Are there any techniques available today that would make it considerably easier to anyone attempting to translate it today (as if they were trying todo it for the first time)? Otherwise would the process be the same as it was for the first person who managed to translate it? Would computers be of use and if yes how?

  • Hello Arjang and welcome to Linguistics! That's actually an interesting question. What was the original method again by the way?
    – Alenanno
    Commented Sep 1, 2013 at 16:58
  • 1
    Sample of script in three orthographies -- hieroglyphic Egyptian, Demotic Egyptian, and Greek. Greek was known. Dunno how well-known Egyptian Demotic was at that time; I doubt Champollion knew it already.
    – jlawler
    Commented Sep 1, 2013 at 20:35
  • @Alenanno : pretty much what jlwler said, there was also Thomas Young that tried a Logical approach. But instead of viewing as a Champollion vs Young, today one should be able to harness the power of computers and advancements in linguistics to do a better job one hopes. There was also a video on youtube. My main interest is to see if there is anything that could turn Champollion-Young works into a body of knowledge with wider applications than just that one of work they did on Rosetta Stone.
    – jimjim
    Commented Sep 1, 2013 at 23:48
  • The character/word frequency stuff is the same today but faster via computer. There are still undeciphered languages today (and probably hoaxes like Voynich) even with computers, but those don't come on handy Rosetta Stones. Commented Sep 2, 2013 at 3:30
  • 2
    @jlawler According to wikipedia, "Demotic is a development of Late Egyptian and shares much with the later Coptic phase of the Egyptian language." I know that Champollion (who spoke a dozen languages) was a specialist of Coptic, and that it was essential in his work on the Rosetta stone. Irrelevant note: copies of the Rosetta stone can be bought (I was offered one as a gift).
    – babou
    Commented Dec 7, 2013 at 21:18

2 Answers 2


I think there are two big innovations that we would look to today.

  1. Computers able to analyze millions of patterns in minutes

  2. Cryptographic techniques that use what is known about patterns of a language to discover regularities in the seemingly random noise of a cipher

However, the fundamentals would be the same. You always have to start from a fundamental understanding of language. You always need a starting point. And you also need a decent corpus to use whatever quantitative techniques you have on. That's why so many scripts with only fragments and no language available remain a mystery.

But even parallel texts may not be enough. The difference between cryptography and Rosetta stone is that in the first case you're dealing with at least one known language. (That's why Germans couldn't crack Navajo communications.) And translations are never exact parallels. So one fact Champollion relied on was that 'cartouche' - oval around some text - was used to indicate royal names. That gave him an anchor from which to work backwards. But he still had to come to an understanding that the hieroglyphs are not strict ideographs and then rely on his in depth knowledge of Coptic (he kept a diary in it) to make something of it.

None of this could be achieved by pure pattern analysis but it could certainly speed up some aspects.

Similar thing happened in the decipherment of Hittite which relied on sound philological work once a large corpus of the language was found in a known script.

However, no computer method can help with the decipherment of scripts like Linear A because of the tiny corpus from which to work.

Even where a much larger corpus is available as with the Indus valley Harappan script, it's hard to know what to apply modern techniques on since it's not even clear the script represents a language in the conventional way.

  • With the deciphering of the Rosetta Stone being 200 years old yesterday, I was about to ask a very similar question on Cryptography SE. This excellent answer (+1) and th eones from linguistics.stackexchange.com/questions/34083/… cover most of the points I think.
    – WoJ
    Commented Sep 29, 2022 at 10:58

I think there are so many avenues to explore in this area that its hard to answer this question. I believe there are many ways computers could be used in the decipherment.

Kevin Knight's research is particularly relevant in my opinion.

You may be interested in these papers, that are in that direction.

  • Deciphering Foreign Language by Combining Language Models and Context Vectors - Nuhn, Mauser & Ney, here
  • A Statistical Model for Lost Language Decipherment - Synder, Barzilay, & Knight - here
  • Unsupervised Analysis for Decipherment Problems - here - Knight, Nair, Rathod & Yamada

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.