Nonmetric multidimensional unfolding as a tool to identify latent perceptual variables: Measuring contemporary bards singing

Transcription

1 Nonmetric multidimensional unfolding as a tool to identify latent perceptual variables: Measuring contemporary bards singing Victor S. Kamensky Albert Einstein College of Medicine, Bronx, New York, U.S.A. kamensky@aecom.yu.edu Vladimir M. Petrov State Institute for Art Studies, 5 Kozitsky per., Moscow , Russia Tel./Fax: +7 [495] vladmpetrov@yandex.ru Abstract The perceptual parameters of works of art are usually hidden, and they cannot be revealed by means of direct questions. Only integral estimations are capable of providing certain primary data adequate to real perception. The best way to process such primary data, is nonmetric multidimensional unfolding. Appropriate procedure was used in an investigation involving 222 participants (visitors of clubs of amateur song ) which were asked to put integral estimations to the creativity of 14 famous bards (singers). Computer processing these scores resulted in two-dimensional configurations permitting to measure the strength of 7 principal parameters of singing. The measurements provided estimations of left- or righthemispheric orientation of bards songs, as well as relative importance of text and music. Keywords: Scaling, music, latent variables, perception, textual parameters, hemisphericity, correlation 1. Introduction Among numerous problems arising when measuring the parameters of works of art, one seems to be rather substantial when dealing with works which belong to the borderline between two kinds of art. It is the problem of comparative importance of features of these two kinds. What are the weights of these features? So works of art which are characterized by dual nature (i.e., combining inhomogeneous constituents), became very attractive for investigators. Recently D.K. Simonton [1], on the basis of the statistical analysis of social success of 911 operas (created by 59 eminent composers), received evidences in favor of rather strong prevalence of their musical components over textual (literary) ones. However, he didn t measure immediate reactions of recipients when estimating these works of art. Meanwhile, the relationships between two kinds of constituents in works of dual nature, are now especially interesting in the light of the concept of hemispheric specialization which was initiated by Maslov [2] and afterwards elaborated by several other investigators [3, 4]: each work of art can be oriented either mainly on right-hemispheric perception or lefthemispheric one. (When combining literary and musical constituents, it means prevalence either of left- or right-hemispheric processes, respectively.) Moreover, as it was shown in the framework of the information approach [5], each kind of art (and sometimes each genre) responds to its own way to train recipients mental life. That is why it was decided to realize measurements of recipients attitudes to one class of such works. 2. The problem Creativity of contemporary minstrels (bards, liedermachern, etc.) is undoubtedly of such dual nature. Textual components of their works deal mainly with analytic processes, 4-254

