Three competing models of word recognition processes currently receive psycholinguistic research support: direct encoding (i.e., word recognition proceeds from graphemes); indirect encoding (i.e., word recognition proceeds from a phonemic translation of graphemes); and dual access (i.e., word recognition can proceed from either graphemes or phonemes). Resolution of the conflict between these models has been hampered by methodological problems common to research techniques in this area. The two experiments reported in this study analyzes the competing models via the SLIP technique, which avoids those methodological problems. Support is yielded for the direct encoding model. Implications are discussed.