OFFICE OF SCIENTIFIC RESEARCH AND DEVELOPMENT
NATIONAL DEFENSE RESEARCH COMMITTEE
Articulation Testing Methods !!
THIS REPORT IS
UNCLASSIFIED
1 November 1944
OSRD NO. 3802.
NORTHWESTERN
UNIVERSITY
LIBRARY
EVANSTON
ILLINOIS
0^ 0^ nk 0^ 0n 0n 0^ #n
(¿- 5. OFFICE OF SCIENTIFIC RESEAECÏÏ AND DEVELOPMENT.
NATIONAL DEFENSE RESEAECH COMMITTEE
DIVISION 17 SECTION I7.3
Contract OEMsr-658
OSRD Report No. 3002
1 Noveiaber ISkk
ARTICULATION TESTING METHODS II
(Relevant to Project NA-Í08)
RESEARCH ON SOUND CONTROL
Psycho-Acoustic Laboratory
Harvard University
Cambridge, Massachusetts
THIS REPORT IS
K E s I R I C T E E .jNCUSSiriED
EEPOET PREPARED BY
J.P. Egan
With the Cooperation of
Members of the Psycho-
Acoustic Laboratory
APPROVED FOE THE CONTRACTOE
S,S, Stevens
Director, Psycho-Acoustic'Laboratory
Harvard University
APPROVED FOR NDEC
Harvey Fletcher
Chief, Section 17.3
P.M. Morse
Project Supervisor
TABLE OF CONTENTS
Page
I INTRODUCTION 1
II TEST MATERIAL 4
I. Représentâtion of fundamental speech sounds. 6
II. Types of test items 6
III. The difficulty and reliability of test lists 10
III AVAILABLE LISTS OF SYLIABLES, WORDS, AND SENTENCES . l4
I. Syllable lists li<-
II. Monosyllabic word lists . l8
III. Spondaic word lists ..... .23
rv. Sentence lists 2^1
V. Other test materials ............ 25
IV METHODS OF CONDUCTING ARTICULATION TESTS 27
I. Selection of testing personnel 28
II. Importance of training and fatigue 33
III. Selection of complementary equipment .... 36
TV. Selection of ambient noise conditions ... 36
V. Intensity of received speech ^<-1
VI. Use of a carrier sentence and time
interval between test items k6
VII. Instructions to testing personnel ij-7
VTII. Statistical methods W
IX. List of factors affecting articulation
scores 51
V ABBREVIATED TESTING METHODS 5^
VI COMPARISONS BASED UPON SUBJECTIVE APPRAISAL^ .... 55
VII THRESHOLD METHODS FOR EVALUATING THE INTELLIGIBILITY
OF SPEECH 58
APPENDIX I ' Syllable lists 6l
APPENDIX II PB word lists 66
APPENDIX III Spondaic word lists 72
APPENDIX IV Sentence lists 7^
APPENDIX V R word lists 92
EESTRICTED
I INTRODUCTION
There are many reasons why oral communication may he In¬
efficient or fall completely. Distortion of speech sounds hy
inferior transducers, msklng of those sounds hy ambient or
electrical noises, Improper use of microphones and headsets
hy communications operators, all conspire to make communication
uncertain and unreliable. In order to emluate the relative ^
Importance of those factors which Influence the Intelligibility
of speech, methods are required hy which to determine quantita¬
tively the degree of Intelligibility of speech. This report
presents a description of some of the methods found useful for
this pvirpose at the Harvard Psycho-Acoustic Laboratory. Its
purpose Is also to make available to other laboratories en^ged
In research on communication the lists of syllables, words, and
sentences constructed and extensively used at this laboratory.
To the extent that these test materials represent em Improve¬
ment over those contained In our earlier discussion of articu¬
lation testing methods,* this report may be regarded as super¬
seding the earlier account.
A quantitative measure of the intelligibility of speech
may be obtained by counting the number of discrete speech units
correctly recognized by a listener. The procedure by which this
quantitative measure Is obtained Is an articulation test. Typi¬
cally, an announcer reads lists of syllables, words, or sen¬
tences to a group of listeners, and the percentage of Items
correctly recorded by these listeners Is called the articula¬
tion score. This percentage Is taken as a measure of the In¬
telligibility of speech.
* NRC Committee on Sound Control report, "Articulation Test-
Ing Methods," 1 February 19^2, OSRD No. 383.
RESTRICTED
-2-
Articulatlon testing methods may he used In the solution
of many problema of communication. These problems may be con¬
veniently classified under four major headings:
1) Comparison of communication devices,
2) Evaluation of the effects of noises on com¬
munication,
3) Measurement of the basic audibility of words
and commands,
U) Eating and training of commimications personnel.
Several factors, such as those listed below, may be evalu¬
ated by controlled articulation tests. For example, communica¬
tion systems differ markedly in their ability to transmit In¬
telligible speech. Under favorable conditions most cofflm\inlca-
tlcsn devices are adequate, but under adverse conditions mny
systems fall to provide reliable communication. Frequency and
amplitude distortion may be Introduced by transducers or ampli¬
fiers. When communication is attempted In noisy surroundings,
the speech may be masked by Interfering noise. Microphones and
headphones differ In the amount of protection they provide
against this noise.
Static and man-made Interference are further hazards to
communication. The detrimental effects of these noises and
the degree to which they may be tolerated In a radio circuit
must be determined and evaluated. Which type of noise Is pre¬
sent makes eonslderable difference In the extent to which
speech is masked. A comparison of the effectiveness of vari¬
ous types of noises by articulation testing methods has Import¬
ant applications in the "Jamming" of enemy messages.
Large differences exist between the basic audibility of
words and commands. Communication can be mde more effective
by selecting those words and phrases which can he heard either
above the din of noise or over systems which distort the sounds
of speech. Again, articulation tests provide the tool for se¬
lecting military vocabularies.
Articulation tests have been applied to the rating and
training of communication personnel. Because listeners may be
tested by these methods In relatively large numbers, the arti¬
culation test may be used to provide a practical and accurate
rating of listening atility.r Since articulation tests are
somewhat time-consuming, however, the rating of talkers can
usually he made more easily hy ahhreviated methods, although
articulation tests have proved useful in hasic studies of
talking ability.
Although a quantitative measure of the intelligibility of
speech is the most completely informative, a qualitative evalu¬
ation is sometimes adequate. Frequently it is sufficient to
have competent observers evaluate the quality of a given sam¬
ple of speech. These observers nay rank order various samples
of speech in terms of the relative intelligibility of the sam¬
ples, or they may describe the speech in terms of some stan¬
dard of reference. If properly used, these "subjective" methods
of appraisal can be employed to advantage. A brief description
of these methods is included below.
For certain other purposes, the results obtained by deter¬
mining the thresholds at which speech becomes just detectable,
just perceptible, or easily intelligible, are of value. In
determining these thresholds the listener adjusts some variable
or condition until the appropriate threshold is reached. These
procedures will be discussed later and typical results presented.
II TEST MATERIAL
In an articulation test a talker pronounces selected
speech items and a listener records the sounds that he hears.
Since the nature of the spoken items helps to determine the
resulting articulation scores, the test material must "be care¬
fully selected if a proper assessment is to he made. For most
testing purposes the speech so^lnds used should be representa¬
tive of convorsational speech. Furthermore/ for cconoiby of
effort and time, it is important to group the speech units used
for the articulation test into convenient test lists, each list
as difficult as each other listi When lists of Comparable dif¬
ficulty are used, differences in articulation scores obtained
with two different microphones may be interpreted as due to dif
ferences in the instruments rather than to differences in the
difficulty of the lists.
In order to construct lists of equal difficulty, repre¬
sentative of conversational speech and suitable for testing,
we must consider the following:
1. Representation of fundamental speech sounds.
a. All, or nearly all, of the fundamental soimds
into which speech can be analyzed should be
represented in each list of test items.«
b. Ideally, the relative frequencies of occui^nce
of those fxmdamental speech sounds should re¬
flect the distribution in normal speech. In
any case, it must be remembered that such fac¬
tors as the equipment and the type of interfer¬
ing noise may alter the relative difficulty of
different speech sounds.
2, Types of test items.
a. The particular comibinations of the fundamental
speech sounds may be nonsense syllables, words,
or sentences.
b. The peychologlcal factors of meaning, context,
rhythm, inflection, etc.. Introduced by these
combinations partially determine the relative
difficulty of the test material. Consequently,
even though the lists are balanced with res¬
pect to phonetic"composition. It Is usually
necessary to demonstrate by actual teste that
each list is as difficult as each other list.
3. Difficulty and reliability of test lists.
a. The range of difficulty provided by the lists
should be such that there will be few items
which are too easy or too dlfflcxilt under the
conditions chosen for testing. Iteias which are
always or never correctly recorded by the
listeners are useless for measuring differences
between two items of equipment or two condi¬
tions of testing.
b. Short lists are desirable so that the articu¬
lation scores will not be unduly influenced
by factors which tend to raise or lower the
articulation scores throughout the reading of
a given tost list. Since articulation scores
of any desired reliability may be obtained by
increasing the number of tests, it is not
necessary to use a large nuniber of test items
for each individual test.
c. For an extensive testing, program, a large num¬
ber of reasonably short and homogeneous test
lists should be available. When too few lists
of words or sentences are used, the crew of
listeners tends to leam the particular group¬
ing of Items, even when the order of the Items
is changed each time that list is read.
If valid and reliable articulation tests are to be con¬
ducted, each of these factors must be considered in select¬
ing suitable lists of speech units. Each of these factors
will now be discussed more fully.
I. BEPEESENTATION OF FUNDAMENm SPEECH SOUNDS
The desirability of a proportional representation in the
test lists of the sounds which occur in everyday speech stems
from a consideration of the problem of validity. A microphone
which passed only one type of sound might test well with a list
of words containing only such sounds, but the test would not be
a valid indication of the usefulness of this instrument for or¬
dinary conversation. For example, if a set of test lists con¬
tained no nasal sounds, it might not provide a fair test of a
microphone worn coupled to the mouth by certain types of noise
shields.
As a practical matter, the actual distribution of sounds
in speech depends upon whose speech it is and what is being
talked about. It is possible,^ne^erthele'ss, to state approxi¬
mately the relative frequencies of occurrence of sounds in
"average" speech, and to approach these frequencies in the
distribution of sounds among the lists of an artic^lp•^■^^ on - '
In general, as the number of different speech sounds in
the test lists is increased and as the relative frequency with
which each sound occurs approaches that found in conversational
speech, the more valid will be the articulation results.
II. TYPES OF TEST ITEIvS
Since some of the phonemes cannot be pronounced as iso¬
lated units, it is necessary that they be combined with other
phonemes. Furthermore,-for many purposes it is desirable to
obtain the articulation not of the individual speech sounds
but of combinations of them. When combinations of speech
sounds are employed, the articulation scores depend not only
upon the difficulty of the individual speech sounds but also
upon the particular combination used.
* For a standard reference on the distribution or speech
sounds, see Godfrey Dewey. Relative Frequency of English
Speech Sounds, Harvard University Press, 1923-
-7-
Tests which measure how well speech sounds are recognized
fall Into three classes:
1) Tests consisting of single syllables made up of
meaningless combinations of speech sounds;
2) Tests consisting of meaningful words given out
of context as isolated units;
3) Teste consisting of meaningful phrases or sen¬
tences, in which there are contextual relations
among the words.
The principal differences amoi^these three classes depend upon
the psychological factors ofvi^aning, inflection, rhythm, etc.
The advantages and disadvantages of each of these types of com¬
binations of phoneDies are mainly determined by such psycholo¬
gical factors.
Whether syllables, words, or sentences are used for test¬
ing purposes depends upon a number of considerations. Articu¬
lation scores obtained by using lists of nonsense syllables
indicate more accxirately the number of nhonemes actually heard
by the listener than when words or sentences are used. For
example, if the nonsense syllable zad is pronounced, and if
the listener hears only he may complete the syllable with
any one of the terminal consonants. In other words, he must
hear the phoneme d in order to record the syllable correctly.
However, the number of terminal consonants and consonant com¬
pounds which can be used in place of the phoneme d in the mean¬
ingful word should is far more limited uhan if the item were
meaningless. In this case the listener could frequently record
the correct word even though he did not hear the phoneme d.
Therefore, when it is desirable to determine accurately the
particular speech sounds transmitted by an Interphone, non¬
sense syllables are superior to words or sentences as tost items.
The use of nonsense syllables, however, requires that the
testing crew be thoroughly trained. The announcer must cor¬
rectly pronounce the speech sounds and the listeners must record
with phonetic symbols the sound^ they hear. When words or sen¬
tences are used, on the other hand, the training program is far
less arduous.
-8-
The extent to which word articulation scores are influ¬
enced "by meaning. Inflection, etc., depends largely upon the
particular words chosen for the test. The accent given a
word may aid the listener in discriminating "between two such
meaningful words as ex'it and ex-ist', even though the final
consonant or consonant compound is not heard. Scores obtained
with nolvsvllables are affected more by these psychological
fact^s than are those obtained with monosyllable3.—Figure 1
shows a relation betweéîmîë number c5r~wórds~correctly recorded
and the number of sounds contained in the words. Short words
which provide loss opportunity for the operation of such fac¬
tors as inflection, meaning, etc., are usually missed more fre
quently than long words. A test list can be made difficult or
easy by varying the proportions of short and long words. When
sentence lists are scored in terms of the meaning conveyed
these psychological factors are still more important. For this
reason, sentence articulation is typically higher than word or
syllable articulation^ ^ "
Since the difficulty of a test list is determined not only
by the fundamental speech sounds but also by numerous psycholo¬
gical factors, it is necessary to demonstrate by actual test
that each test list is as difficult as each other list.
For many purposes it is important to know at least approxi¬
mately the relation between the scores obtained with various
types of articulation tests. Such a relation makes it possible
to predict from scores obtained on one kind of articulation test
the performance to be expected on a different type of test, On
the other hand, since the articulation obtained with any one
type of material is dependent upon the difficulty of the items
selected, there is no unique relation between the scores ob¬
tained with two such types of tests as word and sentence tests.
The larticular relation obtained will also depend upon other
factors, such as, for example, the experience of the crew and
the type of equipment used. However, within the limits of ex¬
perimental error, it has always been found that, when syllable
articulation increases, word or sentence articulation will also'
increase. Figure 2 shows one relation obtained experimentally
between word and sentence articulation. This relation shows
that sentence ajrbiculation is higher than the conrésponding
word articiilation.
-12-
For the particular task of appraising these two inter¬
phones, the third test list, comprised only of items of inter¬
mediate difficulty, is the "best list. This list is economical
in that it contains few items of dead weight. It provides a
sensitive measure of the relative merits of systems A and B.
The foregoing example illustrates these three important
requirements for" the distribution of test items with respect
to difficulty:
1) If test lists are to be sensitive to small dif-
ferences in intelligibility and convenient for
use, the test items must be fairly closely dis¬
tributed along a scale of difficulty.
2) Those items which under the conditions of the
tests are always recorded correctly or always
missed are dead weight and may well be omitted
from the test lists.
3) Ideally, the number of test items in a list
should be sufficiently large and the distri¬
bution of difficulty sufficiently wide to em¬
brace the requisite range as determined by
the types of communication equipment under
study.
It should be pointed out that in terms of reliability,
alone, the ease with which an articulation score could
be repeated CMi a second testing, the second list shown in Fig.*
3 is the most reliable and the third is the least reliable."
With the second list a score of 50 percent would be obtained
regardless of small random fluctuations in the subtle factors
which always affect articulation scores. Scores obtained with
the third list would be quite sensitive to these random factors.
It is plain, therefore, that a reliable, repeatable tost is notv/
necessarily a good one, and it is also obvious that a sensitive
test is apt to be inherently unreliable.
Fortunately, the reliability of an articulation score oan^^**^
always be improved by increasing the number of tests. For this
reason it is important to have available a large number of
equivalent lists of test materials, and to choose judiciously
the number to be administered at any one time.
-13-
As a final matter, it should he noted that in any actual
list of test items, the distribution of difficulty would prob¬
ably never follow the ideal schema presented in Fig. 3« In-»
stead, the items would tond to follow a bell-shaped distribué
tion along a scale of difficulty. For this reason, articula¬
tion tests are ordinarily not unifoimily sensitive over the
complete range of possible scores. The greatest sensitivity
usually results when the testing conditions are so adjusted
that scores near 50 percent are obtained. These testing con¬
ditions also provide the additional practical advantage that
about the 50 percent point there is the most room for varia¬
tion ajnong articulation scores obtained with differen^t com¬
munication systems.
-l4-
III AVAILABLE LISTS OF SYLLABLES, WORDS, AND SENTENCES
The actual task of assembling lists of materials for articu¬
lation testing has proven to be both interesting and endless.
Since the first extensive word lists were conipiled by this labor¬
atory three years ago, there has been continuous tinkering, test¬
ing, and trying by one or another research group in an effort io
improve old lists or devise new ones for special applications.
For our present piirposes it will be sufficient to consider four
main categories of test materials, and to describe briefly the
procedures used in their congoilation.
These categories are:
1) Nonsense syllables,
2) Monosyllabic words,
3) Spondaic words,
4) Sentences. .
In Appendices I to V are listed test materials of thèse
four types. In the paragraphs below we shall discuss some of
the more pertinent considerations affecting their nature and use.
I. SYLLABLE LISTS
When it is desired to determine accurately the effective¬
ness of a device in transmitting particular speech sounds, non¬
sense syllables are superior to words or sentences as test items.
Articulatiön scores obtained with lists of nonsense syllables
may be made to represent the number of phonemes actually heard
by the listener. Furthermore, since lists of nonsense syllables
are constructed by combining into syllables the fundamental
speech sounds, the relative frequency of occurrence of these
speech sounds can be adjusted as desired. On the other hand,
the uso of nonsense syllables as test items requires that the
testing crew be thoroughly trained. The announcer must pro¬
nounce the speech sounds correctly and the listeners must record
in phcaietic symbols the sounds they hear.
-15-
There are several forms of nonsense sylla"blos. One form
frequently used is obtained by permuting the consonants with
the vowels. Exançles of syllables formed in this way are te
and et. Sometimes isolated vowels and diphthongs are Included.
Another form of nonsense syllable consists of an initial con¬
sonant, a vowel, and a terminal consonant, as in the syllables
zed, mpk, and tid.
In constructing lists of nonsense syllables this laboratory
used as test items both isolated vowels and syllables. These
syllables were of the consonant-vowel and vowel-consonant type.
Most of the speech sounds in the Eevised Scientific Alphabet*
were used in forming these syllables.
The phonetic alphabet from which the nonsense syllables
were constructed consists of kO sounds of which 11 are vowels,
1»- are diphthongs, and 25 are consonants. To form the nonsense
syllables these sounds were combined, when possible, into all
of their consonant-vowel and vowel-consonant combinations. In
the construction of the test items the consonant sounds y, w,
h, and ^ were used only in the consonant-vowel form. The con¬
sonant sounds ^ and ng were used only In the vowel-consonant
fom. This procedxire provided 3^5 syllables of the consonant-
vowel type and 315 syllables of the vowel-consonant type. By
using in addition the 15 vowels and diphthongs as test items,
675 test items were available for constructing the test lists.
Since three of the test items (ew, yew, and are pronounced
in the same way, two of these items and yu) were not used.
One other syllable, hwew, was not included, because it is so
difficult to pronounce. The remaining 672 nonsense syllables
were divided into eight lists of 8U syllables each. As nearly
as was possible the lists were constructed so that each of the
IfO speech sounds occurred as often in one list as in any other list.
Since it is obviously impossible to put each specific com¬
bination of vowel and consonant sounds in every list, resort
was had to an alternative procedure in which 18 classes of soimd
combinations were set up on the basis of the categories shown in
Table 1. There the vowels are divided somewhat arbitrarily into
3 categories and the consonants into 6; and the number of possi¬
ble vowel-consonant arrangements amounts to 18 combinations of
categories. In the final lists, then, the number of syllables
from any one of the I8 classes was made, as nearly as possible,
the same from list to list.
* See Funk and Wagnalls' Standard Dictionary.
-16-
ÍDable 1
BASIC CATEGORIES OF THE FUNDAMENTAL SPEECH SOUNDS
VOWELS CONSONANTS
Long
Voiced Stops
Unvoiced Stops
■3. (as
In take or face)
b (as in bold)
P
(as in poor)
a ("
" father or arm)
d ("
" dent)
t
(" " tent)
0 ("
" orh or wall)
J ("
" Joke)
ch (" " choke)
Ô ("
" boat or low)
g ("
" gold)
k
(" " cold)
ë ("
" team or feet)
ü ("
" boot or glue)
Voiced Fricatives
Unvoiced Fricatives
V (as
in vow) ,
f
(as in face)
Short
z ("
" zone)
s
(" " soup)
a ' (as
In tack or cat)
zh ("
" aziire)
sh
(" " shine)
e ("
" let or fed)
th • ( "
" then)
th
(" " thin)
i ("
*' lit or sin)
u' (•♦
" run or sun)
Transitional
u ("
" could or put)
y (as in you)
w ("
" will)
Diphthongs
h ("
" hot)
1 (as
in bite or lime)
hw ("
" what)
ow ("
" cow or ouch)
oi ("
" coil or boy)
Semi-vowel
ew ("
" pew or cute)
1 (as in loaf)
r ("
" root)
m ("
" men)
n ("
" never)
ng ("
" wrong)
-17-
Although the phonetic compoBltlon of the lists has "been
made relatively homogeneous "by the foregoing procedure, there
still remains the possibility that each list may not be as
difficult as each other list. For one thing, about 28 percent
of the 672 syllables form common words. Many of the other syl¬
lables, such as and ^ may readily arouse meaningful asso¬
ciations. In fact, very few of the syllables are devoid of
suggestive meajilngs. The various degrees of meaning which
these syllables may have make It necessary to determine by
actual articulation tests whether or not the difficulty of
the lists Is the same from list to list.
In an experiment designed for this purpose, each of four
announcers read each syllable list one or more times over a
high-fldellty system to a crew of practiced listeners. In most
cases each announcer read each list at least twice. Noise was
mixed electrically with the speech. Ihble 2 shows the results.
Thble 2
AETICUIATION SCOEES OBTAINED WITH SYLLABLE LISTS
List
Annoiuicer
Grand
Average
M
E
w
T
A
7Í+.5
73.5
69.2
77.8
73.8
B
72.9
71.^
67.5
76.4
72.1
C
68.6
64.1
66.6
75.2
68.6
D
73.5
74.6
61.1
74.4
70.9
E
72.h
65.8
68.9
71.2
69.6
F
72.6
66.6
63.7
72.9
69.0
G
70.2
63.3
66.8
73.6
68.5
H
75.0
.
68.3
66.1
78.5
72.0
The difference between the highest and lewest grand average Is
only five score units. Bad more tests been conducted, this
range would probably have been less. A statistical amlyals
of the data presented In Table 2 shows that the differences
among the average scores could reasonably be attributed to ex¬
perimental errors. Of course, such an experiment does not
prove that these syllable lists would be equal In difficulty
"•X8-
when tested by different announcers using different equipment.
However, extensive use of the syllable lists has shown that
under a wide variety of experimental conditions each list may
be considered as difficult as each other list.
The eight syllable lists are Included In Appendix I.
When they are used, the order of the syllables In each list
should be changed from reading to reading.
II. MONOSYLLABIC WORD LISTS
In the course of Its articulation testing program, this
laboratory has constructed several sets of word lists. One of
these sets, consisting of monosyllabic and dissyllabic words,
was published In an earlier report.* Although these earlier
lists have had wide utility In articulation testing, their
chief limitation has been that they contain too many easy words.
