A STUDY OF ISOLATION EFFECTS DURING THE ACQUISITIION AND RECALL 0F SPELUNG WORDS Thesis for the Degree of M. A. MICHIGAN STATE UNIVERSITY Terry Ten Brink 1966 i T P ABSTRACT A STUDY OF ISOLATION EFFECTS DURING THE ACQUISITION AND RECALL OF SPELLING WORDS by Terry TenBrink The problem of studying the effect of isolating some of the letters in a spelling word has for the past two or three decades been dismissed as unfruitful. .Recent studies, however, present evidence that students tend to make the greatest number of spelling errors at the middle positions of a word. This evidence reOpens the question concerning the effect of isolating some of the elements in a spelling word, for it suggests that words might be conceived of as serial lists, exhibiting typical serial-posi- tion phenomena. This study was designed to study the nature of isolation effects on spelling. Specifically, answers were sought to the following questions: If more errors are made in the middle of a spelling word (serial-position phenomenon), will isolating the middle position reduce those errors (Von Restorff phenomenon)? Will the reduction be great enough to cause an overall reduction of errors for the entire word? Is the effect of isolation within a given spelling word independent of variations occurring within the list of words (context variables)? What context variables might influence the isolation within the individual words? Onehundred sixth grade pupils were selected from seven private elementary schools in western Michigan. They were assigned randomly, to one of the following six conditions: control, constant-consistent cuing, constant-consistent cuing control for color per se, constant- inconsistent cuing, intermittent-consistent cuing, and constant- consistent cuing shift. The Ss in all conditions were taught twelve spelling words by rote and learned all words to a criterion of two successive correct responses. A recall task was administered one week after the learning task. The Data were analyzed on an acquisition measure (trials to criterion) and a recall measure (words wrong on recall task). The results of these analyses did not support the hypothesis that isolating the middle three letters of a nine-letter spelling word would significantly aid the learning and recall of the words. The question of whether the effect of isolation within a given spelling word is independent of context variables was not answered by the results of this study. The discovery of the sub-group from a school which taught a "whole" approach to spelling led to the hypothesis that isolation treatments would interfere with learning and recall for these 38, but would have an Opposite effect for the remaining 83. This hypothesis was partially supported by evidence of significant interactions between the treatment effects and the sub-groups on both the acquisition and recall measures. Further clarification was sought by analyzing the sub-groups separately and the only significant main effects disclosed was on the recall measure for the "whole" group. Individual comparisons showed that the control group made significantly fewer errors than the other groups which did not differ significantly from each other. It was noted that although almost all of the differences which existed among the various conditions were not statistically significant, the results for the "part" approach were all in the expected direction and the results for the "whole" approach were almost a mirror image of the "part" approach. It was suggested, therefore, that further research be conducted which would account more precisely for the variable of pretraining. The spelling task does lend itself to experimental manipulation and a number of future research problems were suggested. A STUDY OF ISOLATION EFFECTS DURRING THE ACQUISITION AND RECALL OF SPELLING WORDS By Terry TenBrink A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Education 1966 ACKNOWLEDGEMENT The author wishes to take this Opportunity to express his appreciation to Dr. Clessen J. Martin for his assistance and guidance throughout this study. ii TABLE OF CONTENTS Page A CKNOWIJEmMNT O O O O O O O O O O O O O O O i i LIST OF TABLES . O O O O O O O O O O O O O 0 iv LIST OF FIGURES . C O . C O O O O C O O O O C v LIST OF APPENDICES . . . . . . . . . . . . . vi Chapter I. INTRODUCTION . . . . . . . . . . . . . 1 II D mmOD . O O O O O C C O C O O O O C O 15 A 0 subjects . O O O O O O O C O O C O 15 B. Materials . . . . . . . . . . . . . 15 C 0 Procedure 0 O O O O O O O I O C O O 15 III . RESIJIJTS . O O O O C O O O O O C O O O O 20 IV. DISCUSSION 0 O C O C O C O O O O O O O 33 V0 BIBLIOGMPIIIY . O O O . O C . . . C O O 42 VI 0 APPENDICES O O O O 0 O O O O O O O O O 45 iii Table 10 11 12 13 14 LIST OF TABLES A comparison of the lists used for each of the six conditions showing the isolated elements Means and standard deviations of the number of trials to criterion on the acquisition task Means and standard deviations of the number of words wrong on the recall task . Means and standard deviations of the number of trials to criterion for the four critical words Means and standard deviations of the number of words wrong on the recall task for the four critical words . . . . . . . . . Means and standard deviations of trials to cri- terion for the two sub-groups 2 x 6 analysis of variance summary for sub- groups and treatments on the acquisition measure Means and standard deviations of words wrong on recall for the two sub-groups 2 x 6 analysis of variance summary for sub-groups for treatments on recall measure . . 