The Effects of Humming on the Prephonatory Vocal Fold Motions Under High-Speed Digital Imaging in Nondysphonic Speakers

ARTICLE IN PRESS The Effects of Humming on the Prephonatory Vocal Fold Motions Under High-Speed Digital Imaging in Nondysphonic Speakers *Toshihiko Iwahashi, *Makoto Ogawa, *,†Kiyohito Hosokawa, *Chieri Kato, and *Hidenori Inohara, *Suita, †Osaka, Japan

Summary: Objectives. This study aimed to investigate whether humming affects the adductive motion of the vocal folds and transient glottal closure in the prephonatory adjustment phase of vocal onset using high-speed digital imaging (HSDI) and a motion analysis software program. Methods. Twenty normal healthy adults without any vocal abnormalities were enrolled. While a transnasal flexible fiberscope connected to a high-speed camera was inserted, each participant was asked to perform three phonatory tasks— natural /e:/ phonation, loud /e:/ phonation, and humming /m:/ phonation—and laryngeal HSDI movies (4000 frame/s) were recorded. On each HSDI movie, the duration of the prephonatory glottal closure was measured. In addition, using motion analysis, the changes in the angle between the bilateral vocal folds during vocal fold adduction and the average angular velocity in the ranges of 100%–80%, 80%–20%, and 20%–0% from all of the angular changes were analyzed. Results. The angular changes showed sigmoid and polynomial-like curves during the natural/humming and loud phonation, respectively, and the 80%–20% and 20%–0% average velocities were the highest during the natural/humming and loud phonation, respectively. The humming phonation decreased all of the average regional velocities, eliminated the transient prephonatory glottal closures observed during the natural and loud phonation, and induced a greater value for the minimal angle than the natural phonation. Conclusions. The present study demonstrates that humming encourages easy vocal initiation by decelerating the vocal fold adductive motion throughout the prephonatory adjustment phase and alleviating transient prephonatory laryngeal closure, leading to gradual and smooth vocal fold positioning. Key Words: Humming–Loud phonation–Vocal onset–High-speed digital imaging–Motion analysis.

INTRODUCTION Humming is a well-known vocal training technique for producing a resonant voice.1–5 Previous textbooks on voice therapy1–3 have emphasized the importance of feeling resonance in the nose, cheeks, or lips during humming in order to effectively induce a hum. Yiu et al4,5 also described that speakers should produce the sound /m:/ followed by gliding the pitch to the most comfortable and natural level during humming as if sincerely agreeing with someone. Indeed, voice therapy sessions using humming have been reported to improve the perceptual vocal quality in patients with vocal nodules or laryngitis4 and those diagnosed to have muscle tension dysphonia (MTD) with supraglottic compression,6 and to decrease the computed perturbation parameters of acoustic and electroglottographic (EGG) signals in MTD patients.7 To elucidate the mechanisms underlying the vocal improvement by humming, our research group has assessed the laryngeal changes that occur during sustained humming phonation using objective methodologies.8–10 The results of these studies revealed multifarious effects of humming in both dysAccepted for publication September 8, 2016. Conflict of interest: The authors declare no conflicts of interest in association with the study. Disclosure: The authors alone are responsible for the content and writing of the paper. From the *Department of Otorhinolaryngology—Head and Neck Surgery, Osaka University Graduate School of Medicine, Osaka, Japan; and the †Department of Otorhinolaryngology, Osaka Police Hospital, Osaka, Japan. Address correspondence and reprint requests to Makoto Ogawa, Department of Otorhinolaryngology—Head and Neck Surgery, Osaka University Graduate School of Medicine, 2-2 Yamadaoka, Suita, Osaka 565-0871, Japan. E-mail: [email protected] Journal of Voice, Vol. ■■, No. ■■, pp. ■■-■■ 0892-1997 © 2016 The Voice Foundation. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jvoice.2016.09.008

phonic patients and nondysphonic speakers. The effects were as follows: (1) an immediate decrease in the degree of supraglottic compression in both MTD patients and nondysphonic speakers;8 (2) an immediate decrease in the perturbation parameters of EGG signals, reflecting the irregularity of the contact between the bilateral vocal folds in the vibratory cycles in both MTD patients and nondysphonic speakers;9 (3) an immediate induction of a slight increase in the contact quotient (CQ), ie, the temporal ratio of the vocal fold closed phase to each period during phonation, in MTD patients;9 and (4) an immediate decrease in the perturbation parameters of the glottal area waveforms derived from the laryngeal movies recorded by high-speed digital imaging (HSDI) in a group of nondysphonic speakers and dysphonic patients with benign vocal mass lesions combined.10 These results suggest that humming modulates the spatial interrelationship between the laryngeal structures, leading to the optimization of the phonatory dynamics in the larynx. In addition, a textbook by Colton and Casper2 mentions encouragement of easy vocal initiation as one of the rationales for the use of humming. Werner-Kukuk and von Leden11 define vocal initiation/attack/onset as “a function of vocal cord placement at the start of phonation which in turn depends on the precise coordination between the subglottic pressure and the resistance at the level of larynx.” Recently, Wittenberg et al12 divided the process of vocal initiation into four main events: adduction of the vocal folds, prephonatory glottis closure, onset of the oscillation, and steady-state oscillation. In contrast, Watson et al13 stated that “the initiation of phonation includes two somewhat distinct phase: (1) the prephonatory adjustment phase that is associated with setting the appropriate tension, gross adduction,

