| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Contact author: Sally Collyer, who is now a private singing teacher, P.O. Box 156, Box Hill, Victoria 3128, Australia. E-mail: sallycollyer{at}yahoo.com.au.
 |
Abstract |
|---|
 Top Abstract References |
|---|
Method: Five trained female singers performed an 8-s messa di voce (a crescendo and decrescendo on one F0) across their musical F0 range. Lung volume (LV) change was estimated, and chest-wall kinematic behavior (dimensional change in ribcage [RC] and abdominal [AB] wall) was recorded using triaxial magnetometry.
Results: The direction of F0 influence on LV excursion (LVE) varied among singers, but SPL range appeared to be less important than duration to LVE. LVE was generally evenly divided between crescendo and decrescendo. Kinematic patterns differed markedly among singers, despite task consistency, and RC and AB paradoxing was widespread.
Conclusion: Each singer maintained her characteristic kinematic pattern regardless of F0 or SPL range, although these did influence aspects of RC and AB behavior. Given the essential role of breathing in classical singing, further work is needed to understand how singers develop their highly individual respiratory strategies and the principles by which each singer's breathing strategy can be optimized.
KEY WORDS: singing voice, normal respiration, females
Good breathing is essential to classical singing (Chapman, 2006; Garcia, 1841/1982; Hemsley, 1998; Miller, 1996; Vennard, 1967). Respiratory studies of singing have provided valuable information to singing pedagogy about lung volume (LV) change and the underlying chest-wall kinematic (rib cage [RC] and abdominal wall [AB]) patterns. However, the influence of fundamental frequency (F0) and sound pressure level (SPL) range on respiratory behaviors remains unclear. The purpose of this study was to investigate F0 and SPL range effects on respiratory behavior in classical singing.
LV at the initiation of the sung phrase (LVI) is generally well above end expiratory level (EEL), whereas LV at the termination of the phrase (LVT) is at or below EEL (Bouhuys, Mead, Proctor, & Stevens, 1968; Bouhuys, Proctor, & Mead, 1966; Foulds-Elliott, Thorpe, Cala, & Davis, 2000; Thomasson & Sundberg, 1997; Thorpe, Cala, Chapman, & Davis, 2001; Watson & Hixon, 1985; Watson, Hixon, Stathopoulos, & Sullivan, 1990). Musical phrases, however, do not allow the isolation of F0 and SPL influences. LV change in a repeated task over rising F0 does not appear to have been studied in detail. Titze et al. (1999) studied SPL effect on LV for classical singers performing the messa di voce (a crescendo then decrescendo on one note) at three F0s. Their results suggested that larger SPL range was associated with singers using more air (LV excursion [LVE]). Classically trained singers generally meet increased LVE demands by lowering LVT rather than by raising LVI (Bouhuys et al., 1966). SPL range is greatest in the mid-F0 range and narrower toward F0 extremes (Gramming, Sundberg, & Åkerlund, 1991). These findings suggest that singers would use more air (LVE) in the mid-F0 range by lowering LVT with little change in LVI. If LV patterns differed more between low and high F0, this would suggest a greater influence of F0.
Chest-wall kinematic studies of RC and AB patterns have not supported singing pedagogy's concept of one common, ideal breathing behavior (Miller, 1996; Vennard, 1967). Although kinematic patterns do show high intrasinger consistency in classical singers, they also show high intersinger variability. Changes in F0 and SPL during sung phrases do influence kinematic behavior (Foulds-Elliott et al., 2000; Thorpe et al., 2001; Watson & Hixon, 1985; Watson et al., 1990). Titze et al. (1999) noted the same high intrasinger consistency and intersinger variability in the messa di voce, but their use of only three F0s did not provide sufficient comparison to assess F0 influence. Thus, it is still uncertain how rising F0 might affect kinematic patterns on the same task.
Three elements contribute to the variability in kinematic patterns between singers. First, some singers show RC predominance, some AB predominance, and some a mix (Konno & Mead, 1967; Thomasson & Sundberg, 1997, 1999; Thorpe et al., 2001; Titze et al., 1999; Watson & Hixon, 1985; Watson, Hoit, Lansing, & Hixon, 1989). Second, some singers exhibit paradoxical behavior, with an increase in RC or AB during phonation (expiration; Thorpe et al., 2001; Titze et al., 1999). The third element is sequential patterning, where RC and AB predominance alternate through the phrase. Sequential patterning has been associated with longer phrase length and greater LVE (Winkworth, Davis, Adams, & Ellis, 1995), so it is not surprising that it has also been observed regularly in singing studies (Gould & Okamura, 1974; Thorpe et al., 2001; Titze et al., 1999; Watson & Hixon, 1985; Watson et al., 1989). The effect of F0 and SPL range on these elements is unclear.
This study examined the influence of F0 and SPL on respiratory patterns of female classical singers on a repeated task, the messa di voce. It was hypothesized (a) that LV measures would be uninfluenced by F0; (b) that increased SPL range would be associated with an increase in air used (LVE), achieved by lowering LVT without affecting LVI; (c) that kinematic patterns would be highly consistent for each singer but would show high intersinger variability in RC and AB predominance; and (d) that RC and AB range would increase with SPL range, underlying the increase in LVE. The symmetry of the crescendo and decrescendo sections in the messa di voce is highly prized (Miller, 1996; Vennard, 1967). Thus, we also hypothesized (e) that singers would use the same amount of air (LVE) in crescendo as in decrescendo and that this would show no influence of F0.
