A new local indicator of spatial autocorrelation identifies clusters of high <i>rendaku</i> frequency in Japanese place names

Abstract The methods of spatial statistics have been successfully applied to the study linguistic variation, especially for detecting existence patterns in geographical distribution features. However, use local indicators autocorrelation clusters limited continuous variables, and we propose apply new method Anselin Li (2019) categorical variables data. We illustrate this with case Japanese rendaku , or sequential voicing, whose dialectal variation is still poorly documented. Focusing on regional differences frequency rendaku, examined occurrence four lexemes 4,921 place names from all Japan. A statistical analysis association an unsupervised density-based cluster revealed two areas high centered around Wakayama Fukushima-Yamagata prefectures. This suggests that more frequent those dialects, recommend further studies start by looking at areas.