The main purpose of this paper was to discuss the interrelation among graphemic, phonemic, and semantic processing. In order to make this interrelation clear, we investigated the reading processes of Kanji and Kana script from the viewpoint of cognition. In the first section, the interrelation was explained by both data-driven and conceptually-driven processing, and in the second section, the effects of various factors on these two types of processing were revealed by analyzing the processes by which Kanji and Kana script were read. These factors were as follows: 1) the number of letters in the target, 2 ) the meaningfulness of the target, 3) the presence or absence of prior warning, 4) the existence of contexts, and 5) the "discrimination potential" of the target in sentence.