musical analysis as articulation

17
MARK DEBELLIS Musical Analysis as Articulation The Journal of Aesthetics and Art Criticism 60:2 Spring 2002 What are the purposes of musical analysis? What knowledge, if any, does it convey to us, and what, if anything, does it do to reshape our hearing? Is analysis basically a descriptive or a prescriptive enterprise? Is its fundamental pur- pose to describe how listeners perceive music—or, better, to say what it is they perceive in it—as, perhaps, data for a cognitive science of musical abilities? Or is its purpose rather to give us new ways to hear, to serve as interpretation and criticism? And if there is something right in each of these conceptions of the purposes of analysis, how do those purposes relate to one another? Three main views of the value and purposes of musical analysis have emerged. On the first, analysis has an essential role in explanation: it aims to reveal, as it were, the microstructural basis of such overtly perceptible, manifest prop- erties as tonal coherence in terms of features of the musical fabric that may or may not them- selves be perceived. The classic instance of analysis claimed to have this function is the motivic analysis of Rudolph Reti, which finds “hidden” thematic relationships within a work that allegedly confer unity upon it. 1 There is an analogy here with microstructural explanations of physical properties such as solubility, or perceptible (or psychophysical) ones such as sweetness. A second view of musical analysis finds its value not in explanation but in its cultivation of an enriched mode of hearing on the listener’s part, where the listener hears the music in a new way, or becomes aware of new relationships to hear in it. Peter Kivy argues cogently for this conception of analysis over the explanation view, particularly in the case of Reti-style analy- ses. 2 The aim of such analyses, on Kivy’s view, is to augment and enrich the intentional object of this listener’s experience; they function as in- terpretation rather than naturalistic explanation. When such analyses are successful (which Kivy thinks Reti’s are, by and large, not), they “[open] up to us heretofore unperceived and un- appreciated features of the works they treat of for us to contemplate, understand, enjoy.” 3 In what follows I shall call this the “prescriptive” view of analysis, because on this view analysis invites the listener to hear in new ways, or to hear new things in the music. The third view of musical analysis takes it to be, by contrast with the second, descriptive. On this view, analysis takes as its object of descrip- tion the intentional object of the listener’s musi- cal experience, and moreover an object that, normally, is not a product of the analytic en- counter but something that exists prior to it, something that analysis then illuminates or elu- cidates. 4 The canonical example of analysis that conforms to this view is that of Lerdahl and Jackendoff’s Generative Theory of Tonal Music. 5 Their theory is proffered as cognitive science, a model of the musical cognition of the “experienced listener” of tonal music. 6 Thus we have analysis as explanatory, pre- scriptive, and descriptive; in what follows I will primarily be concerned with the last two (leav- ing what I have called the explanatory aside, for the most part). In both the prescriptive and de- scriptive conceptions there is something impor- tant, I think: each conveys part of the picture of what music theorists are attempting to do in the analysis of music. But it must be acknowledged that the prescriptivist and descriptivist dimen- sions are not mutually exclusive, but interact with one another in important ways. My purpose is to demonstrate that and explain how

Upload: bb7sharp9

Post on 23-Apr-2017

231 views

Category:

Documents


1 download

TRANSCRIPT

MARK DEBELLIS

Musical Analysis as Articulation

The Journal of Aesthetics and Art Criticism 60:2 Spring 2002

What are the purposes of musical analysis?What knowledge, if any, does it convey to us,and what, if anything, does it do to reshape ourhearing? Is analysis basically a descriptive or aprescriptive enterprise? Is its fundamental pur-pose to describe how listeners perceivemusic—or, better, to say what it is they perceivein it—as, perhaps, data for a cognitive science ofmusical abilities? Or is its purpose rather to giveus new ways to hear, to serve as interpretationand criticism? And if there is something right ineach of these conceptions of the purposes ofanalysis, how do those purposes relate to oneanother?

Three main views of the value and purposesof musical analysis have emerged. On the first,analysis has an essential role in explanation: itaims to reveal, as it were, the microstructuralbasis of such overtly perceptible, manifest prop-erties as tonal coherence in terms of features ofthe musical fabric that may or may not them-selves be perceived. The classic instance ofanalysis claimed to have this function is themotivic analysis of Rudolph Reti, which finds“hidden” thematic relationships within a workthat allegedly confer unity upon it.1 There is ananalogy here with microstructural explanationsof physical properties such as solubility, orperceptible (or psychophysical) ones such assweetness.

A second view of musical analysis finds itsvalue not in explanation but in its cultivation ofan enriched mode of hearing on the listener’spart, where the listener hears the music in a newway, or becomes aware of new relationships tohear in it. Peter Kivy argues cogently for thisconception of analysis over the explanationview, particularly in the case of Reti-style analy-ses.2 The aim of such analyses, on Kivy’s view,

is to augment and enrich the intentional objectof this listener’s experience; they function as in-terpretation rather than naturalistic explanation.When such analyses are successful (which Kivythinks Reti’s are, by and large, not), they“[open] up to us heretofore unperceived and un-appreciated features of the works they treat offor us to contemplate, understand, enjoy.”3 Inwhat follows I shall call this the “prescriptive”view of analysis, because on this view analysisinvites the listener to hear in new ways, or tohear new things in the music.

The third view of musical analysis takes it tobe, by contrast with the second, descriptive. Onthis view, analysis takes as its object of descrip-tion the intentional object of the listener’s musi-cal experience, and moreover an object that,normally, is not a product of the analytic en-counter but something that exists prior to it,something that analysis then illuminates or elu-cidates.4 The canonical example of analysis thatconforms to this view is that of Lerdahl andJackendoff’s Generative Theory of TonalMusic.5 Their theory is proffered as cognitivescience, a model of the musical cognition of the“experienced listener” of tonal music.6

Thus we have analysis as explanatory, pre-scriptive, and descriptive; in what follows I willprimarily be concerned with the last two (leav-ing what I have called the explanatory aside, forthe most part). In both the prescriptive and de-scriptive conceptions there is something impor-tant, I think: each conveys part of the picture ofwhat music theorists are attempting to do in theanalysis of music. But it must be acknowledgedthat the prescriptivist and descriptivist dimen-sions are not mutually exclusive, but interactwith one another in important ways. Mypurpose is to demonstrate that and explain how

this obtains, and thereby to embrace bothprescriptivist and descriptivist strains within amore comprehensive, unified account than theyhave heretofore received, one that illuminatesthe role of each and their relation to one an-other.

The central insight behind the view I shall en-dorse is owed to Kendall Walton: that analysisis the articulation of one’s musical experience.7What I propose to do here is to flesh out the ac-count, to see how it should go in the case of aparticularly important kind of musical analysis,one originating with Schenker, and to work outwhat must be the case if music-analytical prac-tice is to fit intelligibly within Walton’s articu-lation model. In so doing, I shall show how theprescriptive and descriptive functions of analy-sis interact with one another in such a way thatthe enriched ways of hearing cultivated bySchenkerian analysis further the articulation ofpreanalytical musical experience. In the courseof this, I shall inquire into the relation betweenperceptions of, and judgments about, musicalstructure; whether Schenkerian structures func-tion like theoretical entities in science; andwhether Schenkerian theory can be reconciledwith a model of musical composition and recep-tion as communication. Finally, I shall point outwhat I think is the most pressing, epistemo-logical problem: to understand how articulationis the expression of music-analytical knowl-edge.

I

The kind of musical analysis with which I willbe concerned is Schenkerian analysis, surely thedominant paradigm today in the analysis oftonal music.8 Originated by the Viennese musictheorist Heinrich Schenker (1868–1935),Schenkerian analysis presents a view of thestructure of a tonal piece or passage of music onwhich certain elements are structurally basic andothers ornamental; the latter are said to “pro-long” the former.9 As an example, consider theopening measures of Mozart’s Piano Sonata inD, K. 311, first movement, shown in Example 1.

The musical surface is shown in level (a), theunderlying structural pattern in (b); that underly-ing pattern (from the third beat on) is a succes-sion of parallel tenths between the highest andlowest voices (a pattern to be conceived of, firstand foremost for Schenker, in contrapuntalterms). The G at the beginning of the third beat,in the right hand, is ornamental, an appoggiaturato the more structurally important F#; the entireturn figure G-F#-E-F# serves to prolong F#.10

(The official Schenkerian term for the relation-ship of ornament to structure is “diminution”:what would otherwise be a quarter-note F# is, asit were, replaced by a figure in shorter notes.)

