关于语音计时中相关性使用的说明。

Note on the use of correlations in speech timing.

作者信息

Benoit C

出版信息

J Acoust Soc Am. 1986 Dec;80(6):1846-9. doi: 10.1121/1.394302.

Abstract

A large number of investigators use Pearson's product-moment correlation in analyzing the temporal organization of speech. Some of their papers present results obtained by correlation between overlapping intervals. The purpose of this letter is to show that a high correlation could be a "statistical artifact" in such a method. Second, it mathematically points out incorrect reasoning by Tuller and Kelso in their reply to such an argument [J. Acoust. Soc. Am. 76, 1030-1036 (1984)].

摘要

许多研究人员在分析言语的时间组织时使用皮尔逊积矩相关。他们的一些论文展示了通过重叠区间之间的相关性获得的结果。这封信的目的是表明，在这种方法中，高相关性可能是一种“统计假象”。其次，从数学角度指出了图勒和凯尔索在对这一论点的回应中存在的错误推理[《美国声学学会杂志》76, 1030 - 1036 (1984)]。