MUSIC AS A FILM VARIABLE THESIS FOR THE DEGREE Oi? FHA). meme»: 5mm: UNWERsm RECHARfl H. GERRERG 1 95$ IININIUIHHHIHI”1|!!!“thlHlNlUllUHIIIHIW 1293 10589 5902 This is to certify that the thesis entitled Music as a Film Variable presented by Richard H Ge rre ro has been accepted towards fulfillment of the requirements for Ph, D, degree in Qurriculum 0-169 i I B R A R 1“ Michigan Q in re University 1 ABSTRACT MUSIC AS A FILM VARIABLE By Richard H. Gerrero Educators, film makers and critics frequently question the function and value of music in the instruc- tional film. A review of the sparse efforts to study music as a film variable revealed that "there is little experimental evidence to suggest that musical background has any marked effect on learning from instructional films." This view is not shared by those who design and produce films; consequently, film makers and media specialists who desire to employ music in their educa- tional presentations must design their messages intui- tively and according to traditional patterns. A more rational use of music in the design of instructional messages requires knowledge of the functions and inter- actions between the film and music elements. This study investigated congruency and structure as two of the con- ditions which systematically influence the relative effectiveness of the music component. The first level of the study, however, simply asked if the perceived Judgments of a filmed visual can be significantly altered Richard H. Gerrero by the music that is made to co-exist with the scene. The second level of the study investigated the usefulness of the point of resolution formula in predicting the Judgments of film with music combinations from knowledge of the component parts. The third level considered the effects of congruency and structure and the interaction of these conditions on component performance. The broadest generalization from the findings was that music appears to exercise considerable influence when made to co-exist with a film scene. The influence was systematic and was predicted with considerable ac- curacy with Osgood's point of resolution formula. Over- all, the film component had a greater influence on the film with music combination than did the music. This influence, however, varied considerably among examples. The precision of the point of resolution predictions as well as the relative influence of the separate components seemed to be highly dependent on the degree of congruency existing between the two components. Structure, an assess- ment of viewer consensus, also affected component relation- ships, but to a lesser degree. Music influence was found to be greater than the film component influence only in an incongruous combination with high film structure and low music structure. MUSIC AS A FILM VARIABLE By v 1 Richard H. Gerrero A THESIS Submitted to Michigan State University in partial'fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY College of Education 1969 ACKNOWLEDGMENTS My debt of gratitude, although acknowledged, can never be repaid. Many have contributed to whatever intellectual merit may reside in this thesis. Their counsel, their guidance, their patience, unselfishness and generosity of time has allowed a curiosity to develop into a researchable idea. In particular, I would like to thank Dr. Charles Schuller for his advise, help and encouragement through- out my graduate career. Professor Edward McCoy will always be remembered for his stimulating and penetrating discussions. I am also deeply indebted to Dr. Erwin Bettinghaus, for without his faith and genuine interest, my return to graduate study would not have been possible. And then there are all the graduate fellows, like Chuck Bollmann, who not only shared the midnight lamp to probe alternatives but also aided immensely in the routine . tasks of collecting and scoring data. For all those who have assisted, helped, and tolerated me, I am grateful. 11 TABLE OF CONTENTS I. THE PROBLEM .................. . ............... Need for the Study ............... ........ Purpose and Scope of the Study ... ......... II. REVIEW OF THE LITERATURE ...... ............... Film Variable Studies ..... ...... . ......... Film Music Studies ... ..... ........... ..... III. METHODOLOGY AND PROCEDURE .................... Procedure .... ........... . ................. Film examples .................... ..... . Music examples ............ ............. Questionnaire construction and administration ...... ..... ....... Experimental populations ............... Equipment .................... ..... ..... Presentation order ..................... Methodology ............................... Level one ..... ..... . ...... .. ...... ..... Level two ......OOIOOOOOOOOOO0.0.00.0... Level three ............. .... ....... .... IV. RESULTS 000.000.0000...00.0.0000. OOOOOOOOOOOOO Prerequisite Statistical Tests ............ Component Influence ....................... Judgment predictions ................... Relative component influence ...... ..... Congruency ................................ Congruency and point of reSolution predictions ....... ..... .. Congruency and component effect ...... .. Structure ................................. Structure and point of resolution predictions .............. Structure and component effect ......... Component influence by levels of congruency and structure ............ Summation Effect .......................... Scales ....................... ...... ....... 111 \0—4 -< uH4 H Scale relevancy ... ....... Direction consensus ..... ...... ......... Factor analysis ........... V. SUMMARY AND CONCLUSIONS ......... Hypotheses Scale Analysis Future ResearCh 000000 OOOOOOOOOOOOOOOOOOOO. Conclusion BIBLIOGRAPHY ....... APPENDICES ......................................... Appendix A Appendix B Appendix C Appendix D Appendix E iv 10. ll. 12. 13. 114. 15. 16. LIST OF TABLES Predicted with observed scores correlated by examples . ....... .................. Predicted with observed scores correlated by scales 00.0.00...OOOOOOOIOOIOOOOO Sum of D2 by types and combination ..... Sum of D2 by levels2of congruency - Point of resolution D , Music D , Film D , and Summation effect ............... Rank order of relative structure ......... 2 D between point of resolution prediction and observed scale mean summed by levels of structure ................ 2 q ....... D between components and combined examples by levels of structure ............. Net difference between music influence and film influence by combined levels of structure and congruency ........... Summation effect averaged by combined levels of structure and congruency ........ Scale relevancy and direction consensus for music examples ..................... Scale relevancy and direction consensus for filmexanlples ......IOOOOOOOOOOOOOOO Scale relevancy and direction consensus for combined film-music examples ....... Scale factor comparisons ................ Proportions of variance ................ Rotated factor loadings — music examples Rotated factor loadings - film examples V 36 36 38 Al A4 “5 146 A8 50 52 53 514 56 57 59 6O 17. 18. I9. 20. 21. 22. 23. 2A. 25. 26. 27. 28. 29. 30. Rotated factor loadings - film with music combinations ..... ............ Varimax rotation analysis of scales used with Population A, music examples ......... Varimax rotation analysis of scales used with Population A, film examples .......... Population B, music example scale means, N=68 ............................... Population B, film example scale means, N=68 ........................ Film with music combination predicted scale means, P.O.R. ....................... Film with music combination observed scale means, N=7l .................. One—way analysis of D2 between point of resolution predictions and observed scale means by levels of congruency One—way analysis of D2 between music component scale means and film with music combination scale means by levels of congruency ............... One—way analysis of D2 between film component scale means and film with music combination scale means by levels of congruency ............... Fisher-Student T-Test values between levels of congruency for film and music components ........... .... One-way analysis of D2 between point of resolution predictions and observed scale means by levels of music structure .......... ........ ... ..... One-way analysis of D2 between point of resolution predictions and observed scale means by levels of film structure ..................... ..... One-way analysis of D2 between music component scale means and film with music combination scale means by high and low levels of structure ... vi 61 81 82 85 86 87 88 89 9O 3 U) UL) w l. R) One-way analysis of D2 between film component scale means and film with music combination scale means by high and low levels of structure ....... 93 Subject D2 scores summed by scales ........... . 9Ll Component values summed by combinations ....... 95 vii 1. LIST OF FIGURES Design of the experiment ...................... viii CHAPTER I THE PROBLEM Need for the Study Awkward as it may be for those who design and pro- duce instructional materials to admit, very few decisions reflected in a film message or a TV lesson are based on either a theory of media design or the findings of em— pirical experiments. For most of the time, film makers and media specialists design their message or illustra- tion intuitively according to traditional patterns, as Lumsdaine's extensive review of media research suggests concerning film: Often the content and sequencing are governed as much by practices and traditions borrowed from the field of theatrical motion pictures as by considerations of pedagogy. The respects in which films are likely to differ instruc- tionally from kinesc0pes or video—tape record- ings of "live" televised instruction probably lie more in philosophy and practice of produc- tion than they do in inherent media differences. Later Lumsdaine proposes: What is heeded are experiments which seek to reveal the influence of specific factors in 1A. A. Lumsdaine, "Instruments and Media of Instruc— tion," Handbook of Research on Teaching, ed. N. L. Gage, (Chicago: Rand McNally & Co., 1963), p. 588. 2 the design characteristics of the media. In this way we can obtain experimental data to support the validity of generalizations on whigh to base future design decisions about media. Among his "specific factors," Lumsdaine lists the use of music in instructional films, a frequent topic among film makers and critics. But the almost complete lack of research into the function or value of film music invites attention. As Travers puts it: If music makes any contribution to learning from film media, and this is an hypothesis yet to be tested, then the problem becomes one of establishing what is suitable and apprOpriate in terms f the effects that music has on the audience. Travers' comment implies two issues: first, that of es- tablishing what effects film music has on an audience, and second, that of Judging the suitability and appro- priateness of those effects. This second question, truly a design question, may be evaluated once we identify a set of media elements that can be manipulated and estab- lish some prediction of how these manipulations affect a target audience. The general task, then, is to determine the principles that govern the effectiveness of film music. This study, it is hOped, will investigate some of 2Ibid., p. 601. 3M. w. Travers et al., Research and Theory Related to Audiovisual Information Transmission, (rev. ed.: Kalamazoo, Western Michigan University, 1967), pp. “7-8. 3 the dimensions of film music and, hepefully, suggest some of the general principles underlying the functions of music as a film variable. Purpose and Scope of the Study This study seeks first to establish that music can influence how a film is interpreted and then to explore some of the relationships between the visual and musical components. More specifically, the study will undertake to answer the following: 1. Can different music cause a change in the perceived meaning for a given film example? 2. Can the Judgments for a combined film with music presentation be predicted from knowledge or measures of the component parts? 3. Is the relative influence of the film and music components systematically influenced by conditions of congruency and structure? To study music as a film variable, one must first determine whether the meaning of a scene sometimes depends upon the music which accompanies it. It certainly does in the movie industry when, for instance, most filmed scenes of wild animals fighting are in fact tamed animals playing; the addition of the "fight" music causes the audi- ence to perceive the scene as a real fight. One show on Walt Disney's Wonderful World of Color TV series, illustrating how films are made, presented a scene with two lions accompanied by playful music, and then repeated the scene with fight music: the two kinds of music produced unmistakably different effects. And A similarly, Saul Bass discussing the famous "Cat Fight" used in the opening of his Walk on the Wild Side described the fight illusion thus: An examination of this footage will reveal that there is very little contact between the two cats. Actually-infact, [sic] there was no "fight." We threw the black cat on the white cat. Camera blurs and skillful editing of isolated cuts showing the cats snarling and pawing the air certainly aided the fight illusion. But it is doubtful that the audience could have perceived a fight without the fierce music used in the sound track. Assuming that music can in fact alter our interpreta- tion of a scene, we must then inquire into the relation— ships of the component parts. If the meaning is known for a film scene and also for the segment of music to accompany it, is it possible to predict the meaning that will be perceived when the two are combined? Will picture and music contribute equally to the combined interpretation, or will the influence of one predominate over that of the other? 5 Just as Kerrick's picture and caption studies suc- cessfully predicted combined effects using Osgood's formula “Saul Bass, "Opening for Walk on the Wild Side," Pagina, International Magazine of Graphic Design, TMilan, Italy, Gennaio 196A7, pp. 16-7. 5Jean S. Kerrick, "The Influence of Captions on Picture Interpretation," Journalism Quarterly, 32:177-82 (Spring 1955). 5 6 for the point of resolution, so the principles of con- gruency and the point of resolution formula might predict with equal success the combined effects of film and music. The studies involving photo—captions concerned the effect of the congruity and incongruity between pictures and captions on the final Judgment.7 Incongruent captions occasionally produced effects opposite to those predicted; and, when both components were on the same side of the scale, their combined effect summed rather than resolved as predicted by balance theory. Kerrick suggests that a caption is better able to change the meaning of a picture if the picture is ambigu— ous,8 although she did not investigate the effects of component ambiguity on interpretation. Such an investiga- tion requires first that some gauge of ambiguity be es- tablished. In a study of producer consensus and audience consensus regarding a series of TV productions,9 Greenberg compared the differences between all possible pairs in Judgments toward the TV programs measured on a Semantic 6C. E. Osgood, G. J. Suci and P. M. Tannenbaum, The Measurement of Meaning, (Urbana: University of Illinois Press, 1957), p. 207. 