HOLLIEN ET AL 1991 Listening preferences for voice types as a function of age

Prévia do material em texto

J. COMMUN. DISORD. 
24 (1991). 157-171 
LISTENING PREFERENCES FOR VOICE TYPES 
AS A FUNCTION OF AGE 
HARRY HOLLIEN 
Institute for Advanced Study of the Communication Processes, Universily of Florida 
MARYLOU PAUSEWANG GELFER 
Department of Speech and Hearing Sciences, Indiana University 
TERRY CARLSON 
Department of Speech Communication, California State University, Los Angeles 
This experiment was designed primarily to generate information about the preferences of 
older listeners for various classes of voices. Speech samples for that purpose were elicited 
from 80 speakers, who provided the desired stimuli (sentences) under frequency and intensity 
control. Specifically, there were eight cells in the design, each represented by 10 speakers 
(5 male and 5 female); all combinations of low, medium, and high speaker fundamental 
frequency (SFF) were combined with soft, middle and loud vocal intensity (VI) produc- 
tions-except for the low-SFFihigh-VI combination, which proved impossible to obtain. 
Listeners were four groups of 20 individuals equally divided as to sex. The two older of 
these groups, designated as the experimental subjects, were: older adults (60-70 years of 
age) and the elderly (80-90 years). The two younger groups served as controls; they included 
young adults (20-30 years of age) and middle-aged adults (aged 40-50 years). Listeners rated 
each speech sample on a 5-point preference scale varying from “like very much” to “dislike 
very much.” The results suggest that most listeners prefer medium intensity voices. Other 
preference tendencies were toward low-pitched voices and (slightly) toward male speakers; 
but these trends were not as marked as the first. Most importantly, there were no systematic 
differences in voice type preferences between or among the older and younger groups-or 
between male and female listeners. 
INTRODUCTION 
This project on voice class preferences is part of a larger effort focused 
on the study of the communication capabilities of older individuals. It is 
the second in a subset (see Tolhurst et al., 1984) developed in response 
to the observation that very little information is available about the pref- 
erences of older adults for various heard stimuli. Questions of this type 
appear to be of significance for a number of reasons. Primary among them 
Address correspondence to Harry Hollien, Ph.D., Institute for Advanced Study of the 
Communication Processes, 50-Dauer Hall, University of Florida, Gainesville, FL 32611. 
0 1991 by Elsevier Science Publishing Co., Inc. 157 
655 Avenue of the Americas, New York, NY 10010 0021~9924191/$3.50 
158 H. HOLLIEN et al. 
is the fact that the number of older citizens in the population is increasing 
steadily and they are likely to seek structured services and experiences 
appropriate to their needs. Knowledge of their preferences (of all types) 
is necessary if the desired responses are to be effective. 
It should be noted also that there are two quite different groups of older 
individuals who are in need of these services: those who are handicapped 
in some manner and, in contrast, those who constitute a potentially active 
resource group. While these subpopulations lie at the ends of the continum 
of older people, together they make up the great majority. As a matter 
of fact, it will be found that the underlying goal of much current geron- 
tological research is to diminish the first group through rehabilitation, and 
to maintain (and/or expand) the second by means of stimulative and pro- 
ductive activity. It is quite possible that the judicious use of listening 
preference information will contribute to these goals as individuals re- 
sponsible for rehabilitative, informative, and recreational activities thusly 
can employ auditory materials designed to be both attractive and effec- 
tive. Unfortunately, the present situation is one where such decision mak- 
ers have but limited data available to them. Indeed, they must rely upon 
their personal opinion as to those vocal features which are the most ap- 
pealing (or at least, nondistracting) when they attempt to select speakers 
appropriate to the cited tasks/procedures. 
LISTENER PREFERENCES 
First, it would appear timely to consider what is known about listener 
preferences-in general and as a function of age. Not a great deal of 
information is available and much of it is noncontemporary. 
