Vocal Effort and Acoustic Analysis of Gargle Phonation

Versus Water Swallow in Patients With Muscle Tension

Dysphonia: A Clinical Trial
⁎ ⁎ ⁎ ⁎
Diana M Orbelo, †Sara A Charney, Elisabeth Renkert, ‡Mary Pietrowicz, David Aka, Semirra L Bayan, and
§ ⁎
Keiko Ishikawa, Rochester, Minnesota, †Phoenix, Arizona, ‡Urbana-Champaign, Illinois, and §Lexington, Kentucky

Summary: Purpose. To determine the effects of gargle phonation (GP) on self-perceived vocal improvement,
vocal effort, acoustic parameters, and speech rate in patients with muscle tension dysphonia (MTD). We hy­
pothesized that GP would improve voice, reduce phonatory effort, and alter acoustic and speech measures.
Study design. Prospective randomized, single-blind cross-over clinical trial
Methods. Thirty-four participants (26 females, 8 males; average age 53 years) who were diagnosed with MTD
completed the Voice Handicap Index-10 (VHI-10) and were assigned three study conditions: Baseline (B), GP,
and Water Swallow (WS; sham), presented in one of two counterbalanced orders B-WS-GP (WS1st) or B-GP-
WS (GP1st). Participants recorded stimuli from the Consensus Auditory-Perceptual Evaluation of Voice
(CAPE-V) and rated their perceived vocal effort and vocal improvement. F0, vocal intensity, cepstral peak
prominence (CPP), and speaking rate were measured.
Results. Average VHI-10 scores by group were 16 (min/max 2–29) for WS1st and 15 (min/max 3–40) for GP1st.
About 73.5% reported more vocal improvement after GP, 17.65% after WS, and 8.8% noted no difference
between conditions. Reduced effort was reported after GP, compared to B (P < 0.001) and WS (P = 0.005).
Lower effort was also reported after the WS condition, compared to B (P = 0.011). Key acoustic findings
included an increase in F0 after GP for sustained /i/ for females. CPP was significantly higher for females
reading CAPE-V sentences after GP, when GP preceded WS, compared to B (P = 0.004) and WS (P = 0.003).
Speech rate was faster for females after GP versus B (P = 0.029).
Conclusions. GP may be beneficial in the treatment of MTD. CPP may be a useful marker for vocal im­
provement after GP for women with mild MTD. Further studies would benefit from having more male par­
ticipants and those with moderate and severe MTD.
Key Words: Gargle phonation–Muscle tension dysphonia–Voice therapy.

INTRODUCTION norms. Acoustically, MTD has been associated with ab­

“Muscle tension dysphonia” (MTD) is a functional voice normal fundamental frequency (F0), vocal intensity, and
disorder often associated with excessive muscle tension in cepstral peak prominence (CPP).10–12
the laryngeal and peri-laryngeal muscles.1,2 It is a common Though MTD is referred to as a voice-specific disorder,
voice disorder seen in voice centers, accounting for up to it may also have consequences that extend beyond the
30% of a voice therapist’s caseload.3 Symptoms of MTD larynx more generally to speech production. The laryngeal
include voice quality changes, vocal fatigue, pain, com­ and perilaryngeal muscles are anatomically connected to
munication difficulties,4–6 and self-perception of increased the muscles involved in the articulation of speech sounds.
vocal effort.7–9 A strained voice quality, excessive effort, This mechanical coupling could restrict articulatory
and abnormal pitch and loudness have also been reported. movements, affecting overall speech output.13–15 In sup­
For instance, Altman et al.’s retrospective study of 150 port of this notion, studies have shown articulatory
MTD patients reported that 82% exhibited inappropriate changes after manual circumlaryngeal treatment, an es­
habitual pitch—79% with an elevated habitual pitch and tablished therapy approach for MTD.13,16 Thus, a treat­
3% with a lowered habitual pitch for their age and sex. ment effect may manifest in measures associated not only
Additionally, 23% had inappropriate vocal loudness, with with voice production but also with speech production,17–20
14% speaking too softly and 9% too loudly compared to such as speaking rate, which has been used to describe
motor speech disorders21 and speech production styles.22
Voice therapy is considered the gold standard treatment
Accepted for publication February 20, 2024.
⁎
From the Mayo Clinic Department of Otolaryngology, Rochester, Minnesota; for MTD.5 The goals of voice therapy are to improve
†Mayo Clinic Department of Otolaryngology, Phoenix, Arizona; ‡National Center phonatory efficiency, tonal clarity, and reduce perceived
for Supercomputing Applications, Carle Illinois College of Medicine, University of
Illinois at Urbana-Champaign, Urbana-Champaign, Illinois; and the §University of vocal effort.5,23 These are achieved by balancing the re­
Kentucky, Department of Communication Sciences and Disorders, Lexington, spiratory, phonatory, and resonatory subsystems, while
Kentucky.
Address correspondence and reprint requests to Diana Orbelo, 200 1ST ST SW, calibrating and reducing unnecessary activation of intrinsic
Rochester, MN 55901. E-mail: [email protected] and extrinsic laryngeal muscles.5,24,25 One therapeutic ap­
Journal of Voice, Vol xx, No xx, pp. xxx–xxx
0892-1997 proach for MTD is gargle phonation (GP). GP, or voca­
© 2024 The Voice Foundation. Published by Elsevier Inc. All rights reserved. lizing while gargling water, is an activity used in many
https://doi.org/10.1016/j.jvoice.2024.02.018
2 Journal of Voice, Vol. xx, No. xx, xxxx