Method
Study Participants
Five female classical singers, ranging in age from 24 to 30 years, participated in this study. Three identified their voice type as lyric soprano, one as lyric coloratura soprano, and one as lyric mezzo. All held bachelor's degrees in singing, and two held postgraduate diplomas in singing performance. Table 1 sets out their years of singing training (ranging from 6 to 12), and professional experience, in terms of the taxonomy proposed by Bunch and Chapman (2000), ranged from 4.1b (Regional/Touring: minor principal) to 2.1 (International: opera principal). The study was approved by the Human Ethics Committee of the University of Sydney, and all participants gave informed consent. The entire protocol, including explanations of procedures, took approximately 1.5 to 2 hr per singer. All singers had warmed up their voices before arriving and were given as much time as they desired to familiarize themselves with the acoustic of the recording room.
|
The audio signal was recorded at 44.1 kHz via a pre-amplifier (Rane MS1 [Rane Corp., Mukilteo, WA] for Singers A through D; Behringer Ultragain Pro Model MIC2200 [Behringer International GmbH, Willich, Germany] for Singer E) to a DAT (Tascam Model DA P1 [Tascam, Montebello, CA]; Singers A through D) or a CD recorder (Marantz CDR 631 [Marantz America, Mahwah, NJ]; Singer E). The audio signal was monitored by an oscilloscope (Hewlett Packard 1741A [Hewlett Packard, Palo Alto, CA]) to confirm that there was no clipping of the audio signal. The DAT-recorded audio signal was transferred to an IBM Aptiva 833MHz computer through a LynXOne Mixer (LynX Studio Technology, Costa Mesa, CA) digital soundcard using CoolEdit 96 (Syntrillium Software Corp., Phoenix, AZ). Acoustic data were analyzed using MATLAB (The MathWorks, Natick, MA), CoolEdit 96, and Microsoft Excel. The signal was calibrated using a 1-kHz tone from a signal generator (Power Acoustik CP 500C; Power Acoustik, Montebello, CA) and a Rion NL 06 (Rion Co. Ltd., Tokyo, Japan) SPL meter set to slow damping with no weighting.
Respiratory recording used a 12-sensor triaxial magnetometry (trimag) system (Fastrak 3sF0002, Polhemus; Colchester, VT) at a sampling rate of 25 times per s. The sensors were attached to the singer with double-sided adhesive disks and single-sided tape in the configuration illustrated in Figure 1. Four sensors were attached by leads to each of three collector boxes. The corresponding three transmitter boxes were mounted on a plinth behind the singer. The plinth was mounted on a floorboard, on which the singer also stood. Because the trimag system is sensitive to field distortion from metal objects (Day, Dumas, & Murdoch, 1998; Nixon, McCallum, Fright, & Price, 1998), singers were asked to remove jewelry, and care was taken to remove metal objects, such as screwdrivers and spanners, from the vicinity.
|
LV Calibration Maneuvers and Singing Tasks
Taking into account the three-dimensional data provided by the trimag system, we based an extended set of calibration maneuvers for estimation of lung volume change on the work of Konno and Mead (1967); Hixon, Goldman, and Mead (1973); McCool (1995); McCool, Kelly, Loring, Greaves, and Mead (1986); and Watson et al. (1990). The maneuvers included quiet breathing, vital capacity (VC), isovolume, relaxation, sustained phonation while following a regimen of bending and stretching, and two messe di voce (on C4 [262 Hz] and C5 [523Hz]). The maneuvers were split into two groups, one before and one after the protocol tasks, and were not randomized.
Each singer sang two 8-s messa di voce tokens per F0 on /
/, beginning at A4 (440 Hz) and moving down by semitone (ST) to the lowest note at which both tokens were judged to be of public performance standard, then returning to A4 and moving up by ST to the highest satisfactory note. A high standard was set for acceptable phonational quality in order to ensure that the results would be truly representative of the classical singing demands of the messa di voce, not just of musical range. Reference pitches were provided on a Yamaha Clavinova CLP 811.
Data Analysis
Acoustic data. Analysis of the acoustic data is reported in Collyer, Davis, Thorpe, and Callaghan (2007). The audio signal was resampled at a rate of 25 times per s to align with the trimag data. For each token, maximum SPL was calculated as an average of the five samples (200 ms) centering around the point of highest SPL, whereas minimum SPLs at the start and at the end were averaged over the first and last five samples of each token's SPL trace, respectively. For each singer, maximum SPL for each F0 was then averaged over the two tokens (four tokens at A4), and minimum SPL was averaged over the start and end of the two tokens (i.e., four minima per F0 and eight minima at A4).
Respiratory data—trimag data correction. Respiratory data were analyzed in three steps: trimag data correction (field distortion correction, rotation, and translation), LV estimation, and chest-wall kinematic plotting. Day et al. (1998) noted that distortion of the electromagnetic field was the greatest source of error in trimag data collection, but field distortion remains constant if the transmitter position and environment remain unchanged (Day et al., 1998; Day, Murdoch, & Dumas, 2000). Thus, it is possible to calculate and compensate for these distortions by measuring a known grid and calibrating the magnetometer data according to the error calculations (Day et al., 2000). Readings were taken across a vertically mounted calibration board marked with a grid at 20-cm intervals (6 columns x 5 rows). The board was positioned at five distances from the transmitter boxes (36 mm, 136 mm, 336 mm, 536 mm, 736 mm). The external boundaries of these measurements were chosen to ensure they would encompass all sensor positions on all singers. The readings taken from the calibration board were compared with a reference grid of the actual distances, and a polynomial equation of best fit was established for correction. The distortion correction coefficients were applied to all trimag data. Next, to align by rotation the x, y, and z axes of the three transmitter boxes, we took axial calibration measurements of three common points (an arbitrary zero point, a point further along the anteroposterior [x] plane, and a point further along the lateral [y] plane) using one sensor from each box. Rotation correction coefficients were calculated from the variations and were applied to all data. Finally, because the four sensors attached to each box reported their positional coordinates relative to the point of origin (0, 0, 0) of their own transmitter box, the axial calibration measurements were used to translate mathematically the point of origin of two of the transmitter boxes to match the third.
