What is the cognitive system's preferred route for deriving phonology from print?

The present study examined the manner in ivliich Turkish readers read aloud words printed in shallow (transparent) Turkish orthography. The first experiment showed that when the set consists of only word stimuli there is a significant word-frequency effect. This indicates that readers had made reference to lexical information for naming. The result of a second experiment, on the other hand, shoived that when an equal number of nonwords are embedded in the stimulus set the word-frequency effect disappears. This indicates that readers had made reference to nonlexical information for naming. These results support the ideas that: (a) the preferred route for naming amongst readers of Turkish script is lexical, regardless of the orthographic-phonological transparency; and (b) the naming process is indeed a flexible one in which task demands may affect the nature of the route used for naming. The above findings are discussed in relation to research on shallow and deep (nonshallow) orthographies