This paper reviews research relevant to the question of whether words are identified through the use of abstract lexical representations, specific episodic representations, or both. Several lines of evidence indicate that specific episodes participate in word identification. First, pure abstractionist theories can explain short-term but not long-term repetition priming. Second, long-term repetition priming is sensitive to changes in surface features or episodic context between presentations of a word. Finally,long-term priming for pseudowords is also difficult for pure abstractionist theories to explain. Alternative approaches to word identification are discussed, including both pure episodic theories and theories in which both episodes and abstract representations playa role.Researchers have repeatedly demonstrated that word identification is facilitated on the second presentation of a word. This phenomenon, known as repetition priming (or savings), occurs in a variety of tasks believed to involve word identification processes. I Repetition speeds lexical decisions (e.g., Scarborough, Cortese, & Scarborough, 1977), word naming (e.g., Durso & Johnson, 1979), and the reading of sentences (e.g., Kolers & Ostry, 1974) and longer text (e.g., Kolers, 1976). Repeated words are also more likely to be correctly identified on a perceptual identification test, in which a word is shown for a very brief duration before being masked (e.g., Jacoby & Dallas, 1981).There has been considerable disagreement about the nature of repetition priming in word identification tasks. At the heart of the debate is the question of whether the effect ofrepetition on word identification is mediated by abstract representations, by specific episodic representations, or both. The goal of this paper is to provide a critical review of the evidence relevant to this question.Distinctions between general and specific information have provoked lively discussions throughout the field of cognitive psychology. General and specific knowledge are often assumed to be used for different purposes and to be very different in nature as well. Two types of distinctions between general and specific information have been made. First, general knowledge of facts and skills Preparation of this paper was partially supported by a Summer Research Stipend from Loyola University of Chicago. I thank Kim Kirsner, Betty Ann Levy, 1. W. Whitlow, and an anonymous reviewer for their insightful reviews. I am especially grateful to Jim Neely for numerous valuable comments. I also thank Gordon Logan for helpful e-mail discussions. Finally, thanks to Ed Shoben for comments on an earlier version of this paper. Correspondence should be addressed to P. L. Tenpenny, Department of Psychology, Loyola University of Chicago, 6525 North Sheridan Road, Chicago, IL 60626 (e-mail: ptenpen@luc.edu).has been distinguished from specific knowledge of episodes in which those facts and skills are learned or used. Tulving (1972), for example, proposed that semantic and episodic information are stored ...

No abstract

Three experiments tested the conceptually-driven/data-driven distinction as an account of memory dissociations. Experiment 1 found that cue typicality had opposite effects on semantic cued recall and category verification, yet performance on both of these conceptually-driven tasks was best when the study and test cues matched. In Experiments 2a and 2b. graphemic cued recall and fragment completion (both data-driven) were compared with semantic cued recall. Cued recall performance was best when the test cues resembled the study cues, but fragment completion priming did not vary. Experiment 3 found opposite effects of frequency on fragment completion and graphemic cued recall. Moreover, the similarity of study cues to their targets affected only graphemic cued recall. These results are consistent with the general transfer-appropriate processing framework but not with the conceptually-driven/data-driven distinction.