Researchers from the Department of Mechanical Engineering at San Diego State University have published a study in which, using artificial intelligence as the primary analytical tool, they demonstrate significant structural similarities between ancient writing systems originating in the Horn of Africa and the Caucasus region of Eurasia.

The results, published in the journal Digital Scholarship in the Humanities, indicate that the Armenian alphabet shows a morphological relationship with the ancient Ethiopian system, known as Ge’ez, that is much closer than the linguistic and historical community had previously considered based on direct visual observations.

The research, led by Sam Kassegne, professor of mechanical engineering, began from the observation that for years various specialists had noted superficial similarities between certain characters of the Armenian, Georgian, and Caucasian Albanian alphabets and those of Ge’ez, a writing system developed in eastern Africa more than 1,600 years ago.

From left, characters in the Ethiopic (portions only), Armenian, Georgian and Caucasian Albanian alphabets. Credit: D. Zemene et al. 2026

The obstacle to validating these observations lay in their subjective nature, as they were based exclusively on the visual inspection of glyphs by scholars. To overcome this limitation, the SDSU team designed a quantitative and reproducible methodology using a computer program trained with more than 28,000 images of Ethiopian characters.

The process involved teaching the algorithm to recognize the fundamental geometric elements of Ethiopian writing, such as curves, straight lines, angles, and the overall structure of each letter. The researchers emphasize that the model lacked any contextual data related to the history, religion, geography, or culture of the analyzed systems; its analysis was strictly limited to form.

Once the learning phase was completed, the program proceeded to compare the morphological characteristics of Ethiopian characters with those of the Armenian, Georgian, and Caucasian Albanian alphabets, calculating through explicit mathematical criteria the degree of similarity among them. As a control, the researchers included the Latin alphabet, the one used in the English language, in order to establish a comparative reference.

The results obtained showed a defined hierarchy in the similarities. Among the three alphabets analyzed, the Armenian alphabet exhibited the highest level of structural correspondence with Ethiopian characters. The Caucasian Albanian alphabet showed a moderate degree of similarity, while Georgian displayed some correspondences, although less consistently.

Historical timeline for the Ethiopic, Armenian, Georgian, Caucasian Albanian, and Odessian (Edessian) scripts and their possible origins. It has to be noted that the exact origins and progressions of particularly the Ethiopic script is not unanimously agreed by scholars in the field and while the depictions are based on the best available sources now, it is not necessarily as linear as shown here. Credit: D. Zemene et al. 2026

The Latin alphabet, used as an external comparison, yielded a substantially lower level of similarity. One of the findings that researchers describe as most surprising is that the correspondence between the Armenian alphabet and Ge’ez turned out to be almost as high as that which exists between the Ethiopian system itself and its preceding versions, suggesting that the observed resemblance responds to a significant structural pattern rather than to a chance coincidence.

Our objective was to go beyond visual impressions that are difficult to prove or replicate, explained Kassegne in the university statement. By making our criteria explicit and mathematical, we introduced an objective computational approach that is easily reproducible. We believe that this reproducibility is the key contribution of our method.

Daniel Zemene, a graduate student and researcher in artificial intelligence and machine learning at the NanoFAB laboratory at SDSU, as well as the study’s first author, emphasized the convergence between the computational results and historical hypotheses. The model had no access to historical records; however, it learned solely from visual and structural data and identified Armenian as the closest structural match to Ethiopian within the same time period that historians have debated for a long time. That convergence between computation and history is powerful, Zemene noted.

The historical context framing these results refers to the origin of the Armenian alphabet, created around the year 405 of the Common Era, a period in which the Ethiopian writing system was expanding and gaining widespread use. The researchers point in their study to the existence of historical records documenting the mobility of populations originating from Ethiopia toward regions such as Jerusalem, Egypt, and Syria during that period.

Likewise, the creator of the Armenian alphabet, Mesrob Mashtots, made documented journeys through different areas of the Middle East. The published work does not establish a direct copying relationship between one system and another, but it proposes that the quantified structural correspondence, together with the historical evidence of cultural contacts in the region during that span, makes the hypothesis of reciprocal influences or exchanges between these cultures in the creation of writing systems more plausible.

The authors of the study emphasize that structural similarity does not automatically imply direct borrowing, but they stress that the use of artificial intelligence applied to the analysis of ancient writing forms makes it possible to address historical questions with an unprecedented level of precision. By stripping the analysis of the subjectivity inherent in human observation and replacing it with explicit and reproducible mathematical criteria, the methodology employed opens a pathway to reevaluate other possible relationships between graphic systems from different civilizations.

The research, beyond its specific implications regarding the links between the Armenian alphabet and Ge’ez, exemplifies how technological tools developed for fields such as autonomous driving or medical imaging can be applied to the study of literary and cultural heritage, providing numerical evidence that complements and refines traditional interpretations based on direct observation.