University of California Press

The Micro- and Macrostructural Design of Improvised Music
Author(s): Jeff Pressing

Source: Music Perception: An Interdisciplinary Journal, Vol. 5, No. 2 (Winter, 1987), pp. 133-
172
Published by: University of California Press
Stable URL: http://www.jstor.org/stable/40285390 .
Accessed: 08/12/2013 20:12
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 [email protected].
University of California Press is collaborating with JSTOR to digitize, preserve and extend access to Music
Perception: An Interdisciplinary Journal.
http://www.jstor.org
This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

All use subject to JSTOR Terms and Conditions
Music Perception ©1988 by the regents of the
Winter 1987, Vol. 5, No. 2, 133-172 university of California
The Micro-andMacrostructural
Design
of ImprovisedMusic
JEFF PRESSING
La TrobeUniversity,Australia
Two short pieces of freely improvised music by the same performer were
recorded in microstructural detail by the use of a specially constructed
automatic transcription apparatus. The apparatus consists of a modified
DX7 synthesizer and 2650 microprocessor which interfaces with other
computers for data processing.
The resultant music is transcribed into a modified form of traditional
notation and subjected to both micro- and macrostructural analysis. Mi-
croanalysis includes the areas of timing (interonset and duration distri-
butions, displacement, chordal spreads, etc.), dynamics (key velocity,
quantization, chordal patterns, etc.), and legatoness (relative, absolute,
pedaling). Macroanalysis uses the full panoply of devices from tradi-
tional music theory (tonal procedures, rhythmic and motivic design,
pitch class sets, etc.). Correlations between microstructural parameters,
and with macrostructure, were found to be highly significant in Improvi-
sation A, which had a supplied external pulse, but largely absent in Im-
provisation B, which had no such pulse. Where pulse was present, rhyth-
mic design was found to be based largely on pulse subdivision and
shifting.
Some performance effects (e.g., chordal spreads) operated over a time
scale of 10 msec or less. Others (e.g., synchronization to an external
pulse) showed less resolution. Differences in the distribution patterns of
interonset times, durations, and legatoness suggest three independent un-
derlying temporal mechanisms that may sometimes link together in coor-
dination with macrostructure. Quantization ("categorical production")
of some variables (interonset times, key velocities) was clearly estab-
lished. The results were also interpreted in relation to an earlier model of
improvisation (Pressing, 1987).
Introduction
Thisarticlereportsthe firstexperimentalresultsof a largeprojecton the

cognitivedesignof improvisedmusic.The approachtakenis thatof collect-
ing and interpretingmicrostructuraldata fromkeyboardperformancesre-
cordedon a speciallybuilt automatictranscriptionsystem.A macrostruc-
tural analysis of the music is performed as well, and micro- macro
Requests for reprints may be sent to Jeff Pressing, Department of Music, La Trobe Uni-
versity, Bundoora, Victoria, Australia 3083.
133

134 Jeff Pressing
correlationsdiscussed.The methodused here is noteworthyin that it fea-

turesthe applicationof scientific(reductionistic)analysistechniquesto a
creative and expressive phenomenon, "spontaneous"human aesthetic
action. The centralinterrelatedquestionsof the projectare consideredto
be:
1. How is improvisedbehaviorproduced?
2. How is improvisedbehaviororganized?
Thereis plenty of precedentfor the use of automatictranscriptionde-
vices in music. Ethnomusicologists,with the melographand other ma-
chines,have recordeddetailedacousticspectraof (primarilynon-Western)
musicin performance,with the primarygoal beingthe resolutionof certain
finedetails(intonation,optimalframeworkfor rhythmicsubdivision,etc.)
in the productionof transcriptionsof the music (e.g., Hood, 1982). Me-
chanicaltranscriptiondeviceshave also been used for Africanpercussion
music(Jones,1959). However,the automaticrecordingof microstructural
performancedetail,which looks at actionswith a resolutiontoo fineto be
includedin normal scores (or notated by repeatedlistening),has largely
formeda separatepsychologicaltradition,beginningwith the pioneering
work of Seashore(1938). He made photographicrecordsof the action of
thehammersinsidea piano, andwas ableto determinekeynumber,relative
noteintensities,andonset andoffsettimesto a resolutionof 10 msec.Other
morerecentstudieshave focusedon timing,usinga varietyof transcription
techniques.Gabrielsson(1974, 1982) has investigatedtiming(accuracy±5
msec) in the performanceof monophonie music, and MacKenzieet al.
(1986) studiedthe effect of tonal structureon rhythmin piano perform-
ance.Studiesof the performancesof specificpiecesusingoscillogramtech-
niquesincludePovel (1976) on the BachC majorPrelude(onset accuracy
1-2 msec,less accuratefor durations),andMichon (1974) on Satie'sVexa-
tions (accuracy1 msec). Tuckeret al. (1977) and Lamb (1978) have re-
porteda keyboard-basedgraphicalnotationdisplaysystemused for teach-
ing. Rasch (1979) measuredthe synchronizationof performedensemble
music (resolution5 msec), finding typical spreadsof onset times in the
range30-50 msec.
The experimentalapparatuscapableof greatesttemporalresolutionis
that of Shafferat the Universityof Exeter.This is a grandpiano-basedsys-
tem capableof accuracyto the orderof microseconds,althoughin a series
of articles(Shaffer,1980, 1981, 1984a, 1984b; Sloboda,1983) examining
microstructuralissues in the interpretationand timing of a number of
piecesof standardpiano repertoire,data are reportedonly to millisecond
resolution.
The issue of what degreeof temporalresolutionis necessaryfor such
studiesis answeredin the firstinstanceby the work of Smalland Campbell
(1962). They found a temporaldifferentialsensitivityof at best 1-2 msec

Micro- and Macrostructural Design 135
for clicks,with largervalues for other kinds of sounds.This has therefore

beentakenas adequatetemporalresolutionin the presentstudy.Two reser-
vations should, however, be noted. First,their data treatperception and
thecurrentstudyinvestigatesproduction.Althoughtheseprocessesseemto
sharecomparabletimeframesin a numberof situations(Pressing,1987), it
remainsunprovenhere.Second,it is possiblethat finertemporaleffectsare
in someway significantbutremainsubconscious(i.e.,not accessibleby lan-
guageor otherconsciousmethodsof givingreference).
In this firststudy, the designof the apparatusand the methodsof data
processingare described.The detailedanalysisthat follows treatsexhaus-
tivelytwo contrasting"free"improvisationsby the sameperformer.Subse-
quentpublicationswill look at referent-basedimprovisations(seePressing,
1984 for a discussionof the use of the referentin improvisation),rhythmic
studies,andgroupingeffectsandwill featurethe work of a numberof other
performers.
Methods
APPARATUS
The experimental apparatus consists of a modified 61-key Yamaha DX7 synthesizer, an

accompanying 2650 microprocessor-based hardware unit, and audio playback equipment.
This enables the recording and playback of key onset, key offset, key velocity, and controller
information. (Controllers are ancillary performance controls and include pitch bend and
modulation wheels, aftertouch, sustain pedal, and breath controller). MIDI information has
not been used directly, as extensive testing and calibration of the apparatus revealed that
such information from the DX7 contains inadequate temporal resolution. For example,
groups of 4 to 8 keys mechanically constrained by tape and wood to produce simultaneous
depression revealed random delays in MIDI key onset times of up to 30-40 msec. There-
fore, parallel data are read directly off the keyboard subprocessor, producing a temporal
resolution of 1-2 msec. This was verified by direct oscilloscope comparison of the timing of
key data and audio output. As a further check, direct MIDI comparisons were made (for
simple melodic material that should not exhibit the polyphonic delays cited) with the output
of a Yamaha QX1 digital sequence recorder, which also confirmed the indicated accuracy.
Unneeded controller information from the keyboard is filtered out by the apparatus, and
relevant data stored in bank memory, to a total of 64K of RAM. These data are then up-
loaded to a PDP 11/23 and Macintosh Plus for interpretation, plotting, etc. The apparatus
can also produce a variable tempo click track for synchronization.
The disadvantages of this system over that of a wired-up grand piano are that it has only
61 keys and that its organ-type keyboard and synthesized sounds can be alienating to a mi-
nority of potential performers. These have not proved to be major stumbling blocks. The
advantages, which are considered to more than compensate, include portability of the sys-
tem, ease and cheapness of construction, and particularly, the flexibility of a synthesizer-
based system. Three possibilities are particularly appealing in this regard. First, the effects of
differentvoices with different envelope and spectral characteristicson playing technique can
be examined. Second, performance detail on non-keyboard instruments and the voice can be
recorded by the use of a pitch-to-MIDI conversion device such as the Fairlight voicetracker
or the guitar synthesizer. Third, musical processes such as pitch glides and bends, which are
impossible to execute at the piano, can be documented and subjected to analysis by using

