The graphic code of the Old Tatar monument: language variability and the challenges of the digital era

The article presents a comprehensive interdisciplinary study of the problems of deciphering, transliterating, and digitally representing the Old Tatar written heritage based on the unique collections of the Center of Written Heritage of the G. Ibragimov Institute of Language, Literature and Art (“Mirashane”). The relevance of the work is driven by the need to integrate a vast array of Arabic-script Turkic texts, numbering over 50 thousand depository items (including manuscripts of the 14th–20th centuries, personal archives of G. Tukay and G. Iskhaki, early printed press, and epigraphic monuments), into the modern global information space. The aim of the research is a deep analysis of the evolution of academic and technological approaches to the machine processing of the Old Tatar graphic code. Structural-linguistic, source study, and contextological analysis, along with advanced digital humanities tools, were applied as research methods. The fundamental problem of Old Tatar script variability, which historically complicates direct optical character recognition (OCR), is revealed. The results of the long-term development of the domestic system of algorithmic transliteration “MirasTat” and large-scale consortium projects on deep learning of neural networks are presented. It is proven that the radical solution to the problem is the use of generative artificial intelligence and large language models (LLMs) trained on reference academic datasets, which led to a tectonic shift in digital Turkology. The conclusion is drawn about the necessity of a conceptual transition from mechanical visual digitization to the paradigm of semantic full-text search, as well as the importance of creating specialized data centers and developing a unified canonical transcription system.