ARTICLE IN PRESS 2 and aerodynamic forces, and (2) the attack phase that is associated with the onset of the vocal fold oscillation and sound generation.” Thus far, variable patterns of vocal onset have been characterized using various experimental methodologies, such as high-speed cinematography,11,14 pneumotachography,15,16 electromyography (EMG) of the intrinsic laryngeal muscles,17,18 EGG,13,19 image analyses of laryngeal video movies,20 and computed motion analyses of HSDI movies of the larynx.21–23 In particular, Werner-Kukuk and von Leden 11expounded on the three main types of physiological vocal attacks, namely the breathy, the hard, and the soft attacks, as follows: (1) in the breathy attack, the vocal folds assume “a paramedian position prior to sound production”; (2) in the hard attack, “the glottis is closed tightly before the onset of phonation”; and (3) the soft attack is “a smooth onset of phonation without any aspirate sound, and is produced through gradual adduction of the vocal cords towards the median line, with the gentle closure of the glottis.” Especially regarding the prephonatory adductive vocal fold movement in the three patterns of phonatory onset, the velocity of the vocal fold adduction in the prephonatory adjustment phase has been reported to be faster and slower in the hard and breathy attacks, respectively.11,20 Colton and Casper2 also describe that rapid and complete adduction of the vocal folds prior to the initiation of phonation characterizes a hard glottal attack. In recent decades, with advancements of digital technology, computed image analyses of the normal laryngeal videos now enable the determination of the kinematics of the vocal fold adductive motion in the prephonatory adjustment phase, but in a discontinuous manner.20,24,25 Later, the use of HSDI has allowed for analyzing the continuous changes in parameters derived from the laryngeal images.26,27 In particular, we recently applied the combination of HSDI and a motion analysis software program to analyze the vocal fold adductive motion during throat clearing (TC) and normal phonation, and confirmed that the kinematics of the angular changes between the vocal folds were different: a sigmoid and a polynomial-like curve during normal phonation and TC, respectively.27 Regarding the patterns of prephonatory transient glottal closure at the three patterns of vocal onset, Wittenberg et al21 reported the digital kymographic characteristics as follows: (1) in the normal phonation, the prephonatory vocal fold adduction movement is followed by a period of prephonatory glottal closure, and during this period the ventricular folds show a discrete adduction movement; however, the view at the vocal folds is never obscured; (2) in the hard attack, a more prolonged period of prephonatory standstill, ie, a transient laryngeal closure where the adductive movement of the ventricular folds that temporarily cover the vocal folds can be observed; (3) in the breathy attack, the vocal folds do not end up in a vocal fold contact, and the glottal closure remains incomplete during the phase of prephonatory buildup of the muscular tension with a shorter prephonatory standstill. In the present study, by applying our previous methodology using the combination of HSDI and a motion analysis software program,27 we aimed to verify (1) whether or not humming affects the velocity of the adductive motion of the vocal folds; (2) whether or not humming changes the minimal vocal fold angle, approximate to the terminal phonatory position of the vocal folds; (3) whether or not humming modulates the duration of

Journal of Voice, Vol. ■■, No. ■■, 2016

the prephonatory transient glottal closure in the vocal onset of nondysphonic speakers; and (4) whether or not loud phonation like shouting, a representative of inappropriate phonation in the view of vocal hygiene, changes the above-mentioned measures. METHODS Participants The protocol of this study complied with the Declaration of Helsinki, and an institutional review board approval was obtained from the Osaka University Graduate School of Medicine (No. 15592). Twenty healthy nondysphonic participants without any musical or theatrical training (medical staff members and postgraduate students) who showed neither functional nor organic laryngeal abnormalities (13 men and 7 women; median age: 32.5 years; range: 23–60 years) were included. Before the recording of the laryngeal HSDI videos, each patient underwent a routine ENT examination to confirm the absence of laryngeal abnormalities. Recording of the laryngeal HSDI videos and EGG signals In a common ENT consultation room without soundproofing, the laryngeal HSDI movies and EGG signals during the phonatory tasks were recorded synchronously by a single laryngologist (TI). After placing the EGG electrodes (Model EG2PCX2; Glottal Enterprises Inc, Syracuse, NY) on the neck skin surface and topically administering 1% lidocaine and 0.02% adrenalin into the nasal cavity, each participant in a seated position underwent the insertion of a flexible rhino-laryngo fiberscope (ENT-P4; Olympus, Tokyo, Japan) connected to a monochrome high-speed camera system (Phantom Miro eX4; Vision Research Inc, Wayne, NJ) and a 300W xenon light source (CLV-S40 Pro; Olympus). The tip of the fiberscope was located at a level just below the uvula. Subsequently, under the observation of the larynx, the laryngologist asked the participant to perform three tasks, namely (1) stable natural phonation of “/e:/” at a habitual pitch and loudness for more than 3 seconds and (2) loud phonation of the vowel “/e:/” like a shout for more than 3 seconds, followed by (3) humming phonation to close his or her lips and hum “/m:/” for more than 3 seconds in a relaxed manner while feeling resonance in the nose or lips, without changing the pitch as possible. In cases where an individual failed to produce a hum appropriately or sufficiently loud voice, the performance of the tasks could be repeated up to three times. Consequently, all of the participants were judged to be able to perform all of the tasks satisfactorily. While the participant was performing the tasks, the HSDIs and EGG signals were recorded synchronously. HSDIs of 256 × 256 pixels in size were captured at 4000 frames/s at 16-bit resolution. The digitalization and synchronous data acquisition were achieved using a data acquisition device (NI-USB 6341; National Instruments, Tokyo, Japan) set at a sampling frequency of 48,000 Hz with 16-bit quantization. The recorded data were transferred to a Windows PC (ThinkPad E540; Lenovo, Tokyo, Japan), converted to the cinefile format and then stored. The data recordings and transfer were completed for all of the participants within 20 minutes, because the transfer of data of each task requires a few minutes.

