Maybe you should just crop the same number of words from the start of each article, and take the same number of articles from each year by random sampling. That

Maybe you should just crop the same number of words from the start of each article, and take the same number of articles from each year by random sampling. That would make things easier to compare.