從校稿失誤探討中文閱讀的單位

胡志偉(1989)引用Healy(1976)的「單位整體化模式」解釋反置詞偵測的詞劣效果。根據這個效果，他並提出詞是中文的閱讀單位，且詞之構形特徵是文字辨識的重要線索。本研究的目的是重新檢驗胡志偉的研究與觀點。在實驗一中，研究者操弄了文章呈現方式（橫排與直排）與目標項類別（形異音似錯別詞、形似音異錯別詞、高頻與低頻反置詞）並同時記錄參與者閱讀文章時的眼動情況。實驗發現：（一）相較於「形異音似錯別詞」其它三類目標項的被偵測率皆出現詞劣效果，（二）反置偵測之詞劣效果不受文章呈現方式與詞頻的影響，（三）眼動資料結果顯示，四類目標項的被跳讀率相似，皆低於3%。實驗二以紙本方式呈現直排文章，重複驗證了實驗一之直排情境的發現。綜合本研究的結果顯示，對於中文讀者而言，雖然詞可能是閱讀的單位，且詞的構形特徵是文字辨識的重要線索，但造成反置詞偵測的詞劣效果的原因並不符合胡志偉的觀點。

關鍵字

中文閱讀歷程；文字辨識；眼動模式；詞劣效果

並列摘要

Hue (1989) reported a word inferiority effect in detecting character-position transposed words embedded in text. To explain the phenomenon, he cited Healy's (1976) unitization model, and argued that ”word” is a reading unit of Chinese. Furthermore, he argued that in word recognition, a word's orthographic configuration is important cue, and that the word inferiority effect he found was caused because a character-position transposed word still possesses its orthographic configuration. The present study was a follow up of Hue's experiment, and in particular, was designed to test his theory concerning how the word inferiority effect is produced. In Experiment 1, the participants' eye movements were recorded while they were detecting misprints in the reading materials which were presented on a computer screen in two different writing directions (vertical vs. horizontal). There were four types of misprinted two-character words: 1. High frequency words with characters transposed in position, 2. low frequency words with characters transposed in position, 3. words with their second character replaced by an orthographically similar one, and 4. words with their second character replaced by a phonologically similar character (controls). The experimental results show that writing direction neither affected the detection rate nor the skip rate of the four types of misprints. Compared with the detection rate of the controls, the detection rates of the other three types of misprints are lower (i.e., word inferiority effect). In Experiment 2, the vertical printing condition of Experiment 1 was replicated, only that in Experiment 2, the reading materials were printed on papers. These results suggest that two-character words can be reading units of Chinese and, in word recognition, a word's orthographic configuration is important cue. However, the results of the present study also indicate that Hue's idea concerning how the word inferiority effect is caused is not correct.