Do Vocal Warm-Up Exercises Alleviate Vocal Fatigue? Rochelle L. Milbrath* Nancy Pearl Solomon** University of Minnesota Minneapolis

Vocal warm-up (WU) exercises of varying types and durations have been suggested as a way of improving vocal function. However, limited research has been conducted to assess the effects of vocal WU exercises on normal or disordered voices. This study attempted to manipulate vocal function, assessed by phonation threshold pressure (PTP) and self-perceived phonatory effort (PPE) at 3 pitches, in 8 young women who reported symptoms of chronic vocal fatigue. Predictions were that PTP and PPE would decrease after 20 min of vocal WU exercises, increase after 1 hr of loud reading, and decrease after 30 min of vocal silence. Furthermore, greater increases in PTP and PPE were expected when the loud-reading task was preceded by a placebo condition of vocal rest than by vocal WU exercises. Results failed to reveal statistically significant changes in PTP or PPE after any of the experimental tasks. High between-subject variability contributed to this result. Removal of 1 outlier from the sample resulted in a statistically significant difference for PTP across tasks, although post hoc pairwise comparisons failed to detect specific effects. Informal inspection of the data indicated that the most obvious difference was an increase in PTP after the loudreading task at the highest pitch. KEY WORDS: vocal fatigue, warm-up exercises, phonation threshold pressure, effort

M

any individuals report experiencing symptoms of vocal fatigue after extended voice use. Symptoms of vocal fatigue generally include complaints of increased effort to talk and to sustain speech, and a tired, weak voice (Kitch & Oates, 1994; Smith, Kirchner, Taylor, Hoffman, & Lemke, 1998). Clinical reports indicate that vocal fatigue is related to overuse or misuse of the voice, using poor vocal techniques, and being emotionally or physically stressed. Prolonged talking using a less than optimal style can lead to vocal fatigue, and this misuse can involve extreme laryngeal muscle tension in the absence of pathology (Hirano, Koike, & Joyner, 1969; Jackson, 1940; Koufman & Isaacson, 1991). Women tend to report vocal fatigue more often than men (Russell, Oates, & Greenwood, 1998; Smith et al., 1998). Furthermore, those in the occupations of teaching, singing, or acting seem most susceptible (Kitch & Oates, 1994; Kostyk & Rochet, 1998; Novak, Dlouha, Capkova, & Vohradnik, 1991; Russell et al., 1998; Smith et al., 1998), but vocal fatigue can affect anyone misusing or overusing his or her voice. Because vocal fatigue is a problem for so many individuals, its prevention is of clinical importance.

* Currently affiliated with the Minneapolis Public Schools, Minneapolis, MN ** Currently affiliated with Walter Reed Army Medical Center, Washington, DC

422

Journal Speech, Language, and Hearing Research• •Vol. Vol. 422–436• •April April 2003 • ©American Speech-Language-Hearing Association Journal of of Speech, Language, and Hearing Research 4646• •422–436 2003 1092-4388/03/4602-0422

