多模态学习分析应用于公众演讲能力评估的 系统性综述

A Review of the Assessment Methods of Public Speaking Competence Based on Multimodal Learning Analysis

学习者公众演讲能力的精准评估是开展个性化演讲教学的必要前提,对提升人才培养质量意义深远。随着多模态学习分析技术的发展成熟,如何实现针对较大规模学习者公众演讲能力的多模态评估,是多模态学习分析规模化教育应用的挑战性课题。本研究采用系统性文献综述的方法,总结了十年来多模态学习分析应用于演讲能力评估的48项实证研究。首先,总体描述了相关研究的学科领域、研究对象的教育背景、样本规模及主要演讲语言。其次,结合多模态演讲教学数据的采集与分析过程,系统梳理了多模态的数据形态、特征提取与融合计算方法、常用的多模态演讲教学数据集及公众演讲能力多模态评估系统。最后,总结了多模态学习分析技术应用于公众演讲能力评估的优势和挑战。文章旨在系统梳理多模态学习分析应用于演讲能力评估的研究基础与应用现状,为较大规模公众演讲教学数据集的构建以及公众演讲能力的多模态评估提供有益参考。

Assessment of learners’public speaking competence is the prerequisite for personalized public speaking instruction,which has far-reaching significance for improving the quality of talent training.Despite of the development of multimodal learning analysis technology,it is still a challenging issue to achieve the multimodal assessment of public speaking ability at a large-scale.This study adopted the method of systematic literature review and analyzed 48 empirical studies on the application of multimodal learning analysis in the assessment of public speaking for the recent ten years.It summarized the disciplinary areas of the 48 studies,the educational background of the research participants,the sample size and the main language used in public speeches.Then,the study synthesized the forms of multimodal data,feature extraction and data fusion methods commonly used in multimodal learning analysis and popular multimodal assessment systems for assessing public speaking competence.Finally,the advantages and challenges of applying multimodal learning analysis technology to public speaking ability assessment are noted.This paper aims to systematically review the research status of multimodal learning analysis in the assessment of public speaking competence.It provides implications for setting up a large-scale public speaking database and improving the multimodal assessment of public speaking competence.