The effect of three different orthographical presentations of a text upon the reading behaviors of native and nonnative readers of Japanese: An eye-tracking stud

The Japanese writing system combines a phonological system (Kana) and an ideographic system (Kanji). This study investigated the effects of Kanji upon reading behaviors of native and non-native readers of Japanese.;Two independent variables, Text and Group, were included. There were four levels of Text; three Japanese texts and an English text. A Japanese text was manipulated into three levels (normal text, spaced-Kana text, and no-space Kana text) in order to investigate Kanji's effects as meaning carriers and visual meaning boundary markers. The second independent variable was Group with three levels; the native group (N), the non-native intermediate group (NNI), and the non-native beginning group (NNB).;The four dependent variables were used; fixation frequency, regressive fixation frequency, average fixation duration, and comprehension scores. A 3 x 4 factorial design was used for both multivariate and univariate analyses.;The MANOVA results revealed that Group effects were significant for 4-text analysis (Japanese and English) and 3-text analysis (Japanese only). Text and Interaction effects were significant only for 4-text analysis.;The ANOVA results revealed that the use of fixations was significantly different between the N and NNI groups, whereas length of average fixation duration was significantly different between the NNB and N groups. The results showed that non-native readers of Japanese must reach a certain proficiency level to become capable of using fixations strategically. The N group also used long fixation duration as an important strategy to cope with the English text. Raw comprehension scores revealed that the NNI group could achieve, if allowed to have necessary processing time, a comparable level of comprehension with the N group whereas the NNB group could not.;The differences of comprehension scores standardized to reading time revealed Kanji's effect for each group: both effects as meaning carriers and meaning boundary markers were minimal for the NNB group; for the NNI group, major effect was as meaning carriers; for the N group, both functions were substantial.;The LaBerge and Samuels L1 reading model was used as a theoretical base for the study, and its effectiveness and limitation for L2 reading study was discussed.