Outcomes of Gender-Affirming Voice and Communication Modification Training for Non-binary Individuals: a Case Series

Abstract

OBJECTIVES:

There is currently no research reporting solely on outcomes of voice and communication modification training (VCMT) in individuals who identify as non-binary and genderqueer (NBGQ) in the English literature. This study aimed to describe the objective and subjective impact of VCMT on the voice of NBGQ individuals undergoing a 12-week gender-affirming VCMT program.

METHODS:

A retrospective consecutive case series of NBGQ individuals enrolled in a VCMT program was performed. Demographics, Transgender Self-Evaluation Questionnaire (TSEQ), fundamental frequency (F0) and frequency range were collected before and after the program. RESULTS: 4 NBGQ individuals enrolled between January 2019 and June 2021; mean age was 27.0 years. While all four participants represented in this case series showed improvement in at least one of their initial goals, only one improved both their F0 and TSEQ score; the other three participants had mixed results.

CONCLUSION:

NBGQ individuals experienced improvements in self-reported outcomes and changes in acoustic measures after completing VCMT in our case series. Individuals experienced significant improvement in subjective outcomes despite small changes in acoustic measures, and vice versa. More research is needed to better understand the voice and communication needs of NBGQ individuals, along with their outcomes with VCMT.

INTRODUCTION

Transgender individuals are those whose gender identity differs from the gender associated with the sex they were assigned at birth. Non-binary and genderqueer (NBGQ) individuals exist outside the gender binary (i.e., women or men) and comprise over 1.2 million people in the United States.¹ Previous research has shown that transgender individuals recognize their voices as an important component of identity^2,3 and their quality of life can be affected by how feminine or masculine their voice is perceived to be.^4,5 Furthermore, gender-affirming voice care delivered via a structured voice and communication modification program has been shown to improve client-reported outcomes in trans women irrespective of the absolute change in acoustic measures.⁶ Although NBGQ individuals make up 35% of the transgender and gender non-conforming population, their experiences with gender-affirming voice therapy have received little attention in the literature relative to transgender men and women. Additionally, it is important to note that while transgender and NBGQ individuals are often grouped into a single category, the experiences of NBGQ individuals are often different from those of binary transgender people and thus merit further study.⁷

Evidence suggests that NBGQ people experience poor psychological wellbeing and social exclusion, including but not limited to harassment, discrimination, and threats of violence.⁸ NBGQ individuals have also been shown to experience anxiety, depression, psychological distress, eating concerns, and self-harm and suicidality at rates exceeding both cisgender and binary transgender individuals, with one study finding nearly 50% of non-binary college students reporting a past suicide attempt.⁹ Further, NBGQ individuals can experience gender dysphoria in ways that are distinct from their binary transgender counterparts (with an emphasis on androgyny over masculinity and femininity).¹⁰ These results emphasize the urgent need for further study to better guide therapeutic interventions for this population.

Previously, our team conducted a retrospective cohort study investigating the effect of VCMT program on 16 trans women. Both self-reported outcomes and acoustic measures improved after the program; however, changes in acoustic measures did not significantly correlate with improvements in Trans Woman Voice Questionnaire (TWVQ) scores.⁶ Although this initial study was limited by the lack of transgender men and NBGQ individuals, it was the first of its kind to evaluate the correlation between acoustic measures and self-reported voice outcomes in response to a gender-affirming voice training program.

To date, there have been few published studies examining the voice-related concerns of NBGQ individuals. One exception is the development of the Voice-related Experiences of Nonbinary Individuals (VENI), a psychometric instrument proposed for this population to capture their self-perception of their voice and communication experience.⁷ Merrick et al. reported on the outcomes of two NBGQ individuals undergoing hybrid VCMT with both in-person and virtual interventions. However, this study was limited by a small sample size of 2 NBGQ individuals who were, as a result, grouped into a larger transgender cohort for data analysis.¹¹ Quinn et al. included eight NBGQ individuals in a prospective study investigating effectiveness of delivery schedules for VCMT, but also included them within a larger cohort of transgender women; all individuals in this study were seeking voice feminization.¹²

