You are on page 1of 17

Set Constellations

Author(s): Mark Hoover


Source: Perspectives of New Music, Vol. 23, No. 1 (Autumn - Winter, 1984), pp. 164-179
Published by: Perspectives of New Music
Stable URL: http://www.jstor.org/stable/832914
Accessed: 15-03-2015 09:31 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/832914?seq=1&cid=pdf-reference#references_tab_contents

You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp

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.

Perspectives of New Music is collaborating with JSTOR to digitize, preserve and extend access to Perspectives of New Music.

http://www.jstor.org

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
SET CONSTELLATIONS

MARKHOOVER

1. TRADITIONALABSTRACTMUSICAL RELATIONSHIPS

Whenwe listento music,notestendto grouptogether, eitherin linesor as


simultaneities.
Suchsimultaneities havecomeundertheterm"chord" inthe
past.A composertoday,however,is morelikelyto use wordslike"color,"
"sonority"or "set'' to describethem.
This paperwilldeal withrelationships betweenpitch-class sets(sets)and
set-types
pitch-class (set-types).However, in order to clarify musicalsig-
the
nificanceoftheserelationshipsitwillbe necessary to look furtherintowhatis
beingdescribed bythe variouswords we use to name theseentities.
The words"chord" and "set" generally describecategories ofthegrouped
notesinmusic.Butcategories ofcolorandsonority arenotusefulbecausethe
applicationofthesetermsis so context-dependent. The usefulness ofgener-

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
SetConstellations 165

alizedconceptslike"chord"and"set" is immense in allowing


us to talk
aboutandthink intermsofcategories
ofsound.
Ifwe setup a continuumwhichranges as in
fromabsolutedistinctness,
"ThatfirstchordinHortimurr'sSymphony tocomplete
lastnight," abstrac-
tion,asin"sound,"wecanposition thevariouswordsandconceptson the
continuum toa certain
andclarify oneaspectofhowtheyworkforus.
degree

major tune
TnI-type Tn-type triad C majortriad
pitchclassset color
abstraction distinctness

C H O R D sonority

Atthefarright "sonority," "color,"andespecially "tune"couldbeused


to describespecificmusical events.As we move towards abstractionwefind
"pitch-classset" and "C major triad"occupying the same space. thefar
At
leftwe findTnI-types whichdo notevendistinguish betweenmajorand
minortriads."Chord" wouldseemto occupya widerangeon thecon-
tinuumsinceitis usedto describe both"anysimultaneity" andspecific
events.Atanyrate,chords andset-types arepretty abstractthings,butwe're
usedtousingthemandweneedtousethem.Herewewillbeforfeiting dis-
inthenameofgeneralization
tinctness anddeeper-level relationships.
Relationshipsamongabstract conceptsarecapableofspawning manyreal-
andthusa single
izations, suchrelationship candescribe many musical
situa-
tionswhichmaydiffer greatlyone to theother,in some or manyparticulars.
In addition,thenature ofeventherelatively lowlevelofabstraction involved
ina pc set(orforthatmatter a pitchset)issuchthatprecisely defined rela-
tionships between setscannot assureconsistentmusical For
relationships.
ifa
example, relationship isdefined between two sets,such as thedominant-
tonicrelationship between setsA = [0,3,5,9]andB = [2,5,10] ina tonal
context,therelationship the of a
represents possibility general musical
event
whicha composer canchoosetoclarify

MP

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
166 ofNewMusic
Perspectives

orto obscure.

I 3 -

: PP

A relationship
betweensets,therefore, ishardlya completedeterminant of
themusicalsenseofallinstancesofthetwosets.In thispaperwe willbe deal-
ingmainlywithcontext-free relationships,and due to thesomewhattypical
responsemusiciansseemto haveto abstract I'm a littledefensive
relationships
about thelevelofabstractionrequiredto deal withtheserelationships. It is
hopedthatthepreceding discussionwillallayfearsand instilltrustthatthere
areindeedgemsto be foundbeneaththeintricately wovenyetsupplecloak-
and notdaggers.

2. RECENT SET RELATIONSHIPS

2.1 COMMON-TONE RELATIONSHIPS

Pitch-classsetshavebeencomparedon thebasisofcommontones,andon
the basisof sharedsubset-types.'Under the common-tonecriterion,the
amountofsimilarity betweenpairsofsetsis basedon thegreatest
numberof
commontonespossiblein a singleconstellation2formedbythetwosets.

