CONCEPT FORMATION AS A FUNCTION OF LOGICAL AND INFRALOGICAL MEDIATION AND STIMULUS IMAGERY-CONCRETEIIESS Thesis for the Degree of M. A. MICHIGAN STATE UNIVERSITY SHELLEY JOHN STOKES 1971 m”II"[IUUII’HIIIIIIIIIII'IIIUIIIIW L ABSTRACT CONCEPT FORMATION AS A FUNCTION OF LOGICAL AND INFRALOGICAL MEDIATION AND STIMULUS IMAGERY-CONCRETENESS BY Shelley John Stokes The role of verbal and nonverbal processes in grouping stimuli into conceptual units was examined. Reaction time data were obtained for subjects instructed to group concrete words or abstract words on the basis of logical or infralogical mediation strategies which re- quired the use of verbal or nonverbal associative proc- esses, respectively. A 2 X 2 X 2 factorial design involving mediational set and order of presentation as between—subjects factors, and stimulus attribute as the within-subjects factor yielded eight conditions, (a) Infralogical mediation with concrete words, (b) logical mediation with concrete words, (c) infralogical mediation with abstract words, and (d) loqical mediation with abstract words, for each of two orders of presentation (concrete words-abstract words vs. abstract words-concrete words). Shelley John Stokes It was hypothesized that: (a) infralogical media— tion would be more difficult with abstract than with con- crete stimuli resulting in longer response latencies, (b) logical mediation would be relatively less affected by variation of stimulus concreteness, and (c) grouping ab— stract stimuli should generally require longer reaction times than would grouping concrete stimuli. The results confirmed the latter hypothesis (i.e. concrete words were generally grouped more rapidly than were abstract words). This finding was particularly interesting since abstract— ness has apparently not been investigated in studies of conceptual behavior. The results, however, did not con- firm the predicted mediational set x stimulus attribute interaction. The lack of significance obtained for this interaction was interpreted on the basis of the subjects' apparent failure to use the mediational strategies required. Approved: Date: 17/5/7Z CONCEPT FORMATION AS A FUNCTION OF LOGICAL AND INFRALOGICAL MEDIATION AND STIMULUS IMAGERY-CONCRETENESS BY Shelley John Stokes A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1971 ACKNOWLEDGMENTS The author expresses his deepest appreciation to Dr. James S. Uleman, adviser and chairman of his thesis committee, for his ever present support, suggestions, and assistance throughout the various stages of this thesis, and to Drs. Henry C. Smith and David L. Wessel for their knowledgeable comment and advice. ii TABLE OF CONTENTS Page ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . . ii LIST OF TABLES . . . . . . . . . . . . . . . . . . . v LIST OF APPENDICES . . . . . . . . . . . . . . . . . vi INTRODUCTION . . . . . . . . . . . . . . . . . . . . 1 LITERATURE . . . . . . . . . . . . . . . . . . . . . 5 STATEMENT OF THE PROBLEM . . . . . . . . . . . . . . 8 HYPOTHESES . . . . . . . . . . . . . . . . . . . . . 12 Major 0 O 0 O O O O O O O O O O O O O O O O O O O 12 Minor 0 O O O O O O O O O O O O O O O O O O O O O 12 METHOD 0 O O O O O O O O O O O O O O O O O O O O O O l 3 Subjects and Design . . . . . . . . . . . . . . . l3 Concept Formation Lists . . . . . . . . . . . . . 13 Procedure . . . . . . . . . . . . . . . . . . . . 15 Time Measure . . . . . . . . . . . . . . . . . . 17 Mediation Instructions . . . . . . . . . . . . . 17 RESULTS 0 O O O O O O O O O O O O O O O O O O O O I O 19 2 X 2 X ANOVA . . . . . . . . . . . . . . . . . . l9 Stimulus-Attribute' . . . . . . . . . . - - . . . 21 Order of Presentation . . . . . . . . - - . - . - 21 Mediation Set . . . . . . . . . . . . - . . . . . 23 Mediation Set X Stimulus Attribute . . . . . . . 23 Order X Stimulus Attribute . . . . . . . . . . . 24 2“X 2 X 5 ANOVA . . . . . . . . . . . . . . . . . 24 Abstract Lists . . . . . . . . . . . . . . . . . 24 ‘Order of Presentation . . . . . . . . . . . . . . 25 Mediation Set . . . . . . . . . . . . - . . - - - 25 Order X Abstract Lists . . . . . . . . . . . . . 27 2 X 2 X 5 ANOVA . . . . . . . . . . . . . . . . . 27 Concrete Lists . . . . . . . . . . . . . . . . . 27 iii Order of Presentation Mediation Set Order X Concrete Lists DISCUSSION Major Hypotheses Minor Hypothesis REFERENCES APPENDICES iv Page 27 28 28 29 29 33 34 36 LIST OF Table 1. Mean latencies (sec.) by order of presentation, for each grouping . . 2. Mean latencies (sec.) by and order of grouping 3. Mean latencies (sec.) by order of grouping . . TABLES Page stimulus-attribute, and mediation set 0 O O O O O O O O O O O 20 stimulus attribute O O O C O O O C O O O O 22 abstract lists and O O C O O O O O O O O I 26 LIST OF APPENDICES Appendix Page A. word Lists 0 O O O O O O O O O O O O O I O O O 36 B. General Instructions . . . . . . . . . . . . . 46 C. Mediation Sets . . . . . . . . . . . . . . . . 48 D. Instructional Set Reminder . . . . . . . . . . 50 E. Mediation Set Reminder . . . . . . . . . . . . 51 vi INTRODUCTION An impressive body of evidence has demonstrated that language and verbal process play an important role in human conceptual behavior. However, little research has investigated the role of nonverbal process in grouping stimuli into conceptual units. Cofer (1960) has suggested that mediating verbal responses are among the most important factors for deter- mining hypotheses in concept formation. Evidence for this position has been obtained by several studies which show the unmistakable influence of verbal process (cf. Cofer, 1960). Judson and Cofer (1955) have reported that when subjects were given a problem solving task of selecting th3"unrelated" word in a series of four, the subjects consistently selected and rejected concepts on the basis of prior verbal habits. Similar evidence for a directional effect on hypotheses in problem solving tasks has been offered by Gelfand (1958). In a study of the effect of reinforcement of response systems on the solution of problems to which those systems were relevant, Gelfand reports an unmistakable influence of prior verbal habits. The subjects consistently chose concepts for which they had been primed. Those subjects who had learned concept-relevant words made few errors in concept learning. The subjects who learned irrelevant words made more errors than subjects who had learned neutral words. Research on the effects of verbal process in con- ceptual behavior has been so compelling that Deese and Hulse (1967) have suggested that "the human ability to invent and utilize concepts is one aspect of linguistic activity." Further, Bourne (1966, p. 21) has observed that the present emphasis onlanguage in conceptual behav— ior has led at least one psychologist who has worked extensively in the area to the conclusion that concepts ". . . are meaningful words which label classes of other- wise dissimilar stimuli" (i.e. Archer, 1964, p. 238). While it may be useful to investigate conceptual behavior as if it were a subset of language, such an ap- proach may exclude a wide range of behaviors which may be functional for conceptual activity. Predilections toward a something-in—addition-to—words approach are inherent in Bourne's (1966, p. 21) suggestion that, not only is it the case that most conceptual groupings have meaningful verbal labels but also some concepts are learned and used almost exclusively in a verbal context. However, while . . . [it may be said that] verbal processes enter into the chain of behaviors appropriate in almost any conceptual problem, there is no reason to assume a strict identity of words and concepts. Thus while the importance of language in concep- tual behavior cannot be denied (cf. Bourne, 1966), the implication that in many instances it may become impos- sible to distinguish a concept from its verbal label (Archer, 1964) is unjustified. The fact that subjects can readily identify instances of a concept without being able to describe the underlying attributes which make any par- ticular object an instance of that concept (Hull, 1920) makes Archer's (1964) position untenable. As Hull has observed, people can learn concepts or stimulus groups without being able to put them into words. Further, it would seem that the difficulty a subject has in describing the basis for his groupings, either because the concept is arbitrary and unnatural, or because the stimulus attributes do not lend themselves easily to verbal associations, illustrates the independence of words and concepts. As Bourne (1966, p. 107) concludes (in reference to the con- cept-language problem), "while the existence of common or at least parallel features may be granted, we should also recognize the essential independence of these processes." What is needed, then, is a View of conceptual activity which explicitly states the relationship between verbal and nonverbal processes. .Generally, the present research will attempt to explore such a relationship based on the assumption that neither verbal nor nonverbal proc- esses are sole determiners of conceptual activity, and based on the suggestion that the functional significance of nonverbal processes may be studied without denying the obvious relevance of verbal mechanisms (cf. Paivio, 1969, p. 243). More Specifically, the present research is con- cerned with the functional significance of nonverbal imagery and verbal processes as mediators for concept formation. LITERATURE While the two-process theory of meaning and media- tion developed by Paivio (1969) has received most attention in analyses of associative strategies in paired-associate (PA) learning, and of mediational latencies within stimulus response (SfR) pairs, it presents a model which may prove useful to the study of conceptual behavior. The possible importance of such a model for conceptual behavior is often implied by psychologists who suggest that there is more to conceptual activity than simply verbal meaning and words. Paivio (1969) has suggested that words derive their meaning through at least two kinds of associative processes. Concrete words (i.e. words which refer to ob- jects, materials, or persons, Paivio, Yuille & Madigan, 1968) presumably derive their meaning through associations with concrete objects and events as well as through asso- ciations with other words. Therefore, such words acquire the capacity to evoke both nonverbal images and verbal processes as associative reactions. Abstract words (i.e. words which refer to abstract concepts that cannot be experienced by the senses, Paivio, et al., 1968) however, derive their meaning primarily through intraverbal ex- periences and more effectively arouse verbal associative than imaginal processes. According to the two process theory, imaginal and verbal processes, then, may function as alternative coding systems which affect meaning, mediation and memory. Imag- ery is postulated to be functionally linked to an abstract- concrete dimension of stimulus meaning such that the higher the concreteness of stimulus items, the more likely they are to evoke sensory images which can function as mediators of associative learning and memory. Verbal mediators, on the other hand, are not assumed to vary functionally with concreteness, but are said to be corre- lated with verbal associative meaning (i.e. as indexed by such measures as association value, Glaze, 1928, and m, Noble, 1952). One prediction which may be derived from the two- process theory is that the latency of an associated image to a stimulus word should be longer when the stimulus is abstract and low imagery (I), than when it is concrete and high I; whereas, verbal associative latency should be less affected by variation of concreteness. Paivio (1966) tested this prediction by obtaining reaction time (33) data for subjects instructed to press a key and give verbal descriptions when either a mental image or an im- plicit verbal associative occurred to a stimulus word. The results of the study supported the two-process model: while longer response latencies were found for abstract than for concrete words under both imaginal and verbal instructions, the difference was much greater under the imagery set. While Paivio's (1966) study established the speed of associative reaction to individual stimulus words for verbal and imaginal sets, it provided no direct evidence for actual mediational efficiency of either of the two processes. Additional support for the validity of the two-process model has been offered by Yuille and Paivio (1967). As a direct test of the speed of establishing mediated linkage between groups of words, Yuille and Paivio measured §T_for subjects instructed to link members of S73 pairs with either verbal or imaginal mediators. It was found that, as predicted, imaginal latencies increased with abstractness of the stimulus attribute. This inter- action of concreteness of stimulus and mediation set sup- ports the hypothesis that verbal symbolic processes and images are equally available as mediators when the stimulus is concrete, but only verbal mediators are readily aroused when'the stimulus is abstract. Further, additional evi- dence for the differential availability of the two proc- esses has been obtained from PA_subjects who have indicated that they used imaginal mediators more often to learn concrete pairs while the reverse was true in the case of verbal mediators (Paivio, Yuille & Symthe, 1966). STATEMENT OF THE PROBLEM Although the results of Paivio (1966) and Yuille and Paivio (1967) have provided relevant data on the dif- ferential availability of imaginal and verbal processes as associative reaction to individual stimulus items, and for mediated linkage between SfR_pairs, no direct evidence has been shown which establishes the availability of verbal or imaginal mediators for members of larger groups of words, or for more complex cognitive process. If nonverbal imagery and verbal processes function as alternative cod- ing systems for PA learning and memory, perhaps they may function similarly in the formation of concepts (i.e. words which have been grouped or classified together on the basis of some common feature or characteristic of each, cf. Bourne, 1966, p. 2). It would seem that a parallel research model may be constructed for concept formation which suggests that the relative Speed with which a conceptual grouping is formed, may, at least in part, be dependent upon the degree of concreteness and the imagery value (I) of the stimulus items, and upon the mediational strategy employed. To provide such evidence, subjects in the present study were required to group concrete or abstract words within multi-item lists on the basis of logical or infra- logical mediation strategies (cf. Reigel, 1965, pp. 4-5). As Reigel has noted, infralogical tasks require responses derived from reference to physical objects and do not rely merely upon verbal abstractions. In order to respond ap- propriately, the subject must mentally perceive the phys- ical object or the attribute or quality denoted by the stimulus; he must make a "conceptual partition" of the perceptualized object and detect its Specific character- istic properties. Thus, infralogical mediation would call for the use of imagery (i.e. a mental picture, or sounds, or other sensory experience, Paivio, gE_§l., 1968) to find a "common feature or characteristic" (i.e. Bourne, 1966) upon which the stimulus words could be grouped. Logical tasks, on the other hand, demand the recognition of classes and their members. Whether spatially and/or temporally combined or separate, the subject need only think of the instances as constituting a group. This definition is quite consistent with the notion that contiguity of stimuli 'is not necessary for the formation of verbal associative response (Asch, 1969). Based on the derivative of two-process theory that images and words are differentially available as associa- tive responses and mediators for abstract stimuli, and on the results of previously cited research (Paivio, 1966: Yuille & Paivio, 1967), it was expected that infralogical 10 mediation would be more difficult with abstract than with concrete nouns as stimulus items, resulting in longer re- sponse latencies. Logical mediation, however, Should be affected less by variation of stimulus concreteness. Of course, inherent in these predictions is the assumption that generalizations from PA research can be applied to conceptual behavior. While it is clear that such appli- cations are not direct due to the differential complexities of behavior involved in moving from PA_1earning to con- ceptual behavior, construction of a parallel model was assumed to be possible for the present research. Garner (1962) has suggested that there is a continuity between paired-associate learning and concept learning: in the usual case of PA_learning, one response is assigned to each stimulus. However, by definition, concept learning depends on having several stimuli to each response because the task is for the subject to discover what is common between two or more stimuli. There are, then, more stimuli than responses in the general concept-learning case. Thus, Garner concludes that the basic experimental paradigm for what is called a concept problem is quite similar to the paradigm for a paired-associate learning task in that a series of stimuli is presented and the subject must learn to use different responses for differ- ent stimuli. Concept formation, however, may be differ- entiated on the basis of two Specific characteristics: 11 first, the subject must use the same response for more than one stimulus (i.e. stimulus equivocation). Second, the stimuli must be multi-variate in nature, so that one or more of the variables can be defined as relevant to the response differentiation required, and others as irrele- vant. While the present research does not attempt to resolve this issue, significant differences obtained would indicate need for an approach to conceptual behavior which is, at least, similar to a two-process PA_approach. Major (1) (2) Minor (3) HYPOTHESES The latency of discovery of infralogical mediators should be slower for abstract, low I stimuli than for concrete, high I stimuli resulting in longer RES for forming the stimulus groupings. The availability and effectiveness of logical mediators should not be similarly affected by variation of abstractness-concreteness. In addition to the differential effects of stimulus concreteness and mediation set, grouping abstract stimuli should generally require longer response latencies than would grouping concrete stimuli. 12 METHOD Subjects and Design Fifty-three students (males and females) from introductory psychology classes at Michigan State Univer- sity volunteered for the experiment as a part of a course requirement. Each was given extra credit for participating in a "study of principles for grouping words." A 2 X 2 x 2 factorial design was used involving mediational set and order of presentation as between- subjects factors, and stimulus attribute as the within- subjects factor. Twenty—six subjects were run under the infralogical set and 27 under the logical set, the assign- ment to conditions being random. The subjects were tested as a group during a 1-1/2 hour period. Each subject worked independently. Concept Formation Lists Two-hundred and forty concrete and 240 abstract nouns were selected from a pool of 925 nouns (Paivio, Yuille & Madigan, 1968) scaled on abstractness-concreteness (g), imagery (I), and meaningfulness (m). The concrete nouns were selected on the basis of high g_and high I scores, while the abstract nouns were selected on the 13 14 basis of low Q and low I scores. On a seven point rating scale, the high I, high g_nouns selected ranged from 6.20 to 6.90 I (mean 6.49) and from 6.20 to 7.70 E_(mean 6.80). The low I, low §_nouns selected ranged from 1.63 to 4.10 I (mean 3.14) and from 1.28 to 4.14 9 (mean 2.60). The mean m values for the concrete and abstract lists were 6.80 and 4.96 respectively.1 The Thorndike-Lorge (1944) frequency for concrete nouns used in the present study ranged from 1—100 or more per million. (With AA and A words assigned values of 100 and 50, the mean E for concrete nouns was 43.46.) The §_for abstract nouns ranged from 0-100 (mean 5" 27.13).2 Five concrete lists of 24 nouns each were con— structed by random selection, without replacement, from the pool of 240 concrete nouns. The same procedure was followed with the abstract pool to get five abstract lists. 1No attempt was made to control for m Since Paivio, Yuille and Madigan (1968) have reported that items which are high on I are also high on m. This relationship is consistent With the view that abstract items derive their meaning largely through intraverbal experience. Further, Paivio (1966) has reported a correlation of .90 between I and m for 32 nouns; and Paivio, Smythe and Yuille (1968)— that- m is ineffective in PA learning. 