As indicated by some studies, the problem of prior knowledge often exists when exploring the outcome of reading comprehension in academic language learning. This pilot study aimed to employ eye tracking technology to explore how students with different levels of prior knowledge processed the content of manga-based interactive E-book while learning Japanese language. Students’ visual behaviors were tracked and recorded when they read a Japanese conversation with the relationship between graphical manga and interactive textual annotations. According to the pretest scores, 6 university students were categorized into high and low prior knowledge (PK) groups. Using EyeNTNU-120 eye tracker to compare including Total Contact Time (TCT), Number of Fixations (NOF), and Number of Clicks on textual annotations of the two PK groups based on areas of interests (AOIs) was measured. After the eye tracking experiment, students received a posttest of reading comprehension. The results revealed that (1) the high PK students showed longer reading time in graphic AOIs than the low PK students, (2) the low PK students showed longer reading time in text AOIs than the high PK students, (3) the low PK students showed longer reading time in annotation AOIs than the high PK students, (4) the high and low PK students had no significant difference in the whole reading time, (5) the low PK students showed more NOF of texts that the high PK students, (6) the low PK students clicked many of annotations AOIs than the high PK students, and (7) the low PK students had a significant outcome of reading comprehension compared with pretest and posttest scores. This suggests that interactive E-book containing graphical manga attracted students’ visual attention and improved students’ outcome of reading comprehension. Suggestions are made for future studies and instructional design for interactive E-book learning.