From orthography to phonetics: ERP measures of grapheme-to-phoneme conversion mechanisms in reading.

Neuroimaging has provided evidence that the first stages of visual word recognition activate a visual word-form center localized in the left extrastriate cortex (fusiform gyrus). Accordingly, neurological cases of patients suffering from pure alexia reported the left posterior occipital lobe as the possible locus of orthographic analysis. There is less agreement in the literature about which brain structures are involved in the subsequent stages of word processing and, in particular, their time course of activation. Functional magnetic resonance imaging and magnetic source imaging studies recently reported data that could indicate a dual route model of reading. These findings are particularly relevant to studies on the functional deficits associated with phonological and surface dyslexia. There is evidence for the existence of two different brain mechanisms supporting phonological processing in visual word recognition: one mechanism subserving "assembled phonology" for reading letter strings and another one subserving "addressed phonology" for reading meaningful words. However, available knowledge on the time course and neural locus of grapheme-to-phoneme conversion mechanisms in reading is still inadequate. In this study, we compared processing of meaningful and meaningless Italian words in a task requiring a phonemic/phonetic decision task. Stimuli were 1152 different orthographic stimuli presented in the central visual field. Half the stimuli were Italian words (with a high or low frequency of occurrence), the other half were meaningless strings of letters (legal pseudowords and letter strings). Event-related potentials were recorded from 28 scalp sites in 10 Italian university students. The task consisted of deciding about the presence/absence of a given "phone" in the hypothetical enunciation of word read: for example, "Is there a /k/ in cheese?". Results showed that lexical frequency and orthographical regularity affected linguistic processing within 150 msec poststimulus. Indeed, the amplitude of a centroparietal P150 varied as a function of stimulus type, being larger in response to high-frequency words than to low-frequency ones and to words and pseudowords than to letter strings. This component might index visual categorization processes and recognition of familiar objects, being highly sensitive to orthographic regularity and "ill-formedness" of words. The amplitude of the P150 was the same in response to well-formed meaningless and to meaningful words, when these latter had a low lexical frequency. This might indicate that highly familiar words are recognized as meaningful unitary visual objects at very early stages of processing, through a visual route to an orthographic input lexicon. Moreover, the amplitude of the negativity recorded between 250 and 350 msec showed an anteroposterior topographic dissociation for access to the phonemic representation of well- or ill-formed strings of characters. Brain responses were larger over the left occipito-temporal regions during reading of words and pseudowords and over the left frontal regions during reading of letter strings.