Could this be normalised against the baseline distribution of languages for the respective platforms / software categories to see if there are any notable deviations?