Respiratory data—LV estimation. A more comprehensive modeling of RC movement and distortion may improve LV estimation from surface movement (Levine et al., 1991; Paek, Kelly, & McCool, 1990; Robertson, Bradley, & Homer, 1980). Eight triaxial distances between the 12 sensors were grouped as set out in Table 2. Triaxial distances were used to ensure that calculations were insensitive to rotation around the third axis (Levine et al., 1991). The error of each weighted sum, using different combinations of the five groupings, was calculated using a least-squares error fit to the spirometer signal. All calibration maneuvers except the isovolumes were included. The correlations for the major combinations are summarized in Table 3. Combining anteroposterior (AP), lateral, and craniocaudal distances raised the correlation for all singers except Singer E to greater than 0.92. The addition of rib cage anterolateral for Singer E made a significant improvement (from 0.838 to 0.905). However, the addition of the sensor over the epigastrium gave little improvement overall. From these results, we decided to use AP, lateral, craniocaudal, and RC anterolateral to calculate the weighted sum of distances for LV estimation, with correlations ranging from .905 to .971.
|
|
Chest-wall kinematic plotting. Kinematic plots RC and AB AP dimension were standardized by constructing a kinematic template for each singer, after Thorpe et al. (2001). The template contained the VC boundaries, the relaxation line, and the average RC:AB configuration at EEL. Two isovolume maneuvers were plotted, the average slope was calculated, and the template axes were adjusted so that this slope was –1 and the axes were rescaled as 100% RC and AB volume (see Figure 2). RC:AB configurations were identified at RV and TLC, and lines of the average isovolume slope were drawn through the configurations. A least-squares regression line was drawn through the relaxation tracing.
|
Results
Acoustic Data
The singers averaged an F0 range of 30.8 ST (SD = 3.8 ST), between E3 (165 Hz) and E6 (1319 Hz). Maximum SPL range on one F0 averaged 32.3 dB. SPL range across the singer's full F0 range averaged 49.6 dB.
Issues Arising From LV and Kinematic Analysis
LV profiles for each singer are set out in Figure 3. Before reporting the LV and kinematic results, certain anomalies merit comment. First, the LV profiles show that 2 singers registered LV less than RV (Singer C) and greater than TLC (Singer E) during the messa di voce. Examination of all kinematic traces found that Singer C's RC:AB configuration at the end of singing (LVT) was below the RV line in 27 of her 63 messe di voce, whereas Singer E's RC:AB configuration at the end of inhalation (LVI) was above the TLC line in 13 of her 76 messe di voce.
|
The second anomaly arose with regard to the relaxation line on the templates. Activity of the chest-wall musculature as inferred from departures from the relaxation line (Hixon et al., 1973) can be highly informative for singing. However, the performance of accurate relaxation maneuvers requires specific training and skill (Hodge & Rochet, 1989; Sharp, Goldberg, Druz, & Danon, 1975; Thorpe et al., 2001), and postural demands mean that a true relaxation of respiratory musculature is not possible in the standing posture (Hixon et al., 1973; Konno & Mead, 1967; Sharp et al., 1975; Thorpe et al., 2001). Our singers were not experienced in performing the relaxation maneuver, received only brief instruction and opportunity to practice, and performed while standing. The singers were asked to inhale to approximately 80% VC, then to relax against a slow-release setting on the spirometer. A line of best fit was extrapolated from the kinematic tracing after Thorpe et al. (2001), but inferences from the relaxation line must be cautious because of the limitations in its execution.
Because the focus of the present study was intraindividual respiratory behavior across F0 and SPL range, the issues of VC range and relaxation line raised here did not affect the results but did inform the discussion.
Lung Volumes
LV profiles for each singer, and averaged across all singers at common F0s, are set out in Figure 3. Average LV behavior was consistent with previous studies, with phrases initiated (LVI) at 70%–80% VC and terminated (LVT) at 30%–50% VC. Individually, LVI was generally well above EEL, with the exception of the lower F0s of Singer D. LVT at most F0s approached or went below EEL. Consistent exceptions were the higher F0s of Singers B and E. LVT was more variable than LVI, as observed by Foulds-Elliott et al. (2000), Thomasson and Sundberg (1997), and Thorpe et al. (2001). However, intrasinger variability of LVT and air used (LVE) was markedly lower than has been shown in other studies and reflected the consistency of the task. Even so, high intersinger variability across comparable F0 ranges showed that each singer used consistent but personalized LV patterns.