Despite the paucity of research in this population, a survey of over 27,000 transgender adults in the United States showed that 36% of NBGQ respondents were either receiving VCMT (2%) or desired it (34%).¹³ Moreover, recent qualitative studies have shown that the healthcare needs and services for NBGQ individuals remain poorly understood. This is in part due to the heterogeneity within this population, as well as the equally varied health and gender perception goals for NBGQ people relative to their binary transgender counterparts. Transitioning for a NBGQ person is often considered a process with goals that diverge from the stereotypical binary transition from one gender to another.⁸

By contrast, the literature on the androgynous synthetic voice is rapidly growing, particularly in the fields of human-computer interaction (HCI), consumer research, and audio engineering.^14–16 Recognizing the inherent problems of a gendered voice in artificial intelligence (AI) and its reinforcement of stereotypes in society, several groups have pioneered efforts to create a gender-neutral synthesized voice. Two of these efforts include Project Q and [multi’vocal], both of which were designed using voice samples recorded from speakers of various gender identities.^17,18 Though still in its infancy, the recently increased interest in the study of the non-binary voice highlights its social importance. It is important to note, however, that not all NBGQ individuals have or desire a gender-neutral voice, and it should not be assumed that the interest in a gender-neutral voice is representative of NBGQ people as a whole.

Though vocal acoustics are primarily dictated by one’s anatomy, acoustic signal production is also subject to voluntary control.^18–24 Individuals can thus exert some control over how their voice is perceived by others and produce a more gender-congruent vocal sound, though scholarship in this area, including Azul and Hancock’s ASSEMBLE model, illustrates the many factors – many outside of the speaker’s control – that influence the perception of gender in communication.^25–27 One evidence-based method of teaching one how to habituate their voice in a healthy manner is through behavioral training with a speech-language pathologist. VCMT can be pursued in conjunction with or separate from medical and surgical interventions. For instance, masculinizing hormone replacement therapy can deepen one’s voice through physiological changes that occur in the larynx that are similar to pubertal changes seen in cisgender men.²⁸ Likewise, surgical techniques have been described in both trans men and trans women (i.e., masculinization and femininization laryngoplasty) with the goals of deepening or raising one’s pitch, respectively.^29,30 Although medical and surgical interventions may result in vocal parameter changes alone, behavioral training remains an important means of voice modification for many patients, both in combination with aforementioned interventions or as a stand-alone treatment modality for individuals who cannot access or do not desire medical or surgical options.

Previous studies have largely focused on trans women and their self-reported ratings of vocal femininity with a very limited number of studies that explored outcomes of voice modification utilizing a validated questionnaire (the Trans Woman Voice Questionnaire).^12,31–36 There are currently no published studies investigating outcomes as they relate to VCMT for NBGQ individuals only. To address this gap, the present study aimed to describe both objective and subjective outcomes among a small cohort of NBGQ individuals undergoing a 12-week VCMT program, using both acoustic measures and validated self-reported outcomes measures. Based on earlier studies that demonstrated a discordance between fundamental frequency and vocal satisfaction,^32,33,37,38 we hypothesized that changes in subjective outcomes measures would not necessarily correlate with changes in acoustic metrics, and vice versa.

MATERIALS AND METHODS

This study was approved by the Ithaca College Institutional Review Board (Protocol #124). A retrospective consecutive case series of non-binary and genderqueer individuals enrolled in the Voice and Communication Modification Program for People in the Transgender Community at Ithaca College (Ithaca, New York) from January 2019 to June 2021 was performed. Before the program, demographic data, Transgender Self-Evaluation Questionnaire (TSEQ) scores, mean fundamental frequency (F₀) of spontaneous speech and reading, and frequency range of spontaneous speech and reading were collected. These acoustic measures were chosen as they are used as a reference point during the training program, and there are well-established normative values for cisgender men and women. Although previous authors demonstrated changes in vocal tract resonance measures such as formant frequency, resonance measures were not included in this pilot treatment program, in part due to constraints related to the COVID-19 pandemic.

