Inner speech, or internal utterance, is fundamental to cognitive activities like language comprehension, planning, and control. However, its psychophysiological mechanisms, particularly the link between verbal and sensorimotor processes, remain poorly understood. Embodied cognition theories suggest that understanding action words involves the reactivation of sensorimotor experiences tied to those words. This study investigates neural differences in the perception, internal utterance, and mental imagery of action words through eventrelated potentials (ERPs). Three experimental series were conducted: (1) word perception and pronunciation, (2) word perception and imagery representation, and (3) sequential perception and utterance of action words. Eight audio stimuli representing distinct actions (e.g., run, stand, up, left) were used. EEG recordings were obtained from 130 participants using a 19-channel Neuro-KM electroencephalograph, with brain activity analyzed via the ‘BrainSys’ software and the ‘Virtual Implanted Electrode’ method. During utterance and representation, similar ERP patterns were observed, differing in amplitude at N140 and N400 latencies across multiple leads (e.g., Cz, Fz, Pz). Significant neural differences were observed, involving regions such as the brainstem, midbrain, right amygdala, cingulate gyrus (BA24), Broca’s area (BA44), parahippocampal gyri, thalamus, and hippocampus. Specific activations during utterance included the left caudate nucleus and right orbitofrontal cortex (BA47).The results highlight distinct neural mechanisms underlying perception, internal utterance, and imagery of action words, supporting embodied cognition theories and offering insights into the verbal-sensorimotor connection during cognitive tasks.