AI finds no relatives for Sumerian

AI scans Sumerian against every language in the world and finds no family

There is a language we have been able to read with ease for more than a century and yet we have no idea where it came from. Sumerian was deciphered in the 19th century from thousands of clay tablets, but the problem was not the alphabet: it was that much of what was being read was in Akkadian, a Semitic language, written with characters invented by an earlier people. When linguists isolated the oldest tablets, the ones that did not match Akkadian, they ran into a language with no known relatives. Not one. Decades of manual comparisons failed to link it to anything. Now, a project backed by artificial intelligence has repeated the exercise on a global scale, and the result, according to the materials circulating, is still the same: Sumerian has no identified linguistic family.

What exactly the Sumerian problem is

The key lies in a detail that is often overlooked. The name "Sumerian" was not coined by the Sumerians themselves: it is what the Akkadians called an earlier people who already lived in southern Mesopotamia when they arrived. Sumerian ruins were located precisely because archaeologists were searching for that earlier people. And when their tablets appeared, the language turned out to be an oddity: no relation to Akkadian, no relation to the neighboring Semitic languages, no relation to anything.

Translation was made possible thanks to a tablet written by an Akkadian schoolmaster who, to teach his students, listed Sumerian words with their Akkadian equivalents. A domestic Rosetta Stone. Sumerian, moreover, continued to be used centuries later as a scholarly language, much like Latin in ecclesiastical documents.

Why it fits no language family

As the thread explains, all the languages of the planet belong to families related to neighbors or to peoples who passed through. Spanish shares a root with Portuguese, French or Italian, and has absorbed Arabic loanwords like almohada (pillow) or alfajor. Akkadian is Semitic and has clear relatives. Sumerian does not. It is an isolate, and that has fueled decades of unconfirmed hypotheses.

According to the thread's account, computing power has made it possible to compare the Sumerian corpus word by word, with meanings and variants, against living and dead languages. The result has not solved the enigma.

Where the Sumerians came from: genetics versus legend

The thread mentions a hypothesis that placed their origin on the northwest coast of India, near what would later be the Indus Valley civilization. As the thread argues, genetic data weaken it: the Neolithic genomes of Mesopotamia, northern Iraq and the Zagros show a West Asian continuum, with components from Zagros, Anatolia, the Levant and the Caucasus, not a population replacement coming from the Indus. Added to this, according to the same source, is a serious chronological problem: the urbanization of Uruk and proto-cuneiform writing predate any plausible connection with the Indus.

Another line mentioned points to the Obeid culture and Eridu as the starting point. And some argue that the Sumerian counting system, base 60 and the origin of the 360 degrees, is actually the sum of two different systems, which would suggest the merger of two peoples. The thread maintains that the hypothesis of an emerged Persian Gulf and of Dilmun as the original paradise belongs to the realm of myth, not archaeology.

The puzzling data point

According to the thread, beer was already being consumed at the Neolithic sites of Tas Tepeler, at least 8,000 years before the Sumerians appeared. The summary of the linked video indicates that Uruk came to house some 40,000 inhabitants around 3000 BC. And Sumerian, meanwhile, still has no relative.