ARTICLE IN PRESS Toshihiko Iwahashi, et al

Humming Decelerates Prephonatory Vocal Fold Adductive Motions

Motion analysis of vocal fold angle and angular velocity from the laryngeal HSDI The changes in the vocal fold angle and the angular velocity were analyzed via the same method as in our previous study.27 Briefly, on the laryngeal HSDI movie of each phonation, the contrast and brightness were adjusted using an image tool in the Phantom Camera Control software (PCC2.6; Vision Research Inc) in order to strengthen the contrast between the glottic areas and the glottal area. Subsequently, the sequential HSDI frames were converted to AVI format. These files were then opened in a motion analysis software program (DIPP-Motion Pro; DITECT, Tokyo, Japan) that allows for the acquisition of the trajectories of several points on a frame-by-frame basis with the automatic detection of the designated points via pixel information, followed by the calculation of the lengths and angles formed by these points. Three tracking points were set on the tips of the bilateral vocal processes and the anterior commissure (Figure 1A). Subsequently, a motion analysis was performed by tracking these three points, followed by the calculation of the vocal fold angle from the tracking information. The changes in the vocal fold angle were then plotted on a graph (Figure 1B). Graphs were also drawn to show the changes in the angular velocity as the time derivative of the vocal fold angle using the same software program. Determination of the borders of each prephonatory event and calculation of the regional average angular velocities from the laryngeal HSDI movies Similar to the method of Freeman et al,22 the borders of the prephonatory events of each phonatory task were determined as follows: the times of (1) maximal glottal opening, (2) the contact

3

between the bilateral vocal folds and between the bilateral ventricular folds, (3) the dissociation between the bilateral vocal folds and between the bilateral ventricular folds, and (4) the beginning of vocal fold oscillation. With regard to the average regional angular velocity, on the graph for the angular change, the region from the time of maximal glottal opening to the time of the maximal approximation (including contact) between the bilateral vocal folds (ie, from the instant when the vocal fold angle value reaches its maximal value to the time when the angle reaches its minimal value [including zero]) was determined. In addition, the maximal and minimal angle values were recorded. Subsequently, we calculated the average angular velocities between 100% and 80%, between 20% and 0%, and between 80% and 20% from all of the angular changes (Figure 1B). Statistical analysis To explore the relationship between the durations of the prephonatory events and the tasks, between the velocities of the different ranges of the prephonatory adjustment phase and the tasks, and between the maximal/minimal vocal fold angles and the tasks, we conducted a two-way repeated measures analysis of variance (ANOVA) with these measures and the three tasks (natural phonation, loud voice phonation, and humming phonation) set as the within-subject factors, and the possible main effects and their interaction were assessed. Subsequently, a posthoc Wilcoxon’s signed-rank test was used to analyze the differences among the velocities of the differential ranges and between the tasks. The significance was set at P = 0.05. The JMP ver. 11.2.0 software program (SAS Institute, Cary, NC) was used to perform the statistical analyses.

FIGURE 1. The method of analyzing the change in the angle between the bilateral vocal folds and calculating the average angular velocities of different ranges in the total amount of the vocal fold angular change during normal phonation. Using a motion analysis software program (DIPPMotion Pro), three tracking points were established on the tips of the bilateral vocal processes and the anterior commissure (A), and a motion analysis of the tracks of these three points was carried out to obtain the pixel information of these points. The vocal fold angle was then calculated from the pixel information, and a graph of the change in the vocal fold angle was drawn (B). The average angular velocities in the ranges of 100%– 80%, 80%–20%, and 20%–0% of the total amount of the change in the vocal fold angle were calculated (B).

ARTICLE IN PRESS 4 RESULTS Figures 2A, 2B, and 2C, respectively, show the sequential frames of the laryngeal HSDIs during the three phonatory tasks of one of the participants. In natural phonation (Figure 2A), the vocal folds and ventricular folds adduct, and the vocal folds closed with paramedian positioning of the ventricular folds. Subsequently, the glottal closure releases, and then the vocal folds position in the median line and start to oscillate. Next, in loud phonation (Figure 2B), the vocal folds and ventricular folds adduct promptly, and then the vocal folds close with severe approximation/contact between the bilateral ventricular folds. Subsequently, the ventricular folds are dissociated slightly, and the vocal folds reposition in the median line, followed by vocal fold oscillation. In contrast, in humming phonation (Figure 2C), the vocal folds and the ventricular folds adduct slowly, and the vocal folds position close to the median line without showing either prephonatory vocal fold closure or contact between the bilateral vocal processes, followed by vocal fold oscillation. First, regarding the velocity of the vocal fold adductive motion, Figure 3 presents the changes in the vocal fold angle and the angular velocity in the prephonatory adjustment phase during the three phonatory tasks of one of the participants, obtained by a motion analysis software program. The changes in the vocal fold angle and angular velocity were smoothed by a moving average method of the motion analysis software program. The changes in the vocal fold angle showed polynomial-like and sigmoid curves for loud and natural/humming phonation, respectively. Furthermore, these graphs demonstrate that the sigmoid curve of humming phonation ranged for a longer time than that

