Professional Documents
Culture Documents
*Lynda Moorcroft, Dianna T. Kenny, and Jennifer Oates, *ySydney, New South Wales, and zMelbourne, Victoria, Australia
Summary: Objectives. This study investigated acoustic change in singers vibrato following imagery and nonimagery tasks.
Study Design. The study used a fully randomized cross-over (six conditions 3 two times) block design, in which
each singer received each intervention in random order. Data were analyzed using the general linear model (GLM).
Main effects for time and condition and interaction effects (time 3 condition) were calculated for each dependent
measure.
Methods. Six classically trained female singers recorded an 8-bar solo before and after three nonvocal, 25 minute
tasks. Each singer performed the tasks in a different randomized order in a single sitting. Task 1 involved imagery of
the breath directed up and down as far from the larynx as possible; Task 2 used Braille music code, enabling the singer
to engage in tactile, kinesthetic and visual imagery related to music but unrelated to breath function; Task 3 was a nonimagery activity requiring the completion of a cloze passage about breath function. From the 11 longest notes in each
solo, spectrograms of the partials were produced and assessed for pre- to post-test changes in vibrato rate, vibrato extent,
and sound pressure level (SPL).
Results. Only the breathing imagery task produced significantly more moderate and regular vibrato rates. Vibrato
extent was not responsive to any intervention.
Conclusions. Findings indicate that breathing imagery regulates singers vibrato in a manner consistent with that of a
more proficient, warmed-up voice.
Key Words: BreathingImageryVibratoWarm-upTone qualityClassical female singers.
INTRODUCTION
Singers trained in the Western classical tradition often use indirect techniques as an aid to soliciting the complex and often
subconscious physiological coordinations that produce optimal
vocal results. Mental imagery is one such technique, and indeed
the discipline of singing and vocal pedagogy . has consistently and historically used mental imaging techniques to
achieve its objectives. (p. 41)1 Although Cleveland1 noted
the need to extend voice research into the science of mental imagery, his call has largely been ignored to date.
Imagery used by singers often draws on the aural, visual, and
proprioceptive senses.24 It may or may not be text-based or
represent some aspect of physiology. However, because sensation in the larynx means lack of freedom in the larynx, (p.
154)5 and the voice tends to be more artistically acceptable if
it feels to the user as although it were produced in almost any
other region of the body than the throat,3,69 much of the
imagery traditionally practiced as an aid to technical control
focuses the vocalists attention away from the throat in a
manner that does not represent reality. Such images have
played an important role in voice teaching for at least five
centuries10 and include those in which sensations of the breath
or the tone are directed far from the larynx and even to some
point outside the body.10,11
Lynda Moorcroft, et al
183
detrimental symptoms of performance anxiety the word
anxiety stemming from Greek and Latin words meaning
constriction, pressing tight and strangling.35 The literature for the spoken voice adds that if breathing imagery is used
pre-performance, it serves as a silent warm-up by creating a
mental blueprint for the sound.24,25
Adding credence to the theory of a mental blueprint, the discovery of the mirror neuron system supports claims that imagery activates neural responses, triggering physical adjustments
that are often beyond conscious control.36,37 Furthermore,
mirror neurons show greater activation the more the
individual has a strong sense of the goal to be achieved38,39;
and pedagogical wisdom suggests imagery of sensations
directed both upward and downward, far from the larynx,
presents the singer with a proprioceptive goal linked to
skilled performance and optimal vocal tone quality. Thus,
Hurley40 proposes that the presence of mirror neurons may account for why musicians often report that imagining a skilled
performance in music improves performance.
In the tradition of Western classical singing, skilled performance requires optimal vocal tone which possesses as much
brilliance and mellowness as possible.8 Vennard writes that
the singers sensations which appear to be directed up and forward are related to a bright brilliance of tone and those directed
down and back are related to a darker mellowness of tone, and
for a chiaroscuro ideal balance of brilliance and mellowness,
both of these directionally opposing sensations must occur
simultaneously.8 Thus, according to Vennard, if imagery is
used optimally then vocal tone quality improves.
