The distance matrices used in the analyses are:
-Linguistic distance, which is based on the Levenshtein algorithm as calculated through GABMAP
-Time since divergence, which is based on Lee and Hasegawa's 2011 maximum clade credibility tree
-Geographic distance, which is based on the geospatial data of the language varieties
-Separation by water, which is a binary variable indicating the presence of an oceanic barrier
These distance matrices are combined into a single long-format file for the mixed model analysis.
The accompanying R script file provides the syntax for the analyses reported:
-Cluster analysis of Japonic based on linguistic distance
-Simple Mantel correlations between the predictors of linguistic distance
-Partial Mantel correlations between linguistic distance and its predictors
-Multiple regression over distances matrices for linguistic distance
-Linear mixed effects model analysis for linguistic distance that controls for language variety as a random variable