Word length in immediate serial recall: Separating effects on order information from effects on item information

Four experiments were conducted to test whether word length effects are confined to immediate serial recall tasks or whether they can be observed in comparative judgement tasks as well. Both tasks measure the accuracy of stored serial information, but they have different output requirements. Experiment 1 and Experiment 2 showed that word length effects which were observed in a serial recall task were not observed in a comparative judgement task. Experiment 3 and Experiment 4 showed that word length effects in serial recall tasks do not depend on a single-item sequential acquisition procedure, but can be observed with transitive inference acquisition procedures as well. A synoptic analysis of subsets of the data from the four experiments revealed an interaction of encoding variations and test variations and an interaction of word length variation and test variation, which can be explained most parsimoniously by a two-component model in which the word length effect is only associated with the access-to-item component.