humansHumanshumansancient ancestors

Decoding Ancient Languages – Modern Methods For Ancient Problems

Cracking ancient languages is a long and tricky process, but will AI technologies help researchers in the future?

Russell is a Science Writer with IFLScience and has a PhD in the History of Science, Medicine and Technology

Dr. Russell Moul

Russell is a Science Writer with IFLScience and has a PhD in the History of Science, Medicine and Technology.

Science Writer

Ancient Egyptian hieroglyphs carved into a wall background.

There are still many mysteries in ancient languages that we have not yet crack, but can AI offer help? Image Credit: Triff/Shutterstock. 

In 1822, French philologist Jean-François Champollion made an announcement that would change our understanding of the ancient world. Champollion had successfully deciphered the enigmatic Rosetta Stone and provided the key to understanding ancient Egyptian text. Now 200 years have passed, and our knowledge of ancient Egypt has grown significantly, but there are still many intriguing mysteries to puzzle out. So how do we translate long-dead languages and can technology help?

Deciphering or Translation?

As with so many things, to understand where we are today, we need to know what came before. More precisely, we need to understand why the Rosetta Stone was so significant and how scholars like Champollion originally cracked its code. 


The first thing to grasp here is a subtle difference between translation and decipherment. On the surface, you may think this is a distinction without difference, as these two terms are often used synonymously in everyday language. However, in the context of ancient languages at least, there is a distinction. 

Dr Camilla Di Biase-Dyson, Senior Lecturer in the Department of History and Archaeology at Macquarie University, explained to IFLScience that to translate something, “implies that there is a working lexicon of words with the translation. And the translator is involved with employing those word lists and any viable grammars of the language in order to devise a translation.” So if there is no lexicon, dictionary, or known grammar, then the act of translation is basically impossible. The task then is to “decipher” the unknown language through other means – and although there are various ways this can be achieved, depending on the nature of the text or code in question, it is nevertheless a challenging process. 

This was the situation that European scholars found themselves in prior to the discovery of the Rosetta Stone.  

The key to knowledge

At the start of the 19th century, no one knew how hieroglyphs worked; researchers were unsure whether they functioned like pictograms or phonetic symbols. There was no known key to this puzzle, so the language was a complete mystery. 


However, in around July 15, 1799, French soldiers with Napoleon Bonaparte’s army were excavating a fort near the port city of Rosetta (Rashid today), which is on the Nile delta, and they found a chipped slab with text on it. As with so many ancient artifacts, the slab was quickly taken out of the country and soon ended up in England as part of the Treaty of Alexandria in 1801, after Napoleon was defeated (the stone has been in the British Museum ever since).

The slab was originally inscribed with text praising the pharaoh Ptolemy V at a time when the Ptolemaic Kingdom was experiencing open rebellion. The stone’s message was basically propaganda assuring the denizens of ancient Egypt that the Ptolemaic regime was legitimate and benevolent. It was written in three languages – formal hieroglyphs, demotic (the standard everyday Egyptian script), and ancient Greek (Ptolemy V was himself Greek and Hellenic culture exerted considerable influence in Egypt at the time). French scholars, and soon after English scholars, were quick to realize what they were dealing with. The tripartite message could help unravel the secrets of the ancient Egyptian language.  

“Of course, everyone knew Greek”, Di Biase-Dyson explained. “So within 30 seconds, the whole world knew what was actually written on the stone because they had the cleverness to assume it's probably the same inscription three times.”

“They were right, there are variations, but they were right. And so this is why the Rosetta Stone is considered such a corner piece. Right? This is why it's so important because it already gives you a key to decipherment, because it gives you the translation already. And then what you have to do is work out how it relates to the demotic directly above it and the hieroglyphs at the top.”


Some scholars noticed that the pharaoh’s name kept appearing in these little lozenge spaces, called cartouches, in each version of the text. This gave them a place to start comparing and mapping out the glyphs by comparing the different words in the three texts. This was a significant step, but it was not enough. Champollion went a step further by analyzing the text’s relationship to Egyptian Coptic, a liturgical language that was (and still is) used by Coptic Christians. Through this, Champonllion gained a far broader view of the language and was able to create an alphabet that allowed for the language to be spoken again. 

By the time he died, Champonllion had essentially deciphered the mysteries of the Ancient Egyptian language into an understandable language, albeit an incredibly complex one. 

Can machines help?

Although Champonllion’s work had huge impacts on our understanding of this ancient language, there are still mysteries that we may never solve. This is because the materials we have are incomplete. This fragmentation can make it difficult to understand some words, but the issue is worse when there are words that only ever occur once. In these instances, there is nothing for us to base our translations on and so the word’s meaning remains unknown. But overall, our understanding of the language we do have is pretty certain and the ease at which we can translate ancient Egyptian is becoming more advanced with developments in AI and deep learning technologies. 

Although tools like Google Translate are specifically optimized for contemporary languages, their increasingly sophisticated AI-powered features will likely help scholars with historical language reconstruction in the future. Google Translate can already translate Latin and it is likely that Ancient Greek may eventually be added to the mix as well. But it is currently unclear if and when ancient Egyptian will be included as the Egyptian script and grammar is a lot harder for the software to understand. This is because hieroglyphs can have many meanings, their usage changes over time, and their appearance can vary due to their author’s stylistic approaches. 


“That’s not to say that it won’t be cracked in a couple of years”, Di Biase-Dyson added. “There are certain parts of Egyptian text which are very formulaic, and it is likely that a computer will be able to at least crack those parts. But the thing is that most Egyptologists learn that stuff first so I’m not sure it is going to help all that much personally.” 

However, this translating ability may have significant value for museums and cultural heritage sites where plaques of information or text on objects can be examined and translated by visitors’ phones. According to Di Biase-Dyson, there are museums that are already trialing this type of technology, though in many instances the translations being offered are pre-learned and not necessarily translated in real-time. “We’re not there yet”, Di Biase-Dyson explained, “though there are a lot of people working on it.”

For a long time, progress in AI’s ability to approach languages like ancient Egyptian were held back by a lack of engagement with trained Egyptologists and experts in ancient languages. This meant the software had insufficient data to be trained on. Essentially, to be able to translate something in the field – for instance, the text written on an Egyptian tomb – a machine needs a lot of data to draw on to make sense of it, but few IT developers worked with scholars to provide detailed and comprehensive corpora. But increasingly this is changing

So at the moment, the extent to which AI will be able to help crack the unsolved mysteries in ancient languages is still unclear, but it is likely that even modest developments will be of value to Egyptologists to help them digitize texts more quickly. But Di Biase-Dyson wants to stress that having tools to help translate languages should not replace the act of learning ancient languages ourselves.


“One thing to remember is that we still get a lot of students coming to learn hieroglyphs because of the joy of discovery and the excitement of decipherment will be lost if we crack the code completely. If we crack it completely, we will never need to use our heads for it again. The joy will be lost.”


humansHumanshumansancient ancestors
  • tag
  • language,

  • ancient egypt,

  • AI,

  • machine learning,

  • ancient ancestors,

  • ancient languages