LV and F0. Correlations of LV measurements with rising F0 are set out in Table 4. Singers D and E showed moderate to strong tendencies for LVI, LV at the start of the decrescendo (LVD), and LVT to increase with rising F0. Singer B showed a mild tendency for LVT to rise. Singer A, by contrast, showed mild to moderate tendencies for LVD and LVT to fall with rising F0. Consequently, she showed a strong tendency to use more air (LVE) with rising F0, whereas Singers B, D, and E showed moderate to strong tendencies for LVE to decrease. Only Singer C showed no influence of F0 on any LV measure.
|
LVE during the crescendo and the decrescendo. To compare air used (LVE) during the crescendo and the decrescendo, we expressed the difference between LVI and LVD as a percentage of LVE (% LVE) for each token (see Table 5). Overall, singers used slightly less air during the crescendo (M = 46.1% LVE, SD = 2.8% LVE). Only Singer D used markedly less air during crescendo (M = 36.7% LVE, SD = 7.6% LVE). No strong correlation was found between LVE and rising F0 or between LVE and SPL range.
|
|
Sequential patterns and paradoxing. All singers used sequential patterning in at least some tokens. The most notable were Singers A and D, whose crescendi included substantial RC paradoxing, which generally ceased around the start of the decrescendo. At higher F0s, Singer A's RC paradoxing in the crescendo often followed an initial RC decline, forming a dip at the start of the messa di voce (see G5 in Figure 4). Singer B's clockwise sequential pattern was due to AB paradoxing in crescendo against a backdrop of RC decline. Decrescendi often included a simultaneous fall in AB and interruption to RC decline. As RC decline resumed, AB often stabilized or resumed paradoxing, with the trace resembling a question mark. Singer C's patterns generally showed a co-contribution of RC and AB, with some RC paradoxing in the crescendo at F5 and higher. Dips, with RC falling and then rising, appeared in some crescendi of Singer C but were not as consistent as for Singer A. Singer E's crescendi tended to be either two-stepped (first AB, then RC decline) or S-shaped (two-stepped with a further AB decline). At the very lowest F0s, very small AB change meant that the trace descended almost vertically. At the highest F0s, the RC decline was accompanied by AB paradoxing, increasing the curve of the S shape. Stepping was much less pronounced in the decrescendi, but most tokens at higher F0s closed with AB paradoxing. The relationship between Singer E's sequencing and the start of the decrescendo was inconsistent: sometimes there was a clear change in kinematic pattern (see G4 and G5 in Figure 4), and at other times there was not (see G3, C4, C5, and E6 in Figure 4).
F0 and SPL range influence. The ranges of RC change and of AB change were not calculated from RC and AB at initiation and termination of phonation. They were calculated as the maximum (RCmax or ABmax) minus the minimum (RCmin or ABmin) measurement during each token. This took into consideration the paradoxical behavior observed in other studies where, for example, the maximal RC dimension occurred during the phrase, not at the start. RCrange and ABrange reflected maximal and minimal dimensions at any point during the messa di voce and so were more indicative of the complexity of muscular coordination used by the singer. Maxima, minima, and ranges were correlated with rising F0, with SPL range, and with LVE (see Table 6).
|
.70; r 
–.70) of rising F0 on RC or AB behavior. Singer A showed a very strong tendency (.92) to increase RCrange by reducing RCmin, but there was no influence on AB behavior. Singers B and D showed a moderate tendency to raise both RCmax and RCmin, so that their RCrange showed no correlation with F0. Their ABrange showed a moderate F0 influence, with Singer B's ABrange tending to increase and Singer D's tending to decrease, due mostly to corresponding changes in ABmin. Singer E showed high correlations in both RCmax (.81) and ABmax (–.87). This related to changes in RC:AB configuration at the onset of phonation, where onset was preceded by a gradually larger isovolume movement increasing RC and reducing AB (compare her G5 and E6 tokens in Figure 4). The concomitant rise in Singer E's RCmin canceled out the effect of the rising RCmax, so that RCrange showed no F0 dependency. Changes in ABmin also reduced the effect of the fall in ABmax, so that ABrange showed only a mild tendency to fall with rising F0. There was no consistent influence of SPL range on RC or AB behavior, reflecting the lack of strong relationship between SPL range and LVE. Comparison of individual RC and AB behavior with SPL range showed only three notable correlations. Two of these were for Singer E, who had a moderate tendency to have smaller RC and larger AB dimensions at the start of messe di voce of larger SPL range. However, Singer E also showed a unique, inverse trend between SPL range and F0 (see Table 4), making it difficult to separate F0 and SPL range influences. Because her RCmax and ABmax correlated more highly with F0 than with SPL range, it would seem that the apparent relationship between SPL range and kinematic ranges was in fact due to F0 influence. The third notable correlation was a moderate tendency for Singer A to increase ABrange with SPL range, with the increase deriving more from a larger AB dimension at the start of the messa di voce than from a smaller AB dimension at the end.
Discussion
Despite performing the same task over similar F0 ranges, the singers in this study showed highly individualized LV and kinematic behavior. Intersinger variability reported in other studies was not reduced by standardization of the task. Individually distinct kinematic characteristics were present irrespective of F0 or SPL range in the messa di voce.
The first hypothesis was that LV measures would be uninfluenced by F0. The results were highly variable. When averaging LV results across all singers, we found that F0 did not affect the LV at which singers began the messa di voce (LVI), but as F0 rose, singers finished the messa di voce at a higher LV (LVT), thus using less air (LVE). Individually, however, LVI rose with F0 for 2 singers (D and E), LVT rose for 2 singers (D and E) and fell for 1 (A), and LVE fell for 3 singers (B, D, and E) and rose for 1 (A). Only Singer C showed no F0 influence.
