THE EFFECN 0F MWAER AND ENTRAPAIR ASSOQA'HON AND [NSTWCTEONS ON WAG; MKREMEMTEOH LEARNENG Tina: €09 “a Deqm a§ M. A. MICHIGAR STATE EINIVEESITY Eileen Eberiein 1967 u‘u‘\“a \ ‘ MlChi'I-r -.‘_ 1- 2 '- tC U'ifl's '. .1 _' '1 [35/ I THESIS ABSTRACT THE EFFECTS OF INTERPAIR AND INTRAPAIR ASSOCIATION AND INSTRUCTIONS ON VERBAL DISCRIMINATION LEARNING by Eileen Eberlein According to the predictions of the frequency theory of verbal discrimination (VD) learning, strong associations between the members of VD pairs should have the same effect as strong associa- tions between correct (C) and incorrect (1) items in different pairs. In both cases the frequency difference is predicted to decrease, creating interference. This experiment was designed to test this prediction and to determine the effects of instructions about the as- sociative relationships on VD performance. Five groups of 16 subjects each were given 10 trials on a 12-pair VD list. In the Paired (P) conditions the items in each VD pair were strong associates. In the Unpaired (U) conditions the C item of each pair was strongly associated with the I item of another pair. Half the subjects in the P and U groups received instructions about the associative relationships; half did not. The Control group received a list in which all of the items were unrelated. Eileen Eberlein . The performance of the U groups was significantly in- ferior to that of the Control and P groups, which did not differ sig- nificantly from each other. The instructions did not affect the total number of errors made. Though the interference found in the U con- ditions was consistent with the frequency theory, the lack of inter- ference in the P conditions was contrary to frequency theory predic- tions. The results were interpreted as supporting the hypothesis of Barch, Lippman, and Whalen (1967) concerning the memory processes involved in VD learning. Approved 4&244/(1 faq [6% Committee Chairman Date 4&2, 21?, W6] 7 THE EFFECTS OF INTERPAIR AND INTRAPAIR ASSOCIATION AND INSTRUCTIONS ON VERBAL DISCRIMINATION LEARNING By ‘ \ Eileen/,Eberle in A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of ' ' MASTER OF ARTS Department of Psychology 1967 ACKNOWLEDGMENTS The author gratefully acknowledges the assistance of Doctor David C. Raskin, without whose help this study would not have been possible. The author also is indebted to the other mem- bers of her guidance committee, Doctors Abram M. Barch and Gordon Wood, for their advice and encouragement. ii INTRODUCTION METHOD Subjects . Design Materials Procedure . RESULTS DISCUSSION . REFERENCES . APPENDICES TABLE OF CONTENTS iii Page 15 20 22 Figure 1. Table 1. Table 2. Table 3. LIST OF FIGURES AND TABLES Mean Errors per Trial for the Five Groups . Summary of Analysis of Variance Comparing the Mean Total Errors for the Five Groups Mean Total Errors and Standard Deviations for the Five Groups and the Results of the Duncan Multiple Range Test Summary of Analysis of Variance Comparing the Mean Total Errors for the Four Experimental Groups . iv Page 12 11 13 13 LIST OF APPENDICES Appendix Page A. LISTS . . . . . . . . . . . . . . . . . . . . 23 B. INSTRUCTIONS . . . . . . . . . . . . . . . 26 INTRODUCTION In the typical verbal discrimination (VD) task pairs of verbal items are presented to the subject, who is told that one item in each pair has been arbitrarily chosen as "correct" by the experi- menter. On each trial the subject chooses one member of each pair and is informed by the experimenter as to the correctness of his choice. The VD task has been used by several investigators to study the phases of the verbal learning process (Runquist 8: Freeman, 1960; Battig, Williams, & Williams, 1962). In these studies the VD task has been regarded as a simpler task which could be used to iso- late variables and processes involved in the more complex paired- associate (PA) and serial tasks. However, the results obtained in a number of these studies indicate that both the processes involved in VD learning and the relationship between VD and PA or serial learn- ing are more complex than was originally hypothesized. A number of conflicting results and complicating factors have been found, particularly in studies of transfer from VD to PA learning (Battig e_t_gl. , 1962; Spear, Ekstrand, & Underwood, 1964; Young 8: Underwood, 1954). One question concerning the processes involved in VD learning is whether any learning of the incorrect (I) item in a VD pair takes place independent of the knowledge or presence of the correct (C) item. Keppel (1966) and Battig _e_t_a_l. (1962) found that C items were remembered significantly better than I items. From an experiment involving transfer from one VD list to another, McClelland (1942) concluded that I items were not effectively learned independent of the C items. In a similar transfer experiment de- signed to answer this same question, Underwood, Jesse, and Ekstrand (1964) found initial positive transfer from one VD list to a second VD list in which the C items were the same as in the first list and also to a second list in which the I items were