Alternating Reading Task (ART) Corpus

Alternating Reading Task (ART) Corpus是一个专为研究语音交流中的趋同和模仿行为而设计的数据集。该数据集包含三种实验条件：单独阅读、交替阅读和故意模仿，以及涵盖法语、意大利语和斯洛伐克语口音英语的三个子语料库。这种设计允许在控制和非自发环境中系统地研究语音趋同。数据集还包括详细的转录、英语熟练度评分、人口统计信息和实验内问卷，用于探究语言、个人和人际因素对趋同的影响。该数据集适用于AI驱动的语言教育、表现性语音合成和说话人识别任务，其灵活的设计可应用于多种语言和对话情境。

Alternating Reading Task (ART) Corpus is a dataset specifically designed to study convergence and imitation behaviors in spoken communication. This dataset includes three experimental conditions: solo reading, alternating reading, and deliberate imitation, as well as three sub-corpora covering accented English with French, Italian, and Slovak accents. This design enables systematic research on speech convergence in both controlled and non-spontaneous environments. The dataset also contains detailed transcripts, English proficiency scores, demographic information, and in-experiment questionnaires, which are used to explore the impacts of linguistic, personal, and interpersonal factors on convergence. This dataset is applicable to AI-driven language education, expressive speech synthesis and speaker recognition tasks, and its flexible design can be applied to multiple languages and dialogue scenarios.