1 x 6 analysis of variance summary for sub-groups on the acquisition and recall measures . . Means and standard deviations of recall errors at positions 4, 5, and 6 for the two sub-groups Means and standard deviations of trials to criterion on the critical words for the two sub-groups Results of "Part" and "Whole" approaches for acquisi— tion and recall ordered from lowest to highest num- ber of trials and errors A sample list of spelling words illustrating different syllable/length ratios . iv Page 16 20 21 23 24 25 26 26 26 27 31 32 37 Figure LIST OF FIGURES Errors per letter position on recall task for six conditions Errors per letter position on recall task for six conditions-Part Approach Subgroup . Errors per letter position on recall task for six conditions whole approach subgroup . Page 22 29 30 CHAPTER I INTRODUCTION The problem of studying the effect of isolating some of the letters in a spelling word has for the past two or three decades been dismissed as unfruitful. Horn (1940, 1950, 1960), Fitzgerald (1951) and Foran (1952) claim that only a relatively few persons make errors in the same position in spelling words and therefore it is not worthwhile to consider the effect of marking "hard spots". However, recent studies (Jensen 1962; Kooi, Schutz, & Baker, 1965; and Guinagh et. al., unpublished, 1966) present evidence to show that second graders through college students all tend to make the most errors in the middle positions of a word. This evidence reopens the question concerning the effect of isolating some of the elements in a spelling word for it suggests that words might be conceived as serial lists, exhibiting serial- position phenomena on a spelling task (listing the letters in the correct order). One of the earliest and most extensive studies which concerned itself with isolation of elements in a spelling task was conducted by Tireman (1929). His study involved over 4,000 pupils of grades four, six, and eight and included over a half million spellings. He concludes his study with the following statements: "...one is impressed with the consistency with which the data shows that marking hard spots is of little or no value. ...the people who advocate marking the hard spots are not only suggesting a useless device but possibly a harmful one." -2- Tireman's study was performed in the classroom and the students studied their words for a period of one week as they were accustomed to. Some 83 studied from lists in which tne words contained isolated elements (capitalized, underlined, printed in bold-face) and others studied from lists which contained words with no isolated elements. This arrangement left the experimenter with virtually no control over the variables of the learning situation. Another problem inherent in Tireman's study arises as a result of his method of locating difficult positions in words. The method he used often yielded several difficult positions in thessame word. In some cases he had the pupils mark their own "hard spots" and again this often yielded more than one difficult position per word. The question arises as to whether or not an element can be considered isolated when there are other elements in different parts of the word which are similarly made to stand out. (e.g. efficiency was used with the sixth graders and printed: efficigngy. Perhaps the pupils were, in fact, diverted from the task of writing all the letters in the right order when they were asked to learn words in which more than one part of the word was 'isolated in this manner. Mednick (1964), in discussing Jensen's 1962 spelling study, suggests that isolating the middle part of spelling words will not decrease the number of errors. According to Mednick, if spelling yields the same results as other serial learning tasks, one would expect a decrease in the number of errors made in the -3- isolated positions but no decrease in the total number of errors for the whole word. This expected result is based on the findings of studies of isolation in serial learning. Jensen (1962b) "...connections are learned in a certain suggests that associative order around an 'anchor' point, which neusually the first item presented." The anchor point, the position around which the other items are learned, has become an important factor in ex- plaining the effect Of isolating some part of a serial list. It has been suggested that when an item in a serial list is isolated, it becomes the anchor point and therefore fewer errors are made in that position and correspondingly more errors in some other position. This seems to be a satisfactory explanation of the effect of isolation when all items are equally easy to learn, but spelling words are not made up of equally easy items, so the effect of isolation in spelling still deserves further