GENERALIZATION OF THE PAIRED-ASSOCIATE
MODEL TO DEFINITION LEARNING

Thesis for 951:2 Degas of Ph. D.
MICHIGAN ETATE UNIVERSITY
SaIaI'a AbduI-Menéen': Hafar
I966

TH ESIS

 

This is to certifg that the

thesis entitled

GENERALIZATION OF THE PAIRED-ASSOCIATE
MODEL TO DEFINITION LEARNING

presented by

Salah Abdul-Moniem Hotar

has been accepted towards fulfillment
of the requirements for

Ph. D degree in Education

(mew/Mam

Major (tofessor

Date and H, me
I

0-169

ABSTRACT

GENERALIZATION OF THE PAIRED-ASSOCIATE
MODEL TO DEFINITION LEARNING

by Salah Abdul-Moniem Hotar

The Problem

Verbal learning psychologists are primarily concerned with study-
ing the basic mechanisms of learning by using nonsense syllables in
highly controlled experimental procedures with adult subjects. The
variable which has received the greatest attention has been the mean-
ingfulness (m) of the verbal material. There is a reasonable consis-
tency between the empirical findings and paired-associate theories,
namely that m of the reSponse member of a paired-associate has a greater
effect upon learning than m of the stimulus member. Most paired-asso-
ciate theories predict the following order of the four basic types of
lists H-H. L-H, HAL. and L-L arranged according to their ease of learn.
ing (H indicates high meaningfulness, and.L low meaningfulness). Howe
ever, paired-associate studies have yielded contradictory results con-
cerning the effect of familiarization on the response members prior to
the actual learning task. The satiation theory, the most promising one.
predicts the following order HAL, L-L, H-H. and.L-H if the subjects are
familiarized.with the reSponses.

Educators, on the other hand, are concerned as to whether the
material, the method, and even the findings have any objective applica-

tions to the classroom. Accordingly, the present study stands between

Salah Abdul-Mbniem Hotar

these two extremes and attempts to test the appropriateness of extend-
ing the paired-associate model to definition learning. Thus, the major
purpose of this study is to test the hypothesis that paired-associate
learning can be used as a model for learning arithmetical definitions.
The acceptance of this hypothesis requires two conditions. First, there
must be a one-to-one correspondence between the elements of a paired
associate and a definition. Second, certain relations must be preserved,
namely that the order of the types of definitions must be the same as
the order of the types of paired-associates under different familiariza-

tion treatments.

The Methodology

Feur arithmetic textbook series were used to gather arithemtical
definitions which appear in grades five, six, seven and eight. Defini-
tions with stimuli composed of more than one word or with symbols in
the reaponse were excluded. Ninety-seven arithmetical definitions were
finally used. Forty-eight of these definitions were numerical, and the
other forty-nine were geometrical. The two kinds of definitions,
numerical and geometrical, were randomly arranged. The stimuli and the
responses were separated from each other. The m of the definitions,
their stimuli, and their reSponses were determined separately by the use
of three rating scales each of which was judged by approximately thirty
sixth grade pupils whose median age was 1h5 months. Meaningfulness was
defined operationally in terms of the students' judged familiarity and
- ease of learning. The reliability coefficients were significantly dif-
ferent from zero (Pa<:.01). The results of this step has been.used

first to study the interrelationships between m of definitions, m.of

Salah Abdul-Moniem Hotar

the stimuli, m of the reSponses, number of letters in the stimulus and
number of words in the response; second, to select the four types of
definitions H-H, L-H, HAL, and.LéL.

Four sets of definitions were chosen. Each set contained four
definitions, two numerical and two geometrical. One definition of each
kind was longer in length of the response than the other. However,
each list had approximately the same total number of words. The verbal
familiarization material was composed of three sentences for each
selected definition. In case of picture familiarization, the numerical
definitions were explained by presenting three number operations, while
the geometrical definitions were explained by drawing three consecutive
pictures. The subjects who did not receive familiarization were desig-
nated as a control group. Thus the four types of definitions and the
three kinds of familiarization yielded twelve treatments.

