H ' llWlHllli l | 1111 1 l 113 127 THS THE APPLECABIUW OF THE SEMANTiC DiFFERENTiAL TO PRESCHOOL AGE CHiLDREN Thesis for the Degree of M. A. MICHIGAN STATE UNWERSETY PHYLLBS J. JONES 1958 mesqs » 1.1853512 Yr 4' Michigan dram W I.n.-I—a- 7“ n" a o ' I ‘ V BINDING BY HOAE & SflNS’ Manny mc. BI In: Dc 1111111111111111111111111111111111111111111111111111111111 PLACE IN RETURN BOX 3 1293 1039 to remove this checkout from your record. To AVOID FINE return on or before date due. MTE DUE DATE DUE DATE DUE ,- i Wm, FEB IVAN} "ii“ 0 we «cramps-p.14 ABSTRACT THE APPLICABILITY OF THE SEMANTIC DIFFERENTIAL TO PRESCHOOL AGE CHILDREN by Phyllis J. Jones Recent investigations in the area of language develOp- ment have evidenced a mutual concern by psychologists and linguists to understand the systematic nature of language to discover the underlying structures or rules behind changes during this deve10pment. One aSpect of this study, semantics, has focused on the investigation into meanings individuals attach to objects and concepts. The Semantic Differential has become the most extensively used method in studying meaning structures, but has had a very limited use with children. Several researchers have suggested this technique be adapted for use with very young children by administering it orally, but no attempt has been made to use the Semantic Differential to measure connotative meanings of concepts across the three factoral dimensions (Evaluative, Potency, Activity) with this group. The main purpose of this investigation, therefore, was to eXplore the feasibility of administering an oral form of this technique to preschool age children. Also an analysis Phyllis J. Jones of obtained results was made in order to discover what meanings very young children attach to concepts as well as to explore the bases of these meanings. The sample included sixty-five children ranging in age from 42 to 66 months. There were thirty-two boys and thirty-three girls, selected from the Michigan State Univer— sity Laboratory Pre-School. The Semantic Differential technique was modified for use with these very young children. Two equivalent forms of an oral individually administered questionnaire including nine concepts rated on six three-point bipolar adjective scales were designed as the instruments for the investigation. The selected concepts and bipolar adjectives were carefully chosen to insure that the children were familiar with them and had meanings attached to these words. Statistical measures used to analyze the data indicated that the number of these familiar task items completed in- creased with age increment. Based on the children's per- formance on this orally administered questionnaire the tech— nique was found to be generally inapprOpriate with subjects under 48 months of age. Children 48 months of age and older were able to reSpond to the questionnaire, however, as was evidenced by the ceiling effect found in the mean number of items completed by the older preschoolers. Analysis done to discover what connotative meanings very young children attached to the selected concepts showed 2 Phyllis J. Jones the following trends: a) Meaning systems similar to those an adult in our society might be expected to use in describ— ing the concepts with the bipolar adjectives provided seemed to be develOping. b) Familial concepts included received similar ratings by boy and girl subjects. c) Sex differ- entiated ratings seemed to appear in relation to the concepts BOY and GIRL. An eXploration of the meanings attached to the selected concepts indicated that the majority of the children used sensory eXperiences as their bases. The findings of this exploratory study, that the Seman- tic Differential is feasible with very young children, in addition to the resulting trends in the connotative meanings shown using this technique, would support the recommendation that investigators make further application of this method in research with preschool age children. THE APPLICABILITY OF THE SEMANTIC DIFFERENTIAL TO PRESCHOOL AGE CHILDREN BY .1 fl Phyllis J? Jones A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Home Management and Child DevelOpment 1968 AL7 -> 1 I . ACKNOWLEDGMENTS Special appreciation is expressed to Miss Phyllis E. Lueck for her guidance, encouragement and patience in directing this study. Because of her supportive atti- tude, this has been a valuable learning eXperience. Appreciation is also extended to Dr. Vera Borosage, Dr. Louise Sause, and Dr. Beatrice Paolucci for their guidance and helpful suggestions. For their assistance with data collection, sincere thanks is extended to Miss HOpe Schweitzer and Mrs. Sharon Stolz. Gratitude is expressed to Dr. Sarah Hervey and Mr. William Logan for their assistance with the statistical analysis. An eXpression of appreciation is also extended to Miss Barbara Neumann, Miss Colleen McNally, Miss Karen McNally, Mr. Gerald Schwab and. Mr. Fred Hahn for their helpfulness and encouragement. ii TABLE OF CONTENTS CHAPTER Page I. INTRODUCTION . . . . . . . . . . . . . . . . 1 Purpose of the Study. . . . . . . . . . 2 Operational Definitions . . . . . . . . 5 Assumptions . . . . . . . . . . . . . . 5 Objectives. . . . . . . . . . . . . . . 4 Hypotheses. . . . . . . . . . . . . . . 4 II. REVIEW OF LITERATURE . . . . . . . . . . . . 5 The Semantic Differential . . . . . . . 10 III. METHOD . . . . . . . . . . . . . . . . . . . 17 Pretesting the Instrument . . . . . . . 20 Administration of the Instrument. . . . 21 Data Collection . . . . . . . . . . . . 25 IV. RESULTS. . . . . . . . . . . . . . . . . . . 26 V 0 DISCUSSION 0 O O O O O O O O O O O O O 0 O O 50 VI. SUMMARY AND IMPLICATIONS . . . . . . . . . . 58 Implications for Research . . . . . . . 40 REFERENCES . . . . . . . . . . . . . . . . . 42 APPENDIX . . . . . . . . . . . . . . . . . . .45 iii LIST OF TABLES AND FIGURES TABLE I. Chi-Square Values Obtained in Analysis of Test- Retest Reliabilities. . . . . . . . . . . . . . II. Percentage of Consistent Test-Retest ReSponses. FIGURE 1. Excerpts from Questionnaire Forms' Score Sheets showing positional rotations of adjective pairs 2. Concepts and bipolar adjectives employed in the stUdy O O I O O O O O O O O O O O O 0 O O O O O 5. Mean number of items completed by groups. . . . iv Page 29 29 19 25 27 CHAPTER I INTRODUCTION Language develoPment has been the object of extensive research by both linguists and child psychologists, because it is the prime communication vehicle among men and is so closely related to thought. Reference to recent reviews of research in this area (Carroll, 1966; Ervin-Tripp, McCarthy, 1954) give evidence of the magnitude of these studies. A current area of research, psycholinguistics, has evolved as a result of a mutual concern about language and verbal behavior by linguists and psychologists. As defined by Osgood, "psycholinguistics is concerned in the broadest sense with relations between messages and the characteris- tics of human individuals who select and interpret them" (Osgood & Sebeok, 1965). One aSpect of this research, focus- ing on the investigation into meanings individuals attach to objects and concepts, is called semantics. The best known method for measuring the meaning of objects and con- cepts is the Semantic Differential devised by Osgood and his associates (Osgood, Suci, and Tannenbaum, 1957). Extensive research has been done with this technique and it has been found to be very useful in studying meaning structures, since a quantitative measure is obtainable (Kerlinger, 