Professional Documents
Culture Documents
MARK DEBELLIS
Drawing on logical positivism and recent philosophy, Matthew Brown, Douglas Dempster, and
Dave Headlam interpret Schenkerian theory as empirical science. They view the theory as a gen-
erative system. An empirical theory of music would have the virtue of being intersubjective and
grounded in sense experience. But Brown et al. do not specify a generative system in terms explicit
or rigorous enough to yield a theory that is observationally adequate. Their test of the #IV(W)
Hypothesis, which relies on the notion of a plausible interpretation, is inconsistent with the generative-
system view. Because that view does not account for what it is for an analysis to be correct or incor-
rect, it fails as a rational reconstruction. If Schenkerian theory is empirical, then its analytical terms
should be regarded not as theoretical, but as observational. That construal is plausible, however, only
if theorists' arrival at judgments about structure is best explained on the hypothesis that the works
really do have that structure.
III
II2 MUSIC THEORY SPECTRUM 32 (20I0)
PRIMITIVE
CONCEPTS
Tonal Generated
speaker. 11 In the case of Schenkerian theory, the property of But why should we regard it as a good thing if a theory of
being tonal—or, more precisely, the property of being judged music is empirical? It is in part because empirical theories are
tonal by qualified listeners—corresponds to grammaticality. "testable" and "explanatory," but the reason runs deeper. It lies in
Brown et al.'s view is equivalent to the claim, then, that the fact that empiricism is centrally concerned with meaning.
Schenkerian theory is observationally adequate, reading "tonal" Empiricism begins with the thought that the content of our
for "grammatical." 12 ideas is derived from sense experience. Language, then, is
Notice that there are two ways in which a theory of tonality grounded in experience: the meaning of an expression is derived
could fail to be observationally adequate: it could generate a from the sense experiences that give rise to its use. Some expres-
piece that is not tonal (Example 1 [see A] ), or fail to generate sions, such as "That's red," are connected more or less directly to
one that is (Example 1 [see B]). In Brown et al.'s view of experience; others, such as "That's an electron," are indirect.
Schenkerian theory, the system does not fail in either of those It follows, then, that a scientific theory has a structure, one
ways: everything that it generates is tonal, and every tonal piece that is determined by the relation between theoretical constructs
is generated (see Example 2). Thus construed, Schenkerian such as "electron," on the one hand, and data—what is given,
theory is straightforwardly empirical: it predicts that a piece will observable, or available to sense experience, comprising the
be judged tonal by qualified listeners if, and only if, it can be theory's epistemological starting point—on the other. In the
generated by the grammar. picture of a scientific theory advanced by the early positivists, a
II On the issue of observational adequacy I follow Allerton (1979, 60-61), theory contains, in addition to data, two components: the "pure
and Botha (1980, 218). The term originated with Chomsky (1964, 28-29). calculus," or theoretical apparatus in the form of an axiomatic
I say "at least" because, as a referee for this journal pointed out, Brown et al. system, and bridge principles. 13 The pure calculus makes state-
may well conceive of Schenkerian theory as adequate in a stronger sense, ments about some class of entities posited by the theory, and
that of descriptive adequacy. It is not necessary to adjudicate the question can be thought of as "hovering" over the data (see Example 3).
because observational adequacy is implied by descriptive adequacy, and the The bridge principles connect the pure calculus to the data.
weaker notion is sufficient for present purposes.
By virtue of that connection, meaning "seeps up" from observ-
12 To be sure, Brown and Dempster (1989, 88) grant that there can be bor-
derline cases, so the claim might be more accurately stated: Schenkerian ables to theoretical terms.
theory is at least approximately observationally adequate. I pass over this
complication. Feigl (1970, 5-7); Suppe (1977, 16-27); Hempel (1970b, 142).
