experimental studies of the elements of expression in music

8/12/2019 Experimental Studies of the Elements of Expression in Music

1/24

Experimental Studies of the Elements of Expression in MusicAuthor(s): Kate HevnerSource: The American Journal of Psychology, Vol. 48, No. 2 (Apr., 1936), pp. 246-268Published by: University of Illinois PressStable URL: http://www.jstor.org/stable/1415746.

Accessed: 04/05/2013 07:40

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at.http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

.

University of Illinois Pressis collaborating with JSTOR to digitize, preserve and extend access to The

American Journal of Psychology.

http://www.jstor.org

This content downloaded from 129.240.0.82 on Sat, 4 May 2013 07:40:21 AMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/action/showPublisher?publisherCode=illinoishttp://www.jstor.org/stable/1415746?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/1415746?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=illinois


2/24

EXPERIMENTAL STUDIES OF THE ELEMENTS OFEXPRESSION IN MUSIC

By KATE HEVNER, University of Minnesota

The experiments to be described here deal with the problem of meaningin music, or, more explicitly, with the affective value and the expressive-ness of music. In some earlier experiments the author had developed anobjective and quantitative method for isolating certain variables in musicstructure and measuring their expressiveness, and by application of themethod had investigated the affective value of the major and minor modes.'It is the object of the present study to improve the processes for treatingthe results from the method, and to discover the effect of three othervariables, the rising and falling of the melodic line, the firm or flowingmotion in the rhythm, and the simplicity or complexity of the harmony.

Obviously, in undertaking such a study, it is assumed that there is an expressive-ness in music which is so generally and consistently associated with it that it willlend itself to experimental variation and control. We find it necessary to refer tothis statement as an assumption because of the fact that in the history of musicalculture, there has been a lengthy controversy over the problem of expressiveness inmusic, and much difference of opinion as to whether music can be made to expressdefinite emotions and concepts and sentiments, or whether it is merely suggestive andmoving in a very general way and to be interpreted quite variously by the differentlisteners who are attending to it. On the one hand, we find the very important andsuccessful use of musical expression in the modern opera, the tone poems, and otherforms of descriptive music which have met with great popular success. The volumesupon volumes of program notes also testify to the popularity of the assumption withconcert audiences. Likewise, in experimental psychology we have evidence that atleast the general mood or affective state suggested by any one composition is fairlyconstant and universal.2 The introspective reports of individuals who try to analyzetheir experience in listening also show that this interpretative activity is of veryfrequent occurrence,8and furthermore that it is often the means by which apprehen-sion or appreciation of a more difficult type is developed. On the other hand, it is asimple matter to demonstrate that listeners often disagree markedly in their specificinterpretations of music, and some of the differences of opinion of eminent criticson important symphonies have become world-famous. It is certainly clear at least,that there are different kinds of music, formal and descriptive, absolute and prag-matic, pure music and music full of symbolism, and also that there are differentkinds of listeners to music. In short, there are many ways of enjoying music, and many

* Accepted for publication July 1, 1934.1K. Hevner, The affective character of the major and minor modes in music,this JOURNAL, 7, 1935, 103-118.2 Max Schoen, editor, The Effects of Music, chaps. VII, VIII.3Max Schoen, The aesthetic attitude in music, Psychol. Monog., 39, 1928, (no.178), 162-183.246

This content downloaded from 129.240.0.82 on Sat, 4 May 2013 07:40:21 AMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp


3/24

ELEMENTS OF EXPRESSION IN MUSICtypesof music o enjoy,andof exquisitelyelightfulmusicof all varieties, ndenthusiasticisteners f all kinds, herehas neverbeenanylack.Since there has been so much confusion and disagreementattendingthe problemof meaning in music, it will be necessary o state clearlyatthe outsetwhat phaseof this complexquestionwe are studying,and whatfurtherassumptions ndtheorieswe endorse.The limitationsof expressive-ness,its psychological xplanation, ts use in understanding ndappreciatingmusic, and the particularkind of symbolismwhich it employs,we shalldiscussat lengthelsewhere4 for the presentwe shall sketchthe underlyingtheoryand methodof our experimentsas brieflyas possible.The presenceor absenceof meaning,and the particularquality of themeaningin anymusic is dependenton a numberof factors: the form andstructureof the music itself, the attitude of the listeners,their previousexperience,their training, talent and temperament,and their momentarymood and physiologicalcondition. A certainflexibility in expressivenessis thereforepostulated,and experimental tudieswill be the meansof de-terminingthe nature and limits of this flexibility.The suggestivenessof music does not dependdirectlyon the amountofformaltrainingwhich the listenerhashad,but is apparentboth to thetrainedand untrained. Previousexperimentshave shown that the perceptionofmeaningis aidedand facilitated o someextentby musicaltraining,experi-ence, and interest,but that the interpretations re affectedby these factorsmuch essthanpopularopinionhassupposed.

