THE EELANQRSHS? 0F ACCUSTECAL
ENVERQNMENT T0 “ii-{E EVALUATION
5F PERCEWED LOUDNESS
OF CCflNKTED DESCOURSE

Thesis Ear {'im Degree of D“. D.
MICHLGAN STATE UNEVERSITY

Patricia Stump Waksh
1966

'j’qut;

LIBRARY

Michigan State

University

 

This is to certify that the

thesis entitled
THE RELATIONSHIP OF ACOUSTICAL ENVIRONMENT TO THE

EVALUATION OF PERCEIVED LOUDNESS OF CONNECTED DISCOURSE

presented by

Pat Walsh (Patricia Stump Walsh)

has been accepted towards fulfillment
of the requirements for

Ph .D . degree in S ECECh

 

Date W355—

0-169

ABSTRACT
THE RELATIONSHIP OF ACOUSTICAL ENVIRONMENT
TO THE EVALUATION OF PERCEIVED LOUDNESS
OF CONNECTED DISCOURSE

by Patricia Stump Walsh

Researchers have realized that the judgment of loud—
ness of various types of acoustic stimuli is "subjective"
and must be measured "indirectly" by observing the effects
of specified stimuli upon the behavior of a group of lis-
teners. Whereas many aspects of loudness have been inves—
tigated, few studies involving subjects' evaluations of
the loudness of stimuli, in terms of listener comfort, have
been carried out. Furthermore, even fewer studies have
investigated the effects of various factors that might
influence these evaluations.

The purpose of this study was to investigate the
effects of intensity level of connected discourse and
acoustical background condition upon listeners' evaluations
of the loudness of connected discourse. The psychophys-
ical method of single stimuli (rating scale method) was
used. Connected discourse was reproduced on low noise
magnetic tape at 13 different intensity levels (45 to 105
dB SPL,lJ15 dB steps) in quiet and with 70 dB SPL wide
band white noise, narrow band white noise, and speech bab-

ble. There were 52 possible combinations of intensity

Patricia Stump Walsh

level of connected discourse and background condition. Two
seven-second segments of each stimulus were presented to the
subjects, making a total of 104 stimuli. The stimuli were
spliced together in a random manner for presentation to

the subjects.

Twenty-one normal hearing subjects were tested in
two groups, each subject listening to all of the stimuli.
For each stimulus he evaluated the connected discourse as
being too soft, at his lower limit of comfortable loudness,
comfortable, at his upper limit of comfortable loudness,
or too loud, using a rating scale of l to 5.

The data for analysis consisted of the means of
each subject's responses for the two presentations of a
particular combination of intensity level of connected dis-
course and background condition.

The data were subjected to an analysis of variance,
the results of which indicated that intensity level of con—
nected discourse and background condition affected a subject's
evaluation of the loudness of connected discourse and that
there was a significant interaction between these two factors.
A critical difference test was performed in order to deter-
mine where the differences lay among subject's mean evalua-
tions of the loudness of connected discourse presented at 13
intensity levels in quiet or with 70 dB SPL wide band white
noise, narrow band white noise, and speech babble. It was
found that differences existed among most of the treatment

combinations.

Patricia Stump Walsh

On the basis of the results, the following conclu—
sions were made:

1. The perceived loudness of connected discourse
increases as the intensity level of the connected discourse
increases, irrespective of the type of acoustical background
condition.

2. The presence of background noise influences
listeners' evaluations of the perceived loudness of con-
nected discourse until the intensity level of the speech
is increased beyond the level of the noise. When the in-
tensity level of connected discourse is increased beyond
the level of the noise, the four background conditions be-
gin to produce quite similar effects on listeners' evalu-
ations of the loudness of connected discourse.

3. The range of comfortable loudness for speech
heard in quiet is 65 to 90 dB SPL. The ranges of comfort-
able listening for speech heard in 70 dB SPL wide band white
noise, narrow band white noise, and speech babble are 45
to 95 dB SPL, 80 to 90 dB SPL, and 75 to 90 dB SPL, respec-
tively.

4. The magnitude of the range of intensity levels
at which connected discourse was presented is not the same
for the three regions of comfort (too soft, comfortable,
and too loud).

Recommendations for further research were made on

the basis of these findings.

THE RELATIONSHIP OF ACOUSTICAL ENVIRONMENT
TO THE EVALUATION OF PERCEIVED LOUDNESS

OF CONNECTED DISCOURSE

By

Patricia Stump Walsh

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Speech

1966

Copyright by
Patricia Stump Walsh
December, 1357

TABLE OF CONTENTS

LIST OF TABLES O O O O O O O O O I O O O O O O O O 0

LIST OF ILLUSTRATIONS O O O O O O O O O O O O O O 0

CHAPTER

I.

II.

III.

IV.

INTRODUCTION 0 O O O O O O O O O O O O O O 0

Purpose of the Study . . . . . . . . . .
Importance of the Study. . . . . . . . .
Definitions. . . . . . . . . . . . . . .
Organization of the Report . . . . . . .

REVIEW OF BACKGROUND LITERATURE. . . . . . .

Comfort Levels for Pure Tones and Speech in

Quiet. . . . . . . . . . . . . . . . . .
Comfort Levels for Pure Tones in Noise . .
Range of Comfortable Loudness Levels for

Pure Tones and Speech in Quiet and in

Noise. . . . . . . . . . . . . . . . . .
Psychophysical Methods Used to Determine

Comfort Levels for Pure Tones and Speech
Reliability of Comfortable Loudness Judg-

ments. . . . . . . . . . . . . . . . . .
Factors Related to Comfortable Loudness

Judgments of Pure Tones and Speech . . .
Clinical and Experimental Applications of

Comfortable Loudness Judgments . . . . .
Summary. . . . . . . . . . . . . . . . . .

SUBJECTS, MATERIAL, INSTRUMENTATION, AND

PROCEDURES 0 O O O O O O I O O O O O O O 0

Subjects . . . .
Material . . . .
Instrumentation.
Procedures . . .

RESULTS AND DISCUSSION 0 O O O O O O O O O O

ReSUltS. O O O O O O O O O O O O O O O O 0
Discussion . . . . . . . . . . . . . . . .

iii

Page

vi

[-1

bqbw

2O
39
41
49
61
78
98
105
107
107
108
108
109
124

124
129

CHAPTER

V. SUMMARY AND CONCLUSIONS. . . . . . . . . . . .

BIBLIOGRAPHY.

APPENDIX

APPENDIX

APPENDIX

.APPENDIX

.APPENDIX

TABLE OF CONTENTS (continued)

sumary C O O O O O O O O O O O O O O O O O 0
Conclusions. . . . . . . . . . . . . . . . .
Implications for Further Research. . . . . .

A.

B.

C.

DESCRIPTION OF THE STIMULUS TAPE . . . .

SUBJECT'S RESPONSE SHEET . . . . . . . .

FIFTY-TWO AVERAGED EVALUATIONS OF THE
LOUDNESS OF CONNECTED DISCOURSE FOR
EACH SUBJECT, TOTALS, AND MEANS. . . .

CRITICAL DIFFERENCES AMONG THE MEANS OF
CELLS O O O O O O C O O C O O C O O O O

CRITICAL DIFFERENCES AMONG THE RANGES
OF CELLS O O O O O O O O O O O O O O 0

iv

Page
153
153
155
156
159
165

172

174

179

182

LIST OF TABLES
Table Page

1. Median and Geometric Mean Values of the Pooled
Decibel Scores for Each of the Comfortable-
Conversational Listening Levels . . . . . . . 37

2. Median and Geometric Mean Values of the Pooled
Decibel Scores for Each of the Comfortable-
Conversational Listening Levels . . . . . . . 47

3. Mean Scores for Replications for 20 Normal
Hearing Subjects. . . . . . . . . . . . . . . 75

4. Test and Retest Scores for Psychophysical
Methods, Judgmental Procedures, and Earphone
and Loudspeaker Presentations . . . . . . . . 76

5. Mean Evaluations of the Loudness of Connected
Discourse Presented at 13 Intensity Levels
and under Four Background Conditions. . . . . 126

6. Summary Table for the Factorial Design Analysis
of Variance . . . . . . . . . . . . . . . . . 127

7. Results of Critical Difference Tests for Each
Intensity Level at which Connected Discourse
Was Presented . . . . . . . . . . . . . . . . 140

8. Summary Table Showing in dB SPL the Intensity
Levels for the Five Regions of Comfort and
the Four Background Conditions. . . . . . . . 147

9. Summary Table of Analysis of Variance . . . . . 148

LIST OF ILLUSTRATIONS
Page

The Variability among Observers of Comfort-
able Listening Levels as a Function of
Frequency. 0 O O I O O O O O 0 O O O O O O O 73

Block Diagram of the Instrumentation Employed
to Measure the Sound Pressure Level of the
Connected Discourse and to Reproduce the
Connected Discourse on the Stimulus Tape . . 114

Block Diagram of the Instrumentation Employed
to Measure the Sound Pressure Level of the
Wide Band White Noise and to Reproduce the
Wide Band White Noise on the Stimulus Tape . 116

Block Diagram of the Instrumentation Employed
to Measure the Sound Pressure Level of the
Narrow Band White Noise and to Reproduce
the Narrow Band White Noise on the Stimulus
Tape . . . . . . . . . . . . . . . . . . . . 117

Block Diagram of the Instrumentation Employed
to Measure the Sound Pressure Level of the
Speech Babble and to Reproduce the Speech
Babble on the Stimulus Tape. . . . . . . . . 118

Block Diagram of the Instrumentation Employed
to Present the Stimuli to the Subjects . . . 122

Subjects' Mean Evaluations of the Loudness of
Connected Discourse. . . . . . . . . . . . . 128

vi

CHAPTER I

INTRODUCTION

Determining the loudness of various types of acous-

 

tic stimuli has challenged researchers for many years.
Whereas many aspects of loudness have been investigated,
few studies involving subjects' evaluations of the loudness
of stimuli, in terms of listener comfort, have been carried
out. Furthermore, even fewer studies have investigated
the effects of various factors that might influence these
evaluations.

Recently the area of loudness has received national
attention, both as a topic for research in acoustics and
in relation to radio and television commercials. (1) In
the program of a recent meeting of the Acoustical Society
of America, one of the sessions concerned with psycholog-
ical and physiological acoustics was subtitled "Mainly Loud-
ness."1 (2) After a two-year investigation of hundreds
of complaints about loud commercials (presumably designed
to reach potential consumers who leave the room during the
advertisements), the Federal Communications Commission is-

sued a policy statement on July 12, 1965, objecting to

 

1The Journal of the Acoustical Society of America,
XXXVII, No. 6 (June, 1965), p. 1174.

”commercials delivered in a 'loud, rapid and strident man-
ner' and loud commercials sandwiched between soft music
or speech."2
The investigation of loudness and other psycholog-

ical aspects of audition belongs to the area of psychoacous-
tics, since there does not yet exist a “loudness meter”
which indicates the effects of a specified acoustic stim-
ulus upon a group of listeners. It is true that

the acoustic engineer can calculate approximately

from the band spectrum levels of a noise how it

will sound in relation to a standard tone of 1000

cps at a sound level of 40 db, but the rules are

rather complicated and have not been fully stand—

ardized. He does not express the answer in deci-

bels but in a subjective unit of loudness called

the some, i.e., the loudness of a 1000 cps tone

at 40 db. But in everyday life we are not accus-

tomed to thinking in sones.

Therefore, the loudness of acoustic stimuli must be deter-

mined by noting the effect of such stimuli upon the listen—
4
er.
Researchers have long realized that loudness is a

“subjective“ phenomenon and that it must be measured

 

2Fred P. Graham, “Ads on TV Found Too Loud by F.C.C.:
Broadcasters Are Warned against Strident Delivery," The New
York Times, CXIV, No. 39, 252 (July 13, 1965), p. 67.

3Hallowell Davis, "Physics and Psychology of Hear-
ing,” Hearing and Deafness, ed. Hallowell Davis and S.
Richard Silverman (2d ed. rev.; New York: Holt, Rinehart
and Winston, Inc., 1960), pp. 54-55.

4Stanley Smith Stevens and Hallowell Davis, Hear—
ing: Its Psychology and Physiology (New York: John Wiley
& Sons, Inc., 1938), p. 4.

”indirectly“ by noting listeners' reactions to specified
acoustic stimuli. Assuming, then, that the loudness of
acoustic stimuli can be evaluated and measured, what are
some of the factors that might affect listeners' evaluations
of the loudness of various auditory stimuli, including con-

nected speech?

Purpose of the Study,

The present study deals with two variables that
may be related to listeners' evaluations of the loudness
of connected speech relative to comfort. Specifically,
the effects of 13 different intensity levels of connectv
ed discourse and four different acoustical background
conditions upon normal hearing listeners' evaluations of
the loudness of connected discourse are investigated.

At the beginning of the experiment, the following
questions were asked:

1. Does the intensity level of connected discourse affect
a person's evaluation of the loudness of connected dis-

course?

2. Does the acoustical background condition affect a per—
son's evaluation of the loudness of connected discourse?

3. Is there any interrelationship between the two afore-
mentioned factors that might affect the evaluation of
the loudness of connected discourse?

In an attempt to answer these questions, the fol-
lowing null hypotheses were formulated for testing in this

study:

1. There are no significant differences in a subject's
evaluation of the loudness of connected discourse

presented at 45, 50, 55, 60, 65, 70, 75, 80, 85, 90,
95, 100, and 105 dB SPL.

2. There are no significant differences in a subject's
evaluation of the loudness of connected discourse pre—
sented under four different background conditions:
in quiet (with no noise other than that inherent in
the system), and with 70 dB SPL wide band white noise,
narrow band white noise, and speech babble.

3. There is no interaction between intensity level of con-

nected discourse and background condition in a subject's
evaluation of the loudness of connected discourse. '

Importance of the Study

 

There have been comparatively few studies made that
involve subjects' evaluations of the loudness of acoustic
stimuli in terms of listener comfort, as was mentioned above.
Likewise, few studies have investigated the effects of vari-
ous factors that might influence listeners' evaluations of
auditory stimuli in terms of comfortableness. No previous
research has dealt with the effects of various intensity
'levels and background conditions upon normal hearing sub-
jects' evaluations of the loudness of speech, which is the
focus of this study. Likewise, no previous study has em—
Ployed the psychophysical method of single stimuli (rating
Scale method), which is utilized in the present research,
as a means of evaluating the loudness of speech in terms
of comfortableness.

Despite the fact that little is known about the
effects of certain factors upon normal hearing persons'
evaluations of the loudness of speech, individuals with

various types and degrees of hearing loss have been

determining their “most comfortable listening levels" in
audiology clinics for at least twenty years. The determina-
tion of a most comfortable listening level for pure tones
or speech is frequently accomplished in order to provide
information regarding the presence or absence of recruit-
ment or to assist in the selection of a hearing aid.5 Ac-
cording to Newby, the most comfortable listening level pro-
vides information concerning the level of amplification

at which the hard-of-hearing person will hear most comfort-
ably6--the level that corresponds to the comfortable lis-
tening level at which persons with normal hearing are ac-
customed to receiving sounds.

Therefore, one of the objectives of a hearing aid
evaluation is to fit the patient with an aid that, insofar
as is possible, will enable him to receive sounds at a level
which is comfortable for him. Presumably one of the reasons
that a hard-of—hearing person who might benefit from ampli-
fication either refuses to wear a hearing aid or discontin-
ues its use is that the aid fails to deliver acoustic stim-
uli at a level that is comfortable for him, as he goes about

his daily routine.

Each day persons with normal hearing and those with

 

5Otto J. Menzel, "The 'Comfort' Aspects of Loud—

ness," The Eye, Ear, Nose and Throat Monthly, XLV, No. 2
(February, 1966), p. 106.

6Hayes A. Newby, Audiology (2d ed. rev.; New York:
Appleton-Century-Crofts, Inc., 1964), p. 132.

various types and degrees of hearing impairment are forced
to attend to acoustic stimuli (mostly speech) in a variety
of situations, with varying amounts of background noise
present. To date the listening conditions of daily life
can be simulated only roughly in the laboratory. It is
conceivable that the results from a properly designed se-
ries of studies may yield sufficient information so that
various types of listening situations may be reproduced
accurately in the laboratory and in the audiology clinic.
Such an achievement would enable researchers to obtain in-
formation regarding the method used by persons with normal
hearing and those with impaired hearing to evaluate the
loudness of various acoustic stimuli in different listen—
ing situations and, also, would enable audiologists to pre-
dict better the successful use of a hearing aid in patients'
everyday lives.

Although such an outcome is not within the limits
of the present study, it is hoped that, if various inten-
sity levels of connected discourse and background conditions
do affect normal hearing listeners' evaluations of the loud-
ness of connected discourse, future studies will be carried
out that will increase the usefulness and efficiency of
the hearing aid evaluation. For example, additional studies
could be planned that would utilize a variety of listening
situations and persons with various types and degrees of

hearing loss.

Harris recently remarked that "the audiologist has
a definite responsibility along with the sensory psycholo-
gist; since he will be seeking departure from norms, he
should contribute to those norms."7 It is hoped that this

study may make a contribution in this direction.

Definitions

 

Several terms appear in the literature dealing with
comfortable loudness. These terms and their definitions
follow.

Loudness. American Standard Acoustical Terminology

 

defines loudness as " . . . the intensive attribute of an
auditory sensation, in terms of which sounds may be ordered
on a scale extending from soft to loud."8 Although loud-
ness depends mainly upon the sound pressure of the stimu-
lus, it also depends upon the frequency and wave form of
the stimulus.9

Comfortable Loudness. Four terms have been applied

 

to the concept of comfortable loudness in the literature:
comfortable loudness, most comfortable loudness, most com-

fortable listening level, and comfort level. The distinction

 

7J. Donald Harris, “Research Frontiers in Audiol-
ogy," Modern Developments in Audiology, ed. James Jerger
(New York: Academic Press, 1963), p. 426.

8American Standard Acoustical Terminology (Includ-
ing Mechanical Shock and Vibration) (New York: American
Standards Association, Incorporated, New York, 1960), p.
45.

91bid.

among these terms is not clear, and various writers have
used them interchangeably, although the last three terms
seem to imply a certain level rather than a range of com—
fortable loudness levels.

In 1961 Lezak formulated a rather comprehensive
definition of comfortable loudness:

In general terms comfortable loudness is the

level at which the listener prefers to hear sounds
of interest. The level may be different for dif-
ferent situations such as background dinner music
at a rather low level or concert music at a rather
loud level to appreciate better the many nuances
present. Also, comfortable loudness is related

to the listening condition. That is, comfortably
loud speech in a noisy factory is at a different
intensity level than in a quiet home. As used for
hearing of speech, comfortable loudness includes
the concept of being loud enough for easy intelli-
gibility but not so loud as to be annoying.

Lezak's definition of "comfortable loudness" has
been adopted for the purposes of the present study and will
be used in its generic sense. The term "most comfortable
listening (loudness) level" (MCL level) or "comfort level”
will refer to a specific level of comfortable loudness,
while ”range of comfortable listening (loudness)" (RCL)
will refer to the range of intensity levels which listeners
evaluate as being neither "too soft“ nor "too loud” but

"comfortable.“

In this study "comfortable loudness" is defined

 

10Raymond Joseph Lezak, ”Some Aspects of Comfort-
able Loudness" (unpublished Ed.D. dissertation, Department
of Speech and Theater, The Pennsylvania State University,
1961), p. l.

operationally as a rating above 2.0 and below 4.0 on the
five-point rating scale used by the subjects in evaluating
the loudness of connected discourse.ll

Lower Limit of Comfortable Loudness (LLCL). Using
pure tones as stimuli, Pollack determined the range of com-
fortable listening in quiet and in noise for normal hear-
ing listeners.12 Each subject controlled the intensity
level of the stimulus and was instructed to ascertain his
most comfortable listening level and his upper and lower
limits of the range of levels that he considered comfort-
able. Pollack defined the lower limit of the range of com-
fortable loudness as " . . . that level at which it is nec-
essary to 'strain' in order to hear the signal clearly. . . ."l3
The results of this study yielded mean lower limits of the
range of comfortable listening levels that were consider-
ably above the normal auditory threshold for pure tones.14
[The lower limit of the range of comfortable loudness should

not be confused with the threshold of auditory acuity.)

For the purposes of this study, the lower limit

 

11This rating scale will be described more fully

in Chapter III..

12Irwin Pollack, “Comfortable Listening Levels for
Pure Tones in Quiet and Noise," The Journal of the Acous-
tical Society of America, XXIV, No. 2 (March, 1952), pp.
158‘62 0

13Ibid., p. 161.

l4Ibid., p. 160.

10

of comfortable loudness is defined as being comfortable
to listen to, without the listener wanting the stimulus
(speech) to be any softer. In the present study the LLCL
is defined operationally as a rating of 2.0 on the five-
point rating scale used by the subjects in evaluating the
loudness of connected discourse.

Upper Limit of Comfortable Loudness (ULCL). In
the same study cited above, Pollack defined the upper limit
of the range of comfortable listening levels as ” . . . that
level at which the sound appeared 'annoying' or 'bother-

some."'15 The results of this study yielded mean upper

limits of the range of comfortable listening levels16 that
were considerably below the thresholds of discomfort, feel-
ing, tickle, or pain reported by other researchers. [The
upper limit of the range of comfortable loudness should
not be confused with the above measures.] Lezak used "up-
per limit of comfortable loudness" in reference to a level
5 dB below a subject's uncomfortable loudness threshold.l7
In the present study, however, the upper limit of
comfortable loudness is defined as being comfortable to

listen to, without the listener wanting the stimulus

(speech) to be any louder. Here the ULCL is defined

 

lsIbid., p. 161.

l6Ibid., p. 160.

l7Lezak, op. cit., p. 41.

11

operationally as a rating of 4.0 on the five-point rating
scale used by the subjects in evaluating the loudness of
connected discourse.

Acoustical Background Conditions. The four differ—
ent acoustical background conditions utilized in this study
involved connected discourse presented in quiet (i.e., with
no noise present other than that inherent in the system),
and with 70 dB SPL wide band white noise, narrow band white
noise, and speech babble. Wide band white noise, narrow
band white noise, and speech babble will be described in
Chapter III.

Sound. This term may be defined as an oscillation
in pressure, stress, particle displacement, particle veloc-
ity, etc., in a medium or as an auditory sensation evoked
by such an oscillation.18 It should be noted, however,
that not all sound waves can evoke an auditory sensation
(e.g., ultrasound).19

Intensity Level. According to American Standard
Acoustical Terminology,

the intensity level, in decibels, of a sound is
10 times the logarithm to the base 10 of the ratio

of the intensity of this sound to the reference

intensity. The reference intensity shall be stated
explicitly.20

 

18American Standard Acoustical Terminology, op,

cit., p. 9.

lgIbid.

ZOIbido, pp. 14‘150

12

It should be noted that a common reference sound intensity

is 10-16 watt per square centimeter in a specified direc-
tion and that in a free progressive plane or spherical wave
there is a known relation between sound intensity and sound
pressure, so that sound intensity level can be deduced from
a measurement of sound pressure level.21

Sound Pressure Level. In the American Standard

 

Acoustical Terminology sound pressure level, in decibels,
is defined as ” . . . 20 times the logarithm to the base

10 of the ratio of the pressure of this sound to the ref-

4

. 2 2 - O
erence pressure.” The reference pressure 2 X 10 micro-

bar is-generally used for measurements concerned with hear-
ing and with sound in air and liquids.23 According to Hirsh,

the sound pressure of 0.0002 dyne/cm2 corresponds to the

-16

sound intensity, in air, of 10 watt/cmz; and this sound

pressure is often used for expressing Sound Pressure Level

in dB.24 In this study the intensity level of connected

discourse, and of wide band white noise, narrow band white
noise, and speech babble, is expressed in dB SPL.

Sensation Level. This term refers to " . . . the
pressure level of the sound in decibels above its threshold

211bid., p. 15.

22Ibid., p. 14.

23Ibid.

24Ira J. Hirsh, The Measurement of.Hearing (New
Ybrk: McGraw-Hill Book Company, Inc., 1952), p. 59.

 

1 I

l.

13

of audibility for the individual observer or for a speci-

fied group of individuals."25 Sensation level does not

provide any direct information concerning the physical in-
tensity of the stimulus.26

Loudness Level. According to American Standard

 

Acoustical Terminology,

the loudness level of a sound, in phons, is numer-
ically equal to the median sound pressure level,

in decibels relative to 0.0002 microbar, of a free
progressive wave of frequency 1000 cycles per sec-
ond presented to listeners facing the source, which
in a number of trials is judged by the listeners

to be equally loud.27

Perhaps this idea is stated more clearly by Stevens and

Davis:

Loudness-level for a given tone is defined as
the intensity level of a 1000—cycle tone which
sounds equal in loudness to the given tone. For
such a tone of 1000 cycles, called the reference-
tone, intensity-level and loudness-level are equiv-
alent.28

Phon. The phon is the unit of loudness level, as
specified in the above definition.29

Sone. The some is defined as a unit of loudness;

 

25American Standard Acoustical Terminology, 22¢
Cit}, p. 450
26 -
Newby, op. c1to, p. 12°

27American Standard Acoustical Terminology, loc.

n
P.
0

28Stevens and Davis, op. cit., p. 111.

29American Standard Acoustical Terminology, loc.

n
l—l.
d-
O

l4

and, by definition, a 1000 cps tone, 40 decibels above a
listener's threshold, produces a loudness of one sone.3O
If a listener judges the loudness of a sound to be 3 times

that of the one-sone tone, the loudness of that sound is

2'sones.

Organization of the Report

Chapter I has introduced the topic of loudness and
has discussed this aspect of auditory sensation with respect
to clinical audiology. The problem to be studied in this
paper was presented, viz., the effects of 13 intensity
levels of connected discourse and four acoustical back-
ground conditions upon subjects' evaluations (in terms of
comfortableness) of the loudness of connected discourse.
Several terms encountered in the study of loudness were
defined and discussed.

Chapter II consists of a comprehensive overview
of the literature related to listeners' evaluations of the
loudness of auditory stimuli in terms of comfortableness.
The following areas will be discussed: (1) comfort levels
for pure tones and speech in quiet, (2) comfort levels for
pure tones in noise, (3) the range of comfortable listening
levels for pure tones and speech in quiet and noise, (4)
psychophysical methods used to determine comfort levels for

pure tones and speech, (5) the reliability of comfortable

 

30Ibid.

15

loudness judgments, (6) factors related to comfortable
loudness judgments of pure tones and speech, and (7) the
use of comfortable loudness judgments for clinical and
research purposes.

Chapter III presents the experimental procedures
related to this study, including the criteria for selection
of subjects, the preparation of stimulus materials, and
the experimental apparatus.

Chapter IV is concerned with the results of the
statistical analyses. The results of this study are dis-
cussed with regard to the hypotheses set forth in Chapter
I. The findings of this study are also related to previous
research.

Chapter V presents a summary of the present study.
Conclusions are drawn on the basis of the analysis, and

recommendations for further research are made.

CHAPTER II

REVIEW OF BACKGROUND LITERATURE

Over the years many of the studies concerned with

audition above threshold have involved loudness. Some re-

 

searchers have addressed themselves to the definition of

31,32

this subjective aspect of acoustic stimuli; others have

attempted to calculate the loudness of various types of aud—

itory stimuli by means of mathematical formulas.33’34’35’36’37

 

31Harvey Fletcher, “Loudness, Masking and their

Relation to the Hearing Process and the Problem of Noise
Measurement," The Journal of the Acoustical Society of
America, IX, No. 4 (April, 1938), pp. 275-93.

32Harvey Fletcher and W. A. Munson, "Loudness, its
Definition, Measurement and Calculation," The Journal of
the Acoustical Society of America, V, No. 2 (October, 1933),
PP- 82-108.

331bid.

 

34H. P. Knauss, "An Empirical Formula for the Loud-

ness of a 1000-cycle Tone,” The Journal 9f the Acoustical
Society of America, IX, No. 1 (July, 1937), pp. 45—46.

35S. S. Stevens, "Calculation of the Loudness of
Complex Noise," The Journal of the Acoustical Society of
America, XXVIII, No. 5 (September, 1956), pp. 807-32.

363. S. Stevens, "Procedure for Calculating Loud-
ness: Mark IV,” The Journal of the Acoustical Society of
America, XXXIII, No. 11 (November, 1961), pp. 1577—85.

37W. A. Munson, “The Loudness of Sounds,“ Handbook
of Noise Control, ed. Cyril M. Harris (New York: McGraw-
Hill Book Company, Inc., 1957), pp. 1-22.

 

16

17

Still other investigators have employed psychOphysical tech—

niques for determining the loudness of sounds.38’39’40’41’42’43

Various experimenters have determined the loudness of spe-

44,45,46,47

cific kinds of acoustic stimuli. The investigation

 

385. S. Stevens, “A Scale for the Measurement of
a Psychological Magnitude: Loudness," Psyghological Review,
XLIII, No. 5 (September, 1936), pp. 405-16.

39E. B. Newman, J. Volkmann,and S. S. Stevens, "On
the Method of Bisection and its Relation to a Loudness Scale,"
American Journal of Ppychology, XLIX, No. 1 (January, 1937),
pp. 134-37.

408. S. Stevens, "The Direct Estimation of Sensory
Magnitudes--Loudness," American Journal of Psyghology, LXIX,
No. 1 (March, 1956), pp. 1-25.

41Lloyd B. Ham, Frank Biggs, and Everett H. Cathey,
Jr., ”Fractional and Multiple Judgments of Loudness,” 222.
Journal of the Acoustical Society_of America, XXXIV, No. 8
(August, 1962): pp. 1118-21.

