Mechanisms underlying data driven retrieval processes

Mary Susan Weldon, Purdue University


Three experiments examined the nature of retrieval processes on implicit, data-driven tests. Data-driven retrieval tests are defined as those that benefit more from sensory encoding (e.g. reading the target word in isolation) than conceptual encoding (e.g. elaborating upon the target's meaning). Implicit retrieval tasks are defined as those that do not require explicit recollection of the encoded material. Examples of implicit, data-driven tasks include perceptual identification (identifying rapidly presented targets) and word fragment completion (supplying missing letters to form a compete word, e.g. lpat as elephant). Three experiments were designed to assess the roles of sensory, lexical, and conceptual processes in priming on the word fragment completion and perceptual identification tasks. Experiment 1 compared priming from four study conditions involving different combinations of encoding processes: Read (shared sensory, lexical, and conceptual properties with the test words); Auditory or Generate (shared lexical and conceptual properties with the test words); and Pictures (shared conceptual properties, but were assumed to be less likely than the word conditions to engage lexical access). Reading produced the most priming on both tests. On the word fragment completion test, the Auditory and Generate conditions produced more priming than Pictures, but this trend was not significant on the perceptual identification test. Experiment 2 examined whether sensory processing alone is sufficient to produce priming. Two groups of subjects received identical sensory input (pronouncable anagrams with interchanged vowels, e.g. tripocs). One group was instructed to interchange the vowels, which produced significant priming. The other group read the anagrams as nonwords, and no priming was obtained. These results indicate that sensory processing alone is not sufficient to produce priming; the study stimulus must access a meaningful lexical representation. Experiment 3 attempted to discriminate the effects of lexical and conceptual processes. Subjects studied target words (e.g. black, scotch) in either "same-meaning" compounds in which targets retained their individual meanings (e.g. blackbird, scotch bottle), or "different-meaning" compounds in which targets lost much of their individual meanings (e.g. blackmail, scotch tape). Encoded meanings did not affect priming on the perceptual identification task, but same-meaning compounds produced somewhat more priming, on the word fragment completion task. (Abstract shortened with permission of author.)




Roediger, Purdue University.

Subject Area


Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server