similarity relations in listening to music

30
http://msx.sagepub.com/ Musicae Scientiae http://msx.sagepub.com/content/11/1_suppl/9 The online version of this article can be found at: DOI: 10.1177/1029864907011001021 2007 11: 9 Musicae Scientiae Irène Deliège Similarity relations in listening to music: How do they come into play? Published by: http://www.sagepublications.com On behalf of: European Society for the Cognitive Sciences of Music can be found at: Musicae Scientiae Additional services and information for http://msx.sagepub.com/cgi/alerts Email Alerts: http://msx.sagepub.com/subscriptions Subscriptions: http://www.sagepub.com/journalsReprints.nav Reprints: http://www.sagepub.com/journalsPermissions.nav Permissions: http://msx.sagepub.com/content/11/1_suppl/9.refs.html Citations: What is This? - Mar 1, 2007 Version of Record >> at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014 msx.sagepub.com Downloaded from at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014 msx.sagepub.com Downloaded from

Upload: mumbaba

Post on 17-Jul-2016

21 views

Category:

Documents


5 download

DESCRIPTION

Article about musical cognition

TRANSCRIPT

http://msx.sagepub.com/Musicae Scientiae

http://msx.sagepub.com/content/11/1_suppl/9The online version of this article can be found at:

 DOI: 10.1177/1029864907011001021

2007 11: 9Musicae ScientiaeIrène Deliège

Similarity relations in listening to music: How do they come into play?  

Published by:

http://www.sagepublications.com

On behalf of: 

  European Society for the Cognitive Sciences of Music

can be found at:Musicae ScientiaeAdditional services and information for    

  http://msx.sagepub.com/cgi/alertsEmail Alerts:

 

http://msx.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

http://msx.sagepub.com/content/11/1_suppl/9.refs.htmlCitations:  

What is This? 

- Mar 1, 2007Version of Record >>

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

Similarity relations in listening to music:How do they come into play?

IRÈNE DELIÈGEUniversity of Liège

• ABSTRACT

In this essay, similarity relations in the perception of music are studied on the basis

of two different standpoints. The first is said to be external and studies comparisons

between distinct and autonomous musical entities, i.e. different works or different

interpretations. The second is internal and looks at similarity relations, which the

listener identifies in the same work or part of a work, and in the same

interpretation. This last procedure is developed in different directions. We show the

importance of the similarity factor: (i) in the actual composition process as the

composer seeks unity and coherence in his or her piece; (ii) in folk music; (iii) as an

essential element for the listener who strives to understand the piece as he listens

to it in real time. This last point is of fundamental importance for the study of

cognition in general i.e. the implicit or explicit role of similarity in perception

processes.

Concrete examples of the main theoretical points described here can be identified

in the course of the different phases of the cue extraction model that I developed

recently, i.e. segmentation, categorization, schematization, imprint formation

(Deliège, 1989, 1991). The SIMILARITY/DIFFERENCE axis is a central element in

the structure of the model. The empirical approaches used in experimenting with

the different phases of the model are analyzed in this two-fold perspective. From

the very beginning, this was the aim of a series of procedures that aimed at

showing the implicit or explicit aspects of the processes that musician and non-

musician participants use during the experimental listening process. An overall view

of the results obtained confirms that the assumptions were correct.

INTRODUCTION

Whatever the perception mode, the concept of similarity refers initially to a processthat compares two or several things, states, events, etc. This process is required toassess “resemblance”; in other words, it must reveal those characteristics by which theobjects that are being compared differ, whatever the degree of difference, or it must

9

© 2007 by ESCOM European Societyfor the Cognitive Sciences of Music

Musicae ScientiaeDiscussion Forum 4A, 2007, 9-37

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

show that there is pure and simple “identity” or a state of maximum similarity (Hahn& Chater, 1998).

In his Vocabulaire de la psychologie (1951), Henri Piéron, the editor, defined theconcept of similarity as being the “psychological assertion of a resemblance that canbe assessed, either objectively, and is thus a property appertaining to a physicaldimension, or subjectively and is an attribute of pairs of stimuli”. This definitiontherefore specifies that, over and above the fact of relating the entities concerned,there is an explicit aspect — the objective estimate, in full awareness, of physicaldimensions that can be observed — or a subjective aspect, due to the perceivedimpression of similarity and thus possibly felt unconsciously. The analysis of thisassumption in the field of musical perception, leads to some fundamental questionsconcerning the cognitive organization of the listening process of a work, as we willsee later on in this article. We must, however, add a distinction to clarify whether weare dealing with relationships that I would call “external”, i.e. that establishcomparisons between distinct and autonomous entities; or whether we areconsidering “internal” relations, i.e. that deal with similarities within one same workor a part of a work.

1. EXTERNAL SIMILARITY RELATIONS

Pierre Froidebise and André Souris wrote in a joint article (1961, pp. 840-41) that amusic lover or a critic might explain at length that he or she prefers Furtwängler’sinterpretation of Beethoven’s Fifth Symphony rather than Toscanini’s. You often hearcomparisons of this kind when people are discussing a performance. It is on the basisof the same principle that instrumentalists are made to include a set piece in theirprogram for a competition. This is a piece that all the competitors must play; thusthe differences in their outlook will stand out and will facilitate the comparisonbetween them.

There are other occasions where a comparison process takes place automaticallywhile listening to music. This is particularly the case for forms that are based on aTheme and Variations structure, where a basic statement is repeated a certain numberof times, but always worked out in different ways. The theme recurs, modified bydifferent compositional procedures such as transposition, new timbres, differentrhythms, harmonies and keys etc.; each variation has its own character. I will justmention a few well known examples: Bach’s Goldberg Variations for keyboard,Beethoven’s 33 Variations on a Diabelli Waltz, Schumann’s Abegg Variations for pianoand Brahms’ Variations on a theme by Haydn for orchestra.

One could possibly include under the heading of “variations” the orchestrationof pieces written for a solo instrument, such as Moussorgski’s Pictures of anExhibition, orchestrated by Ravel and which is often played in concerts. There arealso orchestral pieces transcribed for a solo instrument, such as the Brahms Variationsmentioned above which the composer then rewrote for two pianos. This was oftendone in the nineteenth century. Liszt wrote his Etudes for piano based on Paganini’s

10

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

Caprices for solo violin and also used many excerpts from operas such as Bellini’sNorma, Donizetti’s Lucia di Lamermoor, Halevy’s La Juive and many others. On thebasis of themes from these works, he created arrangements that allowed him todisplay his extraordinary virtuosity.

But can we, in cases such as these, still compare the original piece and its newversion? This is not likely and at any rate, it could only happen if the necessaryconditions for the creation of a cognitive process were organized, that is if the twoversions were played one after the other. This is how Pierre Boulez recently presentedhis Notations 1, a series of twelve short pieces for the piano, some of which havealready been orchestrated. The original piano version was played before eachorchestral version so that the listener was able to compare them. This also made itpossible to see how the composer stepped back from the original piece and translatedhis ideas into a new sound context.

However, in this kind of comparison, similarity relations operate on entities thatare more or less lengthy. On the contrary, if the aim is to experience and/or assess theinfluence of “internal” similarities (within one work or section of a work), then theshort motifs, which Schoenberg called “the ‘germ’ of the idea” (1967, p. 8) willtrigger the mechanism, as we will shortly see.