2It should be noted also that correlational anal- yses have indicated that frequency was not a significant variable in either PA learning or free recall, the highest correlation (that between PA recall scores, and frequency values of stimulus membersT_being only 14 (Paivio, 1967). This would seem feasible Since Noble (1952) based his fre- quency measure on the hypothesis that frequency of occur- rence in written language would be highly correlated with m. 15 These lists were combined into a ten page booklet contain- ing one list per page. Each of the booklets contained a set of general instructions and a set of mediation in- structions. The mediation instructions for each booklet appeared before the first list and again before the fourth and the seventh lists to reinforce the instructional set. List presentation was counterbalanced as follows to control for warm-up and fatigue effects (gj represents a high I, high 9 list: Aj represents a low I, low g_list): grey-uses 51' 9.! ° ' ' 55' 95 Procedure Each subject received a test booklet containing ten concept formation lists. General instructions were then read aloud by the experimenter while each subject followed the reading. The subjects were required to read each list of 24 words, in consecutive order, beginning each list only when given the Signal to do so. After reading a list, the subjects were to pick out the words which they felt should "go together." The groupings were to be based on the mediation instructions given each sub- ject. Thus, those with logical mediation instructions were to form logical groupings; and the others were to form infralogical groupings. 16 To indicate each grouping formed, the subjects were to check a box next to each word selected. Six columns of boxes were provided for each list to allow the subject to group the words in each list Six different ways. At least nine words were to be included in each grouping the subject made. The same list appeared three times on its page to facilitate scanning from list to column (see Appendix A). Every time a subject finished marking a particular grouping column, he was to record a "time" measure at the bottom of that column. The time measure was a coded num- ber Shown on a screen and blackboard at the front of the room. The subjects were informed that they had seven minutes to work on each list (one page). Should a subject occasionally complete six groupings before the seven min- utes for that list had passed, he was to wait for the signal from the experimenter before going to to the next list. The subjects were allowed a three minute rest per- iod before the fourth and the seventh word lists to rest and to reinforce their particular mediation set. At the end of the testing period the subjects were asked to re- cord their sex and level of education on the back of the test booklet. 17 Time Measure The time measure was composed of two parts: (a) a two-digit number (01 to 80) projected on white paper taped to a front blackboard, and (b) a cycle number which counted the number of repetitions of the 01 to 80 sequence. The two digit number was presented by a Kodak 80-Slide Carousel projector. Each number of the series of 80 was shown for 15 seconds. The projector ran continuously providing 1/4 minute divisions of the entire session. The cycle number counted the cycles of 81 units (80 slides plus a space) so that each cycle took approximately 20-1/4 minutes. The cycle number was written on the blackboard by an assistant following the 01 to 80 series resulting in a three-or-more-digit number which appeared to be a whole unit. (For example, a 90 on the screen and a seven on the blackboard would be recorded as a time of 907.) Mediation Instructions Subjects under the infralogical mediation set (Grouping Principle II) were instructed to form groups of words on the basis of images or feelings which each list suggested to them. Their groupings were not to be based on verbal reactions to the words. Examples were presented in the instructions, e.g. FREEDOM, WOODS, SKY might be grouped together on the basis of a mental picture of being 18 away from others while in the woods looking up at the sky. FREEDOM, JUSTICE, PACT would probably not be grouped since these are verbal associates. It was emphasized that these subjects were to find groups based on sensory rather than verbal associations. Logical instruction set (Grouping Principle I) subjects were instructed to form word groups on the basis of other related words which each list suggested to them. The related words could be category names, or words which have the same associations or are associates of each other. The groups were not to be based on sensory reactions to the words. Thus, FREEDOM, JUSTICE, PACT might be grouped together on the basis of verbal association. FREEDOM, WOODS, SKY would probably not. It was emphasized that these subjects were to find groups based on verbal rather than sensory associations. RESULTS The initial step in the analysis of the data was to assess the effects of the three variables, mediation set (logical vs. infralogical), order of presentation (A list-g list vs. C list-A list), and stimulus-attribute (concrete vs. abstract) for each grouping across the ten lists. Because of the differences in defining latencies for groupings 1-6 (i.e. the assumptions that: (1) the subjects defined each group as being different from the previous ones, and (2) an expected increase in the level of difficulty with each additional grouping) an overall analysis of variance was not considered appropriate. However, an analysis of variance of mediation set by order of presentation by stimulus attribute was carried out separately for each grouping. Therefore, the data for each grouping (1-6) were each analyzed by a 2 X 2 X 2 analysis of variance with mediation set and order of presentation as between-subject factors, and with stimulus attribute as a within-subject factor. The mean latencies for the three factors by each grouping are presented in Table l.3 3Preliminary analyses Showed that there were no systematic differences in latencies between males and 19 20 mm.mm mm.mm ma.mm Hm.mm mh.mm mm.mm new: ma.mw mo.vm hm.mw vo.mm m>.>m mm.om «:0 m vm.vm mm.mm mm.mm mm.om nh.mv oo.am Uud hm.ao om.mm mn.mm mm.wm ma.mm ma.>m new: mo.mm mm.mm Ha.vm na.om mm.wm m>.mm «:0 m om.mw Hm.hm mm.mm ma.hm mm.mm vo.mm Ulfi mm.mm um.mm mm.hm mv.mm mo.mo hm.mm new: mh.mw mm.mm mm.nm mm.~o mm.mm Hm.mm «:0 v om.wm mv.mm mh.mm mm.mm mm.om vv.mm Uud mm.ow mm.mm mm.mm mH.mm nm.mo m>.mm com: om.mm oo.mm am.mm mm.mm ma.mm H>.mv «:0 m mw.mm om.>m wv.mm hm.mm mm.mm m>.hm Dim vo.nm m~.mo Hw.mm om.hm mm.mm mm.wm com: wm.mm me.wm mm.mv mm.mm Hm.mm Hm.nm «no N mm.mm mo.mm oo.mm mm.mm ew.am mv.mm Old nm.mm mv.H> oa.mm mo.mm ma.mm mo.mm new: om.am om.mm om.mm mm.aw mm.aw Hm.~m «no a mo.mh mo.mm oo.>o Hm.mm mm.mm mn.mm Old cam: uomupmnd mumnocou new: uomnpmnd mumuocoo coaumusmmmum mcflmsono Dmflq mo “mono mo Hmono Hasfiapm Hasfiflpm pom HMUHOOA pom HecamoamumcH .msflmsonm some now pom coauMHpmfi paw .coflumusmmmum mo umpno .musnflnuumumSHSEflpm an A.oomv mwflocmuma saws .H magma 21 Stimulus-Attribute The analysis yielded Significant main effects for stimulus-attribute in Grouping 1 (F( l, 45) = 5.90, p < .025), in Grouping 2 (F( l, 45) = 5.11, p < .05), and. marginally in Grouping 3 (F( l, 45) = 3.73, p < .10). The mean latencies for these data are presented in Table 2. Specifically, for these Groupings (l, 2 & 3) RIS were faster for lists with concrete stimuli than for those with abstract stimuli. Further, although the effect for stim- ulus-attribute were not significant for Groupings 4 and 5, the differences were in the expected direction. Grouping 6 Showed a marginally significant difference (F( 1, 34) = 3.06, p < .10) in the opposite direction. However, fur- ther inspection of the data indicates that no subject at- tempted a sixth grouping under the logical mediation, Afg_ presentation set for abstract lists. Order of Presentation The only significant effect for order of presenta- tion involved Grouping 6 (F( 1, 34) = 6.08, p < .025), females for any of the conditions. Males and females were therefore pooled in the analysis. It should also be noted that two logical set subjects were excluded from the data because the number of words in the groupings they formed (§'= 4) did not conform to the number required by the in- structional set. One additional logical and one infra- logical subject were also excluded. These subjects re- corded their "time" measures incorrectly. Thus the data analysis was performed on 25 infralogical and 24 logical subjects. 