Preliminary Investigation of Four Voice Therapy Concepts in a Single Subject Design with Four Professional Teachers

Abstract

Purpose:

The Global Voice Prevention and Therapy Model (GVPTM) with the Estill Voice Model (EVM) involves four concepts; 1) training multiple new voice targets to meet all the vocal demands of the client, 2) using an integrated implicit-explicit instructional approach, 3) switching between voice targets, and 4) facilitating voice practice in a bottom-up speech hierarchy. As a preliminary investigation into the four concepts, the current study compared the GVPTM with EVM to a modified version of the GVPTM (MGVPTM) with EVM that eliminated concepts 1 (only one new resonant voice was trained) and 3 (no switching between voice targets) and kept concepts 2 and 4.

Methods:

Four professional teachers were randomized into one of two conditions in a single subject design counterbalanced with alternating treatments and baselines (i.e., A₁BA₂CA₃ or A₁CA₂BA₃). In condition 1 (A₁BA₂CA₃), the GVPTM was first followed by MGVPTM. In condition 2 (A₁CA₂BA₃), the MGVPTM was first followed by the GVPTM. Baseline measures included fundamental frequency (f_o), participant self-ratings of overall voice quality, roughness, and strain, the Voice Handicap Index (VHI)-10, Vocal Fatigue Index (VFI), survey questions, and answers to questions via an exit interview.

Results:

Results indicated significant differences for f_o and self-ratings of overall vocal quality, roughness, and strain from pre to post for the GVPTM and MGVPTM. VHI-10 and VFI factors, 1, 2, and 3 yielded no differences from pre to post. Overall, the participants viewed all four concepts in the GVPTM favorably and suggested it was limiting to learn only one new resonant voice with no switching between voice targets in the MGVPTM.

Conclusions:

Speech-language pathologists may consider using all four concepts in voice therapy sessions represented via targets, ingredients, mechanisms of action of the Rehabilitation Treatment Specification System.

Introduction

Voice problems have been estimated to be present in 11% (29.92 million) of the U.S. adult population¹ and limit a person’s ability to participate in society. Teachers, one of the country’s largest employed groups, have the highest risk among common occupations for developing voice problems.² Estimates indicate that nearly 60% of teachers are affected by voice problems over the course of their careers.³ Overall, teachers reported a significantly higher number of days in which they intentionally reduced their interactions because of voice problems and as many as 20% of teachers reported they missed work days due to their voice.⁴ The need to restrict or adjust teaching activities as a result of voice problems has implications for the quality of teaching as well as the students’ learning experience. In fact, some reports indicate that students’ cognitive functioning is reduced for materials delivered orally by teachers who are only mildly dysphonic.^5,6 Due to the COVID-19 pandemic, teachers in schools were required to wear face masks putting an additional strain on the phonation system. In fact, individuals who wore face masks for professional and essential activities had increased perceptions of vocal fatigue, discomfort, and vocal effort.⁷ Voice disorders in teachers represent a significant public health concern; therefore, studies have addressed the prevention and treatment of voice disorders in professional teachers^8–22 and student teachers.^23–28 Student teaching involves working with a cooperating professional teacher and leading all teaching activities 5-days a week typically across a 14- or 15-week semester, making student teachers the perfect test case for studying prevention of voice disorders.

Subsequently, initial work addressed prevention of voice disorders in 82 student teachers via the in-person and telepractice Global Voice Prevention and Therapy Model (GVPTM) with the Estill Voice Model (EVM) compared to a control condition.²⁹ The in-person and telepractice GVPTM conditions included vocal hygiene, education, and training. The control condition only facilitated vocal hygiene and education. Voice improvements were noted in fundamental frequency (f_o), noise-to-harmonic ratio (NHR), and jitter% for both the in-person and telepractice conditions, but not for the control condition. The in-person and telepractice GVPTM conditions facilitated an improved resonant voice as evidenced by an increase in f_o from pre to post that was maintained during 14 weeks of student teaching. NHR and jitter% for the in-person and telepractice GVPTM conditions were significantly lower at post as compared to pre during fall, while the control condition stayed the same or even increased at post. A lower score for NHR and jitter% is better indicating less vocal fold variability which was facilitated with the resonant voice achieved through vocal training in the in-person and telepractice GVPTM conditions. Additionally, the Voice Handicap Index-10 (VHI-10) and Vocal Fatigue Index (VFI) factors 1 and 2 did not demonstrate any significant differences across the conditions from pre to post in the fall. During student teaching in the spring, however, the VHI-10 and VFI did document an increased negative impact on voice-related Quality-of-Life (QoL) and vocal fatigue in the control condition as compared to the in-person and telepractice GVPTM.

The overall success of the in-person and telepractice GVPTM for student teachers as demonstrated by improving voice quality, voice-related QoL, and vocal fatigue may be due to four major concepts within the GVPTM.²⁹ The concepts include: (1) training multiple voices to meet all the vocal demands of the client, (2) connecting auditory-perceptual prompts (implicit) to anatomy and physiology of the voice production system (explicit) in an integrated implicit-explicit instructional approach, (3) switching between multiple voices at each level of the hierarchy including the old voice (i.e., the voice the client was using before voice therapy), and (4) using a bottom-up speech hierarchy during voice training. Concepts 1 and 2 are not widely represented in the literature. For example, most voice therapy models focus on training only one new voice with no integrated training of implicit auditory-perceptual prompts (e.g., humming, loud voice) with explicit anatomy and physiology that is hypothesized to create the new voice (e.g., True Vocal Folds (TVF): Body-Cover Thin for humming and Aryepiglottic Sphincter (AES) Narrow for loud voice).^30–38 Three exceptions include an implicit-explicit voice therapy approach described by Tellis^39,40, the GVPTM with EVM,²⁹ and VoiceCraft Sob Voice Therapy.^41,42 Concept 3 practicing multiple voices, including the old voice, at each level of the hierarchy was originally described in the Global Voice Therapy Model⁴³ and has since been applied to other voice therapy models in the literature.^33,41,44 Concept 4 involving a bottom-up speech hierarchy is described in many evidence-based voice, fluency, and articulation therapy models in speech-language pathology.^{29,35,37,38,43,45–50}

Historically, clinical research involving efficacy of voice therapy has primarily focused on changes in outcome measures from pre to post with little to no focus on what, why, and how a clinician’s treatment theory leads to changes in client functioning. The Rehabilitation Treatment Specification System (RTSS) has been described as a proposed solution to this problem.^77–80 The RTSS offers a framework that specifies a singular target (i.e., why the client’s aspect of functioning is directly targeted for change), ingredients (i.e., what the clinician does to facilitate the change in client functioning), and mechanisms of action (MoA) (i.e., how the ingredients are expected to influence the target) as one treatment component. A clinician’s therapeutic approach may involve multiple treatment components. A target may include increasing habitual use of a loud voice (i.e., twang) to meet the client’s communication needs when talking over noise of the classroom. Ingredients may include opportunities to practice loud voicing (i.e., twang) in a speech hierarchy, providing feedback on the client’s practice, and discussing how the client can be capable and motivated to perform the practice. MoAs can be known and measurable or hypothesized and not directly measured during routine clinical care. For example, learning by doing and hypothesized changes in aryepiglottic sphincter narrowing to produce a “twangy” vocal quality connect a target of increased habitual use of loud voice via twang with ingredients of providing opportunities for the client to practice voicing in a bottom-up speech hierarchy given clinician feedback.

One goal of applying the RTSS framework is to determine the treatment components (i.e., singular target, ingredients, and MoAs) that are shared generally, across speech-language pathology therapy models and more specifically, across voice therapy models described in the literature. Subsequently, investigations can assess efficacy and effectiveness of these shared treatment components with recommendations for which ones should be present in every therapy session regardless of the specific model. Prior work has defined a methodology to identify and describe unique targets and ingredients in voice therapy^51,52 The four concepts present in the GVPTM will be described considering that methodology.

