Abstract: Unlike alphabetic scripts, such as English, the vast majority of written languages have no space information to delimit words, for example Chinese and Japanese. Chinese text is written without spaces between successive characters or words. There is no obvious visual cue to demarcate words except punctuation marks. Given this, it is intriguing how readers target their saccades and how words are recognized during Chinese reading for Chinese people and Chinese language learners.
Bai et al. (2009) investigated how American international students read Chinese sentences with or without spaces. English, which was the participants’ mother’s language, has interword spaces in the text. Four spacing conditions were included in the experiment: normal unspaced condition; single character spaced condition (text with spaces between every character); word spaced condition (text with spaces between words); and nonword spaced condition (text with spaces between characters that yielded nonwords). The results suggested that American readers’ Chinese reading was facilitated under the word segmentation condition compared to the normal condition because they have no unspaced text reading experience, thus the spaces could help with word segmentation during Chinese reading. Therefore, it is considered that the experience of reading text with visual segmentation cues plays an important role during Chinese reading for people who learn Chinese as a second language as Li et al. (2010) suggested.
Typical Japanese text is a mixture of Kanji, Hiragana & Katakana. There is no interword spacing in ordinary Japanese script. However, Japanese readers are, to some extent, used to interword spacing, as children are initially taught to read spaced Hiragana to aid learning. Sainio, Hyöna, Bingushi, & Bertram (2007) investigated the role of interword spacing in pure Hiragana and mixed Kanji-Hiragana text. The results indicated that interword spacing served as an effective segmentation cue during Hiragana text reading; spacing information in mixed Kanji-Hiragana text was redundant, since the visually salient Kanji characters served as effective cues by themselves. Therefore, Japanese readers are considered to be very familiar with those visual cues provided by the text during reading.
Although Japanese text has no interword spaces, the mixed text can provide readers effective visual word segmentation cues. However, there is no such cue in normal unspaced Chinese text. Therefore, we predicted that interword spaced Chinese text would have a similar facilitatory effect for Japanese students as it did for American students, considering interword spaces could help them to demarcate Chinese words. That is to say, their reading performance under the word spaced condition would be better than that under unspaced condition.
Two experiments were conducted including normal text, word, nonword and character segmentation conditions using spaces or highlighting as the visual segmentation cue. Twenty-four Japanese-Chinese bilinguals participated in these two experiments. An EyeLink 2000 eye tracker (SR Research, Canada) was used to record their right eye movements, sampling every two milliseconds.
Same results were observed in the two experiments. No significant difference was found between word segmentation and normal text conditions on total number of fixations and total sentence reading time for global analyses. However, the local analyses of eye movement measures showed that Japanese students required shorter total reading time and fewer total fixations in the word segmentation condition than in the normal text condition. The results suggested that sentences with word segmentation cues were as easy to read as normal sentences for Japanese students. Thus, word segmentation may facilitate Chinese word recognition for Japanese students.

Key words: word segmentation, Japanese-Chinese bilinguals, Chinese reading, eye movements