- Email: [email protected]
loudness dynamics in voices of normal children, with and without education in singing
International
ELSEVIER
Journal
of Pediatric
35 (1996)
Otorhinolaryngology
107-115
Fo-perturbation and Fe/loudness dynamics in voices of normal children, with and without education in singing P.H. Dejonckere, G.H. Wienekea, D. Bloemenkamp”, J. Lebacqb “The
Institute
of Phoniatrics, European Research Group on Larynx (GREL), Utrecht University, P.O. Box 85500, Nl-3508 GA Utrecht, The Netherlands bDepartment of Physiology, Catholic University of Louvain, B-1200 Brussels, Belgium Received
9 August
1995; accepted
1 October
1995
Abstract
Sustainedphonations were comparedin two groups of children (aged 7-12) one with specialartistic voice education and one from a normal school, without voice complaintsor problems.The hypothesisof specific(better) biomechanicalvocal fold propertiesin the first group is confronted with the hypothesisof differencessolely related to training of voice control. In both groups, Fo-aperiodicity was measuredin a sustainedphonation at 3 different SPL levels.As a generalrule, aperiodicity clearly decreases when the voice becomes louder. Aperiodicity is highly significantly lower, at all SPL-levels,in children with trained singingvoices: this impliesbetter mechanicalpropertiesof the vocal oscillator. The Fo/SPL relation on a sustained/a:/ doesnot differ in trained and untrained children’svoices: out of singing context, trained children do not spontaneouslycontrol the Fo/SPL dynamics differently from untrained children. The higher regularity of vocal fold pulsesis not related to the duration of training. Voice; Singing voice; Education; Fundamental frequency; Jitter; Aperiodicity; Laryngograph; Electroglottography; Loudness Keywords:
* Corresponding 0165-5876/96/$1.5.00 SSDI
author. 0 1996 Elsevier
0165-5876(95)01291-5
Science
Ireland
Ltd.
All rights
reserved
108
P.H.
Dejonckere
et
al. /
Int. J. Pediatr.
Otorhinolaryngol.
35 (1996)
107-115
1. Introduction 1.1. Fundamental frequency perturbation
A certain degree of irregularity is inherent in all of the speech production processes, regardless of how much practice there has been [l]. The term ‘jitter’ refers to short-term (cycle-to-cycle) variability in fundamental frequency (Fo), not accounted for by voluntary changes, as e.g. intonation. It is well known that, even in the voice signal of a (perceptually) perfectly steady sound, small fluctuations in frequency are present: they reflect the internal ‘noises’ of the human body. Sources of Fo micro-perturbation or aperiodicity are biomechanical (vocal folds are a damped oscillator), aerodynamic and neurological [2,3]. Measurements of jitter are primarily thought to assess the stability of vocal fold vibration [4]. There is by now a considerable body of literature that asserts the usefulness of measuring frequency perturbation in evaluating laryngeal and vocal pathology [5]. However, when only normal subjects are considered, is this irregularity phenomenon, as measured on a sustained /a:/, smaller for trained singers voices than for untrained voices? Have singers a laryngeal oscillator with better mechanical characteristics than normal people? And if so can this be considered a consequence of the singing education and training, or is it constitutional? These questions are not answered at present. In order to avoid gender differences as well as variability in singing style and practice, general education, profession etc., we choosed to compare two groups of children: on the one hand 47 children from the Utrecht Cathedral Choir School; on the other hand 54 children from a normal basic school in Utrecht. At the Utrecht Cathedral Choir School, young children, selected on the basis of their musical aptitudes, receive, besides a general education, a special training in (particularly vocal and sacred) music. 1.2. Neuromuscular
control of phonation
Education in singing implies - besides the artistic aspect - acquiring a better neuromuscular regulation of the voice production mechanism, especially a better control of pitch and loudness. Generally speaking, the amplitude of vibration of the vocal folds increases with lung pressure, which is, in normal voices, closely related to sound pressure level. This increased vibration amplitude determines a higher value of the average dynamic strain in the vocal fold, and we can expect, therefore, that the fundamental frequency will also increase with lung pressure [3]. This effect is obvious in elastic models of the larynx, as well as in tracheotomized patients with a bilateral laryngeal paralysis in adduction. In contrast, a singer practising a ‘messa di vote’ exercise is able to produce a long crescendo followed by a long diminuendo on a single note, thus at constant pitch. Between these extremes, when normal adult subjects are asked to produce, in a spontaneous way (thus not in singing voice), successive sustained emissions on /a:/ with increasing loudness, a concomitant slight increase in pitch is usually observed. The rate is about 1.4 Hz/dB (significantly less than in mechanical models or than in pathological cases) [6].