One opinion is that “low-pitched” voices are generally preferred by 
listeners (Austin, 1965; Cooper, 1973; Linke, 1973). Moreover, this pre- 
sumed societal preference seems (anyway) to have originated from public 
reaction to radio/tv announcers and movie celebrities. On the other hand, 
the research relevant to this question-most of which was reported some 
time ago-appears contradictory. For example, relatively high funda- 
mental frequency values sometimes have been associated with voices 
classed as superior. The very early work reported by Lynch (1934) is a 
case in point. She examined a number of acoustic variables in an attempt 
to determine the vocal differences between trained and untrained speakers 
and reported that the speaking fundamental frequency (SFF) levels ob- 
served for her trained (and presumably superior) speakers were higher 
than were those for the untrained subjects. Other authors (Block, 1942; 
Cowan, 1936; Lewis and Tiffin, 1933; Murray and Tiffln, 1933; Pronovost, 
1942; Snidecor, 195 1) working within this time frame report data that agree 
with those of Lynch. Yet these presumed “superior” talkers exhibited 
SFF levels that were significantly higher than those reported for normal 
LISTENING PREFERENCES 159 
healthy (and contemporary) adult males (Benjamin, 1981; Fitch and Hol- 
brook, 1970; Hollien and Jackson, 1973; Hollien and Shipp, 1972; Krook, 
1988; Mysak, 1959; Perello, 1983) and adult women (dePinto and Hollien, 
1982; Gilbert and Weismer, 1974; Hollien and Paul, 1969; Hollien et al., 
1982; Krook, 1988; McGlone and Hollien, 1963; Perello, 1983; Saxman 
and Burk, 1967; Stoicheff, 1981). To be specific, the mean levels for “su- 
perior” and/or “trained” male speakers studied by the earlier investi- 
gators are reported to be (roughly) 130 Hz and above; and those for similar 
women, 230 Hz and higher whereas those for the contemporary males 
and females more or less range from 110 to 125 Hz and 190 to 210 Hz, 
respectively. These contrasts are somewhat difficult to explain. True, the 
methodological approaches vary substantially and these differences may 
be substantive. Nor is it possible to determine what they meant by a 
superior speaker as even the “trained” subjects were selected on a sub- 
jective basis or identified by a simple perceptual task. Moreover, the issue 
is further complicated by what appear to be secular trends in speaking 
fundamental frequency usage. That is, when the SFF levels from the past 
40-50 years are examined, a systematic tendency toward the gradual low- 
ering of the mean (SFF) levels appears extant (see, for example, dePinto 
and Hollien, 1982). How these patterns interact with the cited differ- 
ences-or listener preferences for that matter-is difficult to explain. 
Vocal intensity (VI) is another phonatory factor that plays a role in the 
perceptual rating of voices. For example, Cooper (1973) postulates that 
there is a general negative (public) attitude toward speakers who exhibit 
relative “loud” intensities when speaking. He suggests that actors tend 
to employ this vocal characteristic when they wish to portray a character 
with a dominant and/or aggressive personality. However, very little data 
at all have been reported on the intensity patterns of speakers, vocal 
intensity (VI) preferences, or such preferences as a function of age. In- 
stead of directly attacking this issue, investigators sometimes have re- 
ported data on intensityvariation. However, it is possible that the lack 
of attention to vocal intensity may be due to the fact that this parameter 
ordinarily is varied according to speaking task. That is, it appears that 
the goal of the speaker is to employ intensity levels that are appropriate 
to the situation-at least, if he or she is capable of doing so (King and 
DiMichael, 1978; Moore, 1971). Hence, since loudness (intensity) metrics 
would change markedly from situation to situation (even when psycho- 
logical perception remained essentially the same), it is possible that in- 
vestigators have judged overall vocal intensity as a measure of limited 
usefulness. On the other hand, when listener preferences for voice fea- 
tures are sought (as in this case), speaker differences in average intensity 
can be quite meaningful-especially if the speaking conditions are con- 
trolled. 
Data from studies comparing vocal intensity with personality traits sug- 
160 H. HO,LLIEN et al. 
gest that both positive and negative characteristics may be attributed to 
those individuals exhibiting loud voices. For example, Huttar (1968) found 
that louder voices were judged to be “happier” on a happy-sad contin- 
uum, and Aronovitch (1976) reported that female subjects with higher 
average intensities were judged to be more self-confident, extroverted, 
bold, energetic, mature, and dominant. Interestingly, these relationships 
did not hold for male voices. Here, none of the correlations between 
average intensity and personality variables reached significance. On the 
other hand, Page and Balloun (1978) found that females speaking at high 
intensity levels to a male interviewer were likely to be perceived as ag- 
gressive but lacking in self-assurance. Moreover, ratings of maturity, sin- 
cerity, and desirability in a work partner were judged highest when the 
voice intensity levels were low. While it is difficult to synthesize these 
various findings, primarily owing to the confounding variables of gender 
and situation, it appears that the intensity of a voice may influence pref- 
erence judgments. Unfortunately the direction of this influence cannot be 
predicted. 