1966; Remmers, 1965). This technique has been chiefly used with adults,‘however, and has had a very limited usage with children (Church, 1961; Ervin-Tripp, 1966). Several research- ers have suggested that the Semantic Differential could easily be adapted for preschool age children by administering it orally (Ervin-Tripp, 1966; Lilly, 1965: Small, 1958), but only one researcher (Williams, 1967) investigating racial attitudes has attempted to use this technique with these children. Using only the evaluative factor, this latter study seemed to indicate that the technique was usable with very young children since the results were consistent with adult ratings on a similar investigation. Purpose of the Study Since no systematic attempt has been made to use the Semantic Differential with very young children to measure connotative meanings of concepts across the three factoral dimensions (Evaluative, Potency, Activity) as described by Osgood, this study will explore the administrability of an oral form of this technique with children ranging in age from 42 to 66 months. This study will endeavor to determine if a modified Semantic Differential is feasible with pre- school age children. An analysis of obtained results will also be made in order to discover what meanings young children attach to concepts as well as to explore the bases of these meanings. 1. Operational Definitions Dimensional factors--Refers to aSpects of concept meaning measured by bipolar adjectives as categorized by Osgood and his associates. Factors included in this study are: Evaluative (E) good-bad, clean-dirty, happy—sad; Potency (P) large-small, hard—soft; Activity (A) fast-slow. Neutrality—-Refers to the type of reSponse which indicates that the bipolar adjectives are not relevant to the particular concept. Assumptions The children are familiar with the concepts (nouns) and bipolar adjectives used in the study and have meanings attached to these signs (words) (Church, 1961; Osgood & Sebeok, 1965). The following criteria were used in the selection of words used: (a) The concepts and bipolar adjectives are within the children's experiential back- ground. (b) These words are used in nursery rhymes and storybooks that have been read to the children in nursery school. (c) The words elicited responses and definitive comments in the pilot study. 2. II. Preschool age children can understand oral directions and select from the bipolar adjectives presented in re- lation to the concepts included in this study (Anastasi, 1960). Results of the pilot study seem to indicate this. Objectives To determine if the Semantic Differential as modified for this study is feasible with preschool age children. To discover trends in meanings attached to selected con- cepts by young children. To explore the bases of these meanings. Hypotheses The number of items completed by the children will in- crease as age increases. The incidence of neutral responses to those items irrel- evant to concepts rated will increase as children's age increases. CHAPTER II REVIEW OF LITERATURE Through the years the focus of research on language deveIOpment of the preschool age child has shown much variety. Starting in the 1890's the investigations were chiefly longitudinal diary studies of brief content reports. These studies were of a descriptive nature, and generally neglected any investigation of the relation of thought to language (Berko and Brown, 1960). Examples of this type of research are found in The Twenty-eighth Eggrbook of the National Societyyfor the Study of Education (1929), which includes abstracts of 125 published studies on language de- velOpment of children from three through five years. These studies primarily involved the investigation of vocabulary and sentence structure through verbatim recordings of children's conversations in play as the main source of data. Only Piaget (1926), using the diary method, attempted to study thought processes via a study of language develOpment. A transition in the focus of research from the diary method to quantitatively measurable aSpects of language (e.g., articulation errors, picture-naming) occurred in the latter part of the 1920's and predominated the work of the 1950's. Wellman, Case, Mengert and Bradbury (1951), for example, attempted to study the deveIOpment of articulation in young children. In this investigation children's verbal reSponses were obtained by showing them toys and pictures. Their interest was in the child's phonetic develOpment be— tween two and six years of age. Data collected was tran- scribed into the International Phonetic Alphabet and included 155 sounds (66 consonants, 48 consonant blends, 15 vowels, and 4 dipthongs). About this time seeking improved measures of vocabulary growth a number of vocabulary tests for preschool children were developed including Smith (1926), Van Alstyne (1929). and Williams and McFarland (1957). Smith's test included 205 items from the Thorndike (1921) Word Book. Presenting an object or a picture the examiner questioned the child about the word. Subjects included 244 children ranging in age from eight months to six years. The sample used for standardizing the test has been criticized (Irwin, 1960), because the number of subjects from the upper and lower ends of the mental scale was greater than would be found in a normal distribution. Reliability of the test was obtained by administering two halves of the test to 55 children rang- ing in age from four to six years. Though the number of correct scores increased with age progression, there was no explanation about the discriminating power or the uniqueness of the items. In 1957 Williams and McFarland revised the Smith Vocab— ulary Test. This test had two forms, each comprised of 42 items. It was administered to children between the ages of 27 and 74 months. Standardization was based on 559 sub- jects and there has been criticism of the sample used. Because the children used were either of superior and above intelligence or of low socioeconomic status and below aver- age intelligence, it was felt these subjects were not typical of the pOpulation. Test reliability was determined by the (correlation between scores on Form I and Form II for the sample (.96i.OO5). For this test the order of difficulty of items was determined. Van Alstyne's (1929) vocabulary test was designed for use with three year olds. In this test the child was asked to select the correct item from a card with four pictures. Forty-five cards comprised the test which included 51 nouns and 14 verbs. Eighty children from 55 to 59 months of age were used to standardize the test. The reliability corre- lation of the test was 0.87. Van Alstyne's test also was usable with children between the ages of two and five years. Shirley (1958) did a content analysis study of the Speech of preschool children in an attempt to see which part of language content was an outgrowth from within the child's own body and which was superimposed from without. ,Studies like this contributed to research information, but were quite Specific and limited in content by the means or source of data collection. Shirley, for example, collected data from children's conversation at an all day health clinic. .Such a setting indeed affected the content of the language samples. Not until the 1940's, however, did researchers investi- gate the relationship between language develOpment and intel- lectual development in a theoretical context, thereby provid- ing a framework for the comparison of children's verbal growth across