The visual perception of words is known to activate the auditory representation of their spoken forms automatically. We examined the neural mechanism for this phonological activation using transcranial magnetic stimulation (TMS) with a masked priming paradigm. The stimulation sites (left superior temporal gyrus [L-STG] and inferior parietal lobe [L-IPL]), modality of targets (visual and auditory), and task (pronunciation and lexical decision) were manipulated independently. For both within- and cross-modal conditions, the repetition priming during pronunciation was eliminated when TMS was applied to the L-IPL, but not when applied to the L-STG, whereas the priming during lexical decision was eliminated when the L-STG, but not the L-IPL, was stimulated. The observed double dissociation suggests that the conscious task instruction modulates the stimulus-driven activation of the lateral temporal cortex for lexico-phonological activation and the inferior parietal cortex for spoken word production, and thereby engages a different neural network for generating the appropriate behavioral response.

The phonemic structure of the maternal language determines the way of perceiving speech signals. A typical example is that native Japanese listeners map two English phonemes, /r/ and /l/, onto the same /R/. This perceptual assimilation of speech sounds has been associated with the left and/or right posterior perisylvian region, but the precise functional anatomy is unknown. Using functional magnetic resonance imaging and a repetition priming paradigm, we identified three subregions in the left temporal cortex: an anterior division sensitive to language-specific phonological knowledge, and a midlateral and a posterior division related to other vocal stimuli features. Dynamic causal modeling supports the scheme by which the anterior pathway processes perceptual assimilation; the posterior pathway processes lexico-semantic information.

No abstract