本研究旨在探討高低先備知識與高低英語閱讀能力的讀者在閱讀英語物理文本的推論歷程。為了得知讀者的閱讀歷程，本研究採放聲思考作業，為2(組別：高先備知識低英語閱讀能力組，高英語閱讀能力低先備知識組) x 2(實驗材料：有/無引導插入句)的混合設計。中央大學大學部的學生為受試對象，理工相關科系的學生為物理高先備知識組，文、商學院的學生為物理低先備知識組，並以GEPT中級的閱讀項目區分受試者英語閱讀能力，前30%的學生為高英語閱讀能力組，後30%的學生為低英語閱讀能力組。每位受試者閱讀兩篇實驗材料內容，兩篇內容的差別在於有無引導插入句。插入引導句呈現的目的是在探討引導句的有無可否增加低先備知識讀者的推論。每篇材料有三個閱讀理解題項，讀者在閱讀完文本之後立即回答三個閱讀理解題項，以確認理解程度及引導插入句所發揮的作用。 本研究發現當讀者閱讀英語物理篇章時，學科知識為影響推論的主要因素。閱讀歷程中，高先備知識低英語閱讀能力讀者產生精緻化推論，藉由語境線索觸發先備知識達到正確推論；低先備知識高英語閱讀能力讀者產生解釋推論，採逐字閱讀，遇到學科詞彙會整合前後詞彙並觸發一般知識(real world knowledge)形成推論，但因學科詞彙知識不足，不正確推論次數明顯多於正確推論次數。比較兩組讀者閱讀有卅無引導插入句文本的理解分數發現低先備知識組的讀者閱讀有引導插入句文本的理解分數有顯著的高於閱讀無引導插入句文本。根據本研究結果建議，授課教師可視學習者程度採用合適教材並在教學過程中補充文句與段落間的知識概念，以利學習者有效達到閱讀理解。This study examined the EFL readers with high-low English reading ability and high-low prior knowledge by recording participants’ on-lined reading Physical text processing. The experimental design was 2 ( group: high prior knowledge low English ability, low prior knowledge high English ability) × 2 ( material : with / without bridging inference sentences). The participants were recruited from the undergraduate students of NCU. The Engineering undergraduate students form a high prior knowledge group; whereas, the college of Liberal Arts, and the school of Management students form a low prior knowledge group. In order to check the participants’ English ability, GEPT was conducted. Two Physic articles were read, which differ in the existence of bridging inference sentences. The existence of bridging inference sentences act as support to enhance participants’ making inferences during reading. There are three reading comprehension questions in each article. Every participant has to answer those questions to examine their comprehension of the article, as well as make sure the effect of the bridging inference sentences. The result of the study provided that the domain knowledge is the main factor to affect readers’ making inference. High prior knowledge participants generated elaborative inferences by reading context. Low prior knowledge participants generated explanation-based inferences by reading word-by-word. When reading domain vocabulary, low prior knowledge participants attempted to construct a meaningful referential with real world knowledge. Due to the lack of domain knowledge, the number of incorrect inference which low prior knowledge participants generated was more than correct inference. Comparing two groups reading comprehension question scores, the articles with inference sentences showed to help low prior knowledge participants gained higher scores. The result suggested that the teachers should provide supporting knowledge to suit students’ domain knowledge as well as fill the gap between sentences, which would help learners comprehend the article effectively.
