An investigation of determinants used in deciding fixation positions in reading of Japanese sentences

本研究通过两个实验，探讨了文字类型对日语句子阅读的影响，以及决定句子阅读固定位置的行列式。在实验中，将不同的文字类型作为实验变量，记录阅读时间和眼球运动。实验的结果揭示了以下几点。有间隔的平假名句子比没有间隔的句子阅读速度更快，注视的时间更短，注视的时间也更短。此外，间隔句的词首注视位置比非间隔句的词首注视位置更接近最佳注视位置。这些结果表明，词间的空格有助于句法分析，并可作为平假名句子阅读时决定词首固定位置的线索。在汉字假名混合句的阅读中，间距对阅读时间和眼动测量没有影响。与平假名句子相比，没有间隔的假名和汉字混合句子的阅读速度更快，注视的时间更短，注视的时间更短。此外，在汉字和假名混合句的阅读中，汉字和（或）平假名旁边的平假名比平假名旁边的平假名更常被注视。这些结果表明，汉字字符使句法分析更容易，并作为线索引导眼睛阅读汉字-假名混合句。平假名句和假名-假名混合句中单词的第一个固定位置更接近单词的开头，而先前固定的位置则更靠近单词的左侧。上述结果表明，日语句子阅读时的注视位置不仅取决于低级视觉过程（检测单词和汉字之间的空格），还取决于先前注视的位置。

In this study, two experiments were conducted to investigate effects of script types on reading of Japanese sentences and determinants used in deciding fixation positions in reading of the sentences. In the experiments, the script types were varied as experimental variables, and reading times and eye movements were recorded. The results of the experiments revealed the followings.1. Spaced hiragana sentences were read faster with fewer fixations and shorter gaze durations than unspaced sentences. In addition, first fixation positions in words of the spaced sentences were closer to the optimal viewing position than those of the unspaced sentences. These results suggested that spaces between words made syntactic parsing easier and were used as cues in deciding the first fixation positions in words in reading of hiragana sentences.2. In reading of kanji-kana mixed sentences, effects of spacing were not found on the reading time and the measures of eye movements. Unspaced kanji-kana mixed sentences were read faster with fewer fixations and shorter gaze durations than unspaced hiragana sentences. In addition, kanji characters and/or hiragana characters next to kanji characters were fixated more frequently than hiragana characters next to hiragana characters in reading of kanji-kana mixed sentences. These results suggested that kanji characters made syntactic parsing easier and were used as cues in guiding the eye in reading of kanji-kana mixed sentences.3. First fixation positions in words of hiragana sentences and of kanji-kana mixed sentences were closer to the beginnings of the words when the locations of prior fixations lay further to the left of the words. These and above mentioned results suggested that the fixation positions in reading of Japanese sentences were determined depending not only on low-level visual processes that detected spaces between words and kanji characters but also on the locations of the prior fixations.