The aspects of laryngeal neurophysiology and biomechanics that change with prolonged voice use are little understood. Intrinsic laryngeal muscles of both human (Claassen & Werner, 1992) and canine (Cooper & Rice, 1990) specimens have been identified as predominantly nonfatiguable and fatigue resistant. These findings, derived from standard histochemical fiber-typing procedures, suggest that muscle fatigue is an unlikely mechanism for vocal fatigue, but they require validation from in vivo physiologic investigations of muscle-fiber properties. Changes in biomechanic properties of laryngeal tissue are thought to play a more prominent role in changes in vocal function with prolonged voice use. The molecular structure of the vocal folds seems specially designed to withstand frequent and prolonged vibration (Titze, 1983). Consequences of continual vibration include increased tissue stiffness (Haji, Mori, Omori, & Isshiki, 1992) and heat dissipation (Cooper & Titze, 1985). Increased vocal fold viscosity is a logical outcome of both of these processes. Vocal endurance appears to depend on the preservation of optimal levels of laryngeal tissue elasticity and viscosity. Clinical management of vocal fatigue is based, in part, on the belief that there are preventive or therapeutic methods for reducing its occurrence. Studies have indicated that formal vocal training is beneficial to vocal endurance. That is, trained vocalists appear to be less susceptible to vocal fatigue than persons who have not had vocal training (Gelfer, Andrews, & Schmidt, 1991, 1996; Scherer et al., 1991). Several different vocal exercise routines have been created, primarily based on general principles of fatigue prevention derived from the sports medicine literature. For example, professional voice users often warm up their voices before a performance to combat or delay vocal fatigue or perhaps even injury. Athletes commonly warm up to prepare muscles for prolonged, strenuous activity. Warm-up (WU) exercises and stretching before a sport are considered essential to prepare the athlete’s muscles, mind, and lungs for the activity and to avoid muscle soreness and injury during and after exertion. Physiologically, WU exercises are thought to increase blood circulation and respiration, and literally “warm up” the muscles, so muscle viscosity drops (Safran, Seaber, & Garrett, 1989), allowing smoother muscle contractions and increased muscle elasticity. Following these principles, leading the intrinsic laryngeal muscles through a series of vocal WU and “stretching” exercises should help prepare the vocal folds for vocally demanding tasks. This, in turn, could prevent or delay symptoms of vocal fatigue. Elliot, Sundberg, and Gramming (1995) reported that 10 amateur singers felt that their voices were in better condition after a 30-min session of vocal WU exercises. Exercises involved

singing a descending melodic pattern on /mu…/ as softly as possible and incorporated various pitches, loudnesses, and vowels. Many vocal WU protocols include the use of nasal consonants and nasalized vowels. The basis for this is that improving overall resonance is thought to improve voice quality and provide maximum vocal economy (Verdolini, Druker, Palmer, & Samawi, 1998). Theoretically, the result is an optimal relationship between vocal output and vocal fold impact stress. Specifically, resonatory training exercises are intended to facilitate an optimal glottal width (barely abducted), which should result in improved phonatory ease (Lucero, 1998; Titze, 1988). The specific components and techniques involved in vocal-endurance training have been addressed in recent literature. A key example is the series of vocal-function exercises (VFEs) developed by Stemple, Glaze, and Klaben (2000) to train and condition the voice. These exercises are intended to balance the phonatory, respiratory, and resonatory subsystems. Exercises include maximally sustained vowels on a variety of pitches and pitch glides throughout the pitch range, all performed quietly. VFEs are recommended several times per day for several weeks. Vocal improvement after 4–6 weeks of VFEs has been demonstrated in vocally normal young adults (Sabol, Lee, & Stemple, 1995; Stemple, Lee, D’Amico, & Pickup, 1994) and in teachers with a history of voice problems (Roy et al., 2001). Using another regimen, Blaylock (1999) custom designed a 4.5-month program of vocal exercises that were performed daily and monitored weekly during studio sessions. He reported improvements in voice quality in 4 trained and untrained singers who had various voice disorders. This research indicates that vocal training programs can have therapeutic value. A variety of measures have been used in the attempt to assess changes in vocal function in untrained singers. Studies that have used acoustic measures, such as fundamental frequency or jitter, have reported inconsistent and equivocal changes after vocal loading (Buekers, 1998; Gelfer et al., 1991; Neils & Yairi, 1987; Scherer et al., 1991; Stemple et al., 1994). Less commonly, auditory perceptual ratings have been used to assess vocal fatigue. Neils and Yairi (1987) had listeners rate vocal normalcy before and after 6 vocally untrained women spoke over background noise for 45 min, and they failed to find systematic changes. Thus, previous research on vocal fatigue does not support the use of certain acoustic or auditory–perceptual measures. Because vocal fatigue is defined by the vocalist (i.e., it is self-perceived) and the laryngeal properties presumably most affected are biomechanical, as discussed previously, two particular measures have gained relative