22 Table 2. Mean latencies (sec.) by stimulus attribute and order of grouping. Stimulus Attribute Grouping Concrete Abstract 1 62.61 68.28 2 55.24 60.29 3 56.13 62.91 4 61.63 63.51 5 56.97 60.96 6 58.68 63.13 23 indicating that RIS were faster for Afg presentations than for ng_presentationS. These results were disregarded however due to the fact that N = 0 for the one 579.condi- tion. All remaining groupings with the exception of Grouping 4 had consistent mean differences in the opposite direction (i.e. QfA orders yielded Shorter latencies than did A79 orders), however these differences (i.e. for Groupings 1-5) were not Significant. Mediation Set The main effect for mediation set was not signif- icant, although the mean latencies for infralogical sets tended to be faster than the mean latencies for logical sets with the exception of Groups 2 and 6. Mediation Set X Stimulus Attribute The predicted interaction of mediation set X stim- ulus attribute did not occur. Only a marginal Significant difference appeared in Grouping 4 (F( l, 44) = 3.49, p < .10) indicating that infralogical mediation yielded faster BIS for concrete than for abstract stimuli; while logical mediation yielded faster BIS for abstract than for con- crete stimuli. No consistent trends were found in the remaining data. 24 Order X Stimulus Attribute A main effect for order X stimulus attribute was found only for Grouping 6 (F( l, 45) = 10.73, p < .01) indicating that longer response latencies occurred for the type of list (concrete or abstract) presented first. No other trends were found for this interaction. Mediation Set X Order and the second order inter- action of Mediation Set X Order X Stimulus-Attribute were not significant. The second step in the analysis tested for the effects of lists (1-5 concrete, and 1-5 abstract) by grouping. To do this a 2 X 2 X 5 analysis of variance of mediation set, by order of presentation, by lists was performed for abstract lists and then concrete lists by order of grouping.4 Abstract Lists The difference between abstract lists was signif- icant for all groupings: Grouping 1 (F( 4, 152) = 9.91, p < .001), Grouping 2 (F( 4, 136) = 7.24, p < .001), Grouping 3 (F( 4, 116) = 2.98, p < .025), Grouping 4 (F( 4, 92) = 3.34, p < .025), Grouping 5 (F( 4, 75) = 4Grouping 6 was excluded from this analysis since it contained no members in its logical mediation/£7§_order cell. 25 3.89, p < .01) indicating a strong differential effect for lists. The mean latencies for these data presented in Table 3 suggest that the difference was due to a learning effect (i.e. BIS were faster for each successive list). Order of Presentation The main effect of order of presentation for ab- stract lists (i.e. positions 1, 3, 5, 7, 9 vs. positions 2, 4, 6, 8, 10) was significant for Grouping 1 (F( 1, 38) = 7.16, p < .025), and for Grouping 2 (F( l, 34) = 6.30, p < .025). Specifically, for these groupings, RIS to abstract lists were faster when the abstract lists fol- lowed concrete lists (i.e. positions 2, 4, 6, 8, 10) than when the reverse condition was true. The mean latencies for the remaining groupings were also in the same direc- tion, but the tendency was not significant. Mediation Set No Significant mediation set effect was found for abstract lists either as a main effect or in interaction with other factors with two exceptions (Grouping 3: Med- iation Set X Order X Lists interaction, (F( 4, 116) = 3.09, p < .05); and Grouping 1: Mediation Set X Lists interaction (F( 4, 152) = 1.61, p < .10). Specifically, differences in response latencies between abstract lists cannot be attributed to the mediation set manipulation. 26 Table 3. Mean latencies (sec.) by abstract lists and order of grouping. Lists Grouping l 2 3 4 5 1 93.75 71.75 70.63 57.00 62.51 2 82.47 63.78 58.71 48.31 46.61 3 67.29 63.13 58.13 53.75 51.88 4 68.30 61.36 52.17 52.48 50.44 5 72.22 47.44 52.19 48.41 48.19 27 Order X Abstract Lists The interaction of order X abstract lists was significant for Grouping 1 (F( 4, 152) = 3.42, p < .01), and Grouping 5 (F( 4, 76) = 2.22, p < .10), indicating that latencies for Afg_presentations took longer than those for ng_presentations combined with a differential effect of each successive list. However, since this ef- fect involves an interaction of five different lists for each grouping, the nature of this first order interaction is not entirely clear, nor is the second order interaction found for Grouping 3 (i.e. mediation set X order X lists). Concrete Lists The difference between concrete lists was signif- icant for two groupings: Grouping 1 (F( 4, 156) = 8.24, p < .001), and Grouping 3 (F( 4, 144) = 3.31, p < .025). Specifically, for these groupings (1 & 3), the data sug- gest a learning effect (i.e. RIS were faster for each successive list). No consistent trends were found in the remaining data for this factor. Order of Presentation The only main effect of order of presentation for abstract lists was a marginal effect for Grouping 2 (i.e. RIS to concrete lists were somewhat faster when the 28 concrete lists preceded abstract lists than when the re- verse condition was true, (F( l, 36) = 3.99, p < .10). Thus, the main effect of order of presentation was not a Significant factor in differences between concrete lists. Mediation Set A main effect of mediation set for concrete lists was found in two groups: for Grouping 4, BIS were margin- ally faster under infralogical mediation than under logical mediation (F( l, 31) = 3.26, p < .10). However, for Grouping 6, the opposite results were obtained (i.e. log- ical mediation yielded marginally faster BIS than did infralogical mediation, (F( 1, 15) = 3.84, p < .10)). Further, no consistent trends were observed in the remain- ing data. Order X Concrete Lists The interaction of order X concrete lists was significant for Grouping 1 (F( 4, 156) = 6.02, p < .001), and marginally for Grouping 5 (F( 4, 96) = 2.25, p < .10). Again, the nature of this first order interaction is not entirely clear due to the number of levels of lists in- volved, nor is the second order interaction found for Grouping 4 (i.e. Mediation Set X Order X Concrete Lists, (F( 4, 124) = 3.67, p < .01)). DISCUSSION The analyses of the latency data confirm the hypo- theses that grouping abstract stimuli generally requires longer response latencies than grouping concrete stimuli. However, the analyses failed to support the hypotheses that (a) the latency of discovery of infralogical media- tors is slower for abstract than for concrete stimuli, and (b) the availability and effectiveness of logical media- tors are relatively unaffected