2. INTERNAL SIMILARITY RELATIONS

At the end of the eighteenth century, the English philosopher Adam Smith wrote inhis essay Of the Nature of that Imitation which takes place in what are called TheImitative Arts (1795/1982):

Poetry and Eloquence… produce their effect always by a connected variety and succession

of different thoughts and ideas: but Music frequently produces its effects by a repetition of

the same idea; and the same sense expressed in the same, or nearly the same, combination

of sounds, though at first perhaps it may make scarce any impression upon us, yet, by being

repeated again and again, it comes at last gradually, and by little and little, to move, to

agitate, and to transport us.

Richard Wagner’s views are identical in this respect. In her diary (1977, p. 365),on 25th November 1870, his wife, Cosima, reports a remark from Richard whosaid:

Repetition! There is the absolute difference between music and poetry: a theme may be

repeated because it is a person and not a discourse; on the contrary, in poetry, repetition is

absurd, except when it is a refrain or when it has to produce a musical effect.

11

Similarity relations in listening to music: How do they come into play?IRÈNE DELIÈGE

(1) Concert on October 25th 2004 in Brussels, Palais des Beaux-Arts, Salle Henry Le Bœuf.

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

Heinrich Schenker (quoted in Kivy, 1993, p. 337) says also something similar inhis essay “The spirit of musical technique”, where he writes: “language… prefersexactly the opposite strategy [to music] — that is, a continuous flow, withoutrepetition”. Here, the stress is on a practice that has developed progressively inWestern music. Célestin Deliège noted on this subject that “it is… only at the startof the fiveteenth century that a type of local structure began to appear, repeated inimitation and which perception could readily identify and remember. In otherwords, it is ‘the head motif ’… a first sketch of the statement of the theme” (1989,p. 146 & 2005; Engl. tr. 2000) that will become a musical “object” with theViennese school of the classical period (ibid., p. 147). About a century and a halflater, Arnold Schoenberg insisted strongly, in his books meant for educationalpurposes, Models for Beginners in Composition (1942) and later in Fundamentals ofMusical Composition (1967), on variation technique as a way of repeating withoutnecessarily saying the same thing. He considered that this guarded againstincoherence in musical composition. Webern, like Schoenberg, was a great user ofvariations. He stressed the importance of repetition through variation in his lectures,delivered in Vienna in 1933. “Obviously this (the lack of repetition) doesn’t work, itdestroys comprehensibility. At least it’s impossible to write long stretches of music inthat way” (1963, p. 55). “Let’s learn this lesson… it is from… this idea of sayingsomething twice, more often, as often as possible, in order to make one’s selfunderstood, that the most artful things developed” (ibid., p. 22). Explaining theimportance of the technique of repetition and variation throughout history, he beganby emphasizing the role it played from the very beginning, its roots in folk music.Webern thus agrees with a finding of ethnomusicologists in the second half of thetwentieth century who stated that the method was universal.

In 1956, Gilbert Rouget had already rejected the idea that was then widespreadaccording to which so-called “primitive” music was nothing but repetition and thattherefore its form was crude and even non-existent. He felt that this was a simplisticanalysis, for the use of repetition through variation is often extremely refined in thismusic (1956, p. 133). When Constantin Brailoiu analyzed the structure of folkmusic in 1973, he said that at the start, it is “a brief period that the singer repeats asoften as needed to reach the end of the text. But at each repetition, the folkinterpretation more or less modifies the rhythm, the melody or even the architectureso that one can speak of variations. The study of these variations… is the mostimportant question in the field of musical folklore” (p. 22). Simha Arom points outthat in the music of the Aka Pygmies (1985/1991) “repetition and variation is one ofthe most fundamental principles of all Central African musics, as indeed of othermusics in Black Africa” (1991, p. 17). Giving music its structure through the“repetition/variation” system is therefore a basic psychological need of human beings,whatever their culture. This would be worth studying more thoroughly as it goes farbeyond folk music and is still a fundamental part of learning the profession ofcomposer. Can we discern possible reasons for this?

12

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

The musicologist Leonard Meyer considers that similarity relations within amusical work give the listener a feeling of security. This is why the establishment andthe generalization of this composing technique occurred at the same time as thecreation and development of concert halls open to the public in the nineteenthcentury, where professionals and music lovers could meet. In a recent book, Meyersays: “The technique of developing variation was not an invention of the mid-nineteenth century. It has been employed in the Baroque and Classic periods. Whatwas new in the nineteenth and twentieth centuries was its increased importance as acompositional constraint” (2000, p. 270)… “The idea of coherence and unitythrough similarity relationships prevailed because it was consonant both with theideology of Romanticism and with the needs of the novices in the bourgeoisaudience… the unity created by similarity is a natural one, based upon classlikeidentity rather than upon learned constraints” (ibid., p. 266). This suggests thatsimilarity relationships give the listener the feeling that he understands what he islistening to without having to study the compositional rules on which the music wasbased. Furthermore, the music lover gains satisfaction in return for his effort toparticipate: “In a comparable way, similarities of motive (and also of timbre, texture,and the like) helped the less sophisticated members of the nineteenth century’sbourgeois audience to appreciate the music they were paying for” (ibid., p. 267). Itseems, therefore, that similarity relations can influence the listener, whether he is anexpert or a beginner, and can create an atmosphere conducive to listening. Here, onemust consider the way in which these relations are perceived; more precisely, whetherthe perception process is implicit or explicit. This distinction refers to definitionsgiven in studies of memory. Implicit entails “indirect” cognitive strategies, where theperson is not aware of having access to the information contained in theperformance. On the contrary, explicit means intent and “directness” in the searchfor the information to be processed (Gaonac’h, 1998; Tulving, 2000). Along thesame lines, I will sum up the data and the procedures that I developed in myempirical work on the different phases of my “model of cue abstraction in thepsychological organization in listening to music” (Deliège, 1987a; 1989; 1991). Thesimilarity/difference axis is the central point of this model and special attention hasbeen given to its form of intervention in the cognitive processes.

3. IMPLICIT OR EXPLICIT PERCEPTION OF INTERNAL

SIMILARITY/DIFFERENCE RELATIONS IN A WORK

Different performances were approached in several ways, through processes ofsegmentation, categorization and memory. Let me recall briefly that the model isbased on a central hypothesis according to which a person who is listening withattention will seize cues, that is brief but meaningful and significant structures, whichstand out from the sound background. It is on this basis that the listener draws up aframework of the piece, thus limiting the mnemonic load of the information he orshe receives. Cues are specific to each work; they are a part of the composer’s project

13

Similarity relations in listening to music: How do they come into play?IRÈNE DELIÈGE

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

that he introduces so as to help the listener to reconstruct the piece for himself.Composers, conductors and theoreticians often comment on this idea and itsinfluence on memory. For instance, Pierre Boulez, in his book on Paul Klee, speaksof “kernels of memory [that] help the listener to progress through unknown territorythanks to a few signals that are all the stronger that they were first perceived inisolation. For the difficulty for the listener… is precisely to be able to pick out, inthe very instant, these triggers of memory” (1989, p. 110). The conductor, PatrickDavin, says something similar when he emphasizes that auditory references are “away of training the ear of the listener and constantly triggering memory” (1994,p. 13). These structures recur in a composition; they can be a literal repetition ordifferent levels of variation. The listener “picks them out” spontaneously from thesurface of the musical structures, where they stand out because of the force ofiteration and with their help, tries to approach the composer’s project (Deliège,1989, p. 306). He is not trying consciously to detect the iterations; on the contrary,he is subjected to their impact given by an accumulation effect. As Peter Kivy says:“Musical repeats perform an obvious and vital function in that they are thecomposer’s way of allowing us, indeed compelling us to linger; to retrace our steps sothat we can fix the fleeting sonic pattern; they allow us to grope so that we can grasp”(1993, p. 352).