Concept 1 relates to vocal function targets to increase habitual use or performance accuracy of a new voice quality.^51–52 The clinician and the client determine the voice quality targets that will be trained during therapy, addressing all the vocal needs of the client. Determining the multiple voice quality targets can occur during the voice assessment⁵³ through the initial client interview. For example, teachers are professional voice users who report that they need to communicate in a variety of contexts; therefore, training multiple voice quality targets should be the focus of voice therapy. In the current study with professional teachers, the GVPTM included multiple voice targets: new resonant voice in connected speech, falsetto voice for quiet talking, twang voice for talking over noise of the classroom, and belt voice for healthy yelling.

Concept 2 relates to ingredients and MoAs of the RTSS framework.^51,52 Ingredients are the clinician actions taken to achieve a target. MoAs are the hypothesized or observed ways the ingredients affect a target. For example, concept 2 can be used as an ingredient in opportunities to practice voicing^51,52 via an integrated implicit-explicit approach linking implicit auditory-perceptual prompts with explicit training of the vocal anatomy and physiology that are relevant for accurate production of the prompt. The integrated implicit-explicit approach can also be considered a volitional ingredient to increase a client’s capability, opportunity, and motivation to perform a behavior. For MoA, applying explicit anatomy and physiology to implicit auditory-perceptual prompts describes what is occurring or hypothesized to be occurring within the laryngeal, vocal tract, and support structures, which may lead to the desired change in client functioning (i.e., target). In the GVPTM, EVM is used to link implicit prompts like “moh-you” for resonant voice, “beep-beep” for twang voice, and “Yay!” for belt voice with client training of the relevant Estill anatomical Figures and physiological Options that are hypothesized to create the prompt.²⁹ For example, False Vocal Folds (FVF) Figure and the Option Retract for resonant voice, AES Figure and the Option Narrow for twang voice, and Cricoid Figure and the Option Tilt for belt voice. For an in-depth explanation with clinical case studies and video demonstrations of concept 2 in the RTSS, please see Grillo et al.⁵⁴

Concept 3 involving practicing multiple voices, including the old voice, at each level of the speech hierarchy could be considered a vocal function target related to voice quality to increase performance accuracy of switching between the various voices.^51,52 Concept 3 could also be considered an ingredient for opportunities to practice voicing in a practice schedule of alternating between the voices.^51,52 Related to the MoA of concept 3, the hypothesis is that switching between all the voices at multiple levels of the speech hierarchy facilitates generalization and maintenance of the target voices while learning how to rescue oneself from the old voice. First, the GVPTM facilitates a blocked practice schedule of the new voice quality target with feedback offered by both the clinician and client less often after every 10^th trial at each step of the hierarchy. Blocked practice schedules with less feedback have been demonstrated to be effective in the acquisition of newly learned voice motor skills.⁵⁵ After the new voice is acquired at a specific step, then a random practice schedule, switching between multiple voices, is introduced with feedback offered by both the clinician and the client more often, after every 5^th trial. Random practice schedules with more feedback have been demonstrated to be effective in promoting learning of a new voice motor skill.⁵⁵

Concept 4 using a bottom-up speech hierarchy is an ingredient of RTSS involving opportunities to practice voicing from facilitator syllables/words, phrases, sentences, etc., all the way up to conversation.^51,52 The MoA of concept 4 involves the hypothesis that practicing voice targets from the smallest unit of utterance with the least amount of cognitive complexity (e.g., hum-hum) to the largest unit of utterance with the most amount of cognitive complexity (e.g., monologue or conversation) will facilitate improved generalization and maintenance of target voice qualities. The GVPTM includes the following bottom-up speech hierarchy; facilitator syllables/words, automatic speech, words, functional phrases, sentences, memorized speech acts (e.g., pledge of allegiance for the USA, nursery rhyme, song lyrics spoken not sung, etc.), specific spontaneous speech acts (e.g. tell me how to make a peanut butter and jelly sandwich, describe your bedroom, list transportation directions from home to work, etc.), monologue (e.g., tell me what you did last weekend), and conversation.^29,43

As a preliminary investigation, the current study involved two voice therapy models (i.e., GVPTM with EVM and a modified GVPTM with EVM) in a single subject design (SSD) with alternating treatments and multiple baselines to assess changes from pre to post in acoustic, participant self-ratings of perceptual measures, voice-related QoL, vocal fatigue, survey responses, and open-ended answers during an exit interview. The alternating treatment SSD allows a direct comparison of two treatment models within one participant offering valuable information that is not available in group designs with one participant receiving only one treatment. To the author’s knowledge, the current study is the first to assess voice therapy models using SSD with alternating treatments and multiple baselines.

Professional teachers with no diagnosed voice disorders were chosen because they are aware of the vocal demands of teaching yet have not experienced a voice disorder; therefore, offering insight into prevention. The GVPTM with EVM has all four concepts present; therefore, the GVPTM was compared to a modified version of the GVPTM (MGVPTM) with EVM keeping concepts 2 and 4, while eliminating concepts 1 and 3. The GVPTM trains multiple voices to meet all the vocal needs of the client (concept 1) using an integrated implicit-explicit instructional approach (concept 2) switching between all the voices (concept 3) at each step of the bottom-up treatment hierarchy (concept 4). MGVPTM facilitated only a new resonant voice for connected speech (eliminated concept 1) using an integrated implicit-explicit approach (kept concept 2) with no practice of switching between the new and old voices at each step of the hierarchy (eliminated concept 3) in a bottom-up speech hierarchy (kept concept 4). The hypothesis was that the GVPTM with all four concepts would have improved results from pre to post as compared to the MGVPTM and that the participants would prefer the GVPTM over the MGVPTM.

Methods

Participants

Professional teachers from the West Chester Area School District (WCASD) and Downingtown Area School District (DASD) were recruited to participate in the study, which was approved by West Chester University’s (WCU) Institutional Review Board. The districts were chosen because they collaborate frequently with WCU and are in the same area of Pennsylvania, USA. First, recruitment procedures included seeking permission from the Superintendent of each school district to contact school Principals and second, emailing the Principals to share the recruitment information with all the professional teachers employed within their school. A total of 33 school Principals disbursed study information to teachers across 21 elementary, six middle, and six high schools from end of August 2022 to middle of September 2022.

Following the recruitment process, 12 professional teachers were interested in the study. All 12 completed informed consent, screening questions, and were enrolled. All participants who provided consent met the following inclusion criteria; professional teacher, owner of either an android or iOS smartphone, no current voice disorder diagnosis, and not currently receiving behavioral voice therapy. Four of the 12 participants completed the current study comparing the GVPTM to the MGVPTM. The remaining eight participants completed an additional study that will be described in a future article. The four participants (average age of 37 years with a range of 30–59 years) completed all aspects of the study from September 19 to December 9, 2022. From the four, three were cisgender females and one was a cisgender male. All participants were not Hispanic or Latino and self-identified as White. The four participants ranged from 3 to 10 years of professional teaching representing early grades (kindergarten - 5th), middle grades (6th - 8th), secondary grades (9th - 12th), music education, and family and consumer science (see Table 1). Two of the four participants did not report any current voice issues at the time of screening, while the other two reported voice issues ranging from vocal fatigue, weakness, and tiredness (see Table 1). Three of the four participants reported singing or acting experiences ranging from choral singing, voice lessons during college, and singing and acting in musical theater. None of the participants reported being diagnosed with a voice disorder and none of the participants received voice therapy with a speech-language pathologist (SLP).

Table 1.

Number of Participants by Condition (i.e., Global Voice Prevention and Therapy Model [GVPTM] and Modified GVPTM [MGVPTM]), Cisgender (i.e., female and male), Area of Teaching (i.e., early grades [EG], middle grades [MG], secondary grades [SG], music education [ME], family and consumer science [FCS]), Years of Teaching, Current Voice Issues, and Experiences with Singing or Acting.

	Condition 1	Condition 2
	GVPTM 1st	MGVPTM 1st
	MGVPTM 2nd	GVPTM 2nd
Cisgender	1 female 1 male	2 females
Area	MG FCS & SG ME	EG ME & MG ME
Years of Teaching Current Voice Issues	3 years & 10 years - “My voice feels weaker as I have gotten older. My voice gets very fatigued at the end of a day when I’m giving a lot of new instruction.” - “It feels tired whenever I need to speak a lot.”	7 years & 9 years No current issues
Experience with Singing or Acting	No Yes- sang in choirs as a child and in college.	Yes for both participants. Studied music education in college with voice as the focus. One participant acts and sings in musical theater for fun.

