探討科學文章閱讀任務中，工作記憶廣度、 SRL策略使用及閱讀理解表現間的關聯性

Abstract

閱讀是科學學習的途徑之一，多數學習者透過閱讀來建構科學概念，在閱讀過程中，學習者會監控並調整閱讀行為和狀態，以提升學習成效。然而學習者本身工作記憶廣度的大小會影響其認知資源上的分配，採取不同的SRL策略類型或組合，導致不同的閱讀理解表現，但過去的研究並未深入探討工作記憶廣度與閱讀歷程間的關聯性。故本研究目的在於分析工作記憶廣度如何影響學習者在SRL策略使用上的不同，進而影響其閱讀理解表現。 本研究採質量混合設計，探討非生物主修之大學生於網路平台進行科學文本線上閱讀的自我調節學習行為和表現。本研究以參與者的先備概念測驗表現來確認其對文章內容的概念很有限，接著以分類工作記憶廣度測驗表現進行分組，請參與者進行線上科學文本閱讀任務。在進行閱讀任務的過程中，錄下參與者以滑鼠輔助閱讀等行為及放聲思考所展現的閱讀歷程與理解內容，以分析不同工作記憶廣度的參與者，其SRL策略的使用類型與頻率和閱讀理解表現是否不同?有何不同?除量化資料外，研究者亦以工作記憶廣度高、中、低三組中的高理解和低理解典型個案各兩名，以分析不同廣度的參與者在閱讀任務中使用的SRL策略的模式與類型如何影響閱讀理解的表現。 研究結果顯示，參與者工作記憶廣度的大小與其閱讀理解表現並未呈現顯著正相關，但部分認知策略如總結、推論和連結資訊等，則與閱讀理解表現呈顯著正相關；並另有少部分認知策略如假設、重新閱讀等與閱讀理解表現為負相關。分組比較的結果亦發現，高、中、低工作記憶廣度的參與者，在使用SRL策略的類型與頻率上並無顯著差異，唯中、低工作記憶廣度的參與者使用較多「控制情境」策略輔助閱讀，且中廣度者顯著使用較多「重新閱讀」認知策略來閱讀。質性資料則進一步說明，無論工作記憶廣度大小，當參與者能使用較多有助理解的認知策略，且交互使用後設認知及認知策略、利用後設認知策略來判斷所閱讀的訊息是否有助理解，且將注意力放在重要訊息上加以組織，則能建立較高理解層次的推理命題或延伸概念，展現較佳的閱讀理解表現，唯工作記憶廣度較為不足的參與者，閱讀過程中常輔以「控制情境」策略協助閱讀。然而當參與者使用較多無助理解的認知策略，如假設、重新閱讀等，且較少使用「控制情境」策略幫助閱讀，則大多僅能組織段落內相關訊息產生文意命題，極少整合或連結跨段落內相關訊息，則無法達到較高層次的理解程度。 本研究顯示在閱讀理解過程中，學習者的工作記憶廣度大小會影響閱讀理解過程的認知和後設認知策略使用模式，進而影響閱讀理解表現。中、低廣度的學習者可運用控制情境策略來克服工作記憶廣度較小的限制，達到同樣的閱讀理解表現。

Reading is one of the ways in learning science, and most learners learn science concepts through reading. According to current SRL model, learners automatically monitor and regulate their reading behavior and learning status to understand the text. Previous studies have shown that the learners’ working memory span may affect their allocation of cognitive resources and use of SRL strategies in read, which may result in difference in reading comprehension. However, previous studies did not explore whether working memory span would affect SRL behaviors in reading nor the relationship between and among working memory span, use of SRL and reading comprehension. The purpose of this study is to analyze whether and how the working memory span affect learners’ use of SRL strategies in reading a science text, and whether the difference in working memory span and SRL results in different reading comprehension performance. This study used a mixed-method approach to explore non-biology majored college students’ SRL behaviors and reading comprehension when reading a science text in a computer-based learning environment. A biology concept test was used to assess if the participants possessed little understanding about the topic of the text. A categorization working memory span test was then implemented to assess the participants’ working memory span. During the reading task, tools were given to allow the participants to highlight sentences when they read, and the participants were asked to think-aloud during the entire reading process. The participants’ non-verbal reading behaviors and audio data of self-reported thinking process were recorded and transcribed for further analysis. The researcher analyzed and compared differences in types and frequencies of SRL strategies as well as performance of reading comprehension for groups with high, medium, and low span of working memory span. Qualitative case studies were conducted on high and low performers with high, medium, and low span in order to understand if their SRL patterns affected the reading comprehension. The results showed that there was no significant positive correlation between the working memory span and performance of reading comprehension. Some cognitive strategies, such as summarization, inference, and coordination of information sources were significantly and positively correlated with reading performance. Negative correlations were found between reading performance and few cognitive strategies, including hypothesizing and re-reading. Comparisons between the groups with high, medium, or low working memory span indicated that no difference was found on the types and frequencies of SRL strategies, except that learners with medium span used significantly more re-reading than their counter parts. Participants with medium and low working memory span used more control of context during reading. The qualitative analysis indicated that, regardless the working memory span, the participants could effectively combine metacognitive and cognitive strategies to help them understand the text. The high performers paid attention on inferring and summarizing important information to build their understanding and utilized metacognitive strategies to evaluate whether the information they just read was relevant and adequate to the key ideas of the text. Thus, the high performers generated more propositions about key concepts of the text and achieve the level of situational understanding. Only the participants with the lower working memory span frequently used control of context strategy to facilitate their read process. Low performers, on the other hand, more frequently used strategies such as hypothesizing and re-reading; in addition, less control of context were used while reading the text. They could only coordinate information within the same paragraph and seldom linked the relevant information across paragraphs. Thus, they understood the text at the text-based level and mainly synthesized propositions for substance understanding rather than for conceptual understanding. This study concluded that working memory span affects learners’ patterns of metacognitive and cognitive strategies during reading process and, in turn, may result in different level of reading comprehension. Moreover, learners with medium and low working memory span can achieve equal performance on reading comprehension by using control of context appropriately to facilitate information processing.