7Jean S. Kerrick, "News Pictures, Captions and the Point of Resolution," Journalism Quarterly, 3U:183—89 (Spring 1959). 8Ibid. 9Bradley S. Greenberg, "The Effects of Communication Incompatibility on Children's Judgments of Television Programs,: Journal of Broadcasting, 8:157-71 (Spring 1965.) 6 Differential, and computed a measure of "co—orientation"lo for each evaluated program. Quite likely the same tech- nique could be used to measure the relative ambiguity of film and music examples. Only after thus scaling a series of musical examples and a series of filmic examples from high ambiguity to high consensus could one examine the relative influence of component ambiguity on total effect. For instance, music might function as a film variable only in the pres- ence of a highly ambiguous picture, which would help ex- plain some of the difficulty previous research has had in establishing the functions of music in film. loTheodore M. Newcomb used the term co—orientation to describe the dynamic interaction of a communication system which involved the homogeneity of orientation toward cer- tain objects, the homogeneity of perceived consensus, and the attraction among members. See "An Approach to the Study of Communicative Acts," Psychological Review, Vol. 60, No. 6. 1953, pp. 393-AOA. As pointed out by Newcomb, "The very fact of using language or gesture presupposes the assumption of consensus among communicants as to the information transmitted by the use of symbols." In this present study, co-orientation refers to a static measurement technique that attempts to deal only with the homogeneity of orientation toward music and film examples. This measurement technique was developed by Richard R. Carter in a paper presented at the convention of the Association for Education in Journalism, "A Measure of Communication Effectiveness for Messages to Groups," Chapel Hill, North Carolina, 1962. Operational procedures for deriving levels of structure from measures of co- orientation are presented in Chapter III, pp. 28—9, and in Chapter IV, p. A3. CHAPTER II REVIEW OF THE LITERATURE Film Variable Studies Existing film research can be grouped under five maJor headings. The first and largest, stimuli compari- son, sets one form of presentation against a conventional lecture or some other mode, the variable being recall or some other measure of learning. The second group of studies tests viewer preferences for color vs. black & white, picture vs. drawing, sound vs. silent, etc., and occasionally correlates these with age, sex and other viewer variables. The third classification manipulates one or another element of the medium as the experimental variable: e.g., camera angle, embellishments, or sound. The fourth type investigates utilization and environ- mental conditions such as group size, teacher and peer- group attitudes, viewer comfort, room noise, and other independent variables. And fifth, film may be used for testing postulates from other disciplines such as informa- tion theory1 and learning theory.2 Film studies in 1M. W. Travers et al., Research and Theory Related to Audiovisual Information Transmission, (rev. ed.: Kalamazoo, Western Michigan University, 1967). 2A. A. Lumsdaine, "Audio-visual Research in the U.S. Air fogc§," Audio-Visual Communication Review, 1:76-90, (Spring 9 3 . 7 8 which elements of the media are manipulated are perhaps the type to be found least frequently in the literature. The majority of existing film variable studies grew out of three major research projects: the Pennsylvania State Army-Navy studies, l9A2-50; the Yale Motion Picture Research Project, l9u6-5A; and the Air Force Human Re- sources Research Laboratories Studies, 1950-57. Since these three major efforts, few studies involving film variables have been conducted, and the number has been decreasing. Noting this trend, Greenhill states: This type of research has tapered off in re— cent years possibly as the result of a shift of emphasis to research on television and partly because such research studies require good film production facilities and the kind of adequate financing that can only come from a sustained program of research. Hoban and Van Ormer reported on the Pennsylvania State studies in the course of abstracting over 200 film research studies in Instructional Film Research 1918- 1950,“ and provided introductions, summaries, and implications for each class of research. May and 3Leslie P. Greenhill, "Review of Trends in Research on Instructional Television and Film," Research in In— ‘structional Television and Film, (Washington: U.S. Government Printing Office, 1967), p. 15. ”C. F. Hoban, Jr. and E. B. Van Ormer, Instructional Film Research, 1918-1950, Pennsylvania State University Instructional Film Research Program, (Port Washington, N.Y.: U.S. Naval Training Device Center, Office of Naval Research, Tech. Rept. No. SDC 269-7-19, 1950). 9 Lumsdaine's book, Learning from Film,5 develops and ex- pands the principles investigated by the Yale research project. And a volume edited by Lumsdaine, Student Response in Programmed Instruction,6 reports a large number of film variable studies growing out of the Air Force project investigating such principles of learning as repetition, motivation, feedback and others. Finally, Reid and MacLennan's Research in Instructional Television and Film7 reviews about 100 film studies conducted since the three major research efforts referred to above. Film Music Studies Despite the considerable number of film variable studies, scarcely anyone has studied the music function in film. And the few studies that have been reported offer little evidence for the value of music. Hoban and Van Ormer state "there is little experimental evidence to suggest that musical background has any marked effect on learning from instructional films."8 Lumsdaine and Gladstone find that films presented with musical 5M. A. May‘and A. A. Lumsdaine, eds., Learning from Films, (New Haven:. Tale University Press, 1958.) 6A. A. Lumsdaine, ed., Student Response in Pro- ggammed Instruction, (Washington: National Academy of Science, 1961). 7J. C. Reid and D. W. MacLennan, eds., Research in Igstructional Television and Film, (Washington: U.S. Government Printing Office, 1967). 8 Hoban and Van Ormer, p. 8-33. 10 embellishments are actually less effective for learning 9 than the plain version. And Travers complains of a 10 "dearth of experimental literature," reporting only one study in which music was experimentally manipulated.11 Travers postulated that "experimentation here suggests that music does not add to the communicative effective- ness of an informational film."12 The most systematic effort to identify film music 13 functions, the Zuckerman study, reviewed the writings of film critics, film music composers and film makers to find what they considered the function of music to be in film. He states: Since there is no information provided by re- search on how motion picture music assists learning, if indeed it does, a beginning must be made somewhere. Intuitive conclusions, and opinions of composers, musicians, and film critics (mostly referring to entertainment film music and its functions), have been organized 9A. A. Lumsdaine, and A. I. Gladstone, "Overt Prac- tice and Audio-Visual Embellishments," in M. A. May and A. A. Lumsdaine, eds., Learning from Films, (New Haven: Yale University Press, 1958), pp. 58-71. 10 Travers, p. A7. 11D. M. Neu, "The Effect of Attention-gaining Devices on Film-mediated'Learning," Journal of Educational Psychol— 052, “2, 1951, pp- “79-90. l2Travers, p. A8. 13J. V. Zuckerman, Music in Motion Pictures: Review of Literature with Implications for Instructional Films. Pennsylvania State University Instructional Film Research Program, (Port Washington, N.Y.: U.S. Naval Training Device Center, Office of Naval Research, Tech. Rept. No. SDC 269-7-2, 19A9.) ll here so that they may provide suggestions and even testable hypotheses for research on in- structional films.lu Zuckerman thus assembled a series of functions claimed for music in motion pictures: I. Informational Functions a. Delineation of personality or of an actor's character. Provision of subjective evaluation for an objective image. Emphasis for action. Telling a story. Recalling past events. Foretelling the future. 0‘ H3“) {10 II. Emotional Functions a. Establishment of atmosphere or mood. b. Adding to the emotional tone or mood of incidents. c. Pointing up dramatic or comic high- lights of the film. III. Conceptual and Integrative Functions Unification of dramatic material. Association of ideas. Connection of dialogue sequences. Accompaniment for sequences of silent action. CLOUD) In the same way, certain film music techniques were iden- tified and listed, for instance, the "Mickey Mouse" effect where music tempo and rhythm are synchronized with the film as in cartoons and comedy, and the use of loud or sinister chords to cue the audience to something unusual shortly to take place. Zuckerman's lists, though scarcely exhaustive, cer- tainly provide a starting point. A series of applications 12 to learning also presented by Zuckerman, derived from established learning principles, suggests that music can be used for perceptual direction, motivational reward, conceptual aid, memory reinforcement, and the determina- tion of attitudes and opinion.15 Zuckerman's presumed intention was the testing of proposed relationships and functions, but no references identify the studies that must have followed. Perhaps his suggestions could not be confirmed and the studies were therefore not reported. But it is more likely that succeeding experiments focused on measuring and validating the claimed instructional merits of music, rather than identifying the performance characteristics of music as a film variable. So when the literature reports "music does not add to the instructional effectiveness," we do not know whether the functions claimed for music are in- valid or if the way in which we now use music in educa- tional films is simply inappropriate. l51pm, pp. 12-13. CHAPTER III METHODOLOGY AND PROCEDURE This study investigated music as a film variable. The first attempt was to determine if the perceived judg- ments of a filmed visual with musical accompaniment can be significantly altered by the music that is made to co- exist with the scene. The measuring instrument was the Semantic Differential. Given that an influence can be established for the musical accompaniment, the second level of the study investigated the usefulness of the point of resolution formula in predicting the judgments of the combined film with music from knowledge of the component parts. The third level considered the effects of congru- ency between film and music and the relative influence of the component parts. The final analysis was of the rela- tive co—orientation between subjects on a given component and the effects of this structure or lack of structure on the performance of the component parts. Each level of the experimental design will be discussed after a review of the procedures. Procedure The sampling problem in this study involved the identification and selection of a set of film visuals l3 1A and a set of musical examples with certain specific char- acteristics. That is, there needed to be some visuals that were ambiguous and others that were structured or reflected a high degree of consensus among the viewers. In addition to these needs, the samples had to be suffi- ciently diversified so that incongruous as well as con- gruous combinations could be made. A random sampling procedure was therefore not appropriate. A film editor and music expert were used to make preliminary selections. This insured that examples of the required characteristics and dimensions were at least available in the sample. Final selection was made through measures derived from the pilot population. Film Examples In order to obtain film examples that would reflect a range representative of the assumed dimensions of struc- ture and meaning, about 200 educational films1 were viewed on a high speed Moviola film reader. A variety of repre- sentative scenes were identified and excerpted (cut out of the film).2 The 60 scenes so selected were then viewed repeatedly, each time selecting those which intuitively 1A collection of over 200 16mm sound motion pictures was made available through the courtesy of Dr. Elwood E. Miller, Director, United States Office of Education, Single Concept Film Clip Project. This collection was housed on the campus of Michigan State University. (OE-A-16-030) 2The sequence of procedures is shown in schematic form in Figure l. IS Figure I Design of the Experiment. )- 9“ z 200 300 O . . E g Educational ————————— I Sample ———————— Musical S 8 Films Selections .... d In I» A j a ‘0 | I e. I I I I | I I \l’ < V l9 2 Q = 19 Pro—test Selected 3 g Selected ----------- Selection —————————— Music as. Films Examples 2 ‘ I I I I l I I I I ' I l 02 between --— —-- D2 between I a: V subiects subiects V F7 <' 0 g , Scale > < Scale ,2 .3 Film _9 means __> . ‘ <___ __ means <_ 3 2 Examples A 53:30:30" T Examples g File; Prediction Music 2 D 02 |' P.O.R. 02 I l W L POPULATION C 71 I l I I I I I I I I I I Scale Means A I I l I I I I I I I V and 9 film segments used in two ditterent combinations) II Film with Music Combinations (each of the 9 music l6 seemed to be the most useful. This continued until the 19 most "useful" were selected. These 19 film sequences were then presented to a pilot population of 11 doctoral students in Educational Media, hereafter referred to as Population A, for judgments on a series of Semantic Dif- ferential scales. By calculating the co-orientation for each example as expressed in the judgments of these "experts," the relative structure of the examples was determined. The highest and lowest structured examples 3 were then used for further experimentation. Music Examples Music examples that would reflect a range of type and structure were drawn from the more than 300 musical themes and selections contained in the Chappell” and the Corelli Jacobs5 film music libraries. First, about 60 examples were selected for variety and assumed differences in structure. These were then dubbed from the masters and the selection process continued until 19 examples were finally chosen. Close examination of the musical phrases within each example showed that about 35 seconds was ideal for presenting a complete musical theme. Each example 3A listing of the selected films is given in Appendix A. “Chappell Music Library, 117 West A6 Street, New York, N.Y. 5Corelli Jacobs Film Music Inc., 25 West A5th Street, New York, N.Y. The Corelli and Chappell collections were made available through the courtesy of Professor Edward McCoy, Head, Motion Picture Production, Michigan State University. 