A comparison between phonetically balanced wordlists and a phonetically balanced nonsense sylable
list in the measurement of speech intelligibility
Item Type text; Thesis-Reproduction (electronic)
Authors Larsen, William Guy, 1930-
Publisher The University of Arizona.
Rights Copyright © is held by the author. Digital access to this materialis made possible by the University Libraries, University of Arizona.Further transmission, reproduction or presentation (such aspublic display or performance) of protected items is prohibitedexcept with permission of the author.
Download date 26/05/2018 03:59:17
Link to Item http://hdl.handle.net/10150/318849
A COMPARISON BETWEEN PHONETICALLY BALANCED WORD LISTS AND A PHONETICALLY BALANCED NONSENSE SYLLABLE LIST
IN THE MEASUREMENT OF SPEECH INTELLIGIBILITY
, by ; ;, .
W il l iam G. Larsen
A Thesis Submitted to the Faculty o f the
DEPARTMENT OF SPEECH
In P a r t ia l F u l f i l lm e n t o f the Requirements For the Degree o f
MASTER OF ARTS
In the Graduate College
THE UNIVERSITY OF ARIZONA
1961
STATEMENT BY AUTHOR
This thesis has been submitted in p a r t i a l f u l f i l l m e n t of requirements for an advanced degree at The Un ivers i ty o f Arizona and is deposited in The U n ivers i ty L ib ra ry to be made a v a i la b le to borrowers under rules of the L ib ra ry .
B r ie f quotations from th is thesis are a l lowable w i t h out special permission, provided tha t accurate acknowledgment of source is made. Requests for permission for extended quota t io n from or reproduction of th is manuscript in whole or in part may be granted by the head o f the major department or the Dean o f the Graduate College when in t h e i r judgment the proposed use o f the mater ia l is in the in te res ts o f scholarship. In a l l other instances, however, permission must be obtained from the author.
APPROVAL BY THESIS DIRECTOR
below:This thesis has been approved on the date shown
GENE ENGLAND, Vf\. D. Assistant Professor o f Speech
M L
ACKNOWLEDGMENTS
The wrI t e r wishes to express his g r a t i tu d e to Gene
England, D irec to r o f the U n iv ers i ty o f Arizona Speech and
Hearing C l i n i c , and to Klonda Lynn, Speech Department Head,
fo r t h e i r guidance and encouragement during the w r i t in g of
th is thes is .
The w r i t e r a lso wishes to thank Kenneth Dimmick and
Henry Schmitz for technical advice in the study.
Appreciat ion for t h e i r cooperation in serving as sub
je c ts for the study Is extended to Vicki B e l l , Jim Larrabee,
Max Huss, Gail J a f f e , Mary Rule, Jacqueline Brown, Gerald
Si W a r , Sandra Focht, Jim Burkhardt, and Tom Economidis.
TABLE OF CONTENTS
Page
LIST OF TABLES v
Chapter
I. INTRODUCTION AND BACKGROUND. . . . . . . . . . . 1
Introduction . . . . . . . . . . . . . . . . . 1Background o f the Study. ...................... 2
Psycho Acoustic Laboratory Auditory Tests. . 3Central I n s t i t u t e for the Deaf Auditory
Test W-1 ...................... 3Central I n s t i t u t e for the Deaf Auditory
Test W-22. . . . . . . . . . . . . . . . . 4Test o f Phonemic D i f f e r e n t i a t i o n - The
Rhyme Test . . . . . . . . . . . ................... 5M u lt ip le -C h o ice I n t e l l i g i b i l i t y Tests. . . . 5CNC Word Tests . . . . . . . . . . 6Newspaper Word Test . . . . . . . . . . . . . 7T h re e -D ig i t Test . ....................... . . . . . . . 7In te r ro g a t iv e or D e c la ra t iv e Sentence Test . 7S y l la b le L is t Tests. ....................... 8
I I . STATEMENT OF THE PROBLEM . ................................ 11
Importance o f the Study. . . . . . . . . . . . 11D e f in i t io n o f Terms. . . . . . . ................... » . 12
I I I . PROCEDURE. . . . . . . . . . . . . . . . . . . . 14
Construction o f the T e s ts ...................... 14Select ion o f Subjects. . . . . . . . . . . . . 18Speech Reception Threshold . . . . . . . . . . 19Presentation o f Tests. . . . . . . . . . . . . 20
IV. PRESENTATION OF DATA . . . . . . . . . . . . . . 23
V. SUMMARY AND CONCLUSIONS. . . . . . . . . . . . . 31
Summary. . . . . . . . . . . . . . . . . . . . 31Conclusions. . . . . . . . . . . . . . . . . . 32
APPENDIX: PHONETICALLY BALANCED NONSENSE SYLLABLE TESTS 34
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . 39
iv
L I ST OF TABLES
Table Page
1. The Frequency of Occurrence o f Vowels in W-22Word L is ts . . ....................... . ................................ 15
2. The Frequency o f Occurrence o f Consonants in W-22Word L is ts . . . . . . . . . . . . ................... . . 16
3. Phonet ica l ly Balanced Word Test (PB) andP h onet ica l ly Balanced Nonsense S y l la b le Test(PBNS) Sequence and S ig na l - to -N o is e Ratios . . . 22
4. Ph on et ica l ly Balanced Word (PB) and Phon et ica l lyBalanced Nonsense S y l la b le (PBNS) Test Scores in Percentages o f I n t e l l i g i b i l i t y and S ig n a l - to -Noise Ratios - I n i t i a l Tests. . . . . . . . . 24
5. P h onet ica l ly Balanced Word (PB) and Phon et ica l lyBalanced Nonsense S y l la b le (PBNS) Test Scores in Percentages o f I n t e l l i g i b i l i t y and S ig n a l - to -Noise Ratios - Retests. . . . . ............................ 25
6. Means and Standard Deviations o f I n i t i a l Tests andRetests for Phon et ica l ly Balanced Word (PB) and P hon et ica l ly Balanced Nonsense S y l la b le (PBNS)T e s ts ................................ 26
7. Results o f T Tests for D if ferences Between Meansand F Tests for D if ferences Between Standard Deviations in Comparing Phonetica l ly Balanced Word Tests w ith Phon et ica l ly Balanced Nonsense S y l la b le Tests . . . . . . . . . . . ....................... 28
v
CHAPTER 1
INTRODUCTION AND BACKGROUND
Introduct ion
The transference o f thoughts from one individual to
another by means of speech sounds is a complicated process.
An idea ar ises in the mind of a man, who, w ith the a id of an
inc re d ib le mechanism o f i n t r i c a t e l y designed nerves, muscles,
a r t i c u l a t o r s , and resonators, t ra n s la te s the thought into the
sounds o f speech. These sounds are transm itted through the
a i r , or through various types o f communication systems such
as telephones, magnetic recordings, radio or t e le v is io n , u n t i l
they reach another complicated mechanism, a human ear. The
process of thought t ransference can be ha lted here, however,
unless the sounds make sense to the hearer, unless he compre
hends t h e i r meaning, unless the speech sounds are i n t e l l i g i b l e
to him.