retained. How- ever, in later trials performance on the list in which the I items were retained improved slowly and was surpassed by the performance of control subjects. In order to explain these results, Underwood et al. developed a frequency-discrimination hypothesis, which has been further elaborated and has become the major theory concerning the processes involved in VD learning. According to this theory (Underwood 91.31: , 1964; Ekstrand, Wallace, & Underwood, 1966), the cue for discrimination in VD learn- ing is the subjective difference in frequency of occurrence of the C and I items within a pair. Frequency units are added to an item whenever a subject sees, pronounces, or rehearses the item. As trials progress, the C items accumulate more frequency units than the I items as a result of the subject' 3 rehearsal of the C item and the additional presentation of the C item in some feedback condi- tions. This increase in favor of the C item makes discrimination on the basis of differential frequency possible; the subject is able to choose either the most frequent (Rule 1) or least frequent (Rule 2) item in a pair. In order to demonstrate that increasing the difference in frequency between the items in a pair facilitates performance and decreasing it interferes with performance, Ekstrand §_t_a_l. (1966) performed an experiment using single lists. They found that when the same item was presented as the C item in two pairs, resulting in additional frequency units for the C items, VD learning was facil- itated. When the same item was the I item in two pairs, the I items received additional units, the frequency difference between the C and I items was decreased, and interference occurred as predicted. Even greater interference was found when the same item was the C item in one pair and the I item in another pair. This was in accord with the prediction that a conflict in the frequency rule used, as well as a decreased frequency difference, would occur in that list. A number of other studies have produced results which offer further support for the frequency theory. Raskin, Boice, Rubel, and Clark (in press) performed a transfer experiment designed to provide a more stringent test of the theory than the Underwood et a1. (1964) experiment, and their results supported the theory. Dominow- ski' s (1966) results concerning the effects of first trial guessing and type of feedback are consistent with the frequency theory, as are the findings of Erlebacher, Hill, and Wallace (1967) that C items are better retained than I items following VD learning. The studies of Lovelace (1964) and Kausler and Sardello (1967) also provide support for the frequency theory. Ekstrand eia_l. (1966) also hypothesized that the frequen- cies of items can be manipulated by presenting strong associates of the items. They proposed that frequency units may be transferred between highly associated items by means of the implicit associative reSponse (IAR). Their results supported this hypothesis. Though the effects were not as strong as those found with the use of identical items, the results showed that a list containing associated items as the C items in two pairs was easier to learn than a control list. A list in which an item was C in one pair and its associate was I in another pair was more difficult than the control list. Experiment 11 in the study of Raskin et al. (in press) also provides evidence that frequency units may be transferred by means of IARs. The second list in this transfer experiment contained associates of C items from the first list. When these associates were correct, positive transfer resulted; when they were incorrect, negative transfer occurred. How- ever, these differences were significant only when the subjects were informed about the associative relationships between the items in the two lists. None of these studies has dealt with the effect of associa- tion between the C and I items in the same pair. Ekstrand et al. (1966) predicted, however, that similarity or associative relations between the items in a pair would have the same effect as similarity or association between the C and I items of different pairs, 1. e. , interference would result. Underwood and Viterna (1951) studied the effects of intrapair meaningful similarity of VD learning when inter- pair similarity was minimized. They concluded that the slight disad- vantage for the high similarity list found in one comparison was due to differences in ability and that intrapair similarity had little effect on VD learning. Edwards (1966) used a four-choice VD task to study separately the effects of within-display and between-display similarity. He found that within-display similarity did not affect learning, while between-diSplay similarity caused significant interference. Several studies have compared the performance of normal and deaf adolescents on VD tasks in which intrapair similarity has been varied. Putnam, Iscoe, and Young (1962) found, contrary to frequency theory predictions, that highly meaningfully similar word pairs were easier to learn than unrelated pairs for both groups of subjects. The