Effects of Listener Age 
Even less is known about listener preferences for different types of voices 
when the auditors’ age is considered a factor. One somewhat tangential 
perspective in this regard is provided by Huntley et al. (1987), who in- 
dicate that older listeners (in the 64-go-year range, anyway) may perceive 
voices differently than do young adults (ages 20-30 years) and middle- 
aged adults (ages 40-50 years). For one thing, differences were found 
relative to older and younger listeners’ ability to estimate the age of speak- 
ers. It also is known that the auditory capabilities of older individuals 
tend to decrease-with decrements occurring in higher auditory functions 
as well as sensitivity (see Hollien, 1987, for a review of this literature). 
Given these factors, it might be predicted that older listeners would exhibit 
different preferences for auditory stimuli in general-and voices in par- 
ticular-than do younger listeners. That a listener’s age can influence 
preferences for heard stimuli has been demonstrated (Tolhurst et al., 
1984). It also could be conjectured that, since the older listeners’ pref- 
erences undoubtedly were established some decades ago, the factors that 
influenced them could be different from those that affect the younger 
populations. Yet while certain relationships appear potentially attractive, 
they.are not robust enough to permit the structuring of a reasonably ac- 
ceptable predictive model-at least at this time. 
OBJECTIVES 
This project was designed as one response to the need for a predictive 
model relative to listener preferences for certain vocal characteristics- 
and especially to test the possibility that the preferences observed may 
LISTENING PREFERENCES 161 
be related to listener age. The specific focus in this experiment was on 
speaker type. Since voice quality could not be controlled (except for the 
absence of pathology), the variables studied were talker sex, speaking 
fundamental frequency levels, and vocal intensity characteristics. The 
obtained data were used to permit contrast of the perceptual responses 
of older subjects with those for other age groups. Please note, however, 
that only healthy listeners were used; evaluation of the preferences of 
special populations (individuals handicapped by presbycusis or hearing 
anomalies for example) were deferred for later study. 
METHOD 
Subjects 
Listeners were 80 individuals who were residents of an area including, 
or immediately adjacent to, Gainesville, Florida. They were equally di- 
vided as to sex and assigned (20 to a group) to the following age categories: 
(1) Group A, Young Adults, 20-30 years of age; (2) Group B, Middle- 
aged Adults, 40-50 years of age: (3) Group C, Older Adults, 60-70 years 
of age; and (4) Group D, the Elderly, 80-86 years of age. All subjects 
were ambulatory, alert, and of good health. To be included in the research, 
a volunteer had to pass a multiple-choice speech discrimination test (Grif- 
tiths, 1967) with a score of either 90% correct for the younger two groups 
(actual mean: 98.3%) or 80% correct for the two oldest cohorts (actual 
mean: 94.7%). Moreover, no listener was included who exhibited any 
auditory, visual, or motor pathology of a magnitude sufficient to interfere 
with easy completion of the listening tasks. Finally, the groups were 
roughly matched with respect to educational (high school and/or college) 
and socioeconomic (middle-class) backgrounds,. The biases associated 
with these variables are recognized; however, homogeneous listener 
groups were judged to be desirable at least for the initial experiment. 
Speaker Selection and Recording Procedures 
Eighty talkers (40 male, 40 female) were drawn from a pool of 136 vol- 
unteers; the individuals selected were healthy and free from any speech 
or voice disorders. Only talkers who spontaneously produced a funda- 
mental frequency in one of the desired ranges while maintaining one of 
the target intensities were accepted (see target frequency-intensity in- 
tersects on Table 1). 
Speakers were recorded reading a standard passage (The Rainbow Pas- 
sage, Fairbanks, 1960) under three different conditions: first, in a low 
intensity (60-65 dB), second, at a medium intensity (70-75 dB), and fi- 
nally, at a high intensity (SO-85 dB). A head positioner and a sound level 
162 H. HOLLIEN et al. 
Table 1. Mean Fundamental Frequency for the 
Three Talker Groups for Each Intensity Range 
Mean SFF in Hz 
Condition Men Women 
Low-frequency group” 
Target 
Actual 
Soft intensity (60-65 dB) 
Middle intensity (70-75 dB) 
Mean 
95-l 15 165-190 
95.4 181.2 
107.0 184.8 
101.2 183.0 
Mean-frequency group 
Target 
Actual 
Soft intensity (60-65 dB) 
Middle intensity (70-75 dB) 
Loud intensity (80-85 dB) 
Mean 
120-135 195-215 
124.6 199.8 
124.2 204.2 
131.8 209.8 
126.9 204.6 
High-frequency group 
Target 145-165 220-255 
Actual 
Soft intensity (60-65 dB) 164.6 232.2 
Middle intensity (70-75 dB) 149.6 230.6 
Loud intensity (SO-85 dB) 164.4 246.4 
Mean 159.5 236.4 
* The low-SFF high-VI category is not included as only three sub- 
jects could perform this task. 
