JOURNAL

OF COMMUNICATION

DISORDERS

8 (1975). 97-104

ATTRIBUTES OF LOUDNESS, PITCH AND RATE AMONG MALE CHILDREN HARRY Department of Communication

N. SEYMOUR

Studies, University of Massachusetts,

Amherst, Massachusetts

01002

Ratings, by trained judges, of vocal loudness level, pitch level and rate of speaking for a nonpathological sample population of male children were studied relative to subjective ratings of vocal merit. Averaged physical measurements for loudness level, pitch level and rate of speaking determined points of reference and the ranges within which parameters were rated. Vocal samples of male children, judged to have more acceptable voices, were rated significantly higher in pitch level, greater in loudness and faster in rate of speaking than children judged to have less acceptable voices.

Introduction Research has provided physical data that define limits of appropriate pitch, rate and loudness of voice for the male and female adult (Murray and Tiffin, 1934, Bartholomew, 1934; McGinnis, Elnich and Kraichman, 1951). Studies by Lewis and Tiffin (1934) and Black (1942) applied psychophysical methods to the study of vocal attributes of male adults. These investigators were among the first to differentiate, by means of psychophysical methods, good speaking voices from poor speaking voices, and to identify and subsequently to quantify in objective terms those factors associated with either good or poor vocal production. The research devoted to vocal characteristics of children between birth and pubescence provides little information regarding attributes of voice that identify with vocal merit. Studies by Fairbanks (1942), Fairbanks, Herbert and Hammond (1949), McGlone (1966) and Fairbanks, Wiley and Lassman (1949) represent acoustical investigations of the pitch of children, boys and girls, from birth to 8 years of age. Additional data on pitch of older children may be found in Hollein and Malcik (1967). However, other than the physical data of pitch characteristics, there is little information available on characteristics of rate and loudness or vocal attributes of the child’s voice. In the assessment of vocal merit, it is axiomatic that the trained listener have a preconceived concept of suitability of voice for a speaker’s age and sex. This concept may be more accurate for the adult’s voice than for the child’s since attributes of vocal merit are less definable for children. If this is so, appropriate clinical judgments such as identification of voice problems among children and their subsequent treatment might be questionable. More information is needed 0 American Elsevier Publishing

Company,

Inc.,

1975

97

98

HARRY

N. SEYMOUR

about attributes and parameters of children’s voices that may determine evaluative judgments. Thus the purpose of this study was to investigate the attributes of (a) loudness level, (b) pitch level and (c) rate of speaking associated with vocal merit. For purposes of this investigation, pitch level was defined as the highness or lowness of vocal tone at which an individual was judged to speak. The magnitude of auditory sensation at which an individual was judged to speak defined loudness level. Rate of speaking was defined as the rapidity with which words were spoken during a unit of time as judged by a panel of listeners.

Method One hundred and thirty male children between the ages of 6 and 8 participated in the investigation. The children attended parochial schools in Columbus, Ohio, and represented a variety of racial groups and socioeconomic levels. The criteria for selection of male children were that they not have organic defects of hearing, speech or voice. Also, it was important to select children who had not reached puberty; therefore, a ceiling age of 8 was imposed. The first 130 male children to meet the above criteria were selected for study. All children were screened for hearing and organic vocal disorders. Two minutes of spontaneous and continuous speech of each child was recorded. Speech stimuli consisted of pictures, objects and stories. In their investigation of pitch and nasality, Sherman and Goodwin (1954) prepared 20 second speech samples for the judgmental tasks. Likewise for this investigation, a 20 second sample was chosen. Twenty second vocal samples with 5 second intervals between successive samples were copied from the 2 minutes of continuous speech, randomized and then rerecorded in the backward play mode. The method of recorded speech stimuli played backwards was introduced by Sherman (1954) and used in this investigation as a means of alleviating the influence of extraneous factors such as dialect, contextual matter, etc., when rating vocal characteristics. Forty advanced students majoring in speech pathology participated as judges for this investigation. The 40 judges were divided into four groups (A, B, C and D) of 10 members each, and each of the four groups was assigned one of four listening tasks. The first group (A) was asked to rate the 130 vocal samples on a 7 point equal-appearing intervals rating scale of vocal merit in which I represented extremely acceptable and most meritorious and 7 extremely unacceptable and least meritorius. The 130 vocal samples were then rated on 7 point equal-appearing interval scales of (a) loudness level, (b) pitch level and (c) rate of speaking by judges in groups B, C and D, respectively. The 7 point equal-appearing intervals scale for loudness level included scale values in which I represented extremely soft and 7 extremely loud; for pitch level, the scale values were from I, extremely low to 7,

