JOURNAL
OF VERBAL
Visual
LEARNING
Prompting
AND
VERBAL
BEHAVIOR
7,148-153 (1968,
in a Paired-Associates
Mediation
Paradigm1
MARVIN W. DAEHLERAND JOHNC. WRIGHT University
of Minnesota,
Minneapolis,
Minnesota
55455
Two experiments using 105 fifth-grade children were designed to determine the effects of presentation of various types of prompts on the test list of a paired-associate mediation paradigm, BC-AB-AC. Picturepromptspresented with somepairsduringthe anticipation interval to elicit appropriate mediators facilitated performance, but only for those pairs where the prompts were given. Performance for the unprompted pairs was similar to that of a group given no prompts or random shapes as prompts. Presentation of picture prompts designed to elicit inappropriate mediators also facilitated performance for the pairs so
prompted,but for the remainingunpromptedpairsperformancewassimilarto that of a control group,BC-AD-AC. Sincean inappropriatepromptwasproducedby re-pairingan appropriate mediator from an unprompted pair, it was concluded that the prompt was “captured” by the pair with which it was presented and the additional distinctiveness of the pictures facilitated performance for these pairs. As a result, however, the mediator became
unavailable to thepairit originallyservedto mediate. Using a mediation paradigm, verbal pairedassociatesstudies have generally found facilitated learning on the test list and have assumed such facilitation to be a function of covert occurrence of mediators (e.g., Russell and Storms, 1955; Horton and Kjeldergaard, 1961; Norcross and Spiker, 1958). For example, in the paradigm BC-A&AC the B term is conceptualized as occurring covertly on the final list as a response to stimulus A, and the response-produced stimulus properties of B are said to elicit the correct overt responseC. When no mediator is available, as in the control paradigm BC-AD-AC, fewer correct responsesare given on the test list. One way to examine the assumption that the covert mediating response produces the observed facilitative effects is to manipulate directly the probability of occurrence of that responseby meansother than association. 1 This study is based on a thesis, submitted by the senior author in partial fulfillment of the requirements for the MA degree in Child Psychology, Institute of ChildDevelopment, University of Minnesota. Portions of this paper were presented at theannual convention of the Midwestern Psychological Association, Chicago, May, 1966. 148
This study is concerned with the availability of mediators during test-list performance as a meansof facilitating or interfering with correct responding. Explicit presentation of picture prompts designedto elicit appropriate mediztors during performance should produce facilitation beyond that obtained by assuming the functioning of only a covert mediator. Given the availability of the name resulting from its occurrence on the training lists, the dominant, though covert responseto the picture of a namable object was assumedto be its name. Consequently, the responselearned to that name on the training list would now be expected to follow on the test list, that is, the correct response.Moreover, if the tendency to useappropriate mediating responsesfunctions as a general set, it should be manifest on all pairs for a given S, once it has been experimentally elicited on some pairs in the list. Conversely, presentation of a picture prompt that elicits an equally available but inappropriate mediator during performance, should produce interference on prompted pairs. Two experiments were designedto test these assumptionsby useof visual prompting.
VISUAL PROMPTING IN MEDIATION EXPERIMENT
I
Method Subjects. Sixty fifth-grade children served as Ss.2 Five Ss were replaced because of learning difficulties or because of mechanical errors in the apparatus. No more than two of these replaced’ss had been assigned to any particular condition. Design. The Ss were randomly assigned to one of four groups and members were required to learn three lists of six pairs of words. The control group learned lists BC-AD-AC, and three mediation groups learned lists BC-AB-AC. These latter three groups differed only in the nature of the visual prompts presented on the final test list. One group was provided no prompts (Mediation-No Prompts), another group was provided random shapes as prompts (Mediation-Random Shapes), and the third group received pictures, line drawings of the appropriate mediators, as prompts (MediationPictures). The random shapes were outlines of shapes designed by Munsinger, Kessen, and Kessen (1964) and were presented to control for effects that might have been a function of presentation of material other than the stimulus word during the anticipation interval. The pictures elicting mediators were line drawings taken from the Peabody Picture Vocabulary Test. Lists. For all the groups the stimuli for List 2 and the test list (A-words) consisted of CVC trigrams with association values between 68 and 69 % (Archer, 1960). The remaining verbal materials consisted of common nouns. Those nouns serving as mediators on the test list (B-words) and those serving as responses (C-words) were randomly selected, with the restrictions that the mediator be capable of pictorial representation and that the mediator and response show no association on available word-association norms (Palermo and Jenkins, 1964). For the control group the lists were the same, with the exception of an additional and comparable set of nouns (D-words) used as responses on List 2. Procedrrre. An electrically programmed Dunning Animatic filmstrip projector delivered stimuli to the back of a 3 x 12-in. opal glass window located in a panel partitioning the observation and control rooms of a mobile laboratory. For all lists the anticipation interval, joint presentation interval, and interpair interval were 3 set each. For the Random-Shapes and the Pictures groups prompts on the test list were pre-
2 The authors wish to acknowledge the cooperation of Dr. Rodney Tillman, Assistant Superintendent, and Dr. Robert Rainey, Research Coordinator, Minneapolis Public Schools; and the collaboration of Miss Irene Gjermundboe and Mrs. Helen Tressel, Principals, and the teachers, staff, and students of Clara Barton and Standish Elementary Schools, Minneapolis.
