Perturbation and Harmonics to Noise Ratio As a Function of Gender in the Aged Voice

Perturbation and Harmonics to Noise Ratio as a Function of Gender in the Aged Voice

THESIS

Presented in Partial Fulfillment of the Requirements for the Degree Master of Arts in the Graduate School of The Ohio State University

Meredith Margaret Rouse Hunt

Graduate Program in Speech and Hearing Science

The Ohio State University

2012

Master's Examination Committee:

Michael Trudeau, Advisor

Michelle Bourgeois

Copyrighted by

Meredith Margaret Rouse Hunt

2012

Abstract

The purpose of this investigation was to explore possible differences as a function of gender in perturbation (jitter and shimmer) and harmonics to noise ratio (HNR) among aged male and female speakers. Thirty normal aged adults (15 males; 15 females; over age 60) prolonged the vowel /a/ at a comfortable loudness level. Measures of jitter (%), shimmer (%), and HNR were used to compare vocal function between aged gender groups. No significant differences were found between genders on any of the measures.

Findings are discussed relative to other published studies on similar measures and support data that aged voices exhibit increased variability. Future suggestions for research are discussed.

Dedication

This manuscript is dedicated to my husband, Ryan, for his unfailing patience, support, and humor during the completion of my thesis and in all aspects of my life.

iii

Acknowledgments

I would like to acknowledge Michael Trudeau, Ph. D., CCC-SLP, my academic and thesis advisor, for his gentle and persistent guidance. His dedication to teaching and patience with students has allowed me to become adept at critical evaluations of research and treatment methodology. More importantly, his love of voice science and care for his clients has shaped my future professional career as speech-language pathologist. I cannot thank him enough for the opportunities and mentorship he has provided me.

I would also like to acknowledge Michelle Bourgeois, Ph. D., CCC-SLP, for her unending enthusiasm for research; without whose passion and support I would not have attempted a thesis project. Thank you for making research and methods understandable and attainable. I would like to acknowledge Kerrie Obert, M. A., CCC-SLP, the principle investigator. The document that follows could not have been possible without her generosity in allowing me to take part in her research.

Finally, I would like to acknowledge the Department of Speech and Hearing

Science for graciously providing lab space to complete the data collection process.

Vita

2005...... B.A. in Liberal Studies with a focus in

Communication Sciences and Disorders,

Northern Arizona University

2010 to present ...... Graduate Clinician, Department of Speech

and Hearing Science, The Ohio State

University

Fields of Study

Major Field: Speech and Hearing Science; Speech-Language Pathology

Table of Contents

Abstract ...... ii

Dedication ...... iii

Acknowledgments...... iv

Vita ...... v

List of Tables ...... viii

Chapter 1: Introduction ...... 1

Chapter 2: Methods ...... 10

Screening Procedures ...... 10

Participants ...... 11

Equipment ...... 14

Measures...... 15

Procedures ...... 17

Reliability ...... 17

Chapter 3: Results ...... 21

Chapter 4: Discussion ...... 25

References ...... 31

Appendix A: Informed Consent ...... 34

Appendix B: Participant Questionnaire ...... 40

Appendix C: Screening Protocol Form ...... 41

Appendix D: Study Participant Raw Data ...... 42

Appendix E: Test-Retest Reliability Data...... 43

vii

List of Tables

Table 1. Reported participants, ages (mean), shimmer (dB), jitter (%), harmonics to noise ratio (HNR), and recording conditions for genders only on /a/ phonations...... 7

Table 2. Descriptive male participant characteristics...... 13

Table 3. Descriptive female participant characteristics...... 14

Table 4. Intra-judge ratings comparison for GRBAS scale...... 19

Table 5. Inter-judge ratings comparison for GRBAS scale...... 20

Table 6. Group means (SD) and t-test results for Jitter (%), Shimmer(%), and harmonics to noise ratio (HNR) (dB) ...... 21

Table 7. Revised group means (SD) and t-test results using only subjects where G = 0. 22

Table 8. Group means (SD) for subjects aged 60-69 years...... 23

Table 9. Group means (SD) for subjects aged 70-79 years...... 24

Table 10. Study participant raw data ...... 42

Table 11. Test-retest data for sustained /a/ task averaged across three recorded trials. ... 43

Table 12. Test-retest data for continuous speech sample...... 44

Table 13. Test-retest data for maximum performance tasks ...... 45

viii

Chapter 1: Introduction

As the population in the United States ages and the elderly population growth outpaces growth in other age groups it is necessary to understand the effects of aging on the voice (Brock, 1990). With the increasing number of elderly clients seeking services from speech-language pathologists due to complaints of voice changes or difficulties, clinicians must be able to delineate whether those changes are part of the normal aging process of the voice or a disorder of aging. Many avenues are available for clinicians to judge voice parameters, yet there is a lack of sufficient normative data for the aged voice.

Anatomical and physiological changes in the larynx due to the aging process have been reported and reviewed by several researchers. Established age-related changes to the larynx include ossification and calcification of the laryngeal cartilages, degeneration of the cricoarytenoid joint, decreased blood supply, bowing of the vocal folds, reduction and degeneration of nerve fibers innervating the laryngeal musculature, and a reduction in vital capacity of the lungs (Colton, Casper, & Leonard, 2011; Kahane, 1987 & 1990).

These changes in the aging larynx may alter the perceptual, acoustic, and physiologic characteristics of phonation.

Research in gerontology and the aging of the larynx and peripheral speech mechanism has outlined measurable change in all humans across the life-span. The research has regularly defined that those degenerative changes occur earlier in the life- span in men than in women (Kahane, 1990). Changes are also thought to be of greater

1 extent in male anatomy versus female anatomy. This may lead one to logically deduce that there should be greater perturbation in the aged male voice than the aged female voice. Kahane‟s 1990 review of relevant literature on age-related structural and physiological changes in the speech mechanism provides some of the most comprehensive information regarding the ways this integrated mechanism shows its age.

However, the body of research concerning how anatomical evidence may predict differences in older speakers is equivocal (Awan, 2006; Ferrand, 2002; Bohnenkamp &

Hageman, 2011; Harnsberger, Shrivastav, Brown, Rothman, & Hollien, 2008; Linville &

Fisher, 1985). Structural changes in the larynx and peripheral speech mechanism may not clearly result in functional changes in phonation allowing for age prediction. This study hopes to provide data related to this idea and a basis for future research in gender functions of the aged voice. Within the scope of this study it is necessary to note the term

„aged‟ is used to describe primarily those voices in the range of 60 to 80 years. Further investigation should also be conducted on the very aged voice to include those persons over 80 years.

Voice scientists have researched acoustic aspects of normal phonation in the aged voice population including fundamental frequency (F0), vocal intensity, perceptual characteristics, perturbation, harmonics to noise ratio (HNR), glottal closure patterns and gaps, vocal fold movement, periodicity, maximum phonation frequency range and many others. Most studies have looked at the genders separately, often determining change in phonation parameters across the age spectrum within one gender. With increasing availability of computer programs that automate phonatory measurements via pre- programed algorithms applied to voice samples instantaneously, numerous acoustic

2 measures are quickly and easily attainable. Measures of stability are thought to reflect control of the voice and are often examined aspects of vocal aging (Linville, 2001).