Such relationships, moreover, are typicallyordered hierarchically on several structural lev-els. The F#-G-A motion at level (b) can, in turn,be related to a deeper level, not shown, consist-

120 The Journal of Aesthetics and Art Criticism

EXAMPLE 1. Mozart, Piano Sonata in D, K. 311, first mvt., mm. 1–2

ing of the tonic chord D-F#-A. The G on thefourth beat functions as a passing tone betweenF# and A, themselves more structurally stablethan G because they belong to the tonic chord.

A second example, which illustrates struc-tural levels well, is the beginning of Beetho-ven’s Piano Sonata Op. 2, No. 2, second move-ment (Example 2, score annotated withharmonic analysis).

There are three levels here: a small-scaleI-V-I in mm. 1–2, which is nested in alarger-scale I-V-I, which is in turn part of theoverall I-V of mm. 1–4. The dominant chords onthe first beats of mm. 2, 3, and 4, respectively,do not all have the same structural significanceor function. In particular, the V in m. 3 is not onas high a level as that in m. 4. The m. 3 V is in-ternal to a larger prolongation of I, which occu-pies the highest of the three levels shown, andpersists through the first three measures, movingto V only in m. 4.

II

The central metatheoretical question in musicalanalysis is this: What is an analysis, such as aSchenkerian graph, fundamentally about? Whatis its real subject matter, what information doesit convey, what purposes does it serve? In an im-portant essay on musical understanding KendallWalton takes this view:

Much of [music theorists’] effort, I suggest, goes to . . .recognizing or acknowledging, and articulating, thecontent of their musical experiences. . . . The possibil-ity is open that even the Schenkerian deep structure ofa piece, or the fact that the foreground and middleground are elaborations of the deep structure, is in fact

an unacknowledged part of the content of musical ex-periences even of ordinary listeners. . . . An analysis ofa piece may amount to a speculation about what mightbe in the unacknowledged content of the analyst’s orother listeners’ experiences.11

On Walton’s view, then, an important kind ofunderstanding gained in musical analysis is thatof recognizing and articulating the intentionalcontent of one’s musical experience.

On the whole I shall endorse this view of mu-sical analysis (in the case of Schenkerian analy-sis, at least). But there are matters that warrantinquiry. One important area is epistemological.We want to understand the process of observa-tion, reasoning, or inference by which the musicanalyst arrives at an articulation of his or herhearing. We want, moreover, to understand whythis process issues (if it does) in knowledge; wewant an understanding of how anyone knowsthat a putative articulation of his or her hearingis in fact a correct articulation of it. The follow-ing passage from Walton helps to set the issuesin relief:

I urge that we avoid a narrow conception of what onehears. We are sometimes angry or jealous withoutbeing able to articulate fully what we are angry orjealous about. We also, even more frequently per-haps, laugh without being able to say in much detailwhat it is that we are laughing at. . . . So the content ofa musical experience may include features of themusic the experiencer cannot specify. There may notalways be a good way of ascertaining what the con-tent of any of these intensional states includes, of ac-quiring the capacity to articulate them in their en-tirety. When possibilities are suggested, we mayrecognize or acknowledge them as indeed among the

DeBellis Musical Analysis as Articulation 121

EXAMPLE 2. Beethoven, Piano Sonata in D, Op. 2, No. 2, second mvt., mm. 1–4

objects of the experience (as having been so allalong). But failure to make such an acknowledgmentcannot be taken as conclusive proof that they are notobjects of the experience. Recognition does not al-ways come easily.12

Walton’s suggestions are intriguing, but pointtoward the need for further clarification. Wewill want, first of all, to have a fuller under-standing of the machinery by which we have re-liable, though fallible, epistemic access tomusic-experiential contents. (Ideally, we wouldlike to know why that machinery works when itworks and fails when it fails, why our knowl-edge of music-experiential content is locatedprecisely where it is between the Scylla of incor-rigibility and the Charybdis of inaccessibility.)We will want to understand how genuine recog-nition differs from what only purports to be rec-ognition. Part of the worry is that the notion ofcontent employed here is so indefinite as to pro-vide no meaningful constraint on a correspond-ing notion of the successful employment of ar-ticulation or recognition of that content. And theextreme skeptical worry, which may or may notgrip us but which in any case we should be ableto defuse, is that what purports to be articulationis merely an arbitrary, unconstrained use of lan-guage subject to no standard of correctness.13

In allowing that analysis may be speculationabout experiential content, Walton evinces ad-mirable sensitivity to the need to make room forthe uncertainty attached to musical analysis, toits open-ended revisability. A given analysis canseem right to us (where, for Walton, this meansthat it seems to us to correctly articulate our ex-perience); and then we change our minds, decid-ing that, not this, but that way is how we werehearing the music all along. But to make senseof this we need a fuller understanding of how weknow that a given analysis or description of themusic correctly articulates our experience(when and if we do): talk of speculation has apoint, after all, only if we have some route of re-liable access to the truth. Thus, we need to un-derstand how, at the earlier stages of the processof making up our minds, error can occur, andwhy we have any reason to think that later stagesconstitute an improvement.14

The second area in which things need furtherspelling out is broadly semantic. We need afuller understanding of the relationships of con-

tent among analytical statements, hearing, andjudgment, so as to arrive at a clear conception ofwhat it is for an analysis to be an articulation of ahearing. Thus we need to inquire into the mean-ing of music-analytical statements and their re-lation to the content of musical experience.

Third, we want to show how the articulationaccount coheres with actual music-analyticpractice, and even illuminates that practice.Does it cohere with the facts about how musicanalysts actually arrive at and regard musicalanalyses? Can analysis be articulation if it culti-vates new ways of hearing? If analysis is a mat-ter of discovering the facts about the structure ofmusical works, does the articulation model helpus to understand the nature of such facts andhow music-analytical practice enables us toknow them?

III

Central to and cultivated by Schenkerian theoryis a certain mode of perceptual contact withmusic that I shall call structural hearing, appro-priating the term from the title of a book byFelix Salzer that has the express purpose ofteaching one “to hear not only the succession oftones, melodic lines and chord progressions butalso their structural significance and coher-ence.”15 Structural hearing is a special percep-tual activity, distinct from ordinary, “intuitive”musical cognition (where the latter is whatLerdahl and Jackendoff’s theory, for example,describes).16 Structural hearing goes beyond theintuitive in that it involves explicit awareness of,and attention to, structural relationships con-ceived in Schenkerian terms.

To spell this out a bit, structural hearing con-sists in perceiving the musical fabric as the un-folding of certain kinds of temporally-extendedprocesses. An example is the linear progression(Example 3).

One hears the process commencing at a cer-tain time with a certain pitch (D), progressingthrough a passing tone (C), and eventuallyreaching a goal (B-flat); in this example of pass-

122 The Journal of Aesthetics and Art Criticism

EXAMPLE 3. Linear progression

ing motion one hears a process beginning, con-tinuing, and ending. Now it is important to seethat in any actual piece several processes will begoing on simultaneously at different structurallevels, and, thus, an important element of struc-tural hearing will consist in the determination ofwhich of those processes any given surface ele-ment belongs to and what role(s) that elementplays. Consider, for example, the opening of theDivertimento in B-flat (Chorale St. Antoni), at-tributed to Haydn (Example 4, analysis by Forteand Gilbert).17

There is a linear progression C-B-flat-A in m.9 (ignoring D), which is at a lower structurallevel than the larger-scale D-C-B-flat in mm.6–10, shown in beamed notes; the lower-levelprogression prolongs the second note, C, of thehigher-level one. A structural hearing of the pas-sage corresponding to this reading entails hear-ing the B-flat in m. 10—but not the one in m.9—as the goal of the larger-scale progression.On hearing the m. 9 C, one has an expectation oflarge-scale structural motion to B-flat—onehears the C as issuing a “promissory note,” so tospeak, which one subsequently hears as “paidoff” by the m. 10 B-flat. The B-flat in m. 9, incontrast, is not heard as participating in thathigher-level progression, but rather as a passingnote within the lower-level C-B-flat-A; it carrieswith it its own, lower-level “promissory note,”which is paid off by the motion to A. (It wouldbe a mistake to hear the m. 9 B-flat as belongingto and completing the higher-level progression;that would be a premature assignment of clo-sure.) Typically the completion of a process(payment of a promissory note) is heard as theresolution of tension; and what is a tense or dis-sonant note at one level can itself be prolonged(through notes that treat it as a relatively stable

point, departing from it) at lower level(s). In thisway the ear hears temporally extended processescharacterized by tension and resolution, nestedhierarchically. Structural hearing involves vividattention to such patterns: to the resolution ofmelodic and harmonic tension, to “promissorynotes” issued and paid off, to the nested struc-ture of such processes beginning, progressing,and ending.18

IV

Several aspects of, or assumptions behind,Schenkerian analysis need to be elucidated atthis point; there are five in all. The first assump-tion is that it is meaningful to speak of analysesas correct or incorrect.19 An analysis of the Mo-zart K. 311 passage (Example 1) as a 10-10-10—i.e., as having that intervallic pattern as itsstructural basis—is correct; an analysis of it as11-11-10 is incorrect. I shall take it, moreover,that no important distinction is lost if we takecorrect to mean true and incorrect to meanfalse, for the simple reason that music theoriststreat expressions such as the following as inter-changeable:

“10-10-10” is a correct analysis of that passage

and

That passage is a 10-10-10.

