How to say <i>no</i> and <i>don’t</i>
After reviewing some earlier analyses of negative imperatives, we argue that there is a correlation between the (non-)availability of negative imperatives in a language and the merger of the anaphoric negator ( <i>no </i>) and sentence negator ( <i>not </i>). This shows up not only as lexical identity of these two types of negation (e.g. Portuguese <i>n ã o </i>‘no’/’not’), but also in the syntactic merger of the corresponding functional projections. The interaction between this merger and the imperative will be studied in the light of Hoekstra and Jordens’s (1994) data from Dutch child language, in which there is a lexical distinction between boulemaeic negation ( <i>nee </i>) and epistemic negation ( <i>niet </i>). In the adult language, the corresponding functional projections have grammaticalised to an A-bar and A-projection, respectively. These two projections undergo complete merger in languages where the negators coincide lexically. In this way, the absence of the negative imperative can be viewed as a consequence of relativised minimality, an insight due to Rivero and Terzi (1995) , which will be shown to crucially involve the A-bar properties of the imperative operator (with V-movement skipping negation on its way to COMP) and the A/Abar properties of negation.