Changes in stability of phonation are considered central to the characteristics of aged phonation (Baken, 2005), leading to numerous studies attempting to examine values of perturbation and quantify normal change in acoustic stability related to the aging process.

It has been generally accepted that aged male voices exhibit a higher fundamental frequency, greater variability in fundamental frequency and intensity, increased perturbation with jitter increasing to a greater extent than shimmer, and increased signal noise when compared to young male voices; and that aged female voices present with lowered fundamental frequency, increased variation in fundamental frequency, and particularly increased jitter when compared to young female voices (Linville, 1987;

Linville & Fisher, 1985; Sussman & Sapienza, 1994; Wilcox & Horii, 1980).

The amount of amplitude perturbation, or shimmer, present in a voice depends on the vowel produced and gender. Shimmer is defined as the cycle to cycle irregularity in the amplitude of vocal fold vibration. Jitter, or frequency perturbation, is described as frequency variations produced when attempting a sustained vowel and result from instability of the vocal folds during vibration (Colton, Casper, & Leonard, 2011). These measures are gained from sustained vowel phonation and can be jointly referred to as perturbation. The amount of jitter has been shown to vary based on age, physical condition, and gender.

Jitter and shimmer differences as a function of gender was first discussed by

Sorensen and Horii (1983). They outlined the primary finding that adult female voices were characterized by increased jitter and decreased shimmer when compared to adult

3 male voices (Horii, 1980). The shimmer values only approached significance as a function of gender, with female and male means (standard deviations) of .25 (.11) and .39

(.31), respectively. The increased jitter values in females versus males reached significance with means of .84 (.30) and .64 (.24), respectively. Their data included only adults under age 40 and should not be directly applied to an older population. However, the authors recommended the continued separation of normative values by gender in future research. Evaluating these differences across age has been a focus of research since this time, but most studies have not compared differences as a function of gender in only the aged population (Awan, 2006; Ramig & Ringel, 1983; Sussman & Sapienza, 1994;

Xue & Deliyski, 2001).

Brockmann, Drinnan, Storck, and Carding (2011) discuss that their research demonstrated a decrease of jitter and shimmer in both genders in phonations above a threshold of 80 decibels (dB). However, they also note that “comfortable loudness and pitch” tasks for both genders fall below their critical threshold of 80 dB. Their recent work, while outlining that sound pressure level (SPL) has the greatest impact on jitter and shimmer, still demonstrates effects of gender that may be clinically relevant. These emerging gender data must be further investigated in the aged population to determine if the relationship remains significant. The authors agreed it is still appropriate to delineate gender specific normative data. Sussman and Sapienza (1994) also lend support to gender differences and the establishment of gender specific normative data sets as they reported significant gender differences in fundamental frequency (F0) and jitter in adults.

However, both of these studies included only young adults leaving the need for continued

4 research to determine if significant gender differences are present with the aging of the speech mechanism.

Stathopoulos, Huber, and Sussman (2011) demonstrated the nonlinearity of aging across the life-span in terms of F0 and signal to noise ratio (SNR). Their work further supports aging differences as a function of gender, particularly related to noise characteristics in the voice. Age related changes as a function of gender were discussed as largely resultant from the hormonal changes related to female menopause. The widely accepted changes associated with menopause are known to affect bone, skin, and muscles contributing to increased laryngeal edema and vocal fold thickening. However, the study was large and cross-sectional, with limited and unequal numbers of participants between genders falling into an elderly category. The study also failed to discuss differences in actual measurement of SNR in reference to age groups, only addressing it as differing between genders across the aging timeline.

Many researchers have reviewed acoustic and perceptual characteristics of the aging voice, yet broad agreement does not exist among professionals regarding the type and extent of changes. Acoustic voice measures of jitter, shimmer and noise to harmonics ratio have been found to increase as a function of aging (Ferrand, 2002; Gorhan-Rowan

& Laures-Gore, 2006; Linville & Fisher 1985; Xue & Deliyski, 2001). Evaluation of these characteristics has primarily investigated changes between young and old voices within one gender (see Table 1). Brockmann, Storck, Carding, and Drinnan (2008) evaluated perturbation levels dependent upon vocal intensity finding jitter and shimmer to be significantly affected by voice loudness and gender differences in perturbation.

While the increased jitter measures in male subjects at softer phonation levels supports

5 gender effects of perturbation, only young adults were used for evaluation. Table 1 outlines the results from the authors discussed above.

Author Participants Ages (mean) Jitter (%) Shimmer (%) HNR Recording Conditions Horii 10cm distance in a 31 males 18-38 (26.6) .61 (.20) .47 (.34) (1980)a sound treated booth Sorensen & 10cm distance in a Horii 20 females 25-49 (36.8) .71 (.22) 0.33 (.22) sound treated booth (1983) a Ramig & 26-35 (29.5) .424 (.089) .275 (.127) 48 males, Ringel 25-38 (32.3) .497 (.092) .274 (.178) grouped by (1983) 46-56 (53.0) .501 (.103) .349 (.160) 15cm distance in a age and 42-59 (52.6) .700 (.319) .425 (.298) sound treated booth physical 62-75 (67.5) .596 (.147) .364 (.090) 7 condition 64-74 (69.1) .646 (.142) .426 (.249) 75 females 25-35 5.23 (1.81) Linvilleb, c 25 per group 45-55 5.50 (2.98) not reported (1987) by age 70-80 9.13 (13.09) .590 (.094) .371 (.136) 60-80dB Orlikoff & .467 (.069) .307 (.105) 70-78dB Kahane 10 males 26-37 (31.8) .252 (.074) .209 (.111) 80-88dB (1991) a,f 30cm distance in a sound treated booth Xue & 21 males 70-80 (75.43) 2.10 (1.55) 5.54 (3.51) .18(.08) 20cm in a “quiet Deliyski 23 females 70-80 (74.83) 2.02 (2.03) 5.34 (4.51) .20(.11) room” (2001)d Table 1. Reported participants, ages (mean), shimmer (dB), jitter (%), harmonics to noise ratio (HNR), and recording conditions for genders only on /a/ phonations.

