Undeciphered writing systems
« previous post | next post »
5 Mysterious Writing Systems That No One Has Deciphered: These ancient scripts offer tantalizing clues about civilizations we still don’t fully understand.
Crystal Ponti, History (5/15/26)
The five undeciphered writing systems covered in this post are: Linear A, Indus Script, Rongorongo, Proto-Elamite, and The Phaistos Disk. We have discussed each of these scripts on Language Log, some of them at consierable length and on multiple occasions.
What conditions unite them?
Early civilizations recorded trade, ritual and governance in various written forms, leaving behind systems of communication that shaped human history. Egyptian hieroglyphs and Maya glyphs have been deciphered, but others remain stubbornly encoded, even after centuries of study.
As Marc Zender, associate professor of anthropology and director of the linguistics program at Tulane University, explains, decipherment depends on a specific set of conditions:
-
Script typology: Determining what kind of writing system it is and whether symbols represent sounds, syllables, whole words or a combination of these.
-
Sufficient corpus: Having enough examples of the script for researchers to study and compare.
-
Known or reconstructible language: Identifying the underlying language or being able to reasonably reconstruct it. Without this, decipherment is nearly impossible.
-
Cultural context: Understanding the civilization that produced the script, including known names, places and historical references.
-
Constraint: Having a crucial clue, such as a bilingual inscription, that allows researchers to match meanings across languages.
“Those scripts without any of these pillars will remain undeciphered,” he notes, while even partially supported systems “will never be as well-understood as scholars would like.”
Increasingly sophisticated AI systems may make it possible to compensate for the lack of one or more of these 5 pillars.
Selected readings
- "Are all writing systems equally easy / hard?" (4/17/25)
- "Language trees and script trees" (12/27/21) — guest post by Jim Unger (J. Marshall Unger), who wrote it in response to my invitation to him to draw up a Stammbaum to show the relationships of the world's scripts.
Peter Cyrus said,
May 24, 2026 @ 5:39 am
I'd like more care taken to distinguish between "writing" of any kind, and "written language". Not all writing is written language – think of our arithmetic or musical notations – and I suspect that the earliest writing was definitely NOT written language.
wgj said,
May 24, 2026 @ 9:50 am
Surprised that the Voynich manuscript didn't make the cut.
CuConnacht said,
May 24, 2026 @ 12:59 pm
There seem to be quite a few people who believe that if we ever identify a radio signal as likely originating from an extra-terrestrial intelligence, we will be able to establish communication with the source. Given the conditions listed by Marc Zender, it has always seemed unlikely to me.
Sniffnoy said,
May 24, 2026 @ 1:12 pm
I was going to say "Wasn't there a post here recently about a claimed decipherment of proto-Elamite?", except, oops, no, that was Linear Elamite. Seems proto-Elamite is a different matter…
Scott P. said,
May 24, 2026 @ 2:02 pm
CuConnacht
It depends what you mean by 'communication'. It seems possible we can send and receive mathematical statements, which would be a good start.
Rodger C said,
May 25, 2026 @ 9:29 am
wgj, the difference with the Voynich MS is that it's a cryptogram, not a lost script. It's apparently not written in a natural language, and probably isn't in any language at all.
David Marjanović said,
May 25, 2026 @ 7:38 pm
I wouldn't call Linear A undeciphered, really. Most signs are similar enough to those of Linear B that they can be assigned sound values or meanings (there are logograms) with fairly high probabilities; moreover, most of the documents are spreadsheets, so it's easy to understand what they're about. It's even been possible to decipher the logograms for fractions.
What's not understood is the language.
The script on the Phaistos Disk (and a very small number of other artefacts) seems to be the non-linear version of Linear A, i.e. drawings as opposed to simplified glyphs of just a few strokes each. But what's on the Disk is a text, not a spreadsheet, so we're not going to get far without knowing the language.
I can't see how.
That's a wider definition of "writing" than seems usual in that field. However, whether the Indus Script is writing in this sense, i.e. whether it represents language, is not actually clear.
That said, language creeps in very easily: as soon as there's any rebus, or as soon as you represent homonyms with the same character, you're writing a specific language. That began quite early in the development of Proto-Cuneiform.
I take its absence as an implication that the manuscript is a fake.
Yes, though the paper on the decipherment of Linear Elamite says it seems to be descended from Proto-Elamite, so there's hope.
David Marjanović said,
May 25, 2026 @ 7:41 pm
Indeed the number of languages for which Linear A was used could be anywhere between 1 and 10 or so. Crete was linguistically pretty diverse in the following 1000–1500 years. However, what very little grammar and vocabulary can be inferred from the spreadsheets does seem rather uniform.
Andreas Johansson said,
May 26, 2026 @ 6:54 am
The fact that someone thought it necessary to invent a new script for Mycenaean Greek rather suggests to me that Linear A was used for just one. Or at least was used for only one when Linear B was invented.
Cretan hieroglyphics is another candidate for the non-linearized version of Linear A.
Stephen Goranson said,
May 28, 2026 @ 4:57 am
For another set of undeciphered writing:
"[Christopher] Rollston on the Father Nazario Inscribed Stones, Puerto Rico"
http://www.rollstonepigraphy.com/?p=1015
David Marjanović said,
May 31, 2026 @ 4:09 am
But that's the thing – Linear B is almost the same as Linear A. It's like modern Greek and modern Cyrillic. This has been consensus since about the 1970s.
Cretan hieroglyphics are the hand-drawn version; the Phaistos Disk is stamped. Like elaborate handwriting vs. printing.