Hence the assumption amounts to this: analyseshave truth-values.20

The second assumption is that an analysispredicates, of a passage, a structural property;the analysis is true (correct) if and only if thepassage has that property. Facts about what

DeBellis Musical Analysis as Articulation 123

EXAMPLE 4. Schenkerian analysis of Haydn, Chorale St. Antoni, mm. 1–10Reprinted from Introduction to Schenkerian Analysis by Allen Forte and Steven E. Gilbert ©1982 by W. W. Norton& Company, Inc. With permission of the publisher, W. W. Norton & Company, Inc.

structural properties belong to what passages arethe truth-makers of analyses.

Third, analyses are used to express and con-vey analysts’ judgments about what musicalstructures belong to what passages; those judg-ments are likewise correct or incorrect, true orfalse.

Fourth, there is a clear relationship of corre-spondence between analyses and structuralhearings such that we may speak of a structuralhearing “according to” or “along the lines of” agiven analysis.

Fifth—and this is especially important—predications of, and judgments about, musicalstructure typically receive their epistemic justifi-cation from (certain) corresponding structuralhearings. The analyst reaches the conclusionthat the Mozart passage is a 10-10-10 at leastpartly on the basis of his or her structural hear-ing of the passage; the judgment is supported bythe hearing. (For the cases in which I am inter-ested, it is a necessary but not sufficient condi-tion that said hearings correspond with the anal-yses in the aforementioned way; presently weshall discuss what further condition needs to besatisfied.) Now this point may seem unexcep-tionable and I think it is well borne out by actualanalytical practice; however, the music-theoreti-cal literature does not show unanimity on thispoint. It has been suggested that Schenkerianstructures are like theoretical entities in science,themselves unobservable but having some indi-rect explanatory connection to observables suchas tonality or the property of sounding unified.21

But it seems to me that this overlooks themore-or-less direct connection, in music-analyt-ical practice, between ascriptions of Schen-kerian structures (dominant prolongations, tochange the example) and experiences music the-orists typically characterize as perceptions ofthose structures in the music (e.g., hearing a pas-sage “as a dominant prolongation”).22 One sim-ply does not have, in Schenkerian theory, any-thing like the network of physical laws andtheory that link electrons and cloud-chamber ex-periments indirectly, through complex inferen-tial connections; rather, one has (what for a firstapproximation may be characterized as) the-ory-laden observation, observation that directlysupports the ascription of the observed proper-ties. Schenkerian structures are, in short, muchcloser to observables than theoretical posits.23

Notice, however, that the analyst does nottypically proceed immediately from the fact thata passage is heard in a certain way on a given oc-casion (much less that it can be heard in thatway) to the ascription of the relevant structure tothe music. Normally what goes on is a process Ishall call music-analytical “deliberation,” whereone “tries on” various structural hearings of apassage and asks which of them is (or are) mostnatural, plausible, or convincing. Thus,epistemic support for an analytical judgment isnormally provided by a plausible correspondingstructural hearing. The phenomenology of ob-servation and judgment is interesting here; let usreturn to the Beethoven passage in Example 2and trace through the process. (We will startwith what I shall call interpretation A, considerinterpretation B, and ultimately opt in favor ofA.) Interpretation A corresponds to the analysisgiven earlier: one hears the dominant chord onthe first beat of m. 2 resolve to the tonic on thenext beat, and again, one hears the V on the firstbeat of m. 3 resolve to the tonic on the third beat.On this hearing, the “promissory note” issued bythe m. 3 V is heard as paid off by the return tothe tonic on the next beat. That surely seems anatural hearing. Alternatively, let us consider in-terpretation B, which has the progression (c′) in-stead of (c); here the I in m. 3 is a passing chordwithin a prolonged V chord (Example 5).

This reading treats the V chords in mm. 3 and4 as pillars of a single dominant unit. Now atfirst, one might—especially if one has a skepti-cal turn of mind—doubt whether there is anyreal basis to speak of either A or B as correct orincorrect; perhaps they are just different ways oflooking at the piece as structured. But as one“tries on” structural hearing B, one mentallytraces out a continuity from the first V chord tothe other, hearing the line of the music as pass-ing through the intervening tonic chord, not re-solving to it as a goal (as it did on interpretationA). (On this hearing the tonic chord on beat 3 ofm. 3 does not serve as resolution, as payment ofthe V’s promissory note.) Yet this does not sitright: it seems a falsification of the true struc-tural weight of the tonic chord here. (Intuitivelyone wants to say, “That’s not how I hear it.” Ofcourse one just did hear it that way, but clearlythat is not what is at issue.) Interpretation A, incontrast, gets the structural weights right. Thusthe analyst opts for A over B. (Prior to engaging

124 The Journal of Aesthetics and Art Criticism

in this deliberation, we may have been skepticalabout the very meaningfulness of there being a“correct” analysis of the passage. But by the endof it, we do seem to have arrived at a right an-swer, and talk of a right answer does seem tohave a point.)24

Thus, in arriving at and justifying a structuralinterpretation of a passage, we typically engagein acts of structural hearing that correspond tovarious analyses and select the most plausible(natural, satisfying) one; that hearing, and analy-sis, are those we deem “correct.”25 But whatdoes correctness amount to here: what is itssource and ultimate explanation? It is virtuallytaken for granted in music analysis that plausi-ble structural hearings are a ticket to correctjudgments about musical structure. But whyshould they be? Why should plausibility—ornaturalness or convincingness—be any indica-tor of the facts about what musical structures be-long to what passages? Or, to turn the questionaround, what must musical structure be, if thetruth about it is revealed above all by plausiblestructural hearings?

Perhaps there is no real answer to this ques-tion. Perhaps all there is to say is that plausibil-ity gives us correctness, and that is that. Expla-nations do have to come to an end somewhere.But it is relevant, I think, to point to the fact that,in selecting a preferred analysis, we say, “That’show I hear it” (or in rejecting one we say,“That’s not how I hear it”). But, again, what dowe mean by this, and why is it relevant? Afterall, in considering any analysis, even a wrongone, we can, at least momentarily, hear it thatway. It might be suggested that by “I hear it thatway” we really mean “I prefer to hear it thatway,” or “It is interesting to hear it that way”;but I think that is to go down the wrong path. It

seems to me that in an important range of cases,“how I hear it” refers to some intuitive way ofhearing that precedes analysis and discursivejudgment—that precedes our explicit, theory-laden hearings—and it is in such cases, I argue,that Walton’s idea of articulation comes intoplay.26

It cannot be emphasized enough that struc-tural hearing is something that goes beyond or-dinary listening; it is no exaggeration to say it isthe raison d’être of Schenker’s work, somethingwhich cannot, as a practical matter, be experi-enced by a listener without some exposure toSchenkerian theory. To come back to Kivy’sformulation of the issues, then, it is evident thatSchenkerian analysis does function as interpre-tation, leading one to hear in new ways, aug-menting and expanding the intentional object ofmusical experience.

