- Email: [email protected]
Chest wall dynamics and muscle recruitment during professional flute playing
Available online at www.sciencedirect.com
Respiratory Physiology & Neurobiology 160 (2008) 187–195
Chest wall dynamics and muscle recruitment during professional flute playing Isabelle Cossette a,∗ , Pierpaolo Monaco a , Andrea Aliverti b , Peter T. Macklem c a
Schulich School of Music, McGill University, Canada b Politecnico di Milano c Faculty of Medicine, McGill University, Canada Accepted 18 September 2007
Abstract Respiratory parameters and sound were recorded during professional flute playing in order to assess what physiological processes were associated with the control of sound production that results in ‘breath support’ which in turn is associated with high quality playing. Four standing young professional flautists played flute excerpts with and without breath support. Recordings included optoelectronic plethysmographic measurements of chest wall volume (Vcw ) and its compartments, surface electromyography of the scalene, lateral abdominal, rectus abdominus, parasternal and sternocleidomastoid muscles, mouth pressure, and sound. Flow was estimated from differentiating Vcw during playing. Results showed that flute support entails antagonistic contraction of non-diaphragmatic inspiratory muscles that tends to hold the rib cage at higher lung volume. This relieves the expiratory muscles from the task of producing the right mouth pressure, especially at the end of the phrases, so they can contribute more to the finer control of mouth pressure modulations required for high quality playing. © 2007 Elsevier B.V. All rights reserved. Keywords: Respiratory patterns; Respiratory muscles; Abdominal muscles; Intercostal muscles; Scalene muscles; Sternocleidomastoid muscles; Chest wall configuration; Pulmonary volumes; Mouth pressures; Flow; Musicians; Flute players; Breath support; Flautists; Chest wall; Chest wall dynamics
1. Introduction Although it is widely accepted both in the musical field and respiratory physiology that the way one breathes and blows and the pressures and flows developed are important determinants of wind instrument performance and singing (Bouhuys, 1964, 1968; Bouhuys et al., 1966; Roos, 1936), knowledge of the mechanics of breathing during wind instrument playing is meager. Music educators and performers commonly refer to ‘breath support’ in playing the flute and other wind instruments, yet the term ‘breath support’ is neither well-defined nor consistently used. In order to achieve the required breath support (BS), some students are instructed to contract the diaphragm while others are recommended to use the rib cage muscles. Spillane (1989) reported over 90 different directives for ‘breath support’! The survey showed that the wide variety of instructions given to ∗
Corresponding author at: Music Education Area, Department of Music Research, Schulich School of Music, McGill University, 555 Sherbrooke Street West, Montreal, Quebec, Canada H3A 1E3. Tel.: +1 514 398 4535; fax: +1 514 398 8061. E-mail address: [email protected] (I. Cossette). 1569-9048/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.resp.2007.09.009
singing students to achieve support seem to be based more on opinions and personal experience than on proven facts. Because of this confusion, defining ‘breath support’ accurately is not possible but for the sake of clarity of this article, we refer to, ‘breath support’ as the physiological processes used by the player to control precisely the sound production which is directly linked to the control of air flow, air velocity and pressure required to play a specific note or a musical passage. Even though breath support is not defined, there is general agreement that during flute playing, a low-pressure wind instrument (Brown and Thomas, 1990), and during singing, inspiratory muscles are recruited as antagonists to the expiratory act of playing. At high lung volumes the elastic recoil of the respiratory system provides a pressure (approximately 35 cmH2 O) that is too high for most of the notes produced on the flute. This must be counterbalanced by inspiratory muscle contraction (Bouhuys, 1964). We hypothesize that this is accomplished by non-diaphragmatic inspiratory muscles. The diaphragm is activated during specialized manoeuvres such as vibrati, staccati and abrupt decreases of pressure during flute playing (Cossette et al., 2000) or during singing (Leanderson et al., 1987) but because each wind instrument has its specific pressures, flows and air
188
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
Fig. 1. Score of Poulenc Sonata second movement (Cantilena) first 10 bars as performed by each flautist. The 53 notes are numbered as used in the analysis.
velocities requirements (Brown and Thomas, 1990), generalization of respiratory strategies is impossible (Bouhuys, 1964). The antagonistic activity of inspiratory muscles to produce the appropriate mouth or subglottic pressures for flute playing or singing gives no information on how support is achieved. In fact, we could find no previous studies designed to determine how respiratory muscles provide support during flute playing and which muscles are used. In order to provide this knowledge, we were faced with a major problem: because the ideas about what constitutes support are quite diverse, individual flautists probably ‘support’ in different ways. Nevertheless, we hoped that we might find some features common to all flute players that would allow us to generalize and develop a hypothesis about how support was achieved that might eventually lead to a clear cut definition of what support is. Our aim in the present study therefore, was to identify: (1) the rib cage and abdominal kinematics associated with support; (2) the muscles that were recruited to produce support; (3) how these muscles were coordinated to achieve the respiratory patterns associated with support; (4) how this allowed better control of the required pressures and flows to play the flute. Because the elastic recoil of the respiratory system decreases continually during the playing of a phrase, we performed our measurements as a function of lung volume. 2. Methods 2.1. Subjects and protocol We measured chest wall displacements, electromyography (EMG) of the respiratory muscles, mouth pressure and recorded sound during professional flute playing with (BS) and without breath support (NBS) in four young professional flautists
(three males and one female aged from 29 to 34). Three were recruited at La Scala di Milano or amongst the finishing students of schools associated with La Scala. The fourth subject, the first author of this article, had obtained similar or higher level diplomas in Canadian institutions and had extensive performance experience. Consent sheets were signed by all subjects in accordance to ethics approval granted by the institutions involved. The same flute was used for all subjects (Sankyo, Etude model, with RS1 Cooper head joint). After performing quiet breathing and two vital capacity manoeuvres, the subjects were asked to play an excerpt from the second movement of the Poulenc Sonata (Fig. 1) which is a standard piece of the repertoire of professional flautists. It consists of long soft phrases in the three registers of the flute. Subjects were asked to play this piece with and without support. We did not give any definition of support to the flautists as we wanted them to play in as natural way as possible what they considered to be support and no support. The subjects were also asked to play without vibrato. As musicians train themselves to play with vibrato, it is not obvious to play without it. As a result, vibrato was seen on the sound tracings of both conditions, with and without support, though at different amplitudes. The length of the musical phrases varied between both conditions (from 47 to 56 s long) even though the performers played exactly the same musical passage. For the analysis, we separated each of the 53 notes (Fig. 1) from the recorded sound signals so we could compare similar events. That allowed us to compute average value of the recorded quantities within each note and each musical phrase. 2.2. Measurements and data processing For the first three subjects, recordings included measurements of chest wall volume (Vcw ) and the volumes of its compart-
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
ments, i.e. the lung and diaphragm-apposed parts of the rib cage (Vrcp and Vrca , respectively) and the abdomen (Vab ), by OptoElectronic Plethysmography (OEP, Elite System, BTS), surface electromyography of the respiratory muscles (scalene, rectus abdominus, parasternals, sternocleidomastoids, and lateral abdominal muscles, i.e., internal and external obliques and transversus combined), mouth pressure and sound recording during playing. Flow was measured by differentiation of the volume signals obtained by OEP during flute playing. The measurements in the fourth subject were acquired during two different experiments. The first experiment (referred to as 4a) included OEP measurements of chest wall volumes and flow, measurements of mouth pressure (Pm ) and sound recording. Electromyography of the respiratory muscles and sound recording were acquired during the second experiment (4b). OEP consisted of seven infrared video cameras (five in front and two behind the subjects both at 2 m) tracking 89 hemispherical 6 mm diameter reflective markers. These were applied to the surface of the chest wall, front and back in seven rows between the clavicles and the iliac crest (Aliverti et al., 2000). The 3D coordinates of the markers were converted into volumes using Gauss’ theorem as described in detail by Cala et al. (1996). Because of movement artefacts or of the required holding position of the flute, some of the 89 markers were occasionally hidden. When too many markers were missing for the reconstruction, the data was rejected. This only happened for the vital capacity manoeuvre of subject 3. The data from the OEP was stored on an IBM compatible PC at 50 Hz during flute playing and at 25 Hz during respiratory manoeuvres. Prior to flute playing, subjects breathed quietly and performed two vital capacity manoeuvres (VC) during which flow was measured with a pneumotachometer (HR 4700-A; Hans Rudolph Inc., Kansas City, MO) at the mouth connected to a pressure transducer (Celesco, ±5 cmH2 O) and recorded on the same computer using an analog-to-digital converter (RTI 800; Analog Devices; Norwood, MA). The following parameters were compared during support and without support playing: Vcw , average flow per phrase obtained by dividing Vcw by the time to complete the phrase and the relative contribution of the CW compartments. The functional residual capacity (FRC) was estimated by averaging the end expiratory volumes during quiet breathing. For each phrase, the curves of the chest wall, abdomen and total rib cage (Vrcp + Vrca ) were integrated over time and the area (in units of ml s) above and below FRC was calculated. The area above FRC was then expressed in percentage of the total area (above FRC + below FRC) and compared between support and without support. In addition, we calculated the number of the notes played at lung volumes lower than the FRC level in the two conditions and calculated their percentage in respect to the total number of notes played during the excerpt. We then used a χ2 test (Mantel–Haenszel correction) based on a 2 × 2 contingency table to evaluate the statistical significance of their difference. We refer to Vcw at the beginning of the musical phrases as the initial lung volume; at the end of the phrases, termination lung volume (Thomasson and Sundberg, 1997).
