You are on page 1of 24

Describing the Perception of Tonality in Music: A Critique of the Tonal Hierarchy Theory

and a Proposal for a Theory of Intervallic Rivalry


Author(s): David Butler
Source: Music Perception: An Interdisciplinary Journal, Vol. 6, No. 3 (Spring, 1989), pp.
219-241
Published by: University of California Press
Stable URL: https://www.jstor.org/stable/40285588
Accessed: 12-11-2019 05:27 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

University of California Press is collaborating with JSTOR to digitize, preserve and extend
access to Music Perception: An Interdisciplinary Journal

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
Music Perception © 1989 by the regents of the
Spring 1989, Vol. 6, No. 3, 219-242 university of California

Describing the Perception of Tonality in


A Critique of the Tonal Hierarchy Theo
a Proposal for a Theory of Intervallic R

DAVID BUTLER

The Ohio State University

Strengths and limitations of the tonal hierarchy theory, and of the p


tone testing procedure used to substantiate that theory, are discus
The tonal hierarchy theory is characterized as an important contrib
in that it begins to describe hierarchical relationships of tones in the
tonic set. The tonal hierarchy theory is, however, criticized becau
does not describe the mental process or processes by which the tona
ter of a piece of tonal music is recognized, nor does it account for the
namic perception of tonality as it unfolds during actual musical listen
The probe-tone testing procedure most often used to substantiat
tonal hierarchy theory is criticized for the ambiguity of its response
so that test results could be an artifact of effects of short-term mem
An alternative perceptual theory is proposed to describe the t
dependent nature of pitch relationships in music. In this description
teners are assumed to recognize the tonal center in tonal music on a
evidence basis, and it is asserted that the clearest evidence is carri
the rarest-occurring intervals in the diatonic set. Evidence, gathere
series of experiments, is cited to demonstrate that listeners both with
without extensive formal training in music form strong (and usually
mental representations of unambiguous tonality when tones are arr
across time so as to form meaningful tonal referents.

psychological literature on tonality and tonal harmon


on plenty of mass, weight, and diversity during the past de
spite this diversity there seems to be general agreement on t
points. First, the potential rewards of this research are consi
sic, and particularly in Western tonal music, psychologists
a highly systematic and complex nonverbal auditory doma
study human organizational capabilities and memory capaci
prove our understanding of listeners' capabilities and limitati
ers may find ways to refine current compositional materials

Requests for reprints may be sent to David Butler, School of Music, Th


versity, Columbus, Ohio 43210.

219

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
220 David Butler

or they may discover new on


port or alter their descriptio
pendent on, or independent
theorists have begun to gath
listening to augment evidenc
of introspective examinati
widely recognized that music
complex, hierarchical relatio
physical testing procedures
Shepard's (1964) ground-br
perceptual independence of p
vided solid evidence to sup
chroma helix (Figure 1), wh
1852; Révész, 1913/1954) ha
separable and valid perceptu
12 perceptual theories of pit
ness among psychologists s
could not be described adeq
nevertheless, it took Shepar
was until then based on wi
the quarter century since Sh
joined musicians in recognizi
that remain invariant under
1982a, b) to propose several g
ing a double helix, a four-dim
toroidal helix. "Heightened
found in scales, melodies, a
proximity on the skin of th
map." Adjacent horizontal po
the ascending chromatic set;
to top, trace whole-tone scal

1. Indeed, controlled experiments h


contributed through centuries of
2. It is assumed that an introduct
tone pitch system is unnecessary
(1982), Krumhansl (1979), or Ward
typically are indifferent to pitch r
numbers (B|t, C, and DU> = 0; 0
quently in the text and in figures.
3. Scales, melodies, and harmonie
same abstract level; for example,
of pitch classes to find a "melod
proper bending and stretching o
p. 377). In both cases, the maps ar
relations that are neutral with re

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
Perception of Tonality in Music 221

Fig. 1. Chroma helix: tone "chroma" (equivalent to pitch class) is represented i


dimension and pitch height (octave equivalence) is shown in the vertical dim
Drobisch (1852, p. 121).

has some important limitations. First, it is unclear how the "spec


of the intervals between adjacent members of the diatonic (major
represented by proximity to one another. Rather, it is the scales
ized as nontonal- the chromatic set, the whole-tone scale, the
scale- that are formed along straight lines or by regular ziggi
ging across the map. Also, the model conveys no distinction o
levels commonly thought to exist within the diatonic scale: speci
tonic, subdominant, and dominant are described as primary scale
cause of their positions a fifth above and below the tonic, the do
subdominant are said to be most important in defining the tonal
scale set.