42Bruce Schneider and Harlan Lane, "Ratio Scales,
Category Scales, and Variability in the Production of Loud—
ness and Softness,” The Journal of the Acoustical Society
of America, XXXV, Part 2, No. 12 (December, 1963), pp. 1953-
610

 

43Edith L. R. Corliss and George E. Winzer, "Study
of Methods for Estimating Loudness," The Journal of the
Acoustical Society of America, XXXVII, No. 6 (June, 1965),
p. 1174 (Abstract).

44D. E. Baier, "The Loudness of Complex Sounds,"
Journal of Experimental Psychology, XIX, No. 3 (June, 1936),
pp. 280-308. ~

4SDavis H. Howes, "The Loudness of Multicomponent
Tones,” American Journal of Psychology, LXIII, No. l (Janu-
ary, 1950), pp. 1-30.

46Irwin Pollack, "Studies in the Loudness of Com-
plex Sounds“ (unpublished Ph.D. dissertation, Department
of Psychology, Harvard University, 1948).

47Bertram Scharf, "Critical Bands and the Loudness
of Complex Sounds Near Threshold" (unpublished Ph.D. disser-
tation, Department of Psychology, Harvard University, 1958).

18

of the effect of certain factors upon loudness determinations

has attracted the attention of other researchers.48’4‘9’50’5]"52’S3

Fletcher and Munson, among others, have determined the inten-
sity levels at which pure tones of different frequencies
are judged as equal in loudness to a 1000 cps reference tone

54,55,56,57

and have plotted "equal-loudness contours." More

 

48E. K. Chapin and F. A. Firestone, "The Influence

of Phase on Tone Quality and Loudness: the Interference
of Subjective Harmonics," The Journal of the Acoustical
Sociepy of America, V, No. 3 (January, 1934), pp. 173—80.

9James P. Egan, "The Effect of Noise in One Ear
upon the Loudness of Speech in the Other Ear," The Journal
of the Acoustical Society of America, XX, No. 1 (January,
1948), pp. 58-62.

SOIrwin Pollack, "The Effect of White Noise on the
Loudness of Speech of Assigned Average Level,“ The Journal
of the Acoustical Society of America, XXI, No. 3 (May, 1949),
pp. 255-58.

51Willard Thurlow and Leon Tabory, "Effects of Re-
peated Presentations of a Tone upon Absolute Loudness Judg-
ments,” Journal of General Ppychology, LX (April, 1959),
pp. 161-66.

52William Bevan and Joan Faye Pritchard, "Effect of
'Subliminal' Tones upon the Judgment of Loudness," Journal of
Experimental Psychology, LXVI, No. 1 (July, 1963), pp. 23-29.

53H. N. Wright, "Loudness as a Function of Duration,"
The Journal of the Acoustical Societyyof America, XXXVII,
No. 6 (June, 1965), p. 1174 (Abstract).

54Fletcher and Munson, op. cit.

 

 

 

 

55B. G. Churcher and A. J. King, "Performance of
Noise Meters in Terms of Primary Standard," Journa1_of the
Inspitution of Electrical Engineers, LXXXI, No. 487 (July,
1937), pp. 57—81 (London).

56D. W. Robinson and R. S. Dadson, "A Re-determina-
tion of the Equal-loudness Relations for Pure Tones," British
Journal of Applied Physics, VII (May, 1956), pp. 166-81.

57J. P. A. Lochner and J. F. Burger, “Pure-Tone

19

recently, however, several investigators have attacked the
problem of trying to learn more about how listeners evalu-
ate the loudness of stimuli (mostly pure tones and speech)
in terms of comfortableness and the relationship between
various factors and comfortable loudness level determina-
tions.58’59’60’61

Hardly an hour passes that an adult living in our
modern society does not evaluate the loudness of acoustic
stimuli. He does so, for example, each time he adjusts
the volume control of his radio or television set, although
he may not be aware of it. Likewise, he may not be aware
of the subjective nature of his evaluations, until perhaps
a companion complains that a program is "too loud," although
it is coming to him at a comfortable listening level. It

seems that persons prefer to do their listening at a level

which is "comfortable“ for thenulmit that this level is a

 

Loudness Relations,“ The Journal of the Acoustical Society
of America, XXXIV, No. 5 (May, 1962), pp. 576—81.

58Pollack, "Comfortable Listening Levels for Pure
Tones in Quiet and Noise," pp.-158-62.

59James Francis Kavanagh, "An Investigation of the
Most Comfortable Listening Levels for Speech" (unpublished
Ph.D. dissertation, Department of Speech, University of
Wisconsin, 1960).

60Lezak, op. cit.

61Eugene Walter Loftiss, "An Evaluation of the Ef-

fects of Selected Psychophysical Methods and Judgmental
Procedures upon Comfortable Loudness Judgments of Connected
Speech" (unpublished Ph.D. dissertation, Department of Speech,
University of Illinois, 1964).

20

dynamic rather than a static one and varies with the state
of the listener and with the listening situation. As is
the case with loudness, no instrument yet exists that can
measure the effects of a specified acoustic stimulus upon
a group of listeners in terms of "comfortableness.”

The available literature involving listeners' eval-
uations of the loudness of auditory stimuli in terms of
comfortableness stems from the clinical experience of aud-
iologists, as well as from the results of systematic in-
vestigations carried out in the laboratory. Within this
context, comfortable listening levels for pure tones and
speech, determined in quiet and in noise, have been reported
in the literature. Persons with normal hearing, as well
as individuals exhibiting various types and degrees of hear—
ing loss, have made these judgments. Several psychophys-
ical methods have been utilized in obtaining comfortable
loudness measures. The reliability with which listeners
can determine comfortable listening levels has also received
the attention of various writers. A number of factors that
may affect comfortable loudness judgments have been studied,
and the use of comfortable loudness judgments for clinical
and research purposes has been discussed in the literature.

Comfort Levels for Pure Tones
and Speech in Qpiet
The present study is concerned with investigating

the effects of intensity level of connected discourse and

21

acoustical background condition upon listeners' evaluations
of the loudness of speech in terms of whether or not it
is comfortable. However, pure tones have been used in some
investigations involving listener comfort because these

acoustic stimuli may be specified and reproduced easily,62

and because of their apparent similarity to speech.63 For
the latter reason, studies dealing with pure tones are per-
tinent to this discussion.

Pure Tones. The first attempt to determine com-
fortable listening levels for pure tones reported in the
literature was made by Norman A. Watson. An abstract of
a paper presented at the Thirty—Fourth Meeting of the Acous—
tical Society of America in December, 1947, indicated that
Watson had computed ranges and mean values of most comfort—
able listening (MCL) levels for pure tones of 128 to 8192
cps (at octave intervals) for individuals with normal hear-
ing and for persons who had been exposed to gunfire and air-
plane noise, some of whom exhibited marked hearing 1055.64
Unfortunately, the abstract of Watson's study did not de-

scribe the experimental design used, nor did it summarize

the results obtained. A search of the available literature

 

62Pollack, "Comfortable Listening Levels for Pure
Tones in Quiet and Noise," p. 158.

§3Kavanagh, op. cit., p. 3.
64Norman A. Watson, "Most Comfortable Listening
Levels for Pure Tones," The Journal of the Acoustical Soci-

ety of America, XX, No. 2 (March, 1948), p. 220 (AbstracEI.

22

revealed no further description of Watson's study, and ef-
forts to contact him for this information proved fruitless.

Two years later Watson and Tolan reported that the
most comfortable loudness level for pure tones might cor-
respond to the level at which a person with normal hearing
prefers to listen to soft music: 60 or 70 dB (presumably
re: normal audiometric threshold).65 These authors cited
no study in support of this statement, however.

In the same year that Watson and Tolan's book was
published, a second experimental study involving comfort-
able loudness levels for pure tones appeared. Hedgecock,
in an investigation of the relationships between degree
and type of hearing loss and the performance of hearing
aids, determined, by means of equal-loudness contours, the
relative intensity at which pure tones were judged to be
"just comfortably loud."66 Sixty-one hypacusic subjects
participated in this aspect of the investigation. The
average relative intensity at which pure tones were judged
to be "just comfortably loud” when compared with the ref—
erence tone was computed for all of the frequencies from

250 to 4000 cps. The mean of the averages for each of the

 

65Leland A. Watson and Thomas Tolan, Hearinngests
and Hearing Instruments (Baltimore: The Williams & Wilkins
Company, 1949), p. 473.

66LeRoy Darien Hedgecock, “Prediction of the Effi-
ciency of Hearing Aids from the Audiograms" (unpublished
Ph.D. dissertation, Department of Speech, The University
of Wisconsin, 1949).

 

23

frequencies was 74.9 i .56 decibels.67 Although no refer—

ence intensity was reported by Hedgecock, it may be assumed
that the mean intensity for which the subjects judged pure
tones as just comfortably loud was given in sound pressure
level re: 0.0002 dyne/cm2 since his results are comparable
to those of other researchers who have used this reference
level. Hereafter, Hedgecock's results will be reported
in dB re: 0.0002 dyne/cm2.

A third researcher to investigate comfortable lis-

68

tening levels for pure tones was Pollack. Thirty—three

persons with normal hearing determined monaural most com-

 

fortable listening levels for seven pure tones ranging from
125 to 8000 cps (in octave steps). Two monaural MCL levels
were chosen by each subject (one for each ear), the mean
of which represented his monaural MCL level. Pollack pre-
sented his findings in the form of a graph. As one inter—
polates Pollack's findings, he discovers that the mean of
the averages of the monaural MCL levels for each of the
frequencies was 72.7 dB SPL.69
In addition to obtaining mean monaural most com-

fortable listening levels for pure tones, Pollack also ob-

tained (1) mean binaural MCL levels for the same 33 normal

 

 

67Ibid., p. 102.

68Pollack, "Comfortable Listening Levels for Pure
Tones in Quiet and Noise," pp. 158-59.

69Ibid., Figure 2, p. 159.

24

hearing subjects mentioned above; (2) mean binaural lower
limits of the range of comfortable listening levels (LLCL),
most comfortable listening (MCL) levels, and upper limits
of the range of comfortable listening levels (ULCL) for
29 of these subjects; and (3) mean binaural LLCL, MCL lev-
els, and ULCL for seven trained listeners.7O These meas-
ures were obtained for each of the seven pure tones, which
were presented through earphones. The findings were again
presented in the form of a graph. As one interpolates
Pollack's findings, he discovers: (1) that the mean of
the averages (for each of the seven frequencies) of the
binaural LLCL was 54.1 dB SPL for 29 subjects and 42.4 for
.seven subjects; (2) that the mean of the averages (for each
of the seven frequencies) of the binaural MCL level was
65.9 dB SPL for 29 subjects, 64.6 for 33 subjects, and 62.0
for seven subjects; and (3) that the mean of the averages
(for each of the seven frequencies) of the binaural ULCL
was 80.7 for 29 subjects and 79.3 for seven subjects.71

As indicated previously, Hedgecock found that the
mean of the relative intensity levels at which pure tones
of frequencies 250 to 4000 cps were judged to be "just com-
fortably loud" by persons with impaired hearing was 74.9

dB (presumably re: 0.0002 dyne/cm2)--which is considerably

 

70Ibid., pp. 158-62.

71;p;g,, Figures 2 and 5, pp. 160—61.

25

higher than the mean most comfortable listening (MCL) lev-
els determined by Pollack's normal hearing subjects for

the same frequencies. By use of Pollack's data, the mean
of the averages for these frequencies was found to be 62.7
dB SPL for 29 subjects, 61.2 for 33 subjects, and 62.0 for
seven subjects. This difference is particularly noticeable
in light of the fact that the hard-of—hearing patients de-

termined the level at which the tone was jpst comfortably

 

loud, whereas the normal hearing listeners determined the

level at which the tone was most comfortable. Listening

 

was binaural in both instances, although the measures re-
ported by Hedgecock were determined in a sound field and
those reported by Pollack were determined with earphones.
This fact may also contribute to the discrepancy in the
results of these two studies.

Speech. Although the comfort levels for pure tones
in quiet reported in the literature seem to have been de-
termined experimentally for the most part, comfort levels
for speech have also been derived from the clinical expe-
rience of audiologists or have been implied in their writ-
ings. Carhart, for example, implied that the comfort level
for speech among hard-of-hearing persons was 40 dB above
normal threshold plus whatever gain they derived from their
hearing aids. Prior to obtaining aided speech reception
thresholds for hard-of—hearing persons, his patients were

asked to adjust the volume control of each hearing aid until

26

they judged speech striking the hearing aid at 40 dB above

normal threshold to be ”most comfortable."72

This proced-
ure provided a basis on which the performance of several
different hearing aids could be compared.
L. A. Watson declared that "a normal ear will reach

MCL on the [Maico] E-l audiometer at 40 db on the hearing
loss dial."73 Since Watson described no study from which
this value was derived, Kavanagh contacted him for this
information and found that this value for comfortable loud-
ness was derived from work in Watson's laboratory and from
his clinical experience.74 As reported by Kavanagh, Watson
stated that the MCL value for speech of 40 dB reported above
was for " . . . the average male voice speaking into the
microphone at a distance of about two inches. . . .'”75
Watson added that

we do find, in normal ears, considerable variation

from this point. Occasionally an individual will

prefer a fainter level at 35 db, and occasionally

at 45 db, but 40 db is unquestionably the point and

the most commonly preferred comfortable listening
level for normal ears.

 

72Raymond Carhart, "Tests for the Selection of Hear—

ing Aids,” The Laryngoscope, LVI, No. 12 (December, 1946),
p. 782.

73L. A. Watson, A Manual for Advanced Audiometry,
(Minneapolis: Colorcraft Press, 1948), p. 18.

74Kavanagh, op. cit., p. 5.
7SIbid.
76

Ibid.

27

Elsewhere in his Manual, Watson indicated that the
normal ear listens to speech at intensities 50 to 60 dB
above the speech reception threshold, and that deafened
ears like to listen to speech at levels from 15 to 25 dB
above SRT or 30 to 40 dB above pure tone thresholds.77
Again, no study was cited.

Fowler implied that speech was comfortably loud
for normal hearing persons when it was presented at 40 to

60 decibels above threshold.78

No systematic study was
offered in support of this statement, however. According
to Fowler, “the best hearing for articulate sounds is nor-
mally attained at an intensity of about 40 to 60 db above
threshold. This is the usual intensity of ordinary speech
at the distance of about 4 feet."79
Although various writers, such as those just cited,
have reported comfort levels for speech derived from their
clinical and laboratory experience, this topic has been
the focus of several systematic investigations during the
past twenty years. The first report of a direct attempt
to determine how listeners evaluate the loudness of speech

was made by Eisenberg and Chinn in 1945.80 The purposes

 

77L. A. Watson, op. cit., p. 29.

78EdmundPrince Fowler, "Diseases of the Neural
Mechanism of Hearing: Cochlea, Auditory Nerve, and its
Centers in the Medulla and Cortex,” Medicine of the Ear,
ed. Edmund Prince Fowler (New York: Thomas Nelson & Sons,
1947), p. 292.

791b1d.

80Philip Eisenberg and Howard A. Chinn, "Tonal

28

of this study were (1) to discover what tonal ranges lis—
teners preferred, (2) to investigate what volume levels
listeners preferred, and (3) to evaluate to what extent

the listener was influenced in his judgments by "prestige
suggestion.“ Three different frequency ranges and three
different intensity level settings were employed. The ma-
terials consisted of speech and music, which were presented
at various tonal ranges and volume levels. The listeners
were asked to compare the various combinations of intensity
level and frequency range and to indicate the one of each
pair that they preferred. In regard to volume level pref—
erences, the authors concluded that the listeners preferred
" . . . a peak sound intensity level somewhere between 60
and 70 db above the acoustic reference level."81 As Lezak
comments, this level was not specified but was probably
0.0002 dyne/cm2 because of the engineering background of
the investigators.82 Lezak reasoned that if the reference
level used was indeed 0.0002 dyne/cmz, the sound intensity
that the listeners preferred was about 40 to 50 dB above

83

normal threshold, which is similar to the intensity level

usually cited for conversational speech. Eisenberg and

 

Range and Volume Level Preferences of Broadcast Listeners,"
Journal of Experimental Psychology, XXXV, No. 5 (October,
1945), pp. 374-92.

811bid., p. 390.

82Lezak, 0p. cit., p. 7.

83Ibide , pp. 7-80

29

Chinn found no correlation between volume level preferences
and sex, age, education, amount of musical training, mus-
ical preferences, or whether the listeners played a musical
instrument.84

Davis §£_gl. also investigated the most comfortable
listening level for speech.85 Prior to obtaining various
types of audiometric data by means of five different fre-
quency patterns of electrical amplification with the Mas—
ter Hearing Aid, these researchers instructed l6 hard-of-
hearing subjects (involving 22 ears) to select the loudness
level at which connected speech heard with the Flat pattern
seemed "most comfortable." Listening was monaural, through
an earphone. The mean most comfortable listening level,
derived from these data, was 101.55 dB SPL (r.m.s.).86
This level is considerably above that reported by Eisenberg
and Chinn, perhaps resulting from the use of hard-of-hear-
ing persons as subjects. These researchers also noted that
all of the most comfortable listening levels lay on the
plateaus of the listeners' articulation curves.

A third investigator to study the most comfortable

listening level for speech was Hedgecock. In the investigation

 

84Eisenberg and Chinn, op. cit., p. 390.

85Hallowell Davis et al., Hearing Aids: An Experi-
mental Study of Design Objectives TCambridge, Mass.: Har-
vard University Press, 19477, p. 72.

86Ibid., p. 73.

 

 

 

"\

v‘l

30

of the relationships between degree and type of hearing
loss and the performance of hearing aids mentioned previ-
ously, he reported an unaided mean comfort level for speech
of 81.5 i 1.27 dB SPL for 61 hypacusic subjects.87 In this
study recorded speech was presented at conversational level,
and the investigator adjusted the gain control setting un—
til the subject reported that the speech was just comfort-
ably loud. Binaural free-field listening was employed.88
Despite the fact that both Hedgecock and Davis gp_g£. used
hard-of—hearing persons as subjects, Hedgecock's mean MCL
level for speech is more similar to that obtained by Eisen-
berg and Chinn for normal hearing subjects than to that
reported by Davis et a1.

Sawyer also investigated the most comfortable loud-
ness level for speech in a research project designed to
determine the essential tests and minimum physical equip-
ment needed in private otologic practice to obtain a reli—
able evaluation of a patient's hearing 1055.89 He took a
complete case history, performed a complete otorhinolaryngo-
logical examination, and administered various hearing tests

to each of 105 hard—of—hearing persons. With speech as

the stimulus, Sawyer found the most comfortable listening

 

87Hedgecock, op. cit., p. 102.

881bid., p. 62.

89Leroy L. Sawyer, "Office Procedure in Hearing
Evaluation: A Practical Approach," The Laryngoscopg, LX,
No. 11 (November, 1950), pp. 1061-85.

 

31

level to be “ . . . 12db. above the threshold at which the
patient could, with effort, understand each word."90 Un-
fortunately, no further information was reported regarding
the most comfortable listening level for speech; therefore,
his findings cannot be compared to those of other investi-
gators. Although no mention was made of the psychophysical
method used to obtain his results, it is perhaps reasonable
to assume that the method of minimal changes was employed,
since the setting was a clinical one.

Hedgecock obtained a second set of most comfortable
listening levels for speech as part of an investigation to
determine whether the loudness function for ears affected
by Méniere's disease was similar to that for unaffected
ears that sustained a loss of acuity for the high frequen-
cies comparable to the loss in the diseased ears.91 A sec-
ond purpose of this investigation was to validate the dif-
ference limen as a test of loudness recruitment by comparing
data obtained by this test with data obtained by equal loud-
ness balance tests.

Twenty-two patients who exhibited Meniére's disease
in one ear and a similar high-frequency hearing loss in the

other ear served as subjects. In addition to other tests,

 

901bid., p. 1076.

91Leroy D. Hedgecock, "Recruitment in Ears with
Abrupt Loss of Acuity for High Frequencies,“ Journal of
Speech and Hearing Disorders, XXII, No. 1 (March, 1957),
pp. 91-970

 

L.

\ul

vi

-4-

in".

1.;-

'01

u-‘

‘~

p“

0“

‘5

"v‘

32

two most comfortable listening levels for speech were
obtained for each of the 22 subjects, one for each ear.
Since the setting was a clinical one, it seems reason—
able to assume that the psychophysical method of limits
was utilized, with the researcher controlling the intensity
level of the speech and the subject indicating when his
most comfortable listening level had been reached.

The mean monaural most comfortable listening level
for the ears affected by Méniére's disease was 72 dB; that

for the better ears was 52 dB.92

Although no reference
level was given, it may be assumed that these values were
reported in terms of hearing loss since the setting was a

93 If such is the case, however, the value

clinical one.
for the better ears does not, as Lezak suggests, agree with
the results Hedgecock reported in his dissertation (81.5

dB SPL).94

The discrepancy in results could be related,
however, to the different types of hearing problems exhib-
ited by the two groups of subjects,95 or to the different
modes of stimulus presentation (binaural, in a sound field
vs. monaural, via earphones) employed in the two studies.

Hedgecock's mean monaural MCL level for speech for the bet-

ter ears does agree fairly closely, however, with the results

 

921bid., p. 93.

93Lezak, op. cit., p. 6.

941b1d.

951b1d.

33

reported by Eisenberg and Chinn for normal hearing subjects
(60 to 70 dB SPL). The mean monaural MCL level for speech
for the ears affected by Méniére's disease approaches that
reported by Davis gp_al. for hypacusic subjects [101.55 dB
SPL (r.m.s.)].

In a comprehensive study of most comfortable lis-
tening levels for speech, Kavanagh obtained MCL levels (in
dB re: normal audiometric threshold) for 182 subjects for
each of ten recorded speech samples during the ascending

and descending methods of presentation.96

Listening was
monaural, through an earphone, and involved a subject's

preferred ear. The mean ascending MCL level for all sub-

 

jects for all speech samples was 43.35 dB re: A.S.A. aver-
age normal "O," with a standard deviation of 5.10 dB. The
mean descendipg_MCL level was 48.66 dB, with a standard
deviation of 10.22 dB.97 These values agree to some extent
with those suggested by L. A. Watson (40 dB above the nor-
mal threshold for pure tones) and by Fowler (40 to 60 dB
above threshold). Kavanagh's results also approximate the
experimental findings of Eisenberg and Chinn (60 or 70 dB
SPL) for normal hearing persons.

Lezak also reported most comfortable listening lev-

els for speech. In a preliminary study conducted as part

 

96Kavanagh, o . cit., pp. 60-62.

97Ibid., p. 60.

34

of his doctoral research, he attempted to determine the
number of trials needed for reliable selection of comfort—
able loudness (MCL) levels.98 Six males with normal hear-
ing served as subjects. The psychophysical method of aver—
age error was employed, and testing was done individually.
Recorded speech was presented monaurally through an earphone.
The subject controlled the intensity level of the speech;
and when he had determined his MCL level, he signaled the
experimenter by raising his hand. Each subject determined
his MCL level for speech 22 times during a one and one-half
hour session.

The mean MCL levels for speech for each subject
averaged over 22 trials were 31.77, 36.55, 44.41, 33.28,
12.91, and 32.50 dB re: normal audiometric threshold.99
The mean most comfortable listening level for the six sub—
jects was 31.90 dB, which was considerably lower than that
reported by other investigators.

Lezak also carried out a second preliminary study

as part of his doctoral research, one purpose of which was

to evaluate the method of pair comparisons100 for determining

 

98

Lezak, o . cit., pp. 148—51.
991b1d., p. 152.
100

Lezak refers to this method as the method of
paired comparisons, although it is the stimuli rather than
the comparisons that are paired. See J. P. Guilford, Ps -
chometric_Methods (2d ed. rev.; New York: McGraw—Hill Book
Company, Inc., 1954), p. 5.

35

comfortable loudness.lOl Seven males with normal hearing

served as subjects for this study. Recorded speech (con-
nected discourse) was presented monaurally by means of an
earphone. Testing was done individually. Each subject
first determined his "uncomfortably soft" and “uncomfort-
ably loud" limits for listening to recorded speech. Then
pairs of stimuli were presented, the intensity levels of
which differed by 0, 5, 10, 15, etc., dB SL, until all com-
binations within these limits were exhausted. For each

pair of stimuli the subject was instructed to indicate which
one was the “most comfortable." Analysis of the data yielded
a mean most comfortable loudness level for speech of 65 dB

SL,102

a level which was somewhat above the experimental
findings of Eisenberg and Chinn, who also used the method

of pair comparisons (60 to 70 dB SPL), and those of Kavanagh,
who used the method of limits (43.35 and 48.66 dB SL).

Still another researcher who determined comfortable
listening levels for speech was Loftiss. For his doctoral
dissertation, he investigated the effect that judgmental
procedures, psychophysical methods, test-retest, test dura-
tion, and contrast or time error had upon normal hearing

subjects' estimates of the loudness of speech; and the effect

that the above factors had upon the precision of listener

 

lOlLezak, op. cit., pp. 153—58.

lOZIbid., p. 158.

E;

4th
.-

\l

\

36

judgments.103 Each subject listened to recorded speech

(connected discourse) presented under four experimental
conditions. For each condition, he determined the soft
comfortable, most comfortable, and loud comfortable loud-
ness of the speech. Each subject heard the stimulus mate-
rials presented monaurally via an earphone (involving his
right ear) and binaurally via a loudspeaker. Half of the
subjects used an anchoring procedure, and half of the sub—
jects used no anchoring procedure.104 The test was read-
ministered to each subject after one week had passed.
Loftiss reported the obtained loudness levels in
dB above the normal speech threshold. The median and geo-
metric mean values that the listeners chose for soft com-
fortable, most comfortable, and loud comfortable loudness
judgments, regardless of psychophysical method or anchor-
ing procedure used, were presented in the form of a table.

This table is reproduced in part in Table l.105

 

103Loftiss, op. cit.
104

In the anchoring procedure the subject first
determined his most comfortable loudness level. Then the
stimulus was presented at this level and increased until
his loud comfortable loudness level had been reached or
decreased until his soft comfortable loudness level had
been reached. In the non-anchoring procedure, the subject
was notified which of the three loudness judgments he was
to make. The speech was presented at a randomly selected
level, and the subject made his judgment. See Loftiss,
ibid., pp. 12-13.

105Information taken from Loftiss, ibid., Table
l, p. 20, and reproduced here with the author's permission.

37

TABLE 1

MEDIAN AND GEOMETRIC MEAN VALUES OF THE POOLED DECIBEL
SCORES FOR EACH OF THE COMFORTABLE-CONVERSATIONAL
LISTENING LEVELS‘I

 

 

 

 

Loudness Level Median Mean
Earphone

Soft Comfortable 37 decibels 37 decibels

Most Comfortable 51 decibels 51 decibels

Loud Comfortable 68 decibels 68 decibels
Loudspeaker

 

Soft Comfortable 31 decibels 32 decibels
Most Comfortable 47 decibels 47 decibels

Loud Comfortable 60 decibels 60 decibels

 

‘Based on an N of 80 judgments for each level.
All values are expressed in decibels, i.e., normal speech
threshold, 22 decibels above 0.0002 dynes/cm2.

Summary. Comfort levels for pure tones and speech
determined in quiet have been discussed in the literature.
In general, comfort levels for pure tones have been deter-
mined experimentally, although Watson and Tolan cited no
study in support of their statement that the most comfort-
able loudness level for pure tones might correspond to a
level of 60 or 70 dB (presumably re: normal audiometric
threshold). Two researchers reported MCL levels for pure
tones derived from systematic investigations. Hedgecock

found that the mean of the average intensity levels at which

38

hypacusic subjects judgedwvarious pure tones to be just
comfortably loud was 74.9 dB SPL. Calculations involving
Pollack's data for the same pure tones that Hedgecock used
yielded average MCL levels of 62.7 dB SPL for 29 subjects,
61.2 for 33 subjects, and 62.0 for seven subjects. These
measures indicated that the intensity level judged to be
most comfortable for normal listeners was considerably lower
than that reported by Hedgecock.

A summary statement regarding comfortable listen-
ing levels for speech determined in quiet is difficult to
formulate because the levels reported in the literature
(1) have been derived both from the clinical experience
of audiologists and from experimental studies, (2) have
been determined both for persons with normal hearing and
for those exhibiting various types and degrees of hearing
impairment, (3) have been based upon various definitions
of "comfortable," (4) have been obtained using various psy-
chophysical methods, (5) have been measured using differ-
ent modes of stimulus presentation (monaural or binaural,
through earphones, or binaural, in a sound field), and (6)
have been reported in decibels re: different reference
levels. Despite these difficulties, the most comfortable
listening levels for speech that were reported in the pre-
ceding discussion are presented below. Since many of the
writers used the threshold of normal hearing as the refer-

ence level, all of the values are presented re: this level:

39

Carhart 40 dB
Watson 40 dB

30 to 40 dB
Fowler 40 to 60 dB
Eisenberg and Chinn 38 to 48 dB
Davis et a1. 80 dB
Hedgecock 60 dB

52 dB

72 dB
Kavanagh 43.55 dB

48.66 dB
Lezak 31.90 dB

65 dB
Loftiss 51 dB

47 dB

Comfort Levels for Pure Tones in Noise

 

The preceding section centered upon a discussion
of comfort levels for pure tones and speech that were de-
termined in quiet. This section will deal with comfort
levels for pure tones determined in the presence of noise.
There are no data available concerning comfort levels of
speech determined in noise.

A search of the available literature revealed only
one study in which comfort levels for pure tones were de-
termined in noise. Pollack carried out two series of de-
terminations of the most comfortable listening (MCL) level
for pure tones in noise.106 In addition to determining
his MCL level, each subject was also instructed to deter—
mine his lower and upper limits of the range of comfortable

listening levels.

The lower limit of the range of comfortable listen-
ing levels was specifically defined as that level

 

106Pollack, "Comfortable Listening Levels for Pure

Tones in Quiet and Noise," pp. 159—60.

