W. A. Bousfield

Measurement of Clustering and of Sequential Constancies in Repeated Free Recall

An operational version of Underwoods distinction between response recall and associative hook-up is applied to the analysis of sequences of freely recalled items. This distinction contrasts what are designated respectively as item properties and order properties. Formulas derived from this rationale are proposed for the assessment of clustering and the development of constancies in the sequential ordering of successive free recalls of given stimulus items. Consideration is given to the diversity of identifiable ordering effects in free recall.

Associative Clustering in the Recall of Words of Different Taxonomic Frequencies of Occurrence

Associative clustering may be defined as the occurrence of sequences of related words in the recall of a randomized stimulus word list. As indicated in an earlier study ( 3 ) , the operations for inducing and measuring clustering have typically involved the following steps: (a) the compiling of two or more sub-lists of equal lengths, of words with each sub-list representing words in a different taxonomic category; (b) the randomization of all the words thus chosen to form a stimulus list; (c) the presentation of the stimulus list for learning; (d) the obtaining of free recall; (e) the analysis of the recalled words to determine the incidence of clusters of two or more words falling in the same taxonomic category. The unit for the measurement of clustering is a repetition which is defined as a sequence of two words in the same operational category. In applying this measure, the number of repetitions in a cluster is designated as one less than the number of words in the cluster. The convention established in the original study of clustering ( I) was to rely upon Es judgment for the choice of words in the operational categories. To equate the response strengths of the words in the specified categories use was made of the Thorndike-Lorge (7) frequencies of usage. It was assumed that both the mean and the range of the frequencies of the words in each category should be similar. The feasibility of these operations was demonstrated in a study (2) employing two stimulus lists with the same categories in each. One comprised words with relatively high Thorndike-Lorge frequen.ties and the other words with relatively low Thorndike-Lorge frequencies. The data indicated a significant difference in both clustering and the number of words recalled in favor of the list with the higher frequencies. The present study was based on the choice of words selected from the

The Occurrence of Clustering in the Recall of Randomly Arranged Words of Different Frequencies-Of-Usage

Abstract : The study is closely related to one previously reported in that both deal with the influence of reinforcement on the incidence of clustering in the recall of randomly arranged words. Both employ the same hypotheses. They differ, however, in the techniques used for controlling reinforcement. The present study deals with differences in reinforcement during a pre-experimental phase whereas the earlier study undertook to manipulate reinforcement during the experimental phase. (Author)

Free vs unhibited recall

This study compared five multitrial free recalls of 20 minimally related words using two instruction conditions: Conditions, standard free recall instructions; Condition U, uninhibited recall instructions in which S s were told to emit all words occurring to them during recall. Errors were more prevalent for Condition U and were predominately item repetitions. Although level of correct recall was similar for the two conditions on the first trial, Condition U had significantly higher recall than Condition S on the last test trial. This recall superiority was attributable to less in ter trial forgetting under Condition U and was independent of the amount of error production.

Rhyme as a determinant of clustering

This study dealt with rhyme as a determinant of clustering in free recall. The stimuli comprised 12 rhyming pairs of words. The results from 30 Ss showed significan clustering as well as high variance attribu table to both Ss and word pairs.