Milbrath & Solomon: Vocal Warm-Up and Fatigue

423

acceptance in the scientific community. These are phonation threshold pressure (PTP) and self-ratings of perceived phonatory effort (PPE). PTP has gained popularity in recent studies that address vocal function, including several concerned specifically with vocal WU and/or vocal fatigue (Chang & Karnell, 2001; Elliot et al., 1995; Motel & Fisher, 2001; Solomon & DiMattia, 2000; Solomon, Glaze, Arnold, & Van Mersbergen, 2003). PTP is the minimum lung pressure required to initiate phonation (Titze, 1988, 1992). It can be obtained noninvasively by measuring oral pressure during the closed portion of the phoneme /p/ in slow sequential repetitions of syllables produced as softly as possible, but not in a whisper. According to the theoretical formulation by Titze (1988), PTP is indirectly proportional to the thickness of the vocal folds and directly proportional to tissue viscosity, the velocity of the mucosal wave during phonation, and the prephonatory glottal width. A prolonged, strenuous vocal activity is expected to result in increased friction of the vocal folds, increased heat dissipation (Cooper & Titze, 1985), a reduction in muscle elasticity, and a likely increase in muscle viscosity. This succession of events would lead to an increase in PTP. Vocal WU exercises are expected to increase blood flow to the vocal folds and decrease muscle viscosity and perhaps nonmuscular tissue viscosity, thus decreasing PTP. If glottal width decreases as a result of vocal WU exercises, PTP should decrease as well. PPE is thought to be closely associated with PTP because both measures apparently reflect ease of phonation (Titze, 1988; Verdolini, Titze, & Fennell, 1994). However, this association has not been substantiated for persons with normal and disordered voices (Fisher, Ligon, Sobecks, & Roxe, 2001; Verdolini et al., 1994; Verdolini-Marston, Sandage, & Titze, 1994). Studies of vocal fatigue have included PPE as a dependent measure, but because methods differ as to when it was rated, the results are hard to interpret. Speakers have rated PPE during an oral reading task (Solomon & DiMattia, 2000), immediately after prolonged loud reading (Chang & Karnell, 2001; Stemple, Stanley, & Lee, 1995), and during a PTP task (Solomon et al., 2003). A benefit of rating PPE during the PTP task at each of several pitches is that correlations between the two variables can be examined validly. This is especially true because PTP at high pitches tends to be particularly sensitive to changes associated with vocal fatigue and hydration levels (Solomon & DiMattia, 2000; Verdolini et al., 1994). Studies that have examined the effect of vocal exercises on vocal function have differed substantially in the duration of the activity. Some involved a single session of vocal WU exercises alone and others had participants perform exercises for weeks to months. Exercises that

424

are intended specifically to warm-up the larynx should produce immediate benefits. A 20–30 min session of WU exercises has been shown to improve general muscle function (Kulund & Töttössy, 1983). Likewise, a single session of vocal WU exercises ranging from 10 to 30 min improves most singers’ self-perception of vocal function (Elliot et al., 1995; Welham & Maclagan, 2000). Changes in PTP after brief WU have produced less consistent results, including an increase, rather than a decrease, in PTP at a high pitch (Motel & Fisher, 2001) and considerable variability among participants (Elliot et al., 1995). Because loud talking is thought to be one cause of vocal fatigue, several researchers have attempted to load the voice with prolonged loud talking. The necessary task duration is not known and has varied from 15–20 min (Linville, 1995; Stone & Sharf, 1973) to 45–60 min (Gelfer et al., 1991; Gelfer, Andrews, & Schmidt, 1996; Neils & Yairi, 1987; Scherer et al., 1991) to 2 or more hours (Scherer et al., 1991; Solomon & DiMattia, 2000; Solomon et al., 2003; Stemple et al., 1995). Each of these studies reported some change in laryngeal function after the experimental task, but findings from studies using shorter tasks were more subtle and inconsistent than those using longer tasks. It is important to note that the studies cited here included participants with normal vocal function and no vocal complaints. Buekers (1998) studied vocally healthy participants and those who complained of vocal fatigue and found that both groups’ self-ratings of fatigue, pain, and discomfort in the throat increased significantly after a 30-min series of vocal tasks. More research that includes people who typically experience vocal fatigue is needed. It is logical to assume that these individuals would be most susceptible to a vocally fatiguing task and perhaps more receptive to preventive and therapeutic approaches. Additionally, the findings from this clinical population will have more direct implications for clinical practice. On the basis of the studies reviewed, we speculated that performing a loud talking task for 1 hr should produce changes in vocal function in vocally untrained persons. Furthermore, we expected acceleration of this process in persons who experience chronic vocal fatigue. Scant literature exists on recovery after vocal loading. Advice from the sports medicine literature includes the importance of muscle cool-down exercises after physical activity. Saxon and Schneider (1995) stated that “lowintensity activity during the cool-down phase significantly shortens the time to complete recovery” (p. 70). Light exercise after a fatiguing task can decrease the time for lactic acid removal from muscles from more than 1 hr to 15–20 minutes (Bonen & Belcastro, 1976; Fox, Robinson, & Wiegman, 1969). It should be noted that

