ARTICLE IN PRESS

The Impact of Double Source of Vibration Semioccluded

Voice Exercises on Objective and Subjective Outcomes
in Subjects with Voice Complaints
*,†Marco Guzman, †Gabriela Acuña, †Fernanda Pacheco, †Francisca Peralta, †Constanza Romero,
†Cecilia Vergara, and †Camilo Quezada, *†Santiago, Chile

Summary: Purpose. The present study reports the effects of double source of vibration semioccluded vocal tract
exercises (SOVTEs) on subjective and objective variables in subjects with voice complaints.
Methods. Eighty-four participants with voice complaints were randomly assigned to one of four treatment groups:
(1) water resistance therapy, (2) tongue trills, (3) lip trills, and (4) raspberry (tongue and lip trills at the same time).
Before and after voice therapy, participants underwent aerodynamic, electroglottographic, and acoustic assessments.
Measures for the Vocal Tract Discomfort Scale (VTDS), self-assessment of resonant voice quality, and sensation of
muscle relaxation were also obtained. Three assessment sessions were conducted: (1) before the therapy session (Pre),
(2) immediately after the voice therapy session (Post 1), and (3) 1 week after home practice (Post 2).
Results. Significant differences between baseline (Pre) and both post measures were found for the perception of muscle
relaxation and resonant voice quality. No significant differences between Post 1 and Post 2 for any exercises were ob-
served. This indicates that all voice exercises improved subjective self-perceived voice quality immediately after exercises
and that improvement remained stable after 1 week of practice. Water resistance therapy and raspberry attained the
highest effect. A significant decrease for all exercises was also observed for VTDS values after 1 week of practice.
Although some significant changes were observed in objective variables, no clear patterns could be detected.
Conclusions. SOVTEs with secondary source of vibration may reduce vocal symptoms related to physical discom-
fort in subjects with voice complaints. Objective variables apparently do not fully reflect subjective positive outcomes,
or they are not sensitive enough to capture changes. No significant differences between four observed SOVTEs were
observed.
Key Words: Semioccluded vocal tract–Voice therapy–Vocal Tract Discomfort Scale–Resonant voice quality–Muscle
tension.