The subjects were 434 volunteer students enrolled in the seventh
grade. Their median age was 151 months. The learning task was admin-
istered by the use of a group procedure in which definitions as well as
the familiarization material were presented by an overhead projector.
Rate of presentation of the definition stimulus and reSponse was 5 and
15 seconds respectively in paired-associate like procedure. The course
of experimental session was as follows: Pre-test, presentation of
familiarization material for all groups except the control group, a
learning trial followed by a test trial; two learning trials followed
by the second test trial; two learning trials followed by the third
test trial; two learning trials followed by the fourth test trial; two

learning trials followed by the fifth test trial; post—test. During

Salah Abdul-Moniem Hotar

the pre- and post-tests the subjects were instructed to express the
meaning of the stimulus using their own words, during the test trials,
they were aSked to write the exact words of the response as shown to
them in the learning trials. The subjects answers in the test trials
were classified into eight different categories with the assumption
that these categories represented a continuum which began with a "no
answer” and ended with "recalling the exact reSponse" and covered the
different levels of the answers. 0n the other hand, the pre- and post-
test answers were classified into three categories; wrong, partially
correct, and correct answers.

Concerning the test trials answers, the mean percentage of cor-
rect (exactly similar or slightly idfferent from the actual response)
responses per test trial (in case of the combined definitions, short,
long, numerical and geometrical definitions) was used as a dependent
variable. The types of definitions were arranged according to the
dependent variable. Kendall rank order correlation coefficient was
used to determine the correlation between the actual arrangement of
the types of definitions and the theoretical arrangement which is
predicted by paired-associate theory. Whenever the correlation was
perfect, a second test was applied using 2 score to determine whether
or not there was significant differences between any two prbportions
of correct responses among the types of definitions. The same analysis
was repeated using the mean percentage of the exact reSponses per test
trial as a dependent variable. As for the pro-test answers, the X2
test was used to compare the pro-test score distributions of the levels

of stimulus m, reSponse m, subject matter types, and length variables.

Salah Abdul-Moniem Hotar

Also, the same approach.was repeated using post-test answers. In addi-
tion, the X2 technique was used to test whether or not a significant

difference existed between the pre-test scores and the post-test scores.

The Results

First: The distribution of definiendum m was bimodal, while the
definition or definien m distributions were found approximately normal.
The mean m of the definitions was found to be significantly higher than
m of the response at the .01 level. However, the mean m of the defi-
niendum was not significantly different from the mean m of the definien
at the .05 level. Furthermore, m of numerical definitions was greater
than m of the geometrical definitions at the .01 level. The results
show a significant correlation between the m of geometrical terms and
number of letters. There was significant correlation between the
response m.and the number of words. The high positive correlations
showed that the shorter the response or the geometrical term the higher
was its m. 'While the intercorrelation coefficients between stimulus m
or response m, and their corresponding lengths were not significantly
different from zero at the .01 level, it was found that each one of
these variables correlates significantly with definition m. The signi-
ficant correlations between m of the definitions indicated that when
definition m was high reaponse m was high, stimulus m was high and
response number of words was few. The results showed that partialling
any group of variables out of the correlation of definition m with
other variables did not change the zero order correlation. However,
once the definition m or other variables beside definition m.were par-

tialled out, all the new correlations differed significantly from their

Salah Abdul-Moniem Hotar

zero order correlation. These results emphasized the penetrating
effect of definition m and its relation with reSponse m or stimulus
m. ‘When definition m was partialled out the following results were
obtained: (1) When reSponse m was high the reSponse was not neces-
sarily high or low; (2) when response m was high the stimulus m was
also high. Thus the effect of the variation in the number of letters
in the stimuli, the variation in number of words in the reSponses,
and the proved influence of definition m on stimulus as well as
response m.indicates the presence of variables in definitions which
are not usually studied in paired-associate learning.

