The widespread adoption of technology-enhanced learning in various knowledge disciplines has pushed forward the development of information technology-assisted media for language learning and teaching. However, most of the existing electronic-learning (e-learning) solutions have underexplored and under-addressed given specific characteristics of grammar learning, which is one of the most demanding areas of language education. The lack of pedagogically informed instructional design to enhance learning performance on the current system can result in low motivation and engagement due to an imbalance and excessive increase of the cognitive load. This paper attempts to address these deficiencies posed by the existing systems by proposing smart communication networks that are driven by the student learning experience to manage cognitive load in the context of grammar learning. The e-grammar learning networks serve as a collaborative learning platform that combines a pedagogically informed instructional model named attention, relevance, confidence, and satisfaction (ARCS) and cyber interaction among teaching/learning agents. From the technological perspective, our numerical simulations demonstrate the desirable performance indicators of the proposed networks to facilitate information exchange and learning. From the education perspective, our empirical studies show that the overall smart network-enabled e-grammar learning system has desirable characteristics to motivate learners (<inline-formula> <tex-math notation="LaTeX">m = 3.78 </tex-math></inline-formula>) and manage their overall cognitive load ( m = 1.73), which suggest the promising capability of the proposed system.