We will now try to analyze the processes whereby musical structures arecompared and that are developed on these basic abstractions. The comments quotedabove showing that the implicit strategies are essential during the listening processand we will consider these first.

3.1 Procedures aiming at the implicit interventionof similarity/difference relations

Different tasks can reveal performances based on implicit strategies. Foremost amongthese are the strategies that aim at investigating (i) group formation in the musicalstructure, (ii) the development of a “cognitive map” or schema of the work, and(iii) the formation of imprints or prototypical patterns.

3.1.1 In group formation processesAt the end of the nineteenth century, when experimental psychology was born, thesubjective element inherent in group perception had been emphasized by Wundt. Heoften mentioned the example of a continuous series of metronome clicks that are allabsolutely identical but that are instinctively perceived in groups of twos or threes.This means that perception organizes these accents and creates a “same/different”relationship that is a part of group formation when no objective change can beobserved.

More recently, Gestalt theorists have spelled out a general definition of thispsychological ability; they showed that group formation is not based on a deliberate,or thought out, cognitive action. On the contrary, the boundaries of the groupstructures stand out spontaneously when a different event seems to be separate from

14

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

the surrounding structures. A series of laws governing perceptual organization thatare well known today, define the origin of this process: proximity, similarity,common destiny etc. and identify by their name the motif that generates groupformation.

In their seminal book, “A generative theory of tonal music”, Lerdahl and Jackendoffdevised an application for the perception of rhythm in music based on these laws. Atthe time the book was published, I had undertaken an empirical study (Deliège,1987b) to test their hypotheses. They were clearly confirmed by my results and it wasobvious that the perceptual mechanisms that can be observed in listening to musicare the same as those that are used in visual tasks. The group boundaries appearspontaneously when a contrasting element stands out from the more undifferentiatedenvironment. One could therefore logically expect similar perceptual behavior on ahigher level, i.e. on the level of group formation of groups, periods or sections of awork so as to achieve those segmentation processes that are thought to be activewhen listening to a whole piece. It then became necessary to define the way in whichthis works.

At this juncture, I put forward the idea of cue abstraction (summarized above in§ 3); the cues stand out and operate on the basis of the Principles of the Same and theDifferent and make it possible to organize relations between the perceived structures(Deliège, 1987a; 1989; 1991). According to these principles, there is a processwhereby the structures perceived become related, in other words are grouped, andthis can continue as long as an invariant potential is perceived; the process stopswhen a contrasting structure appears. This last structure limits the ongoing groupingprocess and at the same time, it selects a new cue structure and defines an identicalprocess, based on the same principles, as listening continues.

To highlight the implicit nature of the processes at work, the cognitivemechanisms that I have just described were studied on the basis of two works, eachabout 10 minutes long, that are part of the contemporary repertoire: Luciano Berio’sSequenza VI for viola and Pierre Boulez’s Éclat for 15 instruments (Deliège, 1989),and a few years later, with a shorter piece (2’22”), the English horn solo fromWagner’s Tristan und Isolde (Deliège, 1998). The participants were professionalmusicians and non-musicians. They had been instructed to listen to these pieces asif they were a story, or discourse. Their task was to note the punctuation that theyheard: commas, periods, next paragraphs, etc. by pressing a key on the computerkeyboard. For the tests with the Boulez piece and the Wagner solo, the participantswere asked to grade the punctuation: hitting a key once meant a low level punctuation;twice meant an average level and three times meant a strong punctuation. Thisinstruction was deliberately devised to observe the implicit role of similarity relationsbetween structures while listening to music in real time. Nothing in theseinstructions could show the listener the way to such relations; nothing could leadthem to think that the idea of similarity could have an impact on performances. Thedata we collected showed clearly that the use of a same type of invariant was the

15

Similarity relations in listening to music: How do they come into play?IRÈNE DELIÈGE

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

decisive factor for group formation. For this procedure, the tasks were repeated twiceduring two consecutive experimental runs; these were prepared by playingbeforehand the Berio and Boulez pieces twice and the Wagner solo three times. Weobserved that the punctuations — i.e. the group boundaries — that were identifiedthe first time were perceived the second time in the same places of the musical textfor all three pieces and for all participants, whether professionals or not, andwhatever the number of times the pieces were heard in the preliminary phase.Furthermore, in the case of the Wagner solo, two additional aspects were considered:the level of general and musical education and the prior knowledge of the piece. Wetherefore divided each group into two: students and professionals for the musicians,undergraduates and graduates for the non-musicians. Neither the level of trainingnor the prior knowledge of the work had a significant impact on performance. Wetherefore felt that we were dealing with psychological mechanisms that were basedon a large share of automatism and that, along with the implicit role that has beenmentioned, they were particularly stable from the point of view of perceptualbehavior. If you look at the results at the same time as the score, the mechanismbecomes very clear. I will show this by taking the example of the results for the thirdstudy on segmentation based on the English horn solo from Wagner’s Tristan undIsolde 2.

We noted twenty-one segmentations. These are shown in roman numerals in themusical text, from I to XXI (see figure 1A). The segmentations in italics are thosethat were given the highest rank (I, II, III, VIII, XI, XV, XVIII, XXI). The graph infigure 1B shows the ranking for all categories of listeners: we added the number ofkeys pressed for each segmentation so as to show the relative importance of theperceived segments3. Let us then consider the musical progression between thestrong segmentations so as to identify the implicit role of similarity in theseperformances. Invariant structures, the cues, stand out and function on the basis ofthe principles of the SAME and the DIFFERENT; they trigger the psychologicalprocesses concerned. This is how the segmentations appeared:I and II were perceived before and after the contrasting sequence that lies betweenthem;III at the end of the varied repetition of the initial motif;VIII after the same melodic formula was used and before the stress on the rhythmicmotif that begins with a triplet;XI, XV and XVIII when the melodic motifs change.

16

(2) I did not only choose this example for the need to be brief because the other two pieces are

longer and I couldn’t have reproduced the entire scores. It is also because I wanted to show the

results for a piece that is easier to approach. The other two are less readily accessible to non-musicians.

(3) The data for the different groups were made consistent so as to reach the maximum weight of

30 (10 by 3 keys) for any segmentation.

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

17

Similarity relations in listening to music: How do they come into play?IRÈNE DELIÈGE

Figure 1 A.

Wagner: English horn solo from Tristan und Isolde.

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

3.1.2 Working out a schema of the pieceBased on the approaches to segmentation that have just been discussed, theprocedures known as the “mental line” were meant to investigate the mental contentsthat were memorized during preparatory and experimental listening processes, againin relation to the principles of the SAME and the DIFFERENT. The aim was todetermine the implicit role of the cues used in the formation of a mental schema ofthe work. “There are many signals that the composer can give the listener to help himfollow the development of the music… If the same object or the same type of objectis heard again, memory records it and one can then understand that, for instance,one section of the piece is over and that another is beginning” (Boulez, 1989,pp. 107-8). Thus the tools to ensure a correct understanding of the composer’sproject are there from the very beginning.

