Early vs Late Recursion Instruction in Pre-Tertiary Programming: A Sequencing Synthesis for Vietnamese Context

Across four decades of computing-education research the introductory curriculum question when should recursion be introduced — alongside loops while iteration is still being learned, or only after iteration fluency is established has been answered by two distinct traditions that rarely talk to each other. The first tradition is the procedural CS1 mainstream, descended from the Pascal / C / Java introductory-programming canon, which introduces recursion late — typically in the second half of CS1 or in CS2 — after iteration is fluent and the call-and-return semantics of non-recursive functions is established. The second tradition is the functional-pedagogy line, descended from Scheme / Lisp / ML teaching and codified in How to Design Programs (Felleisen, Findler, Flatt & Krishnamurthi), which introduces recursion early — sometimes before iteration is even available in the chosen language — anchored on the structural-recursion designrecipe and the worked-example-and-fading instructional pattern. The Vietnamese context complicates the question further. The mainstream Vietnamese curriculum (MOET 2018) follows the procedural late-recursion ordering. The Vietnamese chuyên Tin (selective high-school informatics) tradition, by contrast, introduces recursion early and treats it as a foundational competitiveprogramming primitive — the canonical chuyên-Tin texts (Hồ Sĩ Đàm; Lê Minh Hoàng) and the wider Olympiad-training literature (IOI Syllabus / Forišek 2007) place divide-and-conquer, backtracking, and recursive enumeration squarely in the early curriculum. This paper synthesises the ACT-R production-rule account of recursive-skill acquisition, the Wiedenbeck empirical iteration-versus-recursion comparisons, the Kahney / GötschiSanders-Galpin mental-model tradition, the Hamouda Basic Recursion Concept Inventory, the McCauley / Lister / Lewis tracing-difficulty evidence, the HtDP / TeachScheme! designrecipe pedagogy, the Turbak early-recursion CS1 report, the Caspersen-Bennedsen instructional-design framework, Sweller cognitive-load theory, the Sajaniemi roles-of-variables substrate, the Rinderknecht 2014 survey, and the Vietnamese canonical sources. The paper delivers three contributions. First, a literature verdict: for absolute novices working in an imperative-first language (Python in this programme's case), the prerequisitefluency rule plus the misconception-density evidence support late recursion as the default ordering. Second, a bifurcated recommendation: late-recursion default for FS2P / PP Volume 2–3; early recursion permitted and pedagogically defensible at PP Volume 5 (the competitive-programming track), where the prerequisite-fluency conditions are met by curriculum design. Third, a worked-example specification at each entry point: HtDP structural-recursion design-recipe for the early-recursion track, fixed-collection-iteration-to-recursion bridge for the laterecursion track. Per the no-cohort scope, deployment of the