This suggests that breathing imagery may affect singers
vibrato, because it has long been observed that vocal color is
determined above all by vibrato,41 and that the faster or slower
the vibrato, the brighter or darker the tone.12 Furthermore, in
music where a chiaroscuro beauty of tone is of primary
importance, the more highly trained and skilled the singer,
the more moderate and regular the vibrato rate.4246
Excessively fast and unstable vibrato rates are often found in
students at the commencement of vocal training,46,47 but are
also typical acoustic indicators of muscular hyperactivity that
occur in situations of high stress, excessive force, and
performance anxiety, irrespective of singer level.5,8,48,49 In
addition to producing a very bright, sometimes shrill quality,
fast vibrato rates in the 68 cycles/second range may sound
like a bleat,48 with those in the 7 or 8 cycles/second range associated with tremolo.32 Slow vibrato is typical where lethargy or
poor muscle tone is present.5,46 Generally, vibrato rates below
5 cycles/second are considered unacceptably slow,45 produce
a particularly dark tone quality, and tend toward a wobble.50
Vibrato near 4 cycles/second clearly undulates rather than
creating the impression of a constant pitch.45,48
An acceptable tone color may vary depending on the repertoire, and so too may the vibrato rate. Exceptionally fast and
exceptionally slow vibrato rates appear to be important when
particularly intense emotions or extreme psychological states
are portrayed.48,49 For example, Maria Callas has recorded
vibrato rates as fast as 7.1 cycles/second in the mad scene
from Donizettis opera Lucia di Lammermoor,51 where the
184
METHOD
Singer background
Six classically trained female singers, each from a different studio, participated. Three were studying tertiary level singing,
and three had completed tertiary studies and sang professionally (mean age 29 years, SD 7 years; mean years of vocal
study 13, SD 5 years). Singer profiles are presented in
Table 1. The use of only female singers prevented genderrelated variables entering into the vibrato analysis.
The vocal solo
All singers were asked to learn 8 bars from Villa-Lobos Bachianas Brasileiras No 5 Aria. The vocal range spanned a major
10th from D4 (294 Hz) to F#5 (740 Hz). The excerpt called for a
calm, lyrical delivery and considerable vocal beauty for effective performance. The composer required this section to be
sung solely on an ah. The singers were given the print music,
a recording of the accompaniment to be used, and requested not
to warm-up on the recording day. It was considered that if the
singer had already vocally warmed up such that they were
singing their best before the recording session, then perhaps
only if an intervention had a detrimental effect on the voice
could a change in vocal performance be noted.
Recording procedure
All singers had individual appointments to record their voices.
They were fitted with an AKG C-477 miniature condenser
omnidirectional microphone (AKG Acoustics, Vienna, Austria)
which was head-mounted so as to maintain a constant distance
of 7 cm from the corner of the mouth to the microphone. The
microphone was connected to a DAT Marantz compact disc
recorder model CDR-640 (Marantz Japan Inc., Kanagawa,
Japan) via a Behringer Ultragain Pro MIC-2200 preamplifier
(Behringer International, Willich, Germany). Singers were
also fitted with Beyerdynamic DT331 free-field earphones (Beyerdynamic GmbH & Co. KG, Heilbronn, Germany) through
which to hear the pre-recorded accompaniment. This enabled
the voice to be recorded without the accompaniment. It also
ensured that all singers maintained an identical tempo, which
was necessary as vibrato rate may be influenced by note
duration.50
The recording level of the voice was first checked while the
singer sang once through the excerpt. As fluctuations in the
level of performance stress may influence vibrato characteristics, this process also served to allay singer unfamiliarity with
performance conditions. The singers then recorded the excerpt
before and after three non-vocal 25 minute tasks. Each singer
performed the tasks in a different, randomized order. One
task involved imagery of the breath directed up and down as
far from the larynx as possible. Another task used Braille music
code, which enabled the singer to engage in tactile, kinesthetic
and visual imagery related to music but unrelated to breath
function. A third task was a non-imagery activity in which
the singer completed a cloze passage about breath function.
