A Quantitative Analysis of Writing Systems: The N’ko Alphabet

Abstract

The region of West Africa is of interest for the study of the origin and development of writing because a number of scripts were created there for several local languages during the 19th and 20th centuries, especially for the Mande family (the Vai, Mende, Kpelle, Looma, and Bamana syllabaries). In 1949 the Guinean enlightener Soulemayne Kanté developed the N’Ko alphabet for the Manding (Manden) languages, which belong to the Mande family and include, in particular, Bamana (Bambara), Jula (Dyula, Dioula), and Maninka. The name “N’Ko” originates from the phrase N ko ‘I say’ in Manding languages. This script is predominantly used in Guinea for Maninka (Maninka-Mori), which is native to more than 3.5 million people in Guinea, Mali and Sierra Leone. The N’Ko alphabet is also widely used in Liberia, the Côte d’Ivoire, and the African diaspora (mainly in Nigeria and Egypt) by a hundred thousand to a million persons. This article provides information about studies of various aspects of the N’Ko alphabet. First of all, the complexity of the graphic forms of each of the 27 letters is calculated according to certain principles. For example, the point corresponds to 1, the straight line segment is 2, and the arc is 3; also certain weight is given to various types of connections and crossed lines. A frequency analysis of the distribution of letters is undertaken in the corpus of Maninka texts written in N’Ko, with more than 3.1 million words. This made it possible, in particular, to trace the extent to which the complexity of the graphic form of the signs correlates with their frequency. It appears that such a correlation is not very significant: the correlation coefficient is –0.38, whereas, for example, for the Morse code in English texts it reaches –0.82. The full inverse correlation, when simpler characters are always used to represent more frequent letters, corresponds to –1. It has also been shown that frequency analysis can serve as a further justification for certain orthographic principles in N’Ko, particularly of tone notation. The next task was to calculate orthographic uncertainty: in an ideal alphabet, where there is a one-to-one correspondence between phonemes (sounds) and graphemes (signs), this uncertainty is equal to zero. In the N’Ko alphabet, its values are quite small: 0.37 without taking into account the tone notation, and 0.22 with tone notation. For comparison, the values corresponding to some “old” writing systems are as follows: in the Ukrainian alphabet, it equals 1.12, while a slightly simpler Italian orthography provides uncertainty at the level of 0.56. The results obtained in this study can be useful for studying phonotactics, prosodic elements, and the history of writing and lexicography, as well as in comparative and contrastive studies.