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
222 David Butler

Krumhansl (1979, 1986; K


ternative geometrical model,
chical nature of perceived st
The tonal hierarchy theory d
cone) as the most stable and
next-most stable are the dom
the mediant (third member
set are pictured near the ver
tonic set are considered less
and nonmembers of the set
removed from the cone's v
archy of tonal stability thus

The psychological literature r


Krumhansl (1979), Krumha
(1982), Krumhansl & Shepard
understand the kinds of inter
in listening to tonal music. Th
description of the kinds of k
aspects of tonal organization.
ticulated knowledge system a
between chords as they are us
tions consistently reflect th
knowledge system has both w
common within-level organ
tones and chords: a tonality-s
dominating over others. Cer
these domains, with other ele
central core and to each oth
each of the three domains (to
timately tied to structure in
ciations are, thus, generative
them reside the connections
tion and between the musica
of interrelated key regions, (

The tonal hierarchy theory


lationships in Western tona
lationships (e.g., Bharucha
Kessler, 1982; Krumhansl &
tones within actual musical
of the current influence of
creasingly to validate other
perceptual studies (e.g., Sun
1987) have been shown be i
and are therefore thought

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
Perception of Tonality in Music 223

Most experimental studies of the tonal hierarchy have asked lis


rank how well various tones or chords are judged to "fit with" sh
textual patterns such as scales and major triads (e.g., Krumh
Krumhansl &c Kessler, 1982). These studies have shown that t
erarchy, as manifested both in probe-tone ratings and in statisti
butions of tones in compositions thus analyzed, tend to take
shapes shown in Figure 2. The tonal hierarchy is based on a rank
stability: both in the major and minor modes, the tonic is def
most stable member of the set; next in rank order are the do
mediant, the remaining set members, and then nonmembers of t
For all that it improves upon earlier perceptual models for mu
relations, there are several aspects of the tonal hierarchy theory
experimental procedure used most often to verify it, that are qu
bling.
The tonal hierarchy theory is static in two respects. First, it is necessarily
based on a rigid, time-independent conception of musical consonance in
which the tonic is always considered most consonant and most stable, the
dominant is next-most consonant and next-most stable, and so on. Such
a narrow account fails to describe the hierarchic nature of musical (as con-
trasted with sensory) consonance5 that allows, for example, the melodic
note C to be consonant within the key of C at one level, while forming a
dissonance- on a more local level- within the immediate context of, say,
an extended dominant chord.
Second, the model describes intervallic relationships of members and
nonmembers of the diatonic set, but does not describe how listeners rec-
ognize those relationships in situ, during the act of listening to a musical
performance in musical time.6 In a "tone profile" of a major diatonic set,

4. Krumhansl (1986, p. 17) provides a concise summary: ". . . the perceived relations be-
tween different tones can be represented by a conical surface generated by pitch proximity
and the tonal hierarchy. The relations between tones and keys are characterized by the tonal
hierarchy measured in the probe tone study. And, key distances are summarized by the to-
roidal configuration containing the circle of fifths and the relative and parallel major-minor
key relations; ... the tonal hierarchy affects listeners' abilities to recognize tones in tonal
contexts. Tones that are relatively stable in the system are better recognized than those that
are unstable. And, listeners tend to confuse unstable tones with more stable tones rather than
the reverse."
5. See Cazden (e.g., 1962, 1980) for a detailed discussion of the limits of the relationship
between sensory and musical consonance.
6. The reader who is familiar with the tonal hierarchy theory may object at this point
that the literature on the tonal hierarchy theory certainly contains references to its power
to describe how listeners attend to time-variant aspects of musical pitch, and the reader may
observe also that tests of the tonal hierarchy theory have included the perception of mod-
ulations from one key to another. The fact remains, however, that the display of the 12-
to-the-octave equal-tempered pitch set in a conical array says almost nothing about what
pitch(es) might be encountered at any given instant, and in what temporal collocation, in
an actual musical composition. Perceptual tests of the tonal hierarchy theory have indeed
used stimuli with time-order variations among their component tones, and the serious prob-
lems that attend those tests will be discussed shortly.