40

at which it is necessary to “strain“ in order to
hear the signal clearly; and the upper limit as
that level at which the sound appeared “annoying"
or "bothersome.”107

Twenty-seven untrained listeners with normal hear—
ing participated in the first series of determinations.
Each subject determined the lower limit of the range of
comfortable listening levels (LLCL), the most comfortable
listening (MCL) level, and the upper limit of the range
of comfortable listening levels (ULCL) for a 1000 cps tone
presented with 35, 65, and 95 dB SPL white noise. Listen-
ing was binaural, through earphones.

Pollack presented his findings in the form of a
graph. He found that the LLCL, the MCL level, and the ULCL
increased as the intensity level of the background noise
increased.108

Seven trained listeners with normal hearing partic—
ipated in the second series of determinations. Each sub—
ject determined the LLCL, the MCL level, and the ULCL for
seven pure tones (125 to 8000 cps, in octave steps) in quiet
and with 35, 55, 75, 95, and 115 dB SPL white noise. Lis-
tening was binaural, through earphones. Pollack's findings
were presented in a series of graphs, with the effect of

noise on comfortable listening levels plotted as a function

of frequency. Again the LLCL, the MCL level, and the ULCL

 

107Ibid., p. 161.

108Ibid., Figure 4(a), p. 160.

o

.1

s v

Isa“
vv‘

!-.

‘L.

41

increased as the intensity level of the background noise
increased. He noted that the effect of noise was greater
upon the middle and upper frequencies than upon the lower

. 109
frequenc1es.

Range of Comfortable Loudness Levels for Pure
Tones and Speech in Quiet and in Noise

The range of comfortable loudness (RCL) for pure
tones and speech is not discussed extensively in the lit—
erature, and various writers have not agreed upon a defini-
tion of this term. Watson and Tolan, for example, equate
the RCL with the “dynamic range" or the "range of tolerable
loudness."llO These writers hold that the range of comfort-
able loudness should be measured from the listener's ”Thresh-
old of Intelligibility“ (the level at which he can repeat
correctly 50% of the test words) to the Threshold of Dis-
comfort (the level at which the listener first reports the
stimulus to be uncomfortably loud). Watson and Tolan in-
dicate that the range of comfortable loudness varies (1)
with the stimulus materials (pure tones or speech), (2)
with the listener's hearing acuity, and (3) with the type
of hearing impairment a listener exhibits. According to
these writers,

. . . the RCL of a hard of hearing person may vary
from as little as 5 db for certain pure tones to

 

10911616., Figure 5, p. 161.

110Watson and Tolan, o . cit., p. 401.

42

as much as 100 db, and from as little as 10 db for
speech to as much as 50 db. The upper and lower
limits of the RCL in cases of conductive deafness
are generally not as sharp and definite as in nerve
types of deafness. The RCL in conductive cases
of deafness is not a precise figure, and may vary
as a matter of judgment by 10 or 15 db. In nerve
types of deafness it is more precise and exact,
both for pure tones and for speech.111
No study is cited, however, as the basis for the above state-
ment.

Two experimental studies involving the range of
comfortable listening levels are reported in the literature.
Pure tones are employed as stimuli in the first study; speech
serves as the stimulus material in the second.

In 1952 Pollack studied the range of comfortable
listening levels for pure tones in quiet and in noise. He
operationally defined the RCL as the range between the lower
limit of the range of comfortable listening levels (LLCL)
and the upper limit of the range of comfortable listening
levels (ULCL), as determined by the subjects.112 Twenty-
nine normal hearing subjects determined the LLCL, the MCL
level, and the ULCL for seven pure tones (125 to 8000 cps,
in octave steps) in quiet. Listening was binaural, through

earphones. Pollack presented his results in the form of

a graph, with comfortable listening levels for pure tones

 

lllIbid.

112Pollack, "Comfortable Listening Levels for Pure

Tones in Quiet and Noise,“ p. 160.

 

 

5"

ha

'01

"r

gs.

NA“
“V“

n'fl
‘4“.

“I
(I:

A~-
\ .
v“:

43

plotted as a function of frequency.113 He also plotted
the range of comfortable listening levels for each fre—
quency. Pollack's findings indicate that the range was
greatest for the middle frequencies, least for the lower
frequencies, and intermediate for the higher frequencies.114
He noted that the range above and below the MCL level is
approximately symmetrical (on a decibel scale).115
In another series of determinations reported in
the same study, 27 untrained listeners with normal hear-
ing determined the LLCL, the MCL level, and the ULCL for
a 1000 cps tone with 35, 65, and 95 dB SPL white noise.116
Again, listening was binaural, through earphones. Pollack
presented his results in the form of a graph. His find-
ings indicate that as the background noise level increases,
in general the lower and upper limits of the range of
comfortable listening levels and the MCL level increase
and the range of comfortable listening levels decreases,
mainly because of the greater increase of the lower limit
relative to that of the upper limit.117 “Specifically,

as the noise level is increased over a range of 60 db, the

lower limit is increased about 35 db while the upper limit

ll3Ibid., Figure 2, p. 160.

ll4Ibid.

llsIbid.

ll61b1d., p. 161.

ll7Ibid.

.:
U)
'7‘.

..
‘I

44

is increased only about 18 db."118

A third series of determinations reported by Pollack
involved seven experienced listeners. Each subject deter-
mined his LLCL, MCL level, and his ULCL for seven pure tones
(125 to 8000 cps,in octave steps) in quiet and with 35, 55,
75, 95, and 115 dB SPL white noise. Again, Pollack reported
his findings in the form of a graph. He noted that these
results parallel those reported above.119 In addition, the
changes with increased background noise are smaller at the
lower frequencies than at the middle and higher frequencies.

Pollack also found that the range of comfortable
listening levels for pure tones decreases as the intensity
level of the background noise increases from 35 to 115 dB
SPL.120 The presence of background noise has a greater
effect upon the middle and higher frequencies (5000 and
1000 cps, respectively) than upon the lower frequency (125
cps).121

The research by Pollack cited above dealt with lis-

teners' determinations of the range of comfortable listen-

ing levels for pure tones. Two other investigators studied

 

the range of comfortable listening levels for speech.

 

118Ibid.

llgIbid.

lzoIbid., Figure 6.

1211bid.

a:

\

v

'0‘

R“
'.H\-

:u

.,~\
- R

«Q

«.35
sq ‘

45

In 1960 Lezak stated that lying between the thresh-
old of sensitivity and the threshold of discomfort is an

22 which is similar to Watson

area of comfortable loudness,l
and Tolan's definition of the range of comfortable loudness
mentioned previously. Lezak investigated the upper limit
of comfortable loudness (ULCL) as a possible level at which
speech stimuli might be presented for the purpose of deter-
mining a patient's intelligibility score for speech. His
rationale for investigating the possibility of using a dif-
ferent level was based on the results of various studies
which indicated that, although 40 dB above threshold might
be a comfortable level for presenting stimuli to normal
hearing persons, it may not be comfortable for those indi-

viduals with,impaired hearing.123

Before the upper limit

of comfortable loudness could be used for this purpose,
however, the relationship between the threshold of intel-
ligibility for speech and the ULCL needed to be investigated,
which was one of the purposes of Lezak's study.124 The
method of limits was used to obtain the uncomfortable loud-
ness threshold (ULT), and the ULCL was operationally defined
as a value 5 dB below the ULT.

Based upon his implication that the range of com-

fortable listening levels extended from the threshold of

 

122Lezak, op. cit., p. 1.
123Ibid.
124

Ibid., pp. 42"43.

46

sensitivity to the uncomfortable loudness threshold (ULT),
Lezak found that for most comparisons involving clinic sub-
jects the dB difference between the range to uncomfortable
threshold (RUT) observed means and a hypothetical 40 dB

125 For example,

level was not statistically significant.
when subjects were grouped according to audiometric pattern,
those with flat audiogram curves showed a difference of 40
dB SL, those with gradual audiogram curves showed a differ-
ence of 39 dB SL, and those with marked curves showed a
difference of 36 dB SL.126 These differences were not sig-
nificantly different from the 40 dB range.127 However,
Lezak found that the dB difference between the observed
RUT mean for the normal hearing subjects in his study (60
dB SL) and the 40 dB range was significant.128
Loftiss also discussed the range of comfortable
loudness for speech. In a study aimed at investigating
the effects of selected psychophysical methods and judg—
mental procedures upon 1isteners' judgments of the loudness
of connected discourse, he obtained median and geometric

mean values for soft comfortable, most comfortable, and

loud comfortable loudness judgments for twenty normal

 

1251b16., p. 128.

lzerid.

l27Ib1d.

128Ibid.

47

129

hearing subjects. These values, regardless of psycho-

physical methods or judgmental procedures used, were pre—

sented in the form of a table, which is reproduced again

in Table 2 for easy reference.130

TABLE 2

MEDIAN AND GEOMETRIC MEAN VALUES OF THE POOLED DECIBEL
SCORES FOR EACH OF THE COMFORTABLE-CONVERSATIONAL
LISTENING LEVELS'

 

 

 

 

 

Loudness Level Median Mean
Earphone

Soft Comfortable 37 decibels 37 decibels

Most Comfortable 51 decibels 51 decibels~

Loud Comfortable 68 decibels 68 decibels
Loudspeaker

Soft Comfortable 31 decibels 32 decibels

Most Comfortable 47 decibels 47 decibels

Loud Comfortable 60 decibels 60 decibels

‘Based on an N of 80 judgments for each level.

All values are expressed in decibels,
threshold, 22 decibels above 0.0002 dynes/cmZ.

According

to Loftiss,

i.e.,

normal speech

“the dynamic range of loud-

ness level measures for both the earphone and loudspeaker

 

129
130

Loftiss, op. cit.

Information obtained from Loftiss, ibid., Table

l, p. 20,

and reproduced with the author's permission.

48

presentations (30 decibels (3)1 db) is similar to the range

cited for conversational speech."131

This range is some-
what less than that reported by Lezak for normal hearing
subjects (60 dB SL); however, it is possible that variations
in defining the limits of the RCL in these two studies may
account for the discrepancy in results. Perhaps Lezak's
definition of the range of comfortable listening levels

is broader than that of Loftiss.

Loftiss commented that the decibel differences be—
tween the means of each of the comfortable loudness levels
were somewhat similar (i.e., there was a range of 14-15
dB between soft comfortable and most comfortable loudness
judgments, and a range of 17-19 dB between most comfortable
and loud comfortable loudness judgments, for earphone and
loudspeaker presentations). This observation tends to sup-
port Pollack's finding that, for pure tones, the range above
and below the most comfortable listening level is approxi-
mately symmetrical on a decibel scale.132 Loftiss also
noted that the soft comfortable loudness measures were 30-
37 dB above the normal auditory threshold for speech and
the loud comfortable loudness measures were at least 20

dB below speech levels cited for auditory discomfort.

These findings also tend to corroborate those of Pollack.

 

l3lIbid., p. 19.

 

GO

.A s

f-
.-
.n u

49

Pollack found that persons with normal hearing chose lower
and upper limits of the range of comfortable listening
levels that were considerably above and below the thresh-
old for hearing and the threshold of discomfort.133 How-
ever, Loftiss used soft and loud comfortable loudness judg-
ments in place of the lower and upper limits of the range
of comfortable listening levels, respectively.

Psychophysical Methods Used to Determine
Comfort Levels for Pure Tones and Speech

 

A search of the available literature indicated that
listeners' determinations of comfort levels for pure tones
and speech depend to some extent upon the psychophysical
methods used to obtain these measures.134 This relation-
ship will be treated more fully in the section dealing with
factors related to listeners' determinations of comfortable
loudness levels.

In the present section several psychophysical meth-
ods that have been utilized in obtaining comfort levels
in the audiological clinic and in the laboratory will be
described: the Method of Average Error, the Method of Min-

135

imal Changes, and the Method of Pair Comparisons. When

 

133Pollack, "Comfortable Listening Levels for Pure
Tones in Quiet and Noise," p. 160.

134Menzel, op. cit., p. 107.

135Guilford, op. cit.

SO

systematic investigations involving comfortable loudness
are discussed, the following information will be provided
for clarity as necessary: the population (clinical vs.
non-clinical) from which subjects were drawn and the mode
of stimulus presentation (monaural vs. binaural, with ear—

phones or in a sound field) employed.

The Method of Averag§_Error (Method of Adjustment).
This psychophysical method can be described briefly as one
in which the subject adjusts the stimulus until he judges
it to be subjectively equal to, or in some desired relation
to, a criterion.136

A clinical modification of the psychophysical meth—
od of average error has been employed in determining a pa-
tient's most comfortable listening level for speech for
the purpose of comparing the performance of several differ-
ent hearing aids. For example, Davis 33:21, indicated that
the patient was instructed to set the gain control of each
aid to his most comfortable listening level for conversa-

137

tional speech [usually held constant at a level of 60

 

136S. 3. Stevens, "Mathematics, Measurement, and
Psychophysics,” Handbook of Experimental Psychology, ed.
S. S. Stevens (New York: John Wiley & Sons, Inc., 1951),
.p. 43.

137Hallowell Davis et al., ”The Selection of Hear-
ing Aids,” The Lapyngoscope, LVI, Nos. 3 and 4 (March and
April, 19467: p. 138. I[This article is a reprinting of
"PNR-7” report issued by Harvard Psycho-Acoustic Labora-
tory, Cambridge, Massachusetts, December 31, 1945.]

51

dB re: 0.0002 dyne/cm2 or 40 dB above the normal thresh-
old of intelligibility].138 Carhart also advocated the
use of this technique for establishing the comfort level
for speech for purposes of comparing hearing aids.139
To achieve comfort level the patient adjusts the
hearing aid so that a predetermined input signal
[usually 40 dB above the normal threshold] is being
amplified to the point where he judges the signal
to be "most comfortable." Stated differently, it
is the adjustment which he would prefer if he were
going to listen to the sound for a long period.

The psychophysical method of average error has also
been used by Norman Watson and Pollack, both of whom were
interested in determining most comfortable listening levels
for pure tones for normal and hard—of—hearing subjects.

Norman Watson used normal hearing subjects and
those who had been exposed to gunfire and airplane noise,
with resultant marked hearing loss in some cases, in his
study involving the determination of most comfortable lis-
tening levels for pure tones.141 Although a search of the
available literature failed to yield a complete report of

this study, and details of this research are not presented

in the abstract, the following statement implies that each

 

13811616., p. 139.

139Raymond Carhart, "Volume Control Adjustment in
Hearing Aid Selection," The Laryngoscope, LVI, No. 9 (Sep-
tember, 1946), p. 511.

140Ibid.

141Norman A. Watson, op. cit.

C)
.01

EU“

ten

'0
_ .
as

 

,.

52

of Watson's subjects adjusted the intensity level of the
tones to his most comfortable listening level:

By the most comfortable listening level for a pure
tone is meant the level which the listener chooses
as most comfortable for listening when he listens

to a pure tone of which he can change the level

at will.142

Pollack used the method of average error in a sys-

tematic investigation of comfort levels for pure tones.143

He instructed each listener

. . . to adjust an attenuator dial (numerals not
visible to him) "until the sound is at your most
comfortable listening level, that is, at that level
which you would most prefer to listen to the sound
over a moderate period of time if you were compelled

to do so." Each listener was instructed to explore
a wide intensity range before giving his final set-
ting.144

In the same study most comfortable listening levels,
lower limits of the range of comfortable loudness, and upper
limits of the range of comfortable loudness were also de-
termined by two other groups of subjects for various pure
tones in quiet and/or in noise, using the same procedure
described above.145

A search of the available literature yielded three

studies that utilized the method of average error in deter-

mining comfort levels for speech. Lezak, in a preliminary

 

l421bid.

143Pollack, ”Comfortable Listening Levels for Pure
Tones in Quiet and Noise," pp. 158-60.

l44lbid., p. 158.

145

Ibid., pp. 161-62.

53

study conducted as part of his doctoral research, investi—
gated the method of average error as a means of determin-
ing most comfortable loudness levels.146 Six males with
normal hearing served as subjects. The subject controlled
the intensity level of the speech and signaled the research-
er when he had decided upon his MCL level.147 In a second
preliminary study, Lezak also used this method to determine
the uncomfortable loudness thresholds of seven male subjects
with normal hearing.148

Loftiss compared the effects of the method of aver-
age error and the method of minimal changes upon judgments
of the loudness of connected discourse.149 Using the method
of average error, twenty normal hearing subjects determined
soft comfortable, most comfortable, and loud comfortable
loudness judgments for speech. Again, the subject adjusted
the attenuator dial and signaled the researcher when each

of these levels was achieved.150

The Method of Minimal Changes (the Method of Limits).
This psychophysical method can be described briefly as one

where the experimenter increases and/or decreases the

 

146Lezak, O o Cite, pp. 148-51.

147Ibid., pp. 149—51.

l481bid., p. 156.

149Loftiss, op. cit.

1501b16., p. 9.

54

intensity level of the stimulus and the subject ”signals

151 The psychophys-

its apparent relation to a criterion.“
ical method of minimal changes has been employed in the
clinical setting as a means of obtaining patients' aided
or unaided most comfortable listening levels for purposes
of hearing aid selection or to determine whether recruit-
ment is present. For example, Leland A. Watson advocated
obtaining an unaided most comfortable loudness level in
order to determine a patient's hearing loss for speech at
the level at which he should do his hearing.152 According
to Watson, this could be accomplished by talking to the
patient through the micrOphone while increasing the inten-
sity by turning the attenuator dial until the listener in-
dicated the level at which listening was most comfortable
for him.153
Watson and Tolan also used the method of minimal
changes in the audiology clinic. They advocated determin—
ing a patient's most comfortable listening level at each
of the frequencies employed in a hearing evaluation.154

According to these writers, comfortable loudness could be

established by presenting the tone at the threshold

 

15J'Stevens, "Mathematics, Measurement, and Psycho-
physics,” p. 43-

152Leland A. Watson, op. cit., p. 18.
153Ibid.
154

Watson and Tolan, op. cit., p. 86.

SS

intensity and then at an uncomfortable loudness level, in-
structing the patient to indicate the preferred listening
level between these two limits; or by obtaining his most
comfortable listening level for a 1000 cps tone and, using
this level as a reference, finding equal loudness at the
other frequencies.155 Although they do not state so ex-
plicitly, Watson and Tolan imply that these procedures may
also be used to determine comfortable listening levels for
speech.156

As recently as 1964 Newby also indicated that the
method of minimal changes could be used in determining most
comfortable listening levels in a clinical setting. Accord-
ing to this writer, the patient is instructed to signal
when the speech is most comfortably loud for him, as the
examiner varies the intensity of connected speech at supra-
threshold levels.157 The MCL may be determined either mon-
aurally or in a sound field and is measured in dB above
zero SRT.158

The psychophysical method of minimal changes has

also been used by several experimenters to determine com-

fort levels for pure tones and speech for normal and hard—

 

lSSIbid.

156Ibid., p. 401.

157Newby, op. cit., pp. 113—14.

lssIbid., p. 114.

56

of—hearing subjects in the laboratory. Hedgecock determined
by means of equal-loudness contours the relative intensity
at which pure tones of 250 to 4000 cps were judged to be
"just comfortably loud“ when compared with a 1000-cycle
reference tone which had been evaluated as comfortably loud
by each subject.159 The experimenter adjusted the volume
of the test tone until the subject reported that it was
equal in loudness to the lOOO-cycle reference tone.160
This level was recorded (in dB SPL) as the equal-loudness
point on the audiogram for that frequency.161
In the same study Hedgecock also determined the
level at which the same 61 hypacusic subjects judged

162 The re-

spondee words to be just "comfortably loud."
searcher manipulated the intensity level of the words, and
the subject verbally directed him to increase or decrease
the volume.163 The intensity level which the subject eval-
uated to be just comfortably loud was noted.

Several years later Hedgecock published another

study, the main purpose of which was to determine whether

the loudness function for ears affected by Méniére's disease

 

159Hedgecock, ”Prediction of the Efficiency of Hear-
ing Aids from the Audiograms," p. 60.

l6OIbid., p. 61.

lelIbid.

1621bid., p. 72.

163Ibid.

. aim #v \f-

M .

57

was similar to that for unaffected ears which sustained a
loss of acuity for high frequencies comparable to the loss
in the diseased ears.164 In addition to other measures,
two comfortable listening levels for speech (one for each
ear) were obtained for each of the 22 subjects who ex—
hibited unilateral Meniére's disease. Since the setting
was a clinical one, it may be reasonable to assume that
the method of minimal changes was employed, with the experi-
menter controlling the intensity level of the speech, and
the subject signaling when his most comfortable listening
level had been reached.165
In his doctoral research, Kavanagh employed the
method of limits to determine most comfortable listening
levels for various speech samples for 182 normal hearing

subjects.166 The subject was instructed to indicate when

a particular speech sample was "comfortably loud."l67 The

volume of the speech sample was then increased, and the
listener indicated when it was no longer comfortable. Then
the volume was decreased until the subject indicated that

168

the speech was again comfortable. Using this procedure,

 

164Hedgecock, “Recruitment in Ears with Abrupt Loss

of Acuities for High Frequencies," pp. 91-97.

l651bid., p. 94.

166Kavanagh, op. cit.

'167Ibid., p. 28.

1681bid., pp. 28-29.

 

 

~—

A.

.I. ~

.h\
s

58

Kavanagh thus obtained ascending and descendigg_most com—

 

fortable listening levels for each subject for each speech
sample.

Lezak, in a preliminary study conducted as part
of his doctoral research, utilized the method of minimal
changes to determine the threshold of uncomfortable loud-
ness for seven subjects with normal hearing.169 The re-
searcher introduced the signal at a low intensity level
and increased it in 5 dB steps until the subject signaled
that the speech was uncomfortably loud.170

Loftiss also employed the method of minimal changes
in the laboratory. In his doctoral research he compared
the effects of the method of average error and the method
of minimal changes upon subjects' judgments of the loudness
of connected discourse.171 Using the method of limits,
the subjects determined soft comfortable, most comfortable,
and loud comfortable loudness levels for speech. Here the
experimenter controlled the intensity level of the speech,
and the subject indicated when each of these levels had

been reached.172

The Method of Pair Comparisons. When this

 

169Lezak, o . cit., pp. 156-57.

l7OIbid.

l7J'Loftiss, op. cit.
172

Ibid., p. 9.

59

psychophysical method is employed, each stimulus is paired
with every other stimulus and the subject indicates which
one of each pair is greater with respect to a given quan-
tity or attribute.173 Two studies reported in the litera-
ture on determinations of comfort levels employed this method.
It is interesting to note that one of them represented the
first attempt to determine experimentally the volume levels
that listeners preferred. This study was carried out by
Eisenberg and Chinn.174 Three different frequency ranges
and three different intensity level settings were employed.
The materials consisted of speech and music, which were
presented at various tonal ranges and volume levels. Sub-
jects were instructed to compare the various combinations
of intensity level and frequency range and to indicate the
one of each pair that they preferred.175

Lezak, in a preliminary study carried out as part
of his doctoral research, also used the method of pair com—
parisons in order to determine whether this psychophysical
method might yield more reliable MCL judgments than the

176

method of average error. First,each subject determined

his "uncomfortably soft" and "uncomfortably loud" limits

 

173Stevens, "Mathematics, Measurement and Psycho-
physics,“ p. 43-

174Eisenberg and Chinn, op. cit.

l751bid., p. 378.

176Lezak, op. cit., pp. 143-58.

60

for listening to speech. Then various pairs of stimuli
were presented, which differed from each other by O, 5, 10,
15, etc., dB SL, until all combinations within these limits
were exhausted. For each pair of stimuli presented, the
subject was instructed to indicate which one was "most com-
fortable."177
As was indicated in the above discussion, three
psychophysical methods have been employed in clinical and
laboratory determinations of comfortable listening levels
for pure tones and speech: the method of average error,
the method of minimal changes, and the method of pair com-
parisons. No previous research involving comfortable loud—
ness judgments has utilized the psychophysical method of
successive categories; however, this procedure has been
employed in studies dealing with loudness in general. In
the method of successive categories, a subject determines
to which of several limited categories a stimulus belongs.178
The method of successive stimuli corresponds to the rating
scale method (method of single stimuli) described by Stevens,
where a listener rates each stimulus in terms of some at—
179

tribute.

Since this psychophysical procedure had been

 

1771bid., p. 155.
178 . .

Guilford, op. c1t., p. 223.
179

Stevens, "Mathematics, Measurement and Psycho-
physics," p. 43.

61

utilized previously in loudness studies, it was felt that

it merited investigation as a means of evaluating comfort-
able loudness of auditory stimuli. It is possible that

this procedure could be used in the clinic and in the lab-
oratory more efficiently than the other psychophysical meth—
ods.

The method of successive stimuli was utilized in
the present study, the purpose of which was to investigate
the effects of intensity level of connected discourse and
acoustical background condition upon listeners' evaluations
of the loudness of connected discourse. In this study,
subjects were instructed to rate each stimulus in terms
of comfortableness, using a rating scale of 1 through 5,
ranging from "too soft" to "too loud." The application
of this method to listeners' evaluations of the loudness

of speech will be described more fully in Chapter III.

Reliability of Comfortable Loudness Judgments

The reliability with which listeners can determine
comfort levels is important if this technique is to be uti-
lized in hearing and hearing aid evaluations and in research,
and this problem has received considerable attention in
the literature. The reliability with which listeners can
evaluate the loudness of various types of auditory stimuli
has been subjected to systematic analysis by some investi-
gators, but prior to this time various writers mentioned

the reliability of comfortable loudness determinations in

62

rather vague terms.

For example, Carhart indicated that the "comfort
level" procedure he described had proved clinically useful
in comparing the sensitivity of several hearing aids at
a "use“ setting. In the clinical setting Carhart found
that patients who had been indoctrinated in the use of hear-
ing aids could determine most comfortable listening levels
for speech (usually presented at 40 dB above normal audio-
metric threshold) on repeated trials that were

. . . within the limits of accuracy ordinarily ac—
cepted in clinical audiometry. It is the writer's
opinion that the "comfort level" method is a feas-
ible clinical procedure for obtaining reasonable
equivalence in adjusting the volume controls on
different instruments. Extra precision of results
is possible while evaluating hearing aids if two
independent thresholds at the same "comfort level"
are obtained with each instrument.180

On the contrary, however, Davis indicated that many
subjects are unable to duplicate their own most comfortable
listening level settings with a given hearing aid on repeat-
ed trials. In Davis' words, "'Most Comfortable' seems to
cover a range of a good many decibels.“181 Despite this
difficulty, Davis advocated using the comfort level proced—

ure in hearing aid evaluation as a means of equating the

volume control of several different aids. Watson and Tolan

 

180Carhart, "Tests for the Selection of Hearing
Aids," p. 793.

181Hallowell Davis, "Hearing Aids," Hearing and
Deafness: A Guide for Laymen, ed. Hallowell Davisfi(New
York: Murray Hill Books, Inc., 1947), p. 209.

63

also pointed out that " . . . the most comfortable loudness
level . . . is sometimes . . . difficult to determine pre—
cisely. . . °”l82
Still other authorities questioned the validity

of adjusting the volume control of several promising hear-
ing aids to comfort level for speech as a basis for making
comparisons among these instruments. Hirsh commented that
”although this sounds like a very sensible way out of the

problem, it is clear . . . that the range of intensities

that would be judged to be comfortably loud may be consid-

erable."183

Silverman and Taylor also felt that one of
the weaknesses in comparing several hearing aids on the
basis of the degree to which they increase the listener's
ability to understand faint speech lies in the setting of
the gain controls.
The listener is usually instructed to set it him-
. self so that average speech comes to him at "the
most comfortable loudness." Unfortunately, many
subjects on repeated trials do not duplicate their
own settings for most comfortable loudness. "Most
comfortable” seems to cover a range of a good many
decibels.184
According to Silverman and Taylor, tests based upon a per-

son's ability to understand speech with the help of one

 

182Watson and Tolan, op. cit., p. 86.

183Hirsh, op. cit., p. 298.

184S. R. Silverman and S. Gordon Taylor, "Hearing
Aids,“ Hearing and Deafness, ed. Hallowell Davis and S.
Richard Silverman (2d ed. rev.; New York: Holt, Rinehart
and Winston, Inc., 1960), p. 326.

 

64

instrument vs. another instrument

. . . are likely to be tests of how the listener

happens to set the gain control, not of how well

the instrument “fits" or how "efficient" it is.

All that we can usefully find out with faint speech

is whether an instrument has some gain setting that

will allow the wearer to pick up speech that is

as faint as he is likely to encounter in everyday

life. Most present-day hearing aids can do this.185

Although the above writers discussed the reliability

of comfortable loudness determinations without the support
of experimental findings, some researchers have investigated
this area systematically. Research findings involving the
reliability of comfortable loudness settings will be dis—
cussed under two main headings: intra-subject reliability
and inter-subject variability. Since the reliability of
comfortable loudness determinations seems to depend to some
extent upon various aspects of experimental design, the
following information will be given as necessary for clar-
ity: (l) the population (clinical or non-clinical) from
which subjects were drawn, (2) the method used to determine
comfort levels, and (3) the mode of stimulus presentation

(monaural, with an earphone, or binaural, in a sound field)

employed.

Intra-subject Reliability. Several researchers
have investigated the question of intra-subject reliability

of comfortable loudness determinations. Some of the studies

 

laslbid.

65

were carried out in a clinical setting and involved hard-
of—hearing patients; others were performed in a laboratory
setting and employed persons with normal hearing as subjects.
The research involving clinic patients will be discussed
first.

Davis e£_gl.186 reported the results of a project
that dealt with a ". . . theoretical analysis of the gener—
al problem of 'fitting' a hearing aid and a critical review
of several present and proposed 'fitting' procedures."187
These writers reported that the statistical analysis of
the agreement between two "comfort level” settings made
by 40 patients at Deshon General Hospital showed a standard
deviation of 3.7 dB in the aided threshold determined by
a test like Auditory Test No. 9.188 The consistency of
the two settings for comfort level (using recorded contin-
uous speech presented at about 60 dB above 0.0002 dyne/cmz,
or 40 dB above the normal threshold of intelligibility)
served as a guide to the reliance that was to be placed
on differences for that patient between measures made with

several instruments. However, the Aural Rehabilitation

officer who supervised the tests indicated that it was not

 

186Davis et al., "The Selection of Hearing Aids,"
pp. 85—115, 135-63.