ATTRIBUTES

OF LOUDNESS,

PITCH AND RATE AMONG

MALE CHILDREN

99

extremely high; scale values for rate of speaking were from I, extremely slow to 7, extremely fast. Although all 40 judges listened to the same 130 vocal samples, group A judged for overall vocal merit and groups B, C and D each rated the samples on a different dimension of voice. Prior to rating the vocal samples, each of the four groups of judges listened to a training tape depicting samples of voices representative of the middle and extreme scale points for the parameter they were to judge. The vocal samples used in the training tapes were selected on the basis of 100% agreement among ratings by four experienced voice clinicians. Training tapes consisted of three vocal samples that represented the middle and three samples that represented each of the two end “anchor” points on the scales for loudness level, pitch level and rate of speaking. Average physical measures were obtained for these vocal samples (Table 1). An audiometric calibration kit was used to determine dB SPL, “C” scale, (0.0002 dyn/cm*) for average peak intensity levels for the softest vocal sample, 62 dB. The average vocal sample was 72 dB, and the loudest vocal sample was 86 dB. A pitch indicator, which extracts fundamental frequency, was used to determine fundamental frequencies for the middle and extreme end levels on the rating scale of pitch level. The mean fundamental frequency for the lowest pitch level was 2 15 Hz, for the average pitch level 260 Hz and 350 Hz for the highest pitch level. Rate of speaking in words per minute for the speaker sample illustrating the slowest rate was 58, 119 for the average speaker sample and 188 for the fastest speaker sample. Reliability of the judges was tested by computing the average intercorrelation TABLE I Average Physical Measures of Loudness, Pitch and Rate Levels for the Middle and Extreme Scale Points on the Rating Scales Scale values Extremely

1

Average

4

Extremely

7

_ Loudness level Soft Average Loud Pitch level Low Average High Rate level Slow Average Fast

62 dB 12 dB 86 dB 215 Hz 260 Hz 350 Hz 58 wds./min. 119 wds./min. 188 wds./min

100

HARRY

N. SEYMOUR

coefficient using an analysis of variance technique (March et al., 1958). The coefficients of reliability for a panei of 10 judges were 0.92 for vocal merit, 0.98 for loudness level, 0.96 for pitch level and 0.87 for rate. The coefficients of reliability were sufficiently high enough to regard all four groups of judges as reliable. Based on the rating scale values for vocal merit the 130 vocal samples were divided into two primary groups, voices judged to be acceptable and voices judged to be unacceptable. These dichotomized groups were determined by selecting the approximate end values of the 7 point scale and discarding middle values. Thus, 41 vocal samples with mean values from I to 3.5 were selected as acceptable voices and 60 vocals samples with mean values from 4.5 to 7 were selected as unacceptable voices. There were 29 vocal samples with mean values between 3.6 and 4.4, which separated the dichotomized groups, acceptable voices and unacceptable voices and were not included in the analysis that followed. The unacceptable voices represented negative values on a continuum of vocal merit and should not be interpreted as 60 pathological voices. The frequency distribution for mean ratings of vocal merit for backward and forward play modes are illustrated in Fig. 1. In order to test the assumption underlying the presentation of vocal samples in the backward play mode, 10 judges, advanced speech pathology majors, rated all 130 vocal samples in the forward mode. It is important to note that mean ratings of vocal samples for the backward play mode recordings and the forward play mode recordings present similar distributions (Fig. I). The comparable distributions of

40 ~36 532 E28 z24 g20 16 $12 =

z

FORWARD BACKWARD

MODE RATINGS MODE RATINGS

________ -

8-

4-

“’

,rfI~rlI~I~I~I -2 EXTREMELY ACCEPTABLE

3

RATING

Fig. 1.

