Perimetric complexity, a character (letter) complexity metric, can be easily computed with a computer program and thus has the potential for application to characters in a wide range of languages. However, the validity of perimetric complexity for Japanese has only been tested for kana characters. Thus, the validity of perimetric complexity for kanji characters is still an open question that was addressed in the current study. We asked Japanese and English speakers to rate the subjective complexity of different kanji characters, which was averaged into the subjective complexity of each character for each speaker group. On the basis of these ratings, we calculated the correlations between perimetric complexity and subjective complexity. The results revealed three main findings: (a) we found strong correlations between the two factors (rs >.85), (b) the correlation was comparable to that between subjective complexity and other measures of character complexity (i.e., stroke count), and (c) subjective complexity was highly correlated between Japanese and English speakers. These results suggest that perimetric complexity is a valid index of the subjective complexity of kanji and is more useful than stroke count given its multilingual versatility.