The program consisted of weekly one-hour individual training sessions performed either in-person or remotely via a telehealth platform for 12 weeks; the number of appointments conducted via tele-health was different for each participant and was dictated by the onset of the COVID-19 pandemic during spring 2020. Training sessions were provided by speech-language pathology graduate students at Ithaca College, who had all completed a graduate level course of voice; all clinicians asked to participate in this voice modification program, and none were assigned without the expressed interest in working with this population. None had prior experience providing VCMT to transgender or NBGQ individuals. All students were directly supervised by one of the senior authors (M.P.), who has over ten years of experience working with transgender and NBGQ individuals, and another speech-language pathologist who is a transgender woman. Cultural sensitivity education was provided at the outset of the program and continually emphasized throughout regular meetings between student clinicians and supervisors.

At the start of the program, each participant met individually with their clinician to develop personalized goals and an individualized training plan. Each session focused on a different aspect of voice modification. Early sessions included education on vocal health, laryngeal relaxation (e.g., laryngeal massage, yawn-sigh exercises, straw phonation), phonatory breath support, and establishing a target F0 range that more closely aligned with participant goals. Clinicians assessed for dysphonia at each session, and were trained to listen for tension and hoarseness; a portion of each session focused on vocal health, relaxation, and proper breathing to support the voice exercises being performed. We began with a baseline of F0 in reading and spontaneous conversation as well as obtaining an F₀ range. Each client established goals with their clinician, and F0 targets gradually progressed up or down based on individual preferences rather than cisgender normative values. Frequency practice systematically progressed from sustained vowels to words, phrases, sentences, and eventually conversation in a slow and systematic manner to reinforce healthful phonation. As sessions progressed, emphasis shifted to incorporate more gender-affirming resonance (taught through sensory feedback), intonation, and stress patterns while continually working toward a speaking F0 and range more congruent with participants’ goals. Resonant voice therapy was used starting at the single-word level and progressing to sentences and short conversations; for participants seeking a more feminine tone, a focus was placed on forward resonance, and for participants seeking masculinization of their voice, exercises focused on deeper chest resonance. Depending on individual goals, additional sessions targeted articulation, word choice, pragmatics, and non-verbal communication skills. These topics were all introduced as considerations and incorporated in varying degrees based on each client’s individual goals; participants were asked about their degree of satisfaction with certain verbal and non-verbal aspects of their communication, education was provided to participants about characteristics of speech traditionally associated with masculinity or femininity, and clinicians worked with each participant to make adjustments to particular qualities that the participant wanted to address. Throughout the program, participants were given individualized homework assignments that encouraged both repeated practice and generalization of newly learned skills. Participants were encouraged to practice with friends and were given suggestions of ways to generalize learned skills in a safe manner; informal supportive counselling was also provided by clinicians as needed, and additional resources referrals to more dedicated counselling were provided when necessary.

At the conclusion of the program, the multidimensional voice evaluation was repeated. Acoustic analyses were completed using Praat (v. 6.1.40, 2021, Boersma P & Weenink D).

Participants

All participants were adults and self-identified as NBGQ who were seeking voice and communication modification through behavioral training. Individuals with dysphonia based on clinical assessment were not included. Four participants enrolled over four semesters (Spring 2019 to Spring 2021). The initial target voice in all four NBGQ participants was gender-neutral, and two participants ultimately sought a “somewhat feminine” voice. One participant worked towards voice masculinization and three worked towards voice feminization. All participants completed the program with an overall completion rate of 100%. The TSEQ was completed before and after the program by all participants.

