Does the mirror system affect the control of speech? This issue was addressed in behavioral and Transcranial Magnetic Stimulation (TMS) experiments. In behavioral experiment 1, participants pronounced the syllable /da/ while observing (1) a hand grasping large and small objects with power and precision grasps, respectively, (2) a foot interacting with large and small objects and (3) differently sized objects presented alone. Voice formant 1 was higher when observing power as compared to precision grasp, whereas it remained unaffected by observation of the different types of foot interaction and objects alone. In TMS experiment 2, we stimulated hand motor cortex, while participants observed the two types of grasp. Motor Evoked Potentials (MEPs) of hand muscles active during the two types of grasp were greater when observing power than precision grasp. In experiments 3-5, TMS was applied to tongue motor cortex of participants silently pronouncing the syllable /da/ and simultaneously observing power and precision grasps, pantomimes of the two types of grasps, and differently sized objects presented alone. Tongue MEPs were greater when observing power than precision grasp either executed or pantomimed. Finally, in TMS experiment 6, the observation of foot interaction with large and small objects did not modulate tongue MEPs. We hypothesized that grasp observation activated motor commands to the mouth as well as to the hand that were congruent with the hand kinematics implemented in the observed type of grasp. The commands to the mouth selectively affected postures of phonation organs and consequently basic features of phonological units. (Contains 10 figures.)