Journal of Voice, Vol. ■■, No. ■■, 2016

of natural phonation. In addition, during loud phonation, the angular velocity showed a continuous increase and suddenly reached 0. Table 1 shows the average angular velocities in the ranges of 100%–80%, 80%–20%, and 20%–0% of the total vocal fold motion in the prephonatory adjustment phase of each of the tasks. A two-way repeated measures ANOVA revealed a significant interaction between the factors of the velocities and tasks (F [4, 76] = 17.74; P < 0.001), as well as significant main effects for the velocities (F [2, 38] = 61.67; P < 0.001) and task factors (F [2, 38] = 22.89; P < 0.001). On comparison among the regional average velocities within each tasks, during natural and humming phonation, the 80%–20% average angular velocity was higher than those of the 100%–80% and 20%–0% regions (during natural phonation between 100%–80% and 80%–20%, P < 0.001, and between 80%–20% and 20%–0%, P < 0.01; during humming phonation between 100%–80% and 80%–20%, P < 0.001, and between 80%–20% and 20%–0%, P < 0.01), whereas the 20%– 0% average angular velocity was higher than those of the 100%– 80% and 80%–20% regions during loud phonation (between 100%–80% and 80%–20%, P < 0.001; between 80%–20% and 20%–0%, P < 0.05). On comparison of each average regional angular velocity among the three tasks, loud phonation induced the highest 20%–0% average velocities in all of the tasks (between loud and natural phonation: P < 0.001; between loud and humming phonation: P < 0.001). Indeed, the 20%–0% average angular velocity during loud phonation was approximately twofold higher than that during natural phonation. On comparison of the regional velocities between natural and humming phonation,

FIGURE 2. The time course of high-speed digital imaging in the prephonatory adjustment phase during the three phonatory tasks. Event A: from the maximal glottal opening to vocal fold contact; event B: from vocal fold contact to dissociation; event C: from ventricular fold contact to dissociation. NP, natural phonation; LP, loud voice phonation; HP, humming phonation; VF, vocal fold.

ARTICLE IN PRESS Toshihiko Iwahashi, et al

5

Humming Decelerates Prephonatory Vocal Fold Adductive Motions

FIGURE 3. The changes in the vocal fold angle and angular velocity during the three phonatory tasks. NP, natural phonation; LP, loud voice phonation; HP, humming phonation; VF, vocal fold. humming phonation exhibited lower average angular velocities for all of the ranges (100%–80%, P < 0.01; 80%–20%, P < 0.001; 20%–0%, P < 0.001). In particular, the 20%–0% average angular velocity during humming phonation was approximately half that during natural phonation. Table 2 shows the averages of the maximal and minimal angle values during the three phonatory tasks. A two-way repeated measures ANOVA revealed a significant interaction between the factors of the angles and tasks (F [4, 76] = 16.67; P < 0.001), as well as the significant main effects for the angle factors (F [2, 38] = 200.1; P < 0.001) and task factors (F [2, 38] = 4.048; P < 0.05). A posthoc analysis showed that the average maximal angle values did not differ significantly among the three tasks. However, the minimal angle value during loud and humming phonation was significantly smaller and greater than that during natural phonation, respectively (between loud and natural pho-

TABLE 1. Average Angular Velocities of Different Regions During the Three Phonatory Tasks Average VF Angular Velocity (Mean ± SD, °/s) Task NP LP HP

100%–80%

80%–20%

20%–0%

117*** ± 33n.s. 115*** ± 42‡ 91*** ± 42§§

417** ± 178n.s. 391* ± 212‡‡ 208** ± 101§§§

305 ± 156††† 493 ± 199‡‡‡ 135 ± 91§§§

Wilcoxon’s signed-rank test was used for the comparisons in the three tasks between the 100%–80% and 80%–20%, and between the 80%–20% and 20%–0% (statistical significance: *P < 0.05, **P < 0.01, ***P < 0.001). Wilcoxon’s signed-rank test was used for the comparisons in each range of the average VF angular velocity between the NP and LP (statistical significance: †††P < 0.001), and between the LP and HP (statistical significance: ‡P < 0.05, ‡‡P < 0.01, ‡‡‡P < 0.001), and between the NP and HP (statistical significance: §§P < 0.001, §§§P < 0.001). Abbreviations: NP, natural phonation; LP, loud voice phonation; HP, humming phonation, VF, vocal folds; SD, standard deviation; n.s., not significant.

nation: P < 0.05; between natural and humming phonation: P < 0.001). Next, regarding the prephonatory transient glottal closure followed by vocal fold adduction, Figures 4A, 4B, and 4C show the respective durations of each prephonatory event during natural, loud, and humming phonation, respectively. During natural phonation, 15 (75%) of the 20 participants showed prephonatory transient glottal closure without contact between the ventricular folds. In contrast, during loud phonation, 14 (70%) and 5 (25%) of the 20 participants showed prephonatory transient glottal closure without any contact between the ventricular folds, and both glottal and ventricular closure, respectively. In addition, during humming phonation, 17 (85%) of the 20 participants did not show any prephonatory glottal closure. DISCUSSION In the present study, to assess how humming and loud voice phonation modulate the prephonatory adjustment phase of vocal onset, we compared the characteristics of the prephonatory laryngeal motions, including the velocity of the vocal fold adductive

TABLE 2. Maximal and Minimal Angle Between the Vocal Folds During the Three Phonatory Tasks Mean ± SD (°) Task

Maximal Angle

Minimal Angle

NP LP HP

60.3n.s. ± 13.6 58.3n.s. ± 11.5 53.4n.s. ± 13.1

1.0* ± 2.0 0.1††† ± 0.3 7.1‡‡‡ ± 3.7

Wilcoxon’s signed-rank test was used for the comparisons in the three tasks between the NP and LP (statistical significance: *P < 0.05), and between the LP and HP (statistical significance: †††P < 0.001), and between the NP and HP (statistical significance: ‡‡‡P < 0.001). Abbreviations: NP, natural phonation; LP, loud voice phonation; HP, humming phonation; SD, standard deviation; n.s., not significant.

