Effectiveness of Speech Therapy in Trans Women.

One of the greatest hurdles in the transition of transgender persons is that voice, speech and communication are not congruent with the desired gender. Since hormone treatment does not affect the voice in male-to-female transgender persons (trans women), speech therapy is the treatment of choice to develop a more feminine communication. Speech therapy must focus on aspects of communication that play an important role in listener perceptions of the speakers gender. Results of a systematic review and meta-analysis showed that those aspects are primarily fundamental frequency of the voice and resonance. However, effectiveness studies of speech interventions in transwomen are extremely limited and show methodological limitations. The purpose of this project is to investigate the short-term and longterm impact of speech exercises for pitch and resonance on (a) acoustic voice characteristics, (b) listener perceptions of femininity using a visual analogue scale and binary gender identification (male versus female voice), and (c) self-perception and psychosocial functioning in trans women using a randomized sham-controlled trial and cross-over design.

The main aim of this project is to investigate the short-term and long-term effect of speech therapy for pitch and resonance on vocal characteristics, listeners' perceptions and self-perception in trans women using a randomized sham-controlled trial and a longitudinal cross-over design. The goals of speech therapy are selected based on a systematic review and meta-analysis: pitch and resonance. The first purpose is to determine and compare the short-term effect of two therapy programs Pitch Elevation Therapy (PET) and Articulation-Resonance Therapy (ART), in trans women on (a) acoustic voice characteristics investigating vocal quality, pitch and resonance, (b) listeners' perceptions using a combination of (binary) gender identification and femininity ratings using a Visual Analogue Scale (VAS) and (c) self-assessment investigating the client's satisfaction of the therapy and psychosocial functioning, using a randomized controlled trial. Hypotheses: Outcome acoustic voice characteristics: Based on studies and our pilot study investigating the immediate effects of resonance and articulation exercises in trans women, a significant increase in all formant frequencies, and an increase in speaking fo is hypothesized after ART. Based on studies, an increase in speaking fo between 10 and 84Hz is hypothesized after PET. The speaking fo and formant frequencies are primary outcome measures. Vocal training and increasing pitch (during the PET) is associated with increasing vocal load and has a possible negative effect on vocal quality. Monitoring vocal quality in the three groups is therefore important. PET includes gliding tones and consequently can have a possible effect on frequency range. Therefore vocal quality and vocal range parameters are included as secondary outcome parameters. PET and ART are designed to feminize the voice without vocal misuse. Therefore, measurements of vocal quality are hypothesized to remain stable after treatment. After PET, an increase in vocal range is hypothesized. Outcome listeners perceptions: According to the systematic review and meta-analysis of Leung et al., pitch and resonance are the most salient listener perceptions of speaker gender. Based on the results of our pilot study, and other studies, it is hypothesized that both therapy programs, PET and ART, will result in a more feminine speech rated on a VAS. Based on the results of Gelfer et al. we hypothesize that 50% of the participants will be perceived as female after the PET therapy. Outcome self-assessment: Based on a retrospective study, case studies, and another study, significant improvement in psychosocial wellbeing and high levels of satisfaction with the post-treatment voice are hypothesized in both programs. The second purpose of the study is to determine the short- and long-term effects of the whole therapy program, consisting of the combination of both PET and ART, in transwomen on (a), (b) and (c) immediately after therapy and at 3 months and 1 year using a cross-over design. Hypotheses: Outcome acoustic voice characteristics: Based on a retrospective study and prospective studies on small sample sizes investigating long-term outcome of speech therapy, a small decrease of the speaking fo is hypothesized in the long-term follow-up after one year. The long-term outcome of the formant frequencies remains unclear. Outcome listeners perceptions: Based on a pilot study, it is hypothesized that the listeners perceptions immediately after therapy will not differ from the listeners perceptions 3 months and 1 year after therapy. Outcome self-assessment: Based on the results of a study and because speech is a crucial element in the psychosocial functioning of a trans women, an increasing improvement in activity and participation after 3 months and 1 year is hypothesized. Randomized sham- controlled trial group 1 (n=26): pre sham post 1 PET post 2 ART post 3 follow-up 1 follow-up 2 group 2 (n=26): pre sham post 1 ARTpost 2 PET post 3 follow-up 1 follow-up 2 Study design. A randomized sham-controlled trial will be used to determine and compare the short-term effect of PET and ART in trans women. By adding a cross-over design, the effect of a combined treatment (PET and ART) can also be investigated, both on the short- and longer term. Motivation: A randomized controlled trail is considered the most