综合性听说结合任务构念中的语音特征研究  

Investigating pronunciation features of the integrated listening-to-speak task construct

作　　者：徐鹰[1] 章雅青 XU Ying;ZHANG Yaqing

机构地区：[1]华南理工大学外国语学院,广东广州510641

出　　处：《外语测试与教学》2021年第3期39-48,共10页Foreign Language Testing and Teaching

基　　金：教育部人文社会科学研究青年基金项目(编号:17YJC740102);2019年广东省高等教育教学改革项目(编号:粤高教函2019[52]号);2019年广东省大学英语教学改革项目(编号:YDY-2019-11)的部分研究成果。

摘　　要：综合性听说结合任务因能有效测试考生的口语能力而得到广泛应用,然而该任务中的语音维度包括哪些区分性特征还缺乏实证证据。本研究以广东省高考英语听说考试故事复述任务为例,采用语料库的方法对影响综合性听说结合任务表现的语音特征进行描写和诊断分析。结果发现,音素吞音错误、辅音错误、单词读音错误以及语调等4种语音特征能有效区分不同水平的考生。根据语音特征构建的故事复述分数预测模型是考生分数(24分制)=12.427-0.304×单词读音错误标准频数-0.158×辅音错误标准频数+0.216×语调分。本研究结果对综合性听说结合任务评分标准开发和考试备考具有重要意义。The integrated listening-to-speak task has already been widely used as it can measure test-takers'speaking ability validly.However,there is a paucity of research on the distinctive pronunciation features in the task construct.This paper aims to investigate pronunciation features which may influence the retelling task performance based on a corpus of test-takers'performance in Computer-based English Listening and Speaking Test(CELST)of National Matriculation English Test(Guangdong Version).The sampled students'speech records were transcribed and coded within a framework of pronunciation features.It was found that sound swallowing,consonant error,word error and intonation can differentiate test-takers at different ability levels.Based on these features,the following story retelling task score predictive model was constructed:Retelling Score(in a 24-point system)=14.427-0.304×Standardized Frequency of Word Error-0.158×Standardized Frequency of Consonant Error+0.216×Intonation Score.Implications are drawn for rating scale development and test preparation.

关 键 词：综合性听说结合任务 故事复述 语音特征 音素吞音 语调 

分 类 号：H319[语言文字—英语]