Though there is robust morphological evidence for small units of word formation such as roots to account for patterns of lexical relatedness, the syntactic significance of roots and their role in capturing syntactic generalizations are not as empirically well grounded. For example, in non-concatenative morphological systems, such as Arabic, the root has figured prominently as a primitive unit of derivation that provides a mechanism to account for lexical relations that resist accounts in terms of the familiar patterns of morphological concatenation through prefixation or suffixation to a base. In syntax, recent accounts have invoked the notion of root, mostly on conceptual grounds, to account, for example, for the constructional meaning of sentences and the interplay between functional categories and lexical categories. In this talk, I take up the question of the role of the root in syntax and whether there are empirically principled restrictions on access to it. I will present empirical arguments to show that such principled restrictions can be motivated and that they are consistent with recent developments in syntactic theory, and particularly the relation between the syntactic component and the morpho-phonological interface.