P.H.
Dejonckere
et al. / Int. J. Pediatr.
Otorhinolaryngol.
35 (1996)
107-115
109
It could be hypothesized that children with special training in singing - as investigated in this experiment - will show, when producing sustained emissions on /a:/ with increasing loudness, a better neuromuscular control of pitch, and thus less increase in Fo, than children without training. In the present study, microperturbations of Fo and Fo-dynamics as a function of sound pressure level are compared, in standardized voicing conditions, between the two groups of children, those with and those without singing education.
2. Material
and method
2.1. Subjects 101 normal children, aged 7 to 12, were investigated. Exclusion criteria were -abnormal sounding voice -acute, subacute or chronic inflammation/infection of airways -use of inhalation drugs -antecedents of any kind of laryngeal and pharyngeal pathology -chronic aspecific respiratory pathology -antecedents of intubation -incapacity of producing the required voicings for the experiment. After application of exclusion criteria, the following subjects remained: 38 vocally trained children from the Utrecht Cathedral Choir School (14 boys and 24 girls): mean age 10.3 + 1.3 years. 43 vocally untrained children from a normal basic school in Utrecht (17 boys and 26 girls): mean age 10.3 h 1.2 years. The distribution of ages is shown in Fig. 1. The age distributions of the two groups do not differ significantly. All parents where asked to complete a questionnaire concerning their child, and written informed consent was required. 2.2. Protocol In each child, nine recordings were made of a sustained /a:/ (2 to 4 s): three recordings at comfortable pitch and loudness (sound pressure level, SPL, controlled between 63.5 and 66.5 dB(A) at 30 cm), three recordings (2-4 s) at comfortable pitch and louder voicing (SPL controlled between 67 and 72 dB(A)), and three recordings (2-4 s) at comfortable pitch and soft voicing (SPL controlled between 58 and 63 dB(A)). 2.3. Instrumental
material
All recordings were made in a quiet but not a sound-proof room, using a Casio DA-7 DAT-recorder (frequency range: 10-20.000 Hz). The signal was analyzed
110
P.H.
Dejonckere
et al. i Int. J. Pediatr.
Otorhinolaryngol.
35 (1996)
107-115
afterwards using a PCLX PCpitch and Lx2 program (fundamental frequency and waveform analysis software for the Laryngograph: Laryngograph Ltd, London) [7]. The accuracy of the period measurement is normally to within 1 us, at worst 2 us. 2.4. Parameters
For each condition of loudness, a stable 500 ms segment was selected from the best suited /a:/ signal out of the three records. The following parameters were calculated: (1) mean fundamental frequency (Fo) (2) jitter factor The jitter was expressed as the ‘Jitter factor’, that is the mean difference between the frequencies of adjacent cycles divided by the mean frequency, multiplied by 100 151 $1;:
IP-c+,I] x 100
Jitter factor =
where Fi = frequency of the i th cycle, in Hz, and II = number of periods in the sample. The jitter factor is a Fo-related measure.
Age distribution
9
10
11
12
Years Fig. 1. Age distribution in the groups of children with and without training in singing.
P.H.
Dejonckere
et al. i ht.
J. Pediatr.
Otorhinolaryngol.
35 (1996)
III
107-115
Table I Overview of experimental data
Trained
choices
Soft voice
Normal voice
Loud voice
N =38
( * 13.0)
N = 38 10.3 228.9 Hz 3.0 5.2
(k6.1)
N = 38 10.3 278.8 Hz 1.3 0.6%
(IL 1.3) (k21.1) (z!c 0.5) (ix 0.9)
( * 1.2) (i26.1) ( k 5.6) (+- 15.2)
N =43 10.3 226.3 Hz 5.9 15.2%
(k (f (k (&
N = 44 10.3 276.5 Hz 2.4 3.3%
( * 1.2) (i 37.0) (IL 1.5) ( I!Y4.6)
Age Mean Fo Jitter factor Irregularity
10.3 (f 1.3) 206.8 Hz (k 71.8) 4.9 ( + 4.7)
Lintrained
N =43 10.3 223.2 Hz 7.1 I 9.2’%,
voices
Age Mean Fo Jitter factor Irregularity Trained coicrs Age
& untrained
Mean Fo Jitter factor Irregularity
12x41
1.2) 22.9)
5.2) 13.7)
N =82
N = 81
N =81
10.3 215.5 Hz 6.1 16.2%
(5 1.3) ( + 35.6) (i 1.5)
(x!I 1.2) (+ 53.0) (k 5.3) ( f 14.5)
10.3 227.5 Hz 4.5 10.5%
( f 1.2) (i 29.3) (k4.1) (k 11.8)
10.3 ( z!Y1.2) 277.6 Hz (& 30.5) 1.9 (k 1.3) 2.0% ( f 3.7)
(3) ‘irregularity’: this is another measure of the degree of aperiodicity: it is the percentage of fundamental frequency pairs (adjacent cycles) with a cycle to cycle Fo-variation which is larger than 10.96% of the mean Fo value. The individual results were averaged for each of the three loudness levels in each group of children. 3. Results