SCHENKERIAN THEORY'S EMPIRICIST IMAGE 113
As an example, consider electrical experiments and their re- definition in terms of observables: in P. W. Bridgman's view as
lation to physics. Physics asserts laws governing subatomic par- explicated by Frederick Suppe, each concept was to be given an
ticles such as electrons: those laws are the pure calculus. Bridge C`operational definition,"2° for example, length or mass was de-
principles allow us to infer, from those laws and auxiliary hy- fined in terms of certain operations of measurement. Over time,
potheses about electrons in a wire, conclusions about pointer this requirement was seen to be too restrictive and was relaxed
readings on a voltmeter. It is in virtue of such inferential con- to the requirement that theoretical terms be "partially inter-
nections that statements about electrons have an empirical preted" through links to observables.21 Eventually, the notion of
meaning. This example serves to illustrate the positivists' view bridge principles as distinguishable from the rest of the theory
that statements about theoretical entities are meaningful inso- was called into question. In W. V. Quine's view, which is more
far, but only insofar, as they predict and explain observables. or less a natural outgrowth of logical empiricism, we find, in-
But why were the positivists so concerned with meaning? stead of Herbert Feigl's vertical arrangement, the metaphor of a
Influenced by Wittgenstein, they thought that many apparent web or fabric with theoretical constructs at the center and ob-
philosophical problems, such as abstruse questions of meta- servables at the periphery. In Quine, the unit of meaning is not
physics, were only "pseudoproblems" resulting from the misuse the term or sentence, but the theory, which "face[s] the tribunal
oflanguage. Hence they wanted to state conditions that must be of sense experience" as a whole.22
14 Carnap (1959, 69-73). For another example, see Ayer (1946, 36). zo Suppe (1977, 18-19).
Schlick (1936, 341). See also Carnap (1936, 37); Ayer (1946, 5-16) and 21 Ibid. (22-23).
Chapter I; and Suppe (1977, 13). As Reichenbach (1938, 46-49) points 22 Quine (1961, 41-43), Quine (1960, 11); cited in Brown (2005, 15). Quine
out, there is an affinity between positivist conceptions of meaning and is to be regarded as a later empiricist, not a positivist.
those of pragmatists. 23 Ayer (1959, 12). For this notion of a tautology see Wittgenstein (1972,
16 Schlick (1936, 341). 68-69), and, for a worry about a tautology in Schenker, Kerman (1985, 82),
17 This line of thought can be traced back to Hume, who wrote, ca. 1748: "If we cited in Brown et al. (1997, 156, Note 4).
take in our hand any volume; of divinity or school metaphysics, for instance; 24 Popper (1959, §6, 17-20).
let us ask, Does it contain any abstract reasoning concerning quantity or number? 25 Popper (1989, 36).
No. Does it contain any experimental reasoning concerning matter of 'act and 26 Ibid. (34-38).
existence? No. Commit it then to the flames: For it can contain nothing but 27 Hempel (1970a, 673-74); for discussion, see Bogen (2009). Brown and
sophistry and Hume ([1748] 1993, 114, italics are Hume's). Dempster (1989, 70-71) suggest that logical positivism commits one to
18 Carnap (1936, 420-22), Hempel (1965, 3-4). phenomenalism, but it does not; cf. Carnap's remarks in Schilpp, ed. (1963,
19 Feigl (1970, 5-7). 50 and 870).
114 MUSIC THEORY SPECTRUM 32 (2010)
physical objects. Since sensations are necessarily private to, or what sort of fact is that? In what does this fact consist? What
accessible only by, the individual who has them, whereas physi- exactly do we assert or deny when we attribute a certain struc-
cal objects can be observed by several individuals, the latter con- ture to a piece? What, if anything, are theorists who disagree
ception has the advantage (as Hempel tells us) of explaining the about the structure of a piece disagreeing about? Answers to
objectivity of scientific discourse. It explains how scientists these questions are not obvious. This is evidenced by the "dark
mean the same things by their terms when they talk to each suspicions" aired by Joseph Kerman and others, 29 which point
other, and hence how they are genuinely able to communicate. out the need for clarification of the content of Schenkerian
Since there is intersubjective, or public, agreement on the mean- concepts. It is evidenced also by Joseph Dubiel's remark, "The
ing of observation terms, 28 and since the meanings of theoreti- hardest thing in this business . . . is knowing what to introspect
cal statements are rooted ultimately in observables in (at least for in order to tell whether you're having the experience that is
approximately) the same way for different speakers, the mean- supposed to go with some analytical description of a passage." 3 °
ings of theoretical statements are likewise intersubjective. Dubiel's remark suggests that the meanings of Schenkerian
Because the physical-object conception of the observational vo- terms, even if they are about experiences, are far from transpar-
cabulary explains this, I assume it in what follows. ent. Such considerations motivate a critical attitude toward
Schenkerian concepts.
