「學前大班幼兒閱讀發展評定量表」之編製

The Validity and Reliability of Reading Development Scale for Preschoolers

本研究旨在編製「學前大班幼兒閱讀發展評定量表」，並說明其應用效能。本量表以簡單閱讀觀點（Gough & Tunmer, 1986）及幼兒閱讀發展歷程（Perfetti, Landi & Oakhill, 2005）為構念，評定內涵聚焦於閱讀發展核心技能，含解碼識字、詞素理解以及語言理解三個面向，採四點評定。先根據架構編擬評定項目，初稿24題，再請資深幼教教師協助試評與修題。預試階段，請台中某行政區20名教保人員為其大班幼兒（共263名）評定，並根據兩次因素分析進行試題分析。第一次抽取四個特徵值大於1的因素，總變異量71.39%。依各題因素負荷量及原始構念刪修後餘16題。第二次抽取三個因素，總變異量71.58%。構面與內涵題項與前次相同，形成正式量表。正式量表包含（1）解碼識字（5題）、（2）詞素理解（4題）（3）語言理解（7題）三面項。正式評定擴大至台中市北中南各區公立幼兒園，共徵求18名教保人員172名大班幼兒，進行評訂與實際施測。測量能力包括：認字量、快速自動唸名、注音解碼、詞素理解及語言理解。正式評量資料進行驗證性因素分析，並根據資料進行修正，結果進一步確認此閱讀結構能合理適配資料。正式量表共10題。總量表及各分量與效標間達顯著相關。建構信度、內部一致性（各分項及總量表Cronbachα值介於.87～.95）及評分者一致性信度亦高（ρ=.74,p＜.01）。進而根據此量表之應用提出相關建議。

This study aimed to develop a rating scale of reading development for 6-year-old preschoolers. Based on the perspectives of the Simple View of Reading (Gough & Tunmer, 1986) and the reading development (Perfetti, Landi & Oakhill, 2005), the proposed rating scale is composed of 3 core abilities of reading acquisition, which are decoding, morphological awareness and listening comprehension. The scale was evaluated by 5 preschool teachers through trials in their classrooms. A total of 263 6- year-old preschoolers and 20 preschool teachers were recruited from 20 public preschools in one administrative district of Taichung for the pilot study. The scale was then applied to 172 6-year-old preschoolers from 18 public preschools expansively in northern, middle and southern districts of Taichung. The 172 preschoolers' corresponding reading skills were directly assessed, including character recognition, Tzuyin decoding, Rapid Automized Naming (RAN), morphological awareness and listening comprehension. Three factors were extracted through the data analyses and forming a consistent construct with the theoretical accounts of reading development and were the three main construct of the reading scale, including decoding skill, morphological awareness, and listening comprehension. Confirmatory factor analysis was performed on the formal rating scale from the 172 preschoolers. The results showed that over all model fit indices, suggesting that the constructed reading model fit the observed data. The final rating scale consisted of 10 items. Correlation results also showed significant association of rating scores with direct assessing criteria. Coefficients of Cronbachα(ranged from .87 to .95) and Spearman rank correlation (ρ=.74, p＜.01) for internal and inter-rater reliabilities, respectively, were acceptable. The results suggest reasonable validity and reliability of the rating scale.