Precede-and-Command Revisited

The relation of c-command (Reinhart 1976, 1983) is widely believed to be the fundamental relation in syntax, underlying such diverse phenomena as coreference (the binding principles), scope and variable binding, syntactic movement, and so on. Precedence is generally held to be irrelevant. This article argues that this view is mistaken. Syntax does not involve c-command at all, but rather a much coarser notion of command, phase-command, where only phasal nodes matter, not every node in the tree. Precedence also plays an important role. The article argues this point in detail for the binding principles, and shows that the relation that is required is precede-and-command (Langacker 1969, Jackendoff 1972, Lasnik 1976), where command is phase-command. It revisits Reinhart's arguments for c-command and against precedence, and shows that those arguments do not go through. Finally, precede-and-command does not need to be stipulated, but follows from a view of grammar and processing where sentences are built in a left-to-right fashion.