The second hypothesis was that increased SPL range would be associated with an increase in air used (LVE), achieved by lowering LVT without affecting LVI. This followed the observation by Titze et al. (1999) that singers who encompassed smaller SPL ranges in the messa di voce showed greater symmetry in SPL change. They suggested that the singers might be using less of their VC, which they were then able to control more accurately. However, the singers in our study did not use greater LVE for messe di voce of greater SPL range. In fact, 2 singers (B and D) showed a mildly negative trend, indicating that they used less air with greater SPL range.
A key difference between our study and that of Titze et al. (1999) was in task duration. We conducted our procedure in such a way as to standardize duration of the messa di voce to 8 s. Titze et al. asked their singers to target a duration of 10 s but did not use any timing device. Increased LVE with longer breath groups has been observed in speech studies (Dromey & Ramig, 1998; Huber, 2007). These results, and those associated with the first hypothesis, suggest that F0 and duration, rather than SPL range, are significant influences on the amount of air used by singers in the messa di voce.
The third hypothesis was that each singer's kinematic pattern would be consistent across F0 range but that there would be high intersinger variability in RC and AB predominance. Despite performing an identical task across substantially the same F0 range, the singers' kinematic patterns were indeed distinctive, though individually consistent. Primary distinguishing characteristics were extensive RC paradoxing in crescendo for Singer A, AB paradoxing during the crescendo for Singer B, combined RC and AB decrease across the entire messa di voce for Singer C, RC decrease that commenced or markedly accelerated with the start of the decrescendo for Singer D, and alternating RC and AB decline in the crescendo for Singer E. These characteristics were consistent throughout the F0 range.
The fourth hypothesis was that RC and AB range would increase with SPL range, underlying the increase in air used (LVE). This was based on the second hypothesis—namely, a positive relationship between SPL range and LVE—which was not supported. It further assumed that demand for greater LVE would be met by increasing RC and AB range. The use of range (maximum minus minimum) took into consideration the paradoxical behavior observed in other studies of singers, whereas measuring excursion (start minus finish) would underestimate the scale of kinematic behavior. Given the lack of association between LVE and SPL range reported previously, it is not surprising that changes in RC and AB maximum, minimum, and range showed no consistent influence of SPL range.
However, the assumption of a consistent relationship between LVE and RC and AB range was also found to be unwarranted. The singers showed a mild (C), moderate (A, D), or strong (B, E) tendency for greater LVE to be achieved by smaller RC at the end (RCmin). However, this translated into a strong association between LVE and greater RCrange for only 2 singers (A, B). AB results were highly mixed. Overall, correlations of RC and AB measures showed stronger influence of F0 than of LVE. The exception was Singer C, who showed a moderate tendency to increase LVE by lowering ABmin (–.73), thus increasing ABrange (.67): This was the only evidence of any notable influence on Singer C's LV or kinematic behavior.
The final hypothesis was that singers would use the same amount of air (LVE) in crescendo as in decrescendo and that any difference would not be influenced by F0. Only Singer D showed a marked difference in LVE, averaging 36.7% LVE in crescendo. Differences in LVE during crescendo and decrescendo showed neither F0 nor SPL range influence.
F0, rather than SPL range, had a major influence on the breathing patterns of all but one of the singers, but its effect was highly variable. The dramatic differences in LV and kinematic strategies between singers, and the widespread use of RC and AB paradoxing, reflected surprising complexity for an ostensibly simple vocal task. The consistency with which each singer maintained her unique kinematic characteristics in the messa di voce across F0 range suggests that each has developed a basic personal strategy that she then adapted to meet particular vocal challenges more or less successfully. If this is so, an optimal personal strategy may represent a compromise tailored to the singer's primary vocal needs. This would challenge the pedagogical assumption that there is or ought to be one optimal kinematic pattern for all. Work is needed to assess to what degree singers' kinematic patterns are innate or can be altered by instruction and in what ways. Work is also needed to assess whether a change in kinematic pattern can render a perceptible improvement in the voice.
All 5 singers in this study differed substantially in their respiratory behavior, despite the fact that the task was both simple and standardized. A key direction for future research is to document the range of strategies used by singers and to analyze their strategic responses to highly targeted vocal challenges like the messa di voce in order to identify key variables and interacting elements. An important factor in this is the interrelationship between inspiratory and expiratory behavior. The results of this study show that characteristics can be reliably inferred from a sampling across the F0 range. Thorough descriptions of strategy will require a range of measurements, including pressure, muscular force and laryngeal behavior. However, certain observations and suggestions can be made from the results of this study. Singer A used steady AB contraction throughout the messa di voce. RC paradoxing occurred for most or all of the crescendo, when subglottal pressure demand would have been increasing and expiratory recoil force would have been decreasing. LV change on the kinematic plot during the crescendo (see Figure 4) was significantly less than that of the decrescendo, yet LV at the start of the decrescendo (LVD) was close to the LVE midpoint (see Figure 3). At lower F0s, RC dimension was similar at the start and end of the messa di voce, despite the difference in recoil forces at these differing LVs. One scenario is that RC expansion occurred as the diaphragm moved upward under AB pressure (Hixon & Hoit, 2005). At around the end of the crescendo, the diaphragm could ascend no further (noting that RCmax was uninfluenced by F0) and the singer began to contract the RC. Increase in air used (LVE) with rising F0 was achieved by terminating the messa di voce at a gradually smaller RC dimension (RCmin). Such a strategy would avoid the need to begin at a high LV (LVI) and combat high expiratory recoil force for the pp (very quiet) start but might be less effective if, for example, the task were a decrescendo followed by a crescendo.