by variation of stimulus concreteness. No Significant interactions occurred for the mediation set X stimulus-attribute factor for any of the groupings with the exception of a marginal effect in Grouping 4; further, no consistent trends were evidenced in the latency data for this interaction. Major Hypotheses While the significant main effect of stimulus attribute does not confirm the applicability of two- process theory for the present research, the negative results (mediation X attribute) do not necessitate the rejection of the underlying theory. Since the availabil- ity hypothesis (i.e. that images and words are differen- tially available aS mediators for_abstract stimuli, and 29 30 are equally available for concrete stimuli) has been clearly supported by previous research (Yuille & Paivio, 1967), it would seem that alternative explanations for the present results may be apprOpriate. One probable explana- tion for the failure to demonstrate differential effects for the mediation set X stimulus attribute interaction is that the experimental procedure did not arouse persistent mediation strategies. This interpretation is supported by the suggestion that subjects apparently fail to use, or readily abandon, mediation strategies that are inapprOpriate for particular types of word pairs (Paivio & Yuille, 1967, 1969). In- dicatively, in the present study, one subject under the infralogical set reported that it was not possible to group the abstract words on the basis of sensory reactions. Yet the subject grouped these words and recorded reaction times for each grouping. Thus, although the subjects were assigned a mediation set and were given mediation set re- minders after the third and Sixth lists, this manipulation may not have been effective for maintaining the set over lists. If this explanation is true, then the latencies for the lists immediately following the mediation set reminders Should still conform to two-process prediction. Further inspection of the data, however, revealed no Sig- nificant difference in the mean BIS to concrete lists which immediately preceded the reminders, and the 31 differences obtained were in the opposite direction (mean latency under the infralogical set = 3.24; while latency under the logical set = 3.10). Further, contrary to pre— diction, mean latencies of the abstract lists tended to be longer for infralogical than for logical mediation sets (means = 3.47 and 3.19 respectively). Thus control of the subjects' mediational set, even for the lists immediately following the set, was not accomplished (i.e. Paivio & Yuille, 1967 have suggested that associative strategies are only partly controlled by the eXperimental sets and that, over trials, subjects increasingly revert to asso— ciative habits aroused by the semantic characteristics of the stimulus items). In grouping the words the subjects may have used different mediation strategies interchange- ably or one strategy irrespective of its congruence with the required set. Thus, subjects either did not fully understand the task required or they abandoned their med- iation sets as the strategies seemed inapprOpriate (e.g. the subject who reported that he could not group the words under the mediation set required felt free to utilize another strategy). Another alternative explanation for the lack of effectiveness of mediational set may be found in the con- sideration of task difficulty for the present study. Garner (1962) has suggested that the excess of stimuli over responses situation, which characterizes concept 32 learning, is achieved by making certain stimulus attributes irrelevant. Further, it has been found that increasing the number of irrelevant attributes has a markedly dele- terious effect on performance for this type of learning (Archer, Bourne & Brown, 1955; Bourne, 1963). Thus, the more irrelevant attributes there are, the harder it is to discover those attributes that are correlated with the correct response. Hence, the irrelevant attributes become the defining attributes of the concept. This result holds even when the irrelevant stimuli are correlated with one another (i.e. redundant, Bourne & Haygood, 1959), though when relevant attributes are redundant, performance is improved. In the present study, the subjects defined the number of relevant and irrelevant attributes for each list. It would seem, then, that a distinct possibility that the effects of the difficulty of the experimental task (i.e. based on the number of relevant and irrelevant attributes defined within each list) may have overshadowed any mediation X attribute effect. The subjects were essentially required to find a generalization among at least nine items in each list of 24 upon which the group- ing of these items could be based. If generalization be- tween the items in the list could be easily formed (i.e. a large number of relevant attributes defined), concept formation would be greatly facilitated. If generalization 33 was more difficult (i.e. a large number of irrelevant at- tributes existing), which seemed to be the case, formation of concepts would be likewise difficult. If this explana- tion is valid, then the relatively short latencies observed for grouping the stimulus words would seem to be indicative of the subjects' failure to follow their particular in- struction set. Instead the subjects may have grouped words on the basis whatever ideas came "naturally" or quickly. Minor Hypothesis The significance obtained for the stimulus attri- bute factor (i.e. that concrete words were generally grouped more rapidly than abstract words) is particularly interesting because the effect of abstractness has appar- ently not been investigated in studies of conceptual be- havior. It may be argued, however, that abstractness in the present study was confounded with familiarity since the abstract words were generally lower on §_(cf. Thorndike & Lorge, 1944) than were concrete words. However for reasons already discussed in the methods section of this paper (i.e. familiarity in relation to m), it does not appear that this finding can be attributed to differences in familiarity. REFERENCES REFERENCES Archer, E. J. On verbalizations and concepts. In A. W. Melton (Ed.), Categories of human learning. New York: Academic Press, 1964. Archer, E. J., Bourne, L. E., Jr., and Brown, F. G. Con- cept identification as a function of irrelevant information and instructions. Journal of Ex- perimental Psychology, 1955, 42, 153a164. Asch, S. E. A reformulation of the problem of associa- tions. American Psychologist, 1969, I4, 92-102. Bourne, L. E., Jr. Some factors affecting strategies used in problems of concept formation. American Journal Bourne, L. E., Jr. Human conceptual behavior. Boston: Allyn & Bacon, 1966. Bourne, L. E., Jr., and Haygood, R. C. The role of stim- ulus redundancy in concept identification. Journal of Experimental Psychology, 1959, 58, 232-238. Cofer, C. N. Experimental studies of the role of verbal processes in concept formation and problem solving. Annals, New York Academy of Science, 1960, gI, Deese, J., and Hulse, S. H. The PsychoIogy of Learning. New York: McGraw-HilI Book Company, 1967. Garner, W. R. Uncertainty_and structure asgpsychological concepts. New York: Wiley, 1962. Gelfand, 8. Effects of prior associations and task com- plexity upon the identification of concepts. Psy- chological Reports, 1958, 4, 567—574. Glaze, J. A. The association value of nonsense syllables. Journal of Genetic Psychology, 1928, 35, 255-267. Hull, C. L. Quantitative aspects of the evolution of con- cepts. Psychological Monographs, 1920, 28, No. 123. 