meter or SLM (General Radio 1551-C) were used in regulating speakers’ 
loudness level. The SLM microphone was placed 8 inches from the mouth 
at a 45” angle and care was taken to avoid movement from the head 
positioner. Subjects monitored their vocal output on the SLM (i.e., -3 
VU; t5 dB). An experimenter prompted the subject to maintain these 
levels; talkers who could not do so were eliminated. 
All speakers were recorded in an IAC booth, using an Electrovoice 664 
microphone placed at the samedistance from the speaker’s lips as was 
the SLM microphone. This microphone was coupled to a previously cal- 
ibrated Sony TC-353C tape recorder located outside of the booth. Record 
level was held constant across all speakers, in order to maintain accurate 
intensity representation. 
The placement of talkers in each fundamental frequency category was 
determined on a post-hoc basis. First, the recorded samples were sub- 
jected to two complete speaking fundamental frequency (SFF) analyses 
by means of the IASCP Fundamental Frequency Indicator (FFI-8), a dig- 
ital readout tracking device (Hollien, 1981). This system utilizes low-pass 
LISTENING PREFERENCES 163 
filters and high-speed switching circuits interfaced with a PDP 1 l/23 com- 
puter in order to permit identification of the fundamental frequency of an 
acoustic waveform. It does not sample but rather measures each period 
within the series. 
At this point, each subject’s SFF was determined from the first analysis, 
while the second was used to determine reliability. If there was a greater 
than 1% discrepancy between the two analyses, a third analysis was car- 
ried out in order to resolve the difference. Speakers were assigned to the 
eight frequency-intensity categories listed in Table 1 on the basis of the 
combined data. If a cell contained more than the required number of 
voices, ten (five males and five females) were chosen randomly to rep- 
resent that particular combination. In all, 136 subjects had to be screened 
before 80 talkers were found who met both the frequency and intensity 
criteria for the eight cells. Finally, it should be noted that only two males 
and one of the females were able to produce the spoken passage at the 
low-frequency/high-intensity combination. Thus, that particular SFF/VI 
combination was dropped from the experiment. It is of interest to note 
that this particular observation is consistent with prior research (see Cole- 
man et al., 1977; Stout, 1938; Wolf et al., 1935). The data reported by 
these authors demonstrate that it is difficult for individuals to produce 
very soft intensity phonations at very high fundamental frequencies, in 
the modal register anyway, and that it is virtually impossible to produce 
the lowest frequencies at high intensity. 
Listeuiug Procedure 
The stimulus tape was played to the four groups of listeners; as stated, 
it contained a total of 80 voices: five males and five females in each of 
the eight SFF-VI cells. To be specific, they consisted of the second and 
third sentences of “The Rainbow Passage” presented in random order. 
Finally, as would be expected, the playback level was held constant over 
all stimuli in order to preserve the original intensity relationships. 
The equipment utilized for sample presentation consisted of a TEAC 
model A-6100 dual-channel tape recorder which fed a Marantz model 7 
amplifier. The amplifier output was controlled by a Hewlett-Packard 
model 350 step attenuator, which, in turn, was coupled to an ADS-810 
speaker. The attenuator was included as it permitted control of the output 
sound pressure level. 
The listener environment was a large sound-treated room especially 
constructed to permit perceptual research; it accommodates up to 10 
individuals at a given time. All listeners were seated facing the loud- 
speaker and none was positioned more than 75” from the axis of the 
speaker cone. Seating was adjusted to provide a 70-75dB sound pressure 
level (re: 0.0002 dyne/cm’) at the ear level of each auditor for voice sam- 
164 H. HOLLIEN et al. 
ples in the medium intensity range. These levels were calibrated before 
each experimental session by means of a General Radio 1551-C Sound 
Level Meter. Listeners were asked to indicate their preference for each 
voice by marking an equal-appearing interval scale numbered A-E (rep- 
resenting the continuum of “very much liked” to “very much disliked”). 
It was by this means that the auditors were able to indicate the strength 
of their preferences for a particular sample. The entire listening task took 
nearly 1 hour to complete. Hence, the listeners were provided a lo-minute 
break after the first half-hour so as to reduce the effects of fatigue on their 
preference ratings. 
