I ".VNI IRSITY RESEARCH STUDIES IN EDUCATION NUMBER 5
THE EFFECT OF VARIED
AMOUNTS OF PHONETIC
TRAINING ON PRIMARY
DONALD C. AGNEW
..... :.i ii '
:i DUKE UNIVERSITY PRESS
Durham, N. C.
DUKE UNIVERSITY RESEARCH STUDIES IN EDUCATION
THE EFFECT OF VARIED AMOUNT;
OF PHONETIC TRAINING ON
THE EFFECT OF VARIED
AMOUNTS OF PHONETIC
TRAINING ON PRIMARY
DONALD C. AGNEW
Coker College, Hart ville, South Carolina
DUKE UNIVERSITY PRESS
Durham, N. C.
COPYRIGHT, 1939, BY THE
DUKE UNIVERSITY PRESS
PRINTED IN THE UNITED STATES OF AMERICA BY
THE SEEMAN PRINTER, INC., DURHAM, N. C.
For one reason or another, research has not yielded an unequiv-
ocal answer to the place of phonetic training in the teaching of pri-
mary reading. Sometimes experiments have covered but a short
period of time; sometimes programs of measurement by means of
which the effects of instruction have been assessed have been unfor-
tunately limited and have neglected important outcomes in reading;
and sometimes evaluations have been based upon immediate results
only, with little consideration of deferred results of possible value.
Meanwhile, in the absence of consistent and competent research find-
ings the educational theorist and the practical teacher have perforce
continued to follow their own experience, their insights, and their
Dr. Agnew has wisely not attempted a final answer to the ques-
tions he has investigated. He has not sought, on the basis of his
research data, to say whether phonetic training should be emphasized
or minimized in primary reading, for he has recognized that a de-
cision in this matter rests only in part upon data such as he reports,
and rests much more upon one's conception of the purposes and
aims of reading as a whole. What Dr. Agnew has done is to collect
important new data on the effects of large and of small amounts of
phonetic training. In so doing, he has employed a measurement pro-
gram far more comprehensive than has usually characterized studies
of phonetic instruction, and he has included practically all aspects of
reading ability as this ability has been analyzed by others.
The research herein reported should be of interest not alone to
the teacher and the administrator for its practical results and its im-
plications; it should be of interest as well to the student of educa-
tional research. Dr. Agnew has used a number of procedures, in-
volving both statistical method and measuring techniques, which may
be of considerable worth in the investigation of other instructional
problems of the school.
WILLIAM A. BROWNELL.
I am deeply indebted to Dr. William A. Brownell, Professor of
Educational Psychology, Duke University, for his proposal of the
thesis topic, for his aid and guidance in carrying on the investiga-
tions, for his suggestions as to the treatment of the data, and for his
critical advice with respect to reporting the findings. I am indebted
also to Dr. Howard Easley. Assistant Professor of Education,
Duke IUniversity, for his suggestions concerning the statistical treat-
ment of thlI results and concerning certain psychological interpreta-
tions. I kni grateful to Dr. Holland Holton, Head of the Depart-
ment of Education, Duke University, for his encouragement and aid
throughout the investigations.
The investigations were made possible through the generous co-
operation of Superintendent P. S. Daniel and Assistant Superintend-
ent, Miss Mildred English, of the Raleigh, N. C., Schools; and of
Superintendent \V. F. Warren and Mrs. J. A. Robinson (Primary
Supervisor), of the Durham, N. C., Schools. The elementary
school principals and the teachers in Grades I, II, and III, in both
Raleigh and Durham, were very helpful throughout the testing pro-
grams. I appreciate, too, the valuable assistance of the members
of the classes in Experimental Education and in Educational Meas-
urements (Duke University), who gave numerous tests in the Raleigh
and Durham schools.
DONALD C. AGNEW.
October 11, 1938
TABLE OF CONTENTS
I. INTRODUCTION .................. .................... 3
II. A BRIEF ACCOUNT OF THE INVESTIGATIONS TO BE
REPORTED ................. .... ..................... 9
A. The particular problems investigated .............. 9
B. The general theory of the investigations ........... 10
C. A brief account of the procedure employed ........ 12
D. The measures used in the investigations ........... 12
E. The Durham investigation. ...................... 14
III. THE TECHNIQUE OF THE FIRST INVESTIGATION..... ...... 15
A Selection of subjects...................... ..... 15
B. The measures of phonetic experience.............. 15
C. The testing program ........... ................. 18
IV. THE METHODS OF TREATING THE RESULTS.............. 21
A. Comparison based on the Gross Phonetic
Experience Scores ............................. 21
B. Comparisons of patterns of phonetic experience.... 22
V. RESULTS AND CONCLUSIONS OF THE RALEIGH
INVESTIGATION ..................................... 27
VI. THE PURPOSE AND TECHNIQUE OF THE DURIAM
INVESTIGATION ..................................... 35
A. Purpose ................... ................... 35
B. The technique of the investigation............... 35
VII. RESULTS AND CONCLUSIONS OF THE DURHAM
INVESTIGATION .................................... 40
Conclusions ....................................... 42
VIII. BRIEF SUMMARY OF THE RESULTS OF THE INVESTIGATIONS,
AND CERTAIN EDUCATIONAL IMPLICATIONS ............. 44
A. The relation of the investigations to the controversial
issues with regard to phonetic instruction .......... 44
B. Theoretical considerations ....................... 45
C. Educational implications ........................ 47
BIBLIOGRAPHY ................ ............................ 48
1. Studies of the Phonetic Content of English Vocabularies at the
Primary Level ........................................... 4
2. Studies of the Value of Phonetic Training in the Elementary
G rades ................................................. 5
3. Distributions of Scores on the Teachers' Blanks by Grades...... 16
4. Differences Between the Responses of Teacher F. L. and Teacher
R. H. on Ten Items of the Teachers' Blank ................... 17
5. Distri ution of Gross Phonetic Experience Scores (III-A
Pupils in Raleigh) .......................................... 18
6. Distributions of M.A.'s and I.Q.'s for Groups GH and GL...... 22
7. Frequency Distribution of Phonetic Experience Scores of Pupils
in Grade I................................................ 24
8. Distributions of Phonetic Experience Scores for Grades II and
III Combined Before and After Equating..................... 24
9. Distributions of Equated Groups.............................. 25
10. Summary of the Differences in Terms of the P. E. of the
Differences ............................ .. ........ ....... 27
11. Means and Standard Deviations of the M.A.'s for the Various
Patterns ................ ........... ................... 28
12. Correlations Between Test Scores of Groups GH and GL
(Groups Matched for M.A. and I.Q.) ........................ 29
13. Frequency Distribution of Gross Phonetic Experience Scores
(110 Cases, Durham) ...................................... 37
14. Frequency Distributions of M.A.'s and I.Q.'s of the Raleigh and
Durham Groups Selected by Pairing ......................... 38
15. Factors in the Composition of the Durham and Raleigh Groups.. 39
16. Differences Between the Raleigh and Durham Groups in Terms
of the Differences Between the Means of the Test Scores (89
Cases in Each G roup) ...................................... 41
[ viii ]
THE EFFECT OF VARIED AMOUNTS
OF PHONETIC TRAINING ON
Statement of problem.-The investigation to be reported was
undertaken in an effort to determine the effects of varied amounts
of phonetic training on certain reading abilities as measured by a
battery of tests. The particular reading abilities measured were
those which, it was thought, should shed light on some of the con-
troversial questions that arise in connection with the use of phonetic
methods in beginning reading instruction.
The history of phonetic methods of instruction.-The history of
phonetic methods of instruction is too long to be recounted here. It
should be pointed out, however, that since the beginning of the nine-
teenth century when Noah \ebster's blueback speller stressed the
sounding of letters and syllables in reading and -".lliiil. ph.nertic
iih Atrii,-tion ha-. ha.J nlLI.ri r i',:ul i ri.,d .'. *_f t, ularity and unp'-i Aulariti ,.
[Fuilthermiorei certain mindss i, phi I,-tl c I -_,truc.tio- i. If.r exanlImple, tIhe
.llall rentlh,_,'1. rhe alphalber n'lthulod. [hi '.\, 'id metilod. and ';..,ial
niodilcati.i'ni i-.f tlihe:s nnili.h.,'s. hi\.- been -,'.,l :licl l or ic:.ten-
l-\\':l in certain n [.lcahri-s. Dliiri ti : l- 1 ..le,- .-il' r,.treni.l_ ifYer-
ences of opinion have been expressed by various educators as to the
value of phonetic instruction. Thus. ihr Trh.,rv -_f phonetic m nruc-
;il:, iz i.-ni- i irlco:ln-i.iencv aild c ,ntr..i ersy.'
Quantitative studies related to phonetic training.-The quanti-
tative studies directly related to the problem may be considered in
two categories: studies of the phonetic content of vocabularies, and
studies of the value of phonetic training.
The studies of the phonetic content of vocabularies are listed in
Table 1. The results of the studies of sound frequency have had
considerable bearing on the teaching of phonetics. They have shown,
in general, the following results: (a) The English language has a
large number of nonphonetic words. (b) Letters and letter-
combinations often have a large number of possible sounds. Horn
'A more detailed history of phonetic methods of instruction is given in
Chapter I in the unpublished thesis by the writer: "The Effect of Varied
Amounts of Phonetic Training on Primary Reading" (Duke University.
1936). In the same place, in Chapter II, and in considerable detail in Appen-
dix A, are given an account and a criticism of the quantitative studies related
directly and indirectly to the problem. These studies are listed in the Bib-
Varied Amounts of Phonetic Training
STUDIES OF THE PHONETIC CONTENT OF ENGLISH VOCABU.iLsrilE'
AT THE PRIMARY LEVEL
Author Problem Technique
Atkins Constancy of phonetic sounds in Analysis of words and calculatic. .--1 ihe ir.- .ei.-
(3)* the first 2,500 words of the Thorn- of phonetic and nonphonetic sotd.-. i,. Th Tt. .-
dike list. dike list.
Cordts Classification ofsoundsin primary Analysis of words and calculati.:c :1 te e.:.' :f
(I1) vocabulary, phonetic sounds in a large num':r :.f irAder:
Horn Classification of sounds of the let- Analysis of words and calculator I it i-.er:, % f .11
(26) ter "a" in vocabulary of the first sounds of the letter "a" in the vr." i .1 ,Ii pre ,ire
three grades. by Cordts.
Vogel, Jaycox, Classification of list of phonetic Analysis of words and calculati- .'.[ [re pien.: s f
and Washburn elements for Grades I and II. phonograms in Packer's vocabu'ar, ,c.j .eir.e ler
Washburn and Classification of phonetic elements Analysis of vocabulary and calculation of frequency
Vogel in Gates's Primary Reading Vo- of sounds in the Gates list.
(45) . ..j1-, I:. Grade II.
*Numbers in parentheses following the names indicate the numbers of the references given in the
(26),2 for example, finds that the letter "a" has almost fifty different
sounds when used alone or in digraphs. He shows also that "ea,"
for example, may have as many as eight different sound interpreta-
tions when found in different settings. (c) Lists of the most fre-
quently occurring sounds have been made so that emphasis may be
placed upon the sounds according to their deemed importance.
(d) Elementary vocabularies have been analyzed for their phonetic
Various inferences have been drawn from these studies. Some
educators, for example, Horn (26), claim that the studies show
the teaching of phonetics to be an almost impossible task for the
elementary school. Others, for example, Wheat (46), insist that
phonetics is necessary for the acquisition of reading skills.
The results of the experimental investigations of the value of
phonetic training are by no means conclusive. These studies are
listed in Table 2. The organization of the experiments was not
such that the studies could yield unequivocal evidence. The avail-
able data seem to point tentatively to the following conclusions:
(a) Investigators tend to agree that the first few months of reading
should emphasize the "look and say" method.3 Sexton and Her-
SHereafter in this report, numbers in parentheses following the name
indicate the number of the reference listed in the Bibliography at the end
of the monograph.
'Winch (47) holds the opposite opinion.
STUuiiCS OF THE VALUE OF PHONETIC TRAINING IN THE ELEMENTARY GRADES
Au., hor Problem Technique
Currier Value of phonetic training in Grades I and II. Phonetic Control group experiment.
(12)* training given one group for two years.
Garrison and Value of fifteen minutes daily drill in phonetics in Grades Control group experiment.
Heard (15) I and II. Phonetic training given for two years.
Gates Value of fifteen minutes daily drill in phonetics during about Control group experiment.
(22) six months of Grade I (two experiments).
Mosher Value of "look and say" method for beginners without pho- Comparison of groups of
(31) netic training during the first year. of differing ability.
Mosher and Value of "look and say" method vs. phonetic methods, Control group experiment.
Newhall (32) Grade I. Phonetic training given for seven months.
Sexton and Value of fifteen minutes daily drill in phonetics in Grade 1. Control group experiment.
Herron (39) Phonetic training given for seven months.
