**3. Results**

#### **3.1 Videolaryngostroboscopy**

During videolaryngostroboscopy, it was occasionally difficult to synchronize the stroboscopic illumination with the vocal fold vibration during lip and tongue trills. Therefore, during those tasks, we photographed only those moments of the cycles at which the maximum opening was evident. During the phonation of the sustained vowel /ε/ task, we encountered no such difficulty.

Although the otolaryngologist maintained the fiberoptic laryngoscope in position, laryngeal movement caused variations in the distance between the larynx and the laryngoscope, which in turn caused variations in the images of the larynx. Among the 220 images analyzed in the present study, the larynx was too far from the laryngoscope in 163 (74%). Of the remaining 57 images (26%), 11 (5%) were images of sustained vowel phonation, 19 (9%) were images of lip trills, and 27 (12%) were images of tongue trills. The measurements of vocal fold vibration amplitude are presented in Tables 3 through 6 and illustrated in Figures 5 and 6.


#### **3.1.1 Low intensity**

\*Friedman's test, among all three variables

Table 3. Maximum amplitude of vocal fold vibration during phonation of the sustained vowel /ε/ compared with that of vocal fold vibration during tongue and lip trills at low intensity


\*Wilcoxon signed rank test

138 Otolaryngology

For the cases in which the difference was statistically significant, we used the Wilcoxon

For the presentation of the results, the measurements taken automatically by the program were considered to constitute the intraindividual means and standard deviations, whereas the values obtained by statistical analysis of those results were considered to constitute the

During videolaryngostroboscopy, it was occasionally difficult to synchronize the stroboscopic illumination with the vocal fold vibration during lip and tongue trills. Therefore, during those tasks, we photographed only those moments of the cycles at which the maximum opening was evident. During the phonation of the sustained vowel /ε/ task,

Although the otolaryngologist maintained the fiberoptic laryngoscope in position, laryngeal movement caused variations in the distance between the larynx and the laryngoscope, which in turn caused variations in the images of the larynx. Among the 220 images analyzed in the present study, the larynx was too far from the laryngoscope in 163 (74%). Of the remaining 57 images (26%), 11 (5%) were images of sustained vowel phonation, 19 (9%) were images of lip trills, and 27 (12%) were images of tongue trills. The measurements of vocal fold vibration amplitude are presented in Tables 3 through 6 and illustrated in Figures

Vibration amplitude

/ε/ 10 0.11 0.03 0.06 0.16 0.11

Tngue trills 10 0.15 0.04 0.08 0.21 0.16

Table 3. Maximum amplitude of vocal fold vibration during phonation of the sustained vowel /ε/ compared with that of vocal fold vibration during tongue and lip trills at low

Lip trills 10 0.17 0.06 0.10 0.32 0.16 0.002\*

p Mean SD Minimum Maximum Median

signed rank test in order to identify the types that differed.

interindividual means and standard deviations.

**3.1 Videolaryngostroboscopy** 

we encountered no such difficulty.

**3. Results** 

5 and 6.

intensity

**3.1.1 Low intensity** 

Variable N

\*Friedman's test, among all three variables

Table 4. Pairwise comparisons among phonation of the sustained vowel /ε/, lip trills, and tongue trills at low intensity in terms of the amplitude of vocal fold vibration

Fig. 5. Maximum amplitude of vocal fold vibration during phonation tasks at low intensity in relation to the cuneiform cartilage, as assessed by videolaryngostroboscopy. In a, phonation of the sustained vowel /ε/; in b, sustained lip trills; and in c, sustained tongue trills.