cultures. Velten (1945) investigated phonemic structures by recording all of the meaningful Speech forms Spoken by his daughter, Joan, from age 11 to 56 months. Those phonemes used with consistency in relation to referents were considered meaningful. This research, though limited to one subject, offered a technique for comparison of phoneme usage between English Speaking children and adults and also gave a base for studies of foreign languages to investigate similar occurrences between child and adult language. Since Velten's work, research has manifested an even greater linking of psychology and linguistics, and has sought to understand the systematic nature of language to discover the underlying structures or rules behind changes during lan- guage develOpment. Investigators have more actively involved young children in research through conversation and question- ing in an attempt to gain an understanding of the child's thought processes in language. Patrina (1959), for example, found preschool children could use words concretely before they could use them abstractly. Through talking with the children he discovered 79% of his sample understood "deep water" but only 15% understood "deep secret." Studying the young child's morphological system, Berko (1958) devised an interesting technique to explore whether children understood and had developed a systematic base for supplying English inflexions, derivations, and compounds of words, or were merely doing this in their language by chance. Her sample included preschool and first-grade children, whom she asked to supply, for example, a compound word, in relation to a picture She showed them labeled with a nonsense word. Berko believed that the child's ability to do this task would indicate that he had develOped a grammatical base. Giving a correct word form in relation to some other known object might occur due to the experience of hearing the plural word form and not necessarily Show that a child had an internalized system. .Ervin and Foster (1960) conducted an interesting study to investigate the develoPment of meanings in children's descriptive terms. Using physical dimensions of size, weight and strength (i.e., big, heavy, strong) varied in a Specially prepared set of materials, the subjects were asked to dis- criminate between two objects which were identical except for one dimension. The sample included first and sixth graders and it was found that the younger group often used the three dimensions interchangeably as synonyms. 10 Research into the meanings of words, semantics, has become extensive recently via a technique called the Semantic Differential, but little has been done with children. Susan Ervin-Tripp (1966) emphasized this in saying, "It is sur- prising that the best known system for assessing connotative structures, the Semantic Differential, has seldom been used with children." The Semantic Differential The Semantic Differential was a technique devised by Charles E. Osgood and his associates to measure the connota- tive meanings of words. Prerequisite to the development of such a method the assumption must be made that words general- ly have shared meanings. Kerlinger (1966, p. 564) in dis- cussing this idea stated, "Any concept, then has a common cultural meaning. It also has other meanings, some of them Shared by different groups of peOple, some of them more or less idiosyncratic." The technique involves an individual rating a number of concepts (e.g., MOTHER) on a series of 7 point bipolar ad- jective scales (e.g., good-bad). The scales represent a number of dimensions of which the evaluative, potency and activity factors (E,P,A) have been chiefly investigated. The notion of using polar adjectives to define dimensions of a concept deve10ped as a result of research by Theodore Karwaski and Henry Odbert at Dartmouth College in the late 11 1950's (Osgood, Suci, and Tannenbaum, 1957). The idea of bipolar adjectives falling into dimensional factors (E,P,A) occurred early in the develOpment of this technique. When investigators found that many of the ad— jective pairs (e.g., good-bad, clean-dirty) were clearly evaluative in nature, but some of the pairs (e.g., strong- weak, realistic-unrealistic) were independent of the evalu- ative group, this seemed to indicate the existence of other dimensions of the semantic framework (Osgood, Suci, and Tannenbaum, 1957). The bipolar adjectives that were not in the evaluative category were examined for clustering, that is, were certain adjective pairs interrelated (e.g., strong-weak and large- small are generally associated with each other)? Two clus- ters of adjective pairs were found and designated the potency and activity factors. Though the adjective pairs in these categories have been found to be contaminated with the evalu- ative factor, the dimensions have been retained in order to balance the scales for measurement purposes. An example of contamination was found in the adjective pair 'rough-smooth.‘ These adjectives mainly reflected the potency category, but also had evaluative meaning as well. In using the Semantic Differential technique Special considerations were emphasized as important in the creation of an instrument for research. Osgood discussed the need for a careful selection of the concepts to be rated with the 12 bipolar adjective scales. The concepts should be selected on the basis of their relevancy to the research problem. Also an attempt should be made to choose concepts with ratings that will be distributed across the dimensional factors. That is, the investigator Should choose concepts (e.g., PLAYING and CRYING) that are likely to have different ratings on the adjective scales (e.g., good-bad; fast-slow) (Osgood, Suci, and Tannenbaum, 1957). A second important consideration emphasized by Osgood is the selection of apprOpriate scales or adjective pairs. Two criteria are used in the selection: (1) the representa- tiveness of the bipolar adjectives to the dimensional factors, (2) the relevance of the adjectives to the concepts used. Analysis of data was discussed extensively by Osgood. Factor analysis was suggested as the most appropriate method, but this was in relation to a large number of concepts (10 or more) rated on approximately 20 scales by a sample of at least 100 subjects. Most studies cited in the literature that have used the Semantic Differential technique have been with adults, eSpecially college students. These studies have involved attitude assessment, studies in psycholinguistics, in adver- tising and in aesthetics to name a few of the areas of appli- cation. Researchers' application of the Semantic Differential to young children has been very limited. Small (1958) investi- gated the similarities and differences in the semantic 15 structure of children in relation to age and sex. Her sample was comprised of 275 third, sixth and ninth grade boys and girls. The subjects rated 24 concepts on 16 bipolar adjective scales. Nine of the scales represented the three factors defined by Osgood (E,P,A) and seven other scales of theoreti- cal interest to the investigator were included. Results seemed to indicate that a similar factor structure existed over the age and sex groups selected for the study. Small suggested as a result that the Semantic Differential needed to be used with even younger children to study deve10pmental trends. In 1961 Donahue had four grOUpS of 50 subjects each, with mean ages of 7, 9, 12 and 22 years, rate concepts in order to test for differences in meanings as a function of abstraction