The results are presented in Table 1. The main findings can be summarized as follows: (1) Significantly (P < 0.01) lower values (assessed by a Wilcoxon’s rank test) of aperiodicity parameters are found in trained than in untrained voices, in each of the three volume categories (soft/comfortable/loud). This trend is illustrated in Fig. 2 (Jitter Factor) and Fig. 3 (irregularity). (2) In contrast, no significant differences (Wilcoxon’s rank test) are found between trained and untrained voices for the mean Fo, within each volume category. No significant differences (Wilcoxon’s rank test) are found between boys and girls, neither for the mean Fo, nor for the aperiodicity measurements, in all voice conditions, and for both schools. Further analysis of the data also showed the following: There is a strong and highly significant correlation (Spearman’s rho = 0.93) between jitter factor and irregularity. The jitter factor shows a weak significant negative correlation with age for normal voicing (Spearman’s rho = - 0.34, P < 0.05 in trained voices; Spearman’s
112
P.H.
Dejonckere
et al. / Imt. J. Pediatr.
Otorhinolaryngol.
35 (1996)
107-l
I5
rho = - 0.31, p < 0.05 in untrained voices), and for loud voicing (Spearman’s rho = - 0.40, p < 0.02 in untrained voices; Spearman’s rho = - 0.33, P = 0.05 in trained voices).
4. Discussion 4.1. Mean Fo value
An exhaustive review of the literature of Fo in children by Wilson [8] summarizes, as averaged values: for boys from 260 Hz (7 years) to 165 Hz (15 years), and for girls from 260 Hz (7 years) to 220 Hz (15 years), but acceptable limits are very large. The mean values obviously decrease with increasing age. Our data, although they concern sustained /a:/ and not running speech, are in full agreement with the average values of the literature. The mean Fo in normal speech is essentially related to the dimensions of the vocal folds. A quite convenient parameter to measure is the length of the entire vocal fold: in children aged 7 Hirano et al. [9] found a length of 6-8 mm, while in children aged 15 the length reaches 9- 15 mm. 4.2. Fo /sound pressure level dynamics
When phonations are produced at different vocal intensities, adjustments of laryngeal biomechanics, subglottal pressure and glottal airflow occur. Physiologi-
Irregularity
as function of loudness
%
n q
Trained Untrained Total
Normal
Loud
Fig. 2. Irregularity as function of loudness of voicing (mean values for each group: Trained children N = 38, untrained children N = 43, and total N = 81).
P.H.
Dejonckere
et al. / lnt. J. Pediatr.
Otorhinolaryngol.
3.5 (1996)
107-I
15
113
Jitter Factor as function of loudness Jitter Factor
Soft
Normal
Loud
Fig. 3. Jitter factor as function of loudness of voicing (mean values for each group as in Fig. 2).
tally, neuromuscular regulation makes the vocal fold closure tighter (adjustment of glottal impedance and vocal fold tonus) when the subglottic pressure increases, i.e. when the subject phonates louder [6]. A lower vocal SPL is generally associated with a smaller closed quotient, i.e. a reduced duration in vocal fold contact during the vibratory cycle [lo]. The relative glottal impedance as evaluated by an elec-
Fo as function of loudness Fo (Hz)
240
L Normal
Loud
Fig. 4. Fo as function of loudness of voicing (mean values I‘or each group as in Fig. 2).
114
P.H.
Dejonckere
et al. / Int. J. Pediatr.
Otorhinolaryngol.
35 (1996)
107-115
troglottographic surface quotient, increases with the sound pressure level. This supposes an active thickening of the vocal fold edge, and a longer closed phase. Therefore, the increase in transglottal flow is limited. Further, a larger vibrating mass limits to some extent the upward trend of Fo [6]. This mechanism relies upon an integrated complex of reflex processes which tends to maintain the required posture and state of tension of the vocal folds in the face of the upward distorting force exerted on them by the expiratory flow [ll]. The thyroarytenoid muscle has been found to exhibit the greatest variation in activity with changes in vocal SPL, especially at relatively low phonatory frequencies. Higher SPLs are associated with increased overall tension in the muscular body of the vocal fold [12] and there is a positive correlation between thyroarytenoid activity and Fo
PI.
