Chapter 5 SCRIPTS WITHOUT VOICES

Some writing systems remain undeciphered—windows we cannot open.

5.1 Proto-Elamite

In the museums of Paris, Berlin, and Tehran, clay tablets sit in climate-controlled cases. They’re covered with marks—neat rows of signs, clearly organized, obviously meaningful. They date to around 3000 BCE, making them contemporary with the earliest Sumerian writing. They come from southwestern Iran, from the region the Greeks would later call Elam.

No one can read them.

Proto-Elamite is one of the world’s oldest writing systems and one of its most frustrating mysteries. We have about 1,600 tablets. We can identify numerical notations—the accounting format is recognizable even when the language isn’t. We can see that scribes were recording transactions, keeping inventories, managing economic activity.

But the words themselves are opaque.

The script uses about 1,000 distinct signs—far too many for an alphabet, suggesting a logographic or logosyllabic system like Sumerian cuneiform or Egyptian hieroglyphics. Some signs appear frequently; others are rare. The patterns suggest grammar, syntax, meaning. But without a bilingual text—a Rosetta Stone linking Proto-Elamite to a known language—the meaning stays locked.

What language does Proto-Elamite represent? The obvious guess is Elamite, the language spoken in the same region a few centuries later. We can read later Elamite texts, written in borrowed cuneiform. If Proto-Elamite is an earlier script for the same language, we might eventually crack it.

But we’re not certain Proto-Elamite represents Elamite at all. The later Elamite language is itself poorly understood—an isolate, like Sumerian, with no clear relatives. And the gap between Proto-Elamite (c. 3000 BCE) and the earliest readable Elamite texts (c. 2300 BCE) is seven centuries. Languages change. Scripts change. The connection might be there; the connection might not.

Some scholars have proposed that Proto-Elamite isn’t writing at all—not in the full sense. Perhaps it’s a sophisticated accounting system, capable of recording quantities and transactions but not continuous language. Perhaps there are no words to read, only numbers and categories.

This is possible but seems unlikely. The complexity of the sign system, the consistency of usage across sites, the presence of what look like proper names—all suggest something more than bookkeeping. Something linguistic is encoded. We just can’t extract it.

The tablets wait. They’ve waited five thousand years already. They may wait forever.

5.2 Linear A

The palace of Knossos, on the island of Crete, was one of the wonders of the Bronze Age Mediterranean. Sprawling across multiple levels, decorated with vivid frescoes, equipped with plumbing and light wells, it was the center of Minoan civilization—a culture that flourished from roughly 2700 to 1450 BCE, then vanished.

The Minoans left two scripts. One, called Linear B, was deciphered in 1952 by Michael Ventris, an architect and amateur linguist. He proved that Linear B recorded an early form of Greek. The Mycenaeans, Greek-speakers from the mainland, had adopted Minoan writing for their own language after conquering Crete.

But Linear A, the earlier script, remains unread.

Linear A was the Minoans’ own writing system, used before the Mycenaean conquest. About 1,400 inscriptions survive—far fewer than Proto-Elamite, and many are very short. They appear on clay tablets, pottery, stone vessels, and metal objects. The contexts suggest religious dedications, administrative records, and possibly trade.

We can read the sounds of Linear A, at least approximately. Many signs are similar to Linear B, which we can read. When we apply Linear B sound values to Linear A, we get pronounceable sequences: ku-ro, a-sa-sa-ra-me, ja-sa-sa-ra-mu. These are words in some language.

But what language?

Not Greek. The words don’t match Greek vocabulary or grammar. Not any known language, in fact. The Minoan language, like Sumerian and Elamite, appears to be an isolate—unrelated to anything we can identify.

This is the special cruelty of Linear A. We can sound out the signs. We can hear the echoes of Minoan voices. But we don’t know what the voices are saying.

Some words recur in what seem to be religious contexts: a-sa-sa-ra, possibly a goddess’s name. Some sequences look like place names, personal names, commodity terms. Scholars have proposed connections to Anatolian languages, Semitic languages, even Indo-European languages. None have convinced the field.

Without more texts, without a bilingual inscription, without some breakthrough in method, Linear A may remain a script we can pronounce but not understand—the most tantalizing kind of silence.

5.3 Why Some Fail

What does successful decipherment require?