17 was then edited so that a musically logical beginning and end fell within the 35 seconds, that is, each example neither started nor stopped in the middle of a phrase. The 19 selected examples, 35 seconds each in length, were then presented to Population A who judged them with the same Semantic Differential scales used for the film examples. Measures of co-orientation were used to determine the relative structure of the musical examples. The exam- ples with the highest and lowest relative structure were 6 selected for further experimentation. Questionnaire Construction and Administration Since the goal was to determine the degree to which an audience shares in common a given set of judgments toward a series of film and music examples, a "structured response" approach was used. A questionnaire utilizing Osgood's Semantic Differential technique7 was developed. The Semantic Differential is a combination of associa- tional and scaling procedures and is regarded, by those who developed it, as a logic or technique rather than a test in any final form.8 6A listing of the selected musical themes is given in Appendix A. 7C. E. Osgood, G. J. Suci and P. H. Tannenbaum, The Measurement of Meaning, (Urbana: University of Illinois Press, 1957), Chapter 3. 8Roger Ellis Nebergall, "An Experimental Study of Rhetorical Clarity" (Ph.D. dissertation, University of Illinois, 1956) p. 39. 18 Earlier studies used the Semantic Differential tech- nique primarily to assess meaning and attitude toward single words and concepts. Nebergall9 and others have since used the technique for the evaluation and comparison of judgments toward entire messages. More recently, a considerable number of television versus face-to-face attitude studies have employed the Semantic Differential as a measuring instrument.10 The least popular application of this instrument seems to be in the area of non-verbal and cross modal studies, but Osgood has pointed out that the Semantic Differential is quite useful in the study of non-verbal as well as complex sign situations.ll Some representative studies however, which used the Semantic Differential with non-verbal stimuli are: Tucker's aes- thetic judgments toward paintings study;l2 Tannenbaum and Osgood's studies of color meaning;l3 Kerrick's photo caption study;lu Tannenbaum's musical background 9Ibid,, p, 45, 10J. C. Reid and D. W. MacLennan, eds., Research in Instructional Television and Film, (Washington: U.S. Government Printing Office, 1967). llOsgood, p. 2A. 12William T. Tucker, "Experiments in Aesthetic Com- munications," (Ph.D. dissertation, University of Illinois, 1955). 13Percy Tannenbaum and Charles E. Osgood, 1952, re- ported in Osgood, The Measurement of Meaning, p. 299. l“Jean S. Kerrick, "The Influencecof Captions on Picture Interpretation," Journalism Quarterly, 32:177-82, (Spring 1955). 19 study;15 Mehling's photo study;l6 Greenberg's TV production 18 and, Pallett's study 19 study;17 Harrison's pictic analysis; of the connotative dimensions of music meaning. Although no known studies have utilized the Semantic Differential to investigate film music as a film variable, the variety of non—verbal and complex sign studies found in the literature suggested that this approach would be useful. An 18 scale form of the Semantic Differential was used as the measuring instrument. The questionnaire was made up into a 22-page mimeographed booklet consisting of a cover page explaining the use of the rating scales, a practice page, and 20 sequentially numbered response pages.2O 15Percy H. Tannenbaum, "The Effect of Background Music on Interpretation of Stage and Television Drama," Audio- Visual Communication Review, A:92—103, (Spring 1956). l6Reuben Mehling, "Attitude Changing Effect of News and Photo Combinations," Journalism Quarterly, 3Azl69- 198. (Spring 1959). 17Bradley S. Greenberg, "The Effects of Communication Incompatibility on Children's Judgments of Television Programs," Journal of Broadcasting, 8:157-71, (Spring 196A). 18Randall Paul Harrison, "Pictic Analysis: Toward a Vocabulary and Syntax for the Pictorial Code; With Research on Facial Communication," (Ph.D. dissertation, Michigan State University, 196A). 19Earl Marshall Pallett, "Music Communication Research: The Connotative Dimensions of Music Meaning," (Ph.D. dis- sertation, Michigan State University, 1967). 20A copy of the instructions and a copy of the Semantic Differential scales used are reproduced in Appendix B. 20 The response pages were identical except for the example number printed at the top of each page. The instructions printed on the cover were read to the group. The idea that there were no right or wrong answers was stressed. Attention was also directed to the fact that the zero or central space was to be used to indicate indecision or that the respondent considered the adjective pair to be irrelevant for that example. A seven-step bi-polar scale was used. 3 : 2 : l : O : l : 2 : 3 GOOD : : : : : : BAD Assigning a zero to the mid-point implies that though structured, the responses are not equivalent to a "forced choice." The scales were randomly ordered with the polar- ity positions systematically switched. Eighteen adjective scales were picked from those developed in the studies of Harrison,21 Kerrick,22 Nebergall,23 and Pallett.2u A selection was made of those scales which seemed most likely to fit musical as well as film concepts. This first selection of scales was tested with Population A. A four rotation factor analysis of these scales showed the set to be more favorable to the 21Harrison, p. 126. 22Jean S. Kerrick, "News Pictures, Captions and the Point of Resolution," Journalism Quarterly, 3A:183-89 (Spring 1959). 23Nebergall, p. A0. 2uPallett, p. 136. 21 film since 11% greater variance was accounted for in the film analysis than in the music analysis.25 In addition, since five scales loaded on different factors for the film than for the music examples, it was decided to change certain scales. A four rotation factor analysis of the new set of scales showed that only A% greater variance was accounted for in the film analysis, and that only three scales loaded on different factors. Further refine- ments would have been attempted had time permitted; how- ever, with a difference of only A% in the total account— able variance between the film analysis and the music analysis, it was felt that the set of scales was adequate for the needs of the study. Experimental Populations Three populations were involved in the entire study. First, there was a pilot study involving eleven graduate students in Educational Media, Population A, as previous- ly mentioned. Population A was used to evaluate the Semantic Differential scales and to select the film and music examples that would be most useful for the experi- ment. The scales were evaluated by a four rotation factor analysis. The film and music examples were selected so that the highest and lowest structure, as reflected by measures of subject co—orientation, would be included in the experiment. 25A four rotation factor analysis of the scales used with Population A is given in Appendix B. 22 At a later date, these selected examples and scales were presented to Population B. Population B was used to establish the independent measures for the separate film and music examples. Predictions of Population C responses to combined film with music examples were based also on measures derived from Population B. Populations B and C were randomly matched subsets of an undergraduate teacher education class of 1A0 stu- dents. Randomization was accomplished by having each student draw from a shuffled deck of marked cards as he entered the class. The cards were marked 1 or 2. All those drawing "1" were assigned to Population B, and "2's" were assigned to Population C. By tallying the sex of the students as the cards were drawn, it was established that half the males and half the females were represented in each population. However, a few students arrived late and were assigned to POpulation C causing a slight, but not significant, inequality in numbers between the two populations. The students assigned to Population B were sent to a special room where the projectors and sound equipment had previously been set up, and test booklets had been distri- buted to each desk position. The subjects were asked to check their books for missing pages, to fill in their name, and to mark "Group 1" in the appropriate place. The in- structions given on the cover page were read aloud to them. The practice music example was played and the subjects 23 were asked to fill out the practice page. No questions having arisen, the subjects were told to turn the page when they had completed each example so the examiner could determine when all were ready for the next example. The number of each example was announced immediately before it was presented so that the subjects could make sure they were filling out the proper page. The sequence consisted of: 1 music practice example and 9 music test examples, then 1 film practice example followed by 9 film test examples, and finally, a repeat of the first film example as a test for practice effect. The entire pro- cedure with Population B took about A5 minutes. The test books were collected and the subjects were asked to return to their classroom. Students that had drawn cards marked with a "2", plus the few late students, were instructed to go to the testing room as the "1" group returned. During the time that it took for the two groups to exchange places, new test booklets were placed on the desks in the testing room and the projectors and sound equipment were made ready for the new sequence. Group 2 (Population C) was asked to fill in the same information on the test booklets and was read the same instructions. Population C received only film and music combinations. After a practice combination, a series of 18 film and music combinations were preSented. The room lights were turned on after each film combination so that the test booklets could be marked. A new example was presented as 2A soon as all pages were turned. The test with the second group also took about A5 minutes. Equipment The equipment used in the experiment consisted of two Graflex projectors and a Wollensak tape player fed through the testing room's permanent P.A. system. Back-up equip- ment was available but not needed. White leader was used to separate the film and music examples. Synchronization between film and music was accomplished by the timing marks on the leader for each music example and cue marks on the leader of each film example. Two film projectors were used so that film examples could be repeated and re- ordered without delay to the audience. Three persons were needed to run the presentation. One person ran the equipment, the second rewound and ordered the film and tape, and the third person called the new sequence number when all subjects had completed the previous example and turned the room lights on and off as required. Presentation Order The order of presentation for the separate music and film examples was established by random draw. Each music and film example was used twice to make up the eighteen combination film with music examples with Population C. The first order of presentation for the combined film with music examples was developed by using the random sequence established for the music examples and assigning 25 the film examples on a "best match" basis in terms of con- gruency between scales. For the second set, the film order was repeated and music examples switched to produce a "mis-match." In this way, each film example received two music treatments, one treatment matching better than the other. The "mis-matched" combinations were necessary to insure a sample of incongruous examples. Methodology The focus of this study can be summarized in the fol- lowing general hypothesis. General Hypothesis. The effectiveness of music when used with film is dependent on at least two condi- tions: 1. the congruency between the channels, 2. the relative structure of the two message components. The strategy for the investigation of music as a film variable was to view the problem sequentially from three separate levels. The first level simply asked if the perceived judgments for a film are influenced by the music that co-exists with the film scene. The second level asked if the Judgments for the combined film with music sequence can be predicted from measures of the component parts. The third level asked if the structure of and the con- gruency between the component parts systematically affects the relative influence of the components. 26 Level One. Can different music cause a change in the perceived meaning for a given film example? A series of film examples and music examples, selected to include a variety of styles and types as judged by Population A were presented to Population B. The examples, each 35 seconds in length, were measured on a form of the Semantic Differential.26 The mean of each scale for every music example and every film example was calculated. Each film was then matched with two music examples, one being a better match than the other. The film with music combinations were then presented to Population C which was randomly matched with Population B. These com- binations were judged on the same set of Semantic Differ- ential scales and again the means were calculated for every scale for each combination. Since each film segment ex- isted with two music treatments, it was possible to sub- tract, on every scale, the distance or D between scale positions. If the music did not systematically influence the combined judgments, then it would be expected that the D would be randomly distributed between the two music treatments. A statistical test for matched pairs was used to establish significance. Thus, level one dealt with: Hypothesis 1. Given a specific audience and a series of musical accompaniments to a film scene, the music can significantly change the interpretation the audience gives for the com- bined presentation as measured on a form of the Semantic Differential. 26Osgood, p. 19. 27 Level Two. Can the judgments for a combined film with music presentation be predicted from knowledge or measures of the component parts? A series of musical examples and film examples were judged independently on a series of Semantic Differentiil scales by Population B. Combinations of these films and music examples were then given to randomly matched Popula- tion C. Predictions of Judgments for the combinations 27 were made using the congruency point of resolution model. In scale units of the Semantic Differential, the point of __F__ M _ F + M(F) + F_I_M(M)’ where F - the mean resolution equals judgment of the film on a given scale and M = the mean judgment of the music on the same scale. A correlation between the scores predicted from measures of the component parts and scores observed for the combined presentation was used to test: Hypothesis 2. Given a set of film and music examples, the observed scores for the combina- tions will significantly correlate with scores predicted by the point of resolution formula. Level Three. Is the relative influence of the film and music components systematically influ- enced by conditions of congruency and structure? Congruency.--Separate measures were available for the film and the music components. The relationship of these measures on a given scale determines the congruency be- tween the components.28 If the measures for both 27Ibid, p. 207. 28Ibid, Chapter 5. (:1 b ,4?