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
224 David Butler

Fig. 2. Well-flttedness ratings of


(b) C Minor Key Profiles (adapt

the tonic would be accor


highest; the mediant third
durations culled from a m
this tone profile quite close
we hear moment-to-mom
the theory offers no precise
itself, or recognize key m
seen in KrumhansPs (1986)
note durations compiled b
durations of the 12 pitch cla
ments Musicaux (Op. 94 N
correspondence between thi
ratings collected in the stud
One). There is some confus
tion (1986, p. 20) that G m

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
Perception of Tonality in Music 225

sicaux Number One. Hughes (1977) stated that his sums of not
indicated that the music was oriented "toward" G major althou
of the composition was C major, and indeed Schubert seemed
that the piece was in C major, as well.7 It is more important t
that note-duration tallies can vary greatly from one segment of
position to the next. For instance, the notes of the first eight m
Moments Musicaux Number 1 (Figure 3) can produce quite dis
lies.
This may be seen in Figure 4, where summed durations of the 12 pitch
classes, shown in beats, are displayed separately for measures 1-4, 5-6,
and 7-8. The pitch G [pitch class (pc) 7] is ranked highly in all three tallies,
although there is little reason to expect that analysts or listeners would iden-
tify G as the tonal center of the first eight measures (again, see Figure 3)
of the composition; in fact, the strongest tonal cues in this excerpt indicate
that the key is C major, based on a theory to be discussed later in this essay.
Moreover, there are a number of clear dissimilarities among the note tallies
shown in Figure 4, evidence that the tone profile of an entire composition
does little to describe the tones contained in any single segment of the com-
position. As a result, it seems that any perceptual theory of tonality based
upon durational weightings of pitches can be justly criticized for insensi-
tivity to moment-to-moment changes of harmony and key level that occur
in tonal music, and is counterintuitive therefore as a description of percep-
tual behavior. According to the theory that statistical distributions of tones

Fig. 3. F. Schubert, Moments Musicaux (Op. 94 No. 1), measures 1-8.

7. Briefly, the piece adheres to a ternary form as demarked by the key levels of its major
sections: the first and third sections (Measures 1-29 and 67-95) are based in C major, and
the middle section (Measures 30-66) is based in G major, although at various points in the
piece one can hear references to the keys of C minor, D major, E\> major, E minor, G minor,
and A minor. Some of these references are blatant and some are fleeting, but the tone profile
accounts only for durational weightings of pitches within the piece as a whole.

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
226 David Butler

Fig. 4. Summed durations (in beats


Schubert's Moments Musicaux (Op.
by light squares; for measures 5-

exert some control over our


a listener could identify the k
after having heard and summ
in the composition- that is,
Even if it could be argued
tones had an important effec
reason to doubt that the ton
scription of those durational
rations of tones in the very
second Musical Moment in A-
compared by pitch class, the r
did not exhibit the typical
"major key profile" (Figur
While the heavier weighting
inant orientation" that Hug
most obvious deviation from
weak showings of the median
to reconcile with expectatio
the "dominant orientation"
this profile can be redrawn f
Figure 6.

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
Perception of Tonality in Music 227

Fig. 5. Summed durations (in beats) for the 12 pitch classes found in Schuber
Musicaux (Op. 94 No. 2), graphed beginning with Al> (pc 8).

Fig. 6. Summed durations (in beats) for the 12 pitch classes found in Schuber
Musicaux (Op. 94 No. 2), graphed beginning with El> (pc 3).

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
228 David Butler

Redrawing the tone profile


the tonic (pc 3), but in the
with the tone profile: the d
either the subdominant (pc 8
(pc 1), which is not a memb
is weighted lighter still, virt
with the lowered submediant
diatonic set. In summary, th
of durational weightings of
profile possesses the power t
sic. First, neither the tally o
fine-grained enough to descr
particular regions, within a
some doubt that durational-
spond to one another in an
There is good reason to ques
tonal hierarchy, as well. Mos
have asked listeners to rank
scales and major triads fit to
hansl, 1979; Krumhansl &c
textual sequence, followed by
tated in Figure 7. It seems q
rection and contour, repea
serial positions of tones with
fluence on the perceptions of
fitting" probe tone and inhib
dicate that listeners were a
which of these (or other) at

8. One possible solution to this pr


profile, in which probe tones could
the "population" of pitches found
this be a prohibitively time-consum
reasons to be discussed in the ens
9. One of the strongest relations
classes within the two Schubert
called "dominant orientation." A t
the tonic and dominant chords wer
chords; if this were true, one could
of both the tonic and dominant cho
A personal observation here is that
reductionistic, at least as a measur
to time orders of tones is matched
example, the pitch A\> has strong h
94 No. 2, and the pitch Gjt has diff
points in the piece. Yet, one must
when compiling durational weight

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
Perception of Tonality in Music 229