Winch Value of fifteen minutes daily drill in phonetics vs. "look Control group experiment.
(47) and say" methods. Phonetic training given on twenty-five
consecutive school days.
*See note under Table 1.
ron (39), Gates (22), Mosher and Newhall (32), and Garrison and
Heard (15) agree in this conclusion. (b) Gates (22) and Sexton
and Herron (39) found the "intrinsic method" to be somewhat supe-
rior to the special drill periods, at least when the results are measured
after a brief period of time. (c) It seems to be conceded that
phonetic training aids in word recognition. (d) Garrison and Heard
(15) found that such training, however, inhibits fluency in oral
reading. (e) Currier (12) found that some children need phonetic
training more than do others.
A review of the studies indicates that the evidence on the value
of phonetic training is limited. The need for further investigation
Difference of opinion among teachers.--Nearly every elementary-
school teacher has deep-seated beliefs about the importance or weak-
ness of phonetic methods. Many teachers believe intensive phonetic
training desirable. Nila B. Smith points out, on the other hand:
"There is an impression among some teachers that phonetics is a
disgrace, that this phase of instruction is of no value and is generally
Other studies indirectly related to phonetic training are reviewed in the
original thesis. The bibliographical references to these studies are numbered
as follows: 4, 6, 9, 13, 21, 24, 28, 30, 31, 41, 42.
5 From American Reading Instruction, 1934, by permission of the author,
Nila B. Smith, and the publisher, Silver Burdett Company. See Smith (40),
6 Varied Amounts of Phonetic Training
The attitude of textbook writers.-The attitude of textbook
writers is more conservative. Smith summarizes the situation as
While some . textbook writers seem less certain than writers of
former times on this subject, there is nothing to indicate that any one of
them takes the extreme attitude of dispensing with phonetics entirely.
Every manual that has appeared in connection with a basal series of
readers during this period (1925-1934) has recognized phonetics. Various
states of confidence in the value of phonetics are expressed by authors,
but they all discuss this phase of reading and outline procedures for
Disagreement among educators.-McKee summarizes the situa-
tion with respect to the attitude of educators toward the problem
1'he que:tion .:. iii iructiion in plIonic- hai aruuscd 1 gcre.it deal of
or.ntro ery!' S.'., ie educ'it... '. c held co thie t prop.-i,iton i.htt phi:.netic
training is not only futile and wasteful but also harmful to the best in-
terests of a reading program. Others believe that since the child must
have some means of attacking strange words instruction in phonics is
imperative. There have been disputes also relative to the amount of
phonics to be taught, the time when the teaching should take place, and
the methods to be used. In fact the writer knows of no problem around
,. which more disputes have centered.7
Examples of these differences of opinion are numerous in the
literature. O'Briens is highly in favor of phonetic training. Wheat
holds the same opinion. He says: "Phonic analysis is the device
necessary to train pupils to avoid periods of confusion."9
On the other hand, Gates considers phonetic training of doubtful
value, particularly in the first grade, and especially when taught by
ih:- traditional methods. He says:
The great mistake in American teaching has been the assumption that
phonetic skill was all-important and sufficient, that the other types of
training could be neglected, and that the more phonetics the pupil got the
,,.tter. These mistakes have resulted not only in waste but frequently
in the production of a special type of difficulty in reading. So excessive
has phonetic drill often been that pupils have become not only "word-
fi-rm conscious" at the expense of interest in meanings, but, even worse,
they have become "word detail conscious" . . Thus phonetic skill in
moderation is useful; in less degree, it leaves the pupil handicapped; in
t!reater degree it may result in a more serious deficiency.10
McKee (29), p. 191. Quoted by special permission of the author.
SO'Brien (33), p. 225. Wheat (46), p. 199.
1 Gates (16), pp. 125-126. Quoted by special permission of the publishers:
tal.'ll" agrees with Gates and contends that the teaching of
ph:o1.llci,:- i: I'irgely on the defensive.
Tile (quotations cited are only a few of the many diverse opinions
that could be found, but they are enough to bear out the statement
taken from McKee. In the following paragraphs, a summary of the
arguments that are advanced for and against phonetic training is
The case in favor of phonetics.-The following arguments may
Ie said to sum up the case in favor of phonetic training:
1. Phonetic training has had a long history; during this period
of years until quite recently, it has been provided in increas-
ingly large amounts. Procedures that have been used in the
teaching of reading for a century should be scrutinized very
carefully before they are abandoned.
2 Phonetic training gives the pupil independence in recognizing
words previously learned. This ability becomes steadily more
important in connection with silent reading.
3. Phonetic training aids in "unlocking" new words by giving
the pupil a method of sound analysis.
4. Phonetic training encourages correct pronunciation and
5. Phonetic training gives valuable "ear training" in recognizing
and differentiating sounds.
6. Phonetic training improves the quality of oral reading, for
instance, in breath control and in speech co-ordination.
"7. Phonetic training improves spelling.
8. Phonetic training is a valuable background for shorthand.
9. Many cases of reading disability may be traced to deficiencies
in word recognition and sound analysis. These disabilities
are often overcome by remedial procedures involving phonetic
The case against phonetics.-The disadvantages attributed to
:h.~ler'i.c training may be summarized as follows:
1. Phonetic training tends to isolate words from their meaning-
ful function by emphasizing sound.
2 Phonetic training tends to lead to the neglect of context clues.
3. Phonetic training tends to sacrifice interest in the content of
"Cabell (7), pp. 370-373.
S Varied Amounts of IPhi'ltci Trainit,.i
4. Phonetic training leads to unnecessarily laborious r':c.cognitilIi'
of familiar words.
5. Phonetic training is impractical because of the ni:.Iipll.:.neic
character of English.
6. Phonetic training is unnecessary for many pupils since its
advantages can be obtained without formal training,
7. Phonetic training encourages the breaking of word: init uin-
necessarily small units.
8. Phonetic training narrows the eye-voice span.
9. Phonetic training tends to emphasize too explicit artctliot.n.
Need for experimental evidence.-The claims and :objecti:oni
listed above are inferences based largely on a priori co:,nr.lerati..n..
As such they are at best tentative rather than final. I-lo-e phni.eti:
training _suLlt in these outcomes? The answer must com: le Iri m
scientific d-i a rather than from mere speculation. Perhaps Ioln:netic
training is neither as bad as one group claims nor a z'id qc tlI-w
other group insists. The studies reported here were undertaken iI
an effort to help solve some of the problems raised in ihe ir 'r.c'in
.\ BRIEF ACCOUNT OF THE INVESTIGATIONS
TO BE REPORTED
A. THE PARTICULAR PROBLEMS INVESTIGATED
The object of the investigations herein reported was to obtain
data concerning the validity of some of the claims and objections to
phonetics as pointed out in Chapter I. In fact, the two separate
studies were direct outgrowths of an analysis of these claims and
objections. The general question raised by the analysis was that of
the relative value of phonetic training and of nonphonetic training as
a basis for teaching reading abilities. This general question involved
certain particular questions.
(a) What is the comparative effect of phonetic and nonphonetic
reading instructions on speed and comprehension in silent reading?
The advocates of phonetic training claim that instruction which stresses
phonetic training aids the pupil by giving him methods of attack on
unfamiliar words, thus increasing both speed and comprehension.
The oppponents of phonetic training claim that this type of word
n:i, l\i- tends to make the pupil "word conscious" or "syllable con-
.i,-_.s" and thus slows up reading and renders comprehension more
diltfic:lt because of overemphasis on small sound units.
I, i What are the effects of phonetic and nonphonetic training
o:111 -]peed and accuracy in oral reading? The argument for phonetic
traniiigii is that such training leads to recognition of words and thus
i'icreaies speed, and that it also leads to more accurate pronunciation.
-iln the other hand, the argument against phonetic instruction is that
:-ich training slows up oral reading because of emphasis on small
-ro011d units and that it leads to less accurate pronunciation because
-I tlhe nonphonetic character of many English words.
i c What is the effect of phonetic training on eye-voice span?
Tle oprponents of phonetic training claim that to make the pupil
"...r.-l conscious" is to limit his eye-voice span to single words and
parts if words. Too much concern with small units of recognition
l_.et_- the eye and the voice together and thus prevents the eye from
in;.. ingi considerably ahead of the voice as in fluent oral reading.
Eix'-,ni:nts of phonetic training naturally minimize this danger.
Varied Amounts of Phonetic Training
(d) What are the effects of phonetic and nonphonetic training on
reading vocabulary? This question may be stated thus: Does sound
analysis, by giving training in independent word recognition, increase
vocabulary, or is vocabulary increased with greater ease by "word
whole" and "context" methods?
(e) Finally, does or does not phonetic instruction actually result
in greater abilities to use phonetic methods? Opponents of phonetic
training believe that the skills required for systematic sound analysis
are too complex to be learned with any degree of mastery in the
primary grades, but that, on the other hand, simple phonetic methods
may be developed by the pupil as he feels a need for such methods,
without his having been subjected to formal training.
P THE GENERAL THEORY OF THE INVESTIGATIONS
The d:cili:'-' having been made to investigate the problems named
above, the question of the most suitable technique arose.
The control-group technique, as used in most of the previous
experiments, has been open to a number of criticisms: (a) It tends
to set up artificial situations in which the instructional materials and
methods are often new to the teachers. (b) It allows free play for
prejudices on the part of teachers so that they may wittingly or
unwittingly motivate learning by one method and impede learning by
another method. (c) Often it develops a spirit of competition be-
tween the experimental and the check group that makes learning
under these conditions different from that of the typical school
It was decided, therefore, in the investigations to be undertaken
to avoid these limitations through the use of another technique. The
plan was to study the results of different teaching procedures which
had been employed in ordinary school situations. That is to say,
the plan was to locate one group of children who had been given
large amounts of phonetic training and another that had been given
small amounts of phonetic training, and then to compare these groups
with respect to reading skills.
A number of questions then arose: (a) In what grade should the
investigation be made? (b) Where could the investigation be in-
stituted in order to provide a wide variation in phonetic experience?
(c) How could the amounts of phonetic experience be measured?
In order to orient the reader to the discussion that is to follow, a
brief answer to these questions is given in the succeeding paragraphs.
A Brief Account of the Investigations to Be Reported
(a) Grade.-A major criticism of the previous studies has been
that the effects of phonetic training have usually been measured im-
mediately after such training has been given. Little time was allowed
to elapse between training and testing, although it may well be that
the effects of phonetic training may not be appreciable until some-
what later. Even the end of the second grade (if training is given
in the first grade) may be too soon to measure these effects. It
seemed advisable, therefore, to make the present investigations in
the last half of the third grade. By that time the values of phonetic
training (if any) should be apparent and measurable. If no dif-
ference in reading ability appeared at that time, it would be neces-
sary to conclude that the claims and objections of both advocates
and opponents of phonetic training have been exaggerated. It seemed
unlikely that effects of different methods in teaching primary reading
would for the first time emerge at a later period in the child's progress.
(b) Location.-It was thought that the schools selected for in-
vestigation should be large enough to give a good sample of third-
grade pupils. On the other hand, the sample should be small enough
to permit testing throughout the system so that the whole population
could be included, the problem of selection of schools or of sampling
the pupil population thus being avoided. It was necessary also to
choose a school system having a large number of first-, second-, and
third-grade teachers in order to obtain a wide variation in the amounts
of phonetic teaching. As will be shown later, the city of Raleigh,
North Carolina. provided an excellent location for the first inves-
(c) Measures of phonetic training.-The measurement of the
phonetic experience to which pupils had been subjected presented a
difficult problem. The children could give no information on this
point, and interviews with the teachers could yield no real measure.
One could, of course, ask a teacher whether or not she used phonetic
methods, but her reply, "not much" or "a lot," would be altogether
dependent upon her conception of what these terms mean. In other
words, the interview method could not yield a quantitative measure
of phonetic training. The scale finally employed is described in
'The city of Durham, North Carolina, provided a good location for the
second investigation because it is in many ways comparable to Raleigh.
12 Varied Amounts of Phonetic Training
C. A BRIEF ACCOUNT OF THE PROCEDURE EMPLOYED
The procedures used in the investigations fall logically into three
parts: (a) those used to secure data on the pupils' phonetic expe-
rience; (b) those employed in testing; and (c) those involved in
treating the results. These phases of the investigations will be taken
up in order in the succeeding paragraphs.
(a) Securing data on the pupils' phonetic experience.-Data were
obtained in regard to the phonetic experience of the various pupil
subjects by means of two instruments called in this report the Pupils'
Blank and the Teachers' Blank. These measures are described in
detail later. At this point, it is only necessary to say that the Pupils'
Blank was designed to secure a record of the pupils' educational
histories in terms of the schools they had attended and of the teachers
they had hadi in the first three years. The teachers thus designated
were then a:ked to fill out the Teachers' Blank which was designed
to furnish a quantitative measure of the amount of phonetic instruc-
tion they had given to the pupils. The relating of these two sets of
data made it possible to secure a measure of the amounts of phonetic
experience which each child had had in each grade and in the three
(b) The testing program.-Tests were carefully selected to meas-
ure various reading abilities. Group tests of silent reading abilities
and vocabulary, and individual tests of oral reading, word pronun-
ciation, eye-voice span and phonetic abilities were administered. In
addition to these tests of reading abilities, a group intelligence test
was given. Measures of intelligence were intended to relate the
factor of general intelligence to the scores on the reading tests and
thus aid in the interpretation of the results. A list of these tests is
given in Chapter III, together with detailed descriptions of their
content and purpose.