189
Mouth pressure (Pm ) was measured with a catheter placed at the corner of the mouth during flute playing. The catheter was connected to a pressure transducer (SCX01; Sensym, Milpitas, CA, ± 100 cmH2 O) and stored on a IBM compatible PC via the OEP System at 1000 Hz. Previous studies (Cossette et al., 2000; Bouhuys, 1964) used an air-filled balloon placed at the back of the throat. The less invasive method used in the present study was chosen in order to better represent natural performance. In subject 3, Pm measurements were discarded because of blockage of the catheter by saliva. In order to avoid blockage, a very small flow was put through the catheter for the subjects 1, 2 and 4 and the pressure resulting from the flow was deducted subsequently. For both conditions (support and without support), a band pass filter (4–10 Hz) was used in order to quantify the vibrato content and a low pass filter (2.5 Hz) was used to compare slower fluctuations of Pm . An average within each note and per phrase was computed. Surface electromyograms were recorded at 1000 Hz using a telemetric eight-channel system (TELEMG, BTS, Milan, Italy) and stored on the IBM computer via the OEP System. Movements, done in order to check muscle activity, consisted of inward abdominal displacement for the lateral abdominals; flexing the neck against a resistance on the forehead for the scalenes; taking a big rib cage inspiration for the parasternals; lifting the legs when sitting at the edge of a chair for rectus abdominus; and turning the head against a resistance for the sternocleidomastoids. As some muscle acquisitions contained large electrocardiogram (ECG) content, a cross-correlation technique was used for ECG removal before the signals were rectified and integrated. This technique consisted of four steps: (1) identification, with the help of a graphic interface, of the QRS complex of the ECG signal (template) during a period of relative EMG inactivity; (2) search of all the other ECG complexes across the signal by computation of a correlation coefficient between the template and a same-length part of the signal; (3) comparison and adjustment of the template amplitude according to the complexes found; (4) subtraction of the ECG signals. This method was adapted from the one suggested by Levine et al. (1986). The signals were then rectified and low pass filtered (2 Hz, Butter window). For each musical phrase per subject, the activation was expressed as percentage of the maximal activation during support playing. The length of each phrase, generally different between the two conditions, was normalised to 100%. Comparisons of the average activation between with and without support were performed for every 10% sections of the total length of each phrase. The data were then ensemble averaged over the different phrases within each subject and over all the phrases for all the subjects. In addition, for the most representative muscles, namely lateral abdominals and scalenes the EMG activation was plotted against Vcw relative to FRC, calculated over the same time periods and ensemble averaged individually and for the group. In subject 3, the parasternal EMG was clearly contaminated by expiratory intercostals adjacent to the parasternals as evidenced by the strong recruitment at the end of some phrases during without support. Because this contamination seriously
190
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
interfered with the ensemble averaged, we do not show the results for these muscles. During playing, sound was recorded at 44,100 Hz (Sony Tape recorder, TC-D5 no. 11631, Sennheiser Microphone (Phantom Power) MKH-416T no. 17551, Sennheiser Microphone Power supply MZA 16T no. 110030) and also stored on the IBM compatible PC at 1000 Hz as a synchronization reference. The microphone was placed at a fixed position for each subject around 1 m from the opening of the flute. Because they had a catheter measuring the mouth pressure and were filmed by cameras, the subjects who played standing kept a relatively fixed position. The sound was analysed by quantifying the frequency and intensity variation between support and without support. The frequency content of each note was computed by the fast Fourier transform (FFT) (number of samples: 35,790 ± 25,412; flute playing frequency content 349–1480 Hz; sampling rate 44,100 Hz). The sound was then rectified and a linear interpolation function was used to get the envelope of the sound. The mean of this function within each note gave us the averaged intensity for that note. The combined measurements of Vrcp , Vrca , Vab simultaneously with the EMG’s, mouth pressure and sound allowed us to see if there were any differences in muscle activation and action between support and non-support. 3. Results 3.1. Flows, chest wall volumes and thoraco-abdominal coordination The average flows per phrase shown in Fig. 2 were clearly smaller during playing with support in subjects 2 and 3, whereas
Fig. 2. Average flow per phrase with support (black dots) and without support (white dots) for each subject.
in subjects 1 and 4 they did not show consistent difference between the two conditions. Fig. 3 shows volume tracings of the chest wall (Vcw ). The slopes of these traces illustrate that flow tended to be higher and started more abruptly without support (NS) than with support (S) in subjects 2 and 3. The dashed horizontal line in the figures denotes functional residual capacity (FRC) and the open dots indicate the beginning and end of each of the 53 notes. Without support, all subjects consistently ended playing below FRC during almost every phrase. With support, the termination lung volume in subjects 1 and 3 were generally well above FRC, while subjects 2 and 4 did not play as far below FRC with support. Whether starting at higher lung volumes and using the same flow (subjects 1 and 4) or starting at the same
Fig. 3. Tracings of chest wall volumes for all subjects during playing with breath support (left) and without breath support (right). The horizontal dotted line represents the FRC volume and each open circle represents the beginning and end of each of the 53 notes of the excerpt.
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
191
Table 1 Number of phrases for which the area of the tracings over FRC is equal, smaller or larger with support than without
Chest wall Pulmonary rib cage Abdominal rib cage Rib cage total (rcp + rca) Abdomen
Area over FRC BS = NBS
Area over FRC BS > NBS
Area over FRC NBS > BS
2 1 6 1 5
15 18 13 18 8
3 1 1 1 7
Bold shows the condition with the most phrases, i.e. area of tracings larger over FRC with support than without support.
