Brestovci Behlul, Natalija Bolfan - Stosic

Acoustic Laboratory for Speech and Hearing
Department of Logopedics
Faculty of Special Education and Rehabilitation
University of Zagreb

Voice Quality of Hearing-Impaired Children*

Abstract

The aim of this research was to determine voice characteristics of hearing-impaired boys aged 10-12 years by Bruel and Kjaer, Real-time Frequency Analyser, Sound forge 4.0 and EZ Voice (PC programs for spectral analyses) in relation to control group of boys the same age without hearing disorders. In this research we have reported higher than normal Fo perturbations, higher Fo frequency and spectral noise levels between 1 kHz and 2 kHz than in control group for 12 hearing-impaired boys. The results showed shorter vowel and consonant production (“a”and “z”) characterised by over-aspiration in hearing-impaired group. Obtained results of T-test have shown significant statistical differences in seven applied variables among the tested groups.
The research gives spectrograms for each group in three graph types of presenting voice characteristics and audiograms: as a “bar graph type”, oscillogram form and as a jitter and shimmer curve form. Each of voice-graph types is a good for identification and recognition specific voice characteristics as well as a noise levels on “bar graph type”.

Keywords: Fo frequency, jitter, noise level, over-aspiration

1. Introduction

According to Rosenhouse (1988) one of the major problems of deaf speech is intonation, which is often described as 'monotones' or 'too jumpy' or 'not in pitch,' etc. People with hearing losses have inadequate fundamental frequencies (Fo) characterized as monotone (Nickerson, 1975; Youdelman, MacEachron, McGarr, 1989) and higher than the average Fo of people with normal hearing (Higgins, Carney, Schulte, 1994). Hearing-impaired people may also have unusual voice quality, characterized by over-aspiration, spectral noise and so on. The over-aspiration, which may increase the tactile feedback for the speaker, gives a breathy quality to the speech and changes the temporal pattern (Bench, 1992). Radovančić (1995) describes non-balance between optimal biofeedback and receiving and production of speech as a reason for non-standard acoustical voice features of this population. According to the same author the basic types of hearing disorders is conductive impairments (placed in the middle and outside of ear) and sensorial (placed in inside structures of ear and cortical centers). So, the Fo could be standard, tense, loose, low, and unnatural with following noise. The purpose of the paper is in finding, establishing and describing acoustical features of hearing impaired children in research which is not often in our country.

2. Methods

2.1 Selection of variables and instrumentary

In order the following variables were selected to obtain an acoustical evaluation of the parameters: FOHZ and FODB - fundamental frequency in Hz and dB, HD1 – HD7 – intensity of first seven harmonics, NOISE1-NOISE7 - noise level between first seven harmonics, NUM - number of columns around Fo as an indicator of disordered voice, jitter – frequency fluctuations, FONAMAX - sustained vowel “a” productions, FRICS and FRICZ - sustained consonants “s” and “z” productions.
Measures of these voice parameters in 24 school children’s voices were obtained using an acoustic analysis by Bruel and Kjaer, Real-time Frequency Analyzer, Type 2123, Sound forge 4.0, and EZ Voice (Voice Analysis Software, v 1.2). The subject’s voices were immediately recorded on Real-time Frequency Analyzer and PC. The microphone was placed 30 cm from the subject’s lips. We obtained 24 voice pictures of both groups divided to type of graphic presentation in oscillograms (Sound forge 4.0),“ bar graph types” (Real-time Frequency Analyzer, 2123), and jitter and shimmer curve form (EZ Voice). Audiograms of boys from special school for deaf in Zagreb are enclosed, too. They show hearing losses on both ears: on left ear at 60-105 dB intensity range, on right ear at 30-105 dB intensity range.

2.2 Tasks

There was three phonatory tasks: 1) Sustained vowel production in which the subjects were asked to articulate vowel /a/ (as long as they can); The vowel was repeated three times with rest periods between vowels. 2) Sustained consonant production /s/ 3) Sustained consonant production /z/.

2.3 Data

The differences in variables between two groups were established by T-test. The data was processed on PC computer (Program STATISTICA for Windows, Release 4,5 A, (Statsoft, Inc.1993)).

3. Results and Discussion

Results from Table 1 show bold values of parameters – sustained “a”, frequency of F0, sustained consonant production /z/, noise intensity level between harmonics (N3 and N4), number of columns around Fo and jitter in percents which are significant differ groups.

