AhmedElTaher/ancient-egyptian-multilingual-premium

🏛️ Ancient Egyptian Multilingual Corpus

Dataset Description

A comprehensive multilingual corpus of Ancient Egyptian texts combining multiple authoritative sources, including dictionaries and translated texts from German to English. Available in multiple formats for maximum accessibility.

📊 Dataset Summary

This dataset combines:

1. Dictionary Sources (3 sources, ~15,000 entries):

   • Vygus Egyptian Dictionary 2015 (Mark Vygus)
   • Dickson Dictionary of Middle Egyptian
   • Ancient Egypt Dictionary (10,000+ words from Gardiner, Faulkner, etc.)

2. Translated Corpora (4 sources, ~65,000 entries):

   • BBAW Egyptian Corpus (35,503 entries)
   • TLA Demotic (13,383 entries)
   • TLA Earlier Egyptian (12,773 entries)
   • TLA Late Egyptian (3,606 entries)

Total: ~80,000+ entries covering 4,000 years of Egyptian language history.

📦 Available Formats

This dataset is available in multiple formats for maximum compatibility:

Format	Description	Best For	Size
Arrow	Native HuggingFace format	Fast loading, memory-efficient processing	Smallest
JSON	Human-readable, structured	Data exploration, custom processing	Medium
CSV	Tabular format	Spreadsheet software, basic analysis	Medium
ZIP	All formats compressed	Offline use, complete download	Compressed

Loading Examples

Arrow Format (Recommended)

from datasets import load_dataset
dataset = load_dataset("AhmedElTaher/ancient-egyptian-multilingual")

JSON Format

import json
with open('ancient-egyptian-multilingual_full.json', 'r', encoding='utf-8') as f:
    data = json.load(f)

CSV Format

import pandas as pd
df = pd.read_csv('ancient-egyptian-multilingual_full.csv')

🎯 Supported Tasks

• Ancient Egyptian to English translation
• Hieroglyphic text analysis
• Historical linguistics research
• Egyptological studies
• Language model training for ancient languages
• Comparative linguistics
• Digital humanities research

🌍 Languages

• egy: Ancient Egyptian (hieroglyphic, hieratic, demotic scripts)
• en: English (translations)
• de: German (original translations in TLA/BBAW datasets)

Dataset Structure

📝 Data Fields

Field	Type	Description
id	string	Unique identifier for each entry
transliteration_unicode	string	Unicode transliteration (ꜣ, ꜥ, ḥ, ḫ, ẖ, š, ṯ, ḏ, ỉ)
transliteration_mdc	string	MDC ASCII-compatible transliteration
translation_en	string	English translation
translation_de	string	German translation (where available)
hieroglyphs	string	Hieroglyphic representation (Unicode/Gardiner codes)
source	string	Original dataset source
pos	string	Part of speech (where available)
lemmatization	string	Lemma information (where available)
grammatical_info	string	Additional grammatical information

📈 Data Splits

Split	Entries	Percentage
Train	~72,000	90%
Validation	~8,000	10%

Dataset Creation

🔨 Processing Pipeline

1. PDF Parsing: Custom parsers for dictionary formats
2. Transliteration: Bidirectional Unicode ↔ MDC conversion
3. Translation: NLLB-200 neural translation (DE→EN)
4. Standardization: Unified column names and formats
5. Validation: Manual checking of sample entries
6. Multi-format Export: Arrow, JSON, CSV formats

📚 Source Data

Dictionary Sources

• Vygus Dictionary 2018 - R.H. Barnhart
• Dickson Dictionary - Paul Dickson (2006)
• Ancient Egypt Dictionary - ancient-egypt.co.uk

Corpus Sources

• BBAW - Berlin-Brandenburg Academy of Sciences
• TLA - Thesaurus Linguae Aegyptiae

Usage Guidelines

🚀 Quick Start

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("AhmedElTaher/ancient-egyptian-multilingual")

# Access a sample
sample = dataset['train'][0]
print(f"Egyptian (Unicode): {sample['transliteration_unicode']}")
print(f"Egyptian (MDC): {sample['transliteration_mdc']}")
print(f"English: {sample['translation_en']}")
print(f"Hieroglyphs: {sample['hieroglyphs']}")

🔍 Filtering Examples

# Find all entries with pharaoh
pharaoh_entries = dataset.filter(lambda x: 'pharaoh' in x['translation_en'].lower())

# Get only Demotic texts
demotic = dataset.filter(lambda x: 'demotic' in x['source'])

# Get entries with hieroglyphs
with_hieroglyphs = dataset.filter(lambda x: x['hieroglyphs'] != '')

License & Attribution

📜 License

CC BY-SA 4.0 (Creative Commons Attribution-ShareAlike 4.0 International)

This requires:

• ✅ Attribution to original sources
• ✅ Indication of changes
• ✅ ShareAlike distribution

📚 Required Citations

@dataset{egyptian_multilingual_corpus_2025,
  author = {El Taher, Ahmed},
  title = {Ancient Egyptian Multilingual Corpus},
  year = {2025},
  publisher = {HuggingFace},
  url = {https://huggingface.co/datasets/AhmedElTaher/ancient-egyptian-multilingual},
  email = {ahmed.g.eltaher@gmail.com}
}

[Additional citations for BBAW, TLA, etc...]

Additional Information

👤 Dataset Curator

Ahmed El Taher
📧 ahmed.g.eltaher@gmail.com
🤗 AhmedElTaher

🤝 Contributions

Contributions welcome! Please submit issues or pull requests.

⚠️ Limitations

• Machine translations may contain inaccuracies
• Some hieroglyphic representations incomplete
• Chronological periods mixed in combined dataset
• Reflects historical biases in sources

🙏 Acknowledgments

Special thanks to all contributing institutions and researchers.

---

Dataset Version: 1.0.0 | Last Updated: 2025-08-14