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
SetConstellations 167

A, B, andC aresets:

A = [ 0,3,6,9 ] B = [ 0,1,3,6 ] C = 0,4,5,7

r I I T -
V I I -mmm
od
-E-

EXAMPLE 1

Underthecommon-tone criterionC andB aremoresimilar


thanareC and
whichcontainsthreecom-
A, becauseC and B can generatea constellation
withat mosttwo com-
mon tones,whereasC and A generateconstellations
montones.

To(C) = [ 0,4,5,7]
T4(B) = [ 4,5,7,10]
Threecommontones.

()

To(C) = [0 , 4,5 , 7]
TI(A) = 1,4 , 7 , 10]
Two commontones.

L
iAi

EXAMPLE 2

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
168 ofNewMusic
Perspectives

Furtherrefinements
haveindicated
thatA andB areevenmoresimilar than
areC and B, sinceundertransposition
A and B can generatefourconstella-
tions,eachwiththreecommontones,whileC and B cangenerateonlyone
constellation threecommontones.
containing

forA and B:
Constellations

T0(B) = [0,1,3,6 ] T3(B) = [ 3,4,6,9


= [0 ,3,6 , 9] = 0,3 ,6,9 ]

II
T0(A) To(A)

I
rlEwl4o n
TI
VA
tj Ti
Tv rv

T6(B) = [ 0 ,6,7, 9 ] T(B) = 0,3 ,9,10O


To(A) = [0,3,6, 9 } To(A) = [ 0,3,6,9

Ar ff
AFIL
lk Fri's f2
Ap-
Constellation
forB andC:

= [ 0,4,5,7]

r00007
To(C)
T4(B) = [ 4,5,7,10

EXAMPLE 3

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
SetConstellations 169

Developments alongtheselinesbybothRegener(1974)andRahn(1980b)
focuson theformulation of an accuratescaleof measurement forsimilarity
betweensetsbased on a combinationof maximumnumbersof common
tonesandofthegreatestnumberofconstellations maximum
containing com-
montones.
Ifwearelookingonlyforthemaximumnumberofcommontones,wewill
overlookthepossibilities
necessarily ofhighlighting an interesting
diversity
betweenthesetssuchas wouldarisefrom"harmonizing"theone common
tone0, withinthefollowingconstellation.

T0(A) = [0 ,3, 6 , 9]

TI(B) = [0,2 , 5 , 11]

If~if
lP

EXAMPLE 4

These constellations,whichcontainlessthanmaximumcommontones,
are includedin Regener'scommonnote function,3 but at mostit is only
pointedout thattheyexist.Sincepriority is placedon findingthe highest
numberofcommontones,theseconstellations dropfromthepicture.Later
we willbe takingup a specific
and exhaustive
interaction
betweensetsbased
on alloftheirconstellations
whichcontainanycommontones.

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
170 ofNewMusic
Perspectives

2.2 SHARED SUBSET-TYPE RELATIONS

Two or morecommontonesbetweensetsformspecific
subset-types.4

ForsetsX = [0,2,3,5], Y = [0,2,4] and Z = [ 0,5,6];

...I
..
.. . ...

EXAMPLE 5

The maximumnumberofcommontonesbetweenthepairsX andY is2, and


betweenX and Z is2.
The two commontonesbetweenX and Y formthe set-type[0,2], and
betweenX andZ theset-type [0,5]isformed.Merelycountingthenumberof
commontones,andcountingthenumberofconstellations whichcontainthe
maximumnumberofcommontones,does nottakeintoaccountwhichset-
typesareformedby thecommontones.Two pairsof setscould be equally
similar
basedon thenumberofconstellationswithmaximum commontones,
butnotequallysimilarbasedon thekindsofset-types
formedbythecommon
tones.
Undertransposition,eachofthetwopairsofsetsin thefollowing
example
havetwoconstellationscontainingthreecommontones.

