Polyglot is an adaptive language learning application powered by large language models that provides personalized tutoring based on the user's target language and proficiency level (A1-C1). It offers translations, grammar explanations, vocabulary exercises, and supports learning from various file formats according to learner's current CEFR level.

• Dynamic Context Understanding: Uses Large Language Models to understand the nuances of user requests
• Intelligent Topic Detection: Automatically identifies language learning topics in user queries
• Language Detection: Recognizes the language of text with high accuracy
• Content Analysis: Analyzes uploaded files to extract relevant learning materials

• Languages: Support for multiple languages including Finnish, Spanish, French, German, Russian, and more
• Easy Language Switching: Change target language with persistent progress tracking
• Native Greetings: Time-of-day appropriate greetings in the target language

• Adaptive Content: All materials dynamically adjusted to CEFR levels (A1, A2, B1, B2, C1)
• Visual Level Indicators: Clear badges showing current proficiency level
• Progress Tracking: Records level changes across different languages
• Level-Appropriate Guidelines: LLM-generated grammar and vocabulary guidelines for each level

• Translations: Between target language and English with grammar explanations
• Grammar Breakdowns: Detailed explanations with visual aids
• Vocabulary Building: Level-appropriate word lists
• Reading and Writing Exercises: Generated based on current proficiency
• Interactive Quizzes: With immediate feedback

• Text Processing: Extract learning content from uploaded documents
• Image Analysis: Learn from uploaded images containing text in the target language
• File Content Analysis: LLM-powered analysis of file contents for relevant learning material

• Topic Tracking: Identifies user interests to provide relevant examples
• Learning History: Exportable chat history with level indicators
• Exercise Parameter Detection: Understands implicit and explicit exercise requests

Polyglot uses a modern technical stack with:

• Streamlit: For the responsive web interface
• LangChain: To integrate with large language models
• OpenAI API: Powering the intelligent language tutoring
• Function Caching: For performance optimization
• Real-time Content Generation: Streaming responses for better user experience

The application is structured into three main modules:

1. app.py: Main application interface and user interaction
2. chatbot.py: LLM integration and conversation management
3. utils.py: LLM-powered utility functions for language learning

• Python 3.8 or higher
• OpenAI API key

git clone https://github.com/yourusername/language-app.git
cd language-app

curl -LsSf https://astral.sh/uv/install.sh | sh

uv sync

Create a .streamlit/secrets.toml file in your project directory:

mkdir -p .streamlit
touch .streamlit/secrets.toml

Edit the secrets.toml file and add your OpenAI API key:

OPENAI_API_KEY = "your-api-key-here"
MODEL_NAME = "gpt-4.1-mini-2025-04-14"  # Optional - change model

uv run streamlit run app.py

Choose from 15+ supported languages in the sidebar dropdown.

Choose your CEFR level (A1-C1) from the sidebar to ensure all content is appropriate for your current proficiency.

• Type T: [text] to translate text between your target language and English
• Ask grammar questions like "Explain verb conjugation" or "How do possessives work?"
• Request vocabulary on specific topics: "Vocabulary about food in Spanish"

• Ask for level-appropriate exercises: "Give me a reading exercise about hobbies"
• Try writing exercises: "Create a writing exercise about daily routines"
• Test yourself with quizzes: "Quiz me on basic verbs"

• Upload images containing text in your target language
• Upload text documents with content in your target language
• Polyglot will analyze the content, provide translations, and create exercises based on the material

• Reset the conversation at any time using the sidebar button
• Export your chat history as a Markdown file for later review
• Track your level progression in the sidebar for each language

Polyglot replaces traditional rule-based NLP with contextual LLM intelligence for:

1. Topic Extraction: Identifies learning topics in user messages
2. Exercise Parameter Detection: Understands the type of exercise requested
3. Language Detection: Identifies the language of text segments
4. Content Guidelines: Generates appropriate vocabulary and grammar for each level
5. File Content Analysis: Assesses the difficulty level and extracts learning material

The application optimizes the learning experience by:

1. Level-Specific Content: Dynamically generating level-appropriate content
2. Personalized Examples: Using detected topics to create relevant examples
3. Progress Tracking: Monitoring level changes to provide appropriate review
4. File Analysis: Determining if uploaded content matches the learner's level

To add support for new languages:

1. Add the language to the SUPPORTED_LANGUAGES dictionary in app.py
2. Add appropriate greetings to the LANGUAGE_GREETINGS dictionary
3. Add grammar features to get_language_grammar_features() in utils.py
4. Add CEFR level guidelines to get_cefr_level_guidelines() in chatbot.py
5. Add level-appropriate content to get_level_appropriate_content() in utils.py

The modular architecture makes it easy to add new features:

1. UI components can be added to app.py
2. LLM-based processing can be extended in chatbot.py
3. Utility functions can be added to utils.py

Contributions are welcome! Please feel free to submit a Pull Request.

This project is licensed under the MIT License - see the LICENSE file for details.

• Common European Framework of Reference for Languages (CEFR) for level guidelines
• Streamlit for the interactive web application framework
• OpenAI and LangChain for the underlying language model integration