Communication among men, the interchange o f thoughts
and ideas, is la rg e ly dependent upon man's a b i l i t y to recog
n ize and comprehend the sounds o f speech. Hearing a 1 one is
not enough; he must understand what he hears i f communication
is to be accomplished. This comprehension o f the meaning o f
speech is re fe r red to as speech i n t e l l i g i b i l i t y .
This i n t e l l i g i b i l i t y as well as a l l o f the other
factors involved in the o r ig in a t io n , the transference , and
1
2
the receiv ing o f speech are o f primary concern in the f ie ld s
of speech and hearing. A person's a b i l i t y to hear and to
understand the sounds o f speech has been a top ic o f much exper
imentation and research. The c h a r a c te r is t ic s o f the human
hearing mechanism have been studied in i n f i n i t e d e t a i l . Hear
ing aids have been examined to determine the acoustica l prop
e r t i e s necessary for adequate reception.
An important phase o f th is experimentation has been
the construct ion o f various tes ts designed to measure a per
son's a b i l i t y to comprehend the sounds of speech and to evalu
ate the transmission and receiv ing c h a r a c te r is t ic s o f communi
cation systems. These tes ts are known as I n t e l l i g i b i l i t y
Tests. The numerical re s u l t obtained by an ind iv idual taking
th is t e s t is defined as his I n t e l l i g i b i l i t y or A r t ic u la t io n
Score. The stimulus m ate r ia ls used have been words, sentences>
and s y l la b le s which have been presented under d iverse condi
t io n s . A descr ip t ion of some of these tes ts w i l l be presented
as background in form at ion , which w i l l a s s is t the reader in a
c le a re r understanding o f the proposed hypothesis: th a t a te s t
composed of nonsense s y l la b le s could be favorably compared
with a te s t composed o f meaningful words.
Background o f the Study
I n t e l l i g i b i l i t y te s t in g is a r e l a t i v e l y new f i e l d ,
having i ts beginnings in the e a r ly 1900 's. As has been noted,
the stimulus m ate r ia ls used have been e i th e r words or s y l la b le s .
The m a jo r i ty o f the te s ts , however, have used meaningful words
as the s t im u l i . Several tes ts of th is type w i l l be described
in the fo l low ing paragraphs.
Psycho Acoustic Laboratory Auditory T e s ts . These
tes ts were developed at Harvard U n iv e rs i ty and are re fe rred
to as the PAL or Harvard Tests. The stimulus m ater ia l is com
posed o f tw o -s y l la b le words, such as doorway, f o o t s t o o l , and
a ? rp la ne . PAL Tests No. 9 and No. 12 were recorded at Harvard
and provided for a quick and r e l i a b l e measure o f the threshold
o f i n t e l l i g i b i l i t y and i ts re la te d c l i n i c a l measure, the hear
ing loss for Speech (7 :3 2 1 ) .
Central i n s t i t u t e fo r the Deaf Auditory Test W - l . This
t e s t was developed at the Central I n s t i t u t e for the Deaf and
was constructed from the words used in PAL Tests No. 9 and
No. 14 (7 :3 2 2 ) . This o r ig in a l group o f e ig h ty - fo u r words was
presented to a group o f s ix l is te n e rs at s ix d i f f e r e n t inten
s i t y le v e ls . Data were c o l lec ted in terms o f the number of
errors per word for each l i s t e n e r . From these data the judges
rated the words as easy or d i f f i c u l t . An easy word was defined
as one missed no more than once by a l l s ix l is te n e rs ; a d i f f i
c u l t word was one missed f i v e or more times by a l l subjects.
Words f a l l i n g into e i th e r o f these extreme categories were
e l im inated . T h i r t y - s i x words were reta ined from the o r ig in a l
e ig h ty - fo u r a f t e r th is analysis was made. Six d i f f e r e n t word
orders o f the same t h i r t y - s i x words were put together to make
up L is t A through F of C. I .D. Auditory Test W-1.
4
Central I n s t i t u t e fo r the Deaf Auditory Test W-22. This
te s t consists o f a vocabulary o f two hundred monosyllabic words
divided into l i s t s o f f i f t y words each. Each l i s t is said to
be p h on et ica l ly balanced. Hirsh (7 :328 ) s ta tes th a t a l i s t is
ph on et ica l ly balanced i f the speech sounds with in the l i s t
occur with the same r e l a t i v e frequency as they do in a rep re
s en ta t ive sample o f English speech. The plan for phonetic b a l
ance fo r t e s t W-22 was based on the analysis o f spoken English
by French, C a r t e r , and Koenig (5 ) and the ana lys is o f w r i t te n
English by Dewey ( 2 ) . The f i r s t step in s e t t in g up the plan
for phonetic balance was to decide on percentages for the v a r i
ous con sonant- vowe1 arrangements in the words. This decision* :
was based on the ana lys is of the s y l l a b le types in the study
by French, C ar te r , and Koenig w ith the fo l low ing m odif ica t ions:
a l l vowel words were omitted, the percentage o f consonant-
vowel-consonant words was increased s l i g h t l y , and the percent
age o f words contain ing consonant compounds, such as the " t r "
in t rue or the “ i t 11 in f e l t , was increased by 7 .3 per cent.
The f in a l step was to decide on the d is t r ib u t io n o f
the vowels and consonants w i th in each l i s t . The mean o f the
percentages given by Dewey, and by French, C a r te r , and Koenig
fo r the frequency o f occurrence o f each vowel and consonant
was followed c lo s e ly . The sequence o f the words in the o r i g
inal four l i s t s was changed a t random times, thus producing
a to ta l o f twenty-four l i s t s which comprise C . I .D . Test W-22.
5
Test o f Phonemic D i f f e r e n t i a t i o n - The Rhyme T e s t . The
Rhyme Test was constructed by Grant Fairbanks at the Speech
Research Laboratory, U n iv ers i ty o f I l l i n o i s . Fairbanks (3)
re fe rs to th is as a completion type te s t . The vocabulary was
drawn from two hundred and f i f t y common monosyllables of the
Thorndike and Lorge 1 is t in g in The Teachers Word Book o f
30,000 Words (1 1 ) . The two hundred and f i f t y stimulus words
were among the nine thousand most f requent ly used words of the
English language. The stimulus words also involved eighteen
consonant phonemes which accounted for approximately 90 per
cent o f a l l consonant occurrences in the English language,
according to the data by French, C a r te r , and Koenig (5 ) ,
The Rhyme Test vocabulary consists o f f i f t y sets o f
f i v e rhyming words each. The f i v e rhyming words w ith in each
set are spel led a l i k e in the rhyming port ion (stem) and d i f f e r
in the i n i t i a l consonant s p e l l in g by a s ing le l e t t e r . The f o l
lowing is an example o f one rhyming set: hot, g o t , n o t , p o t ,
l o t . The te s t forms presented to the subject show the f i f t y
stems in order o f st imulus, each preceded by a space in which
he is required to enter one l e t t e r to complete the s p e l l in g
of the word. I t is estimated tha t the ty p ic a l stem o f fe rs
the average adu lt a choice o f e igh t or nine a l t e r n a t i v e s .
