Three-Dimensional Grammar Mark de Vries 1. Introduction The current model of generative syntax focuses on hypo/hypertactic relations. This is the result of restrictions on phrase structure and transformations, “binary branching” for instance. Arguably, these restrictions establish a fruitful research strategy. However, at present we face the problem that a whole range of paratactic constructions such as coordination and parentheses cannot satisfactorily be dealt with. 1 I think the scope of this problem can hardly be exaggerated. Therefore we need a solution of a fundamental type, which generalizes in some way over all different paratactic constructions. And this solution may be the introduction of a so-called third dimension into the syntax. In essence, it amounts to the addition of the relation behindance, next to dominance and precedence. In this article I intend to outline the research program on three-dimensional grammar. For reasons of space, the discussion can only be suggestive here. In section 2 I will show why paratactic constructions lead to the assumption of behindance. Section 3 discusses some of the implications for syntactic theory. Section 4 is on linearization and the number of degrees of freedom. Section 5 is the conclusion. 2. Parataxis as behindance The prototypical example of parataxis is common coordination. See (1). (1) Alice saw John, Bill and Richard. A Kayne (1994)- or Johannessen (1998)-type analysis of the direct object is drawn in (2), where CoP is a coordination phrase.