#### **3.1.2 High intensity**

\*Friedman's test, among all three variables

Table 5. Maximum amplitude of vocal fold vibration during phonation of the sustained vowel /ε/ compared with that of vocal fold vibration during tongue and lip trills at high intensity

Comparison Among Phonation of the Sustained Vowel /ε/, Lip Trills,

sustained vowel /ε/, lip trills, and tongue trills at low intensity

**3.2.1 Low intensity** 

Intraindividual mean of the closed quotient

Intraindividual SD of the closed quotient

\*Friedman's test

Intraindividual SD of the closed quotient

\*Wilcoxon signed rank test

**3.2.2 High intensity** 

Intraindividual mean of the closed quotient

Intraindividual SD of the closed

\*Friedman's test

quotient

and Tongue Trills: The Amplitude of Vocal Fold Vibration and the Closed Quotient 141

Measurement Task N Mean SD Minimum Maximum Median p\*

/ε/ 10 47.72 12.27 31.87 67.27 51.64

tongue trills 10 52.25 9.55 37.50 67.76 54.43

/ε/ 10 3.12 3.40 0.88 12.18 1.99

tongue trills 10 7.63 4.41 2.09 13.85 6.54

Table 7. Closed quotient: number of samples, mean, standard deviation, minimum value, maximum value, median, and significance of the differences among phonation of the

Table 8. Pairwise comparisons among phonation of the sustained vowel /ε/, lip trills, and tongue trills at low intensity in terms of the automatic measurements of significant values

Measurement Task N Mean SD Minimum Maximum Median p\*

/ε/ 10 50.71 6.90 36.47 56.07 54.52

tongue trills 10 54.60 9.72 40.63 72.56 52.52

/ε/ 10 1.62 0.84 0.77 3.17 1.39

tongue trills 10 4.75 2.78 1.68 9.50 4.15

Table 9. Automatic measurements, as taken by EGG: number of samples, mean, standard deviation, minimum value, maximum value, median, and significance of the differences among phonation of the sustained vowel /ε/, lip trills, and tongue trills at high intensity

lip trills 10 59.21 12.57 42.85 82.73 54.68 0.007

lip trills 10 6.64 5.36 1.21 15.48 4.23 0.020

Measurement Pairs p\*

0.301 lip trills 10 50.97 12.96 33.93 71.48 53.36

0.020 lip trills 10 6.55 3.12 3.86 12.63 5.34

Lip trills versus sustained vowel /ε/ phonation 0.059 Tongue trills versus sustained vowel /ε/ phonation 0.037 Tongue trills versus lip trills 0.508


\*Wilcoxon signed rank test

Table 6. Pairwise comparisons among phonation of the sustained vowel /ε/, lip trills, and tongue trills at high intensity in terms of the amplitude of vocal fold vibration

The values obtained during phonation of the sustained vowel /ε/ were significantly different from those obtained during lip and tongue trills at high and low intensities (p > 0.01). The statistical tests revealed that the maximum amplitude of vocal fold vibration during phonation of the sustained vowel /ε/ was significantly different from that of vocal fold vibration during lip and tongue trills. However, there were no significant differences between lip and tongue trills in terms of the maximum amplitude of vocal fold vibration.

Fig. 6. Maximum amplitude of vocal fold vibration during phonation tasks at high intensity in relation to the cuneiform cartilage, as assessed by videolaryngostroboscopy. In a, phonation of the sustained vowel /ε/; in b, sustained tongue trills; and in c, lip trills.

#### **3.2 EGG – Closed quotient**

All of the EGG waveforms were either grade 1 or grade 2 in accordance with the criteria proposed by Vieira (1997) and were therefore appropriate for automatic measurements.

The measurements taken automatically by the program for analysis of the EGG waveform were considered to constitute the intraindividual means and standard deviations of the closed quotient. Those values were considered the study variables and placed on the vertical axes of the tables. The interindividual means and standard deviations of the closed quotient were obtained by statistical analysis and were placed on the horizontal axes of the tables.

Tables 7 through 10 show the results related to the intraindividual means and standard deviations of the closed quotient.

#### **3.2.1 Low intensity**

140 Otolaryngology

Table 6. Pairwise comparisons among phonation of the sustained vowel /ε/, lip trills, and

The values obtained during phonation of the sustained vowel /ε/ were significantly different from those obtained during lip and tongue trills at high and low intensities (p > 0.01). The statistical tests revealed that the maximum amplitude of vocal fold vibration during phonation of the sustained vowel /ε/ was significantly different from that of vocal fold vibration during lip and tongue trills. However, there were no significant differences between lip and tongue trills in terms of the maximum amplitude of vocal

a 0.12 b 0.19 c 0.22

Fig. 6. Maximum amplitude of vocal fold vibration during phonation tasks at high intensity

All of the EGG waveforms were either grade 1 or grade 2 in accordance with the criteria proposed by Vieira (1997) and were therefore appropriate for automatic measurements.

