 2018), US (CSTA K–12), and UK statutory computing standards.