“ ‘I (I?) 28 components fall on the same side of the scale, that is, if the music is judged, for example, to be happy and the film is also judged to be happy, then the combination is said to be congruous. If, on the other hand, the film is judged happy and the music sad, then the combination is said to be incongruous. The relative congruency of a combination was determined by the number of incongruent scales found for the combination. Using levels of con- gruency as the independent variable, the accuracy of the point of resolution prediction and also the influence of the components were analyzed as dependent variables. One- way analysis of variance and related statistical tests were used to test: Hypothesis 3. Given measures of film meaning and music meaning, the congruency between com- ponents will systematically affect the accuracy of the point of resolution predictions. Hypothesis A. Given measures of film meaning and music meaning, the congruency between com- ponents will systematically affect the relative influence of the component parts. Structure.--A measure of co-orientation was established for each example by calculating the absolute difference be— tween all possible pairs in the sample and then summing the squares of their differences (D2) across all scales of the Semantic Differential. The number of possible pairs in a (N(N-1)) 2 . The number of D2 scores, using a Semantic Differential sample of 70 people, for example, is 2,A15 pairs with 18 scales, was thus 18 X 2,A15 or A3,A70. Given a range of l - 7 on each scale, there was a maximum absolute 29 difference range of 6, and a D2 range of O - 36 between any given pair on any given scale. So the theoretical range of averaged D2 scores with 18 scales was 0 - 6A8 for each example (36 x 18). The film and music examples were then rank ordered according to their measure of co- orientation. High co-orientation reflected a high consen~ sus within the sample audience and low co-orientation reflected low consensus within the sample audience. When referring to the examples rather than to the audience, it is more convenient to speak of high and low structure. Using levels of structure as the independent variable, the accuracy of the point of resolution prediction and also the influence of the components were analyzed as the de- pendent variables. One—way analysis of variance and re- lated statistical tests were used to test: Hypothesis 5. Given measures of film structure and music structure, the structure of the com- ponents will systematically affect the accuracy of the point of resolution predictions. Hypothesis 6. Given measures of film structure and music structure, the structure of these com- ponents will systematically affect the relative influence of the film and music. CHAPTER IV RESULTS There are three levels of data to be reported in this study. The first level asks if music can cause a change in the perceived meaning for a given film example. The second level asks if the judgments for a combined presenta- tion can be predicted from knowledge of the component parts. And the third level asks if congruency and structure sys- tematically influence the relative influence of the com- ponent parts. For the sake of simplicity, the results will be presented by units of function rather than levels of discourse. In addition, the connotative dimensions of film-music meaning will be reported as found in this study. The data gathered in this study represents a total of more than 55,000 individual responses -- the main study of 1A0 subjects responding on 18 scales to each of 18 items, a retest of 70 subjects on 18 scales to three items, and a pilot study of 11 subjects responding on 18 scales to each of 36 items. These individual responses were analyzed by three separate procedures. First, the responses for each item were averaged across the group for each scale in the set. This established the semantic position of each scale for every item in the film set, the music set, and 30 31 the combined music with film set. Second, each person was paired with every other person in the group and the difference in their responses on each scale to every item was tabulated. This established the relative structure of each item in the set, great differences between pairs being defined as low structure and small differences be- tween pairs being defined as high structure. Third, the responses for each scale were analyzed across all items in the set and factored against all other scales giving the dimensional structures of connotation for the film set, the music set and the combined film with music set. Summarization of the above three classes of informa- tion allows (l) investigation of the similarities between dimensional structure of picture connotation and music connotation in a film presentation, (2) investigation of the prediction of the combinations given knowledge of the elements, and (3) investigation of component conditions that inhibit or contribute to the combined film with music judgments. Prerequisite Statistical Tests Since neither a counterbalance such as a Latin-square nor a control group was used in the major design of this study, some assurance must be given that learning and order effect are not significantly evident. A test—retest situation was built into the overall design to provide for the prerequisite tests. For each mode, the first example after practice was repeated as the final example. A 32 significant change between the two measures of the same example would be an indication Of either learning or order effect. Two separate tests were performed on the retest example in each mode, a correlation of scale means, and a T-test for paired samples. The correlation of scale means between test and retest was .93 for music, .93 for film and .9A for the film with music combination. All correlations are highly significant. The T-test between means paired on each scale was below the .1 level of sig- nificance in all cases, therefore the null hypothesis of no significant difference could not be rejected. Thus it appears that it will not be improper to study the overall results, since neither learning nor order effects seem to have introduced an experimental bias. Component Influence The first task is to establish if music can change the connotative meanings judged for a given film segment. Each of the nine film sequences were presented with two different musical accompaniments. This allows asking if the judgments made for a film when viewed with music A are significantly different than judgments made of the same film when viewed with music B. First, the measures for the film segments and the music examples must be established. The mean for each of the 18 scales across 68 subjects for the 9 film segments and the 9 music examples was calculated, 32A scale means 33 in all.1 Using these scale means, it was then possible to calculate two sets of difference or D scores, each set containing 18 scores, one for each scale. One set would be between the film alone and film with music A, the other set would be between the film alone and film with music B. Squaring these D scores allows adding across scales for each example or across the entire set of examples. If the music did not systematically influence the combined Judgments, then it would be expected that the D2 would be randomly distributed between the two music sets. The null hypothesis to be tested is that there is no signifi- cant difference between the two sets of scores. The Wilcoxon test of Matched Pairs2 was chosen as an apprOpriate statistical test in that the two sets of D2 scores could be paired on each scale. This test is avail- able on the Michigan State University 3600 computer3 with Wilcoxon T approximations for a sample with N larger than 25. Essentially, it is a non—parametric test to determine the probability that two samples, in which the members of one sample are matched with members of the second sample, come from identical populations. lMusic scale means and film scale means are given in Appendix C. 2Sidney Siegel, Nonparametric Statistics for the Behavioral Science, (New York: McGraw-Hill, 1956) pp. 75-83. 3John Morris, "Technical Report A5," Computer Insti- tute for Social Science Technical Reports, (East Lansing: Michigan State University, 1967). 3A The test was taken first across the total set of 162 paired D2 scores (9 x 18 scales). A Wilcoxon T-test value of A018.00 was found and therefore the null hypothesis of no difference was rejected at greater than the .005 level of confidence, two-tailed. Next, tests were calculated between the two music treatments for each of the nine film examples. The null hypotheses of no difference was re- jected at the .005 level of confidence, two-tailed, for six of the nine examples. Therefore, it seems safe to conclude that the music associated with a film scene can cause a difference in the way the film with music combination will be judged. However, it is noted that this difference is not the same for all examples. The task, then, is to determine if the combined judgments can be predicted and to identify the conditions that may inhibit or contribute to the combined music with film judgment. Judgment Predictions It was suggested in Chapter II that Osgood's point of resolution formula might be useful in predicting judg- ments of film with music combinations, given measures of the independent film and music components. The measures for the film segments and the music examples are avail- able and given in Appendix C. They are the means for each of the 18 scales for the 9 music examples and the means of the 18 scales for the 9 film segments, 32A means in all. Utilizing the formula ——E——(F) + F_¥_M F + M (M), where 35 F = the film segment scale mean and M = the music segment scale mean, 32A predictor scores were calculated, 18 scales for each of the 18 film with music combinations. The observed means for each of the 18 scales taken across the 71 subjects for each of 18 film with music combina- tions are also available. The predicted scale means and the Observed scale means are also given in Appendix C. The 32A predicted combination means were then cor- related with the 32A observed combination example means. The correlation between scores predicted from Population B and scores observed in Population C was .87, which is significant at greater than the .005 level. Since the overall correlation was rather high, it seems useful to look at each combination to see if the point of resolution prediction was more successful for some examples than others. The 18 predicted scores were correlated with the 18 observed scale scores for each of the 18 examples. The correlations range from a low of .13 to a high of .95, as shown in Table 1. It is also interesting to look at the scales to see if the predictions are more accurate for some scales than for others by running correlations for each scale across the 18 combinations. These correlations are given in Table 2. Since the point of resolution formula is useful in predicting the judgments that will be made for a com- bined film with music presentation, it seems safe to con- clude that both the film and music components contribute 36 to whatever meanings the combination will be judged to have. However, it cannot be implied, at this point, that the influence is equal. It should also be noted that the predictions are more accurate for some combinations than for others. The next step, then, was to determine the relative influence of the component parts and to identify some factors that might help explain why the predictions are more accurate in some cases, than in others. Table l.-—Predicted with observed scores correlated by examples. Predicted—Observed Predicted-Observed Combination Correlation Combination Correlation l .95 10 .72 2 .91 ll .79 3 .90 12 .13 A .89 13 .95 5 .91 1A .80 6 .75 15 .98 7 .97 16 .68 8 .96 17 .9A 9 .95 18 .86 .68 = P< .002 Table 2.--Predicted with observed scores correlated by scales. Scale Correlation Scale Correlation 1 Kind Cruel .96 10 Danger Safety .91 2 Heavy Light .90 11 Slow Fast .87 3 Sad Happy .89 12 Pleasant Unpleasant .95 A Complex Simple .76 13 Weak Strong .79 5 Good Bad .9A 1A Meaningful Meaningless .AA 6 Passive Active .9A 15 Violent Peaceful .97 7 Ugly Beautiful .9A 16 Dishonest Honest .63 8 Excited Calm .93 17 Serious Humorous .86 9 Worthless Valuable .61 18 Harmonious Dissonant .78 .61 = P< .006 37 Relative Component Influence It has been demonstrated that the meanings judged for a given film sequence change depending on the music used with it, and that this change is, in general, sys- tematic. Therefore, it seems appropriate to ask which of the components contribute more to the combined judgments, the film or the music. In order to answer this question, it is necessary to determine if the judgments for the com- bined presentation are closer to one component than they are to the other. That is, are the judgments for the com- bination more like the judgments made for the film segment when viewed by itself or more like the judgments made for the music example when heard by itself. Given a set of judgments on a series of identical scales for the music example, for the film sequence, and also for the film with music combination, the difference between the com- bination and the music alone can be compared with the dif- ference between the combination and the film alone, scale by scale. By squaring the differences and then summing across the scales, it is possible to determine if they total difference for the film set is greater or less than the music set for any combination or for all combinations. Using the responses for the 68 subjects to separate film sequence and music examples and the responses of the 71 subjects to the film with music combinations, the differ- ence for each component and the combination was calculated and squared. This gave a total of 32A film difference 38 scores and 32A music difference scores. Pairing these scores scale for scale, example by example, a T—test of correlated means was performed. A Fisher-Student T-test value of 2.87 was found and the null hypothesis of no dif- ference between the film and music was rejected with a two—tailed probability significance of greater than .01. The greater differences were found in the music set; that F is, the music had less overall influence on the combina- tion than did the film. In that there is a significant difference between the film and music effects, it seems as shown in useful to look at the D2 summed by examples, Table 3. Table 3.—-Sum of D2 by types and combination. Combination Influence Music Film 1 F 23.3A A.l3 2 F 18.58 2.16 3 F 17.89 3.11 A F 1A.3A 5.2A 5 F 21.71 15.76 6 M 30.93 A2.67 7 M 6.37 7.99 8 M 5.58 7.20 9 F 13.21 5.AA 10 F 57.39 1A.A3 11 F 37.3A 9.93 12 M. 26.26 35.66 13 F 9.A9 A.38 1A M 16.0A 19.70 15 F 7.62 1.86 16 M 19.10 32.A9 17 M 5.59 7.58 18 M 9.78 20.A9 Total F 3A0.56 2A0.22 T between Music and Film = 2.87, P< .01 39 It is interesting to note that although the film was more effective overall, the music was more influential in 8 of the 18 combinations. It seems useful, therefore, to con- tinue further analysis in an attempt to identify system- atic factors contributing to this effect. The two factors investigated are congruency and structure. Congruency {H When any film segment is combined with any music ex— I ample, two component measures are available on each scale. 