INTRODUCTION tension, especially in subjects diagnosed with laryngeal and pha-

In the last decades, evidence supporting both the effectiveness
and the underlying physiology of semioccluded vocal tract ex-
ercises (SOVTEs) has accrued. Consistently, the use of this group
of exercises has importantly increased in voice rehabilitation and
training contexts. SOVTEs can be divided into two different
groups, depending on the number of sources of vibration in-
volved during exercise: (1) SOVTEs with a single source of
vibration (vocal folds only, ie, tube phonation with the free end
in the air, bilabial consonant /ß:/, hand-over-mouth technique,
Y-buzz, and humming) and (2) SOVTEs with a secondary source
of vibration added into the vocal tract (water bubbling from the
tube phonation with the free end into water, lip trill, tongue trill,
and raspberry).
Miller and Schutte proposed that a second source of vibra-
tion at the distal part of the vocal tract produces a fluctuating
oral pressure (Poral) impacting on vocal fold vibration.1 The effect
of a secondary source of vibration has also been attributed to a
vibration of vocal tract tissues.2 The fluctuating Poral during this
type of exercises has been reported to cause a “massage-like
effect” in vocal folds and the vocal tract3 with reduction of muscle
Accepted for publication August 21, 2017.
From the *Department of Otolaryngology, Las Condes Clinic, Santiago, Chile; and the
†Department of Communication Sciences, University of Chile, Santiago, Chile.
Address correspondence and reprint requests to Marco Guzman, Avenida Independencia
1027, Santiago, Chile. E-mail: guzmann.marcoa@gmail.com, mguzman@med.uchile.cl
Journal of Voice, Vol. ■■, No. ■■, pp. ■■-■■
0892-1997
© 2017 The Voice Foundation. Published by Elsevier Inc. All rights reserved.
https://doi.org/10.1016/j.jvoice.2017.08.021
ryngeal hyperfunctionality. This massage-like effect could increase
blood flow in vocal folds. Nevertheless, this hypothesis is not
based on data collected within the voice field but on evidence
from sports medicine.4–6
Previous research has established the presence of Poral mod-
ulation during water bubbling at a frequency of 15–40 Hz.7–10
The magnitude of variation of Poral during tube phonation in water
has also been explored. An in vivo study by Radolf et al10 showed
about two to four times higher peak-to-peak variation in Poral with
the tube immersed 10 cm in water, compared with phonation on
[u:]. Similarly, in a modeling study by Horáček et al,7 phona-
tion at conversational loudness resulted in 2.2 times larger peak-
to-peak pressure variation and 1.6 times larger glottal amplitude
variation with the tube submerged 10 cm in water, compared with
phonation on [u:]. These results suggest that the modulation of
Poral water bubbling could have an effect on vocal fold oscilla-
tion and possibly on vocal fold and pharyngeal tissues. Although
no studies have been carried out to explore the fluctuation of
Poral during other SOVTEs with a secondary source of vibra-
tion (eg, tongue and lip trills), they may have a comparable effect
because of tongue and lip vibration.
The effect of a secondary source of vibration on vocal fold
pattern was reported by Andrade et al11 in an electroglottographic
(EGG) study performed with different SOVTEs. Results showed
that SOVTEs with a secondary source of vibration (lip trills,
tongue trills, and water bubbling) presented a significantly greater
range of EGG contact quotient (CQr) and range of fundamental
 ARTICLE IN PRESS
2 Journal of Voice, Vol. ■■, No. ■■, 2017
frequency (F0r) than the baseline condition. A higher CQr is an
indicator of greater variability of open and closed phases of the
vocal fold vibration. A higher CQr is an indicator of greater vari-
ability of open and closed phases of the vocal fold vibration. A
higher F0r indicates greater variability of the vocal fold vibra-
tion frequencies. No changes were observed for the same variables
when phonating a single source of vibration SOVTEs (humming,
hand over mouth, and straw phonation). Authors labeled these
two groups of SOVTEs as “fluctuating” and “steady.”11 They also
suggest that exercises with a secondary source of vibration could
produce a stronger massage effect because of a larger variabil-
ity in CQr caused by changes in Poral. Therefore, a secondary
source of vibration exercises may be better suited for patients
with excessive tension of laryngeal and pharyngeal muscles than
SOVTEs in a “steady” group.
Enflo et al8 showed that the modulation of Poral during water
bubbling affects phonation. They observed that when Poral in-
creases, there is a reduction on the amplitudes of both audio and
dEGG signals. When Poral decreases (the moment of bubble
release) the opposite occurs. A recent high-speed imaging study
by Granqvist et al12 reported that vocal fold vibration and EGG
signal were affected by Poral variation when the tube was sub-
merged in water.
Even though evidence has shown physiological effects of a
double source of vibration SOVTEs on air pressure, audio, EGG,
and glottal variables, to date there are no data supporting the
hypothesis of a massage-like sensation on the vocal folds or vocal
tract produced by different SOVTEs with a secondary source
of vibration. Moreover, no data related to the effects on objec-
tive parameters of this group of exercises has been published.
The purpose of the present study was to observe the possible
effect of a double source of vibration SOVTEs on subjective and
objective variables in subjects with voice complaints. Possible
differential effects produced by these voice exercises were also
inspected.
Based on previous data and clinical reports, we hypoth-
esized that (1) all double sources of vibration SOVTEs should
produce a reduction in the sensation of muscle tension, an in-
crease in the ease of voice production, and a reduction in sensory
discomfort in the throat (as indicators of a possible massage-
like effect); (2) some objective variables (eg, CQEGG, overall sound
pressure level [SPL], glottal airflow rate, subglottic pressure (Psub),
phonation threshold pressure [PTP]) should vary to reflect a more
economic voice production (high acoustic output and low vocal
fold impact stress); and (3) based on clinical reports, some ex-
ercises might generate a greater improvement than others.
METHODS
Participants
Eighty-four participants were initially enrolled in this study. All
participants were randomly assigned to one of four balanced treat-
ment groups (n = 21): (1) water resistance therapy; (2) tongue
trills; (3) lip trills; and (4) raspberry (tongue and lip trills at the
same time). A simple randomization method was used. When
arriving to laboratory, subjects had to blindly choose one of four
cards containing the name of the treatment to be applied in that
particular session. When someone was unable to produce tongue
trills, lip trills, or raspberries even after multiple attempts and
coaching, subjects were asked to blindly choose another exer-
cise. The mean age for the whole sample was 34 years (range
20–44). Inclusion criteria were (1) age within the range of 18–
50 years; (2) self-reported voice complaints (including vocal
fatigue and muscle tension perception); (3) perceptually dys-
phonic voice; and (4) no current or previous voice therapy.
Subjects were recruited by means of flyers created for this study.
Perceptual auditory assessment of voice for recruitment was per-
formed by one voice pathologist with 15 years of experience in
a voice clinic. This procedure was carried out using the param-
