ABSTRACT THE EFFECT OF SEMANTIC RELATEDNESS AND PRACTICE ON THE COLOR-WORD TEST by Burton L. Alperson A factorial experiment on 220 subjects determined the contribution of practice and semantic relatedness to the production of interference in the color—word test. Each subject was trained in a paired associate learning task to one of three sets of response terms, representing three levels of relatedness to the color naming task (Direct, Irrelevant, and Looking at the stimulus items). Each sub— Ject was also trained to one of three levels of practice (3, 10, or 50 trials). Stimulus items were then used as interfering materials in the color-word test. Analyses of time scores and errors both suggest that semantic relatedness is a more potent variable in the pro- duction of interference than is amount of practice. Amount of interference, in terms of time scores, increased over the course of the experimental interference task. Amount of interference decreased over the course of a standard interference task (one in which the interfering materials were actual color names rather than conditioned nonsense syllables).9 Burton L. Alperson Supplementary analyses failed to support the valid- ity of using tachistoscopic duration thresholds or extra- version scores to predict within subject interference effects. The results support Klein's theoretical analysis of the interference effect. The results also suggest that improvement of performance on the color-word test repre- sents an increase in ability rather than specific learn- ing effects. I’m- A W v '1.. ‘1',” *i’; “- --I—.'-l-I--I-r. Iguana-rad. _ .. r i' 1‘! 7'.‘J-.‘. I?'m_f.~.... I A--no. -r- v.9r9-" *'-V' ' r" .lifi—l W‘J 4 . Iii uranium-men Md' _‘w m?" -- v M‘— THE EFFECT OF SEMANTIC RELATEDNESS AND PRACTICE ON THE COLOR-WORD TEST f By Q0 Burton L. Alperson A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1967 DEDICATION _ to Phinque ii is -7/ -f mg , III. v'l. 4 ACKNOWLEDGMENTS The author would particularly like to thank his major professor, Dr. Paul Bakan, for his support and for contributing his sense of perSpective to the author's Gestalt of a number of areas, one of which is Psychology. Appreciation is also expressed to the writer's thesis committee: Dr. Charles Hanley, Dr. Donald M. Johnson, Dr. Bertram P. Karon, and Dr. William T. Stell- wagen for their helpful suggestions and criticisms con- cerning this study. Finally, the writer wishes to thank his wife, Erma, for her patience and understanding "above and beyond the call of duty." iii DEDICATION . . . ACKNOWLEDGMENTS LIST OF TABLES. . LIST OF FIGURES LIST OF APPENDICES Chapter I. INTRODUCTION II. STATEMENT OF THE PROBLEM . III. METHOD . . . Experimental Design . Subjects. . Procedure . IV. RESULTS. . . Time Data for Individual TABLE OF CONTENTS 0 Interference Within Task Interference Content Data . Tachistoscope Data Extraversion Data. V. DISCUSSION. Interference and Content Within Task Effects Tachistoscope Data Extraversion Data. iv Page ii . . . 15 Cards . . 15 Effects. Data. . . 311 . ... .3. nd' “linker! ’1. E Chapte r Page VI. CONCLUSIONS AND SUGGESTIONS FOR FURTHER RESEARCH . . . 39 REFERENCES. . . . . . . . . . . . . . ’42 APPENDICES . . . . . . . . . . . . . . “5 LIST OF TABLES Table 1. Experimental Design . . . . . . . 2. Summary of Means and Standard Deviations for All Dependent Variables . . . . 3. Means of Within Task Standard Interference for Extraverts and Introverts. . . . Page 16 32 LIST OF FIGURES Figure 1. Word Card Times as a Function of Semantic Relatedness and Amount of Practice . . 2. Experimental Interference as a Function of Semantic Relatedness and Amount of Practice. . . . . . . . . . . 3° Interference as a Function of Thirds of the Task. . . . . . . . . . . A. Mean "Syllable—Conflict" Errors as a Function of Semantic Relatedness and Amount of Practice . . . . . . . 5. Proportion of Subjects Classified as "Color-Direct" Responders in Each Semantic Relatedness Treatment . . . Page 18 20 23 27 3O Appendix A0 B. LIST OF APPENDICES Instructions for All Tasks Statistical Tests Raw Data I I O O O 0 viii Page 116 52 6A r11 ”194-5...“1...” CHAPTER I INTRODUCTION The Stroop color—word phenomenon is generally demon— L strated by presenting a subject with three cards in a fixed order. The first of these cards (the W card) con— sists of a series of color names printed in black ink. S's task is to read these names as fast as possible. This Q card is usually used as a warm up task; reading times on the W card are rarely analyzed. The second card (the C card) contains a series of patches of colored ink. S's task on this card is to name the colors as fast as possible. The final card (the CW card) consists of color names printed in conflicting colors, e.g., the word "red" might be printed in blue ink, "green" might be printed in red ink, etc. On this card, S must ignore the words and name the ink colors as fast as possible. The CW card seems to be universally Inore difficult than the C card. The difference between C and CW is usually taken as a measure of interference. This phenomenon has aroused considerable interest aUKDng psychologists. The interference effect is large and Stéible. Jensen (1965) found no reversals (i.e., CW slower them C) in testing H36 subjects. Bakan and Alperson (1967) fOLuid no reversals in testing 125 subjects. 1 rI‘he scores reveal highly reliable individual differ— ences (Jensen, 1965) which have been correlated with a broad range of events (cf. Jensen and Rohwer, 1966). Used as a measure of individual differences, the Stroop effect has successfully predicted retention of anxiety and non—anxiety materials (Ullman and Saltz, 1965), re- call of intentionally and incidentally learned concepts (Amster, 1965) and errors on a paragraph reproduction task under two motivational conditions (Lazarus, 1957), to name just a few. Jensen and Rohwer review these and a number of other similar studies in their 1966 review. In View of the variety of theoretical approaches assumed by investigators who have used the Stroop effect, it is unfortunate that very few studies have been concerned with analyses leading to increased understanding in dis- cussion of the effect, itself. Most investigators have used face validity as the only criterion of the relation- ship between the Stroop phenomenon and other experimental or theoretical variables (Jensen and Rohwer, 1966). ILazarus (1957), for example, views individual differences 51$ representing a preference for either conceptual or seensori—motor operations. The interference phenomenon is Lissed as an index of "cognitive control." Agnew and Agnew (3.963) feel that individual differences represent the alaility to maintain "narrowed attention." Schwartz and :Skuagass (1960) explain the interference effect by the use of “(magnitive rigidity" and Rorschach type "color—shock." Ullman and Saltz (1965) feel that the interference effect is inversely correlated with "cognitive differentiation." Stroop's original interpretation of the effect (1935a, 1935b) is that many responses are attached to color (e.g., touching, looking, naming, etc.) while only one response (saying) is attached to a word. Word read- ing, therefore, is a more highly practiced task than color naming (Stroop, 1938). The reading habit is stronger than the color naming habit so word reading interferes with color naming to a much greater extent than color naming interferes with color reading. Since 1938 relatively few studies have been directly concerned with the Stropp phenomenon, itself. Dalrymple— Alford and Budayr (1966) showed that the serial nature of the task is not a necessary condition for interference. These investigators used a tachistoscope to present indi- vidual color—word items to SS. The stimulus remained in view until S responded with the appropriate color name. They found that it took longer for