136 Jeff Pressing
built-in or external controllers. Variations in intonation may also be studied. These poten-
tials will be explored in subsequent publications. With a minimum of adaptation the system
can also be made to function with a weighted 88-key master keyboard such as the Yamaha
KX88, removing most of the cited disadvantages.
EXPERIMENTAL PROTOCOL
A number of experienced keyboard musicians of recognized professional standing as im-

provisers in Australia have performed with the system. In practice this has meant almost
exclusively musicians from the jazz or classical organ traditions. The experimental protocol
was as follows. The subject, after a period of keyboard and room environment familiariza-
tion, is asked to improvise under various conditions. These include free improvisation, vari-
ations on given musical materials, and playing with or without a click track. Key data and
audio recordings are made. Immediately after each improvisation, the performer is asked for
an aesthetic evaluation of the piece and to give comments on any thinking processes going on
during it that he or she can recall. Where possible, these recollections were noted in relation
to corresponding locations in the piece, producing a timing track of rememberedimagery or
(occasionally) control processes.
Full analysis of the resultant music is quite time consuming, and this article examines
only two short pieces improvised by the author, with Gregory Troup of the Department of
Psychology acting as the experimenter. His help is gratefully acknowledged.
The two pieces were played with an electric piano voice (E. PIANO 1), and were chosen
from 11 short improvisations recorded in one session in December 1985. They were chosen
as being of contrasting styles, and reasonably free of aesthetic problems, as judged by the
performer. These were the third and fourth improvisations of the session, of respective dura-
tions 55.716 sec and 97.790 sec, and they will be labeled respectively Improvisation A and
Improvisation B. Improvisation A used no sustain pedal and was played to a steady metro-
nome click of J = 60. B used the sustain pedal without a click track. Both improvisations
were free.
It may be objected by some readers that self-analysis of the kind featured in this study
necessarily lacks the foundation for the independence of judgment critical to scientific re-
search. My disagreement with this is unequivocal. First, the analyst needs all the help (here it
is familiarity with the musical style) he or she can get in this difficult task of applying analyti-
cal procedure to creative artistic output; second, the microprocesses analyzed are nearly all
subconscious and hence new to the conscious mind of the analyst in any case, even though he
is the same person; and third, the data were actually collected by an independent experi-
menter. It should also be emphasized that considerable analysis of work by other performers
has been done and will be reported in subsequent publications by this author and Greg
Troup.
DATA FORMAT
The standard data recorded comprise key velocity on a scale of 1 to 127, and key onset,
key offset, and sustain pedal times all to the nearest millisecond. The microprocessor feeds
these data to a PDP 11/43 and a numerical printout is obtained, with automatic calculation
of note durations, likely note duration categories, legatoness, and other parameters. A pre-
cise bar graph in what is effectively proportional notation, with width indicating note dura-
tion, midpoint indicating pitch height (key number), and thickness indicating key velocity, is
also output. The MIDI data representation (as mentioned above, less accurate- particularly
for chords) can be used to drive a Personal Composer software program that automatically
produces an approximate rendition of the piece in traditional notation, which can be useful
as an intermediate step in the production of final precise notation. This last technique was
not used for the pieces described in this article.
Key velocity (KV) data require special comment. All DX7s put out a reduced range of key
velocities relative to other synthesizers. The maximum KV value output by this DX7 was

120, with values above 117 extremely rare regardless of the force applied to the keys. This is
typical. Typical apparently also is a certain fine-grained artifactuality, whereby, for exam-
ple, on this DX7, even KV values are extremely rare in the ranges 94-98 and 112-114, and
odd ones equally rare in the ranges 85-87 and 103-107. This irregularityled to a perceived
need for calibration of the KV data relative to an independent measure of key transit time.
Such a calibration was therefore performed, where, for a large number of random key
strokes, KV data were recorded in conjunction with independent measurement of key transit
time, using an oscilloscope. The results show that KV is roughly linear with measured key
transition time, as expected. An estimate of the range of error in the measured KV values was
obtained as follows. For individual representative KV values across the full dynamic range,
the corresponding weighted distribution of key transit times (KTT) that could have gener-
ated the chosen KV value was noted. The probability that various KV values would be pro-
duced from this distribution of KTTs was then read off the calibration graph. This gives the
error distribution around the input KV value. When this is done, it is found that the error in
KV values is ±2-3 units at at least 80% confidence level throughout the full dynamic range
employed. The error is actually less than this at lower dynamic values (^90).
A final point concerns the relationship of key velocity to sound level. This will vary with
the synthesizer voice, since velocity sensitivity is fully programmable in the DX7 FM al-
gorithms. A calibration of a voice of the type used in this study (PIANO 1) was therefore
performed, using a Bruël and Kjaer Impulse Precision Sound Level Meter, Type 2209, accu-
rate to ±0.5 dB. The results of this calibration are displayed in Figure 1 . Obviously the abso-
lute sound level is dependent on the amplifier gain.
Results and Analyses
The primarydata are analyzedseparatelyfor ImprovisationsA and B.

The analyseslook systematicallyat the followingissues:
1. Macrostructure(traditionalmusicalanalysis):the organiza-
tion of pitch,rhythm,phrasestructure,basicarticulation,ba-
sic dynamics,texture,etc.
2. Microstructure: delays,anticipations,interonsetandduration
distributions, chordal spreads, dynamic detail, legatoness,
quantization.
3. Correlations:correlationsand groupingeffects between mi-
crostructuralparametersand in relationto macrostructure.
Fromthis examinationa numberof inferencescan be drawn,concerning
such things as event partitioning(describedbelow) and the resolutionof
expression(bywhichis meantthe boundarybetweenintendedaestheticef-
fect andmechanicalsystemnoise- cf. Sloboda,1983).
IMPROVISATIONA
Figure2 shows the beginningsof the data file representingthis piece.

Column1 recordsthe eventtype:eithera metronometick (atJ = 60), a key
number (the 61-key keyboard is numberedfrom 0 to 60, with middle
C = 24), or a rest. Keyvelocityis in column2, and event onset and offset

138 Jeff Pressing
Fig. 1. Decibels versus key velocity (KV) for the voice PIANO 1.
timesarefoundin columnsthreeandfourrespectively.Succeedingcolumns
providenote durationin seconds,durationin percentageof the beat,closest
musicallynotateddurationwithin a frameworkof dupleor triplesubdivi-
sion, errorin this notatedapproximation(firstin secondsand then in beat
%), note displacementfrom the beat (firstin secondsand then in beat %),
interonsettime, absolute legato (overlapin seconds between the current
andpreviousnotes), and a relativelegatonessfactor(definedbelow).
Figure3 shows an excerpt from the computer-producedproportional
notationrepresentationof the piece.Pitchheightis on the verticalaxis, time
on the horizontal.Each note is representedby a black rectangularregion
whose length gives its duration,thicknessthe key velocity, and midpoint
the key number.
Macrostructural Analysis
The analysishere is based on a reductionof the complete data into a
slightlymodifiedform of traditionalnotation, as shown in Figure4. This
was done by hand.The legendexplainsthe use of specialsymbols;primär-

t
i <
i «ti
V)
I
I
1
CO
Uh

140 Jeff Pressing
C/5
§
q
r^
o
oo
q
•2'Ì
x s
B °
S§
"C O
^.i
1^
Is
1§
2 a
1
1.2
g
l§
5-.»
1 O <u
S .S

ily theygiveinformationon note displacement[10-30 msec = (- ); > 30

msec = - ], key velocity,chordalspread(in msec),and articulation.The
use of 30 msecas a naturalboundarybetweenqualitativelydifferentinten-
tions or perceptionsof note displacementwas suggestedfrom our own
data, but is also supportedby other experimentalwork (Rasch,1979). In
describingthe piece,metronometick (beat)locationswill be used,with the
piecebeginningwith a pick-upto beat 10, as indicated.
The piece provedto be fundamentallyin two parttexture,with one ex-
tendedpassagein threevoices (beats16-25), and briefchordalexcursions
in four or five parts (beats38, 41, 51, 63, 64, and 65). Contrarymotion
predominatesbetweenparts.The rhythmicdesignshows little syncopation
andonly raredeviationsfromtripleteightnote beatsubdivision,notablyin
thelast threebeats,wheresuddenduplesubdivisionhelpsdefinean ending.
Phrasestructureis not specificallygivenin Figure4, whereslurmarksindi-
cateonly legatoness,but it is clearlyasymmetricaland may be parsedup as
follows: beats 10-14; 15-25; 26-29; 29-36; 37-42; 42-45; 46-51;
52-57; 58-59; 60-65.
Analysisof the pitch materialsshows a numberof features.Firstof all,
althoughtraditionaltonal materialsarenearlycompletelyabsent,thereis a
tonicizationof the note D. This is supportedby the following:
1. D is the most frequentlyoccurringnote in the piece, followed

by G and A respectively,outliningimportantstructuralfunc-
tions of 4 and 5.
2. D is the firstnote of the piece.
3. The lowest octavebasstones areD, FJ,A, Cfl,and (onceonly)
G, which have bass note primarychord functionsin the key
ofD.
4. The last two chordsconstitutethe following half-cadencein
jazzterminology:I^-S-V7*9"5.
Second,therearedistinctiverecurringmotivesandintervallicsequences.
Most noticeableperhapsare threerelatedorderedsets, occurringas indi-
catedin Figure5, foundprimarilyin the bass.Thesemotives(orderedsets)
consistof a minor6th, 5th, or tritone,followedby a perfect4th. In interval
numbernotation, this class of three motives could thus be representedas
[(678),5]. In action representationthis generalizedmotive can be seen to
correspondto the left handfingering521, with slightlyvariablestretchbe-
tweenthe forefingerandlittlefinger.Thisraisesthe possibilitythat the mo-
tive class is stored to a significantdegree in kinestheticrepresentation,
ratherthan aural.