USE OF FREE ASSOCIATIONAL NORMS FOR THE PREDICTION OF GENERALIZATION OF SALIVARY CONDITIONING TO VERBAL STIMULI

In a recent paper, Bousfield, Whitmarsh, and Danick (1958) presented a theoretical analysis of verbal generalization on the basis of what they termed partial response identities. They dealt with the situation in which an observable response having been learned to a given stimulus word could be elicited to varying degrees by a second but related stimulus word used for testing the presence of generalization. Since verbal stimuli reliably elicit free associational responses, they proposed that generalization obtained under these conditions should be predictable from the appropriate use of free associational norms for designated pairs of words. This theorizing led to the development of an index of predicted generalization which received some degree of support from a study in which the so-called observable responses learned to given stimulus words were also verbal in nature. ,The present study was planned as a further test of the validity of the theory of partial response identities as developed by Bousfield, Whitmarsh, and Danick on the basis of determining the extent to which their index of predicted generalization might have predicted the experimental findings reported by Razran (1949) in a study of the generalization of salivary conditioning to verbal stimuli. The basic assumptions of the theory of partial response identities with which this paper is concerned may be summarized as follows. A stimulus word reliably elicits two types of implicit responses, both of which produce feedback stimuli. These responses are: ( a ) a distinctive representational response which may be labelled as the stimulus word itself; ( b ) a group of verbal associational responses which are elicited by the representational response. The specification of the associational responses as well as their relative strengths may be made from the use of free associational norms. During the learning of an observable response to a given stimulus word, multiple S-R connections develop between the observable response and the various feedback stimuli generated by both the representational response and the free associational responses. After this learning has taken place, a second stimulus word may be presented to test for generalization of the observable response. The tested

The Incidental Learning of Associative Responses to Given Stimulus Words

Abstract : This study was undertaken as a test of the following hypothesis: the maintenance of a set to emit a given stimulus word will facilitate the incidental learning of its verbal associative responses. The set to emit the given stimulus word was developed in two stages. The subjects (Ss) first undertook to identify the given word as it was projected on a screen at successive levels of clarity. The Ss were then told to indicate when the already identified word appeared in a series of projected items. These items included a group of associates of the critical stimulus word along with associates having no apparent relationship to the given stimulus word. The results indicated significantly better retention of the associates of the given stimulus word than was the case for unrelated associates. The results appeared to support the experimental hypothesis. The interpretation is in terms of the concept of response priming and what the authors designate as the extent to which stimulus items elicit distinctive meaningful response patterns. (Author)

Sequential Ordering in Repeated Free Recall as a Function of Interitem Associative Strength

A modification of Tulving’s method served for appraising the relationship between sequential ordering in repeated free recall and interitem associative strength, IIAS. Multiple randomizations of three of Deese’s world-lists having zero, low and high IIAS respectively, were presented to separate groups of Ss for a fixed number of presentation-recall trials. Partial support was obtained for the hypothesis that the rate at which ordering develops in repeated free recall as indicated by the occurrence of the same word-bigram units on pairs of successive recalls should vary as a positive function of IIAS. While ordering was substantial for all lists and increased as a function of IIAS, only the zero and high conditions differed significantly. The items constituting the word-bigrams repeated in successive pairs of recalls varied significantly more in the high and low than in the zero conditions. Words from the high were significantly less prone to oscillation in recall than those from the zero List. Possible explanations of these findings are considered.

Immediate Memory Spans for CVC Trigrams

Abstract : Immediate memory spans were obtained for groups of homogeneously meaningful CVC trigrams selected so as to sample the entire range of Archers norms. The items at the lowest level were nonsense syllables of the lowest possible association value. At the opposite extreme were highly meaningful three-letter words. The design of the study involved the use of two types of instruction, the first calling for the ordering of recall and the second calling for free recall. For each type of instruction there were two modes of recall, namely, spelling and pronunciation. Pronunciation yielded appreciably higher memory spans than spelling. In spite of the freedom permitted in the instructions for free recall, the Ss tended to reproduce the items in the order of their presentation. There was an apparent tendency for the spans to increase with meaningfulness at a positively accelerated rate when the Ss spelled the items. An attempt is made to relate these findings to a group of tenable theoretical assumptions. (Author)

The effects of synonymity on item recall and sequential ordering in multitrial free recall

A list of 14 synonymous and one of 14 minimally related adjectives were presented to respective groups for 16 multitrial free recall trials. Analyses of the two sets of data failed to reveal noteworthy differences either in item retention or in sequential ordering of recall. The results indicated that Ss given synonyms relied, to an appreciable degree, on the formal characteristics of the words as memory attributes, whereas those given nonsynonymous words responded more to minimal interitem associations. It was hypothesized that the differential effects of synonymity on performance depends mainly on list length.