4 These measures represent the semantic judgment means of the 68 subjects to the separate components. If these *- measures both fall left or right Of the scale center, then the two components are said to be congruent with each other on that scale. If, however, one component measure is found on one side of scale center and the other on the Opposite side, then the combination is said to be incongru— ous. By counting the number of incongruous scales found in each combination, it is possible to say that some com- binations are more incongruous than others. Taking the six combinations with the least number of incongruouS' scales gives a group of high congruency combinations, the next six gives the medium congruency combinations, and the six with the greatest number of incongruent scales makes up the low congruency group. A0 Congruency and Point Of Resolution Predictions It has already been observed that the point of reso- lution prediction correlated well with the actual observa- tion. It was noted, however, that within this high overall correlation, the correlations of individual examples ranged from .13 to .96. In order to examine this in greater de- tail, D scores between the point of resolution prediction and the observed value were calculated for 18 scales in each combination. The less accurate the point of resolu- tion prediction, the greater the D score between the predicted and observed values. A one-way analysis of variance (see Appendix D) performed on these 32A D2 scores for the three levels of congruency gave a between cell significance of greater than .01. A T-test performed between the High Congruency and Low Congruency cells gave a T value of 2.AA and allowed rejection of the null hy- pothesis of no difference at a two-tailed significance of .02. No significance was found between medium congru- ency and high or between medium congruency and low. The point of resolution D2 scores are given in the Table A summed across scales for each example. The smaller D2 found for high congruency suggests that the point of .resolution predictions are more accurate for combinations that are congruent and becomes less accurate as the number cif incongruent scales is increased. Irr-v' —— Al Hm mm.omH mm.mmH m>.Hm ml mmqul mmqmmu .mmqu :H 2 NH m m:.sH mm.sm HH.AH AA A OH m se.m: mm.om em.HH HH 2 e OH oe.mH so.eH mH.oH OH 2 AH m me.mm OH.mH ms.OH m z eH A ::.m Hm.mH se.H m e m Hozmamozoo 30A ms He.oe AN.HOA sm.mm ml es.mH He.Hm me.: A m m OH mm.m sm.em em.mH m a HH m mm.~ sm.m mm.o m 2 w e m:.om ms.m mm.m : z mH OH eH.m mm.wH mm.e A e m CH mm.z 0:.m m:.m m m mH Hozmpmezoo EpHamz no mH.mm mm.zm m».mm -m. em. me.e Hm.m m m mH mH sm.m sm.HA no.5 m a A a HA.m mw.AH me.m m .e m AH mm.~ mm.m mm.m H 2 AA AA mH.z :m.mm ~>.m H m H m om.m mm.m mo.m o z m wozmamezoo meHm pommmm Q EHHm Q OHmsz Q .m.o.m mOHmom escapwcoocH mucosamcH cofipmcHoEoo cofimeESm m m m mo Looezz hepmopu pommmo COAmeESm new a eHHu . a eHnss . m m m a :oHpsHommn mo choa i mocozpwcoo mo mHm>mH me No esmiu.z meme A2 Congruency and Component Effect Just as the predicted score can be subtracted from the Observed score, it is possible to subtract the mean judgment for the music component from the mean judgment of the combination on a given scale. This gives 32A difference scores for the 18 music segments that can be summed and compared. A set of 32A difference scores was 7 also computed for the film components. These scores, summed across scales for each combination, are given in Table A. The performance of analysis of variance and re- lated T-tests for separate film and music D2 score shows net) " 1 all but one level of congruency significantly different at greater than the .005 level for both film and music (see Table 27, Appendix D). This suggests that as the number of incongruent scales is increased, the distance between each component and the combination increases. Summing the D2 scores by levels of congruency, it is noticed that the film scores are much smaller than the music scores for both the high and medium congruency levels. This is consistent with the overall trend of film component having greater influence on the combination. However, the film'D2 and music D2 scores are about equal for the low congruency level suggesting that film does not have the same primary effect for combinations of low con- gruency. The greater influence has been assigned to each combination by determining which component had the smaller D2 eumi is listed in Table A. There are 10 combinations A3 in which the film component has a smaller D2 and 8 combina- tions in which the music component has a smaller D2. Look- ing at this by levels of congruency, it is noticed that a disproportionate number of music influence combinations fall in the low congruency level. At the high and medium con— gruency levels there are A combinations in which the film predominated and 2 for the music. At the low congruency level, however, the reverse is found. There are A combina- tions in which the music predominates and 2 for the film. This would seem to indicate that the music has a greater opportunity of influencing the film-music combination when the congruency between the components is low. It should be cautioned, however, that this should not be considered a conclusive finding because there are insufficient samples to establish significance. Structure For the purpose of this study, an example is said to have high structure when a high co-orientation is re- flected in the judgments of the sample population. Low structure, then is the label given to an example that is relatively ambiguous to this population. The measure of co-orientation is derived by taking all possible pairs in the sample population and summing the difference between each pair of responses on every scale for each example. About A5,000 D scores are thus generated for each example or combination. Each score can range from 0 for no dif- ference between the.pair, to 6 scale points difference. AA Summing across all A5,000 scores, a large D score re- flects a large difference in judgments of the sample population, a small D reflects relative consensus. The example with the largest D score is said to have the lowest relative structure; the example with the smallest D score is said to have the highest relative structure. The film and music examples ranked according to their relative structure are shown in Table 5. Table 5.-—Rank order of relative structure. MUSIC FILM Example D2 Average Example D2 Average low low 6 72.8A 5 95.3A A 66.50 1 77.25 9 6A.66 2 68.07 7 63.17 8 67.28 1 59.A3 3 65.29 2 58.33 9 6A.58 3 57.81 A 63.70 8 A7.87 7 55.65 5 A7.6A 6 53.78 high high It is interesting to note that, over all, greater ambigu- itqrxwas expressed for the film samples than was reflected by tflie music samples. Struncture and Point of Resolution Predictions The ranked examples can be grouped into three levels of‘satructure: high, medium and low. Using levels of strwuzture derived from subject paired response differences as tflie independent variable and P.O.R. D2 as the dependent A5 variable, it is possible to compare the performance of the point of resolution prediction by component levels of structure. The sum of P.O.R. D2 by component structure level is given in Table 6. Table 6.--D2 between point of resolution prediction and observed scale mean summed by levels of structure. Music Film Eu“ High Structure A5.8l 3A.55 I Medium Structure 32.58 33.19 Low Structure A6.A8 57.13 12u.87 12u.87 WWW It will be noticed that the point of resolution prediction is most accurate for medium structure and least accurate for low structure. Separate one-way analysis of variance (see Appendix D) shows between cell significance (.007) for the film component but not for music. Structure and Component Effect It is also possible to look at the separate film and rmisic D2 scores as was done in the analysis of congruency. This is the distance of each component from the combina- titni in which it appears. It will be remembered that, cyverall, the film component had a greater influence than tflue music. Therefore, the sum of the film D's will be linss than the sum for the music D's. Table 7, below, giines the distance between the component and its combina- ticni for three levels of structure. Separate one-way A6 analysis Of variance failed to establish between cell significance for either film or music. However, using only two levels of structure, high and low, significance of .05 was established for music and .005 for film (see Appendix D). It will be noticed in Table 7 that large D's are found with high structure and small D's with low structure. No satisfactory explanation can be given for this. In fact, the reverse was expected. One would ex- pect an ambiguous component to have less influence than a highly structured one. But apparently the stronger component is used to develop the meaning of the weaker com- ponent and thus moves a greater distance than does the more ambiguous component. Table 7.--D2 between components and combined examples by levels of structure. Music Film High Structure 157.31 9A.63 Medium Structure 89.63 79.A8 Low Structure 93.62 66.11 340.56 2A0.22 Cknnponent Influence by Levels (of Congruency and Structure In order to make a comparison between the film com- ;xndent and the music component for a given combination, 1}; is necessary to have separate measures for each com- ;xnnent. The D scores between each component and its A7 combination have already been calculated for every scale in each example. If the influence of the components is equal, then the D scores will be equal. If, however, the combination is more like one component than the other, the D score of the component having greater influence will be smaller. By summing across combinations, it was earlier determined that the film component had a greater overall influence. In that there have been indications that congruency and structure systematically affect the combination, it seems useful to look at the component influence by combined levels of congruency and structure. If the film D2 score is subtracted from the music D2 score, the difference represents the net amount that the combination favors the component with the smaller D2. Averaging this across all combinations found within a given cell allows comparison of net effect between cells. This is expressed by the following formula: 2M zc2 2F >302 2: Zs(N - N) — zs(N - N) k k Where: M = Music Component Responses F = Film Component Response 0 = Combined Film with Music Responses S = Semantic Differential Scales (18) k = Combination N = Subjects (68 in Population B, 71 in C) For the sake of simplicity, two levels of congruency and two levels of structure are used to determine the cells. Less than five incongruent scales is considered a con- grmMMJs example and five or more incongruous scales an I“ ‘ , “Y A8 incongruous example. High and low structure is determined by the rank order of the components. There are four com- binations of structure; high film with high music, high film with low music, low music with high film, and low music with low film. Table 8.——Net difference between music influence and film influence by combined levels of structure and congruency. Structure Congruency Congruent Incongruent High Music, High Film (HMHF) 5.20 5.76 Low Music, High Film (LMHF) 7.10 -9.02 High Music, Low Film (HMLF) 15.60 25.AA Low Music, Low Film (LMLF) A.25 2.05 In Table 8, 0.00 would indicate equal film and music influence. The positive numbers indicate film influence, the negative number indicates music influence. The larger the number, the greater the influence. Thus it is seen that nmsic has the greatest influence in an incongruent combination with low music structure and high film struc- ture. Film.has the greatest influence in an incongruent cxmnbination with high music structure and low film structure. It will also be noted that very little difference is fourui between HMHF and LMLF structure levels. The film has 2a slightly greater influence than the music for both ccuugruent and incongruent combinations. However, a V ,0... , o-r" .— \ an!“ 1‘ r" .31.” ».-' -gw tar" H... rein" ..- 6A ‘4‘! ...... :.i- é new; .‘ ”I - «...L -.A. <.. :r. 3 I. win. Sea: ’hlrt .4.,,, Of 9;; 7,. “r l I A9 considerable difference is noticed for the mixed structure combinations LMHF and HMLF. Taking only these four cells, a chi-square was used to test the null hypothesis of pro- portionate expectations. A chi-square of 13.3A gave a two—tailed significance of greater than .01. The only condition in which the music component had a greater in- fluence than the film component was in an incongruent combination with low music and high film structure. On the other hand, the greatest film component influence was found in an incongruent combination with high music and low film structure. These results support the notion that a component with strong meaning (high co-orientation) will tend to strengthen the meaning of the ambiguous (low co- orientation) component. It is somewhat surprising, however to find that this tendency is even stronger for incon- gruous combinations. Summation Effect The point of resolution formula predicts that the conflxined value will fall between the two component values. Ckncasionally, however, the observed value will be greater thaxi either of the components. The phenomenon of the ob- serwned combined value falling beyond either component ratruer than between them is called the summation effect. Out (of the 32A scores predicted by the point of resolution “Jean S. Kerrick, "News Pictures, Captions and the POIJTt of Resolution," Journalism Quarterly, 3A:183-89 (Spring 1959) g- 1.. 3»... It»! 50 formula, 139 exceeded one of the values. Of these 139 summations, 93 fell toward scale center and A6 away from the scale center. There were 91 summations found in con- gruous combinations and only A8 summations in incongruous combinations. In terms of structure, 68 summations were found in high structure and 71 found in low structure combinations. The summations favored the film component 75 times and favored the music component 6A times. In general, the summations increased as the number of incon- gruent scales increased and favored the music in low struc— tured combinations and the film in high structured com- binations. The trends, however, were below the .05 level of significance. Table 9 shows summation effects averaged by structure and congruency combinations. Table 9.--Summation effect averaged by combined levels of structure and congruency. Structure Congruency Congruent Incongruent High Music High Film 11 3 Imnv Music High Film 11 3 High Music Low Film 9 7 Low Music Low Film 9 7 Scales There are three considerations that need to be in- veenxigated in terms of scale characteristics and scale 1.5: a I nfl’r nu 51 performance. These considerations are scale relevancy, direction consensus, and inter-scale relationships. Scale Relevancy At the time of the experiment, the subjects were instructed to check the central scale position if they felt an adjective pair was irrelevant. At a normal level, all responses can be placed into two groups (middle step/ non-middle step) for any given scale. The hypothesis was that a significantly greater number of peOple would judge a given scale to be relevant than would Judge that scale to be irrelevant. A binomial (two category) exact test of statistical significance is most appropriate to test this hypothesis. The .05 level of significance (one- tailed) was selected and a total of 6A8 binomial tests were computed5 -- 162 for the music examples, 162 for the film examples, and 32A for the film with music combina- tions. The results are given in Tables 10, 11, and 12 vhqere an irrelevant scale is indicated with an "X". A greater proportion of non-relevant combinations were found .for'the film than for music: 16/162 for film to 6/162 for rmisic. Since scale 16, Dishonest - Honest, was most fre- cyiently irrelevant for music, film and combined mode, its insefulness for this study is somewhat brought into ques- tion. 5U.S. Ordinance Corps, Table of Cumulative Binomial Prmflaabilities, (Washington: ORDP 20-1, 1952). 