Asia Pacific Journal of Speech, Language and Hearing

ISSN: 1361-3286 (Print) (Online) Journal homepage: http://www.tandfonline.com/loi/yslh19

Short-term effect of humming on vocal quality

Edwin M-L. Yiu & Eva Y-Y. Ho

To cite this article: Edwin M-L. Yiu & Eva Y-Y. Ho (2002) Short-term effect of humming on
vocal quality, Asia Pacific Journal of Speech, Language and Hearing, 7:3, 123-137, DOI:
10.1179/136132802805576436

To link to this article: http://dx.doi.org/10.1179/136132802805576436

Published online: 18 Jul 2013.

Submit your article to this journal

Article views: 21

View related articles

Citing articles: 7 View citing articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=yslh19

Download by: [University of California Santa Barbara] Date: 10 May 2016, At: 17:48
 Asia Pacific Journal (~f Speech, Language and Hearing, 7, 123-1372002 123

Short-term effect of humming on vocal quality

Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

Edwin M-L. Yiu and Eva Y-Y. Ho

Voice Researqh Laboratory, Division of Speech and Hearing Sciences, The University of Hong Kong

Abstract
Humming is a conservative voice therapy technique used to facilitate easy and
efficient natural voice production. It is a technique used in treating voice disorders
due to vocal hyperfunction, vocal abuses and/or misuses. However, few efficacy
studies of humming are available in the literature. The present study was a
prospective study which set out to investigate the changes in vocal quality demon-
strated by eight female subjects with hyperfunctional dysphonia (six with vocal
nodules and two with chronic laryngitis) and eight female subjects with normal voice
following two sessions of training using humming per se. Voice recordings were taken
before and after the humming exercises. Three judges rated the roughness and breath-
iness of these samples independently using the GRBAS scheme with a 10-point visual
analogue rating scale. Acoustic analyses were also carried out to measure the average
fundamental frequency, jitter, shimmer and harmonic to noise ratio. Both groups of
subjects demonstrated a significant improvement in perceptual roughness (p <0.05)
but no change in the breathiness rating and the acoustic measures. These preliminary
findings provide some evidence that humming alone can at least bring about short-
term improvement in perceptual rough vocal quality.

Key words: resonant voice therapy, efficacy, vocal improvement.

Introduction
Experts generally agree that conservative (non-surgical) voice therapy is the first
line of treatment for adults with laryngeal pathologies associated with vocal hyper-
function, misuses or abuses with or without organic lesions such as vocal nodules
(Bastian, 1986; Sataloff, 1987). Different forms of conservative voice therapy
techniques for improving voice production and vocal quality in hyperfunctional
voice disorders have been described in the literature (Aronson, 1990; Colton and
Casper, 1996; Harris et aI., 1998; Boone and McFarlane, 1999). According to
Carding et ai. (1999), these techniques can be classified into direct and indirect
therapy. Indirect therapy refers to methods that facilitate a better management of
voice. These include methods such as provision of vocal hygiene and voice care
information (Carding et aI., 1999). Direct therapy, which is usually based on physio-
logical principles, aims at modifying vocal production (Stemple et aI., 1994; Carding
 124 YIU,HO

et aI., 1999). Humming is a physiological exercise that is believed to facilitate easy

phonation with minimal impact force being exerted on the vocal folds. The early
description of humming technique in the literature was closely associated with the
concept of optimal pitch, although the concept of optimal pitch (Fairbanks, 1960)
has been challenged and refuted by some researchers (see Colton and Casper,
1996). Early reports of the use of humming exercise in voice therapy appeared in
the 1960s in Europe (Zaliouk, 1963; Greene, 1964) and later in the 1970s in North
America (Cooper, 1973).
A number of textbooks describe this techniq ue in detail (e.g. Green and
Mathieson, 1989; Aronson, 1990; Morrison et aI., 1994; Colton and Casper, 1996;
Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