187Ibid., p. 86.

1881bid., p. 139.

66

yet known how much difference was significant.189

Davis e£_gl. also studied the reliability of gain
control settings adjusted to comfort level for speech by
patients at Deshon General Hospital and at the Harvard
Psycho-Acoustic Laboratory. Presumably the same statis-
tical analysis described above was carried out. ”The re-
sults indicated a degree of variability, both for hard-of-
hearing patients and for normal subjects, so great as to
vitiate any differences which might reasonably be expected
to appear between instruments.“190

A third person to investigate whether the comfort
level method was precise enough to yield meaningful results
in hearing aid selection was Carhart.191 Specifically,
he considered the following questions:

(1) Do patients with hearing losses show enough reliabil-
ity with the comfort level method to justify its use

for adjusting the volume control of hearing aids?

(2) If so, what precautions in use and interpretation are
necessary?

(3) When aids are compared at the same comfort level, are
the aided thresholds obtained with them great enough
to be due to the performance the individual patient
achieves with the various hearing aids?l92

 

1891bid.

lgOIbido, pp. 139-400

191Carhart, “Volume Control Adjustment in Hearing

Aid Selection,” pp. 510-26.

1921bld., p. 5120

67

In this study 413 patients at Deshon General Hos-
pital served as subjects. To serve as a subject, a patient
needed a hearing aid and had to achieve a 40 dB (above nor-
mal threshold) comfort setting for monitored speech for at
least one hearing aid at less than full gain. Each subject
adjusted each of the three most promising hearing aids to
comfort level twice, and two independent aided thresholds
were obtained for each of these three hearing aids. Spondee
words were presented using the monitored live voice tech-
nique. The difference between these two thresholds (one
for each setting for each aid) was taken as an estimate of
the degree by which the volume control adjustments deviated
from each other.

Based on 1219 comparisons, Carhart found that the
mean difference between these two thresholds was .43 dB,
the median difference was .23 dB, and the standard devia-
tion was 3.91. Therefore, Carhart reasoned, the second
"comfort level threshold“ should not differ from the first
by more than 4 dB, which falls within the i 5 dB limits
of accuracy generally accepted in audiometry.

Thus, it seems fair to conclude that the comfort
level method has sufficient reliability to_justify

its use as a clinical means of setting volume con-
_trol on a_psychophysical basis.l93

 

Carhart also estimated the reliability of the com-

fort level method by comparing the test-retest correlation

 

193Ibid., pp. 515—16.

68

between the first and second measurement of "residual
loss," for which the coefficient of correlation was +0.87,
". . . which may be taken as evidence of high reliability
and as confirmation of the conclusion that the comfort
level method is clinically adequate."194 Carhart's find-
ings are contrary to those of Davis et a1. It is inter—
esting to note that Carhart also found that the position
of a hearing aid in the test sequence did not affect the
reliability of comfort level settings; however, the reli—
ability of the comfort level setting did vary significantly
from one speaker to another.195

Kavanagh, in the pilot study for his doctoral re-
search, also explored the intra—subject reliability of most
comfortable listening levels.196 He selected at random
five men and five women from his group of normal hearing
university students who had participated in his pilot study
and re-tested them two days later with an alternate list
of Spondee words. Listening was monaural, through an ear-
phone, and involved a subject's preferred ear. One of the
questions he sought to answer in his pilot study was, Can
subjects reliably choose their MCL level? Each subject

determined his MCL level (in dB re: normal audiometric

threshold) twice for both the ascending and the descending

 

194Ibid., p. 516.

l951bid., p. 525.

196Kavanagh, op. cit., pp. 31—35.

69

methods of presentation. Kavanagh found that the mean dif-
ference for both the ascending and the descending presenta—
tions of spondees did not exceed 5 dB.197 On the basis of
Hirsh's finding that standard clinical audiometric deter-
minations of threshold yield standard deviations of between
2.4 and 4.25 dB, Kavanagh concluded that under the partic-
ular conditions of his study ". . . the subjects were able
to report their most comfortable listening level for speech
as reliably as they could be expected for their pure tone
thresholds."198 Kavanagh's findings are in agreement with
those of Carhart discussed above.

Lezak, in a preliminary study conducted as part
of his doctoral research, investigated intra—subject reli-
ability of most comfortable loudness level settings, using
the psychophysical method of average error.199 Six males
with normal hearing served as subjects. Each subject
determined 22 most comfortable loudness levels during
a one and one-half hour period. Connected speech (re-
corded) was presented monaurally, through an earphone.
The range of most comfortable listening levels reported

by each subject over 22 trials was equal to or greater

than 16 dB re: normal audiometric threshold. Because this

 

197Ibid., p. 33.

l981bia., p. 35.

199Lezak, OE. Cite, ppo 148-510

7O

variability is considerably greater than the I 5 dB gener-
ally accepted for other audiometric measures, Lezak concluded
that subjects were unable to determine most comfortable
listening levels reliably. His findings were contrary to
those of Carhart and Kavanagh mentioned previously.

Still another researcher to investigate the reli—

200 He ob-

ability of comfort level settings was Loftiss.
tained soft comfortable, most comfortable, and loud comfort—
able loudness judgments for each of his twenty normal hear-
ing subjects during two test sessions separated by one week.
These judgments were obtained under four experimental con-
ditions, involving two psychophysical methods, two judg-
mental procedures, and two modes of stimulus presentation.
Connected speech (recorded) provided the stimulus materials.
Unfortunately, Loftiss did not report measures of intra—
subject reliability; however, inter-subject variability
was discussed.

The research studies just discussed dealt with
intra-subject reliability of most comfortable listening
level determinations for speech. Pollack studied the re-

201

liability of MCL settings for pure tones. The binaural

 

MCL level (in dB SPL) was obtained for 33 normal hearing

subjects for each of seven frequencies (125 to 8000 cps, in

 

200Loftiss, op. cit., pp. 80—91.

201Pollack, "Comfortable Listening Levels for Pure

Tones in Quiet and Noise," pp. 158-62.

71

octave steps) in two experimental sessions, separated by
a minimum of 24 hours. The variability of the set of
difference scores between the two binaural determinations
by each listener for each frequency served as a measure of
the consistency of a given listener's criterion of the MCL
level. Pollack found that the intranobserver test-retest
variability of the binaural MCL level increased as the fre-
quency of the tone increased, ranging from 7.8 dB for the
125 and 250 cps tones to 14.6 dB for the 8000 cps tone.202
According to Pollack,
this variability--either for the entire group, or
for experienced listeners alone--is about twice
(in terms of the standard deviation) as great as
the repeat variability of equal-loudness matches
made by experienced listeners.2
Pollack also found that "there was a consistent, but not
statistically significant, difference between the mean bi-
naural mcl levels of the first and second sessions."204

He found that listeners could determine MCL levels for pure

tones reliably.

Inter-subject Variability. The problem of the re-

 

liability of comfortable loudness determinations among lis-
teners has been investigated less extensively than has intra-

subject reliability. A possible explanation for this fact

 

202Ibid., p. 159.

2031bid.

204Ibid.

72

may be that many of the studies involving comfortable loud—
ness determinations have included listeners with various
types and degrees of hearing impairments, making reliabil—
ity information more difficult to obtain.

Pollack reported the results of an experiment in-
volving the determination of inter—subject variability of
comfortable loudness levels. He obtained the variability
(standard deviation) in decibels of most comfortable lis-
tening level judgments for seven pure tones (125 to 8000
cps, in octave steps) for 33 normal hearing listeners us—
ing both binaural and monaural listening.205 Pollack found
that the variability of the MCL levels from listener to
listener increased as the frequency of the pure tone in—
creased, at least to 4000 cps.206 According to Pollack,

the test-retest variability of the mcl (in terms
of the standard deviation) is about one—half that
of the variability among listeners for the middle
frequencies, but about the same magnitude at the
lower and the higher frequenc1es.2 7

Pollack presented the variability among observers
of various comfortable listening levels as a function of
frequency in the form of a graph, which appears in Figure

1°208

 

205Ibid., p. 160.

206Ibid.

2071bid.

208Ibid., Figure 3 (reproduced with the publisher's
permission).

73

 

 

 

 

._ Range \////// /‘*
__ // _fi

1 5 4"”"1 .1
_ /\“’ _

/
_. /’ .J
/ /

/ j, / / —

__ pper Limit/’/ Lwr __
lO // I / Limit

7' //
_ MCL / / .—

 

Standard Deviation in dB

 

 

 

 

 

 

 

 

 

 

125 250 500 1000 2000 4000 8000

Frequency in cps

Figure l. The variability among observers of com-
fortable listening levels as a function of frequency. Note
that the variability increases as the frequency is increased.
Note also the order of variabilities: lower limit of the
range of comfortable listening levels, most comfortable
listening level, and upper limit of the range of comfort—
able listening levels. The variability of the range as a
function of frequency follows the magnitude of the range
as a function of frequency.

Lezak, in a preliminary study conducted as part
of his doctoral research, also investigated inter-subject
variability of most comfortable loudness level settings,
using the psychophysical method of average error.210 Six

males with normal hearing served as subjects, each sub-

ject determining 22 most comfortable loudness levels

 

209Based on Pollack's description Of this graph,

ibid.

210Lezak, loc. cit.

74

during a one and one-half hour period. Connected speech

was presented monaurally, through an earphone. Lezak tabu—
lated his data and disCovered that one subject had chosen

an MCL level as low as 4 dB (re: normal audiometric thresh-
old) and another had chosen an MCL level as high as 59 dB,
when all of the determinations made by all of the subjects

were considered.211

Much variation oCcurred among listen-
ers' MCL choices within a given trial. The researcher com-
puted mean MCL scores in dB by various combinations of trials
(viz., two's, three's, and four's) and found results simi-
lar to those observed when the entire number of trials was
considered.212
The data were analyzed by using a Friedman two-way
analysis of variance technique, the results of which indi-
cated that comfortable loudness scores varied from subject
to subject. Lezak concluded that most comfortable loudness
levels could not be determined reliably.213
Another researcher who attacked the question of
inter-subject variability of comfortable loudness judgments

was Loftiss.214 Twenty normal hearing subjects determined

 

211Ibid., p. 151.

212Ibid.

213There are some discrepancies in Lezak's discus-
sion of his findings. For more information, see Lezak,
ibido , pp. 39, 148-510

214Loftiss, op. cit., pp. 32e33.

‘_ i

75

soft comfortable, most comfortable, and loud comfortable
loudness for connected speech under four experimental con-
ditions. Half of the listeners used an anchoring proced-
ure, and half the subjects used no anchoring procedure.
The same test was re-administered to each subject one week
later. Loftiss presented the mean scores in dB (re: nor-
mal audiometric threshold for speech) in the form of a

table, which is reproduced in Table 3.215

TABLE 3

MEAN SCORES FOR REPLICATIONS FOR 20 NORMAL HEARING SUBJECTS

 

 

 

EARPHONE LOUDSPEAKER
TREATMENT SOFT MOST LOUD SOFT MOST LOUD
TEST 36 dB 52 dB 67 as 33 dB 47 dB 60 dB
RBTBST 37 dB 52 dB 68 dB 33 dB 46 dB 60 dB

 

As Loftiss noted, and as is apparent from inspection of the

table, the effect of the replications upon the loudness

levels was negligible.216
When the test-retest scores were broken down accord-

ing to loudness levels, psychophysical methods, judgmental

procedures, and mode of stimulus presentation (earphone vs.

 

215
mission).
216

Ibid., p. 32 (reproduced with the author's per-

Ibid.

76

loudspeaker), the differences were also negligible.217

These scores are reproduced in Table 4.218

TABLE 4

TEST AND RETEST SCORES FOR PSYCHOPHYSICAL METHODS,
JUDGMENTAL PROCEDURES, AND EARPHONE AND
LOUDSPEAKER PRESENTATIONS‘

 

METHOD OF ADJUSTMENT METHOD OF LIMITS

NON- NON-
ANCHORING ANCHORING ANCHORING ANCHORING

EARPHONE
Soft Comfortable (36) 35 (38) 43 (34) 36 (36) 39
Most Comfortable (52) 51 (51) 56 (50) 52 (53) 53

Loud Comfortable (69) 69 (68) 66 (69) 69 (66) 66

 

LOUDSPEAKER
Soft Comfortable (31) 31 (34) 34 (31) 32 (32) 36
Most Comfortable (47) 47 (44) 47 (47) 47 (47) 47

Loud Comfortable (61) 61 (59) 6O (61) 61 (59) 59

 

“The numbers in parentheses represent scores ob-
tained during the first testing session. The remaining
numbers represent scores obtained upon retest, one week
later.

Contrary to the Lezak study, Loftiss found that

his subjects could determine the several loudness levels

 

2171bid.

218Ibid., Table 4, p. 33 (reproduced with the
author's permission).

77

reliably. Furthermore, Loftiss observed that the largest
differences were associated with the non-anchoring proced-
ure, which indicates that the anchoring procedure led to

more reliable judgments. However, the statistical analysis

of mean score differences did not support this observation.219
Loftiss also commented that the "variability in the comfort—
able loudness levels associated with judgmental procedures,
psychophysical methods and replications appeared to decrease
as the magnitude of the loudness judgments increased . . . "
(i.e., the Soft comfortable loudness judgments were the

most variable, the most comfortable loudness judgments were
less variable, and the loud comfortable loudness judgments
were the most precise).220 These findings were contrary

to those of Pollack for pure tones; he found that the upper
limit of the range of comfortable listening levels was the
most variable, the most comfortable listening level was

less variable, and the lower limit of the range of comfort-
able listening levels was the most precise--at least for

221

pure tones of 4000 cps or less. Loftiss also observed

that " . . . the loudspeaker scores were comparatively less

variable than the earphone scores."222

 

219

Ibid., p. 32.
220Ibid., p. 34.
221

Pollack, "Comfortable Listening Levels for Pure
Tones in Quiet and Noise,“ p. 160.

222Loftiss, loc. cit.

 

78
Factors Related to Comfortable Loudness
‘Judgments of Pure Tones and Speech

The results of various systematic investigations
involving comfortable loudness levels suggest that a vari-
ety of factors may be related to listeners' determinations
of these levels.. The factors that have been investigated
concerning their effects upon various measures of comfort—
able loudness are as follows: subjects, psychophysical
methods, stimulus materials, mode of stimulus presentation,
direction of stimulus presentation, judgmental procedures,
test duration, magnitude of the preceding judgment, back—
ground noise, speech reception threshold, and equal-loud-
ness contours. Each of these factors will be discussed

in a separate section below.

Subjects. Although most of the systematic investi—
gations involving comfortable loudness judgments have uti-
lized either persons with normal hearing or audiological
clinic patients, several researchers utilized both types
of subjects, thus permitting comparisons of comfortable
loudness judgments on this basis. Norman A. Watson used
listeners with normal hearing and listeners previously ex-
posed to gunfire and airplane noise, with resultant marked
hearing loss in some cases, to determine most comfortable

223

listening (MCL) levels for pure tones. Although the

 

223Norman A. Watson, op. cit.

79

only available reference to this study does not state that
the MCL levels of the two groups of subjects were compared,
such an assumption seems logical. The results of Watson's
study were not reported, however.

Lezak, in a study dealing with the upper limit of
comfortable loudness for speech, used 42 subjects with
various types and degrees of hearing loss and 14 normal
hearing subjects who had been referred for audiological

224 Lezak found that there was no significant dif-

testing.
ference in the upper limit of comfortable loudness (ULCL)
determined by subjects grouped according to hearing sensi-
tivity, functional diagnosis, audiometric pattern, recruit-
ment, and sex, although large differences in individual
scores did appear.225 .
Lezak also found that the ULCL increased as the
speech reception threshold (SRT) increased in some sensi—
tivity groups and decreased as the SRT increased in others,
thereby lengthening or Shortening the range of comfortable
listening levels, as defined by this researcher.226 Lezak
indicated that this question merited further study, using

a larger sample for each group, because of the suggestion

that sensitivity may be an important aspect in the

 

224Lezak, op. cit.

Ibid., pp. 93-94.
226

225

Ibid., pp. 73—91.

80

differentiation of subject performance.227

Kavanagh also investigated the relationship between
choice of most comfortable listening (MCL) level and sub—
ject group.228 The 182 subjects with normal hearing were
divided into six groups: college men, college women, girls,
mothers of girls, boys, and mothers of boys. The mean MCL
levels were computed (in dB re: normal audiometric thresh—
old) for each of these groups, for both the ascending and
descending methods of stimulus presentation. A difference
between means of i 5 dB was considered to be significant,
based upon Kavanagh's finding that subjects could report
their MCL levels for speech as reliably as they could re-
port their pure tone thresholds.229 Kavanagh found that
in general listeners do affect the resultant MCL levels.

His results are summarized below:230 (1) The college men
chose the highest average MCL levels for both the ascending
and the descending methods of presentation. (2) The girls
had the lowest average MCL levels for both methods of pre-
sentation. (3) The adults (college men and women and mothers)
chose higher mean MCL levels than the children for both

methods of presentation. (4) The mothers chose higher mean

MCL levels than did their children, contrary to the popular

 

2271bid., p. 138.

228Kavanagh, op. cit., pp. 62-71.

229Ibid., p. 35.

230Ibid., pp. 85—86.

81

belief that children prefer speech louder than do adults.231

Although Lezak found no significant differences in normal
and hard-of-hearing subjects' choices of the upper limit

of comfortable loudness (which was defined operationally

as 5 dB below a subject's uncomfortable loudness threshold),
Kavanagh found that differences in choice of MCL level oc-

curred among his various groups of subjects.

Psychophysical Methods. Two researchers have in-
vestigated the effects of different psychophysical proced-
ures upon listeners' determinations of comfort levels. In
two preliminary studies conducted as part of his doctoral
research, Lezak compared the method of average error and
the method of pair comparisons as procedures for determin-
ing most comfortable loudness (MCL) levels.232 He found
that the method of pair comparisons yielded a mean MCL lev-
el of 65 dB re: normal audiometric threshold, which was
" . . . considerably different from the individual mean
scores . . ." for the six subjects who used the method of
average error.233 This value is somewhat above the 40 or
50 dB MCL level usually cited for normal hearing persons;

however, Lezak's sample included persons with various types

and degrees of hearing loss, as well as persons with normal

 

231Ibid., p. 66.
232Lezak, o . cit., pp. 148-58.
233

Ibid., p. 157.

82

hearing. On the basis of this finding, Lezak concluded
that the method of pair comparisons yielded a more reliable
mean MCL level than did the method of average error.234
Because he noted that subjects could determine the
uncomfortable loudness threshold (ULT) more reliably than
the MCL level, Lezak also evaluated the method of average
error and the method of limits as procedures for determin-
ing this measure.235 A small sample E_technique was used
for comparing the ULT mean values for the method of aver-
age error and the method of limits. The E_value was 1.98,
which did not reach statistical significance at the 1% lev-
el.236 Lezak interpreted this finding to mean that the
two methods yielded no significant differences in the ob—
tained uncomfortable loudness thresholds.237 However, Lezak
chose the method of minimal changes for the major part of
his research concerning the upper limit of comfortable loud—
ness (which he derived from the uncomfortable loudness thresh-
old) because it could be measured quickly and could be inte—
238

grated easily into his battery of tests.

Loftiss also studied the relationship between

 

234Ibid.

23SIbid., pp. 153—58.

2361bid., p. 159.

2371bid.

238Ibid., p. 158.

83

comfortable loudness judgments and psychOphysical proced-
ures. He investigated the effects of the method of average
error and the method of minimal changes upon judgments of

239

the loudness of connected discourse. The data were sub—

jected to analysis of variance, and none of the F scores
associated with psychophysical procedures was significant.240
On the basis of these results, Loftiss concluded that with-
in the context of the two judgmental procedures these psy-
chophysical methods " . . . do not produce significant dif—
ferences in judgments of the loudness of conversational

speech."241

Stimulus Materials. The relationship of the type

 

of stimulus materials used and listeners' determinations

of comfort levels has been studied by three investigators.
Hedgecock, in an investigation of the relationships between
degree and type of hearing loss and the performance of hear—
ing aids, reported a mean comfort level of 74.9 1 .56 dB

SPL for pure tones (250 to 4000 cps, in octave steps), and

a mean comfort level for speech of 81.5 i 1.27 dB SPL.242

Hedgecock stated that the comfort level for speech was

 

239Loftiss, op. cit.
240Ibid., p. 25.
241Ibid., p. 60.

242

Hedgecock, "Prediction of the Efficiency of
Hearing Aids from the Audiograms," p. 102.

84

" . . . significantly higher than that for pure tones,"

based on a critical ratio of 4.7.243 A value equal to, or

greater than, 3.0 indicated significance.244

Kavanagh also investigated the relationship of stim-
ulus materials to most comfortable listening (MCL) levels.245
One hundred eighty-two children and adults with normal hear-
ing served as subjects. Each subject determined the MCL
level for ten speech samples, which were presented via the
ascending and descending methods of presentation. The stim-
ulus materials consisted of spondaic words (spoken by a
male and by a female), connected discourse (Fulton Lewis,
Jr., broadcasting on the "Housing Program“), excerpts from
six different types of radio broadcasts (a baseball game,
a news broadcast, a drama, etc.), a poem, and nonsense ma-
terial. All of the material had been pre-recorded.246

In order to investigate the effect of the ppeaker's
pe§_upon MCL levels, Kavanagh tabulated (l) the means and
standard deviations for the MCL levels (in dB re: normal
audiometric threshold) chosen by the 45 college men and
37 college women during the ascending and descending pre-
sentations of spondee words, (2) the difference between the
mean MCL levels chosen by the college men for the spondee

2431bid., p. 103.

244Ibid., p. 89.

245Kavanagh, op. cit., pp. 71-83.

246Ibid., p. 37.

85

words spoken by a male and by a female, (3) the difference
between the mean MCL levels chosen by the college women
for the spondee words spoken by a male and by a female,
and (4) the difference between the MCL levels selected by
the total group of students during the ascending and descend-
ing presentations of the spondee words.247 A difference
between means of 1 5 dB was considered to be significant.
Since no differences between the means exceeded 2.33 dB,
Kavanagh concluded that the MCL levels chosen by college
students in his study were not greatly influenced by the
sex of the speaker.248

Kavanagh investigated the effect of content of
speech samples upon MCL levels, too.249 Both standardized
and non-standardized speech samples were utilized in the
analysis of the data. He tabulated the ascending and de-
scending mean MCL levels that listeners chose for nine mean—
ingful speech samples and for nonsense words. These data
were reported for each of the six groups of subjects and
for the total group of 182 subjects. Again, a difference
between means of i 5 dB was considered to be significant.

The difference between ascending mean MCL levels

for the standardized speech samples (spondee words and con-

nected discourse from the Fulton Lewis, Jr., broadcast)

 

2471bid., p. 75.

248Ibid.

2491bido , ppo 77-830

86

for the total group of subjects was not significant. The

mean value for spondaic words was 38.38 dB (with a stand-

ard deviation of 9.66), and the mean value for connected
discourse was 38.83 dB (with a standard deviation of 9.00).250
Neither was the difference between descending mean MCL lev-
els for the standardized speech samples significant. The
mean value for spondaic words was 43.63 (S.D. = 11.68), and
the mean value for connected discourse was 43.98 (S.D. =

10.89).251

Kavanagh noted that the MCL levels for the de-
scending method of presentation for these two speech samples
are similar to the 40 dB value which Watson suggested as
being the "' . . . most commonly preferred most comfortable
listening level for normal ears."'252

With regard to the non-standardized speech samples,
some differences in mean MCL levels occurred. For example,
college men, who had higher mean MCL levels than any other
subject group, chose a mean ascending MCL level only 2.87
dB higher than did the college women for the nonsense words.253
However, the men chose a significantly higher mean ascending
MCL level for the excerpt from a baseball game than did the

women. The difference was 7.73 dB.254 Similar findings

 

25°1bid., p. 77.

251Ibid.

2521bid., p. 78.

253Ibid., p. 81.

254Ibid.

 

Tn:

87

were reported when the MCL levels of the men and the mothers
of girls were compared on the basis of these two speech
samples.255 These findings seem to indicate that MCL lev-
els depend to some extent upon the combination of speech
sample and sex of the listener.

Kavanagh's findings regarding the effect of stimu-
lus materials on MCL levels may be summarized as follows:
(1) The sex of the speaker did not greatly influence the
choice of MCL level. (2) The content of the speech samples
did affect the choice of MCL levels. The obtained mean
MCL levels for the standardized speech samples (spondees
and connected discourse) are considerably lower than the
MCL level for speech reported by Hedgecock (81.5 dB SPL),
for hard—of—hearing subjects.

A third researcher to investigate the relationship
between stimulus materials and comfort levels was Pollack.
In an investigation of most comfortable listening (MCL)
levels for pure tones in quiet and noise, Pollack found
that, for normal hearing subjects, determinations of mon—
aural most comfortable listening levels vary with the Egg:
guency of the tone.256 Using the means of listeners' MCL

determinations for each frequency, he plotted a most

 

255Ibid.

256Pollack, "Comfortable Listening Levels for Pure
Tones in Quiet and Noise," pp. 158-62.

88

comfortable listening level contour.257 He observed that

the shape of this contour had the same general shape as

the equal loudness contours258 of the middle intensity

range.259 He also noted, however, that there was a smaller

difference between the MCL levels at the middle and at the
upper frequencies and that there was a larger difference
between the MCL levels at the middle and at the lower fre—

quencies, than for the equal-loudness contours of the same

260 In the words of Pollack,

intensity range.
the direction of the differences from the equal-
loudness contour--less intensity at high frequen-
cies, more intensity at low frequencies--is pre-
sumably correlated with the finding that the higher
frequency tones are reported as more annoying than
lower frequency tones. 61

Mode of Stimulus Presentation. Pollack investigated
the difference between monaural and binaural most comfort—
able listening levels, both of which were determined with

earphones.262 He found that, at each of the frequencies

 

257Ibid., p. 159.

258Equal-loudness contours graphically present the
relation between loudness and frequency; i.e., they show at
what intensities tones of different frequencies are judged
equal in loudness to a standard (1000 cps tone). See
Stevens and Davis, op. cit., p. 123.

259Pollack, "Comfortable Listening Levels for Pure
Tones in Quiet and Noise," p. 159.

260Ibid.

261Ibid.

262Ibid., pp. 158—59.

89

tested, there was a statistically significant difference
between average monaural and binaural MCL levels for pure
tones determined by normal hearing subjects, the average
binaural MCL level being, on the average, significantly
lower than the average monaural MCL level.263

Loftiss, in an investigation designed to study the
effects of psychophysical methods and judgmental procedures
upon judgments of the loudness of connected speech, found
that, for each of the three loudness levels determined by
the subjects (soft comfortable, most comfortable, and loud
comfortable loudness), the average loudness levels deter—
mined monaurally, with an earphone, were 6 to 8 dB higher
than the average levels determined binaurally with a loud-
speaker in a sound field, for normal hearing subjects.264
Loftiss indicated that these differences are in line with

differences cited for other auditory measures.26S

Direction of Stimulus Presentation. Kavanagh in—
vestigated the relationship between most comfortable lis-
tening (MCL) levels and ascending and descending methods

266 One hundred eighty-two normal

of stimulus presentation.
hearing subjects determined MCL levels (in dB re: normal

audiometric threshold) for ten recorded speech samples which

 

2631bid., p. 159.
264 . .
Loftiss, op. c1t., p. 19.
2651bid.
266

Kavanagh, op. cit., pp. 60-62.

90

were randomly presented with ascending and descending in—
tensity. Kavanagh's results indicated that the mean de-
scending MCL level for all of the subjects for all of the
speech samples was significantly higher than the mean as-
cending MCL level.267 The mean descending MCL level was
48.66 dB and the mean ascending MCL level was 43.35 dB-—

268

a difference of 5.31 dB. A difference of I 5 dB was

considered to be significant.

Judgmental Procedures (Anchoring vs. Non—Anchoring_
Procedures). In a recent study Loftiss investigated the
effects of judgmental procedures upon judgments of the loud—
ness of connected speech.269 Twenty normal hearing persons
served as subjects, all of whom heard the speech presented
monaurally through an earphone and binaurally in a sound
field. Each subject used the psychophysical method of ad-
justment and the method of limits in determining three loud-
ness levels for speech: soft comfortable, most comfortable,
and loud comfortable loudness. Half of the subjects used
an anchoring procedure and the other half used a non—anchor-
ing procedure in making these judgments.

In the anchoring procedure, the subject always de-

termined the most comfortable loudness level first. He

 

2671bid., pp. 60, 62.

268Ibid.

269Loftiss, op. cit.

91

made the soft comfortable loudness judgment as the inten-
sity level of the speech was decreased from this level and
the loud comfortable loudness judgment as the intensity
level of the speech was increased from this level.270 In
the non-anchoring procedure, the subject was told which of
the three loudness level judgments he was to make, the speech
was presented at a randomly selected level, and the subject
made his judgment.271 In the anchoring procedure the most
comfortable loudness level was always determined first;
the other two loudness levels were assigned randomly. In
the non-anchoring procedure, all three loudness judgments
were assigned randomly.272 For each anchoring procedure,
a total of 12 determinations were made at each of two
test sessions: soft comfortable, most comfortable, and
loud comfortable loudness judgments were obtained with the
speech being presented via an earphone and in a sound field
and via the method of average error and the method of min-
imal changes.273

The data were subjected to analysis of variance.