Fig. 4. Scatter plots, for each compartment (chest wall, ribcage and abdomen), of the volume integral above functional residual capacity (FRC), expressed as percentage of the total area (above FRC + below FRC), during with and without support. Each dot represents a single phrase and the solid line represents the identity line.
total area (above FRC + below FRC), for each phrase with and without support. Most of the phrases were played with a higher RC volume above FRC with support than without. This resulted in a similar behaviour of the total chest wall volume despite the scattered distribution of the abdominal compartment. Table 1 shows, the number of phrases in which the area above FRC with support was higher, smaller or equal than without support for all the compartments. Although Fig. 4 shows that the total RC volumes were higher during support, it is worth underlining that the main contribution to total RC behaviour was provided by the pulmonary RC compartment (Table 1). This strongly implicates the nondiaphragmatic inspiratory muscles in the provision of support, because these are the muscles which act directly to expand the pulmonary rib cage whereas the diaphragm acts on the abdomen and abdominal rib cage (Aliverti et al., 1997). How the rib cage and abdomen were coordinated during support and without is demonstrated by a plot of rib cage volume versus abdominal volume (Vab ) as shown in Fig. 5 (Konno and Mead, 1967). With change in rib cage volume the trace moves vertically and with abdominal volume change, horizontally. All subjects showed greater rib cage expansion during support for at least one of the three phrases shown. The inspiration before the note begins ends in a triangle indicating the beginning of the musical phrase. With support 10 of 12 traces are displaced upward showing increased rib cage volume with support. In one of the tracings without upward displacement (phrase 3 in subject 3), the phrase ended with both the rib cage and the abdomen more expanded with support. 3.2. Respiratory muscle activation
lung volumes but using a smaller flow (subjects 2 and 3), the termination lung volumes were consistently higher with support than without. As a result, all subjects played more notes under FRC during no support (9, 25, 18 and 11 in subjects 1, 2, 3 and 4, respectively) than during support (1, 16, 0 and 5 for subjects 1, 2, 3, and 4 respectively). The average percentage of notes played under FRC was then significantly different between support and without support (32% during NBS and 11% during BS with a χ2 of 12.99 and p-value of 0.0003). For the different chest wall compartments of all subjects, Fig. 4 compares, in scatter plots, the volume integral above functional residual capacity (FRC), expressed as percentage of the
The particular behaviour of the rib cage during support is also confirmed by looking at the respiratory muscle activation and coordination during playing. Fig. 6 shows ensemble averaged muscles activation with and without support for all phrases played by the subjects. The non-diaphragmatic inspiratory muscles (scalene and sternocleidomastoids) were clearly more activated during support. The stronger activation of the scalene was striking in subjects 1, 3 and 4, whereas subject 2 recruited the sternocleidomastoids more. The behaviour of the expiratory muscles (lateral abdominals and rectus) was interesting. The lateral abdominal activation
192
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
Fig. 5. Konno–Mead diagrams for all subjects. Rib cage (Vrcp + Vrca ) and abdominal (Vab ) displacements plotted one against the other during the first three musical phrases played with support (thick solid line) and without support (dashed line). Triangles and circles indicate, respectively, initial lung volume and termination lung volume.
Fig. 6. Ensemble average of muscular activation (% of maximal activation with support for lateral abdominals, scalene, sternocleidomastoid and rectus muscles) every 10% of total length of phrases for all the phrases in all the subjects during support (black dots) and without support (white dots).
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
193
Fig. 7. Ensemble average of lateral abdominals and scalene muscular activation (% of maximal activation with support) every 10% of total length of phrases in relation to CW volumes (relative to FRC) during support (black dots) and without support (white dots). Included results: all six phrases for subjects 1 and 3, phrases 1, 4, 5, 6, and 7 for subject 2. The dotted vertical line represents FRC.
during support was higher than without support in the first part of the phrase. Then, activation kept increasing but more without support than with. Toward the end of the phrase, the increase in activation was sometimes higher than 100% of maximal support activation. This pattern can be identified in the rectus as well, but to a smaller extent. Its activation during support seemed to be constant throughout the phrase but remained higher than without support in the first part of the phrase and smaller in the last one. Fig. 7 shows scalene and lateral abdominal muscle activation in relation to CW volume relative to FRC every 10% section of the total length of the phrase. Both volumes and EMG activation are the ensemble average of all the phrases in each subject and in all subjects. The pooled data show that the scalene activation decreases as volume decreases whereas the opposite is the case for the lateral abdominals. In addition, the figure confirms the higher activation of the scalene during support playing (regardless of the volumes at which the flautist plays) and the strong lateral abdominals recruitment at the end of the phrases without support, right after the FRC level is reached (vertical dotted line). Thus both volume
(rather than time only) and condition (support versus without support) are determinants of activation. 3.3. Mouth pressures and sound analysis Table 2 shows that the variations in mouth pressure during the 53 notes, expressed as standard deviations, seemed greater during support than without support, indicating more vibrati during support than without support. Average mouth pressures were greater in subjects 1 and 2 during support but less in subject 4 (Table 3). Sound analysis was carried out to assess whether possible changes in the measured variables might be related to changes in sound parameters. There was no consistent difference between support and without support common to all subjects. A similar analysis carried out on flow computed by differentiating change in Vcw (the major determinant of sound intensity, Cossette et al., 2000), showed that there were no noticeable differences in intensity between support and without support. 4. Discussion The co-authors’ previous studies (Cossette et al., 2000; Cossette, 2002) showed that different professional flautists use
Table 2 Mouth pressure standard deviations (cmH2 O) with and without Phrase
Subject 1
Subject 2
Subject 4a
BS (cmH2 O)
NBS (cmH2 O)
BS (cmH2 O)
NBS (cmH2 O)
BS (cmH2 O)
NBS (cmH2 O)
1 2 3 4 5 6
0.249 0.411 0.383 0.164 0.259 0.295
0.074 0.135 0.174 0.170 0.190 0.097
0.181 0.242 0.276 0.209 0.159 0.088
0.110 0.061 0.102 0.068 0.185 0.100
0.265 0.379 0.435
0.184 0.325 0.312
Mean
0.308
0.135
0.192
0.096
0.362
0.275
Mean Pm values of vibrato content. Bold and underlined respectively show higher vibrato content during breath support and without breath support.
194
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
Table 3 Mean mouth pressure averaged per notes and per phrase with and without support Phrase
Subject 1
Subject 2
Subject 4a
BS (cmH2 O)
NBS (cmH2 O)
BS (cmH2 O)
NBS (cmH2 O)
BS (cmH2 O)
NBS (cmH2 O)
1 2 3 4 5 6
7.258 8.88 8.014 7.872 9.654 7.627
6.649 8.77 7.686 5.997 6.388 7.022
5.972 13.743 8.085 7.65 9.577 10.968
6.885 9.036 7.039 7.018 9.143 9.680
2.493 3.626 3.138
2.86 4.745 4.186
Mean
8.132
7.231
10.058
8.347
3.087
3.939
Bold shows higher mean Pm when playing with support; underlined shows higher mean Pm when playing without support.