Our results show no clear effect of vocal education in young children on this neuromuscular reflex control. This reflex control clearly differs, of course, from volitional control when singing written music.
4.3. Aperiodicity
and acoustic amplitude
The degree of measured perturbation merely provides an indication of the stability of the physiological balance [4]. Orlikoff and Kahane [4] showed that, in normal male adults, the degree of perturbation is inversely related to the acoustic amplitude of the vowel (sustained /a:/ in modal register). A low vocal sound pressure level tends to be associated with a reduced vocal fold contact duration [lo], which might result in increased airflow turbulence and greater randomness in the source spectrum. In louder emissions, the larger vibrating mass contributes to stabilize the oscillator. Clearly this is also observed in children. 4.4. Aperiodicity
and age
Decrease in perturbation means that the vocal fold becomes an oscillator of better quality. The reason of this is probably anatomical: with growth, the vibrating mass increases, and this tends to lower the mechanical damping of vocal fold oscillations [2]. Furthermore, the ratio of the length of the membranous portion to that of the cartilaginous portion of the vocal fold also increases with age: about 3 at the age of 7 and 3.5-5 at the age of 15. Major movement of the vocal fold during vibration takes place in the membranous portion [9], which is probably better suited for vibration. 4.5. Aperiodicity
and duration of education in singing
A hypothesis could be that the quality of the vocal oscillator is enhanced by education and intensive training: a variance analysis demonstrates that, while there is a significant effect on the jitter level of the parameters ‘age’ and ‘training’ (duration of stay in the Cathedral Choir School) separately, this effect totally disappears
P.H.
Dejonckere
et al. / Int. J. Pediatr.
Otorhinolaryngol.
35 (1996)
107-115
115
when ‘age’ and ‘training’ are combined (age x training). The specific (better) biomechanical vocal fold properties seem thus constitutional rather than related to training.
5. Conclusions
1. As a general rule, Fo increases and Fo-perturbation decreases when a normal child (trained or untrained) is voicing louder. 2. As a general rule, Fo-perturbation decreases with age. 3. Children from the Utrecht Cathedral Choir School (with artistic voice education) do not, when spontaneously phonating, control the Fo/SPL dynamics in another way than do normal children from a Utrecht basic school. 4. Children from the Utrecht Cathedral Choir School (with artistic voice education) demonstrate less aperiodicity, that means better mechanical properties of the vocal oscillator than children with normal voices from a Utrecht basic school. 5. This smaller aperiodicity is not related to the duration of stay in the Cathedral Choir School, and is already present at the beginning of vocal training.
References [l] Perkell, J.S. and Klatt, D.H. (1986) Invariance and Variability in Speech Process. Lawrence Earlbaum Associates, Hillsdale, New Jersey. [2] Dejonckere, P.H. and Lebacq, J. (1984) Damping coefficient of oscillating vocal folds in relation with pitch perturbations. Speech Commun. 3, 89-92. [3] Titze, I.R. (1994) Principles of Voice Production. Prentice Hall Inc., Englewood Cliffs, New Jersey. [4] Orlikoff, R.F. and Kahane, J.C. (1991) Influence of mean sound pressure level on jitter and shimmer measures. J. Voice 5, 113-119. [5] Baken, R.J. (1987) Clinical Measurement of Speech and Voice. Taylor and Francis Ltd, London, 1987. [6] Dejonckere, P. (1994) Control of fundamental frequency and glottal impedance with increasing sound pressure level in normal and pathological voices. Voice 3, 10-16. [7] Abberton, E.R.M., Howard, D.M. and Fourcin, A.J. (1989) Laryngographic assessment of normal voice: a tutorial. Clin. Linguist. Phonet. 3, 281-296. [8] Wilson, D.K. (1987) Voice Problems in Children, 3d edn. Williams and Wilkins, Baltimore. [9] Hirano, M., Kurita, S. and Nakashima, T. (1983) Growth, development and aging of human vocal folds. In: Bless, D. and Abbs, J. (Eds.), Vocal Fold Physiology. College Hill Press, San Diego, California, pp. 22-43. [lo] Orlikoff, R.F. (1991) Assessment of the dynamics of vocal fold contact from the electroglottogram : data from normal male subjects. J. Speech Hear. Res. 34, 106661072. [ll] Wyke, B. (1976) Laryngeal reflex mechanisms in phonation. Proceedings XVIth Int. Congr. Logopedics and Phoniatrics, Interlaken 1974, Karger, Basel, pp. 528-537. [12] Hirano, M., Ohala, J. and Vennard, W. (1969) The function of the laryngeal muscles in regulating fundamental frequency and intensity of phonation. J. Speech Hear. Res. 12, 616-628.