So that all tasks might be perceived as equally valid, the singers
TABLE 1.
Singer Profiles
SingerSubject
Age (y)
Years of
Vocal Study
Vocal Education
Performance Background
1
2
3
19
24
25
7
9
9
27
12
5
6
49
29
26
16
Student examinations
Student recitals
Student recitals, semi-professional engagements at
municipal level
Winner of state-wide vocal competitions, state opera
Young Artists training program
State opera company chorus member, solo recitalist
Winner of nation-wide vocal competition, state opera
company soloist
Lynda Moorcroft, et al
185
were told that the project was investigating the effect of varying
levels of relaxation on the voice. At the completion of
recording, singer Sound Pressure Level (SPL) was calibrated
using two different dB readings of a 1000 Hz pure sine wave
tone and a Rion Integrating Sound Level Meter model NL-06
(Rion Co. Ltd, Tokyo, Japan). For a full account of each intervention see Moorcroft.54
Acoustic analysis
All vocal samples were converted into graphic form using the
computer software programs Phog Interactive Phonetography
System Version 2.0 (Hitech, Sweden) and Soundswell Core
Analysis Version 4.0 (Hitech, Sweden). Spectrograms of the
11 sustained notes from each vocal solo (ie, 132 notes per intervention activity, 396 notes in total) were produced and each
vibrato cycle assessed for rate and extent. The 11 notes assessed
per solo are indicated in Figure 2.
The screen resolution for each spectrogram was the result of
settings providing the fast Fourier transform (FFT) size 1066/
2048, frequency range to 5476 Hz, window length 33 ms and
a Hanning Window with a bandwidth of 30 Hz. Each spectrogram displayed the undulations of the partials over the length
of a selected note. These undulations which represented vibrato
cycles were measured by selecting one of the clearly displayed
high partials, manually placing a computer cursor over the peak
and trough of each vibrato cycle in that partial and registering
the corresponding time and frequency from the spectrogram
onto a spreadsheet (Excel, Microsoft Office 2000, Redmond,
WA). High partials were used, as resolution increases with
the partial number.56 From the spreadsheet the registered frequencies were divided by the number of the partial to establish
their related fundamental frequencies.
To ascertain vibrato extent, the maximum departure from the
average fundamental frequency was measured by taking each
vibrato cycle, calculating the distance from peak to trough in
semitones and dividing the result by two. Hertz were converted
into semitones using the formula:
12 log10 F0 log10 16:35
F0 in ST
log10 2
Vibrato rate was measured from the time difference between
adjacent vibrato peaks and expressed in cycles per second. For
both vibrato rate and vibrato extent, the mean values and corresponding standard deviations were calculated for the following:
FIGURE 2. The 11 notes acoustically assessed for vibrato rate and vibrato extent.
186
FIGURE 3. The mean and median vibrato rates for the solo, and the mean fastest and mean slowest vibrato rates from each singers 11 long held
notes per solo.
RESULTS
Pre- to posttest vibrato rate changes for the group as a whole are
presented in Figure 3.
As indicated in Figure 3, after the breathing imagery, although
no significant change occurred in the group mean (mean
change 0.13, P 0.15) or median (mean change 0.06,
P 0.47) vibrato rates, there was a significant reduction in the
range of mean vibrato rates for the group (mean change 0.52,
P 0.02). In contrast, after the Braille music imagery, there
was a significant reduction in the group mean (mean
change 0.17, P 0.01) and median (mean change 0.18,
P 0.01) vibrato rates, and no significant reduction in the range
of mean vibrato rates for the group (mean change 0.19,
P 0.10). After the cloze passage on breathing, no significant
changes were found in either the group mean (mean
change 0.0, P 0.87) or median (mean change 0.03,
P 0.38) vibrato rates, nor in the range of mean vibrato rates
for the group (mean change 0.05, P 0.64).
