Was Sumerian the Language of Babel?

What Genesis means by "the language of the whole world"

This page, like the others on the site, is an exercise in plausibility. The aim is not to prove the Bible, but to ask whether Genesis 1 to 11, read as history in the way this site proposes, can be integrated with what we know of the ancient world.

The Great Ziggurat of Ur, built by Ur-Nammu c. 2112–2094 BC: the kind of tower Genesis 11 describes.

Here the question is language. Genesis 11:1 says that "the whole world had one language and a common speech." Most readers take this to mean that all humanity still spoke a single language at the time of Babel, and that the world's many languages began there, when God confused it. But on the mainstream picture, many peoples and many languages were already in the world long before Babel, which on this site's timeline falls late in the third millennium BC.

We propose a different reading, and ask whether it is plausible. When Genesis 11:1 says the whole world had one language, we take it to be harking back to the ancient world, the world of Adam through Noah, and telling us how things stood there: that whole world really did have one language. The verses that follow then move forward, into our world, to a people who still spoke that old language, and to the story of how it ended. And we propose that the language can be named in our own history. It was Sumerian.

1. One language, carried from the world before the flood

This reading assumes what the earlier guided journeys have already proposed, and does not re-argue it. The Cain and the Cainites journey proposed that Cain was sent out, not into an empty earth, but into a world already full of people: our world, running in parallel to the ancient world he came from, the lost world of the garden. We take that up here only where it bears on language.

In the ancient world, before the flood, there was one language. On this reading, when Cain crossed into our world, near the Zagros foothills east of where the Tigris and Euphrates meet, around 5000 BC, he carried that language with him, and over the following centuries his line plausibly moved west, into the plain of Sumer. When the flood entered the ancient world, on this chronology around 2900 BC, Noah's family carried the same language through it, a second crossing of the ancient world's language into ours, and came to rest at Mount Ararat. Genesis 11 then describes a journey south, until some of them settled on "a plain in Shinar" (Genesis 11:2), the land of Sumer, roughly a thousand kilometres away.

If that is so, then "the whole world" in Genesis 11:1 is the ancient world, the world in which one language really was spoken by everyone. Genesis is not claiming that our world had only one language at the time of Babel. In our world the old language was always one among many. And that is how we read Genesis 11:9 as well: at Babel, the Lord confused what had been the language of the whole world.

The wording of Genesis itself leans this way. The chapter opens by telling us that the whole world had one language and a common speech. Only in the next verse do people set out, moving east, and settle on a plain in Shinar. Read closely, the two verses sit at different moments. The first is a backdrop, the settled state of an earlier time: one language, and no need of a translator, the world in which Methuselah and the others lived out their long lives. The second verse takes up the story later, in our world, on the plain where the tower will rise. In the Hebrew, both verses open with the same word, wayhi, "and it came to pass", a double opening distinctive enough that the commentator Gordon Wenham draws attention to it. The word can introduce the setting of a story, and it can also move a story forward to a new point in time. Read that way here, verse 1 gives the backdrop and verse 2 makes the jump. So the one language of the whole world can be read as the language of the world before the flood, while the scene at Babel belongs to ours.

There is a further hint in the builders' own aim. They set out to make a name for themselves (Genesis 11:4). To make a name is to make it among others. A people entirely alone in the world would have no one before whom a name could be made, and no reason to seek one. The wish implies an out-group, others among whom the builders sought standing. So the text itself need not be read as claiming that those at Babel were the only people on the earth. On this reading they were one people, with one language, among the many peoples and tongues of our world.

The narrator may hint at who these builders were. Genesis 11:5 calls them "the sons of the Adam" (bene ha-adam), an expression found only here in the books of Moses. Elsewhere the phrase would read simply as "mankind"; but in a chapter this careful, following the genealogy of Adam's line through Noah, it can be read as marking the builders out as that line in particular: Adam's descendants, keepers of the old world's one language, now settled among the many peoples of the new.

2. Why we propose that language was Sumerian

If Cain and Noah spoke the one language of the ancient world, and if this reading is also historical, we would expect to find a single major ancient language in Mesopotamia: appearing at about the right time, with no known ancestors, and, because Genesis says it was confused at Babel and its speakers scattered, with no surviving relatives or descendants. We propose that Sumerian fits that description, and several further observations point the same way.