D = [0,3,4,6 , E = [ 0,1,3,4
AND
F = [0,2,4,6], G = 0,1,3,5,6]

ForD and E theconstellations


are

1. To(D) = [0 , 3,4,6 ] 2. To(D) = [0,3,4,6]

00
T0(E) = [0,1,3,4] T3(E) = [ 3,4,6,7

Rp :R
Lh T,
I

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
SetConstellations 171

And forF and G theconstellations


are:

1. To (F) = [0,2,4 , 6] 2. TIo(F) = [0,2,4 , 10

T,,(G) = [0,2,4,5 , 11 } T,,(G) = [0,2,4,5 , 11 ]

loft

wow

EXAMPLE 6

OftheconstellationsformedbyD and E, thefirst hasa sharedsubset-type


[0,1,4]and thesecondhas thesharedsubset-type [0,1,3].Whereasthetwo
constellations
formed byF andG havea singlesharedsubset-type,[0,2,4].On
thisbasisF andG wouldbe considered moresimilar thanD andE.
Measurements ofsimilaritybasedon sharedsubset-types havebeendevel-
oped by RobertMorris,sDavid Lewin6and JohnRahn.7These measure-
mentstakeintoaccountthetypeofthesetformed bythecommontones,and
thevalueisbasedon a combination ofthenumberofsharedsubset-types, and
thenumberofinstances ofthesharedsubset-types in eachset.
Boththecommon-tone criterion
and theshare-subsetcriterionfocuson a
betweensets,andbyextension,
similarity betweenset-types. The variousscal-
ingsenableaccuratemeasurement ofthecommon-tone kindofsimilarity,or
theshared-subset-typekindof similarity.
More importantly, the processof
computingthevaluesforthemeasurements givesbountifulinformation on
how one mightmusically embodytheserelationships.

3. THE DIVERSITY IN SIMILARITY

Certainly, noticingwhatis similarbetweenthingsis a naturaland probably


necessary pointforcomparison.
starting However,therelationshipsbasedon
similaritydo notinanyway account forthoseelementsofa setnotinvolved in
the common-tonetallyor the sharedsubset search. For example, sets
[ 0,1,31and [ 0,3,6,7 1bothcontainthesubset[ 0,3 1.Ifyouwereto imag-
ine an interesting
wayofcomposingout thatsimilarity betweenthesets,the
rolesofpc 1 andofpc 7 couldbecomeveryimportant and/orinteresting since
they willallowdiverse
contexts forthe appearanceoftheshared subset[ 0,3 1.
To beginto approachthissideofsetrelationships we mightpose thefol-
lowingquestion:giventwosetsand a sharedsubset-type, howmanydistinct

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
172 ofNewMusic
Perspectives

whichmaintaina singleinstanceofthatsub-
can be generated
constellations
as
set-type a member ofbothsets?

ForA = [0,4,5,7], B =[0,2,3,5]


S = [ 0,2 ]
and thesharedsubset-type

Owl a

EXAMPLE 7

Each of theseconstellations
thenwillbe a contextforthechosenshared
subset.The sharedsubsetwillbe a focusfora pointof similarity, and the
remaining pitcheswilldeterminethediverse
contexts in which thatsimilarity
canappear.Sinceitmaybe musically fruitful
to displaya singlecommontone
betweensets,the monadicset-type[0] willbe includedas a sharedsubset-
type.
The questionagain:How manyconstellations willcontaina chosensubset
in bothsetsundercomparison?

3.1 FORMAL PRESENTATION

We will use the followingshorthandforthe conceptsand procedures


involvedin finding
boththenumberand contentofeachconstellation.

A andB aresetsundercomparison.
s is a set. However,itwillbe takenonlyas therepresentative
formofany
set-type. In addition
it must a which
representsubset-type is presentin
bothA and B forNUC > 0.

NUC(A,B,s)orsimply,NUC willequal thenumberofconstellations


gen-
eratedbyA and B whichcontaintherepresentative
formofsubsets in
bothsometransformationofA and sometransformation
ofB.
NTTO(A,s) shouldbe readas thenumberoftwelve-tone which
operations
mapsomesubsetofA ontos. Twelve-tone operations maybe definedas
the group of eithertranspositional,inversional,or multiplicative
operations.

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
SetConstellations 173

DS(A) is thedegreeofsymmetry
ofA (Rahn1980a).

intothefollowing
Byplugging valuesforNUC willbe generated.
formula,

NUC(A,B,s) = NTTO(A,s) * NTTO(B,s) / DS(A) * DS(B)

In plainEnglishthisbreaksdownto thefollowing
steps:

1. Choose setsA and B forcomparison:

A= [ 0,3,6,9 ] B= [ 0,1,3,6]

2. Choose somes as definedabove.

s=[ 0,3
3. Countthenumberoftwelve-tone whichmapsomesubset
operations
ofA onto s, and multiply
thisbythenumberoftwelve-tone
opera-
tionswhichmapsomesubsetofB ontos.