Procedures

Following the informed consent process and screening, each participant was randomly assigned into one of two conditions, oriented to the Desire 2 Learn (D2L) academic computing website, and trained with the VoiceEvalU8 app via Zoom videoconferencing during a one-time, 45-minute meeting. Each of the two conditions involved two voice therapy models in a single subject design counterbalanced with alternating treatments and baselines (i.e., A₁BA₂CA₃ or A₁CA₂BA₃). In condition 1 (A₁BA₂CA₃), the GVPTM was first followed by MGVPTM. In condition 2 (A₁CA₂BA₃), the MGVPTM was first followed by the GVPTM. Two participants completed condition 1, while two participants completed condition 2.

After assignment into one of the two conditions, the participants were oriented to the D2L site and trained in the use of the VoiceEvalU8 app. For orientation, participants logged into D2L and the author introduced them to the homepage, content, assignments, and how to navigate the course site. The VoiceEvalU8 training included downloading the app to the participant’s android or iOS smartphone, teaching the participant to measure @4cm from microphone to mouth with index and middle fingers together for all voice recordings, using the app in a quiet room to minimize environmental noise, and running through an app session to review acoustic and perceptual voice measures and procedures for recording (see the videos on https://mobilevoiceapps.com/voiceevalu8/voiceevalu8-videos for recording procedures).

Following orientation to D2L and training of the VoiceEvalU8 app, all four participants completed the first baseline via the VoiceEvalU8 app across 5 days (Monday-Friday) with twice a day logs in the morning from 6–11 a.m. and in the evening from 4–11 p.m. The app sent email and push notifications every 30-minutes in the morning and every 60-minutes in the evening to remind participants to complete the logs. Once the logs were completed for a specific day and time, then the notifications stopped. The app collected f_o in three trials of sustained /a/ for 5 s, the phrase “we were away a year ago,” and connected speech for 15 s; participant self-ratings of perceptual voice measures; VHI-10; and VFI. Acoustic and participant self-ratings of perceptual voice measures were collected twice a day for 5 days, while the VHI-10 and VFI were each collected once over the 5-day period.

After the first baseline was collected, participants began the first voice therapy model delivered via telepractice across four weekly sessions, each lasting 45–60 minutes. The GVPTM and MGVPTM included both asynchronous and synchronous telepractice. The asynchronous component was completed via D2L and included self-study of vocal hygiene and education over one week. Participants completed discussion boards, quizzes, built a paper larynx, and watched videos created by the author. The videos included information about ways to keep the voice healthy (e.g., water intake, vocal naps, etc.) and vocal education (e.g., power, source, filter, and introductory information of relevant Estill Figures and Options). All asynchronous learning was completed before synchronous vocal training for GVPTM and MGVPTM. Synchronous telepractice involved vocal training via Zoom videoconferencing with graduate speech-language pathology students facilitating the voice therapy models with the participants under 100% supervision from the author, who is a licensed and certified SLP. GVPTM and MGVPTM had three weekly voice training sessions lasting 45–60 minutes. The author trained the graduate student clinicians over spring and summer in preparation for working with the professional teachers during fall. In spring, the students took the author’s Graduate Voice Disorders in which they practiced administering the GVPTM and EVM in Zoom Cloud recording assignments and in a voice therapy standardized patient simulation. In addition, the students spent the spring practicing EVM and participated in a pass/fail EVM assignment that required them to produce each Estill Figure with relevant Options and the six Estill Voice Qualities. Over the summer, the students practiced delivering both the GVPTM and MGVPTM via telepractice with volunteers from the community across four weeks with 100% supervision by the author. The students used the RTSS framework to deliver both voice therapy models. Please see the supplemental materials for the detailed approach to each of the voice therapy models in the RTSS framework.

The VoicePractice app was used to facilitate daily voice practice between weekly synchronous sessions. The daily practice occurred Monday-Friday twice a day. The time of day was dependent upon the schedule of each participant. Twice daily practice sessions were available for 2-hour increments. Participants received push and email notifications to complete each practice session. Once completed, the notifications stopped. The specific anchors (voice) and tasks practiced depended upon the specific voice therapy model; therefore, the practice sessions matched the targets, ingredients, and MoAs of each of the voice therapy models.

After conclusion of the first voice therapy model, participants completed the second baseline using the same method and measures described earlier with the VoiceEvalU8 app. Following the second baseline, the participants began the second voice therapy model, which continued for four weeks. At the conclusion of the second voice therapy mode, participants completed the third baseline using the VoiceEvalU8 app again with the same measures and procedures described earlier. At the third baseline, participants also completed a survey via Qualtrics, an online survey tool, with questions related to the two voice therapy models. In addition, an exit interview was held with the participants via Zoom during the last synchronous voice therapy session. The graduate student clinicians asked open-ended questions related to the two voice therapy models completed by each participant. Figure 1 provides a flow diagram of the procedures.

Figure 1.

Flow Diagram

All participants were compensated with an Amazon gift card. For baselines 1–3, the participants were paid $25 for each baseline using the VoiceEvalU8 app twice daily for 5 days totaling $75. In addition, participants were paid $100 for each voice therapy model totaling $200. The total compensation was $275. If participants missed app sessions, synchronous voice training sessions, or asynchronous assignments, then $1-$5 was deducted from the overall payment. Participants did miss some app sessions, but none of the participants missed voice training sessions or vocal hygiene and education D2L assignments.

Outcome Measures

The outcomes measures included f_o, participant self-ratings of perceptual voice measures, the VHI-10,⁵⁶ VFI,⁵⁷ survey questions, and answers to questions via an exit interview. f_o was captured twice daily for 5 days (Monday-Friday) in three trials of sustained /a/ for 5 s, the phrase “we were away a year ago,” and a 15 s connected speech sample via the VoiceEvalU8 app. In all recordings, the participants were instructed to be in a quiet room and measure mouth to microphone distance of 4cm with index and middle finger together, and to use their everyday speaking voice. The files for each utterance were recorded at a sampling rate of 44,100 Hz and saved as .wav files. The VoiceEvalU8 server analyzed the acoustic files through Praat algorithms version 6.0.19, a free voice analysis software program on the web.⁵⁸

The participant self-ratings of perceptual voice measures included 1) overall vocal quality (OVQ), 2) roughness (R), and 3) strain (S) on a scale from 0–100 modeled after the Consensus Auditory-Perceptual Evaluation of Voice.⁵⁹ The numbers corresponded with 0 as no problem, 10 as a mild problem, 35 as a moderate problem, and 75 as a severe problem. Participants were able to slide a square box on the line from 0–100 that represented their current perceptions for each of the three measures in the VoiceEvalU8 app twice a day for 5 days (see Figure 2 for an example). The VHI-10 and VFI were collected via the VoiceEvalU8 app once a week rather than twice a day so that participants were not desensitized to the scales. The VoiceEvalU8 app presented an overview of the scale with the first question and the rubric. Each following question and rubric was presented on a separate screen. The participant rated the question and then clicked “save and next” to move to the next question (see Figure 3 for an example). The survey was conducted via Qualtrics after participants completed both voice therapy models and the questions were asked by the graduate student clinicians during the exit interview, which occurred during the last voice therapy synchronous session.

Figure 2.

Overall Vocal Quality (OVC) Rated in the VoiceEvalU8 App

Figure 3.

Voice Handicap Index (VHI)-10 First and Second Question in the VoiceEvalU8 App

f_o, VHI-10, VFI were outcome measures that were used in prior work with student teachers and were found to capture significant changes in voice quality, voice-related QoL, and vocal fatigue across a semester of student teaching (i.e., 14 weeks).²⁹ In addition, other voice prevention and therapy programs, that have focused on training a resonant voice, used f_o as an outcome measure and demonstrated significant differences in f_o at post^62–64. Qualitative information regarding participant self-ratings of OVQ, R, and S along with survey questions related to the voice therapy models followed by exit interviews were deemed important to capture relevant insights into the participant perceptions of their voice and the two voice therapy models.