Even In the presence of a loud noise, these word lists give
scores of about 90 percent when superior communication devices
are tested. Such scores are too high to reveal readily the
difference between the articulation efficiency of two good In¬
terphone systems, because for maximal sensitivity the test
scores should fall near 50 percent.
Consequently, new sets of word lists containing only nono-
syllablc words have been constructed. Since the editions and
revisions through which these lists have gone are too numerous
to describe here, we shall content ourselves with an account
of the set of lists (EM) most extensively used over the past
year, and a description of the lists (PB) which represent the
most recent -- and, we hope, final -- revision.
)' From a large vocabulary 1200 monosyllabic words were cho¬
sen to make up the EM lists. The exceptionally easy words were
discarded from the original vocabulary and only words In common
use were selected. From this sample of 1200 words 2k lists of
50 words each were constructed. The words were assigned to each
list on the basis of the phonetic conçosltlon of the first part
of the word. No attençjt was made to equate these lists with
respect to the phonetic structure of the final consonant or con¬
sonant coDqsound.
* See NEC Committee on Sound Control report, "Articulation
Testing Methods," 1 February 19^2, OSED No, 383.
-19-
Extenslve use of these PM lists has shovn that under a
wide variety of conditions each list is nearly as difficult
as each other list. In an experiment designed specifically
to test these lists with respect to their relative difficulty,
each list was read seven times to an experienced crew of lis¬
teners. Since the lists of monosyllabic words were constructed
to measure the efficiency of communication in noisy surround¬
ings, these 8.rticu3.ation tests were conducted under severe
acoustic stress. In 'labia 3 is shown the mean score of these
seven tests for each of the 2k lists.
Table 3
AETICULATION SCOPES OBTAINED WITH
MONOSYLLABIC WOED LISTS (EI4)
List
Grand
List
Grand
No.
Average
No.
Average
1
56.0
13
55.9
2
53.9
Ik
3k.6
3
57.2
15
52.2
k
55.8
16
55.9
5
56.0
17
56.8
6
53.6
18
52.1
7-
52.6
19
55.2
8
57.7
20
56.0
9
56.9
21
57.1^
10
56.1
22
55.5
11
57.3
23
53.0
12
55.5
2k
52.5
9
The difference between the highest and lowest average is
less than six score mits. Again it can be said that if more
tests had been conducted the effect of experimental errors on
the avemges would probably have been reduced. A statistical
analysis of the data obtained in this experiment shows that
the differences which were obtained could reasonably be attrib¬
uted to chance fluctuations.
-20-
Although the EM lists were generally satisfactory, there
persisted in the laboratory an undertone of dissatisfaction
with the lack of phonetic balance among the final parts of the
words. Hence the construction of the PB (phonetically baj.anced)
lists. These lists probably satisfy the following criteria
about as well as the English language can be mde to:
1) Monosyllabic structure,
2) Equal average difficulty,
3) Equal range of difficulty,
k) Equal phonetic conçosition,
5) A composition representative of English speech,
6) Words in common usage.
The PB series consists of twenty lists, each containing 50
monosyllables. They cover a wide enough range of difficulty
to make them adequate for most typos of articulation compari¬
sons. The spread of difficulty is approximately the same in
each list and each list has nearly the same average difficulty.
Furthermore, the lists have very nearly the same phonetic
composition, a phonetic composition quite similar to that of
the English language. Eare and unfamiliar words have been
avoided as much as possible, and very few of the words are ex¬
tremely easy or extremely difficult under conditions in which
the whole group of lists would obtain an average articulation
score of about 50 percent. Incidentally, it should be pointed
out that each list contains 50 words because an attempt to"
satisfy the requirements listed above with lists of only 25
words met with failure.
The phonetic composition of the lists is based on Dewey's
frequency count* of the soimds in a sample of 100,000 words.
It was found impossible to adhere strictly to the values given
by Dewey, however, because of the limitations already tnçosed --
namely, that the lists be all different and con.structed of
* Godfrey Dewey, Eelative Frequency of, English Speech Sounds,
Harvard University Press, 1923.
-21-
common monosyllatles. According to Dewey, foi* instance, nearly
25 percent of the words in his sample of 100,000 "began with
short vowels. (This is due to the high frequency of suöh com¬
mon words as and, is, am») There are, however, only some kO dif¬
ferent common English monosyllables beginning with a short vowel.
It was necessary, therefore, to limit this class of words to two
per 50-word list. As a result of this and similar conçromises,
the rosemblance of the phonetic structure of the lists to that
of the language is not entirely exact.
The representation of sounds in each of the PB lists was
made to follow as nearly as possible the distribution of sounds
shown in Table 4. This table was derived from Dewey's coiuits
(in conçromise with the practical restrictions discussed above),
and it shows the frequency of each phonetic class in each posi¬
tion in the monosyllables, initial, medial, and final.
Table k
FREQUENCIES IN TBE PHONETIC CLASSES IN THE PB LISTS
Phonetic Class
Position of sound in word
Initial Medial Final
VOWELS
Long
Short
Diphthong
2 16 2
2 20 2
6
CONSONANTS
Transitional
Somi-vowel
Fricatives
Voiced stop
Unvoiced stop
5 5
9 9
12- 12
9 9
11 11
In each list there are about 10 conçiound consonants in the
initial and about 10 in the final position. In Table h these
compounds are classified in terms of the first consonant of the
compound.
-22-
In the construction of the lists, adherence to the fre¬
quencies of Tahle ij- was found to he most difficult for the
final sovinds, usually because new words were not available
which satisfied all of the requirements. None of the lists,
however, deviates for more than a few sounds from the speci¬
fied pattern.
Earlier experiments had shown that similarity of phonetic
structure is no guarantee of uniform difficulty. Lists of simi¬
lar phonetic'structure can be made up of easy words or of hard
words. As a test of this matter, the words of the preliminary
PB vocabulary were read eleven times vinder several conditions to
a crew of eleven listeners, so that it was possible to obtain
approximate measures of the relative difficulty of each of the
words. A few words almost always missed by all the listeners,
along with several almost never missed by any of the listeners,
were discarded as nondlscrimins.ting. Then, with the words re¬
maining, every effort was made to equate the lists both for mean
difficulty and for spread of difficulty.
Among the original words tested were a number fitting the
phonetic specification but very infrequent in spoken English
(thong, fop, ilk, crass). Even if it comes clearly through an
interphone, an unfamiliar word is likely to be missed. Conse¬
quently, the preliminary vocabulary was read in the quiet to
23 listeners who were instructed to rate each word as 1 (familiar),
2 (somewhat unfamiliar), or 3 (quite unfamiliar). The ratings
for each word were added and all words with a total rating of
35 or over were discarded as too unfamiliar to be used. This
seemed the most reasonable method available for eliminating un¬
familiar words.
Nevertheless, in order to avoid any difficulty with those
words which will inevitably prove unfamiliar to some listeners,
it is recommended that, whenever possible, the listeners be made
familiar with the moaning and spelling of all the words in the
lists before testing is begun.
Finally, it is only fair to point out that although the
Psycho-Acoustic Laboratory places great confidence in the ade¬
quacy of the PB lists, there still remains the possibility that
better lists can be constinxcted. The PB lists are too new to
have had the benefit of months of constant use, as previous
lists have had, and it is certainly not impossible that futvire
testing will bring to light slight differences in average dif¬
ficulty among the lists.
-23-
III. SPOKDAIC WORD LISTS
There are purpoaoa for which It la desirahle to uae liata
of worda of homogeneous audibility, i.e., liste in which each
individual word is as difficult as each other word. In order
to assemble such lists, experiments were conducted in which vari¬
ous types of words were presented to trained listeners under
carefully controlled conditions. It was discovered that the
class of words having the highest homogeneity contained those dis¬
syllables spoken with equal stress on both syllables. These words
are called spondees. Exanples: railroad, iceberg, horseshoe, etc.
The words listed in Appendix III have proved particularly
useful in tests whose purpose is to establish accurately the am¬
plification or power level at which speech can Just be heard.*
These homogeneous words reach the threshold of hearing, all within
a narro\r range of intensity, and thereby serve to determine the
threshold of hearing with precision.
In particular, the lists of spondees have been recorded phono-
graphically for use as an audiometric test in the measurement of
deafness.* The procedure for determining the threshold of a deaf¬
ened oar has been facilitated by recording the spondaic lists at
decreasing levels of intensity on each record. The subject writes
down the worda he hears until the point is reached at which he can
hear the words no longer. The level at which he hears half the
words is usually considered the threshold of hearing for speech.
Similarly, these lists can be used to measure such thijogs as
the relative efficiency of different earphones, or the effective¬
ness of different earphone cushions. The electrical power input
to the earphones required to reach the threshold of hearing of a
given ear is indicative of the sensitivity of the earphone as a
transducer.
Caution: Since the spondaic words are much easier for a lis¬
tener to identify than are the monosyllables of Appendix II, they^,
may sometimes be heard over a "peaked"^transducep,at a lower level
than would be required for words chosen at random. Hence, the
spondees are not recoimended for use in the comparison of two sys¬
tems which differ markedly in frequency response.
* See NDEC Research on Sound Control informal communication,
"Manual of Instructions for Auditory Test No. 9^ Threshold of
Hearing for Words," 20 May 19^^, IC-73.
IV. SENTENCE LISTS
In testing ccrano^icatlon equlpmont. sentence articulation
has ^ limited use. As pointed out ahove, the intelligibility
of sentences is favored to a considerable degree by the psycho¬
logical factors of Tnñanina;, -r.ontfiTrt,, rhythm, etc. As a conse¬
quence, under most test conditions ^ticulation scores obtained
with lists of sentences are so high that communication systems
must differ considerably for a substantial difference in-the
scores to be obtained. The influence of these psyçhslûglcal
factors on the scnrftR TnakfiR tbñ j'ftFnnta difficultjto analyze
ancfto"interpret. Furthermore7~Bince~the listeners easily
remember the sentences, a very large number of sentences is
required in an extensive testing program.
There are, nevertheless, special circumstances in which
sentence lists are of value. They are useful, for example, in
testing the intelligibility of telephone talkers, where the
utterance of a sentence provides a sample of more complex types
of behavior than does the speaking of single words. Eate, in¬
tonation, stress pattern, and maintenance of loudness level- can
be tested adequately only with sentence material. Moreover, it
has been observed in the course of experiments that talkers who
are careless in their manner of uttering sentences may be more
precise when speaking isolated words.*
There are several forms of sentence tests. In one form of
test the listener is required to respond by an appropriate word
or phrase to questions or commands. Each sentence is then scored
as either rigjit or wrong, depending on whether, or not it appears
that the listener has understood its meaning. In another type
of test the listener is required to record the sentence as read
to him. The articulation score is then based upon the number of
key words correctly recorded by the listener. This latter method
provides a more objective measure of the words actually heard.
In order to measure the intelligibility of talkers in noise,
this laboratory constructed 68 lists of 20 sentences each. Each
sentence consists of five key words plus a variable number of con¬
nective words. Four of the key words are monosyllables and one
is a dissyllable. An effort was made to avoid cliches, proverbs,
and other stereotyped constructions as well as the too frequent
use of any one word in the lists of sentences. Since only the
five key words in each sentence are scored, each list of 20 sen-
tences contains 100 scorable items.
* For a more detailed discussion of the use of sentences, see
NDEC Eesearch on Sound Control Informal jjommunication, "The Use
of Words and Sentences in Testing the Intelligibility of Tele¬
phone Talkers," IT April 19^^^, IC-69.
-25-
It íb difficult to determine whether or not these sentence
lists are equal with rospect to relative difficulty. However,
there is seme e3q)erim©ntal evidence which shows that the lists
do not V6u?y greatly in difficulty. In one experiment, each of
2k announcers read two lists of 20 sentences. The correlation
between the individual scores on these two readings was +0,82.
In view of the relatively narrow range of scores obtained in this
experiment, and the known high variability of a single test, a
correlation of this size may be considered good evidence for the
approximate equivalence of the sentence lists.
The 68 lists of sentences are included in Appendix IV. In
using these lists it is well to remember that changing the serial
order of the sentences in a given list will only partially pre¬
vent the listeners from recalling the sentences when they are
read a second time. For most purposes, they should not be read
repeatedly to the same crew of listeners.
Attention should also be called, at this point, to another
set of sentence lists.* These are sentences whose conçrehensicn
can be indicated by the writing of a single word. They, like the
spondaic lists, were especially devised to measure the degree to
which an individual shows a hearing loss for speech. For this
purpose, it is desirable to make each sentence in the list as dif-
,ficult as each other sentence. Ordinarily sentences differ con¬
siderably with respeçt to the intensity level required to make
them audible, and, since it was not feasible to achieve compara¬
bility by a process of selection alone, the sentences^Were made
homogeneous by adjusting the intensity level at which a given
sentence was recorded and reproduced. These adjustments were
made experimentally with the aid of a crew of listeners. Then,
for the final phonographic recording, oach list was divided into
small groups of sentences, and the intensity level of each group
was successively decreased. The result of this procedure is a
test which measures the "threshold of hearing" of a listener in
terms of his ability to hear connected discourse.
V. OTHER TEST MATERIALS
Articulation tests.have been developed and used by several
laboratories and research groups. In particular, no review of
the subject would be complete without casing attention to the
classic work of the Bell Telephone Laboratories. It was the
* KDBC Research on Sound Control informal communication, "Manual
of Instruction for Auditory Test No. 12, Threshold of Hearing for
Connected Speech," 20 December 19^+^, IC-96.
-26-
ploneering studies of this group* whioh estahlished articulation
testing procedures as a "basic method in the science of communica¬
tion.
The group at the Bell Telephone Laboratories placed most
en5)hasis on the development of procedures for testing syllable
articulation, and in the reference cited above there is contained
an excellent scheme for compoimding sounds into syllables of the
consonant-vowel-consonant form. Although syllables of the form
CVC have many merits for articulation testing, syllables of the
foims CY, VC, and V (see Appendix I) have been preferred for the
particuleo* type of testing being carried out in the Psycho-Acous¬
tic Laboratory. This preference was based principally on the
fact that the latter form leads to a much smaller number of dif¬
ferent syllables. It therefore becomes feasible to subject each
device t»sted to the conrolete list of syllables. Only the order
of presentation need be changed from test to test.
Word tests were used in some of the BTL tests, but extensive
balanced lists of words, such as those in Appendix II, were appa¬
rently not developed. Instead, words were merely picked at ran¬
dom. As Fletcher and Steinberg point out, there are difficulties
with this procedure, but it is felt that these disadvantages are
overcome in the word lists presented in this reoort (see Appen¬
dix II).
Another type of word list developed at BTL has considerable
merit for certain purposes. These arc called vowel and consonant
lists. They are useful for determining the relative articula--
tion of separate sounds without having^resort to the use of non¬
sense syllables. Since these lists are composed of English words,
they can be used without subjecting the testing crews to so long
a period of training as is required when nonsense syllables are
used.
The extensive set of sentence lists compiled at BTL has been
widely used by other laboratories. These sentence lists have their
merits as well as their defects. Unfortunately, they vary rather
considerably in difficulty, they are tainted with local color pecu¬
liar to New York City, and they are too much a test of knowledge
imd intelligence for use with some grades of listeners. Neverthe¬
less, for quick appraisals of relatively poor communication sys¬
tems, these lists of sentences provide a useful test material.
* H. Fletcher and J.C. Steinberg, Articulation testing methods.
Bell Syst. Tech. J., VIII, 1929, 806-85^1.
-27-
IV. METHODS OF CONDUCTING AHTICULATION TESTS
The values of articulation scores are dependent upon a
large number of factors which enter as parameters in every
test. For example, the announcer introduces such variables as
vocal quality, regional pronunciation, steadiness or variability
of speech power, proper or inçroper use of microphones, and
other individual characteristics. The interphone equipment in¬
cludes microphone, amplifier, and earphones, and each of these
components has its effect upon the intelligibility of the trans¬
mitted speech. Interfering noise -- its presence or absence,
its type, and whether it is introduced acoustically or electric¬
ally — makes for additional complications. And finally, the
individuals comprising the crew of listeners differ from each
other in their abilities to comprehend words under difficult con¬
ditions, and crews themselves may differ from group to group.
In this array of factors are hundreds of small events. — inter¬
actions between the equipment, the sounds of speech, and the
testing personnel — and each event contributes its small share
to the overall result.
It becomes obvio\xs, therefore, that the results of measvire-
ments of articulation cannot be interpreted in absolute terms.
All articulation scores are relative scores, contingent upon the
use of specific announcers, microphones, amplifiers, earphones,
noises, listeners, and test lists. Little trust can be placed
in absolute statements about articulation, especially if such
statements are based on tests with different voices and differ¬
ent listeners- In general, the only trustworthy statements re¬
garding the effectiveness of communication systems are relative
statements.
All conçarisons of communication devices should be made
with conditions kept as uniform as possible. A comparison of
two microphones, for example, should be made only with regard
to the specific conditions governing the particular compari¬
son. If both microphones are retested with a different ampli¬
fier, or with different earphones, or in a different noise,
they may well give different results. Similarly, in comparing
listeners it is important to make Judgments about their rela¬
tive abilities to hear words or commands only' on the basis of
homogeneous data, i.e., data obtained under comparable conditions.
This great complexity which attaches to articulation testing
makes advisable a discussion of the specific factors contributing
,to the articulation scores and of the techniques used to control them.
-28-
I, SELECTION OF TESTING EEESONNEL
The intelligibility of speech heard over a comniunication
system depends on the vocal quality of the announcer as well as
on the quality of the physical instruments used. The distinct¬
ness of his speech, the einphasis he puts on the separate sounds,
the "resonance" of hie voico, and other factors share in deter¬
mining how well the announcer will be heard. When tests are,
made under very easy conditions, even a poor announcer supplies
enough cues to the listeners so that the difference in quality
between his voice and others usually has very little effect on
the articulation scores. As the listening conditions become
increasingly difficult, however, the differences in vocal quality
become more significant and the differences in articulation scores
become greater. Figure illustrates the differences which may
be obtained among announcers.
A similar situation has been found at the receiving end of
an interphone system. As is shown in Fig. 5^ the members of a
listening crew are not all equally able to hear correctly the
words that are spoken.
In view of these individual differences, a number of factors
should be considered in selecting the testing personnel to be
used in testing communication devices. Announcers should be se¬
lected who are able to enunciate the fundamental speech sounds
in a "normal" manner. Eegional differences in pronunciation are
so numerous and so great that no simple rule, other than common
BenB©i used for selecting "normal" announcers. Since
there are inherent differences in male and female voices, the
sex of the announcer must also be considered, and communication
devices typically used by men should probably be tested with
male announcers.
Listeners having the equivalent of a high school education
and whose eiges range from 17 to 30 years are usually suitable
for this wprk. Sex differences appear to be unimportant insofar
as listening ability is concerned. For most purposes, the lis¬
teners should have reasonably normal hearing. For instance, in¬
dividuals having a marked high-tone deafness would be unable to
detect differences"in the intelligibility provided by an ear¬
phone which doea not transduce frequencies above 3000 cps and an
earphone which transduces all the inçortant frequencies of speech.
A common procedure for ensuring that a listener has normal hear¬
ing is to test his ability to hear pure tones. A practical meas¬
ure of his ability to recognize words vinder noisy conditions of
listening can be obtained by co335)aring his score on a preliminary
iOO
90
80
k
<
cc
<
k
i 60
Ui
o
oc
wso
40
I QUIET
JM
JE TW Di
ANNOUNCERS
F I G . 4
the «hite bars (average scores in ^uiet) sho«
small differences among announcers; the suck
bars (average scores in NOISe) sho« large dif¬
ferences among announcers.
MHR-TA
P-«ä3
AK GS NT PO
LISTENERS
F I G . 5
the white sars (average scores in qulET) show
small differences among listeners; the black
bars (average scores' in noise) show large dif¬
ferences among listeners.
-30-
articulatlon test with estahliahed norms, or by subjecting him
to one or more of the recorded tests designed to measure the
ability to listen in noise.*
Despite differences in abilities among announcers and lis¬
teners, the relative merits of various connaunication systems can
usually be correctly assessed. Figure ,6 shows the articulation
results obtained with two different crews of listeners. Three
different interphone systems were tested with each of the two
crews. One crew consisted of eight Army and Navy servicemen who
had been given very little training before this experiment was
conducted. The other crew consisted of e'xperienced listeners
trained in the laboratory. Although one crew of listeners ob¬
tained higher scores than the other crew, the three systems tested
were ranked the same by both crews.
It is not possible to specify exactly the optimum, number of
listeners, talkers, tests, or test words to be used in articula¬
tion tests in general, since the optimum number of each will vary
according to the purpose of the test and the conditions under
which It is to be run. Rough estimates of large differences in
efficiency may be obtained using very few listeners, talkers,
tests, and words. Precise estimates of small differences in ef¬
ficiency may require a great many listeners, talicers, tests, and
test words. The type of equipment being tested as well as the
typi> and amount of stress used in the tests will affect the num¬
bers required.
Increasing the nmber of talkers and listeners has two ef¬
fects: it Improves the reliability of the test, and it makes
the results of the test more representative of the results that
would be obtained from an unlimited population. Against these
advantages must be weighed the cost in time and effort. The
point of diminishing returns is reached and passed if the nimi-
bere'are compounded too rapidly.
From the thousands of articulation tests that have been
conducted in this laboratory with veu'ious numbers of talkers
and listening crews under a great variety of testing conditions,
it is possible to make an approximate estimate of the optimum
number of listeners, talkers, tests, and test words for the
type of test situation most frequently encountered. When the
time available for testing is limited, it is usually most effi¬
ciently used by tests of about 50 words, which, for purposes of
* NDRC Research on Sound Control Report, "Auditory Tests of the
Ability to Hear Speech in Noise," 1 September 19^U, OSED No. 3516.
100
COMPARISON OF THREE INTERPHONE SYSTEMS
BY TWO ALTERNATIVE CREWS
z
80
3
< 60
Û
oc
o
$
40
o
(r
20
crews: a 1
SYSTEM I
A 1
SYSTEM 2
A 1
SYSTEM 3
F I G . 6
crew 1 was a trained laboratory trsting crew. crew a con¬
sisted of eight army and navy servicemen. all'tests were
made with announcer and listeners in afilamsient noise (|i5
DB). the rank order of the three communication systems as
determined by the articulation scores is the same for both
crews.
T
L-EARNING CURVE FOR
LISTENERS (CREW 6)
t
t
"t
T
t
I 2 3 4 5 6 7 8
SUCCESSIVE PRACTICE DAYS
F I G . 7
each point on the curve represents the average score on
|2 tests for a crew of 10 listeners. the tests were
read over an interphone system by three well practiced
announcers. both the announcer and the listeners were
in an ambient airplane noise ( 1.1 5 db).
-32-
administration, may be divided into two liste of 25 words each.
Since this length has been found very convenient for both ad¬
ministration and scoring, the estimates of the optimum number
of listeners, talkers, and tests are made on the assumption of
50-word lists^
The variability of the individual listener is one factor
contributing to the instability of the final score. The score
of a single listener, due to his own variability, may vary 7 or
8 percent from one tost to another under the same conditions.
This variation can be effectively reduced by reading tests to a
group of listeners rather than to a single listener and using
the mean articulation score of the group. The mean articulation
score obtained from a well trained crew of six to eight listeners
will be sufficiently stable for most experiments. On the other
hand, if differences among talkers are being measured, and each
talker is given only one test, it is well to use as many listen¬
ers as la convenient, since a good deal of confidence must be
placed in the mean score of a single test. Multiplying the num¬
ber of listeners augments the task of scoring, but it does not
materially increase the time and labor required for the adminis¬
tration of the tests themselves.
The variability of the talker is a much more important
source of instability in the final score. The effect of this
variability upon the stability in the final score can be re¬
duced only by having each talker read a number of tests and
averaging the mean articulation scores for the several tests.