results of Youniss, Feil, and Furth (1965) showed no difference between similar and unrelated pairs for deaf adolescents, but greater difficulty with the similar pairs for normal adolescents and normal and deaf college students. On the basis of these results Barch, Lippman, and Whalen (1967) formed a hypothesis concerning the relation between level of linguistic development and the effects of intrapair similarity and association. They found that for children from the fourth to the eighth grade, pairs of highly associated items were easiest to learn, regardless of the age of the subject. These results are directly Opposed to the predictions of the frequency theory. The effects of intrapair association or similarity on VD learning are not yet clear. It was the purpose of the present experi- ment to test further the prediction of the frequency theory concerning intrapair associations, by comparing the effects of association between the C and I items in the same pairs with that of association between the C and I items of different pairs. The effect of instructing the subjects as to the presence and nature of these associations was also studied, since Raskin et al. (in press) found instructions to be of considerable importance in VD transfer, particularly in the transfer of frequency units along associative dimensions. METHOD Subjects. — Eighty-two undergraduate students enrolled in an introductory psychology course at Michigan State University served in the experiment. Two §S were discarded and replaced: one because he failed to understand the instructions, the other be- cause he failed to c00perate. The _S_s were assigned to the conditions in order of appearance at the laboratory, with males and females as- signed separately so as to have the same proportion of each sex in all groups. Design. — Five groups of 16 SS each were established on the basis of the associative relations between the items on the lists and whether or not they were instructed as to this relationship. In the Paired (P) conditions the items in a pair were highly associated. In the Paired-Instructed (P—I) condition the _S_s were instructed about these associations, while in the Paired-Not Instructed condition (P-NI) they were not. In the Unpaired (U) conditions the C items of each pair were highly associated with the I items of different pairs. These §S were also subdivided on the basis of instructions about these associative relationships into the Unpaired-Instructed (U-I) and Unpaired—Not Instructed (U-NI) groups. The Control group received a list in which the items were not related. Materials. — The associate pairs used in constructing the lists, obtained from the Russell and Jenkins (1954) and Bilodeau and Howell (1965) norms, were: queen - king, sweet - sour, high - low, hot - cold, hard - soft, tall - short, hammer - nail, sky - blue, eat - food, ice - cube, green - grass, and table - chair. The re- maining items, used in the unrelated lists, were chosen from the same sources. Each of the lists constructed consisted of 12 word-pairs. The VD pairs in the P lists were the 12 associate pairs listed above. The lists for the U conditions contained the same items, but the two members of each associate pair were presented in different VD pairs. One member of each associate pair was the C item in its VD pair, and the other member of the associate pair was the I item in its VD pair. The lists for the Control condition were constructed by pairing one member of each of the associate pairs with an unrelated item. For half of the Control group one item from each associate pair was used, and for the other half, the other item from each pair was used. For all the lists used in this experiment two sublists were constructed so that half the SS received one item from each pair as the C item, and the other half received the other item from each pair as the C 10 item. The lists are presented in Appendix A. Each list was pre- sented in two different random orders. Each item appeared once above and once below the other item on the two rotations. Procedure. — The lists were presented on a Stowe memory drum. Each pair of items, with one word typed above the other, was presented in the window for 2 sec. , and the C item from the pair was then presented alone for 2 sec. The intertrial interval consisted of a 2 sec. blank. All §s were seated at a table and were read identical instructions, which told them to learn which word in each pair was correct and encouraged them to guess on the first pre- sentation. The _S_s in the P-NI, U-NI and Control conditions received only these instructions. The P-1 and U-I §s received further instruc- tions which explained that the items on the list were associates and gave examples (items not used on any lists) of associated words. The P-I §s were told that each pair would consist of associated items. The U-I §S were told that the associates would not be presented to- gether and were given the information that if an item was correct in its pair, its associate would appear in another pair and would be in- correct, and vice versa. The instructions for all groups are pre- sented in Appendix B. All §s were run for 10 trials. RESULTS The performance of the five groups as measured by the mean number of errors made on the 10 trials is shown in Figure 1. After the first trial the performance of the U groups was consistently poorer than that of the P and Control groups. A one-way analysis of variance performed on the mean total errors for the five groups (Table 1) showed that the groups were significantly different, Table 1. -- Summary of Analysis of Variance Comparing the Mean Total Errors for the Five Groups. Source d. f. Mean Square _F_‘ EXper1.n?ental 4 247. 