ARTICLE IN PRESS 6

Journal of Voice, Vol. ■■, No. ■■, 2016

FIGURE 4. The duration of events in the prephonatory adjustment phase during the three phonatory tasks in each participant. Event A: from the maximal glottal opening to vocal fold contact; event B: from vocal fold contact to dissociation; event C: from ventricular fold contact to dissociation. NP, natural phonation; LP, loud voice phonation; HP, humming phonation; VF, vocal fold. movement, the maximal/minimal angles between the bilateral vocal folds, and the patterns of transient prephonatory glottal closure on the laryngeal HSDI movies among the three phonatory tasks: natural, loud, and humming phonation. Different patterns of the changes in the angle between the vocal folds among the three phonatory tasks Several previous studies11,20,21 have reported that the velocity of vocal fold adduction is faster and slower in the hard and breathy attacks, respectively. In the present study, where the use of HSDI and a motion analysis software program allowed for a more detailed analysis of the changes in the angle and angular velocity between the bilateral vocal folds, the changes in the angle between the vocal folds were manifested as sigmoid and polynomial-like curves during natural/humming and loud phonation, respectively. In addition, the polynomial-like curve of loud phonation suddenly reached 0. An analysis of the average regional angular velocities of the three regions supported this finding: the 80%– 20% and 20%–0% average angular velocities were the highest during natural/humming and loud phonation, respectively. These results indicate that the adductive vocal fold motion in the prephonatory adjustment phase accelerates continuously during loud phonation, and decelerates just before the termination of the vocal fold adduction during natural and humming phonation. Comparison of the velocities of the vocal fold adductive movement between loud voice phonation and throat clearing Of the three average regional angular velocities, the 20%–0% average velocity is considered to be the closest to the terminal velocity just before the termination of the vocal fold adduction or collision. In our previous study investigating the laryngeal movement during TC,27 the 20%–0% average angular veloci-

ties of the vocal fold adductive movement in the precompression closing phase of strong and weak TC were reported to be 597 ± 185°/s and 702 ± 215°/s, respectively. In the present study using the same methodology, the 20%–0% average angular velocities during the natural, loud, and humming phonation were 493 ± 199, 305 ± 156, and 135 ± 91°/s, respectively, indicating that all of the three phonatory tasks induced a lower terminal velocity than TC, and that loud phonation generated the highest terminal velocity, respectively. Colton and Casper2 define a hard glottal attack as “a manner of initiating vowels, usually characterized by rapid and complete adduction of the vocal folds prior to the initiation of phonation.” Indeed, Cooke et al20 reported that of the breathy, hard, and soft vocal onsets, the hard onset phonation induced the highest normalized velocity for the distance between the bilateral vocal processes. These results suggest that loud phonation and hard onset phonation have a similar effect of accelerating the velocity of vocal fold adduction. In the present study, the minimal angle was found to be 0 during loud phonation and the curve for the angular velocity exhibited a continuous increase, followed by suddenly reaching 0, indicating that loud phonation induced not only intense vocal fold collision but also an additional adductive force at the instance of vocal fold contact, leading to subsequent firm glottal closure. According to Titze,28 the impact stress as collision force per unit area σ can be approximated based on the distance from the anterior commissure to the collision point L , the terminal angular velocity ω , the tissue density ρ , the mass depth of a tissue element at the medial surface of the contact Δx , and the impact interval Δt, using the following equation:

σ = LωρΔx Δt

(1)

This equation indicates that the impact stress at the vocal fold collision is proportional to the terminal angular velocity just before

ARTICLE IN PRESS Toshihiko Iwahashi, et al

Humming Decelerates Prephonatory Vocal Fold Adductive Motions

the collision. Indeed, Hess et al 29 reported that the interarytenoid pressures induced during TC and thoracic fixation (<50 kPa and >100 kPa, respectively) were higher than the prephonatory and postonset pressures during loud voice phonation with hard onset (18 kPa and 6.6 kPa, respectively). The present study also demonstrated that the 20%–0% average velocity during loud phonation was lower compared with the above values during strong and weak TC. These results suggest that the harmful effect on the vocal fold at the onset of TC is greater than that during loud voice phonation with respect to vocal hygiene. Comparison of the patterns of prephonatory glottal closure between normal and loud voice phonation With regard to the patterns of the prephonatory laryngeal closure, previous studies using high-speed cinematography11,14 have found that during hard attack phonation, “the glottis is closed tightly before the onset of phonation” and that “the so-called stroke (hard) attack involves more constriction in the superior laryngeal musculature than the breathed and the simultaneous attacks.” In addition, Wittenberg et al 21 reported different patterns of prephonatory laryngeal closure at the levels of both the vocal and ventricular folds during hard attack phonation and only at the vocal fold level during normal phonation. However, no studies have compared the laryngeal closure patterns among different phonatory onset in a series of subjects. In the present study, we compared the laryngeal closure patterns in 20 nondysphonic participants, and found that only 5 (25%) of the participants exhibited double closure at both the vocal and ventricular fold levels during loud phonation, whereas no participants showed double closure during natural phonation. Generally, a hard vocal attack has been recognized as an inappropriate habit with respect to vocal hygiene.2,30,31 Phonation with high intensity has been associated with high glottal resistance due to tight vocal fold closure,32,33 as well as increased subglottic pressure and expiratory airflow.32,34 Accordingly, it appears that the prephonatory double closure plays a compensatory role for augmenting the subglottic pressure just before the beginning of vocal fold oscillation by increasing the glottal resistance, leading to the production of a loud voice with high intensity. We hypothesize that the participants showing the double closure during loud phonation contracts voice disorders well compared with those exhibiting the single closure only at the vocal fold level, as participants with double closure may tend to increase the subglottic pressure by augmenting the laryngeal resistance rather than making an expiratory effort using the abdominal musculature. Regarding the effect of the prephonatory laryngeal closure on produced sound, prephonatory tight glottic closure during hard attack phonation has been deduced to generate the audible impression as an abrupt, explosive, and hard-edged sound.2,11 The abrupt release of the forced laryngeal closure at both vocal and ventricular levels is suspected to produce this explosive sound. In addition, the abrupt opening of tight glottic closure during loud phonation may generate sounds like glottal stops as normal laryngeal articulation35 or a compensatory consonant observed in cleft-palate patients,36 thereby impairing voice quality.