■Mult ip le-Choice, i n t e l l i g i b i l i t y T e s ts . J. W. Black o f
Ohio S ta te U n iv e rs i ty is responsible for compiling two forms o f
m u lt ip le -c h o ic e t e s t s . The vocabulary for th is te s t was taken
from Thorndike data (1 1 ) . The f i r s t to the ten-thousandth
6
most f requent ly used words were screened to remove proper
nouns, homonyms (such as bore and boar) , and homographs (such
as bow ( t i e ) and bow ( to bend)) . The remaining words were
recorded and played to a panel o f e igh t to twelve l i s te n e r s ,
in both qu ie t and noisy condit ions. The c r i t e r i a for se le c
t ion of words to be used in the m u l t ip le -c h o ic e te s t l i s t s
were: (a ) a word must have f a l l e n w i th in the range 15-85 per
cent i n t e l l i g i b i l i t y in both quiet and noise and (b) a word
must have been responded to erroneously with at leas t three
incorrect words a minimum o f ten times in noise and f i v e times
in quie t (1 :2 1 5 ) . The remaining words were arranged into a
to ta l o f twenty - four m u lt ip le -c h o ic e te s t l i s t s .
The te s t forms for these l i s t s are composed o f sets
of four words. The reader speaks one of the four words in
the set , and the subject marks the word in the set which he
thinks he has heard.
CMC Word T es ts . The consonant, nucleus s y l la b le , con
sonant t e s t is a r e l a t i v e l y new i n t e l l i g i b i l i t y t e s t , devised
by Use Lehiste and Gordon Peterson a t the Speech Research
Laboratory o f the U n ivers i ty o f Michigan. The term "nucleus
s y l l a b l e , 66 used in the t i t l e o f th is t e s t , simply re fers to
a vowel. The se lec t ion o f words for th is t e s t was also made
from the Thorndike and Lorge data (1 1 ) . A to t a l o f 1,263
monosyllabic words were analyzed and compared w i th the o r ig in a l
PB Word L is t fo r frequency of occurrence o f vowels and conso
nants, A l l words chosen for th is l i s t met the requirement o f
7
consonant-vowel“consonant construct ion . The complete CMC
Test is composed of ten l i s t s o f f i f t y words each, represent
ing a se le c t io n o f f i v e hundred words from the o r ig in a l 1,263.
Newspaper Word T e s t . Stimulus words fo r th is te s t
were selected by taking the f i r s t word from every t h i r d l i n e
of a newspaper column (4 :8 5 0 ) . A l l proper names and the f o l
lowing f req uent ly used words were excluded: the , o f , and, t o ,
_a, j a . The l i s t s , on the average, contain 46 .3 per cent one-
s y l la b le , 29 per cent two-sy11 ab le , 16.8 per cent three-
s y l la b le , 6 .4 per cent f o u r - s y l l a b le , and 1.5 per cent f i v e -
s y l la b le words, with an average number o f two s y l la b le s per
word.
T h re e -D ig i t T e s t . F le tcher and Steinberg (4:851)
s ta te tha t th is te s t is composed o f numbers: 1, 2, 3, 4, 5,
6, and 8. The numbers are combined a t random into groups o f
three and presented in the manner of an ordinary a r t i c u l a t i o n
t e s t . The d is t ingu ish ing c h a r a c t e r is t i c o f each number is a
vowel sound, and the numbers are in te rp re ted , p r im a r i ly , by
recognizing tha t vowel.
in te r ro g a t iv e or D e c la ra t iv e Sentence T e s t . The sen
tences used in th is t e s t are l im i te d to those which contain
simple ideas or questions. They are designed to te s t the
observer 's acuteness o f perception ra ther than his i n t e l l i
gence (4 :8 4 8 ) . The subjects are asked to rep ly o r a l l y to the
questions which are presented in a normal conversational man
ner. Due to memory e f f e c t s , a set o f sentences can be used
with the same subjects only a very few times.
8
In a l l o f the previously described t e s t s , meaningful
words were used as the stimulus m a te r ia l . Addit ional tes ts
have been constructed using nonsense s y l la b le s in place o f
words. A descr ip t ion o f three such tests fo l low s .
S y l la b le L is t T es ts . G. A. Campbel1 made use of
s y l la b le s in te s t in g communication systems as e a r ly as 1910
( 4 :8 0 8 ) . These s y l la b le s were o f the nonsense type and,
th e re fo re , had no meaning. The s y l la b le s were constructed
by combining various i n i t i a l consonants with the vowel / i / ,
such as / I i / or / z i / . The consonant a r t i c u l a t i o n score was
used as the measure for th is t e s t . This measure is obtained
by counting the number of consonants correct 1y perceived by
the subjects being tested.
I. B. Crandall a lso devised an i n t e l l i g i b i l i t y te s t
using s y l la b le s as the stimulus m ater ia l (4 :8 0 8 ) . The s y l
lables were o f the vowel-consonant and consonant-vowel type.
All o f the common vowels were used and the combinations were
formed in ways which are usual ly found in w r i t t e n speech. The
sounds occurred with the same frequency as they occur in o r d i
nary w r i t te n m a te r ia l . The score obtained on th is te s t was
also ca lcu la ted by considering only the number o f consonants
correct 1y perceived.
H. F letcher and J. C. Steinberg (4 :810 ) describe a
standard i n t e l l i g i b i l i t y te s t which makes use o f consonant-
vowel-consonant s y l la b le s .
9y
They considered the In te rna t iona l Phonetic Association
basic alphabet too complex for te s t in g and so they created a
revised s c i e n t i f i c alphabet of f o r t y - e ig h t simple sounds,
twenty-four consonant, nineteen vowels, and f i v e diphthongs.
Using th is alphabet as the basis , they formed twenty-two simple
s y l la b le s o f the consonant-vowel-consonant type.
In order to make th is type o f s y l l a b le , i t is neces
sary to have an equal number of both vowels and consonants,
where each consonant is used in both the i n i t i a l and the f in a l
p o s it io n . However, some consonants can be used in only one or
the other p o s i t io n . With th is in mind, a l i s t was developed
which used a l l o f the consonants in both i n i t i a l and f in a l
posit ion except Ji, w, and %, which were used only in the f i r s t
po s i t io n , and zh, nq, and St., which were used only in the
1a t t e r .
The te s t in g s y l la b le s were formed by placing the i n i
t i a l consonants, vowels, and f in a l consonants each in a sepa
ra te box. A card was drawn from each to produce a s y l la b le ,
thus forming twenty-two s y l la b le s . This process, repeated
three t imes, produces s i x t y - s i x s y l la b le s in which the i n i t i a l
consonant appears three t im es, each vowel s ix t imes, and the
f in a l consonant three times.
In order to simulate actual speech, the sy l la b le s were
introduced by short introductory phrases, such as: "The f i r s t
group is . . . ” or " I w i l l now say. . . ."
