LiteRaS studies Arabic manuscripts written not in Arabic letters, but in Greek, Coptic, Ethiopian, and Latin scripts, a practice known as Arabic allography. These texts were produced mainly by Christian communities at the margins of the Arabic-speaking world and remain scattered and under-studied. The project asks how Arabic was written and adapted across these different scripts, how features vary across communities, and what these writing practices reveal about the literacy, identity, and cultural belonging of the people who produced them.

The project relies on two computational approaches. NLP supports the analysis of the encoded texts, allowing systematic comparison of linguistic and orthographic features across scripts, regions, and communities. Computer vision supports the study of the manuscripts as visual artefacts, helping to detect scribal patterns, analyse handwriting, and group manuscripts by visual similarity. Together, they help answer how the language behaves across traditions and what scribal practices reveal about the communities behind them.