Studies of a special group of students with no background or training in musichave shown that they differentiate the mood effects of the major and minor modequite distinctly and readily. The measurements show that the typical or median studentin this special group has never had any private music lessons of any sort, does not playany musical instrument, has no musicians in his family and never attends operas orconcerts. He can read music a little for singing, has sung two years in a high schoolchorus and has a piano and a radio in his home; and he is almost as suggestibleto the mood effects of music as his highly trained classmate.6The Os (450 in number) have, in the experiments to be described below, beenmeasured for their musical interest, training, attitude, and appreciation, and theyrepresent a very wide range on all of these scales, with the large majorities falling intothe untrained and untalented groups. They include a few students who are preparingfor a professional career in music (usually teaching), and a few who have anaversion to music and avoid all contact with it; but for the most part they claimonly a casual interest in music, have had a lesson or two, exhibit some little prefer-ence, have no knowledge of the actual form and structure of musical compositions,

4K. Hevner, Expression in music: A discussion of experimental studies andtheories, Psychol. Rev., 42, 1935, 186-204.Hevner, op. cit., this JOURNAL, 7, 1935, 115 ff.

247



4/24

and are strikingly lacking in a vocabulary with which to describe what they hear andwith which to identify musical sounds.The symbolism of music is obviously not as exact or as specific as verbalsymbolism, but we begin with the assumption that it is not chaotic, and notentirely without system. On the contrary, we believe that it may be possible to dis-cover the presence of definite principles together with a certain orderliness in therelation between musical structures and the various emotions and sentiments.

In our experimental search for those elements of musical structure which mustbe at the basis of any such orderly system, there are certain details in methodologywhich must receive careful consideration. Since we are looking for elements of musicwe must be sure that the material provided for observation represents real music andnot merely elements trimmed down for experimental purposes to such an extent thatall the music has been left out. The outline of a rhythmic pattern, to choose one ofthe possible elements, tapped out with a hard wooden stylus, is but the bare skeletonof a rhythm, rattling its dry bones in vast emptinesses, and far different from theliving, throbbing rhythm that pulsates through the whole body of a musical compo-sition. The isolation of the variable to be studied must not be accomplished by anactual isolation of the one particular element from all the relationships which makeit musical, but by some better method; by the study of various compositions in whichthis one element predominates, or by arranging several versions of one musical com-position which differ from each other only with respect to the element under con-sideration.

We must also recognize the importance of the cumulative effect which the succes-sion of stimuli makes possible, and on which the effectiveness of some of the ele-ments largely depends. The repetition of certain stimuli and their natural support byother qualities in the body of the music is necessary in order that the human organismmay perceive them. The sadness which is so strongly associated with the minordoes not reside in the minor chord alone, but only in music written in the minormode, with all the accumulation and interrelation of stimuli which the use of thatmode implies. Again, music is not a spatial, but a temporal art, and distinguishedfrom the visual arts by its dependence on temporal rather than spatial relationships. Inchoosing the elements which carry the meaning of music, and which will form thebasis of a system of musical symbolism we must be careful, therefore, to choose thoseelements which involve successive moments of time, such as rhythm and tempo, ratherthan intervals and chords. It is the masses of harmony, the resolutions and progres-sions, to which meaning attaches rather than to the triads themselves, and it is thetwists and turns of the melody around its keynote, the running of the melodic line,rather than certain momentary skips or intervals, which carry the suggestiveness andmeaning of the music.In securing the responses of the individual listeners, there are also many sourcesof error to be avoided. The responses should be as objective and simple as possiblein order that they may be summarized quickly and accurately by E, and treatedquantitatively. But O must not be forced into a judgment of the meaning when hedoes not spontaneously respond in that particular way, and he must be provided witha wide selection of responses, so that he may be exactly suited in his final choice.From the point of view of both O and E, a most satisfactory method of recordingresponses is by means of a list of 66 adjectives, arranged in 14 groups,6 from which

6 The list of adjectives is given in an earlier study, Hevner, op. cit., this JOURNAL,47, 1935, 109.

248 HEVNER



5/24

ELEMENTS OF EXPRESSIONIN MUSICthe listenerchecks all the adjectiveswhich he findsappropriateo the music.Thismakesthe businessof choosingresponseseasyand convenient or 0, and allows anobjectiveand quantitativereatmentby E. Fromthe original responsesheetsof thelisteners,each of whom had a fresh copy of the adjective ist for every differentmusical composition, t was a simplematterto tabulateall the votes for any oneadjective rom all the Os for each compositionalone, or for all the compositionswhich are alike in certainrespects.Then, by comparing he numbersof votes fordifferentadjectives,hemeaningsor affectivecharacteristicsf the compositionsouldbe ascertained.