149
sented during the final 2 set of the 3-set anticipation interval only and were not exposed during the joint presentation interval. Thus, the stimulus word appeared alone for 1 set followed by the stimulus word and the prompt together for 2 sec. Then the stimulus and response word only appeared during the jointpresentation period. The prompts appeared equidistant from the end of the stimulus word and the beginning of the response word and were presented to the RandomShapes and Pictures groups only on alternate trials of the test list. The alternation of prompt-present and prompt-absent trials was designed to determine if the effect of prompting generalized to prompt-absent trials. The Ss were initially given standard instructions for learning paired-associates. After reaching criterion on the first list, approximately 2 min intervened before second-list learning began, during which time E changed the filmstrip, and each S was instructed to learn this second list as he had learned the first. After reaching criterion on the second list and immediately prior to learning the test list, the following instructions were given to Ss in all groups: Now I want you to learn the final list. Sometimes we may give you a hint in the middle between the words, which may or may not help you remember which word will appear. Again, the first time through the series just read the words out loud when they appear. The first trial on every list served as a familiarization trial. Prompts were delivered to the two specified mediation groups on the familiarization trial of the test list. The two training lists were presented to a criterion of two consecutive errorless trials. If S did not reach this criterion within 20 trials on either training list, he was replaced. All remaining Ss were given 12 trials on the test list.
Results On the first list mean number of errors in the various conditions were : Control, 20.13; No Prompt, 17.73; Random Shapes, 18.20; and Pictures, 19.40. Mean number of errors on List 2 were: Control, 22.40; No Prompt, 26.53; Random Shapes, 26.33; and Pictures, 23.73. Analysis of the number of errors indicated no significant differences among groups on either the first or the second list, as did analysis of trials to criterion, thus supporting the assumption that the groups did not differ in the ability to learn the paired associates.
150
DAEHLER AND WRIGHT
An analysis of variance of correct responses on the test list, including prompted vs. unprompted trials as a repeated measure, yielded an effect for groups which was not quite significant, P(3, 56) = 2.40, p-c .10. However, the interaction between groups and prompted vs. unprompted trials was highly significant, F(3, 56) = 10.66, p < .OOl. Figure 1 illustrates the generally poorer performance of Ss in the control group on all trials, and further suggests that the significant interaction was due to the marked facilitation for the Pictures
FIG. 1. Mean number of correct responses per trial (Exp. I).
group on those trials where the prompts were presented. Further analysis indicated that significantly more correct responses were made on prompted trials (odd, 3-11) than on unprompted trials (even, 4-12) by the Pictures group, F(l, 14) = 17.32, p < .OOl, even though the analysis was such that the general improvement in learning as a function of increasing trials worked against the hypothesis. Apparently the picture prompts, but not the random shapes, had some effect on those trials where they were presented. There was no evidence, however, of generalization of the prompting effect to picture-absent trials. Furthermore, on any given trial either all pairs or no pairs were prompted so that after the first pair in each trial, S could know with certainty whether the remaining pairs on that trial would be prompted or not.