The measurements taken automatically by the program for analysis of the EGG waveform were considered to constitute the intraindividual means and standard deviations of the closed quotient. Those values were considered the study variables and placed on the vertical axes of the tables. The interindividual means and standard deviations of the closed quotient were obtained by statistical analysis and were placed on the horizontal axes of the tables.

Tables 7 through 10 show the results related to the intraindividual means and standard

in relation to the cuneiform cartilage, as assessed by videolaryngostroboscopy. In a, phonation of the sustained vowel /ε/; in b, sustained tongue trills; and in c, lip trills.

Pair p\* Lip trills versus sustained vowel /ε/ phonation 0.005 Tongue trills versus sustained vowel /ε/ phonation 0.005 Tongue trills versus lip trills 0.308

tongue trills at high intensity in terms of the amplitude of vocal fold vibration

\*Wilcoxon signed rank test

fold vibration.

**3.2 EGG – Closed quotient** 

deviations of the closed quotient.


\*Friedman's test

Table 7. Closed quotient: number of samples, mean, standard deviation, minimum value, maximum value, median, and significance of the differences among phonation of the sustained vowel /ε/, lip trills, and tongue trills at low intensity


\*Wilcoxon signed rank test

Table 8. Pairwise comparisons among phonation of the sustained vowel /ε/, lip trills, and tongue trills at low intensity in terms of the automatic measurements of significant values


#### **3.2.2 High intensity**

\*Friedman's test

Table 9. Automatic measurements, as taken by EGG: number of samples, mean, standard deviation, minimum value, maximum value, median, and significance of the differences among phonation of the sustained vowel /ε/, lip trills, and tongue trills at high intensity

Comparison Among Phonation of the Sustained Vowel /ε/, Lip Trills,

as between tongue trills and phonation of the sustained vowel /ε/.

laryngoscope was close to the larynx.

(Garrel *et al.*, 2008; Titze, 2009).

opening was visible.

abovementioned hypothesis.

vibration during trill exercises.

during phonation of the sustained vowel /ε/.

and Tongue Trills: The Amplitude of Vocal Fold Vibration and the Closed Quotient 143

the sustained vowel phonation task, even when the image was indicative that the

There were significant differences among phonation of the sustained vowel /ε/, lip trills, and tongue trills in terms of the maximum amplitude of vocal fold vibration at low and high intensities (p = 0.002 and p = 0.001, respectively). The Wilcoxon signed rank test revealed significant differences between lip trills and phonation of the sustained vowel /ε/, as well

For the performance of lip trills and tongue trills, pulmonary airflow has to increase in order to maintain vocal fold vibration and lip or tongue vibration (Titze, 2009; Warren *et al.*, 1992; Titze 1988, McGowan, 1992). The increase in airflow can lead to an increase in subglottic pressure, which can in turn lead to an increase in the amplitude of vocal fold vibration

In the present study, during the measurement of amplitude, it was difficult to synchronize the flashes of light from the stroboscope with the vocal fold vibration during lip and tongue trills in some cases. We had no such difficulty when we measured the amplitude of vocal fold vibration during phonation of the sustained vowel /ε/. Therefore, during lip trills and tongue trills, we photographed only those moments of the cycles at which the maximum

Increased airflow can also lead to aperiodicity of vocal fold vibration (Jiang, 2001; Tao, 2007) and can destabilize the flashes of light from the stroboscope (Sercarz *et al.*, 1992). In previous studies, we found that the standard deviation of the intraindividual closed quotient was highest during lip trills and tongue trills, the values being higher at high intensities, which is due to increased airflow (Russell *et al.*, 1998; Alku, 2006). Those results might indicate differences in the periodicity of the EGG waveform during trill exercises and corroborate the

