Concerning perceptual evaluation, voices were assessed through the GRBAS scale (41,42) by the phoniatrician and the external experienced judges. Mean values of the perceptual ratings were considered. The GRBAS scale is a widely-used perceptual evaluation tool specifically addressing the following parameters of voice quality: grade (G), roughness (R), breathiness (B), asthenicity (A) and strain (S). Each parameter scores from 0 (normal voice) to 3 (severe dysphonia). Lastly, for the acoustic analysis, the Multi Dimensional Voice Program (MDVP) (43) was used to measure a series of acoustic parameters of the voice quality, such as microperturbations (Jitter, RAP, PPQ, Shimmer, APQ), macroperturbations (SPPQ, sAPQ) and variations of the fundamental frequency (F0), the noise parameters (NHR, VTI, SPI), tremble parameters (FTRI, ATRI) and the subharmonic and broken voice parameters (DVB, DSH, DUV). However, only three of these parameters, the mean fundamental frequency (F0), the frequency variation from cycle to cycle (Jitter) and amplitude variation of the sound wave (Shimmer), were considered as measurements of interest due to their connection with the sound wave quality. The test was carried out by the speech therapist in a soundproofed room (<30 dB). Each participant’s voice was recorded with a Rode® NT1-A microphone in standard conditions, with a mouth-to-microphone distance of 30cm and constant gain. The microphone was connected via USB to a computer with a Windows operating system running Sound Forge® version 9.0 software. Participants were asked to produce a sustained vowel /a/. A mid-vowel selection of 3 seconds was performed for the acoustic analysis (they were asked to repeat the vowel three times but only one vowel was analyzed chosen by the evaluator by his perception)
In the treatment group, each participant was given proper instructions through a tutorial, presented on a 21-inch PC monitor running Experiment Builder® software, and the speech therapists supervised and guided the proper execution of the treatment task that lasted for ten minutes. Previous research showed that vocal warm-up duration is considered to last 10 minutes in order to have some voice perceptual changes (44,45). Note that after 10 minutes performing a task, attention declines (46,47). In the Gauze group, participants had to use sterile gauze (non-woven swabs). They had to soak the sterile gauze in still water at a temperature of around 25°C inside a glass. In order to remove excess water, participants had to squeeze the gauze, and after that, they had to place the damp gauze around the nose, completely covering the nostrils (see Appendix 1). They were then asked to breathe normally through the nose wrapped in the damp gauze and to produce some vocal warm-up exercises (i.e. sustained /i/ and /u/ vowels at different volumes and tones). The following sequence tasks were proposed: breathe normally through the nose (three times); perform an /u/ vowel at a low pitch and high volume as comfortable as possible, sustaining this for three seconds (this helps to create a large and wide mucosal wave); breathe normally through the nose (three times) and perform an /i/ vowel in an ascending-descending glissando one time (this helps to lengthen and narrow the vocal folds). The sequence task lasted around one minute, so participants could repeat the sequences 10 times approximately. Participants were instructed to dampen the gauze after each sequence (lasting for about one minute) in order to always have a wet gauze. They had to repeat the described sequence for ten minutes.
In the Gauze group, vocal warm-up exercises were included to improve the effect of water penetration into the mucosal layers. This aspect was taken into account in order to observe whether some effects were influenced by the vocal warm-up exercises and not simply by the surface hydration itself. Therefore, two more experimental groups were included and compared: the Exercise group, and the Control group.
In the Exercise group, participants had to produce the same vocal warm-up exercises as described above for ten minutes, without the damp gauze. Participants were asked to breathe normally through the nose (three times), to perform a low pitch and high volume /u/ vowel (sustained for three seconds); to breathe normally through the nose (three times) and to perform a sustained /i/ vowel in ascending-descending glissando (one time).
In the Control group, participants had to breathe normally through the nose (three times) and respond to some questions in a normal speaking voice (e.g. “What did you do yesterday?” or “Count from 1 to 20”) for ten minutes.
Statistical analysis
Statistical analysis was carried out with Jamovi software (version 0.9.6.7 for Windows). Mean and standard deviations (SDs) for initial self-assessment and the three tests (laryngostroboscopic, acoustic and perceptual analysis) were calculated. The normality of the distributions was assessed with the Shapiro-Wilk test. An alpha of 0.05 was considered for statistical procedures. Although data did not have normal distributions, analyses were carried out with parametric tests. Non-parametrical analysis is recommended when data are not normal distributed. However, recent literature has shown that a parametrical test, such as ANOVA, was robust in terms of good control of Type 1 errors in non-normal distribution.
Initial self-assessment was used to control the homogeneity of the sample. The different assessments were analyzed separately in order to observe whether participants from the three treatment groups had similar parameters of vocal wellbeing. To do so, one-way ANOVAs were carried out by the three treatment groups.
The inter-rater reliability of the subjective measurements were analyzed through Kappa statistics (48) to observe the reliability between the three external judges’ ratings and the phoniatrician’s rating. In case of excellent agreement between the three external judges’ and the phoniatrician’s ratings, the analysis of the dependent variable was carried out averaging the four ratings, i.e., those of the phoniatrician and the three external judges. Note that the Kappa value above 0.75 reflects excellent agreement (+1 indicates perfect agreement and 0 chance).
The eleven dependent variables (Glottal Closure, Amplitude of mucosal wave and Maximum Opening of the glottic space, Fo, Jitter and Shimmer, G – Global grade of dysphonia -, R – Roughness -, B – Breathiness -, A – Asthenicity -, S – Strain -) were examined separately to predict how each dependent variable could be influenced by direct hydration. Each dependent variable was analyzed separately through two-way ANOVAs with the within-subjects predictor defined as Test Moment and the betweensubjects predictor defined as Treatment Groups. To follow up possible interactions for each dependent variable, consecutive analyses were performed. In order to observe whether the three groups performed equally in the Pre-Test, a one-way ANOVA was carried out. Also, a one-way ANOVA of the Post-Test condition was conducted to assess if significant differences between the three groups could be observed after the treatment task. Then, unpaired-sample t-tests (i.e., if the samples were dependent between them) were carried out to compare the three treatment groups in the PostTest condition and observe which group might modify the vocal quality and the mucosal wave.