level of the Sign. ,Half the subjects at each age level rated pictures and half rated words. Ten concepts including words as SQUIRREL, FLOWER, and LION were rated on nine Semantic Differential scales, three for each major factor (E,P,A). Individual testing was employed with the seven year olds, because of their limited reading ability. Factor analysis and analysis of variance were used and showed Significant effects due to age level. The results suggested that the semantic Space for children was more restricted than it was for adults. Due to the limited number of scales and concepts used, however, it was difficult to draw any firm conclusions 14 with regard to the dimensionality of the semantic Space for children. The abstraction level of the Sign (picture versus word) was found to have no effect on the ratings. Reliability measures were not obtained. Another study utilizing the Semantic Differential to investigate the development of connotative meanings as a function of age was done by Maltz (1965). Seven concepts in- cluding CANDY, SUN, and FRIEND, were rated on 9 five point bipolar scales. Four school grades comprised the sample: 17 second, 25 fourth, 26 sixth and 25 college reSpectively. The grade school children were individually tested. Chi-square was used to test the difference between age levels on each factor for each concept. ,Results showed significant differ- ences between groups and suggested that the Semantic Differ— ential was usable in studying changes in connotative meaning. The study also served to Show that the Semantic Differential could be used with younger children. No reliability indiceS were given. Lilly (1965) attempted to systematically study the de- ve10pmental changes in the dimensionality of the affective meaning system. He examined the dimensions of affective meaning as a function of age and compared the obtained "semantic factors" for cross-age similarity. Twenty concepts were rated on 28 eleven point bipolar adjective scales. Two concepts (MY MOTHER and ELEPHANT) were repeated and Showed that the Semantic Differential ratings were reliable (.96) with children as young as the third grade. A three-mode 15 factor analysis of data was used. The sample included 96 third, 110 fourth, and 107 Sixth graders and 100 high school students. Results seemed to indicate a high consistency with Osgood's three dimensions--evaluation, potency, and activity. There was a representation of factors at each age level. .An_examination of the scaling showed that the older children tended to use more scale factors (seven versus five). Because the children as young as the third grade possessed the major factors of affective meaning, Lilly suggested that still younger age groups should be used in further investigations. Also investigating the developmental emergence of chil- drens affective meaning system, DiVesta (1966) applied the Semantic Differential to children from the second through sixth grades. One hundred subjects from each grade level rated 20 concepts (e.g., TEACHER, LAMB, ENEMY) on the three major meaning factors (E,P,A) plus those of novelty, reality, tautness, and warmth. Results indicated that these factors could be considered as reliable dimensions of the children's use of language. This study seemed to Show that the deve10p- ment of connotative meaning was quite stable by the time the child was in the second grade. The only attempt to apply the Semantic Differential to children below the grade school level was done by Williams (1967). In an investigation of racial attitudes he used the evaluative aSpect of the technique to study children's 16 connotative meanings with regard to the Negro and CaucasifiK/ races. His sample was comprised of 111 Caucasion preschool children ranging in age from 55 to 81 months. Results indi- cated that the Semantic Differential seemed to have potential as a means of investigating evaluative meanings of very young children, that could be compared to adult measures. Williams suggested that this technique might be used in further racial studies with the very young child and that the method might also be employed in the assessment of other attitudes in young children. CHAPTER III METHOD Utilizing the Semantic Differential with very young children necessitated careful planning and a number of modifications in the technique. An individual, oral admin- istration of selected concepts and bipolar adjectives was employed due to the inappropriateness of the usual written format to these nonreaders. In addition, the bipolar adjective scales were modified considerably from the seven point rating scale as designed by Osgood to a possible three point scale. This degree of simplification was done because a prime focus of this research was to discover if the very young child could choose between alternative bipolar ad- jectives. It was felt that the inclusion of a more elabor- ate scale would only serve to increase the difficulty of the task and possibly limit the child's ability to reSpond. The third rating position available, neutrality, was for children's reSponses which indicated that neither of the adjectives presented was relevant to the particular concept being rated (i.e., FLOWER: fast—slow). Special care was taken in selecting the concepts and the bipolar adjectives for this study. Words chosen were within the children's experimential background and were 17 18‘ contained in books for very young children found in the nursery school., In the selection of concepts one additional criterion was used: they could have been experienced through a number of the child's five senses (i.e., MOTHER--touch, sight, hearing). The bipolar adjectives that were selected were divided among the evaluative, potency, and activity dimensions. -An attempt was made to include concepts and ratings that would be distributed across the dimensional factors and rating scales. In order to prevent chance meaning patterns from occur- ring due to the repetitious presentation of words to the children, two equivalent forms of a questionnaire were made. These forms included the same concepts with a reversal of the bipolar adjectives presentation order (Note Figure 1).1 In addition, rotations of adjective pairs were made within one questionnaire form in order to offer variety to the instrument. For example, 'happy-sad' moved vertically from being the second pair presented in relation to BABY to the first pair presented in relation to CANDY. Horizontally, the order of presentation was rotated from 'happy-sad' in relation to BABY to 'sad-happy' in relation to CANDY. 1Figure 1--Entire Questionnaire Form I Score Sheet found in Appendix. Figure 1. 19 Excerpts from Questionnaire Forms' Score Sheets showing positional rotations of adjective pairs. bad happy small clean slow hard sad large clean slow hard good Form I Score Sheet BABY N CANDY good sad large dirty fast soft happy small dirty fast soft bad NR NR good sad large dirty fast soft happy small dirty fast soft bad Form II Score Sheet BABY N CANDY NR bad happy small clean Slow hard NR sad large clean slow hard good Specific criteria for test termination were also established. These were: (1) If the subject gave four "no reSponse' or 'I don't know' replies, or a combination of these in succes- sion, the examiner proceeded to the next concept. If he again elicited two similar reSponses in suc- cession, (2) testing was terminated. Perseverance--If the child selected the final ad- jective Spoken five times in succession for one concept, the examiner proceeded to the next concept. With two additional successive reSponses of the final adjective Spoken, test termination occurred. 