a) In equation (3): The numerator should actually be "LDT - mean(LDT)" instead of LDT only (thanks to Bodo Winter for pointing this out).

b) The p-value for the mixed-effects regression in Table 4 (p=0.0375) is calculated by plugging the t-value (-2.087) into the t-distribution. This method can be anti-conservative for small samples. If our sample of 426 data points in 91 languages is seen as small, then an alternative method is to built a null model (without the fixed effect, but with the same random effects structure) and compare it to the full model by means of a likelihood ratio test (chi-squared statistic). This method yields a p-value that is just about non-significant (p=0.06). However, independent of the exact p-value, we would argue that the interpretation of the results given in the discussion section of the paper is still valid. Namely, that the negative effect of L2 speaker percentages on LDTs does extrapolate to most, but not all, language families and areas in our sample.