RESULTS

Participant 1 (they/them) began voice therapy seeking to achieve a more feminine voice, with a “gender-neutral” voice seen as a stepping stone to a “somewhat feminine” voice. Their goals included increasing mean F0, increasing vocal loudness at their targeted frequency, and improving their TSEQ score. After 12 sessions, their mean F0 greatly increased in paragraph reading (146.2 Hz to 203.6 Hz) and spontaneous speech (140.9 Hz to 195.2 Hz). Their F0 range in paragraph reading (76–499 to 84–502) and spontaneous speech (77–504 to 77–509) remained mostly unchanged, with overall slight increases. Their speech pathologist also determined that their goal of increasing loudness at targeted pitch had been met. However, their TSEQ score increased from a 57 to a 68, though their rating of their voiced changed from “very male” before therapy to “somewhat male” after their final session. Participant 1 acknowledged they had made progress through VCMT, and believed that their increased TSEQ score was explained by their not yet having used their new voice “out in daily life yet.”

Participant 2 (they/them) sought to achieve a lower voice; their goals at outset of therapy were a 5-point improvement in their TSEQ score and a whole-tone lowering of their mean F₀. They attended 9 sessions of VCMT, and achieved a successful lowering of mean F₀ in paragraph reading (219 Hz to 181 Hz) and ongoing conversation (211 Hz to 177Hz); they received a pitched keyboard cue during their final ongoing conversation assessment. Their F₀ range in reading (83–386 to 78–309) and spontaneous speech (91–385 to 76–361) also showed lowering across the board. However, their TSEQ score increased from 65 to 69, reflecting a continuing dissatisfaction with their voice.

Participant 3 (she/her) began VCMT after a year of unspecified gender-affirming hormone therapy (GAHT) failed to resolve negative feelings about her voice. Her goals included an increase in conversational mean F0 and a 10-point decrease in TSEQ score. At her final assessment after 10 sessions, her mean F0 had decreased from baseline (129 Hz to 118 Hz); this was potentially explained by the patient having a “difficult week” leading up to this test, and due to the subject’s chosen subject matter of conversation focusing on an upsetting topic. Improvements were observed, however, in mean paragraph reading F0 (144 Hz to 165 Hz) and sustained phonation on /a/ and /i/ vowels (143 Hz to 230 Hz and 191 Hz to 235 Hz, respectively). Her F0 range in spontaneous speech (85–372 to 88–371) and paragraph reading (87–317 to 90–372) also showed changes congruent with her goals. In addition, her TSEQ score improved from 85 to 60, exceeding the 10-point improvement goal.

Participant 4 (they/them) attended 29 VCMT sessions over 3 semesters, the longest course in this cohort. They separately received estrogen-based GAHT, and sought VCMT with goals of increasing mean F0 and decreasing TSEQ score by 10 points; while a gender-neutral voice was an initial target, they ultimately sought a “somewhat feminine” voice. Notably, their first assessment after 12 weeks took place during Spring 2020, shortly after the outbreak of the COVID-19 pandemic, and incomplete data were collected. Participant 4 self-reported their mean F0, recorded using a phone app of unknown reliability, as 185 Hz, a significant improvement from baselines of 121 Hz in paragraph reading and 112 Hz in spontaneous speech. Their TSEQ score had markedly improved as well (73 to 56). At their final assessment after three semesters, complete data were collected by their speech pathologist. Participant 4’s recorded mean F0 in paragraph reading was 175 Hz (from 121 Hz), and their spontaneous speech mean F0 was 164 Hz (from 112 Hz). Their F0 ranges for paragraph reading (70–309 to 79–326) and spontaneous speech (81–314 to 138–237) also demonstrated improvement. Their final TSEQ was 63 (from 73), thus reflecting success across both F0 and TSEQ goals.

Participants’ individual outcomes are compiled in Table 1. None of the participants developed significant dysphonia requiring further treatment or referrals.

Table 1.

Individual clinical outcomes

ID	TSEQ			Spontaneous speech					Reading
				Mean F0 (Hz)			Range (Hz)		Mean F0 (Hz)			Range (Hz)
	Pre	Post	Δ	Pre	Post	Δ	Pre	Post	Pre	Post	Δ	Pre	Post
1	57	68	+11	141	195	+54	77–504	77–509	146	204	+58	76–499	84–502
2	65	69	+4	211	177	−34	91–385	76–361	219	181	−38	83–386	78–309
3	85	60	−25	129	118	−11	85–372	88–371	144	165	+21	87–317	90–372
4*	73	63	−10	112	164	+52	81–314	138–237	121	175	+54	70–309	79–326

ID = participant identification number

TSEQ = Transsexual Self Evaluation Questionnaire

F₀ = fundamental frequency

Hz = hertz

Pre = Pre-training/Baseline

Post = Post-training

Δ = change

^*:Participant 4 received 3 semesters of VCMT rather than only 12 weeks; their data after 12 weeks were incomplete, and thus the table includes data reflecting their full 3 semesters of VCMT.