Each singers pre- and post-intervention mean vibrato rate is
presented in Figure 4.
Breathing imagery
6.4
1
6.3
6.4
1
6.3
6.2
6.2
6.2
6.1
6.1
6.1
6.0
6.0
5.9
5.9
5.8
5.7
5.6
5
6
5.5
5.4
5.3
5.8
5.7
5.9
5.6
5.5
5.4
5.3
4
6
5.8
5.6
5.4
5.1
5.1
5.1
5.0
5.0
4.9
4.8
After
5.0
4.9
Before
5.3
5.2
4.8
5.5
5.2
5.7
5.2
4.9
6.0
2
Cycles per second
Cloze passage
Braille imagery
6.4
6.3
4.8
Before
After
Before
FIGURE 4. Mean vibrato rates for each of the six singers, pre- and post-intervention.
After
Lynda Moorcroft, et al
TABLE 2.
Results of the Repeated Measures ANOVA for Mean
Vibrato Rates in Each Solo
Condition
187
P Value for
Change Over Time
(Pre vs Post)
P Value for
Differences
Between 11
Notes
0.001
<0.0001
0.95
0.013
0.08
0.11
Breathing imagery
Braille imagery
Cloze passage
Breathing imagery
7.50
7.00
6.50
6.00
Before
5.50
After
5.00
4.50
Singer
1
Singer
2
Singer
3
Singer
4
Singer
5
Singer
6
Braille imagery
7.50
7.00
6.50
Before
6.00
After
5.50
5.00
4.50
Singer
1
Singer
2
Singer
3
Singer
4
Singer
5
Singer
6
Cloze passage
7.50
7.00
6.50
Before
6.00
After
5.50
5.00
4.50
Singer
1
Singer
2
Singer
3
Singer
4
Singer
5
Singer
6
FIGURE 5. Variation in mean vibrato rates for each of the 11 notes assessed per solo.
188
P 0.180). For the non-imagery cloze passage on breathing, no
significant pre- to post-test findings resulted from either
subgroup.
In addition, within individual notes the cyclic undulations
comprising the vibrato became more regular after the breathing
imagery. A significant reduction in notes with vibrato rate SDs
greater than 0.50 cycles/second was found after the breathing
imagery (t 3.90, P 0.002) but not after the Braille imagery
(t 0.13, P 0.902) or cloze passage (t 1.10, P 0.288).
As opposed to vibrato rate, no trends or significant changes
resulted for vibrato extent from any intervention. Mean vibrato
extents for each group as a whole for all six conditions were between 0.91 and 0.95 semitones from the target pitch or
average fundamental frequency. Mean vibrato extents for
each of the solos were between 0.49 semitones (singer 1)
and 1.60 semitones (singer 5). Singer 1 registered the smallest
mean vibrato extent for the solo for all six conditions. Her
smallest value for an individual note, at 0.14 semitones, was
verging on a straight tone. Singer 5 registered the largest
mean vibrato extent for the solo for all six conditions. For individual notes, she exceeded 1.50 semitones in all six conditions, with her largest value being 1.95 semitones. However,
apart from singer 2, all pre-intervention to post-intervention
change in vibrato extent was less than 0.1 semitones.
Results for SPL showed no consistent association with either
vibrato rate or vibrato extent, indicating that pre- to post-test
changes in dynamics were not responsible for the changes
observed in vibrato measurements. Results for both vibrato
rate and vibrato extent were also found to be uninfluenced by
the order of the interventions.
