Comment by cletus | Hacker News Reader

Comment by cletus | Hacker News Reader

Parent Full thread

Cuneiform tables [1] are one of those historical artifacts that I find fascinating. Some fun facts:

- Between 500,000 and 2,000,000 tablets have been found

- Less than 100,000 have been translated

- Cuneiform tablets date back to ~3500 BC

- Roughly 30,000 tablets were recovered from the royal archives at Hattusa (Hittite capital) [2] in central Turkey. This archive and the site probably survived because this ancient city was largely forgotten about until archaeologists discovered it in the 19th century.

- This archive produced the earliest known peace treaty between the Hittites and Egyptians

- These archives contain essentially the personal correspondence between Ramses II and the Hittite king [3]

- Cuneiform tablets have been one of the most important sources of written records from the ancient world [4]

[1]: https://en.wikipedia.org/wiki/Cuneiform

[2]: https://en.wikipedia.org/wiki/Hattusa#Cuneiform_royal_archiv...

[3]: https://www.nationalgeographic.com/history/magazine/2016/07-...

[4]: http://cdli.ox.ac.uk/wiki/doku.php?id=objects1to10

What's the status of ancient Egyptian works? Lots of untranslated maybe-gems floating around? I'm still holding out hope a couple of accidentally-cliffhanger'd stories from Lichtheim's books will have endings discovered for me to read!

“Less than 100,000 have been translated”

Can’t that be addressed with deep learning and crowdsourcing?

Perhaps, but I'm skeptical.

Transcribing the cuneiform script is difficult. There are hundreds of signs. Transcribing a tablet requires some understanding of the language because a sign can take on many different values (polyphony), some signs are so similar that they can only be differentiated by context, and if you're dealing with a language like Hittite, you need to recognize when a sign should be understood as representing an Akkadian or Sumerian word or grammatical construct as a scribal abbreviation. All this changes from language to language and era to era. Also, how signs are written changes over time. And then, getting into the physical aspects of working with tablets, pretty much all of them have damage. Cuneiform is a 3D script and tablets requires varying the light, which won't be captured from a single photograph. Indeed, line drawings are the standard way of publishing tablets. When photographs are used, many are taken with varying light. Some computational techniques are beginning to be used, but very few tablets are recorded in this way. Many tablets are in fragments spread around the museums of the world—figuring out which fragments go together or which tablets are duplicates of each other (to fill in damage) is the largest part of publishing cuneiform texts.

> Many tablets are in fragments spread around the museums of the world—figuring out which fragments go together or which tablets are duplicates of each other (to fill in damage) is the largest part of publishing cuneiform texts.

Although archeologists (an other scholars) may lack access to a horde of fired-up Iranian revolutionary students[0], this one problem has actually been solved. Of course doing it for cuneiform may have to wait until some of the other issues you mentioned may need to be solved first.

[0] https://slate.com/news-and-politics/2009/07/is-it-possible-t...

3D scanning ought to help, then. Even X-ray CT might make sense, if the resolution can be pushed high enough to fully capture the carvings that encode cuneiform script.

The British Museum are working on digitising their collection with 3D scans and high resolution imaging. The catalogue is slowly coming online, an example: https://research.britishmuseum.org/research/collection_onlin...

There are efforts to do exactly that. The Cuneiform Digital Library Initiative [0] has some interesting projects you may wish to browse [1] as an example.

[0]: https://cdli.ucla.edu/

[1]: https://github.com/cdli-gh

Reply on news.ycombinator.com