The article examines the integration of blended learning into the teaching of foreign languages to students of non-linguistic (technical) majors under conditions of reduced classroom time and increasing requirements for communicative competence, with emphasis on aligning pedagogical design with the limitations of working memory and the need to foster learning autonomy. It is shown that transferring materials to a digital environment without considering cognitive load mechanisms provokes overload, intensifies the split-attention effect, and reduces effectiveness, especially with professionally oriented vocabulary that increases the intrinsic complexity of the learning content. A flipped classroom model with elements of adaptive microlearning is presented, implementing content segmentation, elimination of redundancy, dual coding, simplified navigation, and step-by-step task structure; autonomy was supported by route sheets, embedded self-checking, and reflection tools. Comparison with a variant where the electronic environment functioned primarily as a repository was conducted on a sample of 348 first- and second-year students with an initial English proficiency level around B1; psychometric scales for subjective load and self-regulation, standardized language tests, and statistical procedures (t-test, ANOVA, correlations) were applied. At the baseline, groups did not differ (entry test around 58/100; mental and temporal load around 68–72/100; metacognitive self-regulation around 3.4/7), indicating high resource intensity of typical digital tasks. After implementing the cognitive-oriented design in the experimental group, a reduction in mental demand to 54.87 vs. 65.21, temporal to 51.43 vs. 68.92, effort to 58.66 vs. 71.05, frustration to 34.29 vs. 62.47, as well as improvement in subjective performance evaluation were recorded. Final language proficiency improved on the lexical-grammatical test (76.32 vs. 64.89), reading (81.05 vs. 68.14), and listening (74.43 vs. 59.76), with simultaneous reduction in result variability, indicating leveling of achievements. Components of autonomy significantly strengthened: planning (gain 1.72), monitoring (1.67), resource and time management (1.53); a pronounced negative correlation between planning and perceived time pressure (r = −0.68) was revealed. The obtained data are interpreted as a systemic effect of cognitive ergonomics: reducing extraneous load frees resources for generative processing and metacognitive control, transforming the digital environment from a source of barriers into a factor intensifying language learning.