In a previous study, Ehri and Wilce (1985) found that prereaders who had not mastered letters were better at forming logographic access routes than letter-sound access routes into memory to read words by sight. Of interest here was whether prereaders who knew all their letters would reveal the same pattern. Preschoolers and kindergartners who could read few if any words out of context but who knew letters were taught to read either six simplified phonetic spellings whose letters corresponded to sounds (e.g., JRF for "giraffe"), or six logographic spellings whose letters were non-phonetic but were more visually distinctive (e.g., J R F for "candy"). Word reading practice included either naming or counting letters in phonetic spellings or counting letters in visual spellings. Letter naming was expected to draw prereaders' attention to phonetic cues. Letter counting was the control. Results revealed that phonetic spellings were learned better than visual spellings regardless of whether letters were named or counted. We conclude that prereaders become capable of forming letter-sound access routes when they learn letters well enough to take advantage of the phonetic cues they provide.According to various theories, there are at least two processes that readers might use to identify written words: by phonological recoding, and by sight. Phonological recoding consists of applying general knowledge of the orthographic system to transform spellings into pronunciations with recognizable meanings (Baron, 1977; Ehri, in press-a; in press-b;Marsh, Friedman, Welch, & Desberg, 1981;Venezky, 1970). Meanings are recognized from pronunciations of words, not from print cues. Sight word reading involves accessing information that was stored in lexical memory about words when they were read previously. Although sight words 149