Three experiments in which subjects searched for the letter e in printed text were conducted to examine the effects of phonetic factors in silent reading. In Experiment 1, subjects made more errors on silent es than on pronounced es, but silent es always occurred at the ends of words, whereas pronounced es occurred in the middle of words. In Experiment 2, all instances of the letter e occurred in the penultimate location in the words. and no effects of letter voicing were obtained. In Experiment 3, subjects made more errors on es in unstressed syllables than on es in stressed syllables in three-syllable words. However, this effect occurred only for es in the second and third syllables and only for the more common words. All three experiments yielded large effects of word frequency, which were reduced in passages printed in alternating typecase. It was concluded that letter detection is affected by syllable stress but not by letter voicing and that the stress effect depends on whether the subject is able to form reading units at the syllable level.There is much evidence that phonetic recoding of text occurs in the course of silent reading. One of the most influential studies (Corcoran, 1966) demonstrated that subjects searching for instances of the letter e in printed text made more errors on words in which e was silent (as in the word time) than on words in which it was pronounced (as in the word well). The common interpretation of this result, also observed by other investigators (e.g., Chen, 1976;Coltheart, Hull, & Slater, 1975;Locke, 1978;Mohan, 1978), is that subjects silently reading paragraphs of text scan the acoustic image of a word along with the visual stimulus. However, in normal English prose of the type used by Corcoran,