文摘
The 100-language WALS sample is used to create a dataset of complexity variables. Any language in the sample is more complex than any other language in at least one variable. When we calculate Pearson correlation coefficientes between the variables, only 4.8% of them are negative and significant. That proportion increases to 9.15% when we calculate partial correlation coefficients. An additional increase (to 13.7%) occurs if we control for geographic, genetic and demographic variables.