34 35 Judson, A. I., and Cofer, C. N. Reasoning as an associa- tive process: I. "Direction" in a simple verbal problem. Psychological Reports, 1956, 2, 469—476. Miller, G. A. A psychological method to investigate verbal concepts. Journal of Mathematical Psychol- ogy, 1969, 6, 169—191. Nobel, C. E. An analysis of meaning. Psychological Review, 1952, 59, 421-430. Paivio, A. Latency of verbal associations and imagery to noun stimuli as a function of abstractness and generality. Canadian Journal of Psychology, 1966, I0, 378—387. Paivio, A. Mental imagery inassociative learning and memory. Psychological Review, 1969, 26, 241-263. Paivio, A., and Yuille, J. C. Mediation instructions and word attributes in paired-associate learning. ngchonomic Science, 1967, 8, 65-66. Paivio, A., Yuille, J. C., and Smythe, P. C. Concreteness, imagery, and meaningfulness values for 925 nouns. Journal of Experimental Psychology, 1968, 16 (1), pt. 2. Paivio, A., and Yuille, J. C. Changes in associative strategies and paired-associate learning over trails as a function of word imagery and type of learning set.) Journal of Experimental Psychology, 1969, 19, 458-463. Paivio, A., Yuille, J. C., and Smythe, P. C. Stimulus and response abstractness, imagery, and meaningfulness, and reported mediators in paired-associate learn- ing. Canadian Journal of Psychology, 1966, I0, 362-377. Reigel, K. F. The Michigan restricted norms. Report No. I, NIMH Grant MN 07619-01Al 07098, The Univ. of Michigan, 1965. Thorndike, E. L., and Lorge, I. The teacher's wordbook of 30,000 words. New York: Teachers C611ege, Bureau of Publications, 1944. Yuille, J. C., and Paivio, A. Latency of imaginal and verbal mediators as a function of stimulus and response concreteness-imagery. Journal of Ex- perimental Psychology, 1967, 15, 540-544. APPENDIX A WORD LISTS TOMAHAWK SALOON MOSQUITO BUTTERFLY THORN CHURCH CORPSE HURRICANE TOWER BLOSSOM SKULL PIPE TWEEZERS STEAMER HAMMER SHOES DOLLAR VALLEY BEGGAR DIAMOND STAGECOACH YACHT STRING CLAW time 36 LIST ____ TOMAHAWK SALOON MOSQUITO BUTTERFLY THORN CHURCH CORPSE HURRICANE TOWER BLOSSOM SKULL PIPE TWEEZERS STEAMER HAMMER SHOES DOLLAR VALLEY BEGGAR DIAMOND STAGECOACH YACHT STRING CLAW time TOMAHAWK SALOON MOSQUITO BUTTERFLY THORN CHURCH CORPSE HURRICANE TOWER BLOSSOM SKULL PIPE TWEEZERS STEAMER HAMMER SHOES DOLLAR VALLEY BEGGAR DIAMOND STAGECOACH YACHT STRING CLAW time EGO GENDER COMPARISON HINDRANCE INCLEMENCY WELFARE HEREDITY DEMOCRACY SOUL EXPLANATION DISPOSITION GREED ATTITUDE POSITION DECEIT EVIDENCE CAPACITY SUPPRESSION GIST AFTERLIFE MAJORITY HEARING ENTERPRISE DALLIANCE time 37 List EGO GENDER COMPARISON HINDRANCE INCLEMENCY WELFARE HEREDITY DEMOCRACY SOUL EXPLANATION DISPOSITION GREED ATTITUDE POSITION DECEIT EVIDENCE CAPACITY SUPPRESSION GIST AFTERLIFE MAJORITY HEARING ENTERPRISE DALLIANCE time EGO GENDER COMPARISON HINDRANCE INCLEMENCY WELFARE HEREDITY DEMOCRACY SOUL ‘ EXPLANATION DISPOSITION GREED ATTITUDE POSITION DECEIT EVIDENCE CAPACITY SUPPRESSION GIST AFTERLIFE MAJORITY HEARING ENTERPRISE DALLIANCE time EARTH HONEYCOMB BUTCHER JAIL LOBSTER WHALE INK WINDOW SWAMP FOREST DAFFODIL HOTEL BEAVER SCORPION CHIN FLOOD BOY TICKET HOOF ARM OFFICER INFANT FIRE CHAIR time 38 List EARTH HONEYCOMB BUTCHER JAIL LOBSTER WHALE INK WINDOW SWAMP FOREST DAFFODIL HOTEL BEAVER SCORPION CHIN FLOOD BOY TICKET HOOF ARM OFFICER INFANT FIRE CHAIR EARTH HONEYCOMB BUTCHER JAIL, LOBSTER WHALE INK WINDOW SWAMP FOREST DAFFODIL HOTEL BEAVER SCORPION CHIN FLOOD BOY TICKET HOOF ARM OFFICER INFANT FIRE CHAIR time TENDENCY PLEDGE JUSTICE CONCEPT SUPPLICATION DEDUCTION DISCONNECTION HEALTH THOUGHT BELIEF QUALITY MOMENT UNREALITY SCIENCE FORETHOUGHT PREVIEW DUTY VANITY ATROCITY DISTINCTION REPLACEMENT PROXY SHAME DEVELOPMENT time 39 List TENDENCY PLEDGE JUSTICE CONCEPT SUPPLICATION DEDUCTION DISCONNECTION HEALTH THOUGHT BELIEF QUALITY MOMENT UNREALITY SCIENCE FORETHOUGHT PREVIEW DUTY VANITY ATROCITY DISTINCTION REPLACEMENT PROXY SHAME DEVELOPMENT time TENDENCY PLEDGE JUSTICE CONCEPT SUPPLICATION DEDUCTION DISCONNECTION HEALTH THOUGHT BELIEF QUALITY MOMENT UNREALITY SCIENCE FORETHOUGHT PREVIEW DUTY VANITY ATROCITY DISTINCTION REPLACEMENT PROXY SHAME DEVELOPMENT time KETTLE POSTER MEAT ALLIGATOR VEHICLE GEESE SLUSH CANDY SEAT LAKE QUEEN PHOTOGRAPH COFFEE SUNBURN SUGAR RIVER JELLY HOUND APPLE MONEY CLOCK BULLET ROCK ALCOHOL time 40 List KETTLE POSTER MEAT ALLIGATOR VEHICLE GEESE SLUSH CANDY SEAT LAKE QUEEN PHOTOGRAPH COFFEE SUNBURN SUGAR RIVER JELLY HOUND APPLE MONEY CLOCK BULLET ROCK ALCOHOL time KETTLE POSTER MEAT ALLIGATOR VEHICLE GEESE SLUSH CANDY SEAT LAKE QUEEN PHOTOGRAPH COFFEE SUNBURN SUGAR RIVER JELLY HOUND APPLE MONEY CLOCK BULLET ROCK ALCOHOL time lm I. APPEARANCE AGREEMENT EXACTITUDE FORMATION SEMESTER HONOR CONFIDENCE MALICE VIRTUE CAUSALITY ARRAY PROFESSION FREEDOM BANALITY ILLUSION ORIGIN FALLACY HISTORY PACIFISM LOYALTY PERJURY INEBRIETY ABERRATION LIFE time 41 List APPEARANCE AGREEMENT EXACTITUDE FORMATION SEMESTER HONOR CONFIDENCE MALICE VIRTUE CAUSALITY ARRAY PROFESSION FREEDOM BANALITY ILLUSION ORIGIN FALLACY HISTORY PACIFISM LOYALTY PERJURY INEBRIETY ABERRATION LIFE time APPEARANCE AGREEMENT EXACTITUDE FORMATION SEMESTER HONOR CONFIDENCE MALICE VIRTUE CAUSALITY ARRAY PROFESSION FREEDOM BANALITY ILLUSION ORIGIN FALLACY HISTORY PACIFISM LOYALTY PERJURY INEBRIETY ABERRATION LIFE time FOX ELBOW BLISTER FUR HORSE COLLEGE AUTOMOBILE FIREPLACE CANE CATTLE BUILDING TANK HOSPITAL SPINACH PIANO LANDSCAPE LARK SUDS GRASS PAPER FLASK ACCORDIAN MULE KING time 42 List Fox ELBOW BLISTER FUR HORSE COLLEGE AUTOMOBILE FIREPLACE CANE CATTLE BUILDING TANK HOSPITAL SPINACH PIANO LANDSCAPE LARK SUDS GRASS PAPER FLASK ACCORDIAN MULE KING time FOX ELBOW BLISTER FUR HORSE COLLEGE AUTOMOBILE FIREPLACE CANE CATTLE BUILDING TANK HOSPITAL SPINACH PIANO LANDSCAPE LARK SUDS GRASS PAPER FLASK ACCORDIAN MULE KING time FOLLY NONSENSE HARDSHIP FACILITY MEMORY DISCRETION IMPOTENCY RATING OWNERSHIP APTITUDE SATIRE DIRECTION FRANCHISE FUNCTIONARY EXPRESSION MISCONCEPTION INCIDENT CENTENNIAL COOPERATION THEORY IMPULSE VIGILANCE ALLEGORY QUANTITY time 43 List FOLLY NONSENSE HARDSHIP FACILITY MEMORY DISCRETION IMPOTENCY RATING OWNERSHIP APTITUDE SATIRE DIRECTION FRANCHISE FUNCTIONARY EXPRESSION MISCONCEPTION INCIDENT CENTENNIAL COOPERATION THEORY IMPULSE VIGILANCE ALLEGORY QUANTITY time FOLLY NONSENSE HARDSHIP FACILITY MEMORY DISCRETION IMPOTENCY RATING OWNERSHIP APTITUDE SATIRE DIRECTION FRANCHISE FUNCTIONARY EXPRESSION MISCONCEPTION INCIDENT CENTENNIAL COOPERATION THEORY IMPULSE VIGILANCE ALLEGORY QUANTITY time MOSS PENCIL ANKLE DRESS WOMAN ENGINE DOOR LEOPARD MONARCH LIBRARY CRADLE BOWL MONK AMBULANCE ORCHESTRA MAST OCEAN UMBRELLA BOTTLE SNAKE BODY PHYSICIAN LEMON NUN time 44 List MOSS PENCIL ANKLE DRESS WOMAN ENGINE DOOR LEOPARD MONARCH LIBRARY CRADLE BOWL MONK AMBULANCE ORCHESTRA MAST OCEAN UMBRELLA BOTTLE