“Language game” is not a name. It is a picture made to counter the charm of certain other pictures.I think this is right, but there seems to be more than one picture of a language-game. In PI 7 Wittgenstein introduces the term 'language-games' to refer to the "speech-like processes" that constitute exercises one goes through in teaching and learning a language (repeating words after the teacher, pointing to appropriate objects when the teacher says certain words, and so on). He gives a list of examples here, but they all seem to me to be examples of the same thing: games and exercises by means of which one learns a language. Nursery rhymes, for instance (which Baker and Hacker say that Wittgenstein preferred to "games like ring-a-ring-a-roses" as the English translation of Reigenspielen). Then right at the end he adds (my translation):
I shall also call the whole, the language and the activities with which it is interwoven, the "language-game".It isn't clear what he means by "the language" etc. He has already said that he will call a primitive language a language-game, and it isn't really clear what counts as a primitive language. Is "the language" the primitive language, or some other language? The only thing to do is to look and see how he uses the term. In 300, for instance, when he talks about "the language-game with the words "he is in pain"," he seems to mean neither a learning exercise or game nor a whole language in any obvious sense of "whole language." We have to attend, it seems, not to what he says but to what he does. And presumably that's deliberate.
What he certainly does not do is present himself as having unearthed some phenomenon that had previously been overlooked, namely language-games. Instead he has invented a concept as a tool, and what matters is what he does with it. His use of the term develops in the course of the book. So the introduction of the term in PI 7 no more gives us everything we need to grasp its meaning than does the ostensive teaching of a word.