singing studios and voice clinics.26 It is a non-speech task METHODS

that, for those with intact musculoskeletal function, is The present IRB approved (Mayo Clinic IRB #20-004267)
quick to learn and execute. Additionally, it is inexpensive, and registered clinical trial (NCT04766658) consisted of a
as it requires only water to perform. Anecdotally, it has randomly assigned, alternating order, cross-over design.
been asserted that GP may balance muscle effort26; thus, it All participants were tested under three conditions (B, WS,
may reduce maladaptive muscle patterns. and GP). Participants were assigned to one of two condi­
Despite its popularity, the literature on GP is limited. tion orders: B, WS, then GP (WS1st) or B GP, then WS
Albuquerque et al. examined the immediate effects of GP (GP1st). The WS condition was selected as a sham proce­
with individuals with and without vocal complaints.27 The dure, because the only difference between the WS condition
participants were asked to gargle for 3–5 seconds, re­ and the GP condition is the absence of phonation while
peatedly for 1 minute, holding water in the mouth and creating the water turbulence of gargling.
assumedly breathing through their nose between gargle Still water was kept at ambient room temperature in a
episodes. They were to produce the vowel /u/ as they gar­ climate-controlled clinical setting. Medicine cups were used
gled, after which, they could swallow or spit out the water. to premeasure 5 cc of water for both the WS and GP
The electrical activity of the muscles was measured before conditions. For the WS condition, participants were asked
and after GP with surface electromyography during the to hold 5 cc of water in their mouth for approximately
following activities: counting from 1–10 and sustained /Ɛ/ 5 seconds and then swallow, which was performed twice.
at a habitual vocal intensity, elevated vocal intensity, and The 5 seconds of holding the water was intended to ap­
with glissando. The authors reported decreased electrical proximate the duration that a person would gargle. For
activity of the muscles following GP and most participants GP, participants were asked to gargle 5 cc of water while
in this study reported improvement in their voice quality sustaining a neutral vowel (ie, /Ə/) at a comfortable pitch
and increased phonatory comfort after GP. Similar effects for 3–5 seconds before swallowing the water. GP was re­
of GP were observed by Amorim et al, whose participants peated using an ascending and descending pitch glide, also
reported a “more open voice,” a “clearer voice,” and found on /Ə/.
it easier to speak after GP.28 Another study by Amorim Prior to GP, participants were allowed a timed learning
et al found that GP increased F0 and vocal intensity.29 session of up to an arbitrary time of 10 minutes. This time limit
Together, these results provide emerging evidence that GP was set for the purposes of reproducibility and to have a
positively reduces excessive tension in extrinsic laryngeal stopping point in the unexpected event that someone was un­
muscles during phonation. able to adequately learn to gargle in a reasonable amount of
Despite preliminary evidence showing favorable results time. Participants did not phonate, gargle, or swallow con­
of GP, the studies providing this evidence are limited by tinuously during the learning session; therefore, a fatigue factor
their lack of randomization, absence of pre and postvocal was not anticipated. During the learning session, the clinician
effort measures, and the non-use of sham procedures to modeled the pitch glide, usually without water. If the partici­
investigate whether observed effects were due to phonation pant needed further instruction, the clinician would model GP
with gargling or swallowing the water. Furthermore, these with a pitch glide (with or without water) once more. Subjects
studies failed to account for sex as a variable when ana­ were instructed to tilt their heads back for gargling just far
lyzing F0 (fundamental frequency) and vocal intensity, enough not to lose water from their mouths. Based on clinical
which could significantly influence the outcomes. Thus, the practice experience, patients typically learn gargling quickly,
present study was designed to address these limitations to but some require additional modeled examples or cues to in­
enhance our understanding of the effect of GP on MTD crease airflow. Participants were always encouraged to swallow
through a prospective randomized, single-blind cross-over if they felt at risk for choking on the water. The learning ses­
clinical trial with an incorporated sham procedure. The sion was complete when the participant and clinician agreed
following hypotheses were tested: 1) participants will per­ gargling had been adequately mastered.
ceive greater vocal improvement following GP compared to After practice was complete, participants completed two
baseline (B) and a sham intervention using water swallows productions of sustained GP with sustained phonation
(WS), 2) perceived vocal effort will be reduced following (comfortable pitch) and two productions of GP with a
GP compared to other conditions, 3) speaking F0, vocal pitch glide from low-to-high-to-low, both for approxi­
intensity (dB SPL), CPP, and speech rate will change fol­ mately 3–5 seconds. Immediately following each treatment
lowing GP.30–32 In terms of the directionality of the hy­ condition, participants returned their head to a neutral
potheses, we did not set a prior direction regarding F0 or position and were asked to sustain vowels /a/ and /i/ and
vocal intensity based on the previous report, which in­ read the Consensus Auditory-Perceptual Evaluation of
dicated that vocal pitch and loudness could be abnormally Voice (CAPE-V) sentences aloud.33 These speech samples
high or low in these participants.12 On the other hand, we were recorded in a typical clinical office setting, using
hypothesized that if GP reduces degree of dysphonia, then TASCAM-DR-40X with AKG C555L headset micro­
CPP would increase following GP. Lastly, we hypothesized phone. The microphone was placed 5 cm from the corner of
that speech rate would increase after the treatment. the participant’s mouth at a 45-degree angle. The sampling
Diana M Orbelo, et al Gargle Phonation for Mild MTD 3

measure across the six sentences. CPP was obtained using

PRAAT plug-in script by Murray et al.37 Speech rate was
obtained for each sentence with a PRAAT plug-in script by
De Jong and Wempe.38