study. The present study is designed to study the nature Of isola- tion effects on spelling. It will seek first of all to answer the following questions: If more errors are made in the middle of a spelling word (serial-position phenomenon), will isolating the middle positions reduce those errors (von Restorff phenomenon)? And, will the reduction be great enoughtto cause an overall reduction of errors for the entire word? The research which has been done on serial-position and isolation effects leaves many un- answered questions and it is especially difficult to generalize -4- the findings to a study such as this one, for the spelling task is a different kind of task than the serial-learning or paired- associates tasks which have been used thus far in studying these effects. Three facts in particular must be considered which make the study of isolation within a spelling task unique. First of all, each word can be considered a serial list but in a typical spelling situation the person sees the whole word at once, i.e., he is not presented the word letter by letter through an anticipa- tion method and therefore he is not compelled to learn it that way. In fact, Abernathy (1929) recorded the eye movements of good and poor spellers and noted patterns of eye movements which included many regressions and highly unequal fixation times for the various parts of the word. The anticipation method usually used in serial and paired-associates learning studies would re- duce the possibility of unequal fixations on any given item be- cause each item moves out of sight before the next item moves into place, No regressions are possible when the anticipation method is used. In his 1929 study, Abernathy found that the better spellers made repeated surveys of the word as a whole, making several regressions to the difficult part. Perhaps, then, by printing the middle positions (the difficult part) in red the Ss attention would be directed to those letters, causing more re- gressions to that part as his eyes scan back and forth across the word. A second unique feature of the spelling task is that the items (letters) are not of equal difficulty and are not independent -5- of each other. Some letters are more closely "tied" to the sound they represent than other letters. Consonants such as t, b, m, d, and k each represent only one sound, whereas consonants c and g, for example, each have two sounds and furthermore ”share" those sounds with other consonants (k and s in the case of c; j in the case of g). Vowels become more difficult still for they each can represent many sounds. The dependence of letters upon the other letters in a word is easily seen. The quality of the sound for a consonant such as b or p, for example, changes considerably depending upon which letter it preceeds or follows in a word. Some letters join to- gether to form a single sound and the same letter combination may not necessarily represent the same sound from one word to the next (e.g. ough as in tough, though, through, etc.). The presence of silent letters within a word is another example of the inter- dependence among letters. Whether a letter represents a sound or is silent in a word is determined by the context. i.e., the presence or absence of certain other letters or the position of the letter in the word. These unique features of spelling, the inequality of difficulty and interdependence of the letters in the spelling task, are important considerations in light of the various explanations offered for the serial-position and von Restorff phenomena. Many theoretical and empirical explanations have been offered for the serial-position phenomenon (Deese, 1958; Underwood, 1963). Although Hull's attempts to explain this phenomenon (Hull, 1953; Hull et. a1., 1940) are not too widely accepted, an S-R paradigm -6- is still the most widely used basis for analyzing serial learning. A search for empirical evidence to support the idea that each {tum in a series is the stimulus for each succeeding item has yielded mostly negative results and has led researchers to a "search for the functional stimulus”. (Jensen & Rohwer, Jr., 1965). Some experimen- ters maintain that position itself acts as the functional stimulus (Schulz, 1955; Young, 1962; Keppel & Saufley Jr., 1964). Other investigators have pr0posed that some parts of the list are learned predominantly be sequential and other parts predominantly by posi- tion associations. Ebenholtz (1963), for example, states that position learning occurs predominantly at the extremes and sequen- tial learning occurs in the middle of the list. Jensen & Rohwer, Jr. (1965) suggest that although the evidence so far seems to favor a position hypothesis, it is also inadequate to explain fully what is being learned in serial learning. They propose another approach, explained in detail by Jensen (1962c), which regards serial learning as essentially a process of response integration rather than as the acquisition of specific S-R associa- tions. As the authors of this approach have suggested, the idea has not been worked out in enough detail to have much predictive power for specific experimental outcomes. This brief survey of the