8vigorousrobustemphaticmartialponderousmajesticexalting

7exhilaratedsoaringtriumphantdramaticpassionatesensationalagitatedexcitingimpetuousrestless

1spiritualloftyawe-inspiringdignifiedsacredsolemnsoberserious

6merryjoyousgayhappycheerfulbright

2patheticdolefulsadmournfultragicmelancholyfrustrateddepressinggloomyheavydark

5humorousplayfulwhimsicalfancifulquaintsprightlydelicatelightgraceful

4lyricalleisurelysatisfyingserenetranquilquietsoothing3dreamyyieldingtendersentimentallonging

yearningpleadingplaintive

FIG. 1. SHOWING THE ADJECTIVES FROM LIST I ARRANGEDIN 8 RELATED GROUPSBut a still furtheradvantagehas been gained for the use of this adjective-listmethodby a processof treatingthe data after they have been tabulated,a processof combining he individualadjectivesnto homogeneous roups,forminga more orless continuous cale.This processmakes allowance or the errorsarisingeitherfromthe small disagreements n the part of the listeners about the exact definitionsofwords (which are importantperhapsto the Os but certainlynot to E), and fromthe effectsof the momentarymood and physiologicalconditionof the differentOswhichmay modifyslightlythe mood-effectproducedby the music. It also simplifies

249



6/24

HEVNER

tremendously the task of summarizing the data for comparing the results of differentmusical compositions, because of the fact that the adjectives are not treated as 66individual entities, but are arranged into groups so that all those with the samefeeling-tone may be considered together.This grouping was accomplished in consultation with several Os who attemptedto arrange their groups of adjectives around an imaginary circle in such a way thatthe adjectives in one group would be closely related and compatible with eachother; that any two adjacent groups should have some characteristics in common, andthat the groups at the extremities of any diameter of the circle should be as unlikeeach other as possible. The arrangement in Fig. 1 seemed the most satisfactory. Thisarrangement is not the only possible grouping along such a scale, and does notperfectly satisfy all the demands of the opposing groups. It has value merely insimplifying and summarizing the results of these studies. The adjacent groupshave certain characteristics in common, and the transitions between them are madewithout abrupt or awkward changes in feeling tone. A complete circuit of the circlecarries one quite smoothly over the range of the most common affective experiences.In studying the arrangement in the circle, we may note several details which areconsistent with our usual concepts of affective states. Muscular tension is probablymost strongly associated with the vigorous-robust exciting-impetuous groups, and leastwith the groups directly opposite, the sentimental-yearning and quiet-serene groups.Likewise, activity is associated with the four groups beginning with vigorous-robustthrough exhilarated-restless, merry-bright, and humorous-graceful, but quiescence isassociated with the other four groups, from lyrical-soothing to spiritual-serious.For our first experiments, we used phonograph recordings of 5 different musicalcompositions as follows: Debussy, Reflections on the Water, a piano solo played byPaderewski; Mendelssohn, Midsummer Night's Dream, Scherzo, played by the Phil-harmonic Symphony Orchestra of New York under Toscanini; Paganini, Etude in EFlat Major, a Liszt-Busoni arrangement, played as a piano solo by Vladimir Horo-witz; Tschaikowsky, Symphony No. 6 in B minor, the second part of the first move-ment, played by a European orchestra conducted by Albert Gates; Wagner, Lohengrin,Prelude to Act III, played by the Boston Symphony Orchestra. These were all Victorrecords, and the first four were orthophonic recordings.

These 5 records were played to a group of 52 Os, who were directed to checkfrom the printed list all the adjectives which seemed appropriate for describing themusic and to check as few or as many words as they liked. Of course, no hint of thename or nature of the music was given by E, and by inquiry it was found that oneperson in the group could be expected to identify the music exactly, and one personmore indefinitely as "something from Lohengrin," or "a Beethoven Symphony." TheOs were students in the Elementary Laboratory course in psychology; the majoritywere sophomores, and about one-third of them were men. When the votes for eachadjective were tabulated, it was found that in the case of each of the 5 compositionsthere were some Os who could find no adjectives appropriate, or at the most onlyone or two; but most of the Os checked 6 or more adjectives, and found it an easyand agreeable task.The graphs in Fig. 2 show the results of this experiment. In order to plot thediagram for, let us say, the Tschaikowsky Symphony, all the adjectives used fromone group were added, and plotted on a base-line, together with total number of ad-jectives used from all the other groups. In other words, every group of adjectives is

250



7/24

ELEMENTS OF EXPRESSIONIN MUSICtreatedas a whole, as representing ne kind of feeling tone. For convenience nplottingthesegraphs,the groupsare arranged, ot arounda circle as in Fig. 1, but

1 1 1 1 1DEBUSSY

1 1 1 IT S C H AI K OW S K I

m IME N D E L S S 0 H N

P A G A ff I H IAgAI NI

I I 1 IWAGNER12 3 45 6 7 ^GROUPFIG. 2. NUMBEROF OS SELECTINGDJECTIVESROMEACHGROUP

on a straight ine,with the dignified-solemnroupat the extreme eft, and therobust-vigorousgroupat theextreme ight.The readermust be warned hatsuchan arrange-ment does not implythatthe groups actuallyrepresent linear scalewith numerical

251



8/24

HEVNER

units. We have chosen these groupings quite arbitrarily, (1) because it was theopinion of the persons who first studied the adjectives that this was the best arrange.ment to bring out the logical and physiological realtionships between the groups, and(2) because empirical data, collected with these groupings on various kinds ofaesthetic or affective material, poetry, music, and paintings, show a satisfactory'patterning' which very much simplifies the work of presenting and summarizingresults. The graphs in Fig. 2 represent the total number of adjectives used fromeach group, divided by the number of Os making the selection. This process ofdivision is always necessary in order to make these graphs comparable to the graphsfrom later experiments with different groups of Os ranging in number from 20 to100 plus. The graphs are always made in percentages, consequently the results froma group of 40 Os may be compared directly with those from a group of 65, etc.

