Further issues

In the final version of the paper we will discuss how we can implement fine-grained control over the application of lexical rules. We allow regular expressions over lexical rule names. Implementing such a control extension is trivial. On the other hand, such a control extension seems to be vital for practical applications of the methods described, in order to limit spurious ambiguities, ad-hoc features, and the chances that the lexical rule component is not `well-behaved', leading to non-termination (given the result in [3] it is not surprising that such cases exist).