Singer D also used RC paradoxing for all or most of the crescendo, but to a lesser RCmax, using a much larger ABmax and ABrange. She also maintained LVI below 80% VC, above which point Thomasson and Sundberg (1997) have noted that recoil pressure rises sharply. However, her kinematic trace was generally not as steady as Singer A's, with marked amounts of sequencing in the decrescendo. Singer D was unique in using much less air in crescendo (M = 36.7% LVE) than in decrescendo and in having an unusually high EEL. Particularly at lower F0s, kinematic sequencing might reflect a shortcoming in this strategy for this singer in finely controlling subglottal pressure reduction against inspiratory recoil forces as LV descended well below EEL. Analysis of the relationship between spectral balance (0–2 kHz:2–4 kHz) and SPL did find less linearity for this singer (Collyer et al., 2007).
Singer B's kinematic pattern was distinctive with the consistent presence of AB paradoxing. AB paradoxing during phonation has been reported in speech and singing (Hixon et al., 1973; Hodge & Rochet, 1989; Watson & Hixon, 1985; Winkworth et al., 1995). Although it has received much criticism in singing pedagogy (Miller, 1977), judgment on its appropriateness has tended to be withheld by voice researchers (Hixon & Hoffman, 1979; Titze, 1994). The consistency of her kinematic behavior suggested that it was part of Singer B's singing technique rather than unique to the messa di voce, but this is a case where testing other vocal challenges would be of particular interest. It is tempting to speculate that AB paradoxing might represent compensation for overemphasis of RC expansion in her singing training, attained at the expense of AB expansion during inspiration, so that AB paradoxing might represent a clash between pedagogical instruction and predisposition. This speculation is supported by the fact that Singer B's AB paradoxing rarely crossed the relaxation line, suggesting that she did not actively expand her AB wall. Limitations in deriving the relaxation line were discussed above, but reasonable relaxation maneuvers were obtained for Singer B. Also, she used markedly more air (LVE) in tokens below A4 (440 Hz) than above. Although this might have been an artifact of the study design, which divided the F0 range at A4, the contrast is so striking that it may suggest a deliberate change in singing technique for lower and higher ranges.
Singer E's kinematic strategy was distinctive due to the overwhelming predominance of RC movement, often devoid of AB movement, and the initiation of phonation near the TLC line. LVI was above 80% VC with only three exceptions (C#4, D4 and D#4), so that she appears to have adopted a LV strategy involving very high recoil forces at onset. Such forces would need to be counteracted by high inspiratory forces in order to commence the messa di voce at pp. This also implies a high level of glottal resistance, suggesting that her phonation at higher F0s may have moved toward a pressed phonation as described by Sundberg (1987), reducing flow rate and so using less air (LVE). Thus, Singer E appears to have adopted a markedly different strategy from the other singers for meeting the challenges of the messa di voce at higher F0s, involving high subglottal pressures and strong glottal resistance and utilizing inspiratory muscles acting against RC recoil forces more so than expiratory muscles. As F0 rose, maximum RC increased and maximum AB decreased, reflecting the same posturing movement noted by Thorpe et al. (2001) in connection with phrases beginning quietly. Her crescendi are notable for the steep decline in RC with little or no AB movement. This could suggest that she relied entirely upon RC recoil forces at this point, but it has been noted that surface dimensions do not reflect muscular activity, and AB during the crescendo lies to the left of the relaxation line, suggesting AB muscular contraction. It seems that Singer E may have used her AB in a posturing role and left LV change to her RC.
Last, Singer C used a generally steady co-contribution of RC and AB. Her LV and kinematic strategies showed no influence of F0 or SPL range. Greater LVE was achieved by reducing AB more at the end of tokens, in contrast to the other singers, who reduced RCmin. However, interpretation of Singer C's respiratory strategy is complicated by low correlation between anteroposterior dimension and LV estimation. Lateral expansion of the RC has been associated with greater vocal projection in singers (Thorpe et al., 2001). Measuring RC distortion would appear to be a key component in understanding strategies like those of Singer C.
The results of this study indicate that the factors influencing singers' respiratory strategies are highly complex and not well understood. Certainly, the pedagogical concept of one general optimal respiratory strategy appears to be unsustainable. Furthermore, it seems that an individual's optimal strategy may in fact represent a compromise of options, perhaps depending on repertoire or voice type. Much work is needed to understand what interrelationship of nature and nurture determines a singer's respiratory pattern and what constitutes optimization for the individual. Because good breathing is fundamental to good singing, it is of major importance to singing pedagogy to understand what "good breathing" actually means.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Sections of these findings were presented at the Sixth Voice Symposium of Australia, held in October 2002 in Glenelg, South Australia.
Received December 29, 2006
Accepted September 5, 2007
|
References |
|---|
|
TopAbstractReferences |
|---|
Benchetrit, G. (2000). Breathing patterns in humans: Diversity and individuality. Respiration Physiology, 122, 123–129.[CrossRef][Medline]
Boiten, F. A., Frijda, N. H., & Wientjes, C. J. E. (1994). Emotions and respiratory patterns: Review and critical analysis. International Journal of Psychophysiology, 17, 103–128.[CrossRef][Medline]
Bouhuys, A., Mead, J., Proctor, D. F., & Stevens, K. N. (1968). Pressure-flow events during singing. Annals of the New York Academy of Sciences, 155, 165–176.[CrossRef]
Bouhuys, A., Proctor, D. F., & Mead, J. (1966). Kinetic aspects of singing. Journal of Applied Physiology, 21, 483–496.