Ziggurats at the right time and place. Genesis 10:25 says that Peleg was so named because "in his time the earth was divided", the spreading of the nations, in the same era as Babel. On this timeline Peleg's lifetime spans roughly 2372–2033 BC. The Sumerian king Ur-Nammu (reigned 2112–2094 BC) is "for ever associated with the ziggurats . . . which he erected in Ur, Uruk, Eridu, Nippur and various other cities" (the historian Georges Roux). Sumerian speakers were building ziggurats in southern Mesopotamia at about the time and place the account describes.
The rapid end of Sumerian. Roux described the fall of Ur at the close of the third millennium BC as marking "the end of the Sumerian nation and type of society," with Akkadian displacing Sumerian as a spoken language in what he called a "linguistic revolution," though Sumerian continued as a written and liturgical language for centuries. In other words, a spoken language was confused and its speakers were scattered, shortly after they built towers that fit the description, at about the date this site's timeline gives.
The one ancient parallel is Sumerian. The only Mesopotamian text with a possible parallel to Babel, the epic "Enmerkar and the Lord of Aratta," is itself Sumerian. The Sumerologist Samuel Kramer read it as showing the Sumerians "believed there was a time when all mankind spoke one and the same language," though others read it more cautiously. Whichever is right, the one ancient parallel to the Babel story is in Sumerian and about Sumerian.
The oldest flood accounts are Sumerian. The earliest accounts of a great flood resembling the biblical one, including the Eridu Genesis and the Sumerian King List, are in Sumerian. The King List also records pre-flood reigns far longer than those after, echoing the biblical pattern. If Noah's family were Sumerian speakers, this is what we would expect.
A people defined by their language. Genesis 11:6 describes the builders as "one people speaking the same language." Roux noted that "the appellation 'Sumerians' should be taken as meaning 'Sumerian-speaking people' and nothing else". The Sumerians were thus a people defined by their language to an unusual degree, much as Genesis describes the builders.

The tower is also placed in "a plain in Shinar" (Genesis 11:2), a name that may derive from a Sumerian term for their own territory; and the first known writing system was Sumerian. Taken together, we think this makes a reasonable cumulative case that the Babel account can be read as the decline of Sumerian and the dispersal of its speakers: not a proof, but a plausible fit.

An objection: the Mesopotamian flood accounts are older

The point above invites a fair question, and it is one of the most common objections to reading Genesis as history at all. Mesopotamian accounts of a great flood were being written well before Genesis, on any dating of Genesis. The Atrahasis Epic, whose storyline of creation, growing population and flood runs parallel to Genesis 2 to 8, survives in a copy of about 1640 BC and is thought to have been composed near the start of the second millennium; the tradition behind these accounts is generally traced to around 2000 BC or earlier. The Eridu Genesis and the Sumerian King List are in Sumerian. The best known of them is the Babylonian flood story that George Smith read from a broken tablet in 1872, preserved in the eleventh tablet of the Gilgamesh epic, which took its flood account from Atrahasis. Many conclude from this that Genesis is a late retelling of an older story, and that it can therefore contribute nothing to questions about human origins.

Smith himself did not draw that conclusion; he presented the tablet as independent support for events Genesis also records. The argument was first put on a scholarly footing by the German scholar Hermann Gunkel in Schöpfung und Chaos (1895), who proposed not that Genesis copied a Babylonian text but that both drew on a common stock of tradition carried down over centuries. It became a public controversy when the Assyriologist Friedrich Delitzsch turned it against the Bible's claim to be revelation, in his Babel und Bibel lectures of 1902 to 1904. The biblical scholar Peter Enns has argued a recent version of it.

We do not dispute the dating. On the reading set out here, the parallels are what we would expect. If the flood was a real event, and if those closest to it spoke Sumerian, then the earliest written memories of it would be in Sumerian, and later accounts of the same events would resemble them. Shared material would point to a shared history rather than to a borrowed story.

Parallel accounts of the same events can also drift apart, and where they differ, they cannot all be right. On the reading proposed here, some of what looks strangest in the Mesopotamian accounts may even be memory rather than myth. Their dome holding back waters overhead, for instance, may preserve a real feature of the ancient world, the firmament whose floodgates opened at the flood (Genesis 7:11); the drift would lie in assuming that the sky of our world must be the same. The Sumerian King List may well remember that lives before the flood were very long, while overstating how long they were. Shared history, remembered with differing degrees of fidelity, would explain both the parallels and the differences. And those who hold a high view of Scripture can maintain that God preserved the true history of these events through his Spirit, whatever was happening to the memory of them elsewhere.

