SPACER
收藏资源简介:
SPACER数据集是由加州大学欧文分校语言科学系创建的,包含1056个自然发生的单字替换错误及修正的语句,以及5808个理解者对这些初始语句的回应。数据集中的语句从Switchboard语料库中提取,并通过网络文本编辑实验收集理解者的修正。该数据集旨在为研究语言生成和理解中的错误监测和修正提供一个并行数据集,以探究不同修正策略之间的不对称性,并促进语言生成与理解领域中整合性方法的发展。
The SPACER dataset was created by the Department of Linguistics, University of California, Irvine. It contains 1056 naturally occurring single-word substitution errors and their corrected utterances, as well as 5808 responses from comprehenders regarding these initial utterances. The utterances in the dataset are extracted from the Switchboard Corpus, and the comprehenders' corrections were collected via web-based text editing experiments. This dataset aims to provide a parallel dataset for research on error monitoring and correction in language production and comprehension, to explore the asymmetry among different correction strategies, and to promote the development of integrative approaches in the fields of language production and comprehension.

- 1SPACER: A Parallel Dataset of Speech Production And Comprehension of Error Repairs加州大学欧文分校语言科学系 · 2025年