The average of the mean articulation scores of a set of six
50-word tests read by a single talker to a crew of eight ex¬
perienced listeners under conditions permitting about half the
words to be heard correctly will probably be as stable a value
as will be necessary for most purposes. As the number of tests
is increased beyond six, vinder these conditions, the stability
of the average increases only slightly with each test.
In a great many testing situations, the number of different
talkers used is not important, especially when it is desired
only to ascertain the relative merits of two pieces of equipment.
With some kinds of equipment, however, it is important to use a
wide variety of talkers in order to obtain representative re¬
sults. A man with a thick, fat neck may be a superior talker
with an ordinary hand-held microphone, but he may be almost un¬
intelligible over a throat microphone because of the padding on
his throat. A comparison between these two microphones obtained
on his voice alone will underestimate the efficiency of the
throat microphone. In an experiment performed to measvire the
-33-
effect upon articulation score of varying the distance between
the microphone and the lips it was found that the nose and chin
of one of the talkers prevented his getting the microphone closer
than one-quarter inch from his lips. As a result, his scores
for the two locations "touching the lips" and "one-quarter inch
from the lips" were the same since the two positions were the
same for his face. In general, it is with tests of microphones
that the problem of selecting representative talkers is most
pressing.
II. IMPOETANCE OF TRAIUIHG AKD FATIGUE
Articulation scores obtained with ine.xperienced listeners
show improvement with practice. This improvement may be con¬
siderable under listening conditions which involve a high degree
of stress. Therefore, in comparing communication devices by
means of articulation teats, it is important to train the test¬
ing personnel before actual experimenting begins. Typically,
the curves of irapi-ovement which show articulation scores as a
function of practice are negatively accelerated. After a fairly
stable level of performance is i-eachod, improvement is slow and
relatively "smooth," This form of learning curve is the one
most frequently reported in the literature for a wide variety
of learning tasks.
Figure 7 shows a typical learning curve obtained under
severe acoustic stress. After two to three hours of practice
each day for a period of five successive days, this crew had
reached a relatively stable level of performance under the par¬
ticular conditions chosen for the tests.
Although listeners may be well trained under one ^t of
listening conditions, they may show a noticeable amouii|~of learn¬
ing when different conditions are used. For example, after a
long series of tests using high-fidelity earphones, the first
set of articulation scores obtained with such resonant earphones
as E-14's or TH-37's may be abnormally low. Therefore, when
radical changes are made in the testing conditions, additional
training should be given until changes in articulation due to
learning become unimportant. Fortunately, relatively little
additional training is required for a well trained crew to ad-
Just to new conditions.
-31^-
Fatigue from long hours of listoning to word lists, par¬
ticularly in the presence of intense noise, is another factor
which might he expected to introduce systomatic orrors. How¬
ever, the available quantitative data demonstrate that, under
typical conditions of testing and in the presence of an intense
ambient noise, little or no change in the articulation scores
results from several hours of testing.
In on experiment designed especially to measure the effects
of fatigue on articulation scores, tests were conducted through¬
out two experimental sessions of three hours each, one in the
morning and the other in the afternoon of the same day. In these
tests, the listening crew was exposed to the ambient noise only
for the interval required for the reading of a test list, and
the listeners were given two fifteen-minute rest periods in each
three-hour testing session. This entire procedure was repeated
three days later, and the final results are based on both days.
In all, 48 test lists of 100 words each were road to the lis¬
teners under two e3q)erimental conditions which were alternated
from test to test throughout each day. The crew of seven listen¬
ers wore high-fidelity earphones mounted in rubber cushions MC-
162, which provided relatively poor protection against ambient
noise. For one-half the tests, a hand-held T-I7 microphone was,
used, and both the announcer and the listeners were in an intense
ambient noise (II5 db); for the other half, a noise having a uni¬
form spectrum was introduced electricálly into the listeners'
earphones, and the annoimcer used a crystal microphone, equal¬
ized to give a uniform response. By combining the data collected
on the two days of testing, it is possible to compare in Table 5
the articulation scores obtained during each of the six succes¬
sive hours of testing. Each score in the following table is
based upon 400 words — 100 read by each of four announcers.
Table 5
AVERAGE ARTICULATION SCORES,
HOUR BY HOUR, FOR TWO DAYS OF TESTING
T-17
(ambient noise)
BR-2S
(noise in earphones)
A.M
1st hour
2nd hour
3rd hoxir
81
82
82
7T
76
76
LUNCH
P.M.
4th hour
5th hour
6th hour
81
78
80
72
78
75
-35-
The low variability among these scores is evidence that articula¬
tion is reasonably stable throughout the day. Fatigue seems to
have little effect when reasonable provision is made for rest be¬
tween tests.
It should be pointed out that the results of this experiment
were obtained under conditions typical of articulation testing in
this laboratory. The listeners were provided with short rest
periods as veil as a lunch hour. Consequently^ the data should
not be interpreted to mean that under field conditions the effi¬
ciency of communication is not impaired after six hours of con¬
tinuous exposure to trying conditions. That the results of arti-
cvilation tests obtained dviring the early part of a testing ses¬
sion can be compared with those obtained during the latter part
is clearly indicated by these data.
Audiograms taken before and after a typical three-hour ex¬
perimental session show that, when tests are conducted in 120 db
of airplane noise, and when the signal level provided by the lis¬
teners' headsets is not painfully loud, the resulting hearing
loss as measured by pure tones is "negligible." However, under
unusual conditions, as, for example, when the listeners' ears
are directly exposed to an intense ambient noise, a considerable
but temporary hearing loss my result. The extent of this loss,
and therefore its effect on the articulation scores, depends not
only upon the intensity and d^^ration of the exposure, but also
'upon the particular listener. A hearing loss will have its great¬
est effect upon an articulation score when speech is at a weak
signal level. For example, if articulation tests using weak sig¬
nal levels are conducted immediately after tests employing pain¬
fully loud signal levels, the articulation scores obtained with
the weak levels of received speech might be spuriously low.*
On the other hand, a temporary hearing loss acquired by lis¬
tening to lou^ speech in the presence of noise seems to have no
effect on later tests using the same loud speech. In other words,
deafness does not affect articulation scores when the speech is
mde loud enough to exceed the listener's threshold of hearing.
This appears to be true regardless of the nature of the deafness.
As a corollary to this fact, there is apparently no correlation"
between the listener's audiogram (except for extreme degrees of
deafness) and his ability to hear loud speech in the presence of
intense noise.
* See OSKD Eeport under Contract OEMcmr-19^, "Teii^porary Deaf-
ness Following Exposure to Loud Tones and Noise," 30 September
19^3.
-36-
III. SELECTION OF COMPLEMENTARY EQUIPMEJIT
In comparing two devices with respect to the intelligihility
of speech transmitted over them, it is important to consider the
other parts of the communication system used for the tests. One
microphone might "be superior to another microphone when they are
tested with a given type of headset. When they are retested with
a different headset there might he little or no difference between
them. For example, suppose one microphone does not transduce
speech frequencies above 25OO cps and the other microphone trans¬
duces all the important frequencies of speech. If earphones which
do not transduce speech frequencies above 25OO cps are used for
these tests, little or no difference would be found in the articu¬
lation scores provided by the two types of microphones. If, how¬
ever, earphones having a wide frequency response were selected
for this comparison, a higher articulation score would be obtained
with the microphone which transduces all the important frequencies
of speech. Thus it may be stated as a general rule^that the final
evaluation of an instrument cannot be made apart from a considera¬
tion of its associated equipment.
IV. SELECTION OF AMBIENT NOISE CONDITIONS
Everyday experience shows that communication is difficult
when the environment is noisy. A noise reaching the ear of a
listener interferes with the intelligibility of speech by rais¬
ing the threshold of hearing so that soxuids which could be heard
in quiet are no longer audible. This shift in the threshold of
hearing is called masking. 'A quantitative measure of this mask¬
ing effect is obtained by .determining the intensity to which a
tone SJiust be raised before it can just be heard in thepresence
of noise. The amount of masking produced by an interfering noise
is the difference between the normal tlureshold for pure tones and
the masked threshold. If the masking of pure-tones by a noise is
determined at a nvimber of different frequencies, a masked-thresh-
old cvirve may be constructed. Figure 8 shows several masking
curves obtained by using a "white" noise at various intensities.
This noise had a nearly uniform spectrum level for all frequen¬
cies within the speech range. The curves of Fig. 8 show how
effectively a masking noise can reduce the auditory area avail¬
able for the reception of speech.
140
250 500 1000 2000 3000 4000
FREQUENCY IN CPS
F I G . 8
these curves show how effectively a noise of uniform spec¬
trum level (250-5000 CPs) masks pure tones. the area be¬
tween the curves represent itio the yhreshold of hearing in
quiet and the threshold of feeling defines the auditory
area available for communication in quiet. as the inten¬
sity of noise reaching the ear of a listener progressive¬
ly increases, this area available for communication be¬
comes smaller and smaller.
-10 0 10 20 30 00
SPEECH-TO-NOISE RATIO IN DB
F I G . 9
syllable articulation vs. speech-to-noise ratio. the
speech level at the listeners'ears was held constant at
a level of about 103 db. a "whITE" noise (raNDOM spec¬
trum) was mixed electrically with the speech. measure-
were made with a vu meter.
-38-
These masked-threshold curves show why the Intelllgihllity
of speech depends upon the ratio of the level of the speech sig¬
nal and the level of the interfering noise, and Fig. 9 shows
quantitatively how syllable articulation varies as a function of
speech-to-noise ratio. The particular form of this relation de¬
pends upon many conditions. In general, however, as the speech-
to-noise ratio is decreased, speech becomes less intelligible
until, finally, the speech signal is no longer audible and arti¬
culation is zero.
In most communication systems an ambient noise which reaches
the ear enters the communication circuit in two ways:
1) It is picked up by the microphone. The kind and the
amount of ambient noise which enters the system by
this means depends upon throe laain factors: the in¬
tensity and spectrum of the ambient noise; the amount
of acoustic shielding which the construction and
mounting of the microphone provide; and the response
characteristics of the microphone itself.
2) Ambient noise may also enter a communication circuit
by way of acoustic leaks in the headphones or helmet
which the listener wears in a noisy environment. Die
acoustic seal provided at the ear by various headsets
and helmets varies over a wide range. Hence, the in¬
tensity and spectrvim of the Interfering noise leak-
iixg into the ear by this path may vary greatly.
It is clear, then, that the intelligibility of speech pro¬
vided by a commxinication system will depend, to a large extent,
upon the intensity and spectrum of the surrounding ambient noise.
By way of illustrating the inç)ortance of a particular noise. Fig.
10 shows the extent to which an intense ambient airplane noise*
(120 db) reduces the range of levels of received speech available
for intelligible communication.
Certain conclusions can be drawn from these facts. If a
ccmnnunication system is typically used in a noisy environment,
articulation tests designed to measure the intelligibility pro¬
vided by that system should be conducted with the announcer, or
listeners, or both in an ambient noise. Microphones and ear-
phones, as well as overall communication systems, differ markedly
* For a description of methods used to generate various types of
noise, see KHHC Eoaearch on Sovind Control report, "The Electronic
Generation of Airplane Noise for Use in Testing and Training,"
25 May 19^3, OSED No. ikk^.
SENSATION LEVEL OF RECEIVED SPEECH IN DB
FIG. 10
EFFECT OF LOUD AMBIENT NOISE ON THE RELATION BETWEEN
PERCENT WORD ARTICULATION AND RECEIVED SPEECH INTEN¬
SITY. WHEN THE ANWOUNCER AND THE LISTENERS ARE IN
AN INTENSE AMBIENT NOISE (|20 D8), THE RANGE OF LEVELS-
OF RECEIVED SPEECH AVAILABLE FOR INTELLIGIBLE COMMUNI¬
CATION HAS'BEEN REDUCED BY ABOUT 60 DB. (CF. H. FLET¬
CHER,"sPEECH AND hearing" van NOSTRAND, 1929. P. 272)
MHR-7e
P- 4 36
-1|0-
in their ahillty to vithstand acoustic stress. Some instruments
are excellent when tested in a quiet room, hut almost useless in
the presence of an intense sound field. In general, experience
in coHparing communication devices has shown that the louder the
interfering noise, the greater will he the differences in per¬
formance of those devices.
Similarly, as is shown hy the data presented in Figs, h and
5, differences among announcers and differences among listeners
are greater when communication is attempted in noisy svirroundings.
And here too it is important that articulation tests of speaking
and listening ability he conducted under the acoustic stress typ¬
ical of field conditions.
As regards the effects of noise, the most revealing condi¬
tion luider which interphone equipment may he tested places hoth
announcers and listeners in noise. There is then a svtmmation of
stresses due to the noise which is picked up hy the microphone
and the noise which gets directly to the ear hy leakage around
the earphone cushions. This is the condition which simulates
communication among military personnel in many field conditions.
For some purposes, however, it is revealing to isolate the
effects of noise introduced at the microphone. This is done hy
placing the announcer in noise and the listeners in quiet.
For comparing various headsets or helmets it is often desir¬
able to use the condition in which the announcer is in the quiet
and the listeners are in noise. The purpose may then he to dis¬
cover the degree to which the noise which leaks into the ear hy
way of acoustic paths in the headset or helmet interferes with
the comprehension of speech.
Thus, depending upon the purposes of the experiment, there
are four permutations of noise and quiet under which articulation
can profitably ho conducted:
1) Quiet-to-quiet : hoth announcer and listeners in quiet;
2) Quiet-to-noise: announcer in quiet, listeners in noise;
3) Ncise-to-quiet: announcer in noise, listeners in quiet;
Í-) Noise-to-noise: hoth announcer and listeners in noise.
Each of these arrangements has its analogue in one or another
mode of field operation.
-4l-
y. IMENSITY OF RECEIVED SFECCE
Signal level is one of the most iaçortant determinants of
the intelligibility of speech. For convenience, signal level
in interphone communication may be considered a function of five
principal variables:
1) The voice level of the announcer;
2) The efficiency with which the voice of the announcer
is coupled to the microphone;
3) The frequency response and sensitivity of the micro¬
phone;
The gain of the interphone ançlifier;
5) The frequency response and sensitivity of the ear¬
phone when it is coupled to the ear.
These five variables are independent. Therefore, even
though values are assigned arbitrarily to any four of them,
a given level of received speech can be obtained by adjust¬
ing the fifth. Of course, in most experiments, certain of these
variables, such as the sensitivity of the earphones or of the
microphones tested, vary over, only a relatively limited range.
However, other variables, such as the voice level of announcers
and the gain of arçlifiers, are frequently varied over a wide
range. Therefore, in experiments designed to study the rela¬
tion between articulation and the level of received speech it
is more practicable to vary voice level or the gain of the ampli¬
fier than any of the other three variables. As a matter of ac¬
tual practice, it is amplifier gain that is varied in most ex¬
periments .
Exan^les of the relation between articulation and voltage
gain of the aitplifier are shown for three different microphones
in Fig, 11, To obtain these functions the gain of the ampli¬
fier was varied over a wide range and the announcer maintained
a constant voice level for all gain settings.* It is seen that,
as the gain is increased from a very low value, the articulation
improves until it reaches a maximum level for each particular
microphone. As the gain is further increased, the ears of the
listeners are overloaded and articulation declines.
* For a detailed discussion of the data shown in Figs. 11 and'
12, see NDEC Research on Sound Control report, "The Articulation
Efficiency of Nine Carbon Microphones for Use at Low Altitudes,"
1 June 19^4, OSRD No. 3515.
100
90
ANB-M-CI in
D-17 Noitt Shield -—» .
-X
-24
■18 -12 -6 0 6 12 IB
VOLTAGE GAIN OF INTERPHONE IN OB
F I G . II
RELATION BETWEEN PERCENT WORD ARTICULATION AND VOLTAGE
GAIN OF INTERPHONE. THE SI»EECH INPUT TO THE MICROPHONE
WAS HELD AT A CONSTANT LEVEL THROUGHOUT ALL THE TESTS.
BOTH THE ANNOUNCER AND THE CREW OF LISTENERS WERE IN AN
AMBIENT NOISE ( SPECTRUM I, 120 DB ).
100
90 -
80 -
70
3
O 60
I—
o:
<
UJ
o
or
UJ
Û.
50 -
40 -
30 -
20 -
10
ANB-M-CI in
D-Í7 Noise Shield
Hond-held T-17
86
92 98 104 no
VOICE LEVEL IN DB
116
F I G . 12
RELATION BETWEEN PERCENT WORD ARTICULATION AND VOICE LEVEL.
THE VOLTAGE GAIN OF THE INTERPHONE WAS HELD CONSTANT. THE
VOICE LEVELS WERE MEASURED BY A GR SOUND LEVEL METER ( WITH
THE LIPS JUST TOUCHING THE WIRE MESH OF THE MICROPHONE AND
THE METF-R SET TO "SLOW" READING). BOTH THE ANNOUNCER AND
THE LISTENERS WERE IN AN AMBIENT NOISE (sPECTRUM I, 120 DB).
-1+3-
Two examples of the relation "between articulation and
voice level are shown in Fig, 12, For these experiments, the
voice level of the announcer was varied over a wide range,
while the gain of the amplifier was held constant. The form
of this type of relation is generally similar to the curves
in Fig, 11, but the actiial slopes of the functions tend to
differ. This difference is related in a complicated manner
to the relevant parameters of the experiment.
Each of these two methods of obtaining the relation be¬
tween articulation and signal level has advantages, and for a
conplete analysis of the performance of communication devices
both types of relations should probably be obtained.
Experience has shown that for economy of time and effort
certain procedures can be profitably employed in obtaining
functiops of these types. For example, in determining gain
functions, the announcer should maintain his speaking level
constant throughout any one experiment. Care must be taken
that the announcer does not adjust his voice level to compen¬
sate for the differing signal levels provided by the various
equipment under test. Both visual and auditory means of in¬
dicating voice level should ordinarily be provided. The out¬
put voltage of the microphone being used may be euployed to
monitor the voice when the noise level due to microphone pickup
is about 5 or more below the speech level. When the speech-
to-noise ratio measured in terms of the voltage developed by
the microphone is less than 5 this method is inadequate.
Furthermore, carbon microphones do not provide a cox*rect in¬
dication of changes in voice level under typical conditions
of testing. At high voice levels, these microphones distort
the speech signal and an effective compression occurs. Con¬
sequently, changes in the output voltage of some, if not all,
carbon microphones are not proportional to changes in voice
level. Figure 13 shows the relation between the output volt¬
age of several microphones and voice level at the input.*
In most experiments a monitoring system which is inde¬
pendent of the system under test is desirable. By using a
magnetic throat microphone with an amplifier and output meter,
a monitoring system is obtained which is satisfactory over a
wide range of experimental conditions. As is shown in Fig, 13^
* For a more detailed discussion see NDEC Eesearch on Sound
Control informal communication, "The Effect of "Voice Intensity
on the Output Voltage of "Various Microphones," 1 June 19^^^ IC-7^,
o 5 10 15 20 25 30 35 40
RELATIVE OUTPUT OF 633-A MICROPHONE IN DB
85 90 95 100 105 110 115 120 125
SOUND LEVEL OF SPEECH 3 INCHES FROM MOUTH
(Measured by ERPI sound level meter)
FIG. 13
SHOWING THE RELATION BETWEEN VOLTAGE OUTPUT OF VARIOUS MICRO¬
PHONES AS A FUNCTION OF SPEECH INPUT INTENSITY.
CURVES REPRESENT AVERAGE RESULTS FOR THREE SPEAKERS, EACH OF
WHOM READ 10 WORDS AT FIVE VOICE LEVELS WHEN USING THE FOLLOWING
TYPES OF MICROPHONES:
(1) TWO T-30-P AND TWO T-30-V CARBON THROAT
(2) THREE 30NOT0NE MAGNETIC THROAT
(3) TWO T-17 (ONE KELLOGG AND ONE UNIVERSAL) CARBON HAND-HELD
(A) ONE 633-A DYNAMIC
THE CURVES HAVE BEEN MADE TO COINCIDE AT THE LOWEST SPEECH
LEVEL TESTED.
INDICATED BY THE ARROW IS THE AVERAGE SPEECH LEVEL USED BY A
CREW OF ENLISTED MEN IN FLIGHT TESTS AT EGLIN FIELD, FLORIDA.
MHR-78
P - 438
ALL OUTPUT VOLTAGES WERE MEASURED WITH A VU METER
-45-
changes In voice level are quite accurately indicated ty the
voltage output of a magnetic throat microphone. These micro¬
phones also provide a high speech-to-noise ratio.
Further aid may he given to the announcer hy providing, in
the headset he wears, a "side tone" from this magnetic throat
microphone. The level of this side tone can he held constant
at a high level in the annoiincer's headset even when the listen¬
ers arc receiving a weak auditory signal. The chief deficiency
of this method is the difficulty in placing a throat microphone
in the samo position on the throat from test to test, hut this
objection can he overcome hy the exercise of sufficient care.
When it is desirable to determine the complete relation be¬
tween articulation and gain of the amplifier, perhaps the best
procedure is as follows. The speech signal is attenuated until
a listener is Just able to detect the presence of speech. At ,
this level of received speech articulation will he substantially
zero. The upper level for the received speech is then deter¬
mined hy increasing the speech signal until it is painfully loud.
The gain levels chosen for the tests may then he selected within
the range defined hy these two measurements.
When it is desired to determine the gain of the amplifier
which, for a given speech input tg the microphone, will provide
optimal intelligibility over a communication system, experience
has shown that the following procedure is usefulr The gain of
the aii5)lifier is turned up until the speech signal is painfully
loud. The gain is then decreased until the speech is Judged to
he at a comfortably loud level. It has been found empirically
that in most instances a Judgment of this type will determine
the optimal level of received speech.
When articulation tests are conducted at only one speech
level it is frequently useful to know the sensation level of
the received speech signal. The senoation level of a given
sound is the number of decibels that the sound is above its
normal threshold of audibility. The sensation level of speech
may be determined by attenuating the speech level until a lis¬
tener can detect the presence of about 50 percent of the speech
soiuids without identifying any of them. The amount of attenua¬
tion in decibels required to obtain this threshold is used to
determine the sensation level of a given sangjle of speech.
-46-
In actual practice the user of a comimmication system will
tend to adjust his voice level in accordance with the loudness
of the side tone he hears. When the side tone is at a weak level,
the talker will tend to raise his voice in order to provide a more
audible signal. When the signal is too loud, he will tend to
lower his voice. A talker can vary his voice level over a range
of about 35 cLb. This range extends from just above a whisper to
a loud shout. Since changes in voice level not,only change the
level of received speech but also the speech-to-noise ratio pro¬
vided by the microphone, the efficiency of communication will de¬
pend, to a marked degree, upon the particular voice level chosen
by the talker. Therefore, when an attençt is made to compare
communication devices under actual or simulated field conditions,
it may be desirable to permit the announcer to adjust his voice
level on the basis of the side tone he would hear under actual
field conditions. In such cases, an effoi-t should be made to
have each announcer use the communication system as it would be
used by experienced personnel.
VI. USE OF A CARRIER SENTENCE AM)
TIME INTERVAL BETWEEN TEST ITEMS
In an articulation test each test item (syllable or word)
is usually road as part of a sentence. For example, in the sen¬
tence, "You will write car, " the word car is the only word which
the listener is required to record. A carrier sentence is de¬
sirable for several reasons, (l) The listener is prepared for
the presentation of the tost item, and variability in the arti¬
culation scores due to inattention or distraction is reduced.
(2) If cajTbon-button microphones are used, the carrier phrase
preceding the tost item serves to agitate the particles of car¬
bon and reduce tho variability inherent in such microphones.