None of this is a demonstration that Genesis is free of error, and it is not offered as one. It is the narrower claim that the antiquity of the Mesopotamian accounts, and their parallels with Genesis, do not by themselves settle the question. This is argued at length in chapter 5 of A Timeline of Origins, in response to Peter Enns.

3. The Sumerian Paradox: a language isolate with no isolation

There is a further observation that this reading helps explain, and it stands on its own whatever one makes of the argument so far. Many well-known language isolates, tongues with no known relatives, make sense once you put them on a map. Basque survived in the Pyrenees, Burushaski in the remote Karakoram valleys, Korean, an isolate or the whole of a very small family, on its bounded peninsula. Physical isolation often makes linguistic isolation plausible. Not always, though: of the 184 isolates in Glottolog's catalogue, about a fifth stand in open country with no refuge, most of them in Amazonia and on the Gulf coast of North America.

Sumerian is the most extreme case of them all. When terrain is measured around each of the 184, Sumerian's homeland ranks as the most exposed: land varying by less than seven metres across fifty kilometres, nothing above 200 metres within a hundred. It sat in the flat, open plains of Mesopotamia, on major trade routes, in constant contact with its neighbours. And among the refuge-less isolates, only two were languages of literate ancient states: Sumerian, and Elamite next door, which had a highland refuge to fall back on and used it. Sumerian alone had no such refuge. And yet it has no known relatives. Akkadian is Semitic, Hittite Indo-European, Egyptian Afro-Asiatic; the other great river-valley civilisations produced languages that fit known families. Sumerian stands alone.

Of Sumerian's neighbours, Elamite comes closest to being another exception: usually classed as an isolate, with its capital Susa on the same open lowland, second most exposed of the 184 and effectively tied with Sumer at the floor. But Elam was a double kingdom, and its highland half at Anshan in the Zagros, nearly 2,000 metres up with no lowland around it, sheltered the language each time the plain was overrun; Elamite outlived spoken Sumerian by more than a thousand years. The near-exception shows what mattered: a refuge. Sumerian had none.

Compare the map: Take the Isolate Tour

A second anomaly stands beside the first. Dominant languages normally leave descendants: Latin became French, Spanish and Italian; Akkadian's Semitic family lives on in Aramaic, Arabic and Hebrew. Sumerian left nothing. So the record holds two oddities at once: an isolate with no geographic explanation, and a once-dominant language that vanished without descendants.

The reading proposed here addresses both from one starting point. On this reading Sumerian had no relatives because it came from outside our world, and left no descendants because its speakers were scattered into a world already full of other languages, absorbed into families it had never belonged to. We do not offer this as proof. But when a single account, taken as history, resolves two independent anomalies at once, that is the kind of convergence that makes a reading worth taking seriously.

Watch the evidence unfold

The home page opens a guided journey version of this argument on the map, tracing the same sequence: the one language entering our world with Cain, carried again through the flood, and finally confused at Babel, its speakers scattered, a language that leaves no relatives and no descendants.

None of this proves the biblical text, and it is not meant to. It offers a reading of Genesis 11, resting on what the earlier guided journeys proposed about Genesis 4 and 5 and the flood, that fits independently attested features of the linguistic record and coheres with what is known of the ancient Near East. The reading is a proposal. We think the convergence it produces, across the identification of Sumerian and the two standing anomalies of its isolation and its extinction, is plausible enough to be worth serious consideration.

Develops chapter 5 of A Timeline of Origins (Michael D. Russell, 2024). Georges Roux, Ancient Iraq, 3rd ed. (London: Penguin, 1992); S. N. Kramer, "The 'Babel of Tongues': A Sumerian Version," Journal of the American Oriental Society 88 (1968): 108–11; Gordon Wenham, Genesis 1–15 (Word, 1987). The language-isolate argument is developed here beyond the book. The count of 184 isolates, the refuge-less fifth, and the rankings come from a terrain comparison across every isolate in Glottolog 5.0, using the ETOPO 2022 60-arc-second relief grid measured within 50 km and 100 km of each language's Glottolog point; the method is described in the isolate notes. It is one way of measuring refuge, not the only one.

▶ Watch this journey on the map Next argument: Nations after Noah →