20 Pretesting the Instrument A study was conducted to determine whether the direc- tions were easily understood by a preschool age child and whether the selected concepts and bipolar adjectives seemed appropriate for him. Since the attention Span of young chil- dren tendstx>be of short duration, the length of time for testing was noted. Results of pretesting the instrument indicated the need for the following changes: (1) (2) The concept, SPANKING, was eliminated beCause the children evaded rating this word. Reactions to be- ing asked this concept included refusal to talk about it, shifts in the conversation to another subject, and requests to terminate the game. This examiner believed that the intensity of the chil— drenfiemotional identification with this concept was too great for many of them to rate it. The number of bipolar adjective pairs was reduced from nine to six in order to decrease the length of the questionnaire. One pair of adjectives from each of the dimensional factors was eliminated. The average time needed to administer the question- naire was fifteen minutes per child. It was ob- served that most of the children attended to the task for approximately ten minutes without 21 difficulty and then appeared restless and less reSponsive to the questions. Administration of the Instrument Four examiners, experienced in working with very young children, administered the questionnaire. This number of examiners was needed, because the eXperimenter, as teacher of one of the four groups included in the study would possibly bias the data by testing them. Each examiner tested an equal portion of the sample. To insure conformity in testing the following Special measures were taken: (1) Written copies of testing procedures were given to the examiners and (2) practice sessions for administering the questionnaire were conducted. Prior to data collection each examiner Spent time in the nursery school groups in order to establish rapport with the children. When she felt that the children had accepted her she approached the child and said: (Child's Name), I brought a game to school today and I'd like to play it with you. You can have your turn now. Most of the children reSponded to the request and willingly went with an examiner to a testing room. If a child refused an examiner's invitation, a second examiner approached him on another day. If a second refusal was made, the child was deleted from the study. Of a possible sample of sixty-eight children, only three were not included in the study. 22 Sixty-five children from the four groups of the Michigan State University Laboratory Pre-School served as subjects for the study. Ranging in age from 42 to 66 months, the sample consisted of thirty-two boys and thirty-three girls. A pretest was included to assure the children's ability to select between alternatives and to offer them a practice sample. One concept, APPLE, was presented for rating on three bipolar adjective scales not included in the question- naire (little-big, black-red, sweet-sour). To introduce the practice sample and continue with the questionnaire, the examiner said: I have a game I'd like to play with you. This is the way we play. First I say a word, like APPLE. Then I'll ask you a question. Now listen, is an APPLE big or little? Assured that the child understood the task, by his reSponses to the sample item,.the examiner then asked the child to rate nine selected concepts on Six adjective scales (see Figure 22). In order to determine the reliability of the children's ratings, one concept, CANDY, was repeated in the questionnaire administration. Alternative ways used to introduce concepts included: 1. Ready 5. Let's try 2. Next 4. Now In order to maintain interest and encourage the child to continue, supportive reSponses such as hmm, okay, oh, ufifi, uh-huh, and ahh were used. 25 Figure 2. Concepts and Bipolar Adjectives employed in the study. CONCEPTS BABY I CRYING GIRL BOY FATHER MOTHER CANDY FLOWERS PLAYING BIPOLAR ADJECTIVES Sgale Qimensional_§actors good - bad Evaluative (E) clean - dirty Evaluative (E) (happy - sad >Evaluative (E) hard - soft Potency (P) large - small Potency (P) fast - slow Activity (A) 24 Because both bipolar adjectives, were sometimes applic- able to a concept being rated, children selected both ad- jectives (i.e., BOY: large-small). When this occurred the examiner noted this and then asked: Which one is (concept) most of the time? This procedure attempted to have the child select that adjec- tive which seemed most applicable to the concept, as the predominant meaning was sought. In the case of run reSponse the examiner indicated this on the score sheet and then repeated the alternatives once again before proceeding to the next set of bipolar adjec- tives. If a response of “I don't know" was elicited in rela- tion to a set of alternatives, this was noted on the score sheet and the next set of alternatives was presented. A place was included on the score sheet for noting 'neutral reSponses' (i.e., a child replied that neither alternative presented in relation to a concept being rated was applicable). Additional data on the meanings the children attached to the concepts was also recorded. Upon the completion of the rating of a concept, the child was asked in relation to the final selection from bipolar adjectives presented: How do you know (concept, e.g., BABY) is (adjective, e.g., soft)? or why is (concept) (adjective)? 25 Data Collection The data was collected within a four day period, during the morning and afternoon sessions of the nursery school. Three hours each day were used for testing. Approximately Sixteen children were tested each day within two rooms of the nursery school. Each was furnished with a child sized table and two child sized chairs. Retesting was done after a two week time lapse so that the stability of ratings over a short period of time could be examined. Twenty-five of the original sample, who had completed the questionnaire, were retested on five randomly selected concepts. The children were administered the alter- nate forms on the retest. CHAPTER IV RESULTS The hypotheses to be tested were: I The number of items completed by the children will increase as age increaSes. II The incidence of neutral reSponses to those items irrelevant to concepts rated will increase as age increases. The first hypothesis predicting the effect of age in- crement on the child's ability to complete task items was tested by means of the Kruskal-Wallis One Way Analysis of Variance by Ranks. For this analysis the children were grouped by Six month age intervals as follows: Age in Months N Group I 60—66 20 Group II 54-60 15 Group III 48-54 12 Group IV 42-48 18 This statistic was used to test the independence of the four age groups on task performance. The significance level was set at .05, therefore, a chi-square value of at least 7.82 with three degrees of freedom was needed. Because the ob- tained value of 252.425 was Significant, the conclusion that 26 27 the number of items completed varies significantly with age can be drawn. In order to determine the directionality of the rela- tion between age and task performance, the mean number of items completed by each of the four age groups was calcu- lated. These results show that although there is no linear relation, the number of items completed increases with age increment (Figure 5). Note the difference between the mean Figure 5. .Mean number of items completed by groups. 