eter G (grade of dysphonia) from the GRBAS scale. Only subjects
with at least G = 1 were included in the experiment. Out of 84
participants initially enrolled, only 57 completed the entire ther-
apeutic procedure. When designing the experiment, it was
determined that no placebo treatment could be implemented.
Because of the nature of the study, any given exercises would
have generated some kind of physiological effect and therefore
an impact on subjective variables. With a similar logic, includ-
ing a control group would have really meant comparing the four
treatments against some other type of exercise producing po-
tentially different effects. As the goal of this study was restricted
to inspecting these four particular exercises, the decision was
made not to include any other treatment levels and to only inspect
double sources of vibration exercises.
All participants were native speakers of Spanish. This study
was reviewed and approved by the University of Chile, Faculty
of Medicine Review Board. Informed consent was obtained
from all participants. Assessment and therapy sessions were
carried out in the voice research laboratory at the University of
Chile.
Instrumental assessment
Aerodynamic, EGG, and acoustic signals were captured simul-
taneously during all phonatory tasks. Aerodynamic data were
collected with a Phonatory Aerodynamic System (PAS;
KayPentax, model 4500; KayPENTAX, Lincoln Park, New
Jersey). EGG data were obtained with an electroglottograph
(KayPentax, model 6103; KayPENTAX). Both aerodynamic and
EGG systems were connected to an interface (Computerized
Speech Lab, Model 4500; KayPENTAX), which in turn was con-
nected to a desktop computer running a real-time aerodynamic
and EGG analysis software (KayPENTAX, Model 6600, version
3.4; KayPENTAX). All samples were digitally recorded at a sam-
pling rate of 22.1 KHz with 16 bits/sample quantization.
Calibration of the airflow rate and air pressure was performed
before every recording session according to the manufactur-
er’s instructions.
Acoustic signal was recorded using the incorporated con-
denser microphone AKG CK 77 (AKG Acoustics, Vienna,
Austria) that the PAS provides (KayPENTAX). A constant
microphone-to-mouth distance of 20 cm was used. Acoustic
samples (obtained simultaneously with aerodynamic and EGG
signals) were captured to obtain mean fundamental frequency
(F0) and mean SPL. F0 was obtained to ensure that this vari-
able was kept constant between preassessment and postassessment,
 ARTICLE IN PRESS
Marco Guzman, et al Impact of Double Source of Vibration
as was required of participants. Because changes in this vari-
able may affect outcomes for all dependent variables, F0 was
controlled. The baseline value for each subject was initially re-
corded and then presented before asking subjects to engage in
phonatory tasks in Post 1 and Post 2. SPL was calibrated using
a sustained vowel for further sound-level measurements. The
equivalent level of this reference sound was measured with a
sound-level meter (model 2250; Bruel & Kjær Sound & Vibra-
tion Measurement, Nærum, Denmark) also positioned at a distance
of 20 cm from the mouth.
Participants from both groups were asked to produce the same
three assessment phonatory tasks before and after the voice ex-
ercises: (1) a sustained speaking vowel [a:]; (2) repetition of the
syllable [pa:] (speaking voice quality); and (3) read a phoneti-
cally balanced text for 1 minute using a comfortable loudness
level. Repetition of the syllable [pa:] was performed to esti-
mate the Psub from the Poral during the occlusion of the consonant
[p:] (according to manufacturer recommendations). A silicon tube
inserted into the mouth was used to acquire Poral. Participants
were asked not to touch the tube with the tongue or any other
oral structure so as to not block the airflow. To avoid air leakage
through the nose, a nose clip was used for all participants during
data acquisition. Three repetitions of the two first phonatory tasks
were performed by each subject. F0 was required to be kept the
same during preassessment and postassessment for the sus-
tained vowel and repetition of the syllable [pa:] phonatory tasks.
PTP was also obtained. Participants were asked to produce the
same phonatory task they performed to measure Psub estimated
from Poral. They were required to produce a sequence of six syl-
lables [pa:] at the softest possible voice without reaching a
whisper.
All samples were analyzed with a real-time aerodynamic and
EGG analysis software. A criterion level of 25% from the peak-
to-peak amplitude of the EGG signal was used for CQEGG analysis.
Only the most stable sections from the middle part of the samples
were included in the EGG and aerodynamic analysis. Once the
stable sections were selected, the following variables were
obtained:
• F0 (Hz) from the acoustic signal. To control for gender
differences, this variable was converted to semitones. For
each subject, the initial (pre) Hz value was used as the ref-
erence value to later on convert the post-Hz measures into
semitones. Therefore, this variable shows how subjects in
each group deviate, on average, from their own initial F0,
which for all participants is always zero.
• SPL (dB) from the acoustic signal.
• Mean EGG CQ (%) from the EGG signal.
• Psub (cm H2O).
• PTP (cm H2O) from the aerodynamic signal.
• Glottal airflow (Lt/s) from the aerodynamic signal.
• Glottal resistance (cm H2O/Lt/s).
Audio recordings and acoustical analysis
After aerodynamic and EGG assessments, participants were re-
corded when reading a 242-word phonetically balanced text,
which took approximately 90 seconds. A sustained vowel [a:]
3
was also recorded for every participant. A Focusrite Scarlett 8i-6
USB audio interface (Focusrite Audio Engineering, High
Wycombe, United Kingdom) and a Rode condenser-
omnidirectional microphone, model NT2-A (Rode, Long Beach,
California) were used to capture audio signals. The micro-
phone was positioned 30 cm from participants’ mouths, who
remained standing. Recordings took place in an acoustically
treated room, and samples were recorded digitally at a sam-
pling rate of 44.1 kHz and with 16 bits. The recording of voice
signals was made using the software Protools 9.0 (Avid Cor-
poration, Burbank, California).
Acoustical analysis with the Long-Term Average Spectrum
(LTAS) was performed. The acoustical variables in this study
were (1) the sound level difference between the F1 and F0 regions
(L1–L0), that is, the level of difference between 300–800 Hz and
50–300 Hz; and (2) the alpha ratio, which is the sound level dif-
ference between 50–1,000 Hz and 1,000–5,000 Hz. L1–L0 has
been associated with the degree of vocal fold adduction.
Hypoadducted vocal folds present a strong L0 (sound level of
F0) and low L1 (sound level of F1), whereas a weak L0 and strong
L1 are present in voices with higher vocal fold adductions. The
alpha ratio is a measure that represents the overall spectral slope.
This spectral variable has been found to depend on phonation
type (degree of vocal fold adduction) and to be higher in hy-
perfunctional voices. The LTAS spectra for each subject were
obtained by the Praat software, version 5.3.60 (Institute of Pho-
netic Sciences of the University of Amsterdam, Amsterdam, The
Netherlands). For each sample, a bandwidth of 100 Hz and the
Hanning window were used. Before performing an LTAS anal-
ysis, unvoiced sounds and pauses were eliminated from the
samples by the Praat software using the pitch-corrected LTAS
version with standard settings.