2 i.e., left-hemispheric ones. But exactly left-hemispheric activity is tightly connected with social factors. So in the case of such dual works, we may expect the prevalence of textual components, because of strong social role of such songs (at least in Russian sociopsychological life). That is why it seems reasonable to clear up the relationships between different constituents when real perceiving such songs. To do this, it is desirable to build the hierarchy of the features (parameters) of songs, according to their importance for recipients. In turn, to identify these parameters, it seems senseless to resort to the help of direct questions concerning the features to be studied. First of all, usually recipients know nothing about the nature of their preferences in this delicate field. In addition, sometimes the recipients may show rather distorted opinions because of various prestige factors. So, it is desirable to base appropriate empirical investigations on recipients integral preferences, followed by researchers analysis of these estimations. Exactly such methodology is realized when using the procedure of the so-called nonmetric multidimensional unfolding to integral estimations of works of art [6]. Earlier this procedure was applied to identify the perceptual parameters of painting [7], rhymes in poetry [8], style of prose [9], and other features of art. 3. Experimental procedure A sample of 14 famous contemporary bards was compiled: Yu. Vizbor, V.Vyssotsky, A.Galich, A.Gorodnitsky, V.Dolina, A.Dolsky, Yu.Kim, Yu.Lores, V.Luferov, N.Matveyeva, A.Mirzoyan, B.Okudzhava, A.Rozenbaum, and A.Sukhanov. This sample was supposed to represent different types of such songs. 222 recipients were examined in clubs of amateur songs in four cities. Each recipient was given a questionnaire containing names of the above 14 singers to be evaluated according to the recipient s integral preferences (using 10-point scale). The data relating to each club, were processed separately, by means of nonmetric multidimensional unfolding [6, 10]. Twodimensional configurations were obtained both for singers and recipients. Beside, a group of 12 experts estimated singers, over seven hypothetical parameters. These hypothetical parameters were divided into two groups. The first group dealt with integral estimations of singers creativity: a) singer s popularity measured as the share of recipients capable of evaluating his/her songs; b) integral evaluation of his/her songs (estimations averaged over all recipients). The second group concerned semantic and artistic aspects of songs: c) quality of literary texts; d) their social significance; e) humor inherent to their texts; f) quality of performance; g) musical quality of songs. The results of experts rankings are presented in Appendix 1. It is easily seen that the singers of the sample do really reveal very high diversity of their features. As for statistical links between these features, they are presented in Appendix 2. These links vary in a broad range, evidencing of rather good raw material presented by the sample. 4. Results of measurements The preliminary task of the data processing was to establish possible influence of recipient s socio-demographic parameters on their preferences. However, no grouping of recipients was observed responding to their gender, age, or education. Then the main task was to identify the principal parameters of perception using twodimensional configurations obtained. Figure 1 presents one of such configurations, where the axis for one of the parameters (d) is shown, together with projections of all 14 bards onto this 4-255

3 axis. In most cases orderings of such projections onto axes built for each hypothetical parameter, showed rather high correlation with expert rankings (except parameter g musical quality): maximal values of Spearman coefficient of rank correlation ( ) vary in the range from.552 (quality of performance) to.882 (singer s popularity). The hierarchy of these maximal coefficients (for five hypothetical perceptual parameters of the second group) is presented in Appendix 3. It evidences of strong prevalence of textual features over musical ones. Fig. 1. Typical two-dimensional configuration of 14 singers, built on the basis of estimations of 50 recipients in one of Moscow clubs of amateur song. (Coefficient characterizing the quality of the configuration, θ =.214). The axis of Social significance is shown, together with projections, onto this axis, of objects (singers). This hierarchy possesses also high correlation with the proposed hierarchy of songs parameters according to their hemispheric attribution (see the last column of Appendix 3): Spearman coefficient of rank correlation is.975. As well expert estimations (correlation between the above seven parameters Appendix 2) showed rather strong positive correlation between all parameters except the last one (g musical quality of songs), which possessed zero or negative correlation with all other parameters. 5. Conclusion First, the given kind of singing relates to the perceptual pole opposite to opera; the recipients preferences are determined mainly by textual parameters of songs, these parameters being mainly social. Second, the prevalence mentioned is realized by means of appropriate primarily analytic, left-hemispheric orientation of songs. It is possible to establish the place of the genre considered, both in the entire spectrum of musical genres and the spectrum of informational means and devices used by the entire system of contemporary culture. It is important to clear up the mechanisms of enhancement which are functioning when perceiving such works, in the light of the concept of the vertical dimension inherent in the system of culture [11-13]. This phenomenon is also tightly connected with the left-hemispheric prevalence. Probably, the musical accompaniment purifies the perception of the text, eliminating various low everyday associations which could be generated by literary text