different muscle recruitment patterns while playing at similar frequencies and intensities. The diaphragm was generally non-activated but was recruited during specific manoeuvres like staccati and octaves. One subject showed some diaphragmatic activity during vibrati. According to that study and studies of others (Bouhuys, 1964, 1968; Bouhuys et al., 1966; Roos, 1936), flautists use some inspiratory muscles during playing. As playing a wind instrument or singing is an expiratory activity, it is important to underline the involvement of inspiratory muscles during expiration that might be used as antagonist muscles to the elastic recoil of the respiratory system or as antagonistic action to the expiratory muscles. Without the contraction of some inspiratory muscles during the expiration, the elastic recoil would produce higher pressures than required for a low-pressure instrument like the flute. Although these studies have shown that inspiratory muscles are activated during an expiratory task, to our knowledge no previous study has examined how muscular recruitment is coordinated with CW volumes in relation to FRC and how it relates to ‘breath support’ associated with high-level performance. Pettersen and Westgaard (2004) studied neck and shoulder muscles activation in professionally classically trained singers but they did not deal with concept of “breath support”. Griffin et al. (1995) found acoustical and laryngeal differences between singing with and without support. Their results did not support the concept that respiratory muscle activity is related to support during singing (p. 51). However, the only respiratory measurements acquired in their study were done with respiratory inductive plethysmography from which only kinematic information can be obtained. In order to assess muscular activity, dynamics and/or muscle activation must be measured. Thorpe et al. (2001) looked at different levels of singing sound projection and at the respiratory patterns associated with the different levels of projection. There was no consistent change between enhanced and normal projection. The relevance of this to support is not clear, but termination lung volumes were higher during enhanced projection. We found the same during flute playing with support. The goal of our work was to determine features common to each flautist studied while playing with support which were not present in the absence of support. In this way, we hoped to define flute breath support, how it is achieved and which respiratory muscles accomplish it. As expected, we observed considerable
variability between individual. Nevertheless there were clear, common features that should allow generalizations to be made. Thus the primary objectives of this research were achieved. Flute breath support entails antagonistic contraction of nondiaphragmatic inspiratory muscles that tend to hold the rib cage at a higher lung volume. This allows relaxation pressures to provide expiratory pressures over a longer period of the phrase being played. Inspiratory muscle activation may require some increase in the activation of expiratory muscles to counteract the antagonistic inspiratory action as in subjects 1, 2 and 4 (Fig. 6). Nevertheless, the lung volume at which playing a phrase ends, is usually above FRC and rarely far below it (Fig. 3 and Table 1). Thus the strong expiratory muscle pressures required to play the flute at low lung volumes are generally avoided, while advantage is taken of the high relaxation pressures at high lung volumes. In the absence of support, there was strong activation of the lateral abdominals in all subjects at the end of musical phrases when lung volume is lowest and expiratory muscles must develop their greatest pressures. With support this activation was considerably less (Figs. 6 and 7). The avoidance of strong expiratory muscle activation, is helped by the higher rib cage volumes (Fig. 4) which keep total chest wall volume at a higher overall level where relaxation pressures relieve the expiratory muscles of the task. This leads us to speculate that these muscles are being saved for other jobs; that they are used for fine tuning and specialized tasks such as vibrati (Table 2) and staccati. If the expiratory muscles were contracted in order to decrease lung volume and maintain the mouth pressures required to play the flute, the less they would be able to modulate the mouth pressures in finely controlled way. And fine control of mouth pressure is surely a sine qua non of high-quality wind instrument playing. We tentatively put forward the following definition: flute breath support is a mechanism to avoid the recruitment of expiratory muscles in order to decrease lung volume during playing so that they can best exert fine control over the mouth pressure modulations required for high quality playing without being encumbered by other tasks. This is achieved by inspiratory muscle recruitment as demonstrated by their greater electrical activation which keeps the rib cage expanded, lung volumes higher and the expiratory muscles relatively relaxed. This was demonstrated by their decreased activation during support playing at the end of the phrases.
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
Our results may apply to singing as well as flute playing as both are low-pressure expiratory activities. Thus singers may use support mechanisms similar to flautists to allow the fine modulation of subglottic pressures by expiratory muscles. Our tentative definition of support is, at this stage, still a hypothesis that can be further tested. Yet it is virtually axiomatic that a contracted muscle will not be able to modulate force to do another task as well as a relaxed muscle. If one were to put a weight on the back of a piano player’s hands, to keep the extensor digitorum muscle tonically contracted, the pianist would have great difficulty in playing properly. We conclude that there are features in common among flute players playing with support, that allow hypothesis formation and a tentative definition of what support is. If the hypothesis and definition are held up by further research, what is now confused will be considerably clarified. Hopefully music teachers will soon be in a position to teach support based upon sound physiological principles rather than intuition. Acknowledgements Supported by the Social Sciences and Humanities Research Council of Canada and the Canadian Institutes of Health Research. References Aliverti, A., Cala, S.J., Duranti, R., Ferrigno, G., Kenyon, C.M., Pedotti, A., Scano, G., Sliwinski, P., Macklem, P.T., Yan, S., 1997. Human respiratory muscle actions and control during exercise. J. Appl. Physiol. 83, 1256–1269. Aliverti, A., Dellac´a, R., Pelosi, P., Chiumello, D., Pedotti, A., Gattinoni, L., 2000. Optoelectronic plethysmography in intensive care patients. Am. J. Respir. Crit. Care Med. 161, 1546–1552.
195
Bouhuys, A., 1964. Lung volumes and breathing patterns in wind-instrument players. J. Appl. Physiol. 19 (5), 967–975. Bouhuys, A., 1968. Pressure-flow events during wind instrument playing. Ann. N. Y. Acad. Sci., 264–275. Bouhuys, A., Proctor, D.F., Mead, J., 1966. Kinetic aspects of singing. J. Appl. Physiol. 21 (2), 483–496. Brown, S.E., Thomas, M., 1990. Respiratory training effects in wind and brass instrumentalists. Med. Probl. Perform. Artists 5, 4. Cala, S.J., Kenyon, C.M., Ferrigno, G., Carnevali, P., Aliverti, A., Pedotti, A., Macklem, P.T., Rochester, D.F., 1996. Chest wall and lung volume estimation by optical reflectance motion analysis. J. Appl. Physiol. 81 (6), 2680– 2689. Cossette, I., Sliwinski, P., Macklem, P.T., 2000. Respiratory parameters during professional flute playing. Respir. Physiol. 121, 33–44. Cossette, I., 2002. M´ecanique respiratoire des flˆutistes professionnels. Rev. Mal. Respir. 19, 197–206. Griffin, B., Woo, P., Colton, R., Casper, J., Brewer, D., 1995. Physiological characteristics of the supported singing voice. A preliminary study. J. Voice 9 (1), 45–56. Konno, K., Mead, J., 1967. Measurement of the separate volume changes of rib cage and abdomen during breathing. J. Appl. Physiol. 22 (3), 407– 422. Leanderson, R., Sundberg, J., von Euler, C., 1987. Role of diaphragmatic activity during singing: study of transdiaphragmatic pressures. J. Appl. Physiol. 62, 259–270. Levine, S., Gillen, J., Weiser, P., Gillen, M., Kwatny, E., 1986. Description and validation of an ECG removal procedure for EMGdi power spectrum analysis. J. Appl. Physiol. 60, 1073–1081. Pettersen, V., Westgaard, R.H., 2004. The activity patterns of neck muscles in professional classical singing. J. Voice 19 (2), 238–251. Roos, J., 1936. The physiology of playing the flute. Archiv. N´eerlandaises Phon´etique Exp. 12, 1–26. Spillane, K.W., 1989. Breath support directives used by singing teachers: a Delphi study. NATS J. 45 (3), 9–57. Thomasson, M., Sundberg, J., 1997. Lung volume levels in professional classical singing. Logopedia Phoniatrica Vocologia 22, 61–70. Thorpe, C.W., Cala, S.J., Chapman, J., Davis, P.J., 2001. Patterns of breath support in projection of singing voice. J. Voice 15 (1), 86–104.
