JOURNAL or VERBAL LEARNING AND VERBAL BEHAVIOR 12, 324-333 (1973) Semantic versus Acoustic Coding: Retention and Conditions of Organization 1 LARRY L. JACOBY AND PAULA GOOLKASIAN Iowa State University, Ames, Iowa 50010 Incidental learning techniques were employed in two experiments to investigate differ- ences in semantic and acoustic coding with regard to retention and organization. The incidental task required subjects to judge pairs of words as being either related or unrelated on a specified dimension, acoustic or semantic. Intentional learning subjects received the same words but were instructed to remember. Instructions to learn enhanced recall and clustering of acoustically related words but had no affect on the usefulness of semantic relationships. Regardless of learning instructions, semantic relationships were more effective than acoustic ones. Results of Experiment II revealed no effects of acoustic similarity when it was unconfounded with orthographic similarity. An approach that characterizes most con- temporary theories of memory has been that of "divide and (hopefully) conquer." Several theorists have found it useful to divide memory into a number of separate stores, differing in capacity and type of information held (for example, Atkinson & Shiffrin, 1968; Norman & Rumelhart, 1970). A second type of division is concerned with the nature of the memory trace. It appears that the encoded version of a word is composed of several features includ- ing semantic, acoustic, orthographic, and a number of other attributes (Underwood, 1969; Wickens, 1970). Multistore theorists have attempted to identify these attributes with the separate memory stores. Acoustic information has been identified with short-term store, whereas information in long-term store is said to be primarily semantic in nature. Attempts to verify this distinction empirically have met with varying degrees of success (for a review, see Shulman, 1971). There is an additional conceptual problem if it is assumed that in- formation enters long-term store by passing through short-term store. In this case, it would appear impossible for semantic information 1 A paper based on Experiment I in this article was presented at the Thirteenth Annual Meeting of The Psychonomic Society, St. Louis, 1972. Copyright © 1973 by Academic Press, Iac. All rights of reproduction in any form reserved. to enter long-term store if only acoustic information is contained in short-term store. Multistore theories focus on the nature of the stores and assign a secondary role to memory attributes; differences in retention of the separate attributes can be related to the store with which they are associated. An alternative approach would be to disregard the separation of memory into stores and focus on differences in selection and retention of attributes. Selection of attributes for in- clusion in the memory trace might depend on a number of factors including study time restraints (Shulman, 1970) and anticipated recall requirements (Jacoby & Bartz, 1972). With a rapid rate of presentation, there may be insufficient time for semantic processing so that stored information is primarily acoustic. Acoustic information might be sufficient to allow recall immediately after presentation so that a subject is likely to follow the rule of least effort and not store semantic information unless a delayed test of recall is anticipated. Past research suggests that semantic information is better retained (Cermak, Schnorr, Buschke, & Atkinson, 1970) and can be used as a basis of organization across a wider range of conditions (Bruce & Crowley, 1970) than acoustic information. 324