Questionnaire application
All participants were asked to complete the Spanish adaptation
and validation of the Vocal Tract Discomfort Scale (VTDS). The
VTDS is a self-administrated questionnaire designed to measure
the subjective perception of sensory discomfort in the throat (vocal
tract). The assessment is based on the frequency of occurrence
and severity manifestation of eight qualitatively different sen-
sations: burning, tightness, dryness, aching, tickling, soreness,
irritability, and lump in the throat. The frequency and severity
are rated separately on a seven-point Likert scale ranging from
0 to 6 for frequency (0 = never, 2 = sometimes, 4 = often,
6 = always) and for severity (0 = none, 2 = mild, 4 = moderate,
6 = extreme).13
Self-assessment of voice quality
Before recording aerodynamic, EGG, and acoustics param-
eters, all participants were required to self-assess their resonant
voice quality and perception of muscle relaxation in the throat
area. Self-assessment was performed using two different 100-
mm visual analog scales: (1) resonant voice quality, defined as
a voice that feels easy and with the sensation of vibration on
the front part of the face and mouth (0 = not resonant at all,
100 = very resonant) and (2) perception of muscle relaxation
(0 = very tense, 100 = very relaxed).
 ARTICLE IN PRESS
4 Journal of Voice, Vol. ■■, No. ■■, 2017
Voice therapy procedures
One single 30-minute voice therapy session was conducted
with each participant. Therapy sessions were administrated by
four trained speech-language pathologists. This therapy session
was performed in between the first two assessments (Pre and
Post 1). To standardize therapeutic performance, all clinicians
participated in a 10-hour training period (conducted by the
first author of the present study) before administering therapy.
This training period included aspects related to sensorimotor
learning principles applied to voice rehabilitation and the use
of SOVTEs.
For the four different groups (tube phonation in water, tongue
trills, lip trills, and raspberry), exercises consisted of a se-
quence of three phonatory tasks: (1) sustained vowel-like sound;
(2) ascending and descending glissandos throughout a comfort-
able vocal range; and (3) messa di voce. These phonatory tasks
were sequentially included in the therapy session. Participants
were asked to feel the ease of phonation during all voice exer-
cises. Before and during practice, clinicians provided individual
demonstrations and verbal descriptions of each phonatory task.
Water resistance therapy exercises were performed using a com-
mercial plastic drinking straw (5 mm in inner diameter and
25.8 cm in length) submerged 5 cm into water.
All participants were instructed to practice the assigned voice
exercise daily during 1 week. They were required to complete
(six to eight times a day and during 5–10 minutes each time)
the same exercise they were assigned on the first session (tube
phonation in water, tongue trills, lip trills, or raspberry). At the
end of each session, subjects were given detailed written in-
structions for their home exercise programs. Instructions included
all phonatory tasks learned during the session. To monitor patient
compliance, a WhatsApp (WhatsApp Inc. Menlo Park, CA)
message was sent daily to each participant. However, data on
compliance across participants are not presented in the present
study.
Posttherapy assessment
Two posttherapy assessment sessions were performed: (1) im-
mediately after the voice therapy session was completed (Post
1) and (2) 1 week after home practice (Post 2). Each partici-
pant underwent the same assessment procedure performed for
the pretherapy assessment. This procedure included aerody-
namic, EGG, acoustic, and self-assessment of voice.
Statistical analysis
Data were analyzed and plotted with the R software (R Foun-
dation for Statistical Computing, Viena, Austria).14,15 Results
are presented for each one of the three implemented tasks, for
audio signal, for visual analog scale scores, and for VTDS
scores. A 4 × 3 mixed-factor analysis of variance (ANOVA)
was conducted on all tasks’ variables, with treatment as a
four-level between-group factor and Pre and Post measures as
a three-level within-group factor. Generalized eta squared (GES)
is reported as an effect size measure. For repeated measures,
when Geisser-Greenhouse correction’s e > 0.75, reported P values
consider the Huynn-Feldt correction.16 Because of differences
observed for baseline levels (Pre) between treatments, no
between-group comparisons (either post hoc or planned) were
conducted. Because the design considered more than two re-
peated measures, reported post hoc contrasts were conducted
by fitting data for each level of treatment on each dependent
variable into a mixed-effects model (maximum likelihood
method). No between-group factor was entered in these models,
and repeated measures were fitted as a random effect nested
within subjects. Pairwise contrasts were obtained adjusting P
values by means of the Tukey method. Only P values below
the standard 0.05 alpha level are reported. Finally, because
symptoms captured by the VTDS cannot be assessed right
after the treatment (measure Pre 1), a 4 × 2 mixed-factor ANOVA
was conducted for this variable only considering two within-
group levels: Pre and Post 2.
RESULTS
Sustained speaking vowel [a:] task’s variables
Figure 1 plots means for sustained speaking vowel [a:] task’s
variables. Significant ANOVA test results were found only for
the interaction term on glottal airflow (F = 251, P = 0.028,
GES = 0.03). The interaction’s significance is driven by the con-
trast Pre against Post 2 on the raspberry exercise (b = −0.03,
z = −2.68, P = 0.02) and Post 1 against Post 2 on tongue trills
(b = 0.04, z = 3.19, P = 0.003). As for the remaining variables,
only on the CQ was there another significant paired contrast,
namely Pre against Post 1 on tongue trills (b = 3.72, z = 2.92,
P = 0.009). F0 is reported only to show that values remained stable
across measures in all conditions.
Repetition of the syllable [pa:] task’s variables
ANOVA test results on variables from the repetition of the syl-
lable [pa:] task were only significant for the within-group
measure’s main effect on glottal airflow (F = 3.17, P = 0.049,
GES = 0.01). Figure 2 plots these results. This main effect was
inspected by means of post hoc pairwise contrasts on col-
lapsed within measures, which showed a significant difference
between Pre and Post 1 (b = −0.015, z = −2.36, P = 0.047). Post
hoc pairwise contrasts were conducted also on each treatment
level, the only significant contrasts being Post 1 against Post 2
measures on lip trills (b = 0.027, z = 2.38, P = 0.045).
Phonetically balanced text task’s variables
As for the phonetically balanced text task, no significant results
for glottal airflow were observed. Text reading and sustained
vowel [a:] tasks were inspected on two variables (alpha ratio and
L1–L0). No significant results were found.
Self-assessment of voice quality
ANOVA results for the self-assessment of voice quality are pre-
sented in Table 1 and plotted in Figure 3. Highly significant main
effects for within-subject measures were obtained for both de-
pendent variables. Pairwise contrasts on collapsed treatments
showed significant differences between Pre and both Post mea-
sures. No significant differences were found between Post 1 and
Post 2. Because statistical differences between collapsed
 ARTICLE IN PRESS
Marco Guzman, et al Impact of Double Source of Vibration 5