Statistical Analysis

Data analysis was conducted using SPSS (Version 29; IBM, Armonk, NY). The dependent variables (DV) were the outcome measures described in the above section. Independent variables (IV) were condition (i.e., conditions 1 and 2) and baseline (i.e., baselines 1–3). Cisgender females and cisgender males have different anatomy of the vocal tract and vocal folds, which impacts analysis of acoustic measures; therefore, female and male acoustic measures were analyzed separately. For all other outcome measures, the female and male participants were analyzed as a group to assess the effects of the different conditions across baseline. For f_o in females, f_o in males for the 5 sec /a/s, and participant self-ratings of perceptual voice measures, two-way analysis of variances (ANOVA) were used with an alpha level of 0.05 due to the recommendation by Kazdin⁸¹ of at least 20–30 observations or data points per phase for quantitative analysis in SSDs (see Table 2). For post-hoc testing, the Bonferroni method was applied only for significant main and interaction effects. Version 29 of SPSS automatically adjusts alpha level for multiple pairwise comparisons in post-hoc testing. For the ANOVAs, effect sizes were reported to determine the strength of the relationship for the main effects and interactions (e.g., partial eta squared). Similar to the Cohen’s d effect size, the thresholds correspond to a large (η_p² = 0.14), medium (η_p² = 0.06), and small (η_p² =0.01) effect size for partial eta-squared.⁶⁰ For f_o in males for the phrase and speech, VHI-10, VFI, and survey questions, descriptive statistics were used due to <20 data points across the baselines⁸¹ (see Table 2). For the exit interview questions, frequency of responses was recorded related to the four concepts investigated in the current study.

Table 2.

Number of Data Points for Consideration of Quantitative vs Qualitative Statistical Analysis across the Three Baselines.

	Baseline 1	Baseline 2	Baseline 3	Total
fo (Hz) Cisgender Females
5-sec /a/	80	75	67	222*
Phrase	27	25	23	75*
Speech	27	24	23	74*
fo (Hz) Cisgender Males
5-sec /a/	30	30	24	84*
Phrase	10	10	8	28+
Speech	10	10	8	28+
Perceptual Voice Measures
Overall Vocal Quality	39	36	31	106*
Roughness	39	36	31	106*
Strain	39	36	31	106*
Voice Handicap Index-10	4	4	4	12+
Vocal Fatigue Index	4	4	4	12+
Survey Questions	4	4	4	12+

Note.

^*quantitative,

⁺qualitative

Results

Fundamental Frequency (f_o) Cisgender Females

For f_o in 5-sec sustained /a/, the phrase, and speech, the main effects of condition [F(1, 216) = 124.26, p < 0.001, η_p² = 0.36 for /a/; F(1, 69) = 268.21, p < 0.001, η_p² = 0.79 for the phrase; F(1, 68) = 243.34, p < 0.001, η_p^{2 =} 0.78 for speech], baseline [F(2, 216) = 45.22, p < 0.001, η_p² = 0.29 for /a/; F(2, 69)= 23.17, p < 0.001, η_p² = 0.40 for the phrase; F(2, 68) = 31.44, p < 0.001, η_p² = 0.48 for speech), and the interaction of condition and baseline [F(2, 216) = 13.83, p < 0.001, η_p² = 0.11 for /a/; F(2, 69) = 5.12, p = 0.008, η_p² = 0.12 for the phrase; F(2, 68) = 3.94, p = 0.024, η_p² = 0.10 for speech] were significant indicating differences in mean change across the three baselines between the two conditions (see Table 3). Effect sizes were large for the significant condition and baseline interactions (ranging from 0.10 to 0.12). Bonferroni post-hoc tests for the significant interactions between condition and baseline revealed significant and non-significant pairwise comparison across the 5-sec sustained /a/s, the phrase, and speech. The post-hoc results will be presented in the next section.

Table 3.

Means and Standard Deviations with Two-Way ANOVA Statistics for Fundamental Frequency (f_o) across Baselines for Conditions 1 & 2.

fo (Hz) Cisgender Females	Baseline 1	Baseline 2	Baseline 3	ANOVA C			ANOVA B			ANOVA C × B
	M (SD)	M (SD)	M (SD)	F	p	ηp2	F	p	ηp2	F	p	ηp2
Condition 1	Pre	Post GVPTM	Post MGVPTM
5-sec /a/	139 (29)	197 (14)	196 (14)	153.1	<0.001*	0.51	56.4	<0.001*	0.13	8.14	<0.001*	0.10
Phrase	123 (27)	155 (40)	163 (34)	338.5	<0.001*	0.53	69.2	<0.001*	0.08	6.37	<0.001*	0.04
Speech	139 (26)	170 (13)	177 (11)	54.9	<0.001*	0.53	30.3	<0.001*	0.20	2.61	0.005*	0.10
Condition 2	Pre	Post MGVPTM	Post GVPTM
5-sec /a/	208 (27)	222 (33)	227 (31)	153.1	<0.001*	0.51	56.4	<0.001*	0.13	8.14	<0.001*	0.10
Phrase	207 (23)	222 (27)	224 (25)	338.5	<0.001*	0.53	69.2	<0.001*	0.08	6.37	<0.001*	0.04
Speech	200 (10)	217 (7)	218 (6)	54.9	<0.001*	0.53	30.3	<0.001*	0.20	2.61	0.005*	0.10
fo (Hz) Cisgender Males	Baseline 1	Baseline 2	Baseline 3				ANOVA B
	M (SD)	M (SD)	M (SD)				F	p	ηp2
Condition 1	Pre	Post GVPTM	Post MGVPTM
5-sec /a/	101 (4)	112 (7)	124 (3)				106.15	<0.001*	0.72
Phrase	98 (5)	118 (4)	121 (11)
Speech	101 (9)	116 (6)	121 (8)

Note. N=3 for Cisgender Females. N=1 for Cisgender Males. ANOVA = analysis of variance; C= Condition main effects (i.e., Conditions 1–3); B=Baseline main effects (i.e., Pre, Post GVPTM, Post MGVPTM); C x B= Condition and Baseline interaction effects

^*p<0.05

In condition 1, the GVPTM was the first voice therapy model followed by the MGVPTM. In condition 2, the MGVPTM was the first voice therapy model followed by the GVPTM. Baseline 1 reflects f_o values at pre-treatment. Baseline 2 represents f_o values following either the GVPTM (condition 1) or MGVPTM (condition 2) and baseline 3 indicates f_o values following either the MGVPTM (condition 1) or the GVPTM (condition 2). Comparing baseline 1 to 2, there were significant pairwise comparisons for conditions 1 (p < 0.001 for /a/, the phrase, and speech) and 2 (, p = 0.01 for /a/, p = .005 for the phrase, and p < 0.001 for speech). At post-GVPTM and post-MGVPTM (baseline 2) compared to pre (baseline 1), the f_os were higher for /a/s, the phrase, and speech. Condition 1 was 58 Hz higher for the /a/s, 32 Hz higher for the phrase, and 31 Hz higher for speech. Condition 2 was 14 Hz higher for the /a/s, 13 Hz higher for the phrase, and 17 Hz higher for speech. Comparing baseline 1 to 3, there were significant pairwise comparisons for conditions 1 (p < 0.001 for the /a/s, the phrase, and speech) and 2 (p = 0.001 for the /a/s, p = 0.041 for the phrase, and p < 0.001 for speech). For post-MGVPTM (condition 1), f_o was 57 Hz higher for the /a/s, 32 Hz higher for the phrase, and 32 Hz higher for speech. For post-GVPTM (condition 2), f_o was 19 Hz higher for the /a/s, 10 Hz higher for the phrase, and 18 Hz higher for speech. Comparing baseline 2 to 3, there were no significant pairwise comparisons for conditions 1 (p = 0.91 for the /a/s, p = 0.98 for the phrase, p = 0.22 for speech) and 2 (p = 0.40 for the /a/s, p = 0.50 for the phrase, p = 0.85 for speech)

In sum, post-GVPTM and post-MGVPTM resulted in significantly higher f_os for sustained /a/, the phrase, and speech as compared to pre. There were no significant differences between the post-GVPTM and post-MGVPTM higher f_os in the /a/s, the phrase, and speech suggesting that the new, resonant voice learned in the GVPTM was similar to what was learned in the MGVPTM (see Figures 4 and 5).