7

Comparisons of the minimal vocal fold angle between the vocal folds among the different phonatory tasks We observed significant differences in the average minimal angle between the vocal folds among the three phonatory tasks. In particular, the average minimal angle value during humming phonation (7.1 ± 3.7°) was greater compared with that during natural phonation (1.0 ± 2.0°), demonstrating that humming induces a slightly weaker adduction of the vocal folds than normal phonation. As mentioned above, the vocal folds started to oscillate with contact between the bilateral vocal processes during natural phonation, and repositioned close to the midline smoothly and directly without any transient glottal closure, followed by oscillation during humming phonation. Accordingly, these minimal angle values during natural and humming phonation were considered to be coincident with the angles in the phonatory position. Verdolini et al37 hypothesized that resonant voice phonation adjusts the degree of glottal adduction to “a barely adducted or barely abducted state,” and compared the visual perceptual ratings for laryngeal adduction and the CQ calculated from the EGG signals among various sustained phonatory tasks, including normal, resonant, pressed, and breathy voice. However, the authors found no differences in either of the perceptual visual scale or the EGG-CQ value between the sustained normal and resonant voice phonation. Our present results are the first to prove the truth of this hypothesis. Effects of humming on the velocity of vocal fold adductive motion, prephonatory transient glottal closure, and the phonatory glottic position With regard to the effect of humming on the velocity of vocal fold adductive motion, the present study demonstrated that the 20%–0% average velocity during humming phonation (135 ± 91°/ s) was approximately half of that during natural phonation (305 ± 156°/s) and more than one third of that during loud phonation (493 ± 199°/s), and that the minimal vocal fold angle during humming phonation was greater than that during natural phonation. In addition, during humming phonation, the vocal folds in the majority of the participants directly positioned close to the median line without any prephonatory glottal closure followed by vocal fold vibration. These results indicate that humming exerts a beneficial and hygienic effect on the vocal folds by easing the vocal fold collision, which is considered to be coincident with the previous description regarding the soft attack.11 It is hypothesized that this gradual, smooth, and optimal positioning of the vocal folds during humming phonation is due to humming weakening the activity of the adductor intrinsic laryngeal muscles. In addition, the elimination of prephonatory transient glottal closure is considered to delete the plosive noise and to stabilize the subsequent vocal fold oscillation, leading to an improvement in the voice quality. Physiological mechanisms underlying the different patterns of vocal fold adduction during the three types of vocal onsets Hirose and Gay18 investigated the EMG activities of the intrinsic laryngeal muscles, and reported that “in hard attack, the EMG

ARTICLE IN PRESS 8 activities of the lateral cricoarytenoid muscle (LCA) increase markedly long before (more than 700 msec prior to) the onset of the voicing and stay high during the prephonatory phase, followed by a steep fall immediately before the onset of voicing, whereas in soft attack, the EMG activity of the adductor muscles (including the LCA muscle) increases gradually, reaching a peak after the onset of voicing.” Furthermore, Dailey et al24 speculated that the increased vocal fold velocity during the quick repetition of phonation and inspiration is due to the greater control over the recruitment of motor units in vocal fold adduction than during normal phonation. Accordingly, the differences in the velocity of vocal fold adductive motion in the prephonatory laryngeal closure between the three phonatory tasks observed in the present study are assumed to result from excessive or prolonged EMG activities of the adductor muscles and increased number of recruited motor units. Mechanisms of the therapeutic effects of humming in dysphonic patients The present study demonstrated that humming, producing a resonant voice, immediately decreased the velocity of the vocal fold adduction in the prephonatory adjustment phase of vocal onset, leading to the gradual and smooth repositioning of the vocal folds close to the median line and elimination of prephonatory glottal closure. We previously found that in MTD patients, humming immediately improves the degree of supraglottic compression8 and decreases the perturbation parameters of EGG signals9 during sustained phonation, ie, a steady state of vocal fold oscillation. We speculate that the supraglottic compression and disturbed vocal fold oscillation observed in MTD patients are consequences of prolonged prephonatory laryngeal closure by excessive contraction of the adductor muscles. We hypothesize that in MTD patients, humming normalizes the excessive contraction of the adductor muscles, resulting in decreased velocity of vocal fold adductive motion, shortened or eliminated prephonatory laryngeal closure, and a wider distance between the vocal processes in the phonatory position, leading to the correction of the supraglottic compression and the decrease in the EGG perturbation parameters. The onset of vocal nodules is also known to be associated with hyperfunction, including hard glottal attack.38,39 Regarding the long-term therapeutic effects of resonant voice therapy on vocal nodules, Chen et al40 reported that the performance of resonant voice therapy improved voice quality and elimination of hard attack in female teachers with MTD, vocal nodules, or chronic corditis. Verdolini-Marston et al41 also reported the effectiveness of resonant voice therapy on vocal nodules. Indeed, we demonstrated that humming immediately decreased the EGG perturbation parameters in patients with vocal nodules.10 Accordingly, it is reasonable to consider that in dysphonic patients with vocal nodules, humming corrects the phonatory vocal fold position and glottal configuration during sustained phonation by decelerating the vocal fold adductive motion, eliminating the prephonatory laryngeal closure, and correcting the excessively adducted vocal fold position. Regarding the velocity of the vocal fold adduction of the vocal onset in dysphonic patients associated with vocal hyperfunc-