Journal of Speech, Language, and Hearing Research • Vol. 46 • 422–436 • April 2003

the phonation tasks required for postfatigue assessments could serve as cool-down activities. Although the literature has not addressed recovery in laryngeal muscles directly, studies that loaded the voice with a 2-hr loudtalking task reported reductions in PTP after 15–60 min of rest and a return to baseline PPE levels in 15 min to 24 hr (Chang & Karnell, 2001; Solomon & DiMattia, 2000; Stemple et al., 1995). Neils and Yairi (1987) attempted to assess vocal recovery after a 45-min loudreading task, but failed to find changes in voice F0, voice normalcy, or average air flow during vowels, and therefore could not demonstrate recovery. In summary, studies exist that have examined the effects of specific WU exercises on vocally normal and disordered populations, and others have focused on the effects of a fatiguing task on vocal function. This study represents the first known attempt to assess the effects of vocal WU exercises and a subsequent vocal-loading task on any population. The purposes of this study were to examine changes in PTP and PPE after tasks designed to warm up, fatigue, and rest the voice in young women who present with vocal fatigue. A placebo condition, involving vocal rest, was used in place of the WU condition for each participant on a separate day. With this design, it was possible to examine potential ameliorative effects of vocal WU exercises on PTP and PPE after the loud-reading task. The predicted results were that PTP and PPE would not change after vocal rest, decrease after vocal WU exercises, increase after 1 hr of loud reading, and return to levels similar to baseline after 30 min of vocal silence. In addition, PTP and PPE were predicted to have a smaller increase after 1 hr of loud reading when preceded by vocal WU exercises than by vocal rest.

Method Participants Eight women, ranging in age from 20 to 28 years (M = 22.5, SD = 2.5), who presented with complaints of vocal fatigue participated in this study. They were nonsmokers, had no formal speaking or singing training, and had no history of laryngeal pathology, severe upper respiratory infection, asthma, pulmonary disease, or neurological disease. All participants passed screening examinations for laryngeal appearance, pulmonary function, and hearing. Volunteers received $10 per hour for their participation; completion of the study required approximately 7 hr.

Procedures After an initial telephone interview, qualifying candidates participated in a screening and practice session and two experimental sessions. Sessions were

scheduled in the morning, before most daily speech activities. Scheduling also accounted for potential increases in hormonally related edema (Abitbol, Abitbol, & Abitbol, 1999; Davis & Davis, 1993); 3-day periods before and after ovulation and menstruation were avoided. Each experimental session lasted about 3 hr and they were separated by at least 72 hr to allow time for full vocal recovery. Participants were asked to refrain from loud talking for 24 hr and from taking ibuprofen or aspirin for 8 hr before sessions to reduce the risk of vocal fold hemorrhaging (Sataloff, 1997). They were told to consume similar amounts of nourishment before each session because systemic hydration can affect vocal function (Fisher et al., 2001; Solomon & DiMattia, 2000; Verdolini et al., 1994). The screening session included a laryngoscopic examination with rigid endoscopy to confirm normal laryngeal appearance, pulmonary screening (FEV1/FVC ≥ 0.80, where FEV1 is forced expiratory volume in 1 s and FVC is forced vital capacity), and pure-tone screening audiometry (20 dB HL at 0.5, 1, 2, and 4 kHz). Also during the initial session, pitch range (including falsetto, excluding glottal fry) was determined and then segmented into 10th, 50th, and 80th percentiles. The PTP and PPE tasks were trained and then repeated three times, each separated by a 3-min vocal-rest period, at each of the three previously listed target pitches. The order of pitches was counterbalanced between and within participants. This extensive practice regimen was intended to reduce potential learning effects on future baseline measures. Finally, the vocal WU exercises were introduced and rehearsed in their entirety to familiarize participants with the procedures and to increase their comfort level and skill for performing the exercises. Experimental sessions consisted of three sets of baseline data recordings, each separated by 3 min of silence, a 15–20 min vocal-preparation condition, 1 hr of reading loudly, and 30 min of vocal silence. Two types of vocal-preparation tasks preceded the loud-reading task, one for each experimental session: WU exercises and vocal rest. Data were collected eight times during each experimental session: at three baseline recordings, once after the vocal-preparation condition, and midway through and at the end of the loud-reading and vocalsilence segments.