major attempts to explain the serial- position phenomenon discloses two important facts. First, all of the approaches assume that each item is independent and some re- lationship is sought among the items (sequential associations, position -7- associations, interference due to backward associations, integra- tion of the response items, etc.). These explanations would not generalize readily to spelling because independence of items (letters) cannot be assumed. Secondly,none of these approaches to the serial-position phenomenon have proved to be entirely successful and as suggested by Jensen and Rowher (1965), this has caused a search for the functional stimulus. In spelling, the functional stimulus becomes extremely difficult to isolate because in any given word a combination of letters, as well as individual letters could function as a single item and become the functional stimulus. Furthermore, the fact that certain letter combinations within a spelling word (e.g., "per" or "ab" as in Egghaps or 32: sentee) might be differentially meaningful for the Ss greatly re- duces the possibility that a given item would act as the functional stimulus for all Ss. One of the most interesting phenomenon affecting serial posi- tion is the von Restorff phenomenon. In 1913 von Restorff reported a series of studies which showed that isolating an item against a crowded or homogeneous background facilitates the learning of that isolated item. Von Restorff interpreted her original re- sults within a Gestalt framework (Koffka, 1935). The neutral trace of the isolated item stands out as a "figure" against the "ground" of the similar traces of the other, non-isolated items. Wallace (1964), in his review of the status of the von Restorff phenomenon, states that isolation has been manipulated in three major ways. The method giving the experimenter the most control over the manipula- -8- tion of isolation involves an additional Operation on an item within a list. Printing an item in red when the remaining items are in black is an example of this kind of manipulation and it allows the isolated term to be compared with itself in a non- isolated condition. Several explanations have been offered to account for the von Restorff phenomenon. The Gestalt theory has been carefully conceptualized by Koffka (1935). He explains the effect of isolation in terms of figure-ground relationship. Supposedly, the isolated element stands out, becoming the figure, while the rest of the word becomes the ground. The more homogeneous the background against which the isolated element is placed, the more effective the isolation will be. As has already been suggested, the letters in spelling words are not "functionally" similar and the non-isolated elements will probably 225 be perceived by the subject as a homogeneous group of letters. A second major explanation of the von Restorff phenomenon is based on the idea of an S-R association. Various researchers have maintained association as basic to their explanations and yet have come to different conclusions as to the specific mechanism(s) involved. Gibson (1942) posited an interference theory which suggests that isolation increases the differentiation of the items thus reducing the interference from other items in the list. In other words, stimulus generalization is reduced. Horowitz (1962) suggests that isolation affects the associative stage in serial learning whereas Saltz and Newman (1959) suggest that early in -9- learning the influence of isolation occurs mainly during response learning but that as learning progresses it may exert its influence on the associative stage. Smith (1949) and Smith and Stearns (1949) have suggested that the principle benefit of isolation is in the serial organization of the list. They found that an item which had been isolated in red print (the remaining items were printed in black) aided in establish- ing order and therefore the isolated lists were learned considerably better than the non-isolated lists in the later stages of learning, viz. after the responses were known and when the problem was one of getting them in the right order. Jensen (l962d) offered further evidence for this explanation by showing that the learning of items per se is not affected by isolation, but that the order of learning those items is affected. Since the items in a spelling list are familiar (letters and/or letter combinations) and are closely tied to the pronunciation of the word it is reasonable to conclude that the largestpart of the spelling task is the ordering of the units and therefore the spelling task may provide a useful paradigm for looking at isolation. Another explanation of the isolation effect was offered by Green (1956). He suggested that the "surprise" aroused by being unexpectedly presented with a different type item accounts for the isolation effect. The results of Green's study supported the hypothesis that there would be a greater von Restorff effect for the first isolated term than for the second isolated term. -10- A third unique feature of the spelling task is that not only can each spelling word be conceived of as a serial list but it is also an item within a total