Journal of Voice, Vol. ■■, No. ■■, 2016

tion, Stepp et al25 have found that MTD patients and the majority of patients with vocal nodules showed lower and higher velocities of the vocal fold adductive motion, respectively, on vocal onset than normal participants, and attributed these velocities to the stiffness and hyperfunction of the intrinsic laryngeal musculature, respectively. Future studies should investigate the measures regarding the prephonatory phenomena in MTD patients and those with vocal nodules. Differences between humming phonation and soft/breathy attack phonation In a study of Cooke et al,20 during breathy onset, the minimal ratio of the distance between the vocal processes normalized to that when the vocal fold angle was 20° was approximately 0.7. In contrast, the present study demonstrated that the minimal vocal fold angles during natural and humming phonation were 1.0 ± 2.0° and 7.1 ± 3.7°, respectively. Accordingly, it appears that the minimal vocal fold angle during breathy onset is greater than that during natural and humming phonation. However, in the present study, we did not compare the phonatory position of the vocal folds between humming and breathy/soft attack phonation, as we could not ask the present series of participants to perform more than the three phonatory tasks of natural, loud, and humming phonation. This was because it takes approximately 20 minutes to record and transfer all of the three data due to the limited memory capacity of the high-speed camera, and because the participants feel discomfort having a rhinolaryngo fiberscope repetitively inserted and removed before and after the recording of each task. In addition, it seems difficult for participants with no vocal training to perform breathy and soft attack phonation to a standardized degree. That is one limitation associated with the present study. Another limitation is the insufficient standardization of the loud phonation task, as each participant was asked to merely “produce a loud voice like a shout” without measurement of the sound pressure level. However, this instruction is considered to have generated a sufficient difference between natural and loud phonation within each participant. In addition, other variations on command, such as “produce a loud voice using the abdomen” and “project a loud voice distantly,” may generate different results, particularly in the pattern of prephonatory transient laryngeal closure. It is an intriguing theme to elucidate how a loud voice is produced correctly using our experimental model. CONCLUSIONS In the present study, the use of HSDI allowed comparison of the movement of the laryngeal structures in the prephonatory adjustment phase of various vocal onsets, including natural, loud, and humming phonation. We determined the differences in the velocity of vocal fold adduction, the patterns of prephonatory transient laryngeal closure, and the phonatory position of the vocal folds among these three phonatory tasks. In particular, the present results demonstrated that humming decreased the velocity of the vocal fold adductive movement, and smoothly and gradually repositioned the vocal folds close to the median line, ie, “barelyadducted or barely-abducted phonatory position” with elimination of prephonatory glottic closure. In contrast, loud phonation was

ARTICLE IN PRESS Toshihiko Iwahashi, et al

Humming Decelerates Prephonatory Vocal Fold Adductive Motions

found to induce a higher vocal fold terminal velocity just before collision, leading to excessive adduction, and perhaps firm glottal closure, sometimes with the ventricular fold closure. Furthermore, these results suggest that the velocity of the vocal fold adductive motion and the presence/absence of subsequent transient laryngeal closure affect the perceptual impression of the produced voice. Future research using EMG of the intrinsic laryngeal muscles and aerodynamic monitoring may elucidate the mechanisms underlying the control of the laryngeal adjustment on vocal onset during vocal training techniques, leading to improvements in the efficacy of voice therapy in patients with functional and organic voice disorders. Acknowledgment This investigation was supported by JSPS KAKENHI (JP26462604). REFERENCES 1. Cooper M. Modern Techniques of Vocal Rehabilitation. Springfield, IL: Thomas; 1973. 2. Colton RH, Casper JK. Understanding Voice Problems: A Physiological Perspective for Diagnosis and Treatment. 2nd ed. New York, NY: Williams & Wilkins; 1996. 3. Harris S. Speech therapy for dysphonia. In: Harris T, Harris S, Rubin JS, et al., eds. The Voice Clinic Handbook. London: Whurr Publishers; 1998:139–206. 4. Yiu EML, Ho EYY. Short-term effect of humming on vocal quality. Asia Pac J Speech Lang Hear. 2002;7:123–137. 5. Yiu EML. Hong Kong humming. In: Behrman A, Haskell J, eds. Exercises for Voice Therapy. San Diego, CA: Plural Publishing; 2008:62–64. 6. Ogawa M, Yoshida M, Watanabe K, et al. Association between laryngeal findings and vocal qualities in muscle tension dysphonia with supraglottic contraction. Nippon Jibiinkoka Gakkai Kaiho. 2005;108:734–741. 7. Hosokawa K, Yoshida M, Yoshii T, et al. Effectiveness of the computed analysis of electroglottographic signals in subjects with muscle tension dysphonia. Folia Phoniatr Logop. 2012;64:145–150. 8. Ogawa M, Hosokawa K, Yoshida M, et al. Immediate effectiveness of humming in the supraglottic compression in subjects with muscle tension dysphonia. Folia Phoniatr Logop. 2013;65:123–128. 9. Ogawa M, Hosokawa K, Yoshida M, et al. Immediate effects of humming on computed electroglottographic parameters in patients with muscle tension dysphonia. J Voice. 2014;28:733–741. 10. Vlot C, Ogawa M, Iwahashi T, et al. Investigation of the immediate effects of humming on vocal fold vibration irregularity using electroglottography and high-speed laryngoscopy in patients with organic voice disorders. J Voice. 2016;30:http://dx.doi.org/10.1016/j.jvoice.2016.03.010. 11. Werner-Kukuk E, von Leden H. Vocal initiation. Folia Phoniatr (Basel). 1970;22:107–116. 12. Wittenberg T, Moser M, Tigges M, et al. Recording processing, and analysis of digital high-speed sequences in glottography. Mach Vis Appl. 1995;8:399– 404. 13. Watson BC, Baken RJ, Roark RM. Effect of voice onset type on vocal attack time. J Voice. 2016;30:11–14. 14. Moore P. Motion picture studies of the vocal folds and vocal attack. J Speech Disord. 1938;3:235–238. 15. Isshiki N, von Leden H. Hoarseness: aerodynamic studies. Arch Otorhinolaryngol. 1964;80:206–213. 16. Koike Y, Hirano M, von Leden H. Vocal initiation: acoustic and aerodynamic investigation of normal subjects. Folia Phoniatr (Basel). 1967;19:173–182.