19 6. 01* Conditlons Error 75 41. 14 *B < . 01 FM, 75) = 6. 01, p < .001. A Duncan Multiple Range Test (Edwards, 1960) on these means (Table 2) indicated that the P and Control groups performed significantly better than the U groups (p < .05). 11 Mean Errors 12 0——0 Unpaired-Instructed ’——‘. Unpaired-Not Instructed A—-——-A Paired-Instructed H Paired-Not Instructed —————— Control Trials Figure 1. -- Mean Errors per Trial for the Five Groups. 13 Table 2. -- Mean Total Errors and Standard Deviations for the Five Groups and the Results of the Duncan Multiple Range Test. Group P-I Control P-NI U-I U-NI Mean SD Mean SD Mean SD Mean SD Mean SD 12.12 4.30 13.69 4.06 14.31 5.13 20.25 8.36 20.56 7.71 Note: Any two means underlined by the same line are not signifi- cantly different. None of the other differences was significant. To further assess the effects of the experimental conditions, a 2 X 2 analysis of variance was performed on the error scores of the four experimental groups (Table 3). This showed the main effect of Pairing of Associates to Table 3. -- Summary of Analysis of Variance Comparing the Mean Total Errors for the Four Experimental Groups. Source d. f. Mean Square _F_ Between 83 Pairing (P) 1 413. 28 18. 77** Instructions (1) 1 12. 50 .57 P X I 1 6. 10 .28 Error (b) 60 22. 02 **p < . 001 14 be highly significant, EU, 60) = 18.77, p < .01. The Pairing X Instructions were both nonsignificant (I: < 1). Thus, the lists in which the associates were paired were significantly easier for the S3 to learn than the lists in which the associates were presented in different pairs, and instructions to the SS did not significantly affect their total errors. It should be noted that the instructions given the U-1 group not only explained that the items would be associates, but also in- formed the SS that if an item was correct, its associate would be in- correct, and vice versa. This information could have been used on the first trial to determine the C item in any pair containing an asso- ciate of a word which had appeared earlier in the trial. On the lists used in the U-I group, 8 of the 12 pairs contained associates of items presented earlier, and _S_ could have responded correctly on these pairs on the basis of the information given in the instructions. How- ever, the mean number of errors on Trial 1 was 6. 75 for the U-1 group and 6. 88 for the U-NI group. Thus, the SS either ignored the instructions and simply guessed on Trial 1, or interference already was taking place and cancelled any advantage from the instructions. DISCUSSION The finding that strong associations between the C and I items in different pairs created significant interference is in accord with the frequency theory. This result is similar to that of Ekstrand i531. (1967) though their AB condition and the present U condition differed slightly: in the U condition each item in the VD pair was associated with an item in another pair while in Ekstrand 31a}. ' 3 AB condition one item from each pair was associated with another item and the other item in the pair was not related to any other item on the list. Contrary to frequency theory predictions, the results showed that strong associations between the members of a pair did not create interference. This finding is similar to the results of Underwood and Viterna (1951) concerning intrapair similarity and those of Edwards (1966) on between-display similarity. Instructions given to the §S about the associative relation— ships present in the list did not affect the total errors or interact with the Pairing variable. The effects produced by the presence of associ- ates in the list appeared without these instructions, as they did in the study of Ekstrand et al. (1966). Apparently instructions are necessary 15 16 for associations to affect learning only when the associated items are presented in two different lists (Raskin 31:31; , in press). The frequency theory cannot be used to explain the differ- ing effects of intrapair and interpair C and I item associations found in this experiment. According to the theory, whether the associates are in the same or different pairs, the RR to one item is likely to be confused with the IAR made to its associate, and the C to I frequency ratio will be decreased. Interference should result in either case. The results of this experiment seem to indicate that while a subject may use differential frequency as a cue in VD learning, something other than a simple discrimination on a frequency basis must take place. There is