Figure 4.

demonstrates mean fundamental frequency (f_o) (Hz) in sustained /a/, the phrase “we were away a year ago,” and speech for cisgender female professional teachers across baseline 1 (pre), baseline 2 (post-GVPTM), and baseline 3 (post-MGVPTM) in condition 1.

Figure 5.

demonstrates mean fundamental frequency (f_o) (Hz) in sustained /a/, the phrase “we were away a year ago,” and speech for cisgender female professional teachers across baseline 1 (pre), baseline 2 (post-MGVPTM), and baseline 3 (post-GVPTM) in condition 2.

Fundamental Frequency (f_o) Cisgender Male

For the cisgender male, condition 2 was facilitated with the GVPTM being the first voice therapy model followed by MGVPTM. The main effect of baseline was evaluated because only condition 2 was facilitated with one male participant. There was a significant main effect of baseline for sustained /a/ [F(2, 81) = 106.15, p < 0.001, η_p² = 0.72, see Table 3]. Bonferroni post-hoc tests for the main effect of baseline for the 5-sec /a/ revealed significant pairwise comparisons. Comparing baseline 1 to 2, there was a significant pairwise comparison for the 5-sec /a/s (p < 0.001). For post-GVPTM (baseline 2) compared to pre (baseline 1), f_o was 11 Hz higher for the /a/s, 20 Hz higher for the phrase, and 15 Hz higher for speech. Comparing baseline 1 to 3, there was a significant pairwise comparison for the 5-sec /a/s (p < 0.001). For post-MGVPTM (baseline 3) compared to pre (baseline 1), f_o was 23 Hz higher for the /a/s, 23 Hz higher for the phrase, and 20 Hz higher for speech. Comparing baseline 2 to 3, there was a significant pairwise comparison for sustained /a/ (p < 0.001). For post-MGVPTM compared to post-GVPTM, f_o was 12 Hz higher in the /a/s, 3 Hz higher in the phrase, and 5 Hz higher in speech.

In sum, the cisgender male had increased f_o values at both post time points as compared to pre. Interestingly, the f_o values following both the GVPTM and MGVPTM were not that different in the phrase and speech as the new, resonant voice was similar across both voice therapy models.

Participant Self-Rating of Perceptual Voice Measures

For Overall Vocal Quality (OVQ), there were significant main effects for condition [F(1, 100) = 4.74, p = 0.03, η_p² = 0.04], baseline [F(2, 100) = 24.63, p < 0.001, η_p² = 0.33], and the interaction between condition and baseline [F(2, 100) = 6.65, p = 0.002, η_p² = 0.11] (see Table 4). Effect sizes were large (0.11) for the significant interaction between condition and baseline. In condition 1, there were significant post-hoc pairwise comparisons between baseline 1 and 2 (p < 0.001) and between baseline 1 and 3 (p < 0.001), but not between baseline 2 and 3 (p = 0.77). Participants in condition 1 rated their OVQ as 18 (mild problem) at pre (baseline 1). After completing the GVPTM (baseline 2), the OVQ rating dropped to 0.85 (no problem) and stayed as no problem (0.06) after the MGVPTM (baseline 3). Comparing OVQ after GVPTM and MGVPTM, there was no difference in participant ratings. In condition 2, there were significant post-hoc pairwise comparisons between baseline 1 and 3 (p = 0.003) and between baseline 2 and 3 (p = 0.04), but not between baseline 1 and 2 (p = 0.31). Participants in condition 2 rated their OVQ as 14 (mild problem) at pre (baseline 1). There was a significant difference in participant self-rating of OVQ comparing pre (14, baseline 1) to post-GVPTM (4 – no problem, baseline 3) and that decrease achieved following the GVPTM (4, baseline 3) was significantly different from the decrease achieved following the MGVPTM (11 – mild problem, baseline 2). There was no significant difference in participant self-rating of OVQ from pre (14, baseline 1) to post-MGVPTM (11, baseline 2).

Table 4.

Means and Standard Deviations with Two-Way ANOVA Statistics for the Perceptual Measures across Baselines for Conditions 1 & 2.

	Baseline 1	Baseline 2	Baseline 3	ANOVA C			ANOVA B			ANOVA C × B
	M (SD)	M (SD)	M (SD)	F	p	ηp2	F	p	ηp2	F	p	ηp2
Condition 1	Pre	Post GVPTM	Post MGVPTM
Overall Vocal Quality	18.10 (18.27)	0.85 (1.04)	0.06 (0.23)	4.74	0.03*	0.04	24.63	<0.001*	0.33	6.65	0.002*	0.11
Roughness	22 (20)	1.2 (1.5)	0 (0)	0.36	0.54	0.00	12.38	<0.001*	0.19	3.01	0.05	0.05
Strain	21 (20.6)	1.5 (2.7)	0.06 (0.23)	13.9	<0.001*	0.12	34.82	<0.001*	0.41	0.62	0.54	0.01
Condition 2	Pre	Post MGVPTM	Post GVPTM
Overall Vocal Quality	14.11 (5.47)	11.19 (4.75)	4.69 (2.32)	4.74	0.03*	0.04	24.63	<0.001*	0.33	6.65	0.002*	0.11
Roughness	14.3 (17.8)	5.6 (12.2)	8.3 (18.5)	0.36	0.54	0.00	12.38	<0.001*	0.19	3.01	0.05	0.05
Strain	29 (14.8)	13.6 (10.7)	5.8 (1.8)	13.9	<0.001*	0.12	34.82	<0.001*	0.41	0.62	0.54	0.01

Note. N=3 for Cisgender Females. N=1 for Cisgender Male. ANOVA = analysis of variance; C= Condition main effects (i.e., Conditions 1 & 2); B=Baseline main effects (i.e., Pre, Post GVPTM, Post MGVPTM); C x B= Condition and Baseline interaction effects

^*p<0.05

For Roughness (R), there was a significant main effect for baseline [F(2, 100) = 12.38, p < 0.001, η_p² = 0.19], but not for condition [F(1, 100) = 0.36, p = 0.54, η_p² = 0.00] and the interaction between condition and baseline [F(2, 100) = 3.01, p = 0.05, η_p² = 0.05]. Effect size was large (0.19) for the significant main effect of baseline. For Strain (S), there were significant main effects for condition [F(1, 100) = 13.92 p < 0.001, η_p² = 0.12] and baseline [F(2, 100) = 34.82, p < 0.001, η_p² = 0.41], but not for the interaction between condition and baseline [F(2, 100) = 0.62, p = 0.54, η_p² = 0.01] (see Table 4).

Effect sizes were large (0.12 and 0.41) for the significant main effects of condition and baseline. Considering the baseline main effect for both R and S, there were significant post-hoc pairwise comparisons between baseline 1 and 2 (p < 0.001) and between baseline 1 and 3 (p < 0.001), but not between baseline 2 and 3 (p = 0.83 for R and p = 0.11 for S). Participants rated R as 18 (mild problem) and S as 25 (mild problem) at pre (baseline 1). There was no influence of voice therapy model on the three baselines for R and S as the interactions of condition and baseline were not significant. Regardless of the voice therapy model, self-ratings decreased from pre (18 for R and 25 for S) to post (3 for R at baseline 2, 4 for R at baseline 3, 3 for S at baseline 2, and 4 for S at baseline 3) with no differences between R and S self-ratings comparing post-GVPTM and post-MGVPTM.