10
This te s t is administered by supplying blank a r t i c u
la t io n te s t record sheets to the subjects and ins truc t ing
them to w r i t e the sounds which they heard in the appropr ia te
columns. A s y l l a b le is considered to be in c o rre c t ly observed
i f any one o f the speech sounds which i t contains is mistaken.
The percentage o f the. to ta l number of spoken s y l la b le s which
are c o r re c t ly observed is c a l le d the s y l la b le a r t i c u la t io n .
Further analysis o f resu lts may be obtained by determining the
number o f times ind iv idual sounds are observed c o r re c t ly . The
resu lts are thus expressed as " in d iv id u a l sound a r t i c u l a t i o n . "
The terms "consonant a r t i c u la t i o n " or "vowel a r t i c u la t io n " may
also be used to re fe r to the percentages o f the to ta l number
of spoken consonant or vowel sounds which are correc t 1y
observed.
The foregoing described tes ts i l l u s t r a t e tha t the
measurement of speech i n t e l l i g i b i l i t y can be accomplished by
the use of a v a r ie t y o f te s t in g m ate r ia ls , including e i th e r
words, sentences, or s y l la b le s . An analys is o f these tests
indicates that i t may be possible to construct a nonsense
s y l la b le l i s t which w i l l measure i n t e l l i g i b i l i t y with grea ter
accuracy and r e l i a b i l i t y . Chapter I I w i l l discuss th is
p o s s ib i l i t y .
CHARTER I I
STATEMENT OF THE PROBLEM
The present invest iga t ion is designed to te s t the
hypothesis tha t p h on et ica l ly balanced nonsense s y l l a b le l i s t s
are as r e l i a b l e and v a l id for te s t in g i n t e l l i g i b i l i t y as the
ph on et ica l ly balanced W-22 word l i s t s , when presented to a
group o f sophis t ica ted l is te n e rs . The study a lso seeks to
point out tha t to ta l perception from minimal cues is e l i m i
nated when te s t in g w ith these s y l l a b le l i s t s . There is nothing
f a m i l i a r to f a c i l i t a t e the perception as might be the case i f
words were used instead o f nonsense s y l la b le s .
Importance o f the Study
One needs only to ta b u la te the various devices that
are used for t ra n sm it t in g speech to r e a l i z e the importance o f
a q u a n t i ta t iv e method o f ra t ing t h e i r performance. The te s t in g
m ater ia ls used fo r th is measurement were discussed in the
in troductory chapter. The work o f Fletcher and Steinberg
seems e s p e c ia l ly p e r t in e n t to th is study since i t was based
upon nonsense s y l l a b le stimulus m a te r ia l . The te s t ing mate
r i a l s composed for the present study d i f f e r from those of the
Fletcher and Steinberg l i s t s in th a t they are ph on et ica l ly
balanced and meet the requirements set up for the C .L D . W-22
l i s t s .
„
12
Nonsense s y l la b le s are o f p a r t ic u la r value in te s t in g
for i n t e l l i g i b i l i t y because they have no meaning, and one s y l
la b le cannot be recognized more f req uent ly than another simply
because the l i s te n e r is more f a m i l i a r with i t . The perception
of one phoneme in a s y l l a b le does not give a c lue to the other
phonemes in the same s y l la b le . The nonsense s y l l a b le te s t
thus adequately removes expectancy as an aid in id e n t i fy in g
the sounds heard. Emphasis is place on the exact hearing o f
each ind iv idual phoneme, thus increasing the s e n s i t i v i t y o f
the te s t .
D e f in i t io n o f Terms
Phonetica l ly Balanced L i s t . Hirsh (7 :328 ) states
tha t l$a l i s t is p h o n et ica l ly balanced i f the speech sounds
w ith in the l i s t occur w ith the same r e l a t i v e frequency as they
do in a rep resen ta t ive sample o f English s p e e c h . "
Speech I n t e l l i g i b i l i t y . For the purposes o f th is
study, speech i n t e l l i g i b i l i t y is defined as the a b i l i t y to
perceive and comprehend the sounds o f speech.
A r t i c u l a t i o n . A r t ic u la t io n is the percentage of the
to ta l number o f spoken s y l la b le s or words which are correct 1y
perceived and recorded by the subject. This term is synony
mous with i n t e l l i g i b i l i t y score.
Nonsense S y l l a b l e . Any combination o f consonant-
vowel or consonant-vowel-consonant which does not convey mean
ing may be c a l le d a nonsense s y l la b le .
13
Phoneme. While a d e f i n i t i v e discussion o f the term
phoneme is not possible or even necessary for th is study, we
can r e fe r to Kantner and West's (8 :1 8 ) statement tha t "perhaps
the term 1 sound fa m i ly ' is the most adequate simple in te rp re
ta t io n o f the term phoneme; i t implies a group o f sounds that
are c lose ly re la te d to each other in some way yet by no means
iden t ica l , et The reader can, for purposes o f c la r i t y , th ink
of the phoneme as a "speech sound" in the discussion to fo l low .
Speech Reception Threshold. Newby (10:110) states
tha t a s u b je c t 's speech reception threshold is tha t point
( in t e n s i ty le v e l ) a t which the subject can repeat simple words
or can understand simple connected speech.
The hypothesis under consideration in th is study neces
s i ta te d construction o f l i s t s o f p h o n et ica l ly balanced non
sense s y l la b le s . The method o f construction w i l l be discussed
in the fo l lowing chapter.
CHAPTER 11 I
PROCEDURE
The purpose o f th is study was to construct an i n t e l l i
g i b i l i t y t e s t , using nonsense s y l la b le s as the stimulus mate
r i a l , which could be favorably compared w ith C .L D . W-22
Tests la , 2a, 3a, 4a, and lb w ith respect to r e l i a b i l i t y and
v a l i d i t y .
Construction o f the Tests
The f i r s t task in construct ing the nonsense s y l la b le
te s ts was to make a c r i t i c a l analysis o f the exact frequency
o f occurrence o f each phoneme in the selected W-22 word l i s t s .