EXPERIMENT
II
In order to obtain a more sensitive evaluaa second study was designed so that three pairs of the test list were presented with prompts on all trials and three pairs were always presented without prompts. Furthermore, an “interference” condition was added. tion of the effect of prompts,
Method Subjects. Forty-five additional fifth-grade children served as .Ss.Three Ss were replaced because of learning difficulties or because of mechanical errors in the apparatus. Two Ss were replaced in the RandomShapes condition and one in the Appropriate-Pictures condition (see Design). Design. The Ss were randomly assigned to one of three groups. A11 groups received the mediated facilitation paradigm, BC-AB-AC, and differed only in the nature of the prompts presented in half of the pairs on the test list. One group was presented random shapes as prompts (Random Shapes), another group was presented picture prompts that would elicit appropriate mediators (Appropriate Pictures), and the third group was presented picture prompts designed to elicit inappropriate (that is, re-paired) mediators (Inappropriate Pictures). In the test list three of the six pairs were prompted, each pair always with the same prompt. Seven Ss in each group received three of the pairs with prompts and the other three pairs were prompted for the remaining eight Ss in each group. An inappropriate prompt was produced by re-pairing such that the three pictures presented as prompts elicited the appropriate mediators for the alternate three unprompted pairs. Thus, in spite of the use of a classical mediated-facilitation paradigm, the prompts were designed to produce interference for the three prompted pairs. The lists and procedure were the same as those used for the mediation groups in the first experiment.
RESULTS
Results Training Lists. In addition
to the three new groups, the results for the Control and NoPrompt groups from the first experiment were included in analyses of the training lists. Mean number of errors on the first list were : Control (Exp. I), 20.13; No Prompt (Exp. I), 17.73; Random Shapes, 17.53 ; Appropriate Pictures, 18.13 ; and Inappropriate Pictures, 20.40.
VISUAL
PROMPTING
Mean number of errors committed on the second list were: Control, 22.40; No Prompt, 26.53; Random Shapes, 26.53; Appropriate Pictures, 28.93; and Inappropriate Pictures, 29.80. There were no significant differences among groups in number of errors or trials to criterion on either list. Thus, the Ss in the two experiments were considered comparable in learning ability, and the results from the Control and No-Prompt group of the first experiment were therefore included in the analyses of test list data of Exp. II. Test List. For statistical purposes seven Ss in both the Control and No-Prompt groups were randomly assigned as if three of the pairs on the test had been prompted and the other three unprompted. For the remaining Ss in these two groups, those pairs considered prompted and unprompted were reversed. Responses for all groups could then be analyzed in terms of the repeated measure of “prompted” vs. “unprompted” pairs. Table 1 presents the mean number of correct responses per trial for each group for prompted pairs, unprompted pairs, and the two types of pairs combined for all test trials and for the first four anticipation trials only. Over all test trials the best performance was by Ss in the Appropriate-Pictures group on prompted pairs. The classical mediation conditions (No Prompt and Random Shapes) are
intermediate, while the Control group is worse, as predicted. The Inappropriate-Pictures group, however, performed almost as well as the Appropriate-Pictures group when prompted, but slightly worse than the Control group on unprompted pairs, both contrary to expectation. A repeated-measures analysis of variance of correct responses over all trials yielded significant main effects for groups, F(4, 70) = 3.44, p < .05, and pair types (prompted vs. unprompted), F(l,70) = 12.73,~ < .OOl, and a significant Groups x Pair-Type interaction, F(4, 70) = 7.91, p < .OOl. Scheffe’s contrasts performed on simple effects yielded significantly more correct responses on prompted than on unprompted pairs (p < .Ol) for both picture-prompted groups. On “prompted” pairs the Appropriate-Pictures group made significantly more correct responses than the Control group (p < .Ol). Because Ss in the Appropriate-Pictures group showed rapid learning of prompted pairs and approached a ceiling of performance on later trials, a further analysis was performed on the results of the first four anticipation trials only. Significant main effects were found for groups, F(4,70) = 4.87, p < .OOl, and pair types, F(1, 70) = 12.81,~ < .OOl, and a significant Groups x Pair-Types interaction was obtained, F(4, 70) = 11.37, p < .OOl. Scheffe’s contrasts yielded simple effects significant at
TABLE MEAN
CORRECT
151
IN MEDIATION
RESPONSES PER TRIAL
1 FOR GROUPS
AND PAIR
Trials 2-5 Group Control (Exp. I) No prompt (Exp. I) Random shapes Approp. pictures Inapprop. pictures
Prompted pairs AT 1.23” 1.14 2.17 1.65
Unprompted pairs .90 1.32 1.28 1.15 .73
TYPES Trials 2-l 2
All pairs 1.57 2.55 2.42 3.32 2.38
Prompted pairs 1.52” 1.88” 1.92 2.57 2.21
Unprompted pairs 1.59 1.98 1.93 1.93 1.50
All pairs 3.11 3.87 3.85 4.50 3.71
u These groups were presented no prompts; results are for pairs that would have been given prompts had ss in these groups been assigned to a prompted condition.