ELSEVIER
Journal
of Pediatric
35 (1996)
Otorhinolaryngology
107-115
Fo-perturbation and Fe/loudness dynamics in voices of normal children, with and without education in singing P.H. Dejonckere, G.H. Wienekea, D. Bloemenkamp”, J. Lebacqb “The
Institute
of Phoniatrics, European Research Group on Larynx (GREL), Utrecht University, P.O. Box 85500, Nl-3508 GA Utrecht, The Netherlands bDepartment of Physiology, Catholic University of Louvain, B-1200 Brussels, Belgium Received
9 August
1995; accepted
1 October
1995
Abstract
Sustainedphonations were comparedin two groups of children (aged 7-12) one with specialartistic voice education and one from a normal school, without voice complaintsor problems.The hypothesisof specific(better) biomechanicalvocal fold propertiesin the first group is confronted with the hypothesisof differencessolely related to training of voice control. In both groups, Fo-aperiodicity was measuredin a sustainedphonation at 3 different SPL levels.As a generalrule, aperiodicity clearly decreases when the voice becomes louder. Aperiodicity is highly significantly lower, at all SPL-levels,in children with trained singingvoices: this impliesbetter mechanicalpropertiesof the vocal oscillator. The Fo/SPL relation on a sustained/a:/ doesnot differ in trained and untrained children’svoices: out of singing context, trained children do not spontaneouslycontrol the Fo/SPL dynamics differently from untrained children. The higher regularity of vocal fold pulsesis not related to the duration of training. Voice; Singing voice; Education; Fundamental frequency; Jitter; Aperiodicity; Laryngograph; Electroglottography; Loudness Keywords:
* Corresponding 0165-5876/96/$1.5.00 SSDI
author. 0 1996 Elsevier
0165-5876(95)01291-5
Science
Ireland
Ltd.
All rights
reserved
108
P.H.
Dejonckere
et
al. /
Int. J. Pediatr.
Otorhinolaryngol.
35 (1996)
107-115
1. Introduction 1.1. Fundamental frequency perturbation
A certain degree of irregularity is inherent in all of the speech production processes, regardless of how much practice there has been [l]. The term ‘jitter’ refers to short-term (cycle-to-cycle) variability in fundamental frequency (Fo), not accounted for by voluntary changes, as e.g. intonation. It is well known that, even in the voice signal of a (perceptually) perfectly steady sound, small fluctuations in frequency are present: they reflect the internal ‘noises’ of the human body. Sources of Fo micro-perturbation or aperiodicity are biomechanical (vocal folds are a damped oscillator), aerodynamic and neurological [2,3]. Measurements of jitter are primarily thought to assess the stability of vocal fold vibration [4]. There is by now a considerable body of literature that asserts the usefulness of measuring frequency perturbation in evaluating laryngeal and vocal pathology [5]. However, when only normal subjects are considered, is this irregularity phenomenon, as measured on a sustained /a:/, smaller for trained singers voices than for untrained voices? Have singers a laryngeal oscillator with better mechanical characteristics than normal people? And if so can this be considered a consequence of the singing education and training, or is it constitutional? These questions are not answered at present. In order to avoid gender differences as well as variability in singing style and practice, general education, profession etc., we choosed to compare two groups of children: on the one hand 47 children from the Utrecht Cathedral Choir School; on the other hand 54 children from a normal basic school in Utrecht. At the Utrecht Cathedral Choir School, young children, selected on the basis of their musical aptitudes, receive, besides a general education, a special training in (particularly vocal and sacred) music. 1.2. Neuromuscular
control of phonation
Education in singing implies - besides the artistic aspect - acquiring a better neuromuscular regulation of the voice production mechanism, especially a better control of pitch and loudness. Generally speaking, the amplitude of vibration of the vocal folds increases with lung pressure, which is, in normal voices, closely related to sound pressure level. This increased vibration amplitude determines a higher value of the average dynamic strain in the vocal fold, and we can expect, therefore, that the fundamental frequency will also increase with lung pressure [3]. This effect is obvious in elastic models of the larynx, as well as in tracheotomized patients with a bilateral laryngeal paralysis in adduction. In contrast, a singer practising a ‘messa di vote’ exercise is able to produce a long crescendo followed by a long diminuendo on a single note, thus at constant pitch. Between these extremes, when normal adult subjects are asked to produce, in a spontaneous way (thus not in singing voice), successive sustained emissions on /a:/ with increasing loudness, a concomitant slight increase in pitch is usually observed. The rate is about 1.4 Hz/dB (significantly less than in mechanical models or than in pathological cases) [6].
P.H.