Fig. 7. Example of one "tonal context" followed by a probe tone (at asteri
octave-related complex ("Shepard's") tones- which are perceptually ambig
gard to register- are typically used to generate stimulus examples in probe-ton
choice of notated register in this figure is arbitrary.

their choices. The contextual pattern in Figure 7, for instance, c


instances of the pitch C and no instances of tones outside the dia
based on C. One can infer that at least some listeners could have
their ratings on how familiar the probe tones sounded, based on
memory: in the example shown in Figure 7, it seems possible tha
macy and recency effects could contribute to higher "well-fitted
ings for the probe tone C.
The first two experiments by Krumhansl and Kessler (1982) use
textual patterns ("elements") derived from various major and min
These patterns are shown in Table 1 ; to facilitate comparison
been transposed to begin on C (Examples 1-6) or form harmon
in C (Examples 7-12).
Each contextual pattern was presented with each of the 12 tone
equal-tempered set, and listeners were instructed to rate "ho
probe tone fit, in a musical sense, with the element just heard."
and minor "key profiles" that ultimately resulted from this
were derived from average ratings given in response to only som
contextual patterns shown in Table 1. The "Major Key Profile" re
eraged ratings for probe tones following the major triad and
major-mode cadential patterns (see Table 1, Items 3, 7, 8, and
"Minor Key Profile" is drawn from averaged ratings for prob
lowing the minor triad and three minor-mode cadential patterns
Items 4, 10, 11, and 12).

10. Krumhansl and Kessler (1982) justify the omission of the data from th
minor scalar elements by stating that the average correlations between the da
elements and those from the elements that they ended up using were lower than
correlations among the elements actually used. The authors, however, presen
cerning the correlations between any of the individual elements that they e
and the average of the others, which allows for the possibility that correlatio
might have emerged in some of the instances. Further, data from yet other
dropped from the analysis without explanation.

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
230 David Butler

TABLE 1
Contextual Patterns Used
Experiment Onea

1. CDEFGABC 9. E D G

2. CDEbFGAbBC CBE

3. CEG A G C

4. C Eb G

5. CEbGb 10. C D G

6. CEGBb Ab B Eb
F G C

7. C D G

ABE 11. Ab D G

F G C F B Eb

D G C

8. A D G

F B E 12. Eb D G

D G C C B Eb

Ab G C

aScales have been transposed to begin on C, arpeggiated chords have b


to have C as their roots, and harmonic cadences have been transposed to
arranged vertically in Examples 7-12 represent simultaneities.

By summing the occurrences of pitches in the contextual pa


ements") and then plotting the sums, one can derive a grap
of the pitches presented to listeners immediately prior to the
we will call this graph a Stimulus Profile. A comparison of the
file for the major triad and three major-mode cadential pat
striking resemblance to the Major Key Profile (shown earlier i
these are compared in Figure 8. Likewise, the pitches that mak
nor triad and the three minor-mode cadential patterns may be
the sums plotted as a Stimulus Profile that closely resembles t
Profile (also shown earlier in Figure 2); this similarity is sh
9. In comparison to the Stimulus Profiles plotted from the sum
tones, both the Major and Minor Key Profiles tend to hover n
point of the seven-point ranking scale; otherwise, the contour
perfect match. It appears reasonable to conclude that this simi
a stimulus artifact in the design of Krumhansl and Kessler

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
Perception of Tonality in Music 23 1

Fig. 8. Sums of pitches used as contextual tones in deriving the "Major Key Profile
hansl & Kessler (1982, Experiment One), compared with the "Major Key Pro

rather than a reflection of mental representations of pitch relatio


is, the tone profile ratings could be the result of effects of short-t
ory for the pitches of tones presented in the contextual patterns; t
pretation is quite reasonable in light of evidence gathered in other
of short-term memory for pitch (e.g., Deutsch, 1975; Butler
1988).
There is a strong likelihood that short-term memory effects similarly af-
fected the findings reported by Cuddy and Badertscher (1987). These re-
searchers presented auditory patterns consisting of the sequences exempli-
fied in Figure 10, followed by a 0.1 -sec pause and then (at the asterisk) by
each member of the 12-tone equal-tempered pitch set, randomly ordered,
as probe tones. (All examples in Figure 10 are taken from the C major dia-
tonic set, although Cuddy & Badertscher varied the pitch sets from which
they drew their contextual patterns.) In two experiments, children and
adults with various degrees of musical experience were given the typical
probe-tone listening task of ranking on a seven-point scale how well each
probe tone "provided a musical completion" to the preceding contextual
pattern. The results obtained by Cuddy and Badertscher showed several
striking dissimilarities to the expected profile (as shown in Figure 2a). The
children rated E, F, G, and A quite similarly as completions to the major