RESULTS 
Descriptive statistics revealed mean voice ratings for all talkers-re- 
gardless of intensity level, fundamental frequency, or sex-to be 3.08 on 
a 5-point scale; the standard deviation was 1.07. Thus, the mean was about 
at the center and the overall distribution was bell-shaped. These rela- 
tionships suggest that the listeners were reasonably well able to complete 
the task. Data of a general nature also should be of interest-for example, 
information about talkers irrespective of listener’s age or sex. Accord- 
ingly, rankings of the 20 most preferred talkers (out of the total of 80) 
may be found in Table 2. As can be seen, all speaking fundamental fre- 
quency (SF) levels (low, medium, and high) were represented by this set 
of speakers. From consideration of the table, it can be noted that, while 
a few individuals producing speech at high SFF levels scored well, most 
(75%) preferred talkers who spoke at low (45%) to medium SFF levels. 
Surprisingly, middle range intensity productions appeared to be preferred 
(over low) at a better than two-to-one ratio-and not so surprising, none 
of the top-rated 20 speakers spoke at the high intensity level. Finally, the 
raw data revealed preferences for both male and female voices. As can 
be seen in Table 2, there were more males among the 20 most preferred 
voices; however, the female speakers tended to have higher scores. 
As has been stated, the primary goal of this study was to test relation- 
ships between voice preferences and listener age. Mean ratings of the four 
different age groups for the three different intensity levels and the three 
different SFF levels can be seen in Table 3. First off, it may be seen that 
the older cohorts (especially the 60-70 year olds) were more accepting 
than were the younger. This was evidenced by their generally lower pref- 
erence scores, indicating a bias toward the “very much liked” side of the 
continuum. Indeed, with only a couple of exceptions, they demonstrated 
greater approval even of the individual talker groups. On the other hand, 
while scoring all speakers higher relative to frequency preference, they 
did not discriminate as sharply (as did the control groups) among the 
talkers who used low, medium, and high pitches (about 0.38-0.58, re- 
LISTENING PREFERENCES 165 
Table 2. Characteristics of the 20 Most Preferred Talkers 
Averaged Across the Four Listener Age Groups 
Preference 
number 
1 
2 
3 
4 
5 
6 
7 
8 
9 
10 
11 
12 
13 
14 
15 
16 
17 
18 
19 
20 
Sex 
F 
F 
M 
M 
F 
M 
F 
M 
F 
M 
F 
M 
F 
F 
M 
M 
F 
M 
M 
M 
SFF Intensity Average 
Category level score” 
Low Low 1.82 
Low Mid 1.84 
High Mid 1.90 
High Mid 1.91 
High Low 1.91 
Low Low 2.02 
Middle Mid 2.14 
High Mid 2.18 
Low Mid 2.20 
Middle Mid 2.20 
Low Mid 2.24 
Low Low 2.26 
Middle Mid 2.31 
Low Mid 2.31 
High Mid 2.34 
Middle Mid 2.34 
Middle Mid 2.34 
Low Mid 2.42 
Low Low 2.44 
Middle Low 2.48 
” Lower numbers represent more favorable scores 
Table 3. Summary Table of Mean Ratings of 80 Talkers by 80 Listeners as a 
Function of Intensity and Fundamental Frequency Levels 
Listener groups” 
Speaking Young Middle Older 
conditions adult aged adult Elderly Mean 
Intensityb 
Soft 
Medium 
Loud 
Mean 
Frequency’ 
Low 
Middle 
High 
Mean 
2.78 2.91 2.89 3.04 
2.88 2.76 2.51 2.59 
4.r6 4.14 3.63 3.81 
3.27 3.27 3.01 3.15 
2.79 2.76 2.68 2.80 
3.22 3.23 2.98 3.13 
3.36 3.36 3.05 3.18 
3.12 3.12 2.96 3.04 
2.96 
2.68 
3.93 
2.76 
3.14 
3.24 
u For convenience, the values havebeen collapsed across sex (both talker and listener); nor are the 
SFFNI combinations shown. 
b The overall means do not average to 3.08 because of the weighting effect of different cell size. 
166 H. HOLLIEN et al. 
Table 4. Analysis of Variance for Ratings of Voice Samples by Listeners in 
Four Different Age Groupsa 
Source DF Sum of squares F Probability 
Model 33 
Error 6366 
Total 6399 
A. Model 
1822.33 
7332.50 
9154.83 
47.94 0.0001 
B. Factors6 
Intensity 2 1389.91 603.35 
SFF 2 218.73 94.95 
Listener age 3 57.38 16.60 
Speaker sex 1 27.69 24.04 
Intensity x age 6 86.70 12.54 
Intensity x SFF 3 21.21 6.14 
’ The dependent variable is voice rating. 