(c) The treatment of the results.-Groups of pupils who had
been subjected to different amounts of phonetic training were com-
pared in terms of the scores they made on the various tests. In order
to determine the effects of phonetic and nonphonetic training during
the various times in the grade experience of the first two and one-
half years of school, a complex statistical analysis of the data was
necessary. This analysis is explained in Chapter IV.
D. THE MEASURES USED IN TIE INVESTIGATIONS
In the preceding account of the investigation, mention is made
of three types of measures, namely, those secured from (a) the
A Brief Account of the Investigations to Be Reported 13
Pupils' Blank, (b) the Teachers' Blank, and (c) the tests of intel-
ligence and reading ability. In this section, the first two will be
described in detail and a list of the tests will be presented.
(a) The Pupils' Blank.-The Pupils' Blank was designed to
record pupils' phonetic experience. A sample of the blank as filled
in for John Smith is given below.
Name of the pupil John Smith
Grade School Teacher's Name Experience Scores
I1 A Lewis Miss A 45
II B Wiley Miss B 50
II A Murphey Miss C 40
I B Murphey Miss D 39
IA Murphey Miss D 39
Gross Phonetic Experience Score 213
Copies of the blank were distributed to all the third-grade pupils
in the schools for white children in Raleigh. The children were
asked to fill in the columns under "School" and "Teacher's Name,"
thus giving the name of the school attended during each half grade
together with the name of the teacher who had taught them in that
half grade." It will be noted that John Smith, in the case given above,
was in Grade III A in the Lewis School and that his teacher in that
grade was Miss A. In Grade II B, he was in the Wiley School and
his teacher was Miss B, etc.
The last column, "Phonetic Experience Scores," was filled in
later with data obtained from the Teachers' Blank.
(b) The Teachers' Blanks.3-All the teachers who had taught in
Grades I, II, and III during the years 1929-32, and who could be
reached in January, 1932, were given one or more Teachers' Blanks.
These blanks were designed to secure information concerning the
amount of phonetic instruction a teacher had given in each year
covered by the blank. One teacher, for example, had taught in
Grade I one year and in Grade II another year. This teacher filled
out a blank for each year. Thus each teacher filled out a maximum
of three blanks.
2 In case a child was unable to furnish the information, it was obtained
from the school office.
For a copy of the Teachers' Blank, see Agnew (2), Appendix B. This
blank was devised by Dr. William A. Brownell, Professor of Educational
Psychology, Duke University.
14 Varied Amounts of Phonetic Training
The Teachers' Blank is composed of twenty-five questions, each
of which has four possible answers, so worded that they indicate
varying degrees of emphasis on phonetic instruction. For instance,
Item 6 reads as follows: "With respect to consonant blends (tr, bl,
st, etc.) I teach children (a) a very great many, (b) all the common
ones, (c) only a few of the most common, (d) none at all as such."
The teacher answers the question by placing the letter "a," "b," "c,"
or "d" in the space for this purpose. In this case, the answer "a"
indicates thorough instruction in the phonetic method with respect
to this one item, namely, consonant blends, while an answer "d"
indicates practically no such instruction, and "b" and "c" represent
intermediate degrees. For purposes of scoring, "a" was here given
a weight of 4, "b" a weight of 3, "c" a weight of 2, and "d" a weight
of 1. All of the other twenty-four questions were similarly constructed
and similarly scored. These twenty-five item-scores therefore con-
stituted a scale with a possible range of 25 to 100, 25 representing
the least possible amount of phonetic instruction.4
The scores obtained from the Teachers' Blanks were transferred
to the Pupils' Blank to yield a quantitative measure of the pupils'
phonetic experience. Thus, in the case of John Smith,5 the score of
Miss A on the Teachers' Blank was 45, of Miss B, 50, etc. The sum
of the scores is called, in this report, the Gross Phonetic Experience
Score. This score for John Smith was 213.
(c) The measures of abilities.-The abilities which were measured
have been pointed out. Descriptions of the tests used to measure
these abilities are given in Chapter III.
E. THE DURHAM INVESTIGATION
A wide variation in amounts of phonetic training was found in
Raleigh. Nevertheless, in order to test the results of the first ex-
periment, and to secure more data, principally for children who had
had a consistently larger amount of phonetic training than the Raleigh
children had had, it was felt that a check investigation was desirable.
Consequently, a second study was made in Durham where the policy
for the three years (1932-1935) had been one of relatively more
emphasis on phonetics than was the case in Raleigh. Since the
technique in the Durham experiment was essentially the same as
that used in Raleigh, the details are not given here. A full account
of this investigation is given in Chapter VII.
SThe reliability of the scores on the Teachers' Blank (obtained by the
split-half method, odds vs. evens) was found, for 60 cases, to be .96. This
remarkably high coefficient is indicative of the inner consistency of the items.
See Pupils' Blank above.
THE TECHNIQUE OF THE FIRST INVESTIGATION
A. SELECTION OF SUBJECTS
Number.-For the first selection of pupils, Pupils' Blanks (the
nature of which is discussed in the preceding chapter) were dis-
tributed to the pupils in Grade III A (the lower half of the third
grade). Blanks were secured from 356 pupils. Since, in this inves-
tigation, only those pupils were used who had had all their previous
schooling in the city of Raleigh, certain eliminations were necessary
to preserve the homogeneity of the subjects. (a) All pupils who
had repeated grades due to failure or for any other cause were elim-
inated. They numbered 23. (b) All pupils who, for a term or more,
had attended school in any system other than Raleigh were omitted.
These pupils numbered 50. (c) Pupils who had been accelerated
were omitted. These numbered 13. (d) Ten were omitted because
of some ambiguity in their records. (e) Nearly all pupils taught in
Grades I to III by teachers who in the period 1929-31 had left the
Raleigh system were also omitted.' These pupils numbered 30.
Thus, in all, 126 pupils were eliminated. After these eliminations
were completed, there remained 230 III-A pupils.
Homogeneity of subjects.-These 230 pupils who were finally
selected were similar in several respects: (a) They had had all of
their school training in the Raleigh schools. (b) This fact assured
approximately the same course of study and instructional material.
(c) The pupils had all made normal progress in school; that is, they
had been neither retarded nor accelerated.
It should be pointed out that this sample of 230 pupils, therefore,
represents the whole school population of Grade III A. This fact is
of importance to this study, for, while in some of the comparisons
to be reported, the number of cases involved is necessarily small,
these cases really represent a much larger population.
B. THE MEASURES OF PHONETIC EXPERIENCE
The distribution of scores on the Teachers' Blanks.-The dis-
tribution of the scores on the Teachers' Blanks is given in Table 3.
Since it was supposed to be the policy in the Raleigh schools not
SIn the case of one such teacher, it was possible to obtain the necessary
data. Her pupils could therefore be retained.
16 Varied Amounts of Phonetic Training
DISTRIBUTION OF SCORES ON THE TEACHERS' BLANKS BY GRADES*
FREQUENCIES BY GRDES
Scores on the
Grade I Grade II Grade III All Grades
75-79...................... 0 I 1 2
70-74..................... 0 3 1 4
65-69..................... 2 3 0 5
60-64 ..................... 7 3 3 13
55-59...................... 3 0 3 6
50-54...................... 0 2 2 4
45-49...................... 1 2 1 4
40-44..................... 8 6 5 19
35-39...................... 0 1 2 3
30-34..................... 3 3 2 8
Total ............ 24 24 20 68**
*Table 3 should be interpreted as follows: No teacher in the first grade taught phonetics to the extent
represented by a score between 75 and 79. One teacher in Grade II and one teacher in Grade III taught
phonetics to this extent, etc.
**The total number of teachers noted here exceeds the actual number of teachers because some teachers
filled out more than one blank. There were actually 51 teachers. The teachers who filled out more than
one blank changed grades during the three years. This fact has no effect on the data because separate
measures were obtained for each grade taught.
to teach by phonetic methods, it is interesting to note the wide varia-
tion in the scores. The scores range from between 30 and 35 to
between 75 and 80. Thus, the scores represent a range of 45 points
out of a possible 75. It should be borne in mind that the extreme
upper range of possible scores, from 79 to 100. is not represented in
the distribution of scores.2
A consideration of the frequencies for all grades reveals the dis-
tinct bimodal character of the distribution. The difference between
the modes is 20, which represents a significant variation in the total
range of 45.
The meaning of the scores on the Teachers' Blanks.-It is im-
portant to note the facts represented in the scores on the Teachers'
Blanks. A teacher's reaction to a single item may be no index to that
teacher's general practice with respect to phonetics, but her reactions
to the sum of 25 such items very probably does represent her in-
structional procedure. Differences in teachers' scores may, there-
fore, be taken to indicate differences in their practices. In order to
illustrate the differences in meaning between two scores, the responses
of two teachers on ten items of the test are given in Table 4. Teacher
SThe fact that this extreme was not sampled was one of the reasons for
making the second investigation.
The Technique of the First Investigation 17
DIFFERENCES BETWEEN THE RESPONSES OF TEACHER F. L. AND TEACHER R. H.
ON TEN ITEMS OF THE TEACHERS' BLANK
THE TEACHER'S RESPONSE
The nature of the training F. L. R. IH.
involved in the item (Opponent of phonetics. (Advocate of phonetics.
Score: 35) Score: 75)
1. Phonetic training in relation Before any words have been learned
to sight vocabulary .......... None at all............... as sight words.
2. Ear training............... Practically never.......... Regularly.
3. Separate consonant sounds... None at all............... All of them.
4. Separate vowel sounds ..... None at all............... All the sounds of all the vowels.
5. Suffixes ...................... None at all............... All of them possible.
6. Prefixes ................... None at all.............. All of them possible.
7. "Families" of sounds ... .... Pay no attention to them .. Identify all words possible by this
8. The sounding of individuallet- None at all............... Regularly before the word is pro-
ters and combinations in new nounced the first time.
words found in reading......
9. Rules for pronunciation...... Pay no attention to them .. Teach a complete list and require
10. Teach the sounds of letters by
telling stories ............... Never................... Regularly.
F. L. received a score of 35, while teacher R. H. received a score of
75. It is apparent that F. L. gave little instruction in phonetics,
while R. H. gave a considerable amount in such training.
The distribution of Gross Phonetic Experience Scores.-Table
5 presents the distribution of Gross Phonetic Experience Scores for
the 230 pupils of Grade III A. Since each Gross Phonetic Expe-
rience Score is the sum of the scores of a pupil's teachers for each
half grade, it is, in the case of the III-A pupil, the sum of five scores
made by his teachers on the Teachers' Blank. The lowest possible
score a teacher could make was 25. Thus the lowest possible score
for five teachers is 125. Likewise, since the highest possible score
on the Teachers' Blank is 100, the highest possible Gross Phonetic
Experience Score is 500. The possible range of Gross Phonetic
Experience Scores is, therefore, from 125 to 500. The range of
scores for 230 pupils was found to be from 160 to 349. As was to
be expected from the scores on the Teachers' Blanks, the upper ranges
of the distribution which represent extremely large amounts of
18 Varied Amounts of Phonetic Training
phonetic experience contain no scores. Nevertheless, the distribution
indicates a considerable variation in amounts of phonetic experience.
C. THE TESTING PROGRAM
The group testing.-The group testing occupied the first few
days of the testing program. It consisted in giving the Otis Intelli-
gence Scale, Primary Examination: Form A;3 the Gates Silent Read-
ing Test: Types A, B, C, and D;4 and the Pressey Diagnostic Test:
Vocabulary-Grades 1 A-3 A.5 The tests were administered under
DISTRIBUTION OF GROSS PHONETIC EXPERIENCE SCORES
(III-A PUPILS IN RALEIGH)
Total .......... 230
M d............. 267
standard conditions by Dr. W. A. Brownell, the writer, and graduate
and senior members of classes in Experimental Education and
Educational Measurements of Duke University. The members of
these classes had been given full instruction with respect to the ad-
ministration of the tests.
The individual testing.-Pupils were then tested with a battery
of individual tests consisting of four types: (a) tests of phonetic
3 Yonkers-on Hudson, New York: World Book Company.
New York: Bureau of Publications, Teachers College, Columbia Uni-
5 Bloomington, Illinois: Public School Publishing Company.
The Technique of the First Investigation
lilllty. I ,i ;a test of word pronunciation ability, (c) tests of oral
reading .-iand i :) a test of eye-voice span.6
1 lie te:ti- of phonetic ability comprised four of the Tests for
Phonic Abdlines (or extensions of these tests) devised by Arthur I.