Harris et aI., 1998). This technique stresses the importance of easy and efficient
natural voice production using either an 'optimal pitch' or a natural 'urn-hum' as
opposed to the habitual effortful and inefficient voice production. In humming,
speakers are asked to produce the sound Iml with their lips closed in a relaxed
manner. The pitch level of the Iml needs to be maintained at the most comfortable
and natural level of the speaker. Several strategies have been suggested to achieve
the appropriate humming results. One way is to glide a pitch until a maximum
vibration is felt with a finger on the maxillary bone (Moncur and Brackett, 1974).
Another method is to produce an 'urn-hum' as if sincerely agreeing with someone
(Cooper, 1973). Using an appropriate and comfortable pitch with easy phonation
and a correct focus of resonance are considered the keys to a successful hum
(Morrison et aI., 1994; Colton and Casper, 1996).
Humming is incorporated in the resonant voice therapy developed by Verdolini
and her associates (Peterson et aI., 1994; Verdolini-Marston et aI., 1995; Verdolini
et aI., 1998). The therapy involves a proprioceptive sensation of resonance or
vibration of the maxillary bone upon producing a hum with easy phonation
(Peterson et aI., 1994; Verdolini-Marston et aI., 1995; Verdolini et aI., 1998). It has
been observed that in actors and singers with extensive voice training as well as
patients with vocal nodules, their vocal folds barely adducted upon humming as
demonstrated by both endoscopic images and electroglottography (Peterson et aI.,
1994; Verdolini-Marston et aI., 1995; Verdolini et aI., 1998). This easy adduction is
believed to reduce the impact force on the vocal folds and to bring about
improvement in vocal quality. When this phonatory skill using humming is trans-
ferred to daily voice production, the reduced force of impact will give the vocal
lesion a chance to heal (Verdolini-Marston et aI., 1995). In addition to the improved
laryngeal adduction, it is also suggested that the easy phonation facilitates tension
reduction in the supralaryngeal areas and in the extrinsic laryngeal muscles (Harris
et aI., 1998). The relaxation in these structures and areas further facilitates a better
resonance, which results in a clearer voice quality.
The study reported by Verdolini-Marston et a1. (1995) is probably the only
systematic control study on humming therapy. Their study used a combination of
humming technique (direct therapy) and vocal hygiene education (indirect therapy)
to treat 13 female subjects with vocal nodules for two weeks. The combined
methods were shown to be effective in bringing improvement in perceptual and
phonatory effort measures. Since a combination of both direct and indirect methods
were employed in their study, it is difficult to determine the effectiveness of
humming per see
 EFFECT OF HUMMING ON VOCAL QUALITY 125

The present study set out to investigate the effect of humming on the acoustic
and perceptual properties of vocal quality. Subjects who participated in this study
were not given any information about vocal hygiene or methods to eliminate vocal
abusive behaviours prior to the humming training. The hypothesis tested in this
study was:

Following two weekly sessions of humming practice, improvement in the acoustic and
perceptual vocal quality of speakers with vocal pathologies would be observed.

Methods
Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

Subjects
Sixteen female dysphonic subjects were recruited to participate in this study. The
dysphonic subjects were recruited from two public hospital voice clinics in Hong
Kong. The laryngological diagnoses as determined by laryngologists using either
indirect laryngoscopy or fibreoptic nasendoscopy in all these cases were related to
hyperfunctional voice disorders. Hyperfunctional dysphonia in this study is defined
in its broadest sense to include dysphonia resulting from hyperkinetic laryngeal
movement with or without organic lesions such as vocal nodules (Green and
Mathieson, 1989; Morrison et aI., 1994). Subjects with neurological pathology, vocal
fold paralysis, motor speech disturbance, hearing loss of more than 20dB HL and
history of laryngeal surgery or voice therapy were excluded. All the subjects were
non-smokers and non-drinkers. The recording of the subjects' voices started within
two weeks after their first laryngological examinations.
Only eight dysphonic subjects who demonstrated noticeable dysphonia percep-
tually on both the first and second pre-training recording sessions had completed
the whole training with a complete set of data. Data from the other eight subjects
were therefore not included in the final data analysis. Six of these eight subjects had
either unilateral or bilateral vocal nodules and two subjects had chronic laryngitis.
The mean age of the eight dysphonic subjects was 40.0 years (standard deviation =
11.83, range 21-58) with a mean onset time of voice problems of 16.3 months
(standard deviation = 11.33, range 2-36). Table 1 lists the laryngological diagnosis,
occupation, age and voice problem onset time of the eight dysphonic subjects.