Respiratory Physiology & Neurobiology 160 (2008) 187–195
Chest wall dynamics and muscle recruitment during professional flute playing Isabelle Cossette a,∗ , Pierpaolo Monaco a , Andrea Aliverti b , Peter T. Macklem c a
Schulich School of Music, McGill University, Canada b Politecnico di Milano c Faculty of Medicine, McGill University, Canada Accepted 18 September 2007
Abstract Respiratory parameters and sound were recorded during professional flute playing in order to assess what physiological processes were associated with the control of sound production that results in ‘breath support’ which in turn is associated with high quality playing. Four standing young professional flautists played flute excerpts with and without breath support. Recordings included optoelectronic plethysmographic measurements of chest wall volume (Vcw ) and its compartments, surface electromyography of the scalene, lateral abdominal, rectus abdominus, parasternal and sternocleidomastoid muscles, mouth pressure, and sound. Flow was estimated from differentiating Vcw during playing. Results showed that flute support entails antagonistic contraction of non-diaphragmatic inspiratory muscles that tends to hold the rib cage at higher lung volume. This relieves the expiratory muscles from the task of producing the right mouth pressure, especially at the end of the phrases, so they can contribute more to the finer control of mouth pressure modulations required for high quality playing. © 2007 Elsevier B.V. All rights reserved. Keywords: Respiratory patterns; Respiratory muscles; Abdominal muscles; Intercostal muscles; Scalene muscles; Sternocleidomastoid muscles; Chest wall configuration; Pulmonary volumes; Mouth pressures; Flow; Musicians; Flute players; Breath support; Flautists; Chest wall; Chest wall dynamics
1. Introduction Although it is widely accepted both in the musical field and respiratory physiology that the way one breathes and blows and the pressures and flows developed are important determinants of wind instrument performance and singing (Bouhuys, 1964, 1968; Bouhuys et al., 1966; Roos, 1936), knowledge of the mechanics of breathing during wind instrument playing is meager. Music educators and performers commonly refer to ‘breath support’ in playing the flute and other wind instruments, yet the term ‘breath support’ is neither well-defined nor consistently used. In order to achieve the required breath support (BS), some students are instructed to contract the diaphragm while others are recommended to use the rib cage muscles. Spillane (1989) reported over 90 different directives for ‘breath support’! The survey showed that the wide variety of instructions given to ∗
Corresponding author at: Music Education Area, Department of Music Research, Schulich School of Music, McGill University, 555 Sherbrooke Street West, Montreal, Quebec, Canada H3A 1E3. Tel.: +1 514 398 4535; fax: +1 514 398 8061. E-mail address: [email protected] (I. Cossette). 1569-9048/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.resp.2007.09.009
singing students to achieve support seem to be based more on opinions and personal experience than on proven facts. Because of this confusion, defining ‘breath support’ accurately is not possible but for the sake of clarity of this article, we refer to, ‘breath support’ as the physiological processes used by the player to control precisely the sound production which is directly linked to the control of air flow, air velocity and pressure required to play a specific note or a musical passage. Even though breath support is not defined, there is general agreement that during flute playing, a low-pressure wind instrument (Brown and Thomas, 1990), and during singing, inspiratory muscles are recruited as antagonists to the expiratory act of playing. At high lung volumes the elastic recoil of the respiratory system provides a pressure (approximately 35 cmH2 O) that is too high for most of the notes produced on the flute. This must be counterbalanced by inspiratory muscle contraction (Bouhuys, 1964). We hypothesize that this is accomplished by non-diaphragmatic inspiratory muscles. The diaphragm is activated during specialized manoeuvres such as vibrati, staccati and abrupt decreases of pressure during flute playing (Cossette et al., 2000) or during singing (Leanderson et al., 1987) but because each wind instrument has its specific pressures, flows and air
188
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
Fig. 1. Score of Poulenc Sonata second movement (Cantilena) first 10 bars as performed by each flautist. The 53 notes are numbered as used in the analysis.
velocities requirements (Brown and Thomas, 1990), generalization of respiratory strategies is impossible (Bouhuys, 1964). The antagonistic activity of inspiratory muscles to produce the appropriate mouth or subglottic pressures for flute playing or singing gives no information on how support is achieved. In fact, we could find no previous studies designed to determine how respiratory muscles provide support during flute playing and which muscles are used. In order to provide this knowledge, we were faced with a major problem: because the ideas about what constitutes support are quite diverse, individual flautists probably ‘support’ in different ways. Nevertheless, we hoped that we might find some features common to all flute players that would allow us to generalize and develop a hypothesis about how support was achieved that might eventually lead to a clear cut definition of what support is. Our aim in the present study therefore, was to identify: (1) the rib cage and abdominal kinematics associated with support; (2) the muscles that were recruited to produce support; (3) how these muscles were coordinated to achieve the respiratory patterns associated with support; (4) how this allowed better control of the required pressures and flows to play the flute. Because the elastic recoil of the respiratory system decreases continually during the playing of a phrase, we performed our measurements as a function of lung volume. 2. Methods 2.1. Subjects and protocol We measured chest wall displacements, electromyography (EMG) of the respiratory muscles, mouth pressure and recorded sound during professional flute playing with (BS) and without breath support (NBS) in four young professional flautists
(three males and one female aged from 29 to 34). Three were recruited at La Scala di Milano or amongst the finishing students of schools associated with La Scala. The fourth subject, the first author of this article, had obtained similar or higher level diplomas in Canadian institutions and had extensive performance experience. Consent sheets were signed by all subjects in accordance to ethics approval granted by the institutions involved. The same flute was used for all subjects (Sankyo, Etude model, with RS1 Cooper head joint). After performing quiet breathing and two vital capacity manoeuvres, the subjects were asked to play an excerpt from the second movement of the Poulenc Sonata (Fig. 1) which is a standard piece of the repertoire of professional flautists. It consists of long soft phrases in the three registers of the flute. Subjects were asked to play this piece with and without support. We did not give any definition of support to the flautists as we wanted them to play in as natural way as possible what they considered to be support and no support. The subjects were also asked to play without vibrato. As musicians train themselves to play with vibrato, it is not obvious to play without it. As a result, vibrato was seen on the sound tracings of both conditions, with and without support, though at different amplitudes. The length of the musical phrases varied between both conditions (from 47 to 56 s long) even though the performers played exactly the same musical passage. For the analysis, we separated each of the 53 notes (Fig. 1) from the recorded sound signals so we could compare similar events. That allowed us to compute average value of the recorded quantities within each note and each musical phrase. 2.2. Measurements and data processing For the first three subjects, recordings included measurements of chest wall volume (Vcw ) and the volumes of its compart-
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
ments, i.e. the lung and diaphragm-apposed parts of the rib cage (Vrcp and Vrca , respectively) and the abdomen (Vab ), by OptoElectronic Plethysmography (OEP, Elite System, BTS), surface electromyography of the respiratory muscles (scalene, rectus abdominus, parasternals, sternocleidomastoids, and lateral abdominal muscles, i.e., internal and external obliques and transversus combined), mouth pressure and sound recording during playing. Flow was measured by differentiation of the volume signals obtained by OEP during flute playing. The measurements in the fourth subject were acquired during two different experiments. The first experiment (referred to as 4a) included OEP measurements of chest wall volumes and flow, measurements of mouth pressure (Pm ) and sound recording. Electromyography of the respiratory muscles and sound recording were acquired during the second experiment (4b). OEP consisted of seven infrared video cameras (five in front and two behind the subjects both at 2 m) tracking 89 hemispherical 6 mm diameter reflective markers. These were applied to the surface of the chest wall, front and back in seven rows between the clavicles and the iliac crest (Aliverti et al., 2000). The 3D coordinates of the markers were converted into volumes using Gauss’ theorem as described in detail by Cala et al. (1996). Because of movement artefacts or of the required holding position of the flute, some of the 89 markers were occasionally hidden. When too many markers were missing for the reconstruction, the data was rejected. This only happened for the vital capacity manoeuvre of subject 3. The data from the OEP was stored on an IBM compatible PC at 50 Hz during flute playing and at 25 Hz during respiratory manoeuvres. Prior to flute playing, subjects breathed quietly and performed two vital capacity manoeuvres (VC) during which flow was measured with a pneumotachometer (HR 4700-A; Hans Rudolph Inc., Kansas City, MO) at the mouth connected to a pressure transducer (Celesco, ±5 cmH2 O) and recorded on the same computer using an analog-to-digital converter (RTI 800; Analog Devices; Norwood, MA). The following parameters were compared during support and without support playing: Vcw , average flow per phrase obtained by dividing Vcw by the time to complete the phrase and the relative contribution of the CW compartments. The functional residual capacity (FRC) was estimated by averaging the end expiratory volumes during quiet breathing. For each phrase, the curves of the chest wall, abdomen and total rib cage (Vrcp + Vrca ) were integrated over time and the area (in units of ml s) above and below FRC was calculated. The area above FRC was then expressed in percentage of the total area (above FRC + below FRC) and compared between support and without support. In addition, we calculated the number of the notes played at lung volumes lower than the FRC level in the two conditions and calculated their percentage in respect to the total number of notes played during the excerpt. We then used a χ2 test (Mantel–Haenszel correction) based on a 2 × 2 contingency table to evaluate the statistical significance of their difference. We refer to Vcw at the beginning of the musical phrases as the initial lung volume; at the end of the phrases, termination lung volume (Thomasson and Sundberg, 1997).