not know what it is. Brown and Dempster mention Michael Brown et al.'s test produces the following results. The Chopin,
Kassler's formalization of Schenkerian theory, 34 but they pro- Beethoven, and Brahms examples can plausibly be interpreted
vide no argument that it is observationally adequate, and I have in terms of indirect relationships. (The Chopin and Brahms ex-
never seen any evidence that it is. Brown and Dempster do not amples are particularly challenging, since extant analyses by
give an explicit formulation of a generative system rigorous Felix Salzer and Allen Forte represent them as T-successions.)
enough to be tested for empirical adequacy. 35 As far as I am The Berlioz is tonal, but with qualifications. The Schubert can
aware, no one has. Fred Lerdahl and Ray Jackendoff explain be analyzed in terms of indirect relationships—though Brown
why they did not undertake the project of reconstructing et al. acknowledge that the interpretation is "strained"—or
Schenkerian theory as a generative grammar; 36 the complexity deemed ambiguous. Of the remaining examples, the Debussy
of their own system is an indication of what would be involved and Berg are plausibly interpreted in terms of indirect relation-
in the task. ships, and the Bartók and Webern are not tonal. 39 Hence, no
There is another problem. In their article on the #IV(1,V) counterexamples (except perhaps, if I read Brown et al. right,
Hypothesis, Brown et al. claim to subject Schenkerian theory the Schubert).
to empirical test, but their procedure is not in line with the But this testing procedure makes no sense if the theory is a
generative-grammar view. Brown et al. say that they are test- generative system. The notion of a plausible interpretation, in-
is that the grammar might generate too much: it might produce Here, Brown et al. invoke a supposed distinction between
a non-tonal surface, perhaps after sufficiently deep nesting of ‘`weak confirmation," which the #IV(W) Hypothesis is said to
the operations. In order to tell whether it does, we would need receive from a certain correlation of intuitions and "ease" of
definitions of the relevant operations that specify exactly which analytical discovery, and "strong confirmation," which concerns
pairs of passages belong to them. Only then would we be able psychological reality and parsing.42
to tell whether the operations are degned in a way that is inclu- This formulation betrays a basic misunderstanding of the
sive enough to allow all tonal pieces, yet restrictive enough to guiding ideas of empiricism. What Brown et al. are talking
exclude nested or iterated operations sufficiently complex to de- about are not two different ways of confirming the same hy-
stroy a sense of tonality. Doubt whether this could be done is pothesis, but two different hypotheses. One hypothesis says that
raised by the Schubert example, where Brown et al. have it both certain relationships can be found in tonal Music through anal-
ways: though the passage can plausibly be argu' ed to be non- ysis. The other says that certain relationships are grasped or
tonal, the system generates a structural description that repre- mentally represented by auditors. These are quite different hy-
sents it as tonal. This is just one example, of course, but Brown potheses, and empiricism readily captures the difference, in that
et al. give us no reason to think that the problem is not systemic. very different things would confirm them. A hypothesis about
If Schenkerian theory is a generative system, then the way to T-succession confirmed through analysis is not a hypothesis
Brown et al.'s invocation of the notion of a plausible inter- 5. SCHENKERIAN ANALYSES AS OBSERVATION SENTENCES
pretation betrays, I think, the real problem with their construal
of Schenkerian theory as a generative grammar. Their notion of It would be a good thing if Schenkerian theory were empiri-
a plausible interpretation is basically a foreign element in the cal, but Brown et al.'s construal of it as a generative system is
generative-grammar framework. Indeed, Brown points to this unsatisfactory. Can the claim that it is empirical be salvaged?
very problem: Can it be salvaged in a way, moreover, that gives a better ac-
[G]iven the complexity of most functional monotonal pieces, we have count of the notion of a correct analysis? Let me suggest the
every reason to suppose that there may be more than one way to de- outlines of such an account, although I do not know if it will
rive a particular surface from a given prototype, provided that each ultimately prove to be tenable.