In summary, for OVQ, condition 1 with the GVPTM first followed by the MGVPTM demonstrated a significant decrease from pre to post-GVPTM and from pre to post-MGVPTM. Comparing post-GVPTM to post-MGVPTM, there was no difference in OVQ self-ratings. Both voice therapy models were successful in decreasing OVQ. For OVQ, in condition 2 with the MGVPTM first followed by the GVPTM, there was no difference between pre and post-MGVPTM self-ratings of OVQ. However, after participants completed the GVPTM, there were significant differences in self-ratings of OVQ from pre to post-GVPTM and the ratings were significantly different comparing post-GVPTM to post-MGVPTM. The GVPTM, in condition 2, was more influential in decreasing the participants’ self-ratings of OVQ as compared to the MGVPTM. For both R and S, the baseline main effect was significant, indicating that self-ratings decreased from pre to post regardless of the specific voice therapy model.

VHI-10 and VFI

For the VHI-10 total score, the means at pre were 8.5 for condition 1 and 10 for condition 2 (see Table 5). Both conditions resulted in VHI-10 total scores below 11 at pre, indicating a “normal” voice-related QoL, which was not surprising given that the participants did not have diagnosed voice disorders. In condition 1, the means increased from 8.5 at pre to 11 at post-GVPTM and then returned to pre level 8.5 at post-MGVPTM. In condition 2, the means stayed the same at 10 for pre and post-MGVPTM and then decreased to 9.5 at post-GVPTM.

Table 5.

Means and Standard Deviations for the Voice Handicap Index-10 (VHI-10) and the Vocal Fatigue Index (VFI) across Baselines for Conditions 1 & 2.

	Baseline 1	Baseline 2	Baseline 3
	M (SD)	M (SD)	M (SD)
Condition 1	Pre	Post GVPTM	Post MGVPTM
VHI-10	8.5 (4.9)	11.0 (1.4)	8.5 (0.70)
VFI factor 1	23.5 (17.6)	17.5 (4.95)	7.00 (4.24)
VFI factor 2	9.00 (1.41)	4.5 (2.12)	0.50 (0.70)
VFI factor 3	8.5 (4.95)	11.0 (1.41)	8.50 (0.70)
Condition 5	Pre	Post MGVPTM	Post GVPTM
VHI-10	10 (1.41)	10 (1.41)	9.5 (0.70)
VFI factor 1	23.5 (0.70)	17.00 (1.41)	13.5 (7.77)
VFI factor 2	7.50 (4.95)	4.00 (5.65)	5.00 (7.07)
VFI factor 3	10.0 (1.41)	10.0 (1.41)	9.50 (0.70)

Note. N=3 for Cisgender Females. N=1 for Cisgender Male.

For VFI factor 1 (questions 1–11) and VFI factor 2 (questions 12–16), the mean scores decreased from pre to post in both conditions 1 and 2. For VFI factor 3 (questions 17–19), the mean scores did not vary much from pre to post in both conditions 1 and 2.

Surveys & Exit Interview

The participants completed survey questions at the end of the study. For question 1, participants rated their response as follows; 1=dislike a great deal, 2=dislike somewhat, 3=neither like nor dislike, 4=like somewhat, 5=like a great deal. For questions 2–11, the following Likert scale was used; 1=strongly disagree, 2=somewhat disagree, 3=neither agree nor disagree, 4=somewhat agree, 5=strongly agree.

For questions 5–7, there were differences in the mean scores between the GVPTM and MGVPTM (see Table 6). The participants strongly agreed or somewhat agreed that the GVPTM helped them to get loud in a healthy way (question 5, mean of 5.00), helped them to talk quietly in a healthy way (question 6, mean of 4.92), and helped them to talk over noise in a healthy way (question 7, mean of 4.82), while the participants somewhat disagreed or neither agreed or disagreed that the MGVPTM helped them to get loud in a healthy way (question 5, mean of 2.75), helped them to talk quietly in a healthy way (question 6, mean of 3.25), and helped them to talk over noise in a healthy way (question 7, mean of 2.50). For questions 1–4 and 8–11, there were no mean differences between the GVPTM and MGVPTM. The participants liked both the GVPTM and MGVPTM and agreed that both the GVPTM and MGVPTM provided them skills to prevent voice problems in the future. Participants somewhat agreed that the GVPTM provided them the opportunity to address all their vocal needs, while participants somewhat disagreed that the MGVPTM provided them the opportunity to address all their vocal needs. Participants agreed that both the GVPTM and MGVPTM helped them to produce a better “new” voice, both the GVPTM and MGVPTM improved their voice for teaching, the new voice was easy to produce and learn in both the GVPTM and MGVPTM, and that they would recommend both the GVPTM and MGVPTM to others.

Table 6.

Means (M) and Standard Deviations (SD) of the Survey Questions Rated by the 4 Participants for the Global Voice Prevention and Therapy Model (GVPTM) and Modified (MGVPTM).

Questions	GVPTM M (SD)	MGVPTM M (SD)
1) To what extent did you like GVPTM and MGVPTM?	4.92 (0.28)	4.25 (0.50)
2) GVPTM and MGVPTM provided me skills to prevent voice problems in the future.	5.00 (0)	4.75 (0.50)
3) GVPTM and MGVPTM provided me the opportunity to address all my vocal needs.	4.75 (0.62)	2.75 (1.50)
4) GVPTM and MGVPTM helped me to produce a better “new” voice.	5.00 (0)	5.00 (0)
5) GVPTM and MGVPTM helped me to get loud in a healthy way.	5.00 (0)	2.75 (0.50)
6) GVPTM and MGVPTM helped me to talk quietly in a healthy way.	4.92 (0.28)	3.25 (0.95)
7) GVPTM and MGVPTM helped me to talk over noise in a healthy way.	4.83 (0.38)	2.50 (1.0)
8) GVPTM and MGVPTM helped me to improve my voice when teaching.	5.00 (0)	4.75 (0.50)
9) The “new” voice achieved with GVPTM and MGVPTM was easy to produce.	4.50 (0.52)	5.0 (0)
10) The “new” voice achieved with GVPTM and MGVPTM was easy to learn.	4.58 (0.51)	5.00 (0)
11) I would recommend GVPTM and MGVPTM to others.	4.92 (.28)	4.50 (0.57)

Note. N=3 for Cisgender Females. N=1 for Cisgender Male.

In addition, the participants rated the following question, “Which program facilitated the better new, resonant voice?” Two of the four participants (50%) indicated that the GVPTM produced the better new, resonant voice, while the other two (50%) specified that the MGVPTM produced the better new, resonant voice.

During the exit interview, the participants were asked targeted questions related to the voice therapy models (see Table 7). The questions were centered around the four voice therapy concepts within the GVPTM. For concept 1 - training multiple voices to meet all their vocal needs, all participants viewed this positively with all participants suggesting that training only one new voice in the MGVPTM was a problem. For concept 2 - connecting auditory-perceptual prompts to anatomy and physiology, all participants viewed this concept positively. For concept 3 - switching between voice targets, all participants viewed this positively with all participants indicating that not switching between voice targets was a negative in the MGVPTM. For concept 4 - a bottom-up speech hierarchy, all participants viewed this positively.

Table 7.

Example Quotes from the Participants with Frequency Related to the Four Voice Therapy Concepts (N=4)

Voice Therapy Concept	Example Quotes	Frequency, n (%)
1) Training multiple voices	“Liked all the voice options learned, so teachers can decide what voice to use to meet a specific teaching demand.”	4/4 (100) viewed learning multiple voices as positive
	“Oral twang useful for getting classroom attention, while belt can be used to get attention immediately.”
	“Switching between oral twang and belt can be used to convey same information with a different sound.”
	“It was a negative to learn only one new voice in the MGVPTM.”	4/4 (100) viewed learning only one new voice as negative
2) Using an integrated approach that connects auditory-perceptual voice prompts (implicit) to anatomy and physiology (explicit)	“Seeing how the mechanisms move and seeing how all the parts of the larynx work together to make your voice was very helpful. The hand signals and visual aspects were very helpful for a visual learner.”	4/4 (100) viewed as positive
	“Liked the training in anatomy and physiology – how the anatomy is laid out and what I am manipulating as I was making the voices.”
3) Switching between voice targets	“Switching between voices allowed me to feel and hear the differences.”	4/4 (100) viewed switching between voice targets as positive
	“I have more awareness when you’re actively talking and switching (versus just using the new voice).”
	“Shows examples of good and bad.”
	“When I was in the MGVPTM, I only practiced the new voice. When I got to the GVPTM, I liked contrasting all the voices. I know what it sounds like in the wrong way to know that I am doing the new voice the right way.”	4/4 (100) viewed practice of only one voice as negative
4) Bottom-up speech hierarchy	“Using all the steps was easier for me to learn because I would be thinking about what I wanted to say and how to do the voice correctly. I needed small sentences before I could practice it in a conversation.”	4/4 (100) viewed the speech hierarchy as positive
	“When you build complexity in the levels of speech, you can go back to the smaller levels (facilitator phrase) to remind yourself of how to produce the voice.”
	“Breaking it up into smaller steps allowed me to be successful.”
	“So the more practice you have with simpler mental tasks made it easier to use the new voices in more complex mental tasks like conversation and lesson plans.”