Each word was f i r s t t ranscr ibed , using in te rn a t io n a l Phonetic
Association symbols. A Pronouncing D ic t ionary o f American
English by Kenyon and Knott (9 ) was used as a source for
checking the accuracy o f the t ra n s c r ip t io n s . The occurrence
of each phoneme was then ca lcu la ted and entered on a ta b le ,
dupl ica t ions o f which are presented in Tables 1 and 2. Cor
responding nonsense s y l la b le s were formed in the fo l lowing
manner: The f i f t y vowels were f i r s t l i s te d in order of occur
rence, and the consonants were then placed, in an a r b i t r a r y
manner, before and a f t e r the vowel sounds u n t i l the exact num
ber had been used. The s y l la b le s were then randomized in
order to have no sequence of vowels (see Appendix). The
14
/15
TABLE 1
THE FREQUENCY OF OCCURRENCE OF VOWELS IN W-22 WORD LISTS
PB la PB 2a PB 3a PB 4a PB lb
A 4 4 4 3 4
I 7 6 8 8 6
e 6 7 6 6 6
ai 3 3 3 3 3
ae 5 5 5 5 5
a 4 4 5 5 4
i 5 5 3 3 5
u 1 3 2 3 2
e i / e 4 3 4 4 4
ou/o 3 3 3 3 3
0 2 2 1 2 2
u 2 0 2 2 2
arr 1 0 1 1 1
3" 1 1 0 0 1
ju 2 4 2 1 2
OI 0 0 1 1 0
Total 50 50 50 50 50
16
TABLE 2
THE FREQUENCY OF OCCURRENCE OF CONSONANTS IN W-22 WORD LISTS
PB la PB 2a PB 3a PB 4a PB lb
t 1 1 11 11 1 1 1 1
n 10 10 10 10 10
r 8 8 9 9 8
s 6 6 7 6 6
d 6 6 6 6 6
1 6 6 6 6 6
m 5 5 5 5 5
k 4 4 4 4 4
6 2 3 3 2 3
w 3 4 3 2 3
z 4 4 3 4 4
f 2 1 2 1 2
j 1 1 1 2 1
h 3 3 3 3 3
V 3 3 2 3 3
b 2 2 2 2 2
P 2 2 1 1 2
18
completed l i s t was audited by two judges who had had advanced
t ra in in g in phonetic t ra n s c r ip t io n . Any s y l l a b le which con
veyed any degree o f meaning to e i th e r o f the judges was w i t h
drawn from the 1 Ls_t and another meaningless s y l l a b l e was
used as a replacement. This e n t i r e process was repeated for
each o f the f i v e l i s t s .
In order to prepare them for presenta t io n , the f i v e
l i s t s o f p h on et ica l ly balanced nonsense s y l la b le s , together
with the C . i .D . W-22 l i s t s la , 2a, 3a, 4a, and lb were recorded
on magnetic tape, using a Bel 1 and Howell Model Number 300M
tape recorder equipped with a VU m eter . The order o f sequence
was var ied by recording the l i s t s in two d i f f e r e n t ways. In
the f i r s t sequence the word l i s t s preceded the nonsense s y l
la b le l i s t s ; in the second, the nonsense s y l l a b le l i s t s p re
ceded the word l i s t s . Since F le tcher (4 :812) found tha t s y l
la b le a r t i c u l a t i o n scores were somewhat higher when introduc
tory sentences were used, the phrases “W rite the word" or
"Write the s y l la b le " preceded each stimulus on the recorded
tapes.
Select ion o f Subjects
The c r i t e r i a for the se lec t ion of subjects were:
(a) the subject must have had a pure tone threshold of 0 db.
in each ear for a l l frequencies from 128 c .p .s . to 8000 c . p . s . ,
(b) the subject must have had a speech reception threshold in
each ear o f 0 to 10 d b . , and (c ) the subject must have demon
s tra ted p ro f ic iency in accurate ly t ra n sc r ib in g nonsense s y l
lables into phonetic symbols.
19
Pure tone thresholds were obtained by standard audio
m etr ic procedures as o u t l in ed by Newby (1 0 :7 2 -8 0 ) , using a
Model MA 28 Maico audiometer. These audiometric examinations
were conducted in a soundproof room with ambient noise level
of 35 db.
Speech Reception Threshold
To determine,Speech Reception Threshold, the subjects
were f i t t e d with Permoflux Model MA 28 earphones and placed
in a soundproof room with ambient noise level o f 35 db. The
s t im u l i for Speech Reception Thresholds were presented by l i v e
voice and were composed o f spondees o f Auditory L is t W-1. The
in te n s i ty was regulated by a set o f a t tenuators which made
possible a t tenua t ion over a range o f 110 db. , in one db. steps.
The sensation level a t which the subject was able to repeat
f i f t y per cent of the words was reg is te red as his Speech
Reception Threshold.
The subjects ' a b i l i t y to use phonetic symbols was
ascerta ined by req u ir ing them to t ra n scr ib e a short l i s t o f
nonsense s y l la b le s which were presented o r a l l y . No subject
was chosen who missed more than two o f the tw e n ty - f iv e pre
sented s y l la b le s . This te s t in g procedure conveniently served
as a t ra in in g period fo r the t ra n s c r ip t io n o f nonsense s y l
la b le s .
Ten subjects were se lected , f i v e male and f i v e female,
ranging in ages from twenty to tw e n ty - f iv e years. Each o f the
subjects had had a minimum o f one course o f phonetics at the
co l lege le v e l .
20
Presentation o f Tests
The f in a l experimental te s t in g procedures were con
ducted by placing the subjects, in d iv id u a l ly , in a soundproof
room and f i t t i n g them with earphones. Each subject was then
instructed to w r i t e the words o f the W-22 1 is ts and to t r a n
scr ibe the s y l la b le s o f the nonsense s y l l a b le l i s t s into pho
n e t ic symbols. The word l i s t s and the s y l l a b le l i s t s were
presented separa te ly . Subjects were instructed to guess a t
the sounds even though the s y l l a b le was not e n t i r e l y c lear to
them. For the purpose of th is study, the guessing was espe
c i a l l y p e r t in e n t in the perception of the nonsense s y l la b le s .
The stimulus m ater ia l was presented through a console
u n it composed o f a Vi king Model 75 Tape Playback, a two channel
Hewlett Packard 350A a t tenuator system, a two channel VU m eter ,
a Grason S tadler Model 456A noise generator , and two Bogen
Model LOM a m p l i f ie r s .
The in te n s i ty level for speech was a r b i t r a r i l y set a t
60 db. The in te n s i ty o f the w hite noise was a r b i t r a r i l y set
at f i v e d i f f e r e n t le v e ls : 42db., 54 d b . , 60 d b . , 66 d b . ,
and 72 db.
The speech was fed through one channel, w h i le the
noise was fed, simultaneously , through another. In te n s i t ie s
were c o n tro l le d by the two sets o f a t tenuators and VU meters.
Each word l i s t , together w ith i ts corresponding nonsense s y l
la b le l i s t , was presented a t a d i f f e r e n t s ig n a l - to -n o is e r a t i o .
21
Each subject heard a to ta l of. f i v e word l i s t s and
f i v e nonsense s y l l a b le l i s t s a t the f i v e d i f f e r e n t s ig n a l - t o -
noise r a t io s , as charted on Table 3.
A re te s t s i tu a t io n was set up to insure r e l i a b i l i t y
o f t e s t re s u l ts . The same ten subjects heard the same ten
l i s t s under the same circumstances, a t the same s ig n a l - t o -
noise r a t io s , w i th in f o r t y - e ig h t hours a f t e r the i n i t i a l te s ts .
The preceding paragraphs have described the procedures
followed in construction and in presentat ion o f the te s t mate
r i a l . The experimental t e s t re s u l ts and the s t a t i s t i c a l pro
cedures followed in the analysis o f the data w i l l be presented
in Chapter IV.