Gates.7 Test A4 is a test of the ability to translate printed phono-
grams into sounds. Test A5 is similar except that it combines two
phonograms. The tests are described by Gates as "visual stimulus"
tests.8 Tests B2 and B3, on the other hand, are tests of responses to
auditory stimuli. They yield measures of the ability to give letter
equivalents of sounds, B2 of single syllables, and B3 of combinations
of two syllables.9
The test of word pronunciation used was the Gates Graded Word
Pronunciation Test: Form I1.10 In this test, the child pronounces
as many as possible of a list of 100 words of increasing difficulty.
The Gray Oral Check Tests: Sets II and II11 were used to
measure oral reading abilities. The tests provide a useful analysis
of errors. The pupil reads a paragraph aloud, and errors of various
types are recorded. Set II was constructed for Grades II and III,
and Set III, for Grades IV and V. Each of these sets contains three
paragraphs of fifty words. Both sets were used in order to measure
both the upper and lower ranges of reading ability. The paragraphs
in Set III were given in order to make sure that the pupils faced
material representative of a new reading situation. In this study,
Set II and two paragraphs of Set III were given.
A test patterned after the Buswell Eye-Voice Span Test12 was
given in an attempt to determine whether or not phonetic training
SDetailed descriptions and methods of administering these tests are given
in: Agnew (2), Chapters VI, VII, and VIII. Samples of the tests together
with keys for scoring are given in the same place, Appendix B.
'Gates (16), pp. 380-388.
SIn order to make Test A6 more reliable, the number of items was in-
creased from seven to twenty.
Test B2 was modified by adding ten comparable items. This made twenty
items in all. Test B3 was increased from eight to fifteen items.
The reliabilities of the four tests were calculated by the split-half method
and the Spearman-Brown Prophecy Formula. The reliability coefficients (100
cases) are presented in the following table:
Coefficient of Coefficient of
Test Reliability Test Reliability
A4 .88 B2 .91
A5 .88 B3 .89
New York: Bureau of Publications, Teachers College, Columbia Uni-
Bloomington, Illinois: Public School Publishing Company.
2 Buswell (5), pp. 87-88.
20 Varied Amounts of Phonetic Training
shortens the eye-voice span as is sometimes claimed. This test pre-
sents a little story to be read aloud, in which words of the same spell-
ing occur as words of different meaning and pronunciation, e.g.,
"wind" from the verb "to wind," and "wind" as in "the wind blows."
In order to pronounce these words correctly, it is necessary to allow
the eyes to precede the voice so that the context may furnish a cue
to the correct pronunciation.
In all, eight different individual tests were given to more than
300 third-grade pupils. This task was accomplished during the two
weeks following the group testing. The testing was done by the
corps of testers mentioned above as having administered the group
tests. Care was taken to keep the conditions of testing uniform.
Children were given the tests in several sittings so that fatigue was
kept at a minimum.
THIE METHODS OF TREATING THE RESULTS
Sinre the methods of treating the results were somewhat com-
plex. it is ni::essary to devote a chapter to explanation. The pur-
po,:se i.'s tr: divide the cases into groups representing different
amount' o:f phonetic training and to compare these groups in terms
-if ti: re-ts -if reading abilities. Two general methods were used for
ihl, jtirprine. (a) comparison of groups based on the Gross Phonetic
Exerien.: Scores, and (b) comparisons of patterns of phonetic
experience. the different patterns representing different amounts of
Iriniinc. at dilerent times in the pupils' school experience.
.i. (C':IP\RixMi N, BASED ON THE GROSS PHONETIC EXPERIENCE SCORES
I hie cioml.arison of the scores on the tests of reading ability be-
twrren .pil,.iI who had experienced large amounts of phonetic train-
ing and pupils who had experienced little phonetic training was made
p[..sihle l-y a process involving three steps.
I i The Frst step was to select groups at the extremes of the
dlstriluti-in, of the Gross Phonetic Experience Scores.' In order to
do: this. irlitrary limits were set for the extremes. Thus, pupils
..lth 'Ir,.s Phonetic Experience Scores below 230 were included in
tl h:low groi:ij and pupils with Gross Phonetic Experience Scores
.li:' .' 2'e were included in the high group. This method yielded
SQ plu.[ils in the low group, 86 pupils in the high group, and omitted
the 5,5 I tile middle of the distribution.
12 The next step was to equate these extreme groups on the
Ia..is of measures of intelligence. This was done by pairing pupils
in term- -f M.A. and I.Q. Cases that could not be suitably paired
w.ere ,di rc.garled. By this means, two distributions were obtained.
eaich o-.iaining 43 individual scores. The distributions of M.A.'s
and I l.'s ar<: given in Table 6. Hereafter, in this report, the group
.Aith the hih Gross Phonetic Experience Scores is called Group GH.
and tlhe rori:i with the low Gross Phonetic Experience Scores is
cillerd group G'L. The G's indicate that the groups are based on the
ro-:.ss Phi'n .tii: Experience Scores, and the subscripts indicate high
,nal I.:.'. amounts of phonetic experience.
5e Tahle 5
22 Varied Amounts of Phonetic Training
DISTRIBUTIONS OF M.A.'s AND I.Q.'s FOR GROUPS GJI AND GL*
M.A. in Terms I.Q.
of Months Group GH Group GL Group GH G-c..f -.L
130-134........... 1 1
125-129 ............ 1 1 125-129........ 1 I
120-124............ 8 8 120-124........ 3
115-119........... 8 8 115-119........ 4
110-114........... 12 12 110-114......... 12 14
105-109 ............ 4 4 105-109........ 6 4
100-104........... 3 3 100-104........ 5 1
95-99 ........... 4 4 95-99 ........ 7 5
90-94 ........... 0 0 90-94 ........ 3
85-89 ........... 1 1 85-89 ........ 0
80-84 ........... 1 1 80-84 ........ 2
M ean............ 112.04 112.04 ................ 107.50 I' ,'
................ 10.00 10.00 ................ 10.15 I1 'i
*Group GH is made up of pupils with high gross Phonetic Experience Scores, and GL, of ,.,r .I T-.Ih
low gross Phonetic Experience Scores.
(3) The scores of these groups on the tests of reading :l.ihtv
were compared in terms of means, P.E.'s of means, and critical ra-
tios. The results are presented in Table 10 (Chapter V).
Limitation of the method of comparison.-The gross score, g*'.
equal value to phonetic experiences at different grade levels ThiL
effect may not be what is wanted; possibly the method may ',il,
measures so coarse that they obscure real and important diff. r nlce-
There may be, for example, a critical point in the child's edu.:ati-n.al
experience that is particularly propitious for effective phonetic in-
struction. A child who had very little phonetic instruction in 'Irli.
II, but large amounts of phonetic experience in Grades I anil III
might be placed in the phonetic group. But if he had had little ex-
perience with phonetics in Grade II, the critical time for su.li I;n-
struction, this fact would be obscured if the gross scores alone .c:re
considered. In order to bring out the effects of phonetic instructi.o:l
at different grade levels, another method of treating the data .'as
necessary. This method is outlined in the following paragraph
B. COMPARISONS OF PATTERNS OF PHONETIC EXPERIENCE
Method used to isolate the factor of the time at which rr-ir;.ls
obtained phonetic experience.-As has been brought out in the pre-
vious discussion, the object of the investigation was to isolate not
The Methods of Treating the Results 23
only the factor of gross amounts of phonetic experience, but also the
factor of the time at which this experience had been obtained. In
these comparisons, amounts of phonetic experience at different levels
had to be identified. A method involving three steps was devised to
accomplish these ends. According to this method, (1) each grade
was divided into "phonetic," "medium," and "nonphonetic" groups;
(2) the "medium" group was omitted; and (3) the "phonetic" and
the "nonphonetic" groups were equated with respect to the sum of
the phonetic experience scores for the other two grades. The re-
sultant groups represented a pair of patterns of phonetic experience.
In order to illustrate this process, the derivation of one pair of such
patterns will be described in detail.2
The derivation of patterns Ap and An.-In order to measure the
effects of varying amounts of phonetic experience in Grade I, the
amounts in Grades II and III had to remain constant. There were
three major steps in the process of deriving the pair of patterns rep-
resenting the phonetic and nonphonetic groups in Grade I. (1) The
first step was to make a distribution of the phonetic experience scores
of pupils in Grade I as shown in Table 7.
(2) Since only the extremes are important for purposes of com-
parison, the second step was to choose arbitrary limits for the
phonetic and nonphonetic groups and to disregard a group in the
middle part of the distribution. Eighty cases with scores above 120
were selected to make up the phonetic group, and 111 cases with
scores below 90, the nonphonetic group. The 39 cases between 90
and 119 were disregarded. At this stage no attention was paid to
amounts of phonetic training above the first grade.
(3) The third step requires considerable explanation. The di-
vision into phonetic and nonphonetic groups (step 2) was made
purely on the basis of the Phonetic Experience Scores for Grade I.
2Three other methods were thought of in connection with the isolation
of these variables, but these methods were found to be impractical. (1) One
method of isolating the factors, amount and time, is to consider the variable
amount, in terms of large, medium, and small amounts, and to consider the
variable time in terms of the three grade divisions. If these variables were
broken down into their possible combinations it would yield twenty-seven
groups. If there were 230 cases in all, the average number of cases in each
group would be about eight. It is obvious that a comparison of such small
groups would not yield reliable results. (2) Another method might be to
consider the twenty-seven categories mentioned above as an unordered series
and correlate it with the scores on the tests by means of the formula for ?.
[See Holzinger (25), pp. 266-277.] Interpretation of the results obtained
by this method was found to be very difficult. (3) The method of partial cor-
relation, in which amounts of phonetic experience are partialed out for each
grade, was not used because the distributions of variables were by no means
normal. [See Agnew (2), pp. 69-70.]
24 Varied Amounts of Phonetic Training
FREQUENCY DISTRIBUTION OF PHONETIC EXPERIENCE
SCORES OF PUPILS IN GRADE I
90-99 ................. 2
80-89 ................. 4
70-79 ................. 65
60-69 ................. 40
50-59 ................. 2
In order to keep constant the amounts of phonetic training in Grades
II and III, the two groups, phonetic and nonphonetic, were equated
in terms of their phonetic experience scores in these two grades.
Two distributions were made of the scores for Grades II and III
combined, one representing the phonetic group in Grade I and the
other the nonphonetic group in Grade I. These distributions are
shown as "initial distributions" in Table 8. Two new distributions
were then obtained by elimination from one group or the other until
DISTRIBUTIONS OF PHONETIC EXPERIENCE SCORES FOR GRADES II AND III
COMBINED BEFORE AND AFTER EQUATING
Scores Initial distributions Equated distributions
Ap An Ap An
200-209 .................. 14 8 8 8
190-199................. .. 15 1 4 1
180-189................... 6 1 6 I
170-179................... 2 23 2 22
160-169................... 14 3 11 2
150-159................... 14 9 12 7
140-149 ................ 2 0 1 0
130-139 ....... ........... 1 2 1 2
120-129.................. 11 13 7 9
111019 .................. 0 15 0 0
100-109 ................... 1 24 0 0
90-99 .................. 0 11 0 0
80-89 .... ............ 0 I 0 0
N ............... ....... 80 111 52 52
M ean........ ....... . 167.1 166.9
a ....................... 26 .1 25.5
The Methods of Treating the Results
nearly equivalent means and standard deviations were secured. The
final distributions are also shown in Table 8 under the heading
For the sake of ease of expression and reference, the phonetic
group in Grade I was called pattern Ap and the nonphonetic group
An, the subscripts being abbreviations of "phonetic" and "non-
phonetic." Thus pattern Ap means that the group of children so
identified had had a larger amount of phonetic instruction in Grade
I than had An, while both groups had obtained an equal amount of
training in Grades II and III.
Other patterns.-In all, six pairs of patterns were isolated in this
manner. In each case, procedures corresponding to those described
in respect to the derivation of patterns Ap and An were followed.
An analysis of Table 9 indicates in what grades the phonetic experi-
ence was varied and in what grades it was kept constant in the dif-
DISTRIBUTIONS OF EQUATED GROUPS*
Pattern Limits Grades in which (amount of pho- N Mean Standard
phonetics varied netics equated) Deviation
An........ 90 and below III and III 52 167.1 26.1
Ap........ 120andabove III and III 52 166.9 25.5
Bn........ 90 and below II I and III 50 136.2 22.1
Bp........ 120 and above II I and III 50 136.2 21.2
Cn........ 42 and below III I and II 45 228.3 18.7
Cp........ 50 and above III I and II 45 229.3 20.0
Dn........ 130 and below II and III I 56 90.5 20.8
Dp... 165 and above II and III I 56 90.4 20.5
En........ 129 and below I and III II 53 107.9 26.3
Ep........ 165 and above I and III II 53 108.4 27.3
F ........ 195 and below I and II III 47 40.7 7.6
Fp........ 230 and above I and II 11I 47 40.9 7.4
*Table 9 may be read as follows: Pattern An consists of all the cases having a phonetic score below 90
in the first grade that could be equated with cases in pattern Ap which consists of those who had a phonetics
score in the first grade of more than 120. Equating was done on the basis of the sum of the phonetics
scores for Grades II and III. The number of cases in An (and Ap) is 52, the mean is 167.1, and the standard
leviation is 26.1.