The F—ratio was significant only for the earphone, soft

comfortable level and the loudspeaker, soft comfortable

 

270Ibid., pp. 12-13.
271Ibid., p. 13.
272Ibid., p. 14.

273

Ibid., p. 9.

92

level.274 Differences between the mean scores for most

comfortable loudness and loud-comfortable loudness, derived
from the earphone and loudspeaker modes of presentation,
were not statistically significant between the two judg-
mental procedures.275 Loftiss observed, however, that the
magnitude of differences associated with the three loudness
levels for the two judgmental procedures did not exceed 4
dB.276 Thus, "in terms of the decibel scale, it would seem
that the effect of the judgmental procedures on the magni—
tude of the obtained absolute levels was only slight."277
Since it appeared that judgmental procedures did
not affect loudness judgments, Loftiss decided to examine
the effect that the two judgmental procedures had upon the
precision of listener estimates obtained with the two psy-

278 He computed correlation coeffici-

chophysical methods.
ents and standard deviations for each psychophysical method
combined with each judgmental procedure. According to
Loftiss, the correlation coefficients indicated that the

rank order of judgments was Similar from test to re-tests

and, therefore, the reliability of judgments obtained under

 

274Ibid., p. 24.

275Ibid.

276Ibid., p. 35.

277Ibid.

278Ibid., p. 37.

93

these conditions could be evaluated by comparing the appro—
priate standard deviations.279 The results of these analy-
ses indicated that the reliability of listener judgments
associated with the two psychophysical methods was affected
by the judgmental procedure. The anchoring procedure, meth-
od of adjustment was the more precise technique.280

Loftiss found that the most comfortable loudness
judgments were least efficient in terms of reliability (the
standard deviations ranged from 7-9 dB over both judgmental
procedures and psychophysical methods). According to Loftiss,
this finding suggests that the most comfortable loudness
judgment should not be used as a single measure,281 although
it could serve as an anchor, as it did in his study, to
reduce the variability of soft comfortable and loud comfort—
able loudness judgments.282 In general, however, Loftiss
found that soft comfortable and loud comfortable loudness
judgments approximated an acceptable clinical standard (1

5 dB).283

Test Duration. Loftiss also investigated the ques-

 

tion regarding whether the number of presentations within

 

279Ibid.

28°Ibid., p. 38.

281Ibid.

282Ibid., p. 62.

2831bid., p. 40.

94

judgmental (anchoring vs. non-anchoring) procedures produces
Significant differences in judgments of comfortable loud-
ness.284 In order to evaluate this question, he recorded
the order of each loudspeaker and earphone judgment sequence
(each of which involved soft comfortable, most comfortable,
and loud comfortable loudness determinations) as being the
first, second, third, or fourth presentation within each
test session. Decibel scores for each of the three loud-
ness judgments for presentations one and two were combined
and were compared with the decibel scores for each of the
three loudness judgments for presentations three and four.
Loftiss applied tests of Significance (involving mean scores)
for orders of presentation separately for the anchoring

and non-anchoring procedures.285 He found that, for the
separate loudness levels, the means were within 1 1.6 dB

of each other.286 The results of these tests indicated

that " . . . the means of presentations 3 and 4 were neither
consistently higher nor consistently lower than the means

of presentations 1 and 2."287 Loftiss pointed out that
these findings are in contrast to those of other loudness

experiments, which have Shown that as the duration of the

test increases, judgments become more susceptible to whatever

284Ibid., p. 42.
2851bid.
286Ibid., p. 43.
287

Ibid.

95

biases are present.288

Magnitude of the Preceding Judgment. Loftiss in-

 

vestigated whether the relative loudness or softness (mag-
nitude) of the preceding judgment within anchoring or non—
anchoring procedures produced significant differences in
judgments of the loudness of speech.289 He recorded the
order of each soft comfortable, most comfortable, and loud
comfortable loudness judgment as the first, second, or third
judgment within a sequence. Decibel scores for each loud—
ness level were combined according to the preceding loud-
ness judgments (i.e., all soft comfortable judgments pre-
ceded by loud comfortable judgments, etc.). Statistical
analyses were made involving mean values in terms of the
two judgmental procedures and for each loudness judgment.
Loftiss concluded, on the basis of his findings, that the
relative magnitude of the preceding judgment within judg-
mental procedures does not produce significant differences
in judgments of the loudness of conversational speech.290
Loftiss also examined the effect of magnitude (rel—
ative loudness or softness) of the preceding judgment on

the reliability of earphone-loudspeaker judgments.291

 

288Ibid., p. 45.

289Ibid., pp. 48—50.

290Ibid., p. 60.

2911bid., pp. 50-58.

96

Statistical tests involving comparison of the means were
administered. On the basis of his findings, Loftiss con-
cluded that loudness evaluations of connected speech, using
the loudspeaker, anchoring technique, should be made in
this order: most comfortable, loud comfortable, and soft

comfortable.292

Background Noise. The only researcher to study

 

the effect of background noise upon the most comfortable
listening level, the range of comfortable listening levels,
and the upper and lower limits of this range was Pollack.293
Using normal hearing subjects, Pollack carried out two se-
ries of determinations involving the above levels for pure
tones.

In the first series of determinations, 27 subjects
determined the most comfortable listening level and the
upper and lower limits of the range of comfortable listen-
ing levels for a 1000 cps tone in the presence of 35, 65,
and 95 dB SPL. Pollack's findings are presented below.

In general, as the background noise level is in-
creased, (l) the lower and upper limits of the range
of comfortable listening levels and the mcl level

is increased . . . ; (2) the range of comfortable
listening levels is decreased; and (3) the varia-
bility of the upper or lower limits of this range,

of the mcl level, and of the range itself, is de-
creased. . . . The range is decreased at higher

 

2921bid., p. 61.

293Pollack, "Comfortable Listening Levels for Pure
Tones in Quiet and Noise,“ pp. 158-62.

1" ‘\ Vi (I Is A t I c . 15‘ u ’4. 1 T 9514

L... .0 0 LL n 3 at C
d 5 it .L C a l 0 .l at d .l S
R 4 . a O D. e .u -. l O t .r. n1. L1. .3
«a 11 Au n3 .3 S L b n Au :s C «Q ,.N «b

97

noise levels primarily because of the greater in-

crease of the lower limit relative to that of the

upper limit. Specifically, as the noise level is

increased over a range of 60 db, the lower limit

is increased only 35 db while the upper limit is

increased only about 18 db.294

In the second series of determinations, seven trained

listeners determined the most comfortable listening level
and the upper and lower limits of the range of comfortable
listening levels for seven pure tones (125 to 8000 cps, in
octave steps) in quiet and with 35, 55, 75, 95, and 115 dB
SPL white noise. According to Pollack, the results of the
second series of determinations confirmed the general re-
sults of the first series of determinations.295 In addi-
tion, it was found that the changes occurring with increased
noise background were smaller at the lower frequencies than

at the middle and higher frequencies.296

Speech Reception Threshold. Kavanagh investigated
the relationship between the speech reception threshold
and the most comfortable listening level for recorded speech,
which was determined for both ascending and descending meth—
297

ods of stimulus presentation. Forty-five college men

and 37 college women with normal hearing participated

 

294Ibid., p. 161.

295Ibid.

2961bid.

297Kavanagh, op. cit., pp. 57-58.

98

in this aspect of the study. Pearson product—moment coef-
ficients of correlation were computed for the speech recep-
tion thresholds and most comfortable listening levels (in

dB re: normal audiometric threshold) for each group of
subjects for each of the ten speech samples. For each group
of listeners, the MCL levels for the ascending and descend-
ing presentations were compared separately with the listen—
ers' Speech reception thresholds.298 The correlation be-
tween speech reception thresholds and MCL levels for each

of the ten speech samples approached zero, indicating little
relationship between the MCL levels and the speech recep-
tion thresholds.299 Kavanagh concluded that, for these sub-
jects, the choice of the most comfortable listening level

was not dependent upon their speech reception threshold.300

Clinical and Experimental Applications of
Comfortable Loudness Judgments
This section deals with various ways of utilizing
comfortable loudness measurements in the clinical or lab-
oratory setting. In hearing aid selection, comfort levels
have been used most often " . . . to estimate the amount
of amplification required for adequate use of a hearing

aid . . ." and " . . . to study the effects of auditory

 

298Ibid.

2991bid., p. 58.

3001bid., p. 85.

H.~t
I2U|

99

recruitment."301 Comfort levels have also been used in

distinguishing conductive and sensori-neural hearing losses

and in loudness tracking studies.

\

Hearing Aid Selection. Most comfortable listening
levels may be used in evaluating the sensitivity of various
hearing aids. In 1946 Davis e£_gl. advocated that the pa-
tient adjust the volume control of each hearing aid to his
most comfortable listening level for conversational speech,
and that the aids be compared on the basis of his "gain for
speech" (i.e., for each hearing aid, the patient's aided
SRT is compared with his unaided SRT) or on the basis of
his "residual loss for speech" (i.e., the speech reception
thresholds that several aids yield are compared among them-

02

selves)? The instrument yielding the lowest SRT (the SRT

nearest normal) and the smallest "residual loss" was taken
as the most effective instrument for that patient.303
In 1946 Carhart advocated the same procedures, uti-
lizing the comfortable listening level for speech, as did
Davis e£_gl. for comparing the sensitivity of various hear—

ing aids. He added that the patient's "residual loss for

speech" or his "effective gain" for speech could also be

 

301Loftiss, op. cit., p. 5.

302Davis et al., "The Selection of Hearing Aids,"
pp. 135-38.
303

Ibid., p. 138.

 

Al

AU

,.
he

Vs

uh.

h

v.

100

determined with the aid set at full volume, as well as with
the aid adjusted to comfort level.304

In the same study, Carhart also indicated that hear-
ing aids should be adjusted to a most comfortable listening
level for speech in order (1) to make comparisons among
them on the basis of tolerance limits of the various aids,
(2) to determine the relative efficiency of the several
hearing aids in noise, and (3) to determine a patient's
discrimination ability as a function of various hearing

aids.305 Davis et al. also believed that the comfort level

could be used to obtain these measures.306

Prior to obtaining various types of audiometric
data using five different frequency patterns of electrical
amplification with the Master Hearing Aid, Davis ggpgl,
required their subjects to select the most comfortable loud-
ness level for connected speech using the "Flat" pattern
of amplification.307

L. A. Watson also favored the use of the MCL level

in hearing aid selection. He stated that

the MCL gives an important indication of the loss
of an individual for speech at the level at which

 

304Carhart, ”Tests for the Selection of Hearing
Aids,“ pp- 781-830

3051bid., pp. 783—88.

306Davis et al., "The Selection of Hearing Aids,"
pp. 140-52.

307

Davis et al., Hearing_Aids: An Experimental
Stugy of Design Objectives, pp. l7-22.

 

101

he should do his hearing. It is an indication of
the amount of gain or benefit which should be ro-
vided either by a hearing aid or an operation. 08
Newby has also indicated that the MCL provides information
regarding the level of amplification that the patient will
hear most comfortably.309
According to Kavanagh, Utley stated that "' . . . the
MCL will indicate how powerful a fitting is needed, which
model, which receiver and what "B" battery voltage. It
also helps in choosing the ear to fit, and in selecting
the type of ear mold to be used."'310
By the use of pure tones as stimuli, most comfort—
able listening level contours have also been utilized in
hearing aid evaluation. With the level of a 1000 cps tone
as a reference, the patient determines the most comfortable
listening level of various other pure tones. The result
is a most comfortable listening level contour. Watson and
Knudsen fitted a hearing aid to an equal-loudness contour
311

at the level of comfortable listening. Davis et al.,

however, found no relation of the best hearing aid to the

 

308L. A. Watson, op. cit., p. 18.

309Newby, op. cit., p. 132.

310Kavanagh, op. cit., p. 10.

311N. A. Watson and V. O. Knudsen, "Selective Am-
plification in Hearing Aids," The Journal of the Acoustical
Society of America, XI, No. 4 (April, 1940), pp. 406-19.

 

£1:
.4 2
(I)
C)

V

C. DARK
L

(“be r.
IECIL}

higne

'(I

:10

102

patient's audiogram or to his equal-loudness contour.312

Detection of Recruitment. Watson and Tolan suggest

the use of comfortable and uncomfortable loudness judgments

to detect the presence of recruitment.313 These writers

also suggest that the person with recruitment will have

a shorter range of comfortable listening levels than will

the non-recruiting patient.314 Likewise, the person with

recruitment will be able to understand speech only at much

higher levels than would other patients.315

Newby also advocates using measures of comfortable

loudness as "a simple, although not very reliable, means
of determining the presence of recruitment. . . ."316 The
patient's most comfortable loudness level and his uncomfort-
able loudness level are obtained, along with his audiometric
threshold, for pure tones of various frequencies. Accord-
ing to Newby,

these points will tend to be relatively close to-

gether for the patient who presents recruitment,

whereas with the patient with no recruitment the

points may be spread far apart. As a matter of

fact, the patient who has no recruitment should
be able to tolerate the full intensity at each

 

312Davis et al., “The Selection of Hearing Aids,"
p. 102.
313 .
Watson and Tolan, op. c1t., p. 85.
314Ibid.
315Ibid.
316

Newby, op. cit., p. 166.

§

AV

.fi.‘

3‘]

Or

a“
\ ...

y

.\

103
frequency without experiencing physiological dis-
comfort. l7
Newby also states that the range from threshold
to MCL to UCL may vary as a function of frequency. He be-
lieves that the method of comparing threshold, MCL, and
UCL is useful only as a gross indication of whether or not
recruitment is present because the examiner must rely on
his own experience in interpreting the results of these
tests and because it is difficult for patients to determine
most comfortable loudness levels for pure tones reliably.318
Similar criticisms of the most comfortable loudness
level technique were voiced by Bangs and Mullins, although
they also mentioned several factors in favor of this meth-
od.319 These writers compared four different methods for
determining recruitment: the range of comfortable loudness
method, the Békésy test, the Lucher-Zwislocki method, and
the Denes—Naunton method. Bangs and Mullins concluded,

on the basis of their findings, that the range of comfort-

able loudness method was superior.

Detection of Conductive vs. Sensori-neural Impair—

ments. Pollack reported some preliminary results regarding

 

3171bid., p. 169.

318Ibid., pp. 169-70.

319Jack L. Bangs and Cecil J. Mullins, “Recruit-
ment Testing in Hearing and Its Implications," A.M.A. Ar—
chives of Otolaryngology, LVIII, No. 5 (November, 1953),
pp. 582—92.

'(J

.‘n
\nv‘

‘41

A.
Q ,
V Q

104

the notion of using the most comfortable listening level

as a means of differentiating between conductive and sensori-
neural hearing losses.320 For several older subjects who
exhibited sensori-neural types of impairments, MCL levels
were obtained that resembled the decreased range of comfort-
able listening levels for subjects with normal hearing de-
termined in the presence of noise. He found that the nar—
rowed range of comfortable listening levels among these
patients was mainly a result of the elevation of the lower
limit of the range of comfortable listening levels. Accord—
ing to Pollack, a patient with a conductive hearing impair—
ment would exhibit both elevated lower and upper limits of
the range of comfortable listening levels, but the range

would be similar to that of persons with normal hearing.321

Research. Several researchers have utilized the
concept of a most comfortable loudness level in research.
For example, Rintelmann and Carhart investigated the lev-
els at which 12 normal hearing subjects traced loudness

configurations for interrupted and continuous tonal stim-

 

uli.322 Békésy audiometry was utilized, and listening was
320Pollack, "Comfortable Listening Levels for Pure
Tones in Quiet and Noise," p. 162.
321Ibid.
322

William F. Rintelmann and Raymond Carhart,
"Loudness Tracking by Normal Hearers via Bekesy Audiometer,"
Journal of Speech and Hearigg_Research, VII, No. 1 (March,
1964), pp. 79-93.

8310
u“.

q
yu
8

n5

WM

' S

the
0
'4
4“
1

105

monaural. The subjects monitored the tracking by an in-
ternal standard involving a comfortable loudness level and
by means of the recalled loudness of a 1000 cps reference
tone heard at the beginning of the task. For both these
tasks, the writers found that a Significantly greater in-
tensity was required from interrupted than from continuous
tones to accomplish the tracking.

Melnick carried out a similar study involving the
same types of tasks. Each subject was required (1) to track
his threshold for the interrupted signal and the continuous
signal (of a fixed frequency); (2) to make two judgments
of the most comfortable loudness level for each stimulus;
and (3) to make six loudness matches, three for the inter—
rupted Signal and three for the continuous signal, using as
the standard intensity the intensity that he gave as his
323

most comfortable loudness level for the continuous tone.

His results agreed with those of Rintelmann and Carhart.

Summary

In this chapter several aspects of loudness that
have attracted the attention of researchers were mentioned:
(1) the definition of loudness, (2) the calculation of the

loudness of various types of acoustic stimuli by means of

 

.323William Melnick, "Comfort Level and Loudness

Matching for Continuous and Interrupted Signals“ (unpub—
lished paper presented at the 1965 convention of the Amer-
ican Speech and Hearing Association).

 

119‘;

L)

"h

106

mathematical formulas, (3) the utilization of psychophys—
ical techniques for determining the loudness of sounds,
(4) the determination of the loudness of Specific kinds
of auditory stimuli, (5) the effect of certain factors
upon loudness determinations, and (6) the determination
of "equal—loudness contours."

For the most part, however, attention was focused
upon various factors related to the question of how listen-
ers evaluate the loudness of auditory stimuli (pure tones
and speech) in terms of comfortableness. Comfortable lis-
tening levels for pure tones and speech, determined in quiet
or in noise, were discussed; and various psychophysical
methods utilized in obtaining these levels were examined.
The reliability with which listeners can determine comfort-‘
able listening levels was reviewed, and various factors
that may affect comfortable loudness judgments were analyzed.
Some clinical and research applications of comfortable lis-

tening levels were discussed.

CHAPTER III

SUBJECTS, MATERIAL, INSTRUMENTATION,
. AND PROCEDURES
As was mentioned previously, the psychophysical

method of successive categories (rating scale method) has
not been utilized in studies involving comfortable loudness,
although this procedure has been employed in studies deal-
ing with loudness in general. Since this psychophysical
method had been used in loudness studies, it was believed
that this technique merited investigation as a means of
obtaining judgments of auditory stimuli in terms of listen-
er comfort. The purpose of this study was to investigate
the effects of intensity level of connected discourse and
acoustical background condition upon listeners' evaluations

of the loudness of connected discourse.

Subjects
Twenty-one undergraduate students at Michigan State
University served as subjects for this experiment. There
were 17 females and four males, ranging in age from 17 to
22 years. The mean age was 19.29 years; the median age

was 19 years.

107

108

Material

 

Connected discourse served as the stimulus material.
The connected discourse was taken from a commercially avail-
able disc recording of Fulton Lewis, Jr., broadcasting on
the "Housing Problem.“ This disc is used in hearing clinics
for purposes of testing detection threshold and also for
Obtaining the MCL level.

It Should be noted that, when connected discourse
serves as the stimulus material, most comfortable loudness
judgments are related to the intelligibility of the connected
discourse, as well as to the intensity level at which it
is presented. [See the definition of "comfortable loudness"
in Chapter 1.] Hereafter, listeners' evaluations of the
loudness of connected discourse in terms of comfortableness

will be discussed with this relationship in mind.

Instrumentation

 

The following instruments were used for recording
the stimuli and for presenting the stimuli to the subjects:

1. Low noise magnetic recording tape (3M, Type 202) (for
recording the experimental stimuli).

2. Magnetic recording tape (3M, Type 111) (for reproduc—
ing speech babble).

3. Tape recorder (Ampex, Model 601).

4. Tape recorder (Ampex, Model PR-lO).

5. Tape recorder (Wollensak, Model T1500).
6. Phonograph (Newcomb, Model AVlO).

7. Mixer (Ampex, Model MX35).

109

8. Two line amplifiers (Ampex, Model 620).

9. White noise generator (Grason-Stadler, Model 455B).
10. Filter (Allison, Model 25).

11. Sound level meter (Brfiel and Kjaer, Type 2203).

12. Artificial ear (Brfiel and Kjaer, Type 4152).

13. Twelve matched pairs of earphones (Telephonics, Model
TDH39).

14. Disc recording of connected discourse (Fulton Lewis,

Jr., broadcasting on the "Housing Problem,” recorded
by Technisonic Studios, Inc., St. Louis 17, Missouri).

Procedures

 

Subject Selection. The 21 undergraduate stu-
dents who served as subjects were randomly selected from
a voice and articulation class which had been discussing
subjective and objective methods of evaluating the various
aspects of speech. Two criteria were employed in the se—
lection of subjects: (1) Each subject had normal hearing;
i.e., students with known hearing impairments were not se-
lected as subjects. Since hearing screening tests were
administered to all students entering Michigan State Uni-
versity, it was presumed that a student would be aware of
a hearing impairment. (2) None of the selected subjects
had special training in the experimental methods used in
this study.

Just prior to participation, the class instructor
indicated that the group would be involved in a study uti-

lizing specialized methods and equipment as a means of

110

obtaining more objective information about one of the

parameters of speech. It was explained to the group that

they had been chosen to participate in the study in order
that they might have the experience of evaluating the
lxoudness of another person's speech presented at differ-
euut intensity levels under different acoustical background
(2 onditions .

The experimental materials were presented to a
‘tLDtal of 22 students, with 12 in the first group and ten

:irl the second group. One student's responses were not

.illClUded in the analysis of the data because she was ob—

served to be copying another person's responses to the

test stimuli .

Selection and Production of Stimuli. Connected
dtisu:ourse at various intensity levels and under different
Ibéicflcground conditions was presented to the subjects. (The
C<>runected discourse was taken from a commercially avail-
alDlxe disc recording of Fulton Lewis, Jr., broadcasting
<3r1 ‘the "Housing Problem," as was mentioned previously.)
llt: was presented at 13 different intensity levels: 45, 50,
55. 60, 65, 70, 75, 8o, 85, 90, 95, 100, and 105 dB SPL.
T1'lese intensity levels were selected because they repre-

Sel’lted a wide range of intensities, and because they rep-
:Sisented intensity levels at, and somewhat above and some-'

what below, the intensity level of normal conversational

speech. Under the four different background conditions,

111

connected discourse was presented in quiet and in the pres-
ence of 70 dB SPL wide band white noise, narrow band white
noise, and speech babble. These four conditions were se-
lected because (1) they Simulated the kinds of situations
in real life in which persons find themselves trying to
communicate, (2) it was believed that they were dissimilar
enough so that any differences in the subjects' evaluations
of the loudness of connected discourse under these differ-
ent conditions would not be obscured, and (3) they could
be produced readily in the laboratory.

The stimulus materials consisted of 104 seven-second
segments of connected discourse recorded on low noise mag—
netic tape in quiet and in the presence of 70 dB SPL wide
band white noise,324 70 dB SPL narrow band white noise,3:25
and 70 dB SPL speech babble. In a preliminary trial in-
volving three listeners, it was found that a seven-second
Stimulus allowed a subject ample time to listen to the stim-
u1148, evaluate the loudness of the connected discourse,
and record his response on paper.

The 104 stimuli used in this experiment were pro-
duced in the following manner. The calibration tone from
\

324Wide band white noise consisted of all of the

frecIllerlcies in the audible range (20 to 20,000 cps) pres-
ent at about the same intensity level.

. 325Narrow band white noise was produced by filter-
:ng Wide band white noise to remove those frequencies below
and above 2040 cps, leaving those frequencies which

are COntained in the "speech range."

112

time disc recording of Fulton Lewis, Jr., was reproduced
(3r; Channel A of the low noise magnetic tape at an intensity
larvel of 70 dB SPL.326 The purpose of reproducing the cali—
buration tone on the stimulus tape was to provide a refer-
eruze point for calibrating the playback system. The cali—
brwation of the playback system was checked before the stim—
ullls tape was presented to each group of subjects.

After the calibration tone from the Fulton Lewis,
.Ir. , recording had been reproduced on Channel A of the low
:nojgse magnetic tape, twenty-second segments of connected
ditnzourse at intensity levels of 45 to 105 dB SPL were
recxorded in 5 dB intervals on Channel A of the low noise
Inagrnetic tape, making a total of 13 intensity levels.
Prdxlr to the recording of each twenty-second segment of
conruacted discourse at a given intensity level, the cali-
braizion tone from the Fulton Lewis, Jr., recording was
Pla)%3d.while the Channel A record level of the tape recorder
(Ammex, Model PR—lO) was adjusted until the desired output
at true earphones could be read from the sound level meter
(BrHEEL and Kjaer, Type 2203). The desired stimulus was

rePrOduced without changing the record level. A block

 

 

326All calibrations for this study were made di-

reEtIQI from the acoustic output of the earphones using the
BrHel and Kjaer Artificial Ear and the linear scale of the
Brue1_ and Kjaer Sound Level Meter. For example, a reading
Of 70 dB SPL on the Brfiel and Kjaer Sound Level Meter indi-
ggied that the acoustic output at the earphones was 70 dB

113

cjjgagram of the instrumentation used to measure the sound
Furessure level of the connected discourse and to reproduce
time connected discourse on the stimulus tape is shown in
Figure 2.

This procedure of reproducing the 13 twenty-
senzond segments of connected discourse on Channel A of
time low noise magnetic tape was repeated three more times,
yixalding four reproductions of the 13 twenty-second seg-
merrts of connected discourse--one for each of the four dif-
ferxent background conditions.327 Ten seconds of silence
segxarated each reproduction from the one preceding it.

A 1000 cps calibration tone of 70 dB SPL was repro-
ducxad on Channel B of the low noise tape. Wide band white
noigse was then reproduced on Channel B at an intensity
1th£1 of 70 dB SPL. A white noise generator (Grason-
StarLler, Model 455B) was the source of the wide band white
fulises [which consisted of all of the frequencies in the
andilole range (20 to 20,000 cps) present at about the same
inteulsity level]. The wide band white noise was recorded
on Channel B of the magnetic tape in the same relative
positxion that the first reproduction of the 13 twenty-
secondsegments of connected discourse had been recorded

on Ckuannel A. A block diagram of the instrumentation used

327The four different background conditions con-
S¥Stfixi of connected discourse being presented in quiet and
wlfkl 70 dB SPL wide band white noise, narrow band white
noise, and speech babble.

114

.mmmu msaseaum on» so mmusoumeo
Omuumccou may mosuoudmu Op pom onusoumav omuumccou 03» mo Hw>ma whammmum OEDOm
on“ whammme on OOMOHQEO coeumucmesuumcfl on» LG Emummflo xuoam .m musmmm

 

 

 

 

 

 

 

 

 

 

 

 

 

«new:
qm>mq ozoom
_ Tm<moozomm
m<m
q<HOHmHem<
_mmmHqudz< _ _ mmtz _ _ mmomoomm
_ mZHq _ boo «.co _ — <.co F md<e

 

 

 

mzommm<m

m .

$4

115

to measure the sound pressure level of the wide band white
noise and to reproduce the wide band white noise on the
stimulus tape appears in Figure 3.

Narrow band white noise was then reproduced on Chan—
nel B of the low noise magnetic tape at an intensity level
of 70 dB SPL. (Narrow band white noise was reproduced by
filtering wide band white noise to remove those frequencies
below 510 and above 2040 cps, leaving those frequencies
which are contained in the "speech range.") A filter (Alli-
son, Model 25) was used, the roll-off characteristic of
which was 30 dB per octave. The narrow band white noise
was recorded on Channel B of the magnetic tape in the same
relative position that the second reproduction of the 13
twenty-second segments of connected discourse had been
recorded on Channel A. A block diagram of the instrumenta-
tion used to measure the sound pressure level of the narrow
band white noise and reproduce the narrow band white noise
01’1 the stimulus tape is presented in Figure 4.

By the use of a tape recorder (Ampex, Model 601)
for playback, speech babble was then reproduced on Channel
B of the low noise magnetic tape at an intensity level of
70 dB SPL. A block diagram of the instrumentation used
to measure the sound pressure level of the speech babble
and to reproduce the speech babble on the stimulus tape
is Shown in Figure 5. The speech babble was recorded on

ChElnnel B of the magnetic tape in the same relative position

116

.mmmu mSHSEHym may no mmwoc muflzz pawn moflz
map musoouemu on com wwwoc muflsz pawn MUHB mnu mo Hm>ma whammmua UGSOm mnu
mudmmme ou omhoamem coaum»cmesuumcfl on» mo Smummao xuoam .m musmflm

 

 

mmBmZ moadmmzmw
Am>m4 QZDOm mmHOZ MBHIB

 

 

 

 

 

 

 

 

Ch.B

m<m
A<HUHmHBm<

 

 

 

Wide Band
White Noise

 

 

 

 

fiK mmmHqumz< mmamoumm
mZHq use m.co mm<a

 

 

 

 

 

 

 

 

mzommm<m

117

.mamv mSHSEHum mnp co mmHo: muflnz pawn zouumc
mcu musoouamu 0» van mmfloc mpanz pawn zouumc mcp mo Hm>ma mudmmmum oddom mcu
madmmme ou owhoamem coaumucmfisuumCH on“ mo Emummao xuon .w wudmflm

 

mmemz

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mWHO m H.
qm>mq ozpom nmmm wwwmz
moe<mmzmw . mmaqu
mmHoz mesz
mnw
m<m a.i
A<HUHmHam< mum
W
08
.im
at“
NW
mmmHqumz< _ mmomoomm
mZHq use m.zo mmtz m.:o _ mm<e

 

 

 

 

 

 

 

 

mzommm<m

.mmmu msHsEflum mnu no Manama
summam on» muzoouemu on cam Manama summam 02¢ mo Hm>ma mHSmmmuQ bosom map
whammme 0p om>0HQEm coapmucmesuumca may mo Emummwo xuoam .m wusmflm

 

 

118

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mmamz
Am>mq QZDOm
mmnmoumm
mm<B
m<m
A<HUHhHBm<
B
h
C
mmmHthm2< MMXHS mmnmoumm
MZHA #30 m.£U m.£U mm<H

 

 

 

 

 

 

 

 

 

mzommm<m

rt;

KM

v

fi‘

L-

bCi

p;
U)

I .

