Child language and language change : A conjecture and some refutations