* Only main effects and interactions significant at the .Ol level or better are reported. 
0.0001 
0.0001 
0.0001 
0.0001 
0.0001 
0.0004 
spectively). The overall preference patterns for intensity appeared to be 
roughly similar to those for frequency, with differences primarily relative 
to degree. 
The results of an Analysis of Variance (ANOVA) on the voice pref- 
erence data as a function of intensity, SFF, listener age group, and speaker 
sex are presented in Table 4. As can be seen, a significant main effect 
for the variable of intensity was present. The variable of intensity ac- 
counted for 76.3% of the statistical variance-the largest proportion by 
far for all voice rating data. Results of Duncan’s Multiple Range Test, a 
post hoc procedure, showed that middle intensities were significantly pre- 
ferred over high intensities. The second significant main effect was that 
of listener age group. The variable of listener age (by group) accounted 
for an additional 3.1% of the total variance. Post hoc analyses using Dun- 
can’s Multiple Range Test underscored the previously noted relationship 
in the raw data that when voice rating scores were averaged across all 
other variables, the older adult cohort provided significantly more fa- 
vorable voice ratings (for all subjects) than did the other age groups. The 
other groups, however, were not significantly different from each other. 
That these contrasts revealed significant differences-i.e., the older 
groups exhibiting generally better ratings than the younger-may have 
been due to the greater tolerance of older listeners for various types of 
voices, or to their patience with the rating task. 
Other significant main effects in the ANOVA included both speaking 
fundamental frequency (SFF) and speaker sex. In the case of SFF, post 
hoc testing revealed that low SFF levels were significantly preferred over 
LISTENING PREFERENCES 
medium and high levels. The significant main effect for speaker sex in- 
dicates that the somewhat lower mean preference score for the male 
voices was significantly different from the mean score for female voices 
(3.01 on a S-point rating scale, vs. 3.15). Thus, male voices appeared to 
be preferred over female voices. 
Given the significant main effects for intensity and listener age group, 
it is not surprising that a significant two-way interaction between these 
variables was obtained. Application of Duncan’s Multiple Range Test 
revealed that all four age groups disliked high intensities and preferred 
the middle intensities. However, as cited above, the two younger groups 
gave significantly less favorable voice ratings to both the high and middle 
intensity voices than did the older groups. 
Finally, a second significant two-way interaction resulting from the 
ANOVA, frequency by intensity, also related most critically to all sub- 
jects; it was subjected to post hoc analysis. The results here indicated 
that the high-SFF/high-VI combinations were the least preferred by all 
age groups, with mid-SFF/high-VI ratings somewhat (but significantly) 
better. High and middle frequencies, when combined with low intensities, 
were significantly preferred over the two previously cited combinations. 
However, the most favored combinations were any of the SFF levels 
combined with the middle intensity level. Thus, frequency appeared to 
be a more important variable in determining least preferred voices than 
it did in predicting the most preferred ones. 
DISCUSSION 
The results of this study indicate that, other than being generally more 
accepting, older individuals appear to differ but little from younger sub- 
jects in their preferences for voices. Indeed, one of the more striking 
findings in this study was the homogeneity of the ratings across the four 
age groups rather than the expected differences between the older cohorts 
(the 60 and 80 year olds) and their younger controls (the 20 and 40 year 
olds). Thus, medium intensity voices in combination with any pitch level 
were almost always preferred regardless of age of the listener. Preferential 
shifts toward mid-to-low speaking frequencies also were noted but they 
were not very robust. Moreover, all age groups tended to prefer male 
voices although this trend was not very strong. For example, both the 
most preferred and second most preferred talkers were female, and nine 
out of the 20 most preferred voices were female also. 
Several hypotheses can be considered in an effort to explain these gen- 
eral patterns. First, it is possible that voice preferences are established 
in early adulthood and do not change throughout life. Indeed, it must be 
conceded that these patterns may not change even from generation to 
generation. Given these factors, it is possible that preferred vocal char- 
168 H. HOLLIEN et al. 
acteristics are not arbitrary or based on constantly changing standards, 
but rather are more systematic in their nature than has been thought. 
