The authors investigated how 2-digit Arabic numerals are named by looking at the effects of masked primes on the naming latencies. Target numerals were named faster when prime and target shared a digit at the same position (e.g., the target 28 primed by 18 and 21). In contrast, naming latencies were slower when prime and target shared 1 or 2 digits at noncorresponding places (e.g., the target 28 primed by 82, 86, or 72). Subsequent experiments showed that these priming effects were situated at the level of the verbal production of the Arabic numerals. The data point to a nonsemantically mediated route from visual input to verbal output in the naming of 2-digit Arabic numerals.Keywords: naming, two-digit Arabic numerals, masked priming Recent years have witnessed a dramatic increase in the number of studies examining number processing. Unfortunately, for some topics this increase has not been accompanied by a growing consensus about the processes involved. In the current article, we address what arguably should be one of the simplest questions in numerical cognition: How are Arabic numerals named?Four different models have been proposed by various authors, depending on the answers to the following two questions: Is semantic mediation required for the naming of Arabic numerals? Are two-digit numbers processed holistically or through a process of syntactic decomposition into 10s and units?The issue of semantic mediation addresses the question of whether Arabic numerals can be named without first activating their meaning. For visually presented words within an alphabetic language, there is plenty of evidence that such nonsemantically mediated naming is possible. In fact, many existing models of written word naming do not even include an implemented semantic route (e.g., Coltheart, Rastle, Perry, Langdon, & Ziegler, 2001;Plaut, McClelland, Seidenberg, & Patterson, 1996). Such a route is assumed to exist but has not yet been properly examined because it does not seem necessary to account for word-naming latencies. These can be explained by a direct letter-sound conversion route (assembled phonology) or a route in which the visual input activates a presentation in the input lexicon, which, in turn, activates a representation in the phonological output lexicon (addressed phonology).In contrast, most researchers assume that objects and pictures of objects cannot be named without first activating their meaning (e.g., Levelt, Roelofs, & Meyer, 1999). Damian, Vigliocco, and Levelt (2001) provided particularly convincing evidence for the difference between word reading and picture naming. They asked participants to repeatedly name pictures in a semantically homogeneous block (e.g., cat, goat, rat, beaver, tiger, swan) or in a semantically heterogeneous block (e.g., cat, hand, ferry, skirt, broom, leek). Damian et al. observed that participants needed more time to name the pictures in the homogeneous condition than in the heterogeneous condition. They attributed this interference effect to an increased competition in the ret...