If you want to normalize the tokens on-the-fly you can actually write a small java snippet right into the grammar.
An example of this would be the following which removes both the single quote characters from a string inline. The syntax should be in one of your books for this .
setText(getText().substring(1, getText().length() - 1).replace("\\\\", "\\").replace("\\'", "'"));
: '\'' STRINGCHAR* '\''
Edit: Ah, because it's a fragment this is all moot.