If so it would likely be a function of amount of training data in that language.