In this connection, though, a puzzle arises,one that can be best brought out by reference toa recent commentary by Robert Snarrenberg. AsSnarrenberg points out, Schenker regarded hisapproach as one that reveals the “content”(Inhalt) of musical artworks (to the larger end ofrescuing musical culture from its decline).27

Snarrenberg then attempts to locate Schenkeriananalysis within a larger model of musical activ-ity as communication:

Schenker conceives of music as a form of communi-cation in the broadest sense. Composition and inter-pretation are complementary activities centered ontonal content. Composers intend to produce effects orresponses in others by means of configuring tones insuch and such a manner. Listeners hear (or imaginehearing) the presented configuration of tones and re-spond appropriately. The perceptual act of interpreta-tion (the psycho-physical response) may be followed

DeBellis Musical Analysis as Articulation 125

EXAMPLE 5. Beethoven, Op. 2, No. 2, alternative interpretation

up and enriched by a reflective act in which the re-sponse is analyzed (the effects of tones described) andthe requisite intentions imputed to the composer.28

Now surely a model of musical composition andreception as communication is independently at-tractive. But the self-conception imputed toSchenkerian theory against the background ofsuch a model runs a real risk of incoherence.The question is where to locate what Snar-renberg calls the enriched, reflective act relativeto the scope of communication. If the communi-cation model is to be applicable, then the Inhaltof a musical artwork (i.e., content as revealed bySchenkerian analysis) has to be somethinggrasped by composer and intended to be trans-mitted to listener. Now the assumption seems tobe that Schenkerian analysis helps to effectuatethat transmission. But there is a dilemma. IfInhalt can be grasped only in structural hearing,then the communication model fails. For thegreat composers had no expectation that listen-ers would engage in that special perceptual ac-tivity; they had no idea that Schenker or anyoneelse would come along to provide a frameworkto cultivate it. But an utterance cannot plausiblybe taken to be in a code such that a key is evi-dently required, yet the speaker never envisagesthe hearer as having that key. It would violateany reasonable canons of interpretation to sup-pose otherwise. On the other hand, if Inhalt issomething that one does not need structuralhearing to grasp, it is unclear why anyone needsSchenkerian theory.29

Let me be clear that I do not mean this as aninvective against Schenkerianism. At the veryleast, it is valuable to have available to us criti-cally intensified ways of hearing, even if thesewere not intended as communicated content bythe composers.30 But the real challenge is inhow we are to arrive at a coherent conception ofthe purposes of Schenkerian analysis consistentwith Snarrenberg’s insight that something likecommunication does, after all, go on betweencomposers and listeners. We could just throw upour hands, of course, and say that the two do notintersect: that insofar as Schenkerian analysispromotes structural hearing, it falls outside thepurview of the communication model. But Ithink that there is a way of reconciling the intu-itions here, to allow, on the one hand, that thegreat composers were intending to transmit cer-

tain sorts of experience, and, on the other, thatSchenkerian analysis cultivates a new, specialkind of musical experience—structural hear-ing—which is distinct from, yet gives us insightinto, the former kind of experience. To fullymake sense of this is the challenge.

V

An important—and thorny—issue we must nowaddress is this: Are the structural properties wejudge to belong to musical passages, or ascribeto them in analysis, the same as the propertieswe hear them to have, in corresponding acts ofstructural hearing? (This is parallel to the ques-tion of whether the colors we ascribe to objects,or judge them to have, are the same as the colorswe see.)31 If we conceive of musical structuresalong the lines of primary qualities, then we willbe apt to identify the two with one another: theproperty we perceive in a structural hearing cor-responding to the analysis “10-10-10” will bejust the property 10-10-10 ascribed to the pas-sage by the analysis—i.e., ascribed by a judg-ment supported by that hearing. Structural hear-ing, on this conception, amounts to observation,observation of structural properties, which arethen ascribed to the passage via an analysis.

But if we conceive of musical structure insome other way, say as dispositional propertiesdefined in terms of listeners’ intentional states(where these may be either structural hearings orother states), then it will be apt to turn out, oreven be necessary in some sense, that structuresascribed in judgment are distinct from those thatenter into perceptual contents. This will be aconsequence of the need to avoid circularity, orinfinite regress, in the dispositional account.

A fuller explanation of this is in order. Con-sider the question: Which concept is prior to theother, that of a passage being a dominant prolon-gation, or that of its being heard as a dominantprolongation (i.e., being heard in a way corre-sponding to the analysis “dominant prolonga-tion”)?32 (In the remainder of this section, thenotion of a “corresponding hearing” is not as-sumed to be limited to structural hearing, butmay extend to perception more generally.)

If the notion of the structural property is thebasic one, then we will have a primary-qualityconception of the property, and will take hearingto be an intentional state directed toward that

126 The Journal of Aesthetics and Art Criticism

property. Hence the structural hearing corre-sponding to the analysis “dominant prolonga-tion” will be a representation of the passage as adominant prolongation, where the notion of arepresentation of that sort is parasitic on theprior notion of the property. On such an account,structural hearing will be of a piece with ordi-nary perceptual experience of shape properties,on most views of the latter, and of whatever elseone conceives of along primary-quality lines.

If, on the other hand, the notion of the corre-sponding hearing is prior, then that notion can-not, on pain of circularity, be that of an inten-tional state directed toward the structuralproperty. For there is an absurdity in definingwhat it is for a passage to be a dominant prolon-gation in terms of its being heard as a dominantprolongation (where we read this, now, as mean-ing an intentional state directed toward the prop-erty dominant prolongation). For then the ques-tion arises, what is it that one thus hears thepassage as? On such an analysis, to hear a pas-sage as a dominant prolongation is to hear it asbeing heard as a dominant prolongation, whichin turn is to hear it as being heard as being heardas a dominant prolongation, and so on ad infini-tum. The analysis never terminates in a charac-terization of what it is that one hears.33 What it isfor a passage to be structured in a certain waycannot be analyzed, then, in terms of its beingheard as structured in that way, if the latter no-tion is in turn that of an intentional state directedtoward that structural property; for the attemptto answer the question “What is it heard as?”lands one in an infinite regress. Thus, in particu-lar, one cannot analyze a structural property interms of a disposition listeners may have to heara passage in a certain way, if the notion of hear-ing the passage in that way is that of an inten-tional state directed toward that very structuralproperty. The concept of an F cannot, on pain ofcircularity, be defined in terms of a dispositionto be perceived as an F (reading this to mean anintentional state directed toward F).

However—and this is the importantthing—nothing precludes defining a structuralproperty in terms of an intentional state directedtoward some other property, some propertyother than said structural property; no circularityresults there. The way is thus open for atwo-tiered theory, on which a structural property(call it a “thick” property) is to be defined in

terms of a listener’s hearing a passage as havingsome other, correlative (“thin”) property.34

Thick properties, on this account, are the onesjudged to belong to musical works, on the pos-session of which the truth conditions of analysesdepend; the thin ones are perceptual contents.This allows us to take a position that is, I think,independently attractive: it allows us to say thatrival hearings of a piece have for their contentsdifferent aspects that the piece equally well has;yet it will not follow from this that we mustjudge the piece to have all of the thick structuralproperties that correspond to those hearings.

As an illustration of what a two-tiered,dispositional theory might look like, considerthe following hypothetical account of meter.One might analyze meter in terms of the percep-tion of temporal distances certain pitch eventshave to other pitch events. We never perceivetemporal relationships between all events;rather, we (or something in us) as it were “se-lects” certain relationships to perceive. Metermight then be defined in such a way that themetrical pattern assigned to a passage dependson which relationships we actually (or normally)perceive. Suppose, for example, we hear event xfollowed by event y followed by event z. Each ofthose events bears a certain temporal distance toeach of the others, of course, but we are apt notto perceive the temporal distance between eachand every pair; rather, we might perceive onlythe temporal distances that events bear to certainothers that serve as reference points. So wemight (actually or normally) perceive the inter-val in time between x and y, and x and z, but noty and z (or, at any rate, we might not perceivethat interval in the same way we perceive theothers). Then, on the hypothetical account inquestion, x would be a strong beat in the metri-cal framework. If on the other hand we perceivez’s distance from y, but not its distance from x,then y will function as the strong beat. Now thepoint is that on this account the temporal dis-tances perceived in either of the two possiblehearings all belong to the passage—the percep-tions would be veridical in either case—butwhich ones are (actually or normally) perceivedwill determine which of the two meters actuallybelongs to the passage—and, in most cases, onlyone will. In this way, metrical properties mightbe defined in terms of the selective perception ofother properties, viz., temporal relationships;

DeBellis Musical Analysis as Articulation 127

that is the idea of the two-tiered account. (I reit-erate that this is strictly for purposes of illustra-tion of how a two-tiered account might go, notoffered as psychologically real.)

