Festus disc in the Archaeological Museum of Heraklion, Greece.
The Indus Valley Script, Epi-Olmec and Linear A are some of the writing systems of ancient civilizations that were never understood. Can artificial intelligence help decipher the codes of the past?
Imagine receiving an unknown code, with no dictionary, grammar or translation. This is precisely the challenge that archeology and linguistics face in the face of several ancient writing systems that remain a mystery. They reveal advanced civilizations whose writing we can see, but not understand.
A linguist Svenja Bonmannfrom the University of Cologne, Germany, is a specialist in historical-comparative linguistics. Try to decipher historical languages and reconstruct their structures.
“It is incredibly fascinating to have before me such a demanding intellectual enigma that not even the most brilliant minds have been able to solve,” he says. “These written records give us access to a culture that disappeared a long time ago.” He also says that, like a time machine, they allow us to interact, albeit passively, with a foreign culture.
The obstacles to deciphering
Bonmann is currently investigating the writing epiolmecaused on the southern coast of the Gulf of Mexico in Antiquity. Individual inscriptions and symbols in Olmec writing point to an ancient system. However, the evidence is so scarce and the context so uncertain that deciphering it becomes very difficult.
Also enigmatic is the Indus Valley scriptof the Harappan civilization, in present-day Pakistan and northwestern India and of which the ZAP coincidentally this month. It appears in hundreds of seals and pottery fragments, but almost always only in extremely short sequences. Whether this writing represents a fully developed language or a symbolic system is still a matter of debate.
The data is also highly abstract. It resembles pictographic writing composed of birds, human figures and ornamental shapesand survives only on a few wooden boards, often damaged.
The Minoan culture of Crete is more familiar to us. Of its three writing systems, only Linear B has been deciphered, as it is a primitive form of the Greek language. You Cretan hieroglyphics and the Linear Aon the other hand, remain enigmatic to this day.
The famous Festus Disc, dating from the second millennium BC, also originates from Crete. It is a unique clay object with stamped symbols arranged in a spiral which, as it is an isolated artefact, is practically impossible to decipher systematically. However, the “first CD” on the planet.
O Etruscanspoken in central Italy in Antiquity, also remains enigmatic. Although the alphabet is readable because it derives from Greek, the language itself has almost no recognizable relatives. This makes it difficult to understand what is written in the inscriptions.
A Proto-Elamite was the oldest known written and administrative tradition in ancient Elam, a region in the west and southwest of present-day Iran. The signs are well catalogued, but the tablets are often fragmentary. The content appears to consist of administrative records, and the underlying language does not fit into any known language family.
When writing becomes insoluble enigmas
All these writings share a fundamental problem: the lack of calls Rosetta Stones, bilingual inscriptions that contain the same text in a known language and in the riddle script. Without these keys, associating symbols with sounds, syllables or words remains difficult.
But it is not impossible, argues Bonmann, citing the decipherment of Linear B.
“It is not necessary to have bilingual texts, but rather some kind of continuity with historical times. For example, names of places, rulers or gods. So it is certainly possible.”
The problem arises, however, when there are few and very short texts: in this case, patterns are difficult to recognize and hypotheses to test. The same happens when archaeological sites are destroyed or poorly documented.
“We are always working with fragments or pieces of the past,” says Bonmann. Fortunately, Europe has a relatively large number of preserved texts, while in regions like Central America it is necessary to work with what little “the conquerors left behind”, explains the Cologne-based linguist.
For decipherment, it is also crucial that the language can be assigned to a known language family. Without this context, we lack the sound systems, word structures, and typical grammatical patterns that can serve as a basis for testing hypotheses.
Artificial intelligence helps, but…
Artificial intelligence is often touted as a potential “code breaker”. These technologies can identify patterns in character sequences, distinguish variants, fill in gaps in damaged sections and count frequencies.
However, according to Bonmann, AI quickly reaches its limits when there is very small amounts of text. You need a lot of data to do robust analysis. In the case of undeciphered writing systems, there are, in general, very few inscriptions.
“In my opinion, it is relatively unlikely that programs capable of working with so little data will be developed in the near future”, he points out.
Furthermore, AI mainly recombines already known information, rather than “thinking” something truly new, argues Bonmann: “The AI simply varies certain phrases and words, suggesting intelligence. But in reality, it is just a simulation of intelligence. The program is not really thinking.”
This can generate interpretations that seem elegant but are hardly scientifically sound. There is also a risk that the systems reflect the unconscious expectations of researchers, for example, when they “discover” relationships with language families that were particularly frequent in the training material, says Bonmann.
Mysteries will be mysteries
Perhaps this is precisely where the particular fascination of these writings lies: they show that, even in the age of apparently omniscient machines, some voices from the past remain silent — at least, for now.
“We humans are, as far as we know, the only historically conscious species. We think about where we came from and where we are going,” says Bonmann.
For the Cologne-based linguist, reflecting on past societies, how they functioned and the reasons for their disappearance, is essential to understanding the human condition. Therefore, deciphering these languages is an extremely relevant and current issue.