Table 1 continued

~15cm (6in) distance in a 42 females 21-34 (25) .69 (.30) 7.82 (2.65) Ferrand sound treated booth, a 14 per group 40-63 (50) .57 (.27) 7.86 (3.14) (2002) “quiet schoolroom” and a by age 70-90 (77) .66 (.39) 5.54 (2.91) “quiet room” 18-30 (23.3) .36 (.17) 4.47 (1.95) 16.90 (3.59) 50 females 40-49 (43.4) .47 (.30) 4.60 (1.81) 17.80 (2.73) Awan 30cm distance, other 10 per group 50-59 (54.8) .82 (.77) 7.45 (4.66) 15.00 (4.62) (2006) conditions not reported by age 50-59 (65.2) .51 (.40) 6.61 (5.22) 16.42 (3.94) 70-79 (72.3) .46 (.33) 6.57 (3.36) 16.37 (3.89) 28 females 20-39 (28.8) .78 (.71-.86) 1.40 (1.27-1.54) 63.8dB (62.7-64.9) .36 (.32-.39) .50 (.46-.55) 72.5dB (71.4-73.6) Brockmann, 8 .22 (.20-.24) .12 (.11-.14) 96.8dB (95.7-97.9) Storck, 29 males 20-39 (28.11) .94 (.86-1.04) 1.0 (.91-1.10) 68.8dB (67.7-69.8) Carding & .31 (.28-.34) .28 (.26-.31) 79.2dB (78.1-80.2) Drinnan .21 (.19-.23) .13 (.12-.14) 98.8dB (97.8-99.9) (2008) a,e,f 10cm distance in a “soundproof” room Brockmann, Drinnan, 28 females 20-39 (28.8) .37 (.33-.40) .46 (.41-.51) 10cm distance in a Storck, & 29 males 20-39 (28.1) .30 (.28-.33) .31 (.27-.35) “soundproof” room Carding (2011) a,e,f Notes. aThe values reported for shimmer are measured in shimmer dB, not shimmer %. bLinville (1987) did not report the vowel recorded for values reported. cJR (jitter ratio) was used by Linville, 1987. dNHR (noise to harmonics ratio, the inverse of harmonics to noise ratio) was used by Xue & Deliyski (2001). eBrockmann, Storck, Carding, & Drinnan (2008) and Brockmann, Drinnan, Storck, and Carding (2011) reported 95% confidence interval and not standard deviation measures. fBorckmann, Storck, Carding, & Drinnan (2008) and Orlikoff & Kahane (1991) values are reported by phonation levels required of participants. Unless otherwise noted, all measurements were recorded at a “comfortable” or “habitual” pitch and volume level. 8

To date, little research has been completed to evaluate change in the aged voice as a function of gender. Therefore, the purpose of this study was to investigate if significant differences in perturbation characteristics and HNR of the normal aged voice exist as a function of gender. To make this determination the following questions were evaluated:

(1. Is there a significant difference in jitter (%) between male and female aged voices on a sustained vowel task? (2. Is there a significant difference in shimmer (%) between male and female aged voices on a sustained vowel task? (3. Is there a significant difference in the noise to harmonics ratio between male and female aged voices on a sustained vowel task?

Difference in fundamental frequency as a function of age was not evaluated due to the substantial evidence supporting well-established differences in fundamental frequency between male and female voices across the age span. This was not examined as the difference in F0 is largely explained anatomically by differences in the relative length of the vocal folds between genders (Titze, 1989).

Chapter 2: Methods

The data for this study were selected from part of a larger, multi-site project aiming to collect normative acoustics voice samples across the adult life-span. Only the portion of data collected for adults over the age of sixty from the larger project is evaluated here to consider differences in perturbation and HNR as a function of gender in the aged voice.

Screening Procedures

Forty-two volunteers from the community at large were recruited utilizing email lists via local churches and a university faculty club, as well as the investigator‟s family members, friends, and acquaintances. Volunteers were screened by the investigator to determine if inclusion criteria was met. At the time of informed consent (Appendix A), each participant completed a brief voice health questionnaire (Appendix B) that addressed overall vocal health, medication use, smoking history, and vocal training background. Volunteers who self-identified with a history significant for regular use of medications known to have a possible negative effect on the voice (i.e. antihistamines, decongestants, diuretics, some antidepressants, and some antianxiety drugs), recent smoking (within the last five years), severe allergies or asthma, known vocal fold pathology or treatment (surgical or voice therapy) for a previous voice disorder, and/or neuromotor impairment were then excluded from participating further in the study. Those

10 who identified as professional vocal soloists or for whom English was not their primary language used in communication were also excluded. Remaining volunteers included were then seated in a sound treated booth and administered a pure tone hearing screening.

Volunteers were required to demonstrate response by raising a hand when presented with pure tones at 500, 1000, and 2000 Hertz (Hz) and a maximum of 30 dB. Volunteers who did not respond at 30 dB or less bilaterally were excluded from participating further.

Remaining participants were then instructed to describe the “Cookie Theft” picture from the Boston Diagnostic Aphasia Examination (Goodglass, Kaplan, & Barresi, 2001) to obtain a 10 second continuous speech sample from which to rate vocal quality using the

Grade, Roughness, Breathiness, Asthenia, Strain (GRBAS) scale (Hirano, 1981). The

GRBAS scale is a perceptual evaluation tool with a rating scale of 0-3 across each voice quality parameter where a rating of 0 is normal, 1 is mild, 2 is moderate, and 3 is severe.

Volunteers who received a rating greater than 1 on any scale item were excluded from the study. Those volunteers passing all portions of the screening then completed the data collection protocol. See Appendix C for the Screening Protocol Form.

Participants

Of the original 41 volunteers, 11 were excluded: 6 did not pass the hearing screening; 2 with a GRBAS rating >1 in one voice characteristic; 2 were unable to phonate for at least 4 seconds to produce a reliable signal; 1 previously sought treatment for a voice disorder; and 1 reported smoking within the last five years. In total, 30 participants were included in the study. Participants were grouped by gender and age with 15 male subjects (ages 60-85, mean 69.33, SD 8.65) and 15 female subjects (ages

60-78, mean 66.60 years, SD 5.45). Criteria for participating in the larger study included: a.) age 18 or older; b.) speak English as their primary language; c.) no known or suspected voice disorders; d.) GRBAS scale rating <1 on each parameter and overall

Grade rating, as judged by a speech-language pathologist specializing in voice or the graduate research assistant; e.) hearing within normal limits, as determined by a hearing screening performed at the time of participation; f.) a non-singer (recreational choir singers included); and g.) non-smokers for at least the last five years. Exclusion criteria for the larger study included a.) previous voice surgery; b.) previous treatment for a voice disorder; c.) diagnosis of a voice disorder; d.) use of medications with known negative effects on the voice; however, inhalers, birth control and/or hormone replacement therapy were acceptable if voice was minimally affected as judged by GRBAS ratings; e.) self- reported history of severe respiratory allergies, asthma, vocal fold pathology, or neuromotor impairment. Tables 2 and 3 outline the descriptive characteristics of participants.

Subject ID Gender, n=15 Age G R B A S VNMC035 Male 61 0 0 0 0 0 VNMC052 Male 80 0 0 0 0 0 VNMC053 Male 85 0 0 1 0 0 VNMC054 Male 75 0 0 0 0 0 VNMC055 Male 67 0 0 1 0 0 VNMC056 Male 60 0 1 0 0 0 VNMC058 Male 72 0 1 0 0 0 VNMC061 Male 79 0 0 1 0 0 VNMC063 Male 63 1 1 1 0 0 VNMC064 Male 65 0 0 0 0 0 VNMC070 Male 68 0 0 0 0 0 VNMC073 Male 60 0 0 0 0 0 VNMC078 Male 63 0 0 0 0 0 VNMC083 Male 81 1 1 0 0 1 VNMC089 Male 61 1 1 0 0 1 Mean (SD) 69.33 (8.65) Table 2. Descriptive male participant characteristics.