Table 1. T-test between groups with hearing losses and group without hearing losses

Variable

mean
g 1

mean
g 2

t-value

df

p

FONAMAX

12.353

8.250

3.046

22

.0059

FOHZ

244.416

259.166

-2.142

22

.0434

FODB

70.025

70.800

- .316

22

.7547

NUM

1.750

11.083

-6.487

22

.0000

.221

1.045

-3.757

22

.0010

NOISE1

32.675

33.133

- .150

22

.8817

H1DB

62.691

64.083

- .536

22

.5969

NOISE2

31.408

37.216

-2.056

22

.0517

H2DB

65.550

63.516

.723

22

.4772

NOISE3

37.558

46.055

-2.561

22

.0178

H3DB

63.416

69.933

-1.857

22

.0766

NOISE4

36.733

46.216

-2.560

22

.0178

H4DB

57.883

57.258

.171

22

.8653

NOISE5

38.283

38.208

.027

22

.9783

H5DB

55.300

50.716

.929

22

.3629

NOISE6

38.933

39.783

- .304

22

.7636

H6DB

50.025

42.891

1.669

22

.1092

NOISE7

33.208

37.191

-1.412

22

.1718

H7DB

45.233

41.616

1.145

22

.2644

FRICS

9.210

9.333

- .084

22

.9331

FRICZ

10.909

5.500

3.334

22

.0030

Legend:
G1 – group without hearing losses
G2 – group with hearing losses 

Obtained results show bigger oscillation of Fo (NUM and JITTER variable) in group with hearing losses and statistical differences between groups in “noise” variables placed in the middle spectra. According to Bolfan (1996) number of columns around Fo above 40 dB on the spectrogram is an indicator of Fo perturbations or jitter. Number of columns (Num) is just one more of the several ways of Fo variability observation on the Real Time Frequency Analyzer, 2123. Jitter value is above 1 %, which indicates disordered voice quality. It is related to reduced audio control rather than the faulty laryngeal function. The results of investigation by Thomas-Kersting and Casteel (according to Bench, 1992) showed that the hearing-impaired children attained significantly higher spectral noise levels and produced more effort for vocalization than the hearing children, which generated “nosy” speech produce. Results of our research show significant differences between groups in maximal duration of “a” and “z” productions. Boone (1989) considers that sustained time of phonation and friction is the best way for direct view on interaction between laryngeal and respiratory function. The duration of vowel “a” is not so short, but is very variable, without needed variations of voice. The over-aspiration changes the temporal pattern especially in friction of “z” production. That could be the first reason of significant differences between groups. The second one “lay” in fact that hearing-impaired child could not make smooth visual differences between similar places of articulations like in “s” and “z” consonants. In the most causes the friction of “z” has became “s” production. Bench (1992) describes the area of first and second formant (these two formants are equal to spectral field of harmonics 3 to harmonics 5 where we found significant differences in noise variables between groups) as a responsible for the recognition of vowels and voiced consonants. The children with high-frequency hearing losses have problems in perceiving the voiceless consonants characterised by place of articulation. The consonants have less acoustic energy than the vowels in the higher part of spectra for this population. Finally, we found higher Fo frequencies for the group with hearing losses. Measurements of voice fundamental frequency in hearing-impaired people have presented mixed results. Anyway, most of the authors agree with the higher Fo value for this population. Horii (1982) reported higher than normal Fo values for 12 hearing-impaired girls aged 16-19 years. All voice characteristics can be recognized on all types of graph voice pictures (see appendix1, 2 and 3). **

Table 2. S/Z ratio between groups with (G1) and without hearing losses (G2)

S/Z ratio sec.

Boone (1997) proposed the use of sustained production ratio based on the sounds "s" and "z" for respiratory and laryngeal factors contributing to a phonation problem. Kent (1987) reported about normative data on the voiceless/voiced ratio by Tait, Michael, and Capenter from 1980. They reported in their study that is not differences in the parameter between boys ahd girls aged 5-9 years or in age they extracted middle value of 0.85 in sec. for normal voices. Above this value could be pathological. The results of our research from Table 2. show that S/Z values from the group with normal voices are in the most cases less than 1.00 sec. In the other group these values are above 1.00 sec. or child couldn't produce sound "z".

Table 3. Discriminant analyses - Factor Structure Matrix

p-level = .0000

All variables from the Table 3. create obtained discriminant function by values above 0.20 (+ or-) but the "strongest" variable is JITTER, thab FRICZ and FONAMAX. So, we can conclude that the temporal values and oscillaitions of vocal cords vibrations are key points of observing disordered voices of children with hearing losses.