Dejonckere
et al. / Int. J. Pediatr.
Otorhinolaryngol.
35 (1996)
107-115
109
It could be hypothesized that children with special training in singing - as investigated in this experiment - will show, when producing sustained emissions on /a:/ with increasing loudness, a better neuromuscular control of pitch, and thus less increase in Fo, than children without training. In the present study, microperturbations of Fo and Fo-dynamics as a function of sound pressure level are compared, in standardized voicing conditions, between the two groups of children, those with and those without singing education.
2. Material
and method
2.1. Subjects 101 normal children, aged 7 to 12, were investigated. Exclusion criteria were -abnormal sounding voice -acute, subacute or chronic inflammation/infection of airways -use of inhalation drugs -antecedents of any kind of laryngeal and pharyngeal pathology -chronic aspecific respiratory pathology -antecedents of intubation -incapacity of producing the required voicings for the experiment. After application of exclusion criteria, the following subjects remained: 38 vocally trained children from the Utrecht Cathedral Choir School (14 boys and 24 girls): mean age 10.3 + 1.3 years. 43 vocally untrained children from a normal basic school in Utrecht (17 boys and 26 girls): mean age 10.3 h 1.2 years. The distribution of ages is shown in Fig. 1. The age distributions of the two groups do not differ significantly. All parents where asked to complete a questionnaire concerning their child, and written informed consent was required. 2.2. Protocol In each child, nine recordings were made of a sustained /a:/ (2 to 4 s): three recordings at comfortable pitch and loudness (sound pressure level, SPL, controlled between 63.5 and 66.5 dB(A) at 30 cm), three recordings (2-4 s) at comfortable pitch and louder voicing (SPL controlled between 67 and 72 dB(A)), and three recordings (2-4 s) at comfortable pitch and soft voicing (SPL controlled between 58 and 63 dB(A)). 2.3. Instrumental
material
All recordings were made in a quiet but not a sound-proof room, using a Casio DA-7 DAT-recorder (frequency range: 10-20.000 Hz). The signal was analyzed
110
P.H.
Dejonckere
et al. i Int. J. Pediatr.
Otorhinolaryngol.
35 (1996)
107-115
afterwards using a PCLX PCpitch and Lx2 program (fundamental frequency and waveform analysis software for the Laryngograph: Laryngograph Ltd, London) [7]. The accuracy of the period measurement is normally to within 1 us, at worst 2 us. 2.4. Parameters
For each condition of loudness, a stable 500 ms segment was selected from the best suited /a:/ signal out of the three records. The following parameters were calculated: (1) mean fundamental frequency (Fo) (2) jitter factor The jitter was expressed as the ‘Jitter factor’, that is the mean difference between the frequencies of adjacent cycles divided by the mean frequency, multiplied by 100 151 $1;:
IP-c+,I] x 100
Jitter factor =
where Fi = frequency of the i th cycle, in Hz, and II = number of periods in the sample. The jitter factor is a Fo-related measure.
Age distribution
9
10
11
12
Years Fig. 1. Age distribution in the groups of children with and without training in singing.
P.H.
Dejonckere
et al. i ht.
J. Pediatr.
Otorhinolaryngol.
35 (1996)
III
107-115
Table I Overview of experimental data
Trained
choices
Soft voice
Normal voice
Loud voice
N =38
( * 13.0)
N = 38 10.3 228.9 Hz 3.0 5.2
(k6.1)
N = 38 10.3 278.8 Hz 1.3 0.6%
(IL 1.3) (k21.1) (z!c 0.5) (ix 0.9)
( * 1.2) (i26.1) ( k 5.6) (+- 15.2)
N =43 10.3 226.3 Hz 5.9 15.2%
(k (f (k (&
N = 44 10.3 276.5 Hz 2.4 3.3%
( * 1.2) (i 37.0) (IL 1.5) ( I!Y4.6)
Age Mean Fo Jitter factor Irregularity
10.3 (f 1.3) 206.8 Hz (k 71.8) 4.9 ( + 4.7)
Lintrained
N =43 10.3 223.2 Hz 7.1 I 9.2’%,
voices
Age Mean Fo Jitter factor Irregularity Trained coicrs Age
& untrained
Mean Fo Jitter factor Irregularity
12x41
1.2) 22.9)
5.2) 13.7)
N =82
N = 81
N =81
10.3 215.5 Hz 6.1 16.2%
(5 1.3) ( + 35.6) (i 1.5)
(x!I 1.2) (+ 53.0) (k 5.3) ( f 14.5)
10.3 227.5 Hz 4.5 10.5%
( f 1.2) (i 29.3) (k4.1) (k 11.8)
10.3 ( z!Y1.2) 277.6 Hz (& 30.5) 1.9 (k 1.3) 2.0% ( f 3.7)
(3) ‘irregularity’: this is another measure of the degree of aperiodicity: it is the percentage of fundamental frequency pairs (adjacent cycles) with a cycle to cycle Fo-variation which is larger than 10.96% of the mean Fo value. The individual results were averaged for each of the three loudness levels in each group of children. 3. Results
The results are presented in Table 1. The main findings can be summarized as follows: (1) Significantly (P < 0.01) lower values (assessed by a Wilcoxon’s rank test) of aperiodicity parameters are found in trained than in untrained voices, in each of the three volume categories (soft/comfortable/loud). This trend is illustrated in Fig. 2 (Jitter Factor) and Fig. 3 (irregularity). (2) In contrast, no significant differences (Wilcoxon’s rank test) are found between trained and untrained voices for the mean Fo, within each volume category. No significant differences (Wilcoxon’s rank test) are found between boys and girls, neither for the mean Fo, nor for the aperiodicity measurements, in all voice conditions, and for both schools. Further analysis of the data also showed the following: There is a strong and highly significant correlation (Spearman’s rho = 0.93) between jitter factor and irregularity. The jitter factor shows a weak significant negative correlation with age for normal voicing (Spearman’s rho = - 0.34, P < 0.05 in trained voices; Spearman’s
112
P.H.