In studying the graphs, the consistency and uniformity of the interpretation of theOs is immediately apparent. The reader must remember that adjective-groups 1 and8 do not represent the extremes of a scale, but are as closely related as any other twoadjacent groups on the scale, and that the pattern of the Lohengrin music shows asmuch consistency, with its concentration of votes for agitated-impetuous and vigorous-robust, its absence of votes in the central area, and its upthrust on the dignified-majestic group, as does the pattern for the Mendelssohn, where the concentration ismore readily apparent to the eye. The diagrams are valuable, not only for the con-sistency and uniformity which they show, but also for the appropriateness of theinterpretation in consideration of the composer's intention. Mendelssohn's Scherzo,to be played after the first act of Shakespeare's Midsummer Night's Dream, is out-standingly exciting-impetuous and playful-graceful, and there is a total lack of suchaffective states as serenity, sadness, dignity, and solemnity. The Prelude to the thirdact of Lohengrin stands out as impetuous-exciting, robust-vigorous, and it is evenmore interesting to find on the examination of our original tabulations that the heavyvoting for this composition falls on the adjectives exhilarated, triumphant, majestic,and exalting. The Paganini ttude (a composition characterized by its emphaticrhythms and rapid tripping arpeggios, and presented by Horowitz with great bril-liance and precision) lacks entirely the feeling-tone of sadness, dignity, or serenity,and suggests playfulness, gayety, excitement, and vigor. These interpretations areall the more significant because they have been made by listeners distinctly of theuntrained type.The Debussy composition, Reflections on the Water, is of especial interest becauseof the apparent uncertainty of the listeners, and the discrepancies which appear in thedata. The adjectives used are scattered among the 8 groups with the greatest numbersfrom two groups in diametrical opposition to each other, viz. dreamy-sentimental andexciting-impetuous. There are two reasons which might account for these differencesin interpretations. The first is the nature of the subject with which the composer isdealing. Reflections on the water are likely to be at one time clear, still, and distinct,and at another time distorted, restless, fitful, and changing, and the composer mayhave intended several different interpretations to be possible from his music. Fromthis point of view the votes of our listeners may stamp this composition as a par-ticularly successful bit of creative work. The second reason, which is important formany other compositions as well as this particular one of Debussy's, is the presenceof several well-defined sections in the musical structure, sometimes quite differentfrom each other. One of the commonest patterns for musical composition consists in a

252



9/24

ELEMENTS OF EXPRESSION IN MUSICdivision of the material into three parts, with the first and third similar or evenidentical, and the second a contrasting section, only remotely related to the precedingand succeeding parts.In order to separate the various parts of the composition and trace the source ofthe votes for the different adjective groups, the experiments described above wererepeated with a new group of Os, and with slightly different instructions. The new Osbegan to check their adjectives as soon as the suggestiveness of the music became ap-parent, by placing the numeral '1' before each choice. At a given signal from Ewhich coincided with the beginning of a new section of the music, the Os indicatedtheir choices by the numeral '2'; and so on for a third (a fourth or fifth) section, theychanged to numeral '3,' etc. Tabulations were made for the three different sections ofthe piece, and the graphs which show these results are given in Fig. 3. The reactionsoccasioned by the first part of the composition appear in diagonals, for the secondpart in solid black, and for the third part in white areas. The separations for theReflections on the Water were made at bars 22 and 75 and the sources of thedifferent affective patterns are very clearly shown. The first and third sections arequite similar in character: dreamy-sentimental, serene, calm and slightly graceful, andplayful. In sharp contrast, the middle section is very playful, rather merry, andextremely restless, exciting, and agitated.With the two experiments carried out under two sets of directions, it would notbe logical to expect that combining all the adjectives numbered 1, 2, and 3 in thesecond experiment would give the same pattern when plotted as the graph for thesame music from the first experiment (Fig. 2). An inspection of Fig. 3 shows thatif such a procedure were followed, there would be a great preponderance of votesfor certain adjective groups, especially Groups 3 and 4, because the recurrence of thefirst section at the end of the composition makes it necessary for the listeners to votefor those adjectives twice. A comparison of the graphs in Figs. 2 and 3 does reveal,however, the fundamental agreement in the interpretations of the two groups, whilethe differences due to the slightly different experimental conditions are distinctlybrought out. It is quite likely that the Os maintained a much higher degree ofattention to the music in the second experiment than in the first.In Fig. 3, results are given also for several other compositions. The TschaikowskySymphony is of especial interest, for the excitement and agitation are shown tooriginate in the second section which is higher, faster, and thinner than the first,and from the third section, which is a repetition of the first with much more volumeand with a more elaborate orchestration, including a prominent glissando effect in theviolins.