Bueno (2006) observed pronounced movement of the laryngeal framework during tongue trills; such movement can make it difficult for the contact microphone to pick up the fundamental frequency. Because the cycles observed under the stroboscopic light are not the actual vocal fold vibration cycles, which require regularity in order to generate reliable images (Sercarz *et al.*, 1992; Patel *et al.*, 2008), a study involving a high-speed camera would be needed in order to have a better view of vocal fold vibration during tongue and lip trills. To that end, a device that is compatible with the laryngoscope is needed. Studies involving a qualitative analysis of both the periodicity of the EGG waveform and the amplitude of the EGG signal could also be useful in the investigation of the mechanism of vocal fold

The standard deviation of the closed quotient is a numerical representation of the variation in the closed quotient in the EGG waveform. Because the standard deviation of the closed quotient was higher during the trill exercises (at high or low intensity) in the present study, we can affirm that the variation in the closed quotient is greater during trill exercises than

For tongue trills and lip trills to occur, the anterior part of the vocal tract has to be occluded by the tongue or the lips. Intraoral pressure increases and becomes greater than the force


\*Wilcoxon signed rank test

Table 10. Pairwise comparisons among phonation of the sustained vowel /ε/, lip trills, and tongue trills at high intensity in terms of the automatic measurements of significant values

### **4. Discussion**

Our results show that, in general, tongue trills and lip trills were similar in terms of the amplitude of vocal fold vibration. However, there were differences between tongue trills and vowel /ε/ phonation, as well as between lip trills and vowel /ε/ phonation, in terms of the amplitude of vocal fold vibration.

We measured the largest diameter of the region corresponding to the cuneiform cartilage as a reference to the maximum amplitude of vocal fold vibration in order to make the measurements more reliable. The individuals were their own controls, and the amplitude of vocal fold vibration was expressed as a fraction of the cuneiform cartilage size in order to reduce potential errors caused by movement of the laryngoscope and larynx during videolaryngostroboscopy.

Other studies have employed a similar methodology, having used the distance between the anterior commissure and the vocal process as a reference (Omori *et al.*, 1996; Tsuji *et al.*, 2003). We were unable to use that measurement in our study because we noted a change in glottal configuration during tongue trills, as reported in a study by Bueno (2006), who observed anteroposterior constriction of the laryngeal vestibule, without medialization or vibration of the vestibular vocal fold, during tongue trills.

The proximity of the distal end of the laryngoscope to the larynx can generate barrel distortion of linearity (Lee, 1980). In that type of distortion, the proportion of the measurements in the center of the image (represented in the present study as the vibration amplitude measurement) is smaller than is that of those in the periphery (represented as the cuneiform cartilage measurement) for the same millimetric measurement (Lee, 1980). Of the 220 images analyzed in the present study, 163 (74%) were indicative that the laryngoscope was far enough from the larynx to avoid that type of distortion. Of the images that were indicative that the laryngoscope was closer to the larynx and that could therefore show distortion, 21% (46/220) were obtained during the trill exercises and 5% (11/220) were obtained during the sustained vowel phonation task.

We believe that those distortions were not enough to interfere with the present study, given that the amplitude of vocal fold vibration was greater during the trill exercises than during

Lip trills versus sustained vowel /ε/ phonation 0.013 Tongue trills versus sustained vowel /ε/ phonation 0.139 Tongue trills versus lip trills 0.017

Lip trills versus sustained vowel /ε/ phonation 0.017 Tongue trills versus sustained vowel /ε/ phonation 0.013 Tongue trills versus lip trills 0.114

Measurement Pairs p\*

Table 10. Pairwise comparisons among phonation of the sustained vowel /ε/, lip trills, and tongue trills at high intensity in terms of the automatic measurements of significant

Our results show that, in general, tongue trills and lip trills were similar in terms of the amplitude of vocal fold vibration. However, there were differences between tongue trills and vowel /ε/ phonation, as well as between lip trills and vowel /ε/ phonation, in terms of

