Using principal component analysis to explore co‐variation of vowels

Abstract

This paper presents a methodology for exploring systematic co‐variation of vowels using Principal Component Analysis (PCA). As a case study, we examine and build on Brand et al.'s (2021) study of systematic co‐variation amongst the monophthongs of New Zealand English (NZE) across speakers born over a 118‐year time period. We present PCA as a methodology, with information aimed at readers who may themselves want to use it in a related context. We consider tests for the appropriateness of PCA, how to select Principal Components, and how to interpret them once they have been found. At each stage, we provide code in the R programing language to enable readers to both follow our analysis and apply the same methods to their own data.

Analysis of the Kokuho Database to Identify regional differences based on complex Health Data of latter-stage elderly

“…In township-level analysis, the principal component analyses (PCA) is used because the original data are quantitative and meaningful. The PCA is a statistical analysis method that reduces the number of variables in data to minimize information loss and create a small number of uncorrelated composite variables (principal components) [15]. Thus, it is possible to understand information in a lower dimension through dimensionality reduction and find values that fall outside the major components of multidimensional data.…”

Determining The Number of Principal Components with Schur's Theorem in Principal Component Analysis

“…In recent years, important research has been done using PCA. In [8], a method for investigating systematic co-variation of vowels has been presented by using PCA. In [9], an application of principal component analysis has been obtained to reduce the dimensionality of variables representing the speech signal.…”