Statistical analyses
For effort ratings, a repeated measures two-way
Analysis of Variance (ANOVA) and pairwise t tests
were used to evaluate differences between treatment
conditions. For vowels /a/ and /i/, the effects of condi­
tion (B vs GP vs WS), order (B-WS-GP vs. B-GP-WS),
FIGURE 1. Modified BORG-10 scale for measuring perceived and sex of the participants on F0, vocal intensity, and
vocal effort.
CPP were examined with three-way repeated measures
ANOVA. For the CAPE-V sentences, the effects of the
condition, order, and sex on CPP and speech rate were
rate of the recordings was 44.1 kHz with the depth of evaluated with a three-way repeated measures ANOVA.
16 bit. After the speech sample collection, each participant When the omnibus test indicated statistically significant
was asked by the clinician to rate vocal effort on a scale effect of an independent variable, pairwise t tests with
from 0–10 using a modified BORG-10 scale.34,35 Partici­ Bonferroni correction were conducted to identify the
pants then recorded their effort using a pen to circle their pair of conditions that produced statistically significant
chosen number on the printed visual scale (Figure 1). At differences. All statistics were run using R. 39
the end of the protocol, each participant was verbally asked
to choose one of three options: if their voice improved
more after the GP, WS, or if they noticed no difference RESULTS
between the two conditions. Items for selection were pro­ Demographics: A total of 43 individuals with a median age
vided in written form and were listed in the same order as of 58.5 (19−83) were assessed for eligibility. Race and
presented to the participant. ethnicity, as reported in the medical record, were 87%
White-Non-Hispanic or Latino, 7% White-Hispanic or
Participants Latino, and 4% African American. Nine individuals did
All participants were required to be between the ages of not participate; three did not meet inclusion criteria, and
18–89 and to have a diagnosis of either primary or sec­ six, who were screened and otherwise eligible, declined to
ondary MTD. Participants were referred to speech pa­ participate. Ultimately, 34 participants were randomly al­
thology for voice therapy by a qualified practitioner and located to alternating testing order. The final cohort con­
underwent laryngoscopy as part of their regular clinical sisted of 26 females and 8 males with an average age of
care. All participants completed the Voice Handicap Index- 53 years (range 20–83 years). Twenty-nine (85%) of 34
10 (VHI-10).36 Exclusion criteria included severe neurolo­ participants were referred by Ear Nose and Throat (ENT)
gical voice disorders, severe neuropsychiatric conditions practitioner (32% from a laryngologist and 53% from an
likely to affect voice, moderate to severe dysphagia for thin ENT chief resident, ENT-based nurse practitioner, or ENT
liquids based on patient report and/or chart review, and physician’s assistant). An additional five participants (15%)
uncontrolled cardiopulmonary disease (ie, uncontrolled were referred to speech pathology by physicians from non-
asthma and/or chronic obstructive pulmonary disease). ENT departments, two from pulmonology, and one each
Non-native English speakers, extensive laryngeal surgery, from gastroenterology, general internal medicine, and
or other medical conditions that could significantly alter physical medicine and rehabilitation. All participants un­
sensory-motor laryngeal function were also excluded. Any derwent laryngoscopy prior to participating in the study.
participants unable to adequately master GP within the 10- Laryngoscopy for participants referred by non-ENT pro­
minute time limit were also excluded. viders was performed by a speech pathologist and cases
were reviewed with a laryngologist. Twenty-five (74%) of
Acoustic analyses 34 participants were referred for speech pathology eva­
Voice recordings were visually and auditorily inspected luation and therapy with MTD as a primary diagnosis and
for silence and non-speech noise (eg, cough, background nine (26%) were referred with MTD as a secondary diag­
noise, etc) and manually edited to extract target utterances nosis. Non-MTD primary diagnoses, VHI-10, and other
using Goldwave (https://www.goldwave.com/) or Audacity demographics are illustrated by order group in Table 1. All
(audacityteam.org). PRAAT was used to obtain average F0 participants were able to learn how to gargle in a median
for CAPE-V sentences and average F0 for sustained vo­ time of 1:39 minutes (range 0:37–9:34 minutes). Refer to
wels, vocal intensity, CPP, and speech rate. For CPP, Table 1 for median and range of GP learning time by
CAPE-V sentences were concatenated first to obtain the condition order.
4 Journal of Voice, Vol. xx, No. xx, xxxx

Table 1.
Participant Demographics by Treatment Order Group
Swallow 1st (n = 17) Gargle Phonation 1st (n = 17)
Female/male 13/4 13/4
Age (average, range) years 56 (25−83) 50 (20−78)
Baseline VHI-10 (average, range) total score 16 (2−29) 15 (3−40)
Primary MTD/secondary MTD 13/4 12/5
Secondary MTD conditions − Vocal fold paresis (n = 2) − Vocal fold lesion or fibrovascular
− Autoimmune disease (n = 2) change (n = 4)
− Atrophy of vocal folds (n = 1)
Singing voice as a primary concern for 4/13 7/17
consultation
Learning time to master gargle (median, 2:08 (0:39−8:57) 1:34 (0:37−9:34)
range) Minutes:seconds
Effort after baseline speech (average, STD) 3.2 (± 1.8) 2.8 ( ± 1.5)
scale 1−7 (7 = max effort)
STD, standard deviation.

Participant perceptual measures Acoustic measures

Perceived vocal improvement: Participants reported per­ Sustained vowels
ceiving the most vocal improvement following GP (73.5%; Average F0 for vowel /a/: Average F0 for female partici­
n = 25), while 17.6% (n = 6) reported improvement fol­ pants was 189.92 Hz (± 31.20) at B, 207.78 Hz (± 30.47)
lowing WS and 8.8% (n = 3) noted no difference between following GP, and 198.20 Hz following WS (± 26.60).
the two conditions. Average F0 for male participants was 129.58 Hz (± 26.10)
at B, 129.75 Hz (± 15.87) following GP, and 133.34 Hz
(± 28.76) following WS. Repeated measures ANOVA for
the between-participant analysis showed that females sus­
Vocal effort tained /a/ at significantly higher F0 than males, F(1,
The average vocal effort ratings were 3.03 (SD ± 1.64) at 30) = 56.759, P < 0.001. No order effect, F(1, 30) = 3.097,
B, 2.50 (SD ± 1.46) after WS, and 1.85 (SD ± 1.15) after P = 0.0886 and no interaction between sex and order, F(1,
GP. The repeated measure two-way ANOVA indicated a 30) = 0.318, P = 0.5773 were found.
significant main effect of treatment for effort rating (F(2, When considering within-participant F0 across condi­
66) = 19.61, P < 0.01) but the effect of order was insignif­ tions, there was a significant main effect of condition on
icant (F(1, 32) = 0.281, P = 0.6). Pairwise t tests with F0, F(2, 60) = 3.552, P = 0.0349. However, there were no
Bonferroni correction across conditions indicated effort significant interactions for condition and sex (F(2,
ratings for GP was significantly lower compared to B (padj 60) = 1.149, P = 0.3238, condition and order F(2,
< 0.001). The rating was also significantly lower for WS 60) = 0.757, P = 0.4735 or between condition, sex, and
compared to B (padj = 0.011), and for GP was significantly order F(2, 60) = 0.369, P = 0.6928. Despite the significant
lower compared to WS (padj = 0.005) (Figure 2). effect indicated by the ANOVA model, pairwise t test with
Bonferroni correction did not find any significant differ­
ence between the treatment conditions. Results for females
were B vs GP, padj = 0.076; B vs WS, padj = 0.441; GP vs
WS, padj = 0.336; and for males were B vs GP, padj = 1.00; B
vs WS, padj = 0.648; GP vs WS, padj = 1.00.
Average F0 for vowel /i/: Average F0 for female parti­
cipants was 203.14 Hz (± 28.61) at B, 219.47 Hz (± 31.312)
for GP, and 203.56 Hz (± 31.15) for WS. Average F0 for
male participants was 135.37 Hz (± 43.42) at B, 132.11 Hz
(± 17.93) for GP, and 138.16 Hz (± 39.15) for WS.
Repeated measures ANOVA test indicated that females
sustained /i/ at higher F0 compared to males, F(1,
30) = 55.433, P < 0.001. There was no effect of order, F(1,
30) = 0.422, P = 0.5207; however, there was an interaction
between sex and order, F(1, 30) = 3.078, P = 0.0896.
FIGURE 2. Average effort rating for different condition orders. For the within-participants analysis, the main effect of
Error bars indicate standard error. condition on F0 was significant, F(2, 60) = 2.636, P = 0.080;
Diana M Orbelo, et al Gargle Phonation for Mild MTD 5

