In the three millennia before the common era, a highly developed civilization flourished in the Near East, leaving behind a wealth of information: on clay tablets inscribed in cuneiform. In this writing system, wedge-shaped characters were pressed with a stylus into moist clay tablets, which were then dried—resulting in a durable written document.

Most clay tablets have broken over time, and their fragments are scattered across museums around the world. Ancient Near Eastern Studies therefore faces a complex task: it must correctly piece the fragments back together. Only then is it possible to read the complete documents and gain insights into life in the Ancient Near East.

Researchers at the University of Würzburg and the Academy of Sciences and Literature in Mainz have been working on this for many years. Their focus is on the culture of the Hittites, who lived in what is now Anatolia 3,500 years ago. This people’s “alphabet” is extensive: it consists of no fewer than 375 cuneiform signs, which represent syllables as well as entire words.

Setting multiple milestones in Ancient Near Eastern Studies

In its work, the Würzburg-Mainz team has repeatedly set milestones in Ancient Near Eastern Studies that have garnered worldwide attention.

Twenty-five years ago, Gernot Wilhelm and Gerfrid Müller established the Hethitologie-Portal Mainz—featuring a digital catalog of all 30,000 known Hittite clay tablet fragments, along with numerous research materials and texts. The inventory of clay tablets was originally compiled by Silvin Košak. Researchers worldwide contribute to this portal, and it is freely accessible via the Internet.

Ten years ago, the team introduced a digital tool capable of capturing the unique characteristics of individual cuneiform characters in 3D. This significantly facilitated the computer-aided reconstruction of ancient clay tablets. In 2023, the TLH^dig tool followed: it allows users to search texts in cuneiform or in transliteration.

The individual shape of cuneiform signs is important

Now the portal is presenting its latest tool, the “Palaeographicum.” It recognizes the individual shape of cuneiform characters in the digitized photos on the Hethitologie-Portal and searches the entire collection of clay tablets for identically or similarly written characters. It then cuts the characters out of the photos and compiles them clearly in image tables.

The current version of the Palaeographicum provides access to 70,000 photos documenting more than five million cuneiform characters. It was developed in collaboration with the Technical University of Dortmund.

The individual form of the characters is very important for research, as it allows individual scribes to be distinguished. Although cuneiform was pressed into the clay with styluses, individual “handwriting styles” are still recognizable in the texts: each scribe had their own style. Some pulled the stylus out of the clay so vigorously that flourishes formed, while others placed the characters with characteristic spacing.

Palaeographicum facilitates the comparison of handwriting styles

Recognizing a scribe’s distinctive traits makes the work of piecing together clay tablet fragments easier. “With the naked eye, we can usually only do this slowly and with difficulty,” says Professor of Ancient Near Eastern Studies Gerfrid Müller. This is due to the three-dimensional nature of cuneiform—depending on the angle of the light, the characters are more or less legible in the photos.

“The Palaeographicum is radically changing our work; it allows us to save thousands of hours,” says Professor Daniel Schwemer, head of the Department of Ancient Near Eastern Studies at the University of Würzburg. For example, comparing the handwriting of individual characters across five clay tablet fragments used to take three days. Now it can be done in five minutes.

Just as in Europe, handwriting in the Ancient Near East has changed significantly over the centuries. Because Hittite clay tablets never bear a date, the new tool also helps the research community date individual fragments. That is why it was named “Palaeographicum”: It refers to palaeography, the study of the historical development of ancient manuscripts.

AI is continuously being retrained

Work on the tool is not finished: “We are continuously retraining the AI,” says Gerfrid Müller. The team also takes user requests into account when further developing the Palaeographicum—“if they are technically feasible and make the work easier for everyone.”

The global Hittitology community has already provided plenty of positive feedback on the new tool. The development team didn’t have to go to great lengths to publicize their work: the announcement in the news section of the Hethitologie-Portal was more than enough. “All researchers in Hittitology open the portal first thing in the morning; they simply can’t do without it,” says Daniel Schwemer.

Outlook: Social history of writing culture

The researchers have an ambitious goal for the future: they want to train the AI so well that it can automatically recognize the handwriting of individual scribes.

But that is a complex task. One reason is that the scribes produced different handwriting styles—depending on whether they were working in calm conditions at home or hastily drafting a report during an on-site visit, such as an inspection of the sanctuaries of the Hittite Empire.

“If we achieve this goal, we could gain a better picture of what individual scribes produced over the course of their professional careers,” says Daniel Schwemer, “and we could compile a social history of Hittite writing culture.” That would likely be the next milestone the Würzburg-Mainz team would set.

History of the tool’s development

The foundation for the new tool was laid by the DFG-funded CuKa project (Computer-assisted cuneiform analysis): From 2018 to 2023, Gerfrid Müller, in collaboration with TU Dortmund University, developed the AI model that now forms the basis of Palaeographicum.

A web-based demonstrator was also created during the course of the project. Turna Somel from the Mainz Academy handled the philological work for this, annotating the training material and evaluating the training results. The development of the AI model was led by Christopher Rest and Eugen Rusakov from TU Dortmund University.

After the project was completed, Gerfrid Müller, Christopher Rest, and Herbert Baier Saip from the Center for Philology and Digitality at the University of Würzburg further developed the demonstrator’s visual online tool—it had previously been too cumbersome to process the vast amount of available photos. This led to the creation of the first version of the Palaeographicum.

The Festival Rituals of the Hittites

The project “The Corpus of Hittite Festival Rituals” is funded by the Academy of Sciences and Literature | Mainz and forms part of the joint research programme of the academies of sciences. The Academies’ Programme is dedicated to the documentation, preservation and study of cultural traditions worldwide and is currently the largest long-term research programme in the Federal Republic of Germany for basic research in the humanities and social sciences.