142 Jeff Pressing
J = 6° [4msec]
, ,
ft ^T^? <^TTj>. («">7\
==
^ 93 tr; F Ïo3 ,05
DX7 102 ^_
<- ^
104
99
104
concurrent 34 93 102 93
tncnordUl4
trichord
014
thoughts
2 handdescendinggesture fourths
[14 msec] [5 msec] •^"^(->}<-i [25 msec]
C) 102 (<-) (<-) ^ î- L26^"l08 V ff99 98 I 101
10477 82 94 109 H ^ ^ J- 95 I
»' io
102 -Jq ~^^^^™i 106 I 103
*83 ^9-
w ^-97^ 93 6 llOO
102 T T trichord
contrary chordscorrect? grabbedat
motion last moment
[2 msec) [13 msec] [8 msec] [12 msec] [17 msec]
Jffi*> ^ fiL»-| i s ?a
*" 108 m 108 109

1081 10 108109 '09 106~ (?TlO0 ^=p ^^
-
|- ^^ f03
106 104 103 '

93 4g 56 ^-93 „{ I
*ï**g
repeat continue contrary streams to I 1
j
motive contrary motion dissolve activity
[10 msec]
[26 msec] I 3 1
«J 104 in? (->) ÏÔÏ 7^
\.. l^m .^-n, . | ^ («-) «-> ff /r^rff

104 101 R- I ' W
100
C j I' 70
3
41 loi 108HI I1
100 111 I1 i
106 95 92

Fig. 4. Transcription of Improvisation A (duration 55.716 sec). Instrument: DX-7, Voice:
E. PIANO 1. Metronome tick « = 60. [ ] =Chordal spread in msec; numbers near note-
heads give the order of striking. - , ~+- are used to show significant displacements, of at
least 30 msec from the indicated attack points, or, in a few case, of > 10% of the duration
(for brief notes). (-#-), (- ) are used to show slight displacements, in the range 10-30
msec. Displacements of less than 10 msec are not indicated. N.B. The use of such displace-
ment symbols on chords of three or more notes refers to measured displacement of the cen-
troid. ( p) refers to a ghost tone of very low dynamic (often inaudible). Key velocities in MIDI
units are listed below notes. In the case of chords these are stacked vertically in the same
order as the notes in the chord, f, the staccato dot implies only a slight shortening. Slurs
indicate legatoness and not necessarily phrasing. Recorded 10 December 1985 at La Trobe
University.

144 Jeff Pressing
A
* BeatLocations
%yi
jf n 25-26: permuted,D octavedisplaced
\* 37-38
45-46
50-51
52: transposedup a 5th (C(tAD)
63
B
i\\ - 28-29
S 9m 38: containedin r.h. chord
m
-_ 38-39: transposedup a 4th, retrograded,in bass
40-41
43-44: 8va, permuted,in bass
48-49: 8va, retrograded
63-65: top voice, transposed3 octaves +
semitone,intervallicallyinverted
C
fry , 13: transposedup 8va + 4th
JA * \ m= 18
üZZ 41 : transposed, intervallically inverted, rearranged,
IT*" in treble
43: transposed up 8va + M2nd, in bass
47-48: transposed up 8va + M2nd, in bass
51, 63, 65: transposed up 2 or 3 octaves + mi3rd,
intervallically inverted, in treble
Fig. 5. Motives and ordered sets in Improvisation A.
Another recurrentpattern is the intervallic sequence 1211. This occurs in

the following locations, sometimes truncated, sometimes as part of a longer
descending chromatic passage:
-(111211) beats 14-23 treble
1211 19 (last note)-23 bass, retrograde
-(12111) 30-32 treble
2111 46 treble
1211 60-62 treble
-(2111) 61-62 bass
The diversity of contexts here argues against the primary storage of the
motive being in kinesthetic form. Other passage-specific motives with more
variable pitch content include the grace note descending semitone figure in
beats 30-33, and the arpeggiation figure used at beat 51 and to end the
piece.

Third,the pieceis characterizedby the frequentappearanceof two (un-

ordered)pitchclasssets: (056) and (014). Theiroccurrencesareas follows,
where * indicatesthe intervallicallyinvertedform [viz., (056)* = (016),
(014)* = (034)].
(056): beats 10 (last note)-13, 18, 27-28, 38*, 39, 40, 41
(twice),41* (twice),43-44 (treble),43-44 (bass),45,
46-47 (treble),47-48 (bass),51, 51*, 54-55* (bass),
59*, 63, 63*, 64*, 65*.
(014): beats 15-16* (treble),16, 18*, 23-25, 27-28, 29, 31,
34-35, 34-35*, 36-37, 37-38, 41 (bass), 51, 54*
(treble),63, 65, 65* (twice).
Thereare clearlinkagesbetweenintervalor set usage and phrasestruc-
ture. Toward the end of the piece, for example, (056) appearspredomi-
nantly in invertedform. (014) is given agogic emphasisin the phrase of
beats 15-25. The phraseat beats 29-35 has consistentuse of (012) and
(014) sets (except for beat 33), culminatingin the trill at beats 34-35,
where(014) andits inversionaresimultaneouslyrepresentedby a fournote
(0145) set. Othersuchparticularscouldbe spelledout, but theseseemto be
the mainstructuralfeatures.Theprimaryconclusionis the presenceof con-
siderableintervallicand pitch class organizationin what is ostensibly
"free"improvisation.
Microstructural Analysis
Timing
Timing analysis here includes the following: interonsettimes, note and
chorddisplacement,durations,and chordalspread.Interonsettime (IO)is
the time between successivenote attacks. If we label attack and release
timesof the ith note playeda{and r,respectively,then
IO,-= ai+1- ai (1)
(Strictlyspeaking,a{is the time at which the key correspondingto the *th

note is sufficientlydepressedto triggerthe key-sensingmechanismsof the
keyboard).
Threetypes of durationare readilydistinguishable.Key duration(d) is
justthe time a key is held down. Hence
di = ri-ai (2)

146 Jeff Pressing
(Strictlyspeaking,r,-is the time at which the key correspondingto the ith

note is sufficientlyreleasedto triggerthe sensingmechanismsof the key-
board.)A second type of durationmay be called note duration(D). This
simplyaddsthe effectof the sustainpedalto key duration.It maybe defined
as follows:
- = =
D' = f r{: -ai if Pedal = off at t = rf- (3)
1 PR Oi if Pedal on at t r{
wherePRis the firstpedalreleasetime for t > rf-.The thirdtypeof duration
maybe termedsoundingduration(S,-).Thisis the durationof the soundtrig-
geredby the key depressedat time ah includingpedaleffects.If the synthe-
sizersound (voice)decaysimmediatelyto zero afternote release,andhas a
nonzerosteadystate, then S, = D,-.If the voice has no steadystate compo-
nentwe mayhave S,-< D„ while if the voice decaysslowly to zero afterkey
releasewe mayhaveS, > D,-.The performancevoice herewas of electricpi-
ano type, so that S, is nearlyexactlyequalto D, for all of ImprovisationA,
wherenotes were not lengthilysustained.
InteronsetTimes Figure6 shows a plot of the distributionof interonset
times. Severalareasof groupingare apparent,which raisesthe following
generalissue. Namely, to what time (or other parameter)resolutionshall
Fig. 6. Interonset interval distribution in seconds (Improvisation A).

we seekpeaks?Theproblemis a standardone in the fieldof signalanalysis,

where backgroundnoise (hereperformerinaccuracy)is present.The ap-
proachwe adopthereis to sumvaluesoversmall"windows,"chosento be
justlargeenoughto clearlyshow distributioneffectsin the populationsize,
andplot a resultinghistogram.Thisis donein Figure6 for a window size of
33 msec.The choiceof optimalwindow sizewill dependto some degreeon
the natureand structureof the data.
IO quantizationeffects show strong peaks at 0 msec (interpretedas
simultaneity),and 320 msec (interpretedas triplet «h- exact value 333
msec).Weakerpeaksare found at approximately650 msec (interpretedas
tripletJ- exactvalue667 msec),andaround1.00 sec (interpretedasJ).The
datafurthersupportthe possibilitythattheJpeakdividesinto two, peaking
respectivelyat 950 and 1030 msec,whichcouldbe explainedby a cognitive
procedurewhich in practicealways eitherstretchedor contracteda basic
cognitivecategory(J)to produceândJ.Inview of the smallamountof data
involved, this remainsmerely a suggestion,one requiringsupport from
otherimprovisationanalyses.
But,in any case, quantizationof IO timesis clearlybasedon tripletsub-
divisionof the beat.Examinationof the globalfinedetailof IO timesbelow
100 msecdidnot revealanysignificantsubstructure,althoughquantization
thatoperatedonly locallyovera restrictedpassagemightnot be detectedin
such an examination,if rubatoeffects in the same time rangefrom other
partsof the piecewerepresent.
DisplacementFigure7 shows a frequencyplot for note displacementfrom

the beat,in beatpercentage,with 2% windows.Figure8 shows a blowupof
thecentralregionof ±10%, with 0.5% windows.On the basisof thesetwo
plots, the displacementcategoriesshown in Table 1 seemto hold. The last
indicateddisplacementcategoryis weakerthan the othersand formedpri-
marilyfrom data at two locations:the double trill (beats34-35) and the
concludingchords(beats63-65). Overall,the cognitiveprocessesat work
canthereforebe hypothesizedto be:
1. Divisionof the pulseinto 2 or 3 parts
2. Advancementof note placementin relationto the beatby 4%
(tripletsubdivision,on beats); or 2 or 8% (double subdivi-
sion), at a lower level of significance.
The appearanceof both undisplacedand consistentlydisplacednotes on

the Jdownbeatsupportsthe idea that the two processesare independently
operative.(Italso seemsreasonablethatthisindependenceshouldbe easiest
to observein placementof notes on the downbeat.)The possibilityof a re-
latedeffectat the half-beatis raisedby the data, but thereis not sufficient
evidencefor any conclusion.

