In Optimality Theory it is possible to construct complex constraints from simpler elements. The question arises as to how far we can simplify the content of UG by decomposing complex constraints in this way. In this paper we argue that although the goal of a simple UG without overproliferation of constraints may be valid, it should not come at the expense of correct typological predictions. We show that incorrect cross-linguistic predictions are made if certain constraints are considered to be constructed from primitive elements. These incorrect predictions can be eliminated if we place certain restrictions on constraint conjunction and therefore must assume that the constraints in question are part of UG in their complex form. We will discuss the issues of the phonetic motivation for complex constraints and of distinguishing the universal from the language-specific.
ASJC Scopus subject areas