list. This fact immediately broadens the scope of this study and thus the study will also seek to answer the following questidns: Is the effect of isolations within a given spelling word independent of variation occurring within the total list? And if not, i.e. if the effect of isolation within a word lg influenced by variations within the list (context variables), then what context variables might influence the isolation within the individual words? Green (1956), in discussing the "surprise" effect of isolation, suggests that the effect of isolation is due to surprise and that this effect extinguishes when isolation in a list is repeated. His study showed that a second isolated element did not have as great an effect as the first element. In fact only the errors for the first isolated element were significantly reduced in comparison to the other elements in the list. These results suggest that an important context variable might be the number of words in the total list that contain isolation. A possible way of testing this would be to compare the isolation effect on certain "critical" words which contain isolation when these words are contained in two lists - one list in which all the other words in the list also contain isolation (this might be called a constant-consistent list,for all the words are isolated, and each word is isolated in the same way); and a second list in which only the critical words contain isolation and the other words contain no isolation (this might be called an inter- -11- mittent-consistent list for the words which contain isolation occur only occasionally or intermittently throughout the list). 0n the basis of Green's study one would predict that the isolation effect within the critical words wauld be greater when they are contained in an intermittent list than when they are contained in a constant list for the surprise effect would have a greater opportunity to ex- tinguish in the constant list where the subject would be responding to a word containing isolation everytime he made a response. On the other hand, if it is important for the Ss to develop a set to look for the isolated elements in the words, the constant list would produce an Optimum context. Smith and Stearns (1949) manipulated isolation by presenting one item in a serial list in red print and the remaining items in black print. In the later stages of learning, when the responses were known, an advantage for the isolated list was demonstrated. It was assumed that the red item aided in establishing order. Jensen (l962d) also demonstrated that organization of the material is aided by isolation, although he proposed that the organizing aid of the isolated item is quite specific to that item. These studies suggest another possible context manipulation: the position(s) of the color used as a cue. If color is an important cue in helping the S to organize the letters, then shifting the position of the cue-color ought to make it more difficult for the S to use color as a position cue. In order to maintain isolation within a given word while at the same time shifting the position of the cue-color, one could simply switch the colors of the -12- isolated letters (red) and the non-isolated letters (black). One could begin then, with a word printed in black letters with the middle letters isolated in red and by switching the red and black colors end up with a word printed in red letters with the middle letters isolated in black. The isolated element would remain in the same position but its color would change. (This procedure could be represented as follows: BLACK-RED-BLACK/ switch / RED- BLACK-RED). This shift could be made within a list so that from word to word the red color would not be in the same position, but for a learning task of several trials, each trial would be the same and thus any given word would always have the red in the same position(s). This might be called a constant-inconsistent cuing condition for all the words are isolated but the position of the cue color is inconsistent throughout a given trial. Another way of making this shift would be to keep the red in the same position(s) for all words on one trial and then shift it to the other posi- tion(s) on the following trial. NOW, within a given trial the position would remain the same but within a given word it would shift from trial to trial. This would represént a situation where the subjects would be asked to shift from a constant-consistent list with the cue color in one position to a constant-consistent list with the cue color in another position, and it might be called a constant-consistent shift condition. On the basis of a Gestalt explanation of isolation effect, there should be no difference between either of these two conditions, nor should the results from either of them be different from the results _13- of a constant-consistent cuing condition. In all three cases the figure-ground relationship should be maintained. However, on the basis of S-R explanations, which propose that the color aids in organization of the list (Smith & Stearns, 1949; Jensen, l962d), it would be expected that the two conditions where the cue color is shifted would yield