SNAKE BODY PHYSICIAN LEMON NUN time MOSS PENCIL ANKLE DRESS WOMAN ENGINE DOOR LEOPARD MONARCH LIBRARY CRADLE BOWL MONK AMBULANCE ORCHESTRA MAST OCEAN UMBRELLA BOTTLE SNAKE BODY PHYSICIAN LEMON NUN time PROMOTION SAVANT FAULT ATTRIBUTE MERCY HATRED DEBACLE EMANCIPATION CONQUEST FATE MANAGEMENT LEGISLATION INGRATITUDE ABDICATION ECCENTRICITY BETRAYAL INTEREST SENSATION PREDICAMENT EXCUSE LAW COST EXCLUSION time 45 List PROMOTION SAVANT FAULT ATTRIBUTE MERCY HATRED DEBACLE EMANCIPATION CONQUEST FATE MANAGEMENT LEGISLATION INGRATITUDE ABDICATION ECCENTRICITY BETRAYAL INTEREST SENSATION PREDICAMENT EXCUSE LAW COST EXCLUSION time PROMOTION SAVANT FAULT ATTRIBUTE MERCY HATRED DEBACLE EMANCIPATION CONQUEST FATE MANAGEMENT LEGISLATION INGRATITUDE ABDICATION ECCENTRICITY BETRAYAL INTEREST SENSATION PREDICAMENT EXCUSE LAW COST EXCLUSION time APPENDIX B GENERAL INSTRUCTIONS 46 General Instructions This is a study of principles for grouping words. In this booklet you will find 10 different lists of words. Your task will be to pick out words yoo feel go together according to a grouping principle which you will be given. After you have looked over the words on a partic- ular list, ways of grouping the words will occur to you. Indicate your first grouping in the first column provided by checking (/) the words you wish to group. Your second grouping of the same list Should be marked in the second column, your third in the third column, etc. There is room for Six groupings on each page, but you may not think of that many. Just mark the groupings which occur to you most naturally. Some will be common, some original; some will come quickly, others only Slowly. Sometimes you may be able to describe the basis for your groupings; sometimes you may not. Mark each as it occurs to you. If there is a word on any list you don't know, just cross it out and ignore it. Try, however, to include at least 9 words in each groupingyyou make. Please turn to the sample list on the following page. 47 Sample List Notice that the check (/) marks in each column in- dicate one grouping of words. In the sample I1st, the words EPISODET‘MISCHIEF, CHILD, CAMP, MOTHER are in the first group. Words REMINDER, BIRD, PEPPER, CRISIS, ELEPHANT, CHANCE, EVENT are not. The six columns provide space for Six such groups. Also notice that the list of words is repeated three times on the page. This is Simply for your conven- 1ence. 1 2 3 4 5 6 EPISODE / EPISODE / / EPISODE / REMINDER REMINDER / REMINDER / MISCHIEF / MISCHIEF MISCHIEF BIRD / BIRD / BIRD / PEPPER PEPPER / PEPPER / CRISIS CRISIS / CRISIS / CHILD / / CHILD / CHILD / ELEPHANT / ELEPHANT / ELEPHANT / CHANCE CHANCE / CHANCE CAMP / CAMP CAMP EVENT EVENT ___JL_ EVENT h1___‘ MOTHER / / MOTHER MOTHER / time 221 Egg time 128 IIg time 6510 0115 Every time you finish marking a particular grouping column, please record the "time" at the bottom of that col- umn, as indicated in the sample. This is the number shown on the screen and blackboard at the front of the room. For example, a 90 on the screen and 7 on the blackboard would be a time of 907. Remember to record the "time" for each column. OK, the principle you Should use for grouping words is described on the next page. Different people have dif- ferent grouping principles. You will read your grouping principle now. You will also have a couple of chances to rest and refresh your memory on your grouping principle as you work through the booklet. Turn the page now and read your grouping principle to yourself. APPENDIX C MEDIATION SETS 48 Grouping Principle 1 There are at least two ways in which people react to a word. They may think of other related words (a verbal reaction), or they may think of a picture or image or feel- ing (a sensory reaction). In forming your groups of words, you are to use the first type of reaction as much as pos- sible. That is, form groups on the basis of the words the list suggests to you. These may be category names or words which have the same associations, or are associates of each other. The groupings should not be based on sensory reac- tions to the words. For exampIE: 1 2‘ I—F—T PINEAPPLE / / ORANGE BALL FREEDOM / WOODS JUSTICE / GRAPE / OBSESSION HINDRANCE PACT / THOUGHT SKY FREEDOM, JUSTICE, PACT might be grouped together on the basis of verbal association. FREEDOM, WOODS, SKY would probably not be grouped, since these would likely be grouped on the baSIS of a mental picture of being away from others while in the woods looking up at the sky, or on the emo- tional response "joy." ORANGE, PINEAPPLE, GRAPE might be grouped together on the basis being "fruit," a verbal category. ORANGE, BALL, GRAPE would not, if the basis for that grouping was "round." You are to findggroups based on verbal, rather than sensory, assoEIatIOns. Now turn to the next page, and look up. 49 Grouping Principle 2 There are at least two ways in which people react to a word. They may think of other related words (a verbal reaction), or they may think of a picture or image or feel- ing (a sensory reaction). In forming your groups of words, you are to use the second type of reaction as much as pos- sible. That is, form groups on the basis of the images or feelings the list suggests to you. The groupings should noE_be based on verbal reactions to the words. For example: 1 2 PINEAPPLE ORANGE / BALL / FREEDOM / WOODS / JUSTICE GRAPE / OBSESSION HINDRANCE PACT THOUGHT SKY / FREEDOM, WOODS, SKY might be grouped toqether on the basis of a mental picture of being away from others while in the woods looking up at the sky, or on the emo- tional response "joy." FREEDOM, JUSTICE, PACT would prob- ably noE_be grouped, since these are verbal associates. ORANGE, BALL, GRAPE might be grouped together on the basis of being "round." ORANGE, PINEAPPLE, GRAPE would not, if the basis for that grouping was "fruit," a verbal category. Form groups on the basis of emotion, feeling, image, location, sound, color, taste, size, odor, etc. . . . You are to find groups based on sensory, rather than verBal associaEIons. Now turn to the next page, and 100K— up. APPENDIX D INSTRUCTIONAL SET REMINDER 50 OK, remember, using your particular grouping prin- ciple, form groups containing at least 9 words. Cross out any words you don't know and ignore them. Record the time at the bottom of each column when you have finished that grouping. You'll have 7 minutes to work on each list. Should you occasionally get 6 groupings before the end of the time limit, wait for the signal before going on to the next list. APPENDIX E MEDIATION SET REMINDER 51 Reminder Grouping Principle 1 Remember, you are to find groups based on verbal (FREEDOM, JUSTICE, PACT), rather than sensory associations (FREEDOM, WOODS, SKY). Form groups on the basis of cate- gory names, or words which have the same association or are associates of each other. The groupings should not_be based on sensory reactions to words. 52 Reminder Grouping Principle 2 Remember, you are to find groups based on sensory (FREEDOM, WOODS, SKY), rather than verbal associations (FREEDOM, JUSTICE, PACT). Form groups on the basis of emotion, feeling, image, location, sound, color, taste, Size, odor, etc. The groupings Should not be based on verbal reactions to words. JUL 16 I87]