This paper proposes a research framework for graphic digital text Watermarking algorithms and methods. We summarized graphic text watermarking after a brief review of digital text watermarking, and propose the research framework that consists of 8 levels, such as pixel, line, etymon, character, row, paragraph, page, and chapter. According to the embedding methods, each level is divided into three typical characteristics: self-characteristics, similarity and structural feature. At last we proposed an algorithm as case study and discussed further research's possibility.