(3) The carrier sentence permits the annoimcer to modulate his
voice 80 as to keep the level of his voice even from word to
word.
The particular sentence chosen for use is somewhat arbi¬
trary, and, in some cases, several carrier sentences are used
in the reading of a single test list. Care should be exercised,
however, in selecting the word in the carrier sentence which pre¬
cedes the test item, This word should affect the pronunciation
of the tost item as little as possible. Examples of carrier sen¬
tences used in articulation testing are:
-2^7-
"You will write test item."
"You will say test item."
"Number 6 is test item."
"Write down tost item."
"Please write tost item."
In some experiments a carrier sentence is selected to ful¬
fill a more specific pxirpose. When articulation tests are con¬
ducted at high altitudes, it may be desirable to use a fairly
long sentence. In this way, the effect on articulation due to
the difficulty of breathing at high altitudes may be assessed.
One such carrier sentence that has been used successfully is
"One, two, three, four, five, tost item, seven." All the words
in this sentence, including the test item, should be spoken
evenly and with one breath. No attempt should be made to accent
the test item.
For most purposes, the^parrier sentence and not the test
item should be used to monitor voice level. No attempt should
be made to compensate for the typical differences in the speech
power used in pronouncing the different sounds in the test item.
When only the carrier sentence is monitored, the test item
should be spoken with the same general effort as the rest of
the carrier sentence.
The time interval between test items should be Just long
enough to enable the listeners to i-ecord the test item. With
a well trained crew the average rate at which words or syllables
may be read is one item every three to four seconds. With this
interval it is possible to conduct from 25 to 35 tests of 50
words each in a three-hour testing session. With a session this
long, two to four 15-minute rest periods should be given.
VII. INSTRUCTIONS TO TESTING PERSONNEL
The way in which the testing personnel use the microphones
and headsets under test are important conditions of an articula¬
tion test. In order to control the speech input to the micro¬
phone throughout any one test it is necessary to keep the face
of the microphone at a constant distance from the lips of the
announcer. A simple exp)eriment was performed to show the iinport-
ance of instructing the announcer regarding this factor. -A
.l^8-
magnetlc hand-held microphone mounted la a baffle of the same
dimensions as the case for the T-l? microphone vas used« A
narrow bar was attached to the top of the baffle to permit
throe spaclngs. Figure 1^+ chows the results. This experiment
amply domonstrates the importance of properly Instructing the
announcer in the use of a hand-held microphone.* The announcer
should also bo instructed to monitor his voice and pronounce
the test items in a manner and at a rate which fulfills the
purpose of the experiment.
The instructions given to the listeners regarding the use
of headsets also depend • upon the purpose of the experiment.
When the listeners take the tests in an intense ambient noise,
the acoustic seal afforded by the headset is an inçortant factor.
The listeners should be instructed to record a response for
every item called by the announcer. It has been shown in many
experiments that the listener records correctly some test items
even though he maintains that he is only guessing.
VIII. STATISTICAL METHODS
When an attonçt is made to repeat articulation tests under
identical conditions, the articulation scores obtained are usu-
•¿lly somewhat different. If a test is repeated a large number
of times, scores of various magnitude will be obtained. A fre¬
quency distribution of those scores typically shows that the
scores occur most frequently around some central value, and
scores which deviate markedly from this central value are rela¬
tively infrequent. Unless the mean score is very high or very
low the scores tend to be distributed according to the normal
law of error.
The dispersion of the distribution of average scores ob¬
tained under the "same" conditions will depend to some extent
upon how well the experimental conditions have been controlled.
This dispersion also depends upon the number of listeners and
the length of each test list. Under typical conditions of test¬
ing in the Psycho-Acoustic Laboratory, it has been found that
the standard deviation (root-mean-square) of a distribution of
articvilation scores obtained by reading different word lists
* For a more complete discussion of this factor, see NDEC Eesearch
on Sound Control informal communication, "The Effect of Microphone
Position on the Intelligibility of Speech in Noise," 6 December
I9U3, IC-54 (OSBD Eeport No. h023).
100
90
<
-J
O
80
(r
<
o
a:
o
»
70
HJ
o
cc
Ui 60
50
_L
±
0 1/4" 1/2"
DISTANCE FROM LIPS TO MICROPHONE
FIG. 14
SHOWING THE EFFECT ON COMMUNICATION OF VARYING THE DIST¬
ANCE OF THE MICROPHONE FROM THE LIPS. THE TESTS WERE CON¬
DUCTED WITH THE ANNOUNCER AND CREW OF LISTENERS IN AN AIR¬
PLANE NOISE (lis DS). EACH POINT IS BASED UPON THE AVER¬
AGE OF TESTS READ BY THREE ANNOUNCERS.
-20 -10 0 »10
LEVEL OF RECEIVED SPEECH IN DB
MHR-7e
P- 439
FIG. 15
SHOWING THE SPEECH-TO-NOISE RATIOS REQUIRED TO OBTAIN
THREE TYPES OF THRESHOLDS. WHITE NOISE iRANDOM SPECTRUM)
AT VARIOUS LEVELS WAS MIXED ELECTRICALLY WITH THE SPEECH.
THE ABSCISSA LEVEL OF 0 DB CORRESPONDS TO A SOUND PRESS¬
URE LEVEL OF ABOUT 108 DB AT THE LISTENERS' EARS.
-50-
undor the "same" conditions is about five score units. Each
score in such a distribution is based upon the mean scores of
all listoñers (6 to 10) for one test of from 50 to 100 words.
Thus, if the distribution is normal, about 3^ percent of the
scores obtained deviate from the mean by five score vinits or
more. Out of a thousand cases, only two should deviate from
the mean by 15 or more score units.
Because of this variability among individual tests, it
is highly desirable to base conclusions regeirding articulation
efficiency on the average of several repeated tests. In gen¬
eral, the mean of a group of tests is far more stable than is
the score for a single test. This stability increases as the
square root of the number of tests, so that by reading four
word lists, for example, the expected variability of the mean
score can be reduced by a factor of two.
The differences in the articulation scores obtained under
the "same" conditions arise from many sources. Some of the
differences may be due to experimental errors which are present
throughout the reading of a given test list. For example, if a
listener does not obtain a good acoustic seal by adjusting his
headset properly, his score may be spuriously low for that test.
Many of these systematic errors should be eliminated insofar as
possible by proper experimental procedures.
In addition to experimental errors, there are in the array
of basic factors determining the articulation scores many small
events — chance interactions among the equipment, the sounds
of speech, and the testing personnel — which contribute their
sme.ll share to the overall result. The particular combinations
of these chance events determine partially the extent to which
the articulation score deviates from a central value. These
errors arise from numerous sources many of which are not under
the control of the experimenter. For example, the attention of
a listener fluctuates from moment to moment, and he may as a
consequence occasionally miss an easy word. Since different
combinations of these snail events are sampled upon successive
tests, different articulation scores are obtained. Differences
arising from this source may be considered as errors due to ran¬
dom sampling.
-51-
Eecauae of these errors a difference in the articulation
scores provided "by two communication devices may occur even
though there is no difference "between the two devices tested.
Therefore, an experiment must be designed so that the differ¬
ences obtained between the articulation scores may properly be
analyzed by statistical procedures. The basic principle in ex¬
perimental design consists in controlling at known values as
many variables as possible, and in so arranging the remaining
faàtors that their effects are made as random as possible. By
this procedure it is possible to interpret the central or aver¬
age articulation scores in terms of the known conditions, and
to determine the limits of accuracy inçosed by the errors of
random sampling. The dispersion of the articulation scores
then becomes a measure of the precision of the experiment, and
an estimate of the error of measurement may be based upon this
dispersion. It Is then a straightforward matter to assess, by
conventional statistical procedures,* the reliability of the
differences obtained between the scores for different Instru¬
ments, personnel, testing conditions, etc.
In general, the most direct way to Increase the precision
of an experiment. In order that small differences In articulation
may be reliable. Is to Increase the number of Independent articu¬
lation tests. When the amount of time available for testing Is
limited, more reliable results will be obtained with short test
lists than with longer lists. Those errors which tend to raise or
lower the average articulation score throughout the reading of a
"given test list will have less Influence when a larger number of
short tests Is used. Orders and arrangements can then be counter¬
balanced, since with short tests It Is possible to sanple a larger
number of announcers In a given time eind thereby make the test more
representative of conditions that might be encountered In practice.
Finally, if a statistically significant difference is not ob¬
tained in a given experiment. It cannot be concluded that there:.Is
no difference among the devices tested. Furthermore, It cannot even
be concluded that the true difference Is less than the obtained dif¬
ference. An experiment In which more tests are conducted, or an ex¬
periment In which the conditions are more precisely controlled, may
show a reliable difference among the Communication devices compared.
IX. LIST OF FACTORS AFFECTING ARTICULATION SCORES
Seme of the factors which Influence articulation test résultai
are important only under special circumstances, while others shouLd
be considered in every articulation experiment. Only a few of the*
more Inçortant ones have been discussed above. By way pf summary,
then, let us attendit a list of the more obvious factors which may
affect articulation results.
* See any standard text. Particularly reccramended is C.C. Peters
and W.B. "Van "Foorhis, Statistical Procedvires and their Mathematlceil
Bases, McGraw-Hill, 19^01
-52-
Announcer 1.
2.
3.
Speech 1»
Quality and Intensity of voice.
Correctness of pronvinciation.
Manner of holding microphone, etc.
Phonetic conç)osltion and
item difficulty.
Microphone ,1, Frequency-response characteristics,
2. Non-lineox distortion.
3. Efficiency and impedance.
4. Behavior at different altitudes.
5. Directionality (shielding of
microphone from noise --
signal-to-noise ratio).
Amplifier 1. Frequency-response characteristics.
2. Non-lineeu" distortion.
3. Input and output impedances.
Gain.
5. Peak pover limitation.
6. Shielding (noise pickup
and feedback).
Eadio Link .....1. Overall fidelity (response
characteristic).
2. Signal-to-noise ratio.
3. Loudness of side tone chantel
heard hy announcer.
4. 0vermodulation.
5« • Presence or absence of limiters,
Earphones ... 1. Frequency-response characteristics.
2. Non-linear distortion.
3. Efficiency.
4. Behavior at different altitudes.
Earphone Mounting 1.
2.
3.
Design of holmet and headband.
Acoustic seal at the ear (insu¬
lation against noise).
Air volume under receiver.
Listener ...........1.
2.
3.
State of hearing (deafness).
Masking öf speech by noise
entering ear,
Basic ability to understand speech
when distorted and masked.
-53-
ThlB list of factors Is "by no means exhaustive. It is in¬
dicative, hovever, of the extensive matrix of parameters affect¬
ing any attempt to evaluate in a quantitative way the relative
merits of communication systems, the effectiveness of noises in
masking speech, the audibility of words and commands, or the
relative abilities of communications operators. Plainly, the
business of conducting articulation studies that aim at reason¬
able completeness can quickly become laborious to the extreme
if many values of the relevant parameters are sampled.
Although economy of sampling is called for, there is little
to be gained by obtaining only part of the story from too few
tests. As a concrete example of the kind of budgeting of test¬
ing time that is required for a study, let us assume the follow¬
ing experiment:
'Purpose; to compare the effectiveness of two microphones
(carbon and magnetic) for use in airplanes at
high altitude.
Pelevant parameters:
1. Oxygen masks. There are three different
masks whose effects on the microphones must
be studied.
2. Amplifier gain. It may be expected that each
combination of microphone and mask might re¬
quire a different gain for optimal efficiency.
At least gain settings must be sampled.
3. Altitude. The effect of altitude can probably
be Judged from tests at sea level and at 30^000 feet.
it. Headsets. Only one standard type need be used.
5. Announcers. At least three.
6. Listeners. Since the amount of testing is not
multiplied by using more listeners, as many
should be used as can be accommodated in the al¬
titude chambers. This nvimber cüould bo at least six.
If it is assumed that each announcer is to read a 100-vord
test under each combination of testing conditions we can find the
total number of tests required by multiplying the number of an¬
nouncers by the niuuber of masks, etc., etc., i.e., 3x2x3 3ci<-x2,
which equals lil-4. This number of tests would require about 30 hours
of actual testing time.
-5^-
V ABBREVIATED TESTING METHODS
The fermai articulation test "based on groups of announcers
and listeners may prove too cumbersome and inefficient when an
articulation study involves numerous permutations of experimental
conditions. It is then expeditious to devise short-cut methods.
Many schemes are, of course, possible and several variants have
been tried from time to time.
An interesting example of an efficient procedure is one
which was developed to study the effects of amplitude distor¬
tion on articulation.* This method involves a single indivi¬
dual who must be both.conscientious and well practiced. The
test words are recorded phonographically by means of high-fidel¬
ity equipment, and good results have been obtained with the use
of a single list used over and over.
The person conducting the test has before him a written
list of the words recorded on the record. These words he keeps
covered with a blank card. He listens to each word in turn,
and, after he has both heard the word and decided what he thinks
it is, he moves the blank card so as to uncover the correct word.
He then checks whether or not he has heard the word correctly.
This procedure obviously requires care and honest Judgment on
the part of the experimenter, but control checks applied to a
practiced observer have shown that the method can be made to
yield valid results. Since the experimenter, instead of writ¬
ing down the words he hears, merely checks his correct res¬
ponses, he is able to work at a fast pace and the speed-up in
the articulation testing is considerable.
Although a single list of 100 words, of which several re¬
cordings bad been made in altered sequences, has proved adequate
for special purposes, the generality of the method can be ex¬
tended by adding other recorded lists and by using a group of
different announcers to record the words.
The adequacy of a single list is enhanced if the words are
made to resemble one another as closely as possible and at the
same time to sample the various sounds of speech. In order to
satisfy these requirements lOO-vord lists have been compiled in
which 15 vowels and diphthongs are represented by six words each.
The different consonant sounds are distributed among these 90
words. The remaining 10 words of each hundred are used to sam¬
ple some of the compound consonants. Two such word lists are
presented in Appendix Y,
* NDEC Besearch on Sound Control report, "The Effects of Ampli-
tude Distortion upon the Intelligibility of Speech," 15 November
19l^h, OSRD No, 4217,
-55-
VI COMPARISONS BASED UPON SUBJECTIVE APERAISAL
In an articulation teat the evaluation of the intelligibility
of speech is haBocL upon the number of test items correctly recorded
by the listener. In this kind of test, the listener is not re¬
quired to appraiso the quality of the speech, but merely to record
the speech sounds that he hears. By contrast, methods of subject-]
ive appraisal require the listener to evaluate the quality of J
speech itsolf. One variation on the subjective procedure is the
method of rank order. This requires the listener to judge vhich
of two or more samples of speech is more intelligible. Another
variation is the rating-scale method, which requires him to des¬
cribe a sample of speech in terms of its quality. The listener
must make this latter judgment in teims of a standard sanç)le of
speech provided for him, or in teims of his previous experience
in evaluating speech.
If only an approximate asessment of the intelligibility of a
speech sample is desired, these subjective methods may be found
useful. They can be employed to especial advantage in the pre¬
liminary evaluation of a communication system. Also, the experi¬
mental evidence at hand indicates that talkers ms.y be reliably
rated by relatively untrained judges using subjective rating
scales. For detecting the obvious faults of talkers these sub¬
jective methods have important applications.* However, for the
accurate measvirement of small differences in speech quality arti¬
culation testing methods should be employed.
The following experimental evidence indicates that, using
the method of rank ordei", an observer is able to judge with some
accuracy the intelligibility of speech. In this experiment each
of seven observers was instructed to rank order five sarples of
speech obtained by recording the speech transmitted over five
different communication devices. The observer assigned the
numeral "one" to the sample of speech which, in his opinion,
was the most intelligible. The numeral "two" was assigned to
the next most intelligible, and so c«i. The individual results
of this experiment are shown in Table 6. Since it is desirable
to compute a single representative value for each sample, the
mean rank is shown for each device. The mean rank for a given
speech sample is the arithmetic mean of all the ranks which have
been assigned to that sample. In the present case, the mean rank
W8.S computed to the nearest 0.5. Articulation testing methods
were also used in evaluating the five devices, and the average
results of these tests are also shown in Table 6.
* For a discussion of subjective methods which may be used in
rating talkers see NDBC Eosearch on Sound Control informal com¬
munication, "Subjective Ratings of the Intelligibility of Talkers
in Noise," 10 April 19^i<-, IC-67.
-56-
Tatle 6
RANK ORDERS AND ARTICULATION SCORES
OBTAINED WITH FIVE COMMUNICATION DEVICES
Device
Observer
123^567
Mean
Rank
Average
Articulation
A
12 12 112
1.5
66^
B
2133231
2.0
60
C
3 2 1 3 2 3
2.5
56
D
3 14. 4 5 L If 4
4.0
45
E
5 5 5 5 5 5
5.0
44
The excellent agreement "between the mean ranks and the average
articulation scores indicates that methods of subjective appraisal
can be used to obtain an approximate measure of the intelligibility
of speech.
The validity of the rank-order method depends almost entirely^
upon how well the observer avoids basing his Judgment upon the ir-/
relevant aspects of the speech sample. Although different observ¬
ers may base their Judgments upon somewhat different aspects of
the speech samples, a trained listener usually directs his atten¬
tion to the intelligibility of certain speech sounds which he
knows are frequently missed. For this reason the observer should
be well trained and e3q)erienced in the making of such Judgments.
The extent to which different observers assign the same rank to a
given sample of speech may be taken as a measure of the reliabil¬
ity of the procedure.
The rating-s cale methods are frequently used in a simpl^
form to evaluate, in a preliminary way, the intelligibility o:^
speech. This method is essentially the same as the method of
rank order. However, in the method of rank order the intervals
between the various ranlcs have no particular meaning. For ex¬
ample, the difference between a rank of one and two doas not
necessarily have the same meaning as the difference between the
ranks two and three. In the rating-scale methods the observer
is usually instructed to rate the speech samples on a point scale.
He is not required to assign different raides to the speech sam¬
ples, nor is he required to utilize all of the points of the
scale. Typically, an observer is instructed to use a scale which
-57-
ranges from very poor to excellent, such as the following:
6 — Very Poor
5 — Poor
k — Fair
3 — Good
2 — Very Good
1 — Excellent
This method is an attempt to make the eOale of numbers assigned
to the samples of speech indicative' of the actual differences in
intelligibility between the speech samples. However, in order
to evaluate the size of these intervals along the point scale,
it is necessary to obtain an objective measure of the intelligi¬
bility of speech.
-58-
VII THKESHOLD METHODS FOR EVALUATING IBE INTELLIGIBILITY OF SPEECH
It is frequently desirablo to know the conditions under which
speech is Just detectahle or Just intelligible. Three of these
threshold methods will he doserihed:
1) The threshold of detectahility,
2) The threshold of perceptibility,
3) The threshold of intolligibility.*
To determine the threshold of dotectability (sometimos
called the threshold of audibility) the listener adjusts some
variable (usually the speech level or the lovol of a masking noise)
until he is Just able to detect the presence of speech sounds
about half the time. At this threshold level he will ordinarily
be unable to identify any of the sounds themselves. This threshA
old of detectability is particularly useful as the reference J
level for the specification of the sensation level of speech.
The threshojld of perceptibility can be used to determine that
condition under Which the sounds heard at the threshold of detect¬
ability begin to be perceived as words. As a me-chod it is more
direct and loss tedious than articulation testing methods, and the
results obtained by determining the threshold of perceptibility
may be used to supplement articulation test results. The deter¬
mination of the threshold of perceptibility is particularly useful
when the intelligibility of speech can be varied continuously from
a very low to a very high value. In determining this threshold
the listener adjusts some variable until, in his Judgment, he is
Just able to imdorstand with considerable effort the gist of the
connected discourse read to him. In actual practice, the listener
adjusts some variable until the speech is almost completely unin¬
telligible. He then adjusts this variable until he can under¬
stand the speech with little effort. By successive approximations,
the listener finally determines that condition which defines for
hiTn the threshold of perceptibility of speech. If the speech were
made more intelligible, he could, with little or no effort, under¬
stand it, and if the speech were made less intelligible, he could
not perceive a sufficient number of words to allow him to follow
the main ideas of the passage read to him.
* These designations are a revision of the terminology used in a
previous report/from this laboratory (OSRD Report No. 2038), in
which the three thresholds were called detectability, intelligi¬
bility, and immediate appreciation, respectively.
-59-
In an effort to determine the reliability of this procedure
each of three listeners made 30 determinations of the threshold
of perceptibility. A high-fidelity Interphone vas employed and
a random noise with a uniform spectrum was mixed electrically
with the speech. The speech-to-noise ratio was adjusted by the
listener until he obtained the threshold of perceptibility. The
standard deviations of these three distributions of 30 Judgments
each were; I.3 cLb, 1.6 db, and I.7 db. The ranges of these dis¬
tributions were only 6 db, 7 and 6 db, respectively. The
small variability obtained Indicates that the threshold of per¬
ceptibility can be reliably determined.
Furthermore, essentially the same threshold of percepti¬
bility was obtained by different listeners. Eight inexperienced
subjects were given only brief instructions before they made
their Judgments. Table 7 shows the average speech-to-noise ra¬
tios which these listeners required in order to obtain the
threshold of perceptibility. Each entry In the table is based
on the average of three Judgments.
Table 7
THRESHOLD OF PERCEPTIBILITY FOR
INEXPERIENCED LISTENERS
Listener
Speech-to-nolse
ratio In db
AM
-5 *db
SM
-6
PP
-7
RL
-7
MR
-7
IP
-8
RA
-8
JS
L
+8
With the exception of one Individual (JS), the listeners
required the same speech-to-noise ratio within +1.5
average of these eight values is -5 db. This value agrees with
the average threshold obtained with four well trained listeners
who required an average speech-to-noise ratio of -5.5 àb.
-60-
In determining the threshold of Intelligibility the lis¬
tener adjusts some variable until, In his Judgment, he Is Just
able to obtain without perceptible effort the meaning of almost
every sentence and phrase of the connected discourse read to him.
The use of all three of the threshold methods is Illus¬
trated In the following experiment* in which the speech-to-
nolse ratios corresponding to the three thresholds were deter¬
mined. The level of received speech was held constant and the
listener adjusted the level of a random Interfering noise until
he obtained each of the thresholds. This proceiui*e was repeated
using two other levels of received speech. Passages from Adam
Smith's "The Wealth of Nations" were read by ■'he a:.Tno'uncer. This
reading mshorlal Is particularly suitable for cetermining these
thresholds, since It Is uniform in difficulty and Interest. The
results, presented graphically In îlg. 15. slio-w that the speech*
to-nolse ratio req.ulred foi- the tlireshold of perceptibility ls<
about 7 db higher than that required for the threshold of detect-
ablllty. A further Inci-ease In the speech-to-nolse ratio of only
14- db Is required to reach the threshold of Intelligibility
the listener Is able to imderstand easily every sentence.
These three threshold methods are most applicable when the
Intelligibility of speech can be varied by changing the level of
received speech or the level of masking noise. These methods
measure the value of some physical variable, such as the level
of speech, at which detectablllty, perceptibility, or Intelli¬
gibility Is obtained. In this way the threshold methods may
provide useful supplementary Information to the results ob¬
tained by more refined procedures.
* For a more complete discussion of this experiment see NDPC
Eesearch on Sound Control report, "Studies on the Effect of
Noise on Speech Communication," 25 November 19^3^ OSRD No. 2038.