evidence that associations develop between the mem- bers of a VD pair. Battig fl. (1962) and Spear 3t_a_l. (1964) found that from 17% to 33% correct associations were made in recall tests and few subjects failed to produce any correct associations. Keppel (1966) found even greater evidence for the formation of associations in a recognition test following VD learning. Spear e2}. (1964) inferred that associations deve10ped chiefly as a result of the two units in a pair appearing contiguously in the VD task, and it has been assumed (Underwood gig. , 1964) that the VD task does not require the learning of associations between the items. The frequency theory does not deal at all with the l7 formation of intrapair associations or with any relations other than the development of a differential frequency between the items during VD learning. However, the fact that intrapair associations do form in VD learning may indicate that the relations which develop between members of a pair are an important part of the VD process. The model of VD learning proposed by Barch 31:31. (1967) takes this into account. According to their hypothesis S attempts to reduce the memory load by "tagging" or "coding" one member of each VD pair as correct and then collapsing the pair and this tag together for memory storage. When tested on the pair, the subject searches for and retrieves the collapsed unit, unfolds it, andinspects it to find the tagged C item. In this model the combining of the members of a pair is an important part of the process, and any relations conflicting with the intrapair relation interfere with the combining process. As a result, the tag for the correct item may be lost before the pair can be stored; or the members of the pair may have to be stored sepa- rately, leaving the tag highly susceptible to interference. The ease of combination for any pair depends on the strength of the intrapair relation as compared to alternative relations for the items. Thus, strong preexisting relations between the items of a pair should facilitate the combining process, while strong interpair relations should interfere with the collapsing of the items. 18 In their experiment with children Barch _e_t_§_l_. (1967) found that strong intrapair associations produced facilitation of VD learning. They predicted that college students would have more potential alternative links for each item, which would decrease the relative influence of the intrapair relation as compared to alterna- tive relations. Thus, strong intrapair associations would be ex- pected to produce less facilitation for college students than for younger children. Barch e_t_a_l. did predict that intrapair associa- tions would produce enough facilitation to counteract any tendency to confuse or interchange the items of a pair. Therefore, associated pairs would be at least as easy to learn as unrelated pairs for col- lege students. The present experiment indicates that a list with strong intrapair associations is as easy to learn as a list with unre- lated items. In accordance with the model of Barch §t_al. , a list containing associations between C and 1 items in different pairs was found to be more difficult to learn. If the present study had employed the alternating study—trial test-trial procedure used by Barch, e_t_a_l. (1967) rather than the anticipation method, the facilitation might have been more pronounced. The study—trial test-trial procedure would allow the hypothesized collapsing and unfolding processes to occur on separate trials. In the anticipation procedure both would have to 19 occur during the presentation of each pair, and the possibilities for interference in the memory processes would be greater. The Barch et a1. model does not account for all of the previous results which have been explained by the frequency theory. The transfer phenomena (Underwood £11., 1964; Raskin _e_t_a_l. , in press) as well as the finding that strong associations between C items in different pairs (Ekstrand L311. , 1966) facilitate learning, cannot be explained by differences in ease of combination of the pairs but it does appear that at least in cases in which discrimination on the basis of frequency breaks down, other factors make a great deal of difference in VD performance. REFERENCES Barch, A. M. , Lippman, M. , & Whalen, S. Intra-pair relations in verbal discrimination learning: A paradoxical find- ing. Paper read at Midwestern Psychological Associa- tion, Chicago, 1967. Battig, W. F., Williams, J. M., & Williams, J. G. Transfer from verbal discrimination to paired-associate learning. _J_. exp. Psychol., 1962, _6_3_, 258-268. Bilodeau, E. A. , & Howell, D. C. Free association norms by dis- crete and continued methods. Washington, D. C. : Office of Naval Research, 1965. Dominowski, R. L. First-trial guessing and verbal-discrimination learning. Psychon. Sci., 1966, _5, 231-233. Edwards, A. L. Experimental design in psychological research. New York: Rhinehart & Co. , 1960. Edwards, D. C. Similarity in verbal discrimination. Percept. Mot. Skills, 1966, 23, 815-820. Ekstrand, B. R., Wallace, W. P., & Underwood, B. J. A frequency theory of verbal-discrimination learning. Psychol. Rev. , 1966, _7_3, 566-578. Erlebacher, A., Hill, W. F., 8: Wallace, W. P. Differential accrual of frequency in verbal-discrimination learning. J. verbal Learn. verbal Behav., 1967, 6, 420-422. Kausler, D. H., & Sardello, R. J. Item recall in verbal-discrimination learning as related to pronunciation and degree of practice. Psychon. Sci., 1967, 1, 285-286. Keppel, G. Association by contiguity: Role of response availability. J. exp. Psychol., 1966, 11, 624-628. 