The idea of a “line” as a symbol of the progression of a work in time, within thestructures of memory, underlies the experimental approach that we will bediscussing; it suggests that as these signals appear, they leave a mark on that line. Theconcept of a “line” is of course only a metaphor, but it is often mentioned intheoretical studies that concern the way time feels as it goes by, when listening to

18

Figure 1 B.

Wagner: English horn solo from Tristan und Isolde.

Mean perceived hierarchy of the segmentations.

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

music. “It is like the trail left by a jet” says, for instance, Jonathan Kramer in his book“The time of music” (1988, p. 367), suggesting the trace of the mental image createdby the musical events perceived. Xenakis (1971) looks at the duration of the differentsound components and emphasizes that: “Time is considered to be a straight line onwhich it is a question of marking points corresponding to variations of thecomponents” (p. 10). Therefore, either shortly or immediately after the segmentationtests, participants were asked to locate certain sequences of a piece within its totalduration; they had not been informed beforehand. Though we used three differentmethods for these tests, the implicit role of similarity relations in musical memorywas always the aim underlying the three approaches. As for segmentation, theinstructions did not suggest to the listener that he should pinpoint this kind ofrelations. However, the data collected clearly shows its impact.

For the first method, we used Berio’s piece (Deliège, 1989) where six mainsections had been outlined; this was done a few days after the segmentation test. Theparticipants were asked to place about forty excerpts of different durations (5” to10”) in the right section. For the second, we used Boulez’s piece (Deliège, 1993).Fifteen sequences (from 7” to 35”) were played to the participants. They were givena form showing a line divided into fifteen boxes and were asked to place thesequences on the basis of the schema they had memorized. Finally, the thirdprocedure used Wagner’s English horn solo and the participants were asked toreconstruct the whole melody (Deliège, 1998). The musical text had been organizedin seven sequences, played in random order during the test. The sequence subdivisionsfollowed the main segmentations perceived in the previous test.

The results of the three methods had obvious similarities and showed that thememorization of the cues is implemented implicitly within the organization of themental schema during the listening process, and that they imprint marks that are ofan “iconic” nature (Clarke & Dibben, 1997). These are immediate and spontaneousrelations of resemblance — a flying bat looks like a bird — contrary to the reasoningthat forms the basis of abstract classifications, as in biology where the bat would beplaced in the same category as a cat despite the complete lack of “iconic” similarity.This leads us to the distinction between a perceptual cue based on appearance and aconceptual cue which is related to meaning, as used in studies of memory (Tulving,2000).

The participants who were musicians located the excerpts more accurately.However, with the third method where the “familiarity” factor was taken intoaccount, the number of preliminary hearings definitely improved the results of non-musicians. All participants tended to react to the well known effects of primacy andrecency when remembering lists of words (Murdock, 1962; Murdock & Walker,1969; Baddeley & Hitch, 1977). The best performances were found for the segmentsat the beginning and the end of the piece.

19

Similarity relations in listening to music: How do they come into play?IRÈNE DELIÈGE

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

3.1.3 Imprint formationThe methodology used to study the concept of imprint in listening to music, isgenerally based on those inherited from the methods used in the seventies to studythe formation of prototypes (Rosch, 1975; 1978; Brandsford & Franks, 1971;Franks & Brandsford, 1971; Posner & Keele, 1968). Their main principles are basedon the premise that when a series of stimuli are variations of each other, cognitivemechanisms automatically create a pattern that “sums up” the main elementsconsidered. This pattern is therefore “an abstraction, and not simply the most typicalcategory member” (Hampton, 1998, p. 52).

The usual experimental procedures for these studies are related, in that theyconsist in two different phases: the first is the acquisition phase that prepares thesecond, the experimental one, called recognition phase. The specificity of the first isthat the stimuli presented to the participants have a sort of “family resemblance”with a pre-established prototype. The participants are then asked to memorize thedifferent items so as to be able to recognize them during the experimental phase.They do not, however, know that some of the stimuli are not shown, and preciselythat among them the prototype was discarded. During the experimental phase, thecomplete set, including the prototype, is presented. Participants are then asked to saywhich items were, or were not, found in the acquisition phase, and to grade the levelof certainty of each answer. The underlying hypothesis of this experimental methodis that the prototype — and possibly the objects that are very close variations — will be“wrongly” perceived as having been heard during the acquisition phase. The resultsbacked up this assumption.

To approach musical perception from this perspective, I decided to use a shortmonodic piece, built on two main motifs and divided into two parts: the Allegro assaiof Bach’s Sonata for solo violin in C major, BWV 1005 (see annex 1); only the firstpart was played entirely during the acquisition phase. No stimuli were presented inthe shape of short sequences but rather the whole first part of the piece. During theexperiment, participants were asked to listen to it carefully twice. It is important tounderline the difference between this and the older procedures from which this onewas derived. Then, during the recognition phase, three sets of stimuli of one or twomeasures were prepared. They were different variations of the two main motifs (seeexamples in figure 2: 1) and came from the part that had been heard; 2) stimuli takenfrom the second part that had not been heard; 3) stimuli having been slightlymodified rhythmically and/or melodically (Deliège, 1991; 1997; 2001b). As inconventional methodology used to study prototypical effects, participants were thenasked to say whether the items from any original group had been heard during theacquisition phase, by answering YES or NO.

The guiding principle of this research was twofold: first, we expected the cuesabstracted during the acquisition phase to lead implicitly to the formation of apattern called imprint, which would sum up the main characteristics of all thepresentations. Second, the idea of introducing “modifications” of certain items — or

20

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

“mistakes” as compared to the original text — was to test the sensitivity of theimprint to the memory of the style of the piece. This was expected to appear in theresults: the “modified” items being rejected as not having been heard during theacquisition phase. What we tested here was therefore essentially the similarity withthe imprint (or prototype) which in this case might appear to be a sedimentation of thestereotypes of the musical style to which it belongs (Hampton, 1998, p. 54).

As in the previously mentioned approaches, the instructions avoided drawingparticipants’ attention to the processes to be studied so as to see whether the imprintthat was implicitly formed during the preparatory phase was operational during theexperimental phase. To give an example, figure 3 shows a summary of the results ofa first test of our assumption (Deliège, 1991, 1997). As we expected, the participantswere led astray by the items which had not been played, but which they often saidthey had heard. This led to a significant decrease in the number of correct answersfor those items. On the contrary, a high proportion of correct responses for theHeard and Modified items, was observed. The expectations were thus fulfilled: theimprint that was developed acts as a prototype for its category; it also contains thegestures that are a part of the composer’s style. Before we discuss the explicit actionof similarity relations in listening to music, we wish to underline once again that asregards the psychological aspects described so far, the individual does not have directaccess to the processes that led to the cognitive contents studied. This is true of allimplicit processes (Dienes & Perner, 2002).

3.2 Procedures aiming at the explicit action of similarity/difference relationsContrary to the case studied above, the approaches we will now consider are aimedat the opposite aspect, i.e. the processes based on a conscious approach to the task.Under these conditions and procedures, the data acquired during the preparatoryphase, before the actual test, are available to the listener. He can then assemble andadjust some of his strategies to accomplish the tasks set to him. Furthermore, the

21

Similarity relations in listening to music: How do they come into play?IRÈNE DELIÈGE

Figure 2.