Table 1: Background data of the dysphonic subjects

Subject Age (years) Laryngological Occupation Onset of the voice

diagnosis .problem (months)

1 27 Chronic laryngitis Teacher 2

2 46 Chronic laryngitis Kindergarten teacher 6
3 39 Unilateral vocal nodule Housewife 19
4 58 Unilateral vocal nodule Retired 2
5 42 Bilateral vocal nodules Telephone operator 18
6 49 Unilateral vocal nodule Business owner 36
7 21 Unilateral vocal nodule Student 17
8 38 Unilateral vocal nodule Teacher 30
 126 YIU, HO

Eight females with normal voice that matched in age with the dysphonic subjects
were recruited as a non-dysphonic group to undergo the humming training. They
had a mean age of 38.6 years (standard deviation = 9.26, range 22-54). None of
them had any history of voice disorders or hearing problems. They also had no
previous voice or singing training. Their voices were judged normal by themselves
and by four final-year speech pathology students.

Procedure

Humming training programme

Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

The subjects were given two sessions of humming training. Each training session
was approximately 45 minutes long and the two sessions were one week apart. Each
subject was seen individually by a trainer (the second author) with the explanation
of the principle of humming technique given before starting the training. The
subjects were not given any information on vocal hygiene and were not asked
explicitly to eliminate or reduce any possible vocal abusive behaviours before the
training and voice recordings were completed. This eliminated the possible
improvement in voice quality when the subjects changed their vocal habits
following the vocal hygiene advice. For ethical reasons, vocal hygiene information
was given on completion of the training and recordings.
The training was based on a model-practice-feedback-self-monitoring-practice
paradigm. The second author first provided a model demonstration of humming.
The subjects then carried out the practice. Feedback to the subjects on each
practice in relation to gentle onset, resonance and comfortable pitch level was then
provided by the trainer. The subjects then listened to their own voice and improved
on the practice.
In the first training session, each subject was asked:

1. to produce a hum as in acknowledgement of someone asking a question. The

hum had to be produced in a relaxed manner (demonstration was given by the
trainer);
2. to glide the hum up and down a musical scale in order to find a pitch that
resulted in maximum resonance/vibration without increasing effort. The
located pitch was then used as the note for further humming (demonstration
was given by the trainer);
3. to feel the resonance/vibration of the hum over the area of the nasal bridge
using a finger. Tingling sensation around the lip area was another kinaesthetic
feedback that was also explained;
4. to listen to the voice quality of the hum with comments on the performance
given by the trainer;
5. to practise humming three times, each for 15 minutes, on a daily basis at home.

In the second session, each subject was asked:

1. to hum at a natural and comfortable pitch in a relaxed manner and add a

sustained vowel to the end of the /m/. Demonstration of /m/ ...Ia/ production was
given by the trainer. Smooth transition from the sound /m/ to /a/ was emphasized;
 EFFECT OF HUMMING ON VOCAL QUALITY 127

2. to listen to the voice quality of the Im/ ...Ial with comments on the performance
given by the trainer;
3. to practise Im/ .. .Ia/three times, each for 15 minutes, on a daily basis at home.

Voice recordings
Three sets of voice recordings of Im/ ...Ial were carried out. The first recording(R-1)
was carried out a week prior to the humming training. The second recording (R-2)
was carried out at the beginning of the first session just before the humming
exercise. The R-1 and R-2 constituted a no-treatment condition. The third
recording (R-3) was carried out one week after the completion of the second
Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

training session.
All recordings were carried out in a room with background noise level of not
more than 50 dB. A Sony MZ R-55 Mini-Disc (MD) Recorder and a Shure SM 48
professional microphone were used. The microphone was held 5 cm from the
mouth of each subject to avoid recording noise burst (Titze, 1995). No specific
instruction was given for the recording except that each subject was asked to
produce Im/ ...Ial at a comfortable pitch and loudness level. The subjects were also
asked to ensure that the duration for the 1m/and tal sounds was approximately
three seconds each, therefore giving rise to a total duration of at least six seconds
for each Im ... al segment. Five practice trials were allowed before the actual
recordings so that the subject could warm her voice up and the appropriate
recording level could be adjusted.

Acoustic analysis
All the voice recordings were digitized at a sampling rate of 50 kHz using Kay
Elemetrics' Computerized Speech Lab (CSL) Model 4300B, and later analysed by
the Multidimensional Voice Program (MDVP). The middle three-second portion of
the lal from each of the Im/ ...Ial samples was extracted manually using a pair of
cursors on the CSL computer screen. Five acoustic measures, as listed in Table 2,
were obtained for each signal using the MDVP program. These five measures were
related to fundamental frequency, jitter, shimmer and noise measurements. These
perturbation measures were obtained because they have been shown to be useful in
quantifying voice quality and have some correlation with the perceptual roughness
and breathiness (Yumoto et aI., 1984; Wolfe and Steinfatt, 1987; Martin et aI., 1995;
Millet and Dejonckere, 1998).