1 52.50 52.15 50__ 50.40 40-. 34-56 50.1; 20 4.. 101- Age in M08. 42-48 48-54 54-60 60-66 GrouP IV III II I N 18 12 15 20 5 19.117 5.09 5.94 9.01 number of items completed by GrouP IV and the remaining groups. In examining Figure 5 one sees that the mean number of items completed by the oldest subjects, GrOUp I, is slightly less than that of Groups II and III. This discrep- ancy may be accounted for by the small number of subjects in 28 each age category. Generally what is shown is a ceiling effect, considering that the total number of items in the instrument is 54. The second hypothesis was tested using a Pearson product-moment correlation between age in months and number of neutral reSponses elicited. The obtained coefficient of .05 did not meet the .05 level of significance. As a result the second hypothesis was not supported. The reliabilities of internal and test-retest ratings were tested by the chi-square test of independence. A sig- nificance level was set at .05. The number of subjects in the retest sample was 27. In testing the internal reliabilities of the six bi- polar ratings on the concept, CANDY, which was repeated once in the original test administration, only one rating on the adjective pair 'good-bad' was significant at the .05 level. The reliabilities of test-retest ratings on the five concepts readministered to the subjects after a two week time lapse were determines. Table I indicates those ratings significant at the .05 level. Although chi-square was the apprOpriate statistic to use with the ordinal dichotomized measures obtained in this study, it posed a serious limitation. This statistic favors a true dichotomy (i.e., an equivalent chance that one item will occur as another; .5 probability). The dichotomies presented in the Semantic Differential are not of this nature. 29 Table I. Chi-square Values Obtained in Analysis of Test- Retest Reliabilities* h “— _:_ BABY CANDY PLAYING FATHER GIRL good - bad 24.141 clean - dirty happy - sad 4.508 large - small hard - soft 5.821 4.197 10.751 fast - slow 6.750 *= .05, x2 5.84 For example, the probabilities of a child selecting either word from the adjective pair 'large - small' in relation to the concept, FATHER, is not .5. Because of this limitation, a number of highly consistent ratings did not achieve sig— nificance. In order to Show this, the percentage of con- sistent reSponses was calculated (Table II). Table II. Percentage of Consistent Test-Retest Responses Internal Ratings Test-Retest Ratings CANDY BABY CANDY PLAYING FATHER GIRL good - bad 81 68 72 64 85 62 clean - dirty 81 68 96 64 85 74 happy - sad 78 56 67 46 95 70 large - small 65 89 58 64 89 61 hard - soft 72 75 71 68 81 57 fast - slow 59 81 57 70 78 67 CHAPTER V DISCUSSION A purpose of this research was to determine if a modi- fied Semantic Differential is feasible with preschool age children. Of the subjects included in this investigation, ranging in age from 42 to 66 months, this investigator feels this technique as modified for this study is generally not apprOpriate to use with children under 48 months of age. Although the number of children falling into this age cate- gory was limited, 18, the fact that only seven completed the task raises question about the value of administering the instrument to such young children. The attention Span of the three year olds to this orally administered task was quite short. These children became satiated with this game that only offered them a supportive verbal encouragement for their Spoken reSponses to questions. AS a result the chil- dren asked to terminate playing or began to persevere in their responses to questionnaire items. The 48 to 66 month old children were almost all able to complete the task (only 4 out of 47 did not finish the questionnaire). The significant results in support of Hypothesis I concerning the increase in the number of task items completed with age 50 51 increment also seems to indicate the greater ability of the four year old child to attend to this task. The marked dif- ference in the performance of Group IV (42—48 mos.) as com- pared to Group III (48-54 mos.) raises a question of whether there is a significant developmental advance during this time which effects the child's ability to be tested or is this only a chance occurrence. The fact that Hypothesis II was not supported in this investigation may be the result of the sample and/or instru- ment design. Since only 15 of the 65 subjects elicited neutral reSponses, this suggests the possibility of a weak- ness. The instrument as designed for this study did not give the child a choice between bipolar adjective pairs and a neutral reSponse. Rather it offered the child a choice be- tween bipolar adjectives and only provided a place on the score sheet for the examiner to note a neutral reSponse, if the child stated that the alternative adjectives presented were not applicable to the concept being rated. AS a result the investigation does not determine if (1) most of the children were incapable of making neutral reSponses, or (2) the children recognized the inapprOpriateness of some of the adjective pairs to concepts being rated, but felt com- pelled to reSpond within the bounds of the alternatives presented. Future investigations might include the neutral alternative as a possible choice in an attempt to study the very young child's ability to make such a discrimination. 52 Further examination of the neutral reSponse data was made in order to see if the children giving these responses were distinguishable from the total sample. It was found that subjects eliciting neutral reSponses ranged in age from 50 to 66 months. An unusual discovery, however, was the fact that although the sample had an equivalent number of boys and girls distributed over the representative ages, 11 of the 15 subjects giving neutral responses were boys.l This phenomenon offers possibilities for further exten- sive investigation. Could this occurrence possibly indicate a difference in the cognitive develOpment of boys and girls? Is this difference possibly due to the types of behavior fostered by our society in relation to the sexes (e.g., girls may be encouraged to stay within the bounds of Speci- fied directions and rules, but boys may be encouraged to assert themselves and voice Opposition when they do not agree with specified ruleS)? Although analysis showed few Significant test-retest reliabilities on the concept repeated in the original test administration and five concepts included in the retest, the calculated percentages of consistent test-retest reSponses lA Mann-Whitney U-Test was done to determine if this was a chance happening or of Significance. This tested, on the basis of the number of neutral reSponses elicited, if these groups (boys vs. girls) were drawn from the same pOpulation. Analysis of data resulted in the rejection of the null hy- pothesis of no difference at the .025 level of significance. 