*‘J

119

that the third reproduction of the 13 twenty-second seg-
ments of connected discourse had been recorded on Channel
A. Speech babble was obtained in the following manner.
The record level of a tape recorder (Wollensak, Model
T1500) was adjusted so that a 1000 cps calibration tone
was reproduced on magnetic tape at an intensity level of
70 dB SPL and the tape recorder was placed in a classroom
of 18 students. They were instructed to choose at random
all together,

a passage from any book and to read aloud,

until they were instructed to stop.

The interval on Channel B that corresponded to
the space on Channel A where the fourth reproduction of
t:r1e:13 twenty-second segments of connected discourse had
been recorded was left blank.

Prior to the experiment, it was decided that
CDrLly the middle 14 seconds of each twenty-second stimu-
lers would be used in the preparation of the stimulus
t'—‘<il1:>e,328 in order to eliminate any extraneous factors
aS-sociated with turning the record switch on and off.
UDEHE middle 14 seconds of each stimulus was identified
dLlringplayback. Bach l4-second segment was divided into
tMIC) seven-second segments. Thus each of the 52 combina-

tions of intensity level of connected discourse and background

\

 

328A stimulus consisted of connected discourse
Git: one of the 13 intensity levels presented in quiet and

C

in the presence of 70 dB SPL wide band white noise, narrow
and white noise, and speech babble.

120

condition was presented twice, making a total of 104 stim-
uli. A number between 1 and 104 was assigned arbitrarily
to each seven-second recorded stimulus, and these numbers

were marked on the back of each corresponding strip of mag-
ne tic tape.
The 104 seven—second stimuli were randomly selected,

spliced together, and presented to the subjects. The in-

structions to the subjects and the number of each stimulus
(corresponding to its position on the stimulus tape) were

also reproduced on magnetic tape, identified during play-

back, and spliced into the stimulus tape along with the

experimental stimuli.
In a preliminary trial it was found that one sec-
ond of silence between the number of the stimulus and the
Stimulus itself, and two seconds of silence between a stim-
ulus and the number of the stimulus following it, allowed
eI'lough time for a subject to record his responses. These

time intervals remained constant throughout the stimulus

ta~pe. (A more detailed description of the stimulus tape

is presented in Appendix A.)

Manner of Presentation. All testing was done in

the Speech and Hearing Science Laboratory of Michigan State

University. The two groups of subjects were tested during

twO consecutive periods on the same day under the same ex—

perimental conditions. Each subject was seated at one

of the 12 listening stations. The subjects were told that

L.

ah

\,

w

121

they would hear the taped instructions, examples of the

test stimuli, and the experimental stimuli binaurally through
earphones. They were instructed to put on the earphones

so that they fit comfortably but snugly. Assistance was

given as necessary.

The subjects heard the following tape-recorded in-

s tructions:

You are about to hear a news broadcast. Here
are two samples of the speech to which you should
respond. The first one you may not hear. [Con-
nected discourse was presented in quiet at 45 dB
SPL.] Here is the second one. [Connected discourse
was presented in quiet at 105 dB SPL.] Sometimes
you will hear the news broadcast by itself, and
sometimes you will hear the news broadcast in the
presence of background noise, which will in some
cases be speech babble. For each of the 104 items
on your paper, I want you to put a number, 1 through
5, in the space provided. Use 5 if the speech was
too loud; use 4 if the speech was at your upper
limit of comfortable loudness, that is, if you would
not want it any louder; use 3 if the speech was
comfortable; use 2 if the speech was at your lower
limit of comfortable loudness, that is, if you would
not want it any softer; and use 1 if the speech was
too soft. Be sure that you evaluate the loudness
of the speech and not the background noise. Remem-
ber: you are to put a number, 1 through 5, in each
numbered space. Be sure you fill in every space.
Are you ready?

Following the instructions, the stimulus tape was
E5t10pped, and questions regarding the procedure to be fol-
‘l<3vved were answered. Then the test stimuli were presented.
The investigator and an electronics technician re—
Ina:Lned in the Laboratory during the entire experiment. A
13:L<>ck diagram of the instrumentation employed to present

t:}1€2 experimental stimuli to the subjects is presented in

Figure 6.

122

. .mpumnnbm any on Hasaflum
on» ucmmmua on omMOHQEm_COHuMmeEDHmeH 0:» mo EmumMflo xUOHm .o mudmflm

 

 

 

 

 

 

 

 

 

 

 

mmHqumz<
mqu uzo m new <.co m.so
mmamoumm
mmtz
mm49,
mMHqumz< .
quq #50 m ucm <.no . <.no

 

 

 

 

 

 

 

 

 

mmzommm<m

‘a
.l

n.»

123

Response to the Stimuli. The subjects recorded

their responses to each of the stimuli on a Subject Response
Sheet (see Appendix B).
The subjects' responses to the stimuli were tabu—

lated and recorded. The data for analysis consisted of

“the means of each subject's responses for the two presenta-
'tions of a particular combination of intensity level of

<:onnected discourse and acoustical background condition.

a m. t C C LU t.

CHAPTER IV
RESULTS AND DISCUSSION

Each subject evaluated the loudness of connected
ciisa:ourse presented at 13 different intensity levels in
cpiixet and in the presence of 70 dB SPL wide band white
noise, narrow band white noise, and speech babble. The
.si:iJnuli consisted of connected discourse presented under
fVDLLr acoustical background conditions, thereby making 52
<3<Drnbinations of intensity level of connected discourse and
13a<2kground condition. Each of these 52 combinations was
Flrwesented twice, making a total of 104 stimuli.

The data for analysis consisted of the means of
tlhue subjects' evaluations of the loudness of connected dis-
C3<I>L1rse for each of the two presentations of a particular
C3‘-<:>rnbination of intensity level of connected discourse and
1363.ckground condition. Thus for each of the 21 subjects
tidere were 52 averaged evaluations of the loudness of con-
rlected discourse--one for each particular combination of

iJutensity level of connected discourse and acoustical back-

Slround condition. (These averages appear in Appendix C.)

Results

This study involved two factors: intensity level

Of connected discourse (13 levels) and background condition

124

125

Ciaour levels). The means for the various combinations of
‘trne two factors appear in Table 5. The raw scores were
smitxjected to analysis of variance in order to determine
unnerther there was significant variation among mean scores.

A factorial design as described by Lindquist was

Lrtjglized in analyzing the data.329 An analysis of variance

rxatrtine (FACREP, Option 2) for the Control Data Corporation

cxonuouter was employed,330 and the results of this analysis

are found in Table 6.

The results of this analysis showed significant
5‘ x:atios for intensity level of connected discourse and
axzcyustical background condition and a significant F ratio
513:: interaction between these two factors. All of the fol-
1L<>Vving hypotheses, therefore, were rejected:

1-- There is no difference in a subject's evaluation of
the loudness of connected discourse presented at dif-
ferent intensity levels: 45, 50, 55, 60, 65, 70, 75,
80, 85, 90, 95, 100, and 105 dB SPL.

23° There is no difference in a subject's evaluation of
the loudness of connected discourse presented under
four different background conditions: in quiet (with
no noise other than that inherent in the system) and
in combination with 70 dB SPL wide band white noise,
narrow band white noise, and speech babble.

329See 8. F. Lindquist, Design and Analysis of Ex—
Efigriments in Psychology and Education (Boston: Houghton
I‘4«J_fflin Company, Inc., 1953), pp. 20-23, 207-19.

‘ 330D. F. Kiel, A. L. Kenworthy, and w. L. Ruble,
{Analysis of Variance Routines" (East Lansing: Michigan
State University, September 30, 1963), p. 23.

 

126

 

“an:

 

 

 

 

 

oo.m mm.a mm.a mm.m Hm.m am.m aH.N so.w mo.H mo.w oo.H oo.H aH.H
MMHOZ MHHm3
024m 30mm<2

mmow Hm.v mmav vmom HN.m mhom om.H mN.H NO.H bO.H ¢H.H 6H.H OH.H AmUv
MWHOZ MBH$3

024m MQH3

mmow h©.¢ mm.¢ mm.m Hm.m mm.N mN.N om.H HN.H hO.H mO.H OO.H OOoH ANUV
mdmm<m
ZUMMQm

OO.m om.¢ mm.¢ ¢5.m mv.m No.m N®.N m¢.N OO.N mm.H h©.H mvoa fiH.H AHUV
BMHDO

Amec “NHHV “HHHV AoHHc laws AmHV Aswc AoHV Ach xch AmHv ANHV “HHS
ZOHBHQZOU
mOH OOH mm om mm om mm on mm om mm Om mfi QZDomDXU<m
Amm mm ZH mmmDOUmHQ QMBUMZZOU m0 qm>m4 NBHWZMBZH

 

MZOHBHDZOU QZDomOXU<m mbom MMQZD 924 mqm>m4 NBHmszZH

ma B4 Qmazmmmmm mmmboumHQ QMBUHZZOU m0 mmmzoboq mmB m0 mZOHB<DA<>m 24m:

m mqm<B

127

TABLE 6

SUMMARY TABLE FOR THE FACTORIAL
DESIGN ANALYSIS OF VARIANCE

 

 

 

SOURCE: OF SUM OF MEAN F SIGN. AT
VARIANCE quAREs DF SQUARE STATISTIC 5% LEVEL
INTENSITY (I) 2183.25 12 181.94 1378.02 1.75
BACKGROUND

CONDITION (c) 29.93 3 9.98 75.57 2.60
I x c 37.32 22 1.04 7.85 1.00
CELLS 2250.50 51

ERROR 137.31 1040 .13

TOTAL 2387.81 1091

¥

33. There is no interaction between intensity level of con—
nected discourse and background condition in a subject's
evaluation of the loudness of connected discourse.

The subjects' mean evaluations of the loudness

C3f: connected discourse presented at 13 different inten-

ESity levels in quiet and in the presence of 70 dB SPL wide

13&3Lndwhite noise, narrow band white noise, and speech bab-

1311—e are presented graphically in Figure 7. The key to sub-

jeects' mean evaluations is as follows: 1 (too soft), 2

(Ilower limit of comfortable loudness), 3 (comfortable),

4= (upper limit of comfortable loudness), and 5 (too loud).

(The means used to plot the curves in Figure 7 are found

inTable 5 and, along with the subjects' averaged evalua—

tions, in Appendix C.)

128

  

 

 

500 F
.___.i___ Narrow Band White Noise (C4)? E
{a 4.5 23.ngg fQNWide Band White Noise (C3) / /
U)
g [3-—C}--E]Speech Babble (C2)
8 (}--{)~~C)Quiet (Cl)
'3 4.0-
a
m
B
U
m
E 305”
O
U
m
0
m
3.0
8 F
2
O
D
O
A
fig 2.5 —
B
a.
o I
m .
fig 2.0 — _.63
H Q
3 Q.
a 1.5-— GT
‘2: /
Liz: ./’?j///.
g. ‘fit'fi ..
l.O 43u-_.~—t%—-w ’
.4 I I l I I J I l I I I J
45 50 55 60 65 7O 75 80 85 9O 95 100105
(Il)(I )(13)(I )(ISZI )17)(I )(I9(I 5111(1 )113)
2 4 6 8 10 12

INTENSITY LEVEL OF CONNECTED DISCOURSE IN DB SPL

Figure 7. Subjects' mean evaluations of the loud-
neSS «of connected discourse (to the nearest l/lO) plotted as
a:fiu1ction of intensity level of connected discourse. The
Parmneters are the four different background conditions.

129

Discussion

Interaction. As was mentioned above, the inter-

 

action effect proved to be significant. Although the in-
teraction effect is not the primary concern of this study,
this effect merits further consideration because the inter—

pretation of the main effects of each factor depends on

whether or not interaction was present.331 In a factorial

design analysis of variance, interaction refers to that
part of the variance (error) that cannot be accounted for
by constant shifts in row means and column means, i.e.,

which cannot be accounted for in terms of the factors

alone.332 When there is no interaction present, the F

ratio for interaction between the two factors should not

be significantly different from zero.333 As is apparent

from inspection of Table 6, the observed interaction proved
to be significantly different from zero.
The presence of interaction indicates that the row

and column effects have an influence in combination differ-

ent from the sum of their separate effects.334 In terms

of the present study, the presence of interaction indicates

331Lindquist, op. cit., p. 118.

332Wilfrid J. Dixon and Frank J. Massey, Jr., in;
JEJDduction to Statistical Analysis, 2d ed. (New York:
Dkxiraw-Hill Book Company, Inc., 1957), p. 162.

333Lindquist, loc. cit.

334Dixon and Massey, loc. cit.

130

that the effect of the intensity level at which connected
discourse was presented is not the same for the four acous-
tical background conditions.

As is apparent from inspection of Figure 7, the
three curves representing the connected discourse being
presented with 70 dB SPL wide band white noise, narrow band
white noise, and speech babble are parallel, with minor
deviations, indicating that these three acoustical back—
ground conditions do not markedly contribute to the observed
interaction. The curve representing connected discourse
being presented in quiet, however, is not parallel to the
other three curves, indicating that this acoustical back-
ground condition does contribute to the observed interac-
tion. From inspection of Figure 7 it is apparent that in—
teraction occurred mainly when connected discourse was pre—
sented under the quiet background condition at intensity
levels of 45 to 80 dB SPL. It may be further observed that
above 80 dB SPL, the curve representing connected discourse
being presented in quiet becomes parallel to the other three
curves. It is possible that the background conditions in-
‘Volving the presence of 70 dB SPL wide band white noise
and.narrow band white noise may also contribute to the
Sigrdficance of the interaction effect, as indicated by
fine deviation of these curves from parallelism at the upper

intensity levels .

According to Lindquist, an observed interaction

131

may be wholly extrinsic (due entirely to errors involving
the assignment of subjects to treatment groups and to the
errors involving the effects of various extraneous factors

upon different treatment groups) or it may be in part in-
335

trinsic (due to the treatments only). A significant
F ratio for interaction between the factors means that the
observed interaction cannot reasonably be accounted for
in terms of errors involving the assignment of subjects
to treatment groups, but it does not rule out the possibil—
ity that the interaction is wholly intrinsic.

In terms of the present study, it would seem that
the interaction is due to the treatments only, since it
is assumed that the experiment was adequately controlled
and potential errors were eliminated. Specifically, the
errors involving the assignment of subjects to treatment
groups may be ruled out on the basis that the subjects were
not assigned to different treatment groups; instead, all
subjects listened to all of the experimental stimuli. For
the same reason, the errors involving the effects of vari-
ous extraneous factors upon different groups may be ruled
outu It will be recalled, however, that the subjects were
‘Uested in two groups because it was physically impossible
to test all of the subjects at the same time. It is pos—
sible that some extraneous factor or factors in the

335Lindquist, op. cit., p. 209.

132

experimental situation may have affected one group of sub-
jects that did not affect the other group. However, the
researcher was present during both presentations of the
research tape and is not aware of any reasons that would
support this idea.

From the above discussion, then, it seems reason-
able to assume that one--and possibly three-—of the acous—
tical background conditions produced different effects when
combined with intensity level of connected discourse, thus
accounting for the significant F ratio for interaction be-

tween the two factors.

Intensity Level of Connected Discourse Related to
Subjects' Judgments. The curves representing the subjects'
mean evaluations of the loudness of connected discourse
presented at 13 different intensity levels in quiet and
with 70 dB SPL wide band white noise, narrow band white
noise, and speech babble in general follow the same upward
pattern. The shape of the curves indicates, as would be
expected, that the higher the intensity level at which con-
nected discourse was presented, the louder a subject eval-
uated it, for all four acoustical background conditions.
The upward trend of the four curves presented in Figure 7
constitutes a graphic representation of the finding that
a subject's evaluations of the loudness of connected dis-
course is related to the intensity level at which connected

discourse is presented.

133

Although loudness is not dependent only upon inten—

sity,336 nevertheless these findings agree with those of

Ham and Parkinson, who examined the question, Do loudness
judgments vary with the intensity level of the stimulus?337
In their study, warble tones, single frequency tones, and
room noise recordings were presented to untrained listeners
in a sound field. The subjects were given a reference sound
of fixed energy level and were asked to record their judg—
ments as to the relative loudness of a second sound having

338

a different decibel level. These researchers found that

the increase (or decrease) of sound energy measured in dec-
ibels is a logarithmic function of the increase or decrease
in the loudness; however, they could not determine whether
there was any direct relation between loudness judgments
and intensity level because of the degree of variation of

the individual judgments and the errors introduced by back-

ground noise.339

Pollack also studied the relationship between the

loudness of a 1000 cps pure tone, white noise, and speech

336Harvey Fletcher, Speech and Hearing in Communi-

cation (Princeton, New Jersey: D. Van Nostrand Company,
Inc., 1953), p. 176.

337Lloyd B. Ham and John S. Parkinson, "Loudness
and Intensity Relations," The Journal of the Acoustical
§9ciety of America, III, No. 4 (April, 1932), pp. 511-34.

338

 

Ibid., p. 513.

339Ibid., p. 533.

134

and the intensity level of the sound.340 Various types

of loudness determinations were made for these three sig—
nals by trained subjects. The resultant loudness scales for
the pure tone and for white noise agreed closely with each
other for the several methods for determining loudness.341
Large discrepancies were found between the results obtained
by the methods for determining the loudness of speech, how—
ever. These findings indicate that another factor besides
intensity level may be operating in the evaluation of the
loudness of speech.342

It is interesting to note that, in the present study,
subjects evaluated as "most comfortable" the connected dis-
course that was presented at a higher intensity level than
is usually cited by other researchers as the most comfort—
able listening level in quiet. A careful examination of
the data in Figure 7 indicates that the most comfortable
loudness level was reached when connected discourse was
presented at an intensity level of 80 dB SPL in quiet.
The level of conversational speech is usually cited as 60

to 70 dB SPL;343 therefore, the subjects in this experiment

 

340Pollack, "Studies in the Loudness of Complex

Sounds,“ pp. 5-27.

34lIbid., p. 25.

3421bid.

343
ppo 76-770

Fletcher, Speech and Hearing in Communication,

 

135

did not believe that the level of conversational speech
was the most comfortable listening level for speech.

A possible explanation for the above observation
is as follows. Early in the experiment, the subjects were
exposed to both of the extreme intensity levels at which
connected discourse would be presented; and, as a result,
they may have developed some internal criteria for loudness
based upon these extremes. If this is true, 80 dB SPL (the
intensity level they evaluated as the most comfortable lis-
tening level for speech) lies in the middle of the range

of intensity levels employed.

Acoustical Background Condition Related to Subjects'

 

Judgments. Referring again to Figure 7, it is apparent
that the curve representing the subjects' mean evaluations
of the loudness of connected discourse presented at 13
different intensity levels in quiet does not have the

same shape as the three curves representing the subjects'
mean evaluations of the loudness of connected discourse
presented with 70 dB SPL wide band white noise, narrow band
white noise, and speech babble. This observation is par—
ticularly valid when connected discourse is presented at
the lower intensity levels used in this study. In this
area of the graph, the curves representing the subjects'
mean evaluations of the loudness of connected discourse
presented in quiet do not have the same general shape as

do the three curves representing the subjects' mean

136

evaluations of the loudness of connected discourse presented
with 70 dB SPL wide band white noise, narrow band white
noise, and speech babble. The shapes of the curves pre-
sented in Figure 7 constitute a graphic representation of
the finding that a subject's evaluation of the loudness
of connected discourse is related to the acoustical back-
ground condition which accompanies the connected discourse.
Although comparatively little work has been done
regarding the influence of background noise on the loudness
of acoustic stimuli, a few studies may be cited with regard
to the finding in the present study concerning the influ-
ence of background noise upon listeners' evaluations of
the loudness of speech in terms of comfortableness. For
example, Pollack found that the ULCL, the MCL level, and
the LLCL for pure tones increased as the intensity level
of the background noise increased.344 Egan found that white
noise presented at intermediate levels often increased the
loudness of speech heard in the other ear.345 Pollack also
studied the effect of background white noise on the loud-
ness of speech, with both the speech and the noise deliv-

ered monaurally to the same ear.346 He found that the

344Pollack, "Comfortable Listening Levels for Pure

Tones in Quiet and Noise," p. 162.
345 .
Egan, Op. Cit.

346Pollack, "Studies in the Loudness of Complex

Sounds," pp. 72-88.

137

presence of noise decreased the loudness of the speech under

 

the conditions set forth in his study and offered some pos-
sible explanations for this finding,347 which is contrary
to his own findings for pure tones and to Egan's results.
In the present study, speech was delivered via earphones

to both of the subject's ears; yet the findings are simi—
lar to those of Egan.

As is apparent from inspection of Figure 7, the
curve representing connected discourse being presented in
quiet rises more gradually than do the three curves repre-
senting connected discourse being presented with 70 dB SPL
wide band white noise, narrow band white noise, and speech
babble. The relatively flat shape of these three curves,
over a 20 dB range (45 to 65 dB SPL) indicates that loud-
ness judgments remain fairly constant when connected dis—
course is presented at intensity levels of 45 to 65 dB SPL
with 70 dB SPL wide band white noise, narrow band white
noise, and speech babble. Loudness increases rather abrupt-
ly, however, after connected discourse is presented at an
intensity level of 70 dB SPL--the intensity level at which
wide band white noise, narrow band white noise, and speech
babble are presented. This observation indicates that when
Connected discourse is presented at intensity levels greater
iflian 65 dB SPL with 70 dB SPL wide band white noise, narrow

347Ib1d., p. 83.

138

band white noise, and speech babble, loudness increases
more abruptly than it does when connected discourse is pre-
sented in quiet.

It is possible that this finding is related to the
phenomenon of recruitment—-the abnormal increase in loud-
ness of an acoustic stimulus when it is presented above
threshold--observed in certain persons who exhibit hearing
impairments. In the present study, the loudness of speech
increased more abruptly when the speech was presented in
noise than when speech was presented in quiet, in the in-
stance where the speech was presented at about the same
intensity level as the background noise. This observation
appears to be related, also, to Pollack's finding that,
except at the lowest frequency presented, the level of the
lower limit of the range of comfortable listening levels
for pure tones relative to the masked threshold is decreased
approximately 3 dB as the level of the background noise
is increased 10 dB.348 According to Pollack, this result
parallels the subjects' reports that the tone "popped out"
quickly above threshold so that the LLCL was close to the

threshold of detectability at high noise levels.349

Critical Difference Tests. Critical difference

 

 

348Pollack, "Comfortable Listening Levels for Pure
Tones in Quiet and Noise," p. 161.
349

Ibid.

139

tests350 were performed in order to determine the signif-
icant differences among subjects' mean evaluations of

the loudness of connected discourse presented at 13 dif-
ferent intensity levels in quiet and with 70 dB SPL wide
band white noise, narrow band white noise, and speech bab-
ble. (The results of this critical difference test appear
in Appendix D.) It was found that a difference between two
means of more than .22 was significant at the .05 level.

The results of this test reveal that significant
differences existed between 1136 of the 1326 possible com-
parisons. It is apparent, therefore, that many treatment
combinations contributed to the significance of the tests
of the main effects of both factors and to the significance
of the test for interaction between the two factors. It
should be noted, however, that the standard errors are small,
resulting in statistical significance when the practical
significance may be of little or no importance.

Table 7 shows the results of critical difference
tests performed at each of the 13 intensity levels at
which connected discourse was presented. By referring
to Table 7, it is possible to ascertain which of the dif-
ferences among the four means at any given intensity level
on Figure 7 are significant. For example, inspection of

Figure 7 indicates that the loudness of connected discourse

 

350Lindquist, op. cito, PP° 90‘96: 146’ and 214°

TABLE 7

RESULTS OF CRITICAL DIFFERENCE TESTS FOR EACH INTENSITY

LEVEL AT WHICH CONNECTED DISCOURSE WAS PRESENTEDa

 

 

 

 

i’at I (45 dB SPL)
1 c2 c3 c4
c1 1.14 .14 .04 .03
c2 1.00 c2 .10 .17
c3 1.10 c3 .07
c4 1.17
Y'at 12 (50 dB SPL) C c c
2 3 4
cl 1.43 .43. .299 .43°
C2 1000 014 "'-
c3 1.14 .14
c4 1.00
Y'at I3 (55 dB SPL) C c c
2 3 4
cl 1.67 .62‘ .53* .67‘
c2 1.05 .09 .05
c3 1.14 .14
c4 1.00
i’at I4 (60 dB SPL) c C C
2 3 4
cl 1.93 c .86’ .86' .88'
C2 1007 C -"" 002
c3 1.07 c .02
c4 1.05

 

aThe symbols used here to describe the various cell
means are explained in Table 5 and in Figure 7.

‘Differences between two means significant at the

.05 level.

141

TABLE 7 (continued)

 

 

 

 

 

i at I (65 dB SPL)
5 c2 c3 c4
Cl 2.00 .79' .98‘ .95’
c2 1.21 .19 .16
c3 1.02 .03
c4 1.05
3? at I (70 dB SPL)
6 c2 c3 c4
c1 2.45 .55‘ 1.16‘ .78'
c2 1.90 .61‘ .23'I
c3 1.29 .38‘
c4 1.67
I at I (75 dB SPL)
7 c2 c3 c4
c1 2.62 .36‘ .72‘ .48‘
c2 2.26 .36‘ .12
c3 1.90 .24'
c4 2.14
Y at I (80 dB SPL)
8 c2 c3 c4
c1 3.02 .16 .23* .45'
c2 2.86 .07 .29'
c3 2.79 .22
c4 2.57
i at 19 (85 dB SPL) c C C
2 3 4
cl 3.48 .17 .27‘ .27‘
c2 3.31 .10 .10
c4 3.21

 

142

TABLE 7 (continued)

 

NI

at

I

(90 dB SPL)

 

 

 

10 c2 c3 c4
Cl 3.74 .19 .50' .12
c2 3.93 .69‘ .07
c3 3.24 .62‘
c4 3.86
i’at I (95 dB SPL)
11 c2 c3 c4
C1 4029 040. -'- .04
c2 4.69 .40' .36‘
c3 4.29 .04
c4 4.33
Y'at I (100 dB SPL)
12 c2 c3 c4
c1 4.90 .23’ .09 .03
c2 4.67 .14 .26‘
c3 4.81 .12
c4 4.93
I'at I (105 dB SPL)
13 c2 c3 c4
Cl 5.00 005 002 '-
c2 4.95 .03 .05
c3 4.98 .02
c4 5.00

 

143

presented in Quiet is judged consistently higher than the
loudness of connected discourse presented with the three
types of background ngigg, for intensity levels of 50 through
85 dB SPL. Referring to Table 7, however, it becomes ap-
parent that all of these differences are not statistically
significant. It should be noted that, from intensity levels
of 50 to 75 dB SPL, subjects' mean evaluations of the loud~
ness of connected discourse determined in guigt differ sig-
nificantly from subjects' mean evaluations of the loudness
of connected discourse determined with 70 dB SPL wide band
white noise, narrow band white noise, and speech babble.
However, at intensity levels of 80 and 85 dB SPL, subjects'
mean evaluations of the loudness of connected discourse
determined in ggigt_differ significantly from subjects'

mean evaluations of the loudness of connected discourse
determined in the presence of only two types of background
noise: wide band white noise and narrow band white noise.
Moreover, only one such difference is significant at the

90 dB SPL intensity level.

These findings indicate that it is not until con-
nected discourse is presented at an intensity level above
75 dB SPL that the several background conditions begin to
yield loudness evaluations quite similar to those resulting
from the condition under which connected discourse is pre-
sented in quiet. These results appear to be related to

the idea that if the background noise is at a much higher

144

level than the speech (regardless of background condition),
the noise will mask out the speech351 to the point where
it cannot be evaluated as comfortable or not cOmfortable.
The results of the critical difference tests
performed at each of the 13 intensity levels at which
connected discourse was presented (see Table 7) indicate
that the only intensity level at which subjects' mean eval—
uations of the loudness of connected discourse presented
under the four background conditions all differed signif—
icantly from each other was 70 dB SPL--the same intensity
level at which wide band white noise, narrow band white
noise, and speech babble were presented when they occurred
in the experimental design. This finding is apparent from
inspection of Figure 7; at an intensity level of 70 dB SPL
subjects' mean evaluations of the loudness of connected
discourse are more disperse than at any of the other inten-
sity levels. This finding seems to indicate that when con—
nected discourse was presented at an intensity level of
70 dB SPL with a zero signal—to-noise ratio, type of back-
ground noise had the most influence upon subjects' mean
evaluations of the loudness of connected discourse. This
result may also be related to the relative amounts of mask—
ing afforded by the three types of background conditions

which involved the presence of wide band white noise,

 

351Pollack, "Studies in the Loudness of Complex

Sounds," p. 74.

14S

narrow band white noise, and speech babble.

Referring again to Figure 7, it is apparent that
there is less dispersion among subjects' mean evaluations
of the loudness of connected discourse presented in quiet
and with 70 dB SPL wide band white noise, narrow band white
noise, and speech babble when connected discourse is pre-
sented at higher intensity levels. The results of the
critical difference tests performed at each of the 13
intensity levels at which connected discourse was presented
(see Table 7) indicate that, of the six possible compari-
sons of the differences among means when connected discourse
was presented at 90 or 95 dB SPL, only three of these dif-
ferences were statistically significant at the .05 level.
When connected discourse was presented at 100 dB SPL, only
two of these differences were significant. There were £2.
significant differences among the means when connected dis-
course was presented at 105 dB SPL.

These findings tend to indicate that, when connected
discourse was presented at the higher intensity levels,
background condition had less influence upon subjects' eval-
uations of the loudness of connected discourse than when
connected discourse was presented at the lower intensity
levels. In other words, there seemed to be an intensity
level at which connected discourse was presented, above
which the type of background condition did not influence

a subject's evaluation of the loudness of connected discourse.