-62-
S Master
List A
1
du
22
fa
ye
64
ezh
2
da'
23
fî
44
ud
65
uth'
3
bo
24
45
owd
66
owth
bi
25
soi
46
ëb
67
u'f
5
gâ
26
sh'ü
47
ewb
68
Us
6
gol
27
she
48
ag
69
a's
7
Ju'
28
thu
49
Tg
70
ësh
8
pu
29
thow
50
71
ewsh
9
pow
30
lu
51
ej
72
oth
10
të
31
la '
52
u'p
73
ith
11
tew
32
ro
53
art
74
ul
12
ka
33
ri
54
oit
75
owl
13
kl
34
ma
55
ok
76
ër
14
cho
35
moi
56
ik
77
ewr
15
che
36
nu'
57
uch
78
am
16
•vü
37
hu
58
a 'ch
79
Tm
17
va'
38
how
59
av
80
ën
18
zë
39
hwë
60
TV
81
en
19
zew
40
wa
61
äz
82
Ting
20
th'o
41
wT
62
oiz
83
a'ng
21
th'i
42
yô
63
ozh
84
u'
S Master
List B
1
du
22
fu'
^3
yu'
64
oth'
2
dow
23
sü
44
ed
65
ith'
3
bïï
24
se
45
ewd
66
af
4
ba'
25
sha
46
ab
67
if
5
go
26
shoi
47
ib
68
es
6
gi
27
tho
48
^g
69
ews
7
28
thi
49
eg
70
ush
8
joi
29
lu
50
71
a'sh
9
pe
30
low
51
ap
72
uth
10
pew
31
ru
52
oip
73
owth
11
ta
32
ra'
53
ot
74
il
12
tT
33
mo
54
it
75
ewl
13
ko
34
mi
55
uk
76
ar
14
ke
35
na
56
a 'k
77
Ir
15
chu'
36
no i
57
uch
78
om
16
m
37
hê
58
owch
79
em
17
vl
38
hew
59
u'v
80
u'n
18
zu
39
hwa
60
ÖZ
81
äng
19
za '
40
hwr
61
ez
82
oing
20
th'ë
41
WÖ
62
azh
83
u
21
th'ew
42
we
63
oizh
84
GW
-63-
S Master List
G
1
dê-
22
fs-
1+3
yoi
64
arth'
2
dew
23
fe
44
ad
65
oith'
3
"bu
2k
su'
45
td
66
ëf
k
"bow
25
eho
46
Ch
67
ewf
5
gn:
26
shl
47
eh
68
as
6
ga'
27
tha:
48
u'g
69
ts
T
Jo
28
thoi
49
s-J
70
ush
8
J1
29
is-
50
OiJ
71
owsh
9
pa
30
lew
51
op
72
trth
10
pr
31
ru
52
ip
73
a'th
11
tîT
32
row
53
Ht
74
al
12
te
33
mtr
54
a't
75
tl
13
ku'
3^^
m '
55
uk
76
or
11^
char
35
no
56
owk
77
er
15
Choi
36
ni
57
ëch
78
u'm
16
vu
37
ha
58
ewch
79
an
17
vow
38
hr
59
ëv
80
oin
18
za
39
hws"
60
ev
81
ong
19
zT
40
hwe
61
u'z
82
ing
20
th'tr
41
wu'
62
ozh
83
ë
21
th'a '
42
63
izh
84
ew
s Master List
D
1
da
22
fe-
43
yi
64
nth'
2
dr
23
few
44
ëd
65
a'th'
3
he
24
so
45
ed
66
of
1+
hew
25
si
46
u'h
67
ef
5
gu
26
shu'
47
s-g
68
us
6
gow
27
thu
48
oig
69
ows
7
Ju
28
tha'
49
Oj
70
ash
8
Ja'
29
la
50
ij
71
Tsh
9
po
30
ir
51
HP
72
äth
10
pe
31
rë
52
a 'p
73
oith
11
tu'
32
rew
53
ut
74
Ö1
12
ksT
33
mu
54
owt
75
el
13
koi
34
mow
55
Sk
76
u'r
Ik
cho
35
ntr
56
ewk
77
am
15
chi
36
na '
57
ach
78
oim
16
vü"
37
hë
58
ÎCh
79
on
17
ve
38
he
59
ëv
80
in
18
zu
39
hwu'
60
ewv
81
Tng
19
zow
4o
war
61
iz
82
ewng
20
th'a
41
wo i
62
oz
83
ä
21
th'T
42
yo
63
u'zh
84
oi
-6i^-
1
do
2
di
3
de
k
ba
5
bl
6
8©
T
gew
8
9
jov
10
pu'
11
ta
12
toi
13
ko
14
kl
15
chu
16
cha '
17
m
18
voi
19
zô
20
ze
21
th'u
1
du»
2
bo
3
be
4
ga
5
gl
6
7
Jew
8
pa
9
poi
10
to
11
ti
12
im
13
ka '
14
chu
15
chow
16
vu'
17
zâ
18
zoi
19
th'ô
20
th'e
21
fu
S Master List E
22
th'ow
^3
ya'
64
a'zh
23
fo
44
u'd
65
u'bh'
24
fi
45
Üb
66
uf
25
46
oib
67
owf
26
sew
47
og
63
ÖS
27
shû
48
ig
69
es
28
sha'
íi9
uj
70
Fsh
29
thu'
50
a'J
71
cish
30
lô
51
up
72
aT.h
31
le
52
owp
73
tth
32
ra
53
et
7i^
u'l
33
rt
54
ewt
75
Er
3^^
inê
55
ak
76
oir
35
mew
56
ik
77
om
36
nú
57
•üch
78
im
37
now
58
ech
79
TOI
38
hu'
59
ov
80
a 'n
39
hm
60
Iv
81
ung
40
hwoi
61
ez
82
OTOlg
41
wo
62
ewz
83
a
42
wi
63
üzh
84
~
S life,Bter List F
22
fow
43
yow
64
ewzh
23
SU
44
ad
65
ath'
24
sa '
45
oid
66
"ith'
25
she
46
ob
67
of
26
shew
47
ib
68
if
27
tha
48
üg
69
äs
28
thl
49
a'g
70
ois
29
lu'
50
új
71
"ôsh
30
r"5
51
owj
72
esh
31
re
52
■Bp
73
u'th
32
ma
53
ewp
74
El
33
Dîî
54
at
75
oil
34
lîê
55
It
76
or
35
new
56
ök
77
ir
36
hi:
57
ek
78
tSn
37
hoi
58
u'ch
79
a'm
38
hwo
59
UV
80
un
39
hwi
60
owv
81
own
4o
wtr
61
ÜZ
82
§hg
41
m '
62
a'z
83
0
42
yu
63
ezh
84
i
-65-
S Master List G
1
da:
22
fa'
^3
od
6k
ëth'
2
doi
23
SU
kk
id
65
evth'
3
•bu'
2k
sow
1^5
üb
66
af
h
Sô
25
sha
k6
a'^b
67
oif
5
ge
26
shT
k7
ug
68
OS
6
ja
27
the
1^8
owg
69
is
7
jT
28
thew
kg
ëj
70
u'sh
8
po
29
IS
50
ewj
71
(5*-h
9
Pi
30
loi
51
ap
72
eth
10
tu
31
ru'
52
îp
73
ol
11
ta '
32
mö
53
ot
7k
il
12
ku
33
me
5ii
at
75
ûr
13
kow
3^
na
55
u'k
76
a 'r
11^
che
35
nT
56
ach
77
urn
15
chew
36
ho
57
oich
78
owm
16
vc
37
hi
58
üv
79
en
17
vi
38
hwü
59
a 'v
80
ewn
18
zu'
39
hwa '
60
uz
81
ang
19
th'ä
ko
vu
61
owz
82
eng
20
th'oi
kl
wow
62
azh
83
Ö
21
fü
k2
yë
63
Tzh
8k
e
S Master List H
1
do
22
sa
k3
üd
6ii
oth'
2
"ba
23
sT
kk
a'd
65
eth'
3
"boi
2k
shu
kg
uh
66
üf
k
gu'
25
show
k6
owh
67
a'f
5
jo
26
thö
k7
äg
68
u'b
6
je
27
the
kB
ewg
69
oeh
7
pu
28
lo
kg
aj
70
ish
8
pa '
29
Ii
50
Tj
71
ëth
9
tu
30
rä
51
op
72
ewth
10
tow
31
roi
52
ep
73
ül
11
kë
32
mu'
53
u't
7ii
a'l
12
kew
33
no
gk
äk
75
ur
13
cha
3k
ne
55
oik
76
owr
Ik
chT
35
hü
56
och
77
ëm
15
vë
36
ha'
57
ich
78
ewm
16
vew
37
hwu
58
äv
79
an
17
zo
38
hwov
59
oiv
80
Tn
18
zi
39
wë
60
az
81
ông
19
th'u'
i+0
wew
61
Tz
82
u'ng
20
fE
ill
ya
62
uzh
83
ü
21
foi
i|2
yT
63
cwzh
8ii
a '
-66-
APPENDIX II
PB WOED LISTS
In this appendix axe presented the 20 PB lists, each
consisting of 50 common English monosyllahles arranged in
alphabetical order. For use in articulation testing, the
words in each list should he arranged in a different random
order for each presentation. Randomizing may conveniently
he accomplished hy placing the words on separate slips of
paper which are then shuffled and drawn at random. When¬
ever it is more convenient to divide the testing into short
sessions in which only 25 words are read at a time, each
list may he divided into two random groups of 25 words each.
In the confíete experiment hoth parts of each list should
he read hy the same speaker under the same conditions to
the same listening crew, and the two scores added. Conclu¬
sions should not he hased on the score of a single 25-word
test.
1
2
3
k
5
6
7
8
9
10
11
12
13
Ik
15
16
17
18
19
20
21
22
23
2k
25
1
2
3
5
6
7
8
9
10
11
12
13
Ik
15
16
17
18
19
20
21
22
23
2k
25
-67-
PB-50 List 1 PB-30 List 2
are
26
hunt
1
awe
26
nab
bad
27
is
2
bait
27
need
bar
28
mange
3
been
28
niece
bask
29
no
1+
blush
29
nut
box
30
nook
5
bought
30
our
cane
31
not
6
bounce
31
perk
cleanse
32
pan
7
bud
32
pick
clove
33
pants
8
charge
33
pit
crash
3k
pest
9
cloud
3k
quart
creed
35
pile
10
corpse
35
rap
death
36
plush
11
dab
36
rib
deed
37
rag
12
earl
37
scythe
dike
38
rat
13
else
38
shoe
dish
39
ride
11+
fate
39
sludge
end
Uo
rise
15
five
1+0
snuff
feast
ki
rub
16
frog
1+1
start
fern
k2
slip
17
gill
1+2
suck
folk
1^3
smile
18
gloss
1^3
tan
ford
kk
strife
19
hire
1+1+
tang
fraud
k^
such
20
hit
1+5
them
fuse
1+6
then
21
hock
1+6
trash
grove
kl
there
22
Job
1+7
vamp
heap
1+8
toe
23
log
1+8
vast
hid
1^9
use (yews)
21+
moose
1+9
ways
hive
50
wheat
25
mute
50
wish
PB-50 LlBt 3 PB-50 ListT
ache
26
muck
1
bath
26
neat
air
27
neck
2
beast
27
new
bald
28
nest
3
bee
28
oils
barb
29
oak
1+
blonde
29
or
bead
30
path
5
budge
30
peck
cape
31
please
6
bus
31
pert
cast
32
pulse
7
bush
32
pinch
check
33
rate
8
cloak
33
pod
class
31+
rouse
9
course
31+
race
crave
35
shout
10
court
35
rack
crime
36
sit
11
dodgé
36
rave
deck
37
size
12
dupe
37
raw
dig
38
sob
13
earn
38
rut
dill
39
sped
11+
eel
39
sage
drop
1+0
stag
15
fin
1+0
scab
fame
1+1
take
16
float
1+1
shed
far
1+2
thrash
17
frown
1+2
shin
fig
1+3
toil
18
hatch
1+3
sketch
flush
1+1+
trip
19
heed
1+1+
slap
gnaw
1+5
turf
20
hiss
1+5
sour
hurl
1+6
vow
21
hot
1+6
starve
Jam
1+7
wedge
22
how
1+7
strap
law
1+8
wharf
23
kite
1+8
test
leave
1+9
who
21+
merge
1+9
tick
lush
50
why
25
move
50
touch
-68-
1
2
3
k
5
6
^ 7
8
9
10
11
12
13
lil¬
is
16
17
18
19
20
21
22
23
2l|-
25
1
2
3
h
5
6
7
8
9
10
n
12
13
lil¬
is
16
17
18
19
20
21
22
23
2li-
2S
PB-50
List 5
PB-SO
List 6
add
26
pass
1
as
e6
kept
bathe
27
pipe
2
badge
27
leg
beck
28
puff
3
best
28
mash
black
29
punt
1|
hog
29
nigh
bronze
30
rear
S
chart
30
ode
browse
31
rind (rind)
6
cloth
31
prig
cheat
32
rode
7
clothes
32
prime
choose
33
roe
8
ccb
33
pun
curse
3i^
scare
9
Cx-^ib
3i^
pus
feed
/ 3S
shine
10
dad
3S
raise
flap
36
shove
11
deep
36
ray
gape
37
sick
12
eat
37
reap
good
38
sly
13
eyes
38
rooms
greek
39
solve
lil¬
fall
39
rough
grudge
iK)
thick
is
fee
1+0
scan
high
ill
thud
16
flick
i+l
shank
hill
1|2
trade
17
flop
i+2
slouch
inch
i^3
true
18
forge
h3
sup
kid
kh
tug
19
fowl
i+i+
thigh
lend
i^S
vase (vace)
20
gage
l^S
thus
love
k6
watch
21
gap
1+6
tongue
mast
^7
wink
22
grope
1+7
wait
nose
he
wrath
23
hitch
i+8
wasp
odds
h9
yawn
2i|-
hull
1^9
wife
owls
SO
zone
2S
Jag
SO
writ
PB-SO List 7 PB-50 List 8
act
26
off
1
ask
26
hum
aim
27
pent
2
bid
27
Jell
am
28
phase
3
bind
28
kill
but
29
pig
i+
bolt
29
left
by
30
plod
S
bored
30
lick
chop
31
pounce
6
calf
31
look
coast
32
quiz
7
catch
32
night
comes
33
raid
8
chant
33
pint
cook
31^
range
9
chew
31+
queen
cut
35
rash
10
clod
35
rest
dope
36
rich
11
cod
36
rhyme
dose
37
roar
12
crack
37
rod
dwarf
38
sag
13
day
38
roll
fake
39
scout
11+
deuce
39
rope
fling
1+0
shaft
IS
dumb
i+0
rot
fort
1+1
siege
16
each
1+1
shack
gasp
1+2
sin
17
ease
1+2
slid©
grade
1+3
sledge
18
fad
1+3
spice
gun
1+1+
sniff
19
flip
1+1+
this
him
1+S
south
20
food
i+S
thread
Jug
i+6
though
21
forth
1+6
till
knit
i+7
whiff
22
freak
1+7
us
mote
i+8
wire
23
frock
1+8
wheeze
mud
1+9
woe
2i+
front
1+9
wig
nine
SO
woo
25
guess
SO
yeast
1
2
3
k
5
6
7
8
9
10
11
12
13
Ii».
15
16
17
18
19
20
21
22
23
2k
25
1
2
3
k
5
6
7
8
9
10
11
12
13
l^-
15
16
17
18
19
20
21
22
23
2k
25
-69-
PB-50
List 9
PB-50
List 10
arch
26
lit
1
ail
26
line
"beef
27
mass
2
back
27
maze
birth
28
nerve
3
bash
28
mope
bit
29
noose
h
bob
29
nudge
boost
30
nuts
5
bug
30
page
carve
31
odd
6
champ
31
pink
chess
32
pact
7
chance
32
plus
chest
33
phone
8
cl¿ '-ue
33
put
clown
3k
reed
9
cord
3h
rape
club
35
root
10
cow
35
real
crowd
36
rude
11
cue
36
r.Lp
cud
37
sip
12
daub
37
rush
ditch
38
smart
13
ears
38
scrub
flag
39
spud
Ih
earth
39
slug
fluff
ho
ten
15
etch
ho
snipe
foe
hi
than
16
fir
hi
staff
fume
h2
thank
17
flaunt
h2
tag
fuse
h3
throne
18
flié^t
h3
those
gate
hh
toad
19
force
hh
thug
give
h5
troop
20
goose
h5
tree
grace
h6
weak
21
gull
h6
valve
hoof
h7
wild
22
hat
h7
void
ice
h8
wipe
23
hurt
h8
wade
itch
h9
with
2h
Jay
h9
wake
key
50
year
25
lap
50
youth
PB-50
List 11
PB-50
List 12
arc
26
loss
1
and
26
hunch
arm
27
low
2
ass
27
Jaw
beam
28
most
3
ball
28
Jazz
bliss
29
mouth
h
bluff
29
Jolt
chunk
30
net
5
cad
30
knife
clash
31
pond.
6
cave
31
lash
code
32
probe
7
chafe
32
laugh
crutch
33
prod
8
chair
33
ledge
cry
3h
punk
9
chap
3h
loose
dip
35
purse
10
chink
35
out
doubt
36
reef
11
cling
36
park
drake
37
rice
12
clutch
37
priest
dull
38
risk
13
depth
38
reek
feel
39
sap
Ih
dime
39
ripe
fine
ho
shop
15
done
ho
romp
frisk
hi
shot
16
fed
hi
rove
fudge
h2
sign
17
flog
h2
set
goat
k3
snow
18
flood
h3
shut
have
hh
sprig
19
foot
hh
sky
hog
h5
spy
20
fou^t
h5
sod
Jab
k6
stiff
21
frill
h6
throb
Jaunt
h?
tab
22
gnash
kl
tile
kit
h8
urge
23
greet
h8
vine
lag
h9
wave
2h
hear
h9
wage
latch
50
wood
25
hug
50
wove
-70-
PB-50 List 13 PB-50 LiB-TÎTÎ
1
bat
26
muff
1
at
26
muss
2
beau
27
mush
2
bam
27
news
3
change
28
my
3
bust
28
nick
k
climb
29
nag
k
car
29
nod
5
corn
30
nice
5
clip
30
oft
6
curb
31
nip
6
coax
31
prude
7
deaf
32
ought
7
curve
32
purge
8
dog
33
owe
8
cute
33
quack
9
elk
3k
patch
9
dam
3k
rid
10
elm
35
pelt
10
dash
35
shook
11
few
36
plead
11
dead
36
shrug
12
fill
37
price
12
douse
37
sing
13
fold
38
pug
13
dting
38
slab
1^
for
39
scuff
Ik
fife
39
smite
15
gem
ho
side
15
foam
1^0
soil
l6
grape
ki
sled
16
grate
kl
stuff
17
grave
k2
smash
17
group
k2
tell
18
hack
k3
smooth
18
heat
k3
tent
19
hate
kk
soap
19
howl
kk
thy
20
hook
k^
stead
20
hunk
k^
tray
21
tîig
k6
taint
21
isle
k6
vague
22
made
kl
tap
22
kick
kl
vote
23
mood
kd
thin
23
lathe
k8
wag
2k
mop
k9
tip
2k
life
k9
waif
25
moth
50
wean
25
me
50
wrist
PB-50
List 15
PB-50
List 16
1
bell
26
morn
1
aid
26
map
2
blind
27
naught
2
barge
27
nap
3
boss
-28
ninth
3
book
28
next
k
cheap
29
oath
cheese
29
part
5
cost
30
own
5
cliff
30
pitch
6
cuff
31
pup
6
closed
31
punp
7
dive
32
quick
7
crews
32
rock
8
dove (duv)
33
scow
8
dame
33
rogue
9
edge ^
3^^
sense
9
din
3k
rug
10
elf
35
shade
10
drape
35
rye
11
fact
36
shrub
11
droop
36
sang
12
flame
37
sir
12
dub
37
sheep
13
fleet
38
slash
13
fifth
38
sheik
Ik
gash
39
so
Ik
fright
39
soar
15
glove
1^0
tack
15
gab
ll-O
stab
16
golf
kl
teach
16
gas
14-1
stress
17
hedge
k2
that
17
had
k2
suit
18
hole
k3
time
18
hash
k3
thou
19
Jade
tinge
19
hose
kk
three
20
kiss
k9
tweed
20
ink
k3
thresh
21
less
k6
vile
21
kind
k6
tire
22
jnay
kl
weave
22
knee
kl
ton
23
mesh
k8
wed
23
lay
k8
tuck
2k
mitt
k9
wide
2k
leash
k9
tum
25
mode
50
wreck
25
louse
50
wield
1
2
3
k
5
6
T
8
9
10
11
12
13
11+
15
l6
17
18
19
20
21
22
23
2lj-
25
1
2
3
1+
5
6
7
8
9
10
11
12
13
ll+
15
16
17
18
19
20
21
22
23
2l+
25
-71-
PB-50
List 17
PB-50
List 18
all
26
mist
1
aims
26
gray
apt
27
myth
2
art
27
grew
"bet
28
ox
3
axe
28
gush
big
29
paid
1+
bale
29
hide
"booth
30
pare
5
bless
30
his
"brace
31
past
6
camp
31
hush
"braid
32
pearl
7
cat
32
lime
"buck
33
peg
8
chaff
33
lip
case
3^
plow
9
chain
31+
loud
clev
35
press
10
chill
35
lunge
crush
36
rage
11
chip
36
lynch
dart
37
reach
12
claw
37
note
dine
38
ridge
13
claws
38
ouch
falls
39
roam
11+
crab
39
rob
feet
1+0
scratch
15
cub
1+0
rose
fell
1+1
sell
l6
debt
1+1
sack
fit
1+2
ship
17
dice
1+2
sash
form
^3
shock
18
dot
h3
share
fresh
1+1+
stride
19
fade
1+1+
sieve
gum
tube
20
fat
ii5
thaw
hence
1+6
vice
21
flare
1+6
thine
hood
1+7
weep
22
fool
1+7
thorn
if
1+8
weird
23
freeze
1+8
trod
last
1+9
wine
2l+
got
h9
waste
ma
50
you
25
grab
50
weed
List 19 PB-50 List 20
age
26
hut
1
ace
26
pad
bark
27
lad
2
base
27
pew
bay
28
led
3
beard
28
puss
bough
29
lose (lôôz)
1+
brass
29
quip
buzz
30
lust
5
cart
30
ramp
cab
31
notch
6
click
31
retch
cage
32
on
7
clog
32
robe
calve (cav)
33
paste
8
cork
33
roost
cant
perch
9
crate
31+
rox:ige
chat
35
raft
10
did
35
rout (rowt)
chose (choz)
36
rote
11
duke
36
salve
crude
37
rule
12
eye
37
seed
cup
38
sat
13
fair
38
sigh
dough
39
Bhy
1I+
fast
39
skid
drug
1+0
Sill
15
flash
1+0
slice
dune
1+1
slid
16
gang
1+1
slush
ebb
1+2
splash
17
get
1+2
soak
fan
1+3
steed
18
gob
h3
souse
find
1+1+
thief
19
hunç)
1+1+
theme
flank
1+5
throat
20
in
1+5
through
fond
1+6
up
21
Joke
1+6
tilt
gin
1+7
wheel
22
judge
1+7
walk
god
1+8
white
23
lid
1+8
wash
gyp
1+9
yes
2I+
mow (mo)
1+9
web
hike
50
yield
25
pack
50
wise
-72-
Appendix III
SPOKDAIC WORD LISTS
These two lists of spondees contain k2 words each, a
number dictated by their use on phonographic recordings.
These words have been found to be very homogeneous in
difficulty, and they are useful, therefore, in experiments
designed to determine with high precision the threshold of
hearing for speech.
The two lists may be combined or used separately. In
either case, the order of the words should be changed from
reading to reading.