FIGURE 3. Average F0 of vowel /a/ and /i/. Error bars indicate standard errors.

however, no interactions were significant condition and sex Within-participants analysis showed no main effect for
F(2, 60) = 1.630, P = 0.205; condition and order F(2, condition on vocal intensity, F(2, 60) = 1.595, P = 0.211;
60) = 0.291, P = 0.748, and condition, sex, and order, F(2, and interactions were insignificant (condition and sex, F(2,
60) = 2.224, P = 0.117. Pairwise t test showed no difference 60) = 0.653, P = 0.524, condition and order, F(2,
in F0 between B and GP (padj = 0.134) or B and WS (padj 60) = 0.306, P = 0.737, and condition, sex, and order, F(2,
= 1.00) for female participants. However, there was a sig­ 60) = 0.236, P = 0.791) (Figure 4).
nificant difference between GP and WS (padj = 0.038), in­ CPP for vowel /a/: Average CPP values for female par­
dicating that F0 after GP was significantly higher than F0 ticipants were 13.65 dB (± 2.85) at B, 13.95 dB (± 2.83)
after WS. No differences were found for male participants following GP, and 12.62 (± 2.85) dB following WS.
(B vs GP, padj = 1.00; B vs Swallow, padj = 1.00; GP vs WS, Average CPP values for male participants were 13.36 dB
padj = 1.00) (Figure 3). (± 3.76) at B, 15.38 dB (± 3.54) for GP, and 13.03 dB
Vocal intensity for vowel /a/: Average vocal intensity for (± 4.12) for WS. Repeated measures ANOVA, though not
female participants was 67.73 dB SPL (± 4.51) at B, significant, trended toward higher CPP values for males
67.75 dB SPL (± 4.89) for GP, and 66.49 dB SPL for WS compared to females F(1, 30) = 3.961, P = 0.056. There was
(± 4.63). Average vocal intensity for male participants was no main effect of order on CPP, F(1, 30) = 1.631, P = 0.211,
71.13 dB SPL (± 4.60) at B, 71.93 dB SPL (± 4.92) for GP, and there was no interaction between sex and order, F(1,
and 72.25 dB SPL for (± 5.16) WS. Repeated measures 30) = 0.234, P = 0.632. Within-participants tests did reveal
ANOVA showed that males produced significantly higher a significant main effect of condition on CPP, F(2,
dB SPL compared to females, F(1, 30) = 6.314, P = 0.018. 60) = 3.965, P = 0.024, and there was a significant interac­
There was no main effect of order, F(1, 30) = 0.189, tion between condition and sex, F(2, 60) = 3.759, P = 0.029.
P = 0.667, and no interaction between sex and order, F(1, Alternately, there were no significant interactions between
30) = 1.807, P = 0.189. condition and order, F(2, 60) = 0.969, P = 0.385, or condi­
Within-participants analysis showed no main effect of tion, sex, and order, F(2, 60) = 1.063, P = 0.352.
condition on vocal intensity, F(2, 60) = 1.521, P = 0.227 A pairwise t test with Bonferroni correction was used to
and interactions were insignificant (condition and sex, F(2, examine the effects of condition on CPP of vowel /a/. For
60) = 1.823, P = 0.170, condition and order, F(2, female participants, no significant differences were found
60) = 0.774, P = 0.466, and condition, sex, and order, F(2, between B and GP (padj = 1); however, CPP for WS was
60) = 0.646, P = 0.528.). significantly lower when compared to B (padj = 0.040) and
Vocal intensity for vowel /i/: Average vocal intensity of when compared to GP (padj = 0.011). For male participants,
female participants was 67.05 dB SPL (± 3.74) at B, no significant differences were found among any of the
66.01 dB SPL (± 4.52) for GP, and 66.79 dB SPL for WS condition comparisons: B and GP (padj = 0.139), B and WS
(± 4.40). Average vocal intensity of male participants was (padj = 0.618), and GP and WS (padj = 0.99).
70.73 dB SPL (± 4.13) at B, 69.99 dB SPL (± 5.36) fol­ CPP for vowel /i/: Average CPP values for female par­
lowing GP, and 70.48 dB SPL (± 3.57) following WS. ticipants were 12.89 dB (± 2.85) at B, 13.44 dB (± 2.48)
Repeated measures ANOVA showed that males again following GP and 12.61 dB (± 3.12) following WS. Average
produced higher dB SPL than females, F(1, 30) = 5.481, CPP values for male participants were 14.51 dB (± 4.21) at
P = 0.026. There was no main effect of order on vocal in­ B, 14.81 dB (± 3.18) following GP and 13.43 dB (± 3.47)
tensity, F(1, 30) = 0.144, P = 0.707, and no interaction be­ following WS. Repeated measures ANOVA indicated no
tween sex and order, F(1, 30) = 1.501, P = 0.230. main effect of sex, F(1, 30) = 1.126, P = 0.297 or order on
6 Journal of Voice, Vol. xx, No. xx, xxxx

FIGURE 4. Average vocal intensity of vowel /a/ and /i/. Error bars indicate standard errors.

