In Chapter 7, we warn readers against the use of Google Books as a source on the historical use of words: Google Books does not contain everything that has ever been printed; it is biased in many directions (e.g. towards the English language), most of which are unknown. There are, however, corpora, generally built by linguists, that allow you to know inside which group of texts exactly you plot the use of this or that word, or you investigate which words were often used together. You can also choose to restrict your analysis to a part of the corpus (by author, say, or period). Those are also texts that have been checked and indexed by humans: they do not include errors due to optical character recognition and in most cases, they allow an analysis of grammatical categories. The drawback is, of course, that you can only explore what is in the corpus, i.e., often, something like the canon. But anyway, you cannot seriously count if you have not defined a perimeter of what you include and do not include. Continue reading “Analyzing texts – structured historical corpora”