"Selfishness lasts a day; Civilization endures forever"

January 17, 2019 1:01 PM

Tablets from some of the world’s oldest civilisations hold rich details about life thousands of years ago, but few people today can read them. New technology is helping to unlock them. How AI could help us with ancient languages like Sumerian (BBC Future), focused on Machine Translation and Automated Analysis of Cuneiform Languages (Github). There are tens of thousands of Mesopotamian administrative records from the 21st Century BC, and more than 50,000 Mesopotamian engraved seals, yet some 90% of cuneiform texts remain untranslated. Additionally, computers can detect marks too faint for human eyes, and compare stamp marks to identify related works in different collections.

When done by people, the process can be slow (Brown University Library article with translations), so automation could open up a vast amount of information for broader study.

There is already an effort to digitize thousands of cuneform tablets, via UCLA's Cuneiform Digital Library Initiative, which also has its own Wiki to provide additional information and links to resources. If you want to try your hand at translating tablets yourself, you can crack open this Sumerian Cuneiform English Dictionary (PDF, via Archive.org), or visit the Sumerian Cuneiform Dictionary Mugsar Online.

If you just want to write your name or a message in Sumerian Cuneiform, Paleo Aliens has an online English to Cuneiform translater (give it a moment to generate the image), and the University of Pennsylvania Museum of Archaeology and Anthropology has a web tool to write your monogram like a Babylonian.

posted by filthy light thief (10 comments total)

... records from the 21st Century BC ...

To put that in perspective, in the middle of the timeline between then and now ... is Jesus.

Wow.
posted by ZenMasterThis at 1:15 PM on January 17, 2019 [6 favorites]

Okay Google, read the Nam Shub of Enki.
posted by suetanvil at 1:33 PM on January 17, 2019 [3 favorites]

Write like a Babylonian
posted by infini at 1:34 PM on January 17, 2019 [1 favorite]

This is awesome.
posted by xarnop at 2:05 PM on January 17, 2019

It's an interesting idea. My understanding is that the vast majority of untranslated cuneiform is stuff like receipts and other non-narrative records, which probably makes them ideal for machine translation. They're probably not using more than a handful of words, they're not conveying complicated ideas, and if something unexpected does show up on one, it's probably a good sign that it's something a human would want to take a look at. The main issue I can see based on my limited knowledge is that several different unrelated languages used cuneiform, so if you get something that's Old Persian or Akkadian mixed up with the Sumerian, it'll be total gibberish, but I'm guessing that the people working on this are already aware of that fact and aren't just going to shove whatever they happen to have lying around into the machine translation.
posted by Copronymus at 2:25 PM on January 17, 2019 [2 favorites]

here's a fun fact: cuneiform is part of the Unicode standard, can be treated like any other digital text (copy-and-pasted etc.) and on many modern devices you already have a built in cuneiform font.

𒁣𒌨𒌢𒌑

(that's just nonsense because i don't speak babylonian)
posted by vogon_poet at 2:32 PM on January 17, 2019 [4 favorites]

You can have your tweets immortalized to cuneiform clay tablets here
posted by monotreme at 3:24 PM on January 17, 2019 [3 favorites]

This is probably a good place to note that The Electronic Text Corpus of Sumerian Literature is the source of my email sig file of the last 20 years or so. It translates as "half a shekel is half a shekel wherever you go; discarded, it is a shekel belonging to the place of wild cattle and serpents."

As true now as it was millennia ago. Thanks for the post.
posted by talking leaf at 6:33 AM on January 18, 2019 [2 favorites]

I had the good fortune to hold one of these clay tablets in my hand 2 days ago, in the British Museum. It was one of several copies of a text that survived for over 1000 years, written in Sumerian, but reproduced in that form as the ambient language switched to Akkadian. It was the Kesh Temple Hymn (translation), a liturgical ritual text that I think is the earliest evidence for chanting we have: the oldest fragment goes back to 2,600 BCE. The translation makes it clear that the text has a verse-chorus structure, and the assumption is that the chorus was the bit many people joined in. The curator who sat with me was just as wonderful as you might hope for, and we had great fun scanning the wedge marks to try to find the chorus.
posted by stonepharisee at 7:17 AM on January 18, 2019 [3 favorites]

As humanity lurches full-tilt into Hot Venus territory, my mind sometimes wanders towards the challenges archaeologists will have deciphering our modern-day culture. Not much that we create today is durable into the deep future the same way as these clay tablets.

It's hard to think of anything we create now that will last thousands of years, let alone hundreds - groundwater contamination and atmospheric Carbon-14 pollution only communicate a single bit of information about our morals and society.

Historians gasp with relief when there remains a written record of a civilization - even when undeciphered. On the flipside, the unknowable past of humanity must be a gargantuan percentage of all history. The stories untold must be a special kind of infinity.

Cuneiform tablets that are mundane (“11 nanny goats for the kitchen on the 15th day”) are the most interesting and humanizing: Assyrians loved eating roast goat meat - just like me!
posted by Enkidude at 6:55 PM on January 21, 2019 [1 favorite]