However, a contrasting postulate could be that preferred vocal charac- 
teristics are quite arbitrary-and are based on styles adopted by current 
personalities and/or on social stereotypes. Thus, the finding that all age 
groups had similar preferences suggests that (1) the stereotypes have re- 
mained constant, (2) all subjects have been exposed to them and (3) they 
have come to accept them. Yet a third possibility should be considered; 
it is that listeners in this research were not attending to the experimental 
variables but rather to vocal characteristics not measured in this inves- 
tigation-such as general voice quality, emotional “warmth,” intonation 
patterns and so on, and that these parameters covaried consistently with 
those measured (i.e., SFF, intensity and speaker sex). Further research 
would appear needed in order to determine if this hypothesis has merit. 
Another striking relationship observed was the importance of vocal 
intensity as a predictor of voice preference. The results of our research 
suggest that loud voices are negatively viewed. These data are not con- 
sistant with those reported by a number of authors who argue that positive 
personality traits are associated with high intensity (Huttar, 1968; Aron- 
ovitch, 1976). Indeed, the trend in our research was found to be a con- 
sistent one, regardless of the age of the listener. However, it should be 
noted that the basis for the strength of the intensity variable is not entirely 
clear. On the one hand, it may be hypothesized that, “loud” voices evoke 
negative listener attitudes because the speaker is assumed to be unac- 
ceptably aggressive (Cooper, 1973), or lacking in self-assurance (Page and 
Balloun, 1978). On the other hand, this relationship may have been in- 
fluenced simply by changes in the quality of the voice being judged. That 
is, increased intensity may have added some sort of “strident” or “harsh” 
qualities to the speakers’ voices, and listeners could have found such 
precepts objectionable. Indeed, informal listening to such samples by the 
investigators resulted in the conclusion that at least some of them exhib- 
ited voice qualities that could be classed as marginal.Without a doubt, 
the variable of voice quality is an important factor in perceptual voice 
studies; regrettably, it could not be controlled in this case. Finally, it 
appears that perceived pitch is not an especially salient factor in pref- 
erences for voices. That is, even though there was a very mild trend 
toward preferences for low-to-middle SFF, not very much variation either 
overall or between cohorts was observed. This finding is consistent with 
data from the very early studies (Murray and Tifftn, 1933, in particular), 
which suggest that, if there are no other deficiencies in speech, the use 
of high vocal frequencies is not a handicap. 
In conclusion, it appears that older individuals (and individuals of all 
ages for that matter) prefer to hear speakers producing phonations at 
medium intensity and that the fundamental frequency level and sex of the 
LISTENING PREFERENCES 169 
talker are not as important vocal parameters as is this one. Perhaps most 
important, it must be concluded-from this research anyway-that pref- 
erences for particular classes of speakers generally transcend age. 
This research was supported by a grant from the Max and Victoria Dreyfus Foundation. 
The authors also wish to thank Drs. P. A. Hollien, L. H. Leeper, and E. T. Doherty for 
their assistance; it is dedicated to the late Drs. G. C. Tolhurst and R. E. M&lone. 
REFERENCES 
Aronovitch, C. (1976). The voice of personality: Stereotyped judgments and their 
relation to voice quality and sex of speaker. .I. Sot. Psychol. 99:207-220. 
Austin, W. (1965). Some social aspects of paralinguistics. Can. J. Linguis. 11:31- 
39. 
Benjamin, B. L. (1981). Frequency variability in the aged voice. J. Gerontol. 
36:722-726. 
Black, J. W. (1942). A study of voice merit. Q. J. Speech 28:67-74. 
Coleman, R. F., Mabis, J. H., and Hinson, J. (1977). Fundamental frequency 
sound pressure level profiles of adult male and female voices. J. Speech Hear. 
Res. 20: 197-204. 
Cooper, M. (ed.). (1973). Modern Techniques of Vocal Rehabilitation. Spring- 
field, IL: Charles C. Thomas. 
Cowan, M. (1936). Pitch and intensity characteristics of stage speech. Arch. 
Speech (Suppl. 1). 
dePinto, O., and Hollien, H. (1982). Speaking fundamental frequency character- 
istics of Australian women: Then and now. J Phonet. 10:367-376. 
Fairbanks, G. (1960). Voice and Articulation Drillbook (2nd ed.). New York: 
Harper and Brothers. 
Fitch, J. L., and Holbrook, A. (1970). Modal fundamental frequency of young 
adults. Arch. Otolatyngol. 92:379-382. 
Gilbert, H. R., and Weismer, G. G. (1974). The effects of smoking on speaking 
fundamental frequency of adult women. J Psychol. Res. 3:225-23 1. 