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
232 David Butler

Fig. 9. Sums of pitches used as cont


hansl & Kessler (1982, Experimen

Fig. 10. Contextual patterns used


serial position of probe tones, w

triadic context (Figure 10b),


rating. In the scalar conditio
equivalent to A, and judge
Adults rated F higher than
the triadic contextual patte
probe tones following the sc
tent with expectations based
high ratings by both adults
and major-triadic patterns (F
when it followed a diminish
that the note C is not only a
10a and 10b, but it is the o

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
Perception of Tonality in Music 233

is not present at all in the pattern shown in Figure 10c. These resu
that listeners could have been responding on the basis of short-t
ory for in-pattern and out-of-pattern tones, and they give reaso
that both primacy and recency effects confounded response p
A related problem exists in the "harmonic hierarchy" study re
Krumhansl and Kessler (1982). Participants in that study eva
well-fittedness of probe tones that followed chordal sequences of
lengths. At the outset, listeners heard a single chord followed
probe tone; in the next trial another probe tone followed the
chord, so that eventually all members of the 12-tone equal-tem
in random order, were heard as probe tones. Listeners then
chords followed by all probe tones, then three chords and so o
chordal series that formed the context for each probe tone was n
long. The experimenters asserted that rankings of the well-fi
probe tones in this test indicated "how strongly each possible key
tation [was] felt at each point in time" (Krumhansl, 1986, p. 2
results, however, seem to show that a sizeable segment of the lis
tified the final chord in each series as a new tonic. Without h
to the test data, it would be imprudent to conjecture that list
guided more strongly by recency effects than they were by a se
It certainly appears, however, that there was a confounding of c
key in the "harmonic hierarchy" study, either by some listeners m
time or by most listeners some of the time.
To summarize, neither the tonal hierarchy theory, nor the
technique used to substantiate it, is sufficiently sensitive to the
activity of identifying, confirming, and revising one's cognitive
of pitch relationships from one musical moment to the next. Th
question that this essay raises is whether it is the purpose of cog
entists to find out how tonally enculturated listeners become per
oriented to the tonality in a piece of tonal music, or whether it is
to describe pitch relations in a precompositional entity such
chord, or pitch set. Although the extent to which the theory-bu
methodology of psycholinguistics can serve as a model for th
study of music is still unclear, it certainly is clear that major re
"deep structures" in grammar did not issue from repeated and ca
ies of the alphabet.
There is an alternative to the tonal hierarchy theory. It is pred
the recognition of critical intervallic relationships as they unfold
out the musical performance. Butler (1980) pointed out that th
tone" or "tendency-tone" successions within tonal music (i.e.,
tones found in the major diatonic set, or transpositions) car
information about local harmonic goals- both in diatonic and
harmony- and about key identity in aurally presented music,

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
234 David Butler

therefore that these critical


important perceptual cue
"structure" in music.11
Browne (1981) presented an independent, congruent account of several
important characteristics of the diatonic pitch set, demonstrating that in-
tervals in the diatonic set are found in unique multiplicities (see Figure 11);
that is, interval class 1 (the minor second and its octave complement, com-
pounds, and enharmonic spellings) occurs twice, whereas interval class 2
(the major second, octave complement, compounds, and enharmonic spell-
ings) occurs five times, and so on.
Browne proposed that it is the intervals that occur rarely in the diatonic
set- the tritone and the minor seconds- that are distinctive enough to in-
form us of the location of the tonal center as we listen to a piece of tonal
music.
Brown and Butler (1981) tested Browne's rare-intervals hypothesis by
soliciting identifications of major-mode tonal center when scant tonal
cues- such as those in Figure 12- were presented aurally to musically so-
phisticated listeners. Because the rarest interval to be found in Figure 12a
is the major second, which occurs five times in the diatonic set, listeners
might select any one of five plausible tonics: C, F, Bt , El? , or Ak The rarest
interval in Figure 12b, however, is the tritone formed by F and B; in the
context of the third tone, G, the only plausible tonic (given major mode and
absence of chromaticism) is C. Test results indicated that listeners can and
will identify tonal centers given such minimal cues, and also that there is
strong agreement on tonic when listeners are presented with rare intervals,
but agreement erodes when rare intervals are absent. Further, test results
strongly indicated that serial orders of stimulus tones affected perceptual
clarity or ambiguity of the listening patterns: levels of consensus were no-
ticeably higher when the leading tone followed the subdominant-
producing an implied subdominant- to-dominant harmonic progression-
than in the reverse case, which produced the harmonic implication of
dominant to subdominant, a harmonic succession found less often in tra-
ditional Western tonal music.
Butler (1982) found more evidence that listeners' judgments of tonal cen-
ter are strongly influenced when rare intervals are arranged differently
across time. Both musically trained and untrained listeners were swift and
confident in identifying the tonal centers of paired dyads such as those in