As a second example, consider the construalof Schenkerian theory along the lines of an axi-omatic system.35 On this construal, one regardsthe musical surface as generable from a back-ground structure via the application of diminu-tion operations, i.e., transformational rules. Theproblem is, any given surface will be generablein lots of ways from lots of background patterns,and most of these will result in incorrect analy-ses of the passage: there is, within Schenkeriantheory, no rigorous formulation of putativetransformational rules, as far as I am aware, re-strictive enough to preclude analyses that assignincorrect background structures, or incorrect it-erations of diminution operations, to passages.36

The music analyst’s intuition has not, to date,been reduced to an expert system. Now onecould argue that this is a defect, to be removedby a more restrictive formulation of thetransformational rules. But I suggest instead thatwe can leave the rules relatively unrestrictive.Instead we can construe what it is for a passageto have a certain background pattern as its struc-tural basis, in terms of whether a listener’s per-ception of the passage, at some level, has for itscontent the property of being generable, by therelevant set of rules, from that backgroundstructure. On this account, what it is for a pas-sage to have a structural basis—the thick prop-erty—is distinct from, though defined in termsof, the perceptual content, i.e., the thin property,the property of being generable insuch-and-such way from such-and-such back-ground structure.37

On a primary-quality conception, thetruth-values of ascriptions of structural proper-ties and of correlative structural hearings are intandem with one another. Now since the MozartK. 311 passage (Example 1) is not a prolonga-tion of the pattern 11-11-10, a graph showing itas such would be a false description of it; and astructural hearing of the passage as an 11-11-10would be nonveridical, would represent the pas-sage as having a property that it does not in facthave. Graph and correlative structural hearing,on the primary-quality view, would both havethe truth-value false in this case. On adispositional, two-tiered view, on the other

hand, this close correspondence may be relaxed.It is open to us to say that, while the passage isnot an 11-11-10 but is a 10-10-10, the correla-tive perceptions are all veridical as far as theygo: each has for its content an aspect of the pas-sage, a property that indeed belongs to it.Truth-values (and truth-conditions) are in tan-dem on the primary-quality view, on whichidentity holds between structures ascribed andcontents of structural hearings—but may comeapart on the two-tiered view.

I am not altogether sure which conception ofstructure, primary-quality, or dispositional, tofavor. The primary-quality view is simpler andof a piece with an across-the-board direct real-ism that I find independently attractive. But theprimary-quality conception has the consequencethat, when we deliberate over analyses and en-tertain interpretation B of the Beethoven pas-sage, we are hearing it falsely, that is, we arehaving a nonveridical, illusory perception of thepassage, hearing it as it is not. The two-tieredconception need not have that consequence, andthat is in its favor, I would think. It seems appro-priately liberal to allow that, in analytical delib-eration, we are simply entertaining differentreadings, none of which need be maligned as il-lusory. There does not seem to be a tenable anal-ogy between interpretation B and the “bent”stick in water, say; and it is not at all clear whyveridicality must enter at this point into an ac-count of aesthetic perception and distinctions ofinterpretation. But, in any event, given the real-ist stance characteristic of Schenkerianism, nomatter whether we take the primary-quality ordispositional view, it will still turn out that onejudges the Beethoven falsely if one judges it tohave the structure attributed in interpretation B.This realism invites, I think, a certain vision ofprogress on the part of the Schenkerian commu-nity, progress toward an ever-greater accumula-tion of knowledge about the musical structure ofthe canon of masterworks.38 Now perhaps tosome (in that community) this is a hearteningthought, but I am not completely sure that suchprogress has to be, or should be, part of our con-ception of art. In any event, perhaps the choicebetween the two attitudes is ultimately an empir-ical matter, a matter of what our best cognitivepsychological theory will look like. (I do think,though, that cognitive psychologists who look tomusic theory should be careful about the distinc-

128 The Journal of Aesthetics and Art Criticism

tion between thick and thin properties.) At anyrate, I shall not come down conclusively on ei-ther side.

What the foregoing has shown is that thequestion of the relationship between the con-tents of analytical judgment and analytical per-ception is a complex one. Since it is overwhelm-ingly plausible to maintain that analysis expressanalytical judgment, complications are createdby the idea that analysis also articulates percep-tion. But I suggest we cut through this complex-ity via a useful simplification. There is latitudein whether we are to regard analysis as express-ing analytical judgment or as giving the contentof structural hearing; let us then regard analysesin the latter way. Hence, no matter whether wetake a primary-quality or dispositional view ofstructure, it will turn out that an analysis ex-presses the content of a corresponding structuralhearing (and, thus, if we also insist that it ex-presses the content of the relevant judgment,and if we have a dispositional conception ofstructure, then we will regard analyses as sys-tematically ambiguous—but what is wrong withthat?). We need not focus, moreover, on the con-cept of a correct analysis qua expression of acorrect judgment about musical structure, butmay instead simply appeal to the notion of a cor-rect structural hearing; that, or rather the con-tent of such a hearing, is what we shall take acorrect analysis to express.

We then arrive at precisely the conclusion wewant, by dint of appropriate interpretation, onetailor-made for a conception of analysis as artic-ulation: musical analyses ascribe to passages theproperties perceived in corresponding structuralhearings (i.e., structural hearings that corre-spond to those analyses).

VI

Let us now come back to Walton’s notion of ar-ticulation. What is the role of structural hearingin the articulation of one’s musical experience?My proposal is this. When we listen to a passageof music, we hear it first of all on an intuitivelevel. By this intuitive level I mean what Waltoncalls musical experience simpliciter, a level ofhearing that does not itself embody articulationbut is the object of it.39 The intuitive is theas-yet-unacknowledged level, the content ofwhich is there “all along,” waiting to be articu-

lated and acknowledged. Now the sorts of struc-tural hearings we can entertain are constrainedby our intuitive hearing. In particular—and hereis the crucial claim—a structural hearing will,normally, seem plausible to us only if its contentis identical with (part of) the content of our intu-itive hearing. Hence, an analysis that expressesthe content of a plausible hearing will also re-flect the content of intuitive hearing. In this waythe analysis articulates the content of intuitivemusical experience, precisely in line withWalton’s suggestion. Structural hearing thus hasan essential mediating role in articulation. Ouraccess to, and ability to acknowledge, the con-tent of intuitive hearing is mediated in an impor-tant way by our access to the content of struc-tural hearing.

Hence, the music analyst’s process of deliber-ation, the search for a correct analysis—whichconsists in “trying on” different structural hear-ings and choosing the most plausible—amounts(at least in an important range of cases) to find-ing a structural hearing that matches the contentof one’s intuitive hearing. Thus, in choosingamong various Schenkerian graphs, an analystis, in effect, evaluating them for how well theyarticulate intuitive content.

How are we to conceive of the relation be-tween intuitive and structural hearing on this ac-count? Is it even coherent to suppose that therecan be different ways of hearing that neverthe-less have the same content? Yes, it is indeed co-herent. The model to invoke here is that of dis-tinct beliefs, under different Fregean modes ofpresentation, ascribable via the same contentspecification. One may believe that Venus is aplanet thinking of Venus under the mode of pre-sentation expressed by “The Evening Star” orthinking of it under the mode of presentation ex-pressed by “The Morning Star”: those beliefstates are cognitively distinct, though in eithercase the content specification “Venus is aplanet” applies. In like manner, the intuitivehearing of a passage as a dominant prolongationis cognitively distinct from a structural hearingof it as a dominant prolongation, though thesame content specification, “as a dominant pro-longation,” applies to each.40

A further analogy may be drawn: to opaqueand transparent belief ascription. It is one thingto ascribe to the child the belief that the cath-ode-ray tube (reading transparently) is shiny,

DeBellis Musical Analysis as Articulation 129

another to ascribe a belief in those words (read-ing opaquely) to the scientist. The scientist, afterall, thinks of the instrument under the descrip-tion “cathode ray tube,” but the child does not.Different psychological states underlie the as-criptions. In the musical case, by analogy, theordinary listener’s pretheoretical, intuitive hear-ing of a note as a dominant is one thing, the ana-lyst’s structural hearing of that note as a domi-nant—where it is heard under the description“dominant”—quite another. The ascription ofstructural hearing is like an opaque ascription inthat the terms in which it is couched denote con-cepts the listener must possess if the ascriptionis to be true, the ascription of intuitive hearinglike transparent ascription in that the music-the-oretic concepts used to specify content are notones the listener must possess.

This goes some way toward explaining whywe find it intuitively right to say of certain anal-yses, “That’s the way I hear it!” What thismeans is not just that, say, I have come to have acertain structural hearing of the piece, one that Iendorse, or that I prefer such a hearing or find itinteresting. It is that the analysis, and concomi-tant structural hearing, are true to some way Ihave been hearing the piece all along, evenprior to the structural hearing. It is a matter offidelity to the content of intuitive hearing.

But recall that deliberation involves thechoice of a plausible structural hearing. We re-turn now to the question: Why should we thinkthat plausibility in a structural hearing is a ticketto correctness? What is the reason for thinkingthat plausibility serves somehow to indicate thatone has successfully detected which structuralproperties a passage is correctly heard as hav-ing? (On a primary-quality conception, read:“Structural properties that in fact belong to thepassage.”) As noted earlier, we need not insiston an answer to this. But one possible kind ofanswer is now in the offing: plausibility in astructural hearing is a sign of correctness be-cause it indicates congruence of content be-tween structural and intuitive hearing, and be-cause a property’s belonging to the content of(normal) intuitive hearing makes it correct toperceive in structural hearing. More baldly: weshould think that plausible structural hearingspoint to the truth because plausibility indicatescongruence of content with intuitive hearing,and intuitive hearing points to the truth; intuitive

hearing is the fons et origo, the ultimate sourceand explanation, of correctness.41 Thus, wemight conceive of musical structure in such away that facts about (normal) intuitive hear-ing—or, rather, facts about its content—are con-stitutive of facts about musical structure: whatmakes a passage a dominant prolongation is ulti-mately a matter of how it is heard on an intuitivelevel.

