Glyph Norming: Human and Computational Measurements of Shape Angularity in Writing Systems

Writing systems are an underused source of stimuli for behavioral and computational experiments in cognitive psychology, psycholinguistics, and anthropology, despite being ecologically relevant, and systematically different in shape, structure, and orientation. One possible reason that glyphs of writing systems are not commonly used in behavioral research concerns their profound complexity. Yet, recent developments in computer vision (i.e., geometric shape analysis) offer tools to automatically assess their visual dimensions. The current work describes an open-access database of 3,208 glyphs from diverse writing systems that have been normed by computational analyses in terms of shape angularity using an array of measurements. We further validate these norms by obtaining human judgements of angularity for a subset of 400 glyphs and show that they correlate highly with computational measures, specially with First-Order Entropy of Edge Orientation. Additionally, we provide methods for standardized glyph generation based on Unicode ranges, a straightforward example of computational shape analysis, and a demonstration of automated transliteration of glyphs from Unicode strings using a pre-existing Python library. These procedures should ease the characterization of angularity of new glyphs and any other kind of visual shape by independent researchers. The present work will be helpful to scientists working across different topics in the various cognitive science subdisciplines.