55 generally showed a notable degree of stability in ratings (see Table II, page 29). For example, the reSponses on the three evaluative ad- jective pairs (good-bad, clean-dirty, happy-sad) in relation to the concept CANDY, which tested the internal reliability of the original test administration ratings, showed approxi- mately an 80% retest consistency. Also, noteworthy, were all Six of the test-retest reSponses on bipolar adjective pairs in relation to the concept FATHER. These reSponses ranged from 78-95% consistent from the first to the second adminis- tration. Considering the subjects highly consistent retest reSponses on the concept FATHER after a two week time lapse, it is likely that the reliabilities of ratings on concepts in the original test are higher than might be eXpected on the basis of measures obtained on the concept selected to determine internal test rating reliabilities. Content analysis of the reSponses to the Semantic Differ- ential scales was made in an endeavor to discover what con- notative meanings these very young children attach to the selected concepts in this investigation. All subjects com- pleting the questionnaire between 48 and 66 months of age were included.1 Analysis was made with reSpect to sex, age differences and the total group. Only those findings which help to lThis investigator felt the small number of subjects between 42 and 48 months of age completing the questionnaire was too few to include in this analysis. 54 eXplain meaning trends found through this analysis are re- ported. One trend was that the children seemed to be estab- lishing meaning systems that are similar to those an adult in our society might be expected to use in describing the selected concepts with the bipolar adjectives provided. For example, the concept PLAYING was rated as 'happy' by .718 of the subjects but only .199 rated the concept CRYING in this way. The prOportion of subjects rating BABY as being 'Slow' (.977), 'small' (.905), and 'soft' (.725) was very high. .Similarly a high prOportion of the children viewed MOTHER as being 'clean' (.929), 'large' (.955), and 'soft' (.714). This is not surprising, as Carroll (1960) points out that a language contains a system of socially shared meanings which must of course be learned by the child. Also these results might be expected since Vygotsky (1959) has stated, "Many words therefore, have in part the same meaning to the child and the adult, eSpecially words referring to concrete objects in the child's habitual surroundings." Another trend that seemed to appear was that all the subjects generally made similar reSponses in relation to the familial concepts MOTHER and FATHER. ~For example, data on the proportion of 'good' ratings on these concepts showed: (1) In relation to MOTHER, .955 of the boys and .905 of the girls gave 'good' ratings. 7(2) Similarly .952 of the boys and .900 of the girls rated FATHER as 'good.' Also these concepts rated on the bipolar pair 'large-small' Showed 55 similar results. MOTHER was rated as 'large' by .905 of the boys and girls. FATHER received a 'large' rating from .905 of the boys and .952 of the girls. The data seemed to Show sex differentiated ratings in relation to the concepts GIRL and BOY. Only .474 of the boys compared to .810 of the girls rated GIRL as 'good.' However, .800 of the boys compared to .524 of the girls rated BOY as 'good.' Only one of the bipolar pairs 'large-small' Showed a variation from this apparent trend. The concept GIRL was rated as 'large' by .455 of the boys and .571 of the girls. In contrast BOY was rated as 'large' by .810 of the boys and .710 of the girls. Could this difference indi- cate that the children have internalized a possible cultural- ly differentiated value in male bigness that does not apply to females? Also worth noting is the fact that some of the children when asked how they knew CANDY was 'good' reSponded, "Because." Such a reSponse could indicate a number of things. Inhelder and Matalon (1960, p. 454) in discussing this say, "Children do not generally eXpress their beliefs because they think that everyone believes as they do, because they are afraid of making a mistake or, finally, because their ideas are not sufficiently systematized to be formulated." In looking at this data, although some of the bases for mean- ings attached to the selected concepts were in information probably acquired from adults or inexplicable due to the very 56 limited nature of the reSponses, the majority of the chil- dren's meanings seemed to be based on personal sensory experiences. Concerning limitations of this investigation, the sample, selected concepts and bipolar adjectives, and instrument de— sign will be discussed. A high percentage of subjects used in this exploratory study, had parents with more than four years of college training. Because of this, results may not be representative of what the majority of preschool age chil- dren would attain on the task. The number of concepts and bipolar adjectives included in this investigation was small. Future investigators might want to extend this instrument by including additional concepts and adjective pairs to be rated in a number of questionnaire administrations so it might be used to learn about the deve10pment of connotative mean- ings in these very young children. wAlso it would be valu- able to extend the rating scale by adding one or more pos- sible alternatives. AS discussed earlier, one possible extension of the scale might be the inclusion of the neutral alternative. As a result of this study, further research application of the Semantic Differential to preschool age children is recommended. This experimenter feels the application of this technique could yield pertinent information regarding a child's feelings about his family relationships. Considering the high proportion of similar ratings on the familial 57 concepts MOTHER and FATHER included in the present study, an extension of the number of family members and the addi- tion of other concepts associated with family life (e.g., HOME) rated on apprOpriate bipolar adjectives, might be of great value in learning how the very young child views his family milieu. .Further studies with this technique in the area of attitude assessment in regard to race, sex, and/or prejudices could also be designed. CHAPTER VI SUMMARY AND IMPLICATIONS The purpose of this research was to determine if an orally administered Semantic Differential was feasible with children ranging in age from 42 to 66 months. Also an en— deavor was made to discover what meanings young children attach to the selected concepts as well as explore the bases of these meanings. Results of this study seemed to indicate that this tech- nique is usable with most children 48 months of age and older. The majority of children younger than this were not testable. The potential applicability of this technique to a high per— centage of preschool age children is of importance, however, considering that first grade children have been the youngest known group to whom the Semantic Differential including the major dimensional factors has been administered. Two hypotheses were tested in this study. I. The number of items completed by the children will increase as age increases. Results of the study supported this hypothesis. Although no linear relation existed, a significant difference in the mean number of items completed by the youngest subjects 58 59 (42-48 mos.) and the next oldest group of subjects (48-54 mos.) was found. II. The incidence of neutral reSponses to those items irrelevant to concepts rated will increase as children's age increases. This hypothesis was not supported by investigation find- ings when tested using a Pearson product-moment correlation. .Test-Retest reliability was examined and showed a sig- nificant stability in regard to a few of the ratings. The percentage of consistent reSponses on test-retest items was also calculated and seemed to indicate that a greater reli— ability probably existed, than might be eXpected by looking at the results of the statistical analysis on test items. Apparent trends in the connotative meanings attached to the selected concepts, that were found in the content anal- ysis of the data included: (1) Children seemed to be estab— lishing meaning systems that are similar to those an adult in our society might be exPected to use in describing the selected concepts with the bipolar adjectives provided. (2) Boys and girls seemed to give similar reSponses in rela— tion to the familial concepts included in the study. (5) Sex differentiated ratings seemed to appear in relation to the concepts BOY and GIRL. Exploration of the bases of the meanings attached to the selected concepts by young children indicated that the majority of the subjects in this study used sensory experi- ences . 