CPP, F(1, 30) = 0.121, P = 0.73; and no interaction between and GP (padj = 0.017) and for GP and WS (padj = 0.005),
sex and order, F(1, 30) = 0.003, P = 0.956. Within-partici­ but not for B and WS (padj = 1). For male participants, no
pants tests showed a significant main effect of condition on significant differences were found among any of the con­
CPP, F(2, 60) = 5.798, P = 0.005. However, there were no dition comparisons: B and GP (padj = 1), B and WS (padj
significant interactions between condition and sex, F(2, = 1), and GP and WS (padj = 1). Another pairwise t test
60) = 0.755, P = 0.475, condition and order, F(2, with Bonferroni correction was conducted to examine the
60) = 0.935, P = 0.398, or condition, sex, and order, F(2, effects of order on CPP. In the GP1st group, a significant
60) = 0.791, P = 0.458. difference was found between B and GP (padj = 0.004) and
Pairwise t test with Bonferroni correction examining the between GP and WS (padj = 0.003), but not between B and
effects of condition on CPP of vowel /i/ showed, for female WS (padj = 1). Alternatively, in the WS1st group, no sig­
participants, that CPP values for GP were higher when nificant differences were found among condition compar­
compared to WS (padj = 0.026), but not when comparing isons (B and GP (padj = 1), B and WS (padj = 1), and GP and
GP and B (padj = 0.073), or B and WS (padj = 0.93). For WS (padj = 1)) (Figure 5).
male participants, no significant differences were found Speech rate: For female participants, the average speech
among any of the condition comparisons (B and GP (padj rate was 3.30 syllables per second (sps) (± 0.74) for B, 3.51
= 1), B and WS (padj = 0.774), and GP and WS sps (± 0.78) for GP, and 3.42 sps for WS. For male parti­
(padj = 0.549). cipants, the average speech rate was 3.00 sps (± 0.72) for B,
3.21 sps (± 0.68) for GP, and 3.32 sps (± 0.81) for WS.
Repeated measures ANOVA for the between-participants
Sentences analysis for speech rate showed no significant main effect of
CPP: For female participants, average CPP values across condition (F(1, 28) = 0.405, P = 0.529), sex (F(1,
six sentences from the CAPE-V were 9.32 dB (± 1.50) at B, 28) = 2.259, P = 0.144), or order (F(1, 28) = 0.287,
9.71 dB (± 1.18) for GP, 9.32 dB (± 1.04) for WS. For male P = 0.597). Likewise, no significant interactions were noted
participants, average CPP values were 8.69 dB (± 1.80) at between condition and order (F(1, 28) = 0.877, P = 0.357),
B, 8.89 dB (± 1.36) for GP, 8.70 dB (± 1.64) for WS. par­ or order and sex (F(1,28) = 1.515, P = 0.229). The within-
ticipants, Repeated measures ANOVA for the between- participants analysis showed a significant effect of condi­
participants analysis revealed no effect of sex on CPP, F(1, tion on speech rate (F(2, 567) = 5.155, P = 0.006). There
30) = 1.896, P = 0.179, Similarly, there was no effect of was no significant effect of order F(1, 567) = 2.014,
order on CPP, F(1, 30) = 1.376, P = 0.250, and no interac­ P = 0.156). Likewise, no significant interactions were noted
tion between sex and order, F(1, 30) = 1.514, P = 0.228. between condition and sex F(2, 567) = 0.988, P = 0.373),
Within-participants tests revealed a significant main effect order and condition (F(2, 567) = 1.298, P = 0.274), and
of condition on CPP, F(2, 60) = 4.494, P = 0.015. While condition, order, and sex (F(2, 567) = 0.818, P = 0.442).
there was no significant interaction between condition and Pairwise t tests with Bonferroni correction were con­
sex, F(2, 60) = 0.241, P = 0.787, there was a significant in­ ducted to investigate differences in speech rate across
teraction between condition and order, F(2, 60) = 3.592, conditions for both male and female participants. For
P = 0.034, and a significant three-way interaction between female participants, a speech rate was significantly faster
condition, sex, and order, F(2, 60) = 3.224, P = 0.047. for GP compared to B (padj = 0.029). No significant dif­
Pairwise t test with Bonferroni correction was used to ferences were observed between B and WS (padj = 0.573)
examine the effects of conditions on CPP. For female or between GP and WS (padj = 0.462). For male partici­
participants, a significant difference was found between B pants, speech rate was significantly higher for WS
Diana M Orbelo, et al Gargle Phonation for Mild MTD 7

FIGURE 5. CPP for vowel /a/ and /i/ and CAPE-V sentences. Error bars indicate standard error. CAPE-V, Consensus Auditory-
Perceptual Evaluation of Voice; CPP, cepstral peak prominence.

compared to B (padj = 0.003), while there was no sig­

nificant difference between the B and GP (padj = 0.132)
and between GP and WS (padj = 1.00) (Figure 6).

DISCUSSION
This study aimed to evaluate the potential effect of GP on
self-perceived vocal improvement and phonatory effort as
well as on acoustic measures (F0, vocal intensity, CPP, and
speech rate) in individuals with MTD. The study utilized a
clinical trial format, which to our knowledge, is the first
study to do so while assessing GP and incorporating a
“sham” procedure. The importance of using clinical trial
study designs is well documented40 and ideally, study de­
signs should include a sham procedure and blinding.41 FIGURE 6. Speech rate for CAPE-V sentences. Error bars in­
Blinding to testers and participants was not possible for dicate standard error.
GP; however, a crossover design using WS was used to
address this limitation. The WS condition incorporated all
elements of GP, including the presence of water and act of with primary MTD. The predominance of females in the
swallowing, with the act of gargling itself being the only present study mirrors the demographic trends reported in
difference between the two, in order to isolate the act of the literature, which show a higher incidence of MTD and
gargling from simply swallowing water. Most of the par­ voice disorders in females compared to males.12,42 The re­
ticipants in this study were females, who were diagnosed sults of this study showed that most participants reported
8 Journal of Voice, Vol. xx, No. xx, xxxx