Two of Beethoven's Sonatas were also used in this part of the study, the Seventh,in A major, third movement, first half, and the Fifth, in C minor, first movement,first half,' and the results are shown in Fig. 3. In both of these compositions thefirst and third parts are alike and the second part is different in tempo, rhythm,theme, and orchestration. The graphs of the mood-effects bring out these factors veryclearly, especially in the excerpt from the Fifth Symphony, where the first and thirdsections were identical in every way. The more complicated graph for the Mendels-

7 The music is played by Felix Weingartner and symphony orchestra, Columbiarecord. In the Fifth Symphony the music is divided at the 71st and 125th bars;in the Seventh at the 145th and 237th.

253



10/24

sohn Scherzo is not included, although its five different parts showed equally inter-esting and consistent results.

r

DEBUSSY-.2 10- t- 1 -"- -.

I I

B E B U NS Y-wf -,v e,, ,

aU~ , / ,"_ o 1

,1 2 3 4 _ 5 6 ? aOROUP

FIG. 3. NUMBER OF OS SELECTINGADJECTIVESFROM EACH GROUPSince the adjective-group method was shown to be successful in handling the datafrom these longer compositions, the same process has been applied also to the study

of the affective character of certain elements of musical structure. The method ofisolating the element to be studied follows the method described in an earlier experi-ment, on the affective qualities of the major and minor modes.8 Two versions of thesame composition were presented to two different audiences who were asked to ex-press their attitude toward the music by checking adjectives. The music in the twoversions is arranged in such a way that the difference between them lies only in theone element to be studied, and the listeners indicate their interpretation of the music

8For a complete discussion of the advantages and disadvantages of composingexperimental music in this way cf. the author's previously cited article. The musicmust be carefully chosen, and rewritten by the most expert musicians. For invalua-ble assistance in this work the author is indebted to Miss Hariett Johnson, pupilof Rubin Goldmark, Lecturer for The Layman's Music, Inc., New York; and toMr. John Verall of the Music Department, Hamline University, St. Paul.

B E E T H V Ef~~~~~~~~~~~o

.olo 11~~~~~~~~~~~~~~c~~~I "Ln~~~~~~~~~~0

GROUP~~FIG 3MBR S SLETIG DJCTVESFRm Ac Gou

Mr. John Verall of the Music Departmtnt, Hamline University, St. Paul.

254 HEVNER



11/24

ELEMENTS OF EXPRESSIONIN MUSIC 255MENDELSSOHN SONG WITHOUT WORDS

(Descending melody)FLOWING MOTION, (ORIGINAL),I II I11 ,Te I IL' i I I

1r ,i d

_r- a e1 11

-~ ~

tfILI r-I i; I I I r

rt r- I I I - X II I l | l

.il ? '' . - e I. -

I



12/24

256 HEVNERFIRMRHYTHM

T ;Il- I I Irn111470 ro Ba

-ft__ ^ I i r

r- j j-? ?*


13/24

257LEMENTS OF EXPRESSIONIN MUSICMENDELSSOHNONGWITHOUTWORDS

(Ascending melody)

~~~~~~~I~~~~~~~~~~~~~rr~~~~~~~~~~~~

~~~1 1 ~~~~~~~~~~~~a--J, a

11 1 I

Jt lI I -I I III X- , X .J- X



14/24

HEVNERBEETHOVEN, P. 10. SIMPLEHARMONY, ORIGINAL)

IIv

- -zh-i - ' g~- -S- '-5 ;-'

.V0 o-


15/24

ELEMENTS OF EXPRESSION IN MUSIC

COMPLEX HARMONY

-* l+I1 1 1'I~ -II 1 I_,_ Ss t_s~ -- -k . I I

It- bbbb- j a- 1: |-I L-a- -a- b-5e- -z'- -Z')

F b z -s ii i? ebe bb r - f -- S f3^b~g r i 1 h-I -?- e-l

had to be made between the demands of the harmony and the exactness of theinversion of the melody.(3) It was sometimes difficult to keep the relation of the melody to the tonicor keynote exactly parallel throughout the length of the two versions. The keynote isthe point of reference for all parts of the melody and returns to it are importantbecause they give a feeling of satisfaction and finality which helps to establish theform of the composition and give it balance and unity. The parallelism was alwaysstrictly maintained, however, at the ends of the phrases, since they mark off themelody into definite parts and determine the kind of symmetry it shall have. It isunderstood then that the inversions of the melodies were never perfect, but thatthey were as exact as these other considerations could allow.'In the third group of 9 compositions the modifications were made in the harmonicstructure of the music, changing it from simple, consonant harmonies, resolved in a

very smooth and satisfying manner, to complex, dissonant harmonies, more harshand rough, and not always reaching a satisfactory resolution. The versions whichwe call complex have augmented and diminished intervals and modern effects. Itwill be surmised from the discussion of the difficulties encountered in making melodic9The difficulties encountered in inverting melodies were so great that this part ofthe experiment was entirely revised and repeated, with new music and new Os.The music for the second experiment was much more satisfactory to the musiccritics, but the results from the two series are very little different, and thereforeonly the second experiment is reported here.

