Graphic Compensation in Ancient Greek Documentary Hands: A Computational Paleographic Analysis from Handwritten Character Recognition

Paraskevi Platanou, Lavinia Ferretti, Isabelle Marthot-Santaniello, Giuseppe De Gregorio, Spiros Barbakos, Maria Konstantinidou, Asimina Paparrigopoulou, John Pavlopoulos

Keywords: Digital Paleography, Handwritten Character Recognition, Representation Learning

Domains: Vision and Learning, Other

TL;DR: Using representation learning on character images, we treat neural-network misclassifications as paleographic signals that reveal script similarity and the internal organization of graphic systems.

Abstract:

Historical scripts are structured visual systems whose internal organization has traditionally been explored through qualitative
paleographic analysis. However, many of the principles shaping their evolution remain difficult to test empirically across corpora. One
such principle, formulated by Jean Irigoin, is the hypothesis of graphic compensation, according to which graphic systems tend to
maximize visual distinctiveness between letters that share similar phonetic functions while allowing greater visual proximity in the
other cases. In this work, we introduce a computational approach for exploring the visual structure of historical scripts and apply it
to the cursive Greek writing of Hellenistic papyri. Using representation learning on a corpus of character images, we analyze the
classification behavior of neural networks in order to extract the similarity relations that structure the script. Rather than treating
classification errors as poor technical results, we interpret them as paleographically meaningful patterns that reveal the internal
organization of the graphic system. The resulting similarity matrix provides quantitative support for Irigoin’s hypothesis and suggests
that the principle of graphic compensation continues to operate even in the less controlled domain of cursive handwriting. More
broadly, this study proposes a methodological framework for computational paleography, in which model behavior can be used to
extract interpretable descriptions of script structure from large collections of cultural heritage material. Finally, we suggest that the
learned relationship between phonetic difference and visual confusion may itself be used as paleographic background knowledge
and incorporated in future HTR pipelines, helping recognition systems resolve visual ambiguity when visual evidence and linguistic
expectations compete during decoding.

Submission Number: 164