vocal improvement and reduced vocal effort after GP. on vocal improvement. This result supports the improve­
These self-perceived changes were reflected in some of the ment reported by the participants. Interestingly, CPP for
acoustic measures. Overall, our findings support the use of the vowel samples did not change after GP but decreased
GP as a treatment tool for MTD. after WS. The results were also vowel-specific: CPP was
The results for the self-reported measures were in support of lower for WS compared to B and GP for the sustained
our hypothesis: participants reported improvements in voice vowel /a/, while for the sustained vowel /i/, CPP was lower
quality and decreased vocal effort following GP. These findings for WS compared to GP only. The reason for this decrease
are consistent with those of Albuquerque et al27 who observed is unclear. Because CPP is known to be affected by the
a reduction in electrical activity in suprahyoid and infrahyoid intensity of a signal, it is possible that the decrease in CPP
muscles. While our study did not measure the treatment effect is explained by a decrease in vocal intensity from B to WS.
on the change in muscle physiology, the self-reports may reflect However, this intensity change was observed only with the
the reduced extrinsic muscle activity elicited by GP. Ad­ vowel /a/. One potential explanation is that the 5 seconds
ditionally, the reduction of vocal effort following GP observed water hold incorporated in the WS condition required the
in this study also aligns with the reports by Amorim et al. 28 participants to hold their breath, or swallow aggressively,
Our acoustic results indicated that GP elicited notable which may have perturbed the balance between the pho­
changes in both vocal and speech production character­ natory and respiratory systems. Though these statistical
istics. There was an increased F0 for female participants in differences were found, CPP values for the vowels hovered
the sustained vowel /i/ after GP, which aligned with the around established cut-off points for vocally-normal
increased F0 reported by Amorim et al.29 Amorim et al adults.45,46 Therefore, interpreting these differences for
posited that GP may have increased F0 due to the chin their clinical significance should be approached with cau­
lifting to produce GP. Our finding challenges this view, as tion. Similarly, caution is also warranted when interpreting
participants did not maintain this posture and were in a the changes in CPP elicited by GP or WS.
neutral head position for audio recordings following GP. As predicted, the effect of GP on speech rate was noted but
Additionally, our participants always swallowed water only in female participants. For this group, the speech rate
after gargling, suggesting a probable reset of laryngeal increased from B to GP, but not from B to WS. The speech
height. It may be that pitch increased as the voice was re­ rate for GP was significantly higher compared to WS, in­
turning to a higher, more optimal F0 due to increased dicating that GP had a distinctive effect on speech production
muscle relaxation. However, it should be noted that this that was not elicited by WS. For male participants, GP did not
interpretation is not definitive, as our study did not involve elicit a change in speech rate; however, there was a significant
monitoring the electrical activity of the muscles, which increase from B to WS. There was no difference between GP
would provide more direct evidence of increased muscle and WS. The difference between the female and male groups is
relaxation. Additionally, considering that the baseline F0 difficult to explain due to the small sample size of the male
values were within normal limits, it is essential to ac­ group. Regardless, this observation may underscore a close
knowledge that the observed vocal change, while sig­ relationship between the larynx and supraglottic articulators.
nificant, remains subtle in its nature. Excessive tension in the laryngeal and peri-laryngeal muscles
Our results also highlight the importance of considering might have restricted the movements involved in articulation.
both sex and vowel type in assessing the treatment effects By reducing this tension, GP and WS may have enabled par­
of GP. Unlike the study by Amorim et al, our study ana­ ticipants to speak more quickly. It should be noted that speech
lyzed male and female voices separately given the potential rate was the only objective measure sensitive to the treatment
sex-related differences in F0.43 Additionally, we examined effects of GP in both sexes, suggesting its potential utility in
the treatment effect with two vowels. This approach re­ tracking clinical progress in MTD.
vealed that the effect of GP was specific to women and The effect of WS on some outcome measures was un­
particularly evident for the vowel /i/. The absence of a expected, including the decrease in CPP for the CAPE-V sen­
noticeable treatment effect in male participants may be tence in the female participants and an increase in speech rate
attributed to the relatively small sample size of this group. seen in the male cohort. Though this may indicate that the
This finding also underscores the need for larger sample effects of swallowing impacted the speech production system,
sizes in future research to fully understand the differential the precise mechanism behind these findings is yet to be de­
impacts of GP across sexes and vowel types. For vocal termined and beyond the scope of our current study. Although
intensity, we did find that males spoke louder than females; WS did demonstrate some effects, our findings demonstrated
unlike Albuquerque et al,27 we found no relative change in greater positive effects following GP.
vocal intensity following GP.
CPP, an objective measure of breathiness and general Possible mechanisms of action and gargle phonation
dysphonia,44,45 showed the effect of GP only with the as a voice therapy tool for MTD
sentence level samples. An increase in CPP indicates a de­ The clear findings in the present study were the self-per­
crease in aperiodicity, suggesting reduced dysphonia se­ ception of improved voice and decreased vocal effort after
verity. GP significantly raised CPP for the sentence-sized GP. There are several potential explanations for these im­
samples from the B and WS conditions, implying an effect provements. As mentioned, perilarygeal muscles were likely
Diana M Orbelo, et al Gargle Phonation for Mild MTD 9

