National characteristics and variation in Arabic handwriting

Abstract

From each of four Arabic countries; Morocco, Tunisia, Jordan and Oman, 150 participants produced handwriting samples which were examined to assess whether national characteristics were discernible. Ten characters, which have different configurations depending upon their position in the word, along with one short word, were classified into distinguishable forms, and these forms recorded for each handwriting sample. Tests of independence showed that character forms used were not independent of country (p < 0.001) for all but one character-position (this was dropped from subsequent analyses). A correspondence analysis ordination plot and analysis of similarity (R = 0.326, p = 0.0002) showed that whole samples were discernibly grouped by country, and a tree analysis produced a classification which was 71% accurate for the original data and 83% accurate for 80 new handwriting samples that underwent ‘blind’ classification. When the countries were combined into two regions, North Africa and Middle East, the grouping was more marked. Thus, there appears to be some scope for narrowing down the nationality, and particularly the wider geographical region of an author based upon the character forms they use in Arabic handwriting.