FIGURE 1. Means plot for objective variables from sustained speaking vowel [a:] task. Error bars represent one unit of standard error.

FIGURE 2. Means plot for objective variables from syllable repetition [pa:]. Error bars represent one unit of standard error.
 ARTICLE IN PRESS
6 Journal of Voice, Vol. ■■, No. ■■, 2017

TABLE 1.
Mixed-factor ANOVA Test Results for Resonant Voice Quality and the Perception of Muscle Relaxation
Pre/Post 1/Post 2 Treatment Interaction
F P GES F P GES F P GES
Resonant voice quality 44.6 <0.001 0.22 2.16 0.10 0.07 0.45 0.8 0.008
Perception of muscle relaxation 47.28 <0.001 0.26 0.67 0.5 0.02 0.45 0.87 0.008
GES provided as Effect Size Measure.

FIGURE 3. Means plot for subjective variables from self-assessment of voice quality. Error bars represent one unit of standard error.

TABLE 2.
Tukey Post Hoc Paired Contrasts for Resonant Voice Quality
Pre/Post 1 Pre/Post 2 Post1/Post 2 GES
b z P b z P b z P d
Raspberry −14.22 −4.43 <0.001 −17.53 −5.47 <0.001 3.3 1 0.5 1.67
Water resistance −20.07 −4.48 <0.001 −23.57 −5.22 <0.001 −3.5 0.78 0.71 1.21
Lip trills −20.42 −3.69 <0.001 −19.85 −3.53 0.001 0.57 0.1 1 1.08
Tongue trills −14.42 −3 0.005 −22.07 4.74 <0.001 7.6 1.64 0.22 0.9
Cohen’s d is based on unadjusted P values of one-tailed paired t tests comparing Pre and Post 1 measures.

