Mental Models as Contexts for Interpreting Texts: Implications from Studies of Anaphora