22
TABLE 3
PHONETICALLY BALANCED WORD TEST (PB) AND PHONETICALLY BALANCED NONSENSE SYLLABLE TEST (PBNS) SEQUENCE
AND S IGNAL-T O-NOISE RATIOS
Subj ect T estIn te n s i ty Levels
Speech No i se S/N Rat io
Subjects 1, 3, 5, 7, 9 PBNS la 60 42 + 18
PB la 60 42 + 18
PBNS 2a 60 54 + 6
PB 2a 60 54 + 6
PBNS 3a 60 60 0
PB 3a 60 60 0
PBNS 4a 60 66 6
PB 4a 60 66 6
PBNS lb 60 72 - 12
PB lb 60 72 - 12
Subjects 2, 4, 6, 8, 10 PB la 60 42 + 18
PBNS la 60 42 + 18
PB 2a 60 54 + 6
PBNS 2a 60 54 + 6
PB 3a 60 60 0
PBNS 3a 60 60 0
PB 4a 60 66 6
PBNS 4a 60 66 6
PB lb 60 72 - 12
PBNS lb 60 72 - 12
CHAPTER IV
PRESENTATION OF DATA
This study is designed to t e s t the v a l i d i t y and r e l i
a b i l i t y o f a ph on e t ica l ly balanced nonsense s y l l a b le te s t
which can be used for te s t ing speech i n t e l l i g i b i l i t y . In
order to t e s t v a l i d i t y , a s t a t i s t i c a l comparison was made
between the a r t i c u l a t i o n scores obtained on the ph on et ica l ly
balanced nonsense s y l l a b le te s ts and the widely used phoneti
c a l l y balanced word te s ts . To te s t r e l i a b i l i t y , each te s t was
given to each subject a second time under id en t ic a l condit ions.
Tables 4 and 5 show the a r t i c u l a t i o n scores for the
i n i t i a l tests and for the re te s ts . The a r t i c u l a t i o n scores
for the p h o n et ica l ly balanced word tes ts ranged from one
hundred per cent to s ix per cent on the i n i t i a l te s ts , and
from one hundred to fourteen per cent on the re te s ts . The
range o f a r t i c u la t io n scores for the p h on et ica l ly balanced
nonsense s y l l a b le tes ts was from seventy-e ight per cent to
zero per cent on the o r ig in a l t e s t , and from seventy-four per
cent to two per cent on the second te s t .
The means and standard deviat ions obtained from the
above scores are shown in Table 6. In order to determine
whether or not there was a s ig n i f i c a n t d i f fe re n c e between the
mean scores on the p h on et ica l ly balanced word te s ts and the
23
24
TABLE 4
PHONETICALLY BALANCED WORD (PB) AND PHONETICALLY BALANCED NONSENSE SYLLABLE (PBNS) TEST SCORES IN PERCENTAGES
OF INTELLIGIBILITY AND SIGNAL-TO-NOISE RATIOS -INITIAL TESTS
Test S/N Ratio 1 2 3 4Subj ect
5 6 7 8 9 10
PB la + 18 100 92 100 94 94 92 94 96 96 94
PB 2a + 6 90 82 98 74 68 78 82 82 82 78
PB 3a 0 96 74 98 64 68 78 84 68 92 74
PB 4a - 6 34 40 48 36 40 62 56 46 54 62
PB lb - 12 6 10 24 6 20 26 30 20 42 26
PBNS la + 18 66 64 58 64 74 72 78 66 66 62
PBNS 2a + 6 52 50 72 58 56 48 64 50 52 46
PBNS 3a 0 52 40 64 44 48 54 44 52 58 44
PBNS 4a - 6 14 20 18 18 20 26 30 22 32 24
PBNS lb - 12 4 4 6 4 6 2 10 0 10 8
25
TABLE 5
PHONETICALLY BALANCED WORD (PB) AND PHONETICALLY BALANCED NONSENSE SYLLABLE (PBNS) TEST SCORES IN PERCENTAGES
OF INTELLIGIBILITY AND SIGNAL-TO-NOISE RATIOS -RETESTS
T est S/N Ratio 1 2 3
Subj ect
4 5 6 7 8 9 10
PB la + 18 96 98 100 92 94 92 92 98 100 96
PB 2a + 6 92 86 88 84 82 80 66 90 84 84
PB 3a 0 86 80 78 68 68 70 62 86 76 76
PB 4a 6 54 44 50 42 48 56 34 52 40 50
PB lb - 12 20 24 36 22 16 18 16 14 28 22
PBNS la + 18 68 66 60 56 74 72 68 72 72 64
PBNS 2a + 6 56 60 58 56 60 50 56 52 58 34
PBNS 3a 0 52 44 48 52 46 58 48 60 56 42
PBNS 4a 6 28 18 32 10 18 28 16 24 30 24
PBNS lb - 12 12 18 6 8 6 6 2 2 10 6
26
TABLE 6
MEANS AND STANDARD DEVIATIONS OF INITIAL TESTS AND RETESTS FOR PHONETICALLY BALANCED WORD (PB) AND PHONETICALLY
BALANCED NONSENSE SYLLABLE (PBNS) TESTS
Means Standard Deviations
I n i t i a l Test Retest I n i t i a l Test Retest
PB la 95.2 95.8 2.87 3.19
PB 2a 8 1 .4 83 .6 8 .22 7.15
PB 3a 79.6 75.0 12.29 7.96
PB 4a 47.8 47.0 10.34 6.88
PB lb 21.0 21.6 11.31 6.58
PBNS la 67.0 67.2 5.98 4 .90
PBNS 2a 54.8 54.0 8.01 7.71
PBNS 3a 50.0 50.6 7 .50 6.02
PBNS 4a 22.4 22.8 5.63 7.07
PBNS lb 5 .4 7 .6 3.27 5.83
27
corresponding p h o n et ica l ly balanced nonsense s y l l a b le t e s t s ,
the T te s t was app l ied . The formula used for th is c a lc u la
t io n , as presented by Underwood (1 2 :1 2 7 ) , is as fol lows:
The F te s t was used to determine whether there was a s i g n i f i
cant d i f fe re n c e between the standard deviat ions o f the two
te s ts . The fol lowing formula used fo r the F te s t was pre
sented by G arre t t ( 4 :3 0 3 ) .
Results of the T and F tests are shown in Table 7.
