Self-Directed Language Learning: An Integrated Adaptive Method

Self-directed language learning outside everyday target-language interaction requires a coherent pathway from available materials to purposeful practice and later return. This conceptual study develops an integrated adaptive method for adults, with English learning in Vietnam as its focal application. The method unifies learner-supplied material preparation, learner activity, source-qualified feedback and persistent planning. Selective evidence synthesis grounds four design propositions concerning assistance, task-conditioned dependencies, access constraints and revisable source links. A competency/context network connects linguistic resources, regulation, motivation and practical conditions. A multi-session module and re-entry case specify continuation under changing support and access; an auditory case separates candidate difficulties and their evidence requirements. Six authored case pairs expose decisions that change with those conditions. A book example traces relationships among source objects and an adapted task. A worked material-to-review pathway connects an attempt, correction of a mislinked answer key, focused repair and a later task. Synthetic Bayesian knowledge tracing illustrates sensitivity to observation assumptions, while cash/time grids expose utilization and checking-burden thresholds. Two synthetic scenario waves used 5,000 request entries and 200 fictional profiles to examine response coverage, output variation and conditional package judgments. These results characterize the instrument and generator; whole-method learning effectiveness remains outside the observations. The contribution is a unified, theoretically grounded method with inspectable learning pathways, conditional feasibility and reproducible analytical artifacts. Version 1.0 contains the same conceptual and methodological study in English and Vietnamese (35 pages each, 46 references, 13 figures and 18 tables). The English PDF is the primary preview file. Independently buildable LaTeX sources and selected analytical artifacts accompany the manuscripts. This research was self-funded by the author. The author is the founder of and affiliated with Elixverse Labs, which develops AI-related systems relevant to the discussion. The author directed the research questions, methodological choices and revisions and takes responsibility for the work. AI tools assisted research, drafting, analytical code and figure preparation, and translation. The model-generated scenario ratings are explicitly synthetic; no human learner survey or intervention is reported. This is a conceptual preprint, not an independently peer-reviewed article. The synthetic scenario waves characterize the instrument and generator; they do not establish population preferences or whole-method learning effectiveness. Authored manuscripts, original figures and authored analytical materials are CC BY 4.0. The internal workspace snapshot, credentials, raw third-party reference publications and the source book are not part of this public deposit.