On the Formalization of Glyph in the Chinese Language

謝清俊、張仲陶、黃克東

Abstract

Although the word "glyph" is a newly created word, the concept of a "glyph" has been used in China for thousands of years. Traditional morphological studies of the Chinese language include numerous documents that study the "forms" of characters. The knowledge about the "forms" includes the originality, the expressional structure, the content structural phenomena, the structual law, and the functional rules of Chinese characters. The 2000 years old morphlogical work, LIU-SUE (六書), still provides a keynote quide-line for studying the "forms" of character. This knowledge includes how a character was formed, how a character was composed of by its components/roots, how a charracter appears, what is the original meaning of a character, etc. This information helps people to understand the "forms" of a character, but it doesn't help computers. Therefore, we need to formalize this related knowledge in order to let the computer handle characters, glyphs, and fonts more elegently than before.

In this paper, the authors are trying to formalize the related phenomena of glyphs for Chinese characters, according to the characteristics of the Chinese language. There will be no invention concerning language, but the formalized system will provide the computer with logical data structures and procedures so that it can carry out the basic functions needed to create glyphs and characters in Chinese language.

From now on, for simplicity's sake, the paper will not mention to any morphological works from Chinese linguistics. This assumption does not mean that the work in this paper ignored the traditional Chinese linguistical aspects. On the contrary, all the proposed mechanisms are based on the morphological nature of the Chinese language, although it will not be explicitly cited. For those who are not familiar with the morphological characteristics, or for those who can not read Chinese, the book "Formalization of Natural Languages" by P. Kummel (ISBN 3-540-08271-9) is highly recommended as a reference.