bilingual (Arabic— English) subjects to name the ink color when the inter- fering stimulus was an Arabic or English color name than Inhen the interfering stimulus was a nonsense "squiggle." fPhey also found that serial presentation contributes to tzhe interference effect and that variation in the serial cxrder of items can produce variations in interference, when true standard Stroop cards are used. 7 u;- 'y .3 if; 31"; ‘ " ‘2 . . , "(‘31 Klein (196A) manipulated interference by varying the verbal text in which the colors are imbedded. He found increasing amounts of interference as the nature of the verbal text went from nonsense syllables to rare words, to common words, to words that imply color (lemon, grass), to color names different from the ink colors, to color names the same as the ink colors. He also found that allowing §S to say the word response before naming the color improved performance on the color naming task. Klein's interpretation of these results is that there are two factors which produce interference. The first is the semantic relationship of the interfering words to the color naming task. The closer the words to color in mean- ing, the greater the interference. The second component is the "attensity" or attention catching power of the word. His measure of attensity was frequency of occurrence of the word in English. These two components produce a response (saying the word) which competes with the color naming re- sponse for a single response channel. Thus, interference is produced and there is an increase in time necessary to name the colors on the CW card in comparison to the C card. The effect of attensity was later confirmed by Bakan 51nd Alperson (1967). In this study, verbal materials at fYDur levels of pronounceability were used as interfering sizimuli. Three groups of subjects were tested with non— seense syllables, each with a different level of pronounce— alaility and two groups of subjects were tested with words, each Vtith a different level of pronounceability. Two of the groups worked with material equated for pronounce- ability, but varying in meaningfulness (words and nonsense syllables). There was general support for the hypothesis that amount of interference increases with pronounceability. Of the two conditions equated for pronounceability but varying in meaningfulness (words and syllables), the inter- ference was greater for the more meaningful material. A tachistoscopic measure of the attensity of the stimuli showed a better relationship to interference than the mea— sure of pronounceability. Schiller (1966) administered a modified form of the Stroop test to students in grades 1, 2, 3, 5, 8 and college freshmen. He found little interference in the first and maximal interference in the second and third grades. Naming colors was faster than reading color names in the first grade. From the second grade on, the reverse was true. Some doubt is cast on the simple differential practice explanation of the interference effect since the difference between C and CW remained constant across all grade levels alter the first grade. Reading is increasingly more garacticed in comparison to color naming as the child ad- vsinces in grade level. If differential practice accounts fwar the Stroop effect, there should be greater interference iri the higher grades than in the lower grades. CHAPTER II STATEMENT OF THE PROBLEM Both the Klein and Schiller studies suggest that the simple differential practice explanation of the Stroop ef- fect is insufficient to account for the interference. The implication of the Klein study is that it is not general practice in reading, but rather it is the specific kind of practice which is important in the production of inter— ference. This interpretation is necessary to explain the effects of semantic relatedness in this study. There are, however, two problems in the interpretation of Klein's data. The first is that different words are used in each class. Although this variable probably accounts for very little variance, there remains a possibility that it does have an effect. It should, therefore, be controlled. The second problem is a more serious one. It is clear that both practice (if it can safely be assumed that practice :is highly correlated with word frequency and tachistoscopic digration threshold) and semantic relatedness influence aJnount of interference. All of the studies done on this ngcoblem (Klein, Schiller and Bakan and Alperson) have con— fkjunded these two variables. Words have been selected on 1| (1) .(1 L n the tflisis of their relatedness to color naming and this automatically fixes the level of practice for each word. For example, if the word "lemon" is selected as an inter— fering stimulus because of its relationship with the color naming task, no manipulation of practice is possible since the frequency of occurrence of the word "lemon" is already fixed. This problem makes it desirable to devise a study analagous to Klein's but one in which semantic relatedness and amount of practice may be independently manipulated. In such a study, a clearer indication of the relative contribution of practice and semantic relatedness may emerge. The approach employed in the present study is to take stimuli which are initially neutral with respect to the color naming task (i.e., nonsense syllables). These syllables are then paired with different semantic classes of words. The classes vary in their relatedness to the color meaning task. Thus, one group of subjects is given paired associate training on a set of nonsense syllables paried with color names. Another group of subjects is trained on the same nonsense syllables, but the syllables sire paired with a set of words which are irrelevant to ccalor naming. A third group is instructed to look at tide set of syllables, in order to control for the effects cyf‘ exposure to the stimuli. Practice is independently man1¥nllated by varying the number of training or looking trials. Following training, the nonsense syllables are used as the interfering stimuli on the CW card. An ad- vantage of this technique, in addition to the independent manipulation of practice and relatedness, is that the interfering stimuli for all groups are identical. Two supplementary variables are also included in this study. Bakan, Belton and Toth (1963) have shown that extraversion predicts within task decrement of performance in a vigilance task. It is of interest to see is a simi- lar relationship holds for within task Stroop performance. The second supplementary variable is tachistoscopic duration threshold of the conditioned syllables. This variable predicts interference when the interfering stimuli are real words (Bakan and Alperson, 1967). The present study attempts to assess the predictive validity of this variable when the interfering stimuli are conditioned non- sense syllables. CHAPTER III METHOD Experimental Design The experiment employed a factorial design with two independent variables (semantic relatedness and amount of practice) which assumed three levels each. These vari- ables represent the content and amount of paired associate training undergone by each S. The levels of semantic re— latedness were labeled "direct," "irrelevant" and "looking." The levels of practice were 3, 10 and 50 trials. An additional control group which received no paired associate training was added to the experiment. The basic design is, therefore, a 3 x 3 + l factorial (Winer, 1962, pp. 263-267) as represented in Table 1. TABLE 1.