189
Mouth pressure (Pm ) was measured with a catheter placed at the corner of the mouth during flute playing. The catheter was connected to a pressure transducer (SCX01; Sensym, Milpitas, CA, ± 100 cmH2 O) and stored on a IBM compatible PC via the OEP System at 1000 Hz. Previous studies (Cossette et al., 2000; Bouhuys, 1964) used an air-filled balloon placed at the back of the throat. The less invasive method used in the present study was chosen in order to better represent natural performance. In subject 3, Pm measurements were discarded because of blockage of the catheter by saliva. In order to avoid blockage, a very small flow was put through the catheter for the subjects 1, 2 and 4 and the pressure resulting from the flow was deducted subsequently. For both conditions (support and without support), a band pass filter (4–10 Hz) was used in order to quantify the vibrato content and a low pass filter (2.5 Hz) was used to compare slower fluctuations of Pm . An average within each note and per phrase was computed. Surface electromyograms were recorded at 1000 Hz using a telemetric eight-channel system (TELEMG, BTS, Milan, Italy) and stored on the IBM computer via the OEP System. Movements, done in order to check muscle activity, consisted of inward abdominal displacement for the lateral abdominals; flexing the neck against a resistance on the forehead for the scalenes; taking a big rib cage inspiration for the parasternals; lifting the legs when sitting at the edge of a chair for rectus abdominus; and turning the head against a resistance for the sternocleidomastoids. As some muscle acquisitions contained large electrocardiogram (ECG) content, a cross-correlation technique was used for ECG removal before the signals were rectified and integrated. This technique consisted of four steps: (1) identification, with the help of a graphic interface, of the QRS complex of the ECG signal (template) during a period of relative EMG inactivity; (2) search of all the other ECG complexes across the signal by computation of a correlation coefficient between the template and a same-length part of the signal; (3) comparison and adjustment of the template amplitude according to the complexes found; (4) subtraction of the ECG signals. This method was adapted from the one suggested by Levine et al. (1986). The signals were then rectified and low pass filtered (2 Hz, Butter window). For each musical phrase per subject, the activation was expressed as percentage of the maximal activation during support playing. The length of each phrase, generally different between the two conditions, was normalised to 100%. Comparisons of the average activation between with and without support were performed for every 10% sections of the total length of each phrase. The data were then ensemble averaged over the different phrases within each subject and over all the phrases for all the subjects. In addition, for the most representative muscles, namely lateral abdominals and scalenes the EMG activation was plotted against Vcw relative to FRC, calculated over the same time periods and ensemble averaged individually and for the group. In subject 3, the parasternal EMG was clearly contaminated by expiratory intercostals adjacent to the parasternals as evidenced by the strong recruitment at the end of some phrases during without support. Because this contamination seriously
190
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
interfered with the ensemble averaged, we do not show the results for these muscles. During playing, sound was recorded at 44,100 Hz (Sony Tape recorder, TC-D5 no. 11631, Sennheiser Microphone (Phantom Power) MKH-416T no. 17551, Sennheiser Microphone Power supply MZA 16T no. 110030) and also stored on the IBM compatible PC at 1000 Hz as a synchronization reference. The microphone was placed at a fixed position for each subject around 1 m from the opening of the flute. Because they had a catheter measuring the mouth pressure and were filmed by cameras, the subjects who played standing kept a relatively fixed position. The sound was analysed by quantifying the frequency and intensity variation between support and without support. The frequency content of each note was computed by the fast Fourier transform (FFT) (number of samples: 35,790 ± 25,412; flute playing frequency content 349–1480 Hz; sampling rate 44,100 Hz). The sound was then rectified and a linear interpolation function was used to get the envelope of the sound. The mean of this function within each note gave us the averaged intensity for that note. The combined measurements of Vrcp , Vrca , Vab simultaneously with the EMG’s, mouth pressure and sound allowed us to see if there were any differences in muscle activation and action between support and non-support. 3. Results 3.1. Flows, chest wall volumes and thoraco-abdominal coordination The average flows per phrase shown in Fig. 2 were clearly smaller during playing with support in subjects 2 and 3, whereas
Fig. 2. Average flow per phrase with support (black dots) and without support (white dots) for each subject.
in subjects 1 and 4 they did not show consistent difference between the two conditions. Fig. 3 shows volume tracings of the chest wall (Vcw ). The slopes of these traces illustrate that flow tended to be higher and started more abruptly without support (NS) than with support (S) in subjects 2 and 3. The dashed horizontal line in the figures denotes functional residual capacity (FRC) and the open dots indicate the beginning and end of each of the 53 notes. Without support, all subjects consistently ended playing below FRC during almost every phrase. With support, the termination lung volume in subjects 1 and 3 were generally well above FRC, while subjects 2 and 4 did not play as far below FRC with support. Whether starting at higher lung volumes and using the same flow (subjects 1 and 4) or starting at the same
Fig. 3. Tracings of chest wall volumes for all subjects during playing with breath support (left) and without breath support (right). The horizontal dotted line represents the FRC volume and each open circle represents the beginning and end of each of the 53 notes of the excerpt.
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
191
Table 1 Number of phrases for which the area of the tracings over FRC is equal, smaller or larger with support than without
Chest wall Pulmonary rib cage Abdominal rib cage Rib cage total (rcp + rca) Abdomen
Area over FRC BS = NBS
Area over FRC BS > NBS
Area over FRC NBS > BS
2 1 6 1 5
15 18 13 18 8
3 1 1 1 7
Bold shows the condition with the most phrases, i.e. area of tracings larger over FRC with support than without support.