reliable form of scientific evidence in efficacy studies. The use of a sham therapy will take into account placebo an therapist effects. A cross-over design is used for ethical reasons, so that each subject will receive the whole voice feminization approach (PET and ART) in an acceptable time frame. Follow-up data are included to determine long-term effects of the intervention and track which subjects still need phonosurgery or further behavioral intervention. Participants. Sample size. To date, no randomized controlled trials investigated the effect of speech therapy in transwomen. Therefore, sample size calculation was based on the study of Dacakis. This pilot study gave a minimal sample size of n = 18 participants per group (SAS, Power and Sample Size). Minimal sample size was calculated for the fundamental frequency with an α level of 0.05, a power of 80%, and a SD of 20 and 15. Taking into account a mean drop-out rate of 30%, minimal sample sizes are increased from 18 to 26 participants per group. It guarantees a large sample size to perform data-analysis with sufficient power. Taking into account the research group's expertise in the field and the accessible recruitment via the gender clinic of UZ Gent, a total of n = 78 participants is a very feasible number. Inclusion criteria and exclusion criteria for the trans women are described further in the application. The multidimensional voice assessment will be performed by an SLP (different from the therapist) with more than 5 years of experience of the research group who is blinded to group allocation and study evolution. Assessments will be planned before the treatment (pre), immediately after sham therapy (post 1), immediately after therapy 1 (post 2), immediately after therapy 2 (post 3) and at three months (follow-up 1) and 1 year (follow-up 2) follow-up. All measurements will be performed in a sound treated room (speech lab) at Ghent University Hospital at the department of Rehabilitation Sciences. Listening experiment. Listeners: An equal number of male and female naïve, native Dutch speaking laypersons (n=60), aged (and equally spread) between 20 and 49 years old, with a normal hearing will be recruited for the listening experiment. All listeners will be blinded to the purpose of the study and the gender of the speakers of the speech samples. The hearing levels of all listeners will be evaluated by the audiologists of the research team using audiometry. Table 1: Number of listeners per age group, gender and task. VAS-scale Binary rating M (n) F (n) M (n) F (n) 20-29y 5 5 5 5 30-39y 5 5 5 5 40-49y 5 5 5 5 Method: The listeners, who are blinded to the purpose of the study, will be randomly assigned to 2 groups to rate audio samples on masculinity/femininity using a VAS (group 1, n=30) and gender identification (male voice versus female voice) (group 2, n=30). Speech samples: From the speech samples of continuous speech during reading (using a phonetically balanced text) the middle sentences (+- 15 seconds) will be selected and used in the experiment. For the listening experiment E-prime 3 software is used to randomize and present the samples. Ratings: The raters will listen once to the samples presented by E-prime 3 software during a PowerPoint presentation using a laptop and over ear headphones in a quiet sound treated room. They can include pauses to maintain concentration levels. Intensity levels will be presented at a comfortable level for the listener (approximately at 70dB at ear level). The listeners will also receive 2 other questions (distracting them from the purpose of the study) regarding the age and vocal quality of the speakers. They will rate the audio samples of all participants (n= 78) at all test moments and of 13 cis-male and 13 cis-female control persons. Inter-rater reliability will be calculated. 15% of the samples will be randomly repeated to calculate intra-rater reliability. Speech therapy procedures. Trans women will be randomly assigned to one of the 2 groups (PET, ART) using randomization. All groups will receive the same dosage of individual therapy: 10 sessions of one hour during 5 weeks. The interventions will have an equal duration (5 sessions) and will be provided in a different order in both groups: Pitch Elevation Therapy (PET) and Articulation Therapy (ART). All therapies, including the sham therapy, will be provided by the same experienced therapist of the research group, having more than 10y of experience in this field, to control for therapist effects and assure optimal treatment fidelity. Interventions will take place in a sound treated room (speech lab) at Ghent University Hospital (dep. of Rehabilitation Sciences). PET and ART will be described in detail for the publications of this project to ensure replicability. The control group will receive information sessions regarding transgender logopaedic and general health care and, communication aspects, an active intervention for non-verbal communication skills and breathing patterns. Therefore, it is feasible to provide the same duration as the PET and ART interventions. The same hierarchical structure will be used for both therapy programs and is based on a comprehensive literature searching. Statistical analysis. Linear mixed model analyses (LMM) (incl. baseline analyses) will be used to determine whether there is a difference in evolution between the therapy groups over time (interaction effect time*group). Within- and between-group differences will be analyzed using post-hoc pairwise comparisons (with Bonferroni correction).

