One key issue for any computational model of visual-word recognition is the choice of an input coding scheme for assigning letter position. Recent research has shown that pseudowords created by transposing two letters are very effective at activating the lexical representation of their base words (e.g., relovution activates REVOLUTION). We report a masked priming lexical decision experiment in which the pseudoword primes were created by transposing/replacing two consonants or two vowels while event-related potentials were recorded. The results showed a modulation of the amplitude at an early window (150-250 ms) and at the N400 component for vowels but not for consonant transpositions. In addition, the peak latencies were faster for transposed than replaced consonants. These results suggest that consonants and vowels play a different role during the process of visual word recognition. We examine the implications for the choice of an input coding scheme in models of visual-word recognition.