-73-
Spondee List 1
1
airplane
15
doorstep
29
oatmeal
2
armchair
16
dovetail
30
outlaw
3
■backbone
17
drawbridge
31
playground
k
bagpipe
18
earthquake
32
railroad
5
baseball
19
eggplant
33
shipwreck
6
birthday
20
eyebrow
3^^
shotgun
7
blackboard
firefly
35
sidewalk
8
bloodhound
22
hardware
36
stairway
9
bobwhite
23
headlight
37
sunset
10
bonbon
2l|-
hedgehog
38
watchword
11
buckwheat
25
hothouse
39
whitewash
12
coughdrop
26
inkwell
1^0
wigwam
13
cowboy
27
mousetrap
kl
wildcat
ll4-
cupcake
28
northwest
k2
woodwork
Spondee List 2
1
although
15
hotdog
29
playmate
2
beehive
16
housework
30
scarecrow
3
blackout
17
iceberg
31
schoolboy
1+
cargo
18
jacknife
32
soybean
5
cookbook
19
lifeboat
33
starlight
6
daybreak
20
midway
3h
sundown
7
doormat
21
mishap
35
therefore
8
duckpond
22
mushroom
36
toothbrush
9
eardrum
23
nutmeg
37
vampire
10
farewell
21)-
outside
38
washboard
11
footstool
25
padlock
39
whizzbang
12
grandson
26
pancake
ko
woodchuck
13
greyhound
27
pinball
41
workshop
ll<-
horseshoe
28
platform
42
yardstick
-71^-
Appendix IV
SENTENCE LISTS (TYPE Si)
The following 68 lists contain a total of I360 sentences.
They are designed for articulation tests in which the listeners
write down the sentence as heard. Scoring is in terms of the .
five underlined words in each sentence. Count one point for
each of these five words written correctly and ignore the re¬
maining words of the sentence. In each sentence the five
key words consists of four monosyllables and one dissyllable.
1> The birch oanoe elld on the snooth
2. Glue the sheet to the dark blue background»
eaay to tell the depth of a veil.
U. These dAys a chicken leg le a rare dish.
often served in round bowls.
6. John is Just e dope of long standing.
7* The .luice of lemons makes fine punch.
8. The chest was thrown beside the parked truck.
9* The hogs were fed chopped com and garbage.
10. A cry in the night ch1 lis ag marrow.
11. Blow high low but follow the notes.
hours of steady work faced us.
13* A large size in etocklngs Is herd to sell.
Ik. Many are tauAt to breathe throwdi the nose.
1^. Ten days leave Is coning up.
l6. The Frenchman was shot when the sun rose.
17* A rod is used to oat^ p1nk salmon.
1^. He smoked a pipe until it burned hie tongue.
19« The li^t flashed the message to the eyes of the watchere.
20. ühe source of the huge river is the clear spring.
1. For quick cleaning, buy a hemp rug.
2. The beauty of the view stunned the young boy.
3» Two blue herring swan in the sink.
U. Her purse was full of useless trash.
The colt reared and threw the sick rider.
6. It snowed, rained, and hailed the same morning.
7» An eel tastes sweet but looks awful.
8. Read veres out loud for pleasure.
9* Hoist the load to your left shoulder.
10. He was bribed to cause the new motor to fail.
11. Take the winding path to reach the lake.
12. Red pencil the words spelled wrong.
13« A plung) hen is well fitted for stew.
Ik. The tempo was slow but pioted ig) soon.
1^. Bote closely the size of the gas tank.
l6. Haste nay cause a loss of power.
17* The coast was guarded by field guns in the hills.
iB. Cold, damp rooms are bad for romance.
A true saint is lean but quite h'™»"-
20. Wipe the grease off your dirty face.
List
2
List
k
1.
Death marks the end of our efforts.
1.
Mend the coat before you go out.
2.
The gift of 0pee<^ was denied the poor child.
2.
The wrist was badly strained and hung limp.
3.
Never kill a ansite with your bare hands.
3.
The stray cat tore green kittens.
k.
£ick the ball straight and follow tfaroufdi.
k.
a pest may be a man or a disease.
%
Help the wCTnaa get back to hm: feet.
5.
The coy girl gave no clear response.
6.
Put a dot on the i end sharpen the point.
6.
The meal was cooked before the bell rang.
7.
The hum of bees made Jim sleepy.
7.
What .loy there is in living.
8.
a Pint of tea helps to pass the evening.
a.
a king ruled the state in the early days.
9-
Smolor fires lack flame and heat.
9.
The ship was torn apart on the sharp reef.
10.
The soft cushion hroke the man'e feúl.
10.
Soldiers poured through the wide breach in the wall.
11.
While he spoke. the others took their leave.
11.
The deep cave wound left then etrai^t.
12.
The core of the apple housed a gre^ worm.
12.
He quoted the book by the hour.
13.
The salt breeze came across from ths sea.
13.
a frog grunts loudly if he wants food.
Ik.
The ^^1 at the booth sold fifty bonds.
Ik.
Sickness kept him home the third week.
15.
The purple pup gnawed a hole in the sock.
15.
Give her the gun. he shouted then.
16.
The fish twisted and turned on the bant hook.
16.
The In^ad road shimmered in the hot sun.
17-
a lot of fat sieve a mile racer.
17.
The lazy cow lay in the cool grass.
18.
^esB the pants sew a button on the vest.
18.
Joe blew his base horn wildly.
19-
The swan dive was far short of perfect.
19.
Lift the equare stone over the fence.
20.
Jamas tried his best to gain ground.
20.
The rope will bind the sev^ mice at once.
10-69
P-281
Ugt
ml' OTW th» fno# and plmw la.
^ d—d dog !• no ttao for hunting duojn.
3. Hilt ooup taatoB lito attvod buaiard»
t. The age grinned and gnnehed his yellow teeth»
3. The friendly gang le gone fron the drug atore»
wire toepa ohloto Inalde»
7» Sm the bank under a falae naae.
8» The froaty air paaeed through the coat.
9» to drank a ooto vlth rum therein»
10» The crooked maae failed to fool the mouae.
11» Print her name beeide the oroea.
12. Adding faat leada to wrong auaa.
13» The ehow waa a huge flop at the rery etart.
IK, The berry hung anl ewayed on the bare atem.
loTee his aour and grouchy wife.
16. to the taak quickly or you fail»
17* A eav la a tool uaad for mnlrlng boards.
18» Sha hotned In ^ the goaalp of the girls»
19» The plague killed ^Ir^ cows In a week»
20. Weedo atop the plante frcan getting big.
LI at
1. The dune rose troa the edge of the water.
2. Those words ware the ow for the actor to leaya«
3« Famera hats to use a hoa or rake.
U. A yacht alld around the point Into the bay.
^» The two mat while playing on the sand.
^b*a foolish to mnto a pass at Jane.
7- The ink etaln dried on the finished page.
toll once on thle .job and to dlecharged.
9» Scotch can't be bourfrt today at all.
10. The walled town waa seized without a flAt.
11. The lease ran out In alxteen veaks.
They pulled o fast one on the deacon.
13. The lewd face stared out of the window,
llf. A fine starry night greets the pair.
1^. I am speaking dumb and words.
16. A tame squirrel matos a nice pet.
17. The throb of car voto the sleeping cop.
18. George the second was then queen of the May.
19. Ch^eat men are the worst husbands.
20. The heart beat strongly and with firm strotos.
Uat 6
1. The wagon moved on veil oiled wtoeto.
2. The fleas bopped on both the cat and the chair.
3. Bever buy a blind pig In a bag.
V. March the aoldlere peat the next hill.
A CV9 of sugar makes aweet fudge.
6. Place a rosebush neer the stags.
7* George gave hie sister a lot of ooln.
6. Both lost their Uves In the raging atom.
9. ^ of the eida show In the clroue.
10. ^ a pencil to write the rough draft.
11. to rm way to the hardware atore.
12. Elfdit cope viait the now cook.
13. A cute baby is ^ or croea.
Ik. The clock etruok to mark the third, period.
15. CoUega gflrlB are full of zip and verve.
16. A small creak cut aoroae the field.
17. Boye thrive on ronrfi gamee and candy.
18. Care iniaaaa a^-^^^^ed In anow drifts.
19. TSie eet of ^[na Ut the floor with a craah.
20. toy la • grand aeaeon for hltoe on the road.
Idet 8
1. Tto pearl was worn In a thin sliver ring.
2. The ftult peel was cut In thick Blicea.
3. The Wavy attacked the big task force.
k. See the cat glaring at the acarad mouae.
The crest of the wave vae eight feet below.
6. There are more than two factors here.
7. Breed doge until you win the prize.
8. The climb was warm and done without water.
9* Ann tore her blonde hair in anger.
10. The hat brim was wide and too droopy.
^^le chat and gossip all day.
12. The lawyer tried to lose his case.
13. The Isah curled around the fence poet.
14. Cut the pie into large parte.
1^. Pot a big cravllng bug in her ear.
16. The bait waa aoapped and the black fox captured.
17. Man Btrlva but eeldon get rich.
18. Alwaya cloaa the bam door tlAt.
19. to Imr prone and hardly moved a llab.
20. Soothe the with end dro^a.
IC-69
F-262
M5L2
1. Bé told th« ftu—n to ■end b«(r hârah 1555».
2. Tbo ilush lay d—p along tba etraat»
3* Saldo» hay I aaoo auoh a dopa.
t. A vlap of oloud hung la tha hlua air.
% Shay tha idilakara az^ ahoy jour own faoa.
0P» !■ rlAt. tha dootor aald.
7. A pound of augar ooata aora than agga.
8. Tha fin van aharp and 0^ tha olaar wtar.
9* Tha play aaaaia dull and ^ulta atupld.
10. Bail tha boat to atop it trooi ainhing.
11. Tha tarn andad in lata Juna that yaar.
12. !Z3ia taan foil into a aaan h^^fy than.
13* Tha hoary rock vaa huga and rosa hi^.
It. A tuafc ia uaad to naka ooatly gifta.
1^. Ten pina vara eat in order.
16. Tha bill vas paid every third vaak.
17* is strong end also givae ahada.
Cate Snga asch hata tha othar.
19. The park etratohed fToB the pond to tha ancient house.
20. Hoomb to let go begging this spring.
Uat 11
1. Bill out hla toe on the bottom of tha pool.
2. Sign tha vlll using your ovn nana.
3> Qyder peach pie vlth Ipe oream.
^he oarpat on tha right elda.
Heap ia a vaad found in parta of tha tropice.
6. Thaae flaas muet ba detftroyad eoon.
7* Hlca girla often anoka nov.
^ l*"» back kept hla ecora lov.
lata in tha day drugs tha body.
10. "Thua" ia a vord used to Join clausas.
11. find Joy in tha slaplast thinga.
12. Type out three lists of orders.
13. Lace ^e ehoe vlth the flat leather strings,
lli^. The harder ha tried the leas ha got dona.
1?. You f^got to note this fact.
16. You can hardly keep a vlfa on that pay.
17. The taste vas bland end flat to scmbs extant.
18. Ice has a nasty vet feel.
19. Wa seen to lose too much for oonfort.
20. Ovia hitfi in air make veird noieea.
Ust 10
1. Tha bat flaw straight for seven gdlae.
2. The pipe began to rust ^diile nev.
3* Stro^yp a skunk is a dumb vay to act,
k. Open the crate hut don*t break the filase.
Yea, the base rnmnwr has reached bona.
6. Add tha sum to tha product of thaaa three.
7* Thievaa \AiO rob friends daeerva Jail.
8. Tha ripe taeta of cheese inprovas vlth age.
9. Act on those orders vlth great speed.
10. The hog crawled xpder the hl^ fence.
the vat over tha hot fire.
12. Kftaen men stood on his bare cheat.
13. Tha bark of tha pine tree was allag- and dark.
Ik. She p"*» 1 her full aklrt over her head.
15. Leaves turn brown and yellov In the fall.
16. The pennant waved when the vlnd blew.
^7. Split tha with a efaarp, ^ulck blow.
18. People stub toes on din walks.
19. Tha boya In the old ho\isa vara silent aa • tomb.
20. Bum peat after tha loga ^y out.
List 12
^^3-a gape at deads of great daring.
2. The hOBB ran the ehov vl'Ui s watchful giaa.
3. Tha cup CTackad and spilled ite contanta.
k. Pasta can cleanse tha nost dingy teeth.
The slang vord for raw \dilBkay Is booze.
6. Be hia h1nfl pay in a maty trap.
7. Ghoets lurk in dark comera of tha cava.
8. It*e easy to vln if you bat vlth care.
9. BlrmJa Base vora an avaning gown.
10. Es Beamed a dunce until ha warmed to tha task.
11. Keep your faoa turned to the panthar.
12. Jóbn la mora fond of praoiaa facts.
13. I felt tha touch of a smooth, silky akin.
Ik. Eat turnips and greens to stay veil.
^a*ll use three more coplee.
16. The vharf could be ee«i at the farther shore.
17» Eeel the heat of the vaak dying flsma.
Iß. Tha tiny girl took off her hat.
19. What fata viU meet tha slovast boat?
20. A craap is m small danger on a svln.
IC-89
P-263
Uet
1. The fteaa refueed to grant hin leaTe to ¿o.
2. A ahrev'a tongue la aharper than b aword.
3- To earn a dime tatea endleaa vork.
He aald the aame nhraae thirty timea.
5. Pluck the brirtit roae without fear.
6. Two nlua aeven la leaa than ten.
7- The aguare waa void of ueoule■ nice, or doge.
8. Duak fell before the drunk did.
9- llie glow deepened In the eyea of the aweet girl.
10. Brl^ your prohlena to the wlae chief.
11. A gnart la .luat enouA for a anall (goup.
12. Write a fond note to the friend you oherlah.
13. What knaye has atolen w rare hook?
lU. Clothea and lodging are free to new man.
15. Do you feel colder In the late fall?
16. A œag caught the line In a sturdy grip.
17. The preacher yelled at the aly old goat.
18. We frown idien eyenta take a tad turn.
19. The cold yolee made Ita sharp coanenta.
20. Port la a strong vine with a anoky. taste.
List 15
1. The spot on the blotter was oauaed by the green pen.
2. Twlat the tall to make the Hon roar.
3. 'Hie man from Spain flew In a allyer blimp.
h. Mud was spattered on the front of his white ahlrt.
3. The cigar burned a hole on the deak top.
6. A mouse la tiny but sgueake loudly when acared.
7. Can you rtm a faat hundred yarda?
8. ISie empty flaak stood on the bare tray.
9. l^e lltdita flickered tdille they came on.
10.. A speedy man can best thla track meev
11. Coated In grease she awam the vide bay.
12. He broke a new shoelace that day.
13. The coffee atand la too high for the oouch.
lb. The urge to write short atorlea la rare.
13. Put s nickel on the drum and you'll be aayed.
16. The pencils have all been used.
17. Hie pirates aelted the crew of the lost ship.
18. We tried to replace the tea but failed.
19. She aeved the torn cost gulte neatly.
20. The cushion la fat, red, and of llrfit velidit.
List lb
1. Hhe young kid .lumped the rusty gate.
2. He felt no gualm about saying yea.
3. Gueaa the results from the first acorea.
b. nie rowdi fliditer needs a cloee guard.
3. nie c»?'" words he read were aage adylee.
6. Beer la better with deep idilte foam.
7. A salt pickle tastes fine with ham.
8. nie .lust claim got the rlrfit verdict.
9. Oírle and women must vork at the .lob.
10. Bablee fear a fall and a loud noise.
H. nie sand scattered like chaff In the wind.
12. Short bathing suite should be filled.
13. Joyce waa abort but packed a heavy punch,
lb. Hire a hack since no tails are here.
13. Theae thlatlas bend In a hlidi wind.
16. nie face la strange but I know the perfume.
17. TvplBta chew gum to keep sane.
18. Pure bred poodles have curls.
19. His answer aounda like "yes".
20. nie tree top waved In a graceful way.
List 16
1. The .lacket hung on the back of the vide chair.
2. At that hlidi level the air la pure.
3. Ihcip the two whan you add the figures,
b. Take a letter to the boss. Mtas Jones.
3. A filing case Is now hard to buy.
6. An abrupt start does not win the prize.
7. never la a short word but a long time.
8. Wood ,1s beat for toys and blocks.
9. nie Offlee paint was a dull. sad tan.
10. He knew the skin of the great young actress.
11. A film of dust was on Its rough crooked wann.
12. A blotter will soak up spilled Ink.
13. A shower of dirt fell from the hot pipes,
lb. Does the absurd fact make him Hu¿h?
13. Steam hissed from the broken valve.
16. nie child cruelly hurt the small dog.
17. niere was a sound of dry leaves outside.
18. Abe gulckly took the chance to play.
19. nie sky that morning was deer and brldbt blue.
20. ^ .ludg» can excuse such aots.
IC-69
P-28b
!• Tom BcrapB llttBrnJ. tbs BtonB floor.
5. Sanflay Ib the T)»«t part of the veelc.
3. Who here upeet the trayt
H. The doctor oared hlB with eweet plUe.
?• Ele tody vae foand epravled the length of the ted.
6. He expoeed hie ahite ekln to the hot ean.
7. The sea girl vae fired today at noon.
8. They felt gay vhan the ehlp arrived In port.
9. Add the etore'B aoooant to the laat cent.
10. To aoouae In haate le to aek for a Qalok tloa.
11. Add tame holee In wool cloth.
12. Marder talee ehoald te to write.
13. From aeed to tree le a natter of yeare.
Ik. It le beet to nee a clear way of egeaMng.
1?. A toft of grey hair was grcwlng on hie
16. El^t ndlefl of woodland tamed to waate.
17. Toar decayed tooth will ache.
18. The third act vae d"n and tired the playera.
19. A joang child ehoald not Buffer frirfit.
20. Add the celui and put the aun here.
Idet 19
1. Where were they whan the nolae etarted.
2. The paper tog la full of thanb taeke.
3. The Jape fly Zeroa through the oloude.
k. A red noae telle of feet Urlgg.
3. Sell yoar gift to a turer at a good gain.
6. A plane nakee huiwina Into tlrde.
7. The tonga lay tealde the Ice pall.
8. The affair had gone much too far.
9. ^nie pétala fell with the next p\iff of wind.
10. Sturdy love waa what aha craved.
11. The rape caae vae In court for thirteen weeka.
12. I deny the alara of the laat apeeoh.
13. Bring your teat compaea to the third elaee.
Ik. Saddened ae they were, they coald l'arfi
13. famere caae In to threah the oM crop.
16. The trown hoaae waa on fire to the attic.
17. A lure la aaed to catch trout and flounder.
18. Float the aoap on top of the tath water.
19. The atate can toaat of Ita rural dirt roada.
20. Age la no atop to te^"° apry.
Llat IB
1. It la the etyle to telong to aone new cult.
2. I ad^re and love a good cook.
3. There the flood nark la ten Inchea.
k. Tell nothing end guard her fair naiae.
9. Ee carved a head irom the round tlock of nartle.
6. Keep a smart way of wearing clothea.
7. The horae trotted around the flald at a trlek pace.
8. She lovea telng made a fool.
9. The deuce le an cm of lU luck In carde.
10. An «Aiilt. la laat a child wown
11. The fruit of a ftg tree la apple-ahaped.
12. Put the vert atrlotly In the rlrfit place.
13. Com cöba are ueed to Irlndle flrea.
Ik. ÎS advice la to ahut your trap.
15. QT a atory with a acand punch Una.
16. m «"I the door olceed ehe whlaperad.
17. The doctor aald he waa etruck da»ib.
IB. Ladlea pin their hair In thin neta.
19. The devil poked the poor nan Into the fire.
20. My ££ wlkl Boon wear oat writing.
Uat 20
1. Can you afford to have that tale told?
2. A tlue crane le a tall wading tlrd.
3. Bot feared to sea Um alone.
k. The Blrtit of an eU la the J02 of ag^ child.
9. A freah atart will work auch wondera.
6. The club rmted the rink for the fifth night.
7. After the dance, they a^ com flakea.
8. ^le hoeteaa taarfit the deaf nald to aerve.
9. The llaht went dla amid strange events.
10. Ee wrote hla laat novel there at the Inn.
11. High fever meana tad health.
12. A girl ourtit never to seem a prude.
13. Even the woret will teat hla hlidi score.
Ik. Boll the cores to make a JeU2 tase.
19. The cernant had dried when he moved.
16. The loss of the second ship waa hard to take.
17. The fly made hla along the wall.
18. Do that with a wooden stick.
19. Uve wires should te kept covered.
20. The large houae had hot water tape.
10-69
P-289
Uat 21
1. Watch for min buga in aoft applea.
2. It la hard to eraaa blue or rod 1^.
3. Write at once or you'll eure regret it.
U. ihe doortoob vaa inade of bright clean braaa.
The vreok ooounred by the bank on Main Street.
A pencil vith yhite paint vritea beat.
7« ÏOU ndfidit hare to ooajc a calf to drink from a bucket.
8. Sohoola for ladiea teach charm and grace.
9. Hla b1 wnk gaze apoke of sheer vonder.
10. 'Ehe ahone vith a ateadj green flame.
11. "They took the mm and the bw to the foreat.
12. Lover the aiok boy to the first deck.
13. A atorm came vith the ianenae b3nok olouda.
Ik* A true stitch sayee time and vorry.
1^. ihe ancient coin vas a\ilte dull and vom.
16. Shout as loud as yoinr lungs allov.
17. Ihe ahalgr bam fell vltii a, loud craah.
18. Seep the man you have. I advlee.
19. Jazz and svlng fans like fast music.
20. They opened the deek and s^ a b^ gun.
Liât 23
1« Some ada aerve to cheat buyers.
2. Jerk the rope and the bell rings veaklj.
3* A vaxed floor helps us lose balance.
French art la not seen on poatoards.
Sell the dam auto, it's d^ as a bone.
6. Madam, this is the best brand of 00m.
7« On the islands in the sea tha breeze is soft and mild.
8. The play began as soon as ve sat dovn.
9. Tou'U have a share of pie in the s^ in the futinre.
10. This vlU lead the world to more sound and fury.
11. ^a ahade trees tovared large in the gloom.
12. The pleat ripped and zipped down to her ankXea.
13* Be nice, dcm't snap your rubber gum.
1^* Cacar is a sly raka from way back,
^5» Before you fry the egg, add aalt.
16. The rush for funda reached ita peak Tuesday.
17. In dull Bad Yoicea they chant the dirge.
18. The birch looked atark ^dilte and lonesome.
19. The box ia held by a bright red enapper.
20. To make pure ice, you freeze water.
List 22
1. Health suffers when food goes bad.
2. Brftfls ahlnes, but its cost is little.
3. Rake the rubbish up and bum It too.
Slaah the gold cloth into fine ribbons.
5. The silt flows Into the pond until It is filled.
6. Hov can you cuastion the fact that we will win?
f, Try to have the court decide the case.
8. IlMy are puahed back each time they attack.
9. We Inylte the Duke to elt In your box.
10. He broke his ties with groupe of former friends.
11. Thay floated on the raft to swa their srtilte backs.
12. The ash drifts across the drear Isvns.
13. The buzzards swam to one strange spot.
ll(. She bared each arm and coyly bowed.
1^. The nap ^ an X that meant nothing.
16. dog's tail almost talks.
17, Bob lobster stuffed with crumbs.
18« Yo^^"g Sue was betrothed to a sour puss.
19. One par is quits good.
20. Whitings are aman fish cauritit in nets.
List 2k
Sot stories sell the worst trash.
2. The first worm gets snapped early.
3. Jump the creek emd hurry up the bank.
Boot seas are fuH of fi^iy brine.
Tell and clap as the curtain slides back.
6. They are men wfti^ the of the road.
7* Both brothers wear the same size.
8. The work la easy, the hours short, but no pay.
9. In son» form or other we need fun.
10. He ran into a swarm of mad little bees.
11. Put it on the cuff, and serve a second round.
12. The prince ordered his head chopped off.
13. Whm bells toll, foT-kp flock to chapel,
ll^. ^ heated chums swim in the creek.