—-Experimenta1 design. Amount of Practice Semantic Control .Relatedness Trials 0 3 10 50 IDirect n=22 n=22 n=22 Jirrelevant n=22 n=22 n=22 Invoking n=22 n=22, n=22 n=22 10 Subjects Two hundred twenty Ss were unsystematically assigned to the 10 treatment groups. These Ss were students in Introductory and Advanced General Psychology courses at Michigan State University. They received course credit for their participation in experiments. Procedure Each S was tested individually in one 20—30 min. experimental session. An experimental session consisted of four tasks which were administered in the same sequence for all Ss. These tasks are described below. Task I: Tachistoscope Pretest Nine nonsense syllables were presented in an ascend- ing method of limits on a screen approximately 5 ft. in front of S. These syllables were shown on a Polymetric model V-1459—A projection tachistoscope. Each syllable was shown at a duration of 10 msec., and was repeated with a 10 msec. increment in duration on each repetition until S correctly identified the syllable. The first four sylla— bles (pim, fod, fet, and bot) were "warm-up" syllables. (These syllables were followed by three "experimental" ssyllables (dap, lar, and fon) and two "filler" syllables (rel and sog). S was not informed of the distinction be- ‘tvveen these syllables; he was simply told that he would be sluown a series of syllables (see Instructions, Appendix A). 11 The syllables were shown in the same order to all Ss. The room was not darkened for the presentation of the syllables and neutral density filters were used in front of the projector lenses to decrease the illumination. The use of the lighted room and the filters was an attempt to reduce figure—ground and thus to reduce the probability of having S see the syllable on the first (10 msec.) pre- sentation. A "pre—exposure" slide consisting of two verti- cal black lines which framed the area in which a stimulus would appear, was projected on the screen continuously both before and after stimulus presentation during this task. Task II: Paired—Associate Training Three pairs of items were presented in a 1:1 antici- pation method (Runquist, 1966) on a Stowe model A59B memory drum. Ss were required to say aloud both stimulus and re— sponse terms. All Ss had either 3, 10 or 50 trials of training, with a trial defined as one complete presentation of the three pairs of items. At each level of practice there were three levels of semantic relatedness as described below. All pairs were typed in black ink. Direct Conditioning.--The stimulus items dap, lar and fon, were paired with the response items red, blue and green, respectively. Irrelevant Conditioning.--The stimulus items dap, lar and fon were paired with the response items girl, boy and man, respectively. Looking.--Only the stimulus items dap, lar and fon were shown at a rate of one syllable every 2 sec. Ss were told to say each item aloud as soon as it appeared. Ss in the control group1 had no paired-associate training. They went directly from Task I to Task III. Task III: Color—Word Test This task consisted of four 9 x 12 cards presented in the same sequence to all Ss. The Word (W) card consisted of 80 color names (red, blue and green) printed in black ink, eight to a line. These words were randomly ordered with the restriction that no word could follow itself. Ss were instructed to read the names as fast as possible. The Color (C) Card was made up of 80 colored rectangu- lar patches, approximately .15" x .50", of red, green or blue ink. These patches were randomly ordered with the re- striction that no color could follow itself. S was told to name the colors as fast as possible. The Experimental Color Word (EXPCW) Card consisted of the three stimulus items used in the paired—associate training printed in red, blue or green ink. The order of ink colors was identical to the order of colors on the C card. Two restrictions were placed on the syllables. First, a syllable could not follow itself. Second, a 1It should be noted that the first 12 Ss in this group were run consecutively since the group was added after the study was in progress. l3 syllable could not appear in a color with which it had been paired in the Direct Conditioning groups (i.e., fon never appeared in green, lar never appeared in blue, and dap never appeared in red). There were 80 items on the card, eight to a line. S was told to ignore the words and name the colors in which the words were printed. The Standard Color Word (STDCW) Card consisted of the words red, blue and green, printed in conflicting colors of red, blue and green ink. The order of ink colors was identical to the order of ink colors on the C and EXPCW cards. The order of color names corresponds to the order of nonsense syllables on the EXPCW card. Ss were told that the instructions for this card were the same as the in- structions for the card they had just finished. Each card was placed on an easel which was a com- fortable reading distance from S. The complete instructions for this task may be found in Appendix A. The intended ef- fect of these instructions was to emphasize both speed and accuracy, but to place the primary emphasis on speed. Re— sponses on all four cards were tape recorded for later analysis. Task IV: Tachistoscope Posttest With the exception of the selection of warm-up syllables, this task was identical to Task I. Four new warm-up syllables (rec, ish, fid and dal) were employed here. 14 Extraversion Ninety—five of the Ss in this study had taken the Maudsley Personality inventory as part of their Intro— ductory Psychology course. Their scores on this test comprise the Extraversion variable in the present study. CHAPTER IV RESULTS Table 2 is a summary table consisting of the means of all dependent variables to be discussed in this section, along with their associated standard deviations. 55; Time Data for Individual Cards Time data were measured by listening to the tape recordings of S's performance on the four cards (W, C, EXPCW, STDCW), and marking times with an Esterline-Angus event recorder. Chart speed of the event recorder was .lO"/sec. and subsequent measurements on the records were rounded to the nearest tenth of an inch (i.e., to the nearest second). The first line was omitted from the time measurements of all cards; thus, the time measure— ments are on lines 2 through 10. Since some investigators have used individual card times instead of difference scores, separate analyses were performed for each card. No specific predictions were made for the W and C cards. The analyses of the CW cards, however, may be used to supplement the analyses for the difference scores. 15 Av.|'|||lllllll ' TABLE 2.--Summary of means and standard deviations for all dependent variables.1 Type of Conditioning: Direct Irrelevant Looking Trials: 3 10 50 3 10 50 3 10 Dependent Variable ' 50 Control W-Card x 27.19 26.82 26.27 28.00 27.50 26.09 29.00 27.05 26.92 27.23 s 3.73 2.91 3.56 9.29 3.79 9.61 9.92 9.59 3.22 3.01 C-Card X 39.86 35.95 37.18 39.91 39.99 39.73 36.00 36.00 36.86 35.00 S 9.76 5.81 9.98 5.70 9.55 6.78 5.99 6.13 9.82 3.79 EXPCW-Card X 92.82 93.32 96.55 90.50 91.00 93.09 93.55 92.73 92.82 92.18 s 6.06 7.35 7.16 7.00 9.83 9.96 6.76 6.59 5.65 5.23 STDCW-Card I 70.19 76.73 79.36 69.91 70.36 69.18 73.95 73.731 79.73 71.32 3 13.03 19.93 17.17 16.62 13.98 18.20 19.88 15.63 11.02 12.20 Experimental X 7.95 7.86 9.36 5.59 6.95 8.36 7.55 5.86 5.31 7.18 Interference S 9.87 9.05 . 9.82 3.92 3.31 6.98 5.03 3.93 3.97 3.85 Standard 2 35.27 91.27 92.18 35.00 35.82 39.95 39.86 36.86 37.23 36:32 Interference 3 12.18 16.90 16.15 13.19 12.87 19.90 9.75 12.09 9.91 10.70 Experimental X 1.68 2.73 2.86 1.59 1.50 2.59 1.95 1.50. 