40 Implications for Research The results of the present study indicating that the Semantic Differential is feasible with very young children, would SUpport the recommendation for further research with these children. By extending the number of concepts and bipolar adjectives in a series of questionnaire administra- tions, investigators might use this technique to learn about the deve10pment of connotative meanings in these young children. An extension of the rating scale could also be made for an investigation of the children's ability to cope with this more complex task. Further study into the significant difference found between the number of boys (11) and girls (4) giving neutral reSponses might be undertaken. Investigators could possibly discover if this was only a chance occurrence or possibly indicated a difference in the cognitive development or societal role eXpectations of the sexes. This technique Shows potential value in providing im- portant information about a child's attitudes toward his family and environment. Considering the high proportion of similar ratings obtained on the concepts MOTHER and FATHER included in the present investigation, an extension of the number of familial concepts and inclusion of other concepts associated with family life (e.g., HOME) rated on apprOpriate adjective scales, might prove to be of value in learning 41 how the very young child views his environment and its members. An assessment of the young child's attitudes toward other concepts as his peers or races might also be of value. REFERENCES Anastasi, Anne. Standardized ability testing. In Paul H. Mussen (Ed.), Handbook of Research Methods in Child Development. New York: John Wiley & Sons, Inc., Pp. 456-458. Berko, Jean and Brown, Roger. Psycholinquistic Research Methods. In Paul H. Mussen (Ed.). »Handbook of Research Methods in Child Development. New York: John Wiley & Sons, Inc.,-1960. Pp. 517-557. Carroll, J. B. Language development in children. In EncyclOpedia of Educational Research, 1960. Pp. 744- 752. Church, Joseph. .Language and the Discovery of Reality, New York: Random House, 1961. Pp. 185-187. .DiVesta, Francis J. A deve10pmental study of the semantic structures of children. Journal of Verbal Learning and Verbal Behavior, 1966,5L Pp. 249-259. Donahue, John W. Changes in meaning as a function of age. The Journal of Genetic Psychology, 1961, 99, Pp. 25-28. Ervin, Susan.M. and Foster, Garret. The development of meaning in children's descriptive terms. Journal of Abnormal and Social Psychology, 1960, 61, Pp. 271-275. Ervin-Tripp, Susan. Language deve10pment. In L. W. Hoffman and M. L. Hoffman (Eds.). Review of Child DevelOpment Research. .New York: Russell Sage Foundation, 1966. Pp. 55-105. Inhelder, Barbel and Matalon, Benjamin. The study of prob- lem solving and thinking. In Paul H. Mussen (Ed.). Handbook of Research Methods in Child Development. .New York: John Wiley & Sons, Inc., 1960. Pp. 421-455. Irwin, Orvis, C. Language and communication. In Paul H. Mussen (Ed.). Handbook of Research Methods in Child Development. ,New York: John Wiley & Sons, Inc., 1960. Pp. 487-516. 42 45 Kerlinger, Fred N. Foundations of Behavioral Research. New York: Holt, Rinehart and Winston, Inc., 1966. Lilly, Roy S. A deve10pmental study of the Semantic Differential. Unpublished doctoral dissertation, Princeton University, 1965. McCarthy, Dorthea. Language development in children. In L. Carmichael (Ed.). .Manual of Child Psychology. New York: John Wiley & Sons, 1954. Pp. 492—650. Maltz, Howard E. Ontogenetic change in the meaning of concepts as measured by the Semantic Differential. Child DevelOpment, 1965, 54, Pp. 667-674. Osgood, Charles E. and Sebock, Thomas A. (Eds.). Psycho- linguistics: A survey of theory and research problems, Bloomington, Indiana: Indiana University Press, 1965. Osgood, Charles Egerton, Suci, George J. and Tannenbaum, Percy H. The Measurement of Meaning. Urbana, Illinois: University of Illinois Press, 1957. Patrina, K. T. On the formation of linguistic deve10pments in the course of Speech development. Voprosy Psikhogie, 1959, No. 5, Pp. 112-125. Piaget, J. The Language and Thought Of the Child (Trans. by M. Warden), New York: Harcourt, Brace, 1926. Remmers, H. H. Rating methods in research teaching. In N. L. Gage. Handbook of Research on Teaching, Chicago: Rand McNally & Company, 1965. Pp. 529-578. .Shirley, M. M. Common content in the Speech of preschool children. Child Development, 1958, 9, Pp. 555-546. Small, Edna R. Age and sex differences in the semantic structure of children. Unpublished doctoral disserta— tion, University of Michigan, 1958. Smith, Madora Elizabeth. An investigation of the deve10p- ment of the sentence and extent of vocabulary in young children. University of Iowa Studies Child Welfare, 1926, No. 5, Vol. III, p. 92. Van Alstyne, D. The environment of three year old children: Factors related to intelligence and vocabulary tests. Teachers college Contr. Education, No. 56, 1929. 44 Velten, H. V. The Growth of Phonemic and lexical patterns in infant language. Language, 1945, 19, Pp. 281-292. Vygotsky, L. S. Thought and Language. Cambridge, Massa- chusetts: The M.I.T. Press, 1962. Wellman, Beth L., Case, Ida M., Mengert, Ida G., and Bradburg, Dorothy E. Speech sounds of young children. Iowa City: University of Iowa Studies Child Welfare, No. 2, 1951. Whipple, Guy Montrose (Ed.). Studies in language deve10p- ment. The Twenty—eighth Yearbook of the National Society for the Study of Education, 1929, 28, Pp. 495- 568. Williams, H. M. and McFarland, Mary L. A revision of the Smith Vocabulary test for preschool children in deve10p- ment of language and vocabulary in young children. Iowa City: University of Iowa Studies Child Welfare, 1957, 15. Williams, John E. and Roberson, J. Karen. A method for assessing racial attitudes in preschool children. Education and Psychological Measurement, 1967, 27, -Pp. 671-689. APPEND IX 45 46 30Hm Dmmm* 0mm mmmm£* :mmHo human ooom omn HHmEm mmuma umom ohm: mmmm: omm ummm 30am omn ooom wunflp cmoao puma umom wmuma HHmEm m2 2 m2 2 mmmaoz Mom muuflo cmmao* omn ooom* momma HHmEm Dmow ohm: omm mmmmn ummm 30am ooom omn Sumac Gmmau umom puma HHmEm momma ummm 30am momma omm 111 m 11%! oszdqm wazgo HHmEm omnma* umom oum£* mamms omm ummm 3on omfl ooom munflo cmmao Gums umom momma HHmEm 30am ummm Umm magma cmoao huuflo Uoom omn m2 2 mz z mmmzoqm Mmmm noon ummBm .oEH IIIIIII|.. mm o .llllll . no men 11 5183 . .e .u u m L 9 men mauufla "HmcHmem "Honomma m2 2 mamm< unmououm "xom "8mg "oEmz 47 111111 m 2 m 2 m 2 omm oooo umom ummm mmmmm momma omn omm: 3oam cmmao HHmEm wmmmn umom mmmm Somme momma omm oooo . 111111 AMHO Z mfimfifim . 11111 UZHMMU . mamm£* omn ommz 30Hm cmmao HHmEm oooo* umom ummm wummm momma omm Ummz* 30am Gmmau HHmEm mammn man umfimz I111111111111111111111 r