This paper is an attempt to sketch the outlines of an approach to ‘negative-raising’ and related problems which differs both from the traditional syntactic account in terms of a transformational rule, and from the traditional pragmatic account in terms of a pragmatic inference schema. The proposal is cast in terms of the theories of Functional Grammar and Functional Procedural Grammar, and the discussion is not only intended as a contribution to the solution of the negative-raising problem per se, but is also meant as an attempt to show that solving this problem requires levels of representation in a grammar ‘deeper’ than those available in the current model of Functional Grammar. This means that developing a more encompassing model such as Functional Procedural Grammar is not merely a matter of luxury.