derivational scheme follows the prescribed laws. In practice, however, Matters are complicated; the goals of theory and analysis are
it is clear that Schenker endorsed some derivations and not others." many and varied. What I propose to sketch is, as it were, an
The generative-system model, then, does not capture certain empirical strain in Schenkerian theory. What follows, then, is
distinctions between analyses significant to Schenker and an account meant to capture a substantial chunk of theoretical
Schenkerians. Such distinctions, or at least many of them, are activity, but not necessarily all of Schenkerianism.
not just peripheral to Schenkerian theory, but are centrally im- In order to proceed, however, we need to rethink one of
44 Brown (2005, 91). In saying that we have "reason to suppose" this, Brown 47 Brown and Dempster (1989, 97).
oddly distances himself from the theory; presumably it is a fact that can be 48 Interpretations are sometimes justified in other ways, for example, by ap-
checked straightforwardly by computation. peal to the principle of motivic parallelism or considerations of coherence.
45 Brown (2005, 91-92). But such principles are insufficient to decide more than a fraction of the
46 Brown (2005, 92) attributes to Milton Babbitt the claim that the principle cases in which there is, nonetheless, a right answer. Aural intuition is indis-
of hidden repetition is both normative and explanatory, but it is hard to see pensable.
how it can be both. 49 Quine (1960, 5-8 and 31-46).
118 MUSIC THEORY SPECTRUM 32 (2010)
• 0. • • ..• 0
*
d • dP°
(11
EXAMPLE 4. Giovanni Battista Bovicelli, Regole passaggi di musica (1594). Reproduced by permission from
Forte and Gilbert (1982, 8), 1970 W W Norton
typically rooted in exposure to paradigmatic cases, such as—I Hempel's distinction between "two levels of scientific systemati-
there are tables and chairs; but that is not science (in Hempel's claim, look to its observable consequences, to see how it is
second sense). Insofar as its constructs are limited to the obser- tested. We may be living in a post-positivist era, but there is stil
vational, then, Schenkerian theory does not ascend to Hempel's much to be learned from empiricism.
second level of scientific systematization. 56
The second reason we should hesitate to declare Schenkerian WORKS CITED
theory empirical science is this: Schenkerian analyses are plau-
sibly understood as observation reports only if the fact that an Allerton, D. J. 1979. Essentials of Grammatical Theory: A Consen-
analyst judges that the piece has a certain structure is best ex- sus View of Syntax and Morphology. London: Routledge and
plained on the hypothesis that the piece really does have that Kegan Paul Ltd.
structure. 57 This reflects a general requirement on scientific ob- Ayer, A[lfred]. nules]. 1946. Language, Truth and Logic. 2nd ed.
servation. "Round" is observational only because the best expla- London: Victor Gollantz.
nation of things looking round to us is, in part, that they really . 1959. Editor's introduction to Logical Positivism. New
are round: the judgment results from a certain causal operation York: The Free Press.
of the roundness of the object on our senses. For this reason, Beaney, Michael. 2004. "Carnap's Conception of Explication:
Gilbert Harman has argued that ethics is not empirical: it is not From Frege to Husserl?" In Carnap Brought Home: The View
DeBellis, Mark. 2009. "Perceptualism, Not Introspectionism: . 1989. Conjectures and Refutations: The Growth of Scien-
The Interpretation of Intuition-Based Theories." Music Per- tific Knowledge. 5th ed. London: Routledge Classics.
ception 27 (2): 121-30. Quine, W[illard]. V[an Oman]. 1960. Word and Object. Cam-
Dempster, Douglas, and Matthew Brown. 1990. "Evaluating bridge [MA]: Technology Press of the Massachusetts Insti-
Musical Analyses and Theories: Five Perspectives." Journa/ of tute of Technology.