Bunch, M., & Chapman, J. (2000). Taxonomy of singers used as subjects in scientific research. Journal of Voice, 14, 363–369.[CrossRef][Medline]
Chapman, J. L. (2006). Singing and teaching singing: A holistic approach to classical voice. San Diego, CA: Plural Publishing.
Collyer, S., Davis, P. J., Thorpe, C. W., & Callaghan, J. (2007). Sound pressure level and spectral balance linearity and symmetry in the messa di voce of female classical singers. The Journal of the Acoustical Society of America, 121, 1728–1736.[CrossRef][Medline]
Day, J. S., Dumas, G. A., & Murdoch, D. J. (1998). Technical note: Evaluation of a long-range transmitter for use with a magnetic tracking device in motion analysis. Journal of Biomechanics, 31, 957–961.[CrossRef][Medline]
Day, J. S., Murdoch, D. J., & Dumas, G. A. (2000). Technical note: Calibration of position and angular data from a magnetic tracking device. Journal of Biomechanics, 33, 1039–1045.[CrossRef][Medline]
Douglas, N. J., White, D. P., Weil, J. V., & Zwillich, C. W. (1983). Effect of breathing route on ventilation and ventilatory drive. Respiration Physiology, 51, 209–218.[CrossRef][Medline]
Dromey, C., & Ramig, L. O. (1998). Intentional changes in sound pressure level and rate: Their impact on measures of respiration, phonation, and articulation. Journal of Speech, Language, and Hearing Research, 41, 1003–1018.
Foulds-Elliott, S. D., Thorpe, C. W., Cala, S. J., & Davis, P. J. (2000). Respiratory function in operatic singing: Effects of emotional connection. Logopedics Phoniatrics Vocology, 25, 151–168.[CrossRef]
Garcia, M., II (1982). A complete treatise on the art of singing, in two parts (D. V. Paschke, Trans. & Ed.). New York: Da Capo. (Original work published 1841).
Gilbert, R., Auchincloss, J. H., Jr., Brodsky, J., & Boden, W. (1972). Changes in tidal volume, frequency, and ventilation induced by their measurement. Journal of Applied Physiology, 33, 252–254.
Gould, W. J., & Okamura, H. (1974). Respiratory training of the singer. Folia Phoniatrica, 26, 275–286.[Medline]
Gramming, P., Sundberg, J., & Åkerlund, L. (1991). Variability of phonetograms. Folia Phoniatrica, 43, 79–92.[Medline]
Gribbin, H. R. (1983). Using body surface movements to study breathing. Journal of Medical Engineering & Technology, 7, 217–223.[CrossRef][Medline]
Grimby, G., Bunn, J., & Mead, J. (1968). Relative contribution of rib cage and abdomen to ventilation during exercise. Journal of Applied Physiology, 24, 159–166.
Han, J. N., Stegen, K., Cauberghs, M., & Van de Woestijne, K. P. (1997). Influence of awareness of the recording of breathing on respiratory pattern in healthy humans. European Respiratory Journal, 10, 161–166.[Abstract]
Hemsley, T. (1998). Singing and imagination: A human approach to a great musical tradition. Oxford, United Kingdom: Oxford University Press.
Hirsch, J. A., & Bishop, B. (1982). Human breathing patterns on mouthpiece or face mask during air, CO2, or low O2. Journal of Applied Physiology, 53, 1281–1290.
Hixon, T. J., Goldman, M. D., & Mead, J. (1973). Kinematics of the chest wall during speech production: Volume displacements of the rib cage, abdomen, and lung. Journal of Speech and Hearing Research, 16, 78–115.[Medline]
Hixon, T. J., & Hoffman, C. (1979). Chest wall shape in singing. In V. Lawrence (Ed.), Transcripts of the Seventh Symposium Care of the Professional Voice held June 1978 at the Juilliard School, New York City: Part 1: The scientific papers (pp. 9–10). New York: The Voice Foundation.
Hixon, T. J., & Hoit, J. D. (2005). Evaluation and management of speech breathing disorders: Principles and methods. Tucson, AZ: Redington Brown.
Hodge, M. M., & Rochet, A. P. (1989). Characteristics of speech breathing in young women. Journal of Speech and Hearing Research, 32, 466–480.[Medline]
Huber, J. (2007). Effect of cues to increase sound pressure level on respiratory kinematic patterns during connected speech. Journal of Speech, Language, and Hearing Research, 50, 621–634.
Iwarsson, J., & Sundberg, J. (1998). Breathing behaviors during speech in healthy females and patients with vocal fold nodules. Speech Transmission Laboratory—Quarterly Progress and Status Report, 4, 23–38.
Konno, K., & Mead, J. (1967). Measurement of the separate volume changes of rib cage and abdomen during breathing. Journal of Applied Physiology, 22, 407–422.
Konno, K., & Mead, J. (1968). Static volume-pressure characteristics of the rib cage and abdomen. Journal of Applied Physiology, 24, 544–548.
Levine, S., Silage, D., Henson, D., Wang, J., Kreig, J., LaManca, J., & Levy, S. (1991). Use of a triaxial magnetometer for respiratory measurements. Journal of Applied Physiology, 70, 2311–2321.