Degree to which equating was possible.-Table 9 shows the de-
gree to which equating was possible. For instance, in the case of pat-
terns Ap and An, there is only .2 difference between the means of the
two distributions, and there is only .6 difference between the standard
deviations. For patterns Bp and in,' tlhe difference .between the
,I : J ,
Varied Amounts of Phonetic Training
means is .0 and that between the standard deviations .9. For pat-
terns Cp and Cn, the difference between the means is 1.0 and be-
tween the standard deviations 1.3. Thus, it is evident that the sets
of scores for each pair of patterns were equated closely.
Defense of the derivation of patterns.-It should be pointed out
in this connection that the groups so equated were relatively small.
However, it is to be borne in mind that the small size of the groups
is more apparent than real. The description of the method employed
in selecting these cases has emphasized the similarity of the subjects
with respect to chronological age, level in school achievement, and
similarity of instruction. That is, the subjects employed possess a
really unusual degree of homogeneity. When groups of from 45 to
56 are compared in this study, therefore, they represent. in reality.
much larger populations.
Comparison of groups and patterns in terms of the scores on the
tests of reading ability.-The groups GH and GN were compared
by calculating the differences between the means of the test scores
of reading abilities. Since the groups had been equated in terms of
intelligence, the following formula was used :
P-E.diff= J_- + ( I -r) .674
Differences between each pair of patterns (in terms of the differ-
ence between the means of the scores of the various tests) were simi-
larly calculated except that in these cases the usual formula for the
critical ratio was used.
SSee E. F. Lindquist, "The Significance of a Difference Between Matched
Groups." Journal of Educational Psychology, XXII (March, 1931), 197-204.
RESULTS AND CONCLUSIONS OF THE RALEIGH
The results of the Raleigh investigation are presented in Table
10.1 Before these results are analyzed, it is thought wise to consider
certain factors that might have influenced the test scores.
SUMMARY OF TIE DIFFERENCES IN TERMS OF THE P. E. OF THE DIFFERENCES*
GROUPs AND PATTERNS
Tests G A B C D E F
(1) (11) (111) (11-111) (1-111) (1-11)*
Gates A4........................ 1.66 .42 2.15 .71 1.68 4.56 2.53
Gates A5........................ .84 1.12 .52 1.36 1.15 .34 3.69
Gates B2........................ 1.22 .21 .34 1.25 .96 1.88 3.97
Gates B3........................ 3.02 1.27 .75 -1.88 .26 6.41 -5.74
Gates Word Pronunciation ........ -2.40 .85 1.21 1.79 2.30 1.98 -3.72
Type A........................ -2.50 1.68 .89 1.22 1.82 1.18 -8.49
Type B.........................-2.73 -- .04 .75 .15 -3.17 1.98 -5.72
Type C ......................... -1.92 -1.40 .61 .16 .49 2.84 -6.35
Type D........................ .87 1.23 .39 .75 -1.35 -2.83 -6.63
Pressey Vocabulary .............. .12 .26 .90 2.03 -1.33 -8.25 3.73
Gray II (errors)................ .30 -1.16 1.02 .91 .15 1.32 -3.54
Gray IIl (errors)................ 1.43 1.16 2.10 .48 1.40 2.55 -5.13
Gray 11 (time).................. .32 .80 2.46 .00 1.60 1.50 -3.20
Gray 111 (time) ................. .85 .33 2.46 .02 2.11 4.52 -5.17
Eye-voice Span................... 1.68 1.22 .89 1.95 .09 3.54 -3.53
M ean............................- .08 .00 .65 .33 .18 1.56 4.17
*Negative differences favor the nonphonetic groups.
**Numbers in this row represent grades of variation.
Factors other than phonetic experience that might influence the
test results.-A list of the chief factors other than phonetic training
that might influence the test scores is given below:
1. Lack of homogeneity of subjects in age, school experience, etc.
'A complete exposition of the basic data (except those on word pronun-
ciation) including the distributions of scores, the actual means and standard
deviations, is given in: Agnew (2), Chapters VI, VII, and VIII. A similar
exposition of the data on word pronunciation is given in: Agnew (1), Chap-
ters III, IV, and V.
28 Varied Amounts of Phonetic Training
2. Biased sampling of schools and teachers.
3. Differences in the intelligence of the groups compared.
4. Differences in methods of teaching.
5. Differences in instructional material.
6. Differences in the motivation of reading.
7. Other unknown factors.
The factors listed above are briefly discussed in the following para-
1. The previous discussion has emphasized the homogeneity of
the subjects.2 It will be recalled that all the pupils selected had made
normal progress in school, were in the same grade, and lived in the
same city where they had received all of their school experience.
2. The pupils used in the study came from nine different schools
(all the schools for white children in the city). It is interesting to
point out in this connection that each pattern or group was composed
of pupils from at least four different schools, and in many cases, from
as many as six or seven schools. This fact insures a wide sampling
of teachers, so that the chances are against the differences being
due to the peculiar excellences or deficiencies of the teachers.
3. Table 11 presents the mean mental ages of the pupils repre-
sented by the patterns, together with the standard deviations of the
means. The largest difference is in the case of the difference be-
tween the patterns Bp and Bn. This represents an apparent differ-
ence of five months. The difference is just great enough to be re-
liable, that is, it is assured that it is greater than 0. The correlation
MEANS AND STANDARD DEVIATIONS OF THE M.A.'s FOR THE VARIOUS PATTERNS
Pattern Mean Standard deviation
Ap 108.0 10.15
An 111.9 10.20
Bp 115.0 9.85
Bn 110.0 11.95
Cp 111.6 9.05
Cn 115.6 10.85
Dp 115.5 8.45
Dn 111.8 10.30
Ep 112.8 8.75
En 115.5 11.55
Fp 117.9 10.05
Fn 117.8 10.50
2 See Chapter III.
Results and Conclusions of the Raleigh Investigation
between intelligence and the test scores would have to be rather high
for this small difference to play an important role in causing differ-
ences in the test scores. There is some evidence that this correlation
is low. Table 12 shows the coefficients of correlation between the
test scores and the intelligence of the matched groups GH and GL.
Since these coefficients were really obtained by correlating the test
scores of the two groups matched for intelligence, the true correla-
CORRELATIONS BETWEEN TEST SCORES OF GROUPS GH AND GL
(GROUPS MATCHED FOR M.A. AND I.Q.)
Tests r Tests r
Gates A4....................... .06 Gates Types A, B, C, and D..... .16
Gates AS....................... -- .11 Pressey Vocabulary ............. .08
Gates B2 ....................... .06 Gray II, errors................. .13
Gates B3.................. ..... .21 Gray III, errors................ .07
Gates Type A .................. .06 Gray II, time .................. .09
Gates Type B .................. .20 Gray III, time.................. .07
Gates Type C .................. .30 Eye-voice Span ................ .20
Gates Type D.................. .02
tions between the scores and intelligence may be somewhat larger. It
is doubtful, however, that the correlations would be large enough to
modify appreciably the reliability of the differences.
It is not proposed to rule out intelligence as a possible factor in
influencing test scores, but it is suggested that it is unlikely that the
small differences in intelligence found between the patterns could
have been responsible, to any great extent, for the differences be-
tween the means of the measures of reading ability.
4. There is a lack of information with regard to the methods of
teaching reading, other than the degree to which it was phonetic or
nonphonetic. The methods, however, were under the same super-
vision throughout the schools; and there is no reason to believe that
the methods were causative in producing the differences in the means
of the reading tests.
5. Since all the schools were in the same system, there is no rea-
son to believe that the instructional materials varied significantly, or
if so, that they tended to favor one pattern or group more than an-
6. Nothing is known about differences in motivation or other
7. It is assumed that "unknown factors" operate equally upon all
30 Varied Amounts of Phonetic Taiii:
The influence of phonetic experience on phon'tic ailities -The
tests of phonetic abilities, Gates A4, AS, B2, and EB have been
shown to have high reliability coefficients (see Chaptcr 11. noiti 91.
They test phonetic abilities as these are measured b, tihe ielliI arid
pronunciation of nonsense letter groups. The d iere'ncre that are
statistically reliable, as shown in Table 10, are c:niii-tiet :or each
pair of patterns, as far as the four tests are coreornrel. bunt inoi:i-
sistent as between patterns. Thus, the two reliable d'iffercijc:- be-
tween the E-patterns favor the phonetic group. (,n tli .l.-hler haind
the differences between the F-patterns, although siinIarl coni.istent
with each other, favor the nonphonetic group. If thl E-piattern- are
taken alone, one would be led to conclude that plilncietic training in
Grades I and III results in higher scores on the Gal',- Tct;s .-\4 and
B3. If the F-patterns are considered alone, it might be o':niclud.ed
that, when little phonetic training is given in Grades I and II. hiiher
scores are obtained on the Gates Test A5. B2, B3. anl [:.-.-.1 .4 A4-
In the latter instance it might have been concluded that [Ipupll I, larnied
phonetic abilities without having had formal train: in ph l)netiic;.
While these facts may actually represent the situation, the appIarent
inconsistency between the results of the E- and F-I:attrni caLt'
doubt on the significance of the differences obtained.
The influence of phonetic training on word p-' ..!'itr-ncml. -The
comparison of Gp with Gn on the Gates Word P'rr.unrc:at,:,iin Tetc
reveals a small difference in favor of the nonphonetl grI:U. co.iii-
parison of the patterns reveals: (a) phonetic traiinlni itn 'ra.de I
seems to have a slight detrimental effect on word prcuincaiti,,r abil-
ity; (b) phonetic training in Grade II seems to ru-ilt in a slight
increase of ability to pronounce words; (c) phljoitti trai!iill II
Grade III has a slightly greater tendency to incrca-i al.Dili.t to. pro-
nounce words; (d) the differences show a general lack '-, rcalbility.
The influence of phonetic training on silent 'ciiinll ,I!i''ft- -
In the comparison of the patterns in terms of tl:- res-Ilt- fronm ilte
Gates Silent Reading Tests, Types A, B. C and D. live ,ditTerienicsi
appear that are greater than three times the P. E. uf tlhei df-rcntce.
All of these favor the nonphonetic groups, one favo:iinme i..ttiern Dn.
and four, pattern Fn. The difference in the case -,I- tlic D-pane[in
(for Type B) may not be significant because it is iic:t ullpported -1.
the other measures of silent reading ability (TypeA .\ iandi Di.
but the large and consistent differences favoring' tlc iinl.ihounetcI
group in the comparisons of the F-patterns appear to .- Inur mr -ig-
nificant. If this is the case, it may be concluded that there i- s,,um
Results and Conclusions of the Raleigh Investigation
c i iJlce that large amounts of phonetic training in Grades I and II
are not so advantageous to silent reading abilities (as measured by
these tests) as are small amounts of phonetic training in these grades.
The influence of phonetic training on vocabulary.-In the com-
parisons of the patterns on vocabulary attainment, as measured by
the Pressey test, one large difference appears to favor the non-
phonetic group in the E-patterns, and one fairly large difference ap-
pears to favor the phonetic group in the F-patterns. It is possible
that the phonetic group in pattern Ep was so accustomed to the
phonetic attack on unfamiliar words that much time was taken in
analyzing the nonsense words of the Pressey test. If this is true,
the test may not have measured the actual vocabulary of the pupils
in this group. However, if this were the case, there should be some
evidence of this phenomenon in the comparisons of the C-patterns
in which phonetic training in the third grade is isolated. Unreliable
differences between the C-patterns favor the phonetic group. The
C-patterns and the F-patterns are thus seen to be inconsistent with
the E-patterns. This inconsistency tends to reflect doubt on the sig-
nificance of the differences obtained.
The Iflii i,, of phonetic training on oral reading.-In the com-
parisons of the patterns with respect to speed and accuracy on
the Gray Oral Reading Check Tests (Gray II and III), there are
five reliable differences. One, in the E-patterns, favors the phonetic
group in the speed of reading Set III. The fact that the other dif-
ferences between the E-patterns on both speed and accuracy are un-
reliable casts doubt on the significance of the one difference found.
Reliable differences are found to favor Fn as opposed to Fp in both
the measures of speed and accuracy. These differences alone would
indicate a superiority on the part of nonphonetic training in Grades
I ;nd II in speed and accuracy on the Gray tests. However, no
thler comparisons of patterns bear out these conclusions, and again
the differences seem to be of doubtful significance.
