Cuneiform 101 · Chapter 11

Decipherment

Everything in this course — every sign, every reading, every claim about how the system works — passes through you because a small number of people, working mostly without knowing whether they would succeed, worked out how to read it again. Cuneiform was not a script handed down through unbroken tradition, the way the Latin alphabet was. It went out of use, entirely, and for well over a thousand years no living person could read a line of it. This chapter tells that recovery in outline, the way the standard histories of decipherment tell it — coarse on dates, careful on the logic that actually cracked the problem.

But first, two signs. They are not decoration: they are the kind of sign that made decipherment possible.

Why open a history chapter with two syllable signs? Because signs like these — signs that carry pure sound — are what let cuneiform spell names. You met that in chapter 07 (your own name can only be spelled, not word-signed), and names, as you are about to see, were the decipherers’ crowbar: the one thing a lost script and a known history could hold in common. A sign like 𒈾NA · [na] · [na]; stone in some words keeps a spelled name honest, syllable by syllable.

A script nobody could read

By the early centuries of the Common Era, cuneiform writing had stopped. The languages it had served — Sumerian long dead as a spoken tongue, Akkadian fading after it — were gone from use, and the script went with them. The stone-carved and clay-pressed wedges that remained, at ruined mounds across Iraq and Iran, were visible to anyone who traveled there. Nobody could read them. They were not even reliably recognized as writing; travelers described the marks on the ruins of Persepolis as ornamental, or as some kind of patterning, without a clear sense that they encoded language at all.

Recovering the script therefore had to start from zero: not translating a known language written in an unfamiliar alphabet, but working out that these wedge-shaped marks were writing in the first place, and only then asking what they said.

Careful copies before careful reading

Progress began with better copies. Earlier European travelers had brought back sketches of the Persepolis inscriptions, but the signs were too imprecisely rendered to analyze. The turning point, as the standard histories tell it, was the German traveler and surveyor Carsten Niebuhr, whose late-eighteenth-century copies of the Persepolis monuments were accurate enough to work from. Niebuhr made an observation that later proved essential: the same monuments carried not one cuneiform script but three distinct ones, carved side by side. Whatever these inscriptions said, they said it in three different systems at once — a strong hint, in hindsight, that the same content had been rendered for three different audiences or languages.

That observation split an impossible problem into three smaller ones. One of the three scripts had, comparatively, very few distinct sign-shapes, and appeared to mark word boundaries — features that made it look far more approachable than the other two. It became the door decipherers went through first.

The first wedge: Old Persian

That simplest of the three scripts recorded Old Persian, and its decipherment is credited, in the standard telling, largely to Georg Friedrich Grotefend, working around the turn of the nineteenth century. Grotefend reasoned from context rather than from any prior key: royal inscriptions of this general kind, in languages already known, tended to follow a set formula naming a king, his royal title (a phrase along the lines of king of kings), and his father’s name. By matching repeated groups of signs in the Persepolis texts against that expected pattern, and against known Persian royal names from Greek and other historical sources, he proposed readings for a handful of signs and for royal names such as that of Darius. It was a narrow beachhead — a proposal about a few signs and a few words, not a working knowledge of the language — but it held up, and later scholars, above all Henry Rawlinson, extended it into a far fuller reading of Old Persian cuneiform over the following decades.

Behistun

The single find the standard histories treat as decisive is the Behistun inscription: a monumental relief and text carved by order of the Persian king Darius I high on a cliff face in western Iran, far above the plain and difficult to reach. Behistun mattered because it repeated one long inscription in three languages — Old Persian, Elamite, and Babylonian — carved together, exactly the multilingual pattern Niebuhr had already noticed at Persepolis, but longer and richer than anything found there.

A trilingual text of real length, in a script already partly broken for one of its three languages, was the closest thing cuneiform studies had to a Rosetta Stone. Rawlinson copied it — by the standard accounts, at real physical risk, working from a narrow ledge on the cliff face over more than one expedition in the 1830s and 1840s — and that copy, combined with the Old Persian reading already achieved, gave scholars a long, reliable bridge into the other two languages of the inscription, Babylonian above all.

Babylonian gives way

Old Persian used comparatively few signs and worked in an alphabet-like way. Babylonian cuneiform was a different order of difficulty entirely: hundreds of signs, most of them syllabic rather than alphabetic, and many signs readable with more than one sound value or more than one meaning depending on context — polyvalence, in the field’s term. Cracking it took the combined work of several scholars across the 1840s and 1850s, working partly in competition and partly building on one another’s results: Rawlinson, extending the Behistun bridge from Persian into Babylonian; Edward Hincks, an Irish clergyman whose close analysis of sign patterns established, against real resistance, that Babylonian signs were syllabic and routinely polyvalent — a structural insight that unlocked far more of the script than any single word-reading could; and Jules Oppert, who brought comparative rigor to the language once behind the signs. Between them, the case that cuneiform could genuinely be read, and not merely guessed at, became strong enough to state as a claim worth testing.

The test of 1857

That test came in 1857, and the standard histories present it as decipherment’s decisive public moment. The Royal Asiatic Society in London had a newly discovered inscription — a prism text of the Assyrian king Tiglath-pileser I — sent independently to four scholars working on Babylonian: Rawlinson, Hincks, Oppert, and William Fox Talbot. Each translated it separately, under sealed conditions, without comparing notes. When the four translations were opened and compared, they agreed with one another closely enough, in substance, to settle the question: this was not four people independently inventing compatible fictions. Cuneiform, and the Akkadian language written in it, could be read.

A second language nobody expected

One more discovery followed, and it surprised the field more than the others. Beneath the Babylonian material, texts kept surfacing that were written in the same cuneiform signs but did not read as Akkadian at all — a different language entirely, older, with its own grammar. Some scholars doubted at first that this second language was real, suspecting it might be some scribal code or cipher layered onto Babylonian rather than an independent tongue. The doubt did not hold: ancient bilingual word-lists and literary texts, pairing this language line by line with Akkadian translations, let scholars reconstruct it as what it was — Sumerian, the language the script had originally been built to write, surviving Babylonian and Assyrian texts. Cuneiform, it turned out, had not served one ancient language behind its wedges. It had served at least two, layered across three thousand years, and Sumerian had been sitting underneath Babylonian the entire time, waiting to be told apart from it.

What decipherment did not finish

It is worth closing honestly, because that is the standard the field holds itself to and the standard this course tries to meet. The work begun by Niebuhr, Grotefend, Rawlinson, Hincks, Oppert, and their contemporaries did not finish cuneiform. Three things remain open today.

The earliest phase of the script, the proto-cuneiform of the fourth millennium BCE touched on in earlier chapters, still resists reading as language in any full sense — its earliest tablets can often be counted and categorized, but not read the way a sentence is read. Sumerian grammar and phonology, despite long study built on those bilingual lists, still carry real points of scholarly disagreement — how certain forms worked, how the language actually sounded, are not closed questions. And the practical arithmetic is stark: hundreds of thousands of cuneiform tablets survive, museum collections still hold large numbers never transliterated, and the world’s working cuneiformists number only in the hundreds. New texts surface faster than they can be read.

That last gap is not a footnote — it is a large part of why a school like this one exists. Reading ability is the scarce resource in this field, more scarce than the tablets themselves. Every reader this course trains, however modestly, is a small answer to a real and current shortage, not just an exercise in history.

Next: the reference — every sign taught, every convention used, gathered in one place.