Second: The analysis of answers of the subjects in the test
trials using the correct responses as a dependent variable were used
to test both the theory which emphasizes the stimulus m, as well as
the theory which emphasizes the response position. Each theory was
tested fifteen times (with the combined definitions, long, short,
numerical and geometrical definitions--per control, verbal and pic-
ture familiarization). The theory which emphasizes the role of stimp
ulus m was accepted twice in case of control and verbal familiariza-
tion of numerical definitions. The theory which stresses the response
m was accepted twice--for the control treatments with the combined and
the long definitions. In each of the confirming cases there was some
overlap among the types of definitions when both theories predict no
overlap. Possibly the nature of the dependent variable was responsible
for the failure to confirm either one of the theories with a high degree
of consistency because verbal learning psychologists assume the exact
reproduction of the response as their criterion measure of learning.

Thus the general hypothesis proved to be untenable.

Salah Abdul-Moniem Hotar

Third: When the exact response was used as a dependent variable,
the results did not confirm the theory which favors the stimulus posi-
tion. The theory which emphasizes the m of the reSponse was confirmed
in three cases (combined, short and long definitions) with control treat-
ment, and once (numerical definitions) with verbal familiarization. So
the choice of the exact responses as a dependent variable improved the
chances of confirming the general hypothesis.

Fourth: The data concerning the familiarization conditions led
to the rejection of the satiation hypothesis. Such deviation might
have occurred as a result of using a familiarization procedure not com-
pletely analogous to the familiarization procedures used by verbal
learning psychologists. The proactive inhibition theory explained the
results better than the satiation theory.

Fifth: Analysis of pro-test scores showed that stimulus m.was
more critical than response m. Thus definitions with higher m stimuli
were better known in advance than definitions with lower m stimuli. The
pre- and post-scores supported the hypothesis predicting an increase in
definition attainment as an increasing function of stimulus m. Finally
the analysis of post-test scores showed the presence of controversial
findings between m of the stimulus and m of the response. While the
arrangement of the types of definitions indicated that m of the stimulus
was more critical in the post learning than response m, the statistical
results were insignificant.

These facts emphasized the difficulty in generalizing from the

paired-associate model to definition learning.

GENERALIZATION OF THE PAIRED-ASSOCIATE

MODEL TO DEFINITION LEARNING

By

Salah Abdul-Moniem Hotar

A THESIS

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

DOCTOR OF PHILOSOPHY

Department of Educational Research, Measurement and Evaluation

1966

ACKNOWLEDGEMENTS

The investigator wishes to express his deep gratitude and
acknowledgements to Dr. Clessen J. Martin, Chairman of the Guidance
Committee who assisted in the initiation and completion of this study
and whose relationship has provided.professional growth and maturity.

The other three distinguished members of the investigators'
Guidance Committee: Dr. Benard Corman. Dr. Wilber Brookover, and
Dr. Jay Artie have offered valuable suggestions.

The investigator is indebted to Dr. Henry Kennedy, Dr. Clessen
Martin, Dr. Leland Dean, and Dr. Clyde Dow, who convinced the princi-
pals to participate in the experiment. Many principals, teachers and
pupils have shown interest in this study. Their cooperation is appre-
ciated.

The investigator acknowledges the assistance of his colleagues:
Mr. Rogers Bruce efficiently discussed the statistical procedure;

Mr. Munir Morcos kindly drew the figures; Mr. Taha Hussein and Mr. Kamal
El-Ganzoury have done considerable computer work.

The investigator's wife, Nabilah, offered continuing support
through her patience, understanding. and sacrifice. The investigator
is grateful to her.

The investigator and his wife extend their thanks to many American
families. The families of the distinguished Professor Dr. Nelson Bossing,
Dr. Irving Weyth, and Mr. James Brand made our stay in the United.States

of America a happy and memorial experience.

ii

The investigator and his wife express their gratitude to their
families and friends in the United Arab Republic (Egypt), eSpecially
our fathers, Mr. Abdul Moniem Motwalley Hotar and Dr. Naguib Saleh Arwad,
who gave us their encouragement.