1.86 2.91 Interferencel s 2.10 2.39 2.19 2.02 1.91 2.15 2.06 1.87 2.01 1.76 Experimental 2 3.05 3.00 3.91 1.82 2.95 2.91 3.05 2.69 1.91 2.55 Interferencez S 2.92 1.77 1.89 1.92 1.53 3.11 3.11 1.99 1.89 2.09 Experimental X 3.23 2.19 2.82 2.18 2.50 3.32 2.55 1.73 2.05 2.23 Interference3 S 2.29 2.03 1.99 2.90 2.06 3.75 1.95 2.69 2.57 1.85 Standard X 11.82 15.55 19.95 12.69 12.27 12.91 13.82 11.91 19.18 13.00 Interferencel S 9.69 6.69 5.72 9.63 5.98 5.33 9.19 9.83 9.53 3.89 Standard x 11.86 12.68 13.95 11.55 12.36 11.91 13.36 12.77 12.59 12.32 Interference2 S 5.07 6.09 6.89 5.60 5.06 5.97 3.99 9.75 5.90 9.21 Standard x 11.59 13.05 13.27 10.82 11.18 10.69 12.68 12.18 10.95 10.55 Interference3 s 9.59 6.71 5.23 9.76 9.72 5.55 9.72 5.77 9.90 5.00 Syllable Conflict X .09 .27 .59 .09 .09 .19 .19 .09 0.00 .09 Errors S .93 .77 1.18 .92. .29 .97 .69 .29 0.00 .93 Tachistoscope x 9.86 10.18 10.68 10.90 9.72 9.18 10.22 10.95 9.95 8.09 PreTest S 9.51 5.39 9.92 3.89 2.39 9.52 9.55 6.01 3.59 3.36 Tachistoscope X 9.36 8.91 8.36 8.55 7.59 8.50 8.27 7.95 8.23 7.00 PostTest S 9.15 3.10 2.06 2.87 1.71 9.09 3.09 2.80 2.62 2.60 1 Each mean is based on 22 observations. m Card A 3 x 3 + 1 analysis of variance was performed on the time taken to read lines 2 through 10 of the W-card (Apendix B, Table l). The main effect for the amount of practice is reliable (F2,210 = 3.56, p < .05). Neither the semantic relatedness main effect nor the Relatedness by Practice interaction reached the .05 level. Figure 1 shows the means for the 10 groups in this analysis. This figure indicates that as amount of practice in paired associate learning and naming nonsense syllables increases, the time taken to read color names decreases. This re- lationship probably indicates nothing more than a warm up effect. C-Card An analysis of variance on the time taken to name the colors in lines 2 through 10 of the C—Card revealed no reliable main effects or interactions (Appendix B, Table 2). Thus there is no evidence in the present study for an ef- fect of either amount of practice or semantic relatedness on the speed of color naming. EXPCW-Card Analysis of the EXPCW—Card revealed no reliable main effects or interactions, although the main effect for relatedness approached significance (F2,210 = 2.61, .10 > p > .05; Appendix B, Table 3). ~——4 Direct Conditioning Mean Time in Seconds O———o Irrelevant Conditioning 26 —— *——-% Looking Control I l l l 3 10 Trials Figure l.--Word card times as a function of semantic re- latedness and amount of practice. re] COH Pel 00m fer Pele dept \\ W No main effect or interaction reached the .05 level in the analysis of the STDCW—Card. The main effect for relatedness, however, approached significance (F2 210 = 2.36, 3 .10 > p > .05; Appendix B, Table 9). Interference .\ The scoring formula chosen to represent interference in this study was CW-C. Jensen's (1965) factor analysis suggests that this formula is the "purest" interference 6} formula of the 16 proposed formulas he studied. '1 If the independent variables have any effect, they should be expressed in the experimental interference scores (EXPCW—C). An analysis of these scores reveals a reliable main effect (F = 9.05, p < .05) for related- 2,210 ness (Appendix B, Table 5). Neither the practice main ef- fect nor interaction are reliable. Figure 2 is a graph of the means for all groups on this variable. Scheffe's test (Winer, 1962, pp. 209—210) reveals no reliable difference for the looking vs. irrelevant condition comparison collapsed over trails. Direct conditioning is reliably different from looking and irrelevant conditioning combined (F = 7.59, p < .05 where F = 6.08). .05,2,200 A similar analysis was performed on the standard inter- ference scores (STDCW—C) to determine if the effect of relatedness was specific to the experimental interference dependent variable. This analysis revealed no significant 20 lo-w- r—-——OIM1ect Conditioning O——-43Irrelevant Conditioning 3 S9-t g X————7XLooking O o W Control a H an) 8 —-4- ,_ E H B . \ C.) C. ‘\ m '7-+- \ ///// CH $4 \ m 4.) s H 53 m (5" z I L 1 l I l 3 10 50 Trials Figure 2.--EXperimental interference as a function of semantic relatedness and amount of practice. main Effects or interactions (Appendix B, Table 6). Thus the effect of relatedness seems to be confined to the specific interfering items conditioned in the paired associate task. Within Task Interference Effects In addition to measuring total times on each card, the time necessary to complete each 1/3 of a card was also measured. Thus, C1 = the time necessary to name the colors in lines 2, 3, and 9 of the C-card; C2 = time necessary for lines 5, 6 and 7; and C3 = time necessary for lines 8, 9 and 10. Measurements were made in a similar fashion for the CW-cards. Interference scores were then calculated using the formulas: Experimental Interferencei = EXPCWi — C1 and Standard Interferencei = STDCWi — Ci This method of scoring yielded three experimental inter- ference scores and three standard interference scores for each S, i.e., one score for each third task. The analysis for these scores were two 3 x 3 x 3 (Relatedness x Practice x Thirds) analyses of variance with repeated measures on one Variable (Appendix B, Tables 7 and 8). The between subjects' portion of this 22 analYSis is only partially redundant with the previous analyses of total interference scores since the control group is omitted. The omission of the control group does not alter the interpretation of the previous analyses as the main effect for relatedness remains reliable in the experimental interference condition (F2,189 = 3.99, p < .05). The main effect for thirds of the task is reliable for both the experimental (F = 9.32, p < .05) and 2.378 standard (F = 9.03, p < .01) conditions. 2,378 Figure 3 shows the within subjects effect for both the standard and experimental interference conditions. The linear trend in this figure for standard inter- ference is significant (Fl,131 = 19.92, p < .0005), and the linear trend for experimental interference is also significant (F = 5.55, p < .05). The quadratic trend 1,131 for experimental interference approaches significance (F = 3.35, .10 > p > .05). In general, Figure 3 1,130 shows that standard interference declines as the task pro— gresses while experimental interference increases. Two within subject analyses of variance were per- formed on the control group alone. These analyses show the same within task effect as the analysis of the other groups in the standard interference condition (F2 42 = 9.39, 5 p < .01). However, there is no reliable within task effect for experimental interference in the control group (Appen— dix B, Tables 9 and 10). Mean Interference in Seconds 11.5-~ 11.0—— / j/ 300‘— 205—— 2.0n—- o-—~4OStandard Interference e————oExperimental Interference 105-b- ._ - 1 , ._ 1- i_ n J l I First Second Third Third Third Third Thirds of the Task Figure 3.--Interference as a function of thirds of the task.* *Note discontinuity of ordinate. Content Data Data on the types of overt errors made by Ss in the experimental interference condition were also gathered from Ss tape recorded performance. The scoring classifi- cation is analagous to the classification system origi- nated by Rand et al. (1963). Some modifications were necessary since the interfering items are nonsense sylla- bles rather than color names. The following categories were employed. lo "Syllable Conflict": If S responds by reading the syllable rather than naming the color of the syllable, he has made a syllable conflict error. If "dap" is printed in green and S says "dap" or "da...," he has made a syllable conflict error. "Color Direct": S says the name of a color which had been conditioned to the particular syllable in the direct conditioning groups. For example, "dap" was always paired with "red" in the direct conditioning groups. If the appropriate response to "dap" is "green" on the CW—card (i.e., "dap" is printed in green ink) and S says "red," this response would be scored as a color-direct error. Partial response, e.g., "re...," are also included in this category. 