Functional imaging using PET and fMRI typically focuses on the functional neuroanatomy of language. Due to the inherent temporal limitations of hemodynamic techniques, however, the connection to cognitive models of representation and processing is rather indirect. Magnetoencephalography (MEG) is a noninvasive technique that provides the temporal resolution of an electrophysiological method while also providing useful spatial information. The data generated in MEG studies connect the excellent spatial resolution of hemodynamic techniques with the temporal resolution required to investigate online processing. I will provide an MEG primer and discuss MEG research on the neural basis of speech perception and language comprehension, showing the role temporal information (of various types) plays.