Perceptual evaluation
The stimuli edited for th~ acoustic analysis were used for the perceptual evaluation.
Each stimulus was presented twice. With a total of 16 subjects and three recording
sessions, the stimulus set therefore contained a total of 96 stimuli. All the stimuli
were presented in random order to three listeners, who had qualifications in speech
pathology and training in voice evaluation. They were required to rate the severity
of the roughness and breathiness qualities of the GRBAS scale (Hirano, 1981)
using a 10 cm long visual analogue scale. None of the judges was familiar with the
nature of the study nor knew the subjects. Roughness and breathiness of the
GRBAS scale were chosen for rating because of their wide popularity and relia-
 128 YIU,HO

..-....-..
\0 lr)
~~
----
00
o 00
00
oo
r-

Q.,
~
o
1-;
00
u
..-....-..
lr) ~
'2
----
('f') ('f')

o 00
..s::
Q., ~('f')
en
~ ~~
"'0 00
Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

C:
o
Z
..-....-..
lr) N

----
('f') ('f')

00
lr) ~
\0 \0
00

..-....-..
\0 \0

----
r- r-
oo
r- lr)
0"1 0"1
00

..-....-..
lr) 0

----
00 r-
oo
0"1 00
C'!~
~ ~

..-....-..
~o

----
lr) lr)
00
r- \0
0"1 0"1
00
 EFFECT OF HUMMING ON VOCAL QUALITY 129

bility for clinical purposes (Dejonckere et aI., 1996; Dejonckere, 1998). Written
definitions of roughness ('lack of clarity') and breathiness ('audible air escape
during voice production') were given to the judges before the evaluation began.
Prior to the actual rating, the three listeners were given a training session which
consisted of listening to 10 standard voice samples representing different degrees of
severity of breathiness and roughness.
The ratings were carried out individually in a sound-treated room using a Sony
MD player (MDS-S39) and a pair of Sony MDR-E565 earphones. The judges were
allowed to listen to each recording as many times as they liked. The evaluation
session took approximately an hour for each judge to complete. A break was given
halfway through the evaluation session.
Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

The severity ratings made by the judges on the scale for breathiness and
roughness were measured. The mark on each rating scale was measured from the
left end of the scale in centimetres to the nearest millimetre. A total of 15% of the
recordings were presented to the three judges for rating on a separate occasion one
week later. These data were used to determine the intra- and inter-judge reliability.
Two judges were considered in agreement when their ratings fell within one
centimetre of each other.

Results

Acoustic analysis
The acoustic measures obtained using the MDVP included the average funda-
mental frequency, two frequency perturbation measures, an amplitude perturbation
measure and a noise-related measure. The mean values of these acoustic measures
across the three recording sessions are listed in Table 2.

Between-group comparison
The dysphonic and non-dysphonic groups were significantly different on the
relative average frequency perturbation, pitch perturbation quotient and amplitude
perturbation quotient (Mann-Whitney U = 11, P = 0.028, in all three cases) only in
the voice recordings carried out in R-2. No significant differences were found in the
samples in the other two recording sessions or the other acoustic measures.

Within-group comparison of baseline measures

Non-parametric repeated Wilcoxon signed ranks tests were carried out to
determine whether the acoustic measures changed between R-1 and R-2 in both the
dysphonic and non-dysphonic groups. No significant difference was found in any of
the acoustic measures in either the dysphonic or non-dysphonic groups (p >0.05).

Within-group comparison of pre- and post-humming training

After the humming training, the dysphonic group showed a significantly higher
average fundamental frequency (Wilcoxon signed ranks test Z = -2.1,p = 0.02) in R-
3 when compared with R-2. Other acoustic measures were not significantly
different after the humming training (p>0.05). A similar trend in the average
fundaJ;11entalfrequency was also demonstrated by the non-dysphonic group after
the humming training (Wilcoxon signed ranks test Z = -2.2, P = 0.01). Interestingly,
 130 YIU,HO

both the relative average frequency perturbation (Wilcoxon signed ranks test Z = -
2.1, p = 0.02) and pitch perturbation quotient (Wilcoxon signed ranks test Z = -2.1, .
p = 0.02) were significantly higher after the humming training.

