Abstract
Different regions of the brain have to interact to perform language processing. Such neural integration processes can be studied by measuring synchronization of oscillations. Coherence is the best known algorithm to study synchronization between two sites of the cortex with regard to selected oscillation frequencies. Phase coherence quantifies phase synchronization, in particular. In addition, cross phase (or simply phase) gives information about the direction and speed of the spread of oscillations. We use a new method to study short-time phases between different sites of the cortex in order to explore transient neural networks during word processing. Particularly, processing of abstract and concrete nouns was investigated by dynamic phase analysis of the alpha1 frequency band (8-10 Hz). Abstract and concrete nouns turned out to induce different dynamic networks of information transfer. Whereas processing of concrete nouns excites a widespread network for about 800 msec, the interhemispheric interactions during abstract noun processing are restricted to 300-500 msec after word presentation. Further, the direction of cerebral information transfer differs for the two word categories. Additionally, it could be shown that the propagation speed of information is slower for concrete nouns than for abstract nouns.