# 🗣️ English–Kumaoni Translation Dataset (Romanized, 9538 Pairs)

This dataset contains **9538 sentence pairs** of English and their corresponding translations in **Kumaoni** language written in **Roman script** (Latin characters).  
Kumaoni is a low-resource Indo-Aryan language spoken in the Kumaon region of Uttarakhand, India.

---

## 📂 Dataset Format

- **Format:** JSONL (newline-delimited JSON)
- **Encoding:** UTF-8
- **Total entries:** 9538
- **Fields:**
  - `english`: Sentence in English
  - `kumaoni`: Kumaoni translation in Roman (Latin) script

### 📄 Example:
```json
{"english": "give me water.", "kumaoni": "mank paani diyo."}
{"english": "what are you doing?", "kumaoni": "tum ki kar rahe chho?"}
```

---

## 🏔️ About Kumaoni

**Kumaoni** is a regional language mainly spoken in the **Kumaon region** of Uttarakhand, India.  
Despite having over a million native speakers, it is **digitally underrepresented** in the field of NLP.  
This dataset contributes to preserving and enabling AI systems to work with Kumaoni through:

- Romanized scripts (for ease of typing and modeling)
- Clean and natural sentence pairs
- Practical usage in translation, voice systems, and chatbots

---

## 🔍 Use Cases

- Fine-tuning translation models on low-resource languages
- Evaluating LLMs for English–Kumaoni code-switching
- Building regional chatbots or bilingual assistants
- Input for TTS (Text-to-Speech) or ASR if paired with audio
- RAG pipelines for Kumaoni-language applications

---

## 📜 License

**CC BY 4.0**  
You are free to use, modify, and distribute this dataset with proper credit to the original author.

---

## 🙏 Acknowledgements

Created and maintained by **[Sourab Singh Bora](https://www.kaggle.com/sourabsinghbora)**  
Special focus was given to:
- Romanized consistency
- Natural daily-use phrases
- Clean Unicode formatting

---

## 💡 Citation

```
@dataset{english_kumaoni_roman_2025,
  author       = {Sourab Singh Bora},
  title        = {english-kumaoni_translation_dataset},
  year         = 2025,
  url          = {https://huggingface.co/datasets/sourabsb/english-kumaoni_translation_dataset},
  note         = {Romanized parallel corpus for low-resource translation}
}
```