Perceptual evaluation
The inter- and intra-judge agreement measures are listed in Table 3. The intra-
judge agreement measures were all statistically significant and were at more than
90°!c>(see Table 3). The inter-judge agreement measures were moderate and were
statistically significant (between 50% and 59°!c>,p <0.01). Although they were only
at a moderate level, they were of similar degrees to those reported in other
Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

perceptual studies (Kreiman et aI., 1990; Kreiman et aI., 1992). The mean ratings of
the three judges were averaged for the purpose of further analysis in the present
study.

Table 3: Percentage of inter- and intra-judge agreement within 1 em on the visual

analogue scale (VAS)

Judges Inter-judge agreement (% ) Intra-judge agreement (0,10)

1 and 2 52.5 90.0

2 and 3 50.0 93.0
land 3 59.2 95.0

Between-group comparison
The boxplots in Figure 1 show the median ratings, ranges and outliers of each of the
three recording sessions. The dysphonic group demonstrated significantly higher
roughness and breathiness in both recording sessions before the humming training
(R-1: roughness, Mann-Whitney U = 5, p = 0.003; breathiness, Mann-Whitney U =
4, p = 0.002; R-2: roughness, Mann-Whitney U = 5, p = 0.003; breathiness, Mann-
Whitney U = 1, P = 0.0001) than the non-dysphonic group. After the humming
training, the dysphonic group still demonstrated significantly higher perceptual
roughness than the non-dysphonic group (Mann-Whitney U = 9, P = 0.01).
However, the dysphonic group was not significantly different (Mann-Whitney U =
14, p = 0.06) from the non-dysphonic group on the breathiness quality, possibly
because of highly variable ratings in the dysphonic group (see Figure 1).

Within..;group comparison of baseline measures

The dysphonic group demonstrated no significant differences between the two
recordings (R-1 and R-2) before the humming training in both the roughness
(Wilcoxon signed ranks test Z = -0.7, P = 0.48) and breathiness (Wilcoxon signed
ranks test Z = -0.63, p = 0.53) ratings.

Within-group comparison of pre- and post-humming training

The median roughness rating (1.9) shown by the dysphonic group after humming
training (in R-3) was significantly lower than that before the humming training
(R-2) (median rating 3.9; Wilcoxon signed ranks test Z = -2.38, p = 0.02). Never-
 EFFECT OF HUMMING ON VOCAL QUALITY 131

10

9
Dysphonic Non-dysphonic

o
Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

o
1 2 3 2 3
Recordings

10

9 Dysphonic Non-dysphonic
8

3
2.9 o
2.1
2

o
2 3
0~
1 2 3

Recordings
maximum
3rd quartile
median (value is listed)

~ minimum

o outlier

Figure 1: Median, quartile, range and outliers of perceptual ratings.

theless, the breathiness rating did not show any significant changes between
Recording 3 (median rating 2.1) and Recording 2 (median rating = 3.0; Wilcoxon
signed ranks Z = -0.56, p = 0.57), possibly, again, due to the highly variable ratings
 132 YIU,HO

of breathiness in Session 3 (see Figure 1). The improvement in the roughness

quality was also found in the non-dysphonic group. Median roughness rating of
samples from Recording 3 (0.6) was significantly lower than that from Recording
2 (1.9; Wilcoxon signed ranks test Z = -2.5, p = 0.01). Like the dysphonic group, the
median breathiness ratings of the ,non-dysphonic subjects were not significantly
different between R-3 (0.5) and R-2 (0.3, Wilcoxon signed ranks test Z = -0.42, p =
0.67).

Data on individual subjects

As the fundamental frequency and roughness quality were shown to increase and
Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

improve significantly after the humming training, individual subject performance

was examined. Figure 2 shows the subjects who demonstrated at least an increase of
10 Hz after the humming training. Four dysphonic subjects and five non-dysphonic
subjects showed this trend in the fundamental frequency.
Six of the eight dysphonic subjects showed at least a one-point improvement in
their vocal roughness between R-2 and R-3 (see Figure 3). Two subjects showed an
improvement of more than two points while four subjects showed an improvement
of one to two points (Figure 3). Interestingly, all the non-dysphonic subjects showed
an improvement in rough quality between R-2 and R-3 (Figure 4). Three subjects
showed an improvement of more than one point while the other five subjects
showed minimal changes.

