All grammars must include some notion of movement accounting for discontinuous constituents which have been recognized since the very first post-Bloomfieldian syntactictians. What triggers movement? In the standard generative theory, movement is inherently associated with interpretation of morphological features: elements would move to cancel features which could not be interpreted by the semantic module. In the alternative approach defended here, instead, movement is just the consequence of the fact that hierarchical structures encoding all grammatical relations must be linearized for purely physical reasons when the sentence is uttered. Some consequences of this alternative proposal are discussed in this paper starting from two simple challenging questions that do not receive a unitary answer in the standard approach.