According to the Self-Regulated Language Processing model (Stine-Morrow, Miller, & Hertzog, 2006), text understanding depends on the construction of a multifaceted mental representation (dark gray box) of the surface form (words), semantic content or ideas expressed by the text (textbase), and the situation described by the text (e.g., Kintsch, 1998). Each of these three levels of representation can be constructed more or less completely depending on criteria (reference values) that are influenced by the cognitive and socioemotional goals of the reader (top). At each level, attention to relevant processes is regulated by a negative feedback loop, in which the perceived current representation (i.e., perceived word comprehension, perceived textbase comprehen-sion, perceived situational coherence) is compared (C) to the reference value; when the representation is not ‘‘good enough’’ (Christianson et al., 2001), the processor at that level demands more resources, until equilibrium is reached (i.e., perceived comprehension satisfies reference value). In our research, allocation of effort to specific processes at each level is often operationalized as the relative change in reading time as a function of text variables assumed to engage those processes.