Abstract: In multi-line text reading, multi-character words at the end of line are often divided across two lines. Unlike English, when Chinese words are displayed across two lines, hyphens are not used after the first part of line-final words. Many studies have shown that words play an important role in Chinese reading. So, does presenting a word on two lines interfere with reading? Li, Zhao, and Pollatsek (2012) found the word boundary effect—the processing times were longer in the divided-word condition than in the word boundary condition. This study further explores the influence of word frequency and reading mode on the word boundary effect. High-frequency words are often stored as a whole in memory, but low-frequency words are more likely to be stored in the form of single characters. It is therefore predicted that the word boundary effect for high-frequency words will be larger. Silent reading and oral reading are two common reading modes. The need for pronunciation consumes more working memory resource in oral as compared to silent reading. When a word is presented across two lines, the reader needs to store the first character information in the working memory. Because the working memory load of oral reading is higher than that of silent reading, it is anticipated that the word boundary effect is stronger in oral reading than silent reading. The experimental used a 2 (presentation condition: word boundary condition vs. divided-word condition) × 2 (word frequency: high vs. low) × reading mode (silent reading vs. reading aloud) within-subject design. The target words were of either high frequency or low frequency, embedded into a same sentence frame. In the word boundary condition, two characters belonging to same word were located at the end of the line. In the divided-word condition, the first character was at the end of the line, and the last character was at the beginning of the next line. Using the EyeLink Portable Duo eye tracker, 48 participants’ eye movements were recorded during reading sentences of different conditions silently and orally. The target word and before and after it was defined as a region of interest. The experimental found the main effect of presentation condition, word frequency, and reading mode for gaze duration, total fixation duration, and total number of fixations. Gaze durations and total fixation durations were longer, and total number of fixations were greater in divided-word, low frequency, and oral reading condition. The presentation condition by word frequency interaction was significant on gaze duration, with a larger word boundary effect for high-frequency compared to low-frequency word. There were also significant interactions between presentation condition and reading mode on gaze duration and total number of fixations, with a larger word boundary effect during oral reading than silent reading. The results show that the presenting a word across two lines interferes with reading, and this interference is affected by word frequency and reading mode. When constructing a more ecologically valid models of eye movement control for multi-line text reading, it is necessary to consider the word boundary effect and the influence of word frequency on this effect. The discovery that dividing a word across two lines interferes with silent and aloud reading has certain application value. In typesetting, the presentation of a word across two lines should be avoided by adjusting the inter-character spacing to improve reading efficiency.

Key words: word boundary effect, word frequency, silent reading, oral reading, eye movement

摘要： 通过使用眼动追踪技术，采用2（呈现条件：同行呈现、跨行呈现）×2（词频：高频、低频）×2（阅读方式：朗读、默读）的被试内实验设计，探讨了词频和阅读方式对词跨行呈现效应的影响。结果发现，呈现条件、词频和阅读方式的主效应显著，跨行呈现、低频和朗读的凝视时间和总注视时间更长、总注视次数更多；呈现条件之间在凝视时间上的差异，高频词大于低频词；呈现条件之间在凝视时间和总注视次数上的差异，朗读大于默读。结果表明，词跨行呈现干扰了阅读，且这种干扰作用受词频和阅读方式的影响。

关键词: 词跨行呈现效应, 词频, 默读, 朗读, 眼动