Goal: mean fundamental frequency of 180Hz (range 140 and 300Hz); Doses and duration: 5 one hour sessions (5h), 5 weeks Content: auditory discrimination of lower and higher tones pitch glides pitch glides using straw phonation and resonance tube phonation practice of higher pitch voice (target: 180Hz) in isolated vowels - syllables - mono and bisyllabic words -phrases - sentences and texts - conversation (generalization to spontaneous speech)

Goal: increase in lipspreading, forward tongue carriage, larynx elevation, forward resonance, precise articulation; 5 sessions Doses and duration: 5 one hour sessions (5h), 5 weeks Content: auditory and visual discrimination of forward vs backed tongue carriage and lip spreading vs lip rounding) practice of technique on isolated vowels -syllables -mono and bisyllabic words - phrases - sentences and texts - conversation (generalization to spontaneous speech) practising larynx elevation through the twang technique, isolated vowels -syllables -mono and bisyllabic words - phrases - sentences and texts - conversation (generalization to spontaneous speech) forward resonance, focusing on oral forward resonance rather than chest resonance, isolated vowels -syllables -mono and bisyllabic words - phrases - sentences and texts - conversation (generalization to spontaneous speech) clear, precise articulation and cork exercise

Goal: active sham treatment, 4 sessions Doses and duration: 4 one hour and 15 minutes sessions (5h), 4 weeks Content: providing information about vocal function, vocal surgery, vocal hygiene, transgender health care discussing communication aspects and patterns, possible exercise contexts, communication partners, expectations of voice modification, conversational subjects active intervention on nonverbal communication (Kinesics, gestures and sign language, facial expression, haptics, proxemics, occulsics, and physical appearance) practising costo abdominal breathing patterns

pre, post 1 (immediately after sham), post 2 (immediately after intervention 1), post 3 (immediately after intervention 2), follow-up 1 (after 3 months), follow-up 2 (after 1 year)

pre, post 1 (immediately after sham), post 2 (immediately after intervention 1), post 3 (immediately after intervention 2), follow-up 1 (after 3 months), follow-up 2 (after 1 year)

pre, post 1 (immediately after sham), post 2 (immediately after intervention 1), post 3 (immediately after intervention 2), follow-up 1 (after 3 months), follow-up 2 (after 1 year)

pre, post 1 (immediately after sham), post 2 (immediately after intervention 1), post 3 (immediately after intervention 2), follow-up 1 (after 3 months), follow-up 2 (after 1 year)

The DSI is a multiparametric approach designed to establish an objective and quantitative correlate of the perceived vocal quality [19]. It is based on a weighted combination of 4 voice parameters: maximum phonation time (MPT, s), highest frequency (F-high, Hz), lowest intensity (I-low, dB) and jitter (%). The DSI is constructed as 0.13 MPT + 0.0053 F-high - 0.26 I-low - 1.18 jitter +12.4. The index ranges from -5 to +5 for severely dysphonic to normal voices but values higher than +5 are possible in subjects with excellent vocal capacities.

