Hypothetically, words can be named by spelling-sound translation rules or by looking up a phonological code in a lexicon. Following J. Baron and C. Strawson (1976), naming performance was measured as a function of skill with each route, using sets of stimuli varying in reliance on either route. ''Phoenicians'' were defined to be better with rules than with look-up; ''Chinese'' were better at look-up than with rules. As predicted by J. Baron and C. Strawson, Phoenicians named low-frequency regular words and nonwords faster than Chinese. Contrary to predictions, Phoenicians were also faster at naming irregular words of various frequencies. Implications of these results for various dual-route models versus single-route models are discussed.