TABLE 3.
Tukey Post Hoc Paired Contrasts for the Perception of Muscle Relaxation
Pre/Post 1 Pre/Post 2 Post1/Post 2 GES
b z P b z P b z P d
Raspberry −23.67 −4.71 <0.001 −24.53 −4.97 <0.001 −1.26 −0.25 0.96 1.2
Water resistance −29.07 −4.73 <0.001 −26.14 −4.25 <0.001 2.92 0.47 0.88 1.5
Lip trills −23.35 −4.24 <0.001 −22.92 −4.16 0.001 0.42 0.08 1 1.15
Tongue trills −17 −3.48 0.001 −21.42 −4.38 <0.001 −4.42 0.07 0.63 0.98
Cohen’s d is based on unadjusted P values of one-tailed paired t tests comparing Pre and Post 1 measures.

repeated measures followed exactly the same pattern when in- interpretation and to gauge the improvement attained by each
specting single exercises, we only report disaggregated effects treatment, effect sizes for one-tailed, paired t tests for all treat-
(Table 2 and Table 3 summarize Tukey post hoc contrasts ments are also reported based on unadjusted P values for contrasts
for all treatments on repeated measures). To ease the results’ between Pre and Post 1 measures.
 ARTICLE IN PRESS
Marco Guzman, et al Impact of Double Source of Vibration 7

TABLE 4.
Mixed-Factor ANOVA Test Results for Vocal Tract Discomfort Scale Scores
Pre/Post2 Treatment Interaction
F P GES F P GES F P GES
VTDS 93.85 <0.001 0.27 1.22 0.31 0.05 0.67 0.57 0.008
GES provided as the effect size measure.

FIGURE 4. Means plot for the Vocal Tract Discomfort Scale (VTDS). Error bars represent one unit of standard error.