259



16/24

TABLE ISHOWING THE VOTES FOR EACH ADJECTIVE GROUP, FOR 6 COMPOSITIONS IN FIRM RHYTHM

Adj., Beethoven, Schumann,Adj.? Beethoven,Group Op. I3 N.No. 5I II 79I4 42 30 32II 73 5 o13 124 2 46 335 5 2i6 0oo6 ii 326 867 143 68

91 328 68 II629 8

Mendelssohn,Op. I9No. iI2682010

375773641432

1I44

I384

Schubert,Sonata,No. I202769o

I733725i42424568II12226

Prokofieff,Op. I23I7

1026

1539213594465028I327

412

FirmFlowingFirmFlowingFirmFlowingFirmFlowingFirmFlowingFirmFlowingFirmFlowingFirmFlowing



17/24

TABLE IISHOWING THE VOTES FOR EACH ADJECTIVE GROUP, FOR 9 COMPOSITIONS IN SIMPLE

Adj.. TirJohnson, ach, luck, Mendels- Beethoven,Adj.- Johnson, Bach, Gluck , Vigorous Veral sonaGroup Gigue Chorale Andante y eChorale o Sonata& Lively Op. 64 Op. xoI o 75 9 2 149 4 14Io 59 34 3 82 39 832 o 88 9 o x35 13 29o 70 24 I I05 45 743 0 49 84 o 6 143 31I 4 83 o 19 99 274 6 41 139 6 8 114 272 o 96 I 7 71 II5 i26 3 I22 9191 o 48 326 I66 0 37 9675 0 9 327 39 o x 88Ix4 14 4 1948 8 8 o 21

28 32 2 55

00002

102326

19 29 22 o0 01 48 42o 245 37

SimpleComplexSimpleComplexSimpleComplexSimpleComplexSimpleComplexSimpleComplexSimpleComplexSimpleComplex



18/24

TABLE IIISHOWING THE VOTES FOREACH ADJECTIVE GROUP, FOR9 COMPOSITIONSIN ASCENDING A

Beet' i Grieg, Verall, Schumann, Mendels- rlAdj, GCluck, Verall,d hoven, u Wedding Prel Op. 155 sohn,Group o 3 Cprice Day ude No. 5 Op. 9 tuAscending I 5 56 27 15 31 5 5Descending I 8 I 15 2 9 I5Ascending 2 2 10 25 28 5 6 7Descending 10 7 4 43 2 5 I7Ascending 3 o 22 19 34 o 40 26Descending 2 32 30 24 o 44 52Ascending 4 x 28 14 20 o 66 17Descending 1 40 17 20 o 77 63Ascending 5 o 53 14 10 o o 7 31Descending 0 20 9 42 0 34 65Ascending 6 2 I 8 6 5 8Descending 4 4 6 9 9 14 I5Ascending 7 65 i 10 o 78 0 2Descending 178 o o 1 42 2 IAscending 8 38 3 12 I 130 1 oDescending 115 o 3 2 6i 2 I



19/24

TABLE IVSHOWING THE VOTES FOR EACH ADJECTIVE GROUP, FOR IO COMPOSITIONS IN MAJOR

Schu, Beet.Adj. mann, hoBeet Bach, Martini, Bach, Rameau, Durand, Gluck,Group Folk Mroup Folng Minuet Minuet Gavotte Musette RigaudonChaconne AndanteSongMajor I 20 20 14 8 13 10 I 95Minor 43 74 20 28 46 21 4 108Major 2 7 2o 4 o 7 7 o 38Minor 67 95 17 30 III 56 3 122Major 3 33 68 Io 13 43 i6 4 75Minor 107 140 32 68 III 42 9 125Major 4 53 99 49 45 82 33 30 131Minor 52 124 6o0 4 7 29 22 94Major 5 152 104 225 210 139 xo8 287 47Minor 90 87 97 107 83 87 134 25Major 6 174 69 215 244 93 128 335 IMinor 64 21 67 93 25 6i 155 4Major 7 42 27 I02 59 17 47 84 i6Minor 43 24 66 22 35 95 96 32Major 8 i6 6 32 31 5 34 30 15Minor 20 1 23 12 9 41 22 23