Data Collection Participants performed tasks that provided PTP and PPE data. The procedures for training and recording PTP data have been described in detail previously (Solomon & DiMattia, 2000). Briefly, a commercially available aerodynamic measurement system (Glottal Enterprises MS100-A2, MCU-4 Calibration Unit) was used to acquire oral pressure and airway-opening flow

Milbrath & Solomon: Vocal Warm-Up and Fatigue

425

signals, and each was digitized at 1 kHz onto a laboratory computer. Participants repeated the syllable /pi/ seven times at a rate of 1.5 syllables/s, following the method of Smitheran and Hixon (1981), using the quietest possible phonation. At least three acceptable syllable trains, determined by online visual and auditory monitoring, were recorded at each of the three target pitches (10th, 50th, and 80th percentiles of the pitch range). Typically, five or six syllable trains were recorded to ensure adequate data collection. Productions were monitored for continuous phonation during vowels and the cessation of airflow during the closed phase of /p/. Occasionally, nose clips worn within the face mask were required to prevent nasal air leakage, an occurrence that results in spuriously low tracheal pressure estimates (Fisher & Swank, 1997; Smitheran & Hixon, 1981). Pitch levels were counterbalanced across participants. Pitch order also was counterbalanced within participants for baseline measures but kept consistent for the remaining data-collection trials. In this way, pitch order within participants was varied for multiple measures and fixed for single-trial measures. This was done to maintain the same time course across pitches so that same-pitch data could be directly compared for each participant. Participants rated PPE on a visual-analog scale immediately after performing the PTP task at each pitch. Ratings involved marking a 20-cm undifferentiated line labeled no effort and extreme effort at the endpoints. For informal assessment of vocal fatigue and general wellbeing, the participants were asked to qualify “How do you feel?” at 15-min intervals throughout the loud-reading and vocal-silence tasks. These responses were used as an aide in judging the participant’s willingness to continue with the protocol and in interpreting the results of the study.

Vocal-Preparation Conditions Participants were informed that there are two general approaches to preparing the voice for strenuous activity: resting and conserving the voice, and warming up the voice with vocal exercises. Each participant engaged in both of these conditions, one during each experimental session; the order was alternated between participants. During the vocal-preparation conditions, participants drank 250 ml of water. The first vocal-preparation condition consisted of 20 min of vocal rest and general relaxation, intended as a placebo condition. The participant was instructed to sit quietly in a chair and focus on her breathing and relaxing the body. Soft music was played in the background to provide a peaceful atmosphere and to help her acclimate to the environment. The second condition consisted of 15–20 min of vocal WU exercises, which are listed in their entirety in 426

the Appendix. The WU exercises focused on respiration, resonance, and phonation. They were meant to be simple and targeted to the nonsinger. Each activity was performed twice. Briefly, the WU routine began with a series of static 5-s stretches targeting the arms, chest, back, neck, mandible, and tongue. Participants then sat with well-aligned posture and inhaled and exhaled slowly and deeply. They then sustained pitches (1–2 s) on nasal consonants or nasalized vowels; they were instructed to “feel the nose or upper lip buzz.” The starting pitches were adjusted for each individual so that the task encompassed 10% and 80% of her pitch range. The last set of exercises focused on phonation while incorporating the improved respiration and resonance just practiced. These tasks followed the four vocal-function exercises described by Stemple et al. (2000) and included maximally sustained vowels or syllables and ascending and descending pitch glides. All exercises were performed as quietly as possible, without becoming breathy and without strain.