Discussion
This study showed that humming facilitated a change in the quality of voice
production. The technique significantly facilitated voice production with a relative
increase in fundamental frequency and also less perceptual roughness when
compared with voice production before learning the technique. This observation
was noted in both the dysphonic and non-dysphonic individuals. On the other hand,
none of the acoustic measures (jitter and shimmer) showed any significant change
following humming. This suggests that the perturbation measures were not as
sensitive as the perceptual rating in detecting voice quality changes in these
subjects. It has indeed been demonstrated that perturbation measures are not
always accurate in identifying abnormal voice quality (Bielamowicz et aI., 1996) nor
in discriminating normal from abnormal voice quality (Yiu, 1999).
Approximately half of the subjects in the dysphonic and non-dysphonic groups
showed an increase in fundamental frequency following humming training. Such
change was noticed regardless of whether there was a vocal problem or not. As the
recordings were carried out on the phrase fm ...af, it was not possible to determine
whether this increase in fundamental frequency would have been found in daily
voice use as well. Becaus.e this finding isnotju$t found in the dysphonic group,
improvement in the vocal condition that has led to an increase in fundamental
frequency is not an appropriate explanation. Nevertheless, it can be interpreted
that a humming task facilitates an increase in fundamental frequency. This slight
increase in fundamental freq uency might have been necessary to produce a
comfortable and natural pitch, which resulted in a noticeably louder and richer
voice with better resonance (Moncur and Brackett, 1974). The improvement in
resonance might have been related to the strong spectral energy at around 2.5-3.0
kHz (Singer's formant) as a result of possible adjustment of the vocal tract configu-
 EFFECT OF HUMMING ON VOCAL QUALITY 133

Dysphonic subjects

With 10Hz or more increase

300

250 _.•... 53
-54
-.- 55
.-..-•...
Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

200 ••.... . - .._ ..- ..__ ..... -e-57

150
1 2 3
Recordings

Non-dysphonic subjects

With 10Hz or more increase

300

250
~ ....
- .... - ..• - N3
-N4
~N6
-M-N7
200 -N8

150
1 2 3
Recordings

Figure 2: Subjects with at least an increase of 10 Hz in the fundamental frequency after humming
training in Recording 3.

ration (Harris et aI., 1998). This strong spectral energy is responsible for the loud
and rich resonated perceptual quality.
It should be noted that the perceptually significant improvement in voice quality
was noticed in both the dysphonic and non-dysphonic subjects following humming.
The improvement, however, was restricted only to roughness and not breathiness. It
might well be that local tissue changes as a result of the humming exercise helped to
 134 YIU,HO

With more than 2 points improvement

8
C)7
~ 6
eu
=5 -'-51
=c 34 -r-57
"5
g 2
a: 1
Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

o
2 3
Recordings

With 1 to 2 points improvement

8
7
en
.5 6
--tJ-53
! 5
f/) -'-55
:: 4 -·-56
c
a 3 _. -S8
g
a:
2
1
o
2 3
Recordings

Figure 3: Changes of ratings of perceptual roughness in dysphonic subjects after humming training
(Recording 3).

improve the roughness. However, this factor alone could not explain why the non-
dysphonic group also demonstrated similar changes in the roughness ratings. The
data from the present study did not allow us to explore the possible explanations for
the improvement in roughness. However, findings from other studies may provide
some possible explanations. It has been shown that humming facilitated a laryngeal
configuration with the vocal folds barely adducted in normal subjects (Peterson et
aI., 1994; Verdolini et aI., 1998) and in subjects with vocal nodules (Verdolini-
Marston et aI., 1995; Verdolini et aI., 1998). It is hypothesized that the minimal
intraglottal impact under such configuration helps to bring about better voice
quality. Another mechanism put forward by Harris et al. (1998) is also hypothesized
 EFFECT OF HUMMING ON VOCAL QUALITY 135

With more than 1 point improvement

3

Ir-._--~
--,',~, -e-N2
-X-N4
-6- N8

o
Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

2 3
Recordings

With less than 1 point change

-+-N1
+- -. - - - _.- - +. -6--N3
. ..• -- N5
---N6
-I--N7
o
2 3
Recordings

Figure 4: Changes of ratings of perceptual roughness in non-dysphonic subjects after the humming
training (Recording 3).