Of course, one might object, “Why resort tothe intuitive level at all here? If you want to givea subjectivist account of musical structure youcan define musical structure just as well withreference to the structural-hearing level, interms of what is perceived in plausible structuralhearings.” But it is not as satisfying to do so.One feels that one reaches, if not bedrock, thensomething more primitive and founda-tional—and less arbitrary—in intuitive than instructural hearing. Compare: facts about linguis-tic syntax are constituted not by theorists’ ex-plicit intuitions about structures of sentences,but by speakers’ largely unreflective capacitiesto understand and use sentences.42

To return now to the issues raised bySnarrenberg about content (Inhalt) and commu-nication, we are now in a better position to sortthings out. What falls within the scope of thecommunication model—what the composer in-tended to transmit—is intuitive musical experi-ence, not structural hearing or the articulation ofintuitive experience. The revelation of Inhalt(which revelation, on the present view, is articu-lation in Walton’s sense) is not to be understoodas the simple furtherance of the composer’s“message,” as ensuring that it gets through; it israther the revelation of what the content of thatmessage, the content of that intended experience,truly is: through a better, more articulate grasp ofthat content (than one has in that experience).The idea is that in articulation one has a certainkind of grasp of the content of intuitive musicalexperience, a grasp that is not available in intu-itive musical experience itself. This will soundparadoxical if we assume that the only grasp wecan have of the content of an experience is in thatvery experience; but not so. We must recognizethat content and experience can come apart tosome extent: the content of an experience is suchthat it can be grasped via an experience of anothertype. Perhaps it takes some hubris to claim thatany grasp of content other than that intended by

130 The Journal of Aesthetics and Art Criticism

the composer can be, in any relevant sense, abetter one; but that, for better or worse, is a fun-damental attitude behind Schenkerianism andmuch technical analysis of music.

VII

Walton’s notion of articulation leads to the pros-pect, then, of a unified picture of the intuitive andstructural ways of hearing, and of the prescrip-tive and descriptive dimensions of analysis.Analysis involves articulation, most directly, ofa structural hearing; the analysis expresses thecontent of that hearing. The structural hearing, inturn, has the same content as an intuitive hearingthat precedes the process of analysis (that “wasthere all along”). The process of deliberation,“trying on” different structural hearings and se-lecting the most “natural” or “plausible,” enablesone to arrive at a structural hearing that has thesame content as one’s intuitive hearing; and thatis why (at least in these cases) analysis has a cor-rectness condition: its purpose is to articulate in-tuitive hearing, which in turn reflects, or ratherconstitutes, the truth about musical structure.

We come, then, to a somewhat tempered un-derstanding of what it is for analysis to be “pre-scriptive”: it gets us to hear in a new way, to besure, but this new way is strongly constrained.What we have is not unbridled access to newcontents, but a new mode of access to content al-ready in place.

What seems to me, now, to be the real chal-lenge to this synthesis of prescriptive and de-scriptive functions is to make sense of it all froman epistemological point of view: to explain howintuitive hearing and structural hearing are re-lated in our cognitive economy, and how it is wehave knowledge of what the content of our intu-itive musical experience is (if we do). On thepresent view, we make a mental transition fromintuitive hearing to acknowledgment of the con-tent of that hearing. Now I would suggest that themental transition from intuitive musical experi-ence to an articulation thereof is a transition froma certain perceptual representation of a structuralproperty, involving a certain perceptual mode ofpresentation of that property, to the thought thatone hears the passage as having that prop-erty—where, in the thought, the property is givenunder a mode of presentation distinct from theperceptual one. It is now given under an m. p.

(mode of presentation) that has as constituentsthe music-theoretic concepts contained in theanalysis.43 This is unlike ordinary cases ofself-attribution of intentional states, in which, asChristopher Peacocke has pointed out, “con-ceptual redeployment” occurs. Normally, “whena thinker self-ascribes an attitude with the con-tent that summers are becoming hotter, he . . . re-deploys the very same concepts summer, hotter,and the like as he would employ in having athought about the world, to the effect that sum-mers are becoming hotter.”44 Musical articula-tion is not like that: instead of redeployment ofthe same concept (i.e., mode of presentation),one has a shift from an intuitive-perceptual m. p.(of a certain property) to a music-theoretic m.p. (of that same property).

I would argue, moreover, that there are twologically (not necessarily psychologically) sepa-rable phases to this transition: (1) a redescriptionof musical properties, i.e., a translation from anintuitive-perceptual m. p. to a music-theoreticalm. p., and (2) a reflective act whereby one ap-plies the relevant music-theoretic description asa label to one’s intuitive experience. It is the firstof these that I now want to stress. I take it thatthis translation is precisely what is carried out asone moves from an intuitive hearing to a struc-tural hearing with the same content.

The question is how we know to make thattranslation. The translation is carried out, if I amright, via the process of deliberation and selec-tion of a plausible structural hearing; but what isthe correct cognitive explanation of this—whatknowledge underlies our ability to carry out thisprocess?

I suggest that the best explanation of this abil-ity will entail that we rely on an item of back-ground knowledge such as the following: thatthe property presented under the relevant intu-itive-perceptual m. p. is identical to the propertypresented under the relevant music-theoreticalm. p. That is, we know a proposition of the fol-lowing form:

P � F

where “P” expresses an intuitive-perceptualm. p., “F” expresses a music-theoretical m. p.,and what each presents is a property (in cases ofsuccessful articulation, the same property).45

For instance: this property—the property pre-

DeBellis Musical Analysis as Articulation 131

sented in this perceptual way—is the propertydominant prolongation.

But here is the real philosophical perplexity.We have no real understanding, as far as I cansee, of how anyone knows, or comes to know,any such thing. In particular, it is entirely un-clear how such knowledge, the knowledge thatP � F, could possibly be empirical. We cer-tainly do not know it by any route of inductivegeneralization or hypothesis confirmation. For itis not as if we have some independent grasp onwhat it is for a hearing to present the propertydominant prolongation that allows us to induc-tively confirm a hypothesis of the form: hear-ings of that perceptually presented sort are onesthat present the property dominant prolonga-tion. There is a pointed contrast here withknowledge of identities such as The EveningStar � The Morning Star or Water � H2O,where we can observe (or infer the existence of)the phenomenon under each mode of presenta-tion independently, enabling us to discover theidentity. That is precisely what is missing in themusical-structure case.

But if our knowledge of the identity of intu-itively presented and theoretically presentedstructural properties is not empirical, then itmust be a priori. At the basis, then, of music-analytical activity—structural hearing, articula-tion, and music-analytical judgment—there lies,I submit, a highly interesting species of a prioriknowledge.

I think there is clearly a comparison to bemade with mathematical knowledge here, al-though I am not sure how far it can be pushed; itremains to work out the comparison in detail.But consider, for example, analytic geometry:we are capable of knowing that the perceptuallypresented shape of a circle, a shape given ingeometric intuition, is specified by the equationx2 � y2 � 1. Algebraic concepts may be used toconstruct a discursive equivalent, a “transla-tion,” of spatially apprehended shapes where acoordinate system provides a basis for measure-ment. In like manner, music-theoretic conceptsallow for the measurement and discursive repre-sentation of events and relationships presentedin auditory intuition. A meter felt by tappingone’s foot gets measured through counting:ONE, two, three, ONE, two three. . . . Intu-itively apprehended pitch distances are mea-sured and represented discursively by means of

concepts of pitch intervals (major third, perfectfifth).46

This aspect of the foundations of musicalanalysis is one in which, I submit, further philo-sophical investigation is most urgently needed.What are the character and origin of such a pri-ori knowledge of musical structure? How is itpossible that we have such knowledge? Whatpsychological account is to be given of the cog-nitive machinery of musical-analytical delibera-tion, in which such knowledge plays a role?What are the workings of the music analyst’s“intuition,” such as to generate a successful ar-ticulation of the content of intuitive musical ex-perience? I have attempted to outline a frame-work within which such questions may beapproached, but much more needs to be done inorder to work out adequate answers to them. It ison this that a fully satisfying view of musicalanalysis, and its role in articulation, will de-pend.47

MARK DeBELLIS225 West 83rd StreetNew York, New York 10024

INTERNET: [email protected]

1. Rudolph Reti, The Thematic Process in Music (Lon-don, 1961), treated at length by Peter Kivy, in Music Alone(Cornell University Press, 1990). The present discussion isindebted to, though diverges on some points from, Kivy’s.