The influence of phonetic training on eye-voice span.-The test
.-if eye-voice span is of questionable validity because it was neces-
i-rily short and seemed, in many cases, to involve too difficult read-
ini material. On the other hand, although the instrument was some-
...iat crude, it was hoped that it might show extreme differences in
'.\r-voice span (if they existed). The results, in no case, present
lifferences as great as four times the P. E. of the difference.3 Small
'The difference must be at least four times the P. E. of the difference in
order to insure complete reliability. See Garrett (14), p. 136.
Varied Amounts of Phonetic Training
differences appear within the E-patterns and within the F-patterns.
These differences are in opposite directions so that they are of ques-
The influence of phonetic training on the battery of tests as a
whole.-Table 10 presents (in the last row) an average of all the
differences. This is, of course, a crude method of summary because
it is impossible to say how the tests should be weighted. Also, any
isolated reliable differences are obscured by being averaged with dif-
ferences of small reliability. Yet these averages may aid in bring-
ing out some general characteristics of the table as a whole.
1. Perhaps the most striking feature of these averages, and, in
fact, of the comparisons as a whole, is the paucity of reliable differ-
ences. (a) Not a single difference as great as four times the P. E.
of the difference appears in the comparison of Groups GH and GL.
Only one difference (Gates B3) is as great as three times the P. E.
of the difference. (b) In the comparisons of the phonetic and non-
phonetic groups in the patterns A, B, and C (in which phonetic ex-
perience varied in one grade), only five differences are as great as
three times the P. E. of the difference. (c) In the D-patterns (in
which phonetic training varied in Grades II and III), only one dif-
ference is as great as three times the P. E. of the difference. (d)
In the E-patterns (in which phonetic training varied in Grades I
and III), five differences out of the fourteen are as great as three
times the P. E. of the difference; and, of those, four favor the
phonetic group, and one favors the nonphonetic group. Thus, in the
above comparison there is no consistent evidence that the differences
in phonetic training measured in the Raleigh study affected the test
2. Only in the case of the comparisons of the F-patterns (in
which phonetic training varied in Grades I and II), do differences
appear consistently reliable. (a) With but one exception, the differ-
ences are as great as three times the P. E. of the difference. (b)
All but one of these differences (that in vocabulary scores) favors
the nonplionetic groups. This is true even in the phonetic tests. (c)
The average of these differences is 4.17 P. E. It would seem that
the comparison of the F-pattern in terms of the battery of reading
tests presents rather consistent evidence that phonetic experience (in
the first two grades) is not so beneficial to reading abilities, as meas-
ured by the tests, as is nonphonetic experience. How great this ben-
Results and Conclusions of the Raleigh Investigation 33
efit of nonphonetic training may be, however, is not answered from
3. The inner consistency of the differences found in the E-pat-
terns suggests that these differences have some significance. If the
difference favoring the nonphonetic group on the Pressey Vocab-
ulary Test is omitted and the average of the other differences com-
puted, the resultant mean is found to be 2.35. This difference seems
to indicate a fairly adequate advantage on the part of the phonetic
Various interpretations that may be made of the inconsistencies
between the directions of the differences in the E- and F-patterns
are mentioned in the last paragraphs of this chapter.
General conclusions of the Raleigh investigation.-The conclu-
sions suggested as a result of the Raleigh investigation may be sum-
marized as follows:
1. The comparisons made failed to reveal a significant advantage
or disadvantage (in terms of reading test scores) arising from dif-
ferent amounts of phonetic experience as measured by the Gross
Phonetic Experience Scores.
2. The effort to find a critical grade in which phonetic experience
is particularly effective for training in reading was unsuccessful.
3. There seems to be a tendency for large amounts of phonetic
experience in Giades I and II (as is indicated in the F-patterns)
to affect the reading abilities adversely.
SA search was made for factors that might account for the differences
between patterns Fp and Fn.
(a) It was observed, as is indicated in Table 11, that the average M. A.'s
of these two patterns were practically identical. The same is true of the
standard deviations. The range of M. A.'s in 'both patterns was found to be
approximately from 90 to 140. Both groups were somewhat above the average
in M. A. It is difficult to say what significance (if any) this fact may have.
It is possible that more intelligent children tend to be inhibited by phonetic
(b) It was thought that one or both of these groups might happen to be
composed of pupils who had been in particular schools or who had had par-
ticular teachers. No facts were disclosed to indicate that any other factor
than phonetic experience operated in the selection of groups. The table below
shows the distribution of the groups of schools.
Schools F FSchools F
Fp Fn Fp F,
Hayes Barton ........ 19 16 Thompson.......... 0 5
Boyland Heights...... 11 4 Lewis.............. 2 0
Olds ................. 0 16 Pool............... 1 0
Murphy............. 8 6 Wiley.............. 6 0
Total............ 47 47
34 Varied Amounts of Phonetic Training
4. The direction of the differences found in the E-patterns sug-
gests that large amounts of phonetic experience in Grades I and III
are beneficial to most of the reading abilities measured.
Interpretation.-These conclusions may have a number of inter-
pretations, the most reasonable of which appear to be the following:
1. The measures of phonetic experience may have been faulty.
2. In general, the differences between the amounts of phonetic
experience as between groups or between members of patterns may
not have been great enough to affect measurable differences in the
reading test scores.
3. The inconsistency between the E- and F-patterns may be due
to unknown factors in selection or training. It is perfectly possible,
however, that the inconsistency is due to the actual differences in
the patterning of the phonetic training. The differences in scores
may be the effect of the interference of one type of training with an-
other. Thus, phonetic training in the first two grades may be in-
effective if, in the third grade, other types of training are stressed.
The training in the third grade (immediately before the testing)
might tend to interfere with the earlier training and thus cause the
relatively low scores observed in pattern Fp. On the other hand,
since pattern Fn represents a fairly consistent nonphonetic training,
the interference would be less and the scores higher, as was actually
In pattern Ep, the interference effect would not be so apparent,
since the phonetic training in the third grade would tend to coun-
teract any interference effect that might have occurred in the sec-
ond grade. If this is true, the data suggest that consistent phonetic
training might have beneficial effects on the abilities measured.
In view of the high reliability and the facts presented previously
concerning the validity of the Teachers' Blank, it seems reasonable to
assume that the measures of phonetic experience were not at fault.
The second and third of the four explanations given above seem
worthy of further investigation. The study to be reported in the
subsequent chapters was undertaken to determine the effects of con-
siderably larger and more consistent amounts of phonetic experience
on the scores of the same battery of tests. In this way, it was an-
ticipated that the explanations of the results of the Raleigh investi-
gation might be checked.
THE -'Ur:POSE AND TECHNIQUE OF THE DURHAM
'llic .ic..id investigation was undertaken (1) in order to check
the re-uli ..Il..tained in the Raleigh investigation, and (2), in order
: to ir,-,'.:. nw data on the effects of larger and more consistent
ami.units- ,f i-honetic experience than those found in Raleigh.
It % ill Bc recalled that the upper ranges of the possible scores on
til T.-:achl,-r' Blanks were not sampled in Raleigh. Table 3 presents
tle: .ijitrib.utii.n of teachers' scores in Raleigh and shows the highest
.Crt_- : for the various grades to have been: for Grade I, between
'd. a:l r:,9: and for Grades II and III, between 69 and 75. Since the
liltrsl I "'-.-.ile scores on the Teachers' Blank is 100, a possible
rain:l,: -f st,.res consisting of some 20 points at the upper range of
thie ,ltribLutr..n was not sampled in Raleigh.
i-iurilih-rnii.re, when the Gross Phonetic Experience Scores are
.'.n:sierelj, I ee Table 5), it will be observed that the upper limit
i. ic.'-rc I f-ll in the interval 340-349. The possible upper limit of
Ili;ic ,:c.r- is 500. Thus, a possible range of 150 points was un-
;saln[pled I- the Raleigh data. A search was made, therefore, for
ilr.l-.r.lde Ipupils whose phonetic experience might be great enough
i.. ;amrle this upper range. It having been the policy during the
IrcJdnIig thrce years (1931-35) to teach large amounts of phonetics
n Durliani, that city was selected for the second investigation.
B. THE TECHNIQUE OF THE INVESTIGATION
L.i,''r nation of amounts of phonetic experience.-Teachers'
ClanL.- ..>-r- -ubmitted to the first-, second-, and third-grade teachers
[i a inmbil-r of the Durham elementary schools. The blanks were
-.,.rol a- c' Ire those used in Raleigh.
F-i.u.lh I-.timates of the pupils' phonetic experience were made by
adding t,:.etlier the scores of trios of first-, second-, and third-grade
irachciir .\.t er the schools had been selected (in which there seemed
i, Ile thle ,-rcatest likelihood of high scores) accurate records were
made .-i tile pupils' phonetic experience, as was done in the Raleigh
36 Varied Amounts of Phonetic Training
Selection of schools.-Two schools were selected for investiga-
tion. In the selection of these schools, two principles were borne in
mind: first, it was desirable to obtain subjects who had experienced
large amounts of phonetic training; and second, it was desirable to
obtain a distribution of subjects comparable to those used in the
Raleigh investigation. The North Durham and Edgemont schools
were finally selected as best meeting these criteria. Both of these
schools had consistently emphasized phonetic instruction during the
training of the pupils who were in the third grade at the time of the
investigation. Neither school represents an extreme in the eco-
nomic and cultural levels of the city. The Edgemont School derives
a majority of its population from the homes of mill workers, people
in small business concerns, etc. The population is probably some-
what below the average for the city in cultural advantages. The
pupils of the North Durham School come, in general, from more
advantageous home environments, perhaps slightly above the average
for the city. Thus the schools furnished subjects not significantly
different from, and certainly not superior to, the subjects of the Ra-
Determination of amounts of phonetic training.-An analysis of
the school records revealed the fact that, out of a third-grade popu-
lation of about 200, there were 110 pupils who had made regular
progress through the grades, and who had received all three years of
their school training in these selected schools. Since there had been
no changes in the teaching personnel during these years, and since
Teachers' Blanks (covering the time when the pupils had been
taught by the particular teachers) had been secured from all the
first-, second-, and third-grade teachers, it was possible to compute
the amounts of phonetic experience for each pupil in terms of a
Gross Phonetic Experience Score.
Table 13 presents a frequency distribution of the Gross Phonetic
Experience Scores of the 110 selected pupils in Durham. It is ap-
parent that the pupils had received consistently large amounts of
phonetic training. When it is recalled that, in Raleigh, no Gross
Phonetic Experience Scores exceeded 350, and only a scattered few
exceeded 320 (see Table 5), it is clear that these Durham pupils
had received considerably more phonetic instruction than had the
pupils in Raleigh.
Administration of the tests.-The same battery of tests that had
been previously administered in Raleigh was given in the selected
Durham schools. A class in Educational Measurements from Duke
7Th Purpose and Technique of the Durham Investigation 37
FREQUENCY DISTRIBUTION OF GROSS PHONETIC EXPERIENCE SCORES
(110 CASES, DURHAM)
390-399 .................. 20
University, composed of seniors and graduate students, administered
the tests. The administrators were carefully instructed in the tech-
nique of giving the tests in order that the conditions of testing that
had been set up in Raleigh might be duplicated as nearly as possible.
The group tests, the intelligence test, the vocabulary test, and the
tests of silent reading abilities were given to all the third-grade pupils.
The individual tests of phonetic abilities, eye-voice span, oral read-
ing and word pronunciation were given to the 110 selected pupils.
Treatment of the results.-It will be recalled that the primary
purpose of the Durham investigation was to provide data on the
reading abilities of pupils who had received large consistent amounts
,-f phonetic training, in order that these data might be compared with
160 180 200 220 240 260 280 300 320 340 360 380 400
It Cs----< Durham I
FIG. 1. Histogram showing the frequency of the Gross Phonetic Expe-
rience Scores from which groups were selected in the comparison of the
Raleigh and Durham data.
38 Varied A-moiu, is of F'hn;,'c Ti inii,.
the data obtained in Raleigh, where the amounts of phon,:tii tra.i'inc
had been smaller and less consistent. In order to rr.al;k tlih croup-
compared represent similar ranges of intelligence, puipl t'i ol Io
Phonetic Experience Scores in Raleigh were paired \v.uh Duriha;n
pupils (whose Phonetic Experience Scores were high ) .,:. tl bi.,-i-~
of M.A.'s and nearly equivalent I.Q.'s. A detailed ac:ourt ,'-. tl!ii
Figure 1 indicates graphically the distribution of Phlho:utic Ex-
perience Scores from which the pairs were drawn.
All the Raleigh cases had Phonetic Experience Sc.irc,- lithat f1.ll in
the range from 160 to 270, and all the selected Durhamii caT:e- iha
Phonetic Experience Scores ranging from 370 to 400.'
