From Papyrus to Pixels: Optical Character Recognition Applied to Ancient Egyptian Hieratic

In this thesis, Optical Character Recognition (OCR) was used to investigate the morphology of Middle Egyptian Hieratic, a cursive form of the ancient Egyptian hieroglyphic script used primarily in the ancient Egyptian Middle Kingdom and after. A data set of 13,134 individual hieratic signs was created from existing and new facsimiles and a novel program was developed to analyze them using an Image Deformation Model. It was determined that the program operates with high accuracy, can distinguish separate signs, and can perceive variation from the original texts beyond that of modern facsimiles. When the program was used on large portions of the data, some signs previously thought to be effectively identical have been shown to be morphologically distinct. Also, the "tails" of A1 signs were shown to be largely irrelevant in the analyses, implying that they do not convey significant information. Beyond this, the program was used to answer questions about various well-known texts: it validated the established hypothesis that the Shipwrecked Sailor and Papyrus Prisse were written in the same hand, isolated sign forms unique to the town of Lahun, demonstrated similarities between Papyrus Ebers and the Rhind Papyrus, and determined sign forms unique to Papyrus Westcar. The work provides a starting point for future study in hieratic digital paleography, allowing far more signs to be compared at once than ever before, and offers a free, open-source tool and data set.