52 Table 10.--Scale relevancy and direction consensus for music examples. Scale Example Totals 2 3 A 5 6 7 9 X 0 1. Kind Cruel 0 0 2. Heavy Light 0 0 3. Sad Happy 0 O A. Complex Simple 0 O O 0 3 5. Good Bad 0 l 6. Passive Active 0 0 l 7. Ugly Beautiful 0 0 l 8. Excited Calm 0 0 9. Worthless Valuable X 0 3 l 10. Danger Safety 0 0 11. Slow Fast 0 0 12. Pleasant Unpleasant 0 0 13. Weak Strong 0 0 0 2 1A. Meaningful Meaningless 0 0 1 15. Violent Peaceful 0 0 l 16. Dishonest Honest X E 3 1 17. Serious Humorous 0 0 18. Harmonious Dissonant _g__l Non-relevant 0 0 0 0 l l l 6 No Direction 2 0 2 0 l 3 3 13 X = irrelevant 0 = no direction consensus E = irrelevant and non— directional 53 Table ll.--Scale relevancy and direction consensus for film examples. Scale Example Totals 1 2 3 A 5 6 7 8 9 X 0 l. Kind Cruel 0 H X 2 2 2. Heavy Light 0 0 l 3. Sad Happy 0 X l l A. Complex Simple 0 0 0 0 0 A 5. Good Bad 0 0 l 6. Passive Active 0 0 l 7. Ugly Beautiful 0 X l l 8. Excited Calm 0 0 0 2 9. Worthless Valuable 0 0 l 10. Danger Safety X l 0 11. Slow Fast 0 0 0 2 12. Pleasant Unpleasant 0 0 l3. Weak Strong 0 H H 0 H 3 5 1A. Meaningful Meaningless 0 0 15. Violent Peaceful 0 0 l6. Dishonest Honest H X X X X X X 7 l 17. Serious Humorous 0 0 18. Harmonious Dissonant 0 H _l _2 Non-relevant l 2 2 l 2 2 2 l 3 16 No Direction 3 l 3 2 8 l 2 2 2 2A X = irrelevant 0 = no direction consensus E = irrelevant and non- directional 5A HMCOHQOOLHUICOC Ucm pcm>mH®eHrHH u E mzmcmmcoo COHpomeHfiU 0C H O ucm>®HmefiHH n X Ne A m A H A m m A m m m A m m m m m m eOHeeetHe oz mm m m m m H H H m H m m m m N m m m m ece>eAetueoz 0 O o o o o o o quCOmmHD mSOHCOEeme .wH m o o o o mSOhOESm meHeHmm .NH e AA A x x x e x x e a e x x e x A x x e eneeom eneeoean .eH m o o o o Asceeeem eeeH0H> .mH A o o o o o o o o mneHweHeeez HsemeAenez .AA A e o x o e o o o x e x x o museum Ase: .mH m o o o o ecsneeHee: pesmseHm .mH H o 0 been zOHm .HH m m x x x o o x A esteem smegma .0H CA A o e A o o o o e o o o A eHnesHe> mneHeetoz .m m o o o e sHeo eeeHexm .m H A o x HseHessem AHma .A m o o o o e>Hee< esAnnem .e m o o 0 ppm eooe .m m o o o o o o eHnsAm ernsoo .A H o o annex new .m H o e HemAH A>sem .m m A x e x e Hesto ecHe .H o A AA AA eH mH AH mH NH AH OH A m A e m A m m H mHmpoe OHQmem moamom .mOHQmem OAmSEIEHHm eocheOo pom newcomcoo coHpooaHU new mocm>maop OHmomui.mH manme 55 Direction Consensus Direction consensus means that a significant number of peOple marked a scale in one direction or the other. Eliminating the central or non-relevant position, it is then possible to treat at a nominal level the responses as being right or left of center. The hypothesis was that a significantly greater number of subjects will judge the example in one direction or the other. A two—tailed binomial exact test of statistical significance is appro- priate to test this hypothesis. The results of these 6A8 binomial tests (using the .05 level of significance) are given in Tables 10, 11 and 12. For ease of coding, a 0 is used to indicate a non-significant finding for any given scale in any given example. More non-directional cases were found for the film examples than for the music. The lack of direction consensus as well as irrelevancy was proportionately higher for the combination mode than for the film or music mode. It may well be that when the two symbol systems are combined, sufficient informa- ‘tion is available to make some dimensions, such as honest - (dishonest, less useful. Factcw’Analysis In order to compare the dimensional nature of film axui music as well as the combination of film with music (n1 a given set of scales, three factor analyses were ap- plififli to the three modes of presentation. A four factor 56 solution was chosen in that only one variable was found to have its highest loading on additional rotations in each mode. Over all, the scales factored in about the same way for film, music, and the combined film with music. Only 8 out of a possible 5A differences were observed as shown in Table 13. A binomial exact probability test shows this significance at greater than the .001 level, one- tailed. Table l3.-—Scale factor comparisons. MUSIC FILM F/M COMBINED l. Kind Cruel I I I 2. Heavy Light I I I 3. Sad Happy 11* I 1* A. Complex Simple IV IV IV 5. Good Bad I I I 6. Passive Active II II II 7. Ugly Beautiful I I I 8. Excited Calm II II II 9. Worthless Valuable III III III 10. Danger Safety I I I 11. Slow Fast II II II 12. Pleasant Unpleasant I I I 13. Weak Strong IV* III III* 1A. Meaningful Meaningless III III III. 15. Violent Peaceful I I I 16. Dishonest Honest I I* IV* 17. Serious Humorous 11* I I* 18. Harmonious . Dissonant I I I *Differences in factoring between modes. An orthogonal varimax rotation was used in order to emphasize the simple structure of the data. The four :notations accounted for the greatest amount of variance in a _>.I,‘ .a Ts 57 the music mode and the least variance in the film with music mode. However, the amount of variance accounted for was very close between modes for each factor as shown in Table 1A. Table lA.-—Proportions of variance. Music Film F/M Combination Factor I .3AA3 .3539 .339A Factor II .1867 .IAOO .1330 Factor III .0739 .0939 .1179 Factor IV .0832 .0559 .0A93 .6881 .6AA3 .6396 The percent of total variance reported here is some— what higher than was found in the original Osgood studies,6 but is rather close to that found in Harrison's Pictic Analysis7 and Pallett's Music Meaning study.8 The fact that the combined film with music extractable response 6C. E. Osgood, G. J. Suci and P. H. Tannenbaum, The MeEHMIrement of Meanipg, (Urbana: University of Illinois Press, 1957), p- 37- 7Randall Paul Harrison, Pictic Analysis: Toward a Vtmxabulary and Syntax for the Pictorial Code; With Research cuiIFacial Communication, (Ph.D. dissertation, Michigan State University, 1967) p. 85. 8Earl Marshall Pallett, Music Communication Research: TTMB Connotative Dimensions of Music Meaning, (Ph.D. dis- serflnation, Michigan State University, 1967), p. 85. 58 variance is less than found for the music or film alone, reinforces Pallett's suggestion that we should expect higher explainable variance when investigating a single symbol system.9 The higher music proportion suggests that this set of scales is more relevant to the music examples, a condition that has already been noted in the discussion of scale relevance. The values for each scale are reported in Tables 15, 16, and 17. Although this study was not designed to map the dimension of film-music meaning, it may be useful in terms of future research to look at the way meaning is struc- tured, given this set of scales. The first factor is clearly what Osgood has labeled the evaluative factor: good, kind, beautiful, pleasant, etc. It seems to hold up well across all three modes. The second factor is the activity factor: fast, active, excited. It will be noticed that serious - humorous, and sad — happy factored in this category only for the music examples. This may well be due to the tendency for sad- ness and seriousness connotations to be influenced pri- marily by music tempo.lO 91bid. 10Paul R. Farnsworth, The Social Psychology of Music, (New York: Holt, Rinehart and Winston, 1958), p. 99. u '3' ' ‘7' Jed—am Table 15.--Rotated factor loadings - music examples. 10 12 l5 l6 18 Ugly Kind Danger Pleasant Violent Dishonest Good Heavy Harmonious Slow Excited Calm Serious Sad Meaningful Worthless Complex Weak Beautiful Cruel Safety Unpleasant Peaceful Honest Bad Light Dissonant Fast Calm Active Humorous Happy Meaningless Valuable Simple Strong II—I .8A28 .8303 .8302 .8239 .8231 .8009 .7A27 .651A .552A .27A1 .1A76 .036A .361A .5583 .0659 .3520 .30A2 —.2All II .1036 .16A2 .lA26 .20AA .1020 .0373 .3033 -.A080 -.0017 .8522 .8A09 .7866 .6936 .6338 II .0672 .0157 II -.1635 -.1567 III .1598 .0265 -.0333 .1919 -.0003 .079A .2197 .01A0 .2216 III .1087 .0283 .lA62 .1896 .0A30 III .7602 .6629 III .0173 .2690 IV .1235 .250A .2698 .1279 .2766 .1116 .1A95 .A591 .0707 IV .0769 .0860 .061A .355A .2913 IV .1079 .0539 .6589 .5720 Table l6.—-Rotated factor loadings — film examples. 1A 13 Pleasant Sad Good Ugly Kind Violent Danger Heavy Harmonious Dishonest Serious Excited Slow Passive Meaningful Worthless Weak Complex Unpleasant Happy Bad Beautiful Cruel Peaceful Safety Light Dissonant Honest Humorous Calm Fast Active Ieaningless Valuable Strong Simple II—I .8939 .86A3 .8619 .8299 .82A7 —.79A2 .7659 .6588 .5692 -.5629 .3677 .3A77 .2560 .0965 .0Al9 .255A .1198 .09A7 II .1599 .0118 .06A6 .1769 .208A .A25A .3251 .0A87 .2760 .0907 .2739 .7889 .7850 .7A25 II .1223 .llAA .3A88 II .193A III .0383 .1169 .08A2 .1086 .0309 .OAOI .0396 .16A9 .2380 .3068 .3A09 III .1118 IV .0503 .2106 .028A .0289 .1935 .0162 .1360 .3971 .0677 .0351 .3506 IV .0568 .0365 .2235 IV .0172 .1563 .2A31 .6906 61 Table l7.--Rotated factor loadings - film with music combinations. 3 II III IV 3 Sad Happy .90A5 —.Ol69 -.05A3 .1727 12 Pleasant Unpleasant .8731 .1853 -.0988 .060A l Kind Cruel .820A -.2033 -.0137 .l757 7 Ugly Beautiful -.810A .1972 .1539 .0063 2 Heavy Light -.7815 .09Al -.l262 .2913 5 Good Bad .7752 .0786 -.2575 .1678 15 Violent Peaceful .7603 .A319 .0291 .123A 10 Danger Safety .7558 .2923 .0502 .2178 17 Serious Humorous -.563A .2632 -.3677 .073A 18 Harmonious Dissonant .A3A3 .lAAO .3666 .088A I 9;; III IV 11 Slow Fast .07A5 .8290 .0A37 .0519 8 Excited Calm .3392 .7969 .0659 .0753 6 Passive Active .0A78 .709A .2605 .07A0 I II I_I_I Iv 1A Meaningful Meaningless .039A .1031 -.7760 .0053 9 Worthless Valuable .1A8A .1010 .7591 .00A8 l3 Weak Strong .150A -.2l66 .5033 .0925 I II III I_V_ A Complex Simple .3875 0313 —.l801 .6121 16 Dishonest Honest .2066 086A .A388 .A861 62 The remaining three scales are of questionable value in terms of film-music measures. Table 32 in Appendix C gives the averaged D2 scores between all possible pairs summed across examples for each scale. Complex - Simple, Meaningful - Meaningless, and Worthless - Valuable will be found to be the most ambiguous scales across all three modes. Perhaps they factor together because they are equally useless. The main dimensions in this study, then, are evaluation and activity. CHAPTER V SUMMARY AND CONCLUSIONS The broadest generalization is that music appears to exercise considerable influence when made to co-exist with a film scene. The influence is systematic and to a great extent it can be predicted with Osgood's point of resolution formula. Overall, the film component has a greater influence on the film with music combination than does the music component. This influence, however, varies considerably between examples. The precision of the point of resolution predictions as well as the relative influ- ence of the separate components seem to be highly depend- ent on the degree of congruency existing between the two components. Structure, which is an assessment of viewer consensus, also affect component relationships, but to a lesser degree. Music influence was found to be greater than the film component influence only in those incongru- ous combinations with high film structure and low music structure. This finding was somewhat unexpected and no satisfactory explanation is to be found in the present data as to why the judgments of the film-music combina- tions were influenced by the more ambiguous component. It might be conjectured that the known moves to structure the 63 'r 6A unknown. However, except for this one unexpected finding, all hypotheses were supported, as summarized below. Hypotheses Hypothesis 1. Given a specific audience and a series of musical accompaniments to a film scene, the music can significantly change the inter- pretation the audience gives for the combined presentation as measured on a form of the Semantic Differential. A matched, two-sample, non—parametric test between the two music treatments across all film samples resulted in the rejection of the null hypothesis of no difference at a two-tailed confidence level greater than .005. There- fore, it seems safe to conclude that the music associated with a film scene can cause a difference in the way the film with music combination is judged. Separate tests between the two music treatments for each film example, however, showed significance for only six of the nine ex- amples. It is evident that the component influence is not equal in all conditions. Hypothesis 2. Given a set of film and music examples, the observed scores for the combina- tions will significantly correlate with scores predicted by the point of resolution formula. Predictions of judgments for the combined film with music examples were calculated according to the Osgood point of resolution formula utilizing the mean response on each scale derived from independent assessment of the separate components. Overall, a highly significant correl- ation of .87 was found between the predicted and observed 'I' 65 scores. This suggests that the point of resolution formula is useful in this context and that both the film and music components contribute to whatever meanings the combination is judged to have. Separate correlations for each example, however, showed a correlation range from a low of .13 to a high of .95. It is again evident that the component influ- ence is not equal in all conditions. Hypothesis 3. Given measures of film meaning and music meaning, the congruency between com- ponents will systematically affect the accuracy of the point of resolution predictions. By counting the number of incongruous scales in each combination, it is possible to say that some combinations are more incongruous than others. The high congruency group was the set of 6 film and music combinations that had 0 to 2 incongruent scales between them. The medium con- gruency group was the set of 6 combinations that had from 3 to 7 incongruent scales. The 6 low congruency examples had from 8 to 1A incongruent scales in each example. The distance, or D score, between the point of resolution pre- diction and the observed scale mean was calculated and summed by levels of congruency. If levels of congruency did not influence the precision of the point of resolution predictions, it would be expected that the D2 scores would be randomly distributed among the levels of congruency. A one-way analysis of variance performed on the three levels of congruency gave a between cell significance greater than .005. Related T-tests allowed rejection of 66 the null hypothesis of no difference between high and low congruency at the .01 two-tailed level of confidence. No significance was found between medium congruency and high or between medium and low congruency. It is safe to con- clude that predictions made with the point of resolution formula are more accurate for combinations that are con— gruent and that the predictions become less accurate as the number Of incongruent scales increases. Hypothesis A. Given measures of film meaning and music meaning, the congruency between components will systematically affect the relative influ- ence of the component parts. Assessment of component influence was made by calcu— lating the distance, scale by scale, between the mean judg- ments of the combination and the mean judgment of each component. The component that is less distant from the combination measure is assumed to have the greater influ- ence. Overall, film had a greater