Validity of perimetric complexity index for Japanese Kana character complexity

Abstract

Perimetric complexity, which is a simple metric of character (letter) complexity defined by an image’s area and peripheral length, has been widely used, especially in alphabetic orthographies. We examined whether perimetric complexity is also a valid index for Japanese kana characters (hiragana and katakana) by comparing it with subjective complexity. We obtained evaluations of subjective complexities from Japanese and English speakers and calculated the mean of each character for each type of speaker for character-based analyses. The analyses revealed three main findings: (a) Perimetric complexity was highly correlated with subjective complexity (rs > .85), and its correlation was higher than that between the subjective complexity and other measures for character complexity (i.e., stroke count). (b) The perimetric complexities were highly correlated across different typefaces, except for significantly different typefaces. (c) Subjective complexity was highly correlated between Japanese and English speakers. These findings suggest that perimetric complexity can also be used as an index for Japanese kana character complexity.