The T te s t res u l ts show that there is a s ig n i f ic a n t
d i f fe re n c e , above the ten per cent confidence le v e l , between
the mean of each of the p h o n et ica l ly balanced word tests and
the mean o f each corresponding phon et ica l ly balanced nonsense
s y l la b le t e s t , with the exception o f ph on et ica l ly balanced
word te s t lb and p h o n et ica l ly balanced nonsense s y l la b le te s t
lb . Results o f the F te s t show tha t in the i n i t i a l te s t in g
there was a s ig n i f i c a n t d i f fe re n c e between the standard d e v i
at ions on three o f the matched sets o f te s ts . Phonet ica l ly
balanced word tes ts la and 4a and t h e i r corresponding phonet i
c a l ly balanced nonsense s y l la b le te s ts showed a s ig n i f ic a n t
d i f fe re n c e at the f i v e per cent confidence le v e l . P honet ica l ly
balanced word te s t lb and i ts corresponding phon et ica l ly
OTJiff
28
TABLE 7
RESULTS OF T TESTS FOR DIFFERENCES BETWEEN MEANS AND F TESTS FOR DIFFERENCES BETWEEN STANDARD DEVIATIONS IN COMPARING
PHONETICALLY BALANCED WORD TESTS WITH PHONETICALLY BALANCED NONSENSE SYLLABLE TESTS
T Test^ F Test%
I n i t i a l T est Retest
I n i t i a lTest Retest
PB la - vs - PBNS la 4 .26 4.90 4 .35 2.35
PB 2a - vs - PBNS 2a 2.31 2.82 1.04 1.16
PB 3a - vs - PBNS 3a 2.05 2.44 2.68 1.74
PB 4a - vs - PBNS 4a 2. 16 2.45 3.37 1.05
PB lb - vs - PBNS lb 1.32 1.59 11.95 1.27
1. A T value as high as 3 .25 indicates the 1% level o f confidence; 1.83 indicates the 10% le v e l .
2. An F value as high as 5.35 indicates the 1% level of confidence; 3 .18 indicates the 5% le v e l .
29
balanced nonsense s y l la b le te s t showed a d i f fe re n c e at the
one per cent confidence le v e l . No s ig n i f i c a n t d i f fe re n c e
between the standard devia t ions were found for the retests
at any 1e v e l .
G arre t t suggests that a t e s t - r e t e s t procedure be used
for computing the r e l i a b i l i t y o f a t e s t (6 :3 3 7 ) . The t e s t -
re te s t procedures used in the present study made i t possible
to obtain a measure of r e l i a b i l i t y o f each type o f te s t under
considerat ion. The Pearson product moment c o e f f ic ie n t of cor
r e la t io n was ca lcu la ted for i n i t i a l t e s t and re te s t o f both
the ph on et ica l ly balanced word te s ts and the p h o n et ica l ly b a l
anced nonsense s y l la b le te s ts . The formula used for th is com
putat ion , as presented by G a rre t t (6 :1 2 6 ) , is as fol lows:
r _ s f e ~N
C o rre la t io n for the p h o n et ica l ly balanced word te s t
and re te s t was 99 .2 , wh i le c o r re la t io n for the phonet ica l ly
balanced nonsense s y l la b le te s t and re te s t was 97 .8 .
The W-22 ph on e t ic a l ly balanced word l i s t te s t has been
previously accepted as a v a l id tool for i n t e l l i g i b i l i t y t e s t
ing; th e re fo re , the v a l i d i t y o f the newly constructed phonet i
c a l l y balanced nonsense s y l la b le tes t was determined by com
puting the c o e f f ic ie n t o f c o r re la t io n between the mean scores
of the two te s ts . Appl icat ion o f the formula above y ie lded a
c o r re la t io n o f 99.6 between the two te s ts , thus ind icat ing
30
tha t the p h o n et ica l ly balanced nonsense s y l l a b le tes ts are as
v a l id a measure o f i n t e l l i g i b i l i t y as are the phon et ica l ly
balanced word te s ts .
This chapter has been concerned with the s t a t i s t i c s
employed in the determination o f the r e l i a b i l i t y and v a l i d i t y
of the newly constructed p h o n e t ica l ly balanced nonsense s y l
lab le t e s t . The resu lts o f the inves t iga t ion were also p re
sented. The conclusions drawn from these data w i l l be sum
marized in the fo l low ing chapter.
CHAPTER V
SUMMARY AND CONCLUSIONS
Summary
The present experimentation was undertaken in order
to construct a new te s t for determining speech i n t e l l i g i b i l
i t y , using ph on et ic a l ly balanced nonsense s y l la b le s rather
than meaningful words as stimulus m a te r ia l . I t was f e l t th a t
the use o f ph on et ic a l ly balanced nonsense s y l la b le s would pro
vide a more accurate estimate o f i n t e l l i g i b i l i t y by reducing
perceptual cues. The s p e c i f ic questions to be answered con
cerning t h is t e s t were:
1. Is th is new te s t as r e l i a b l e and v a l i d for te s t in g i n t e l l i g i b i l i t y as the ph on et ica l ly balanced W-22 word test?
2. Does th is new te s t y ie ld lower a r t i c u l a t i o n scores by e l im in a t in g perceptual cues inherent in word test ing?
A review of the l i t e r a t u r e revealed tha t there are
many types o f I n t e l l i g i b i l i t y Tests in ex is tence, most of
which use meaningful words as t e s t m a te r ia l . A notable excep
t io n is a t e s t devised by F le tcher and Steinberg (4 ) using
nonsense s y l la b le s instead of words. The descr ip t ion of th is
t e s t indicates that the nonsense s y l la b le s are not phoneti-
c a l 1y balanced.
31
32
The I n t e l l i g i b i l i t y Test constructed for the present
study employed the developmental procedures fol lowed in the
construct ion o f two d i f f e r e n t types o f the tes ts now in e x i s t
ence. The stimulus m ater ia ls were o f the nonsense s y l la b le
type as described by F le tcher and Steinberg ( 4 ) . These non
sense s y l la b le s were ph on et ic a l ly balanced by using the same
phonemes the same number o f times as they were used in the
accepted p h o n e t ic a l ly balanced C * I .D . Auditory Test W-22.
A d i r e c t comparison between the newly formed phoneti
c a l ly balanced nonsense s y l la b le t e s t and the p h o n e t ica l ly
balanced W-22 word te s t was made by presenting both tests to
ten l is te n e rs at f i v e d i f f e r e n t s ig n a l - to -n o is e r a t io s . The
scores obtained by the ten subjects were s t a t i s t i c a l l y analyzed
in order to t e s t the v a l i d i t y and r e l i a b i l i t y o f the phoneti
c a l l y balanced nonsense s y l l a b le te s t as well as the overa l l
leve ls o f a r t i c u la t io n scores for the two types o f te s ts .
Conclus ions
The r e s u l ts o f th is in ve s t iga t ion a l low the fo l lowing
conclusions:
1. The ph on et ic a l ly balanced nonsense s y l l a b le te s t
constructed for th is study is as r e l i a b l e and v a l i d for t e s t
ing i n t e l l i g i b i l i t y as are the C . I .D . W-22 te s ts . The Pearson
product moment c o e f f i c i e n t o f c o r re la t io n ca lcu la ted between
the two tes ts produced a c o r re la t io n o f 99 .6 , ind ica t ing high
v a l i d i t y . A high degree of r e l i a b i l i t y was ind icated through
t e s t - r e t e s t procedures which showed a c o r re la t io n of 97 .8 .
33
2. Total perception from minimal cues is la rg e ly
e l im inated when te s t in g w ith nonsense s y l la b le s . A compari
son o f the a r t i c u la t i o n scores obtained by the subjects on the
two types o f te s ts points out tha t f a m i l i a r i t y with tes t in g
m ate r ia ls does produce markedly higher scores. The scores
obtained on the p h o n et ica l ly balanced nonsense s y l l a b le tests
were considerably lower than those obtained on the correspond
ing p h on et ica l ly balanced word t e s t .
