One of the key controversies in current syntactic and morphological analyses revolves around the basic units of derivation, including derivations that feed word formation. Some analyses assume that the basic unit is the root which maybe devoid of categorical features. However, most of the debate has been driven by theoretical and conceptual factors. Semitic languages provide a rich empirical ground for this debate because they have derivations that involve the consonantal root and derivations that access the entire stem (both consonantal and vocalic melodies). In this talk, I provide empirical arguments that challenge the role of the root in syntactic and post-syntactic derivations but that does give it a prominent role in lexical derivations. One case study will be number marking. It will be shown that number marking which can realize both verbal plurality (at the level of event structure) and number agreement with subjects and objects differ in how they are derived (and which prosodic units they can access) depending on their syntactic and semantic properties. Important questions about the interface between morphology, the lexicon, and syntax will be discussed as well as first, second, and heritage language acquisition of the basic units of word formation.