Subject ID Gender, n=15 Age G R B A S VNFC012 Female 78 0 0 0 0 0 VNFC051 Female 70 0 1 0 0 0 VNFC060 Female 78 0 1 0 0 0 VNFC062 Female 63 0 0 0 0 1 VNFC066 Female 62 0 0 1 0 0 VNFC071 Female 62 0 0 0 0 0 VNFC072 Female 70 1 1 0 0 1 VNFC074 Female 62 0 0 0 0 0 VNFC075 Female 63 0 0 0 0 0 VNFC076 Female 62 1 1 0 0 1 VNFC079 Female 63 0 0 0 0 0 VNFC082 Female 67 0 0 0 0 1 VNFC084 Female 68 0 0 0 0 0 VNFC088 Female 64 0 0 0 0 0 VNFC090 Female 67 0 0 0 0 0 Mean (SD) 66.60 (5.45) Table 3. Descriptive female participant characteristics.

Equipment

Each recording took place in a sound treated booth (Industrial Acoustics Company,

Inc., Model # 402) and was obtained using a headset mounted microphone

(omnidirectional Shure Beta 53) ensuring a constant 3 cm microphone-to-mouth distance.

The microphone signal was digitized directly into a Dell Inspiron laptop using the

TASCAM US-122L 16 bit 48 kHz digitizer with a 20 Hz to 20 kHz +/- 1dB frequency response. Calibration of the microphone to digitizer signal was completed each recording session using a SPL meter set for C-weighting. The digitized signal was immediately analyzed by the Voice Evaluation Suite (VES) (“Voice Evaluation,” 2006). VES is a computer program developed by Vocal Innovations as a clinical tool to automate collection, analysis, storage, and retrieval of significant voice measures.

Measures

Absolute jitter and shimmer measures have been demonstrated to be affected by mean F0 and SPL whereby a decrease in F0 results in an increased absolute jitter ratio

(Orlikoff & Baken, 1990). Since significant average F0 differences are expected between genders, percent jitter and percent shimmer are used to allow for comparison.

The VES program provides measures of jitter and shimmer as a percent of change relative to the average period. VES examines each waveform to determine the period. To express jitter as a percent of change of period relative to the average period, as was used in this study, the period of each vibratory cycle is measured and subtracted from the previous or succeeding period. The differences are averaged and then divided by the average period, then multiplied by 100 to produce a percentage. Jitter is then the average absolute value of change in period from one cycle to the next divided by the average period, expressed as a percentage. The mathematical formula used is:

Where N is the number of cycles across the entire sample and P(i) is the period of the ith cycle.

The VES program also examines each cycle of the waveform to determine peak- to-peak amplitude. Shimmer is then the average absolute value of change in peak amplitude from one cycle to the next divided by the average peak amplitude, expressed as a percentage.

The mathematical formula used is:

Where N is the number of cycles across the entire sample and A(i) is the peak to peak amplitude of the ith cycle.

Harmonics to noise ratio (HNR) in the VES program is defined as the average harmonic signal amplitude divided by the average noise signal amplitude over the frequency range of the first ten harmonics of F0 expressed in dB. The VES program excludes higher frequency harmonics as they are likely to vary much more widely as a function of SPL of the voice. The system uses a comb filter, as defined by F0 and its multiple harmonics, to separate the harmonic signal and the noise signal. HNR is computed every 10msec across the entire audio sample. The HNR value reported is then the average value of this HNR measurement across the length of the audio sample. The mathematical formula used is:

Where M is the number of 10msec time increments across the audio sample, H(i) is the average harmonic sample amplitude, and N(i) is the average noise signal amplitude at the ith time increment.

Procedures

Each participant was seated in a sound-treated booth (Industrial Acoustics

Company, Inc., Model # 402) and wore a headset mounted microphone (omnidirectional

Shure Beta 53) ensuring a 3cm microphone to mouth distance. Participants were instructed to sustain the vowel /a/ for four seconds at a “comfortable pitch and loudness level.” Three trials of this task were completed with the average of acoustic measures for the three samples then calculated by the VES program and utilized for evaluation.

Participants were allowed to repeat the task to a maximum of three recorded trials.

Multiple trials were completed when the participant failed to sustain /a/ for the required four seconds or if there was an uncharacteristic break in voicing during the recording, as identified by the participant. Recorded tasks were then analyzed by VES. Standard measures were derived from the three trial segments averaged together and included: a.) average F0; b.) average SPL (dB); c.) average jitter(%); d.) average shimmer(%); and e.)

HNR. See Appendix D for subject raw data from all measures.

Additional tasks completed by each participant as part of the data collection process, but not evaluated for this investigation, included: maximum phonation time, s/z ratio, dB

SPL range, F0 range in semitones, and laryngeal diadochokinetic rate.

Reliability

The VES program has not been used in previously published studies, so a test group of five non-study volunteers was used to determine test-retest reliability. Two immediately consecutive protocol trials were completed to reduce the possibility of confounding factors including task practice, vocal fatigue, difference in hydration levels,

17 medication effects, etc. The intra-subject reliability was high with correlations of r >

0.980 across all tasks. See Appendix E for all test-retest data demonstrating the reliability of the VES program.

To obtain intra-judge reliability for voice ratings of participants, the investigator analyzed all participant recordings of the “Cookie Theft” picture description task a second time, in random order. Intra-judge reliability was high with % agreement at

94.48% (Table 4).

First rating Second rating Difference Subject ID G R B A S G R B A S FC051 0 1 0 0 0 0 1 0 0 0 - FC060 0 1 0 0 0 0 1 0 0 0 - FC062 0 0 0 0 1 0 0 1 0 1 1 (B) FC066 0 0 1 0 0 0 0 1 0 0 - FC071 0 0 0 0 0 0 0 0 0 0 - FC072 1 1 0 0 1 1 1 0 0 1 - FC074 0 0 0 0 0 0 0 0 0 0 - FC075 0 0 0 0 0 0 0 0 0 0 - FC076 1 1 0 0 1 0 1 0 0 1 1 (G) FC079 0 0 0 0 0 0 0 0 0 0 - FC082 0 0 0 0 1 0 1 0 0 0 2 (R, S) FC084 0 0 0 0 0 0 0 1 0 0 1 (B) FC088 0 0 0 0 0 0 0 0 0 0 - FC090 0 0 0 0 0 0 0 0 0 0 - MC035 0 0 0 0 0 0 0 0 0 0 - MC052 0 0 0 0 0 0 0 0 0 0 - MC053 0 0 1 0 0 0 0 1 0 0 - MC054 0 0 0 0 0 0 0 0 0 0 - MC055 0 0 1 0 0 0 0 1 0 0 - MC056 0 1 0 0 0 0 1 0 0 0 - MC058 0 1 0 0 0 0 1 1 0 0 1 (B) MC061 0 0 1 0 0 0 0 1 0 0 - MC063 1 1 1 0 0 1 1 1 0 0 - MC064 0 0 0 0 0 0 0 1 0 0 1 (B) MC070 0 0 0 0 0 0 0 0 0 0 - MC073 0 0 0 0 0 0 0 0 0 0 - MC078 0 0 0 0 0 0 0 0 0 0 - MC083 1 1 0 0 1 1 1 0 0 1 - MC089 1 1 0 0 1 0 1 0 0 1 1 (G) % Agreement 94.48 Table 4. Intra-judge ratings comparison for GRBAS scale. Note. G=Grade, R=Roughness, B=Breathiness, A=Asthenia, S=Strain

To determine inter-judge reliability a second judge, who is a speech-language pathologist specializing in voice disorders, rated a random sample of 30% of participants‟ recordings using the GRBAS scale after listening to the recorded continuous speech sample as provided via the “Cookie Theft” picture description task. Overall inter-judge reliability for GRBAS ratings was high with % agreement at 94% (see Table 5).

