The Ancient Origins of Writing: From Cave Symbols to Clay Tablets

More than 40,000 years ago, early humans carved strange symbols onto bones and figures. What could these ancient marks possibly mean?

Writing is a powerful technology for recording ideas and events. Its creation was a pivotal moment in human history, allowing knowledge to accumulate across generations. Experts widely believe that writing was invented independently in at least four separate regions of the world: Mesopotamia, Egypt, China, and Mesoamerica. The earliest known writing systems appeared in Mesopotamia and Egypt. Ancient Egyptian hieroglyphs and Mesopotamian cuneiform script are both approximately 5,200 years old.

The archaeological record reveals a clear series of innovations that eventually led to writing. In Mesopotamia, around 9,500 years ago, merchants began using small clay tokens to track the value of traded goods. Soon, they started enclosing these tokens inside hollow clay spheres, possibly to create a permanent record of a completed transaction. Because these sealed balls were opaque, officials began pressing marks onto the outer surface to indicate what was inside.

A critical third step happened roughly 5,500 years ago. Traders stopped using the tokens altogether. Instead, they flattened the clay balls into tablets. These flat surfaces still bore the impressions that once represented the tokens inside. These marked tablets are considered direct ancestors of cuneiform, the world's oldest confirmed writing system.

All these developments occurred within the last 10,000 years. However, in early 2026, a team of researchers made a startling announcement. They argued that there is evidence humans from a much older culture—the Aurignacian culture of Paleolithic Europe—developed a system of signs that shares key features with a precursor to cuneiform. The Aurignacian culture existed between approximately 43,000 and 26,000 years ago.

Some observers interpreted this claim as proof that the roots of writing are tens of thousands of years older than anyone had imagined.

Archaeologists have known for a century that Aurignacian people carved marks onto bones and small statues. These marks include sequences of dots and crosses. They were clearly made on purpose, but according to Christian Bentz, a computational linguist at Germany's Saarland University, no one had ever done a complete analysis to interpret them.

He and his colleague, archaeologist Ewa Dutkiewicz at the Berlin State Museums, decided to perform such an analysis. Their goal was to determine if the marks had the potential to encode information. This process involves measuring the variety within a sequence of symbols. A row of identical crosses carries little information. A row of varied, different symbols can be dense with information. This is precisely how an alphabet works to convey complex messages.

Bentz and Dutkiewicz compared the Aurignacian marks to 5,500-year-old Mesopotamian inscriptions in proto-cuneiform, the immediate forerunner of true cuneiform. To their astonishment, they found both sets of symbols had a similar mathematical potential to carry information.

“I couldn’t believe it,” Bentz said. “I thought I must have messed up the analysis somehow.” But after repeated checks, the finding remained.

Linguistic anthropologist Piers Kelly from the University of New England in Australia considers the comparison valid, but not because the Aurignacians had a true script. Instead, it is because proto-cuneiform inscriptions themselves can hardly be classified as full writing.

Proto-cuneiform looks like writing, especially since it was created using the same stylus and clay tablet method as later cuneiform. Yet the earliest proto-cuneiform used a very limited set of distinct signs representing a narrow range of concepts, like sheep or grain. It could encode only very basic information. By about 5,200 years ago, it had developed a richer collection of signs, gaining the potential to store more data. Bentz and Dutkiewicz’s analysis confirms this, showing later proto-cuneiform had a far greater information-carrying potential than the Aurignacian marks.

“It can look as if there is a neat continuity between the early and late proto-cuneiform inscriptions,” Kelly said. “But they are categorically different systems.” Cuneiform is writing; proto-cuneiform is not. Kelly and many other scholars avoid calling even the richer, later examples of proto-cuneiform “writing.”

Kelly explains that writing is a system for encoding linguistic information—a tool for recording speech. This requires assigning specific sound values to signs. Proto-cuneiform never did this. Its sign for “sheep”—a cross inside a circle—reveals nothing about how the word “sheep” was pronounced.

It was not until about 4,900 years ago that Mesopotamians began encoding speech by repurposing signs to represent spoken syllables. Kelly identifies this moment as the actual invention of writing in that region. “The various notational systems that preceded and influenced this innovation were simply part of the symbolic raw material that made writing possible,” he says.

Therefore, the Aurignacian artifacts do not necessarily prove that writing has roots stretching back 43,000 years. They show only that humans were slowly and iteratively experimenting with symbolic technologies. These experiments, tens of thousands of years later, would create the conditions necessary for writing’s eventual invention.

Why were humans experimenting with signs 43,000 years ago?

“It is generally accepted that the signs have a communication function,” said archaeologist April Nowell at the University of Victoria in Canada. “We have known for a long time that they are not random.”

However, not all experts are convinced the marks were meant to communicate in a structured way.

Archaeologist Silvia Ferrara at the University of Bologna in Italy and her colleagues recently analyzed even older marks. These were made by humans in southern Africa roughly 65,000 years ago. The marks, including parallel lines and crosshatched grids, were carefully scratched onto fragments of ostrich eggshell.

Ferrara said their analysis deliberately avoided assumptions about the marks’ purpose. Instead, they explored the cognitive behaviors needed to produce them. “Writing was not part of the picture,” she said. This suggests humans were making intentional marks long before they assigned them any specific communicative meaning.

These ideas complicate our understanding of the Aurignacian marks. While Bentz and Dutkiewicz showed they had the potential to store information, that does not mean they actually did. “I’m not convinced they have the same kind of information content as the proto-cuneiform signs, for which we can often assign definite meanings,” said computational linguist Richard Sproat.

Kelly highlights another important factor. From a modern viewpoint, writing often acts as a direct substitute for speech, conveying every detail. Non-writing sign systems can be far more complex. For example, Indigenous Australians have long used message sticks—pieces of wood engraved with marks. “Message stick communication is highly multimodal, involving spoken language, gesture, and body paint,” Kelly says. “Message sticks can be thought of as an almost supplementary prop within a multi-channel communicative routine.” They cannot be understood in isolation.

This does not mean the Aurignacian artifacts were part of a similarly elaborate system. But it is a crucial reminder that we may be missing the vital context needed to understand how—or even if—these objects conveyed messages.

“Anthropologists have always championed the importance of ‘being there’ as an active participant-observer in the communities they want to understand,” says Kelly. Because archaeologists cannot travel back to “be there” with the Aurignacians, we may never fully understand the true purpose of the strange symbols they left behind.