Examples of the 4 categories of items set up for the imprint experiment.

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

instructions to the participants explicitly mention the similarity/resemblance axis asthe operative reference for the answers expected of them; this is an importantdifference (Dunn & Kirsner, 1989). These procedures are fundamental to themethods that we are about to discuss and point to the classification of the musicalmotifs into thematic categories and to their recognition.

3.2.1 In the categorization processesThe two motifs in the J.S. Bach piece that we have been discussing (see annex 1) wereused in three other tests (Deliège, 1996). In the first, participants first heard thewhole piece. Then the two reference motifs (called A and B) were played once. Theparticipants, musicians and non-musicians, were asked to assess the frequency of theappearance of each one in the whole piece, on a scale of 1 to 3: 1 was not frequent,2 was relatively frequent and 3 was very frequent.

The second test aimed at classifying the variations of these two motifs intofamilies (see figure 4). Before this, they were played eight times to allow participantsto memorize them. Then each one was played again once separately. For the actualtest, all the variations were played three times in random order and participants wereasked to place them in the right family, A or B.

For the third test, participants were required to assess the degree of similarity ofpairs of motifs on a scale of seven degrees from 0 (no similarity) to 6 (totalsimilarity). These pairs were made of one of the two reference motifs, followed by itsexact repetition (similarity 6) or by a variation based on the family of the otherreference motif (similarity 0), or by one of its own variations (where similarity couldgo from 1 to 5).

For these three tests, the musical material presented to the participants was takenfrom a solo piece, so that no change in timbre could interfere with theirperformance. Thus the memorizing of cues and their impact were the only aspectsconcerned by the changes in motivic variations.

The impact turned out to be particularly reliable, though the instructionsregarding the tasks were not equally explicit in all three. In the first, no question was

22

Figure 3.

Percentages of correct responses for Heard (1 – 12), Unheard (13 – 24) and Modified (25 – 36)

items (from Deliège, 1991, 1997 & 2001).

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

23

Similarity relations in listening to music: How do they come into play?IRÈNE DELIÈGE

Figure 4.

Allegro Assai of the Sonata for violin solo in C major by Bach. Motifs A and B and variations.

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

asked during the preliminary hearing. It is after that that the two reference motifswere played separately, and that explicit similarity relations could be identified onthe basis of the preliminary hearing. Results showed that not only were thecharacteristics of the two motifs selected and memorized, but that during thelistening process, they formed a cognitive trace that matched their role andimportance in the piece. Motif A, that occurs less often, was graded 2 by musiciansand 2.16 by non-musicians, while motif B, which occurs more frequently, wasgraded 3 and 2.56 respectively by musicians and non-musicians.

In the second and third tests, the instructions were much more explicit as regardthe answers that were expected. Participants had been told that they would be asked:a) to place the different variations (see figure 4) in the right family and b) to assesstheir degree of similarity after having memorized the reference motifs. Results weresimilar to the ones for the first test. Therefore it seems that the degree of accuracy ofinstructions had no influence. Musicians gave 100 % correct answers as regardsclassification. The score of non-musicians was 94.5 %. As for the degree of similarity,figure 5 shows the average levels found by the musicians and the non-musicians forthe ten variations of motif A and the fourteen of motif B. It is interesting to note thatthe assessments of both groups of participants — musicians and non-musicians — werenot so far apart. Not surprisingly, the variations of motif B were considered to bemore similar than was the case for motif A: the original cues were not changed asmuch in the variations of motif B than in the ones of family A. It will suffice toconsider the text of the sequences (see figure 4) to realize that the composer hadmodified several variations of motif A much more thoroughly; in particular, there isthe inversion of the rhythmic group (sequence 2) and the modification of themelodic contour (sequences 3, 4, 8 and 9). This explains that the assessment of thedegree of similarity with the reference motif was strongly affected (see figure 5.)

3.2.2 In motif recognitionFor the impact of the cues abstracted while listening, we still needed to study motifrecognition processes in the case where the instructions explicitly mentionedsimilarity relations for the requested performances. For this project, we used a25 minutes excerpt from the second scene of Wagner’s opera Das Rheingold (Deliège,1991; 1992), which was particularly well adapted to this kind of approach. We knowthat Wagner builds strong links between his motifs and symbols, images or charactersthat they identify and recall during the whole opera and which are therefore called“Leitmotifs”. The cue in the leitmotifs thus acquires a semantic and even emotionalnature; yet at the same time it remains a “signal” (Peirce, 1974) and is able to triggerreminiscences every time it is heard, though it may appear more or less often, moreor less far between and often in an unforeseeable way. This differs from the otherapproaches we have discussed so far, where cues connected structures that were closerto each other in time. Furthermore, over and above the variations of motivicstructures, the composer has many more possibilities since in an opera he can varyinstrumental and vocal timbres.

24

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

Participants had to memorize ONE single target motif which they were asked torecognize later on in a long sequence where this target motif was played in differentforms. The experimenter, on the contrary, focuses his investigation on two or moremotifs so as to compare performances with the internal qualities of the cue structuresin the recognition process.

In this study, we compared the results of musicians and non-musicians inrecognizing three different Leitmotifs, Walhall-motif, Vertrags-motif and Riesen-motif(see figure 6 A). Three different phases were planned:- 1st phase: an acquisition session during which participants listened as many times asthey wished to the target motif; the number of repetitions was recorded by theexperimenter.- 2nd phase: a recognition trial with a small excerpt of the opera to make sure thatparticipants had understood the instructions. After this trial, participants were able

25

Similarity relations in listening to music: How do they come into play?IRÈNE DELIÈGE

Figure 5.

Average degrees of similarity recorded for Motifs A and B and variations.

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

to listen again to their target motif if they felt unsure of succeeding in theexperimental task.- 3rd phase: in the experimental session itself, participants had to recognize their targetmotif in whatever varied form in the selected excerpt of the opera by hitting a key ofthe computer keyboard while they were listening.

Results showed that the explicit recognition processes were very reliable for all

26

Figure 6 A.

Wagner: Three leitmotifs from Das Reingold.

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

participants. The only difference was the number of repetitions of the target motiffor non-musicians. Also, contrary to the musicians, they asked to hear the motifseveral times after the preparatory phase so as to feel more secure during the testphase (see figure 6 B).

Aside from these general comments, we would point out that the speed ofmemorizing during the acquisition phase depended greatly on the fact that the cuespresent in the leitmotifs were striking in themselves. The number of repetitionsrequired (see figure 6 B), showed that the Treaty-motif, was more difficult to dealwith, both for musicians and non-musicians. Its descending melodic line reached thelow register with a slow tempo and without there being any prominent event thatwas susceptible of striking the attention of the listener.

27

Similarity relations in listening to music: How do they come into play?IRÈNE DELIÈGE

Figure 6 B.

Mean number of listenings requested by musicians and non-musicians for each motif during the

1st and 2nd phases of the procedure.

4. IMPLICIT OR EXPLICIT PERCEPTION OF EXTERNAL SIMILARITY RELATIONS

So far, our discussion has mainly dealt with internal similarity relations, thosetherefore that occur within one same piece. The study was largely based on a seriesof empirical results found through the testing of my model of cue abstraction whilelistening to music. In this case and with this model, only internal similarity relationswere considered. In concluding, the question arises as to whether one can extend itsscope to external similarity relations, in other words between different works or partsof works.