OPOtt the cask and let .ley reigi.
16. The brawny Scot flapped hie tweed kilts.
17* Houaes are built of red clay briete.
IB. Ducks fly north but lack a compaes.
flavOTB are xiaed in fizz dritte.
20, Sweet pills do leaa good than bitter.
ic-69
F-286
LlBt 25
1* Salee rlee vJien ceaeore gay no.
2. It*B hard to dlaguiae a cute ahape.
3* Canned peara lao¿ full flavor.
The darte pot hung In the front oloeet.
CMrry the pall to the veil and epill It there.
6* Gravi up and perch youreelf on top of the
7. The shell idiletled past hie dated head.
8. He left ner with a hlaote eye yeare ago.
9* The train brought our hero to the big povn.
10. I am sure that one auch var Is enough.
11. Grajf paint etretched tox" ailes around.
12. Soft noises sound on the bare stage.
13. Roete the boat and end life like a moron.
Ih. mctt-al life vore her dovn f<iet.
1^. Cats provled in the etorage place.
16. The rude laugh filled the empty room.
17« Mgh seats are beet for football fans.
18. Mitee paid the toll for vile nurdw.
19* Roam the vorld, boys« don't miss the chancea.
20. Tea served from the brovn .lug is tasty.
List 27
1. The .Ixmte yard had a mouldy amell.
2. One volt von't ahocte much, you aiaay.
3* .The flint sputtered am lit a pine torch.
U. If you must punish San, uae the ehed.
Soate the cloth and drovn the aharp odor.
6. A anil an boor is bound to ba left in the cold.
7* The ahelvea were bare of both Jam or cracker a.
8. Dotty has twelve jara of beane to can.
9« A Joy to every child is the avan boat.
10. Yea and no ai'e both the right ansver.
11. All eat frozen em watched the screen.
12« The braggart caused a pain in my etiff necte.
13. A swarm of dxist attmg his temer eyea.
lU. Teiapt ma not, my strength has limita.
1^. I owe him a hefty aock on the chin.
16. To reach the end needa much courage.
17. The shell buret the tip of the taH ateeple.
18. The hlBB, bang, and boom came from far away.
19. A hand and a bunch of fingere are the same thing.
20. Shape the clay gently into block form.
Uat 26
1. Cub looae and the deuce with the shore patrol.
2. Gtaaw on bones so yoxzr teeth get pearly.
3. The man's left Hiwh dangled from the chair.
k. A toeh of pepper epoila beef etsw.
Ihe WAC will win the war by tender means.
6. Wrath poured on othwe, comea home.
7. Mob acta aie yioioua forms of ein.
8. Gate hae been a comoon alang term.
9* Poor punch is the death of aome partlea.
10. A zeatful food is the hot-croee bun.
11« the twin pinched the pearl necklace.
12. Cut the cord that binds you bo tifditly.
13. 71s for the colored spot in a bee line.
1^« Purple hats come out on top.
1^. The red tape boxmd the ainuggled food.
h<M?se or beef, ia ^ deaire.
17. Look in the comer to find the tan ahirt.
18. The cold drizzle will halt the bond drive.
19. nine were hiyad by the firm to dig the ruins.
20« A swell chef ia dear to big kitchens.
Ust 28
Slip na a cigar or I'll aock you one.
2. followB guilt in auch a case.
3. A ridge on a smooth surface la e. bump or flaw.
The hedge apples ataln your hands green.
"My life end timea" is a trite column nead.
6. Quench your thirst, then eat qrackera.
7. She Jumped from the burning plane in her sleep.
worde ia a tar on my feeble brain.
9. Men remain weak from birth to death.
10. The pipe of peace amella of dying leavea.
11. TiAt curie get Ump on rainy days.
12. The mute muffled the hl^ tones of tne horn.
13. The gold ring fits only a pierced ear.
Youth ia gone, dry age renaina.
1^. Harah «nd pleasant eo\mdB make a tone poem.
16. Hold your horses, the last comes eoon.
17. Tbe old pen was covered with hard fudge.
18. It is fun to play fox and hounds at recesa.
19* Wat^ the log float in the diemal swamp.
20. The node on the atalk of ^eat grew daily.
IC-69
P-287
List 29
1* nie heap of fallen leaTee vaa eet on fire.
2. Write fast If you want to end early.
3* She bought vex for the oar, tablee. and floors.
Els ehlrt van clean but one button vas gone,
nie barrel of beer vaa a brev of Belt and hope.
6. Tin oana are absent froBi store shelves.
7* A half pint of ale Is hardly a hip.
8. Slide the box Into that enpty space.
9* The plant grew large and green In the window.
10. He irtiooped and yelled, but no voice replied.
11. The bean dropped down on the vorkaaa's head.
12. rev men are coarse enough for her.
15. The boy's story took a queer,.tvlst.
1^* Say your piece. stranger, or I shoot.
1^. Pink clouds floated with the breeze.
16. She danced like a swan, tall ana graceful.
17* The tube vas blown and the tire flat and useless.
Id. It Is late aomlng on the slov wall clock.
19* Carry lae back to my hosae state.
20. Let all Join as we sing the last chorus.
List 31
1. Screw the sans cap on as tight as needed.
2. Tins brings us many changes.
The purple tie was ten years old.
k. Metal which la coarse Is low In price.
Men think and plan and sometimes act.
6. The stork la a strange bird by repute.
7« To tease a mad dog Is to ask for trouble.
8. Pill the Ink Jar with sticky glue.
9* Short straight skirts are the vogue In dresses.
10. Ee smokes a big pipe with strong contents.
11. He will twit her about her poor speech.
12. Ve need grain to keep mules healthy.
13. The sick and pallid man felt cold and faint.
lU. Pack the records In a neat thin case.
1^. Pox and geese is a game played In winter.
16. The gun spoke and spun him right over.
17* The crunch of feet In the snow was the only sound.
18« Cars are noisy, but make trips nice.
19. The copper bowl shone in the sun's rays.
20. Boards will warp unless kept dry.
List 30
1. The last switch cannot be turned off.
2. The fight will «d In Just six seconds.
3. Don't sprain yoinr tonsils on the bass note.
Her first real try at singing was a howl.
5. The tough kid knocked the sissy on the bricks.
6. Rubber shoes are swell in a damp marsh.
7. The store walls were lined with colored frocks.
8. Villains are stock roles Inigror shows.
9. Ihe peace league met to discuss their plana.
10. Hide the bottle; here comes Jim.
11. Two plus one Is too many.
12. -Seek Joy, you die soon eno\>gh.
13. rise of a person to fame takes luck,
llf. The oyster died on the edge of the salt sea.
15. The otter Is caught for its rich dark fur.
16. Paper Is scarce, so write with much care.
17. The quick fox Jumped on the sleeping cat.
18. Hope springs endless In the male breast.
19. The nozzle of the fire hose was bright brass.
20. Itast I say "uncle" a score of times?
List 32
1. The plitsh chair leaned against the rail.
2. The clink of glass when stioick by metal Is nice.
3. Bathe and relax In the cool green brine.
Mine rows of soldiers stood In line.
3« Toll and squeal, and you still get a. beating.
6. The beach Is dry and shallow at low tide.
7. A chip of steel cut deeply Into the skin.
8. The Idea Is to sew each edge straight.
9. A hen In a roost is worth two turkeys.
10. The kitten chased the dog down the street.
11. Tea beats coffee In a pinch.
12. Pages bound in cloth make a book.
13. Try to trace the fine lines of the painting.
lU. jo>'"»y won't fetch a pall of suds.
15. Fall, and you lose both gifts and respect.
16. Vomen foim less than half of the group.
17> The zones merge J.n the central part of town.
18. A gem In the rough needs work to polish.
19« A buck less Is worth• "Oie extra time.
20. Code Is txsed when secrets- are sent.
ie-69
F-ses
Uat 55
1* Ploua hopea may cauae the loea of life#
2. Moat of the aeve la eaey for vb to hear.
3* He aenaed a sharp pang of sheer dlanay.
He uaed the lathe to nahe a hraaa gadget«
übe mare ie sleeker than pure allk»
6. Ihe vane on top of the pole revolved In the vlnd.
7* Mince pie la a dleh served to children.
8. Hie tone nade each girl retreat.
9» TtAt fact la the key to a lurid tale.
10. He didn't apeak, nor vould you.
11. The clan gathered on each dull night.
12. Let It hum. It glvea ue vamth cmd comfort.
13« A dome built on eand falla to endure.
Ik, Ho bullet hole could aar that face.
15» The fit was tight, and the coat confined his arms.
l6. It eeema quite odd to have Arthur here.
17» A child * s vit retrieved the day for us.
18. To fend against the vor Id vaa hie aim.
19. Tack the atrip of carpet to the vom deck.
20. Hart Tuesday ve vote.
Uat 3^
1. The thav early and freed the etreaa.
2. He took the le^ and kept It the vhole distance «
^ velt roas vhere the vhip had fallen.
U. Ihe key you dealgied vlll fit the lock.
5» Plead to the council to free the starved thief.
6. With a laet moan he fell over.
7» Youth is a stuff vhlch lasts a sumer.
8. Better hash Is made of rare beef..
9. Did he claim I took the little child there?
10. You deem It a. duty veil done.
11. A real scoop keeps the public's faith.
12. This plank vas nade for valklng.
13. A cop yelled, and the riot stopped at once.
lU. The lake sparkled In the red hot sun.
15. He cravled with care along the ledge.
16. Tend the eheep vhlle the dog vandere.
17. It takee a lot of help to finish these.
18. Mark the spot vith a algi painted red.
19. Take tvo shares as » fair profit.
20. The fur of cats goes by many names.
List 3^
1. A tvltch of the tall acatteared the dense files.
2. When he sings It shakes the loose rafters.
3* A pike vas a veapon of var long back.
4. Pour the etev from the pot into the plate.
bhe child to the volves. you rascal.
6. Xach penny shone like nev.
7. Then there la no cauae for vaste to occxtr.
8. Bie hag vent to the vooda to gather sticke.
9. Hear the vren sing of varmth.
10. What this sentence needs Is a noun.
11. A suitor Is a rere bird In these times.
12. The dirt piles vere lined along the road.
13« Qie logs fell and tumbled Into the clear stream.
1^. Juflt hoist it and take it avay.
15. A ripe plum la fit for a k^»g'« palate.
16. Our plans right nov are hagy.
17. Bi-asa rings am sold to natives.
id. It takes a good trap to capture a bear.
19. Peed the vhlte mouse vith flover seeds.
20. If vend, ecnecoe mist spend.
List 36
1. East winds bring colds and fevers.
2. He asks no person to vouch for him.
5* A trite Joke evokes real tears.
Go nov and come here later.
3. Bold as a lion, Jim threv the turf.
6. He let the vlndov close on his thumb.
7. The peevish miss said harsh vords.
8. A sash of golden silk viU trim her dress.
9. The Japs bound his ankles vith thick stripe.
10. Soap can vash some filth avay.
That move means the game is over.
12. He vrote down a long list of Items.
13. A siege will fall to crack the strong defense.
Popp 'tbe stuff that makes for happy dreams.
15. Grape .luice and vster mix veil.
16. A louder sound could fade and yet be heard.
17. Do the Job and question me not.
18. Pave the road vith nice sticky tar.
19. Pake stones shine but cost little.
20. Hie scalp begins to look quite pink.
IC-69
P-a09
LlBt 57 Llgt 39
1.
The drip of the rain made a ceaseless sound.
1.
The wraith drifted from the gloom of the cave.
2.
Smoke poured out of every crack.
2.
Three for a dime the young peddler cried.
3.
Serve the hot rum to the tired heroes.
3.
The sense of smell is sharper than that of touch.
k.
It*s free today, so try your luck.
1^.
No hardehip seemed to keep him sad.
5.
Hear the captain's words, if you want to die.
5.
Grace makes up for lack of beauty.
6.
Dare not the anger of old gods.
6.
Nudge gently but wake her now.
7.
Though he sought peace, he found sorrow.
7.
The news struck doubt into restless minds.
8.
A part is not the whole thliig. that's logic.
8.
Once we stood beside the shore.
9.
Take your lips away from his.
9.
A chink In the wall allowed a draft to blow.
10.
Much of the story makes bad sense.
10.
Fasten two pins on each side.
u.
My fist will teach him to stop lying.
11.
He left CI mess of spinach in the deep sink. ,
12.
The sun came up to light the eastern sky.
12,
A man's creed is his own affair.
13.
Heave the line over the port side.
13.
A.cold dip restores health end zest.
Ik.
The hue is blue, the mad-painter said.
Ik.
He takes the oath of office each March.
15.
A lathe cuts and smooths any wood.
15 b
The sand drifts over the sill of the blasted house.
16.
I've known its true all the time.
16.
The knell was sounded for our false hopes.
17.
Teach him manners or spoil his life.
17.
The point of the steel pen was bent and twisted.
18.
Fold it once, then stow it inside.
18.
There is a lag between thought and act.
19.
A sifäit to gladden hearts is the new moon.
19.
Seed is needed to plant the spring com.
20.
Xt'e a dense crowd in two distinct ways.
20.
Draw the chart with heavy black lines.
LlBt 38
1. Hie hip a truck the knee of the next player«
2. The stale smell of old heer lingers.
^« The desk vas fina on the shaky floor.
k. It takes heat to hring out the odor.
Beef is soaarcer than fine gold.
6. A yiggift and a. shake hotii mean the same.
7. Furl the sail and steer the ship northward.
A sad Jew sold Persian rugs.
9. Who called him a sullen brute?
10. It had the taste of veil bumt leather.
U, A cone cost five cents Monday.
12. A £od is what peas always grow in.
15. Jerk the dart from the cork target.
Ik. dW hair deftly.
15. Ho cement will hold hard wood.
16. We now hare a new base for shipping.
17. Lust fires his heart at the shameless eight.
18. A list of names is carved around the base.
19. men lose their souls with ease.
20. The sheep were led to slaughter by a goat.
List ho '
1. The boy owed his pal thirty cents.
2. The chap slipped Into the crowd and vas lost«
5. Bats are vom to tea and not to dinner.
k. The ramp led up to the vide highway.
5. Beat the dust from the rug onto the lawn.
Lean meat helps sxistaln a diet.
7b Say it slowly but make it ring clear.
8. The straw nest housed five robins.
9. Talking through » gag is very hard work.
10. Screen the porch with woven straw mats.
11b Ihis horse will nose his way to the finish.
12. The dry scab protects the deep scratch.
15. ^ picked up the dice for a second roll.
1^. Ihe b^^Hf of the coins wüd be needed to pay his debts.
1^« The lad must le€im to shoot a pistol.
16. Ruth would cheat her brother to gain fame.
17* A slug of strong whiskey turns the trick.
18. We miss the note of Joy his presence brought.
19* 3he nag pulled the frail cart along.
20. Strange pets are found in iCTialy spots.
ic-69
p-290
List kl
1« Tirl«t the yelTe end releeee hot
2. Cie reap of the ehoe had a gold "bookie.
3* Bie reek of "burned rage itches my noee.
V, Qm keen ragor eliced the skin off.
Ihe rent Is due, "but ufao has nopey.
6, Hall the chair to the sloping floor of the roen.
7* Sie Blllj young nald hong her head.
8. ?or Pete's s۟ce. let hin scatter his oatst
9* tvlce and batter dowi the door.
^0. lew pants lack cuffs and pockets.
11. Bet all the noney you can scrape 19.
12. A cleft chin Is no sign of courage.
1^. The MELrsh vlll freeze vhea cold enough,
lb. Braye souls suffer death but <aice.
1^. They slice the sausage thin in night clubs.
16. The bloon of the rose lasts a brief instant.
17* A gray aare walked before the colt.
18. Breakfast buns are fine with a. hot drink.
19« Show TlM frc« nine until noon.
20. Hers is a surface charm and has no depth.
List 45
1. The clothes dried on a thin wooden rack.
2. Bouth of Maine the land extends to the west.
3. Turn on the lantern which gires us light.
A. Sie cleat sank deeply Into the soft turf.
Die bills were mailed promptly on the tentii of the month.
6» To hare Is better than to wait and hope.
7. Salt Is less rare than money to most.
8. The price is fair, for a good antique clock.
9« The music played irtille they talked.
10. A great feat. In Tiew of her tender age.
11. The war has made many folks poor.
12. Dispense with a yeet on a day like this.
13« The bunch of grapes was pressed Into wine.
lA. Be sent the figs, but kept the ripe cherries.
13« The hinge on the door creaked with rusty age.
16. Use wealth wl^ care, to arold harm.
17. The screen before the fire kept In ^e sparks.
18. A chunk of struck him squarely.
19. The whistle cm the dark etraet was » good sound.
20. A pœ is email, but holds yaat power.
List A2
1* Bottles hold four kinds of rum.
2. Tom left at the fork of the concrete way.
3, The man wore a feather in his felt hat.
A* Be Triieeled the bike past the winding road.
3. T*** small bomb explodes amid the trucks.
6. Drop the ashee on the worn old rug.
7. The desk both chairs were painted tan.
8. Help me. If you wish to a^ yourself.
9« Thin brunettes hawe few cuiree.
10. "WiTywr out the used paper cup and plate.
11. freedom is the thing to be sought long years.
12. A clean neck means a neat collar.
13. sat on her big back porch.
lb. The couch coyer and hall drapes were blue.
15. John asked Jane to run away.
16. The stems of the tall glasses cracked and broke.
17. Dwarfa. too, haye stcmache aches.
Id. file your nails before you put on your hose.
19* Hell, the darl^**gi walked on epll^ heele.
20. The wall phone rung loud and often.
List Ab
1. fly by night, and you waste little time.
2. Click glasses helped him read the print.
3* Blr^ and death amrk the limita of life,
b. He pulled no punch in each fast attack.
Just pick fights with pigmies.
6. The chair looked strong but had no bottom.
7* A curt answer turns wrath your way.
8. It was staged by real Ilye moyle stars.
9* A real earl la Bnglleh and a rare bird.
10. The shade flew wildly In the high yln^.
11. The harp, when struck, yields angel's tones.
12. The only thing to hope for le a qulok win.
13* A fur muff is stylish, once more.
lb. lis aim hung queer limp and lifeless.
13» The thigh extends from hip to knee.
16. Think no thoughts except nice clean ones.
17« The tin box held priceless stones.
l8. Ve need an end of all such chatter.
19* The case was puz«HTtg to the old and wise.
20. ihe br^gh-fc laatems were gay cn the dark lawn.
P*291
Uat 45
!• Ye don't get much »oney but ve hare fun.
2. The youth drore vlth «eet. but little eklll.
3* Fire years he llTed vith a shaggy dog.
A fence cuts through the comer lot.
3» The iray to saye noney Is not to spend auch.
6. Shut the hatch before the vayes caye it In.
7* Cxe states vere all of one mind on the question.
8. The odor of spring malees young hearts jump.
9« He aelca about the dead old men.
10. These days reylye the horse and coach.
11. Grab him by the left ankle If he shouts for help.
12. Punk emits smoke vhlch bugs hate.
13« Cmck the ualnut vlth your sharp side teeth.
Ih. He offered proof In the foim of a large chart.
13« Send the stuff In a thick paper bag.
16. The thick fur svsjaed vlth hopping fleas.
17. Bound all the vorld the var Is being von.
iS. Her tone Is not Ideal but can be heard.
19. & vas let out on ball the day the bond vas posted.
20. A quart of milk Is vater for l^e most part.
List U7
1. A force equal to that voxild moye the earth.
2. Her practised arm held the blade firm.
3. Cut the end vord and It cqsms out eyen.
like to see clear veather.
3* Gate vlth ave at this Instance of zeal.
6. The vork of the deyll Is seen on each side.
7« Find a phrase to sell this to the public.
8. toep mum or else you'll be sorry.
9. A cough may herald dire th^"g" to ccsie.
10. !fake a chance and vln a china doll.
11. The shell bxirst near the spravllng dock.
12. Shake the dust from your shoes. stranger.
13« Froth and foam are no sign of fine liquor.
1^. Be kind to sick old people all the time.
13* Duck first, and then ask questions.
16. ^e square vooden crate vas packed to be shipped.
17. A leap In the dark vlH land you In trouble.
18. The dusty bench stood by the stone vail.
19. We dress to suit the veather most days.
20. 1» vhen you say nasty vords.
List K6-
1. They told vlld tales to frlghtwi him.
2. T2m smell of skunk, once felt, you von't forget.
3« The three story house vas bullt of brick.
h. The race vent to the svlftest of the flye yachts.
3. The right vlng of the bird vas iCTiger than the left.
6. Where are the full meals of a year ago?
7. A large passage vas In the rear of the ground floor.
8. Carrots and peiuB I loathe vlth a full heart.
9. At the desk sat a mw In a blue sveater.
10. Oats are a food eaten by horse and men.
11. Tell me vhat goes on today, please.
12. Their eyelids droop for vant of sleep.
13. That's fine, he replied, come here,
ll»-. A sip of tea revives his tired frame.
13. Poor dear John has business on his mind.
16. A cent In the slot vlU produce a stick of gun.
17. I never thought you vould turn out that sort.
18. Be vas rigid and slim as a bean pole.
19. There are many vaya to ^ these things.
20. Tuok the sheet under the edge of the mat.
List kS
1. A bovl of rice Is free vlth chicken stev.
2. Her skirt dragged, her Jav vagged. she vas messy.
3. A sturdy crutch vlU help cure the llap.
h. Take my advice arid fait tiie road, quick.
3. Plod on, dumb herd, and ask nothing.
6. The vater of this veil Is a. source of sound health.
7. The volf yelped, and sprang Into the brush.
^ felt at a loss on hov to begin the scene.
9. An eye for a tooth Is the code of^ the heathen tribe.
10. I'm happy to learn that at last he's gone to Jail.
11. Take shelter In this tent, but keep still.
12. That guy Is the vrlter of a few bad books.
13* A mere child and yet so knovlngl
1^. The little tales they tell are false.
13. IF ^ fall, ve can at least be sportsmen.
16. The door vas barred locked and bolted as veil.
17. Ripe pears are fit for a queen's table.
18. A big round vet stain vas on the carpet.
19« The kite dipped and svayed. but stayed aloft.
20. The pleasant hoiors fly by too soon to suit me.
IC-69
P-292
Liât U9
1* The rooa vas crowded vlth a vlld aob.
2. Here, puae. came get your supper.
3« I know It Is not their fault or guilt alone.
This strong arm shall shield your honor.
She blushed when he gave her a vhlte orchid.
6. Wrap the victim in the thick wool shavl.
7. Step \xp closely now and vatch this.
Ô. Ve are six, all told, and hungry too.
9» He loves to fly a fighter ship at night.
10. The beetle droned in the hot June sun.
11. It is a shameful act to wipe your mouth on a sleeve.
12. Care and stealth will get you safely back.
13• Press the pedal vlth the left foot thus,
li»^. They fed and clothed the poor hungry valf.
15. It is far from true to say the fight is over.
16. Change the tube, and fíx the po£ dial.
17. AH you mist do is sit on the sofa and cry.
18. The cT^n^ of the chain sounded loud in the cell.
19. What shall I read next Sunday?
20. Bat this crust and a piece of aouldy cheese.
List y)
1. Heat Plans fail vithout luck.
2. "Bß at Boss, but paid court to Martha.
3. The black trunk fell from the landing.
U. Croak and groan, you vill ge-c nothing.
3. The bank pressed for paynmit of the debt.
6. Ite eeiged the thin vrlst in his iron grasp.
7. The theft of the pearl pin was kept secret.
8. She lived a fast life, ana died merry.
9. He hurled a clod in the face of the raging beast.
10 T Give your paw to this friendly child.
H. The ant hacked and chewed at the spider's leg.
12. The thick neck was in contrast wltii the gaunt face.
13. The vast space stretched into the vague distance.
li^. A rich farm is rare in this sandy waste.
