AthDGC: An Open Diachronic Greek Treebank with Indo-European Parallels
AthDGC is the first openly licensed, end-to-end dependency-parsed treebank of Greek spanning eight diachronic periods under a unified PROIEL schema, featuring verse-level cross-alignment with Latin, Gothic, Old Church Slavonic, and Classical Armenian through a multilingual annotation workflow.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine the Greek language as a single, incredibly long movie that has been playing for 3,000 years. It started with the ancient epics of Homer, moved through the Bible, the Byzantine emperors, and continues today with modern news and novels. For a long time, linguists could only watch this movie in short, disconnected clips. They had a great analysis of the "Koine" (Bible Greek) section, but they couldn't easily compare how a specific sentence changed from the time of Homer to the time of a modern Greek speaker.
AthDGC is like building a massive, unified cinema screen that plays all 3,000 years of this movie at once, with a special "director's commentary" track running underneath every single word.
Here is a breakdown of what this project does, using simple analogies:
1. The "Universal Translator" for Ancient Texts
Think of the PROIEL Treebank as a very strict, high-quality rulebook for analyzing how sentences are built. Previously, this rulebook was only used for a small section of the Greek movie (mostly the New Testament) and a few other ancient languages like Latin and Gothic.
AthDGC takes that same rulebook and applies it to the entire history of Greek. It creates a single, consistent map for:
- Archaic & Classical Greek (Homer, Plato)
- Koine Greek (The New Testament)
- Byzantine & Late Byzantine (Medieval monks and scholars)
- Early Modern & Modern Greek (Today's language)
It's like taking a single GPS system and using it to navigate a car from ancient Rome all the way to modern-day Athens without ever changing the map format.
2. The "Family Reunion" of Languages
Greek is part of the Indo-European language family, which is like a giant family tree. Greek has cousins like Latin, Gothic, Old Church Slavonic, and Armenian.
AthDGC doesn't just look at Greek in isolation. It lines up specific verses from the Bible (like the New Testament) side-by-side with their translations in these "cousin" languages.
- The Analogy: Imagine a family reunion where everyone is reading the same poem. AthDGC creates a chart that shows exactly which word in the Greek version corresponds to which word in the Latin, Gothic, or Armenian version. This allows researchers to see how the "family" spoke differently over time.
3. The "Time-Traveling Detective" Tool
The paper introduces a concept called a "Retelling and Retranslation Chain."
- The Scenario: Take the first line of the Iliad ("Sing, Goddess, the wrath of Achilles").
- The Chain: AthDGC tracks this exact line as it appears in:
- The original Homer (Ancient Greek).
- A summary by a Byzantine scholar named Tzetzes (Medieval Greek).
- A modern prose translation by Kazantzakis (Modern Greek).
The system acts like a detective. It can instantly tell you: "In Homer, the goddess was being shouted at directly. In the Byzantine summary, that shouting stopped. In the modern version, the shouting is back." This helps researchers see which parts of the sentence structure stayed the same for 2,800 years and which parts changed.
4. How It Was Built (The Assembly Line)
The team didn't just read these books by hand; they built a digital assembly line:
- Discovery: They scoured the internet and libraries to find open-source texts (like finding old manuscripts in a digital attic).
- Cleaning: They used filters to remove footnotes, page numbers, and foreign languages, keeping only the pure Greek text.
- Annotation: They used smart computer programs (AI) trained specifically on ancient languages to label every word with its job (e.g., "This is the subject," "This is the object," "This is a verb").
- Alignment: They used advanced math to match words across different languages (Greek to Latin, etc.).
5. What You Can Do With It (Right Now)
The project is currently in a "preview" phase (v0.4).
- For Researchers: It's a massive database where you can ask questions like, "How did the way Greeks expressed 'hearing' change from the 5th century BC to the 19th century?"
- For Students: It provides a visual way to see how language evolves, turning dry grammar rules into a visual story of change.
- For Developers: They have released the "tools" (the software code) so others can build their own language analysis projects using this same high-quality data.
The Bottom Line
AthDGC is the first time the entire history of the Greek language has been mapped out in a single, consistent, and open format, allowing us to see the "DNA" of the language as it evolved over three millennia, and how it relates to its ancient linguistic cousins. It turns a library of thousands of books into a single, searchable, interactive timeline.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.