We measured the largest diameter of the region corresponding to the cuneiform cartilage as a reference to the maximum amplitude of vocal fold vibration in order to make the measurements more reliable. The individuals were their own controls, and the amplitude of vocal fold vibration was expressed as a fraction of the cuneiform cartilage size in order to reduce potential errors caused by movement of the laryngoscope and larynx during

Other studies have employed a similar methodology, having used the distance between the anterior commissure and the vocal process as a reference (Omori *et al.*, 1996; Tsuji *et al.*, 2003). We were unable to use that measurement in our study because we noted a change in glottal configuration during tongue trills, as reported in a study by Bueno (2006), who observed anteroposterior constriction of the laryngeal vestibule, without medialization or

The proximity of the distal end of the laryngoscope to the larynx can generate barrel distortion of linearity (Lee, 1980). In that type of distortion, the proportion of the measurements in the center of the image (represented in the present study as the vibration amplitude measurement) is smaller than is that of those in the periphery (represented as the cuneiform cartilage measurement) for the same millimetric measurement (Lee, 1980). Of the 220 images analyzed in the present study, 163 (74%) were indicative that the laryngoscope was far enough from the larynx to avoid that type of distortion. Of the images that were indicative that the laryngoscope was closer to the larynx and that could therefore show distortion, 21% (46/220) were obtained during the trill exercises and 5% (11/220) were

We believe that those distortions were not enough to interfere with the present study, given that the amplitude of vocal fold vibration was greater during the trill exercises than during

Intraindividual mean of the closed quotient

Intraindividual SD of the closed

**4. Discussion** 

\*Wilcoxon signed rank test

the amplitude of vocal fold vibration.

vibration of the vestibular vocal fold, during tongue trills.

obtained during the sustained vowel phonation task.

videolaryngostroboscopy.

quotient

values

the sustained vowel phonation task, even when the image was indicative that the laryngoscope was close to the larynx.

There were significant differences among phonation of the sustained vowel /ε/, lip trills, and tongue trills in terms of the maximum amplitude of vocal fold vibration at low and high intensities (p = 0.002 and p = 0.001, respectively). The Wilcoxon signed rank test revealed significant differences between lip trills and phonation of the sustained vowel /ε/, as well as between tongue trills and phonation of the sustained vowel /ε/.

For the performance of lip trills and tongue trills, pulmonary airflow has to increase in order to maintain vocal fold vibration and lip or tongue vibration (Titze, 2009; Warren *et al.*, 1992; Titze 1988, McGowan, 1992). The increase in airflow can lead to an increase in subglottic pressure, which can in turn lead to an increase in the amplitude of vocal fold vibration (Garrel *et al.*, 2008; Titze, 2009).

In the present study, during the measurement of amplitude, it was difficult to synchronize the flashes of light from the stroboscope with the vocal fold vibration during lip and tongue trills in some cases. We had no such difficulty when we measured the amplitude of vocal fold vibration during phonation of the sustained vowel /ε/. Therefore, during lip trills and tongue trills, we photographed only those moments of the cycles at which the maximum opening was visible.

Increased airflow can also lead to aperiodicity of vocal fold vibration (Jiang, 2001; Tao, 2007) and can destabilize the flashes of light from the stroboscope (Sercarz *et al.*, 1992). In previous studies, we found that the standard deviation of the intraindividual closed quotient was highest during lip trills and tongue trills, the values being higher at high intensities, which is due to increased airflow (Russell *et al.*, 1998; Alku, 2006). Those results might indicate differences in the periodicity of the EGG waveform during trill exercises and corroborate the abovementioned hypothesis.

Bueno (2006) observed pronounced movement of the laryngeal framework during tongue trills; such movement can make it difficult for the contact microphone to pick up the fundamental frequency. Because the cycles observed under the stroboscopic light are not the actual vocal fold vibration cycles, which require regularity in order to generate reliable images (Sercarz *et al.*, 1992; Patel *et al.*, 2008), a study involving a high-speed camera would be needed in order to have a better view of vocal fold vibration during tongue and lip trills. To that end, a device that is compatible with the laryngoscope is needed. Studies involving a qualitative analysis of both the periodicity of the EGG waveform and the amplitude of the EGG signal could also be useful in the investigation of the mechanism of vocal fold vibration during trill exercises.