than effects for objective variables. Although some significant

TABLE 5.
Tukey Post Hoc Paired Contrasts for Vocal Tract Discom-
changes were observed in objective variables, there were no
fort Scale Scores systematic effects that could be associated with changes in
subjective parameters. Our data constitute the first empirical
Pre/Post2 GES attempt to provide information about the possible massage-
b p d like effect attributed to SOVTEs with a secondary source of
Raspberry 4.08 <0.001 1.31 vibration.
Water resistance 4.1 0.04 1.36 Positive outcomes were obtained for both subjective self-
Lip trills 6.21 <0.001 1.97 assessment scales (resonant voice quality and perception of muscle
Tongue trills 6.86 <0.001 2.17 relaxation). Results showed significant differences between base-
Cohen’s d is based on P values of one-tailed paired t tests comparing Pre line (Pre) and both post measures, but there were no significant
and Post 2 measures. differences between Post 1 and Post 2. This indicates that all
voice exercises produced a higher perception of muscle relax-
ation and a more resonant voice quality immediately after
Finally, a 4 × 2 mixed-factor ANOVA was conducted on the exercising, with change remaining stable after 1 week of prac-
VTDS scores. Results are presented and plotted in Table 4 and tice. Effect sizes for the level of change produced by all exercises
Figure 4. A significant main effect for pre/post measure was ob- were large (d > 0.8). Thus, all exercises produced a significant
tained. Results for one-tailed, paired t tests are reported in Table 5, improvement in voice self-perception, water resistance therapy,
along with Cohen’s d for effect size. and raspberry, generating the highest effects.
Because resonant voice quality improved after all voice ex-
DISCUSSION ercises, it can be said that subjects felt an easier voice production
The present study assessed the effects of a double source of and more vibratory sensations in the front part of their faces.
vibration SOVTEs on subjective and objective variables in Paes et al17 studied the immediate effects of phonation into a
subjects with voice complaints. Results seem to support the traditional Finnish glass tube submerged into water among teach-
role of this group of voice exercises as a potentially effective ers diagnosed with behavioral dysphonia. Authors found that
tool for reducing symptoms of vocal tract discomfort (eg, sen- most of the subjects (68%) experienced an increased phona-
sation of muscle tension in laryngeal and pharyngeal areas). tory comfort (easy voice production) after practicing. Data from
Observed effects for subjective variables were more manifest a recent study aimed to assess the effect of vocal warm-up using
 ARTICLE IN PRESS
8 Journal of Voice, Vol. ■■, No. ■■, 2017
a commercial plastic stirring straw (with the free end in the air)
also demonstrated a significant increase in self-perceived reso-
nant voice quality immediately after the exercise.18 Based on
results from the present study and previous research, it seems
to be the case that different types of SOVTEs promote reso-
nant voice production. Further studies could be conducted to
explore the possible differential effect on this subjective param-
eter when comparing SOVTEs with one and two sources of
vibration.
Our results showed a significant decrease for VTDS values
when comparing baseline measures against measures obtained
1 week into practice. This clinically positive result was the same
for all voice exercises. This implies that physical throat symp-
toms associated with voice disorders can be effectively reduced
by SOVTEs with a secondary source of vibration. An earlier study
by Mailänder et al19 explored the effectiveness of a 3-week train-
ing program with water resistance therapy for teachers. Authors
found a significant improvement in the physical subscale of the
Voice Handicap Index (VHI). Furthermore, comparable out-
comes in VHI physical subscales were found by Guzman et al20
in subjects diagnosed with behavioral dysphonia after eight ses-
sions with water resistance therapy. Additionally, several earlier
studies using physiological programs for voice rehabilitation
(based on different semiocclusions with a single source of vi-
bration) have shown possible results in the VHI.21–26 Because the
VHI physical subscale is related to the degree of physical vocal
discomfort (eg, “I use a great deal of effort to speak,” “I feel as
though I have to strain to produce my voice,” “my voice sounds
creaky and dry,” “my voice ‘gives out’ on me in the middle of
speaking”), this subscale can be considered as akin to the VTDS.
In fact, a strong correlation between the VTDS and the VHI has
been reported in literature.27
The reduction of perceived effort and other voice symptoms
(eg, VTDS, VHI, perception of muscle relaxation, and percep-
tion of resonant voice quality) should relate with some objective
variables. Specifically, the PTP should be expected to decrease
during SOVTEs because of increased vocal tract inertance.28–30
To the best of our knowledge, only two investigations have
used PTP as an objective outcome to observe changes after a
physiological voice therapy program (based on SOVTEs).20,25
Chen et al25 reported a significant reduction of PTP after an
8-week period of voice treatment with resonant voice therapy
in teachers with voice disorders. Guzman et al20 found similar
results after an 8-week period of voice treatment with water
resistance therapy. Even though all double sources of vibration
exercises tended to decrease PTP in the present study, no
significant reduction on this variable was found either immedi-
ately after voice exercises or 1 week into the treatment. This
lack of change could be explained by the short therapy period
implemented in our study (1 week), if compared with the
8-week period of voice treatment implemented in these two
related studies.
Another objective variable included in the present study that
may be associated with observed subjective changes is the EGG
CQ. Although three exercises (lip trills, tongue trills, and rasp-
berries) generated a decrease immediately after practicing, a
significant change was observed only for tongue trills. Previous
works have also reported changes on the EGG CQ when using
tongue trills. A recent study designed to compare the effect on
the EGG CQ of different SOVTEs showed that tongue trills
tended to decrease the CQ during exercises compared with vowel
phonation.31 Andrade et al11 found that tongue trills produced
the lowest values of the CQ when compared with the other
semioccluded postures. Gaskill and Erickson32 also found a de-
crease in the CQ during phonation with lip trills. Because the
EGG CQ is positively correlated with the degree of vocal fold
adduction,33 one may expect an increase on glottal airflow during
exercising when Psub is kept constant. A trend of increasing glottal
airflow (during both the repetition of the syllable [pa:] and the