NTTO(A,s) = 8 NTTO(B,s) = 4: 8 * 4 = 32

4. Since symmetrical
setswill generateduplicateconstellations,the
degreeofsymmetryofbothsetsmustbe multiplied byone another,
andthisproductmustbe dividedintothenumbergenerated instep3
above.

DS(A) = 8, DS(B) = 1: 8 * 1= 8
32/8 =4

NUC(A,B,s) = 4.

SinceA istranspositionally
andinversionally
symmetrical,theeightopera-
tionswhichmapone ofitssubsetsonto s generateonlyone set.Thisis what
necessitatesthe divisorin the formula.The fourconstellationsthen are
formedbyapplying thoseoperations,countedin stepthree,whichformdis-
tincttransformationsof bothsets,and combiningeach transformation
ofA
witheachtransformation ofB.

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
174 ofNewMusic
Perspectives

To(A) = O, 3 , 6,9] T(A) = [0,3,6,91


To(B) = [0,1 ,3 , 6] Tg(B) = [0 , 3 9,10]

To(A) = I0 , 3 ,6, 9] To(A) = 0, 3 , 6, 9]


T31(B) = 0 , 2,3 , 9 T6I(B) = 0 , 3 , 5,6 ]

EXAMPLE 8

in example8 actuallyshare
It willbe noticedthatall oftheconstellations
threecommontones.Thisapparentredundancy bya consid-
willbe clarified
erationofallconstellations
generated byA and B as s over
ranges the represen-
tativeformsofall sharedsubset-typesofA and B. Thesesharedsubset-types
includethemonad[0], thedyads[0,3] and [0,6],and thetrichord [0,3,6].

Fors = [0 1; NUC = 8
Fors = [ 0,3 ]; NUC = 4
Fors = [0,6 ]; NUC = 4
Fors = [ 0,3,6]; NUC = 2

EXAMPLE 9

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
SetConstellations 175

Sinceanyconstellation withone ormorecommontonescouldfunction as


a contextfora singlecommontone,theconstellations generatedbys = [ 0 1
willbe a listingof all possibleconstellations
foranysetsundercomparison.
The eightconstellations forA andB aregivenbelow.

1. To(A) = I0 , 3, 6, 9]

To(B) = [0,1, 3 , 6

2. To(A) = 0 , 3 , 6, 9]

TI(B)= [0 ,2, 5 , 11

3. To(A) = 0, 3, 6, 91
T9(B) = 0 , 3 , 9,10 1

4. To(A)= [0,3,6 , 9
T6(B) = 0 , 6, 7, 9]

5. To(A) = 0 , 3 , 6 , 91
ToI(B) = 0 , 6, 9, 111
6. To(A) = [0 , 3 , 6 , 91

TI(B) = [0,1 , 7, 10

7. To(A) = [0 , 3 , 6, 9

T31(B) = [0 ,2,3 , 91
8. To(A) = [0, 3, 6, 9
T6I(B) = [0 , 3, 5, 6}

EXAMPLE 10

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
176 ofNewMusic
Perspectives

3.2 TOPOGRAPHIC CONSIDERATIONS

The constellations
givenbynumberbelowareorderedaccordingto their
to s.
relationship

s= 0 s= [ 0,3 s= [ 0,6 s= 0,3,6

3 31

4 A4

5 5

8 Igi
rg8
7 7

8 8 8 8
EXAMPLE 11

The boxedconstellation numbersindicateTn-relatedconstellations and


linesconnectconstellationswhichcanactas a contextfordifferent s subsets.
Noticethatonlytwooftheconstellations generated bys = [0](numbers 2 and
6) containonlyone commontone.
EntireconstellationswhichareTn-related mightseemredundant.After
all, anyconstellation
could be transposed and/orinverted and it would still
havethesamenumberofcommontonesand thesamesharedsubset-types.
However,ifwe considerconstellations 7 and 8 forexample,it is usefulto
knowthattheyconstitute different contextsforthesamepitchclasses,i.e. 0
and 3, butin orderto highlightthe[0,3,6}subsettheywouldmergeintoone
context.In otherwords,to viewconstellations 7 and 8 as differentcontexts
forsubset[0,3,6}wouldrequireredefining s as a set-typeratherthanas a set,
sinceconstellation7 containssubset[0,3,9] (thatis, T31[0,3,6])but not
[0,3,6),thechosenreference set.Certainlytheoptionoftransposing s iscom-
attractive,
positionally but forit to function as a focusfor similarity it must
the
"stayput" during juxtaposition process.

