Stimulus concreteness, response characteristics, and the recognition-recall method in paired-associate learning.

Abstract

Stimulus-word concreteness and type of response were varied in two studies using different versions of the recognition-recall method for paired-associate learning. With stimulus-recognition ratings, criterion shifts were suggested that were a joint function of stimulus concreteness and type of response (noun, number, or consonant-vowel-consonant [cvc] trigram). When criterion shifts were controlled with a forced-choice recognition procedure, stimulus concreteness was found to affect response recall strongly even when it did not affect stimulus recognition for noun-cvc pairs as well as noun-noun pairs. The rating-bias effects and a triple interaction among distractor similarity, stimulus concreteness, and type of response suggested that response terms may affect stimulus processing even with highly codable stimuli and representative responses. Despite this and other potential artifacts of the recognition-recall method, results were seen as providing strong support for Paivio's interpretation of stimulus-concreteness effects with noun-cvc pairs. Since Martin's (1967) pioneering studies, the attempt to disentangle stimulus-recognition and associative processes in pairedassociate learning (PAL) by measuring stimulus recognition and response recall concurrently and by examining recall made conditional upon recognition (referred to here as the "recognition-recall method") has been much used but also much criticized. It has been used to help pinpoint the effects of such variables as word concreteness or imagery value (Nelson, Brooks,

Prepassage questions: The influence of structural importance.

“…The decision to select randomly one item from the no-question passage segment to serve as a comparison rather than to use the average number of recalls from the noquestion segment was based on the consideration that a comparison of recall from a small class of items with recall from a larger class of items can be biased in favor of finding a superiority for the smaller class of items. Such a bias is expected on the assumption that the larger the class of items the greater the number of difficult items contained in that class (see Runquist, 1974;Wicker, Thorelli, & Saddler, 1978). Thus, using the average recall measure for the noquestion condition could have biased the comparison in favor of a superiority for the question conditions, in which recall involved only one specific item.…”

Hierarchical Importance of Pre-Passage Questions: Effects on Cued Recall

“…The decision to select randomly one test question as the basis for assessing recall, rather than to use the average number of test questions answered correctly, was based on the consideration that a comparison of recall from a small class of items with recall from a larger class of items can be biased in favor of finding a superiority for the smaller class of items. Such a bias is expected on the assumption that the larger the class of items the greater the number of difficult items contained in that class (see Runquist, 1974;Wicker, Thorelli, & Saddler, 1978). Consequently, using the average recall measure for the no pre-passage question condition could have biased the comparison in favor of the question conditions, in which overall the number of possible indirect recall opportunities was smaller.…”