DISCUSSION

In the present study, outcomes for NBGQ people undergoing a voice modification program are described. Notably, while all four participants represented in this case series showed improvement in at least one of their initial goals, only one improved both their F0 and TSEQ score; the other three participants showed improvements in controlled speech tasks such as reading, but had mixed results with generalization to spontaneous speech and with achievement of TSEQ score improvement goals. Thus, while VCMT proved at least moderately effective for all four participants, these results emphasize the fact that approaches to gender-affirming voice training have historically been developed for trans women rather than NBGQ people. While recently published guidelines^39–41 and textbooks^42–44 aim to provide an improved framework for voice clinicians in their work with NBGQ people, there is a clear gap in research investigating the outcomes associated with these techniques. Especially given the unique needs of this population, further study is required to establish an evidence base for VCMT strategies targeted towards NBGQ individuals. In addition, the observed discrepancy between TSEQ and F0 in three participants may also suggest that the TSEQ is an inadequate assessment tool to measure voice perception in NBGQ people. This highlights the overwhelming need to develop new validated assessment tools of vocal self-perception that are more inclusive to all gender identities.

Shefcik and Tsai (2021) recognized the need for a validated instrument to query voice-related concerns in the NBGQ population and proposed the first tool (VENI) to address this gap. They conducted a qualitative study with 10 NBGQ participants who desired voice modification and developed a questionnaire to capture their voice-related experiences. Though their initial findings supported the instrument’s content validity, use of this instrument is limited by the need to evaluate its construct and criterion validity in addition to its test-retest reliability. Of note, their study population also varied in their goals for voice modification, reiterating the importance of individualized, client-centered interventions for VCMT.⁷

Our findings build upon those of prior studies showing a discordance between acoustic measures and voice satisfaction. For instance, in earlier studies involving trans women, F0 correlated with self- and listener-perception of vocal femininity but not voice satisfaction as determined by overall voice happiness³² or TWVQ scores.^37,38 Moreover, Hardy et al. demonstrated that F₀ did not correlate significantly with TWVQ scores or overall quality of life.²⁹ While our study participants were NBGQ, we hypothesized that changes in acoustic measures would not necessarily correlate with changes in self-reported outcomes (TSEQ scores) given F₀ is one of numerous factors that contribute to voice satisfaction; this was confirmed by three participants realizing an improvement in either F0 or TSEQ, with a worsening in the other outcome.

Our results further demonstrate that F0 alone is an insufficient endpoint to determine the success of voice training. Bensoussan et al. previously showed that fundamental frequency alone is insufficient for the perception of a “female” voice using a validated artificial intelligence tool, reinforcing the complex nature of conversational speech that involves multiple acoustic and prosodic measures that go beyond pitch.⁴⁵ The degree to which factors besides pitch can influence perceived gender of human speech has also been studied; Leung et al. reported in a meta-analysis that fundamental frequency contributed to only 41.6% of the variance of gender perception.⁴⁶ Within our cohort, those with minimal improvement in F₀ – or even changes in F₀ in the opposite direction from their goals – experienced improvements in voice satisfaction. For example, participant 3 did not achieve her desired increase in F₀ (decrease of 11 Hz in spontaneous speech and 6 Hz in the frequency range for speech), but saw significant improvements in TSEQ score (decrease of 25 points). Conversely, participant 1 demonstrated a sizable increase in F₀ (increase of 54 Hz in spontaneous speech), but this was accompanied by worsening of their TSEQ score (increase of 11 points). These results again highlight the importance of an individualized approach to VCMT for each client based on their distinct goals.