sustained vowel [a:]) was observed during tongue trills in the
present study. However, this change did not reach statistical
significance.
CONCLUSION
SOVTEs with a secondary source of vibration may reduce
symptoms related to throat physical discomfort in subjects
with voice complaints. They may also contribute to an im-
proved self-perceived resonant voice quality and a reduction in
the sensation of muscle tension. These changes could be asso-
ciated with the so-called massage-like effect attributed to this
type of SOVTE. Water resistance therapy and raspberry seem
to produce the greatest effects. Changes occur immediately
after exercises and remain stable after 1 week of practice.
Objective variables apparently do not reflect subjective posi-
tive outcomes, or they are not sensitive enough to capture
effects after short periods of practice.
REFERENCES
1. Miller DG, Schutte HK. Effects of downstream occlusions on pressures near
the glottis in singing. In: Gauffin J, Hammarberg B, eds. Vocal Fold
Physiology. Acoustic, Perceptual and Physiological Aspects of Voice
Mechanisms. San Diego, CA: Singular; 1991:91–98.
2. Schwarz K, Cielo CA. Vocal and laryngeal modifications produced by the
sonorous tongue vibration technique. Pro Fono. 2009;21:161–166.
3. Radolf V, Laukkanen A-M, Horáček J, et al. In vivo measurements of air
pressure, vocal folds vibration and acoustic characteristics of phonation into
a straw and a resonance tube used in vocal exercising. In: Proceedings of
the 19th International Conference Engineering Mechanics, Czech Republic.
2013:478–483.
4. Mori H, Ohsawa H, Tanaka TH, et al. Effect of massage on blood flow and
muscle fatigue following isometric lumbar exercise. Med Sci Monit.
2004;10:CR173–CR178.
5. Hinds T, McEwan I, Perkes J, et al. Effects of massage on limb and skin
blood flow after quadriceps exercise. Med Sci Sports Exerc. 2014;36:1308–
1313.
6. Weerapong P, Hume PA, Kolt GS. The mechanisms of massage and effects
on performance, muscle recovery and injury prevention. Sports Med.
2005;35:235–256.
7. Horáček J, Radolf V, Bula V, et al. Air-pressure, vocal folds vibration and
acoustic characteristics of phonation during vocal exercising. Part 2:
measurement on a physical model. Eng Mech. 2014;21:193–200.
8. Enflo L, Sundberg J, Romedahl C, et al. Effects on vocal fold collision and
phonation threshold pressure of resonance tube phonation with tube end in
water. J Speech Lang Hear Res. 2013;56:1530–1538.
9. Guzman M, Castro C, Testart A, et al. Laryngeal and pharyngeal activity
during semioccluded vocal tract postures in subjects diagnosed with
hyperfunctional dysphonia. J Voice. 2013;27:709–716.
 ARTICLE IN PRESS
Marco Guzman, et al Impact of Double Source of Vibration
10. Radolf V, Laukkanen A-M, Horáček J, et al. Air-pressure, vocal fold
vibration and acoustic characteristics of phonation during vocal exercising.
Part 1: measurement in vivo. Eng Mech. 2014;21:53–59.
11. Andrade PA, Wood G, Ratcliffe P, et al. Electroglottographic study of seven
semi-occluded exercises: LaxVox, straw, lip-trill, tongue-trill, humming,
handover-mouth, and tongue-trill combined with hand-over-mouth. J Voice.
2014;28:589–595.
12. Granqvist S, Simberg S, Hertegård S, et al. Resonance tube phonation in
water: high-speed imaging, electroglottographic and oral pressure
observations of vocal fold vibrations—a pilot study. Logoped Phoniatr Vocol.
2014;28:1–9.
13. Mathieson L, Hirani SP, Epstein R, et al. Laryngeal manual therapy:
a preliminary study to examine its treatment effects in the management of
muscle tension dysphonia. J Voice. 2009;23:353–366.
14. R Core Team. R: A Language and Environment for Statistical Computing.
Vienna, Austria: R Foundation for Statistical Computing; 2015 Available
at: https://www.R-project.org/.
15. Wickham H. ggplot2: Elegant Graphics for Data Analysis. New York, NY:
Springer-Verlag; 2009.
16. Girden ER. ANOVA: Repeated Measures. Newbury Park, CA: Sage; 1992.
17. Paes SM, Zambon F, Yamasaki R, et al. Immediate effects of the Finnish
resonance tube method on behavioral dysphonia. J Voice. 2013;27:717–722.
18. Portillo P, Rojas S, Guzmán M, et al. Comparison of effects produced by
physiologic versus traditional vocal warm-up in contemporary commercial
music singers. J Voice. 2017;doi:10.1016/j.jvoice.2017.03.022. In press.
19. Mailänder E, Mühre L, Barsties B. Lax vox as a voice training program
for teachers: a pilot study. J Voice. 2017;31:262.e13–262.e22.
20. Guzman M, Jara R, Olavarria C, et al. Efficacy of water resistance therapy
in subjects diagnosed with behavioral dysphonia: a randomized-controlled
trial. J Voice. 2017;31:385.e1–385.e10. In press.
21. Roy N, Gray S, Simon M, et al. An evaluation of the effects of two treatment
approaches for teachers with voice disorders: a prospective randomized
clinical trial. J Speech Lang Hear Res. 2001;44:286–296.
9
22. Sauder C, Roy N, Tanner K, et al. Vocal function exercises for presbylaryngis:
a multidimensional assessment of treatment outcomes. Ann Otol Rhinol
Laryngol. 2010;119:460–467.
23. Kaneko M, Hirano S, Tateya I, et al. Multidimensional analysis on the effect
of vocal function exercises on aged vocal fold atrophy. J Voice.
2015;29:638–644.
24. Roy N, Weinrich B, Gray SD, et al. Three treatments for teachers with voice
disorders: a randomized clinical trial. J Speech Lang Hear Res.
2003;46:670–688.
25. Chen SH, Hsiao TY, Hsiao LC, et al. Outcome of resonant voice therapy
for female teachers with voice disorders: perceptual, physiological, acoustic,
aerodynamic, and functional measurements. J Voice. 2007;21:415–425.
26. Kapsner-Smith MR, Hunter EJ, Kirkham K, et al. A randomized controlled
trial of two semi-occluded vocal tract voice therapy protocols. J Speech Lang
Hear Res. 2015;58:535–549.
27. Niebudek-Bogusz E, Woznicka E, Wiktorowicz J, et al. Applicability of the
Polish Vocal Tract Discomfort (VTD) scale in the diagnostics of occupational
dysphonia. Logoped Phoniatr Vocol. 2012;37:151–157.
28. Titze I, Story B. Acoustic interactions of the voice source with the lower
vocal tract. J Acoust Soc Am. 1997;101:2234–2243.
29. Titze I. The physics of small-amplitude oscillation of the vocal folds. J Acoust
Soc Am. 1988;83:1536–1552.
30. Conroy ER, Hennick TM, Awan SN, et al. Effect of variations to a simulated
system of straw phonation therapy on aerodynamic parameters using excised
canine larynges. J Voice. 2014;28:1–6.
31. Guzman M, Calvache C, Romero L, et al. Do different semi-occluded voice
exercises affect vocal fold adduction differently in subjects diagnosed with
hyperfunctional dysphonia? Folia Phoniatr Logop. 2015;67:68–75.
32. Gaskill C, Erickson M. The effect of a voiced lip trill on estimated glottal
closed quotient. J Voice. 2008;22:634–643.
33. Verdolini K, Chan R, Titze I, et al. Correspondence of electroglottographic
closed quotient to vocal fold impact stress in excised canine larynges. J Voice.
1998;12:415–423.