15. His vide grin earned many foes.
16. S^^ act, but others talk and vatch.
17. Count the pebbles .on the beach, if you can.
18. Plug the leaky pipe with » vad of gum.
19. y lax Bmxes a fine brand of paper.
20. Hurdle the £it with the aH of a long pole.
Liât 31
1. A strong bid may acare your partner stiff.
Charley lost his nerve and screaasd.
3« If you hear a whistle, seek a safe place.
and shake, you timid mloel
3. A bed to 1st is scarce In booming tovna.
6. The quick word vas tinder to his hot rage.
7. £ven a Just cause needs power to vln.
8. Peep under the tent and see the clovae.
Between the sharp eyes was a long nose.
10. The leaf drifts along with a slow spin.
11. All authors are in the soup to their necks.
12• The hard clod dropped from the stately cliff.
13. Please stop this silly talk and act.
Ik, Debts are made to t>e paid, but rarely are.
15« All my kin are scattered here and there.
16. Plot and plan as you desire, the die is cast.
17. Cheap clothes are flashy but don't last.
18. A thing of small note can cause despair.
19. Half a. peach is good, but so are peanuts.
20. Sing a soft tune and lull my anguish.
List 32
!• Flood the mails with requests for cheap books.
2. He talked with a thin soft soothing sound.
5. A thick coat of black paint covered all.
if. The pencil vas cut to be snarp at both ends.
3* Those last words were a strong stateaasnt.
6. He wrote his name boldly at the top of the sheet.
7* Dill pickles are sour, but taste fine.
Poto that roan is the way to the grain farmer.
9« Ward bosses seek the loose votes.
10. The flashy shade of his tie vas bright in the sun.
11. Blther mud or dust are found at all times.
12. The best method is to fix it in place vlth clips.
13• If you mumble, your speech vill be lost.
1^. At night the alarm roused him from a smooth sleep.
15« Rlnee your mouth with this tart liquid.
16. A bushel of wheat is worth a child's life.
17» Give your oath that the worst la over.
18. Bead Just what tíie meter says.
19« Fill your pack with bright trinkets for the poor.
20. The email red neon lamp went out.
10-69
P-293
iâîLa
1. are »^*11 ^ round, eoft, and taety.
2« Bich olotha filiad the l<tfty ahalTae,
3* The fan vhirled its round hladee sviftly.
Hy niece la plain, so she draasea to anlt the aoda,
3« The line vhere the edges .joined vas clean,
6, Breathe deep, and saall this piny air.
7* It nattera not if he reads these vorda or those.
8. Lift the aaah, and shov the hidden face.
9« A brovn leather hag hung fron its a^?ap.
10. A toad and a frog are hard to tell apart.
11. A-iAite eilk jacket goes vlth any shoes.
18. Ask the cop to let the waif escape.
13* Stiff frozen clods vill serre as bricks.
Ik» He sped like an arrov straight to the nark.
13. Ss did not notice the roar and drone of the noise.
16. Prunes are black and vrinkled, vith soft sklna.
17* A break in the dyke alnost caused a flood.
18. Attract his eye, and nake hla see us.
19. Paint the sockets in the vail dull green.
20. The child crawled into the dense grass.
List 33
1« The bou^ snapped, and tunbled straight on his head.
2. The weight of the package was seen on the coarse paper.
3. An apt and strong serrant la what I want.
Qyaaoa la a pawn In a big and tricky game.
3« Wake and rise, and step Into the green outdoors.
and silly boys last a day.
7* The sot e»J^"*»*red and gaped at the droad sight.
8. The green light In the brown box flickered.
9. The brass tube circled the high wall.
10. The lobes of her eare were pierced to contain rings.
11. Hold the bajiwftr near the end to drive the nail.
12. Hext Sunday is the twelfth of the month.
13* Ihe headlines in the neve scream of fierce ver»
lU, ^ posed for a picture wi^ his hands on «■ clean cloth.
15. Every word and phrase he speaJcs Is stale.
16. They know more about the tough life than most.
17« 1 have orders to lead you out of this place.
18. He pat his last cartridge into his gun and fired.
19. They took their kids from the public school.
20. Drive the screw straight Into the wood.
List 3^
1. Bribes fail vhere honest threats work.
2. The gun barked twice, then quiet reigned.
3. Trample the spark, eine the flames will spread.
k» Be ready to frown and prowl to lose friends.
3. Sport is fun, but w need money.
6. This jear*B custom in hose is not eilk.
7. Mourn the loss of taste in these lurid times.
8. The hilt of the sword was carved with fine deaigns.
9. Chaste itids are last, and never first.
10. A round hole was drilled through the thin board.
11. Dumb guys fall for a snappy line.
12. Footprints showed the path he took up the beach.
13. ^ was waiting at m£ front lawn.
1^^. A vent near the edge freshened the stale air.
15. Prod the.o^ mule with a crooked stick.
16. It is a band of ateel three inches vide.
17. The boat tipped, and the fat lady screamed.
18. The pipe ran the lengüi of the ditch.
19. That stuffy hag craatpa my style a bit.
20. He was hidden from sight by a mass of leaves and shrubs.
List 56
1. Keep the hatch tight and the watch constant.
2. Sever the twine vith a quick snip of the knife.
3« ^11 the lad who tries but misaes fire.
dries out when wet.
3. Sell the copper at a good rate of cash.
6. The scalding tea slid into the deep drain.
7. Slide the catch back and open the desk.
8. Bark into the mike and get bad results.
9. The dark future held no hope.
10. Flirts and flappers wear short skirts.
11. Sneak off end the deuce with the dizzy blonde.
12. Trap the tackle by letting him through.
13. Jerk tíie knee irtien you feel him tapping.
1^. Save coins and get filthy rich.
guys spun loaded tops.
16. Help the weak to preserve their strength.
^7* A sullen smile gets few friends.
18. Shake the mud from your foot, silly.
19. See much, talk not, and stay alive.
20. Stop whistling and watch 'Uie boys march.
10-69
P-29>i
Liât 57
1. Deal the carda fraa the top. 70U bully.
2. Jerh the cord, and out tunblea the gold.
Slide the tray acroea the glaea top.
V. Heat the rod and tap It agalnat the ahln.
The cloud moved In a atately way and vas gone.
6. Light aaple neJcea a avell room.
7. There vere high .llnha on the gala day.
6. Set the piece here and aay nothing.
9. Gay ladfl make hay In the aowiBhlne.
10. Drop that notion, and ^ a eport.
11. Stale bread la a waate in pudding too.
12. ^e gray paint claahed vlth the gaudy tinta.
13. The atlff drink hit like a bolt of lightning.
IV. Sell It ahort. and loee your money.
13. Dull atorlea aaJce her laugh.
l6. The clown moulded hla pale face into a emlrk.
17* Sev and knit. It helpa evenlnga paas.
18. Man are but Infanta to a anart girl.
19. Dip the pall once and let It settle.
20. A atlff cord vlll ^ to faaten your ahoe.
1. Sveet vorda vork better than flata.
2. A thin atrlpe runa down the middle.
3. A Bix ocmea up more often than tan,
h. Suck the egg dry in thla ahortage.
3. A Turk la a man and a bj^ bird alao.
6. Luah fern grev on the lofty rook.
7. Follov ault. came the aharp voice.
6» Buga drive off city gueata.
9. A gnome danced long on the graaay knoll.
10. The ram acarad the aohool children off.
11. The team vlth the beat timing looka good.
12. Bonea are the atuff the human frame la built of.
13. The fanner avapped hla horae for a brown ox.
Sit on the porch and tell the o'ÜierB what to
1^* A steep trail la painful to lame feet.
16. The blondea played havoc with the stag line.
17. The early phaae of life moves fast.
Id. The tazl avung in and the ahot mlaeed.
19. Fill the alloa full ao the cowfe may feast.
20. Bloody clave dripped In the glocm and mist.
Uat 38
1. Get the trust fund to the bank early.
2. A tight bow la not foui¿. In vom ahoe-strlnga.
5. A mean child pulls vlngs from Insects.
i*-. There la plenty of pork, but spring lamb is short.
3. Jean Is a man trap In silk stockings.
6. Can such falsehoods be proved true?
7. Die nude men was Jailed by the angry cop.
8. She vorks in a boiled shirt factory.
9. A Tiimiher of tunes tell of June nights.
10. Men dare not scorn a dab of lipstick.
11. Choose between the high road and the low.
12. A plea for funds seems to come again.
13. He lent hla coat to the tall gaunt stranger.
Ik. There la a etrong chance It will happen more than once.
13. The line will march until they halt or drop.
16» The duke rldea the park la a ailver coach.
17. The donkeys bray as the ghoata flit la the fields.
18. As the boys fought the cops brou^t the wagon.
19. Greet the new gueata and leave quickly.
20. When the frost has come it la tima for turkey.
Uet 60
1. A ^8 leak explodes In your face.
ToBPor likes a new bride each year.
3. A draped ault fita a slim body.
k. Green moaa grows on the northern aide.
3. Tea in thin china has a aveet taste.
6. Pitch the straw throu^ the door of the stable.
7. Freah boye often have slapped cheeks.
8. The latch on the back gate needed a nail.
9. Girls are a very petty lot, he smiled.
10. Vice Is called crime's young brother.
11. It's a cold grey drab day, isn't It?
12. The goose was brought straight from the old market.
13. The elnk la the thing In which we pile dlahea.
lU, A whiff of It will cure the moat etubbom cold.
13» The facta don't always show ^o la rl^t.
16. ühe that clapped had lost the ring finger.
17* She flaps her cape as she parades the atreeta.
18. Die lose of the cruiser was a blow to the fleet.
19* A mad rage flushed her pallid face.
20. Die lump in her throat availed her slender neck.
ic-69
p-295
llBt 61
IiHt 63
braid to the left and then orer.
2. Plead vith the lavyer to drop the loot cagae.
3« Calves thrive on tender spring graaa.
Poet hllla on the first office vail.
3, Hie fuse hlev end darkneaa clairoed down.
6. Recite "Who knova a brown cow".
7* Smooth the edge of the card before you aeal it.
8. Hie Blgn pointed right, but the oar vent cm.
9* (rift dlah counters are thick with people.
10. Tear a thin sheet fron the yellow pad.
11. A alack In the line will cauae a Jerky anap.
12. A orulae In warm waters In a sleek yacht la fun.
13. A daub of color ran down the left edge,
lb. A proBpt answer la what ve don't get.
13* It was done before the hoy coiild see It.
16. The coke bottle was a cinch to shoot from the log.
17. Ihe nymph Is s myth of the devil to tenpt men.
10. (ïrouoh before you spring or mlas the mark.
19* Pack the kits and dwi't forget the salt.
20. ^e aguare peg will settle In the round hole.
1. Hie ateed balked and threw the tall rider.
2. Hilok aoles keep feet from puddlea.
3« The crowd hoota and aneara at the sly umpire,
iv. Whips and ropes fall to phase lovers.
3« The tax rides higher than a box kite.
6. Hie hitch between the horae and cart broke.
7* Pile the coal high In the ahed comer.
8. Cure or kill was the motto of that firm.
9* The nub of com hit the wagon wheel.
10. A gold vase la both rwre and costly.
11. The knife was hung Inside Its bright sheath.
12. Dirty the floor vith vet shoee. you oaf Î
13* The rarest spice comes from the far East,
li^. Hie roof should be tilted at a sharp slant.
13« A small bit of banter brings a smile.
l6. They tied the girl to the rack for a long torture.
17» The rude bore told crude stories.
10. He put his finger In the deep dent In his head.
19. A smatter of French Is worse than none.
20. Some folks dwell In an abyss all their days.
List 62
1» Twist and wriggle, poor fool, you are caught.
2. Pant the ball quick, and far away.
3« A thin skin needs o brazen front,
b. Fine soap eaves tender skins.
3. Poached eggs and tea must suffice.
6. Jever Jest, lest you be socked.
7. Bad nerves are Jangled by a door slam.
8. Ship maps are different from those for planea.
9« Ouch Is a strange word to cry In surprize.
10. Dimes showered down from all aides.
11. They sang the sane tunes at each party.
12. Hie sky in the west la tinged with orange red.
13. The waltz la » light quick dance If perfoimed well.
lU. The boor alaained the door and stalked away.
13. Peat la cut In the bogs as fuel for the poor.
16. Sow ^e seeds at wrcog and reap your reward.
17. Hie blood races through veins to the heart.
l6. Take a plunge and pray to (Sod for blessing.
19. Get the ecoop, and ¿ou eell the paper.
20. The pods of peas feimant In bare fields.
List 6k
1. Hie uniTe trod the treadmill day and night.
2. The aim of the contest Is to raise a great fund.
3* It was merely a miss, though close to a hit.
k. It's quicker to put k^Tiv« into a rope than take them out.
3. To spend it now in large amounts Is bad.
6. The bat needs no eyes to fly around In dark.
7. There Is a fine hard tang In salty air.
8. We looked long to find where the verb was written.
9* The cod is 'Uie main business of .the north shore.
10. The slab was hewn from » heavy block of slate.
Id* Bawdy Jokes shock prim folks.
12. Dunk the stale biscuits In strong drink.
13« Peep and pry longer, and your name Is mud.
1^. Tassels swing from hate, for ots thing.
13* Sheep bleat, but deer remain mute.
16. Hang tinsel from bo^ brenches.
17. Accept the dare and knock his block off.
18. Crom are solenm and fly by day.
19« Bight and might Join In the hot battle.
20. the with a tight brass cover.
IC^
U«t 63
1. The poor eap aieaed the hoat again,
2« Be aure to eat the la^ flialy in the <hole.
3« Hop to ^ and finiah the old Joh,
HI ok a card and allp It under the pack,
A round nat viU ooTar the dull apot,
6, The firat part of the plan naeda ohanglng,
7, A good book Infoma of vhat ve ought to know,
8, The Mail coaea in three batchea per daj,
9« Sveatera look irell en young glrla,
10, You cannot bree tea in a cold pot,
11, Bake in the oren or you fall to haTe a cake,
12, Dote of light betrayed the black cat.
1?, The havk la a Ticioua bird aa it aooota paat,
Ik, Jump and alide through the alaty bog,
1^. The beat of the dmaa drlTea natlTea nad.
l6. Put the chart on the nantie and tack it down.
17« <»tay akiea turn vatera to lead.
l8. Bend 'Uie lard fron the pork blubber,
19> aioot the bolt, ala, vlthhold your fire,
20, The night ahlft rate extra pay.
Liât 67
1. A thin book fita in the aide pocket.
2. The gloaa on top wade it unfit to read,
3* The houae la a neaa. but thia la atill waahday,
k. Ihe hail pattered on the burnt brown graaa.
BeTen aeala vere ataaged on great aheeta.
6. Our troopa are aet to atrike heayy blowa.
7« Where haa the anow gone vhen arrivée7
8, The atore vas jeinwKni before the sale could start.
9, It vae a bad error on the part of the nev Judge.
10, A kiaa vas wice thought a alnful act.
Btep more and the cliff vlU collapae.
Ihey grope in the dark and atumble on the dead.
PBah to the frt»t and burrov a dark trench,
Ik. When the heel leers at you, beware.
1^. She showed a mild front, but ateved beneath,
16. A velrd light flashed among the trees«
17. To singe feathers first is a good trick,
18. The girl laid her pistol down to sing,
19. The Tnina shaft occurs In a dread spot,
20. I am confused by all the noise of great crowds.
Hat 66
1. Cie red paper brlfäitened ^e din stage.
2. See the rascal scoot to third base.
Full pails aean happy lads.
k. Slide the bill between the two leaves.
Help the tipsy nan cross the street,
Mepy hands help get the Job done,
7- like to our small faults,
8. Increase the fund of coins to buy tanks,
9, Ho doubt about tíie way the wind blowa.
10. Dig deep in the earth for pirate's gold.
11. The steady drip is worse than the drenching rain.
12. You cad, ahe acreeched. and flung hla his diamonds.
^3* A flat pack takes leas luggage apace,
Ik, A hive of buay bees talked as they worked,
1^. Dii«p the Junk oulckly or get caught,
16, gvll means are nixed with good ends,
17, Gr^n ice frosted the punch bowl,
18, A stuffed chair ellpped from the moving van.
19, The stitch needs to be shortened, but will serve.
20, Begin the game and slip a card to Jin.
List 68
1. Hard knocks build tough people.
2. Take ti» match and atrike it against your shoe.
Die pot boiled, but the contents failed to Jell,
k. Sock tíie goof and tell him nothing.
3« Drop the whole affair if you want to be viae,
6, Ihe baby puta hie right foot In hla mouth.
7, The bemba left moat of the town in ruins.
8, Stop to stare at the men working hard,
9* The etreeta are narrow and full of sharp turns.
10, They froze with a glance and turned their faces,
11. The drone buzzed on but said nothing,
32, The pup Jerked the leash as he saw a feline shape,
13* They walk erect and chill to meet their «ad,
Ik, Fish evade the net and flit .
13. Form sheets are a major need for our stock,
16. Glee and grief look alike at first.
17. A gnat is food for the dark houae spider.
18. Her had a fine beauty spot.
19. Fat boye make the Jolly type,
20. Dice games are fun, but beware of tricks.
IC-69
P-297
-92-
Appendix V
E WCED LISTS
The following word lists are useful when a limited
nvimber of different words are req.uired for ahhreviated
articulation tests. They are of particular value for pre¬
liminary tests in which the listener is permitted to check
his own responses as he hears the words. The words were
selected so that they résemhle one another as much as
possible and at the same time sample the various soiuids of
speech.
In order to satisfy these requirements the 15 vowels
and diphthongs are represented by six words each. An
effort was made to distribute all the different consonant
sounds among these 90 words. The remaining 10 words of
each hundred are used to sample some of the compound con¬
sonants. When these lists are used, the serial order of
the words in each list should be changed from reading to
reading.
dBiC
OOT
iiqnj:
61
OAasx
06
q.qnop
6s
3{T37C
66
qsnj
daex
6+1
qoqip
+I3
86
81
3iJCT3t
8+1
auTp
83
2fA
16
q.ooJC
3l
Sjfof
l+T
3TP
33
eiTfiiA
96
9OJ:
Tl
^oC
9+1
papp
13
9ITTIA
66
ictaojc
01
qop
6+1
9q.n0
03
dOdA
^6
69
+T+I
qjno
61
QA-BQA
£6
qota
89
qanq
e+T
osano
91
SiCoq.
Z6
92aa
19
3+Î
pjno
It
^oq.
16
99
uoq
T+I
qjno
91
XSAoq.
06
q.nd
69
nioq
0+1
ono
61
q.noq.
68
XXtiá
+19
ujajS
68
otiejo
+ll
i[q.ooq.
88
d99d
£9
9uo2
Q£
tim
81
q.ooq.
¿8
9saqd
39
1X^3
18
uxauq
31
Suoq.
98
9A99d
19
iinj
9^
SiCoq
it
©UTq-
68
noto
09
qooj
68
iCoq
01
9TT^
^8
qo
66
eoj
+ie
TOAoq
6
uoxiq.
£8
qA9U
86
qoj
88
qnoq
Ô
BiQiiq.
38
A9U
16
raaoj
38
qqooq
1
d^aq.
18
9q.niif
96
JBJ
18
qooq
9
Iiaq.
08
qSTÏDI
66
sp^j
08
qooq
6
^ioaq.
61
jjrira
+T6
ue
63
2uoq
+f
ufaxs
8i
A9HI
86
tno
83
qoq
8
das
ii
9zxani
36
upajp
l3
jaq
3
3ioaB
91
spam
T6
XOAOp
93
ii^q
T
3 q-sxx a
d-pz
OOT
IIT3IS
61
q.stn
06
qoej
63
3u-pz
66
dxs
+îl
BBOin
6+1
pooj
+I3
q.e^
86
2uxs
81
doox
8+T
qnp
83
08iC
16
XITJt{s
3l
uoox
1+Î
auop
33
UAa^
96
pxnoqB
II
OATX
9+Î
ourôp
13
AajC
66
doqs
Ol
ops XX
6+1
opap
03
PTT^OA
■ti6
poqs
69
UAax
+Î+T
AOO
61
93I0A
86
snraB
89
Aap
8+T
ppnoo
91
90A
36
9X^S
19
map
s+T
qonoo
It
raai[A
T6
xmj:
99
qoaP
I+Í
doo
91
3toat{A
06
qnjc
69
oSnq
o+i
dooo
61
ADA
68
dxj
+I9
Aoq
68
uooo
+ll
i[onoA
88
Suxj
£9
doq
ee
Tpoo
8t
qnq.
18
ucaj
39
pooq
18
poo
31
ittjq-
98
5[oaj
19
poq
9£
Quoqo
it
uoq.
68
Bjmd
09
OAxq
68
oqoqo
OT
TToq.
+18
jnod
66
SAoq
+ï£
TToq
6
9niBq.
e8
93{0d
86
SBQ-nS
88
oSpoq
9
0Xaq.
38
BA9d
16
poo9
38
qooq
1
TTTi^B
18
UAad
96
q.oS
18
paoq
9
XITds
08
Aad
66
Qsnj
08
oqsaq
y
6
dnos
61
qono
+16
oranj
63
qjaq
+í
q.oos
81
xto
86
ITOJ
83
eSjaq
8
XIOOB
11
P99U
36
QApj
13
qjraq
3
TTOS
91
q.a9U
t6
QITJ
93
ops pa
T
i qsfi a
-£6-
100 T ♦ r«00
A B
SYSTEMS
TESTED
2 .„J
Uj 50 "
<
O
CO
o
K
I
ITEMS IN
LIST 2
SYSTEMS-
TESTEDo
a
Ï ,
■ 75 5
z
UJ
25 o
ITEMS IN
LIST 3
A B
SYSTEMS
TESTED
F I G . 3
THIS nsURE ILLUSTRATES HOW TWO COMMUNICATION SYSTEMS WOULD
8E APPRAISED BY THREE IDEALIZED LISTS OF 20 WORDS EACH. THE
DISTRIBUTION OF THE ITEMS WITH RESPECT TO THEIR INHERENT DIF¬
FICULTY IS INDICATED FOR EACH LIST. EACH SYSTEM IS REPRESENT¬
ED BY A BAR WHOSE HEIGHT, AGAINST THE SCALE OF DIFFICULTY,
SHOWS WHAT ITEMS THE SYSTEM IS ABLE TO TRANSMIT INTELLIGIBLY.
THE RESULTING ARTICULATION SCORE IS SHOWN ABOVE EACH BAR.
3 4 5 6
AVERAGE NUMBER OF SOUNDS PER WORD
F I G . t
FIVE GROUPS OF 20 WORDS, HAVING DIFFERENT AVER¬
AGE NUMBERS OF SOUNDS PER WORD, WERE READ 20
TIMES TO A GROUP OF LISTENERS. THE GRAPH SHOWS
HOW ARTICULATION IMPROVES WHEN THE NUMBER OF
SOUNDS PER WORD IS INCREASED.
PERCENT WORD ARTICULATION
F I G . 2
LISTS OF WORDS AND LISTS OF SENTENCES WERE COM¬
PARED AS TO INTELLIGIBILITY UNDER A WIDE VARIETY
OF CONDITIONS. EACH POINT ON THE GRAPH REPRE¬
SENTS ONE TEST OF 100 WORDS AND ONE TEST OF 50
SENTENCES. THE CURVE PASSING THROUGH THE DATA
MHR-78 DERIVED,FROM RESULTS OBTAINED AT THE BELL
P- 431
TELEPHONE LABORATORIES.