The history of decipherment offers lessons. Egyptian hieroglyphics yielded to Champollion in 1822 because of the Rosetta Stone—a decree inscribed in three scripts, including Greek. Linear B yielded to Ventris in 1952 because he guessed, correctly, that the underlying language was Greek, and because enough texts existed to test patterns systematically.

In each case, certain conditions were met:

Sufficient text. You need enough inscriptions to identify patterns—which signs are common, which are rare, which cluster together. A single short inscription is almost impossible to crack. Thousands of texts offer statistical leverage.

External clues. A bilingual text is ideal. Failing that, proper names can help—if you can identify a sign sequence as the name of a known king or city, you have a foothold. Contextual clues matter too: if a text appears on a wine jar, it probably discusses wine.

A known language—or a related one. Ventris succeeded with Linear B because he tried Greek. If the Minoans had spoken Greek, Linear A would be read by now. The problem is that we don’t know any language related to Minoan.

Luck. Champollion was brilliant, but he was also lucky that the Rosetta Stone existed, that it survived, that it was found. Ventris was lucky that earlier scholars had done groundwork he could build on. Decipherment often depends on fortunate accidents.

Proto-Elamite and Linear A lack these conditions—or lack enough of them.

Proto-Elamite has sufficient text (1,600 tablets) but no bilingual, no securely identified names, and no known related language. The script is complex, the underlying language unknown. Every attempt to crack it has failed.

Linear A has some external clues—the similarity to Linear B provides approximate sound values—but too little text and no known related language. We can guess at words, but we can’t confirm.

Other undeciphered scripts face similar problems:

Script	Date	Location	Problem
Proto-Elamite	c. 3000 BCE	Iran	Unknown language, no bilingual
Linear A	c. 1800–1450 BCE	Crete	Unknown language, limited text
Indus script	c. 2600–1900 BCE	Pakistan/India	Very short texts, unknown language
Rongorongo	c. 1200–1860 CE	Easter Island	Very few texts, unknown language

The Indus Valley civilization left thousands of inscriptions—but most are just a few signs long, perhaps names or labels. Without longer texts, statistical analysis fails. Rongorongo, from Easter Island, survives on only about two dozen wooden objects, and the civilization that created it was devastated by European contact before anyone recorded how to read it.

Each script is a locked door. We don’t know if the key exists.

5.4 The Silence

What does it mean that some voices are permanently lost?

Consider what we’re missing. Proto-Elamite tablets contain administrative records—but administration implies a society, a government, a system of laws and customs. Behind the accounts are people making decisions, resolving disputes, organizing labor. We can’t hear them.

Linear A tablets include what seem to be religious texts—dedications, offerings, invocations. The Minoans had gods, rituals, beliefs about death and meaning. The frescoes at Knossos show bull-leaping, processionals, figures that might be priests or deities. But without the texts, we’re guessing.

The Indus Valley civilization was one of the largest Bronze Age cultures—contemporary with Egypt and Mesopotamia, trading with both, building cities with sophisticated drainage and standardized weights. Five million people, perhaps more. And we cannot read a single complete sentence they wrote.

This is the silence of the past.

It’s easy to forget, when reading translated texts, that translation was once impossible. Every readable ancient language was once unreadable. Scholars spent lifetimes on hieroglyphics before Champollion. Cuneiform was mysterious marks until Rawlinson and Hincks. Each breakthrough felt miraculous.

But the breakthroughs depended on conditions that don’t always hold. Not every lock has a key. Some doors stay closed.

The silence is informative in its own way. It reminds us that literacy was rare in ancient societies—most people left no texts because they couldn’t write. It reminds us that preservation is accident—the vast majority of ancient texts are lost. It reminds us that decipherment is not guaranteed—intelligence and effort aren’t always enough.

Most of all, the silence reminds us that human diversity was once far greater than it is today. Thousands of languages have existed; most are extinct. Hundreds of writing systems have been invented; most are forgotten. The homogenization of modernity—the spread of a few dominant languages and scripts—is recent and historically unusual.

When we look at an undeciphered tablet, we’re looking at a form of human thought that has been lost. Not just the specific content—the record of grain or the prayer to a goddess—but the entire framework of concepts, categories, and associations that the language embodied. A language is a way of being human. When a language dies completely—when no one can read it, when no descendants carry its echoes—a possibility dies with it.

We know this happened. We don’t know how much we’ve lost.

The tablets sit in museums, patient as stone, waiting for readers who may never come.