California State University, San Diego, USA, conducted a comparison between ancient writing systems in Africa and the mountainous Caucasus region using artificial intelligence techniques.
The study results showed that the Armenian alphabet is structurally closest to the ancient Ethiopian script (Ge’ez), which originated in the Horn of Africa about 1600 years earlier than previously thought.
At the University of California's School of Mechanical Engineering, researchers used artificial intelligence techniques instead of human evaluation. The computer program was trained on a database of over 28,000 images of Ethiopian symbols, enabling it to recognize the curves, straight lines, angles, and overall structure of each symbol simply by analyzing its shapes. The program then compared these symbols to letters from the Armenian, Georgian, and Caucasian Albanian (Agawan) alphabets.
Of the three alphabets tested, Armenian letters showed the highest degree of similarity to Ethiopian characters, followed by the Ighanian alphabet with a moderate degree of similarity, while Georgian script showed the lowest level of correspondence. The Latin alphabet showed only limited similarity.
Professor Sam Cassinia of the University of California said:
"Our goal was to go beyond visual impressions, and by adopting clear criteria based on mathematical foundations, we have provided an objective computational approach, the reproducibility of whose results is one of its most prominent advantages."
He added:We did not expect the Armenian alphabet to be so close to the Ethiopian alphabet, and to the same degree that the latter is close to its oldest version. The Armenian alphabet originated around 405 AD, when Mesrop Mashtots, its creator, traveled in the Middle East.”
Although the study does not prove the existence of direct borrowing, it suggests the possibility of cultural exchange and mutual influence between these societies.
For his part, Daniel Zimini, an expert in artificial intelligence and machine learning and the lead author of the study, explained:
“The importance of this research lies in the convergence of the results of computational engineering with historical science. The algorithm did not rely on historical documents, but only on visual and structural data, yet it concluded that the Armenian alphabet is structurally closest to the Ethiopian alphabet, which has remained a point of contention among historians.”