pre, post 1 (immediately after sham), post 2 (immediately after intervention 1), post 3 (immediately after intervention 2), follow-up 1 (after 3 months), follow-up 2 (after 1 year)

The AVQI is a robust and valid method to quantify the severity of overall dysphonia based on both continuous and sustained vowel recordings [20]. This multiparameter index consists of a weighted combination of 6 voice parameters: smoothed cepstral peak prominence (CPPS), harmonics-to-noise ratio (HNR), shimmer local (SL), shimmer local dB (SLdB), general slope of the spectrum (slope) and tilt of the regression line through the spectrum (tilt). The formula is constructed as 9.072 - 0.245 × CPPs - 0.161 × HNR - 0.470 × SL + 6.158 × SLdB - 0.071 × Slope - 0.170 × Tilt and ranges from 0 to 10. The higher the score of the AVQI, the worse is the overall vocal quality.

pre, post 1 (immediately after sham), post 2 (immediately after intervention 1), post 3 (immediately after intervention 2), follow-up 1 (after 3 months), follow-up 2 (after 1 year)

The frequency and intensity range of the voices will be measured with the Voice Range Profile software of the Computerized Speech lab of Kay following the procedure of Heylen et al.

pre, post 1 (immediately after sham), post 2 (immediately after intervention 1), post 3 (immediately after intervention 2), follow-up 1 (after 3 months), follow-up 2 (after 1 year)

Study population are trans women or AMAB (assigned male at birth). Inclusion criteria are established diagnose and gender identity confirmed by the gender team, under hormonal treatment (confirmed by blood analysis), female gender role, perceived as male speaker at the beginning of the therapy by the investigators and speaking fo within the male or ambiguous zone (<180Hz).

Inclusion Criteria: established diagnose and gender identity confirmed by the gender team, aged between 18 and 70 years, self-reported normal hearing, native Dutch speaker, under hormonal treatment (confirmed by blood analysis), female gender role, perceived as male speaker at the beginning of the therapy by the investigators speaking fo within the male or ambiguous zone (<180Hz). Exclusion Criteria: history of neurological disorders, previous phonosurgery or speech therapy pathology of the vocal folds (observed by videolaryngostroboscopic examination of the vocal cords), smokers. to form a homogeneous study population gender non-binary persons will not be included in this study.

Gelfer MP, Tice RM. Perceptual and acoustic outcomes of voice therapy for male-to-female transgender individuals immediately after therapy and 15 months later. J Voice. 2013 May;27(3):335-47. doi: 10.1016/j.jvoice.2012.07.009. Epub 2012 Oct 22.

Dacakis G. Long-term maintenance of fundamental frequency increases in male-to-female transsexuals. J Voice. 2000 Dec;14(4):549-56. doi: 10.1016/s0892-1997(00)80010-7.

Leung Y, Oates J, Chan SP. Voice, Articulation, and Prosody Contribute to Listener Perceptions of Speaker Gender: A Systematic Review and Meta-Analysis. J Speech Lang Hear Res. 2018 Feb 15;61(2):266-297. doi: 10.1044/2017_JSLHR-S-17-0067.

Gelfer MP, Mikos VA. The relative contributions of speaking fundamental frequency and formant frequencies to gender identification based on isolated vowels. J Voice. 2005 Dec;19(4):544-54. doi: 10.1016/j.jvoice.2004.10.006.

Carew L, Dacakis G, Oates J. The effectiveness of oral resonance therapy on the perception of femininity of voice in male-to-female transsexuals. J Voice. 2007 Sep;21(5):591-603. doi: 10.1016/j.jvoice.2006.05.005. Epub 2006 Jul 5.

Soderpalm E, Larsson A, Almquist SA. Evaluation of a consecutive group of transsexual individuals referred for vocal intervention in the west of Sweden. Logoped Phoniatr Vocol. 2004;29(1):18-30. doi: 10.1080/14015430310021618.