First judge Second judge Difference Subject ID G R B A S G R B A S FC062 0 1 0 0 1 0 0 0 0 1 1 (R) FC074 0 0 0 0 0 0 0 0 0 0 - FC075 0 0 0 0 0 0 0 0 0 0 - FC084 0 0 0 0 0 0 0 0 0 1 1 (S) FC090 0 0 0 0 0 0 0 0 0 0 - MC035 0 0 0 0 0 0 0 0 0 0 - MC053 0 0 1 0 0 0 0 1 0 0 - MC054 0 0 1 0 0 0 0 0 0 0 1 (B) MC064 0 0 0 0 0 0 0 0 0 0 - MC070 0 0 0 0 0 0 0 0 0 0 - % Agreement 94.48 Table 5. Inter-judge ratings comparison for GRBAS scale. Note. G=Grade, R=Roughness, B=Breathiness, A=Asthenia, S=Strain

Chapter 3: Results

The participant data were analyzed for differences in gender groups on the dependent variables of jitter, shimmer and HNR. Individual means were calculated immediately across the three trials of /a/ and only the individual average was saved for evaluation.

Those individual averages were then combined for each gender and averaged to determine the gender group mean. The group means (standard deviations) for males and females for jitter were 6.71 (3.09) and 6.01 (3.50), respectively. Shimmer means (SD) for males and females were 3.46 (2.41) and 4.29 (2.17). Group means (SD) for HNR were

12.24 (4.02) and 10.78 (3.50), respectively. Gender group means for each measure were then compared using t-tests with significance level set at p < 0.05, where the t value is the probability of type one error. The results for the t-tests between genders were jitter (%) p= 0.566, shimmer (%) p= 0.332, and HNR p= 0.298 (see Table 6). The results across all three measures revealed no statistically significant differences between genders in the aged population when measuring perturbation and HNR.

Variable Male M (SD), n=15 Female M (SD), n=15 Age 69.33 (8.65) 66.60 (5.45) Jitter % 6.71 (3.09) 6.01 (3.50) p=0.566 Shimmer % 3.46 (2.41) 4.29 (2.17) p=0.332 HNR (dB) 12.24 (4.02) 10.78 (3.50) p=0.298 Table 6. Group means (SD) and t-test results for Jitter (%), Shimmer(%), and harmonics to noise ratio (HNR) (dB)

While inclusion criteria allowed participants with an overall GRBAS scale G rating of

1 to participate in the study, post-hoc evaluation excluding those participants was done to determine if their exclusion affected any trends in the data. Evaluating only the data where G = 0 left a sample of 12 males (ages 60-85, mean age 70 years old) and 13 females (ages 62-78, mean age 67 years old). The revised group means (standard deviations) for males and females for jitter were 6.78 (3.42) and 6.67 (3.27), respectively.

Shimmer means (SD) for males and females were 3.41 (2.69) and 4.47 (2.25), respectively. Group means for HNR were 13.28 (3.67) and 10.59 (3.65), respectively.

Gender group means for each measure were then compared using t-tests with significance level set at p < 0.05. The results were jitter (%) p = 0.933, shimmer (%) p = 0.294, and

HNR p = 0.079 (see Table 7) which trended towards, but still failed to reach, significance between genders.

Variable Male M (SD), n=12 Female M (SD), n=13 Age 69.58 (8.53) 66.69 (5.64) Jitter % 6.78 (3.42) 6.67 (3.27) p=0.933 Shimmer % 3.41 (2.69) 4.47 (2.25) p=0.299 HNR (dB) 13.28 (3.67) 10.59 (3.65) p=0.079 Table 7. Revised group means (SD) and t-test results using only subjects where G = 0.

Data were then parsed further by age to examine the possible effects of increased age variability. The revised decade group age 60-69 included 9 males, 63.11 (2.98), and 11 females, 63.91 (2.30). The revised decade group age 70-79 included 3 males, 75.33

(3.51), and 4 females, 74.00 (4.62). Increased variability within both genders was

22 demonstrated through larger standard deviation for each measured variable when comparing subjects aged 60-69 to subjects aged 70-79. As there were only male participants over age 80 a third group by decade was not included. Gender differences in perturbation were inconstant when comparing groups by decade. While jitter trended toward significant gender differences in the 70-79 age group it did not reveal much change in the 60-69 age group when compared to the data in entirety. Shimmer trended toward significance in the 60-69 age group (p=0.077), but away from significant in the

70-79 age group. HNR trended away from significant gender differences in both age groups when compared to the full data set (see Tables 8 and 9).

Gender Male, n=9 Female, n=11 Age (SD) 63.11 (2.98) 63.91 (2.30) %Jitter (SD) 5.73 (2.74) 6.65 (3.57) p=0.526 %Shimmer (SD) 2.82 (2.82) 4.17 (1.98) p=0.077 HNR (dB) (SD) 12.44 (4.06) 10.94 (2.83) p=0.362 Table 8. Group means (SD) for subjects aged 60-69 years.

Gender Male, n=3 Female, n=4 Age (SD) 75.33 (3.51) 74.00 (4.62) Jitter % (SD) 7.97 (3.10) 4.25 (3.02) p=0.181 Shimmer % (SD) 6.37 (4.29) 4.60 (2.97) p=0.579 HNR (dB) (SD) 12.07 (6.26) 10.35 (5.48) p=0.724 Table 9. Group means (SD) for subjects aged 70-79 years.

The VES formula used to determine HNR results in a narrower spectrum for lower mean F0 because it only includes the first ten harmonics of F0. Harmonics are whole number multiples of the F0 (Borden, Harris, & Raphael, 2003); therefore, the harmonic range evaluated in determining the HNR will be narrower for men than for women as the average F0 for males and females in this data set were 122.71 Hz and 214.11 Hz, respectively. Logical deduction would result in the expectation that a wider spectrum of harmonics will produce a lower HNR value in women. In fact, the data above do reflect a reduced average HNR with higher F0 in the female subjects.