more errors than a constant- consistent condition in which the cue color maintained its pos- ition. If the Ss can develop a set quickly enough (within a given trial) and learn to shift that set fromitrial to trial, then it would be expected that the constant-consistent shift condition would yield fewer errors than the constant-inconsistent condition. It is possible that effects which might occur in the con- dition involving shifts in cue color could be due to the fact that one cue color is more effective than another rather than to the experimental manipulations. In order to control for this a constant-consistent list could be devised which would contain all words in red print with black isolation. If this condition would yield results not significantly different from the condi- tion in which all the words were in black print with red isola- tion, then it could be assumed that any "shift" effect would not be due to differences in the strength of the color of the isola- tion. The objectives of this study are: l. to determine whether or not the von Restorff phenomenon can be produced by isolating the middle letters of spelling words during the learning task. -14- to determine whether or not spelling words will be learned faster and recalled better a week later if the middle letters of the words are isolated during the learning task. to determine the nature of the effect of context (list manipu- lation) upon the isolation-effect within individual words when the list is manipulated in the following ways: a. all of the words in the list are isolated in the same manner (constant-consistent cuing) b. one—fourth of the words are isolated in the same manner, and the other three fourths of the words contain no isolation (intermittent-consistent cuing) c. all the words are isolated -- one half of them with the cue color in the end positions, the other half with the cue color in the middle positions (constant-inconsistent cuing) d. all the words are isolated with the cue color in the middle positions on the odd trials (1, 3, 5, etc.) and with the cue color in the and positions on the even trials (2, 4, 6, etc.) (constant-consistent cuing shift) CHAPTER II METHOD A. Subjects A total of 120 sixth-grade Ss (54 females, 66 males) were tested in this study. They were selected randomly from the sixth-grade classes of seven private elementary schools in the western Michigan area. B. Materials Six 12-item lists of nine-letter words were constructed. Each list contained the same 12 words but the lists varied according to the amount of isolation. Table 1 presents the six lists. The words containing isolation in lists 4 and 5 were selected randomly from the 12 words. C. Procedures A separate group of sixth-grade pupils were tested on a large list of nine-letter words. From those words which these pupils were unable to spell, 12 words were chosen for this study. Each word was printed in large print, in all lower- case letters, on a 3 x 5 index card. A11 83 learned the twelve words to a criterion of two correct responses. Each word was presented for five seconds and the experimenter (E) simultaneously gave the correct pro- nunciation. An unlimited amount of time was allowed the S to write the word immediately after it had been exposed. The entire list was presented in this manner and whenever an S spelled a -15- mHODDOH xOMHm ... muoquH pom ...-ll ... ... ... .... . OI . .alll'I .. ... ... .... Mao osmoowo haosmoomo waosmoomo Ndosmoomo qusmoomo w: dud mo waoamoomo ...llll... .... .... ... I. .Ill ... .. .... .... mHoowmson mHooston mHomwnsoa mHoowuson mHooston mHowwu om mHoowuson I'll-...! ...III-Iu.. .. . . IIIIII III Illll ........ HOmmOmmom HOmmOmmo “Ommowmmm HOmmwmmom HOmmemom mmeOm.mm HOmmOmmom ll...lll ...-Ill. . ...-II . . I'll . ...... .. momOHstm oomOHs mm OOmOHs mmwo.swmm wowmm.mom wmmenmmfi oomOHsmom IIIIIII...-lllll ...lll... ......... l...ll|l Ill...l'| ...ll... ........0 umH00poo> umHoocoo> umH00poo> umHoowoo> umH00poo> umHoovoo> umHoowoo> .IIIIII O..I|Il.. . O. I'lllu .. . .... mum: oumx ouhsmoumw wmwsmowmx fiwN .. muNmmmme wmhfimo.mw ouwcmoumx EOOOE ...E... ....O.... ...E... El...nlllll ...-Ill... ......... ooOwOOmOH ooSmOOmou mochOme monOOmOH ooOMOOmOH mo «COmOH moamOOme III-Ill.....lllll o..'lllll... ......... IIIIIII...III.II ll...llll ...l'lll... .....O... mOOHumomm mDOHuoowm mOOHuoowm mSOHumowm mDOHuoomH mOOHuoomm mSOHuwomH [DOE ...lllll... ...-Ill... . .. I .ll ... ..... nonwosoxo nonwosoxo nonwoaoxo H..mo nmxm Hmsmmsuxm wmmmmsmww Honmosoxo Eooolllll ...IIIIII ...... .0 E...IIIIIII ll... Cool 00.... 0. EmHEonmsw EmHEos no EmHEOL so EmHEosmso EmHEwama EmHEO: so EmHEmn so level-III ooollllllooo 000E000 000E000 Ecol]- co. 00. on... 000 HOHOmeHn HOHpmeHn HOHpmeHn Hmemeun Hmem Hun HOHpmeun HmewwHHn EooolllI-I coo-I'll... 000000000 000 .00 0.0!... 00000000. muoEHomss mooEHcmos mooEHamas mooEHcmcm meEHammm mnoEHams: muoEHcmcs om mom nHoHoo mHmHuu co>w mHmHuu pwo How Houu:00v mumHam ucoumHmcoo ucoumHmuooaH uaoumHmsoo acoumHmaoo ucoumHmaoouuamumooo nucouuHEHouOH nuamumaoo nuawumcoo nuamumaoo Houuaoo mum Hum mum mum Hum o muamaon woumHOmH Osu waHzonm .mGOHquaoo me ecu Ho 50mm Mom womb mumHH msu mo :OmHHmmEOU 4 H mam