DISCUSSION
The calm lyrical nature of the Villa-Lobos excerpt called for a
moderate vibrato rate for optimal performance. Group mean
vibrato rates for each of the six conditions were between 5.46
and 5.63 cycles/second, and were well within acceptable limits
for vibrato as proposed by Titze.48 Nevertheless, assessment of
the individual solos revealed a broad range of vibrato rates
which, for the least experienced singers in the group, was often
problematic. Singers 1 and 2 often had very fast, unstable
vibrato, and singer 3 had a particularly slow vibrato. In accordance with reports that vibrato rates tend to be more moderate
and regular the more skilled the singer,4247,57 the other, more
experienced singers exhibited no problems of excessively
fast, slow, or unstable vibrato. Yet for all six singers, vibrato
rates adopted patterns of change according to the intervention
undertaken and according to each singers pre-existing
strengths and weaknesses.
Following the breathing imagery, a number of changes closely
associated with skilled singing were noted. The cycles-to-cycle
regularity of vibrato undulations and note-to-note vibrato rate
stability improved significantly. The average standard deviation
of note-to-note vibrato rates, at 0.24 cycles/second, matched
that reported by Prame50 for professional recordings of Schuberts Ave Maria. Pre-intervention values of 6.00 cycles/second or faster become significantly slower, and those slower
Lynda Moorcroft, et al
189
for the solo showed the least change of any intervention, and the
note-to-note progression of vibrato rates and cycle-to-cycle regularity although fluctuating somewhat, showed no specific
trends.
Unlike vibrato rate, the values for vibrato extent showed no
significant changes following any intervention, nor were any
pronounced trends observed. For all six conditions, the group
mean vibrato extents (between 0.91 and 0.95 semitones
from the target pitch or average fundamental frequency) were
well within acceptable limits considering mean vibrato extents
of between 0.71 semitones from professional Lieder recordings56 and 1.4 semitones from professional opera singers67
have been documented.
Supporting the observation of Titze et al46 that beginner
singers often display minimal vibrato extent, the youngest
singer in the group, at aged 19 years, consistently had the smallest vibrato extent. By contrast, the oldest singer, at aged
49 years, consistently had the largest vibrato extent, exceeding
1.50 semitones for individual notes in all six conditions. This
brought her large vibrato extent into that area (1.5 semitones
and above) where Howes51 noted that listener-judges tended
to rate singers negatively. Such large values, however, appeared
to be beyond the control of the singer who expressed discontent
with her own vibrato, which according to her self-observation
had increased because of the onset of menopause. Whether
her perceived vocal change was associated with menopause
or other factors was beyond the scope of the present article.
Nevertheless, the perception of the singer was consistent with
the findings of Sundberg et al68 that professional singers often
experience increased vibrato extent as a result of aging.
To date, studies investigating whether a link exists between
dynamic intensity (as indicated by SPL) and either vibrato
rate or vibrato extent have produced varying conclusions. It remains unclear whether links between SPL and vibrato may
depend on a singers individual vocal technique, amount or
type of study and proficiency. However, in general, this study
supports the findings of Michel and Myers,69 Michel and Grashel,70 and Shipp et al67 that changes in vibrato extent are not
consistently related to changes in dynamic intensity, and findings as reported by McLane,71 Michel and Grashel,70 Murbe
et al47 and Shipp et al67 that vibrato rate is not intrinsically a
function of dynamic intensity.
CONCLUSION
Although the breathing imagery consistently had the greatest
impact on the less experienced singers, the vibrato rate of all
singers improved after the breathing imagery in a manner normally observed in voices following vocal warm-up54,55 and
after extensive training.47 That is, singers displayed (i) more
evenness in the cycle-to-cycle undulations comprising the
vibrato rate of a note, (ii) more note-to-note stability when
the vibrato rates of sustained notes in a solo are compared,
and (iii) a moderating of excessively fast and excessively
slow vibrato rates. By contrast, the Braille music imagery produced significantly slower and generally more erratic vibrato
rates, whereas the non-imagery task produced no significant
190
24. Linklater K. Freeing the Natural Voice. New York, NY: Drama Book Publishers; 1976.
25. Rodenburg P. The Right to Speak. Working with the Voice. London, UK:
Methuen; 1992.