Chapter 4: Discussion

Results of this study demonstrated no significant differences between gender for any of the dependent variables of perturbation or HNR. Possible reasons for similarity between genders may be related to the changes in hormone balances in both genders, with females producing less estrogen and males producing less testosterone (Abitbol,

Abitbol, & Abitbol, 1999; Gugatschka, Kiesler, Obermayer-Pietsch, Schoekler, Schmid,

Groselj-Strele, & Friedrich, 2010). The decline of laryngeal function in both genders may also contribute to the acoustic similarity in aged voices outlined in this data. Hormonal changes and physiologic decline may contribute not only to changes in F0 in both genders, but also to the less efficient functioning of the laryngeal system with a resultant decrease in HNR values (Ferrand, 2002). Reduced efficiency and increased instability in phonation is widely accepted as a result of aging across both genders. When compared to other studies (Ferrand, 2002; Stathopoulos, Huber, & Sussman, 2011) evaluating the same parameters in younger voices, the results discussed here support previous findings of reduced stability in aged voices versus young voices across genders, as well as the increased variability in the aged voice. Reduced stability is primarily indicated by lower measurement results of HNR in the aged voice; or increased measurement results of the inverse, noise to harmonics ratio (Ferrand, 2002; Xue & Deliyski, 2001; see Table 1).

This study‟s full set of data included subjects with GRBAS scale ratings where G=1.

When participants perceptually graded G=1 were removed from the data set, jitter

25 trended even further away from significance and became increasingly similar between genders. Conversely, shimmer and HNR both trended toward significance in gender differences with the comparison of HNR nearly reaching significance (p= 0.079). This finding would provide additional support for Stathopoulos, Huber, and Sussman (2011) who determined signal to noise ratio was lower in older females than in older males beginning after age 50 (see Table 7).

When subjects were parsed into groups by age, the data from the present study also support previous findings that variability across acoustic measures increases with age.

Increased variability was demonstrated by larger standard deviations for each variable when comparing subjects aged 60-69 to subjects aged 70-79 (see Tables 8 and 9) with the exception of females age 70-79 for jitter. The increased variability within chronological age groups may also be related to variable physiologic aging. Ramig and Ringel (1983) discussed differences found within age groups of males when they were evaluated based on physical condition in young, middle aged, and old subjects. Replication of the present study may be warranted to determine if significant differences in acoustic voice measures are present when the combination of gender and physiological aging are controlled. It will likely benefit future research to include both chronological and physiological age for the purpose of establishing normative values.

Results of the present study are inconsistent with previously published jitter and shimmer measurements (see Table 1). Several factors may confound valid measurements of perturbation including SPL, mean F0, and analysis system used (Linville, 2001) which may be related to these differences, as well as the wide variation of reported values for these measures. First, data collected in previous works often used only one recorded

26 vowel to extract the desired measures (Awan, 2006; Ramig & Ringel 1983; Sorensen &

Horii, 1980; Xue & Deliyski, 2001). Sorensen and Horii (1980) and Ramig and Ringel

(1983) also both chose the single vowel production that demonstrated the least amount of jitter to include for statistical analysis. This artificial sample selection may contribute to the lower values of jitter and shimmer reported by other authors. Higher jitter and shimmer values reported in the present study may also be related to the increased intra- subject variability introduced when analyzing a sample that is averaged across three trials. Future research should address this difference in sampling methodology to determine which provides increased accuracy for representation of normative values.

Results of the present study for measures of jitter differ significantly from previously published work with means of 6.71 (3.09) and 6.01 (3.50) for males and females, respectively. With the exception of Linville (1987) and Xue and Deliyski (2001), all other authors presented in Table 1 reported jitter values <1. However, Linville (1987) reported jitter ratio values, which uses a multiplier of 1000 rather than 100 as used for

%jitter. When the results presented in Linville (1987) are converted to %jitter, then the data reflect values <1. Xue and Deliyski (2001) reported %jitter values >2 in their elderly population, which is elevated in relation to other authors outlined in Table 1 but still significantly less than the data reflected in the present study. Rationale for these differences in jitter may be related to the increased variability seen in the voice with increasing age, differences in sample, and data collection methodology as addressed above.

Shimmer results from the present study are similar to those reported by Xue and

Deliyski (2001) and Awan (2006), but significantly elevated from all other authors listed

27 in Table 1. However, both investigators with similar shimmer values reported higher

%shimmer and greater standard deviations than was described in the present study. It should be noted that neither of the investigations address the significant difference of their findings for shimmer values against others. While the differences for shimmer values are also likely related to the greater variability with vocal aging and sample differences, future investigation should look to reconcile such discrepancy in order to provide normative values and evaluate possible mechanisms of change.

The values reported in Table 6 for measures of HNR fall between those reported by

Ferrand (2002) and Awan (2006). Similar to Awan (2006), when the data were separated by decade very slight reductions in HNR values with increased age were observed (see

Tables 8 and 9). However, the change between decades in both data sets is not statistically significant. It may be possible that the increased jitter values as reported in the present study contributed to the lower HNR values as compared to Awan (2006).

Possible differences in the values reported by Ferrand (2002) may be related to recording conditions, as recordings for the two older groups were not completed in a sound treated booth. This possible lack of control for ambient noise may have contributed to lowered

HNR values.

An additional factor likely contributing to variation in the present study‟s results was not controlling for SPL. Gender effects on perturbation in this study are inconsistent with previous work by Brockmann, Storck, Carding, and Drinnan (2008) that suggested significant gender effects on perturbation. However, they demonstrated those effects via tasks controlled for vocal loudness. Brockmann, Storck, Carding, and Drinnan (2011) found that women had higher %jitter values in some vowels, but no difference in

28 shimmer measurement between genders after correcting for SPL and F0. Their study was completed using young adults and therefore the results cannot be directly applied to the population studied here. However, their data may serve as a basis for adult measurement outlining that the acoustic voice measure of shimmer remains similar between genders with age as this study has shown. The microphone to mouth distance difference in the present study (3cm) compared to most others presented in Table 1 (10cm) may have also had an effect on the SPL of the sample collected even when SPL is controlled for within the task. The combination of difference in microphone to mouth distance and lack of control for SPL make direct comparison of the present study to previous research problematic. As demonstrated by Brockmann, Drinnan, Storck, and Carding (2011) and

Orlikoff and Kahane (1991), SPL is currently thought to have the greatest effect on jitter and shimmer measurements. Replication of the present study measures to include SPL controls and adjusted microphone to mouth distance would allow for better direct comparison to other authors and to reveal possible gender effects.

Further limitations of this study include absence of the very aged population. The study inclusion criteria of general good health may have contributed to only enrolling persons with aged voices and not very aged voices. Additional limitations include a relatively small sample size, given the known increased variability in the aged population as discussed previously.

Values reported across studies of acoustic voice measurements vary considerably, hindering reliable comparisons. The results contributed by this study further support that research in the area of voice acoustics may benefit from an accepted standardized protocol for data collection. Standardization may allow for greater comparison across

29 studies, strengthening future establishment of normative data. However, we must recognize as clinicians that our perspective of any variable is limited by our methodology. Greater development of normative data across the life-span, especially in the aged voice, will only aid clinicians in distinguishing processes of normal aging from a disorder of aging in acoustic measurements of phonation. Further research evaluating the aging progression is warranted to determine repeatability and clinical significance in voice evaluation.

References

Abitbol J, Abitbol P, Abitbol B. (1999). Sex hormones and the female voice. Journal of Voice, 13(3), 424-446.