The compiled s t a t i s t i c a l data fo r th is study a l low the
f in a l conclusion tha t the ph on e t ic a l ly balanced nonsense s y l
la b le te s t constructed for th is experimentation compares favo r
ably with C. I .D . Auditory Test W-22 in v a l i d i t y and r e l i a b i l
i t y . However, i t appears tha t te s t in g w ith words measures
perception as well as i n t e l l i g i b i l i t y . Since the ph on et ica l ly
balanced nonsense s y l l a b l e t e s t e l im inates to ta l perception
from minimal cues, i t is a more v a l id te s t o f i n t e l l i g i b i l i t y .
APPENDIX
PHONETICALLY BALANCED NONSENSE SYLLABLE TEST la
1. ta l 21. hoxid 41. tes
2. sib 22. zam 42. di
3. WjUV 23. r i l 43. wout $
4. SI 24. - tod 44. ta iy
5. kge 25. ts1 45. hwawn
6. PA 26. rera 46. mes
7. zin 27. msez 47. nouk
8. hev 28. Irr 48. l id
■9. na 29. 9ju 49. na
10. reel 30. di v 50. reb
11. Ses 31. l A
12. t A 32. jeed
13. dsaik 33. BA
14. rem 34. te
15. k i f 35. , ren
16. hun 36. toz
17. dait 37. ri6
18. nid 38. t TJ
19. r i 39. tees
20. f ep 40. wig
34
35
APPEND IX (C o n tin u e d )
PHONETICALLY BALANCED NONSENSE SYLLABLE TEST 2a
1. ta d 21 . kon 41. n e s
2. rju b 22. d si 42. . hod
3. t l V 23. rem 43. we
k. r e t 24. t i v 44. lo g
5. r e 25. l e 45. Z33S
6. z ju 26. wed 46. 8 At
7. S^p 27. 6ud 47. ta
8. g i t 28. pom 48. m lz
9. dAf 29. r in 49. h ju s
10. n ot % 30. 611 50. meb
11. vai t 31 . r j u l
12. m z 32. tse
1.3. ka 33. rA
14. wut 34. kus
15. HAS 35. j e
16. r i 36. han
17. l e 37. kai
18. di 38. Wl
19. t e n 39. meb
20. Isen 40. n a id
36
APPENDIX (C o n t in u e d )
PHONETICALLY BALANCED NONSENSE SYLLABLE TEST 3a
1. dA 21. t e f 41 . ni v
2. tod 22. 42. mu6
3. sjus 23. SI 43. sen
4. dssl 24. rse 44. run
5. to ip 25. he 45. 6 i d
6. tov 26. jed 46. tffil
7. djun 27. wem 47. t a i g
8. kA 28. mdy 00 non
9. dank 29. Is 49, wi
10. na 30. web 50. ru
11. t id s 31. 1®
12. f et 32. hi
13. reb 33. r i z
14. hAS 34. r a ik
15. ko 35. tA
16. r a l 36. zlra
17. hwi s 37. re
18. InS
00on san
19. nai 15 39. r i ■
20. z i t 40. taQi
37
APPENDIX (C o n t in u e d )
PHONETICALLY BALANCED NONSENSE SYLLABLE TEST 4a
1. h s t 21. mi 4 ) , kAn
2. d i 22. dsz 42. dut
3. k o it 23. ' dom ; 43 . dgi s
4. ni 24. ruS 44. ro5
5. ma 25. je z 45. jeen
6. wai v 26. ne 46. lu
7. rae 27. tee 47. I i 5
8 o r ju t 28. mas 48. hw ol
9. t a r f 29. lAn 49. d l t
10, hAS 30. l a ib 50. t e
11. ta z 31. wot
12. mev 32. h iy
13. VI 33. l a l
14. t^un 34. se
15. hwe 35. 01
16. k ig 36. rap
17. nok 37. nes
18. r i z 38. rem
19. dm s 39. rxi
20. natrt 40. ret)
38
APPENDIX (C o n t in u e d )
PHONETICALLY BALANCED NONSENSE SYLLABLE TEST lb
1. dap 21. ta1! 41. nai d2. lib 22. nez 42. gan3. nu 23. kA 43. dA4. mi 24. ro$ 44. ron5. wi© 25. ved 45. wai s6. re 26. la 46. haut7. t ju 27. r it 47. veb8. VIS 28. lep 48. zut9. hin 29. wid 49. lot
10. raik 30. zit 50. to811. IQA 31. jAS12. tuio 32. ral ,13. ke5 33. t$i14. ne8 34. nok15. fi 35. hwat16. ta 36. ma17. rse 37. di18. zem 38. dges19. njuf 39. run20. mas 40. hes
REFERENCES
1. Black, J. W, , M u l t ip le -c h o ic e i n t e l l i g i b i l i t y te s ts .J. Speech Hearing P is . , 22, 1957, 213-235.
2. Dewey, G . , R e la t iv Frequency o f English Speech Sounds.London: Humphrey M i l fo r d , Oxford U n iv e rs i ty Press.Cambridge: Harvard U n ivers i ty Press, 1931.
3. Fairbanks, G . , Test o f phonemic d i f f e r e n t i a t i o n : theRhyme Test. J. Acoust. Soc. Amer. , 30, 1958, 596-600.
4. F le tcher , H. and J. C. S te inberg , A r t ic u la t io n tes t ingmethods. Bell Svst. Tech. J . , 8 , 1929, 806-854.
5. French, N. R . , C. W. Carter and W. Koenig, The words andsounds o f telephone conversations. Bel 1 * Svst .Tech. J . , 9, 1930, 290-324.
6 . G a r r e t t , H. E . , S t a t i s t i c s in Psychology and Education.New York, London: Longmans Green and Co., 1953.
7. H irsh , I . J . , H. Davis, S. R. Silverman, E. G. Reynolds,E. E ld e r t and R„ W. Benson, Development o f m ater ia lsfor speech audiometry. J. Speech Hearing P is . , 17, 1952, 321-337,
8 . Kantner, C. E. and R. West, Phonetics. (Rev. E d , ) NewYork: Harper and Brothers, 1960 .
9. Kenyon, J. S. and T, A. Knott, A Pronouncing D ic t ionaryo f American Engl ish . S p r in g f ie ld , Massachusettst G. and C. Merriam Co., 1944.
10. Newby, H . , Audioiogy; P r in c ip le s and P r a c t ic e . New York:Appleton-Gentury-Crofts , I n c . , 1958.
11. Thorndike, E. L. and 1. Lorge, The Teachers Word Book o f30,000 Words. New York: Columbia U n iv e rs i ty Press,l a t e ---------------
12. Underwood, B, J . , C. P. Duncan, J . A. Taylor and J. W.Cotton, Elementary S t a t i s t i c s . New York: Appleton-C en tu ry -C ro f ts , I n c . , 1954.
39