Mador, M. J., & Tobin, M. J. (1991). Effect of alterations in mental activity on the breathing pattern in healthy subjects. American Review of Respiratory Disease, 144, 481–487.[Medline]
Masaoka, Y., & Homma, I. (2001). The effect of anticipatory anxiety on breathing and metabolism in humans. Respiration Physiology, 128, 171–177.[CrossRef][Medline]
McCool, F. D. (1995). Noninvasive methods for measuring ventilation. In C. Roussos (Ed.), The thorax: Part B. Applied physiology (2nd ed., pp. 1049–1069). New York: Marcel Dekker.
McCool, F. D., Kelly, K. B., Loring, S. H., Greaves, I. A., & Mead, J. (1986). Estimates of ventilation from body surface measurements in unrestrained subjects. Journal of Applied Physiology, 61, 1114–1119.
Miller, R. (1977). English, French, German and Italian techniques of singing: A study in national tonal preferences and how they relate to functional efficiency. Metuchen, NJ: Scarecrow Press.
Miller, R. (1996). The structure of singing: System and art in vocal technique. New York: Schirmer.
Nixon, M. A., McCallum, B. C., Fright, W. R., & Price, N. B. (1998). The effects of metals and interfering fields on electromagnetic trackers. Presence, 7, 204–218.[CrossRef]
Paek, D., Kelly, K. B., & McCool, F. D. (1990). Postural effects of measurements of tidal volume from body surface displacements. Journal of Applied Physiology, 68, 2482–2487.
Paek, D., & McCool, F. D. (1992). Breathing patterns during varied activities. Journal of Applied Physiology, 73, 887–893.
Perez, W., & Tobin, M. J. (1985). Separation of factors responsible for change in breathing pattern induced by instrumentation. Journal of Applied Physiology, 59, 1515–1520.
Ritz, T., Dahme, B., Dubois, A. B., Folgering, H., Fritz, G. K., Harver, A., et al. (2002). Guidelines for mechanical lung function measurements in psychophysiology. Psychophysiology, 39, 546–567.[CrossRef][Medline]
Robertson, C. H., Jr., Bradley, M. E., & Homer, L. D. (1980). Comparison of two- and four-magnetometer methods of measuring ventilation. Journal of Applied Physiology: Respiration, Environment, and Exercise Physiology, 49, 355–362.
Sackner, J. D., Broudy, M. J., Davis, B., Cohn, M. A., & Sackner, M. A. (1980). Ventilation at rest and during exercise measured without physical connection to the airway. In F. D. Stott, E. B. Raftery, & L. Goulding (Eds.), ISAM 1979: Proceedings of the Third International Symposium on Ambulatory Monitoring (pp. 341–351). London: Academic Press.
Sharp, J. T., Goldberg, N. B., Druz, W. S., & Danon, J. (1975). Relative contributions of rib cage and abdomen to breathing in normal subjects. Journal of Applied Physiology, 39, 608–618.
Shea, S. A. (1996). Behavioral and arousal-related influences on breathing in humans. Experimental Physiology, 81, 1–26.[Abstract]
Shea, S. A., Walter, J., Pelley, C., Murphy, K., & Guz, A. (1987). The effect of visual and auditory stimuli upon resting ventilation in man. Respiration Physiology, 68, 345–357.[CrossRef][Medline]
Strömberg, N. O. T., Dahlbäck, G. O., & Gustafsson, P. M. (1993). Evaluation of various models for respiratory inductance plethysmography calibration. Journal of Applied Physiology, 74, 1206–1211.
Sundberg, J. (1987). The science of the singing voice. Dekalb: Northern Illinois University Press.
Thomasson, M., & Sundberg, J. (1997). Lung volume levels in professional classical singing. Logopedics Phoniatrics Vocology, 22, 61–70.
Thomasson, M., & Sundberg, J. (1999). Consistency of phonatory breathing patterns in professional operatic singers. Journal of Voice, 13, 529–541.[CrossRef][Medline]
Thorpe, C. W., Cala, S. J., Chapman, J., & Davis, P. J. (2001). Patterns of breath support in projection of the singing voice. Journal of Voice, 15, 86–104.[CrossRef][Medline]
Titze, I. R. (1994). Principles of voice production. Englewood Cliffs, NJ: Prentice Hall.
Titze, I. R., Long, R., Shirley, G. I., Stathopoulos, E., Ramig, L. O., Carroll, L. M., & Riley, W. D. (1999). Messa di voce: An investigation of the symmetry of crescendo and decrescendo in a singing exercise. The Journal of the Acoustical Society of America, 105, 2933–2940.[CrossRef][Medline]
Vennard, W. (1967). Singing: The mechanism and the technic. New York: Carl Fisher.
Watson, P. J., & Hixon, T. J. (1985). Respiratory kinematics in classical (opera) singers. Journal of Speech and Hearing Research, 28, 104–122.[Medline]
Watson, P. J., Hixon, T. J., Stathopoulos, E. T., & Sullivan, D. R. (1990). Respiratory kinematics in female classical singers. Journal of Voice, 4, 120–128.[CrossRef]
Watson, P. J., Hoit, J. D., Lansing, R. W., & Hixon, T. J. (1989). Abdominal muscle activity during classical singing. Journal of Voice, 3, 24–31.[CrossRef]
Winkworth, A. L., Davis, P. J., Adams, R. D., & Ellis, E. (1995). Breathing patterns during spontaneous speech. Journal of Speech and Hearing Research, 38, 124–144.[Medline]
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| All ASHA Journals | AJA | AJSLP | JSLHR | LSHSS |
= start of crescendo; 
= start of decrescendo; 
= end of decrescendo.