Aronson, A. E., & Bless, D. M. (2009). Clinical voice disorders (4th ed.). New York: Thieme.

Awan, S. N. (2006). The aging female voice: Acoustic and respiratory data. Clinical Linguistics & Phonetics, 20(2), 171-180.

Baken, R. J. (2005). The aged voice: a new hypothesis. Journal of Voice, 19(3), 317-325.

Baken, R. J., & Orlikoff, R. F. (2000). Clinical measurement of speech and voice. San Diego: Singular Thomson Learning.

Binstock, R. H., George, L. K., Cutler, S. J., Hendricks, J., & Schulz, J. H. (2011). Handbook of aging and the social sciences (7th ed.). Amsterdam; Boston: Elsevier/Academic Press.

Bohnenkamp, T. A., & Hageman, C. F (2011). Voice and motor speech abilities. In A.Burda (Ed.), Communication and swallowing changes in healthy adults (109-126). Sudbury, MA: Jones & Bartlett Learning.

Boone, D. (1997). The three ages of voice: the singing/acting voice in the mature adult. Journal of Voice, 11(2), 161-164.

Borden, G., Harris, K., & Raphael, L. (2003). Speech science primer: physiology, acoustics, and perception of speech (4th ed.). Baltimore: Lippincott Williams & Wilkins.

Brock, D. (1990). Characteristics of the elderly population in the United States. In E. Cherow (Ed.), Proceedings of the Research Symposium of Communication Sciences and Disorders and Aging (ASHA Rep. No. 19). Rockville, MD: American Speech- Language-Hearing Association.

Brockmann, M., Drinnan, M.J., Storck, C., & Carding, P.N. (2011). Reliable jitter and shimmer measurements in voice clinics: the relevance of vowel, gender, vocal intensity, and fundamental frequency effects in a typical clinical task. Journal of Voice, 25(1), 44-53.

Brockmann, M., Storck, C., Carding, P.N., & Drinnan, M.J. (2008). Voice loudness and gender effects on jitter and shimmer in healthy adults. Journal of Speech, Language, and Hearing Research, 51(5), 1152-60.

Colton, R., Casper, J., & Leonard, R. (2011). Understanding voice problems: a physiological perspective for diagnosis and treatment (4th ed.). Baltimore, MD: Lippincott Williams & Wilkins

Ferrand, C. (2002). Harmonics-to-noise ratio: an index of vocal aging. Journal of Voice, 16(4), 480-487.

Goodglass, H., Kaplan, E., & Barresi, B. (2001). Boston Diagnostic Aphasia Examination, 3rd ed. Austin, TX: Pro-Ed.

Gorham-Rowan, M., Laures-Gore, J. (2006). Acoustic-perceptual correlates of voice quality in elderly men and women. Journal of Communication Disorders, 39(3), 171-184.

Gugatschka, M., Kiesler, K., Obermayer-Pietsch, B., Schoekler, B., Schmid, C., Groselj- Strele, A., , & Friedrich, G. (2010). Sex Hormones and the Elderly Male Voice. Journal of Voice, 24(3), 369-373. doi:10.1016/j.jvoice.2008.07.004

Harnsberger, J. D., Shrivastav, R., Brown, W. S. Jr., Rothman, H., & Hollien, H. (2008). Speaking rate and fundamental frequency as speech cues to perceived age. Journal of Voice, 22(1), 58-69.

Hirano, M. (1981). Clinical examination of voice (Vol. 5). New York: Springer-Verlag.

Horii Y. (1980). Vocal shimmer in sustained phonation. Journal of Speech and Hearing Research, 23(1), 202-9.

Kahane, J. (1990). Age-related changes in the peripheral speech mechanism: structural and physiological changes. In E. Cherow (Ed.), Proceedings of the Research Symposium of Communication Sciences and Disorders and Aging (ASHA Rep. No. 19). Rockville, MD: American Speech-Language-Hearing Association.

Kahane, J. (1987). Connective tissue changes in the larynx and their effects on voice. Journal of Voice, 1(1), 27-30.

Kent, R. D., & Ball, M. J. (2000). Voice quality measurement. San Diego: Singular Pub. Group.

Linville, S. E. (2001). Vocal aging. San Diego: Singular Thomson Learning.

Linville, S. E. (1987). Acoustic-perceptual studies of aging voice in women. Journal of Voice, 1(1), 44-48.

Linville, S. E., & Fisher, H. B. (1985). Acoustic characteristics of women‟s voices with advancing age. Journal of Gerontology, 40(3), 324-330.

Orlikoff, R., Kahane, J. (1991). Influence of mean sound pressure level on jitter and shimmer measures. Journal of Voice, 5(2), 113-119.

Ramig, L. A., & Ringel, R. L. (1983). Effects of physiological aging on selected acoustic characteristics of the voice. Journal of Speech and Hearing Research, 26(1), 22-30.

Sorensen, D., & Horii, Y. (1983). Frequency and amplitude perturbation in the voices of female speakers. Journal of Communication Disorders, 16(1), 57-61.

Stathopoulos, E., Huber, J., & Sussman, J. (2011). Changes in acoustic characteristics of the voice across the life span: measures from individuals 4-93 years of age. Journal of Speech, Language, and Hearing Research, 54(3), 1011-1021.

Sussman, J., & Sapienza, C. (1994). Articulatory, developmental, and gender effects on measures of fundamental frequency and jitter. Journal of Voice, 8(2), 145-156.

Titze, I. R. (1989). Physiologic and acoustic differences between male and female voices. Journal of the Acoustical Society of America, 85(3), 1699-1707.

Voice Evaluation Suite (Version 4.1) [Computer Program] (2006). Pittsburg, PA: Vocal Innovations.

Wilcox, K., & Horii, Y. (1980). Age and changes in vocal jitter. Journal of Gerontology, 35(2), 194-198.

Xue, S., Deliyski, D. (2001). Effects of aging on selected acoustic voice parameters: preliminary normative data and educational implications. Educational Gerontology, 27(2), 159-168.