20/24

HEVNERversions that the new harmonies sometimes wrought changes in the melodic line, sothat it became a slightly different melody, even though the sequence of tones re-mained identically the same. To meet this argument, we can but repeat that wewere aware of such dangers, and have guarded against them as far as possible in therearrangements of our music. In order that the reader may have the opportunity ofjudging for himself in these matters we quote one typical example of our musicalmaterial, eight measures from Beethoven, Op. 10.Table I shows the number of times each adjective group was selected for boththe firm and flowing versions of the 6 different compositions. The total number ofvotes for all 6 compositions is also shown, and the ratio of the difference to itsprobable error for each group.l0 These ratios, shown in the last column of the table,are the most important figures for estimating the difference in the affective value orexpressiveness of the two different rhythms, and therefore are expressed graphicallyin Fig. 4. Tables II and III present similar data for the rising and falling of themelodic line and for the simplicity and complexity of the harmony. Table IV presentsdata from an earlier experiment with the major and minor modes.In Fig. 4, the significant data from all four tables are presented in graphic form.The octagonal figures represent the 8 different adjective groups designated here by oneor two characteristicadjectives from each group. (The reader may refer to Fig. 1 forthe full list of adjectives in each group.) There is a separate figure for each of thevariables studied, modality, rhythm, harmony, and melody. The ratios of the differ-ences to their probable errors (D/PED) are represented for each group in anappropriate symbol, so that the reader may see at a glance the importance of eachelement in differentiating the mood effects of the music.

It is apparent that the use of the major or minor mode is of the mostclear-cut significance in the expression of four different mood effects, thatthe major mode is very strongly associated with happiness, gayety, play-fulness and sprightliness, and the minor with sadness, and with senti-mental yearning, tender effects. The expressiveness of modality, eithermajor or minor, is more stable and more generally understood than that ofany other element which we have studied. It is of equal interest to note alsothat the modality has no discoverable effect on the expressiveness of someof the affective states, on dignity, vigor, excitement, or calm. Those statescan be produced equally well by means of either mode.The firm or flowing motion in the rhythm divides the circle of adjectivesexactly in two, and has its most important effects on a different axis ofthe circle, on the dignified and vigorous vs. the happy, playful, sprightlygroups. These differences fall where musical tradition and custom would

'0 Calculated by Yule's formula and the Edgerton-Paterson tables, J. AppliedPsychol., 10, 1926, 378-391. The total number of adjectives used is taken as apercentage of the total number possible, i.e. the number of Os multiplied by thenumber of adjectives in the group, usually a total of between 3000 and 4000. Theratios would all be much larger save for this very conservative way of estimatingthe percentages, for use by every O of every adjective is not, after all, a real possi-bility.

264



21/24

ELEMENTS OF EXPRESSION IN MUSIClead us to expectthemand on the whole seem quitelogical and acceptable.The differencesbroughtabout by simple vs. complex harmony,or wemight say by consonancevs. dissonanceor cassic vs. modernharmony,arefor the most part effectiveon the 5 adjectivegroups on the upper half

Ascendingelody,Descendinqelody/ Harmonqimple-,Harmonqomplex^

-97 v 7 w- VSad Sad

MajorMode-,MinorMode^ Rhythmirm-,Rhythmlowing^FIG. 4. SHOWING THE RELATIVE IMPORTANCE OF MODE, RHYTHM, HARMONY, ANDMELODY ON THE AFFECTIVEVALUE OF MUSICof the circle.The simpleconsonantharmoniesarehappy,graceful,serene,and lyrical.The dissonantmodern harmoniesareexciting, agitating,vigor-ous, and somewhat sad. The differencesare substantial,but not as clear-cut as those for the majorandminormodes.In comparisonwith the other three figures,the circle representing heeffectsof the rising and falling of the melodic line is particularlynote-worthy for its lack of significantdifferences.The largest difference,in-dicatingthat the descendingmelodies are more exciting and impetuous,

265



22/24

is only 6 times its probableerror. The importanceof this finding is mini-mized by the difference ikewise in favor of the descendingversions,forthe adjectivegroupcalm, lyrical,serene,etc. The descendingversionsgiveonly one significantdifference,on the dignified-serious roup.The first inferenceto be drawnfrom a studyof these results s the rela-tive importanceof these four elements in expressingthe affective one ofmusic.Apparently he melody,at least the risingor falling of the melody,is of much less importancehan the harmony,or rhythmof the music,andthe modalityis most importantof all. It must be remembered hat thefigures indicate the expressivenessof these elementsonly in relation toone another.Compositionswrittenin the majormode are not always gayandplayfulandnothingelse.Theyareonlymoregayand moreplayfulthanthe samecompositionsplayedin the minormode. Likewise,a compositionwith a firm rhythmwill be much more serious, dignified and vigorousthan the same compositionwith a flowing rhythm.But a compositioninwhich the harmonyand rhythmremain identicaland only the melody ischanged by inverting it, apparentlydoes not suffer a change in its ex-pressivenesswhich distinguishesit in any constantand predictablewayfrom the original.To some extent thesedifferencesn the definitenessof the expressivenessof these elementsare the function of the complexityof the element. Achangefrom majorto minor is verysimple, definiteand limited,but thereis an infinitevarietyof ways in which a melodic line mayascend-a longslow ascent, or a successionof short rises, reaching a high point veryquickly,a slowly rising patternof scale-likefigures or a series of largeskips with repeatedclimaxes.There is also a varietyof ways in whichrisingand falling of the melodymaybe combinedso as to give prominenceto either one quality or the other. In our studies we have chosen 9compositionswhich displaya varietyof melodic patterns,and with thesecompositionswe do not find consistencyin the results. More elaboratestudiesin which certainaspectsof rise and fall in melodiesare isolatedinindependenceof each other, or studies of some other qualities of themelody, e.g. its circularor pendularmotion, may demonstratea muchgreater mportanceor the expressiveness f melody.

