Based on clinical and functional imaging data, the left anterior insula has been assumed to support prearticulatory functions of speech motor control such as the ''programming'' of vocal tract gestures. In order to further elucidate this model, a recent functional magnetic resonance imaging (fMRI) study of our group (Riecker, Ackermann, Wildgruber, Dogil, & Grodd, 2000) investigated both overt (aloud) and covert (silent) production of highly overlearned word strings (''automatic speech''), based on the suggestion that ''inner speech'' might provide a ''window'' into preparatory motor activities (Jeannerod, 1994). As a control condition, subjects were asked to reproduce a nonlyrical tune. In contrast to hemodynamic responses within motor cortex and cerebellum, activation of the intrasylvian cortex turned out to be bound to overt task performance. Rather than prearticulatory processes, these findings suggest the left insula to contribute to the actual coordination of the up to 100 muscles engaged in articulation and phonation. Conceivably, the association of speech production with intrasylvian cortex might have evolved within the framework of phylogenetically older connections between the insula and limbic structures, on the one hand, and nonspeech functions of the upper midline musculature such as swallowing, on the other. Whereas (overt) speech tasks predominantly elicit activation within left anterior insula, reproduction of a nonlyrical tune yielded an opposite response pattern. Conceivably, the opposite distributional pattern of speaking and singing at the level of intrasylvian cortex reflects operation of the two hemispheres across different time domains (''double filtering by frequency theory'': left HEMISPHERE=segmental information, right HEMISPHERE=intonation contours of verbal utterances and musical melodies; Ivry & Robertson, 1998). In line with this suggestion, a further study of our group (Ackermann et al., 2001) provided first evidence that differential hemispheric filtering might be bound to insular cortex.