20 21 Lovelace, E. A. Knowledge of correct and incorrect items in verbal discrimination. Psychon. Sci., 1964, _1_, 363-364. McClelland, D. C. Studies in verbal discrimination learning. II. Retention of reSponses to right and wrong words in a transfer situation. J. exp. Psychol., 1942, _3_1_, 149- 152. Putnam, V. , Iscoe, I., & Young, R. K. Verbal learning in the deaf. J. comp. physiol. Psychol., 1962, E, 843-846. Raskin, D. C., Boice, C., Rubel, E. W., & Clark, D. Transfer tests of the frequency theory of verbal discrimination learning. J. exp. Psychol., In press. Runquist, W. N. , & Freeman, M. Roles of association value and syllable familiarization in verbal discrimination learn- ing. J. exp. Psychol., 1960, 5_9, 396-401. Russell, W. A. , & Jenkins, J. J. The complete Minnesota norms for reSponses to 100 words from the Kent-Rosanoff Word Association Test. Technical Report No. 11, 1954, Contract N8-onr-66216, Office of Naval Research. Spear, N. E., Ekstrand, B. R., & Underwood, B. J. Association by contiguity. J. exp. Psychol., 1964, _6_7, 151-161. Underwood, B. J., Jesse, F., & Ekstrand, B. R. Knowledge of rights and wrongs in verbal-discrimination learning. J. verbal Learn. verbal Behav., 1964, _3_, 183-186. Underwood, B. J., & Viterna, R. 0. Studies of distributed practice: IV. The effect of similarity and rate of presentation in verbal discrimination learning. J. exp. Psychol., 1951, _4_2_, 296-299. Young, R. K., & Underwood, B. J. Transfer in verbal materials with dissimilar stimuli and response similarity varied. J. exp. Psychol., 1954, 47, 153-159. Youniss, J., Feil, R. N., & Furth, H. G. Discrimination of verbal material as a function of intrapair similarity in deaf and hearing subjects. J. educ. Psychol. , 1965, 56, 184-190. APPENDICES APPENDIX A LISTS 24 mfix :9 me; my hsom smmsv snow mmmmm undo :mmpm maze a a hogan: .393 a Boo em“: 38 SE a co“ m mmnw .mloM 6:3 s8 33 mm flaw new: 80% 22 a $6.3m 33 a :8 839 :8 oSS 908 be 808 film slow so: sow E a a a Madam :9 ms} 9556 a Goopm in: “madam: NE :9: 6350 wow poem ewe tom 23 snow Home: mzlswm 839 mafia hxm Boo «0: 3m . :86 mm mag a mmmnw a in: a 32 nmE @950 “ml“ coon mmm ”now new: msom mmumlsmm 3.30 633 6:3 New. 38 mm a 8:3 25 Molm. 89 :8 been: “0|: 00353 a Em 83 0.82 gm 2%: a somfiopm a “sums: mm mammo Hmm mnemmfio m poo.“ 8.853.: .Blom. and 3m :8 Wm. :2 g mo: “a fimno a 83 n83 :85 Em: :mmnm g soon: a :9 a so: a coca Miami 33 8pm mmmmlsm 300 Drama mmwhm mnow winfix. .38 ma EB 950m #50 a Sommog 63.3 hopfinonm a 2:15.... a seam? mafia oxoam P: as 13»an NMIm :0335 a 33:00 38 3% mmwum ones we} 32 team EB .39... «law ammo g 639 a £3: “:5 goddamn a case a new: 3 ha clouzlsflm a APPENDIX B INSTRUC TIONS 27 In this experiment, we are interested in the way peOple learn to discriminate between words. In the window on the machine a list of words will be presented. First a pair of words will be pre- sented for 2 seconds. Your task is to guess which word in the pair is correct. Since one word in each pair has been chosen arbitrarily as correct, you will be guessing at first. Please choose one of the words anyway. After the pair has been presented for 2 seconds the correct word will be shown to you for 2 seconds. As we go over the list again try to remember which word in each pair was correct. There are 12 of these pairs on the list, and we will go through the list 10 times. Please be sure to tell me which word in each pair you think is correct l_)_e_f_o_r_e_ the correct word appears alone in the list. Do you have any questions ? Paired-Instructed (P -I) Group: Each pair of words on this list is a pair of words which are associated with each other. By associated, we mean any kind of connection between the words. They may be synonyms like "ocean - sea"; opposites, like "man - woman"; or words that often appear together, like "salt - pepper. " 28 Unpaired-Instructed (U -I) Group: This list of words is made up of words which are associ- ated with each other. By associated, we mean any kind of connection between the words. They may be synonyms like "ocean - sea"; opposites, like "man - woman"; or words that often appear together, (like. ”salt - pepper. " These associated words are not paired with each other on the list. That is, one word will be in one pair and the word associated with it will be in another pair. But if one of the words is correct in one pair, say, "boy, " then when the pair containing "girl" appears, "girl" will be wrong. If "boy" is wrong, then "girl, " when it appears, will be correct. >- tr. ) u‘ yu‘ ‘ MICHIGIAN STATE UNIVERSITY LIBRARIES W I III III IIIIII II III 129 03071