Dejonckere
et al. / Imt. J. Pediatr.
Otorhinolaryngol.
35 (1996)
107-l
I5
rho = - 0.31, p < 0.05 in untrained voices), and for loud voicing (Spearman’s rho = - 0.40, p < 0.02 in untrained voices; Spearman’s rho = - 0.33, P = 0.05 in trained voices).
4. Discussion 4.1. Mean Fo value
An exhaustive review of the literature of Fo in children by Wilson [8] summarizes, as averaged values: for boys from 260 Hz (7 years) to 165 Hz (15 years), and for girls from 260 Hz (7 years) to 220 Hz (15 years), but acceptable limits are very large. The mean values obviously decrease with increasing age. Our data, although they concern sustained /a:/ and not running speech, are in full agreement with the average values of the literature. The mean Fo in normal speech is essentially related to the dimensions of the vocal folds. A quite convenient parameter to measure is the length of the entire vocal fold: in children aged 7 Hirano et al. [9] found a length of 6-8 mm, while in children aged 15 the length reaches 9- 15 mm. 4.2. Fo /sound pressure level dynamics
When phonations are produced at different vocal intensities, adjustments of laryngeal biomechanics, subglottal pressure and glottal airflow occur. Physiologi-
Irregularity
as function of loudness
%
n q
Trained Untrained Total
Normal
Loud
Fig. 2. Irregularity as function of loudness of voicing (mean values for each group: Trained children N = 38, untrained children N = 43, and total N = 81).
P.H.
Dejonckere
et al. / lnt. J. Pediatr.
Otorhinolaryngol.
3.5 (1996)
107-I
15
113
Jitter Factor as function of loudness Jitter Factor
Soft
Normal
Loud
Fig. 3. Jitter factor as function of loudness of voicing (mean values for each group as in Fig. 2).
tally, neuromuscular regulation makes the vocal fold closure tighter (adjustment of glottal impedance and vocal fold tonus) when the subglottic pressure increases, i.e. when the subject phonates louder [6]. A lower vocal SPL is generally associated with a smaller closed quotient, i.e. a reduced duration in vocal fold contact during the vibratory cycle [lo]. The relative glottal impedance as evaluated by an elec-
Fo as function of loudness Fo (Hz)
240
L Normal
Loud
Fig. 4. Fo as function of loudness of voicing (mean values I‘or each group as in Fig. 2).
114
P.H.
Dejonckere
et al. / Int. J. Pediatr.
Otorhinolaryngol.
35 (1996)
107-115
troglottographic surface quotient, increases with the sound pressure level. This supposes an active thickening of the vocal fold edge, and a longer closed phase. Therefore, the increase in transglottal flow is limited. Further, a larger vibrating mass limits to some extent the upward trend of Fo [6]. This mechanism relies upon an integrated complex of reflex processes which tends to maintain the required posture and state of tension of the vocal folds in the face of the upward distorting force exerted on them by the expiratory flow [ll]. The thyroarytenoid muscle has been found to exhibit the greatest variation in activity with changes in vocal SPL, especially at relatively low phonatory frequencies. Higher SPLs are associated with increased overall tension in the muscular body of the vocal fold [12] and there is a positive correlation between thyroarytenoid activity and Fo
PI.
