In the better-developed sciences it is the departures from symmetry rather than the symmetries that are typically taken to be in need of explanation. Mirror theory is an attempt to look at some of the central properties of syntactic representations in this spirit. The core hypothesis of this theory is that in syntactic representations complementation expresses morphological structure: X is the complement of Y only if Y-X form a morphological unit'a word. A second central assumption is the elimination of phrasal projection: a head X in a syntactic tree should be taken to ambiguously represent both the zero-level head(s) and its (their) associated phrasal node(s).
Perfect Syntax dispenses with the idea of externally forced imperfections in syntax. This article presents a system of principles relating (L) LF representations and lexical items that aims to be compatible with this assumption. The core of this theory is that phrase structures are viewed as projection lines (lexical items and their projections) linked by an Insert relation. This explains uniqueness and locality of projection, the fact that phrases and nonphrasal elements can immediately dominate each other only when they are part of the same projection line, and most effects of the “target projects” requirement. I attribute a residue to the Generalized Projection Principle, for which I also provide an explanation. In addition, I explore various consequences of the present approach for the Move/Chain relation.