4 SCALE

5

6’7 EXTREMELY UNACCEPTABLE

Frequency distribution of mean scores for ratings of vocal samples of 130 male children recorded in the forward and backward play mode.

ATTRIBUTES

OF LOUDNESS.

PITCH AND RATE AMONG

MALE CHILDREN

101

both play modes suggest that either procedure may be used in global ratings of voice. However, since the reliability of judges for the forward play mode was 0.83 as opposed to 0.92 for the backward play mode, the latter mode might be a more valid rating procedure. Also, the slightly greater skewness for ratings in the forward play mode could be a function of the “halo effect.” Extraneous factors such as dialects, misarticulations and so on, may influence ratings of vocal samples played in the forward mode.

Results The t test for a difference between independent mean values was used to determine differences between children with acceptable voices and children with unacceptable voices in ratings of (a) loudness level, (b) pitch level and (c) rate of speaking. The children with acceptable voices had significantly higher means than the children with unacceptable voices in these ratings (Table 2). All t values were significant at the 0.0 1 level. The frequency distributions of mean scores for ratings of loudness level, pitch level and rate of speaking are illustrated in Figs. 2,3 and 4, respectively. It is apparent in Table 2 and Figs. 2, 3 and 4 that children rated as having acceptable voices differed from those judged unacceptable in the following ways: (1) rated as louder, (2) rated as higher in pitch and (3) judged as speaking faster. Discussion Differences between children with more acceptable voices and children with less acceptable voices in ratings of (a) loudness level, (b) pitch level and (c) rate of speaking suggest that there are characteristics of these parameters of voice that may be identified with vocal merit. Thus, children rated as having more acceptable voices were judged also to have higher pitched voices, louder voices and faster

TABLE r-Values for Differences

2

between Children with Acceptable and Unacceptable Loudness Level, Pitch Level and Rate of Speaking

Voices for Ratings of

Means

Loudness level Pitch level Rate

AV

uv

ta

4.61 4.69 4.98

3.58 2.98 3.68

3.35 6.61 5.50

‘Significant at the 0.01 level 2.36 (df, 99). AV-children with acceptable voices; UV-children

with unacceptable

voices,

102

HARRY

20 w ‘8 S 16 g 14 gs 12 10 8i =

N. SEYMOUR

ACCEPTABLE ______ UNACCEPTABLE

VOICES VOICES

RATING Fig. 2.

SCALE

Frequency distribution of mean scores of ratings of loudness level for children with acceptable voices and children with unacceptable voices.

speaking rates. Conversely, children judged to have less acceptable voices were judged to have lower pitched voices, softer voices and slower speaking rates. Some of the findings by Murray and Tiffin ( 1934)) Bartholomew (1934), McGinnis, Elnich and Kraichman (195 1) and Black (1942) suggest that within appropriate vocal limits adult voices tend to be preferred when pitch is relatively low and

z

20 18

-

%? 16g 14 E 12=f 10 z 8

z

“4-

,”

2’

-ACCEPTABLE -----UNACCEPTABLE

VOICES VOICES

/ i / i a’

L 1234

567

RATING Fig. 3.

SCALE

Frequency distribution of mean scores of ratings of pitch level for children with acceptable voices and children with unacceptable voices

ATTRIBUTES

OF LOUDNESS,

PITCH AND RATE AMONG

201816 14 121086420

:

i

I 1

:: :: : : :

I 2

:‘R : : ‘I‘\ \

-ACCEPTABLE -----UNACCEPTABLE

103

VOICES VOICES

:\

L 3 RATING

Fig. 4.

