本研究目的是探討閱讀能力高低不同的國小學童在篇章閱讀時，其所產生的朗讀錯誤情形，以深究閱讀能力不同的讀者在閱讀時其識字策略的使用。本研究以學童口語錯誤及自我解釋的資料為主，結果發現國小五年級學童的口語朗讀錯誤可以分成七類：「替換」、「自我更正」、「重複」、「省略」、「贅加」、「顛倒」及「其他」，尤以「替換」類型為最常發生。而根據字形、字音及字義線索又可將「替換」類型之錯誤細分為九個次類別。從分析各種線索在替換類別上次數所佔的比例發現：學童較易運用字義線索，進而在上下文線索的影響下，替換相同字義的字詞。若比較高、低能力組學童的錯誤類型，則發現兩組在「替換」和「其他」類別上的錯誤數量有顯著差異。此外，兩組學童在其他類別上也有質的差異存在，例如在「省略」或是「贅加」類別的分析比較可以發現高能力組所產生的朗讀錯誤較不會與原文有很大的差異，也較不會影響文意，反觀低能力組的朗讀錯誤，較產生文意的改變。

The study investigated the nature and patterns of errors in the oral reading of Chinese elementary school students. Participants were 32 Chinese fifth graders recruited from 4 different classes. Before the oral reading session, students were divided into high and low reading ability groups based on their performance on a reading comprehension test. The task for the students was to read aloud two expository texts, each containing approximately six to eight hundred characters and was selected by experienced teachers to ensure that the materials were age-and grade-appropriate. During each individual oral reading session, data were audio-recorded and later transcribed verbatim. The errors made by the students can be classified into 7 types: Substitution, Self-correction, Repetition, Omission, Insertion, Reversal, and Others. The three most commonly found reading errors were Substitution, Self-correction, and Repetition, each accounted for at least 18% of the total errors made by students in both high and low reading ability groups. Additional analysis focused on the predominant error type, Substitution, by further dividing the errors into 9 subtypes based on the cues (graphic, phonetic, or semantic） that students used. It is noteworthy that when students encountered unfamiliar characters, they tended to derive the pronunciations by using context-related semantic cues, which accounted for more than 40% of the total Substitution error type. When comparing the performance of students in high and low ability groups, results indicated that low ability students made significantly more Substitution type of errors, and their errors, in other types (e.g., Omission and Insertion） were also less semantically acceptable.