11. The suggestion that compositional conventions regarding the time-ordering of "ten-
dency" tones might be central to psychological theories of harmony and tonality was not
widely accepted within the community of music perception researchers in 1980, and does
not enjoy much wider acceptance now. Nevertheless, an awareness of the importance of
"tendency" tones can be found among the earliest musical theories of tonality (e.g., Fétis,
1844).

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
Perception of Tonality in Music 235

Fig. 11. Index of occurrences of the intervals of the major diatonic set when
equivalents and octave complements are excluded, after Browne (1981).

Fig. 12. (a) Trichord formed by common intervals, with five possible major-m
terpretations and (b) trichord with only one possible major-mode tonal inter

Figure 13a, but accuracy levels dropped markedly when the s


were presented in different time orders, shown in Figure 13b. Un
teners turned out to be almost as adept at this task as were traine
in fact, there was a great amount of overlap in the accuracy l
sponses between the two groups.
More recently, Brown (1988) has produced test results dem
that the same group of tones that elicits nearly unanimous ident
of one key can be temporally reordered such that it evokes stron
rence in the identification of a second key. Then, the same patter
be reordered to violate compositional conventions of implied
successions with the result that listeners exhibit little agreement
key of the pattern. For example, listeners were 86% accurate in i
the tonal center of an excerpt, shown in Figure 14a, from Movem
of Schubert's Sonata in D Major D. 664. When the excerpt w
to one instance each of its component pitches and reordered to p
identification of the original key (Figure 14b), agreement that D
increased to 95%. Brown then reordered the pitch string with th
eliciting the choice of G major as the key (Figure 14c); althoug
tinued to identify D as the tonal center, 45% identified G. W
again reordered the pitch string such that temporal orders of ra
did not imply clear harmonic progressions (Figure 14d), choic
plausible key were much more diffuse: 45% identified D, but
tifications were scattered among seven other pitches, three of w
not in the tonal pattern (indicated by whole notes).
Butler (1988) reported convergent results in a study of key
that presented listeners with beginning extracts from actu
atonal compositions. In this test, listeners heard rare-interval pat

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
236 David Butler

Fig. 13. (a) Paired dyads arranged t


in a conventional and strongly ton
ambiguous time order. From But

Fig. 14. (a) Excerpt from Schubert


of each interval in the diatonic set a
extracted from Schubert excerpt an
center; (c) same pitches reordered
tonally ambiguous. Levels of identi
centages; whole notes in (d) show o
(1988).

tracted from the first 5-sec-


atonal compositions. Tones we
rare intervals (tritones and
tones sounded simultaneou
formed unisons or octaves, m
fourths, their octave compo
alents were replaced with sile
excerpt, from Babbitt's Semi
first 5 sec of performance tim
what listeners heard. A gr
intervals- semitone (s) and
tically and horizontally align
tial arrangements of each.
Each pattern was presented 1
so that every excerpt was ma

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
Perception of Tonality in Music 237

Fig. 15. (a) Milton Babbitt, Semi-Simple Variations, Var. 1, Measures 1-3; (b
derived from the same excerpt. Simultaneous and sequential semitones (s) are
low part b; no tritones were present in the example.

Fig. 16. (a) Joseph Haydn, Piano Sonata in El» Major, Hob. XVI/52, moveme
1-2; (b) test pattern derived from same excerpt. Simultaneous and sequenti
and semitones (s) are indicated below part b.

tempered set. The 72 items on the test were intermixed and pres
each in random order. Listeners were asked to indicate whether o
tones were judged to be appropriate tonal centers for the pattern
lowed. Responses to patterns derived from the tonal selectio
strong agreement in identifying the correct tonic; in fact, levels
were as high on this portion of the test as for a pretest that solici
judgments in response to 20 unmodified examples of tonal mu
which was the same music as any test item). Listeners tended to
the most appropriate tonics of the patterns derived from atonal
were identical in pitch either to the highest tone in the pattern or
repeated most often in the pattern.
Everyday musical experience tells us that tonally enculturated
can recognize the tonal center in an unfamiliar tonal composi