Fig. 4. Scatter plots, for each compartment (chest wall, ribcage and abdomen), of the volume integral above functional residual capacity (FRC), expressed as percentage of the total area (above FRC + below FRC), during with and without support. Each dot represents a single phrase and the solid line represents the identity line.
total area (above FRC + below FRC), for each phrase with and without support. Most of the phrases were played with a higher RC volume above FRC with support than without. This resulted in a similar behaviour of the total chest wall volume despite the scattered distribution of the abdominal compartment. Table 1 shows, the number of phrases in which the area above FRC with support was higher, smaller or equal than without support for all the compartments. Although Fig. 4 shows that the total RC volumes were higher during support, it is worth underlining that the main contribution to total RC behaviour was provided by the pulmonary RC compartment (Table 1). This strongly implicates the nondiaphragmatic inspiratory muscles in the provision of support, because these are the muscles which act directly to expand the pulmonary rib cage whereas the diaphragm acts on the abdomen and abdominal rib cage (Aliverti et al., 1997). How the rib cage and abdomen were coordinated during support and without is demonstrated by a plot of rib cage volume versus abdominal volume (Vab ) as shown in Fig. 5 (Konno and Mead, 1967). With change in rib cage volume the trace moves vertically and with abdominal volume change, horizontally. All subjects showed greater rib cage expansion during support for at least one of the three phrases shown. The inspiration before the note begins ends in a triangle indicating the beginning of the musical phrase. With support 10 of 12 traces are displaced upward showing increased rib cage volume with support. In one of the tracings without upward displacement (phrase 3 in subject 3), the phrase ended with both the rib cage and the abdomen more expanded with support. 3.2. Respiratory muscle activation
lung volumes but using a smaller flow (subjects 2 and 3), the termination lung volumes were consistently higher with support than without. As a result, all subjects played more notes under FRC during no support (9, 25, 18 and 11 in subjects 1, 2, 3 and 4, respectively) than during support (1, 16, 0 and 5 for subjects 1, 2, 3, and 4 respectively). The average percentage of notes played under FRC was then significantly different between support and without support (32% during NBS and 11% during BS with a χ2 of 12.99 and p-value of 0.0003). For the different chest wall compartments of all subjects, Fig. 4 compares, in scatter plots, the volume integral above functional residual capacity (FRC), expressed as percentage of the
The particular behaviour of the rib cage during support is also confirmed by looking at the respiratory muscle activation and coordination during playing. Fig. 6 shows ensemble averaged muscles activation with and without support for all phrases played by the subjects. The non-diaphragmatic inspiratory muscles (scalene and sternocleidomastoids) were clearly more activated during support. The stronger activation of the scalene was striking in subjects 1, 3 and 4, whereas subject 2 recruited the sternocleidomastoids more. The behaviour of the expiratory muscles (lateral abdominals and rectus) was interesting. The lateral abdominal activation
192
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
Fig. 5. Konno–Mead diagrams for all subjects. Rib cage (Vrcp + Vrca ) and abdominal (Vab ) displacements plotted one against the other during the first three musical phrases played with support (thick solid line) and without support (dashed line). Triangles and circles indicate, respectively, initial lung volume and termination lung volume.
Fig. 6. Ensemble average of muscular activation (% of maximal activation with support for lateral abdominals, scalene, sternocleidomastoid and rectus muscles) every 10% of total length of phrases for all the phrases in all the subjects during support (black dots) and without support (white dots).
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
193
Fig. 7. Ensemble average of lateral abdominals and scalene muscular activation (% of maximal activation with support) every 10% of total length of phrases in relation to CW volumes (relative to FRC) during support (black dots) and without support (white dots). Included results: all six phrases for subjects 1 and 3, phrases 1, 4, 5, 6, and 7 for subject 2. The dotted vertical line represents FRC.
during support was higher than without support in the first part of the phrase. Then, activation kept increasing but more without support than with. Toward the end of the phrase, the increase in activation was sometimes higher than 100% of maximal support activation. This pattern can be identified in the rectus as well, but to a smaller extent. Its activation during support seemed to be constant throughout the phrase but remained higher than without support in the first part of the phrase and smaller in the last one. Fig. 7 shows scalene and lateral abdominal muscle activation in relation to CW volume relative to FRC every 10% section of the total length of the phrase. Both volumes and EMG activation are the ensemble average of all the phrases in each subject and in all subjects. The pooled data show that the scalene activation decreases as volume decreases whereas the opposite is the case for the lateral abdominals. In addition, the figure confirms the higher activation of the scalene during support playing (regardless of the volumes at which the flautist plays) and the strong lateral abdominals recruitment at the end of the phrases without support, right after the FRC level is reached (vertical dotted line). Thus both volume
(rather than time only) and condition (support versus without support) are determinants of activation. 3.3. Mouth pressures and sound analysis Table 2 shows that the variations in mouth pressure during the 53 notes, expressed as standard deviations, seemed greater during support than without support, indicating more vibrati during support than without support. Average mouth pressures were greater in subjects 1 and 2 during support but less in subject 4 (Table 3). Sound analysis was carried out to assess whether possible changes in the measured variables might be related to changes in sound parameters. There was no consistent difference between support and without support common to all subjects. A similar analysis carried out on flow computed by differentiating change in Vcw (the major determinant of sound intensity, Cossette et al., 2000), showed that there were no noticeable differences in intensity between support and without support. 4. Discussion The co-authors’ previous studies (Cossette et al., 2000; Cossette, 2002) showed that different professional flautists use
Table 2 Mouth pressure standard deviations (cmH2 O) with and without Phrase
Subject 1
Subject 2
Subject 4a
BS (cmH2 O)
NBS (cmH2 O)
BS (cmH2 O)
NBS (cmH2 O)
BS (cmH2 O)
NBS (cmH2 O)
1 2 3 4 5 6
0.249 0.411 0.383 0.164 0.259 0.295
0.074 0.135 0.174 0.170 0.190 0.097
0.181 0.242 0.276 0.209 0.159 0.088
0.110 0.061 0.102 0.068 0.185 0.100
0.265 0.379 0.435
0.184 0.325 0.312
Mean
0.308
0.135
0.192
0.096
0.362
0.275
Mean Pm values of vibrato content. Bold and underlined respectively show higher vibrato content during breath support and without breath support.
194
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
Table 3 Mean mouth pressure averaged per notes and per phrase with and without support Phrase
Subject 1
Subject 2
Subject 4a
BS (cmH2 O)
NBS (cmH2 O)
BS (cmH2 O)
NBS (cmH2 O)
BS (cmH2 O)
NBS (cmH2 O)
1 2 3 4 5 6
7.258 8.88 8.014 7.872 9.654 7.627
6.649 8.77 7.686 5.997 6.388 7.022
5.972 13.743 8.085 7.65 9.577 10.968
6.885 9.036 7.039 7.018 9.143 9.680
2.493 3.626 3.138
2.86 4.745 4.186
Mean
8.132
7.231
10.058
8.347
3.087
3.939
Bold shows higher mean Pm when playing with support; underlined shows higher mean Pm when playing without support.