Finally, the investigator is greatly indebted to the people and
the government of the United Arab Republic (Egypt). Their moral and
financial support have carried us through these years of academic effort

and achievement.

iii

TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS. . . . . . . . . . . . . . . . . . . . . . . . . ii
LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . iv

LIST OF FIGURE 0 O O O O O O O O O O O 0 0 O O O O O O O O O O 0 Vi

INTRDWCTION . O O O O O O O O O I O O O O O O O O O O O O O O O 1
Related Literature 0 e e o o e o o e o e o o e e o e o a a 7
Significance of the Study and the

Emerimnta]. Hypotheses O O O O O O O O O O O O O O O O 28

WHOD O O O O O O O O O O O O O 0 O O O 0 O O O O O O O O O O O 36
Determination of Definitional Meaningfulness . . . . . . . 36
mermntd materifl-s O O O O O O O O O O O O O O O O O O 39
Design and Statistical Procedure . . . . . . . . . . . . . #5

REULTS . O O O O O O O O O O O O O O O O O O O O O O O O O O C O 52
Definitional meaninnglneSS e e e e e o a a a a a e o e e a 52
Differential Effect of Meaningfulness of

Definition Components on.Subjects' Learning . . . . . . 54
Subjects' Pre- and Post-Definition Attainment . . . . . . . 76

DISCUSSION AND CONCLUSIONS . . . . . . . . . . . . . . . . . . . 91

BIEOIWRAPIH O O 0 O O O O O O O O O O O O O O O 0 O O O O O O O 107
APPENDICE O O O O O O O O O O O O O O O O O O O O O O O O O O O 112

iV'

LIST OF TABLES

Table Page

1 Methodolical Information Concerning the Reliability
of the Meaningfulness Scales . . . . . . . . . . . . . . 37

2 Rank Order of the Mean Percentage of Correct ReSponses
for the Types of Definitions in the Control Treatment. . 56

3 z Values for the Differences Between Proportions of
Correct ReSponses in the Control Treatment . . . . . . . 58

4 Rank Order of the Mean Percentage of Correct Reaponses
for the Types of Definitions in the Verbal Familiar-
iz‘tion O O O O O C O C O C O O C O C 0 C O C O . O O O 61

5 z Values for the Differences Between Proportions of
Correct Responses for Numerical Definitions With
vorb‘l F‘miliarization I O O O O O O O O O O O O O O O C 63

6 Rank Order of the Mean Percentage of Correct Heeponses
for the Types of Definitions in the Picture
Familiarization . . . ... . . . . . . . . . . . . . . . 67
7 Rank Order of the Mean Percentage of Exact Responses
for the Types of Definitions in the Control Treatment. . 71

8 z Values for the Differences Between Proportions of
Exact ReSponses for the Control Treatment . . . . . . . 72

9 Rank Order of the Mean Percentage of Exact Reaponses
for Types of Definitions in the Verbal Familiarization . 73

10 z Values for the Differences Between Proportions of
Exact Responses for Numerical Definitions With
VerbalFamiliarization................o7’+

11 Rank Order of the Mean Percentage of Exact Responses
for Types of Definitions With the Picture Familiar-
1811:1011........................75

12 x2 Values of Pre- and Post-Tests' Scores With Definition
Variables in Case of Control Treatment . . . . . . . . . 78

13 X2 Values of Pre- and Post-Tests' Scores for Types of
Definitions Under Control Treatment . . . . . . . . . . 80

14 12 Values of Pre- and Post-Tests' Scores With Definition
Variables in Case of Verbal Fandliarization . . . . . . 82

‘v

LIST OF TABLES (CONTINUED)

Table Page

15 X2 Values of Pre- and Post-Tests' Scores for Types of
Definitions Under Verbal Familiarization . . . . . . . 8h

16 12 Values of Pre- and Post-Tests' Scores with Definition
Variables in Case of Picture Familiarization . . . . . 86

17 12 Values of Pre- and Post-Tests' Scores for Types of
Definitions Under Picture Familiarization . . . . . . . 88

vi

LIST OF FIGURES
Figure Page

1 The Percentage of Correct ReSponses Per Test Trial
forcontrOlTreatmenteeeeeeeeeeeeeeeee59

2 The Percentage of Correct Responses Per Test Trial
for Verbal aniarization e e e e e e e e e e e e e e 64

3 The Percentage of Correct Responses Per Test Trial
for Pieture Familiarization e e e e e e e e e e e e e e 68

vii

ZKHP‘:

0

LIST OF APPENDICES
Page

I O O O O O O O O O O O O O O O O O O O O O O O O O O O 112

eeeeeeeeeeeeeeeeeeeeeeeeee00113

O O O O O O O ........ O O I O O O O O O O O O O 114

O O O O O O O OOOOOOOOOOOOOO O C O O C O O 116

O O O O O O O O O O O O O O O O 0 O O O O O O O O O O O 127

..... . . . . . . . . . . . . . . . . . . . . . . . 155

O O O O O O O O O O O O O O O O 0 0 O O O O O C O O O O 162

eeeeeeeeeeeeeeeeeeeeeeeeeeee163

eeeeeee eeeee eeeeeeeeeeeeeeee165

O O O O O O O O O ..... O O O O O O O O O O O O O O 166

eeeeeeeeeeeeeeeeeeeeeeeeeeee167

O O O O O O O O 0 O O O O O O 0 O O O O O O O O O O O O 169

O O O O O 0 O O O O O O O O O O O O O O O O O O O O O O 170

I O O O O O O O O O O O O O O O O O O O O O O O O O O O 171

O O O O O O O O O 0 O O O O O O O O O O O O O O O O 0 O 172

eeeeeeeeeeeeeeeeeeeeeeeeeeeei73

O O O O O O O O O O C 0 O O O O O O O O O O O C O O O O 17“
O O 0 O 0 O O 0 O O O O O O O O O O O O O O O 0 O O O O 175

O I O O O O O O O O O O O O O O O O 0 O O O O O O O O O 176

viii

LIST OF

U1 0 O O O O O O O O O
02 O O O O O O O O O 0
v1 0 O O O O O O O O O

APPENDICES (CONTINUED)

Page
.180
. . . . . . . . . . . . . . . . . . . . . 181

O O O O 0 O O O O O O O O O O O O O O O O 183

CHAPTER I

INTRODUCTION

Verbal learning psychologists are concerned with identifying
the variables which affect learning. The general procedure involves
the establishment of a miniature learning situation in which the
learner is presented a number of verbal units for memorization. A
wide range of variables has been studied within such learning situa-
tions. The variables have involved manipulations of the meaningful-
ness of the material, method of presentation. rate of presentation
and a host of other such variables. The many possible manipulable
variables have created a wide range of verbal learning experiments.
Each one of them is fairly well controlled and designed to test
Specific hypothesis concerning effects of the variables in question.

However, this experimental approach has raised many objections.
Learning psychologists (Deese, 1961; Rothkopf, 1963) claim that it
tends to ignore or neglect verbal characteristics and skills in addi-
tion to being restricted to examining elementary associations. Edu-
cational psychologists have also objected to such learning situations
as being too restricted and not applicable to meaningful materials.

On the other hand, the advocates of this type of learning experi-
ment believe that it stands squarely in the center of all human learn-
ing. They claim that research in verbal learning creates phenomena
and theories related to the study of the processes involved in the

learning situation (Underwood. 196“). Such basic principles of

learning make it possible to understand retention, forgetting, dis-
crimination, generalization, transfer and problem solving behavior.
Hilgard (1964) summarized the relationship between the latter
points of view of learning theory and educational practices as that
between any pure science and its technological applications. By
pure science research, he meant that research which is guided by the
problems which the researcher sets himself without regard for the
immediate applicability of the results to practical situations. In
applied research. on the other hand, the researcher is concerned with

a practical problem directly relevant to classroom learning. The

road from.pure science research to established educational practice

has been classified into six steps. The first three of these steps
relate to pure science research in learning and the others relate to
the applied or technological research. The steps may be described

as follows:

Step 1. Research on learning with no regard for its educational rele-
vance. e.g. animal studies. physiological, . . .

Step 2. Research on learning which is not concerned with educational
practices but which is more relevant than that of Step 1
because it deals with human subjects and with content that
is nearer to that taught in school, e.g. nonsense syllable
memorization and retention. . .