Our results show no clear effect of vocal education in young children on this neuromuscular reflex control. This reflex control clearly differs, of course, from volitional control when singing written music.
4.3. Aperiodicity
and acoustic amplitude
The degree of measured perturbation merely provides an indication of the stability of the physiological balance [4]. Orlikoff and Kahane [4] showed that, in normal male adults, the degree of perturbation is inversely related to the acoustic amplitude of the vowel (sustained /a:/ in modal register). A low vocal sound pressure level tends to be associated with a reduced vocal fold contact duration [lo], which might result in increased airflow turbulence and greater randomness in the source spectrum. In louder emissions, the larger vibrating mass contributes to stabilize the oscillator. Clearly this is also observed in children. 4.4. Aperiodicity
and age
Decrease in perturbation means that the vocal fold becomes an oscillator of better quality. The reason of this is probably anatomical: with growth, the vibrating mass increases, and this tends to lower the mechanical damping of vocal fold oscillations [2]. Furthermore, the ratio of the length of the membranous portion to that of the cartilaginous portion of the vocal fold also increases with age: about 3 at the age of 7 and 3.5-5 at the age of 15. Major movement of the vocal fold during vibration takes place in the membranous portion [9], which is probably better suited for vibration. 4.5. Aperiodicity
and duration of education in singing
A hypothesis could be that the quality of the vocal oscillator is enhanced by education and intensive training: a variance analysis demonstrates that, while there is a significant effect on the jitter level of the parameters ‘age’ and ‘training’ (duration of stay in the Cathedral Choir School) separately, this effect totally disappears
P.H.
Dejonckere
et al. / Int. J. Pediatr.
Otorhinolaryngol.
35 (1996)
107-115
115
when ‘age’ and ‘training’ are combined (age x training). The specific (better) biomechanical vocal fold properties seem thus constitutional rather than related to training.
5. Conclusions
1. As a general rule, Fo increases and Fo-perturbation decreases when a normal child (trained or untrained) is voicing louder. 2. As a general rule, Fo-perturbation decreases with age. 3. Children from the Utrecht Cathedral Choir School (with artistic voice education) do not, when spontaneously phonating, control the Fo/SPL dynamics in another way than do normal children from a Utrecht basic school. 4. Children from the Utrecht Cathedral Choir School (with artistic voice education) demonstrate less aperiodicity, that means better mechanical properties of the vocal oscillator than children with normal voices from a Utrecht basic school. 5. This smaller aperiodicity is not related to the duration of stay in the Cathedral Choir School, and is already present at the beginning of vocal training.
References [l] Perkell, J.S. and Klatt, D.H. (1986) Invariance and Variability in Speech Process. Lawrence Earlbaum Associates, Hillsdale, New Jersey. [2] Dejonckere, P.H. and Lebacq, J. (1984) Damping coefficient of oscillating vocal folds in relation with pitch perturbations. Speech Commun. 3, 89-92. [3] Titze, I.R. (1994) Principles of Voice Production. Prentice Hall Inc., Englewood Cliffs, New Jersey. [4] Orlikoff, R.F. and Kahane, J.C. (1991) Influence of mean sound pressure level on jitter and shimmer measures. J. Voice 5, 113-119. [5] Baken, R.J. (1987) Clinical Measurement of Speech and Voice. Taylor and Francis Ltd, London, 1987. [6] Dejonckere, P. (1994) Control of fundamental frequency and glottal impedance with increasing sound pressure level in normal and pathological voices. Voice 3, 10-16. [7] Abberton, E.R.M., Howard, D.M. and Fourcin, A.J. (1989) Laryngographic assessment of normal voice: a tutorial. Clin. Linguist. Phonet. 3, 281-296. [8] Wilson, D.K. (1987) Voice Problems in Children, 3d edn. Williams and Wilkins, Baltimore. [9] Hirano, M., Kurita, S. and Nakashima, T. (1983) Growth, development and aging of human vocal folds. In: Bless, D. and Abbs, J. (Eds.), Vocal Fold Physiology. College Hill Press, San Diego, California, pp. 22-43. [lo] Orlikoff, R.F. (1991) Assessment of the dynamics of vocal fold contact from the electroglottogram : data from normal male subjects. J. Speech Hear. Res. 34, 106661072. [ll] Wyke, B. (1976) Laryngeal reflex mechanisms in phonation. Proceedings XVIth Int. Congr. Logopedics and Phoniatrics, Interlaken 1974, Karger, Basel, pp. 528-537. [12] Hirano, M., Ohala, J. and Vennard, W. (1969) The function of the laryngeal muscles in regulating fundamental frequency and intensity of phonation. J. Speech Hear. Res. 12, 616-628.