25 3. "External Color": This class includes any inappropriate color response which is not a color direct error if it is one of the three colors on the card. Considering "dap" printed in green, again, if S had responded "blue," he would have made an external color error. Partial responses are again included in the total. Four other types of errors, articulate utterances, inarticulate utterances, omissions, and irrelevant external colors (e.g., orange, brown, purple, etc.) were also scored. The frequencies of these errors were insufficient to permit analysis (raw data may be found in Appendix C). It was felt that independent judgments of error scores were unnecessary since these errors are quite obvious on the tapes. There is practically no ambiguity in select— ing the appropriate category for an error. The only ambiguous classification in this respect is the inarticu— late vocal utterance category. It is difficult, at times, to discriminate heavy breathing from a sigh. Syllable Conflict A 3 x 3 + 1 analysis of variance on syllable con— flict scores (Appendix B, Table 11) revealed a reliable main effect for relatedness (F = 3.98, p < .05). 2,210 ‘ueither the practice main effect nor the Practice by Relatedness interaction reached the .05 level. The mean number of syllable conflict errors for each group is shown in Figure 9. Scheffe's test for all comparisons within a logical grouping shows that the direct conditioning 50 trial mean is reliably different from all other means at at least 7.2:."3; p < .05 (Appendix B, Table 12). There is some question about the Relatedness x Practice interaction. The over- all F test reveals no reliable interaction while Scheffe's test suggests that practice interacts with relatedness on this dependent variable. Further research may resolve this ambiguity. This interaction is not crucial to the inter- pretation of this study. When collapsed over trials, the irrelevant and look- ing treatments are not significantly different from one another. Direct conditioning is significantly different from the other two treatments combined (F = 6.89, where F.05,2,200 = 6°08)' Color Direct vs. External Color "Color direct" and "external color" are not neces- sarily independent scoring categories. It may be that, in some cases, an error scored as a color direct error may actually be an external color error. Consider the consecutive items "dap" and "fon" which are printed in .55‘” ° ' Direct Conditioning .50’" O”"’"0 Irrelevant Conditioning .95‘—- X————* Looking .90nfl Control Mean Errors LA) 7 .25~— / / .20—— ,/ / r / .15'7“ ,/ .lOf— :2E:::::::::::gr”////’/’///o 7 .57 i i r 3 10 50 Trials Figure 9.--Mean "syllable conflict" errors as a function of semantic relatedness and amount of practice. 28 blue and red ink, respectively. If S responds to "dap" with "red," he would be scored as having committed a color direct error. Actually, S may have been anticipating the response for the next item ("red"), meaning that his error is not produced by the syllable "dap." With only three colors being used it is impossible to construct a card which is unambiguous in this respect and still meets the criteria of randomness used in the construction of stimulus materials in this study. Consequently, the analysis of these scores must be concerned with the relative prOportion of color direct and external color errors rather than mean number of errors of each type. Further, "color-direct" is a meaningless classification for the irrelevant conditioning and looking groups. The nonsense syllables were not conditioned to colors for these groups. Color direct scores in these groups, however, can be used to form a baseline against which to compare the direct conditioning groups. The appropriate question here is, "Does the direct conditioning group show a greater prOportion of color direct errors to external errors than do the other two groups?" Each S who made color direct and/or external color errors was classified as a color direct responder or an external color responder depending upon the dominant type of error made. Thus a S who made five color direct errors and four external color errors would be called a color 29 direct responder. Two frequency analyses were performed. S_had to make at least one dominant error to be included in one analysis and at least two dominant errors to be included in the other. Both analyses lead to the same conclusions. Figure 5 shows the proportion of SS classi- fied as color direct responders in the three relatedness conditions using the criteria of "one or more" and "two or more." In both the "one or more" (X: = 7.32, p < .05) and the "two or more (x; = 10.77, p < .005) there are reliable differences among groups (complete contingency tables may be found in Appendix B, Tables 13 and 19). Tachistoscppe Data Three tachistoscope variables were analyzed: the sum of thresholds of the conditioned syllables on the pre— test, the sum of thresholds of the conditioned syllables in the posttest and the pretest-posttest difference score (Appendix B, Tables 15, 16 and 17). It was expected that the practice variable would be reflected in either the posttest or the pretest-posttest difference scores. Correlation coefficients were then calculated between these two variables and experimental interference as a further test of the attensity variable. In the analyses for the pretest and posttest, only the comparison for "Control vs. all other Groups" proved reliable (PretestzFl,210 = 9.01, p < .05; Posttest: F = 9.99, p < .05). In both pretest and posttest 1,210 Percent of "color-direct" subjects in each group. "At least one" criterion. least two" criterion. A//'Direct Conditioning Irrelevant Conditioning 95 9O 35 Looking 3O 25 2O 15 10 N = 126 N = 81 Figure 5.--Pr0portion of subjects classified as "color-direct" responders in each semantic relatedness treatment. 31 the Control group displayed £2123 thresholds than the mean of all other groups combined (Table 2). There were no reliable effects in the analysis of the difference scores and none of the variables were reliably correlated with experimental interference. Extraversion Data Using the data from the Maudsley Personality inven- tory, two groups of Ss were formed for the purpose of analysis. These groups consisted of the lowest 99 and the highest 33 Maudsley Ss. A low score on this test repre- sents interversion while a high score represents extra- version. These groups were then compared in a 2 x 3 analysis of variance on their standard interference scores. Thus, the two dimensions in this analysis represent extra- version and within task interference (i.e., interference in thirds of the task). The cell means for the analysis are presented in Table 3. This analysis (Appendix B, Table 18) produced a reliable main effect for thirds of interference (F2,227 = 9.07, p < .05). Neither the extraversion main effect nor the Extraversion by Thirds interaction was reliable. The direction of the main effect for thirds is similar to the direction of the main effect for thirds for all SS combined (c.f. section on "Within Task Inter- ference"). 32 TABLE 3.—-Means of within task standard interference for extraverts and introverts. First Second Third Third Third Third Introvertsl 13.6 13.3 13.2 Extraverts2 19.3 12.3 12.2 1N = 99. 