We believe that refinements in the method of these experiments, especially in therewriting of the music, will yield results even more striking and clear-cut than thosewhich we are quoting here. Although every possible precaution is taken to eliminatefrom the experimental material every composition whose two versions differ in morethan one respect, it is difficult to anticipate every possible deviation which might affectthe results. The two compositions must be matched for tempo, rhythm, harmony, andmelody, and the musical beauties of both versions must not be sacrificed in any way.

266 HEVNER



23/24

ELEMENTS OF EXPRESSION IN MUSICThe resultsfor our studiesof rhythmwould give muchmore clear-cutdifferentiationwere it not for one error which occurredn the use of the ProkofieffGavotte.Oneverysecondand fourth beat in the firmversionthere s a grace-notentroducinghechord.This grace-notewas not included in the flowing version,and its presenceundoubtedlyauses he increasen playfulnessandgayetywhichthis one compositionaloneshows for the firmversion(see TableI). A furtheranalysisof the studiesinharmonywould also lead to some furtherdetailsof expressiveness.A comparison four complexversions n which the chordswere diminishedand those in which theywere augmentedeads us to the conclusion hat the augmenteddissonance, .g. thenervousmoderndiscordsn the modern movie'style,makesthe musicmoreexcitingand impetuous,and that diminishedand minorseventh dissonancemakesthe musicmore depressedand yearning.A more detailed study of these two factorsmightestablishthese differencesquite exactly.There are certainlyadditional elements of expressivenesswhich we have notinvestigatedhere, and there are other aspectsof these four elementswhich are nodoubtof great importancen the expressiveness f music. We have left entirelyun-touchedthe effect of tempo,for example,an element which could be investigatedbythismethodwith thegreatest ase. We have studiedonlyoneaspectof rhythm, ndleft unexplored wo aspectswhich would probablyprove to be of importance, heduple vs. the triple rhythm,and the smooth vs. the jerkyrhythms.In our studiesof harmonywe have still to explore the effectof differentkindsof resolutionsanddifferentstyles of consonanceand dissonance,augmentedand diminished,etc. Theinterrelationsbetween these harmonicstyles, augmenteddissonace in major vs.augmenteddissonance n minor,may also be explored.The relativeimportance f the variouselements,tempo,rhythm,harmony,etc.,would best be studiedby an extensionof our method, n which eight or morever-sions of the samecompositionwould be written,embodying ll the possiblepermuta-tions of several variables.A half-dozencompositions,very carefullyselected andtreated n this way,wouldnot completely olve all of the problemsof expressivenessof music,but would certainlyextendour knowledgeof the field beyondits presentrestrictedimits. SUMMARY

(1) In these experiments, we have assumed that there is some systematicsymbolism underlying the expressiveness of music, and we have been ableto show that the meaning of both short and long compositions, (shortthemes of 8 to 12 measures and complete musical compositions such asopera preludes and symphonic movements) is generally understood byboth trained and untrained listeners. There is striking uniformity andconsistency in the interpretations.

(2) The Os report their reactions by checking adjectives from a printedlist, and the process of combining the adjectives into homogeneous groupsfor statistical treatment has greatly simplified and clarified the results, andhas emphasized the consistency and the systematic character of musicalmeanings.

(3) We have tried to isolate four of the elements of musical form in

267



24/24

order to discover the meaning or character hat is associatedwith each.The method is that of preparingtwo versionsof a musicalcompositionwhich differin one respectonly, for example,the rhythm.In this way, theelement to be investigated s left in its propersetting, and the stimuluswhich the 0 hears is real music, not isolated chords, or unsupportedmelodies.(4) The elementsstudiedand their associatedmeaningsas determined

by thismethodareas follows: (a) The majormodeis happy,merry,grace-ful, and playful,theminor mode is sad, dreamy,and sentimental,and suchqualitiesas excitement,vigor, dignity, serenity,etc. arenot determinedbyeither mode. (b) Firm rhythms are vigorous and dignified; flowingrhythmsarehappy,graceful,dreamy,and tender,and neither s particularlyuseful in determiningsuch characteristics s excitement,satisfaction,andserenity. (c) Complex,dissonantharmoniesare exciting, agitating, vigor-ous, and inclined towardsadness; simpleconsonantharmoniesare happy,graceful,sereneandlyrical.(d) Differences n expressivenessausedby therisingor falling of the melodicline are not clear-cut,distinct,or consistent.Therearetendencies owardthe expressionof both exhilarationand seren-ity by the descendingmelodies, and towarddignity and solemnityby theascending.

268 HEVNER

experimental studies of the elements of expression in music

Documents