第二語言閱讀理解之後設路徑模式分析: 譯碼、單字知識、文法知識、第一語言閱讀理解

Abstract

本研究的目的有三：（一）探討第二語言閱讀理解與其相關因素的主要內涵。（二）以統合分析法分析第二語言閱讀理解調節模式中各預測變項之間直接與間接的預測關係。（三）以結構方程模式驗證第二語言閱讀理解模式。 本研究旨在透過統合分析，了解第二語言閱讀理解與其四個預測變相之間的相互關係--- 解碼、單字知識、文法知識、第一語言閱讀理解，並以解碼、單字知識、文法知識為模式的中介變項以觀察變項之間直接與間接影響的效果。在各個觀察變項間的關係上，本研究經由58篇個別研究報告，以統合分析技術蒐集5個觀察變項間的相關係數後，發現這些相關係數均達.05的顯著水準，產生一個5×5的相關矩陣，並以此相關矩陣作為第三階段模式驗證時的輸入資料。 在第三階段中，本研究所提的理論模式在各方面適配度的考驗上皆達理想的適配。故路徑分析的結果證明了解碼、單字知識、文法知識以及第一語言閱讀理解皆為預測第二語言閱讀理解表現的重要因素。且觀察變項之間亦具有顯著的因果關係，尤其字彙知識為其他變項與第二語言閱讀理解之間重要的中介變項。本研究對於第二語言閱讀理解的構成因素進行統合性的分析，透過模型驗證，了解第二語言閱讀理解主要及間接的影響因素，進而提供教師在診斷閱讀理解困難或者設計教學內容時能夠更有系統地瞭解與閱讀理解有關的因素之間相互存在的影響。

The present meta-analysis study aimed to examine the predictive roles of L2 decoding, L2 vocabulary knowledge, L2 grammar knowledge and L1 reading comprehension on L2 reading comprehension of passage level, and also encapsulated the mediation relationships among the variables through path analysis. To this end, a meta-analysis was conducted to aggregate the accumulated correlation evidence from k=58 primary studies and a multi-variate structural equation modeling approach (MASEM) was implemented to test the fit of the hypothesized model on the full samples (n=7147) and observe the mechanism how L2 reading comprehension could be directly or indirectly predicted by the four high evidence reading variables (Jeon& Yamashita, 2014). The results showed that grammatical knowledge had the largest predictive power on L2 reading comprehension, followed by decoding knowledge and vocabulary knowledge. With regard to the mediation effects, L1 reading comprehension could exert significant indirect effects on L2 reading comprehension via the mediation by decoding knowledge and grammatical knowledge. Even though vocabulary knowledge failed to serve as mediator between L1 and L2 reading comprehension, it could significantly mediate the relationships between decoding knowledge and L2 reading comprehension, and grammatical knowledge and L2 reading comprehension, respectively. The best fit of the data supported the assumption of the hypothesized model in which L2 reading comprehension can be strongly predicted by decoding and L2 linguistic knowledge in either direct and indirect fashions. Besides, L1 literacy can be drawn on while developing L2 reading skills. The provisional model also provided educators in second language reading a more systematic and holistic understanding of learners’ reading difficulties.