Context in semantic information retrieval

Abstract

Retrieval processes in word recognition were studied by systematically varying context conditions during testing. Ss received a single presentation of 30 triplets of semantically related words followed by a recognition test, with one word from each triplet being tested either in its original context, with the context deleted, in a new context of related words which primed a different meaning, or in a context of new unrelated words. Recognition performance was highest when the context remained unchanged and lowest when the target item was embedded in a new and unrelated context. No difference was found between the unchanged context and the semantically related context condition when d’ values were used as a measure of recognition performance. © 1971, Psychonomic Journals, Inc.. All rights reserved.