more relaxed following GP. It is also likely that GP im­ References

proves coordination of the airflow associated with phona­ 1. Balata PM, Silva HJ, Pernambuco LA, et al. Electrical activity of
tion. As is commonly known within the field, breath is the extrinsic laryngeal muscles in subjects with and without dysphonia. J
Voice. 2015;29:129 e9–17. https://doi.org/10.1016/j.jvoice.2014.03.012.
power source for phonation. Improving breath and voice 2. Aronson AE, Bless DM. Clinical Voice Disorders. 4th ed. New York,
coordination can therefore improve voice quality and re­ NY: Thieme; 2009.
duce vocal effort by improving overall vocal efficiency.47 In 3. Van Houtte E, Van Lierde K, D’haeseleer E, et al. The prevalence of
teaching GP in the present study and in the clinic, it is laryngeal pathology in a treatment-seeking population with dys­
sometimes necessary to cue an individual to “use their air” phonia. Laryngoscope. 2010;120:306–312.
4. Oates J, Winkworth A. Current knowledge, controversies and future
or other cue to facilitate increased airflow. It may be directions in hyperfunctional voice disorders. Int J Speech-Lang
helpful in future studies to obtain aerodynamic measures, Pathol. 2008;10:267–277.
such as mean airflow during phonation following GP. 5. Van Houtte E, Van Lierde K, Claeys S. Pathophysiology and treat­
Another influence of breathing could be that relatively ment of muscle tension dysphonia: a review of the current knowledge.
longer exhales with sustained GP as compared inhalation J Voice. 2011;25:202–207. https://doi.org/10.1016/j.jvoice.2009.10.009.
6. Verdolini K, Rosen CA, Branski RC. Classification Manual for Voice
may drive parasympathetic tone,48,49 helping individuals Disorders-I. New York, NY: Psychology Press, Taylor & Francis
with MTD to feel more relaxed. Group; 2014.
Limitations: The present study has several limitations. 7. Morrison MD, Rammage LA. Muscle misuse voice disorders: de­
Most of the current participant cohort had mild dysphonia, scription and classification. Acta oto-laryngologica. 1993;113:428–434.
as evidenced by baseline VHI-10 scores, vocal effort rat­ 8. McKenna VS, Diaz-Cadiz ME, Shembel AC, et al. The relationship
between physiological mechanisms and the self-perception of vocal
ings, and acoustic measures. Also, males were under­ effort. J Speech Lang Hear Res. 2019;62:815–834. https://doi.org/10.
represented, which likely has resulted in the lack of 1044/2018_JSLHR-S-18-0205.
statistical significance in this group. In future studies, it 9. Solomon NP. Vocal fatigue and its relation to vocal hyperfunction.
would be beneficial to investigate a more evenly distributed Int J Speech-Lang Pathol. 2008;10:254–266.
group of mild, moderate, and severe dysphonia cases to 10. McDowell S, Morrison R, Mau T, et al. Clinical characteristics and
effects of vocal demands in occupational voice users with and without
determine the impact of GP across different degrees of primary muscle tension dysphonia. J Voice. 2022.
dysphonia severity. Future work should include audio- 11. Toles LE, Shembel AC. Acoustic and physiologic correlates of vocal
perceptual measures, additional acoustic and aerodynamic effort in individuals with and without primary muscle tension dys­
measures, as well as select speech or linguistic measures to phonia. Am J Speech-Lang Pathol. 2024;33:237–247.Jan 3.
fully understand the scope of potential changes related 12. Altman KW, Atkinson C, Lazarus C. Current and emerging concepts
in muscle tension dysphonia: a 30-month review. J Voice.
to GP. 2005;19:261–267. https://doi.org/10.1016/j.jvoice.2004.03.007.
13. Dromey C, Nissen SL, Roy N, et al. Articulatory changes following
treatment of muscle tension dysphonia: preliminary acoustic evidence.
J Speech Lang Hear Res. 2008;51:196–208. https://doi.org/10.1044/
CONCLUSION 1092-4388(2008/015).
The present study suggests that GP is a quick, easy, and 14. Higgins MB, Netsell R, Schulte L. Vowel-related differences in lar­
accessible tool for treating MTD, and reveals that GP can yngeal articulatory and phonatory function. J Speech Lang Hear Res.
robustly reduce vocal effort and improve vocal quality. 1998;41:712–724. https://doi.org/10.1044/jslhr.4104.712.
Participants overwhelmingly noted that GP improved their 15. McClean MD, Tasko SM. Association of orofacial with laryngeal and
respiratory motor output during speech. Exp Brain Res.
voice more than the sham. Furthermore, GP can alter 2002;146:481–489. https://doi.org/10.1007/s00221-002-1187-5.
acoustic and speech measures. In the present study, both 16. Roy N, Nissen SL, Dromey C, et al. Articulatory changes in muscle
CPP and speech rate increased following GP in women tension dysphonia: evidence of vowel space expansion following
with mild MTD. Further work is warranted to explore manual circumlaryngeal therapy. J Commun Disord. 2009;42:124–135.
potentially favorable changes in speech rate or articulation, https://doi.org/10.1016/j.jcomdis.2008.10.001.
17. Bottalico P, Murgia S, Puglisi GE, et al. Intelligibility of dysphonic
and future studies should include more severe cases speech in auralized classrooms. J Acoust Soc Am. 2021;150:2912.
of MTD. https://doi.org/10.1121/10.0006741.
18. Jacobson MC, Franssen E, Birt BD, et al. Predicting post­
laryngectomy voice outcome in an era of primary tracheoesophageal
fistulization: a retrospective evaluation. J Otolaryngol.
Declaration of Competing Interest 1997;26:171–179.
None 19. Evitts PM, Starmer H, Teets K, et al. The impact of dysphonic voices
on healthy listeners: listener reaction times, speech intelligibility, and
listener comprehension. Am J Speech Lang Pathol. 2016;25:561–575.
https://doi.org/10.1044/2016_AJSLP-14-0183.
Acknowledgments 20. Ishikawa K, Boyce S, Kelchner L, et al. The effect of background
noise on intelligibility of dysphonic speech. J Speech Lang Hear Res.
This publication was made possible by the Mayo Clinic 2017;60:1919–1929. https://doi.org/10.1044/2017_JSLHR-S-16-0012.
CTSA through grant number UL1TR002377 from the 21. Utianski RL, Duffy JR, Martin PR, et al. Rate modulation abilities in
National Center for Advancing Translational Sciences acquired motor speech disorders. J Speech Lang Hear Res.
(NCATS), a component of the National Institutes of 2023;66:3194–3205. https://doi.org/10.1044/2022_JSLHR-22-00286.
Health (NIH). Aug 17.
10 Journal of Voice, Vol. xx, No. xx, xxxx