influence on the com- bination judgments since the sum of D2 for the film com- ponent was significantly smaller than the sum of D2 for the music component. In general, each component had a greater influence in high congruency combinations than it had in low congruency combinations, significant at greater than the .001 level. This suggests that as the number of incongruent scales is increased, the distance between each component and the combination increases. However, it was also noticed that the influence of the music component was about equal to that of the film component in cases of low 67 congruency. This suggests that music has a greater oppor- tunity to influence film-music combinations when the con- gruency between components is low. Hypothesis 5. Given measures of film structure and music structure, the structure of the com- ponents will systematically affect the accuracy of the point of resolution prediction. Structure is an assessment of viewer co-orientation. High structure reflects relatively high consensus and low structure, relatively low consensus in the viewer sample toward a specific example. Structure is determined by calculating the difference between all possible pairs with- in the sample for each response item. Summing across the squares of differences between pairs gives an index of structure for each example. The rank order of these ex- amples is used as the independent variable while analyzing the point of resolution D2 scores. Using three levels of component structure, the point of resolution predictions were found to be most accurate for the medium structure and least accurate for low structure. Separate one-way analysis of variance showed between level significance of better than .01 for the film but not for the music. The fact that the point of resolution formula was the most accurate for both film and music at the medium level of structure points out the "normalizing" characteristic of the formula. The formula does not seem to take into con- sideration the extremes of structure and congruency. a... A 68 Hypothesis 6. Given measures of film structure and music structure, the structure of these components will systematically affect the rela- tive influence of the film and music. In general, there seems to be an inverse relationship between structure and component influence. That is, com- ponents that achieved high agreement among subjects shift more when combined with the second channel than do com- ponents that are relatively ambiguous. In its broadest generalization this finding may suggest that people are quite willing to change their judgments about something when new or different information is added. However, this finding was somewhat unexpected for which the data does not provide a satisfactory explanation. Apparently, the known makes a greater shift in order to structure the unknown. Osgood's congruency concept may not be the most useful approach in answering the above unexpected finding for as Ivan Prestonl points out, Osgood obtains great precision but in so doing he limits the generality of his predictive device. Under unbalanced circumstances, the evaluation of both components must change to a specified extent. Balance theory, of course, is not the only approach to studying communication conflicts. Rokeach2 deals with resolving informational conflicts in terms of the 1Ivan Preston, "A Review of Balance Theories," (un- published mimeographed paper, Michigan State University, 196A). 2M. Rokeach, The Open and Closed Mind, (New York: Basic Books, 1960). 69 individual's personality as reflected by his belief system. Festinger3 has looked at conflicting information and changes in belief as influenced by social or group pres- sures. From an information viewpoint, Broadbentu suggests that channel constraints limit informational input and therefore influences multi-sensory effectiveness. From a perception viewpoint, the human mind is seen to adjust incoming information to fit past expectations or basic "natural" principles. The Gestalt principle of closure is but one of the several concepts used to explain the ways in which informational parts are perceived, the way in which subwholes emerge, and the way in which the grouping 5 Marshall McLuhan,6 on the other of information occurs. hand, is more concerned with how mediation of environmental information structures basic dimensions of cognition much as Benjamin Whorf7 suggests that man's mental abilities develop within the constraints of his available language. From the McLuhan or Whorf view, the human mind is seen to 3Leon Festinger, Henry W. Riecken, Jr., and Stanley Schachter, When PrOphecy Fails, (Minneapolis: UniverSity of Minnesota Press, 1956). “D. E. Broadbent, Perception and Communication, (New York: Pergamon Press, 1958). 5William N. Dember, Psychology of Perception, (New York: Holt, Rinehart and Winston, 1960). 6Marshall McLuhan, Understanding Media: The Exten- sions of Man, (New York: McGraw-Hill, 196A). 7B. L. Whorf, Four Articles on Metalinguistics, ‘(Washingtonz Foreign Service Institute, 1950). 70 evolve according to the classes of information it has de- velOped a necessity to recognize. It is impossible to suggest, at this time, whether the shifts toward the more ambiguous component as observed in this present study are a tendency to adjust to some greater whole and thus "close" on the unknown, whether the existing information available in the music and visual com- ponent has provided the necessary dimensions for struc- turing a new meaning in the ambiguous area, or whether additional investigation will provide a suitable explana- tion from a balance theory viewpoint. Scale Analysis Although this study was not designed to map the dimen- sions of film-music meaning, it may be useful in terms of future research to look at the way meaning is structured, given this set of Semantic Differential scales. A four rotation factor analysis was applied to each of the three modes of presentation. Overall, the scales factored in about the same way for film, music, and the combined film with music. Two main dimensions appeared in this selected set of scales -- evaluation and activity. In general, it seems quite possible to select a set of scales that are useful in judging separate film and music and film with music modes of presentation. Future Research The most obvious need is for research that looks sys- tematically at the interaction between structure and 71 congruency. This study has pointed to some trends but there were insufficient combinations available at the various levels of structure and congruency to establish the existing relations. To conduct such experiments a more refined method of pre-selecting the film and music samples is needed so that adequate representation of the sample space is insured. However, before a serious ef- fort of sample selection can be made, a careful evaluation of the co-orientation measure of structure is also needed. The data in this study suggests that the rank order of example structure was not stable between the pilot and ex- perimental populations. But since the pilot population was much more sophisticated in terms of media and non- verbal exposure, it is not known whether the measure lacks validity or whether the two populations were actually quite different in their perceptions of certain film and music examples. For whatever reason, this instability of the co-orientation measure caused certain inconveni- ences in the treatment of data and loss of significance for most of the tests involving measures of structure. Certainly, further use of this measure should not be en- couraged until either its validity is established or its extreme sensitivity is understood. This study was exploratory in nature. Some trends were identified and a few hunches confirmed. But very little has been done toward the development of a theo- retical framework for instructional design. The next 72 step would be to look at the congruency and structure interaction concept from a multi-channel, information theory viewpoint. It is very possible that the informa- tion overload problems found in multi-channel studies are indeed similar to the high structure, high incongru- ency interactions found in this study. Conclusion This study investigated congruency and structure as two of the conditions which systematically influence the relative effectiveness of the music component in a simul- taneous film visual and music combination. The first level of the study simply asked if the perceived judgments of a filmed visual can be significantly altered by the music that is made to co-exist with the scene. It was found that the perceived judgments of the visual could be altered. The second level investigated the usefulness of the point of resolution formula for predicting the judg- ments of film with music combinations from knowledge of the component parts. It was found that the point of resol- ution was useful and, in general, extremely accurate.’ The third level considered the effects of congruency and struc— ture and the interaction of these conditions on point of resolution predictions and component performance. The point of resolution predictions became less accurate as the number of incongruent scales increased. The music com- ponent was more influential when the congruency between the components was low. In terms of structure, the point of 73 resolution predictions were more accurate at the medium level of structure for both film and music components. The low structured film component was most influential on combined judgments and the high structure least influen- tial. For music, medium structure was most influential and high structure least influential. This tendency for meanings to shift toward the ambiguous component was some- what unexpected and suggests some kind of perceptual "filling-in" or evolving of new meaning in these ambiguous or low structure areas. Finally, although the film com- ponent had a greater overall influence on the combination, congruency and structure seemed to interact so that the music component had a greater influence on the combined judgment under conditions of incongruent components with high film structure and low music structure. Music, of course, is but one element in the total product called a film. Congruency and structure are but two of the many conditions existing between the variables, and film music does not function solely to alter the visual message. So, although congruency and structure have been shown to systematically influence the relative effective- ness of the music component, much remains undone. But as the flexibility and constraints of the film elements are identified, principles of design will emerge to serve the intentions of the educator. ‘J-an . BIBLIOGRAPHY BIBLIOGRAPHY Bass, Saul. "Opening for 'Walk on the Wild Side'." Pagina, International Magazine of Graphic Design. Milan, Italy: Gennaio, 196A. Broadbent, D. E. Perception and Communication. New York: Pergamon Press, 1958. Dember, William N. Psychology of Perception. New York: Holt, Rinehart and Winston, 1960. Farnsworth, Paul R. The Social Psyphology of Music. New York: Holt, Rinehart and Winston, 1958. Festinger, Leon; Riecken, Henry W.; and Schacter, Stanley. When Prophecy Fails. Minneapolis: University of Minnesota Press, 1956. Greenberg, Bradley S. "The Effects of Communication Incompatibility on Children's Judgments of Tele- vision Programs." Journal of Broadcasting, VIII (Spring, 1965), pages 157-171. Greenhill, Leslie P. "Review of Trends in Research on Instructional Television and Films." Research in Instructional Television and Film. Washington: U.S. Government Printing Office, 1967. Harrison, Randall Paul. "Pictic Analysis: Toward a Vocab- ulary and Syntax for the Pictorial Code; With Research on Facial Communication." Unpublished Ph.D. dis- sertation, Michigan State University, 196A. Hoban, C. F., Jr., and Van Ormer, E. B. Instructional Film Research, 1918-1950. Pennsylvania State University Instructional Film Research Program. Port Washington, N.Y.: U.S. Naval Training Device Center, Office of Naval Research, Tech. Report No. SDC 269-7-19, 1950. Kerrick, Jean S. "The Influence of Captions on Picture Interpretation." Journalism Quarterly, XXXII (Spring, 1955), pages 177-I82. 7A 75 Kerrick, Jean S. "News Pictures, Captions and the Point of Resolution." Journalism Quarterly, XXXIV (Spring. 1959). pages 183-189. Lumsdaine, A. A. "Audio-visual Research in the U.S. Air Force." Audio-Visual Communication Review, I (Spring. 19533. pages 76-90. Lumsdaine, A. A., ed. Student Response in Programmed Instruction. Washington: National Academy of Science, 1961. Lumsdaine, A- A- "Instruments and Media of Instruction." Handbook of Research on Teaching. Edited by N. L. Gage. ChICago: Rand McNally & Co., 1963. Lumsdaine, A. A., and Gladstone, A. I. "Overt Practice and Audio-Visual Embellishments." in Learning from Films. Edited by M. A. May and A. A. Lumsdaine. New Haven: Yale University Press, 1958. May, M. A.,and Lumsdaine, A. A., eds. Learning from Films. New Haven: Yale University Press, 1958. McLuhan, Marshall. Understanding Media: The Extensions of Man. New York: McGraw-Hill, 196A. Mehling, Reuben. "Attitude Changing Effect of News and Photo Combinations." Journalism Quarterly, XXXIV (Spring. 1959), pages 169-198. Morris, John. "Technical Report A5." Computer Institute for Social Science Technical Reports. East Lansing: Michigan State University, 1967. (Mimeographed.) Nebergall, Roger Ellis. "An Experimental Study of Rhetorical Clarity." Unpublished Ph.D. dissertation, University of Illinois, 1956. Neu, D. M. "The Effect of Attention-gaining Devices on Film-mediated Learning." Journal of Educational Psychology, XXXXII (Spring, 195177pages A79-A90. Newcomb, Theodore M. "An Approach to the Study of Commun- icative Acts." Psychological Review, LX (No. 6, 1953), pages 393-AOA Osgood, C. E.; Suci, G. J.; and Tannenbaum, P. H. The Measurement of Meaning. Urbana: University of Illinois, 1957. 76 Pallett, Earl Marshall. "Music Communication Research: The Connotative Dimensions of Music Meaning." Unpublished Ph.D. dissertation, Michigan State University, 1967. Preston, Ivan. "A Review of Balance Theories." (Unpub— lished mimeographed paper, Michigan State Univer- sity, 196A.) Reid, J. C., and MacLennan, D. W., eds. Research in Instructional Television and Film. Washington: U.S. Government Printing Office, 1967. Rokeach, M. The Open and Closed Mind, New York: Basic Books, 1960. Siegel, Sidney. Nonparametric Statistics for the Behavioral Sciences. New York: McGraw-Hill, 1956. Tannebaum, Percy H. "The Effect of Background Music on Interpretation of Stage and Television Drama." Audio-Visual Communication Review, IV (Spring, 1956), pages 92-103. Tannebaum, Percy H. and Osgood, Charles E. "Effects of Color on Meaning of Advertised Products." Reported in Osgood, et al. The Measurement of Meaning. Urbana: University of Illinois Press, 1957. Travers, R. W., et a1. Research and Theory Related to Audiovisual Information Transmission. rev. ed. Kalamazoo: Western Michigan University, 1967. Tucker, William T. "Experiments in Aesthetic Communications." Unpublished Ph.D. dissertation, University of Illinois, 1955. U.S. Ordinance Corps. Table of the Cumulative Binomial Probabilities. Washington: The Corps, 1952. Whorf, B. L. Four Articles on Metalinguistics.