英語口說評量中的同儕互動：後設分析
No Thumbnail Available
Date
2014-09-??
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
英語學系
Department of English, NTNU
Department of English, NTNU
Abstract
近年來，針對英語口說評量中的同儕互動所做的研究，蓬勃發展；有鑑於此，本研究透過量化方式統合現有文獻，以增加該領域新知。在廣泛文獻搜尋後，九個原始研究被納入本研究計算Cohen’s d，據以估計效果量。本研究首先辨識出常被用於測量同儕互動性的互動特徵（例如，話題管理、發出理解信號、合作及使用疑問句等），這些特徵有助於界定英語口說評量構念。本研究結果顯示：隨著效果量與時遽增，對話者對互動表現的整體效果量為中度到高度（d 值為0.96）。語言精熟度對同儕互動表現的效果量也是中度到高度（d 值為0.89），而測驗方式的效果量則為高度（d 值為2.29）。除此之外，本研究亦辨識出數個情境（例如，語言和機構背景、第一語言數量）及研究方法變數（例如，精熟度標準、口說測驗類型、與信度測量方式），因而有助於瞭解該領域中最常應用的研究方法。本文以對未來研究的啟示與建議做結，期能藉此促進英語口說評量中互動研究之進展。
Research on peer-peer interaction in English speaking assessment has been expanding rapidly in recent years, which motivates this researcher to quantitatively summarize existing literature and add new knowledge to the domain. A comprehensive literature search was conducted and nine primary studies were included to calculate Cohen’s d as the effect size estimate. Firstly, a set of interaction features (e.g., topic management, signaling comprehension, collaboration, and using questions) were identified as the most frequently researched measures for peer-peer interaction, which help define the construct in English speaking assessment. A medium to large overall effect of interlocutor variation on interaction performance was found (d = 0.96) along with the report of the trend of effect sizes over time. A medium to large effect of proficiency level (d = 0.89) and a large effect of test methods (d = 2.29) on peer-peer interaction were also found. Additionally, several contextual (e.g., the language and institution settings, and the number of L1s) and methodological variables (e.g., proficiency standards, speaking task types, and reliability measures) were identified, providing a better understanding of research methods that have been typically applied in this field. The article concludes with implications and suggestions for future studies in the hope of facilitating the advancement of interaction studies in speaking assessment.
Research on peer-peer interaction in English speaking assessment has been expanding rapidly in recent years, which motivates this researcher to quantitatively summarize existing literature and add new knowledge to the domain. A comprehensive literature search was conducted and nine primary studies were included to calculate Cohen’s d as the effect size estimate. Firstly, a set of interaction features (e.g., topic management, signaling comprehension, collaboration, and using questions) were identified as the most frequently researched measures for peer-peer interaction, which help define the construct in English speaking assessment. A medium to large overall effect of interlocutor variation on interaction performance was found (d = 0.96) along with the report of the trend of effect sizes over time. A medium to large effect of proficiency level (d = 0.89) and a large effect of test methods (d = 2.29) on peer-peer interaction were also found. Additionally, several contextual (e.g., the language and institution settings, and the number of L1s) and methodological variables (e.g., proficiency standards, speaking task types, and reliability measures) were identified, providing a better understanding of research methods that have been typically applied in this field. The article concludes with implications and suggestions for future studies in the hope of facilitating the advancement of interaction studies in speaking assessment.