Despite the dearth of literature on NBGQ experiences as they relate to voice and clinical outcomes, much has been written in recent years on the “genderless” synthetic voice in the fields of HCI, consumer research, and audio engineering. Indeed, the HCI and AI communities have realized the potentially harmful implications of gendered voice AI assistants. For instance, the default gender for commercial voice-activated assistants is female (e.g., Amazon’s Alexa or Apple’s Siri), raising the issue of whether this decision mirrors traditional and problematic societal stereotypes of women playing a subservient role.⁴⁷ Though gender perception and implicit biases are deeply rooted in our society, Baird et al. (2018) found that when listeners were not given strict binary options to rate the perceived gender of gender-neutral synthesized voices, they did not assign the voices to a binary gender, suggesting that the experience of gender in synthesized voice is not merely a decision made by listeners alone.⁴⁸ Additionally, Lee Turner et al. (2020) argued that companies have the freedom to create a gender-neutral synthetic voice so long as they are careful to avoid gender-framing.¹⁵ One study demonstrated that all gender non-conforming participants wanted the option to change their voice assistants’ voice to an “ambiguous” voice or to one that “sounds like me.”⁴⁹ However, the wealth of literature on the “genderless” voice in the technology world, compared to the relative dearth of scholarship about NBGQ individuals in the medical literature, creates an artificial and simplistic embodiment of what some may conceive as a non-binary voice. As illustrated by this study, the vocal goals of NBGQ individuals are myriad. Therefore, research is important to highlight the diversity of voice identities in this community, in contrast with the simplistic parameters set up by technology; “genderless” voice is not the same as the NBGQ voice.

Our study has several limitations. The primary limitation to this study is sample size, preventing statistical analysis. Due to the individualized nature of the program and finite resources, recruitment capabilities were limited. We also did not collect data on prior voice training, singing experience, homework compliance, or individuals’ perceptions of progress toward their voice goals, nor did we assess for additional social factors such as individuals’ confidence in their gender identity or progress in their transition (if any) outside of their voice. These factors, and others not mentioned here, may have impacted treatment outcomes, and future work should aim to control for these variables; this analysis was retrospective and limited by previously collected data, and further study is planned to address these gaps. Measurements of some outcomes were also limited by the COVID-19 pandemic, which forced sessions in spring 2020 to be conducted via tele-health. It is possible that including more acoustic measures may better capture elements that collectively contribute to voice satisfaction. This may be particularly true of resonance measures, such as formant frequency, which will be a target of future work. Future directions for this work include a greater sample size allowing for statistical analysis, potentially including additional clinical outcomes and acoustic measures across different vocal tasks to assess for correlations between specific objective outcomes and subjective self-assessments.

Based on our findings and the work of prior researchers, including healthcare professionals who are advancing guidelines for voice care for the gender-diverse community, we recommend tailoring voice and communication modification techniques to each individual and their goals rather than a “one size fits all” program.^39,40 More clinical science is needed to further address the voice needs of the growing NBGQ population. The target outcome should be based entirely on the individual’s vision for their voice and how it can better serve to affirm their gender identity rather than aiming to reach voice parameters associated with norms based on the gender binary.⁵⁰ Practitioners should remain flexible in their approach to an individual’s care and not project their own personal biases while providing therapeutic feedback. An individualized approach further highlights the importance of subjective voice and quality of life measures to judge success of interventions as opposed to arbitrary acoustic parameter cut-offs.

CONCLUSION

Transgender and NBGQ individuals’ voices are tied closely to their sense of gender identity and quality of life. NBGQ individuals can experience significant improvements in patient-reported outcomes measures and changes in acoustic measures after completing a gender-affirming VCMT program. Changes in acoustic measures do not necessarily correlate with self-reported outcomes measures, emphasizing the need for individualized training based on the client’s goals – which may or may not include achieving a voice perceived as gender-neutral – as well as psychometric tools validated for this unique population. The demand for client-centered approaches will only continue to increase as we continue to move towards personalized care in the healthcare setting.