BEYOND AUTOMATED TEXT GENERATION:

Abstract

The rapid integration of Generative Artificial Intelligence (GenAI) has transformed instructional design in English as a Foreign Language (EFL) education. While Large Language Models (LLMs) offer unprecedented efficiency in generating reading texts, vocabulary lists, and lesson plans, their mainstream educational usage often relies on superficial, ad-hoc prompting. This erratic approach frequently yields outputs that lack pedagogical scaffolding, cultural relevance, or alignment with established language acquisition frameworks. Consequently, there is an urgent need to shift the paradigm from passive, automated text generation to structured, intentional human-AI communication in curriculum development. This conceptual paper introduces a robust, teacher-facing framework designed to guide EFL educators in communicating systematically with GenAI tools. By treating prompt engineering as a form of specialized, pedagogical dialogue, the framework aims to ensure that AI-generated instructional materials strictly adhere to specific linguistic levels, task-based objectives, and communicative language teaching (CLT) principles. Utilizing a design-based conceptual approach, this study outlines a structured system-prompting matrix. The framework operates on four core linguistic constraints: target pedagogical framework alignment, lexicogrammar profiling, localized contextualization, and interactive scaffolding. To evaluate the framework's viability, a qualitative walkthrough is presented, illustrating how an instructor iteratively prompts an LLM to design a hyper-personalized, context-specific reading unit and an accompanying vocabulary assessment for advanced EFL learners. The generated artifacts are evaluated against traditional instructional design rubrics to determine their pedagogical soundness. The paper demonstrates that when educators engage in structured, constraint-driven dialogue with GenAI, the resulting materials exhibit higher instructional alignment and fewer algorithmic errors or generalizations. Ultimately, this framework contributes to the discourse on digital language policy by repositioning the teacher not as a bystander to automated automation, but as an active, agentic mediator of educational technology.