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
238 David Butler

instantly, and without visibl


the role of perceptual groun
ent ease with which key reco
tual orientation must be ba
process. There is one explan
most reasonable: a perceptu
which takes the dynamic for
of this theory would be that
tonal center- until a better c
ceptual choice of most-plau
tions in the time ordering of
set; that is, minor seconds
carries with it the assumptio
gages with music by impr
unconsciously- of most-plaus
formation is at hand. This th
in the important respect that
which is recognized both wit
intervallic rivalry theory, is
text found within pitch set
to provide for every other to
assumes a second type of co
lations of tones in actual mu
tones in patterns that are m
poser's intent) to the skilled
"Tonal context" in real musi
can have several points of ha
of reference whatever in ato
ies offer compelling eviden
of tonality based on small nu
of scalar or chordal configu
pitch G, for example, what m
the perceptual pitch frame
When one hears both G and
to the At, inasmuch as half
them. To carry the illustrati
as the best candidate for ton

12. The term "rivalry" is not used


although of course this is a comm
that intervals heard in succession

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
Perception of Tonality in Music 239

semitone- the tritone- was found to exist above/below the G


identifying that as the leading tone.13
When time orders of tones within a musical composition a
tonally meaningful rare-interval patterns are not discerni
evidence response seems to prevail. Such was the apparent cas
ple, in response to the listening example (Butler, 1988) deri
atonal composition by Babbitt, illustrated in Figure 15. Althoug
were somewhat diffuse, the strongest concurrence of most
tonal center was found for the test tone Bt, the pitch of the t
most salient by virtue of its registral position and duration in
In contrast, responses to patterns derived from tonal com
proached unanimity for every pattern but one, that shown in F
this case, the strongest concurrence was that E\> (the correct t
of the composition) was the most appropriate tonic; however, a
nority held that A\> was also a good choice. An examination
will quickly show that two conflicting sets of rare intervals are
beginning of the Haydn sonata. The rare intervals of the Ah
are heard first (Measure 1), followed by two presentations of th
tervals of the E\> diatonic set (Measures 2 and 3). In both cases, t
are presented in their clearest temporal arrangement, a cadenti
in which the vertical (harmonic) tritone is linked to sequen
semitones. Music students are taught early in their undergra
of tonal harmony that this process is called tonicization- th

13. To illustrate how the listener might pick up important tonality cues f
let us examine a melodic incipit taken from the fourth movement of Beetho
7 in D Major, Opus 10 (No 3).
12 3 4 5 6 7 8 9 10 11 12 13 14

This description is much more laborious and complex


Although the key signature has been withheld, the
or plays through this excerpt will swiftly deduce that
and the experienced listener who hears (but does not
hum or whistle a D when asked to identify the tona
down tremendously and verbalized thus: the first to
center only until the listener heard T2, a semitone
might represent the first or fourth degree in major, or t
T3, positioned a major third above T2, would elimin
minor-mode tonal center, but these three tones fit
as in E and B minor. T4, T5, T6, and T7 introduce n
a major second below T7, reduces the likelihood that
mation is encountered in T9 and T10. T10 stands a
from T5 : the situation of the tritone within the conte
rule out all major-mode key interpretations other than
implication of the tritone (T5, T10) in either major o
tervallic context provided by Tl, T3, and T8.

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
240 David Butler

transposition of a dominant-
trolling tonic to a secondary
position with a harmonic no
than toward tonic. Although
important compositional p
intervallic-rivalry theory cle
Browne's (1981) description
finding"- determining the
pitches of those tones being
tion of an "orienting sch
intervallic-rivalry theory p
knowledges the critical impo
listener engages with actual
pressions of tonal center. A
tonality as a static hierarchy
discussed in this essay argu
through the explicit or impl
musical time, is intrinsically

References

Bharucha, J., & Krumhansl, C. Th


archies of stability as a function
Brown, H. The interplay of set con
nality perception. Music Percep
Brown, H., & Butler, D. Diatonic
1981, 5(6 & 7), 37-55.
Browne, R. Tonal implications of the diatonic set. In Theory Only, 1981, 5 (6 & 7), 3-21.
Butler, D. Some remarks on tonality cues and tonal stimuli. Paper presented to the Third
Workshop on Physical and Neuropsychological Foundations of Music, August 8-12,
1980, Ossiach, Austria.
Butler, D. The initial identification of tonal centers in music. Paper presented to the NATO
Conference on the Acquisition of Symbolic Skills, University of Keele, England, July 6-9,
1982. In J. Sloboda & D. Rogers (Eds.), Acquisition of symbolic skills. New York: Ple-
num Press, 1983.
Butler, D. A study of event hierarchies in tonal and post-tonal music. Poster presentation
at the Symposium on Music and the Cognitive Sciences, Institut de Recherche et Coor-
dination Acoustique/Musique, Paris, March 14-18, 1988.
Butler, D., & Ward, W. D. Effacing the memory of musical pitch. Music Perception, Spring
1988, 5(3), 91-109.
Cazden, N. Sensory theories of musical consonance, journal of Aesthetics and Art Criticism,
1962, 20, 301-319.

