語音研究中的貝葉斯多變量線性混合效應模型：基於brms的教學

// Bayesian Multivariate Linear Mixed-Effects Models for Speech Research: A Tutorial Using brms

本研究運用Bayesian多變量線性混合模型及R套件brms，同時分析多重語音參數｜以希臘語語句為例，探討升調特徵如何受前置重音及重音位置影響｜多變量方法提升估計精確度，並揭示效應間關聯，提供實務操作指引與優於單一變量分析的證據。

// This study demonstrates the use of Bayesian multivariate linear mixed-effects models with the R package brms to jointly analyze multiple phonetic outcomes. Using Greek utterance data, it investigates how rising accent characteristics depend on preceding accents and stress location. The multivariate approach improved estimate accuracy and enabled examination of correlations between effects. The tutorial offers practical guidance and highlights the advantages of this modeling method over separate univariate analyses.

Published 2026年7月31日