The standard deviation of the closed quotient is a numerical representation of the variation in the closed quotient in the EGG waveform. Because the standard deviation of the closed quotient was higher during the trill exercises (at high or low intensity) in the present study, we can affirm that the variation in the closed quotient is greater during trill exercises than during phonation of the sustained vowel /ε/.

For tongue trills and lip trills to occur, the anterior part of the vocal tract has to be occluded by the tongue or the lips. Intraoral pressure increases and becomes greater than the force

Comparison Among Phonation of the Sustained Vowel /ε/, Lip Trills,

recommended by Aydos and Hanayama (2004) and Nix (1999).

*J Acoust Soc Am.* 120(2):1052-62. ISSN 0001-4966

the wavelike motion of the vocal folds.

muscles during the exercises are needed.

2000).

warranted.

**5. Conclusion** 

**6. References** 

terms of the closed quotient.

and Tongue Trills: The Amplitude of Vocal Fold Vibration and the Closed Quotient 145

Because patients with vocal fold pathologies present with structural changes (and, consequently, biomechanical changes), trill exercises probably produce different effects in those individuals than in individuals without morphological changes. According to the literature, bulging caused by vocal fold pathologies interfere with glottal flow resistance, glottal width, glottal area, and mean glottal volume velocity (Alipour & Scherer,

Studies involving vocal exercises (including tongue and lip trills) in various settings should be conducted in order to provide a deeper understanding of the physiology of vocal exercises in each of those situations and therefore assist speech-language pathologists in prescribing the exercises. The present study can support some of the theories that underlie the use of trill exercises in the clinical practice of speech-language pathology, as well as in the voice training of professional voice users. According to McGowan (1996), the variations that occur in the pharynx during lip trills can increase the force of mucosal vibration during

The greater amplitude of mucosal vibration and the higher standard deviation of the closed quotient during trill exercises reflect changes in the wavelike motion of the vocal folds; this can explain, at least in part, the improvement in voice quality (Rodrigues, 1995) after the use of those exercises, as well as warranting the use of trill exercises in patients with vocal fold pathologies, such as nodules (Bueno, 2006), edema (Pinho e Pontes, 2008), and sulci. The probable need for airflow control and the source-filter interaction caused by the articulatory oscillation warrant the use of trill exercises during the training of professional voice users, as

However, further studies are needed in order to provide a deeper understanding of the effects of trill exercises on the vocal fold mucosa. To that end, a more in-depth analysis of the EGG waveform and, if possible, videolaryngostroboscopy with a high-speed camera are

For a better understanding of the effect of trill exercises and their indications, studies analyzing the blood flow in the region and the mechanics of laryngeal and vocal tract

On the basis of our results, we can conclude that, in professional singers, the maximum amplitude of vocal fold vibration is greater during lip and tongue trills than during natural phonation. In addition, we observed considerable variation in the closed quotient during the exercises employed. Lip trills differed from tongue trills only at higher intensities and in

Alku P.; Airas M.; Bjorkner E. & Sundberg J. (2006) An amplitude quotient based method to

analyze changes in the shape of the glottal pulse in the regulation of vocal intensity.

that maintains the anterior part of the vocal tract closed; therefore, the anterior part of the vocal tract opens and is "sucked out" by the speed of the airflow (McGowan, 1992), closing the vibration cycle, as occurs in the vocal fold vibration model. Therefore, supraglottic pressure and, consequently, vocal tract impedance oscillate.

The theory of source-filter interaction (Titze, 2008; Titze *et al.*, 2008) states that the acoustic pressure in the vocal tract changes the phonation threshold pressure and interferes with vocal fold vibration. The theory of source-filter interaction states that, when the vibratory motion of the point of articulation makes the acoustic pressure in the vocal tract oscillate, the phonation threshold pressure also oscillates, which might explain why the standard deviation of the closed quotient was higher during trill exercises in the present study.