152
DAEHLER
the .Ol level for the same contrasts that were significant, in the previous analysis. In addition, differences approaching significance (p < .lO) were obtained on “prompted” pairs between the Appropriate Pictures and No-Prompt groups, Appropriate-Pictures and Random-Shapes groups, and the Inappropriate-pictures and Control groups.
AND
WRIGHT
words between which it appears, but also the excision of the same word from its original chain, so that it is no longer available to mediate that association. The rapidity with which capture took place strongly suggests a mediational process, but one that depends in part on the perceptual vividness and uniqueness in this context of a picture rather than a word. The timing and spatial position of the prompt appear to be DISCUSSION such as to induce S to rely on this perceptually supported link rather than on the more slowly The results of Exp. I indicated that picture prompts eliciting appropriate mediators pro- established and memory-dependent verbal vided facilitation beyond that typically ob- link. To the extent that the perceptual link tained from the mediation paradigm alone. must have continual perceptual support, however, it is a more limited form of mediation, A comparable group provided appropriate prompts in Exp. II also showed facilitation on and indeed the results of Exp. I indicate that it had no effects on the alternate trials where prompted pairs compared to unprompted pairs and facilitation on prompted pairs com- it was not presented. The performance of the inappropriately prompted group on unpared to a classical mediation group. However, the mechanism by which facilitation occurred prompted pairs indicates that availability of perceptual prompts does not compete with, may not have been the arousal of appropriate but instead suspends, the operation of premediators. The group presented inappropriate viously established mediational links. Thus on prompts also appeared to benefit on prompted pairs, although not to as great a degree. Per- the unprompted pairs, performance was at the level of the control group rather than at that of haps the improved performance on prompted the classical mediation group. Whether such a pairs was due to the additional distinctive and suspension would be general or limited to the meaningful cues provided by the pictures, captured mediators can only be established by whether or not they aroused appropriate mediators. The effect of re-pairing prompts, so an additional experiment, but the effect is as to elicit inappropriate mediators was rather probably limited. The use of heterogeneous lists and a comparison of verbal (perhaps to provide a substitute for the original mediauditory) prompts with picture prompts would ator in the associative chain, one that facilitated performance almost at once. We call this indicate whether prompting of mediators is process “capturing;” that is, the pictures and simply a perceptual crutch to performance or any responses that they elicit were captured by actually intervenes in the availability of prethe pairs in which they appeared. Unlike the viously established mediating responses. In conclusion, providing prompts to elicit random shapes, the pictures strengthened the associative chain by serving as a distinctive and assumed mediators does facilitate correct meaningful link, occurring temporally between responding, but this may be due to increased differentiation of the stimulus complex by A and C, and enhancing their association. Such a facilitating effect could not occur, how- adding distinctive perceptual cues. Even when ever, if the picture prompt were inserted in a prompts are re-paired to elicit the wrong new pair without being disconnected from the mediators, the latter can be captured to facilipair in which it originally occurred. Thus the tate correct responding. Disruption of medioccurs when the assumed term capture is designed to suggest not only ated facilitation the rapid assimilation of the prompt to link the mediator is captured as a distinctive cue for
VISUAL
PROMPTING
another pair and thus is made unavailable as a simple mediator for the original pair. Contrary to expectations, appropriate prompts did not produce a general increment in the effective use of mediators on unprompted pairs or unprompted trials. It appears, therefore, that prompting affects performance rather than strategies or learning sets. It is suggested that direct intervention during performance is a potentially fruitful method of determining some of the operating characteristics of mediators. REFERENCES ARCHER, E. J. Re-evaluation of the meaningfulness of all possible CVC trigrams. Psychol. Monogr., 1960,73, No. 10.
IN MEDIATION
153
HORTON, D. L., AND KJELDERGAARD,P. M. An experimental analysis of associative factors in mediated generalization. Psycho!. Monogr., 1961, 75, No. 11. MUNSINGER, H., KESSEN, W., AND KESSEN, M. L. Age and uncertainty: Developmental variation in preference for variability. J. exp. child Psycho/., 1964,1, l-15. NORCROSS,K. J., AND SPIKER, C. C. Effects of mediated association on transfer in paired-associates learning. J. exp. PsychoI., 1958,55,129-134. PALERMO, D. S., AND JENKINS, J. J. Word association norms: Grade school through college. Minneapolis: Univ. of Minn. Press, 1964. RUSSELL, W. A., AND STORMS, L. H. Implicit verbal chaining in paired-associate learning. J. exp. Psychol., 1955,49,287-293. (Received August 19, 1966)