to contribute to the improved phonation. Harris et aI. (1998) contended that the
relatively neutral position of the larynx during humming induces very little
muscular activity in the supra-laryngeal and extra-laryngeal areas. Onset of
phonation is therefore easier without excessive muscle tension. This therefore
results in a more periodic vibration and clearer voice quality. The changes in pitch
level (as indicated in the present study), laryngeal and supra-laryngeal configu-
ration (Verdolini et aI., 1998) constitute a new motor control pattern. According to
Stemple et al. (1994), vocal exercises facilitate the learning of the new motor
control pattern.
The lack of improvement in breathiness in the dysphonic subjects following
humming training might be attributed to two possible reasons. First, the time frame
of the humming therapy was relatively short (two weeks only). There was probably
not enough time for the vocal lesions to disappear. The persisting lesions might
have prevented complete closure of the glottis during phonation. This resulted in
 136 YIU,HO

persistent breathiness.Second, the hypothesized physiological mechanism

(Peterson et al., 1994; Verdolini et aI., 1998) which facilitates a barely adducted
vocal fold vibration would have no doubt allowed more airtlow through the glottis,
which was perceived as breathiness.
Some cautions are warranted in interpreting the results of the present study.
First, the study did not include a no-treatment or a placebo-treatment dysphonic
control group for ethical reasons. Although the two baseline measurements before
the humming training (R-1 and R-2) were designed to determine whether there was
a lack of change when treatment was withheld, it was, however, different from the
inclusion of a no-treatment or placebo-treatment dysphonic group in which placebo
Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

and Hawthorne effects could have been controlled for. One way to overcome the
ethical issue of withholding treatment in a clinical group is to select clinics that have
a long waiting time for patients seeking voice evaluation and therapy. Subjects
recruited through these clinics could be allocated to treatment, no-treatment and
placebo-treatment group randomly. The no-treatment and placebo-treatment
groups will be assessed during the usual waiting period that the clinics impose. It is
often a difficult issue to decide what activities would constitute placebo treatment
in voice therapy. Unlike drug treatment, behavioural therapy has to be designed in
a form that has no direct effect on· the target behaviour but, at the same time, the
subjects should feel they are receiving some form of therapy. In voice therapy, one
form of placebo treatment could involve some motor exercises that do not require
laryngeal function, such as limb or body movement exercises.
A second caution in interpreting the results from the present study is that only
the short-term effect of humming performance was demonstrated. Long-term
learning effect or the ability to transfer the humming skills outside the clinic was not
examined. Further studies are needed to determine what should be included in the
therapy programme to ensure generalization in using humming.
The third caution is that only a small number of subjects were included in this
study. This sample size was not large enough for us to examine factors (such as
occupations, personality, emotional status, amount of voice use and motivation)
that might have contributed to the success or failure of this method. A larger scale
study would have allowed this to be investigated.
In summary, the present study showed that humming is effective in improving
the voice quality in speakers with normal voice or dysphonia due to hyperfunction.
This provides further support to previous findings (Peterson et aI., 1994; Verdolini
et at, 1998) that humming is an effective treatment approach for dysphonia with
hyperfunctional causes. The change in voice quality, even in the subjects with
normal voice, suggested that the method rhight be useful from a preventive
perspective. Further study will be needed to determine whether the humming
exercise is useful in the prevention of voice problems.

Acknowledgement
This study was supported in part by a grant from Hong Kong Health Service
Research Committee (HSRC 821007). A Universitas 21 fellowship awarded to the
first author, which supported the preparation of this manuscript, is also gratefully
acknowledged. Thanks are also due to the two anonymous reviewers for their
helpful and constructive comments on the initial draft of the manuscript.
 EFFECT OF HUMMING ON VOCAL QUALITY 137

References
Aronson AE. Clinical voice disorders. New York: Thieme, 1990.
Bastian RW. Benign mucosal disorders saccular disorders and neoplasm. In Cummings C, Frederickson F,
Krause C, Schuller D, eds. Otolaryngology - head and neck surgery. St Louis: CV Mosby, 1986, pp
1965-1987.
Bielamowicz S, Kreiman J, Gerratt BR, Dauer MS, Berke GS. Comparison of voice analysis systems for
perturbation measurement. J Speech Hear Res 1996; 39: L26-134.
Boone DR, McFarlane SC. The voice and voice therapy, 6th edn. Needham Heights, MA: Allyn & Bacon,
1999.
Carding PH, Horsley lA, Docherty GJ. A study of the effectiveness of voice therapy in the treatment of 45
patients with nonorganic dysphonia. J Voice 1999: L31: 72-104.
Colton R, Casper J. Understanding voice problems: a physiological perspective for diagnosis and
treatment. New York: Williams & Wilkins, 1996.
Downloaded by [University of California Santa Barbara] at 17:48 10 May 2016

