Structured English Learning Program Using Gemini: A Complete Guide

The byoungd/English-level-up-tips repository provides a Markdown-based static site that integrates Google Gemini into a five-stage learning loop (Live → Guided Learning → Canvas → Quiz → Flashcards) for systematic English acquisition.

This open-source knowledge base demonstrates how to combine static documentation with AI tooling to create a self-sustaining English learning environment. The project is hosted on GitHub Pages and organizes content into logical chapters covering understanding, vocabulary, listening, reading, speaking, writing, and AI-assisted learning.

Repository Architecture and Content Structure

The project follows a documentation-centric architecture where all learning materials reside in version-controlled Markdown files.

Markdown-Based Content Layer

All educational content is stored as Markdown files segmented by skill area. The root README.md provides the high-level overview and licensing information, while the docs/ directory contains the browsable site content. Chapter 7 (docs/threads/part-1/7-ai.md) specifically details the Gemini integration and modern AI workflow implementation.

Localization and Static Site Generation

The repository supports bilingual content through the docs/en/ and docs/ directories for English and Chinese respectively. Navigation hierarchies are defined in docs/SUMMARY.md and docs/en/SUMMARY.md, which GitBook and GitHub Pages use to render the static site. Visual assets including CEFR diagrams and AI illustrations are stored in docs/assets/ and referenced throughout the lessons.

The Gemini-Powered Learning Workflow

The AI chapter (7-ai.md) establishes Gemini as the primary engine due to its multilingual capabilities and low latency, structuring the learning process into a repeatable pipeline.

Core Engine and Capabilities

Gemini serves as the central processing unit for generating personalized learning materials. Unlike static textbooks, the system leverages Gemini to produce real-time listening scripts, comprehension questions, and vocabulary explanations adapted to the learner's current level.

The Five-Stage Training Pipeline

The repository defines a concrete loop for active learning:

1. Live – Real-time conversation practice with Gemini generating contextual responses
2. Guided Learning – Structured lessons with specific objectives and explanations
3. Canvas – Visual note-taking where Gemini generates mind-maps of vocabulary clusters
4. Quiz – Spaced-repetition testing through automatically generated multiple-choice questions
5. Flashcards – Micro-review sessions using distilled key concepts from previous stages

Fallback Strategy with Complementary AI Models

When Gemini encounters niche queries or specific refinement needs, the guide recommends ChatGPT for creative writing alternatives, Claude for nuanced grammatical explanations, Perplexity for research-backed answers, and DeepL Write for final essay polishing and academic tone adjustment.

Practical Implementation and Code Examples

The following snippets from the repository demonstrate how to implement the Gemini-driven workflow in practice.

Gemini Prompt for Listening Exercises

Use this prompt structure in the Live stage to generate on-the-fly listening material:

**Prompt**  
You are an English listening tutor. Give me a short (≈ 30‑second) audio script about daily life, then ask three comprehension questions.  

**Expected output**  
- Audio script (text)  
- Question 1  
- Question 2  
- Question 3

Guided Learning Recipe with Canvas and Quiz

This workflow combines visual learning with immediate assessment:

**Step 1 – Canvas**  
Create a mind‑map of today’s vocabulary (e.g., “commute, schedule, deadline”).  

**Step 2 – Quiz**  
Generate a 5‑question multiple‑choice quiz that tests the new words.  

**Gemini call**  

{ "model": "gemini-1.5-pro", "messages": [ {"role":"system","content":"You are an English tutor"}, {"role":"user","content":"Create a mind‑map for the words commute, schedule, deadline and then a 5‑question MCQ quiz."} ] }

*The response can be copied into a note‑taking app or a flashcard system.*

DeepL Write Integration for Essay Polishing

For final output refinement, the repository recommends this DeepL Write workflow:

**Prompt**  
Rewrite the following paragraph in academic style and correct any grammar errors:  

> I think learning English is fun because I can watch movies without subtitles.  

**Result (example)**  
Learning English is enjoyable because it enables me to watch movies without the need for subtitles, thereby enhancing my cultural immersion.

Key Resources and File Reference

The following files comprise the core documentation engine:

• docs/threads/part-1/7-ai.md – Detailed AI integration chapter describing the Gemini workflow and alternative model usage
• docs/threads/word-list/Common.md – Central vocabulary repository used across all learning modules
• docs/SUMMARY.md – Navigation hierarchy defining the GitBook/GitHub Pages structure
• docs/en/README.md – English-language entry point with resource recommendations
• docs/assets/ – Directory containing CEFR diagrams, icons, and AI illustrations referenced in lessons
• README.md – Root-level project overview, licensing (MIT), and external links

Summary

• The byoungd/English-level-up-tips repository provides a static-site-based English learning framework powered by Gemini
• The five-stage pipeline (Live → Guided Learning → Canvas → Quiz → Flashcards) creates a self-sustaining AI-augmented study loop
• docs/threads/part-1/7-ai.md contains the specific implementation details for integrating Gemini with fallback options for ChatGPT, Claude, and DeepL Write
• Content is organized as Markdown files in docs/ with localization support and automatic GitHub Pages deployment
• Word lists in docs/threads/word-list/ provide domain-specific vocabulary for technical English learning

Frequently Asked Questions

How do I start using the Gemini learning workflow from this repository?

Fork the repository and navigate to docs/threads/part-1/7-ai.md to read the AI chapter. Copy the prompt templates provided in that file into your Gemini interface, beginning with the Live conversation stage for real-time speaking practice, then progress through Canvas for vocabulary mapping and Quiz for retention testing.

Can I use this system without knowing Chinese?

Yes. The repository maintains separate English-language content in the docs/en/ directory. The docs/en/README.md serves as the English entry point, and docs/en/SUMMARY.md provides the complete navigation hierarchy for English-only learners.

What is the difference between the Canvas and Quiz stages?

The Canvas stage focuses on visual organization where Gemini generates mind-maps and conceptual connections for new vocabulary. The Quiz stage implements active recall through automatically generated multiple-choice questions based on the Canvas material, creating a spaced-repetition mechanism for long-term retention.

Are there specific vocabulary lists for technical fields?

Yes. The docs/threads/word-list/ directory contains specialized vocabulary collections including Python.md, Go.md, and Rust.md. These files provide domain-specific terminology that can be imported into the Canvas stage for technical English acquisition alongside general language learning.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}