Music Theory 34 (2): 247-79. . 1961. From a Logical Point of View: Nine Logico Philo- -
Dubiel, Joseph. 1996. "Hearing, Remembering, Cold Storage, sophical Essays. 2nd ed. rev. Cambridge [MA]: Harvard Uni-
Purism, Evidence, and Attitude Adjustment." Current Musi- versity Press.
cology 60-61: 26-50. Ray, Christopher. 2000. "Logical Positivism." In A Companion
Feigl, Herbert. 1970. "The 'Orthodox' View of Theories: Re- to the Philosophy of Science. Ed. H. W. Newton Smith. 243
- -
marks in Defense as well as Critique." In Analyses of Theories 51. Malden [MA]: Blackwell Publishers.
and Methods of Physics and Psychology. Vol. 4 of Minnesota Reichenbach, Hans. 1938. Experience and Prediction: An Analysis
Studies in the Philosophy of Science. Ed. Michael Radner and of the Foundations and the Structure of Knowledge. Chicago:
Stephen Winokur. 3-16. Minneapolis: University of Min- The University of Chicago Press.
nesota Press. Salmon, Wesley C. 2000. "Logical Empiricism." In A Compan-
kerian Analysis. New York: W. W. Norton. 233 42. Malden [MA]: Blackwell Publishers.
-
Hanson, Norwood Russell. 1958. Patterns of Discovery: An In- Schilpp, Paul Arthur, ed. 1963. The Philosophy of Rudolf Carnap.
quiry into the Conceptual Foundations of Science. Cambridge: La Salle [IL]: Open Court Publishing Company.
Cambridge University Press. Schlick, Moritz. 1936. "Meaning and Verification." Philosophical
Harman, Gilbert. 1977. The Nature of Morality: An Introduction Review 45 (4): 339-69.
to Ethics. New York: Oxford University Press. Suppe, Frederick, ed. 1977. The Structure of Scientific Theories.
Hempel, Carl G. 1965. Aspects of Scientific Explanation, and 2nd ed. Urbana: University of Illinois Press.
Other Essays in the Philosophy of Science. New York: The Free Wittgenstein, Ludwig. [1921] 1972. Tractatus Logico Philosoph--
. 1970a. "Fundamentals of Concept Formation in Em- ophische Abhandlung. 2nd ed. London: Routledge and Kegan
pirical Science." In Foundations of the Unity of Science. Vol. 2. Paul.
Ed. Otto Neurath, Rudolf Carnap, and Charles W. Morris.
651-745. Chicago: The University of Chicago Press.
. 1970b. "On the 'Standard Conception' of Scientific
Theories." In Analyses of Theories and Methods of Physics and
Psychology. Vol. 4 of Minnesota Studies in the Philosophy of Sci-
ence. Ed. Michael Radner and Stephen Winokur. 142-63.
Minneapolis: University of Minnesota Press.
Hume, David. [1748] 1993. An Enquiry Concerning Human
Understanding. 2nd ed. Ed. Eric Steinberg. Indianapolis:
Hackett Publishing Company, Inc.
Kassler, Michael. 1968. "A Trinity of Essays: Toward a Theory that
is the Twelve-Note Class System, Toward a Development of a
Constructive Tonality Theory Based on Writings by Heinrich
Schenker, Toward a Simple Programming Language for Musi-
cal Information Retrieval." Ph.D. diss., Princeton University.
. 1977. "Explication of a Middleground of Schenker's
Theory of Tonality." Miscellanea Musicologica 9: 72-81.
Kerman, Joseph. 1985. Contemplating Music: Challenges to Mu-
sicology. Cambridge [MA]: Harvard University Press.
Kuhn, Thomas S. 1970. The Structure of Scientific Revolutions.
2nd ed. Chicago: The University of Chicago Press.
Lerdahl, Fred, and Ray Jackendoff. 1983.A Generative Theory of
Tonal Music. Cambridge [MA]: The MIT Press.
Music Theory Spectrum, Vol. 32, Issue 2, pp. 111-120, ISSN 0195-6167,
Passmore, John. 1967. "Logical Positivism." In The Encyclopedia
electronic ISSN 1533-8339. 2010 by The Society for Music Theory.
of Philosophy. Ed. Paul Edwards. 5: 52-57. New York: Mac- All rights reserved. Please direct all requests for permission to photocopy
millan Publishing Company. or reproduce article content through the University of California Press's
Popper, Karl. 1959. The Logic of Scientific Discovery. London: Rights and Permissions website, at http://www.ucpressjournals.com/
Hutchinson. reprintinfo.asp. DOL 10.1525/mts.2010.32.2.111