<p>The visual word form area (VWFA) has been proposed to function as a multiplex hub integrating both attentional and linguistic processing, yet how its functionally distinct subregions dynamically reconfigure their network connectivity across cognitive contexts remains unclear. This study examined resting-state and task-evoked functional connectivity of two VWFA subregions—posterior VWFA-1 and anterior VWFA-2—with predefined attention and language network regions of interest in 20 healthy adults performing a visuospatial flanker task and a Korean lexical decision task. We hypothesized that the posterior-to-anterior anatomical gradient corresponds to an attention-to-language functional role differentiation, observable both in intrinsic connectivity and in task-dependent reconfiguration. At rest, the VWFA exhibited a posterior-to-anterior intrinsic gradient, with VWFA-1 preferentially coupled to the attention network and VWFA-2 to language regions. Across both tasks, task context differentially shifted attention-network connectivity between VWFA-1 and VWFA-2, broadly following the intrinsic gradient. The alignment was only partial, however, as during lexical decision it was the posterior VWFA-1 rather than the anterior VWFA-2 that increased its coupling with the language network. Behaviorally, attention connectivity predicted accuracy across tasks from both subregions equivalently rather than from the gradient itself, suggesting a shared attention-based mechanism rather than gradient-specific contributions. Together, these findings provide preliminary support for the view that intrinsic VWFA organization shapes task-specific routing of attention network interactions, while behavioral performance converges on a shared attention-based mechanism.</p>