Discussion

The current study assessed the effects of the GVPTM and MGVPTM across f_o; participant self-ratings of OVQ, R, and S; VHI-10; VFI; surveys; and exit interviews for four professional teachers. The GVPTM includes four concepts: concept 1- training multiple new voice targets, concept 2- using an integrated approach that connects auditory-perceptual voice prompts (implicit) to anatomy and physiology (explicit), concept 3 – switching between voice targets, and concept 4 – bottom-up speech hierarchy. The GVPTM was compared to the MGVPTM that eliminated concepts 1 and 3, while keeping concepts 2 and 4 in an alternating treatment SSD with multiple baselines (i.e., A₁BA₂CA₃ or A₁CA₂BA₃). Subsequently, condition 1 began with the GVPTM followed by the MGVPTM, while condition 2 began with the MGVPTM and ended with the GVPTM.

In sum, post-GVPTM and post-MGVPTM resulted in significantly higher f_os for sustained /a/, the phrase, and speech as compared to pre. In cisgender females, there were no significant differences between the post-GVPTM and post-MGVPTM higher f_os in the /a/s, the phrase, and speech suggesting that the new, resonant voice learned in the GVPTM was similar to what was learned in the MGVPTM (see Figures 4 and 5). For OVQ, the GVPTM was more successful in decreasing self-ratings of OVQ from pre to post as compared to the MGVPTM in condition 2. For both R and S, self-ratings decreased from pre to post regardless of the specific voice therapy model. VHI-10 and VFI factors 1, 2, and 3 yielded no clear differences from pre to post, which was a similar finding with student teachers immediately following voice therapy.²⁹ However, when the VHI-10 and VFI factors 1 and 2 were collected over a 14-week student teaching semester, there were significant differences for two treatment conditions as compared to a control condition.²⁹ The prior study included 14 data collection time points for the VHI-10 and VFI during student teaching.²⁹ The current study only had three data collection time points. Perhaps if the professional teachers were followed across a semester of teaching, results may have reflected a significant difference similar to what was found with student teachers during the 14-week student teaching semester. Overall, the participants viewed both the GVPTM and MGVPTM and the four concepts embedded in the GVPTM positively. Specifically, participants agreed that the GVPTM helped them to talk quietly, talk over voice of the class, and yell in a healthy way addressing all the voices needed for teaching (concept 1), while the MGVPTM only trained a new resonant voice. In addition, switching between all the voices (concept 3) at each level of the speech hierarchy in the GVPTM was positive, while only practicing the new resonant voice in the MGVPTM was a negative.

Concept 1 – Training Multiple New Voice Targets

In concept 1, the GVPTM trains multiple new voice targets to meet all the vocal needs of the client. In the current study, the following new voice targets were trained to address all the vocal needs of the teachers: 1) a new resonant voice for connected speech, 2) falsetto for quiet talking, 3) oral twang for talking over noise, and 4) belt for healthy yelling. The MGVPTM trained only the new resonant voice in connected speech.

Results from survey questions demonstrated that the GVPTM addressed all the participants’ vocal needs more so than the MGVPTM. Specifically, the participants indicated that the GVPTM taught them a better new resonant voice, a voice for quiet talking (falsetto), a voice to talk over noise (oral twang), and a voice to get loud (belt). During the exit interviews, all participants (100%) indicated that learning multiple new voice targets was positive, while all participants (100%) indicated that learning only one new voice target was negative (see Table 6). For example, one participant mentioned, “….learning only one new voice in the MGVPTM was a negative.” In contrast, a second participant, offering a comment related to the GVPTM, said, “liked all the voice options learned, so teachers can decide what voice to use to meet a specific teaching demand.” A third participant stated that using the different voices learned in the GVPTM helps with differing classroom attention needs, “oral twang useful for getting classroom attention, while belt can be used to get attention immediately.”

The current study was the first to assess concept 1 - training multiple new voce targets. The design of the study allowed participants to experience two voice therapy models, one model training multiple voice targets (GVPTM) and the other training just one (MGVPTM). Thus, participants were able to provide valuable reflections on their experiences after completing both models. Overall, the participants indicated that learning multiple new voice targets in the GVPTM was positive, while learning only one new resonant voice in the MGVPTM was negative. Conversely, most of the voice therapy literature does not account for training multiple new voice targets, rather the focus is typically on one new voice for all communication tasks.^30–38 Considering evidence-based practice (EBP), SLPs have a responsibility to integrate client wants/needs, current best evidence, and clinician expertise when designing prevention, assessment, and treatment programs.⁶¹ Concept 1 relates directly to client wants/needs by considering all the vocal demands of the client when determining the voice targets that are necessary to address during voice therapy. During the interview, the SLP can simply ask the client, “What are your vocal needs?” “How do you typically use your voice throughout the day, week, month?” “Do your voice demands change daily, weekly, or monthly?” From the answers, the SLP can integrate client wants/needs with evidence and clinician expertise to design a voice therapy program grounded in EBP that may address training multiple new voice targets.

Concept 2 – Integrated Implicit-Explicit Approach to Voice Training

In concept 2, the GVPTM trains the new voices in an integrated implicit-explicit approach linking auditory perceptual prompts (e.g., “moh-you,” “beep-beep, etc.) to anatomy and physiology of the voice (e.g., False Vocal Fold Retract for “moh-you” and Aryepiglottic Sphincter Narrow for “beep-beep”). The MGVPTM kept this concept when training only the new resonant voice.

The GVPTM and MGVPTM conditions facilitated a new resonant voice in connected speech. For the GVPTM and MGVPTM, forward, facial resonance in the auditory-perceptual prompts of “moh-you,” “oh-you,” or “humming,” was trained by focusing on the following anatomical Estill Figures and physiological Options; True Vocal Fold (TVF) Body-Cover (Slack, Thick, Thin, Stiff), False Vocal Folds (FVF) (Constrict, Mid, Retract), Thyroid (Vertical, Tilt), and Head & Neck (H&N) (Relax, Anchor). When participants recorded their voice using the VoiceEvalU8 app, they were to use their normal, everyday speaking voice. As the participants advanced through the voice therapy models, the normal everyday speaking voice changed to a new, resonant voice in connected speech following the GVPTM and MGVPTM. f_o results demonstrated that the GVPTM and MGVPTM facilitated a significant increase in f_o from pre to post. When comparing post-GVPTM and post-MGVPTM, there were no significant differences noted in f_o. Thus, both the GVPTM and MGVPTM facilitated a change in voice that was evidenced by an increase in f_o as compared to pre. For both the GVPTM and MGVPTM, the voice was more resonant.

An increase in f_o has been documented in other studies training a new resonant voice with teachers.^62–64 Prior work with student teachers has also demonstrated an increase in f_o following the telepractice and in-person GVPTM conditions, but not for a control condition that did not facilitate vocal training.²⁹ The reason for the increase in f_o related to the new resonant voice trained in the GVPTM and MGVPTM may be due to the Estill Figures and Options used in the integrated implicit-explicit approach. Many of the teachers were using a baseline voice quality that involved TVF Body-Cover Slack, FVF Mid to Constrict, and Thyroid Vertical, similar to the Estill Speech Quality with variations in TVF Body-Cover as Slack and FVF as Constrict. With this baseline voice pattern, lower f_os have been documented.^29,66–68 When the participants were trained to produce a new resonant voice in GVPTM and MGVPTM (i.e., TVF Body-Cover Thick, FVF Retract, Thyroid Tilt, and H&N Anchor), then the result was an increase in f_o.