To my knowledge, no empirical approach has been used to study this type ofsimilarity relations. However, a few assumptions can be suggested. When I spokeabove of the perception of internal similarity relations, I stated that in this kind ofsituation, the listener does not strive consciously to detect iterations; rather, he or sheimplicitly submits to their impact because of accumulation. On the other hand, when

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

external similarity relations are perceived, one can consider that the psychologicalsituation of the listener is very different and is mainly geared to a kind ofpremeditated search — which is therefore explicit — for comparisons while listening.But as we stated above (see § 1), such a premeditated search is only possible if thelistener has been able to hear the reference elements beforehand, as in the case ofpieces built on the “theme and variations” type of composition, and this is notnecessarily the case. Without a set-up of this kind, cognitive processes of the externalsimilarity comparisons are impossible except by referring to the memory of pasthearings. One may therefore consider that lacking such references, the psychologicalorganization of similarity relations, is relatively similar to that described in the caseof internal processes.

CONCLUSION

These findings show that in listening to music, similarity relations in cognitiveprocesses play by far the most important role; this is accepted and is, from the verystart, a part of the act of composing. Comparisons that we have called external orinternal were analyzed, and we stressed the determining factor in the perceptualorganization of the listener, i.e. similarity. Those who are involved with music, bethey composers, teachers, theoreticians or interpreters, accept that the use of thesimilarity axis in a musical work ensures its homogeneity, its unity and itsconsistency. However, in the empirical approaches described here, these conceptswere not included in the instructions given to the participants and were notinvestigated because it was obviously difficult to achieve a level of definitionunaffected by subjectivity. It is definitely inadequate to ask someone how theyexperience the unity, the consistency and the homogeneity of a piece when you areanalyzing the cognitive processes involved in listening. This would also generatevague answers, influenced by a number of factors such as the level of knowledge ofthe subjects, their intelligence and training as well as the intrinsic quality ofthose works selected for the experiment. All this would make the analysis andinterpretation of the results highly uncertain and might not yield useful results. Thatis why our experimental design did not include a consideration of the unity,consistency or homogeneity of a piece of music. Rather, we studied that elementwhich, according to knowledgeable practitioners is basic in achieving such unity, inother words, that which, in a piece, makes subjects receptive to similarity relations.Participant’s response showed that this was true in all the approaches, whatever theirlevel of musical culture and training. Our procedures intentionally includedinstructions that aimed at studying both implicit and explicit modes. It may not besurprising to see that the impact of similarity relations was greater in thoseperformances where the instructions were of an explicit nature, as in the categorizingand recognition approaches. It may be more surprising when the questions were

28

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

aimed at answers based on the influence of similarity relations, as in the perceptionof segmentations, of the schema of the work and of imprint formation. But all resultsshowed the impact of similarity. Thus, the role of similarity in music, in the presentdebate within cognitive sciences, is a matter of those models that are stronglydependant on similarity (Rips, 1989, p. 24): for instance, the “Prototype and familyresemblance models” (Rosch, 1975; 1978) and the “Exemplar models” proposed byMedin and Schaffer (1978) as I suggested in the conclusions of a recent study(Deliège, 2001). My results had emphasized the interaction of both these models inthe organization of the listening process: prototypes are formed though the impactof unexpected traits that are specific to certain sequences are preserved:

Rosch’s concepts alone, i.e. “prototype” theory, might be too narrow a framework to

circumscribe the cognitive processes at work in listening to music. A more plausible outlook

might be that we are perhaps facing a mixed situation also involving theoretical aspects put

forth by Medin and Schaffer in their “exemplar model”, which preserves and respects the

effect of uncommon traits in perceptual categorization. As far as music perception is

concerned, this seems plausible. Indeed, the results of this study showed true imprint

effects. Yet demonstrating this cognitive reality does not erase the presence and the audible

effect of possible unexpected and surprising variations in the musical discourse: in these

might reside the most important part of our enjoyment in listening to music (ibid., p. 401).

The contrary views voiced by some philosophical schools — Goodman (1970),for example — that consider the concept of similarity as being too vague to explainadequately the cognitive processes involved, do not seem acceptable if you considerthe psychological organization of listening to music. Such a view does not seem ableto define the cognitive processes concerned.

Address for correspondence:

Irène Deliège

CRFMW

5 quai Banning

B-4000 LIÈGE

e-mail: [email protected]

29

Similarity relations in listening to music: How do they come into play?IRÈNE DELIÈGE

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

30

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

31

Similarity relations in listening to music: How do they come into play?IRÈNE DELIÈGE

Ann

ex1.

Alle

gro

Ass

ai o

f th

e So

nata

for

vio

lin s

olo

in C

maj

or b

y B

ach.

Mot

ifA

and

mot

ifB

are

und

erlin

ed.

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

• REFERENCES

Arom, S. (1985/1991). Polyphonies et polyrythmies instrumentales d’Afrique centrale (deux volumes).Paris, Selaf. Engl. Transl. African polyphony and polyrhythm. Cambridge, UK: CambridgeUniversity Press.

Baddeley, A. D. & Hitch, G. J. (1977). Recency re-examined. In S. Dornic (ed) Attention andPerformance VI. London: Academic Press.

Boulez, P. (1989). Le pays fertile. Paul Klee. Paris: Gallimard.Brailoiu, C. (1973). Problèmes d’ethnomusicologie. (Textes réunis et préfacés par G. Rouget).

Genève: Minkoff reprint.Brandsford, J. D. & Franks, J. R. (1971). The abstraction of linguistic ideas. Cognitive Psychology,

2, 331-50.Clarke, E. & Dibben, N. (1997). An ecological approach to similarity and categorisation in music.

In M. Ramscar, U. Hahn, E. Cambouropoulos & H. Pain (eds). Proceedings of theInterdisciplinary workshop on similarity and categorisation, pp. 37-41. University ofEdimburgh, UK.

Davin, P. (1994). La Ronde (Reigen) de Philippe Boesmans. Commentaire musical et littéraire.L’avant-Scène Opéra no 160, 10-67.

Deliège, C. (1989). De l’œuvre musicale comme discours et comme texte. Semiotische Berichte:Text & Theory. Belgian approaches in semiotics, 2/3, 137-51. Trad. angl. MusicæScientiæ, 2000, IV-2, 213-26. Repris in Sources et ressources d’analyses musicales: Journald’une démarche (Mardaga: 2005, Introduction).

Deliège, I. (1987a). Le parallélisme, support d’une analyse auditive de la musique: Vers un modèledes parcours cognitifs de l’information musicale. Analyse musicale, 6, 73-9.

Deliège, I. (1987b). Grouping conditions in listening to music: An approach to Lerdahl &Jackendoff ’s grouping preference rules. Music Perception, 4 (4), 325-60.

Deliège, I. (1989). A perceptual approach to contemporary musical forms, Contemporary MusicReview, 4, 213-30.

Deliège, I. (1991). L’organisation psychologique de l’écoute de la musique. Des marques desédimentation — indice, empreinte — dans la représentation mentale de l’œuvre. Thèsedoctorale, Université de Liège, non publié.

Deliège, I. (1992). Recognition of the Wagnerian leitmotiv. Experimental study based on anexcerpt from “Das Rheingold”. Musik Psychologie, 9, 25-54.