Cooper M. Modern techniques of vocal rehabilitation. Springfield, IL: Charles C. Thomas, L973.
Dejonckere P. Effect of louder voicing on acoustical measurements in dysphonic patients. Logopedic
Phoniatrs Voco11998; 23: 79-84.
Dejonckere PH, Remacle M, Fresnel-Elbaz E. Reliability and relevance of differentiated perceptual evalu-
ation of pathological voice quality. In Clement MP, ed. Voice update. Amersterdam: Elsevier, 1996.
Fairbanks G. Voice and articulation drillbook. New York: Harper & Row, 1960.
Greene MCL. The voice and its disorders, 2nd edn. London: Pitman, 1964.
Green MCL, Mathiesen L. The voice and its disorders, 5th edn. London: Whurr, 1989.
Harris T, Harris S, Rubin JS, Howard DM. The voice clinic handbook. London: Whurr Publishers, 1998.
Hirano M. Clinical examination of voice. Vienna: Springer Verlag, 1981.
Kreiman J, Gerratt BR, Precoda K. Listener experience and perception of voice quality. J Speech Hear
Res 1990; 33: 103-115.
Kreiman J, Gerratt BR, Precoda K, Berke GS. Individualdifferences in voice quality perception. J Speech
Hear Res 1992; 35: 512-520.
Martin D, Fitch J, Wolfe V. Pathological voice type and the acoustic prediction of severity. J Speech Hear
Res 1995; 38: 765-771.
Millet B, Dejonckere PH. What determines the differences in perceptual rating of dysphonia between
experienced raters? Folia Phoniatrica et Logopedie 1998; 50: 305-310.
Moncur Jp, Brackett IP. Modifying vocal behavior. New York: Harper & Row, 1974.
Morrison M, Rammage L, Nichol H, Pullan B, May P, Salkeld L . The management of voice disorders.
San Diego: Singular, 1994.
Peterson KL, Verdolini-Marston K, Barkmeier JM, Hoffman HT. Comparison of aerodynamic and
electroglottographic parameters in evaluating clinically relevant voicing patterns. Ann Otol Rhinol
Laryngol 1994; 103:335-346.
Sataloff RT. The professional voice: Part III Common diagnoses and treatments. J Voice 1987; 1:
283-292. .
Stemple JC, Lee L, D' Amico B, Pickup B. Efficacy of vocal function exercises as a method of improving
voice production. J Voice 1994; 83: 271-278.
Titze IR. Workshop on acoustic voice analysis: summary statement. Iowa City: National Center for Voice
and Speech, 1995.
Verdolini K, Druker DG, Palmer PM, Samawi H. Laryngeal adduction in resonant voice. J Voice 1998;
12: 315-327.
Verdolini-Marston K, Burke MK, Lessac A, Glaze L, Caldwell E. Preliminary study of two methods of
treatment for laryngeal nodules. J Voice 1995; 91: 74-85.
Wolfe VI, Steinfatt TM. Prediction of vocal severity within and across voice types. J Speech Hear Res
1987; 30: 230-240.
Yiu EM-L. Limitations of perturbation measures in clinical acoustic voice analysis. Asia Pacific J Speech
Lang Hear 1999; 4: 157-168.
Yumoto E, Sasaki Y, Okamura H. Harmonic-to-noise ratio and psychophysical measurement of the degree
of hoarseness. J Speech Hear Res 1984; 27: 2-6.
Zaliouk A. The tactile approach to voice placement. Folio Phoniatrica 1963; 15: 147.

Address correspondence to: Dr Edwin Yiu, Associate Professor, Voice Research

Laboratory, Division of Speech and Hearing Sciences, The University of Hong Kong,
5/F Prince Philip Dental Hospital, 34 Hospital Road, Hong Kong. Fax: (852) 25590060;
email: edwinyiu@hku.hk