Two lexical decision experiments examined the joint effects of stimulus quality, semantic context, and cue-target associative strength when all factors were intermixed in a block of trials. Both experiments found a three-way interaction. Semantic context and stimulus quality interacted when associative strength between cue-target pairs was strong, and the interaction was eliminated when the strength was weak. These results support a role for a "local" mechanism that relies on "trial specific" information, in addition to a mechanism that makes use of "global" information available "across" a block of trials. The absence of an interaction between the joint effects of semantic context and stimulus quality is attributed to blocking the feedback from the semantic system to the orthographic system, functionally separating the orthographic and semantic modules. (Contains 5 footnotes, 2 tables, and 3 figures.)