Four experiments were conducted to evaluate the proposition that although prior exposure to a printed word facilitates identification of a corresponding picture, exposure to a picture does not facilitate subsequent word recognition (Durso & Johnson, 1979).Word identification was used, rather than naming latency, in order to avoid the range limitations in adult reading data. Word identification was facilitated by intermodal priming (prior exposure to a corresponding picture), although to a lesser extent than by intramodal (i.e., word-word) priming; the magnitude of intermodal priming was insensitive to strategy; and, as with priming from spoken to printed language, the major impact of word frequency occurred under intermodal, as distinct from intramodal, conditions. Following Scarborough, Gerard, and Cortese (1979), a fifth experiment compared word identification and episodic recognition. Intramodal performance was superior in word identification, whereas intermodal (i.e. picture-word) performance was superior in episodic recognition, a reversal which suggests that episodic recognition involves access to a distinct memory trace.

Two experiments were conducted to evaluate the role of context in word identification and episodic recognition. Each experiment involved a study phase and a test phase. During the study phase, subjects were presented with a mixture of scrambled, semantically anomalous, and semantically integrated sentences. During the test phase, subjects were presented with individual printed words and were required to answer either an episodic-recognition test (Experiment 1) or a wordidentification test (Experiment 2). The results of the word-identification experiment showed that repetition printing, although present under each condition, was insensitive to the contextual manipulation. Performance under episodic-recognition conditions, however, was sensitive to contextual variation; the actual scores ranged from 43% correct for words presented in scrambled sentences to 60% correct for words presented in semantically integrated sentences. The results constitute a single dissociation, and they are consistent with the proposition that performance in episodic recognition is sensitive to a form of information that (1) does not influence word identification, and (2) is influenced by variation in contextual factors during encoding.An important and contentious issue in recent research concerns the extent to which performance in wordidentification and episodic-recognition tasks can be explained by reference to a common memory system. In operational terms, the tasks are quite distinct. Subjects in word-identification tasks are typically presented with a degraded stimulus and instructed to name it. They are not required to judge whether or not the stimulus is old or new with reference to an earlier study session, and memory is therefore tested implicitly rather than explicitly. Subjects in episodic-recognition tasks, by contrast, are given an explicit memory test in which they are instructed to indicate whether or not each test word was present in an earlier study session.The issue is important because evidence that performance in each task is mediated by a common system provides a formal link between perception and memory. According to Jacoby and Brooks (1984), for example, there is no clear distinction between the processes involved in perceptual identification and episodic recognition. They claimed that performance in each task depends on access to information about earlier occurrences of the same stimulus or similar stimuli. There is provision in their account for task-specific effects-where this involves dependence on different types of information-but their basic premise is clear. Performance in each case follows access to memory about specific instances or episodes. The alternative account, that word identification and episodic recognition involve different systems, is widely reflected in the literature, if only in the sense that a wide range