4. Conclusion

The results of acoustical analyses of 12 school children voices showed differences between groups in sustained “a”, frequency of F0, sustained consonant production /z/, noise intensity level between harmonics (N3 and N4), number of columns around Fo and jitter in percents. Children with hearing losses have problems with inadequate Fo and intonation. The last one could be monotone or “not in pitch” like voices presented by our group of children with hearing losses. These children can produce fonations of vowel longer than children with dysphonia and shorter than normal hearing children. The main problem is in controlling during voicing from auditive, respiratory and laryngeal level. In this research in children with hearing losses voice showed less variations, spectral noise, over-aspiration and higher value of Fo. Distorted articulation of consonants “z” is defined by problems in perceiving the voiceless consonants, insufficient visual compensatory mechanisms to recognise different articulatory movements in “s” and “z” productions and by slow movements. Monsen (1978) reported about this slow artuculatory movements in hearing-impaired children. We can describe these movements in hearing-impaired children like “clumsy”.
Despite all that problems, deaf children can improve their speech and voice quality by voice and speech training of course. The accent in voice therapy of children with hearing losses is on which aspect we must focus our attention first: intonation, respiration…..?

5. References

[1] Bench, R. J. (1992): Communication skills in hearing-impaired children. Whurr Publisher Ltd. Singular Publishing Group Inc. San Diego, California.

[2] Bolfan-Stošić, N. (1996): Some essential differences in the vocal characteristics of children with and without voice disorders, Croatian Review of Rehabilitation Research, 32, pp. 37-49.

[3] Boone, D., R. (1977): The voice and voice therapy, Prentice Hall, Englewood Cliffs

[4] Boone, D., R. (1989): The voice and voice therapy, Prentice Hall, Englewood Cliffs.

[5] Conway, D. (1992): Understanding and improving speech production. The Volta Review, 94, pp 241-242.

[6] Higgins, M. B., Carney, A. E., Schulte, L. (1994): Physiological assesment of speech and voice production of adults with hearing loss. Journal of Speech and Hearing Research, 37, pp. 510-521.

[7] Horii, Y. (1982): Some voice fundamental frequency characteristics of oral reading and spontaneous speech by hard of hearing young women. Journal of Speech and Hearing Research, 25, 608-610.

[8] Kent, R. D., Kent, J. F., Rosenbek, J. C. (1987): Maximum performance tests of speech production. Journal of Speech and Hearing Disorders, 52 22, pp. 270-288. 

[9] Monsen, R. B. (1978): Acoustic qualities of phonation in young hearing-impaired children. Journal of Speech and Hearing Research, 22, pp. 270-288.

[10] Nickerson, R. S. (1975): Characteristics of the speech of deaf persons. The Volta Review, 77, 6, 342-363.

[11] Radovančić, B (1995): Basics of rehabilitation of speech and hearing. Faculty of Defectology. Association of deaf people in Croatia.

[12] Rosenhouse, J. (1988): Computer-aided teaching of intonation to Hebrew-speaking hearing-impaired children. 4^th International Congress for the Study of Child. Language. Paper No. 2.

[13] Ryalls, J., Michallet, B., Dorze, G. (1994): A preliminary evaluation of the clinical effectiveness of vowel training for hearing-impaired children on IBM's Speech Viewer. The Volta Review, 96, pp. 19-30.

[14] Ryalls, J., Dorze, G., Boulanger, H., Laroche, B. (1995): Speech therapy for lowering vocal fundamental frequency in two adolescents with hearing impairments: A comparison with and without Speech Viewer. The Volta Review, 97, pp 243-250.

[15] Shukla, R. S. (1989): Phonological space in the speech of the hearing impaired. Journal of Communication Disorders, 22, 5, pp. 317-327.

[16] Wirz, S. L., Subtenly, L. D., Whitehead, R. L. (1981): Perceptual and spectrographic study of tense voice in normal hearing and deaf subjects. Folia phoniat., 33, pp. 23-36.

[17] Youdelman, K., MacEachron, M., McGarr, N. (1989): Using visual and tactile sensory aids to remediate monotone voice in hearing-impaired speakers. The Volta Review, May, 197-207.

[18] Zimmerman, G.., Rettaliata, P. (1981): Articulatory patterns of an adventitiously deaf speaker: Implications for the role of auditory information in speech production. Journal of Speech and Hearing Research, 24, pp. 169-178.

* In: Lehmann,T., Palm, C., Spitzer, K., Tolxdorff, T. (Eds.)  Advaces  in  Quantitative  Laryngoscopy, Voice  and  Speech  Research. Proceeddings  of  the   3rd  International  Workshop. Aachen  University  of   Technology, RWTH  Aachen, June  19-20, 1988,  pp. 35 - 44.

Appendix 1

Audiograms of group with hearing losses

Figure1. Fonation of vowel “a” (S.F.4.0)

Figure2. Fonation of vowel “a” (S.F. 4.0)

Figure 1a. Fonation of vowel “a” (S.F. 4.0)

Figure2a. Fonation of vowel “a” (EZ V.)

Figure 3. Fonation of vowel “a” (S.F. 4.0)

Figure3a. Fonation of vowel “a” (EZ V.)

Appendix 3

"Bar grsph type" of vowel "a" production (Real - Time Frequency Analyzer, type 2123) of the group with hearing losses