According to Titze (2006), the objective of voice training is to promote the interaction between the source and the filter and therefore increase vocal intensity, efficiency, and economy (Titze, 2006). According to the author, lip trills and tongue trills are among the semi-occluded vocal tract exercises. Because semi-occluded vocal tract exercises promote a mechanical interaction between the source and the filter (Titze, 2008; Titze *et al.*, 2008) they change vocal fold impedance and therefore inhibit vocal fold vibration (Story *et al.*, 2000).

Vocal tract pressure during voiced fricatives must be constant, which distinguishes voiced fricatives from the exercises analyzed in the present study. New clinical studies comparing the two types of exercises should be conducted.

In our studies, the mean closed quotient was highest during lip trills at high intensity, which distinguished lip trills from tongue trills and phonation of the sustained vowel /ε/. According to the literature (Story *et al.*, 2000; Titze, 2006), a more anterior obstruction translates to greater vocal tract impedance. Although lip trills are slightly more anterior than are tongue trills, the influence of supraglottic and subglottic pressure on vocal fold vibration is not linear (Zhang, 2009; Titze, 2008; Hatzikirou *et al.*, 2006, Titze, 2008), meaning that the airflow changes caused by increased intensity can lead to different vocal fold vibration proportions and differentiate between high- and low-intensity vibrations (Tao *et al.*, 2007; Becker *et al.*, 2009).

Gaskill and Erikson (2008) found systematic differences between the closed quotient of lip trills and that of phonation of the sustained vowel /ε/; as we did in the present study, the authors argued that those differences might be due to the interaction between the source and the filter. In addition, the authors found differences between trained and untrained individuals in terms of the results obtained; the differences were attributed to the fact that trained individuals have better control over glottic closure.

Some authors have reported that the sound produced by the larynx is not linear (Jing *et all,* 2001) and depends on numerous factors. Therefore, any difference in biomechanics, structure (such as tissue geometry, density, and viscosity), airflow control, or vocal tract control can cause differences in vocal fold vibration. Trained individuals have more control over those factors, and this can actually result in differences between trained and untrained individuals, as well as between trained individuals and patients with morphological changes in the vocal folds, in terms of vocal fold vibration during a given phonation task.

Because patients with vocal fold pathologies present with structural changes (and, consequently, biomechanical changes), trill exercises probably produce different effects in those individuals than in individuals without morphological changes. According to the literature, bulging caused by vocal fold pathologies interfere with glottal flow resistance, glottal width, glottal area, and mean glottal volume velocity (Alipour & Scherer, 2000).

Studies involving vocal exercises (including tongue and lip trills) in various settings should be conducted in order to provide a deeper understanding of the physiology of vocal exercises in each of those situations and therefore assist speech-language pathologists in prescribing the exercises. The present study can support some of the theories that underlie the use of trill exercises in the clinical practice of speech-language pathology, as well as in the voice training of professional voice users. According to McGowan (1996), the variations that occur in the pharynx during lip trills can increase the force of mucosal vibration during the wavelike motion of the vocal folds.

The greater amplitude of mucosal vibration and the higher standard deviation of the closed quotient during trill exercises reflect changes in the wavelike motion of the vocal folds; this can explain, at least in part, the improvement in voice quality (Rodrigues, 1995) after the use of those exercises, as well as warranting the use of trill exercises in patients with vocal fold pathologies, such as nodules (Bueno, 2006), edema (Pinho e Pontes, 2008), and sulci. The probable need for airflow control and the source-filter interaction caused by the articulatory oscillation warrant the use of trill exercises during the training of professional voice users, as recommended by Aydos and Hanayama (2004) and Nix (1999).

However, further studies are needed in order to provide a deeper understanding of the effects of trill exercises on the vocal fold mucosa. To that end, a more in-depth analysis of the EGG waveform and, if possible, videolaryngostroboscopy with a high-speed camera are warranted.

For a better understanding of the effect of trill exercises and their indications, studies analyzing the blood flow in the region and the mechanics of laryngeal and vocal tract muscles during the exercises are needed.
