In my talk I will present a general cognitive architecture of spoken language processing. This is followed by an account of how this cognitive architecture is instantiated in the human brain. Both the spatial aspects of the networks for language are discussed, as well as the temporal dynamics and the underlying neurophysiology. A distinction is proposed between networks for coding/decoding linguistic information and additional networks for getting from coded meaning to speaker meaning, i.e. for making the inferences that enable the listener to understand the intentions of the speaker. I will argue that the classical Wernicke-Lichtheim-Geschwind model about the brain organization for language is largely wrong. Instead I will defend a multiple networks view on language. Finally I will show why mirror neurons fail to account for human language communication.