This meta-analysis of 79 studies evaluates the effectiveness of high variability phonetic training (HVPT) for the development of second language (L2) speech perception and explores learner-related and methodological variables that influence training effects. The overall medium-to-large effects of HVPT on L2 speech perception support the effectiveness of HVPT, for both pretest-posttest comparison (g = 0.92, k = 96) and treatment-control comparison (g = 0.67, k = 32), confirm long-term retention of perception gains, and, to some extent, indicate generalization of learning to novel stimuli. Training effects are influenced by several key variables (length of L2 learning, response labels, type of training task, type of testing task, total training time, target phones, and number of talkers). The findings provide compelling evidence to support the efficacy of HVPT for L2 perceptual learning and suggest circumstances under which training effects are optimized.

这项对 79 项研究的荟萃分析评估了高可变性语音训练 （HVPT） 对第二语言 （L2） 语音感知发展的有效性，并探讨了影响训练效果的学习者相关变量和方法变量。HVPT 对 L2 语音感知的总体中到大型影响支持 HVPT 的有效性，因为前测后测比较 （g = 0.92， k = 96） 和治疗对照比较 （g = 0.67， k = 32），证实了感知增益的长期保留，并且在某种程度上表明学习对新刺激的泛化。训练效果受几个关键变量（L2 学习时长、响应标签、训练任务类型、测试任务类型、总训练时间、目标电话和说话者数量）的影响。这些发现提供了令人信服的证据来支持 HVPT 对 L2 感知学习的有效性，并提出了优化训练效果的情况。