-Washington: Foreign Service Institute, 1950. Zuckerman, J. V. Music in Motion Pictures: Review of Literature with Implications for Instructional Films. Pennsylvania State University Instructional Film Research Program. Port Washington, N.Y.: U.S. Naval Training Device Center, Office of Naval Research, Tech. Report No. SDC 269-7-2, 19A9. APPENDICES 1. APPENDIX A Film examples Music examples 77 zoCmHQ moafimm mmCmHQ wowccmpfihm mfioowaoaozocm mmCmHQ mmCmHQ onCCMQHLm mfioomaofiomocm Lozomodmmz smcmfio paw; BOHCCMuHLm mfiommqoaozocm Loosoopm =EOB< map ccmfipm “so: =mwwHHH> wcfinmfim cacaxmzz =capoz pcmflm was go mumpomm= zzpfiCSEEoo p30: :Ucozmm Ucm mama: =mmmcpmnflfiz mpficz= =cmEmOHHom may: :ocmafiuomz mpcmpsmpmom CD: :EMHW: :phmmmfl $58 CH mhfid: meHB EHHm mmqmzHpmz .mporoam cam swamp pcfima vmmeHc< .Hamnmmmn mcfizman whom .maanm momam nonmefic¢ .mom on» oucfi mesh meHEEmH wcficcsm .man pm smEmOHHom .ocoom pmccfio Hospom .mmoocmo cmfimocfifimm .egfin mp emxomppw wows camfim mcoom 78 Hmmm 3Q monm 3Q mom 0 mabm 3G < < m < ¢ mm» o m wmw o < map 0 m Ham 0 m :mum 30 < mzwm 3Q ambasz whoomm mnoomh fifiampoo mooomh HHHmpoo Hamaamco mpoomh Haampoo Hamaamno Hamaamno Hamaamno Hflmaamno mooomh Haampoo mooomh HHHmpoo zpmpofiq mmqmzfiufisfimm= meHB H oxmc-wm APPENDIX B Instruction Form Used with Population A Response Form Used with Population A Scale Analysis, Population A, Music Scale Analysis, Population A, Film Revised Instruction Form Revised Response Form 79 INSTRUCTION FORMUSED WITH POPULATION A MUSIC — FILM INFLUENCE STUDY Instructional Media Center Michigan State University NAME MAJOR STUDENT NUMBER CLASS MALE FEMALE INSTRUCTIONS: You will be given a series of music and film examples. After each example is shown or played, you will be asked to respond to a series of adjective scales. A new page is to be used for each music/film presentation. Each page contains eighteen adjective pairs. The word pairs are divided into three groups with directions for each group repeated on every page. Respond to all eighteen scales immediately after each presentation. All pages are the same. The adjective pairs are separated by a scale contain— ing seven positions. Please check at a point along each scale to indicate where you rate the music/film sample. Here is an example: 3 2 1 O l 2 3 NICE : : : : : : AWFUL This rating scale is bound by the words "NICE" and "AWFUL." The more "NICE" you feel the music/film is, the farther to the left of the scale you would place your check; the more "AWFUL" you feel the music/film is, the farther to the right you would place your check. If it is hard to decide if it is "NICE" or "AWFUL," or you feel the adjective pair is not relevant to the particular music/ film sample, place a check in the central space, under the zero (0). This means "undecided" or "irrelevant." There are no right or wrong answers. The best response is what you feel is appropriate. Do not spend too much time on any one item. PUT DOWN YOUR FIRST IMPRESSION. Be sure to place a check on every scale. When you have finish— ed each page, turn the page, and the next music/film example will be given. FILM SAMPLE A THE SUBJECT OF THE FILM IS: KIND SAD GOOD UGLY JOYFUL AGITATED B. THE SITUATION INDICATES: DANGER SECURITY PLEASURE EXCITEMENT VIOLENCE FRUSTRATION NO. —- -————— ... ~— 3 : 2 : l O C. THE PICTURE IS: REALISTIC WHIMSICAL SERIOUS DECEPTIVE HONEST SOLEMN 3 : 2 : l RESPONSE FORM USED WITH POPULATION A CRUEL HAPPY BAD BEAUTIFUL MOURNFUL RESTFUL SAFETY JEOPARDY PAIN CALMNESS PEACEFULNESS SATISFACTION DISTORTED SINCERE HUMOROUS TRUSTWORTHY FALSE PLAYFUL 81 SCALE ANALYSIS, POPULATION A, MUSIC Table 18.--Varimax rotation analysis of scales Used with population A, music examples I II III IV 1 Kind Cruel .6608 —.A679 .325“ .0521, 2 Sad Happy .92u3 -.0819 .1566 .0951 3 Good Bad .7u93 -.3ull .3383 .1355 A Ugly Beautiful .SSAO .A872 .3AAO .3A92 5 Joyful Mournful .9008 .0909 .181“ .1692 6 Agitated Restful .1276 .8362 .1378 .1011 7 Danger Safety .7015 .A755 .35A2 .222“ 8 Security Jeopardy .68A5 -.A9A0 .AOAO .1958 9 Pleasure Pain .8805 -.2A51 .1787 .05A2 10 Excitement Calmness .2510 .8709 .0297 .0079 11 Violence Peacefulness .3982 .7932 .2202 .0881 12 Frustration Satisfaction .7085 .U33l .3076 .0887 13 Realistic Distorted .0893 .0151 .7205 .1986 1A Whimsical Sincere .7406 .195“ .3399 .2981 15 Serious Humorous .9056 -.1307 .0202 .1715 16 Deceptive Trustworthy .0927 .2259 .8A7A .07A1 l7 Honest False .INU2 -.2171 .8193 .0610 18 Solemn Playful .935“ -.1368 .OA3A .0270 PrOportions of Variance “3% 20% 16% 03% 82 SCALE ANALYSIS, POPULATION A, FILM Table l9.-—Varimax rotation analysis of scales used with population A, film examples. I II III IV 1 Kind Cruel .6810 .5115 -.0042 -.1807 2 Sad Happy .8u57 .1925 -.0852 .2599 3 Good Bad .8018 .2031 -.1924 -.0549 4 Ugly Beautiful .6621 .2739 -.O209 .0102 5 Joyful Mournful .8121 .0890 .0905 —.3167 6 Agitated Restful .2170 .8469 .0535 .0562 7 Danger Safety .5249 .7283 —.0530 .0925 8 Security Jeopardy -.5767 .6622 -.0210 -.1427 9 Pleasure Pain .7449 .3800 .0144 -.2306 10 Excitement Calmness .1381 .8560 .0273 -.0229 11 Violence ' Peacefulness .4818 .7567 .0003 .1631 12 Frustration Satisfaction .6314 .4888 .1177 .0835 13 Realistic Distorted .0763 .0485 —.7761 -.0014 14 Whimsical Sincere .0918 .0381 .6967 -.4271 15 Serious Humorous .1903 .0890 —.2282 .8063 16 Deceptive Trustworthy .0839 .1119 .7947 —.0685 17 Honest False .1014 .0115 —.7734 .1108 18 Solemn Playful .2959 .0899 -.1069 .8009 Proportions of Variance 27% 21% 14% 10% 83 REVISED INSTRUCTION FORM MUSIC - FILM INFLUENCE STUDY Instructional Media Center Michigan State University NAME MAJOR STUDENT NUMBER CLASS MALE FEMALE INSTRUCTIONS: You will be given a series of music and film examples. After each example is shown or played, you will be asked to respond to a series of adjective scales. A new page is to be used for each example. Each page contains eighteen adjective pairs. Respond to all eighteen scales immediately after each presentation. All pages are the same. The adjective pairs are separated by a scale contain- ing seven positions. Please check at a point along each scale to indicate where you rate the example. Here is an illustration: 3 : 2 : l : 0 : l : 2 : 3 NICE : : : : : : AWFUL This rating scale is bounded by the words "NICE" and "AWFUL." The more "NICE" you feel the example is, the farther to the left of the scale you would place your check; the more "AWFUL" you feel the example is, the far— ther to the right you would place your check. If it is hard to decide if it is "NICE" or "AWFUL," or you feel the adjective pair is not relevant to the particular ex- ample, place a check in the central space, under the zero (0). This means "undecided" or "irrelevant." There are no right or wrong answers. 'The best response is what you feel is appropriate. Do not spend too much time on any one item. PUT DOWN YOUR FIRST IMPRESSION. Be sure to place a check on every scale. When you have finish- ed each page, turn the page, and the next example will be given. 84 REVISED RESPONSE FORM E X A M P L E N U M B E R PRACTICE KIND CRUEL HEAVY LIGHT SAD HAPPY COMPLEX SIMPLE GOOD BAD PASSIVE ACTIVE UGLY BEAUTIFUL EXCITED CALM WORTHLESS VALUABLE DANGER SAFETY SLOW FAST PLEASANT UNPLEASANT WEAK STRONG MEANINGFUL MEANINGLESS VIOLENT PEACEFUL DISHONEST HONEST SERIOUS HUMOROUS HARMONIOUS DISSONANT APPENDIX C Music Scale Means Film Scale Means Predicted Scale Means Observed Scale Means MUSIC SCALE MEANS 85 Table 20.-—Population B, music example scale means, N=68. Example Scale 1 2 3 4 5 6 7 8 9 1 5.53 2.26 1.90 1.98 1.94 5.41 2.56 2.88 5.00 2 1.87 6.35 5.15 5.20 6.79 1.72 6.07 5.25 1.65 3 2.22 6.03 3.23 5.13 6.84 1.59 6.14 5.85 1.59 4 2.41 4.25 4.47 3.63 5.93 2.10 4.45 3.04 2.40' 5 4.45 2.04 2.31 2.18 1.78 5.12 2.57 2.28 4.75 6 4.85 5.57 2.88 3.29 6.03 3.38 5.72 6.12 3.29 7 2.95 5.76 6.07 6.23 5.44 2.65 5.18 5.26 3.69 8 3.03 3.38 6.19 5.73 2.44 4.75 2.29 1.81 5.13 9 4.38 5.00 5.16 5.16 4.84 4.19 4.16 4.66 4.43 10 1.66 5.19 5.48 5.73 5.75 1.93 5.06 4.76 3.32 11 2.51 5.10 1.98 2.59 5.82 1.70 5.73 6.40 2.22 12 4.78 2.04 1.98 1.87 1.68 5.28 2.45 2.43 4.47 13 5.85 4.32 3.45 4.28 4.40 5.35 4.07 5.19 4.90 14 2.66 2.90 2.70 2.78 3.28 3.18 3.76 3.32 32.97 15 2.32 5.50 6.32 6.25 5.50 3.01 5.20 4.48 4.04 16 2.82 4.94 5.44 5.32 5.38 2.90 4.76 4.84 3.95 17 1.79 4.73 2.47 3.07 6.07 1.78 5.44 4.90 1.66 18 4.28 3.00 2.26 1.95 2.41 4.15 2.85 2.82 3.48 86 FILM SCALE MEANS Table 2l.--Population B, film example scale means, N=68. Example Scale 1 2 3 4 5 6 7 8 1 6.34 2.62 3.01 2.10 4.32 2.20 4.07 2.09 3.70 2. 2.16 4.76 4.76 5.28 4.22 5.50 4.63 5.94 3.50 3 1.93 4.82 5.68 5.54 3.78 5.41 4.57 6.29 4.22 4 3.53 3.25 4.91 5.03 4.29 5.25 3.98 3.68 1.98. 5 5.43 2.56 2.35 2.37 3.76 2.19 3.07 2.01 2.97 6 6.31 4.69 3.90 4.43 5.85 2.57 6.51 5.76 5.34 7 2.62 5.88 5.09 5.06 4.10 6.56 4.44 6.35 5.01 8 1.44 5.28 4.97 4.62 2.53 6.32 1.82 3.88 3.82 9 5.32 5.13 4.26 4.69 4.57 5.20 5.12 5.22 5.45 10 1.43 4.84 5.29 5.26 2.88 5.90 3.63 5.25 3.60 11 6.04 2.20 3.10 3.88 5.26 2.25 6.12 4.66 3.86 12 5.81 2.37 2.09 2.48 4.47 1.60 3.37 1.81 3.31 13 5.62 4.15 3.87 4.20 4.57 4.28 5.23 4.82 5.29 14 1.84 2.84 3.40 2.94 3.18 2.69 2.66 2.78 2.73 15 1.63 5.84 5.73 5.38 3.16 6.54 3.32 6.01 ‘4.88 16 3.82 4.98 4.97 5.44 4.40 5.20 4.44 5.03 4.60 17 1.84 2.51 3.48 3.12 3.35 2.97 3.35 4.48 2.32 18 4.53 2.70 2.88 2.78 4.03 2.37 4.18 2.60 2.98 87 50.0 50.0 55.: 00.0 05.0 00.0 05.0 00.0 50.0 00.0 55.0 :0.0 00.0 0:.0 ::.0 50.0 00.0 5:.: 05.0 50.: 50.0 :5.0 05.0 00.: 50.0 00.0 00.: :0.0 05.: :0.: 00.0 05.0 05.0 00.0 00.0 00.5 00.: 00.: 00.0 00.0 05.: 05.0 05.: 05.0 00.: 00.: :0.: 50.: 00.: :0.: 00.0 00.0 00.: 0:.0 00.0 55.0 55.0 0:.0 00.0 00.0 05.: 00.0 05.0 00.: 00.0 5:.: 0:.0 00.: 00.0 :0.0 00.0 40.0 05.0 :0.0 :0.0 05.0 00.0 0:.0 00.0 00.0 05.0 00.0 50.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 :0.: 00.: 00.0 50.0 :5.: :5.: 00.0 00.: 5:.0 05.0 50.0 05.: 50.: 0:.: :0.: 50.0 :0.: 05.0 05.0 00.5 :0.: 50.5 5:.: 5:.0 00.0 00.0 50.: 00.0 55.0 00.0 0:.:. 55.0 00.0 :0.0 00.0 00.0 00.0 00.: 05.0 05.0 00.: 00.: :0.0 00.: 00.0 00.0 50.0 00.0 50.0 00.0 00.0 00.0 00.: 00.0 50.: 00.0 :0.0 05.0 05.0 05.0 0:.: 00.0 00.0 5:.0 00.0 0:.: 00.: 05.: 50.0 00.0 00.0 00.5 50.0 55.0 05.: 05.0 0 .: ::.: 00.: 00.: 50.0 00.: 00.: 00.: 05.: 55.: :0.: 05.: 54.0 00.: 50.: 00.0 :0.: 00.0 00.: 00.0 :0.: 00.: 00.5 50.: 00.0 00.0 00.0 0:.0 00.0 00.0 :0.: 00.0 00.0 50.0 55.0 00 0 50.0 05.0 00.: 50.0 00.: 0:.: 00.0 :0.: 0:.0 00.: 55.0 00.0 -0.0 05.0 00.: 50.0 ::.0 05.0 00.: 05.0 0:.: ::.0 00.0 00.: 00.0 :5.0 00.0 :0.0 :0.0 5:.0 55.0 00.0 :0.0 00.0 00.: 05.0 50.: 5:.0 05.0 :0.0 00.: 50.: 05.0 00.0 :0.: 05.0 00.0 00.0 00.0 00.: 00.0 00.0 00.0 n-.: 50.0 05.: 00.: 00.: 00.0 50.0 00.0 00.: 00.: :0.0 ::.: 05.: 00.0 00.0 05.: 05.0 00.0 00.- 05.0 00.0 55.: 00.0 00.: 00.0 00.0 5:.0 :0.: 05.0 :0.0 05.: 0:.0 00.0 00.: 05.0 :0.0 00.0 00.0 05.0 :0.0 00.0 00.: 50.0 00.0 0:.0 00.: 00.0 :0. 00.: [0.0 00.0 05.0 05.0 00.: 00.0 00.: 00.0 :0.: 00.0 00.0 0:.: 00.0 0:.0 0:.: 00.0 :0.0 00.0 0:.0 00.0 05 55 05 05 :5 .05 05 55 05 0 0 5 0 0 : 0 0 5 650 mHQmem .m.O.nm «WCNOE ®HMOW UmDOHUQcHQ COHmeHDEOO 0.“de Spit) EHfihll.NN mHD, mz¢mz mammmmo l. Con-ruency and P.O.R. Predictions 0’2 2. Congruency and Component Effect 3. Structure and P.O.R. Prediction; 4. Structure and Component Effect 89 CONGRUENCY AND P.O.R. PREDICTIONS Table 24.—-0ne-way analysis of D2 between point of resolution predictions and observed scale means by levels of congruency. Source D.F. S.S. M.S. F Probability Between Groups 2 3.28 1.63 4.95 .008 Within Groups 321 106.15 .33 TOTAL 323 109.43 9O CONGRUENCY AND COMPONENT EFFECT A. MUSIC Table 25.--One-way analysis of D2 between music component scale means and film with music combination scale means by levels of congruency. Source D.F. S.S. M.S. F Probability Between Groups 2 44.64 22.31 9.15 <.0005 Within Groups 321 782.21 2.43 TOTAL 323 826.85 B. FILM Table 26.-—One-way analysis of D2 between film component scale means and film with music combination scale means by levels of congruency. Source D.F. S.S. M.S. F Probability Between Groups 2 67.04 33.51 33.3 <.0005 Within Groups 321 323.01 1.00 TOTAL 323 390.05 91 CONGRUENCY AND COMPONENT EFFECT — continued C. T VALUES Table 27.--Fisher-Student T—Test values between levels of congruency for film and music components. Levels Music Film High-Low 3.91 7.45 Medium-Low 2.88 4.62 Medium-High 1.17 (N.S.) 3.71 T 2.88 = P<.005 (.01 two tailed) 92 STRUCTURE AND P.O.R. PREDICTION A. MUSIC 2 Table 28.--One-way analysis of D between point of resolution predictions and observed scale means by levels of music structure. Source D.F. S.S. M.S. F Probability Between Group 2 1.23 .61 1.82 0.16 (N.S.) Within Group 321 108.20 .34 TOTAL 323 109.43 B. FILM Table 29.--One-way analysis of D2 between point of resolution predictions and observed scale means by levels of film structure. Source D.F. S.S. M.S. F Probability Between Group 2 3.35 1.67 5.06 .007 Within Group 321 106.08 .33 TOTAL 323 109.43 93 STRUCTURE AND COMPONENT EFFECT A. MUSIC Table 30.--One-way analysis of D2 between music component scale means and film with music combination scale means by high and low levels of structure. Source D.F. S.S. M.S. F Probability Between Group 1 9.46 9.46 3.72 .05 Within Group 322 817.39 2.53 TOTAL 323 826.85 B. FILM Table 31.--One-way analysis of D2 between film component scale means and film with music combination scale means by high and low levels of structure. Source D.F. S.S. M.S. F Probability Between Group 1 16.10 16.10 13.8 <.0005 Within Group 322 373.95 1.16 TOTAL 323 390-05 94 SUBJECT D2 SCORES Table 32.--Subject D2 scores summed by scales. MUSIC EXAMPLES FILM EXAMPLES 1 Kind Cruel 22.88 34.87 2 Heavy Light 29.46 43.66 3 Sad Happy 24.16 38.19 4 Complex Simple 44.00 60.95 5 Good Bad 30.77 45.15 6 Passive Active 39.88 38.95 7 Ugly Beautiful 27.38 41.44 8 Excited Calm 33.24 38.98 9 Worthless Valuable 32.05 47.74 10 Danger Safety 23.06 36.12 11 Slow Fast 21.10 40.28 12 Pleasant Unpleasant 32.61 41.71 13 Weak Strong 30.32 36.62 14 Meaningful Meaningless 35.94 51.72 15 Violent Peaceful 25.16 26.72 16 Dishonest Honest 26.55 42.35 17 Serious Humorous 21.76 37.12 18 Harmonious Dissonant 37.22 54.32 95 COMPARATIVE COMPONENT VALUES Table 33.--Component values summed by combinations. mCOapmsssm mHmom mo 509832 0:00: 0500m oo>pmmpo one oopOHompm cmmzumm New .mompsoo< coHpSHOmmm mo pcfiom .wucocoqsoo UoCHnEoo opp mo :00: mamom om>pmmno 0:0 :00: mamom concaooum COHpSHOmmm mo ucfiom cmozpmm coapmngnoo memo: sawem :0550209800 ocm memo: mamom pcocoasoo smmzpmm mam .mocmSHmcH ucmcoasoo Edam memo: mHMom COHpmcHnsoo 0:0 0:00: mamom pcmcoasoo cmmzumm New .mocosamcH pcmcoasoo 00052 055mm poonnsm somspmm mo mwmpm>< .05550355m pcmcoqsoo 855m 0550m pomnnsm comzumm ma mwmpm>¢ .oLSposppm pcmcoasoo 000:2 mucocoasoo cmozpom 000000 mSOSLwcoocH 50 509532 Lonssz maasmxm ucmcoqsoo 6055 geneaz maasmxm neocoasoo 000:2 .HQQESZ COHQGCHDEOO OHmSZ EUH3 25.—”Hm 10935228.“.60“ 009576 11 1 1 11 1 78222686716u39u282 7360u3906120u13u 98 5657u1031731202055 1 111 1 1 5109158652935188u6 99989799977198 96 98 361.“.6790“. 3368 06989 11127692.“..u. 96378.“.5“ u. 23552775.“. 95491270 1.“. 1 3 1 3 2 “89u137819u69u2098 358379352332“. 06157 38 7.“. 106 53776 96 7959 211123 1532 1 1 5790u85885790u8588 202737625202737625 78535357.“. 78535357.“. 766695566766695566 3310.“..“776 7u3761u30 .4 38 56818686.“.168835 98 767237u7793u7286 5556u76u6hfihfi5665756 1&127150815u30291u 1 1 1 1 123u56789123u56789 123.“. 5678 98 5179362“. 123u567890123u5678 111111111 HXCH IGRN STRTE UNIV. LIBRARIES \ “WINl||1|\|\lllllMIIIWIIIWIIHIIWIIIHIHII 1293105895902 H 3