By the method of pairing, 89 cases were selected fri:.m hil,: R;,leili
group, and a similar number from the Durham gronI Tal-. 14
presents the distributions of the M.A.'s and I.Q.'s of the two selected
groups. Most pairs had identical M.A.'s, and in no case did the
M.A.'s of a pair differ more than two months. Care was exercised
FREQUENCY DISTRIBUTIONS OF M.A.'s AND I.Q.'s OF THE RALEIGH AND
DURHAM GROUPS SELECTED BY PAIRING
M.A. in Terms I.Q.
of Months Raleigh Durham Raleigh Durham
140-149.......... 1 1
130-139.......... 7 7 130-139........ 1 1
120-129.......... 27 27 120-129........ 18 15
110-119.......... 37 37 110-119........ 37 37
100-109.......... 12 12 100-109........ 20 22
90-99 .......... 5 5 90-99 ........ 10 11
80-89 ........ 2 2
70-79 ........ I 1
Total............ 89 89 ............... 89 89
Mean ............ 117 117 ................ 112 111
'The two groups differ, not only in terms of gross amounts of phone-',:
experience, but in terms of the patterns of that experience. In the Raleigli
group, the training, although fairly consistently nonphonetic, in some instance:
varied considerably in individual grades. Thus a pupil in Grade I might hast
a score of 50 or 140 and still be included in the group. Similar variation,
sometimes occurred in other grades. It is possible, therefore, that the inter-
ference factors, which may have operated to lower the reading scores in pat-
tern Fp, operated in some cases to lower the reading scores of the Raleigh
The Durham group, on the other hand, represents highly consistent phonetic
training for all three grades. Thus, in this group, interference was probabi,
not a factor.
The Purpose and Technique of the Durham Investigation 39
in order to avoid wide discrepancies between the I.Q.'s of the mem-
bers of a pair. In no case did the I.Q.'s of a pair differ more than
In order to express simply the nature of the groups, Table 15 has
been prepared. A glance at the table will reveal the constants and
the variables in the known conditions of the two groups.
FACTORS IN THE COMPOSITION OF TIE DURHAM AND RALEIGH GROUPS
Conditions Raleigh Durham
Number of cases.............. 89 89
Number of years in school...... 2.6-2.7 2.6-2.7
Intelligence................... Equated by pairing with Durham Equated by pairing with Raleigh
Time spent on reading (daily)*
Grade I...................... 40 to 70 minutes ............... 60 to I00 minutes
Grade II ..................... 60 to 100 minutes.............. 60 to 95 minutes
Grade III.................... 60 to 80 minutes.............. 60 to 80 minutes
Number schools represented.... 7 2
Number teachers represented... 32 10
Class organization ............. Home Room................... Platoon
Supervisor's attitude toward Opposed to direct phonetic Favored considerable phonetic
phonetics .................. teaching.................... teaching
Range of Phonetic Experience
Scores .................... 160-270 360-400
*Time devoted to reading in Raleigh.-The basis of the estimates of the time spent in reading in the
Raleigh primary grades was an analysis of the sample daily programs given in Curriculum Bulletins issued
by the Raleigh Public Schools. The teachers, whose samples are given, were all teachers who filled out
blanks in the Raleigh investigation.
Time devoted to reading in Durham.-The estimates of time spent in reading activities in the Durham
schools were obtained through the kindness of Mrs. Robinson, Supervisor of the Durham Elementary
Schools, who analyzed the daily program for this purpose.
RESULTS AND CONCLUSIONS OF THE DURHAM
Table 16 presents the results of the comparison of the Raleigh
scores with those from the Durham schools. An analysis of the
table reveals a remarkably consistent picture of superiority in the
Durham scores except in certain cases in which silent reading and
speed in oral reading were measured.
The phonetic tests.-The comparative scores of the four phonetic
tests, Gates A4, A5, B2, and B3, indicate a definite superiority on
the part of the Durham group. In every case the difference between
the means is reliable. The differences vary from seven to more than
twelve times the P.E. of the difference. Thus, in so far as they
measure phonetic abilities, the tests indicate that the pupils in the
Durham schools had developed phonetic skills distinctly more than
had the Raleigh pupils.
The word pronunciation tests.-That these phonetic abilities carry
over into the pronunciation abilities is borne out by the fact that the
Durham pupils obtained much higher scores on the Gates Word
Pronunciation Test than did the Raleigh pupils. Here the difference
is seen to be more than eight times the P.E. of the difference.
The silent reading tests.-It will be recalled that the Gates Silent
Reading Tests, Types A, B, C, and D, were given to test specific
types of silent reading abilities. In Table 16 the scores of these tests
have been reduced to grade equivalents. In Types A and B (Read-
ing to Appreciate the General Significance, and Reading to Predict
the Outcome of Given Events), no reliable difference appears be-
tween the two groups. In Type C (Reading to Understand Precise
Directions), a small reliable difference favors the Durham group. A
somewhat less reliable difference favors the Durham group in Type
D (Reading to Note Details). The differences are the smallest that
appear in the table. It will be observed that if the norms are reliable,
the averages of both the Durham and Raleigh groups are slightly
above the expected grade-equivalent of about 3.8.1
The vocabulary tests.-A difference of more than ten times the
SThe norms used were taken from Arthur I. Gates's Manual of Directions
for Gates Silent Reading Tests (revised January, 1934), Bureau of Publications,
Teachers College, Columbia University.
Results and Conclusions of the Durham Investigation 41
DIFFERENCES BETWEEN THE RALEIGH AND DURHAM GROUPS IN TERMS OF THE
DIFFERENCES BETWEEN THE MEANS OF THE TEST SCORES
(89 CASES IN EACH GROUP)
Test Group* Mean between P.E. of Critical
Means* Differences*** Ratio
Gates A4..... D 79.50
R 63.31 16.19 1.61 10.05
Gates AS..... D 32.17
R 23.85 8.32 1.17 7.11
Gates B2..... D 29.29
R 18.11 11.18 .93 12.02
Gates B3 ..... D 15.20
R 9.29 5.91 .70 8.44
Word Pronunciation D 70.17
R 53.15 17.02 1.92 8.86
Gates Type A...... D 4.08
R 4.03 .5 .09 .55
Gates Type B...... D 4.18
R 4.18 .00 .11 .00
Gates Type C ...... D 4.61
R 4.11 .50 .12 4.16
Gates Type D...... D 4.38
R 4.15 .23 .08 2.87
Pressey Vocabulary.. D 71.85
R 59.26 12.57 1.21 10.39
GraySet l ........ D 2.35
(errors) R 8.79 6.44 .76 8.47
Gray Set IlI........ D 7.05
(errors) R 17.50 10.45 .83 12.54
Gray Set 1 ......... D 73.04
(time) R 38.78 -40.26 2.34 17.20
Gray Set III ........ D 77.48
(time) R 52.87 -26.61 3.09 8.61
Eye-Voice Span..... D 37.94
R 31.69 6.25 .64 9.76
*D and R refer to Durham and Raleigh respectively.
**Negative differences favor the Raleigh group.
***The Lindquist formula for matched groups was used to determine the P.E. of the difference.
P.E. of the difference favors the Durham group on the Pressey
Diagnostic Test of Vocabulary. The average score for the Raleigh
pupils, 59, represents a vocabulary of 1,200 words; and the average
score of the Durham pupils, 71, represents a vocabulary of 1,400
The oral reading tests.-The Gray Oral Check Tests yielded two
types of measures, the number of errors and the time consumed in
S. L. Pressey and L. C. Pressey, Directions and Class Record Sheet for
Pressey Diagnostic Reading Test.
Varied Amounts of Phonetic Training
reading the passages. The results indicate that the Raleigh pupils
made considerably more errors on both of the tests. The differences
between the means of the Raleigh and Durham groups are statistically
reliable, as is indicated by the critical ratios. On Set II the differ-
ence is more than eight times the P.E. of the difference; and on
Set III, the more difficult of the tests, the difference is more than
eleven times the P.E. of the difference. The differences are in the
opposite direction in the case of the time taken in reading the pas-
sages. Differences, eight and nine times the P.E. of the differences,
show that the Raleigh pupils read more rapidly than did the Durham
pupils. Thus, the Durham pupils appear to be slower, but more
accurate oral readers.
The eye-voice span test.-The averages of the eye-voice span test
scores favor the Durham group by a difference of more than nine
times the P.E. of the difference.
Speed and accuracy on the silent reading tests.-In the original
study (Agnew 2, Chapter XI) a study was made of speed and ac-
curacy on the silent reading tests. These data showed no consistent
evidence that large amounts of phonetics made silent reading slower
but more accurate, as appears to be the case in oral reading.
Methods used in word pronunciation.-An analysis of the meth-
ods used in pronouncing words in the Gates Pronunciation Tests
showed that approximately 70 per cent of the Durham subjects used
phonetic methods. On the other hand, it was found that only 30 per
cent of the subjects in Raleigh used phonetic methods. These data
indicate that the Durham subjects actually used phonetic methods in
The conclusions of the Durham investigation may be summarized
1. The comparatively large and more consistent amounts of
phonetic training received by the Durham pupils seem to have re-
sulted in greater phonetic abilities as measured by the Gates phonetic
2. The Durham pupils were superior to the Raleigh pupils in
word pronunciation ability.
3. The study of methods used in word pronunciation on the
Gates Graded Word Pronunciation Test revealed the fact that the
Durham pupils used phonetic methods of word pronunciation to a
much greater degree than did the Raleigh pupils.
Results and Conclusions of the Durham Investigation 43
4. Comparatively little difference appears between the Durham
and Raleigh pupils in the silent reading abilities measured. Small
differences on two of the four tests favored the Durham group.
5. No consistent differences appear between the two groups with
respect to speed and accuracy on the silent reading tests.
6. The greater phonetic training of the Durham group seems to
have resulted in the acquisition of greater vocabulary.
7. The Durham group appeared to be slower but more accurate
on the oral reading tests.
8. The Durham pupils seem to have developed greater eye-voice
span than the Raleigh pupils. This conclusion tends to refute the
argument that phonetic training decreases the eye-voice span.
BRIEF SUMMARY OF THE RESULTS OF THE INV\'ESTI-
GATIONS, AND CERTAIN EDUCATION. [L
A. THE RELATION OF THE INVESTIGATIONS TO THE CONTF,.'\FE .hi \L
ISSUES WITH REGARD TO PHONETIC INSTRUCTION: '
In Chapter I, the arguments for and against phon.tic training
were summarized. The investigations reported in the p.ri:-.int -tudy
present evidence that has direct bearing on a number of the argu-
The investigations have tended to support four of the arguments
in favor of phonetic training. These arguments are that phonetic
training when given consistently in large amounts (as in Durham):
(a) increases independence in recognizing words previously learned;
(b) aids in "unlocking" new words by giving the pupil a method
of sound analysis; (c) encourages correct pronunciation; and (d)
improves the quality of oral reading. The investigations provided no
evidence on the other arguments in favor of phonetic training.
The study tends to show that a number of the objections to
phonetic training have been exaggerated. In other words, although
the investigation offered opportunity for evidence in support of these
objections, such evidence did not appear. There was no evidence
that large consistent amounts of phonetic training tend: (a) to sac-
rifice interest in the content of reading; (b) to result in the neglect
of context clues; (c) to result in unnecessarily laborious recognition
of unfamiliar words; and (d) to be unnecessary because, tit :.lani-
tages attributed to phonetic training might be obtained u ihii.:out fr-
mal training. Some positive evidence indicated too that I ph'ltkic
training does not narrow the eye-voice span.
On the other hand, there are some data to show that large ;,miolunts
of phonetic training tend to slow up oral reading. This is. in Ia -:n-c.
counteracted by greater accuracy in oral reading.
The investigations did not reveal striking differences iln ilnt
reading ability as between groups having large difference, iin aillmiunr:
of phonetic training. There was no evidence that phoncti: IniIinIm
decreases efficiency in silent reading. This may be due i.: the fact
that speed in silent reading is largely acquired in the gradles :iabL:
Brief Sunmnary of the Results of the Investigations 45
the primary level. Further investigation would be necessary in order
to determine the effects of this early training on silent reading in the
B. THEORETICAL CONSIDERATIONS
Interference as an explanation of the Raleigh and Durham re-
sults.-Numerous comparisons were made in the Raleigh investiga-
tion between phonetic and nonphonetic patterns of phonetic training.