2. Kivy, Music Alone, pp. 130–145.3. Ibid., p. 144.4. The contrast with the first, explanatory view of analy-

sis is that on the former view, what the analysis describesneed not be perceived by the listener, whereas on the presentview it must. I am indebted to David Temperley for conver-sations on this matter and follow some (though not all) of histerminology; see his “The Question of Purpose in MusicTheory: Description, Suggestion, and Explanation,” CurrentMusicology 66 (2002).

5. Fred Lerdahl and Ray Jackendoff, A Generative The-ory of Tonal Music (MIT Press, 1983).

6. Ibid., p. 3.7. Kendall Walton, “Understanding Humor and Under-

standing Music,” The Journal of Musicology 11 (1993): 39.8. From Kivy’s discussion of Reti, it is clear that

Schenker is the real figure waiting in the wings; see esp. pp.126 and 137. There is a vast literature in and on Schenkeriantheory; see, above all, Heinrich Schenker, Free Composition(Der freie Satz), 2 vols., trans. and ed. Ernst Oster (NewYork: Longman, 1979). For introduction and bibliography,see Allen Cadwallader and David Gagné, Analysis of TonalMusic: A Schenkerian Approach (New York: Oxford Uni-versity Press, 1998).

9. The purpose of this article is not Schenker exegesis,

132 The Journal of Aesthetics and Art Criticism

but to give a satisfactory philosophical account, from a pres-ent perspective, of the analytical approach derived from hiswork; it is part of the “Americanization” of Schenker (philo-sophical wing); see William Rothstein, “The Americaniza-tion of Heinrich Schenker,” in Schenker Studies, ed. HediSiegel (Cambridge: Cambridge University Press, 1990), pp.193–203.

10. I am intentionally using an edition that notates thesefigures as four sixteenth notes, the better to show the con-trast between surface and underlying structure. The moreusual, “authentic” notation that renders G as grace note itselfconveys the present analytical interpretation, which makes itless useful for present purposes.

11. Walton, “Understanding Humor and UnderstandingMusic,” pp. 41–42.

12. Ibid., p. 42.13. When in the grip of such extreme skepticism about ar-

ticulation, or music theory generally, one thinks of this pas-sage from Wittgenstein: “Imagine someone using a line as arule in the following way: he holds a pair of compasses, andcarries one of its points along the line that is the ‘rule’, whilethe other one draws the line that follows the rule. And whilehe moves along the ruling line he alters the opening of thecompasses, apparently with great precision, looking at therule the whole time as if it determined what he did. Andwatching him we see no kind of regularity in this openingand shutting of the compasses” (Ludwig Wittgenstein,Philosophical Investigations, 3rd ed., trans. G. E. M.Anscombe [New York: Macmillan, 1958], §237, p. 87e).One wants to arrive at a conception of the music analyst’sactivity on which it turns out not to be arbitrary in this way.

14. Walton says that the analyst knows the content of hisor her experience primarily through introspection (p. 39),but clearly this invites further explication of how and why,and to what extent, introspection constitutes reliable accessto one’s own mental states in this sort of case. Walton alsoallows that an analyst may ascribe intentional contents onthe basis of more general considerations about what contentsare likely to belong to a listener’s experience (p. 42); but thisleads directly to the question of how such general claims areknown to be true.

15. Felix Salzer, Structural Hearing: Tonal Coherence inMusic, 2 vols. (New York: Charles Boni, 1952), p. xvi.Salzer was a pupil of Schenker; this book was the centraltext for teaching Schenkerian theory in English for over twodecades. My use of the term “structural hearing” may be ashade different from Salzer’s, for where (he says) he uses itto denote an “approach” to training the ear, I use it to denotethe product of that training. There is as far as I know no sys-tematic, general term for structural hearing in Schenker’sown writings; in a way it is the subject of his work as awhole, and his rich terminology is oriented rather toward theobjects of that perceptual capacity: linear progressions, un-folding, and so on. The German term that comes closest is“Fernhören” (employed by Oswald Jonas and WilhelmFurtwängler; see Hellmut Federhofer, Akkord undStimmführung in den Musiktheoretischen Systemen vonHugo Riemann, Ernst Kurth und Heinrich Schenker [Vi-enna: Österreiche Akademie der Wissenschaften, 1981], p.70, and Wilhelm Furtwängler, Ton und Wort [Wiesbaden:Brockhaus, 1955], pp. 201–202), but that term tends to con-note large-scale relationships in particular. (I am indebted toWilliam Drabkin for the reference.)

16. “Anyone who has not heard music as linear progres-sions of this kind has not heard it at all.” Heinrich Schenker,The Masterwork in Music vol. II, ed. William Drabkin(Cambridge: Cambridge University Press, 1996), p. 1 (thetranslation is by John Rothgeb). Schenker’s passage is citedin William Pastille, “Music and Morphology: Goethe’s In-fluence on Schenker’s Thought,” in Schenker Studies, p. 42;Pastille argues that “Schenker’s graphs do not record theperceptions of ordinary normal hearing—not even of themost acute normal hearing. On the contrary, they record theperceptions of an elevated sense of hearing” (ibid.). (As willbecome clear, I agree with Pastille’s claim only up to a point.I agree that graphs are produced in conjunction with, and inthat sense “record,” only perceptions of the elevated sense,i.e., structural hearing. But it does not follow from this thatthey do not contain information about normal perceptions; toallow that they do both is the essential insight of the articula-tion view.)

In speaking of the ordinary listener’s musical cognition as“intuitive” I follow Ray Jackendoff, Patterns in the Mind:Language and Human Nature (New York: Basic Books,1994), pp. 167–168; also Fred Lerdahl (personal communi-cation). In using the term “special perceptual activity” (orcapacity) I have in mind Richard Wollheim’s usage, albeitfor a rather different phenomenon: see his essay “Seeing-as,Seeing-in, and Pictorial Representation,” in Art and Its Ob-jects, 2nd ed. (Cambridge: Cambridge University Press,1980), p. 217. (If pressed, I would be inclined to say thatstructural hearing is closer to seeing-as than seeing-in.)

17. Allen Forte and Steven E. Gilbert, Introduction toSchenkerian Analysis (New York: W. W. Norton, 1982), p.133.

18. The foregoing characterization of structural hear-ing—the special perceptual capacity—is not, of course,meant to be complete, but only to present the most importantaspects for our purposes.

19. This realist assumption is fairly well in evidencethroughout Free Composition and Schenker’s other writ-ings, as well as in Carl Schachter, “Either/Or,” inUnfoldings: Essays in Schenkerian Theory and Analysis, ed.Joseph N. Straus (New York: Oxford University Press,1999); Allen Forte, “Schenker’s Conception of MusicalStructure,” in Readings in Schenkerian Analysis and OtherApproaches, ed. Maury Yeston (New Haven: Yale Univer-sity Press, 1977), pp. 22–23; and throughout Forte andGilbert, Introduction to Schenkerian Analysis.

20. Some music theorists (e.g., Jonathan Kramer, in con-versation) may balk at this initial assumption, saying thatanalyses are not to be regarded as true or false, but rather asuseful or not useful, illuminating or not illuminating, etc. Itseems to me that this response scants the need for semanticaccount of music-theoretical statements, without which it isunintelligible why they should be useful or illuminating. I donot say the assumption that analyses are truth-valued is ab-solutely indispensable; perhaps an emotivist construal oftheoretical language is possible, say. But it would have to beshown how such an account could be defended.

21. Matthew Brown and Douglas J. Dempster, “The Sci-entific Image of Music Theory,” The Journal of MusicTheory 33 (1989): 96. See also Matthew Brown, Douglas J.Dempster, and Dave Headlam, “The #IV(bV) Hypothesis:Testing the Limits of Schenker’s Theory of Tonality,” MusicTheory Spectrum 19 (1997): 155–183, and, for discussion,

DeBellis Musical Analysis as Articulation 133

Joseph Dubiel, “‘When You Are a Beethoven’: Kinds ofRules in Schenker’s Counterpoint,” The Journal of MusicTheory 34 (1990): 291–340.

22. Or simply “hearing a dominant prolongation,” wherethis calls for an opaque rather than transparent reading.

23. This is something of a simplification (see ensuing dis-cussion of primary-quality vs. dispositional conceptions ofstructure). What I say here about observation may be less ap-plicable to higher (background) levels of structure than thosecloser to the surface (I am indebted to Justin London for thispoint). But I still cannot see how the background levels canbe plausibly regarded as theoretical entities, or properties,related in a perceptually remote predictive/explanatory wayto “observables,” since no one, as far as I am aware, hasclearly spelled out what the complex network of connectionsbetween background structures and observables is supposedto be.