Step 3. Research on learning that is relevant because the subjects
are school-age children and the material learned is school
subject matter or skill, though no attention is paid to the

problem of adapting the learning to school practices, e.g.

foreign language vocabulary learned by paired-associate
method with various lengths of list and with various Spac-
ing of trials.
Step 4. Research conducted in Special laboratory classrooms, with
selected teachers. . . .
Step 5. A tryout of the results of prior research in a "normal"
classroom with a typical teacher. . .
Step 6. Developmental steps related to advocacy and adoption.
Hilgard added that too much of the research in the past several
decades has rested at Steps 1 and 2. Educational psychologists have
tended to work at this end of the spectrum and then to jump, by
inference, to Step 6, without being sufficiently patient about Steps
3, 4. and 5. He then emphasized the significance of the tasks to be
done all along the six steps.
Accordingly, the experiment reported in this study aims at being
a kind of bridge between the two endpoints: Pure learning psychology
and applied educational practices. The subjects are seventh grade
school children, and the material consists of arithmetical defini-
tions. The methodological approach is very similar to the laboratory
methods of Hilgard's Step 2. However this study can be located in
Hilgard's Step 3 where it is basically concerned with testing the
validity of extending the existing theoretical notions in what is
called paired-associate learning to the learning of the definitions.
Verbal learning psychologists usually use nonsense syllables.
two consonants separated by a vowel. in their experiments. These

nonsense syllables are scaled according to their familiarity and ease

of learning. This dimension of scaling is called "meaningfulness"
(m). Pairs of verbal units are presented to the subject (S). The
left hand member of the pair is designated as the stimulus term; the
right hand member, the response term. The §fs task is to learn to
be able to recall the reSponse term when the stimulus term is pre-
sented alone. An example of a paired-associate unit is "LAJ-NOV"
where "LAJ" is the stimulus term and "NOV" is the response term.
Similarly, a definition is also composed of two parts, the
definiendum and the definien, which are equivalent to the stimulus
and response of a paired-associate. For example: "Face: A region
of a plane enclosed by a polygon" is a definition.whose stimulus,
i.e. definiendum, is the word "Face," and whose re5ponse, i.e.
definien, is the phrase, "A region of a plane enclosed by a polygon."
However, this similarity is not perfect and the differences
between a paired-associate and a definition are numerous. (1) The
components of a paired-associate might be nonsense syllables, unre-
lated words, or letters. In the definition, the elements are mean-
ingful words related contextually to each other. (2) The definition
components are associated with a common meaning, while the components
of a paired-associate unit are not necessarily related. (3) Almost
all the studies in paired-associate learning use only one nonsense
syllable with a constant number of letters as a reSponse while in the
definition study the response is a phrase. Such definition responses
usually vary in.length. For example, the two phrases: "The numbers
zero and one," and "The process of finding how many times a number is

contained in another number," are responses for the stimuli "Bigits"

and "Division" respectively. While the first reaponse is composed of
five words, the second is composed of fourteen words. (4) Stimuli'
number of letters in paired-associate learning is a constant, that is
each stimulus is composed of either two or three letters, but the num-
ber of letters of the definition's stimulus is a variable. For
example, "Pi" and "Multiplication" are stimuli of two arithmetical
definitions. However each has a different number of letters. Because
of these similarities and dissimilarities, it is difficult to make any
generalization from paired-associate learning to definitions other
than to suggest conservatively that there may be some correspondence
between them.

Again, the first objective of this study is to test the hypoth-
esis that paired-associate learning can be used as a model for learn-
ing definitions. The acceptance of this hypothesis requires two con»
ditions (Brodbeck, 1963). First, there must be a one-to-one corres-
pondence between the elements of a paired-associate and a definition.
Second, certain relations must be preserved. That is to say we must
know whether definition.learning is influenced by the same variables
as paired-associate learning. The correspondence between a paired
associate item and a definition is fairly obvious. The second condi-
tion, i.e. that paired-associate and definitions are influenced by
the same variables requires research. The variable selected for
study and research in the present investigation involves the meaning-
fulness (m) of the definiendum and definien. Then, the keystone is
the coincidence of the findings of paired-associate experiments and

the findings of the definition experiment with regard to the effects

of m on learning. If manipulations in meaningfulness have similar

effects on definition learning, then the correSpondence between this

type of learning and paired-associate learning has been demonstrated
with reapect to the m variable.