4 So the results of measurements make rather substantial contribution into understanding of real functioning of works of art of dual nature. As well, we see the constructive character of the method of nonmetric multidimensionsl unfolding when measuring perceptual parameters of works of art. References 1. D.K. Simonton. The music or the words? Or how important is the libretto for an opera s aesthetic success? Empirical Studies of the Art. 2000, 18, 2, pp S.Yu. Maslov. Assimmetriya poznavatelnykh mekhanizmov i yeyo sledstviya (Asymmetry of cognitive mechanisms and its consequences). Semiotika i Informatika. 1983, 20, pp (in Russian). 3. G.A. Golitsyn. Informatsiya i tvorchestvo: Na puti k integral noy kul ture (Information and creation: On the path to the integral culture). Moscow: Russky Mir (in Russian). 4. V.M. Petrov, & L.G. Boyadzhiyeva. Perspektivy razvitiya iskusstva: Metody prognozirovaniya (Perspectives of art development: Methods of forecasting). Moscow: Russky Mir (in Russian). 5. G.A. Golitsyn, & V.M. Petrov. Information and creation: Integrating the two cultures. Basel; Boston; Berlin: Birkhauser Verlag V.M. Petrov. Kolichestvennyje metody v iskusstvoznanii: Uchebnoye possobiye dlya studentov. (Quantitative methods in art studies: Handbook for students). Moscow: Academic Project (in Russian). 7. V.M. Petrov, V.S. Gribkov, & V.S. Kamensky. Poverit garmoniyu eksperimentom (To prove harmony by an experiment). In: Chislo i mysl, issue No. 3. Moscow: Znaniye. 1980, pp (in Russian). 8. O. Kamensky, L. Mazhul, & V. Petrov. The phenomenon of rhyme: Information-based deductive model and its empirical verification. In: Culture and communication: Proceedings of the XIX Congress of the International Association of Empirical Aesthetics. Ed. by H.Gottesdiener & J.-C.Vilatte. Avignon: Universite d Avignon et des Pays de Vaucluse. 2006, pp V.M. Petrov, V.S. Kamensky, & S.N. Shepeleva. Prozrachnost stilya prozy: Opyt eksperimental nogo issledovaniya vospriyatiya (Transparency of the style of prose: A trial of experimental investigation of perception). In: Problemy strukturnoy lingvistiki Ed. by V.P. Grigoriev. Moscow: Nauka. 1978, pp (in Russian). 10. O.N. Danilova. Muzykal ny mir lichnosti (Musical world of personality). Moscow: Russian Institute for Cultural Studies (in Russian). 11. G.A. Golitsyn. High art and low art: The systemic role of an elite subculture. Journal of Russian and East European Psychology. 2000, 38, 3, pp V.M. Petrov. High against Low : Hierarchies in Nature, Culture, and Art (Information approach). Bulletin of Psychology and the Arts. 2002, 3, 1, pp G.A. Golitsyn, & V.M. Petrov. Sotsial naya i kulturnaya dinamika: Dolgovremennyje tendentsii informatsionny podkhod (Social and cultural dynamics: Long-range tendencies information approach). Moscow: KomKniga (in Russian)

5 Appendix 1. Singers and their ranks over seven hypothetical parameters (expert estimations). Singer a b c d e f g Yu.Vizbor V.Vyssotsky A.Galich A.Gorodnitsky V.Dolina A.Dol sky Yu.Kim Yu.Lores V.Luferov N.Matveyeva A.Mirzoyan B.Okudzhava A.Rozenbaum A.Sukhanov Appendix 2. Spearman coefficients of rank correlation between the above seven hypothetical parameters. Parameter a b c d e f g a) Popularity *.551*.538*.513* b) Integral **.811**.700**.459*.168 evaluation c) Quality of literary texts **.739**.846**.035 d) Social **.692**.207 significance e) Humor **.324 f) Quality of performance g) Musical quality of songs Designations: * 5%-level of statistical significance; ** 1%-level of statistical significance. Appendix 3. Perception of singers: hierarchies of five hypothetical parameters over their maximal correlation with expert rankings ( ), and over their supposed importance for left-hemisphericity (L). Parameter Maximal Rank on Rank on L c) Quality of literary texts d) Social significance e) Humor f) Quality of performance g) Musical quality of songs