All participants (100%) viewed the integrated implicit-explicit approach favorably. The favorable comments included, “seeing how the mechanisms move….. how all the parts of the larynx work together to make your voice was helpful,” “hand signals and visual aspects were very helpful,” “liked the training in anatomy and physiology – how the anatomy is laid out and what I am manipulating as I was making the voices.” In the surveys, two participants (50%) indicated that the GVPTM produced the better new resonant voice, while the other two participants (50%) said that the MGVPTM produced a better new resonant voice. The results related to the question of which voice therapy model facilitated a better new resonant voice is consistent with other findings in the study. For example, participants rated GVPTM and MGVPTM as “strongly agree (5)” that both models helped them to produce a better new resonant voice. f_o increased post-GVPTM and post-MGVPTM providing acoustic evidence that a more resonant voice was created following both the GVPTM and MGVPTM. In addition, the new resonant voice created from both the GVPTM and MGVPTM did not differ as the methods were the same across both models via the integrated implicit-explicit instructional approach. Therefore, it is not surprising that a similar new resonant voice was created across both voice therapy models. Related to OVQ, for some participants in condition 2, their self-ratings significantly decreased only following the GVPTM and not following the MGVPTM; therefore, perhaps the training of the multiple voices (concept 1) and switching between all the voices (concept 3) at each level of hierarchy may have influenced their perceptions in a positive way. Another important aspect to the survey involved questions related to ease of learning and producing the “new” voice in the GVPTM and MGVPTM. There were no significant differences for learning and producing the “new” voice across the voice therapy models. Even with the addition of the anatomy and physiology of voice in the GVPTM and MGVPTM, the participants did not perceive that the “new” voice achieved in these models was hard to learn or hard to produce.

Concept 3 – Switching between Voice Targets

The GVPTM included concept 3 of switching between voice targets. In the GVPTM, participants switched between the old voice, new resonant voice, falsetto, oral twang, and belt. The MGVPTM did not offer concept 3 and only facilitated the new resonant voice. Exit interview responses from the participants indicated that all participants (100%) viewed concept 3 as positive with all (100%) saying it was a negative to only work on one voice in the MGVPTM. Participant comments included, “switching between the voices allowed me to feel and hear the differences,” “I have more awareness when you’re actively talking and switching vs just using the new voice,” “shows examples of good and bad,” and “when I got to the GVPTM, I liked contrasting all the voices.”

When clients switch between voice targets, they use auditory and kinesthetic feedback systems to develop voice awareness in voice therapy⁶⁹ with research suggesting that auditory feedback may be critical to voice training.⁷⁰ Negative practice, meaning the patient deliberately produces appropriate (new voice) and inappropriate voice productions (old voice), may help facilitate voice awareness. The literature related to fluency treatment suggests that negative practice helps to stabilize a new behavior and increases the rate of learning.^49,50 The GVPTM was one of the first voice therapy models to incorporate negative practice by production of “new” voice (i.e., improved vocal output achieved in therapy) vs “old” voice (i.e., poor vocal output before therapy) in a systematic approach.⁴³ The GVPTM has evolved beyond negative practice to include contrasting all the other “new” voices the client is learning in voice therapy with the “old” voice at each step of the speech hierarchy.²⁹ Production of “new” voice vs “other/old” voice immediately follows the achievement of 90% accuracy of “new” voice at a certain level of the hierarchy. For example, the client first achieves the “new” voice at 90% accuracy across sentences. The SLP then moves to the “new” vs “other/old” voice component by asking the client to switch between the voices at the sentence level. In addition, the SLP may ask the client to switch between the voices and the SLP guesses the correct voice. This pattern will continue at each step of the hierarchy until conversational speech is achieved. In prior work, using the GVPTM, participants were successful in contrasting all the voices (i.e., new resonant voice, falsetto, oral twang, belt, and old voice) at each step of the hierarchy by producing them on command or independently.²⁹ The telepratice and in-person GVPTM were successful in improving the voices of student teachers and maintaining the voice improvements for months during student teaching.²⁹ Considering that vocal training occurred during three 45–60 minute sessions and that training was shared among the voices, the GVPTM was successful and efficient in facilitating voice changes; therefore, the same training plan was used in the current study. Negative practice applied by the GVPTM was later included in other voice therapy models.^33,41,44 The importance of switching between voice targets, as mentioned by the participants, provides evidence for incorporating concept 3 into voice therapy sessions.

Concept 4 – Bottom-up Speech Hierarchy

Voice therapy models in the literature focus on a bottom-up hierarchical approach to generalize voice targets into speech.^{11,12,29,35,37–40,43,45–47,71–75} A bottom-up hierarchy builds in complexity from smallest length of utterance with decreased cognitive load (e.g.,words) to greatest length of utterance with increased cognitive load (e.g., monologue or conversation). Consistent with motor learning theory, the client acquires a new skill in a basic motor pattern first before increasing to a more complex pattern.⁷⁶ Seminal literature in fluency, articulation, and voice treatment supports the use of a bottom-up treatment hierarchy.^{35,37,38,48–50} Prior work demonstrated that a bottom-up speech hierarchy involving words, phrases, sentences, memorized speech acts (e.g., pledge of allegiance), specific spontaneous speech (e.g., how to make a peanut butter and jelly sandwich), monologue, and conversation was quick and successful in training four new voices with vocally healthy student teachers in just three 45–60 minute weekly training sessions.²⁹

The GVPTM and MGVPTM incorporated a bottom-up speech hierarchy during voice training. In the participant responses to the exit interview questions, 4/4 (100%) indicated that the speech hierarchy was positive. Participant comments included, “using all the steps was easier for me to learn because I would be thinking about what I wanted to say and how to do the voice correctly,” “breaking it up into smaller steps allowed me to be successful,” and “so the more practice you have with simpler mental tasks made it easier to use the new voices in more complex mental tasks like conversation and lesson plans.”

Conclusions

Overall, the four concepts present in the GVPTM with EVM of training multiple new voice targets (concept 1), an integrated implicit-explicit approach to voice training (concept 2), switching between voice targets (concept 3), and a bottom-up speech hierarchy (concept 4) all hold promise as potential concepts that may be represented as targets, ingredients, and/or MoAs during voice therapy sessions. All four teachers agreed that training multiple new voice targets (concept 1) in the GVPTM met all their vocal needs, while all teachers said it was negative only training one new voice in the MGVPTM. SLPs may consider evaluating the wants/needs of their clients and adapting voice therapy methods to potentially train multiple new voice targets meeting all the vocal needs of their clients. An integrated implicit-explicit approach to voice training (concept 2) was used in the GVPTM and MGVPTM. A new voice for connected speech was more resonant following the GVPTM and MGVPTM as evidenced by an increase in f_o. All teachers perceived the integrated implicit-explicit approach as positive agreeing that the new resonant voice was easy to learn and produce in both the GVPTM and MGVPTM. These results suggest that SLPs may consider using an integrated implicit-explicit to voice training. All teachers viewed switching between voices (concept 3) at each level of the speech hierarchy as positive with all teachers indicating only practicing one voice as a negative in the MGVPTM. Switching between voice targets in the GVPTM functions as a random practice schedule facilitating fast and efficient learning gains; therefore, SLPs may consider adapting voice therapy methods to include switching between voice targets. A bottom-up speech hierarchy (concept 4) from smallest utterance with decreased cognitive load (e.g., words) to largest utterance with increased cognitive load (e.g., conversation) was present in both the GVPTM and MGVPTM. All teachers indicated that the speech hierarchy was positive because practicing in smaller units helped them to be more successful. SLPs may consider using a bottom-up speech hierarchy when facilitating new voice targets into speech.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.