Closeness, Distance, and Identification of Writers in Latin Paleography
Résumé
After intense debates in the 1980s and 1990s regarding paleography as the “art of measurement,” paleographers of Latin scripts developed quantitative and metrological approaches in the 2000s. These approaches encompassed both the single graphical phenomena described and enumerated by researchers (abbreviations, allographs, scripts, and their uses) and the graphical characteristics analyzed in their dimensions based on digital images (Gurrado and
Stutzmann, 2013). Accompanied by new classificatory paradigms, Latin paleography was able to quickly embrace and benefit from the contributions of computer vision (Muzerelle, 2011).
In this presentation, we aim to explore how the concepts of “closeness” and “distance” apply to the Latin domain and how they have been combined, both in image analysis – where "local" and “global” analysis techniques were initially proposed in parallel –, and in results encompassing a wide range of questions and granular levels. These levels range from the white space between letters and words to the characterization of different allographs and the
measurement of graphical distance (between two occurrences of the same symbol or between two allographs of the same letter), to defining the “rhythm” of a handwriting, and to dating, localizing, and classifying the handwriting of entire manuscripts.
Among recent research works, we will provide a more detailed view of writer identification (including within a homogeneous environment) as evidenced in the recent competitions (Christlein et al., 2019; Seuret et al., 2020), the current revival of letter-level approaches (e.g., Siglidis, 2023), and the impact of image analysis on the categories used in the history of writing.