Precede-and-command revisited revisited

Basing his argumentation on an analysis of condition C effects, proposes to replace the familiar notion of c-command underlying dependency relations with a precede-and-command condition, which defines dependency relations as precedence relations within a local domain (phase). In this reply I argue that condition C effects cannot be used to show the relevance of phases for the definition of syntactic dependency, and I question the conceptual necessity of the notion phase as currently defined.