The second objective of the present research is to test the
appropriateness of the paired-associate model in relation to the

findings on familiarization training. Verbal learning psychologists

have studied the effect of pre-training or pre-differentiation on the

differential effect of the stimulus and response members of a paired

associate list. In the familiarization study, the material is pre-

sented frequently and in advance of the learning trials for subjects.
Teachers and textbook authors also try to familiarize their readers

with either the definition stimulus, definition reaponse or both.

HOWever, the educators' approach in familiarization is different from
that of verbal learning psychologists. While the former tend to use

ekartxples from experience, pictures, and models; the latter use fre-

quent repetitions. For this reason, this study will introduce two

C1ifferent methods of familiarization, verbal and picture explanation,
which are similar to the educators' approach. The differential effect

or verbal familiarization will be compared with that of picture
familiarization.

A third objective is to obtain the meaningflllness values for a

number of arithmetical definitions. These values should have two

advantages. First, they help to equate the experimental conditions

of definition learning with the experimental conditions of paired

associate learning. For example, to conduct an experiment in

paired-associate learning, it is the usual procedure to choose a num-
ber of nonsense syllables whose m values are predetermined and to
relate the findings of the experiment to the level of the m values.
But. in the case of definition learning the m values of either the

stimuli, responses or definitions are not available. Therefore to

make a comparison between the experimental findings of paired-asso—
ciate learning and definition learning is questionable without con-
trolling the m of definitions' stimuli and reSponses. Second, there

are many elaborations on m of nonsense syllables. Verbal learning

PSYChologists have done extensive work to understand, for example,
the relationship between m of nonsense syllables and letter sequence.
In definition learning while there is a lack of such studies, there
is a. need to understand, the relationships between length of word

(Stimulus), length of a sentence (response) and their correSponding

“1 vGlues. Also there is a need to understand the relationship between

the m of a sentence and the mean m of the vocabulary which forms the
Sehtence. The knowledge of such relationships, and others to be
reported, are of significance to the educational enterprise in gen—

81‘&1 and to understanding the results of this study in particular.

RELATED LITERATURE

The review of the literature will include:
(a) The methods of scaling meaningmlness of the nonsense
material.

(b) The studies of the differential effect of the meaningful-

ness of paired-associate components.

(c) The studies of familiarization and its differential effect
on meaningfulness of paired-associate components.

Each part of the literature will include the theoretical consid-

erations and the empirical studies.

The Scalng of Meaningfulness

The studies of verbal learning have shown that a verbal unit
such as "MEX" will be learned much more rapidly than "XYJ." The dif-

ferences between the rate of learning of the verbal units have been

attributed to a factor called meaningmlness (m). The words which

are easy to learn are supposed to have higher meaningfulness (m)
Value than the words which are difficult to learn. Definitions of m
h“’6 involved different approaches and sometimes different names.
The earliest operational definition of m was based upon the
munber of associates reported for an item in a certain fixed time.
§3 Were asked to state in a word or phrase what the item meant for
than. If the syllable meant something but the subject could not
”sprees its meaning within the time limit, he was to say "yes." The
1r'Qtal number of items presented ranged from 4534 to 320 per study,
with the items presented by a tachistoscope or a memory drum for an
e3l‘posure time which varied from two seconds to seven seconds. Prin-
cipal contributors to this method were Glaze (1928), Hull (1933).
KI'ueger (1934). Witmer (1935). Archer (1960), and Hilgard (1951).
This m value has been designated as the association value.

The second method of rating the m of the items is called the

production method, devised by Noble (1952). and used by Mandler (1955).

Meaningmlness was defined as the mean number of responses written

during a certain time. However, the time in the production method
is typically longer than the time for the association method. The
maximum reported time using the production method is one hundred
twenty seconds. The items were either dissyllables or nonsense syl-
lables, presented on paper. The subject responded by writing all
the different words elicited by the item, within a certain time.
The third method of rating m has employed a rating scale.
The subjects were asked to rate the item in terms of either ease of
learning, familiarity, or pronunciation. This approach has been
used by Haagen (1949), Noble, Stockwell and Pryer (1957), Underwood

and Schulz (1960) and others. This m value has been referred to as
the familiarity value.

