Describing the neural network that subserves sentence processing, and what differentiates this network from single word processing, is still a major challenge in brain research. During my PhD project, I tested the hypothesis that recurrent interactions between frontal and temporal areas are necessary for sentence processing but not for single word processing. I addressed this issue by interfering with neural feedback by means of visual masking and comparing the masked and unmasked processing of sentences and single words. My results showed that recurrent interactions are required for sentence processing but not for single word processing. This provides an important contribution to ongoing debates about the specific roles of different brain regions in a distributed language network.