2N = 33. CHAPTER V DISCUSSION Interference and Content Data Without exception, the analyses performed in this study suggest that semantic relatedness is a more power- ful variable than practice in terms of the amount of interference produced. This conclusion is not particu- larly surprising in light of the data published by Klein (1969). The magnitude of differences in interference among nonsense syllables, rare words, and common words in his study is on the order of two to four seconds, while the difference in interference between common words and directly conflicting color names is about 25 seconds. Only two effects may be attributed to practice in the present study: (1) practice in reading and making verbal responses increases the speed of reading aloud, and (2) practice may interact with relatedness in the production of syllable conflict errors. As demonstrated by both Klein and Bakan and Alperson, there is a practice effect on interference Or at least an effect of word frequency and duration threshold, variables which should be analagous to practice. The failure of the present study to detect this effect probably means that the effect is too subtle for the paired associate technique which was used. 33 39 Klein's analysis of the reason for interference is that seeing a word causes a rise ". . . in excitatory level of [the] representational structure of the word, in part or whole, which includes reverberations to the word's motor facilitations within the schema" (p. 585). In other words, the word stimulus evokes not only an in- j creased tendency to say the word, but also an increased tendency to say other words which are in the same "associative network." The tendency to say these words competes with the tendency to say the appropriate color name (i.e., the color in which the word is printed) for the single vocal response channel. The content analyses of the present study strongly support this analysis. The groups showing the greatest amount of experimental inter- ference, the direct conditioning groups, also showed the greatest tendency to say both the interfering syllable ("syllable conflict" errors) and its associated response ("color-direct" errors). Within Task Effects The fact that standard interference declines through the task while experimental interference increases is an unexpected finding. Unfortunately, the design of this experiment is not appropriate for this comparison since the standard condition always followed the experimental condition. Although it is difficult to explain how a 35 secnlence effect could account for the direction of the trends, the possibility of such an effect cannot be ex— cluded. Apparently there are no published accounts of the trend within task interference. However, both Jensen (1965) and Smith and Nyman (1959) report that inter- 7 ference decreases with successive presentations of the same CW-card. These studies reduce the probability that 1 a sequence effect accounts for the trend of standard inter- ference. The trend of within task standard interference in this study clarifies the confusion on the nature of im- provement in color—word performance. Jensen and Rohwer (1966) point out that it is not clear whether improvement is a function of specific practice with a particular CW- card or if it represents an increase in whatever ability accounts for superior performance. The results of the present study suggest that an increase in ability accounts for at least some part of the improvement observed by these authors. Performance improves even though the order of reSponses varies unsystematically from line to line. The interpretation of increasing within task experi- mental interference is difficult. A reasonable inter— pretation is that naming colors on the EXPCW—card is a relatively easily mastered task (the magnitude of inter- ference here is about 7 sec. while on the STDCW-card 36 interference is about 37 sec.). It may be that having mastered this task in the early portion of the card. There is a temptation to ascribe this increment to a specific learning effect, since analysis of the control group did not show a reliable within task experimental interference effect. This temptation should be tempered by the relaization that the difference between the con— trol and conditioned groups may represent nothing other than a loss of power in the control group analysis, i.e., the use of 22 SS as opposed to the use of 198 SS. Tachistoscope Data It was expected that posttest thresholds of the conditioned syllables or pretest-posttest difference scores would Show a regular relationship with amount of practice. The failure of this study to find such a relationship may indicate that the tachistoscope technique is not sensitive enough to reflect changes based on about 5 minutes of paired associate learning. The relationship which has been found between duration threshold and word frequency presumably develops over years of familiarity with the language. Winnick and Nachbar (1967) have recently published a study similar to the present one. They trained three groups of subjects with a paired associate learning task (using real words) to a 50% learning criterion, a 100% criterion and a 150% criterion. These authors also failed 37 9C) find a reliable relationship between tachistoscopic duration threshold of the stimulus items and amount of practice. The Bakan and Alperson study relating duration threshold to interference was an independent groups de— sign. Consequently, this study did not deal with possible within subjects relations between these vari— ables. The function of the correlational analyses in the present study was to determine if such relations do, indeed, exist. In view of the insensitivity of the threshold scores to the practice variable, however, these correlations between interference and the threshold measurements are of questionable value. Since the difference between the control group and all other groups in this study exists in both the pretest and the posttest, it is reasonable to assume that this difference represents nothing more than sampling bias. It will be remembered that the first 12 SS of the control group were run in sequence after the study was in pro- gress. The order of testing Ss in this group could have easily allowed their observed scores to be biased either by sampling error or any systematic change which might have been present in the administration of the tachisto- scope task. 38 Extraversion Data Bakan, Belton and Toth (1963) have demonstrated that extraverts and normals display a greater performance decrement than introverts in a vigilance task. The pre— sent study was an attempt to see if a similar relation- ship holds for Stroop Interference. The fact that the present study failed to find such a relationship may reflect the considerable difference in the statistical power of these two studies. The Bakan g£_§;. study used 155 subjects while the present study used only 77. A study using a larger sample might be a better test of the relationship between Stroop performance and Extraversion. CHAPTER VI CONCLUSIONS AND SUGGESTIONS FOR FURTHER RESEARCH Three conclusions may be drawn from the results of the present study. 1. Although the nature of the relationship be- tween practice and semantic relatedness has not been clarified by this study, it is clear that semantic relatedness is the more potent of the two variables in the production of interference. As Klein's theoretical analysis suggests, both specific effects of the inferfering stimulus and effects due to associative connections of this stimulus are implicated in the production of interference. Some authors have noted an improvement in Stroop performance with repeated testing. The within task decrement in interference found in this study suggests that at least some part of this improvement may be accounted for by an in- crease in ability, as Opposed to Specific practice effects. 