Appendix A: Informed Consent

Appendix B: Participant Questionnaire

Voice Norms Study

Voice Health Questionnaire

1. Do you currently use any medications shown to have a negative effect on voice?

______

2. Have you had any previous voice surgery? ______

3. Have you ever sought treatment for a voice problem or voice disorder? ______

4. Do you have a history of severe allergies, asthma, vocal fold pathology, or neuromotor impairment that may impact the voice? ______

5. Have you been a nonsmoker for the past five years? ______

6. Are you a professional vocal soloist? ______

7. Is English your primary language for communication? ______

Appendix C: Screening Protocol Form

Screening Protocol Form

Subject ID: ______Date of Birth: ______Gender: ______

Date of Screening: ______Examiner: ______

ID Key:

 VN, Gender, Age Group, enrollment # (Ex: VN MB 22)  Gender (M = Male, F = Female)  Age Group (A = 18-39, B = 40-59, C = 60 & up)

Consent Form Signed

GRBAS Perceptual Scale: Subjects description of the Boston Diagnostic Aphasia Examination picture of the Cookie Theft. Rating scores over one are excluded. Ratings: 0 (normal) – 3 (severe)

Grade: ______

Roughness: ______

Breathiness: ______

Asthenia: ______

Strain: ______

Pure Tone Hearing Screening: (must pass at 20dB until age 59, 30dB for over age 60 with correction). Hertz Right Ear Left Ear 500 1000 2000

Appendix D: Study Participant Raw Data

Avg Avg Subject ID DOB Age Jitter (%) Shimmer (%) HNR (dB) F0 (Hz) SPL (dB) VNFC012 8/28/31 78 196 4 94.8 3.2 10.6 VNMC035 11/17/49 61 137.9 0.9 81.2 1.2 14 VNFC051 6/27/40 70 245.4 8.6 93.8 8.9 3.7 VNMC052 12/12/31 80 117.3 7.8 96 1.7 13.1 VNMC053 4/18/26 85 113.1 12.7 84.7 2.5 13.7 VNMC054 11/29/35 75 154 8.6 79.3 7 10 VNMC055 4/28/44 67 136.6 5.5 74.8 3 14.9 VNMC056 6/7/51 60 121.6 1.4 81.3 1.2 17.9 VNMC058 11/23/39 72 104.3 10.7 88.2 1.8 19.1 VNFC060 7/15/33 78 248.3 2.3 84.8 2.2 17.1 VNMC061 4/22/32 79 149.5 4.6 74.4 10.3 7.1 VNFC062 5/22/48 63 206.1 6.1 86.1 2.3 11.6 VNMC064 3/31/46 65 137.9 7 87.4 2 13.2 VNFC066 3/11/49 62 211.2 4.1 73.7 7.4 8.9 VNMC070 9/13/43 68 99.9 7 84 3 16.4 VNFC071 7/22/49 62 242.6 4.5 86.3 3.5 9.6 VNFC072 8/18/41 70 236.3 2.1 85.2 4.1 10 VNMC073 3/1/52 60 116.4 8.7 84.6 4.1 7.8 VNFC074 5/31/49 62 231.1 9.1 90.9 2.3 15.4 VNFC075 5/28/48 63 145.3 14.4 80.6 6.3 11.5 VNFC076 7/23/49 62 239.4 1.3 84.9 2.1 14 VNMC078 7/12/48 63 98.4 6.5 88.8 3.1 12.2 VNFC079 1/9/49 63 150.4 8.7 78.2 6.5 9 VNFC082 7/5/44 67 203.5 8.5 82.6 5.3 9.3 VNMC083 5/20/30 81 140.4 4.6 83 3.2 8.6 VNFC084 6/29/43 68 221.5 3.7 89.7 2.1 14.3 VNFC088 10/24/47 64 266 4.8 88 3.2 10.9 VNMC089 5/5/50 61 100.4 7.7 85.2 4.7 5.3 VNFC090 9/15/44 67 168.5 7.9 80.5 4.9 5.8 Table 10. Study participant raw data

Appendix E: Test-Retest Reliability Data

Subject.Recording# Avg. F0 Jitter % Avg. SPL Shimmer% HNR Tasks combined 1.1 230.2 10.3 84.4 2.8 8 1.2 217.3 8.2 84.6 4 12.6 Intra-subject reliability r=0.999 2.1 251.7 4.7 92 1.8 12.9 2.2 253.8 4 94.6 1.6 14.1 Intra-subject reliability r=0.999 3.1 219.8 11.1 85.7 3.2 10.5 3.2 206.3 4.7 93 1.6 11.6 Intra-subject reliability r=0.997 4.1 210.7 9.2 88.7 2 12.4 4.2 211.8 6.1 92.5 1.8 11.3 Intra-subject reliability r=0.999 5.1 215 6.4 88.2 4.1 13.5 5.2 209.4 6.5 86.3 4.2 16 Intra-subject reliability r=0.999 Overall group reliability r=0.929 r=0.402 r=0.615 r=0.666 r=0.507 r=0.999 Table 11. Test-retest data for sustained /a/ task averaged across three recorded trials.

Appendix E: Test-Retest Reliability Data

Subject. Avg. Min Max Avg. Min Max Combined Recording# F0 F0 F0 SPL SPL SPL tasks 1.1 190.8 142.8 284.2 82.4 76.3 89.7 1.2 197 146.3 305.8 85.8 75.9 94.8 Intra-subject reliability r=0.999 2.1 196.5 152.5 292.9 86.1 76.5 95.7 2.2 209.8 164.8 417.4 84.4 77.1 93.5 Intra-subject reliability r=0.986 3.1 191.9 132.2 286.3 84.3 75.6 92.1 3.2 181.1 135.6 264.4 86.9 78.2 92.9 Intra-subject reliability r=0.999 4.1 191.4 140.5 305.8 88.7 77.9 96.7 4.2 178.8 105.3 237.1 89.3 78.1 98.9 Intra-subject reliability r=0.980 5.1 193.4 128.1 345.3 88.3 78 95.9 5.2 182.9 132.5 287.7 88.2 79.8 96.6 Intra-subject reliability r=0.997 Overall group reliability r=0.654 r=0.509 r=-0.200 r=0.640 r=0.561 r=0.541 r=0.934 Table 12. Test-retest data for continuous speech sample.

Appendix E: Test-Retest Reliability Data

Subject. Recording# Phonation S/Z Min Max Time S Time Z Time Ratio SLP SPL 1.1 26.9 22.7 32.1 0.71 75 108.3 1.2 27.4 17.7 28 0.63 76.3 107.6 Intra-subject reliability 2.1 15.4 24.2 20.2 1.2 82.9 109.6 2.2 19.2 34.2 39.7 0.86 80.5 112.1 Intra-subject reliability 3.1 14.2 27.9 18.3 1.52 80 105.5 3.2 12.2 26.7 20.2 1.32 83.3 110.6 Intra-subject reliability 4.1 14.8 16.7 14 1.19 79.8 107 4.2 9 16.5 12.2 1.35 79.5 107.4 Intra-subject reliability 5.1 31.7 48.8 31.7 1.54 78.4 114.3 5.2 27.6 39.5 42.3 0.93 77.1 116.4 Intra-subject reliability Overall group reliability r=0.897 r=0.811 r=0.645 r=0.607 r=0.688 r=0.807 Table 13. Test-retest data for maximum performance tasks

Appendix E: Non Study Reliability Data

Table 13 continued

SPL Min Max F0 Laryngeal Combined Range F0 F0 Range DDK tasks 33.3 155.3 536.7 21 4.5 31.3 120.8 507 25 4.6 r=0.998 26.7 209.2 566.7 17 5.1 31.6 167.4 533.2 20 5.5 r=0.997 25.5 165.2 480.6 18 4 27.3 201.7 530.6 17 3.9 r=0.999 27.2 163.7 461.2 18 4.7 27.9 171.2 557.6 20 3.5 r=0.998 35.9 63.5 888.9 46 4.5 39.2 62.8 1007.4 48 4.3 r=0.999 r=0.843 r=0.823 r=0.954 r=0.989 r=0.614 r=0.993