MALE CHILDREN

1 4

I 5

I 6

-;I

SCALE

Frequency distribution of mean scores of ratings of rate of speaking for children with acceptable voices and children with unacceptable

voices.

variable, and rate of speaking is relatively slow. A disparity appears to exist between what are considered favorable attributes for the adult and child’s voice. What is known about some vocal attributes of adults does not appear to apply for the child’s voice. There is little doubt that the relatively lower pitch level of a man would be undesirable for a child, but within an appropriate pitch range of children a lower pitch level is not preferrable to a higher pitch level. Similarly, within certain determinable limits, a greater loudness level appears to be preferred over a softer loudness level, and preference is for a faster rate of speaking than a slower rate of speaking. The training tapes comprised vocal samples that were representative of end and middle rating scale values. The training samples provided the judges with points of reference for making their subjective ratings. Averaged physical measures of the training samples represented, in objective terms, the limits within which the subjective ratings were made (Figs. 2-4). Vocal samples that were perceived to be below 72 dB, 260 Hz and 119 wds./min. were less acceptable than vocal samples perceived to be above those values. Also, it should be concluded that ratings of vocal samples that approximated extreme ends on the rating scales were least characteristic of vocal merit, such as 215 and 360 Hz, lowest and highest pitch level for this sample population. The results of this study indicate that ratings of vocal merit among male children between the ages of 6 and 8 may be characterized by predictable patterns of vocal behavior with regard to judged loudness level, pitch level and rate of speaking. Also, the extent of difference between acceptable and unacceptable vocal groups

104

HARRY

N. SEYMOUR

varied among the three parameters. The t values for differences between children with more acceptable voices and children with less acceptable voices revealed that judged pitch level represented a greater distinguishing factor between the two vocal groups than either loudness level or rate of speaking. Thus, in the determination of vocal merit among children, pitch level may represent a more affective dimension of voice than rate or loudness level. When one considers the elusive vocal features that,attract attention, either negatively or positively, to a child’s voice, pitch level may relate closely to the recognition of aberrant vocal behavior more so than other single vocal features.

References Bartholomew, W.T. A physical definition of “good voice quality” in the male voice. J. Acoust. Sot. Am., 1934, 6, 25-33. Black, J.W. A study of voice merit. Quart. J. Speech, 1942, 28, 67-74. Fairbanks, G. An acoustical study of the pitch of infant hunger wails. Child Development, 1942.14, 227-232. Fairbanks, G., Herbert, E.L., Hammond, J. M. An acoustical study of vocal pitch in seven and eight-year-old girls. Child Developmenf, 1949, 20, 7 l-78. Fairbanks, G., Wiley, J.H., Lassman, F.M. An acoustical study of vocal pitch in seven- and eight-year-old boys. Child Development, 1949, 20, 6349. Hollein, H., Malcik, E. Evaluation of cross-sectional studies of adolescent voice change in males. Speech Monographs, 1967, 34, 8C-84. Lewis, D., Tiffin, J. A psychophysical study of individual differences in speaking ability. Arch. Speech, 1934, 1, 43-60. March, N., Weaver, C.H., Morrison, S., Black, J.W. Observed and predicted estimates of reliability of aspects of a speech articulation rating scale. Speech Monographs, 1958, 25, 296304. McGinnis, C.S., Elnich, M., Kraichman, M. A study of the vowel formants of well-known male operatic singers. J. Acoust. Sot. Am., 1951, 23, 44@446. McGlone, R. E. Vocal pitch characteristics of children aged one to two years. Speech Monographs. 1966, 33, 178-181. Murray, E., Tiffin, J. An analysis of some basic aspects of effective speech. Arch. Speech, 1934, 1, 61-82. Sherman, D., Goodwin, F. Pitch level and nasality. J. Speech Hearing Dis., 1954, 19, 423-428. Sherman, D. The merits of backward playing of connected speech in the scaling of voice quality disorders. J. Speech Hearing Dis., 1954, 19, 3 12-321.

Attributes of loudness, pitch and rate among male children.

JOURNAL OF COMMUNICATION DISORDERS 8 (1975). 97-104 ATTRIBUTES OF LOUDNESS, PITCH AND RATE AMONG MALE CHILDREN HARRY Department of Communication...
463KB Sizes 0 Downloads 0 Views