Speaking Tasks Immediately following the vocal-preparation condition, PTP and PPE data were collected, and the participant began loudly reading from an entertaining book (Rowling, 1997). SPL was monitored by the investigator with a sound level meter (Extech 407740, C-weighting, slow-response) placed 45 cm from the participant’s mouth. SPL was targeted at 75–80 dB; verbal feedback and encouragement was provided to help the participant maintain loud phonation. After 30 min of loud reading, PTP and PPE measures at the three target pitches were recorded. The participant then continued to read loudly for an additional 30 min, after which data were collected again. To examine whether PTP and PPE ratings returned to baseline levels (suggesting vocal recovery) after the loud-reading task, the participant remained silent for 30 min. After 15 min and again after 30 min of vocal silence, PTP and PPE data were collected.

Data Reduction and Analysis The first author reduced the PTP data by examining the pressure waveforms using data-analysis software (Windaq; DATAQ Instruments, Akron, OH). Nine pressure peaks, three from the middle of each of three syllable-repetition strings, were selected as accurate estimates of tracheal pressure during the quietest possible phonation. Criteria for selection included cessation of airflow during the closed phase of /p/; the presence of alternating air flow (indicating phonation) during the vowels; not the first or last syllables produced in the string; and consistency across peak values. On occasion,

Journal of Speech, Language, and Hearing Research • Vol. 46 • 422–436 • April 2003

three adjacent pressure peaks did not meet selection criteria; in these cases, valid pressure peaks were selected from an extra syllable train. Once selected, peak pressure values were determined using a peak-picking software feature. The nine peak values were averaged to provide a single value for each set of productions. PPE was determined by measuring with a ruler the proportion of the visual-analog scale marked by the participant. PTP and PPE data were analyzed with repeatedmeasures analyses of variance (RM-ANOVAs). The three within-subjects factors were session (vocal rest or WU), pitch (10%, 50%, or 80% of pitch range), and the datacollection trial (median baseline, after vocal-preparation, after 1 hr of loud reading, or after 30 min of vocal silence). Although data were collected after 30 min of loud reading and after 15 min of vocal silence, these values were not included in the formal analysis. Rather, they were collected to monitor the time course of the results. The baseline value used was the median value of the three repeated trials collected before the vocal-preparation condition at each pitch. Post hoc pairwise comparisons were completed to examine significant effects. Because separate RM-ANOVAs were performed for PTP and PPE, a Bonferroni correction was applied; effects were considered statistically significant at p < .025. Pearson product–moment correlations examined associations between PTP and PPE. Correlation coefficients were calculated for each pitch tested (10th, 50th, and 80th percentiles of the pitch range) and each par-

ticipant. Significant correlation coefficients (p < .05) were interpreted to be moderate if between .51 and .85 and to be weak if between .3 and .5 (Silverman, 1993).

Reliability Sixteen blinded samples (13%) of the PTP and PPE data were randomly selected and measured by both authors at least 2 months after the original data analysis for determining inter- and intrarater reliability. For PTP, intrarater agreement was 90% within 0.25 cmH2O and 94% within 0.50 cmH2O; interrater agreement was 81% within 0.25 cmH2O and 98% within 0.50 cmH2O. For PPE, intrarater and interrater agreement were both 98% within 1% of the scale. Pearson product–moment correlation coefficients were greater than .98 for PTP and greater than .999 for PPE for all samples (all pitches combined) and for each pitch individually for both inter- and intrarater reliability.

Results Table 1 lists complete RM-ANOVA results. PTP differed significantly across pitch, F(2, 14) = 49.31, p < .0001. Pairwise comparisons for pitch revealed that PTP at the 80% pitch was significantly greater than that for both the 10% and 50% pitches. No significant differences were detected for session, F(1, 7) = 0.05, p = .8281, or for the trial within each session, F(3, 21) = 1.81, p = .1764. PPE did not differ significantly across session, F(1, 7) =

Table 1. Repeated-measures within-subjects analysis of variance results for phonation threshold pressure (PTP) and perceived phonatory effort (PPE). PTP

PPE

df

MS

F

p

MS

F

p

1 7

0.29 5.68

0.05

.8281

326.16 395.79

0.82

.3942

Pitch Error

2 14

390.13 7.91

49.31