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
SetConstellations 177

The topography ofexample11suggests


varioushierarchical ofall
orderings
constellationsbetweenA and B. Perhapsthemostobviouswouldgivecon-
stellations
1 and 8 priority,
sincetheyformcontextsfortheappearanceofall
formsofs, andtheyarethe"representativeforms"fortheconstellations
con-
tainingthe most common tones.
The following examplemodelsa fewoptionsofthissortforsetsA and B,
and theirconstellations.

s=O0,3,6
Level1: L - /8I

Level2: [4 - 5 - 1 - 81 1
8-7 - 3
s= [ 0,6 / s= 0,3
\\

Level3: 4 -5 - 7
- - - - -
s= O0
EXAMPLE 12

The secondmodelgiveshighest to thoseconstellations


priority withfewer
commontonesandthushighest numberofpitch-classes.

Level1: 2 - 6

Level2: 1 - 3- 4-5- 7- 8

The procedures andrelationships herecouldbe appliedsystemat-


discussed
icallyatvarious structural or
levels, they couldworkto stimulatetheimagina-
tionina morerandomfashion.Atanyratethemotivation forthisworkarose
as a desireto illuminateone ofthemanypossiblepathswhichleadfromthe
recentliterature on pitch-class
setrelationships
towardsactualmusicalsurfaces
andstructures.

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
178 ofNewMusic
Perspectives

NOTES

1. Sharedsubset-types. Subset-typeswhichappearin morethanone set.


For example,it willbe saidthat thesets[ 0,1,3 ] and [ 0,2,5 ] contain
thesharedsubset-type [0,2]. More theycontaininstancesof
precisely,
theset-type
[0,2].
2. ofpitchesorpc's comprised
A pattern
Constellation. ofatleasttwosets,
in whichbothsetsremainseparate,but bywhicha comparison can be
made.Forexample,setsM = [ 0,1,2 1andN = [ 0,2,4]canbe combined
amongothers.
constellations
intothefollowing

To(M) = [0,1,2] To(M) = [0,1,2]

To(N) =0 , 2 , 4 T2(N) = [ 2, 4, 6]

= 10,11 ]
To(M) [ 0,
T, (N) = 1 , 3 , 5]

3. Regener1974. p. 202.
4. Unlessotherwisespecified typeswillbe assumed.
inversional
5. Morris1980.
6. Lewin1980b.
7. Rahn1980b.

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions
SetConstellations 179

BIBLIOGRAPHY

Forte,Allen.1973. TheStructure
ofAtonalMusic.New Haven: YaleUni-
Press.
versity
Lewin,David. 1977. "Forte'sInterval
Vector,My IntervalFunction,and
Regener'sCommon Note Function." JournalofMusic Theory21, no. 2
(1977):194-237.
1980a. "Some New Constructs InvolvingAbstract
PCSets,and
Probabilistic ofNewMusic18: 433-44.
Applications."Perspectives
1980b."A Responsetoa Response:On PCSetRelatedness."Per-
ofNewMusic18: 498-502.
spectives
Morris,Robert.1980. "A Similarity
IndexforPitch-Class
Sets." Perspec-
tives
ofNewMusic 18: 455-60.
1984. "Set-TypeSaturationAmongTwelve-ToneRows." Per-
New
spectives
of Music22: 187-217.
Rahn,John.1980a.BasicAtonalTheory.New York:Longman.
1980b."RelatingSets." Perspectives
ofNewMusic18: 483-98.
1981."Towarda TheoryofChordProgression." et
Informatique
Musique: Actes du Second Symposium International Orsay, 1981.
ELMERATTO, CRNS, 27, ruePaul Bert,94200 Ivry(France).
Regener,Eric.1974. "On AllenForte'sTheoryofChords." Perspectives
of
NewMusic13,no. 1: 191-212.

This content downloaded from 128.235.251.160 on Sun, 15 Mar 2015 09:31:06 UTC
All use subject to JSTOR Terms and Conditions