148 Jeff Pressing
Fig. 7. Note displacement distribution in beat percent (Improvisation A).
Fig. 8. Note displacement: detail of central region (Improvisation A).

TABLE 1
Displacement Categories
Displacement
Center Strength Interpretation
-37% Moderate -33 V3% (triplet J>before beat) with 32/3%

anticipation
-4.5% Strong 0% (on beat) with 4.5% anticipation
0.0% Strong 0% (on beat)
29% Moderate 33V3% (triplet J>after beat) with 41/3% antici-
pation
42%, 48% Weak 50% (J>)with 2% or 8% anticipation
Thereis anothertype of displacementas well. This refersto displace-

mentsin relationto the notatedcategoriesof the transcription.Sincethis is
verymuchpassage-specific,it is treatedunderthe correlationssection.
DurationsWe areconcernedherewith key durations,di = r,-- Uj.Figure9
shows the key durationdistributionfor this piece. There are clear peaks
only at about75 msec (7.5 beat %) (correspondingto gracenotes, the trill,
and staccatofigures)and acrossthe range950-1120 msec, corresponding
to theJpulse.An examinationof the detailbelow 20 beat % discoveredno
finerresolutioneffects. There are no significantpeaks correspondingto
tripletor dupletsubdivision.Longerdurationsnot shownin Figure9 occur
mostlyas roughmultiplesof the Jpulse.
Fig. 9. Key durations (Improvisation A).

150 Jeff Pressing
The differencesbetweeninteronsetand key durationtime distributions

point to the operationof independenttimingmechanismsfor durationand
interonset.At the same time, the sharedJ categorypoints to the fact that
theycanbothoperatein conjunctionwith an externalpulse,or presumably,
an internallygeneratedone. The lack of cleardurationalpeaks in the mid
rangesshows that releasetime is a far less importantvariablethan onset
timefor thisperformer.Thisis in linewith the acknowledgedhabitsof most
piano (as opposed to organ)players,due to the piano's sustainpedal and
decayingsoundenvelope.
ChordalSpreadsA chordis definedhereby indicatedsimultaneousattacks
on the transcription(Figure4). These reflect intuitive decision-making
about what is perceivedas synchronous.Subsequentconsiderationof the
full data set (and particularlythe four arpeggiated,hence asynchronous,
chords)leadsto the conclusionthatfor this synthesizervoice, the boundary
for synchronousperceptionof chordslies in the range50-75 msec.
Certainpatternswithin the chordalspreadsseemto show that different
productionproceduresare operatingfor chords of differentsize. Thus,
77% of the 3 1 two-note chords(X2two-tailedtest p < .005) beginwith the
highernote (r.h.)and50% of fourthree-notechordsbeginwith the highest
note (and50% with the lowest),while 100% of 2 four-noteand4 five-note
chordsend with the highestnote. The four arpeggiatedchordshave been
includedin this tally.
The3 1 two-notechordsarenumerousenoughto justifytheirown analy-
sis. If we definethe spreadX of a two note chord as (highernote attack
time - lower note attack time), the averagevalue for X is X = 6.5 msec,
which representsthe mean time intervalthat the righthand leads the left.
Standarddeviationa = 17.7 msec,with a largecomponentof this due to a
few largevalues.The averageabsolutespreadX is foundto be IXI= 13.6
msec,and the medianabsolutespreadis 7 msec.Thesefiguressuggestthat
two-notechordalspreadis intentionallypositiveand controllableto an ac-
curacyof about 10 msecor perhapsless.
The mean displacementof the two note chordsin relationto theirindi-
catednote positionsin the transcriptionmaybe computedto be -30.6 msec,
correspondingto an anticipationof 3.1%, consistentwith the earlierfind-
ings. The fact that the displacementis so muchgreaterthan meanchordal
spreadstronglysupportsthe intentionalityof the displacementand raises
the question of whether the ability to internallysynchronizeoutputted
actionsexceedsthe abilityto coordinatethem to an externalcue, as seems
likely.
Dynamics
QuantizationFigure10 shows a plot of key velocityincidence.KVvalues
rangefrom 1 to 112, with a meanof 91.8 andstandarddeviation18.9. The

Fig. 10. Key velocity distribution (Improvisation A).
figureshows clear"quantization"effects,with peaksat about93, 100, and

108, and a muchmorediffuseband from76 to 88. It may be hypothesized
that these four peaks, which correspondaccordingto our calibrationsto
keytransittimesof about 10-14 msec (KV76-88), 9.5 msec (KV93), 7.5
msec (KV 100), and 5.0 msec (KV 108), representseparatecategoriesof
intendeddynamicsby this performer,correspondingto a process which
may be termedcategoricalproduction,by analogy with the well-known
phenomenonof categoricalperception.The presenceof noise is considered
to be responsiblefor broadeningthe sharpnessof the peaks. Althoughthe
data here are hardlyunequivocalin this interpretation,the effect will be
seento be morepronouncedin ImprovisationB, and work with other im-
provisersseemsto show it to be a generalphenomenon(Pressing&cTroup,
to be submitted).
Correlationswith KeyNumberNo generalcorrelationbetweendynamics
andkey numberwas observed.Inparticular,therewas no consistentdiffer-
encebetweenblackand white keys.This was as expected.
ChordalDynamicsA look at the 42 struck(transcribedas simultaneously
articulated,includingarpeggiated)chordsin the pieceshows that in all but
three,the key velocityof the highestnote is greaterthan or equalto that of
thelowest.Thehighestnote is also louderthanall innervoicesin all but two
chords.Thus this is clearlya treble-dominatedtexture,with the mean dy-
namicdifferencebetweenoutervoices 11 KVunits.

152 Jeff Pressing
Legatoness
One advantageof microstructuralrecordingis the possibilityof examining
degreesof legatoness.In the followingdiscussion,this is definedin termsof
the overlapof adjacentnotes, where adjacentmeans adjacentin orderof
attacktime. Thus therehas not been, as would be expectedin traditional
musicanalysis,a separationinto parts,whose legatonesswould then each
be consideredseparately.We are lookingthereforeat "global"ratherthan
"line"legatoness.Becauseof the thinnessof textureand the natureof the
figurationsin the two improvisationshere, this differenceis not a major
one. But it is also not always possible to reliablydistinguishindependent
partsin the transcriptionof improvisedmusic.In any case, the questionof
globallegatonessis a legitimateone on its own.
AbsoluteLegatonessabsolute (key) legatonessL, of any pair of adjacent
notes may be definedas theiroverlaptime, viz.
Li = min {rhri+1)-ai+1 (4)
Note that it is possibleto definethreesuch typesof legatoness:key legato-

ness (L), note legatoness (NL), and soundinglegatoness(SL).The above
definitionis for key legatoness.As in the previoussectionon IO times,note
legatonessincludes the effects of the sustain pedal in note overlap, and
soundinglegato includesthe sound decayeffectsof the synthesizervoice in
determiningoverlap.Sincethe sustainpedalwas not usedin Improvisation
A and only one synthesizervoice has been used throughout,we consider
hereonly key legatoness.
Figure11 shows the absolutekey legato distributionfor this piece. The
quantizationhere is very noticeable,with peaks at approximately-250
msec, 0 to 80 msec, and weak effects at 660 msec, 960 msec, and 1060
msec.Highervalues(not shown)clearlyreflectmultiplesof the underlyingJ
pulse.Interpretationsof the peaksareshownin Table2. It is surprisingthat
despitethe fundamentaltriplet«hIO quantization,restquantizationis clos-
est to a divisionof the beat into 4 Jis.Thismightbe consideredto point to a
quaternarytemporaldesignof rests,independentfromothertimingvalues
(IOtimes,durations).A closerexaminationof the passagescorresponding
to this peak, however,shows that this moreprobablyresults,in this case,
fromthe superpositionof two processes: triplet«hsubdivisionplusstaccatis-
simo. Thus, these legatonessvalues are found nearlyexclusivelyin places
(e.g., beats 18-19, bass register)where suchprocessesare readilyinferred
on musicalgrounds (see Fig. 4), and we do not find legatonesspeaks at
-0.50 or +0.25, which would independentlysupportthe hypothesisof
quaternarysubdivision.

