In Experiments 1 and 2, we investigated long-term repetition priming effects in Serbian under crossalphabet and cross-modal conditions. In both experiments, results followed the same pattern: significant priming in all conditions and no significant reduction in priming in the cross-modal as opposed to the cross-alphabet condition. These results are different from those obtained in English (Experiment 3), in which a modality shift led to a reduction in priming. The findings are discussed within a theoretical framework, in which long-term priming is a by-product of learning within the language system. A full list of word stimuli for all three experiments presented in this article can be found at www.psychonomic.org/archive.
Priming occurs when previous exposure to a stimulus modifies the subsequent response to a related stimulus. The duration of priming depends on the priming technique used, from milliseconds (masked priming) to minutes or hours (long-term priming). In the long-term priming paradigm, exposure to items during a study phase affects performance in the test phase. However, the test phase does not make direct reference to the previous study episode, and participants are often unaware of the connection between the two phases of the experiment.
A number of studies have obtained reliable and comparable long-term priming effects within the visual modality when the study and test words are presented in different formats, such as lowercase and uppercase in English (e.g., Bowers, 1996; but see Tenpenny, 1995), Hindi and Urdu scripts in Hindustani (Brown, Sharma, & Kirsner, 1984), hiragana and katakana in Japanese (Bowers & Michita, 1998), and Cyrillic and Roman scripts in Serbian (Feldman & Moskovljevic, 1987). By contrast, changes in the modality of the study and test item (e.g., auditory presentation at study and visual at test) reduces priming, typically, by 50% or more (e.g., Bowers, 2000b; Rajaram & Roediger, 1993), suggesting that a significant component of priming is visual. This pattern of results has led a numger of authors to argue that long-term priming is mediated in part by preexisting and abstract orthographic codes, and that priming reflects structural changes that affect the later processing of the repeated items (e.g., Morton, 1979). Bowers, Damian, and Havelka (2002) simulated various aspects of long-term priming within the orthographic component of Seidenberg and McClelland's (1989) distributed model of word identification, lending further support to the view that long-term priming is a by-product of learning within the word recognition system.
Although priming is reduced following modality shifts, it is rarely eliminated, suggesting a contribution from nonorthographic codes. The phonological basis of crossmodal priming is suggested by a number of findings, including the robust priming obtained between homophones (Rueckl & Mathew, 1999) and imaging studies that reveal that cross-modal priming is associated with activation in brain areas involved in processing of phonological information (e.g., Badgaiyan, Schacter, & Alpert, 1999; Schacter, Badgaiyan, & Alpert, 1999).
In summary, priming may be viewed as a by-product of learning within the language system, or, more specifically, as a result of structural changes within orthographic, and to some extent, phonological and even semantic systems (e.g., Becker, Moscovitch, Behrmann, & Joordens, 1997). Because these systems are also involved in visual word recognition, factors that influence word recognition may, hypothetically, also influence long-term priming, and vice versa.
One issue in visual word recognition research that has long been a subject of debate is the relative importance of visual versus phonological processing. According to one of the predominant visual word recognition models, the dual-route cascaded (DRC) model (e.g., Coltheart, Rastle, Perry, Langdon, & Ziegler, 2001), there are two possible mechanisms for deriving a phonological representation from print. …