146

This finding is probably related to the effect of masking,
as was mentioned above. However, since connected discourse
was not presented at intensity levels greater than 105 dB
SPL, this observation should be regarded as tentative.

An additional question developed as the data were
analyzed. Table 8 presents a summary of the ranges of in-
tensity level of connected discourse for three regions of
comfort (too soft, comfortable, and too loud) and for the
four background conditions. The information presented in
Table 8 was obtained by interpolation from Figure 7.

The question to be answered was, Does the range
of the intensity level of connected discourse affect the
region of comfort?

In an attempt to answer this question, the follow-
ing null hypothesis was formulated: There is no difference
in the magnitude of the range of intensity levels of con-
nected discourse for three regions of comfort: too soft,
comfortable, and too loud.

In order to analyze the data presented in Table 8,
an analysis of variance technique for a single variable

352 The results of this

of classification was employed.
analysis are found in Table 9.
The results of this analysis showed that the effect

of the range of intensity level of connected discourse was

 

352Dixon and Massey, op. cit., pp. 145—52.

147

I 9 (ANIIIIMIYIJ .IIWI ”I‘ll?

.mmcmm.

 

.AOHO

.Aomv

 

 

 

 

moa 0» mm om om 0» Ow om mu 0» ma A UV
mmHoz MBHIZ
Qz<m 30mm<z
ommv oAmHV uAOmv m
mOH 0b OOH mm mm 0b ow om mm 0b ma A UV
mmHoz mBHmz
Qz<m MOHz
SAOHV SAmHV cAmNV N
moa ou mm om om Ob me me on on ma A UV
mamm<m
mummmm
{AOHV Sfimmv oAmHv H
mOH ou mm om om 0b mm mm om 0b ma A UV
BMHDO
Ammo lame Ammo Ammo lame
6:04 008 mmmCOSoq pomuuomeou whoopsoq bmom ooe
mHQmpuomEou mHQMpuomeou
mZOHBHQZOU
mo bflEHA Home: mo bflEHq umzoq QZDomwa<m

 

BmOMSOU ho mZOHomm

Amman AmmHquIm>Hm Bmmm<mz mmB OB QMBzmmmmm Q24 5 mmeHh

20mm ZOHB<AOQMMBZH Mm QMZH<BmOV WZOHBHQZOU QZDOMOXU¢m mDom mmB D24 BMOhZOU

m0 mZOHomm m>Hm Ema mom mqm>mq MBHWZMBZH mmB Amm

m mdm<B

ma ZH OZHBOmm m4m<8 Nm<zzbm

148

TABLE 9

SUMMARY TABLE OF ANALYSIS OF VARIANCE
FOR A SINGLE VARIABLE OF CLASSIFICATION

 

 

SOURCE OF SUM OF MEAN F SIGN. AT
VARIANCE SQUARES DF SQUARE STATISTIC 5% LEVEL
CATEGORY

MEANS 529.17 2 264.59 8.28 4.26
WITHIN 287.50 9 31.94

TOTAL 816.67 11

 

significant at the .05 level, and this finding indicates
that the magnitude of the range of comfortable listening
levels is not the same for the three regions of comfort.
Therefore, the following hypothesis was rejected: There
is no difference in the magnitude of the range of intensity
levels of connected discourse for three regions of comfort:
too soft, comfortable, or too loud.

Since the effect of the range of intensity level
of connected discourse factor proved to be significant,
Ia decision was made to administer a critical difference
test in order to determine where the differences lay among
the ranges of the intensity level of connected discourse
for the three regions of comfort and four background con-
ditions. (The results of this critical difference test
appear in Appendix E.) It was found that a difference be-
tween any two ranges of more than 23.99 wOuld be significant

at the .05 level. The results of this test indicated that

 

149

significant differences existed between only two of the
66 possible comparisons.

Ranges of comfortable listening levels for speech
have been reported by other researchers, as indicated in
Chapter II. For example, Lezak found that the range of
comfortable listening levels for speech for normal hearing

353 Lezak measured the RCL from

listeners was 60 dB SPL.
the threshold of sensitivity to the uncomfortable loudness
threshold. Although Loftiss did not report ranges of com-
fortable listening levels in his study, these values could
be determined from his data. Regardless of psychophysical
methods and judgmental procedures used, the range of com-
fortable listening levels for speech determined in quiet
was 31 dB re: normal audiometric threshold for earphone
presentations and 29 dB for loudspeaker presentations.354
These values were obtained by subtracting the value for
soft comfortable loudness from the value for loud comfort-
able loudness for each mode of presentation.

Referring to Table 8, it is interesting to note
that the range of comfortable listening levels determined
in quiet in the present study is 25 dB SPL, which is con-

siderably below the value reported by Lezak for normal hear-

ing persons and slightly below the values derived from the

353Lezak, op. cit., p. 128.

3S4Loftiss, Op. cit., p. 20.

150

Loftiss study. It is possible that variations in psycho-
physical procedures and/or variations in the definition

of the range of comfortable listening levels may account
for this discrepancy. Nevertheless, the range of comfort-
able loudness levels for speech determined in quiet (65

to 90 dB SPL) is comparable to the values for the most com-
fortable listening level that are most often reported in
the literature. It is also interesting to note that the

LLCL and the ULCL for speech determined in quiet are con-

 

siderably above and somewhat below the values reported by
other researchers for the thresholds of sensitivity and
of discomfort, respectively.

Referring again to Figure 7, it is interesting to
note that there was less variation in subjects' determina-
tions of the ULCL than in their determinations of the LLCL,
for the four background conditions. Obtaining the four
values for the LLCL and ULCL by means of interpolation from
the figure, it becomes apparent that, when connected dis-
course was presented in quiet, subjects evaluated the LLCL
for connected discourse to be at 65 dB SPL. However, when
connected discourse was presented with 70 dB SPL wide band
white noise, narrow band white noise, and speech babble,
subjects evaluated the LLCL for connected discourse to be
higher (at 76, 74, and 71 dB SPL, respectively). Such was
not the case when subjects determined their ULCL. When

connected discourse was presented with 70 dB SPL speech

151

babble, subjects evaluated their upper limit of comfortable
loudness to be at 91 dB SPL, the lowest level for the four
background conditions. When connected discourse was pre—
sented in quiet, however, subjects evaluated their ULCL

to be at 92 dB SPL. Their upper limits of comfortable
loudness for connected discourse presented with 70 dB SPL
narrow band white noise and wide band white noise were 91
and 94 dB SPL, respectively.

These findings parallel those of Loftiss, although
it is possible that background noise accounts for the vari-
ation among these levels in the present study, as was dis-
cussed in Chapter II. Loftiss found that subjects could
determine three loudness judgments for speech with varying
degrees of precision: soft comfortable loudness judgments
were the most variable, most comfortable loudness judgments
were less variable, and loud comfortable loudness judgments
were the most precise. 55

It is interesting to note that the upper limits
of the range of comfortable listening levels obtained in
the present study for the four background conditions were
within 1 5 dB, which is the degree of accuracy usually ac-
cepted for audiometric measurements. This finding would
seem to indicate that the ULCL remains fairly stable, re-

gardless of background condition. Pollack reported similar

 

355Ibid., p. 34.

5’...

 

1".” V

152

findings for pure tones presented in quiet and in background

noise.356

356Pollack, "Comfortable Listening Levels for Pure

Tones in Quiet and Noise," pp. 158-62.

 

CHAPTER V

SUMMARY AND CONCLUSIONS

Summary

Researchers have long realized that loudness judg-
ments are "subjective" and must be measured "indirectly"
by observing the effects of specified auditory stimuli upon
the behavior of a group of listeners.

Many different aspects of loudness have attracted
the attention of researchers. They are (l) the defini—
tion of loudness, (2) the calculation of the loudness of
various types of acoustic stimuli by means of mathematical
formulas, (3) the utilization of psychophysical techniques
for determining the loudness of sounds, (4) the determina-
tion of the loudness of specific kinds of auditory stimuli,
(5) the effects of certain factors upon loudness determina-
tions, and (6) the determination of "equal-loudness con—
tours." However, comparatively few studies involving sub-
jects' evaluations of the loudness of stimuli in terms of
listener comfort have been carried out. Furthermore, even
fewer studies have investigated the effects of various fac-
tors that might influence these evaluations. In addition,
no previous study has utilized the psychophysical method

of single stimuli (rating scale method) as a means of

153

154

evaluating the loudness of speech in terms of listener
comfort, which was utilized in the present study.

The purpose of this study was to investigate the
effects of intensity of connected discourse and acoustical
background condition upon listeners' evaluations of the
loudness of connected discourse. The following questions
were asked: Does intensity level of connected discourse
or acoustical background condition influence a subject's
evaluation of the loudness of connected discourse? Is
there an interrelationship between these two factors that
might influence a subject's evaluation of the loudness of
speech?

Connected discourse was reproduced on low noise
magnetic tape at 13 different intensity levels (45 to 105
dB SPL,in 5 dB steps) in quiet and with 70 dB SPL wide band
white noise, narrow band white noise, and speech babble;
thus there were 52 possible combinations of intensity
level of connected discourse and background condition.
Two seven-second segments of each stimulus were presented
to the subjects, making a total of 104 stimuli. A number
between 1 and 104 was assigned arbitrarily to each seven-
second stimulus. The stimuli were spliced together in
randomlorder for presentation to the subjects.

Twenty-one normal hearing subjects were tested in
two groups, each subject listening to all of the stimuli.

For each stimulus he evaluated the connected discourse as

155

being too soft, at his lower limit of comfortable loudness,
comfortable, at his upper limit of comfortable loudness,

or too loud, using a rating scale of l to 5. The subject
recorded his responses on a Subject Response Sheet.

The data for analysis consisted of the means of
each subject's responses for the two presentations of a ’
particular combination of intensity level of connected
discourse and acoustical background condition.

The data were subjected to a factorial design an-

 

alysis of variance, the results of which indicated that E
intensity level of connected discourse and background con-

dition affected a subject's evaluation of the loudness of

connected discourse and that there was a significant inter-

action between these two factors. A critical difference

test was performed in order to determine where the differ—

ences lay among subject's mean evaluations of the loudness

of connected discourse presented at 13 intensity levels

in quiet and with 70 dB SPL wide band white noise, narrow

band white noise, and speech babble. It was found that dif—

ferences existed among most of the treatment combinations.

Conclusions
Within the limitations of the present study, the
following conclusions seem warranted:
l. The perceived loudness of connected discourse

increases as the intensity level of the connected discourse

156

increases, irrespective of the type of acoustical background
condition.

2. The presence of background noise influences
listeners' evaluations of the perceived loudness of con-
nected discourse until the intensity level of the speech
is increased beyond the level of the noise. When the in-
tensity level of connected discourse is increased beyond
the level of the noise, the four background conditions be-
gin to produce quite similar effects on listeners' evalu-
ations of the loudness of connected discourse.

3. The range of comfortable loudness for speech
heard in quiet is 65 to 90 dB SPL. The ranges of comfort-
able listening for speech heard in 70 dB SPL wide band white
noise, narrow band white noise, and speech babble are 45
to 95 dB SPL, 80 to 90 dB SPL, and 75 to 90 dB SPL, re-
spectively.357

4. The magnitude of the range of intensity levels
at which connected discourse was presented is not the same

for the three regions of comfort (too soft, comfortable,

and too loud)°

Implications for Further Research

 

Although several studies have dealt with listeners'

 

3S7The ranges of comfortable listening levels de-
termined in quiet by this investigation generally agree
with those reported by other researchers. The range of
comfortable listening levels for speech in noise had not
been reported previously.

157

evaluations of the loudness of auditory stimuli (mainly
pure tones and speech) in terms of comfortableness and
various factors that may affect such evaluations, further
research in these areas is needed. Despite the fact that
relatively little is known about how persons with normal
hearing evaluate the loudness of speech in terms of com-
fortableness and what factors may influence their evalua-
tions, the determination of most comfortable listening lev-
els is routinely included in hearing aid evaluations for
various reasons, as discussed previously. Specifically,

the following areas are suggested for further study.

Subjects. Persons who represented a wide age range,

 

and who displayed normal hearing and various types and de-
grees of hearing impairments could be tested in order to
ascertain whether these factors may be related to loudness
judgments in terms of listener comfort. Other factors could
also be examined, such as sex, occupations, and personality
factors. It is possible, for example, that persons with
certain psychological impairments evaluate the loudness

of speech in terms of comfortableness differently than do

persons without such problems.

Psychophysical Methods. More study needs to be

 

made of the rating scale technique applied to the area of
evaluating comfortable listening levels. The results ob-

tained by this method should be compared with the results

158

obtained using some of the more common methods, such as

the method of adjustment and the method of limits. It is
possible that the rating scale method may yield comfortable
loudness data which are just as reliable and valid as the
data obtained with some of the other methods but that the
rating scale method may be more efficient and better suited

to the hearing aid evaluation.

Background Conditions. In addition to the types

 

of background conditions utilized in the present study,
background conditions that more closely simulate everyday
environments might be utilized in order to analyze the ef—
fects of these environments upon listeners' evaluations

of the loudness of speech in terms of comfortableness.

The information gleaned from studies of this nature would
seem to have great possibilities for increasing accuracy
in fitting a person with a hearing aid relative to the en-

vironment in which he would be wearing his aid.

BIBLIOGRAPHY

Books

American Standard Acoustical Terminology (Including Mechan—
ical Shock and Vibration). New York: American
Standards Association, Incorporated, 1960.

Davis, Hallowell. "Hearing Aids." Hearing and Deafness:
A Guide for Laymen. Hallowell Davis, editor. New
York: Murray Hill Books, Inc., 1947.

 

. “Physics and Psychology of Hearing." Hearing
and Deafness. Hallowell Davis and 5. Richard
Silverman, editors. 2d ed. revised. New York:
Holt, Rinehart and Winston, Inc., 1960.

 

 

Davis, Hallowell, Stevens, 3. S., Nichols, R. H., Jr.,
Hudgins, C. V., Peterson, G. E., Marquis, R. J.,
‘ and Ross, D. A. Hearing Aids: An Experimental
Study of Design Objectives. Cambridge, Mass.:
Harvard University Press, 1947.

 

 

Dixon, Wilfrid J., and Massey, Frank J., Jr. Introduction
to Statistical Analysis. New York: McGraw-Hill
Book Company, Inc., 1957.

 

Fletcher, Harvey. Speech and Hearing in Communication.
Princeton, New Jersey: D. Van Nostrand Company,
Inc., 1953.

Fowler, Edmund Prince. "Diseases of the Neural Mechanism
of Hearing: Cochlea, Auditory Nerve, and Its Cen—
ters in the Medulla and Cortex.“ Medicine of the
Egg. Edmund Prince Fowler, editor. New York:
Thomas Nelson & Sons, 1947.

 

Guilford, J. P. Psychometric Methods. 2d ed. revised.
New York: McGraw-Hill Book Company, Inc., 1954.

 

Harris, J. Donald. ”Research Frontiers in Audiology."
Modern Developments in Audiology. James Jerger,
editor. New York: Academic Press, 1963.

 

Hirsh, Ira J. The Measurement of Hearing. New York:
McGraw-Hill Book Company, Inc., 1952.

 

159

160

Lindquist, E. F. Design and Analysis of Experiments in
Psychology and Education. Boston: Houghton Miff—
lin Company, 1953.

Munson, W. A. "The Loudness of Sounds." Handbook of Noise
Control. Cyril M. Harris, editor. New York:
McGraw-Hill Book Company, Inc., 1957.

Newby, Hayes A. Audiology. 2d ed. revised. New York:
Appleton-Century-Crofts, 1964.

 

Silverman, S. R., and Taylor, S. Gordon. "Hearing Aids.”
Hearingyand Deafness. Hallowell Davis and S. Rich-
ard Silverman, editors. 2d ed. revised. New York:
Holt, Rinehart and Winston, Inc., 1960.

Stevens, S. S. ”Mathematics, Measurement, and Psychophys—
ics." Handbook of Experimental Psychology. 5. 8.
Stevens, editor. New York: John Wiley & Sons,
Inc., 1951.

Stevens, Stanley Smith, and Davis, Hallowell. Hearing:
Its Psychology and Physiology. New York: John
Wiley & Sons, Inc., 1938.

Watson, L. A. A Manual for Advanced Audiometpy. Minneap-
olis: Colorcraft Press, 1948.

Watson, Leland A., and Tolan, Thomas. Hearing_Tests and
Hearing Instruments. Baltimore: The Williams and
Wilkins Company, 1949.

Articles and Periodicals

Baier, D. E. "The Loudness of Complex Sounds," Journal of
Experimental Psychology. XIX, No. 3 (June, 1936),

Bangs, Jack L., and Mullins, Cecil J. "Recruitment Testing
in Hearing and Its Implications," A.M.A. Archives
of Otolaryngology, LVIII, No. 5 (November, 1953),
pp. 582-92.

 

Bevan, William, and Pritchard, Joan Faye. "Effect of 'Sub-
liminal' Tones upon the Judgment of Loudness,"
Journal of Experimental Psychology, LXVI, No. l
IJuly, 1963), pp. 23-29.

Carhart, Raymond. "Tests for the Selection of Hearing
Aids,“ The Lapyngoscope, LVI, No. 12 (December,
1946) , pp. 780-940

161

. "Volume Control Adjustment in Hearing Aid Selec-
tion," The Laryngoscppe, LVI, No. 9 (September,
1946), pp. 510-26.

 

Chapin, E. K., and Firestone, F. A. "The Influence of Phase
on Tone Quality and Loudness: the Interference of
Subjective Harmonics,“ The Journal of the Acoustical

 

Society of America, V, No. 3 (January, 1934), pp.
173-80.

 

Churcher, B. G., and King, A. J. "Performance of Noise
Meters in Terms of Primary Standard," Journal of
the Institution of Electrical Engineers, LXXXI,
No. 487 (July, 1937), pp. 57-81 (London).

Corliss, Edith L. R., and Winzer, George E. "Study of
Methods for Estimating Loudness,“ The Journal of
the Acoustical Society of America, XXXVII, No. 6
(June, 1965): p. 1174 (Abstract).

Davis, H., Hudgins, C. V., Marquis, R. J., Nichols, R. H.,
Jr., Peterson, G. E., Ross, D. A., and Stevens,
S. S. "The Selection of Hearing Aids," The Laryn-
goscope, LVI, Nos. 3 and 4 (March and April, 1946),
pp. 85-115, 135-63.

Egan, James P. "The Effect of Noise in One Ear upon the
Loudness of Speech in the Other Ear," The Journal
of the Acoustical Society of America, XX, No. 1
(January, 1948), pp. 58—62.

 

Eisenberg, Philip, and Chinn, Howard A. "Tonal Range and
Volume Level Preferences of Broadcast Listeners,"
Journal of Experimental Psychology, XXXV, No. 5
(October, 1945), pp. 374-92.

 

Fletcher, Harvey. "Loudness, Masking and Their Relation
to the Hearing Process and to the Problem of Noise
Measurement," The Journal of the Acoustical Society_
of America, IX, No. 4 (April, 1938), pp. 275-93.

 

Fletcher, Harvey, and Munson, W. A. “Loudness, Its Defini-
tion, Measurement and Calculation," The Journal
of the Acoustical Societyyof America, V, No. 2
(October, 1933), pp. 82-108.

 

Graham, Fred P. "Ads on TV Found Too Loud by F.C.C.:
Broadcasters Are Warned Against Strident Delivery,"
The New York Times, CXIV, No. 39, 252 (July 13,
1965), p. 67.

 

162

Ham, Lloyd B., Biggs, Frank, and Cathey, Everett H., Jr.
"Fractional and Multiple Judgments of Loudness,"
The Journal of the Acoustical Society of America,
XXXIV, No. 8 (August, 1962), pp. 1118-21.

Ham, Lloyd B., and Parkinson, John S. "Loudness and Inten—
sity Relations," The Journal of the Acoustical So-
ciety of America, III, No. 4 (April, 1932), pp.
511—34.

Hedgecock, LeRoy D. "Recruitment in Ears with Abrupt Loss
of Acuity for High Frequencies," Journal of Speech
and Hearing Disorders, XXII, No. 1 (March, 1957),
pp. 91-97.

 

 

Howes, Davis H. "The Loudness of Multicomponent Tones,"
American Journal of Psychology, LXIII, No. 1 (Jan-
uary, 19507, pp. 1-300

Journal of the Acoustical Society_of Americay The. XXXVII,
No. 6 (June, 1965), p. 1174.

Knauss, H. P. "An Empirical Formula for the Loudness of
a 1000-cycle Tone," The Journal of the Acoustical
Society of America, IX, No. 17(July, 1937), pp.
45-46 0

Lochner, J. P. A., and Burger, J. F. “Pure-Tone Loudness
Relations," The Journal of the Acoustical Society
of America, XXXIV, No. 5 (May, 1962), pp. 576-81.

Menzel, Otto J. "The 'Comfort' Aspects of Loudness," The
Eye, Eary Nose and Throat Monthly, XLV, No. 2 (Fe -
ruary, 1966), pp. 106—07.

Newman, E. B., Volkmann, J., and Stevens, S. S. "On the
Method of Bisection and Its Relation to a Loudness
Scale," American Journal of Psychology, XLIX, No.
1 (January, 1937), pp. 134-37.

Pollack, Irwin. "Comfortable Listening Levels for Pure
Tones in Quiet and Noise," The Journal of the Acous-
tical Society_of America, XXIV, No. 2 (March, 1952),
pp. 158—62.

. "The Effect of White Noise on the Loudness of
Speech of Assigned Average Level," The Journal of
the Acoustical Society of America, XXI, No. 3 (May,
1949), pp. 255-58.

 

 

 

163

Rintelmann, William F., and Carhart, Raymond. "Loudness
Tracking by Normal Hearers via Bekesy Audiometer,"
Journal of Speech and Hearing Research, VII, No. l
‘TMarch, 1964), pp. 79-93.

Robinson, D. W., and Dadson, R. S. "A Re-determination
of the Equal-Loudness Relations for Pure Tones,”
British Journal of Applied Physics, VII (May, 1956),
pp. 166—81.

Sawyer, Leroy L. "Office Procedure in Hearing Evaluation:
A Practical Approach,“ The Laryngoscope, LX, No.
11 (November, 1950), pp. 1061-85.

Schneider, Bruce, and Lane, Harlan. "Ratio Scales, Cate-
gory Scales, and Variability in the Production of
Loudness and Softness,“ The Journal of the Acous-
tical Society of America, XXXV, Part 2, No. 12
TDecember, 1963), pp. 1953-61.

Stevens, S. S. ”Calculation of the Loudness of Complex
Noise," The Journal of the Acoustical Society of
America, XXVIII, No. 5 (September, 1956), pp. 807-
32.

. "The Direct Estimation of Sensory Magnitudes--
Loudness," American Journal of Psychology, LXIX,
No. 1 (March, 1956), pp. 1-25.

 

. "Procedure for Calculating Loudness: Mark IV,"
The Journal of the Acoustical Society of Amggica,
XXXIII, No. 11 (November, 1961), pp. 1577—85.

 

. “A Scale for the Measurement of a Psychological
Magnitude: Loudness," Psyghological Review, XLIII,
No. 5 (September, 1936), pp. 405-16.

 

Thurlow, Willard, and Tabory, Leon. "Effects of Repeated
Presentations of a Tone upon Absolute Loudness
Judgments," Journal of General Psychology, LX
(April, 1959), pp. 161-66.

Watson, Norman A. "Most Comfortable Listening Levels for
Pure Tones," The Journal of the Acoustical Society
of America, XX, No. 2 (March, 1948), p. 220 (Ab-

 

stract).
Watson, N. A., and Knudsen, V. O. "Selective Amplification
in Hearing Aids,“ The Journal of the Acoustical

 

Society of America, XI, No. 4 (April, 1940), pp.

164

Wright, H. N. "Loudness as a Function of Duration," The
Journal of the Acoustical Societyyof America,
XXXVII, No. 6 (June, 1965), p. 1174 (Abstract).

Unpublished Materials

 

Hedgecock, LeRoy Darien. "Prediction of the Efficiency
of Hearing Aids from the Audiograms." Unpublished
Ph.D. Dissertation, Department of Speech, Univer-
sity of Wisconsin, 1949.

Kavanagh, James Francis. "An Investigation of the Most
Comfortable Listening Levels for Speech.“ Unpub-
lished Ph.D. Dissertation, Department of Speech,
University of Wisconsin, 1960.

Kiel, D. F., Kenworthy, A. L., and Ruble, W. L. "Analysis
of Variance Routines." East Lansing: Michigan
State University, September 30, 1963 (mimeographed).

Lezak, Raymond Joseph. "Some Aspects of Comfortable Loud-
ness." Unpublished Ed.D. Dissertation, Department
of Speech and Theater, The Pennsylvania State Uni-
versity, 1961.

Loftiss, Eugene Walter. "An Evaluation of the Effects of
Selected Psychophysical Methods and Judgmental Pro—
cedures upon Comfortable Loudness Judgments of Con-
nected Speech." Unpublished Ph.D. Dissertation,
Department of Speech, University of Illinois, 1964.

Melnick, William. “Comfort Level and Loudness Matching
for Continuous and Interrupted Signals." Unpub—
lished paper presented at the 1965 Convention of
the American Speech and Hearing Association, Novem-
ber l, 1965.

Pollack, Irwin. “Studies in the Loudness of Complex Sounds."
Unpublished Ph.D. Dissertation, Department of Psy-
chology, Harvard University, 1948.

Scharf, Bertram. "Critical Bands and the Loudness of Com-
plex Sounds Near Threshold." Unpublished Ph.D.
Dissertation, Department of Psychology, Harvard
University, 1958.

APPENDIX A

DESCRIPTION OF THE STIMULUS TAPE

166

The calibration tone from the commercially avail-
able disc recording of Fulton Lewis, Jr., broadcasting on
the ”Housing Problem," was reproduced on the low noise mag-
netic tape at an intensity level of 70 dB SPL.

INSTRUCTIONS TO SUBJECTS

You are about to hear a news broadcast. Here
are two examples of the speech to which you should
respond. The first one you may not hear. [Con-
nected discourse was presented in quiet at 45 dB
SPL.] Here is the second one. [Connected discourse
was presented in quiet at 105 dB SPL.] Sometimes
you will hear the news broadcast by itself, and
sometimes you will hear the news broadcast in the
presence of background noise, which will in some
cases be speech babble. For each of the 104 items
on your paper, I want you to put a number, 1 through
5, in the space provided. Use 5 if the speech was
too loud; use 4 if the speech was at your upper
limit of comfortable loudness, that is, if you would
not want it any louder; use 3 if the speech was
comfortable; use 2 if the speech was at your lower
limit of comfortable loudness, that is, if you would
not want it any softer; and use 1 if the speech
was too soft. Be sure that you evaluate the loud-
ness of the speech and Egg the background noise.
Remember: you are to put a number, 1 through 5,
in each numbered space. Be sure you fill in every_
space. Are you ready?

STIMULUS MATERIALS

 

 

RANDOM- INTENSITY LEVEL
IZED STIMULUS OF CONNECTED DIS- BACKGROUND
ORDER NO. COURSE (DB SPL) CONDITION'
1 90 80 Quiet
2 81 100 Quiet
3 98 60 Quiet
4 72 60 Speech Babble

 

‘The intensity level of wide band white noise, nar—
row band white noise, and speech babble was held constant
at 70 dB SPL where present.

167

 

 

RANDOM- INTENSITY LEVEL -
IZED STIMULUS OF CONNECTED DIS— BACKGROUND
ORDER NO. COURSE (DB SPL) CONDITION
5 64 80 Speech Babble
6 52 45 Narrow Band White Noise
7 51 45 Narrow Band White Noise
8 _ 28 105 Narrow Band White Noise
9 19 60 Wide Band White Noise
10 63 80 Speech Babble
ll 96 65 Quiet
12 44 65 Narrow Band White Noise
13 9 85 Wide Band White Noise
14 74 55 Speech Babble
15 15 70 Wide Band White Noise
16 2 105 Wide Band White Noise
17 l 105 Wide Band White Noise
18 35 85 Narrow Band White Noise
19 61 85 Speech Babble
20 10 85 Wide Band White Noise
21 60 90 Speech Babble
22 103 45 Quiet
23 24 50 Wide Band White Noise
24 6 95 Wide Band White Noise
25 31 95 Narrow Band White Noise
26 32 95 Narrow Band White Noise
27 62 85 Speech Babble

28 5 95 Wide Band White Noise

168

 

 

RANDOM- INTENSITY LEVEL
IZED STIMULUS OF CONNECTED DIS- BACKGROUND
ORDER NO. COURSE (DB SPL) CONDITION
29 73 55 Speech Babble
30 91 75 Quiet
31 95 65 Quiet
32 79 105 Quiet
33 104 45 Quiet
34 43 65 Narrow Band White Noise
35 99 55 Quiet
36 12 80 Wide Band White Noise
37 36 85 Narrow Band White Noise
38 14 H 75 Wide Band White Noise
39 94 70 Quiet
40 67 70 Speech Babble
41 87 85 Quiet
42 ll 80 Wide Band White Noise
43 4 100 Wide Band White Noise
44 101 50 Quiet
45 77 45 Speech Babble
46 69 65 Speech Babble
47 45 60 Narrow Band White Noise
48 25 45 Wide Band White Noise
49 68 70 Speech Babble
50 16 70 Wide Band White Noise
51 54 105 Speech Babble

52 4O 75 Narrow Band White Noise

169

 

 

RANDOM- INTENSITY LEVEL

IZED STIMULUS OF CONNECTED DIS- BACKGROUND

ORDER NO. COURSE (DB SPL) CONDITION
53 65 75 Speech Babble
54 84 95 Quiet
55 13 75 Wide Band White Noise
56 38 80 Narrow Band White Noise
57 33 90 Narrow Band White Noise
58 18 65 Wide Band White Noise
59 58 95 Speech Babble
60 29 100 Narrow Band White Noise
61 82 100 Quiet
62 80 105 Quiet
63 48 55 Narrow Band White Noise
64 71 60 Speech Babble
65 83 95 Quiet
66 56 100 Speech Babble
67 7 90 Wide Band White Noise
68 3 100 Wide Band White Noise
69 27 105 Narrow Band White Noise
70 89 80 Quiet
71 53 105 Speech Babble
72 8 90 Wide Band White Noise
73 76 50 Speech Babble
74 86 90 Quiet
75 17 65 Wide Band White Noise

76 37 80 Narrow Band White Noise

170

 

I

 

RANDOM- INTENSITY LEVEL

IZED STIMULUS OF CONNECTED DIS- BACKGROUND
ORDER NO. COURSE (DB SPL) CONDITION

77 7O 65 Speech Babble

78 59 90 Speech Babble

79 97 60 Quiet

80 39 75 Narrow Band White Noise
81 41 70 Narrow Band White Noise
82 78 45 Speech Babble

83 26 45 Wide Band White Noise
84 46 60 Narrow Band White Noise
85 85 90 Quiet

86 34 90 Narrow Band White Noise
87 88 85 Quiet

88 49 50 Narrow Band White Noise
89 75 50 Speech Babble

9O 93 70 Quiet

91 22 55 Wide Band White Noise
92 21 55 Wide Band White Noise
93 57 95 Speech Babble

94 50 50 Narrow Band White Noise
95 30 100 Narrow Band White Noise
96 55 100 Speech Babble

97 20 60 Wide Band White Noise
98 42 70 Narrow Band White Noise
99 92 75 Quiet

100 66 75 Speech Babble

171

 

 

RANDOM- INTENSITY LEVEL
IZED STIMULUS OF CONNECTED DIS- BACKGROUND
ORDER NO. COURSE (DB SPL) CONDITION
101 102 50 Quiet
102 100 55 Quiet
103 47 55 Narrow Band White Noise

104 23 50 Wide Band White Noise

 

APPENDIX B

SUBJECT'S RESPONSE SHEET

173

 

 

 

 

 

 

Name Subject No.