9

17. Hirano M. Laryngeal adjustment for different vocal onsets. An electromyographic investigation. Nippon Jibiinkoka Gakkai Kaiho. 1974;74:1572–1579. 18. Hirose H, Gay T. Laryngeal control in vocal attack. An electromyographic study. Folia Phoniatr (Basel). 1973;25:203–213. 19. Orlikoff RF, Deliyski DD, Baken RJ, et al. Validation of a glottographic measures of vocal attack. J Voice. 2009;23:164–168. 20. Cooke A, Ludlow CL, Hallet N, et al. Characteristics of vocal fold adduction related to voice onset. J Voice. 1997;11:12–22. 21. Wittenberg T, Tigges M, Mergell P, et al. Functional imaging of vocal fold vibration: digital multislice high-speed kymography. J Voice. 2000;3:422– 442. 22. Freeman E, Woo P, Saxman JH, et al. A comparison of sung and spoken phonation onset gestures using high-speed digital imaging. J Voice. 2012;26:226–238. 23. Kunduk M, Yan Y, McWhorter AJ, et al. Investigation of voice initiation and voice offset characteristics with high-speed digital imaging. Logop Phoniatr Vocol. 2006;31:139–144. 24. Dailey SH, Kobler JB, Hillman RE, et al. Endoscopic measurement of vocal fold movement during adduction and abduction. Laryngoscope. 2005;115:178–183. 25. Stepp CE, Hillman RE, Heaton JT. A virtual trajectory model predicts differences in vocal fold kinematics in individuals with vocal hyperfunction. J Acoust Soc Am. 2010;127:3166–3176. 26. Schuberth S, Hoppe U, Döllinger M, et al. High-precision measurement of the vocal fold length and vibratory amplitude. Laryngoscope. 2002;112: 1043–1049. 27. Iwahashi T, Ogawa M, Hosokawa K, et al. A detailed motion analysis of the angular velocity between the vocal folds during throat clearing using high-speed digital imaging. J Voice. 2016;30:http://dx.doi.org/10.1016/ j.jvoice.2015.11.004. 28. Titze IR. Mechanical stress in phonation. J Voice. 1994;2:99–105. 29. Hess MM, Verdolini K, Bierhals W, et al. Endolaryngeal contact pressures. J Voice. 1998;12:50–67. 30. Morrison MD, Nichol H, Rammage LA. Diagnostic criteria in functional dysphonia. Laryngoscope. 1986;96:1–8. 31. Andrade DF, Heuer R, Hockstein NE, et al. The frequency of hard glottal attacks in patients with muscle tension dysphonia, unilateral benign masses and bilateral benign masses. J Voice. 2016;2:240–246. 32. Isshiki N. Regulatory mechanism of voice intensity variation. J Speech Lang Hear Res. 1964;7:17–29. 33. Tanaka S, Tanabe M. Glottal adjustment for regulating vocal intensity. An experimental study. Acta Otolaryngol. 1986;102:315–324. 34. Rubin HJ, LeCover M, Vernard W. Vocal intensity, subglottic pressure and air flow relationships in singers. Folia Phoniatr (Basel). 1967;19:393–413. 35. Stepp CE, Heaton JT, Stradelman-Cohen TK, et al. Characteristics of phonatory function in singers and nonsingers with vocal fold nodules. J Voice. 2011;25:714–724. 36. Wallen DW. Compensatory speech behavior in individuals with cleft palate: a regulation/control phenomenon? Cleft Palate J. 1986;23:251–260. 37. Verdolini K, Drucker DG, Palmer PM, et al. Laryngeal adduction in resonant voice. J Voice. 1998;12:315–327. 38. Hillman RE, Holmberg EB, Perkell JS, et al. Objective assessment of vocal hyperfunction: an experimental framework and initial results. J Speech Hear Res. 1989;32:373–392. 39. Hillman RE, Gress C, Hargrave J, et al. The efficacy of speech-language pathology intervention: voice disorders. Semin Speech Lang. 1990;11:297– 310. 40. Chen SH, Hsiao TY, Hsiao LC, et al. Outcome of resonant voice therapy for female teacher with voice disorders: perceptual, physiological, acoustic, aerodynamic, and functional movements. J Voice. 2007;21:415–425. 41. Verdolini-Marston K, Burke MK, Lessac A, et al. Preliminary study of two methods of treatment for laryngeal nodules. J Voice. 1995;9:74–85.