Speech perception theories differ on whether the process of sub-lexical speech perception is autonomous or is influenced by contextual knowledge such as semantic or lexical context. A better understanding of sub-lexical speech perception would be useful for building cognitive models of speech perception and clinically, in understanding speech perception deficits after neurological injury. We employ the N1 EEG-ERP waveform to measure the sub-lexical processes involved in stop-consonant voicing categorization and test whether voicing encoding is altered by lexical status (e.g. /d/-/t/ in dape-tape versus date-tate). This project improves on previous work investigating sub-lexical perception (e.g. Ganong, 1980) by including an online measure of sub-lexical encoding. We demonstrate effects of lexical context on N1 encoding, indicating interaction of lexical and sub-lexical information. This finding supports online feedback accounts of speech perception. Practically, it also confirms the N1-EEG is a useful candidate for measuring deficits in speech perception in neurologically damaged populations.