39 , .1- .3; 90 The fact that within task standard and experimental interference display opposite trends is a finding which deserves further research. The possibility that implicit rehearsal accounts for the trend of experimental inter— ference might be explored by simply making it clear to a group of subjects that they will not have to recall the 5 paired associates. If this technique reverses or elimi— nates the trend, then the finding is trivial. On the other hand, if the opposition of trends turns out to be specifically related to paired associate learning, such a finding would have broad implications for verbal learn- ing in general. Comparisons of laboratory production of verbal events with the effects produced by "real" verbal units are not often made in psychological studies. Such comparisons should prove fruitful in assessing the limits generalization from the laboratory to the real world. A more powerful study of the effects of extraversion on within task interference is called for. A study coupling a larger number of subjects with more extreme criteria for assigning subjects to groups might clarify the effect of this variable as it did in the Bakan g£_§l. study. Finally it should be pointed out that the Klein study, the Bakan and Alperson study, and the present study all avoid the central question concerning the nature of interference. It is easy to imagine that variables such as attensity, response—competition for a single vocal 91 Chaxnlel, and increase in the tendency to say inappropri- ate responses account for differences in interference. However, none of these variables can account for the tremendous difference in magnitude of interference when the interfering stimuli are color related as opposed to when the interfering stimuli are irrelevant to the color naming task. In other words, why is it so much more difficult for a subject to overcome response competition when the competing responses are similar than when they are dissimilar? A promising direction for research on this question lies in the use of content analyses of errors made by subjects. Comparison of different types of errors pro- duced by different interfering stimuli may shed some light on this problem. REFERENCES 92 '9 REFERENCES Amster, H. The relation between intentional and inci- dental concept learning as a function of type of multiple stimulation and cognitive style. J. Pers. Soc. Psychol., 1965, S, 217-223. Agnew, N. and Agnew, M. Drivelevel effects on tasks ‘~, of narrow and broad attention. Quart. Jg'exE. 43' Psychol., 1963, SS, 58—62. Bakan, P. and Alperson, B. Pronounceability, attensity, and interference in the color-word test. Amer.'J. Psychol., (Sept.) 1967. Bakan, P., Belton, J. A. and Toth, J. C. Extraversion- introversion and decrement in an auditory vigilance task. In Bruckner, D. N. and McGrath, J. J. Vigilance: A Symposium, 1963, New York: McGraw— Hill, 22-27. Dalrymple—Alford, E. C. and Budayr, B. Examination of some aspects of the Stroop Color-Word Test. Percept. mot. Skills, 1966, SS, 1211-1219. Jensen, A. R. Scoring the StrOOp Test. Acta Psychol., Amsterdam, 1965, S5, 398-908. Jensen, A. R. & Rohwer, W. D., Jr. The Stroop Color- Word Test: A review. Acta Psychol., Amsterdam, 1966, SS, 36—93. Klein, G. S. Semantic power measured through interference of words with color naming. Amer. J. Psychol., 1969, 11, 576—588. Lazarus, R. S. Motivation and personality in psychological stress. Psychol. Newsltr., 1957, S, 1959—193. Rand, G., Wapner, H., McFarland, J. H. Age differences in performance on the Stroop Color-Word Test. J. Personal., 1963, 3;, 539-558. 93 99 ‘39nfiplist, W. N. Verbal Behavior. In Sidowski, J. B. E§perimenta1 methods and instrumentation in Psy- chology, 1966, New York: McGraw-Hill, 987-591. Schiller, P. H. Developmental study of color-word interference. J. exp. Psychol., 1966,‘1S, 105-108. Smith, G. J. W. and Nyman, G. E. Psychopathic behavior in a serial experiment. Lunds U. ArSskr.,'Avd.‘2, 1959. 56. StrOOp, J. R. Studies in interference in serial verbal reactions. J. exp. Psychol., 1935. $9, 693-661. Stroop, J. R. The basis of Ligon's theory. Amer. J. Psychol., 1935, 91, 999-509. Stroop, J. R. Factors affecting speed in serial verbal reactions. Psychol. Monogr., 1938, SS, no. 5, 38-1980 Schwartz, M. and Shagass, C. Responses to colored and conflict-inducing stimuli in a psychiatric pOpu— lation. Percept. mot. Skills, l960,‘l£, 295-252. Ullman, L. P. and Saltz, E. Retention of anxiety material as a function of cognitive differentiation. J. Person. soc. Psychol., 1965, 1, 55-62. Winer, B. J. Statistical principles in experimental design. New York: McGraw-Hill, 1962. Winnick, W. A. and Nachbar, S. Tachistoscopic recognition thresholds following paired-associate learning. J. verb. Learn. verb. Behav., 1967, S, 95-100. uni ‘ue:.< .-.‘.-‘.o . “Ad-flit I v hue-r I'.. .0". V" ’ 'w v .- ' HG'C _' _'- Ac-O"”—'Qf " . I ' . ,fi, ' . - ‘ a I ”,1... '3 0 come“ . *1” f, APPENDICES 95 , .‘. J 't 5*... <7 .- 1:1-'- .... Vlavrwv—w. v—vv 68.4): _ :11wa , f ("9.1. . ..: ‘- APPENDIX A ' INSTRUCTIONS FOR ALL TASKS 96 .. H80“, 2...! .l. a- .. v I . LR. 97 T—Scope InstructiOns This part of the experiment is a test of how keen your vision is when a word is briefly flashed before your eyes. When you press this button, a three letter non- sense syllable will appear very briefly on the screen, between the two vertical black lines. When I say "NOW," press the button firmly and then release it. After each presentation, tell me what you have seen by both spelling and pronouncing the syllable. If you see only part of the syllable, spell whatever you have seen. There is no objection to guessing, but if you have absolutely no ideas of what was presented, please say the word "nothing." The first presentations will be very brief and you will probably be unable to recognize the syllable. Each presentation will be slightly longer than the one before it. This will continue until you have correctly recognized the syllable. We will then move on to another syllable. I will notify you before I change the syllable which is being flashed. Do you have any questions? -,-‘. _ -Q—r I "M- q. -: ~o a‘- ”I. .. 98 Paired-Associate Instructions This part of the experiment is a learning eXperi- ment in which you will learn to associate nonsense sylla- bles and words. It is very important that you follow the instructions to the best of your ability. Should you fail to follow any instruction, be sure to tell me since the interpretation of the results may be affected. The list will consist of three pairs of items like the pair in this window. When we begin, the nonsense syllable will always appear in the window alone, while the word is covered by a piece of metal called a shutter. After a short time, the shutter will lift and reveal the word. Your task is to associate or connect the word with the nonsense syllable so that you will be able to say both the word and syllable while the syllable alone is in the window, that is, before the shutter goes up. Since the order in which the pairs follow each other will not always be the same, you must learn these pairs as pairs and not in the particular order in which the pairs follow each other. Always try to anticipate the word just after the sylla- ble has appeared. If you are able to say the syllable and word before the shutter goes up, I will count it as correct; on the other hand, if you say nothing or say the syllable or word after the shutter goes up, I will count it as incorrect. Do you have any questions? . .. tibia 99 W—Card Instructions For this part of the experiment I will give you a page with color names printed on it. When I tell you to begin, read the color names aloud. Please read rapidly as I will be timing you. If you make a mistake, please correct it before going on but remember you are working for speed. Read the page from left to right as though you were reading the page of a book. Do not pause at the end of lines as you are being times on the whole page rather than for individual lines. Please do not point to the words you are reading, and do not use a singsong voice. Read the names as fast as you possibly can. The faster you can read the names, the better your score will be. When you finish the whole page say the word "Stop." Do you have any questions? SO C-Card Instructions This part of the eXperiment is similar to the pre- ceding part, except that the page contains a series of patches of color rather than words. You are to name the colors from left to right after I tell you to begin. Again, do not pause at the end of lines; work through the entire page. I will be timing you again so remember to work for speed as well as accuracy. If you make a mis- take, please correct it before going on, but remember you are working for speed. Name the colors as fast as you possibly can. The faster you can name them, the better your score will be. When you reach the end of the page, say the word "StOp." Do you have any questions? 