Fig. 11. Absolute legato distribution in seconds (Improvisation A).
TABLE 2
Interpretations of Peaks
Peak Location Interpretation
-250 msec ì ? (see text)

0-80 msec tenuto - slight legato
660 msec triplet J overlap
960 msec ^[overlap
1060 msec Joverlap
Theinterpretationof the othermainpeaksappearsunproblematic,as in-

dicatedabove,but two pointsareworthyof comment.First,it is surprising
that thereis no peak at 333 msec (triplet^). Second,we see again a weak
bifurcationof temporaldataaround1.00 sec, the pulserate,which maybe
meaningful,sincethis also occurredwith IO timesand durations.
The finestructureof the verylargepeakin the range0-80 msecis given
in Figure12. In interpretingthe results,the issue of the meaningfulnessof
resolutionagain arises, and a straightforwardansweris not easy. In the
broadsensethereis a peakin the range0-80 mseccorresponding,presum-
ably,to the tenuto or weak legato process.Withinthis, thereare two une-
quivocalbroad bands in the ranges -10 to 25 msec and 40 to 80 msec.

154 Jeff Pressing
Fig. 12. Absolute legato distribution detail (Improvisation A).
These might be consideredto correspondrespectivelyto a tenuto and a

weaklegatoprocess.Differentiatingfurther,the secondbandseemsto have
clearpeaksat about 43 and 63 msec.If theseare not chancearrangements
of data, they may correspondto distinguishablemicroscopiclegato proc-
esseswhose naturehasyet to be elucidated.It maybe notedthatnearlyone-
thirdof all legatonessvalues in the range -10 to +80 msec occur in the
doubletrillpassageof beats34 to 36. However,this doesnot seema serious
objection,as even if these valuesare omitted,the relativepeak amplitudes
in the passagein questionremainin proportion.Furtherstudyof otherim-
provisationsis clearlyrequiredhere,and is in progress.
RelativeLegatonessThis quantitymeasuresthe extent of overlapbetween
two successivenotes in relationto the durationof the first one. In other
words,
A,-= Li/di (keyrelativelegato) (5)

It is plotted in Figure13. There are strongpeaks around0.00-0.05 and
0.95-1.00, and a weakerpeak at 0.50. Thesemay be consideredto repre-
sent tenuto to weak legato, 50% overlaplegato, and completeoverlaple-
gato.

Fig. 13. Relative legato (Improvisation A).
Correlations
Here we discuss correlationsbetween microstructuralparametersand
withmacrostructure. To facilitatethiscomparison,Figure14 shows a num-
berof microvariablesplottedagainsttime, scaledto allow inclusionon the
samepage.Somenoteswith extremeabsolutelegatovalueshavebeenomit-
ted for clarity.
An additionalvariableis also occasionallydiscussed.This is notational
shift,which refersto smalldeviationsfrom the attacktimes of transcribed
notation,wheresimple(experientially,"felt")beatsubdivisionsareusedby
the performer.This reducesto note displacementfor on-beats.
The discussionwill not treatthe entirepiece in full detail, but will look
selectivelyat significantdetailswithinchosenphrasesto highlightthe kinds
of correlationsfoundin this improvisation.
1. Openingphrase,beats 9-13. This passageemphasizesthe in-

tervalsof 4th andtritone.Ignoringthe veryshortandverydy-
namicallyweak G, it dividesinto threedyads:DCJt,BFtf,and
CF. These pairs have remarkablysimilarkey velocities:(93,

Fig. 14. Selected microvariables versus onset time (Improvisation A). O = key number,
= key velocity, A = absolute legato. Absolute legato is scaled for comparison, and ex-
treme values are excluded.
The absolute key

102), (93, 102), and (93, 104) respectively.
the seven notes
legato values for the six transitions between
and
(now including the G) are .064, .053, .070, .009, .062,
.050. Except for the fourth value, these show a considerable

consistency that identifies them as belonging to the same

phrase.The last two notes of thispassage(C andF) andthe Et
of beat 14 show what is seen below to be a characteristicat-
tributeof notationalshift:as a phraseor concatenationof mo-
tivesprogresses,attacksmoveprogressively(usuallyforward)
in relationto the ostensiblenote positions.Here the notes C,
F, and Et have respectivenotationalanticipationsof 35, 42,
and 51 msec.
2. Beats18-25. The bass tripletfiguredisplaysthe samefeature
of progressivenotational shift. Here are the bassline nota-
tional deviations,in milliseconds:
at +37 \
G -13
C -24 tripletfigure,
Et -58 I increasinganticipation
G -62
B -81 >
C -41 \
Qt + 22 I changeto crotchets,
D)t +59 I increasingdelay
E +81/
Hencethe changein note durations,texture,and bass legato-
ness at beats 20-21 is reinforcedby a changeof sign and di-
rectionof changeof the notationalshift,as well as a reduction
in key velocity.This is also the locationwhere a contextually
large chordalspreadoccurs (14 msec) which contrastswith
the remainingpassageaverageof 3.7 msec.
At the same time as these clear correlations,there are iso-
lated phenomenawhose significancecannot properlybe as-
sessed due to the uniqueness (hence unreliablysmall data
base) of improvisation.A simple example occurs with the
middletripletbass notes of beats 18 and 19, where both Gs
have identicalkey velocitiesin a variableenvironment.Very
likelycoincidence,but. . . .
3. Beats29-33. Herewe finda consistentorderof strikingof the
handson downbeatsand a consistentlysmallchordalspread.
Therighthandsequentialsemitonemotivesbeginningon beat
30 show a consistentdynamicrangeand pairing:(108, 111),
(108, 110), (108, 109), and (109, 106). Note that the KV dif-
ferencesare (firstnote KV - second note KV): -3, -2, -1,
+ 3, showing a clear progression;furthermore,the last ap-
pearance of the motive, where relative note key velocity
changessign, is the end of the phrase,follows the only exam-

158 Jeff Pressing
pie of disjunctsopranomotion,andmarksthe appearanceof a

new pitch class set (pc set progression014-012-014-012-
015). Intervals1, 2, and 4 areemphasizedthroughout.As the
phrasenears its end, as well, delays in relationto the down
beat change to increasinganticipations.Thus the successive
displacementsof the secondnotes of the motivefromthe beat
are + 114, + 118, + 89, and + 13 msec.Consistentlegatoness
valueslink the motivesas well, as shown in Table3.
4. Beats 42-54. Here are the suggestiverelationsbetweenkey
velocityand the use of motives.Thusthe three-notemotiveof
both partsof beat 43 representedintervallicallyby ([6, 7], 5),
seenin fourlocations,alwaysoccurswith minimumKVin the
middle position. J J>motives (7 occurrences)always have
higher on-beat key velocities, and this patternis even more
regularwherethe motiveshave intervalliccommonstructure,
as in the trebleof 47-50, which consistsof arpeggiationsof
two diminishedchords, with key velocity pattern 106-97-
106-98-101-95-106. Key velocity also unifies the treble
motivesequencedin beats52-54, as follows:
Note: F D G ; E Cfl F
KV: 110 108 106 111 108 108
This occurs even though the rhythmchanges in the second
statementof the motive.
5. Beats 55-57. This passageis singledout on the basis of the
subjectivesense of surprisereportedby the performerat the
double octave unison on beat 55. The two following beats
show evidence of cognitive programmingcontrol change.
Pitchmaterialshiftsto a tonal chordcenter(C minor),rhythm
is simplified,and 50 msec anticipationshifts to slight delay
(beat56^then backtowardveryslightanticipation(3-5 msec
on beat 57). Thereis also a cleardropin treblekey velocityat
table 3
Absolute Legatoness of the Motives
Beat
30 31 32 33
Firstmotivenote .075 .044 .044 .032

Bass .149 .126 .151 .106
Secondmotivenote .951 .894 .847 .953

beat56. It is possibleto inferfromthis that the performerdis-

orientationat beat55 causedhim to
(a) choose less sophisticatedpitchmaterials;
(b) delaystrikingthe next notes by some 60 msec,possibly
in orderto have moretime to think;
(c) strikethe keysat the followingbeatwith moretentative-
ness, henceless force;
(d) choose less sophisticatedrhythmicmaterials.
Thatthe disorientationwas not completelyresolvedin the fol-
lowingtwo beatsis evidencedby somesloppinessof execution
aroundbeat 59 (largechordalspread).By beat 60, however,
this has passed.
6. Beats 63-65, the endingphrase.The harmonicimplications
of thisphraseweredescribedearlier.Otherfactorssupporting
the climaxingof the phraseon the finalchordinclude:
(a) increasingmean key velocities per beat: 93.0, 94.5,
104.2; and for the bass line, 90, 100, 108;
(b) increasing normalized chordal spread, defined as
chordalspreaddividedby (n - 1), where n is the num-
berof notes in the chord:14.8, 25.3, 32.0 msec;
(c) the occurrenceof the trebleset EAEton beats63 and65 ;
(d) contrarymotionto registraiextremes.
Other generalpairs of microstructuralvariablesshowed little correla-
tion. Thus, no consistentrelation between chordal spread and dynamic
spreadof chordswas found.
Event Partitioning
Ina recentpaper(Pressing,1987), I presentedwhat is apparentlythe first
cognitivemodel of the improvisationprocess.The model is basicallysyn-
thetic(as opposedto analytic)in that it describeshow improvisationsmay
be generated.It can only be analyticto the extent that sufficientperform-
anceinformationis availableto allow the unambiguousparsingof the mu-
sic into groupsof musicalevents,underspecifiedcriteria.This provedpos-
sible for ImprovisationA and gives support for the applicabilityof the
theory.
The model considersthat any improvisationI can be brokenup into a
seriesof nonoverlappingeventclustersE, suchthat
I=[EUE2...EN], (6)
whereeachE,is an integrallyconceivedmotorieandmusicalunitconsisting