26. Hanrahan C, Tetreau B, Sarrazin C. Use of imagery while performing dance
movement. Int J Sport Psychol. 1995;26:413430.
27. Sweigard L. Human Movement Potential: Its Ideokinetic Facilitation. New
York, NY: Dodd, Mead and Co; 1975.
28. Nordin S, Cumming J. Professional dancers describe their imagery: where,
when, what, why, and how. Sport Psychol. 2005;19:395416.
29. Alcantara P. Indirect Procedures. A Musicians Guide to the Alexander
Technique. Oxford, UK: Clarendon Press; 1997.
30. Bartley J, Clifton-Smith T. Breathing Matters. Auckland, New Zealand:
Random House; 2006.
31. Stoyva J. Stress management: what can we learn from the meditative disciplines? In: Kenny D, Carlson J, McGuigan F, Sheppard J, eds. Stress and
Health. Sydney, Australia: Harwood Academic Publishers; 2000:179193.
32. Doscher B. The Functional Unity of the Singing Voice. 2nd ed. Lanham,
MD and London, UK: The Scarecrow Press Inc; 1994.
33. Yurisich G. Gregory Yurisich: ANATS Australian tour. Voice of ANATS.
2000;13:56. 14.
34. Patenaude-Yarnell J. Sensations of singing. A look at time honoured
maxims, descriptions and images. J Sing. 2003;60:185189.
35. Marks M. Fears, Phobias, and Rituals: Panic, Anxiety, and Their Disorders.
New York, NY: Oxford University Press; 1987.
36. Aziz-Zadeh L, Ivry R. The human mirror neuron system and embodied representations. Adv Exp Med Biol. 2009;629:355376.
37. Filamon F, Nelson J, Hagler D, Sereno M. Mirror neurons for execution,
observation, and imagery. Neuroimage. 2007;37:13151328.
38. Gazzola V, Rizzolatti G, Wicker B, Keysers C. The mirror neuron system
responds to human robotic actions. Neuroimage. 2007;35:16741684.
39. Johnson-Frey S, Maloof F, Newman-Norlund R, Farrer C, Inati S, Grafton S.
Action or hand-object interactions? Neuron. 2003;39:10531058.
40. Hurley S. The shared circuits model (SCM): How control, mirroring, and
simulation can enable imitation, deliberation, and mindreading. Behav
Brain Sci. 2008;31:158.
41. Bartholomew W. A survey of recent voice research. M.T.N.A. Proceedings.
Report of the Music Teachers National Association. 1937; 115138.
42. Bartholomew W. A physical definition of good voice-quality in the male
voice. J Acoust Soc Am. 1934;6:2533.
43. Robison C, Bounous B, Bailey R. Vocal beauty: a study proposing its acoustical definition and relevant causes in classical baritones and female belt
singers. J Sing. 1994;51:1930.
44. Sundberg J. The Science of the Singing Voice. Dekalb, IL: Northern Illinois
University Press; 1987.
45. Sundberg J. Acoustic and psychoacoustic aspects of vocal vibrato. In:
Dejonckere P, Hirano M, Sundberg J, eds. Vibrato. London, UK: Singular
Publishing Group, Inc; 1995:3562.
46. Titze I, Story B, Smith M, Long R. A reflex resonance model of vibrato. J
Acoust Soc Am. 2002;111(pt 1):22722282.
47. M
urbe D, Zahnert T, Kuhlisch E, Sundberg J. Effects of professional
singing education on vocal vibrato - a longitudinal study. J Voice. 2007;
21:683688.
48. Titze I. Principles of Voice Production. Englewood Cliffs, NJ: PrenticeHall Inc; 1994.