The literature on the electrophysiology of language comprehension has recently seen a very prominent discussion of "semantic P600" effects, which have been observed, for example, in sentences involving an implausible thematic role assignment to an argument that would be a highly plausible filler for a different thematic role of the same verb. These findings have sparked a discussion about underlying properties of the language comprehension architecture, as they have generally been viewed as a challenge to established models of language processing and specifically to the notion that syntax precedes semantics in the comprehension process. In this paper, we review the literature on semantic P600 effects and discuss a number of challenges--both conceptual and empirical--to existing approaches in this domain. We then provide a new perspective on these effects by showing how they can be derived within an independently motivated, hierarchically organised neurocognitive model of language comprehension in which syntactic structuring precedes argument interpretation (the extended Argument Dependency Model, eADM; Bornkessel and Schlesewsky, 2006). In addition to straightforwardly deriving the phenomenon of a "semantic P600," the basic architectural properties of the eADM account for existing empirical puzzles within the semantic P600 literature.