160 Jeff Pressing
of one or moreevents.The relationsbetweensuccessive£,s areconsidered

to be characterizedby either associativeor interruptgeneration,corres-
ponding to continuity or discontinuityof musical design. AdjacentE,s
linkedby associativegenerationare consideredto sharecommonfeatures,
processes, and/or musical objects (notes, motives, etc.). A set of E,s so
linkedformsa distincteventclusterclassK;,separatedby interruptgenera-
tion from the next event clusterclass X;+1.For furtherdetailsthe readeris
referredto the originalpaper(Pressing,1987).
A partitioningof this improvisationinto 33 distinctevent clusters(£,-)
and 11 distincteventclusterclasses(Kj)is shownin Table4. The numberof
membersof the Kjvaries from 1 to 6 and their durationsfrom 1.305 to
table 4
Partitioning of Improvisation A
EventCluster BeatLocation EventClusterClass
E! 9-13 (first triplet) Kx = (Eu E2)

E2 13 (last 2 triplets)-14
E3 15-17 K2 = (£3,£4,E5)
E4 18-20
E5 21-25 (except last Ji)
E6 25(lastJ))-27 K3 = (£6,£7)
E7 28-29
Eg 30 K4 = (£8,£9,E10,£ii)
E9 31
E10 32
En 33
En 34-35 (except last triplet^) K5 = (£12)
Eu 35 (last triplet JO-36 K6 = (£13)
Em 37 K7 = (E14,E15,E16,£17,E18)
Eis 38-39 (except last J>)
E16 39 (last Ji)-40
Ely 41 (except last triplet J>)
Eis 41 (last triplet J^)-42
Ei9 43-44 Ks = (£19, £20, Elu E22,£23)
E20 45-46 (triplet^)
E2i 46 (last 2 triplet «hs)
E22 47-48 (except last triplet .h)
E23 48 (last triplet «h)-50
E24 51 K9 = (£24)
^25 52-53 K\o = (£25, £26, £27, £28, £29, £30)
E26 54-55
E27 56-58 (except last i)
E28 58 (last » -59
E29 60
E30 61-62
E31 63 K„ = (£3i,£32,£33)
E32 64
E33 65

10.780 sec, with mean 5.065 sec. The choices for the boundariesbetween
the Ei and Kfwere madeon musical(phrasestructures,motives,etc.), mo-
torie (inferredfingering,hand position, etc.) and cognitive(knowledgeof
the performer'sstyle,recordedcommentsin score)grounds.
IMPROVISATIONB
Thiswas recordeda few minutesafterImprovisationA, with the author

as performerand Greg Troup as experimenter.No click track was used.
Theinstructionsto the performerwereto producea (completely)freepiece
of about a minute'sduration.The resultingmusic was characterizedby
GregTroup as "a sustainedsombrepiece, befittingthe mood of the per-
former."
Macrostructural Analysis
A full transcriptionof this piece is seen in Figure15, wherepedal mark-
ings are given along the horizontalaxis. Sinceno click trackwas used to
"drive"note production,such a transcriptioninto traditionalnotation is
onlypossibleif somepulseis chosenas fundamental.In the examinationof
note durationand IO time quantization(see below) no dominantfunda-
mentalpulseemerged,nor did the performerclaimto be usingone. There-
foreJ = 60 was chosenfor transcriptionconvenience,whichis at leastcon-
sistentwith an observedweak IO peak at about 1 sec and allows ready
comparisonwith ImprovisationA. However, it must be rememberedin
lookingat Fig. 15 that no implicationsof tempo, pulse or metershouldbe
drawn.Therefore,except where rapid passageworkdemandsshort note
values,durationsareroundedfor clarityby as muchas 150 msec,to avoid
complicatedandunreasonableliteralness.In all cases,however,local musi-
cal structure(orderof attacks,presenceof legatoness,etc.) is preservedin
the notation.
ImprovisationB in manyways displaysfarless extensivemacrostructure
thanImprovisationA. Therearefew recurrentmotives,andpitchmaterials
show less homogeneity.The usage of sets shows no simple centralunity.
The lack of referenceto a definedpulse has presumablycontributedto
make this overall a "freer"improvisation.Nevertheless,the following
pointsdo hold.
In generalterms,the piece is slow moving, and dominatedby sustained
textures,pedal points, and conjunctmotion. Phrasestructureearlyin the
piece(beats6-19, 20-35 and36-50) is balancedandfairlylong.Thisthen
breaksinto fasternote values and shorterasymmetricalphrases,particu-
larly in the passage of beats 69-81. The final passage (beats 81 -end)
shows a simplerdurationalstructureand a returnto somethinglike the
originaltexture.

162 Jeff Pressing
(J = 60)
6 7 I , 9 « h. i3 . is | [-12msec]
[-18 msec]
,oi7, i9, g^T177. .- ^3^-k~ "^rr-rp^7, i.

jr~S~iz^ io2#-în ^Vîo6 \7y- yr^-^^_ - 'T ^86 ^^1 ^^^
29 1-30msec) 3 33 35 37
À*L\ ]J^ JJ
^-^
'
99
^ "
IJ l>^r~li .8899
IM1 f
, r"'
-^ iL^Jj 106
^
33% 11- înd-1 *J83

s I' ^ 65
68 90 |'w |^- 86 J| I l_94
99
msecn [9 msec]
^
[9:llmH 4i 43 45 47 [17 msec] & msecl
h [
j m iœ
W^|S^
'
i;j:r
^^.^"'l 91
^ ^[jj^
68 ^ '
[17 msec]
1^ I I I ILg* ffI "-^-"1 ^- ^ ^J^

-^ H
81 J I I 79
I M I ^8
Fig. 15. Transcription of Improvisation B.

Fig. 15. Continued.
A motive based on the interval of a rising 6th followed by conjunct

(rarely,minorthird)descent,roughlyof the form I i
is found at beats 6-8 (treble),6-14 (bass),60-63 (top line), 61-66 (top

line-G Et D), and 75-77 (bass-D Bl>A). Relatedrising6th motivesalone
arefoundat beats48 (bass),69 (bass),and 84 (treble).

164 Jeff Pressing
Othermelodiceventsmay be interpretedas transformationsof this cen-

tralmotive.Forexample,the size of the firstintervalis shrunkin the seem-
inglyrelatedmotivesat beats42-44 (trebleEt At G) and48-50 (D F#Ft|).
More distanttransformationsmay exist, as at beat 20, with the patternC
Bt A, omittingthe firstA of the treble;at beats29-26 in the bass,wherethe
leapAt Bt is followedby the stepwisedescentE Et D; the bassfigureat beat
40, thendescendingto F(tat 44; the treblepatternGttAGjtat t56-60; and
beat71 (treble),wherethe notesAt G C couldbe a permutationof the basic
motivethat preservesrelativenote durations.In addition,a relatedrising
M7 motiveis used four timesin beats 89-93 to end the piece.
An independentquick chromaticmotive moving from F to At occurs
also: at beats29 and40, and transposeda tone higherat beat 35; the chro-
maticflurryat beats69-70 is clearlyrelatedas well.
One pervasiveprocess seems to be the use of slowly rising or falling
diatonicor chromaticlines,sometimesdecoratedby gracenotes, frequently
emphasizing(by convergenceor duration)the notes G, B, or D. This is
foundin beats 10-21 (EF G A Bt), 21-29 (Bt A G F G), 29-55 (At Bt C
Bt-bass), 31-37 (E Et D), 35-55 (Bt At G FitFl)E), 41-59 (Et D C B),
57-67 (A G«G), 70-83 (At G F Ffl),71-83 (C C(tD Et Ei)-treble), 71-75
(C(tD Et El)D-bass), 80-87 (F(tG At All),and 85-89 (A A(tB-top line).
This processtends to establishthe notes G and D as tonal focal points.
This is also supportedby the high incidenceof these notes, at least one of
which sounds virtuallycontinuouslythroughoutbeats 6-50 and 62-85.
Otherspecificfigurespromotethis, as for examplethe passage67-72 (tre-
ble) which moves by chromaticmeanderingfrom D to G. It is also pro-
moted by the use of the notes F(f,At and Qt, Et as long term chromatic
neighborsto G and D. This may be seen for exampleat beats41-49 (G F#
G-bass), 72-84 (GF(tG-treble) and 55-72 (C(tD-bass). A relatedproc-
ess is visiblein the closingthreebarsof the piece,wherethe ascendingM7
motivesstarton F# and end with the motionsEt D At G. Finally,the first
two notes of the piece, and the last and thirdlast, areD and G.
There is also a verticalemphasisthroughouton the intervalsM7 and
mi9: in other words, intervalclass 1. This has a high incidencebetween
outervoices, and also betweenan outerandinnervoice, as maybe seen for
examplein beats23, 29, 41, 44, 61, 77, and 89-93.
A separateanalysiswas also performedof the held sonorities of this
piece.No strongstructuralfeaturesemergedfromthis.
Microstructural Analysis
Timing
InteronsetTimesIO timesin this piece,as mightbe expected,show weaker
quantizationeffectsthanin ImprovisationA. Figure16 shows that only the

Fig. 16. Interonset interval distribution in seconds (Improvisation B).
peak at 0 msec is clear-cut, obviously corresponding to intended synchrony.

