Semantic priming effects in naming Italian and English words were investigated. Experiments 1 and 2 were in Italian. In Experiment 1, the subjects named a target word, which was either associated with or unrelated to a preceding prime. The results showed semantic priming effects. However, in Experiment 2, in which the same materials occurred in a list that also included pseudowords, priming effects were obtained with the lexical decision task, but not with pronunciation. In Experiment 3, the inclusion of pseudowords in the materials prevented priming effects from occurring in Italian, but not in English. Finally, Experiment 4 indicated that, even in Italian, nonlexical reading was abandoned when a few of the to-be-pronounced items required lexical knowledge for correct stress assignment. The findings suggest that reading normally occurs lexically. The characteristics of the various writing systems, however, are relevant in determining the strategies that people may adopt in unusual circumstances.Reading a word is the process by which a string of letters makes contact with the corresponding lexical representation, activating different types of information, including its semantics and phonology. An important issue in the study of reading is the nature of the code that allows this process of matching to occur. To date, many researchers agree that most common words are read on the basis of a visual code that activates the information relative to those words directly via orthography (Coltheart, 1978;Davelaar, Coltheart, Besner, & Jonasson, 1978;Henderson, 1982;Patterson, 1982; Seidenberg, 1985aSeidenberg, , 1985bSeidenberg & McClelland, 1989).Part of the experimental evidence in support of this view comes from studies done with the naming task. It is well known, for instance, that frequent words are pronounced faster than less frequent words (Forster & Chambers, 1973;Frederiksen & Kroll, 1976;Glanzer & Ehrenreich, 1979; Rubenstein, Lewis, & Rubenstein, 1971). This is true not only for regular, but also for irregular words. Various studies have shown, in fact, that regular words are pronounced faster than irregular ones when they are low-frequency words but that the difference becomes smaller or even disappears for high-frequency words (Andrews, 1982;Seidenberg, 1985b; Seidenberg, Waters, Barnes, & Tanenhaus, 1984;Waters & Seidenberg, 1985).A commonly accepted explanation of this phenomenon, based on the distinction between whole-word and assembled phonology, is that the lexicon is involved in the pronunciation of common words, including irregular ones, This research was supported by Grant 8700322 from the Ministero Pubblica Istruzione. Fondi 60%. We would like to thank Derek Besner, Cristina Burani, Cristina Cacciari, Corrado Cavallero, Lucia Colombo. and Donia Scott for their useful suggestions. We are also grateful to three anonymous reviewers who helped us to improve the paper considerably. Requests for reprints should be sent to Patrizia Tabossi, Dipartimento di Psicologia, Viale Berti-Pichat 5.40127 Bologna, I...