From left, characters in the Ethiopic (portions only), Armenian, Georgian and Caucasian Albanian alphabets (Photo: phys.org)

The research, conducted by scientists at San Diego State University and reported by Phys.org, used artificial intelligence to examine similarities among ancient writing systems. The findings were published in the journal Digital Scholarship in the Humanities.

Using a dataset of more than 28,000 Ethiopic characters, researchers trained a computer model to recognize structural features such as curves, straight lines, and angles. The system, which had no access to historical or cultural context, then compared these patterns with letters from Armenian, Georgian, and Caucasian Albanian alphabets.

The analysis showed that Armenian letters exhibited the strongest structural similarity to Ge’ez, followed by Caucasian Albanian with moderate resemblance, while Georgian showed weaker and less consistent similarities. By contrast, the Latin alphabet demonstrated significantly lower similarity, reinforcing the distinctiveness of the observed patterns.

“Our aim was to move beyond visual impressions that are difficult to test or replicate,” said Sam Kassegne, the study’s lead investigator. “By making our criteria explicit and mathematical, we introduced an objective computational approach that is easily reproducible.”

One notable finding is that the Armenian script appears nearly as similar to Ge’ez as Ge’ez is to its own earlier forms, suggesting the resemblance may not be coincidental. Both writing systems developed around the 4th to 5th centuries AD —a period marked by documented travel between Ethiopia and parts of the Middle East, including Jerusalem, Egypt, and Syria. Historical accounts also indicate that Mesrop Mashtots, credited with creating the Armenian alphabet, traveled within the region.

“What makes the research significant is that computational geometry and historical scholarship converge on the same scripts and time period,” said Daniel Zemene, the study’s first author.

While the researchers caution that structural similarity does not prove direct borrowing, the findings strengthen arguments that cultural contact and exchange may have influenced the development of writing systems across regions. The study also highlights the growing role of artificial intelligence in uncovering patterns in historical and linguistic research.