Deliège, I. (1993). Mechanisms of cue extraction in memory for musical time. In I. Cross &I. Deliège (eds), Proceedings of the 2nd symposium Music and the Cognitive Sciences.Cambridge, septembre 1990. Contemporary Music Review, 9, 1993, 191-207.

Deliège, I. (1995). The two steps of the categorization process. In R. Steinberg (ed), Music and theMind Machine, Psychophysiology and Psychopathology of the Sense of Music, pp. 63-73.Springer: Heidelberg.

Deliège, I. (1996). Cue abstraction as a component of categorisation processes in music listening.Psychology of Music, 24 (2), 131-56.

Deliège, I. (1997). Similarity in processes of categorization. In M. Ramscar, U. Hahn,E. Cambouropoulos & H. Pain (eds). Proceedings of the Interdisciplinary workshop onsimilarity and categorisation, pp. 59-65. University of Edimburgh, UK.

32

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

Deliège, I. (1998). Wagner “Alte Weise”: une approche perceptive. Musicae Scientiae, special issue1998, 63-90.

Deliège, I. (2001a). Introduction: Similarity perception <-> categorization <-> cue abstraction.Music Perception 18 (3) special issue (I. Deliège, guest editor), 233-43.

Deliège, I. (2001b). Prototype effects in music listening: The notion of imprint. MusicPerception 18 (3) special issue (I. Deliège, guest editor), 371-407.

Dienes, Z. & Perner, J. (2002). A theory of the implicit nature of implicit learning. In R. French& A. Cleeremans (eds) Implicit learning and consciousness. An empirical, philosophicaland computational consensus in the making, pp. 69-91. Hove: Psychology Press.

Dunn, J. & Kirsner, K. (1989). Implicit memory: task or process? In S. Lewandowsky, J. Dunn &K. Kirsner (eds) Implicit memory. Theoretical issues, pp. 17-31. Hove: Lawrence Erlbaum.

Franks, J. J. & Brandsford, J. D. (1971). Abstraction of visual patterns. Journal of ExperimentalPsychology, 90, 65-74.

Froidebise, P. & Souris, A. (1961). «Variation». In Encyclopédie de la musique, pp. 860-61. Paris:Fasquelle.

Gaonac’h, D. (1998). «Mémoire». In O. Houdé, D. Kayser, O. Koenig, J. Proust & F. Rastier(eds), Vocabulaire de Sciences cognitives, pp. 256-58. Paris: Presses Universitaires deFrance.

Goodman, N. (1970). Seven strictures on similarity. In L. Foster & J.W. Swanson (eds), Experienceand theory, (pp. 19-29). Amherst: University of Massachusetts Press.

Hahn, U. & Chatter, N. (1998). Similarity and rules: empirically distinguishable? In S.A. Sloman& Rips, L.J. (eds), Similarity and symbols in human thinking, pp. 111-44. Cambridge,London: MIT Press.

Hampton, J. A. (1998). Similarity-based categorization and fuzziness of natural categories. InS.A. Sloman & Rips, L.J. (eds) Similarity and symbols in human thinking, pp. 51-79.Cambridge, London: MIT Press.

Kivy, P. (1993). The fine art of repetition. Essays in the philosophy of music. Cambridge, UK:Cambridge University Press, pp. 327-59.

Kramer, J. D. (1988). The time of music. New York: Schirmer books.Lerdahl, F. & Jackendoff, R. (1983). A generative theory of tonal music. Cambridge, MA: M.I.T. Press.Medin, D.L. & Schaffer, M.M. (1978). Context theory of classification learning. Psychological

Review, 85, 207-38.Meyer, L. B. (2000). The spheres of music. Chicago & London: University of Chicago Press.Murdock, B. B. Jr (1962). The serial position effect in free recall. Journal of experimental psychology,

64, 482-88.Murdock, B. B. Jr & Walker, K. D. (1969). Modality effects in free recall. Journal of verbal learning

and verbal behavior, 8, 665-76.Peirce, Ch. S. (1974). Collected papers. Cambridge, MA: Harvard University Press.Piéron, H. (1951). Vocabulaire de la psychologie. Paris: Presses Universitaires de France.Posner, M.I. & Keele, S.W. (1968). On the genesis of abstract ideas. Journal of experimental

psychology, 77, 353-63.Rips, L. J. (1989). Similarity, typicality, and categorization. In S. Vosniadou & A. Ortony (eds)

Similarity and analogical reasoning, pp. 21-59. Cambridge, UK: Cambridge UniversityPress.

33

Similarity relations in listening to music: How do they come into play?IRÈNE DELIÈGE

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

Rosch, E. (1975). Cognitive reference points. Journal of Experimental Psychology: General, 104,192-233.

Rosch, E. (1978). Principles of categorization. In E. Rosch & B. Lloyd (eds) Cognition andcategorization, pp. 28-49. Hillsdale, N.J.: Lawrence Erlbaum.

Rouget, G. (1956). À propos de la forme dans les musiques de tradition orale. Les colloques deWégimont (pp. 131-44). Bruxelles: Cercle international d’études ethno-musicologiques.

Schenker, H. (1895/1988). Der Geist der musikalischen Technik. Musikalisches Wochenblatt 26,pp. 245-246, 257-259, 273-274, 285-286, 297-298, 309-310 et 325-326. 1988, Trans.by William Pastille: The spirit of musical technique. Theoria 3, p. 86-104.

Schoenberg, A. (1942). Models for beginners in composition. New York: Schirmer.Schoenberg, A. (1967/1973). Fundamentals of musical compotision. G. Strand & L. Stein (eds).

London: Faber and Faber.Sloman, S. A. & Rips, L. J. (1998). Similarity as an explanatory construct. In Sloman, S. A. &

Rips, L. J. (eds) Similarity and symbols in human thinking, pp. 1-15. Cambridge,London: MIT Press.

Smith, A. (1795/1982). Of the nature of that imitation which takes place in what are called theimitative arts. In Essays on philosophical subjects, pp. 131-84. Hildesheim, New York:Georg Olms Verlag.

Tulving, E. (2000). Concepts of memory. In E. Tulving & F. Craik (eds), The Oxford handbook ofmemory, pp. 33-43. Oxford, New York: Oxford University Press.

Wagner, C. (1977). Journal, vol. I (1869-1872). Paris: Gallimard. V.O. Die Tagebücher (1976).Munich: R. Piper.

Webern, A. (1963). The path to the new music. Vienne: Universal.Xenakis, I. (1971). Musique, Architecture. Tournai: Casterman.

34

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

• Relaciones de similitud en la escucha musical : ¿ Cómo funcionan ?

En este ensayo se estudian las relaciones de similitud en la percepción musical

partiendo de dos perspectivas diferentes : unas de tipo externo, que establecen

comparaciones entre entidades musicales distintas y autónomas, es decir, entre

obras o ejecuciones diferentes ; y otras de tipo interno, que consideran las

relaciones de similitud que el oyente identifica en la misma obra o sección de una

obra, en la misma interpretación. Este último procedimiento se desarrolla en

diferentes direcciones. Se muestra sucesivamente la importancia del factor de

similitud : (i) desde el acto compositivo mismo, cuando el compositor persigue

unidad y coherencia en su obra ; (ii) en las músicas de tradición oral ; y (iii) como

un elemento fundamental para el oyente que aspira a comprender la obra en el

tiempo real de la escucha. Este último punto es de fundamental importancia para

el estudio de los procesos cognitivos en general, es decir, el papel implícito o

explícito del factor similitud en los procesos perceptivos.