Griffhhs, J. (1967). Rhyming minimal contrasts: A simplified diagnostic articu- 
lation test. J. Acoust. Sot. Am. 42:236-240. 
Hollien, H. (1981). Analog instrumentation for acoustic speech analysis. In J. 
Darby (ed.), Speech Evaluation in Psychiatry, pp. 79-103. New York: Grune 
and Stratton. 
Hollien, H. (1987). Old voices: What do we really know about them? J. Voice, 
1:2-17. 
Hollien, H., and Jackson, B. (1973). Normative data on the speaking fundamental 
frequency characteristics of young adult males. J. Phonet. 1: 117-120. 
Hollien, H., and Paul, P. A. (1969). A second evaluation of speaking fundamental 
frequency characteristics in post adolescent girls. Lang. Speech 12: 119- 124. 
170 H. HOLLIEN et al. 
Hollien, H., and Shipp, T. (1972). Speaking fundamental frequency and chron- 
ological age in males. J. Speech Hear. Res. 15:155-159. 
Hollien, H., Tolhurst, G. C., and McGlone, R. E. (1982). Speaking Fundamental 
Frequency as a Function of Vocal Intensity. Progress Report to the Dreyfus 
Foundation, pp. 23. 
Hudson, A., and Holbrook, A. (1981). A study of the reading fundamental vocal 
frequency of young black adults. J. Speech Hear. Res. 24~197-201. 
Hudson, A., and Holbrook, A. (1982). Fundamental frequency characteristics of 
young black adults: Spontaneous speaking and oral reading. J. Speech Hear. 
Res. 25125-28. 
Huntley, R., Hollien, H., and Shipp, T. (1987). Influences of listener character- 
istics on perceived age estimations. J. Voice 1:49-52. 
Huttar, G. (1968). Relations between prosodic variables and emotions in normal 
American English utterances. J. Speech Hear. Res. 11:481-487. 
King, R. G., and DiMichael, E. M. (1978). Articulation and Voice: Improving 
Oral Communication. New York: Macmillan Publishing. 
Krook, M. I. P. (1988). Speaking fundamental frequency analysis of normal Swed- 
ish subjects obtained by glottal frequency analysis. Folia Phoniutr. 40:82-90. 
Lewis, D., and Tifftn, J. (1934). A psychophysical study of individual differences 
in speaking ability. Arch. Speech 1:43-60. 
Linke, C. E. (1973). A study of pitch characteristics of female voices and their 
relationship to vocal effectiveness. Folia Phoniatr. 25:175-185. 
Lynch, G. (1934). A phonophotographic study of trained and untrained voices 
reading factual and dramatic material. Arch. Speech 1:9-25. 
McGlone, R. E., and Hollien, H. (1963). Vocal pitch characteristics of aged 
women. J. Speech Hear. Res. 6:164-170. 
Moore, G. P. (1971). Organic Voice Disorders. Englewood Cliffs, NJ: Prentice 
Hall. 
Murray, E., and Tiftin, J. (1933). An analysis of some basic aspects of effective 
speech. Arch. Speech 1:61-83. 
Mysak, E. D. (1959). Pitch and duration characteristics of older males. J. Speech 
Hear. Res. 2~46-54. 
Perello, J. (1983). Characteristics of the elderly voice. Bull. Audiol. Phonol. 
16: 103-l 14. 
Page, R., and Balloun, J. (1978). The effect of voice volume on the perception 
of personality. J. Sot. Psychol. 105:65-72. 
Pronovost, W. (1942). An experimental study of methods for determining natural 
and habitual pitch. Speech Monogr. 9: 11-123. 
Saxman, J., and Burk, K. (1967). Speaking fundamental frequency characteristics 
of middle-aged females. Folk Phoniatr. 19:167-172. 
Snidecor, J. C. (1951). Studies in the pitch and duration characteristics of superior 
speakers. J. Speech Hear. Disord. 1644-52. 
LISTENING PREFERENCES 171 
Stoicheff, M. (1981). Speaking fundamental frequency characteristics of non- 
smoking female adults. J. Speech Hear. Res. 241437-441. 
Stout, B. (1938). The harmonic structure of vowels in singing in relation to pitch 
and intensity. J. Acoust. Sot. Am. 10:137-146. 
Tolhurst, G. C., Hollien, H., and Leeper, L. (1984). Listening preferences for 
music as a function of age. Folia Phoniatr. 36:93-100. 
Wolf, S., Stanley, D., and Sette, W. (1935). Quantitative studies on the singing 
voice. J. Acoust. Sot. Am. 6:255-266.