Other weak peaks appear to exist at about 180 msec, 500-650 msec, and
950-1150 msec. Examination of the passages where these IO times occur
sheds some further information. The 180 msec unit appears primarily in
beats 70-80. Hence it appears to be a real but local time unit. The 500-
650 msec units are only found in the second half of the piece, while the val-
ues around 1 sec occur throughout.
Durations As seen in Figure 17, distinctive key durations occur only in two
peaks, centered around 250 msec and 550-700 msec. These peaks are not
changed by the inclusion of pedaling effects (i.e., producing note rather
than key durations), since pedaling durations were overwhelmingly at least
1 sec in length and most commonly far longer. No consistent pattern of long
durations (>2 sec) was observed for either key or note durations.
We then turn to the general consideration of time quantization in this
piece. A comparison of the IO and durational peak shows possible com-
monality only in the ranges 180-250 msec and 550-650 msec. As men-
tioned before, 180 msec seems to be peculiar to a particular passage and
hence is unlikely to be globally fundamental. As in Improvisation A, sepa-
rate processes, which sometimes coordinate, seem to govern the IO and du-
rational time structures.
The existing time peaks do not suggest any simple numerical ratios im-
plying division or concatenation of an underlying pulse, with the possible

166 Jeff Pressing
Fig. 17. Key durations (Improvisation B).
exceptionof the 180-250 and 550-650 msecbands,which mightbe con-

struedto be noise-broadenedpeaksin 1:3 ratio.Whilea numberof timing
schemescan be constructed,usingadditive,divisive,andmultiplicativeop-
erations,capableof explainingthe data, the peaks were found to be too
diffuse to unequivocallydecide between them. Hence no furtherconclu-
sionsprovedpossible.
ChordalSpreadsOf the 23 chordsin this piece, 12 have a negativechordal
spreadand Ila positiveone. Contextualanalysisalso shows no significant
localphenomenawith regardto chordalspreadX. Statisticsareas follows:
mean chordal spread X = 0.2 msec; mean absolute chordal spread=
IXI= 20.9 msec;medianabsolutespread= 20 msec.Thiswas
apparently
not a finelycontrolledvariable,as it was in ImprovisationA.
Dynamics
QuantizationFigure18 shows that thereare clearquantizationeffectsop-
eratinghere. Keyvelocitiesvariedfrom 1 to 111 and theirmeanwas 87.0,
with standarddeviation19.5. Largepeaks are found, in the two-unitwin-
dow plot, centeredat 91, 99, and 106. These correspond
closely with the
majorpeaksfoundfor ImprovisationA, at 93, 100, and 108 KVunits,and

Fig. 18. Key velocity distribution (Improvisation B).
suggeststronglythatkey velocityquantizationis operatinggloballyfor this

performer.Other small peaks may also be significant,such as that at 77,
which is also found, weakly, in ImprovisationA, and the peak at 111,
whichmay only be a saturationeffect.
Therearealso cleardifferencesbetweenthe two improvisations.For ex-
ample,88-89 is stronglypresentin B but nearlycompletelyabsentin A.
Overall, there is support for the hypothesis of categorical production
(quantization)that is characteristicof this performerin general,or at least
thisperformerover a certainextendedperiodof time, or withincertainsty-
listic constraints.Here this seemsto describekey velocitiesof 77, 91-93,
99-100, and 106-108. In additiontheremaybe piece-specifickey velocity
quantization,which we see impliedby significantdifferencesin detail for
the two distributions(e.g., at KV 88).
ChordalDynamicsThere are 23 struck chords in this piece, if struck is
taken to mean simultaneousattack times differingby no more than 60
msec.If largergrace-notetypefiguresareincluded(atbeats75 and 89-93)
there are five others. Twenty-six of these 28 chords have KV (top note)
> KV(bottomnote), andthe meandynamicdifferencebetweentheirouter
voices is 14 KV units. The same tendencyfor a treble-dominatedtexture
holds for sustainedchordswith staggeredentriesas well.

168 Jeff Pressing
Legatoness
Absolute Legato Significant detail for absolute (key) legato was only found
in the region below 200 msec, which is plotted in Figure 19. There are clear
peaks centered at 45 and 105 msec. The 45-msec peak was also found in
Improvisation A, but a weak peak around 105 msec there was not judged
significant. This suggests weakly that absolute legato quantization can be
based on pulse and also operate independently of it. Further research is
clearly necessary, but apparently legatoness forms a third type of quantiza-
ble time framework, after IO time and duration.
Relative Legato Relative (key) legato Xis plotted in Figure 20. This does not
include any pedaling effects. There are three weak peaks: at 0.08, 0.19, and
0.54. A comparison with Improvisation A suggests that the peaks at 0.08
and 0.54 correspond to peaks there, shifted slightly (0.04-0.05) to the
right. As there, they may be ascribed to a tenuto and 50% legato process.
The absence of the peak at 1 .00, as in A, is clearly due to the use of the pedal.
The peak at 0.19 does not have an obvious explanation.
Pedaling The presence of extensive pedaling should emphasize that the sec-
tions on duration and legatoness just discussed have referredexclusively to
the actions of the hand. The major effects of pedaling may be gleaned di-
Fig. 19. Absolute legato distribution in seconds (Improvisation B).

Fig. 20. Relative legato (Improvisation B).
rectlyfrom the musicaltranscription.The pedal is used to sustain signi-

ficanttones, andis turnedoff at endsof phrases,at changesof harmony,or
severalnotes afterthe startof rapidnote figuresto limit melodicblurring.
Meanpedaldurationdecreasesafterbeat 69, and gaps increasein size, be-
fore a returnto the openingbehaviorsin the endingpassagebeginningat
beat85. Thus an ABA'formis supported.
Correlations
In contrastwith ImprovisationA, only veryfew micro-microor micro-
macro correlationswere found in ImprovisationB. IO times, durations,
and absoluteand relativelegatonessshowedno clearmicrostructuralrela-
tions to any of the passage-specificmotivesor processesdescribedas mac-
rostructure.No correlationwas foundbetweenchordalspreadandchordal
dynamicspread.Dynamiccorrelationswere found to be absent from the
M6 motivein its variousforms,andfromthe long conjunctstructurallines.
The only correlationsfound were those associatedwith the dynamicsof a
few stronglyidentifiablemotives of short duration:the chromaticgrace
note figureand the closingM7 motives.Specifically,the threenotes of the
chromaticallyascendinggracenote figureand their final main note were
alwaysplayedwith crescendo.The four M7 motives showed a consistent

170 Jeff Pressing
patternof dynamics,respectively(86, 99), (88, 101), (76, 82), (77, 80),

whichpartitionsthe motivesinto two groupsof two andshows a monoton-
icallydecreasingKVdifferencebetweenmotivenote pairs.
Event Partitioning
ImprovisationB canreadilybe parsedup in a musicallyplausiblefashion

into eventclustersE,-and eventclusterclassesKjas was done previouslyfor
ImprovisationA. One such analysisyielded24 event clusters,but only six
event clusterclasses, due to the texturalhomogeneityof this piece. How-
everthe presenceof polyphonicallyoverlappingphrasestructuresand the
lack of unifyingmotivesmakesfor a multiplicityof possiblevariationson
this analysiseven using microstructuraldata, without additionalcognitive
informationfromthe performer,whichwas not available.Sincethesealter-
nativedeterminationscould not reasonablybe excluded,full partitioning
dataarenot givenhere.
Discussion
These two short improvisationscontraststronglyin style, yet share a

numberof microstructuraltendencies.Thus interonsettime distributions
show commonpeaks around0, 650, and 1000 msec (but not elsewhere),
dynamicquantizationscenterin both pieceson the ranges91-93, 99-100,
and 106-108, chordaldynamicsshow top note dominanceby 11(A) and
14(B) KV units, absolutelegato featuresa common peak at 45 msec and
relativelegato sharesthe ranges0.00-0.08 and 0.50-0.54. On the other
hand,the distributionof durationsand chordalspreadsshows no similari-
ties, and most aspectsof absolutelegatonessdifferconsiderablyin the two
cases.Most strikingly,however,the micro-microand micro-macrostruc-
tural correlationswere extensive for ImprovisationA and nearly com-
pletelyabsentin ImprovisationB. Thisis despitethe factthatboth improvi-
sationswere judgedto be comparablysuccessfulin aesthetictermsby the
performer.
How much of this variabilityis due to the whimsicalityof intentionin-
herentin improvisationand how much is due to the presenceof a metro-
nome click in only one piece cannot be assessedwithout furtherstudy of
this performer.However,we may safelyassume,in line with commonmu-
sicalopinion,that both factorswill be important.(Theuse by a performer
of an internallygeneratedpulse is an intermediatecase that will form the
subjectof subsequentinvestigation.)
Butanotherfactoris probablyinvolvedhereas well. Thisis the extentto
which aestheticallysuccessfulart in generalwill be found to have strong