24. Carl Schachter discusses similar examples in “Ei-ther/Or,” pp. 123–124. In the present example, for reasons ofexposition, the analysis we finally settle on is the same as theinitially plausible one; but often this is not the case: frequentlywe discard our initial reading when confronted with an over-whelmingly more compelling graph in, say, Free Composi-tion (for an interesting example of a seasoned Schenkerian’schange of view, cf. Forte’s reading of the Handel Air in B-flatin “Schenker’s Conception of Musical Structure” with that inIntroduction to Schenkerian Analysis, p. 150).

25. Perhaps sometimes we have as the basis of our judg-ment not an actual hearing or imagining, but knowledge of acertain counterfactual, viz., whether a certain type of struc-tural hearing, one we do not in fact entertain, would seem tous natural or plausible were we to entertain it. But that is notalways the case: often we do, for analyses we subsequentlyreject, engage in acts of structural hearing corresponding tothem.

26. Let me stress that what I mean to be doing in thispaper is to characterize merely one interesting subclass ofanalytical activity. “I hear the piece that way” might, whenuttered by some music theorists, have little to do with per-ception at all, functioning rather as an honorific, as endorse-ment of an analysis arrived at on the basis of purely theoreti-cal considerations, say. (I owe this point to Justin London,who, as noted earlier, thinks that this is particularly likely tohappen at higher, less audible structural levels.) Such talkmight plausibly be construed along emotivist lines. I do notdeny, moreover, that musical analysis is sometimes valuedsimply for the way it engenders new hearing (structural orintuitive), not for its being true to some way we were hearingthe piece all along.

27. Heinrich Schenker, Beethovens neunte Sinfonie (Vi-enna: Universal Edition, 1912), vii/4, cited in RobertSnarrenberg, Schenker’s Interpretive Practice (Cambridge:Cambridge University Press, 1997), p. 2 (see also pp.139–161). Schenker’s notion of content is evident in the titleof his essay “Beethoven’s Third Symphony: Its True Con-tent Described for the First Time” [Beethovens dritteSinfonie zum Erstenmal in ihrem wahren Inhalt dargestellt],trans. Derrick Puffett and Alfred Clayton, The Masterworkin Music vol. III, ed. William Drabkin (Cambridge: Cam-bridge University Press, 1997), p. 10, and figures promi-nently in the essay “Rameau or Beethoven? Creeping Paral-ysis or Spiritual Potency in Music,” in the same volume, esp.pp. 7–9.

28. Snarrenberg, Schenker’s Interpretive Practice, p. 7.29. I am not sure that Snarrenberg fully acknowledges the

difficulty. He notes, “The idea that . . . listeners . . . have topossess the same rule book as the composer if they are to un-derstand . . . the composer’s works is a theme that Schenkersounded throughout his career” (p. 142). But surely a moreplausible idea, one still fully in the spirit of the communica-tion model, would be that listeners need to possess the rulebook that composers intend them to have, rules comprisingonly a subset of those the composers themselves have mas-tery of.

30. It may turn out that these ways of hearing are ones thegreat composers themselves enjoyed, as Schenker envi-sioned; cf. Snarrenberg, pp. 8 and 143.

31. See Barry Stroud, The Quest for Reality: Subjectivismand the Metaphysics of Colour (New York: Oxford Univer-sity Press, 2000).

32. This is an instance of a general sort of question aboutthe relation between aesthetic, moral, or perceivable fea-tures on the one hand, and characteristic mental responseson the other; see Warren S. Quinn, “Moral and OtherRealisms: Some Initial Difficulties,” in Values and Morals,ed. Alvin I. Goldman and Jaegwon Kim (Dordrecht andBoston: D. Reidel, 1978), pp. 257–274. In initially posingthe question in terms of being F versus being heard “as” F, Ipresuppose no particular reading of “hears as” that wouldbeg the question (though elsewhere I favor an inten-tionalistic reading).

33. Cf. G. E. Moore (now in terms of belief rather thanhearing): “It is in all cases totally impossible that, when webelieve a given thing, what we believe should merely be thatwe (or anybody else) have the belief in question. . . . Always,when I try to state, what it is that [one] believes, I shall find itto be again merely that somebody believes . . . and I shallnever get to anything whatever which is what is believed. Butthus to believe that somebody believes, that somebody be-lieves, that somebody believes . . . quite indefinitely, withoutever coming to anything which is what is believed, is to be-lieve nothing at all” (Ethics [London: Oxford UniversityPress, 1947], p. 77). Elsewhere, Moore writes: “What I think,namely that something is true, is always quite distinct fromthe fact that I think it. . . . The object of a cognition must bedistinguished from the cognition of which it is the object”(Principia Ethica [Cambridge: Cambridge University Press,1954], pp. 132 and 141). Or, as Ralph Barton Perry puts it: Ifthe claim is that “value consists in being thought to be valu-able,” then “the fundamental difficulty with this view lies inthe fact that one would then have nothing to think about. . . .There can be no judgment about value, or anything else, un-less there is some content or object other than the act of judg-ment itself,—a judged as well as a judging” (General Theoryof Value [Harvard University Press, 1954], pp. 129–130).

34. Here, and in the illustrations to follow, it is not as-sumed that the definiendum involves structural hearing.

35. Brown and Dempster, “Scientific Image,” pp. 88–91;also Michael Kassler, “A Trinity of Essays” (Ph.D. disserta-tion, Princeton University, 1968), and “Explication of aMiddleground of Schenker’s Theory of Tonality,” Miscella-nea Musicologica (1977), cited in Brown and Dempster, pp.104–105 (n. 74).

36. I do not include Lerdahl and Jackendoff’s generativetheory, which does receive rigorous formulation, under“Schenkerian theory” here.

134 The Journal of Aesthetics and Art Criticism

37. Throughout this article I follow what I perceive to bethe generally realist stance of Schenkerian theory. But oneinteresting property of the two-tiered view is that it permitsone to be an antirealist about structure while nonetheless re-maining a realist about the “thin” properties belonging toperceptual contents. I hope to expand on this elsewhere.

38. This issue of knowledge was suggested to me by Eu-gene Narmour.

39. Again, I follow Lerdahl, Jackendoff, and others inusing “intuitive” to refer to a certain basic stratum of musicalcognition; again I suggest, parenthetically, that this is thelevel Lerdahl and Jackendoff’s generative theory is attempt-ing to capture.

40. It seems to me that Temperley, in stressing the oppo-sition between (what I call) descriptive and prescriptiveanalysis, neglects the ways in which they are compatible orcomplementary.

41. This may seem a very un-Schenkerian claim, sinceSchenker thought so little of ordinary listeners’ capacities.But remember that Schenker’s notion of real musical com-petence is essentially that of structural, not intuitive, hear-ing. And that leaves it open that the ultimate explanation ofstructural hearing, and its capacity for correctness, is rootedin more primitive cognitive abilities. (Whether this sugges-tion can actually be squared with Schenker’s own thinking issomething I shall leave to others to determine. But it is, Ithink, an idea to which theorists such as Lerdahl would beamenable.)

42. I am indebted to Christopher Peacocke for underscor-ing the importance of the linguistic analogy.

43. For more on the nature of such properties and theirmodes of presentation, see my Music and Conceptualization(Cambridge: Cambridge University Press, 1995), p. 52.

44. Christopher Peacocke, Being Known (Oxford: Clar-endon, 1999), pp. 245–246.

45. Here, I take propositions to be along the lines ofFregean thoughts.

46. I have argued along these lines already in the specificcase of Lerdahl and Jackendoff’s generative theory, in my“What Is Musical Intuition? Tonal Theory as Cognitive Sci-ence,” Philosophical Psychology 12 (1999): 495, but thepoint applies more generally to Schenker and other kinds ofmusical analysis.

It might be objected that our knowledge of the identity P� F is, while a priori, nonetheless trivial, because the verymeaning of the term “F” (e.g., “dominant prolongation”) isbound up with intuitive experiences of kind P. For—the ob-jection goes—we learn to apply the term only in conjunctionwith the articulation of experiences of that sort. This objec-tion is, I think, mistaken. It is a discovery that the propertypresented perceptually is the same as the property presentedunder a music-theoretic description; the identity is not trivialat all.

47. I am greatly indebted to Charles Burkhart, WilliamDrabkin, Joseph Dubiel, Justin London, and Kendall Waltonfor criticism and advice.

DeBellis Musical Analysis as Articulation 135