Does word naming involve grapheme-to-phoneme translation? Evidence from Hebrew

Are grapheme-to-phoneme rules automatically applied when a word pattern is encoded? This question was examined using Hebrew words. In written Hebrew vocalic information is sometimes transmitted by vowel signs written below the letters, but most often is omitted altogether. Hebrew speakers were asked to name Hebrew words by their letters only, disregarding vowel signs. Naming was equally fast for words with no vowel signs, with correct vowel signs, and with incorrect ones that nevertheless preserved word sound. Naming was slower when the vowel signs were incompatible with the word sound. That this is mediated by visual mechanisms was ruled out by a further experiment. It is concluded that grapheme-to-phoneme translation is a natural response to written words, at least when naming is required.