different muscle recruitment patterns while playing at similar frequencies and intensities. The diaphragm was generally non-activated but was recruited during specific manoeuvres like staccati and octaves. One subject showed some diaphragmatic activity during vibrati. According to that study and studies of others (Bouhuys, 1964, 1968; Bouhuys et al., 1966; Roos, 1936), flautists use some inspiratory muscles during playing. As playing a wind instrument or singing is an expiratory activity, it is important to underline the involvement of inspiratory muscles during expiration that might be used as antagonist muscles to the elastic recoil of the respiratory system or as antagonistic action to the expiratory muscles. Without the contraction of some inspiratory muscles during the expiration, the elastic recoil would produce higher pressures than required for a low-pressure instrument like the flute. Although these studies have shown that inspiratory muscles are activated during an expiratory task, to our knowledge no previous study has examined how muscular recruitment is coordinated with CW volumes in relation to FRC and how it relates to ‘breath support’ associated with high-level performance. Pettersen and Westgaard (2004) studied neck and shoulder muscles activation in professionally classically trained singers but they did not deal with concept of “breath support”. Griffin et al. (1995) found acoustical and laryngeal differences between singing with and without support. Their results did not support the concept that respiratory muscle activity is related to support during singing (p. 51). However, the only respiratory measurements acquired in their study were done with respiratory inductive plethysmography from which only kinematic information can be obtained. In order to assess muscular activity, dynamics and/or muscle activation must be measured. Thorpe et al. (2001) looked at different levels of singing sound projection and at the respiratory patterns associated with the different levels of projection. There was no consistent change between enhanced and normal projection. The relevance of this to support is not clear, but termination lung volumes were higher during enhanced projection. We found the same during flute playing with support. The goal of our work was to determine features common to each flautist studied while playing with support which were not present in the absence of support. In this way, we hoped to define flute breath support, how it is achieved and which respiratory muscles accomplish it. As expected, we observed considerable
variability between individual. Nevertheless there were clear, common features that should allow generalizations to be made. Thus the primary objectives of this research were achieved. Flute breath support entails antagonistic contraction of nondiaphragmatic inspiratory muscles that tend to hold the rib cage at a higher lung volume. This allows relaxation pressures to provide expiratory pressures over a longer period of the phrase being played. Inspiratory muscle activation may require some increase in the activation of expiratory muscles to counteract the antagonistic inspiratory action as in subjects 1, 2 and 4 (Fig. 6). Nevertheless, the lung volume at which playing a phrase ends, is usually above FRC and rarely far below it (Fig. 3 and Table 1). Thus the strong expiratory muscle pressures required to play the flute at low lung volumes are generally avoided, while advantage is taken of the high relaxation pressures at high lung volumes. In the absence of support, there was strong activation of the lateral abdominals in all subjects at the end of musical phrases when lung volume is lowest and expiratory muscles must develop their greatest pressures. With support this activation was considerably less (Figs. 6 and 7). The avoidance of strong expiratory muscle activation, is helped by the higher rib cage volumes (Fig. 4) which keep total chest wall volume at a higher overall level where relaxation pressures relieve the expiratory muscles of the task. This leads us to speculate that these muscles are being saved for other jobs; that they are used for fine tuning and specialized tasks such as vibrati (Table 2) and staccati. If the expiratory muscles were contracted in order to decrease lung volume and maintain the mouth pressures required to play the flute, the less they would be able to modulate the mouth pressures in finely controlled way. And fine control of mouth pressure is surely a sine qua non of high-quality wind instrument playing. We tentatively put forward the following definition: flute breath support is a mechanism to avoid the recruitment of expiratory muscles in order to decrease lung volume during playing so that they can best exert fine control over the mouth pressure modulations required for high quality playing without being encumbered by other tasks. This is achieved by inspiratory muscle recruitment as demonstrated by their greater electrical activation which keeps the rib cage expanded, lung volumes higher and the expiratory muscles relatively relaxed. This was demonstrated by their decreased activation during support playing at the end of the phrases.
I. Cossette et al. / Respiratory Physiology & Neurobiology 160 (2008) 187–195
Our results may apply to singing as well as flute playing as both are low-pressure expiratory activities. Thus singers may use support mechanisms similar to flautists to allow the fine modulation of subglottic pressures by expiratory muscles. Our tentative definition of support is, at this stage, still a hypothesis that can be further tested. Yet it is virtually axiomatic that a contracted muscle will not be able to modulate force to do another task as well as a relaxed muscle. If one were to put a weight on the back of a piano player’s hands, to keep the extensor digitorum muscle tonically contracted, the pianist would have great difficulty in playing properly. We conclude that there are features in common among flute players playing with support, that allow hypothesis formation and a tentative definition of what support is. If the hypothesis and definition are held up by further research, what is now confused will be considerably clarified. Hopefully music teachers will soon be in a position to teach support based upon sound physiological principles rather than intuition. Acknowledgements Supported by the Social Sciences and Humanities Research Council of Canada and the Canadian Institutes of Health Research. References Aliverti, A., Cala, S.J., Duranti, R., Ferrigno, G., Kenyon, C.M., Pedotti, A., Scano, G., Sliwinski, P., Macklem, P.T., Yan, S., 1997. Human respiratory muscle actions and control during exercise. J. Appl. Physiol. 83, 1256–1269. Aliverti, A., Dellac´a, R., Pelosi, P., Chiumello, D., Pedotti, A., Gattinoni, L., 2000. Optoelectronic plethysmography in intensive care patients. Am. J. Respir. Crit. Care Med. 161, 1546–1552.
195
Bouhuys, A., 1964. Lung volumes and breathing patterns in wind-instrument players. J. Appl. Physiol. 19 (5), 967–975. Bouhuys, A., 1968. Pressure-flow events during wind instrument playing. Ann. N. Y. Acad. Sci., 264–275. Bouhuys, A., Proctor, D.F., Mead, J., 1966. Kinetic aspects of singing. J. Appl. Physiol. 21 (2), 483–496. Brown, S.E., Thomas, M., 1990. Respiratory training effects in wind and brass instrumentalists. Med. Probl. Perform. Artists 5, 4. Cala, S.J., Kenyon, C.M., Ferrigno, G., Carnevali, P., Aliverti, A., Pedotti, A., Macklem, P.T., Rochester, D.F., 1996. Chest wall and lung volume estimation by optical reflectance motion analysis. J. Appl. Physiol. 81 (6), 2680– 2689. Cossette, I., Sliwinski, P., Macklem, P.T., 2000. Respiratory parameters during professional flute playing. Respir. Physiol. 121, 33–44. Cossette, I., 2002. M´ecanique respiratoire des flˆutistes professionnels. Rev. Mal. Respir. 19, 197–206. Griffin, B., Woo, P., Colton, R., Casper, J., Brewer, D., 1995. Physiological characteristics of the supported singing voice. A preliminary study. J. Voice 9 (1), 45–56. Konno, K., Mead, J., 1967. Measurement of the separate volume changes of rib cage and abdomen during breathing. J. Appl. Physiol. 22 (3), 407– 422. Leanderson, R., Sundberg, J., von Euler, C., 1987. Role of diaphragmatic activity during singing: study of transdiaphragmatic pressures. J. Appl. Physiol. 62, 259–270. Levine, S., Gillen, J., Weiser, P., Gillen, M., Kwatny, E., 1986. Description and validation of an ECG removal procedure for EMGdi power spectrum analysis. J. Appl. Physiol. 60, 1073–1081. Pettersen, V., Westgaard, R.H., 2004. The activity patterns of neck muscles in professional classical singing. J. Voice 19 (2), 238–251. Roos, J., 1936. The physiology of playing the flute. Archiv. N´eerlandaises Phon´etique Exp. 12, 1–26. Spillane, K.W., 1989. Breath support directives used by singing teachers: a Delphi study. NATS J. 45 (3), 9–57. Thomasson, M., Sundberg, J., 1997. Lung volume levels in professional classical singing. Logopedia Phoniatrica Vocologia 22, 61–70. Thorpe, C.W., Cala, S.J., Chapman, J., Davis, P.J., 2001. Patterns of breath support in projection of singing voice. J. Voice 15 (1), 86–104.