In these comparisons, with two noteworthy exceptions, no reliable
differences were found. The two exceptions, those found in the E-
and F-patterns, furnish clues to an explanation of the lack of reliable
differences between the other pairs of patterns and the differences
between the Raleigh and Durham results. The differences between
the members of the E- and F-patterns may have been due to the
factor of interference. If interference operated in these pairs of
patterns unequally to produce the differences between the members
of the patterns, it is possible that interference operated equally in
the other pairs of patterns. This fact would tend to account for the
lack of reliable differences between the members of the A-, B-, C-,
Furthermore, the inconsistency of instruction represented in the
Raleigh group may have been responsible, to an extent, for the lower
reading scores of the Raleigh pupils as compared to the Durham
pupils. Since the amount of phonetic training to which the Durham
group had been subjected was consistently high for each grade, the
factor of interference probably did not operate to lower the Durham
Phonetic abilities a function of factors other than amounts of
iaining.-The foregoing considerations suggest that phonetic abil-
ities are not only a function of amounts of phonetic training, but also
..f the consistency of phonetic training. If phonetic abilities were
merely a function of amounts of training, it would be expected that
the phonetic groups in the Raleigh investigation would manifest
greater phonetic ability than did the nonphonetic groups. This is
not the case, however. Although the phonetic groups in Raleigh
received much more phonetic training than did the nonphonetic
groups the phonetic groups, in general, showed no superiority over
'An effort was made to study the effects of interference statistically by
comparing the mean deviations from the means of the individual phonetic
.cores of the Raleigh pupils with the corresponding means of the Durham
pupils. The comparison indicated that the Durham pupils' phonetic scores
i for the various half-years) varied less from their means than was the case
, ith the Raleigh pupils.
46 Varied Amounts of Phonetic Tr.:.iin':T
the nonphonetic groups in phonetic abilities. The situation illus-h
treated by the chart which follows (see Figure 2). In this chiirt the
dotted line represents the expected growth of phonetic abit, if tlic
amount of training were the only causative factor. The lih-Iv line
represents the relationship that seems actually to exist.
Theoretical relationship ,...
lb0 200 250 300 350 400
Amount of phonetic experience in terms of Gross
Phonetic Experience Scores
FIG. 2. Theoretical presentation of the Raleigh and Durham data.
The difference between these two lines may be theoretically ex-
plained by either or both of two hypotheses. First, it is possible that
the factor of interference in the phonetic groups in Raleigh tended
to keep the increasing amounts of phonetic training from increasing
phonetic abilities. The facts to support this hypothesis have been
presented in the foregoing paragraphs.
The second hypothesis (somewhat related to the first) is that
phonetic abilities are highly complex, and that it is necessary, there-
fore, to have large amounts of practice in order to insure the ac-
quisition of phonetic skills. Thus, the learning of phonetic skills
may be a function of the difficulty of those skills, as well as of amount
of training. That these skills, the ability to pronounce letter groups
phonetically and to spell the sounds phonetically, are complex has
long been recognized. That the many sounds of letters could be
learned without a large amount of practice and drill seems improb-
able. Since these abilities are complex and require considerable
practice to insure their use, it follows that, if phonetic methods are
only partly learned, other methods of attack on words might be used
in the testing situation; and the half-learned habits of phonetic
analysis might serve rather to hinder successful responses than to
improve those responses.
In the case of the Durham subjects, the methods of phonetics had
been learned to the relative exclusion of other methods; in other
Brief Summary of the Results of the Investigations 47
words, phonetics had become functional. This possibility is borne
out by the very marked superiority of the Durham pupils on the
Gates phonetic tests and by the common use of phonetic methods by
the Durham pupils on the Gates Word Pronunciation Test.
C. EDUCATIONAL IMPLICATIONS
Should phonetic methods be employed in the teaching of primary
reading? The answer to this question can be given only when the
purposes of teaching primary reading have been agreed upon. If
the basic purpose in the teaching of primary reading is the establish-
ment of skills measured in this study (namely: independence in
word recognition, ability to work out the sounds of new words,
efficiency in word pronunciation, accuracy in oral reading, certain
abilities in silent reading, and the ability to recognize a large vocab-
ulary of written words), the investigations would support a policy
of large amounts of phonetic training. If, on the other hand, the
purposes of teaching primary reading are concerned with "joy in
reading," "social experience," "the pursuit of interests," etc., the
investigations reported offer no data as to the usefulness of phonetic
It is possible that the aims of primary reading should embrace
all these purposes. If this is true, the relation of phonetic training
(and the abilities resulting from phonetic training) to these other
purposes would have to be determined before the place of phonetic
training in primary reading instruction can be ascertained.
1. Agnew, Donald C. "A Study of Word Pronunciation by Third-
Grade Pupils." Unpublished Master's Thesis, Duke University, 1932.
2. Agnew, Donald C. "The Effect of Varied Amounts of Phonetic
Training on Primary Reading." Unpublished Doctor's Thesis, Duke
3. Atkins, Ruth E. "An Analysis of the Phonetic Elements in a Basal
Reading Vocabulary," Elementary School Journal, XXVI (April,
4. Bowden, Josephine H. "Learning to Read," Elementary School
Teacher, XII (September, 1911), 21-33.
5. Buswell, Guy T. An Experimental Study of Eye-Voice Span in
Reading. Supplementary Educational Monographs, No. 17. Chi-
cago: Department of Education, University of Chicago, 1920. xii,
6. Buswell, Guy T. Fundamental Reading Habits: A Study of Their
Development. Supplementary Educational Monographs, No. 21.
Chicago: Department of Education, University of Chicago, 1922.
xiv, 150 pp.
7. Cabell, Elvira D. "The Status of Phonics Today, Part I," Chicago
Schools Journal, XIII (April, 1931), 370-373.
8. Cabell, Elvira D. "The Status of Phonics Today, Part III," Chicago
Schools Journal, XIII (May, 1931), 430-434.
9. Carroll, H. A. "Effect of Intelligence upon Phonetic Generalization,"
Journal of Applied Psychology. XV (April, 1931), 168-181.
10. Cordts, Anna D. "An Analysis and Classification of the Sounds of
English Words in a Primary Reading Vocabulary." Unpublished
Doctor's Thesis, College of Education, State University of Iowa,
Iowa City, Iowa, 1925.
11. Cordts, Anna D. "Facts for Teachers of Phonics," Elementary Eng-
lish Revtiew, III (April, 1926), 116-121.
12. Currier, Lillian B. "Phonetics or no Phonetics," Elementary School
Journal, XXII (February, 1923), 448-452.
13. Dreis, Theme A. "A Case of Remedial Reading," Elementary School
Journal, XXXI (December. 1930). 292-300.
14. Garrett, Henry E. Statistics in Psychology and Education. New
York: Longmans, Green and Company, 1926. xiii, 317 pp.
15. Garrison. S. C. and Heard, M. T. "An Experimental Study of the
Value of Phonetics," Peabody Journal of Education, IX (July, 1931),
16. Gates. Arthur I. The Improvement of Reading. New York: The
Macmillan Company, 1927. xii, 440 pp.
Since this monograph was written, the two following articles have appeared:
Dolch, E. W., and Bloomster, Maurine, "Phonic Readiness," Elementary School
Journal, XXXVIII (November. 1937), 201-205; and Tate, Harry L., "The In-
fluence of Phonics on Silent Reading in Grade I," Elementary School Journal,
XXXVIII (June, 1937), 752-763.
17. Gates, Arthur I. "Intrinsic Method in Reading," Educational Out-
look, V (January, 1931), 105-114.
18. Gates, Arthur I. Manual of Directions for Gates Silent Reading
Tests. Bureau of Publications, Teachers College, Columbia Univer-
sity. (Revised, January, 1934.)
19. Gates, Arthur I. New Methods in Primary Reading. Bureau of
Publications, Teachers College, Columbia University, 1928. ix,
20. Gates, Arthur I. A Reading Vocabulary for the Primary Grades.
Bureau of Publications, Teachers College, Columbia University,
1926. 23 pp.
21. Gates, Arthur I. Psychology of Reading and Spelling with Special
Reference to Disability. Contributions to Educator, No. 129, Teach-
ers College, Columbia University, New York, 1922. vii, 108 pp.
22. Gates, Arthur I. "Studies of Phonetic Training in Beginning Read-
ing," Journal of Educational Psychology, XVIII (April, 1927), 217-
23. Gray, William S. Directions for Giving and Scoring Standardized
Oral Reading Check Tests. Bloomington, Illinois: Public School
24. Gray, William S. Remedial Cases in Reading: Their Diagnosis and
Treatment. Supplementary Educational Monographs, No. 22. Chi-
cago: Department of Education, University of Chicago, 1922. vii,
25. Holzinger, Karl J. Statistical Methods for Students of Education.
New York: Ginn and Company. viii, 372 pp.
26. Horn, Ernest. "The Child's Early Experience with the Letter A,"
Journal of Educational Psychology, XX (March, 1929), 161-168.
27. Housh, E. T. "An Analysis of the Vocabulary of Ten Second-Year
Readers," Seventeenth Yearbook of the National Society for the
Study of Education, Part I. Bloomington, Illinois: Public School
Publishing Company, 1918. 40-45.
28. Judd, Charles H. Reading: Its Nature and Development. Supple-
mentary Educational Monographs, No. 4. Chicago: Department of
Education, University of Chicago, 1918. xiv, 192 pp.
29. McKee, Paul. Reading and Literature in the Elementary School.
Houghton Mifflin Company, 1934. x, 591 pp.
30. Morphet, Mabel V. and Washburne, Carleton. "When Should Chil-
dren Begin to Read?" Elementary School Journal, XXXI (March,
31. Mosher, Ramon M. "Some Results of Teaching Beginners by the
Look and Say Method," Journal of Educational Psychology, XIX
(March, 1928), 185-193.
32. Mosher, Ramon M. and Newhall, Sidney M. "Phonic Versus Look-
and-Say Training in Beginning Reading." Journal of Educational
Psychology, XXI (October, 1930), 500-506.
33. O'Brien, John A. Reading: Its Psychology and Pedagogy. New
York: Century Company, 1926. xxviii, 308 pp.
50 Varied Amounts of Phonetic T,aini:';
34. Osburn, W. J. Bulletin on Remedial and Follow-Up Work 1n Siti,
Reading, No. 1. Bloomington, Illinois: Public School Fubl-hirr,
Company, 1925. 23 pp.
35. Packer, J. T. "The Vocabulary of Ten First Readers." T:,,caui:
Yearbook of the National Society for the Study of Educant .i P'art
II. Bloomington, Illinois: Public School Publishing Comp..1, I'. 'I
36. Pressey, S. L. and Pressey, L. C. Directions and Class Rie ,, S'. i
for Pressey Diagnostic Reading Test. Bloomington, Illin.:.i : Public
School Publishing Company.
37. Raleigh Public Schools. Curricudum Bulletin No. 3. 1..aii',.; in
the First Grade. (1928, republished, 1931.)
38. Raleigh Public Schools. Curriculum Bulletin No. 4. Tta~i ,i f'i1
Grades II and 111. (1928.)
39. Sexton, Elmer K. and Herron, John S. "The Newark Phonics Ex-
periment," Elementary School Journal, XXVII (May, 1928), 690-
40. Smith, Nila B. American. Reading Instruction. New York: Silver
Burdett and Company, 1934. x, 287 pp.
41. Smith, Nila B. "The Construction of First Grade Reading Material,"
Journal of Educational Research, XVII (February, 1928), 79-89.
42. Tinker, Miles A. "Visual Apprehension and Perception in Reading,"
Psychological Bulletin, XXVI (April, 1929), 235-236.
43. The Twenty-Fourth Year Book of the National Society for the Study
of Education, Part I. Bloomington, Illinois: Public School Publish-
ing Company, 1925. 363 pp.
44. Vogel, Mabel; Jaycox, Emily; and Washburne, Carleton. "A Basic
List of Phonics for Grades I and II," Elementary School Journal,
XXII (February, 1932), 436-443.
45. Washburne, Carleton and Vogel, Mabel. "A Revised List of Phon-
ics for Grade II," Elementary School Journal, XXVIII (June, 1928),
46. Wheat, Harry G. The Teaching of Reading. Boston: Ginn and
Company, 1923. ix, 346 pp.
47. Winch, William H. Teaching Beginners to Read in England: Its
Methods, Results, and Psychological Basis. Journal of Educational
Research Monographs, No. 8. Bloomington, Illinois: Public School
Publishing Company, 1935. 186 pp.
U .1 1 EE-A:.r C; 1lilli'A l
3 1262 05237 9897 -,
/., j -
DUKE UNIVERSITY RESEARCH STUDIES
No. 1 THE EFFEcrs OF UNFAMILIAR SETTINGS ON PROBLEM-
SOLVING by \V. A. Bro\nell and Lorena B. Stretch.
No. 2 VA\'RI -IITY IN RESULTS FROM NEWv-TYPE ACHIEVE-
NMEnr Trsr. by Earl \. Pullias.
No. 3 LEARNING AS REORGANIZATION: AN EXPERIMENTAL
STUDY IN THIRD-GR.ADE ARITHMETIC by W. A. Brow-
nell 'ith the assistance of Kenneth G. Kuehner and
Willian C. Rein.
No.-4 GROrV.Tl It; UN['RSTr.a\N[ING OF GEOGRAPHIC TERMS
it; GRAD.t IV t.' VXll by T. J. Eskridge, Jr.
No. 5 THE ErrFCT or \'.\IED .AMOUNTS OF PHONETIC
T7rAINING ON P'RIMAR1 READING by Donald C. Agnew.