Date Sex

MSU Class Age ___
Major

 

Note: After each presentation, put a number, 1
through 5, in the space provided. Use 5 for "Too Loud,“
4 for “Upper Limit of Comfortable Loudness (would not want
it any louder)," 3 for "Comfortable," 2 for "Lower Limit
of Comfortable Loudness (would not want it any softer),“

 

l for “Too Soft.“ Be sure to fill in everyyspace.
l. 22. 43. 64. 85.
2. 23. 44. 65. 86.
3. 24. 45. 66. 87.
4. 25. 46. 67. 88.
5. 26. 47. 68. 89.
6. 27. 48. 69. 90.
7. 28. 49. 70. 91.
8. 29. 50. 71. 92.
9. 30. 51. 72. 93.

10. 31. 52. 73. 94.

11. 32. S3. 74. 95.

12. 33. 54. 75. 96.

13. '34. 55. 76. 97.

14. 35. 56. 77. 98.

15. 36. 57. 78. 99.

16. 37. 58. ' 79. 100.

17. 38. 59. 80. 101.

18. 39. 60. 81. 102.

19. 40. 61. 82. 103.

20. 41. 62. 83. 104.

21. 42. 63. 84.

APPENDIX C

FIFTY-TWO AVERAGED EVALUATIONS
OF THE LOUDNESS OF CONNECTED DISCOURSE
FOR EACH SUBJECT, TOTALS, AND MEANS

CONNECTED DISCOURSE PRESENTED IN QUIET AT VARIOUS INTENSITY LEVELS
(i.e., with no noise present other than that inherent in the system)

 

SPL

INTENSITY LEVEL OF CONNECTED DISCOURSE IN DB

60

 

SUBJECT

70 75 80 85 90 95 100 105

65

50 55

45

NO.

 

0000000000000
0

175

mmmmmmmmmmmmm

OOOOOLOLOOOOOO
O O

O

mmmmmd’q‘mmmmmm

LOCI-DOOOOOLOOOOOO
O O

fi'fl'fl'fl'd‘fl'd‘d’fl'd‘fl'fi'm

oommoo

O O O 0 O O

Q'V'MMQ‘M

LnOLnLnOO
0.0000

mammmm

OLDOOOO
mmmmmm

000000
0 o o
MMNNNM

LDOOOOO
o o o
NMNNNN

CO 000

chrIchOI

OOl-OOLOO
o o o
NNI—INHN

Inu1m<3c>m
Hrqrqmrari

moomom
O O O
rifirfiriflrfi

000000
0
t-Ir-‘Ir-lr-II-II—i

HNMQ'LnkO

OLDOLDLOO
O

mmvmmd‘

mmmmo
o

O O O O O

mmmmmm

mmOOOO
o

Nmmmmm

LOOOLOOO
O

NMNNMM

000mm
000000

NMNNNM

OOLOOOOOO
O O

NNr-INNN

OmmOOO
0.0000

Nr-IHNNN

H

13
14
15‘

O
0
LO

L0

1.0

OOOOOOO

mmmmmmm

OOOLOOOLn
0000000

LOLDIDQ‘LDLOQ‘

OLOOLnLnOO
0000000

Lnfi‘d'MQ'Lnfl'

OOLOOOOO

D O O O O O O

Lnd'mmfi‘fi'fi‘

LnOLnOOLnLn
0000000

aammmmm

OOOOLOOLO

MMMMMNM

OOOmOOO
o o O o o o O
MMMNMNM

0000000
0 o o o o o o
mmmmmNm

LDOOOOLOO

NNNNNr-lm

OOOOOLnO
o o o 0 o o o
NNNNNr—Im

OOOOOOO

r-INNNNHN

0000000
0 O O O O O O
HNHNNF‘IN

 

51.5 55.0 63.5 73.0 78.5 90.0 103.0 105.0
2.62 3.02 3.74 4.29

42.0

35.0 40.5

24.0 30.0

TOTALS
MEANS

 

5.00

4.90

3.48

1.43 1.67 1.93 2.00 2.45

1.14

 

'The responses of Subject No. 15 could not be used in the analysis of the data.

H, .'.

 

SPL SPEECH BABBLE

CONNECTED DISCOURSE AT VARIOUS INTENSITY LEVELS
PRESENTED WITH 70 DB

 

INTENSITY LEVEL OF CONNECTED DISCOURSE IN DB SPL

60

 

SUBJECT

70 75 8O 85 9O 95 100 105

65

45 50 55

NO.

 

176

OOOOOOLOOOOOOOO
00000000000000

mmmmmmvmmmmmmm

OOLDLDLOOOLOOLOLDOOU')

Lnl-Ofi'fi'fl'd'fi'vmfi'fi‘mmfi'

OOLOOOOOOOOLOLBOO
O O

mmvmsr'd'd'mvmd'd'mm

d'fi'fi'fi'fi'MMV'fi‘d‘fi'fi'd‘fi'

OOOOOLDOOOOOOOU)
0 0 0 0 0 0 0 0 0 0

LOOOOUWOOLOOOOOLDLD

MQ'MMCOMMC‘OMMMMMM

ommmmomoomoomo

MNNNNMNMMNMMNM

OOOLOOOOLDOOOOOLO
0 0 0

MNNNNNNNNNMMNN

OLnLOOOOOOOOOOOO
00000000000000

NHHNNNNNNNNNNN

CLOOLOLOOOOOOLDLDLD

O
O O O O O O O O O O O O O 0
H

HHHHHHHHHHHHH

OQOLOOOOOOOOLDOO

HrarIHr4r+Hr4r4Hr4r+Hr4

OOOOOOOOOOLOOOO
O O

HrfifiifirfiriflrflriflrfiriflrH

00000000000000
00000000000000

Hrfiriflrfiriflrflriflr4riflr4

00000000000000

0 O O O O O O O O O O O 0 O

r-It-‘lr—Iv-ir-lr-IHI-‘ll-iv-lt-lt-Ir-lr-l

HNMQ‘LOKOFCDCDOHNMQ‘
r-{r-lr-lr-ir-‘l

15‘I

OOOmOOO
O O

LnLnLnfl'LOLnLO

OOLnLnOOLn
0 0

mmfid‘mmv

CODINOOLD

mmmd'tnmv

OOLnoOOO

O O O O O O O

mfi'mmfl'fl‘d'

mmmooom
O

srmmmmmm

LnOOOOI-nm
0000000

MMMMMNM

mmOOOOO
0000000

HNMNNHM

OOLOOLOOO
O

HNNNt-lt-im

OOOOLDOO
0 0 0 0 0 0 0
r—II-‘Iu-lr-{t-II—IN

OOOOOOLD

Araraarqrda

00000011.)
0 0 0 0 0 0 0
HHHHHHH

0000000

0 O O O 0 O O

HrfiriHr4r1H

0000000

0000000
HI—lr-Ir-lv-it-lr-I
0
H

[\COQOHN
HHr—INNN

 

69.5 82.5 98.5 98.0 104.0
3.93 4.69

22.5 25.5 40.0 47.5 60.0

22.0

21.0

TOTALS 21.0

 

4.95

4.67

1.00 1.05 1.07 1.21 1.90 2.26 2.86 3.31

1.00

MEANS

*The responses of Subject No. 15 could not be used in the analysis of the data.

 

 

100 105

65 70 75 80 85 90 95

INTENSITY LEVEL OF CONNECTED DISCOURSE IN DB SPL

CONNECTED DISCOURSE AT VARIOUS INTENSITY LEVELS
60

PRESENTED WITH 70 DB SPL WIDE BAND WHITE NOISE

55

50

 

45

NO.

 

 

SUBJECT

177

 

 

m 0
OOOOOOLOOOOOOOO OOOOOO Ln Ox (0
00000000000000 0000000004.)
mmmmmmommmmmmm LnLnLnLnLnLnLn V. V' (6
O 'U
r-l
Q)
v-I .C
oooomomooomooo OOOOOOLD O G) JJ
0 O O O O O O O O O O O O O O O O O O O O O O
tnLnLnLnsrv'd'LnLnLn-d'mmm LnLnLnd'Lnl-nd' H V. “a
O
H
U)
(h H
OU‘IOOOLDOOOLOOLDOO LOOOOOLD O N U)
00000000000000 000000000»
d‘fi'md'd'MQ'fi'fi'd‘Sfd'mm d'md‘d'd‘fl'fi' g Q' 'r—‘d
C1
<1' (6
LOOOO OOOOOOOOO OOOOOOLD O N
00000000000000 0000000000
mmmmmmmmmmmvmfi 00mm¢NmEBm 5 ;
‘2‘.
:4 C A
mmmoommmmoooom oomommouvm H ;
00000000000000 000000000
cnawmrnwuncudunnumcnwnn acndnnavm<w b-cn U
to 0)
U)
m :3
OOLOLOOOOOLOOOOMO OOanOOO Ln 1‘
00000000000000 0000000000
MMNNMMNMNMMMNM MMNMMNM CD N ,Q
L“
4.)
O O
OOOOOOOOLOOOOOO OLOOOOOO O Ch I:
O O O O C O O O O O O O O O O O O O O O O O O
NCVQHVOJNCVOHAOQNCVQHV Hr4nuuouflcn 0 r4 '0
d' r-I
:3
m 0
OOOLO LOLOOOLDLDOOO 0000000 0 N U
0 O O O 0 O O O O O O O O O O O O O O O O O O
N 0
00000000000000 OOOOOOLO Ln 0 0
00000000000000 0000000002
HHHr-lr-ir-ir-ir-{r-{r-lc-lr-IHH v-lr-lI-lc-lr-lr-lr-l a H 4.)
U
l‘ (I)
OOLOOOOOOOOOOOO OOOOOOO Ln 0 'f'fi
00000000000000 000000000'Q
d‘ 'H
OOOOOOOOOOOOOO 0000000 O H 0
00000000000000 000000000
rariarAFIHrAPIHrqriararq r4F1Hr4FiHr4 a r4 3
0')
<1‘ £1
OOOOOOOOOOOOOO 000000 O H 0
00000000000000 000000000Q
O 54
00000000000000 0000000 O H (1)
00000000000000 000000000£
HrfiCIHrarIHrarIHrquHra Hrariararia $3.4 B
Q
U)
0 Q U)
u—INMfl'LnKOl‘CDmOr—INMVLDKOFQDOOHN E
rariararaararaarAOJNcu B
O (:1
E4 Z

 

 

 

85 90 95 100 105

80

SPL NARROW BAND WHITE NOISE
75

7O

INTENSITY LEVEL OF CONNECTED DISCOURSE IN DB SPL
65

CONNECTED DISCOURSE AT VARIOUS INTENSITY LEVELS
60

PRESENTED WITH 70 DB

55

50

 

45

NO.

 

 

SUBJECT

178

00000000000000
0

mmmmmmmmmmmmmm

OOOOOLOOOOOLOOOO

mmmmm-d'mmtnm-d'mmm

OOOLnLnOOOLnLnOLnOO
00000000000000

fifi'd‘fi‘fi‘d‘d‘mfl'fl'fi'd‘MQ'

COLDOLOLOOOOOLOOOO
O

O O O O 0 O O O O O O O O

V'Q'MQ'MM Q'Q‘Q‘MV'Q‘SI'

OOLnOOOOOLnOOOOO
00000000000000

mommmmmmmmmmmm

OOLOOLOOOLOLOLDOLDLDLO
O O

MMNMNNNNNNMNNN

OOOOLOOOLOIDOLDLOOO
O. O.
NNNN NNNHNNNNN

2.

OLnOLnLnOOOOLnOOOO
0000 0
Nt—INHHHNNt-‘r—INNNN

O
O
O
O
O
O

OOOOOOOOOOLOOOO
O C O O O O O O

0000

OOLOOOOOOO
0000 0
r-lt-{r—lc—‘I

HHHHHHHHH

1.0

00000000000000
0 O O O O
Hr4r4ar4r+araraarar4ar4

00000000000000
0 0°00
HHHHHHHHHHHI—lr—IH

OOl-OOOOOOOOOOOO
0 000°
HHMNHHHHHHHHHH

HNMfl'LnkOPCDO‘O
H

11

. . 1.0 1.0 1.0

1.0

0000000
0 I O O

InanLnLnLn

OOOOO

lnLnLnLnLnQ‘

OOOOLOO
0 0 0 0 0 0

Ln'd'd'd'fi‘d‘

000000
0
Q'V'MV‘V‘fi‘

LDOOOOO
0 0 0 0 0 0
mmmmmv

OLnOLnOO
O
MNNNNM

OOOLDLD
O

NNNNr-im

LOOOOU)
00000

HHNr-IN

OOOOOLO

r-Ir-ir-iu-lr-{H

22.0

22.0

1.0
21.0

1.0
21.0

1.0

18
19
20
21
22
TOTALS 24.5

105.0
5.00

103.5
4.93

’ ._T’a.'_.. '-.

“.0.

3.21 3.86 4.33

 

67.5 81.0 91.0

35.0 45.0 54.0
1.00 1.00 1.05 1.05 1.67 2.14 2.57

17

1.
‘The responses of Subject No. 15 could not be used in the analysis of the data.

 

 

 

MEANS

APPENDIX D

CRITICAL DIFFERENCES AMONG
THE MEANS OF CELLS

 

 

 

 
 

 

 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  
  
 
  
 
 

c1 I8 3.02
c2 18 2.86
C3 18 2.79
C4 I8 2.57
CI I9 3.48
c2 I9 3.31
c3 19 3.21
c4 19 3.21
C1 110 3.74
02 110 3.93
c3 :10 3.24
C4 110 3.86
CI 111 4.29
C2 111 4.69
c3 111 4.29
c4 111 4.33
Cl 112 4.90
c2 112 4.67
c3 112 4.81
04 112 4.93
C1 I13 5.00
C2 I13 4.95
4.98

 

  

various cells are explainec in Table 8
and in Figure 12.

Any difference between two means greater
16‘x"*1 .

than .22 is significant at the

.('D

.2

c2 *c3 ‘C4 c1 c2 c3j c4 c1 c2 c3 c4 c1 02’ c3’ c4 c1 c2' c3 c4 c1 c2 c CI——
Cells Mggpgfi I1 11' I1 12 12 12 I2 13 I3 I3 I3 14 14 14 I4 IS I5 IS IS I6 I6 16 I: 75% 757 I; 17
C1 11 1.14 .14 .04 .03 .29 .14 —— .14 .53 .09 —— .14 .79 .07 .07 .09 .86 .07 .12 .09 1.31 .76 .15 .53 1.48 1.12 76 1 00
c2 11 1.00 .10 .17 .43 -- .14 -— .67 .05 .14 —— .93 .07 .07 .05 1.00 .21 .02 .05 1.45 .90 .29 .67 1.62 1.26 '90’1'14
c3 11 1.10 .07 .33 .10 .04 .10 .57 .05 .04 .10 .83 .03 .03 .05 .90 .11 .08 .05 1.35 .80 .19 .57 1.52 1 16 .80 1'0
c4 11 1.17 .26 .17 .03 .17 .50 .12 .03 .17 .76 .10 .10 .12 .83 .04 .15 .12 1.28 .73 .12 .50 1.45 1:09 '73 '9:
c1 I2 1.43 .43 .29 .43 .24 .38 .29 .43 .50 .36 .36 .38 .57 .22 .41 .38 1.02 .47 .14 24 1.19 .83 '47 '71
c2 I2 1.00 .14 —- .67 .05 .14 -— .93 .07 .07 .05 1.00 .21 .02 .05 1.45 .90 .29 .67 1.62 1.26 .90 1'14
c3 12 1.14 .14 .53 .09 —— .14 .79 .07 .07 .09 .86 .07 .12 .09 1.31 .76 .15 .53 1.48 1.12 .76 1.00
c4 12 1.00 .67 .05 .14 -- .93 .07 .07 .05 1.00 .21 .02 .05 1.45 .90 .29 .67 1.62 1.26 .90 1.14
CI I3 1.6 .62 .53 .67 .26 60 60 .62 .33 .46 .65 .62 .78 .23 .38 -— .95 .59 23 '47
02 I3 1.0 .09 .05 88 .02 02 -— .95 .16 .03 -— 1.40 .85 .24 .62 1.57 1.21 85 1'09
c3 13 1.14 14 .79 .07 .07 .09 .86 .07 .12 .09 1.31 .76 .15 53 1.48 1.12 76 1'00
c4 I3 1.00 .93 .07 07 .05 1.00 .21 .02 .05 1.45 .90 .29 .67 1.62 1.26 .90 1'14
c1 I4 1,93 .86 86 .88 .07 .72 .91 .88 .52 .03 .64 26 .69 .33 .03 :21
c2 I4 1.07 —— .02 .93 .14 .05 .02 1.38 .83 .22 60 1.55 1.19 .83 1.07
c3 I4 1.07 .02 .93 .14 .05 .02 1.38 .83 .22 60 1.55 1.19 .83 1.07
c4 14 1,05 .95 .16 .03 —— 1.40 .85 .24 62 1.57 1.21 85 1.09
CI 15 2.00 .79 98 .95 45 .10 71 33 .62 .26 .10 .14
c2 IS 1.21 19 16 1 24 .69 08 46 1.41 1.05 69 .93
c3 15 1,02 03 1.43 .88 27 .65 1.60 1.24 88 1.12
c4 15 1.05 1 40 .85 .24 62 1.57 1.21 .85 1.09
C1 16 2.45 .55 1.16 78 .17 .19 .55 .31
02 I6 1.90 .61 .23 .72 .36 -— .24
c3 I6 1.29 38 1.33 .97 .61 .85
c4 I6 1.67 .95 .59 23 .47
c1 I7 2:62 .36 72 .48
c2 I7 2.26 .36 .12
c3 I7 1.90 .24
c4 I7 2.14
NOTE: The symbols used here to describe the

 

 

 

 

 

 

 

C1 C2 C3 c4 c1 C2 C3 C4 C1 C2 C3 C4 C1 ”C2 C3 C4 ”’CI’ C2 C3 4
18 IS 18 18 I9 I9 I9 I9 110 110 110 110 11]. Ill 111 .111 112 112 112 (1:12 §i3 $3 $3.33 :3:
1.88 1.72 1.65 1.43 2.34 2.17 2.07 2.07 2.60 2.79 2.10 2.72 3.15 3.55 3.15 3.19 3.76 3.53 . 7
2.02 1.86 1.79 1.57 2.48 2.31 2.21 2.21 2.74 2.93 2.24 2.86 3.29 3.69 3.29 3.33 3.90 3.67 3.31 3.93 2.38 g'gé 3'3: 2'33
1.92 1.76 1.69 1.47 2.38 2.21 2.11 2.11 2.64 2.83 2.14 2.76 3.19 3.59 3.19 3.23 3.80 3.57 3.71 3.83 3.90 3:85 3.88 3'90
1.85 1.69 1.62 1.40 2.31 2.14 2.04 2.04 2.57 2.76 2.07 2.69 3.12 3.52 3.12 3.16 3.73 3.50 3.64 3.76 3.83 3.78 3.81 3.83
1.59 1.43 1.36 1.14 2.05 1.88 1.78 1.78 2.31 2.50 1.81 2.43 2.86 3.26 2.86 2.90 3.47 3.24 3.38 3.50 3.57 3.52 3 55 3'57
= 2.02 1.86 1.79 1.57 2.48 2.31 2.21 2.21 2.74 2.93 2.24 2.86 3.29 3.69 3.29 3.33 3.90 3.67 3.81 3.93 4.00 3.95 3.98 4.00
1.88 1.72 1.65 1.43 2.32 2.17 2.07 2.07 2.60 2.79 2.10 2.72 3.15 3.55 3.15 3.19 3.76 3.53 3.67 3.79 3.86 3.81 3.84 3'86
2.02 1.86 1.79 1.57 2.48 2.31 2.21 2.21 2.74 2.93 2.24 2.86 3.29 3.69 3.29 3.33 3.90 3.67 3.81 3.93 4.00 3.95 3.98 4.00
1.35 1.19 1.12 .90 1.81 1.64 1.54 1.54 2.07 2.26 1.57 2.19 2.62 3.02 2.62 2.66 3.23 3.00 3.14 3.26 3.33 3.28 3.31 3.33
1.97 1.81 1.74 1.52 2.43 2.26 2.16 2.16 2.69 2.88 2.19 2.81 3.24 3.64 3.24 3.28 3.85 3.62 3.76 3.88 3.95 3.90 3.93 3.95
1.88 1.72 1.65 1.43 2.34 2.17 2.07 2.07 2.60 2.79 2.10 2.72 3.15 3.55 3.15 3.19 3.76 3.53 3.67 3.79 3.86 3.81 3.84 3.36
2.02 1.86 1.79 1.57 2.48 2.31 2.21 2.21 2.74 2.93 2.24 2.86 3.29 3.69 3.29 3.33 3.90 3.67 3.81 3.93 4.00 3.95 3.98 4.00
1.09 .93 .86 .64 1.55.1.38 1.28 1.28 1.81 2.00 1.31 1.93 2.36 2.76 2.36 2.40 2.97 2.74 2.88 3.00 3.07 3.02 3.05 3.07
1.95 1.79 1.72 1.50 2.41 2.24 2.14 2.14 2.67 2.86 2.17 2.79 3.22 3.62 3.22 3.26 3.83 3.60 3.74 3.86 3.93 3.88 3.91 3.93
1.95 1.79 1.72 1.50 2.41 2.24 2.14 2.14 2.67 2.86 2.17 2.79 3.22 3.62 3.22 3.26 3.83 3.60 3.74 3.86 3.93 3.88 3,91 3,93
1.97 1.81 1.74 1.52 2.43 2.26 2.16 2.16 2.69 2.88 2.19 2.81 3.24 3.64 3.24 3.28 3.85 3.62 3.76 3.88 3.95 3.90 3.93 3.95
1.02 .86 .79 .57 1.48 1.31 1.21 1.21 1.74 1.93 1 24 1 86 2.29 2.69 2.29 2.33 2.90 2.67 2.81 2.93 3.00 2.95 2.98 3.00
1.81 1.65 1.58 1.36 2.27 2.10 2.00 2.00 2.53 2.72 2.03 2.65 3.08 3.48 3.08 3.12 3.69 3.46 3.60 3.72 3.79 3.74 3.77 3.79
2.00 1.84 1.77 1.55 2.46 2.29 2.19 2.19 2.72 2.91 2.22 2.84 3.27 3.67 3.27 3.31 3.88 3.65 3.79 3.91 3.98 3.93 3.96 3.98
1.97 1.81 1.74 1.52 2.43 2.26 2.16 2.16 2.69 2.88 2.19 2.81 3.24 3.64 3.24 3.28 3.85 3.62 3.76 3.88 3.95 3.90 3.93 3.95
.57 :41 .34 .12 1.03 .86 .76 .76 1.29 1.48 .79 1.41 1.84 2.24 1.84 1.88 2.45 2.22 2.36 2.48 2.55 2.50 2.53 2.55
1.12 .96 .89 .67 1.58 1.41 1.31 1.31 1.84 2.03 1.34 1.96 2.39 2.79 2.39 2.43 3.00 2.77 2.91 3.03 3.10 3.05 3.08 3.10
1.73 1.57 1.50 1.28 2.19 2.02 1.92 1.92 2.45 2.64 1.95 2.57 3.00 3.40 3.00 3.04 3.61 3.38 3.52 3.64 3.71 3.66 3.69 3.71
1.35 1.19 1.12 .90 1.81 1.64 1.54 1.54 2.07 2.26 1.57 2.19 2.62 3.02 2.62 2.66 3.23 3.00 3.14 3.26 3.33 3.28 3.31 3.33
.40 .24 .17 .05 .86 .69 .59 .59 1.12 1.31 .62 1.24 1.67 2.07 1.67 1.71 2.28 2.05 2.19 2.31 2.38 2.33 2.36 2.38
.76 .60 .53 .31 1.22 1.05 .95 .95 1.48 1.67 .98 1 60 2.03 2.43 2.03 2.07 2.64 2.41 2.55 2.67 2.74 2.69 2.72 2.74
1.12 .96 .89 .67 1.58 1.41 1.31 1.31 1.84 2.03 1.34 1.96 2.39 2.79 2.39 2.43 3.00 2.77 2.91 3.03 3.10 3.05 3.08 3.10
.88 .72 .65 .43 1.34 1.17 1.07 1.07 1.60 1.79 1.10 1.72 2.15 2.55 2.15 2.19 2.76 2.53 2.67 2.79 2.86 2.81 2.84 2.86
.16 .23 .45 .46 .29 .19 .19 .72 .91 .22 .84 1.27 1.67 1.27 1.31 1.88 1.65 1.79 1.91 1.98 1.93 1.96 1.98
.07 .29 .62 .45 .35 .35 .88 1.07 .38 1.00 1.43 1.83 1.43 1.47 2.04 1.81 1.95 2.07 2.14 2.09 2.12 2.14
.22 .69 .52 .42 .42 .95 1.14 .45 1.07 1.50 1.90 1.50 1.54 2.11 1.88 2.02 2.14 2.21 2.16 2.19 2.21
.91 .74 .64 .64 1.17 1.36 .67 1.29 1.72 2.12 1.72 1.76 2.33 2.10 2.24 2.36 2.43 2.38 2.41 2.43
.17 .27 .27 .26 .45 .24 .38 .81 1.21 .81 .85 1.42 1.19 1.33 1.45 1.52 1.47 1.50 1.52
.10 .10 .43 .62 .07 .55 .98 1.38 .98 1.02 1.59 1.36 1.50 1.62 1.69 1.64 1.67 1.69
-— .53 .72 .03 .65 1.08 1.48 1.08 1.12 1.69 1.46 1.60 1.72 1.79 1.74 1.77 1.79
.53 .72 .03 65 1.08 1.48 1.08 1.12 1.69 1.46 1.60 1.72 1.79 1.74 1.77 1.79
19 .50 12 55 .95 .55 .59 1.16 .93 1.07 1.19 1.26 1.21 1.24 1.26
69 .07 .36 .76 .36 .40 .97 .74 .88 1.00 1.07 1.02 1.05 1.07
.62 1.05 1.45 1.05 1.09 1.66 1.43 1.57 1.69 1.76 1.71 1.74 1.76
.43 .83 .43 .47 1.04 .81 .95 1.07 1.14 1.09 1.12 1.14
.40 -— .04 .61 .38 .52 .64 .71 .66 .69 .71

.40 .36 .21 .02 .12 .24 .31 .26 .29 .31
.04 .61 .38 .52 .64 .71 .66 .69 .71

.57 .34 .48 .60 .67 .62 .65 .67

.23 .09 .03 .10 .05 .08 .10

.14 .26 .33 .28 .31 .33

.12 .19 .14 .17 .19

.07 .02 .05 .07

.05 .02 ——

.03 .05

.02

 

 

 

 

 

 

APPENDIX E

CRITICAL DIFFERENCES AMONG
THE RANGES OF CELLS

.m 0Hnma cH

00CH0HQX0 mum mHHmu OSOHum> mcu mnauuwwn ou mum: Umm: mHonemm 0:9 «maoz

.mucmvfimcou mo Hm>ma flm um pCMUHMHcmHm mamme 03¢ cmmzumn mucmummmaoo

183

 

 

m 0 OH
O Ommu O
I: O mmNu OH
I: O n: mmHu OH
1: O In I: Nmeu OH
O OH O O O Nmmu OH
O OH O O O I: NmNu OH
OH ON OH OH OH OH OH NmHu ON
ON .ON ON ON ON OH OH O Hmvu Om
ON .ON ON ON ON OH OH O I: Hmmu OO
OH ON OH OH OH OH OH I: O O HmNu ON
O OH O O O I: I: OH OH OH OH HmHu OH
«6 Omm mN mHu mvu NO ON mH Hmwu Hmmu HmNu OHHmu Ommcmm

 

f