51 CW—Card Ingtructiogg On this page you will find a series of words which are printed in different colors. Your task is to ignore the words and name the colors in which the words are printed. If, for example, the word, "Door" was printed in yellow and the word "Chair" was printed in red, you would say "yellow, red" and so on. Again, do not pause at the end of lines; name all of the colors on the page before stopping. There are certain rules we would like you to follow. You are to name the colors one by one. Do not squint or de—focus your eyes to blur the words; do not point, and do not use a singsong voice. If you make a mistake, please correct it before going on again remembering you are work- ing for speed. When you finish the page say the word "Stop." Remember, I will be timing you again, so work for speed as well as accuracy. Name the colors as fast as you possibly can. The faster you can name the colors, the better your score will be. Do you have any questions? APPENDIX B STATISTICAL TESTS 52 53 TABLE l.-—Ana1ysis of variance of the word (W) card. Source SS df MS F Control vs. All Others .196 l .196 < l Relatedness (R) 18.99 2 9.295 < 1 n Practice (P) 105.52 2 52.760 3.56* R x P 26.21 9 6.553 < 1 Within 3111.68 210 19.818 *p < .05. ' TABLE 2.--Analysis of variance of the color patch (C) card. Source SS df MS F Control vs. All Others 12.193 1 12.193 < l Relatedness (R) 190.37 2 70.185 2.503 Practice (P) 51.07 2 25.535 < 1 R x P 39.17 9 9.793 < 1 Within 5889.05 210 28.09 59 TABLE 3.—-Analysis of variance of the experimental color word (EXPCW) card. Source SS df MS F Control vs. All Others 11.07 1 11.07 < l Relatedness (R) 291.09 2 120.52 2.61 Practice (P) 198.01 2 79.01 1.60 R x P 129.05 9 31.01 < 1 Within 9699.19 210 96.19 TABLE 9.-—Ana1ysis of variance of the standard color word (STDCW) card. Source SS df MS F Control vs. All Others 69.99 1 69.99 < l Relatedness (R) 1133.59 2 566.77 2.36 Practice (P) 338.59 2 169.27 < l R x P 683.10 9 170.78 < 1 Within 50930.68 210 290.15 55 TABLE 5.——Analysis of variance for experimental infer- ference. (Experimental Color Word Card minus Color Patch Card). Source SS df MS F Control vs. All Others .02 l .02 < 1 Relatedness (R) 169.93 2 82.22 9.05* Practice (P) 31.90 2 15.70 < l R x P 197.60 9 36.90 1.82 Within 9262.59 210 20.30 *p < .05. TABLE 6.--Ana1ysis of variance for standard interference. (Standard Color Word Card minus Color Patch Card). Source SS df MS F Control vs. All Others 30.10 1 30.10 < 1 Relatedness (R) 682.93 2 391.22 2.03 Practice (P) 69.61 2 39.81 < l R X P 689.18 9 172.30 1.03 Within 35232.59 210 167.77 56 TABLE 7.—-Ana1ysis of variance for within task standard minus Color interference. (Standard Color Word Third Patch Third). Source SS df MS F Between Subjects 11923.66 121 Relatedness 9 (R) 227.98 2 113.79 .97 " Practice (P) 23.20 2 11.60 1 R x P 229.73 9 57.93 1 Subj. w. Gr. [error (be— ,1 tween)] 10993.25 189 57.90 Within Subjects 5200.67 326 Thirds (T) 228.83 2 119.92 .03* R x T 19.96 9 9.86 1 P x T 38.13 9 9.53 l R x P x T 120.95 8 15.05 .19 T x Subj. W. Gr. [error (with— in)] 9793.80 378 12.68 *p < .01. 57 TABLE 8.--Ana1ysis of variance within task experimental interference. (Experimental Color Word Third minus Color Patch Third). Source SS df MS F Between Subjects 1969.85 ;_1 Relatedness I) (R) 50.16 2 25.08 3_u9* 1 Practice (P) 8.19 2 9.07 < 1 f R x P 96.98 9 11.75 1.63 } I /; Subjects Within Gr. [error (between)] 1359.57 189 7.19 Within Subjects 1635.67 3_6 Thirds (T) 39.99 2 17.50 9.32* R X T 19.09 9 3.51 < 1 P x T 26.78 9 6.70 1.65 R x P x T 30.79 8 3.85 < l T x Subj. w. Gr. [error (Within)] 1529.07 378 9.05 *p < .05. 58 TABLE 9.——Ana1ysis of variance of within task standard interference (control group only). Source SS df MS F Between people 1091.53 21 Within people 237.39 99 Thirds 70.69 2 35.32 8.90* Residual 166.70 92 3.97 Total 1278.87 65 *p < .01. TABLE 10.—-Ana1ysis of variance for within task experimental interference (control group only). Source SS df MS F Between people 103.76 21 Within people 122.00 99 Thirds 1.12 2 .56 < 1 Residual 120.88 92 2.88 Total 225.76 65 59 TABLE 11.-—Ana1ysis of variance for syllable-conflict errors. Source SS df MS F Control vs. All Others .119 1 .11 l Relatedness (R) 2.30 2 .15 .98* Practice (P) .69 2 .32 l R x P 2.92 9 .605 .83 Within 69.96 210 .33 *p < .05. mom. *nma. pcm>mamth mes. mmm. *mmz. mzo. Hompcoo Hmm. msz. mmm. *omm. mma. Hmo. .hp om .wcHxOOA mmoconmmmfio O omm. mmH. Hmo. o mcomfiAMQEoo ,o madame om mummz .au om .hp om Hoopsoo .ap om modam> HMOHBHLO .pommfia .pcm>onLnH .mcfixooq .Apmoe m.moomnomv whoppm poafioeoO magmafiamui.ma mqm . 1 1 . .1 . 1 cl 7 . 1 1 1 1 . 1 1 1. . 1 1. 1: 1. 1: a 1 1. 1 1 1. -1 . 1 . .1 I 1. .1 . t; . 1 1 . A,“ . 1 . 1 1 . < [A \ 1 q .- 1 1 1 .1 1 . . .l 1 1 11... 2 1. 1. 111 . 1 1| 21 . .. 1.1 .1 1 . . v . 1 1 . 1 . m~ r— «H :1 con 0: cc 1c: c o 0 cc w c 1 11CLAIF I #:1F :c. 1c cc 1:: c o c cc cc c c 1;:1nnk u 121* so: mc ac cc c o 0 cc cc 1 w 11Cumlr I ~11» 3cl :P NC 0:! c c a CC Fe c m LLCLnIk I P21h ma NP fir CH c m b u u 3 r AUCCUWIIFV LECLLC .JyCL 1.1..11111111HPHC11CU LC .1uacgvlk1 1>m :uzc1h1czcu 101 11.1u 111111 .uuzua1u11kum 1u 11chknggu mc AmOOUMIF1 uuwc 1>u ou1qhu uc 23w ULLHE gupa A»; :129Hkpczpu 10 ..vZQHUIAZC .0 Cu :::1 .00 m::: CU RSI/1 .cu _::: 11.111u 1; LuHL .wH 52% Ha SZJ 1.1a .czc1cu <11>1 h1<>111cz~ .m mh<139 .mupgc11»»: uk<1:u1»1u «kc; wcoumnb ;:< kgmk;cu -_<1y.1& ”Fag .h 1qu .m 1J>J .1 ncI ncI Hcl mcI mcI moI ocI mcI co :cI mcI HcI ncI HcI :cl :cI mcl co :cI cc noI mcI "cI cc Ho+ co co mo+ :c+ ncI Ho+ HoI ncI Nsl cc nc "— cw mac cc LC ac Kc ac :c cc mg mc ac c. «H «C cc cc HF mc cc cc rc 3p ch :c+ cc rcl rc+ mc+ pc+ HcI Hc+ cc cc mcI cc nc+ mcI “cl Nc+ rc+ _c+ rcl pcI mcl 3c+ ficI Hc+ rcI cc cc mc+ Hc+ mc+ :c+ mc+ mcI nc+ mc+ —cI 3: V CC HCCCKCCCCCCCCC’CCNOCCCC/CCCCCCCCLCCCCCCC COOOO OCCOCCOOCO F-I u DOOOOOCOOOOCOOOOCCCOOOOC’ DOOOCCCOCOOC‘OOOOCCOGOOCCJOCCOOCC‘CCCOCO—HC' c: cc c: _c cc «c cc cc Kc «c 7 ac cc 2c cc .c cc fic ac ac cc Hc mc cc cc I CCC‘CCC‘OCCQC-v-‘CJCCCC‘CCCICCCC‘CI—CCC‘CC‘v—‘CCCCCCCC ov—IcwQOOQH‘OCHHI—Ic-cr‘JooH—«cccc:.—-—I HHI—ICCSCr—lm—I—H’? ,— ; Lccchr‘I ngF cu mzcw cDCufiIk I Hzch CU m:;H cLCchk I FZJF cu Mcca uccVfilr I cmh.cu mm_w LcprLk I kzqh GU HthI c;Cc£Ik I hzgh_ou cmzw mcchvaHivzmhwculmaflw cnccfllr I kcuh.ou c—_H ccchIk I PLUFIOU b—SHI ccCcnIh I wzuh.cu {fitH cccchr I Hthrcu nan c;CunIr I kLJh cu a—eH mgncfiIk.r P24».ou nHmHIIII uLcchk I k24k.ou ufizm uLCUMIk I FZJ .00 ndzH cucumlr I FZLF.CU QHFH .cccunI» I hqu.ou c::H;,I:I\ h;Cchh I kzwk CU {CFH c;rcmu» HIFZJF cu:~::g ccccnI» I cmh‘cu ccwH cccwmI» I HZmP.cu acrH ccCcnI» I pad» ou azcc anchr I ham» cu mcqm. claumI» I bch.ou cc;H checmrw.n czupucu-gc:wIIII cchmI» I bzup cu «arc cccumlh I hZUF cu .mlw cccunIh I PZUF ou cmcH c;cch» I k:u»_ou ccrH caccnI» I hchpcu xfirh ccgcmupsn:»;cpnou.Ndflw cacch» I kzgk,cu c.:H [I ccccnI» I P;m+ cu mflcw: cccUnI» I quk ou :crH c;chIk I pzuh cu acta chch» I quk;ou N_:H “cocnp»;nsp2JH.cu:wAdAIII cucch» I pzup cu ccrH ‘ccccnIF I F2MP cu c§cfi Frivmlh I ....th OU Ac5...” hm ..wCII :C nr WC] c O C 'C C; C a Li .ufh‘ IF‘ I _ J 12‘ I“; I/IiZI : co m— : :c+ c o a ..c c; c e fincWWIrr I #:er row 9UP c. co u: c“ cc c c H cc a; c c c;chIk I szk co ,5,” cc ch cc rc rcI c c o cc cc c c cchrIr I k;cr.cu ma,~ «5 nc+ ac dc rc+ c c c c ac c c .c EufiIr I Cir co #9.“ @c o.c+ rc cc mc+ c o o rc c; c c cgccfllk I Cjk. Cd :99 c_. KC! CC rc wcl C : c cc c; c r c? a I (Z \Z' 3. 3. .ZF ._ . .. . . If I PLJF C‘v‘ CwTH III: “I. I ,7; .uJ .;,c