Ejemplos concretos de los puntos teóricos principales descritos aquí pueden ser

identificados en el desarrollo de las diferentes etapas del modelo de extracción de

índices -procesos de segmentación, de categorización, de esquematización, de

formación de marcas- que he desarrollado recientemente (Deliège, 1989, 1991). El

eje SIMILITUD/DIFERENCIA es un elemento central en la estructura del modelo.

Las aproximaciones empíricas empleadas en el curso de experimentaciones de las

diferentes fases del modelo han sido analizadas en esta doble perspectiva. Desde el

comienzo del proceso, éste era el objetivo de una serie de procedimientos que se

planteaban poner en evidencia los aspectos implícitos o explícitos de los procesos

que los participantes músicos y no músicos emplean durante los procesos de

escucha. Una visión de conjunto de los resultados obtenidos confirma que las

hipótesis de partida eran correctas.

• Relazioni di similarità nell’ascolto musicale : come entrano ingioco ?

Nel presente saggio si studiano le relazioni di similarità nella percezione della musica

sulla base di due prospettive differenti. La prima è detta esterna, e studia le

comparazioni fra entità musicali distinte e autonome, ossia brani o interpretazioni

differenti. La seconda è interna ed esamina le relazioni di similarità che l’ascoltatore

identifica nello stesso brano o parte di un brano. Quest’ultima procedura si

sviluppa secondo direzioni differenti. Mostriamo l’importanza del fattore di

similarità : (i) nell’effettivo processo della composizione, quando l’autore ricerca

unità e coerenza nel proprio brano ; (ii) nella musica popolare ; (iii) quale elemento

essenziale per l’ascoltatore che tenti di comprendere il brano durante l’ascolto in

tempo reale. Quest’ultimo punto è di fondamentale importanza per lo studio della

cognizione in generale : la questione del ruolo implicito o esplicito della similarità

nei processi percettivi.

Esempi concreti dei principali punti teorici qui descritti si possono identificare nel

corso delle diverse fasi del modello di estrazione dei segnali da me recentemente

35

Similarity relations in listening to music: How do they come into play?IRÈNE DELIÈGE

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

sviluppato : segmentazione, categorizzazione, schematizzazione, formazione di

impronte (Deliège, 1989, 1991). L’asse SIMILARITÀ/DIFFERENZA è un elemento

centrale nella struttura di tale modello. Gli approcci di tipo empirico usati negli

esperimenti sulle differenti fasi del modello sono stati analizzati in questa duplice

prospettiva, alla quale si è mirato fin dall’inizio con una serie di procedure che

avevano lo scopo di evidenziare gli aspetti impliciti o espliciti dei processi usati da

partecipanti musicisti e non musicisti durante l’ascolto sperimentale. Una visione

d’insieme dei risultati ottenuti conferma la fondatezza delle ipotesi.

• Relations de similarité dans l’audition musicale :comment interviennent-elles ?

Dans cet essai, les relations de similarité dans la perception musicale, sont

envisagées selon deux perspectives : celles, dites de type externe, qui établissent

des comparaisons entre entités musicales distinctes et autonomes, c’est-à-dire entre

œuvres ou excécutions différentes ; celles de type interne qui considèrent les

relations de similarité au sein de l’audition d’une même œuvre ou section d’œuvre,

c’est-à-dire au sein d’une même interprétation. Cette dernière perspective est

développée en différentes directions. On montre successivement l’importance du

facteur similarité (i) dès l’acte d’écriture, c’est-à-dire dans la recherche d’unité et de

cohérence que le compositeur vise à introduire au sein même de son œuvre ;

(ii) dans les musiques d’origine populaire ; (iii) comme élément fondamental pour

l’auditeur dans sa compréhension de l’œuvre dans le temps réel de l’écoute. Ce

dernier point préoccupe au premier chef le domaine de la cognition en général, la

question du rôle implicite ou explicite du facteur similarité dans les processus

perceptifs est dès lors posée.

Les principaux points théoriques décrits trouvent des exemples pratiques dans les

différentes étapes du modèle d’extraction d’indices — processus de segmentation,

de catégorisation, de schématisation, de formation d’empreintes — que j’ai

développé récemment (Deliège, 1989, 1991) : l’axe SIMILARITÉ/DIFFÉRENCE

étant un élément central dans l’articulation du modèle. Les approches empiriques

menées au cours des expérimentations des différentes phases du modèle ont été

analysées dans cette double perspective, laquelle a été ciblée dès le départ dans un

ensemble de procédures visant précisément à mettre en évidence les aspects

implicite ou explicite des processus lors de l’écoute expérimentale chez des

participants musiciens et non-musiciens. Une vue d’ensemble des résultats

enregistrés confirme le bien-fondé des hypothèses.

• Ähnlichkeitsbeziehungen beim Musikhören : Wie entstehen sie ?

In diesem Aufsatz werden Ähnlichkeitsbeziehungen in der musikalischen

Wahrnehmung aus zwei verschiedenen Perspektiven untersucht. Die erste

Perspektive kann als extern bezeichnet werden und vergleicht distinkte und

autonome musikalische Einheiten, d. h. verschiedene Stücke oder verschiedene

36

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from

Interpretationen. Die zweite Perspektive ist intern und berücksichtigt

Ähnlichkeitsbeziehungen, die der Hörer in einem Musikwerk oder einem Teil eines

Musikwerks und in derselben Interpretation identifiziert. Diese letzte Perspektive

wird in verschiedene Richtungen entwickelt. Wir zeigen die Wichtigkeit des

Ähnlichkeitsfaktors (i) beim Kompositionsprozess, indem der Komponist nach

Einheit und Kohärenz des Stückes sucht, (ii) in der Volksmusik, (iii) als ein

essentielles Element für den Hörer, der während des Hörens in Echtzeit nach einem

Verständnis des Stückes sucht. Für Untersuchungen zur Kognition generell ist dieser

letzte Punkt von großer Bedeutung, d.h. für die implizite oder explizite Rolle von

Ähnlichkeit in Wahrnehmungsprozessen. Die hier beschriebenen theoretischen

Hauptpunkte können mit konkreten Beispielen für die verschiedenen Phasen des

Hinweisreiz-Extraktions-Modells (Prozesse der Segmentierung, Kategorisierung,

Schematisierung, Eindrucksbildung), das ich kürzlich entwickelt habe, belegt

werden (Deliège, 1989, 1991). Die Achse Ähnlichkeit/Unterschied ist ein zentrales

Element in der Struktur des Models. Die empirischen Herangehensweisen, die bei

Experimenten zu den verschiedenen Phasen des Modells Verwendung finden,

werden unter dieser zweifachen Perspektive analysiert. Von Anfang an war dies

das Ziel einer Reihe von Untersuchungen, die impliziten oder expliziten Aspekte

der Hörprozesse in Experimenten mit Musikern oder Nichtmusikern als

Versuchsteilnehmern sichtbar machen sollten. Insgesamt bestätigen die erzielten

Ergebnisse die Richtigkeit der Hypothesen.

37

Similarity relations in listening to music: How do they come into play?IRÈNE DELIÈGE

at UNIV ESTDL PAULISTA DE MESQIT on August 29, 2014msx.sagepub.comDownloaded from