structuralunderpinnings.The situation occurs very clearlyin composed

music:somegreatworksdon't"analyzewell." Comparethepublishedpro-
lixitieson Stravinsky'sLe Sacredu Printempswith the paltryanalyticalof-
feringson Schönberg's(pre-systematic)Pierrot Lunaireyboth acknowl-
edgedtwentiethcenturymasterpieces.Speakingon a much more modest
front,ImprovisationB doesn't"analyzewell," comparedto Improvisation
A. In other words, the musicalmeaningis not well describedby existing
analyticalprocedures.What has been demonstratedin this articleis that
suchmusicalmeaningdoesnot necessarilyresidein theperformancemicro-
structureeither,althoughcertainlysituationscan arise where meaningis
predominantlypresent there. It seems increasinglylikely to this writer
thereforethat many currentlyineffableaspects of music will remain in-
definitelyundescribableby words or othersymbolicnotation.
However,manystructuralaspectshave been documentedhere. Quanti-
zation,or categoricalproduction,has beeninferredfromdistributionplots
of key velocity.Thereis also a strongsuggestionthat this quantizationis
largelyindependentof the presenceof pulse in the music. If supportedby
furtherinvestigations,this shouldhaveimplicationsfor performancepeda-
gogy and computer-controlled musicproduction.Likewise,the integrated
microstructural "support"givento the executionof distinctivemusicalmo-
tives (foundprimarilyin ImprovisationA) providesinsightinto the impor-
tanceof gesturaland kineticimageryused by performers,sincesuchimag-
ery presumably acts to coordinate microstructure (based upon past
movementexperience)in a way that focusingon individualdetails could
not, withinthe severecognitiveconstraintsof real-timeperformance.Here
again there are presumablyimplicationsfor computercomposition and
performance.
Othersuggestionsfromthe datainclude:
1. Trebletexturaldominancefor thisperformeris consistentand
pulse-independent.
2. Rhythmicdesign where pulse is present is based upon the
processesof pulsedivisionandshifting(predominantlyantici-
pationin thesepieces).
3. Consistentperformanceeffects may operateover time scales
of 10 msec or less (e.g., chordalspreads).Othereffects (e.g.,
synchronizationto an external pulse) show less resolution,
with 30 msec perhapsrepresentinga useful dividingline be-
tween qualitativelydifferenteffects.Simultaneityof tone per-
ceptionseemsto lie in a rangeabouttwice this (50-75 msec).
4. Differencesin the distributionpatternsof IO times,durations,
andlegatonesssuggestthreeindependentunderlyingtemporal
mechanismsthat may sometimeslink togetherin coordina-
tion with macrostructure(pulse,motives).

172 Jeff Pressing
5. Microstructural recording is only sometimes (Improvisation

A) sufficiently detailed to uniquely determine the event pars-
ing necessary to apply an earlier model of improvisation
(Pressing, 1987).
Clearly many more data are required before the issues raised here about
improvisation design can be satisfactorily addressed.1
References
Gabrielsson, A. Performance of rhythm patterns. Scandinavian Journal of Psychology,

1974, 15, 63-72.
Gabrielsson, A. Perception and performance of musical rhythm. In M. Clynes (Ed.), Music,
mind, and brain. The neuropsychology of music. New York: Plenum Press, 1982.
Hood, M. The ethnomusicologist. Kent, OH: Kent State University Press, 1982.
Jones, A. M. Studies in African music. London: Oxford University Press, 1959.
Lamb, M. R. Andre Tchaikovsky meets the computer: a concert pianist's impromptu en-
counter with a musicianship teaching aid. International Journal of Man- Machine Stud-
ies, 1978, 10,593-602.
MacKenzie, C. L., VanEerd, D. L., Graham, E. D., Huron, D. B., & Wills, B. L. The effect of
tonal structure on rhythm in piano performance. Music Perception, 1986, 4, 215-226.
Michon, J. A. Programs and 'programs' for sequential patterns in motor behaviour. Brain
Research, 1974, 71,413-424.
Povel, D. J. Temporal structure of performed music. Acta Psychologica, 1977, 41, 309-
320.
Pressing, J. Cognitive processes in improvisation. In W. R. Crozier and A. J. Chapman
(Eds.), Cognitive processes in the perception of art. Amsterdam: North Holland, 1984,
pp. 345-363.
Pressing,J. Improvisation: methods and models. In J. Sloboda (Ed.), Generative processes in
music. London: Oxford University Press, 1987, pp. 129-176.
Rasch, R. A. Synchronization in performed ensemble music. Acustica, 1979, 43, 121-131.
Seashore, C. E. Psychology of music. New York: McGraw-Hill, 1938.
Shaffer, L. H. Analysing piano performance: a study of concert pianists. In G. E. Stelmach Oc
J. Requin (Eds.), Tutorials in motor behaviour. Amsterdam: North Holland, 1980.
Shaffer, L. H. Performances of Chopin, Bach, and Bartok: Studies in motor programming.
Cognitive Psychology, 1981, 13,326-376.
Shaffer, L. H. Timing in musical performance. Annals of the New YorkAcademy of Science,
1984a, 423, 420-428.
Shaffer, L. H. Timing in solo and duet performances. Quarterly Journal of Experimental
Psychology, 1984b, 36A, 577-595.
Sloboda, J. A. The communication of musical metre in piano performance. Quarterly Jour-
nal of Experimental Psychology, 1983, 35 A, 377-396.
Small, A. M., 6c Campbell, R. A. Temporal differential sensitivity for auditory stimuli.
American Journal of Psychology, 1962, 75,401-410.
Tucker, W. H., Bates, R. H., Frykberg,S. D., Howarth, R. J., Kennedy, W. K., Lamb, M. R.,
& Vaughan, R. G. An interactive aid for musicians. International Journal of Man-
Machine Studies, 1977, 9, 635-651.
1 . The apparatus was built by the La Trobe University Department of Psychology work-
shop, with software written by Russell MacDonnell and Glen Lawrence of the same depart-
ment. I am also grateful to the La Trobe University School of Humanities for a grant for a
purchase of the DX7 synthesizer and to Greg Troup for experimental assistance and useful
discussions.


University of California Press

Uploaded by

Copyright:

Available Formats

University of California Press

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

University of California Press

Uploaded by

Copyright:

Available Formats

The Micro- and Macrostructural Design of Improvised Music

Author(s): Jeff Pressing

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

Thisarticlereportsthe firstexperimentalresultsof a largeprojecton the

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

correlationsdiscussed.The methodused here is noteworthyin that it fea-

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

for clicks,with largervalues for other kinds of sounds.This has therefore

The experimental apparatus consists of a modified 61-key Yamaha DX7 synthesizer, an

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

A number of experienced keyboard musicians of recognized professional standing as im-

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

Results and Analyses

The primarydata are analyzedseparatelyfor ImprovisationsA and B.

Figure2 shows the beginningsof the data file representingthis piece.

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

ily theygiveinformationon note displacement[10-30 msec = (- ); > 30

1. D is the most frequentlyoccurringnote in the piece, followed

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

[14 msec] [5 msec] •^"^(->}<-i [25 msec]

C) 102 (<-) (<-) ^ ^i- L26^"l08 V ff99 98 I 101

[2 msec) [13 msec] [8 msec] [12 msec] [17 msec]

*" 108 m 108 109

106 104 103 '

«J 104 in? (->) ÏÔÏ 7^

\.. l^m .^-n, . | ^ («-) «-> ff /r^rff

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

Another recurrentpattern is the intervallic sequence 1211. This occurs in

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

Third,the pieceis characterizedby the frequentappearanceof two (un-

IO,-= ai+1- ai (1)

(Strictlyspeaking,a{is the time at which the key correspondingto the *th

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

(Strictlyspeaking,r,-is the time at which the key correspondingto the ith

Fig. 6. Interonset interval distribution in seconds (Improvisation A).

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

we seekpeaks?Theproblemis a standardone in the fieldof signalanalysis,

DisplacementFigure7 shows a frequencyplot for note displacementfrom

The appearanceof both undisplacedand consistentlydisplacednotes on

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

Fig. 7. Note displacement distribution in beat percent (Improvisation A).

Fig. 8. Note displacement: detail of central region (Improvisation A).

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

-37% Moderate -33 V3% (triplet J>before beat) with 32/3%

Thereis anothertype of displacementas well. This refersto displace-

Fig. 9. Key durations (Improvisation A).

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

The differencesbetweeninteronsetand key durationtime distributions

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

Fig. 10. Key velocity distribution (Improvisation A).

figureshows clear"quantization"effects,with peaksat about93, 100, and

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM

Li = min {rhri+1)-ai+1 (4)

Note that it is possibleto definethreesuch typesof legatoness:key legato-

This content downloaded from 193.227.1.127 on Sun, 8 Dec 2013 20:12:15 PM