22. Krause JC, Braida LD. Acoustic properties of naturally produced clear 35. Baldner EF, Doll E, van Mersbergen MR. A review of measures of
speech at normal speaking rates. J Acoust Soc Am. 2004;115:362–378. vocal effort with a preliminary study on the establishment of a vocal
23. Behlau M, Madazio G, Oliveira G. Functional dysphonia: strategies effort measure. J Voice. 2015;29:530–541.
to improve patient outcomes. Patient Relat Outcome Meas. 36. Rosen CA, Lee AS, Osborne J, et al. Development and validation of
2015;6:243–253. https://doi.org/10.2147/PROM.S68631. the voice handicap index-10. Laryngoscope. 2004;114:1549–1556.
24. Boominathan P, Mahalingam S, Arunachalam R, et al. An eclectic https://doi.org/10.1097/00005537-200409000-00009.
voice therapy program for the treatment of hyperfunctional voice 37. Murray ESH, Chao A, Colletti L. A practical guide to calculating
disorders (HFVD). J Voice. 2023;37:969-e1. cepstral peak prominence in Praat. J Voice. 2022. https://doi.org/10.
25. Meerschman I, Van Lierde K, D’haeseleer E, et al. Immediate and 1016/j.jvoice.2022.09.002.
short-term effects of straw phonation in air or water on vocal fold 38. De Jong NH, Wempe T. Praat script to detect syllable nuclei and
vibration and supraglottic activity of adult patients with voice dis­ measure speech rate automatically. Behav Res Methods.
orders visualized with strobovideolaryngoscopy: a pilot study. J 2009;41:385–390.
Voice. 2021. https://doi.org/10.1016/j.jvoice.2021.09.017. 39. R Core Team. R: a language and environment for statistical com­
26. Muckala J. Rosenberg M, LeBorgne W, eds. The Vocal Athlete puting. R Foundation for Statistical Computing, Vienna, Austria. R
Application and Technique for the Hybrid Singer. 2nd ed. San Diego, Core Team; 2019, Published online 2015.URL https://www.R-project.
CA: Plural Publishing; 2021. 89–90. org/.
27. da Rocha Albuquerque AA, Balata PM, de Amorim GO, et al. Effects 40. Sibbald B, Roland M. Understanding controlled trials. Why are
of Voiced Gargling on the Electrical Activity of Extrinsic Laryngeal randomised controlled trials important? BMJ. 1998;316:201.
Muscles and Vocal Self-assessment. J Voice. 2022;36:515–522. https:// 41. Miller FG, Kaptchuk TJ. Sham procedures and the ethics of clinical
doi.org/10.1016/j.jvoice.2020.06.008. Jul 1. trials. J R Soc Med. 2004;97:576–578.
28. Amorim GO, Morais EPG, Bastos FRA, et al. Autopercepção vocal 42. Sheyona V, Devadas U. The prevalence and impact of voice problems
após a realização do exercício do gargarejo sonorizado. Anais do 19° in nonprofessional voice users: preliminary findings. J Voice.
Congresso Brasileiro de Fonoaudiologia. p.1046, 2011b. Available 2022;36:383–388.
from: http://www.sbfa.org.br/portal/anais2011/trabalhos_select.php? 43. Whiteside SP. Temporal-based acoustic-phonetic patterns in read
tt=Busca&id_artigo=1046 [Last accessed March 11, 2024]. speech: some evidence for speaker sex differences. J Int Phon Assoc.
29. Amorim G, Morais E, Bastos F, et.al. Comportamento da frequência 1996;26:23–40.
fundamental antes e ap os o exercício do gargarejo sonorizado. In: 44. Patel RR, Awan SN, Barkmeier-Kraemer J, et al. Recommended
Vol Annals of 19° Congresso Brasileiro de Fonoaudiologia.; protocols for instrumental assessment of voice: American Speech-
2011:1047, 2011a. Available at: http://www.sbfa.org.br/portal/ Language-Hearing Association expert panel to develop a protocol for
anais2011/trabalhos_select.php?tt=Busca&id_artigo=1047. Accessed instrumental assessment of vocal function. Am J Speech-Lang Pathol.
September 11, 2020. 2018;27:887–905.
30. Maryn Y, Roy N, De Bodt M, et al. Acoustic measurement of overall 45. Murton O, Hillman R, Mehta D. Cepstral peak prominence values
voice quality: A meta-analysis. J Acoust Soc Am. 2009;126:2619–2634. for clinical voice evaluation. Am J Speech-Lang Pathol. 2020;29:
https://doi.org/10.1121/1.3224706.Nov 1. 1596–1607.
31. Mehta DD, Hillman RE. Voice assessment: updates on perceptual, 46. Buckley DP, Abur D, Stepp CE. Normative values of cepstral peak
acoustic, aerodynamic, and endoscopic imaging methods. Curr Opin prominence measures in typical speakers by sex, speech stimuli, and
Otolaryngol Head Neck Surg. 2008;16:211–215. software type across the life span. Am J Speech-Lang Pathol.
32. Roy N, Barkmeier-Kraemer J, Eadie T, et al. Evidence-based clinical 2023;32:1565–1577. https://doi.org/10.1044/2023_AJSLP-22-00264.
voice assessment: a systematic review. Published online; 2013. Jul 10.
33. Zraick RI, Kempster GB, Connor NP, et al. Establishing validity of 47. Jiang JJ, Titze IR. Measurement of vocal fold intraglottal pressure
the Consensus Auditory-Perceptual Evaluation of Voice (CAPE-V). and impact stress. J Voice. 1994;8:132–144.
Am J Speech Lang Pathol. 2011;20:14–22. https://doi.org/10.1044/ 48. Russo MA, Santarelli DM, O’Rourke D. The physiological effects of
1058-0360(2010/09-0105). slow breathing in the healthy human. Breathe (Sheff). 2017;13:
34. van Leer E, van Mersbergen M. Using the Borg CR10 physical ex­ 298–309. https://doi.org/10.1183/20734735.009817.
ertion scale to measure patient-perceived vocal effort pre and post 49. Chung YM, Lou SL, Tsai PZ, et al. The efficacy of respiratory reg­
treatment. J Voice. 2017;31:389 e19–389 e25. https://doi.org/10.1016/j. ulation on parasympathetic nervous system appraised by heart rate
jvoice.2016.09.023. variability. J Med Biol Eng. 2019;39:960–966.