14. The author gratefully acknow


four anonymous reviewers and th
Ohio State University for support
study were presented to the Symp
July 1, 1988, held at the Eastman

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms
Perception of Tonality in Music 241

Cazden, N. The definition of consonance and dissonance. International Review


ics and Sociology of Music ■, 1980, 11, 123-168.
Cuddy, L. , &: Badertscher, B. Recovery of the tonal hierarchy: Some comp
age and levels of musical experience. Perception and Psychophysics, 1987
620.
Deutsch, D. The organization of short-term memory for a single acoustic attribute. In D.
Deutsch & J. Deutsch (Eds.), Short term memory. New York: Academic Press, 1975.
Deutsch, D. The processing of pitch combinations. In D. Deutsch (Ed.), The psychology of
music. New York: Academic Press, 1982.
Drobisch, M. Uber musikalische Tonbestimmung und Temperatur. 1852. In Abhandlungen
der Ko'niglich sàchsischen Gesellschaft der Wissenschaften zu Leipzig. Vierter Band: Ab-
handlungen der mathematisch-physischen Classe. Zweiter Band. Leipzig: S. Hirzel, Au-
gust 1855, pp. 3-121.
Fétis, F. Traité complet de la théorie et de la pratique de V harmonie. Brussels: Au Conser-
vatoire, 1844.
Hughes, M. [Analysis Symposium II: Schubert, Op. 94 no. 1;] A quantitative analysis. In
M. Yeston (Ed.), Readings in Schenker analysis and other approaches. New Haven: Yale
University Press, 1977.
Krumhansl, C. The psychological representation of musical pitch in a tonal context. Cog-
nitive Psychology, 1979, 11, 346-374.
Krumhansl, C. Perceptual structures for tonal music. Music Perception, 1983, 1, 28-62.
Krumhansl, C. Tonal and harmonic hierarchies. In J. Sundberg (Ed.), Harmony and tonality.
Music Publication No. 54 1987 by the Royal Swedish Academy of Music, Stockholm,
1986.
Krumhansl, C, Bharucha, J., & Kessler, E. Perceived harmonic structure of chords in three
related musical keys. Journal of Experimental Psychology: Human Perception and Per-
formance, 1982, 8, 24-36.
Krumhansl, C, & Keil, F. Acquisition of the hierarchy of tonal functions in music. Memory
& Cognition, 1982, 10, 243-251.
Krumhansl, C, & Kessler, E. Tracing the dynamic changes in perceived tonal organization
in a spatial representation of musical keys. Psychological Review, 1982, 89, 334-368.
Krumhansl, C, & Shepard, R. Quantifications of the hierarchy of tonal functions within
a diatonic context. Journal of Experimental Psychology: Human Perception and Perfor-
mance, 1979, 5, 579-594.
Neisser, U. Cognition and reality: principles and implications of cognitive psychology. San
Francisco: Freeman, 1976.
Révész, G. Introduction to the psychology of music (C.I.C. de Courcy, Trans.). Norman,
OK: University of Oklahoma Press, 1954. (Original work published in 1913 by Veit in
Lepzig.)
Rich, G. A study of tonal attributes. The American Journal of Psychology, 1919, 30, 121-
164.
Shepard, R. Circularity in judgments of relative pitch. Journal of the Acoustical Society of
America, 1964, 36, 2346-2353.
Shepard, R. Geometrical approximations to the structure of musical pitch. Psychological
Review, 1982a, 89(4), 305-333.
Shepard, R. Structural representations of musical pitch. InD. Deutsch (Ed.), The psychology
of music. New York: Academic Press, 1982b.
Sundberg, J., 8t Frydén, L. Melodic charge and music performance. In J. Sundberg (Ed.),
Harmony and tonality. Music Publication No. 54 1987 by the Royal Swedish Academy
of Music, Stockholm, 1986.
Ward, W. D. Musical perception. In J. Tobias (Ed.), Foundations of modern auditory the-
ory. New York: Academic Press, 1970.

This content downloaded from 38.102.53.132 on Tue, 12 Nov 2019 05:27:26 UTC
All use subject to https://about.jstor.org/terms

You might also like