Goss and Nodine in 1965, called the association method and the
Production method the single-association technique and the multiple-
assOciation technique, respectively. Using the first method the
"ulnber of association by a single subject to each stimulus may be
limited to at most one association. With the second method the sub-

ject may respond with as many associations as he can within inter-

Vale. They called the scaling method as the experimenter-supplied

stimuli for responses because each stimulus may be accompanied by

One or more experimenter-supplied continuous stimuli in the form of

graphic rating scales. Moreover. they assumed a direct relationship

between frequency and m and that frequency can be considered as a

basis for inferring m of stimuli. In this context frequency refers

to the frequency of occurrence of stimuli as counted in samples of

words in written texts.

10

The association, production and rating methods all appear to
have some variance in common. The study reported by Mandler (1955)
showed a correlation coefficient of .65 between the results of the
production method and the number of associations method, using 100
syllables. Noble Stockwell and Pryer (1957) showed a correlation
co efficient of .81 and .86 between the m values obtained by rating
scale method and the values previously reported by Glaze and Krueger
using the number of associations method for 100 syllables. Underwood

and Schulz (1960) used Noble's items in three independent experi-

ments. They found correlation coefficients between m values reported

by Noble (who used the production method) on one side, and m values

I‘eceived by scaling the item's ease of learning, familiarity and

Pronunciation to be .90, .92, and .78. All the reported inter-corre-

lation coefficients between the results of different techniques are
sigl’iificantly different from zero.

Rate of learning has been shown to be functionally related to

In. Studies have shown that high m learning material is more readily

learned than low m material. This relation is confirmed by the

ex‘periments conducted by Lyon (1914), Reed (1929), Davis (1930),
MeGeoch (1930), Sisson (1938), Noble (1952), Underwood and Richardson
(1956), Dowling and Brown (1957), Sarason (1957). and Braun and
Heyman (1958). The first two studies dealt with educational materials
unscaled for m.

The others had scaled items in the form of a serial

learning task where the units were presented to the S in a constant

order on each learning trial and he was required to learn them in the

order presented.

11

Kimble and Dufort (1955), handler and Huttenlocher (1956), Noble
and McNeeley (1957), and Noble, Stockwall and Pryer (1957) proved that

the same relationship namely, that high m material is easier to learn

than low m material, held in the case of paired-associate learning

task where both components were of the same 111 value. An explanation

was offered by Underwood (1949) that m of material facilitates learn-

ing because of greater familiarity of such material.
The previous studies have been concerned with scaling either

nonsense syllables, nonsense figures, numbers, adjectives or nouns.

There are no studies applying the concept of m for educational material,

or even sentences.

.___The Diffegential Effect of Meaniggi‘ulness of Paired-Associate Como-

Lie nts on Legrning

It has been shown that nonsense syllables can be scaled accord-

ing to their m and that the higher the m, the easier the learning.
This part of the literature survey will show the development of the
tIMO-step theory and the empirical findings relevant to the role of

the stimulus and response m on paired-associate learning.
The two-step theory has been mentioned implicitly within the

fI‘amework of information theory. Miller (1951) defined the amount

of information conveyed by an item as dependent on the number of

alternatives from which that item is chosen. For example, it is

known that the twenty-six English letters occur with different rela-

tive frequency, and it is possible to predict a letter in a word if

the one or ones preceding it are known. However, the precision of

prediction is dependent on the number of alternatives from which

12

the predicted letter is chosen. Given the letter "Q" in a word, then
the only predicted letter to follow it is "U." The amount of infor-
naaaision of "U" is dependent on one because it is the only alternative
ifzrc>m which the choice is made. Again, given the letter "E" in a word,
then to predict the letter following it is to make a choice from the
other twenty-five alphabetic letters. Suppose that the letter which
is chosen to follow the letter "E" is "N." The amount of information
<>j?' "N" is then dependent on twenty-five, and is assumed to be greater
i.r1 'value than the amount of information of "U" which has been shown
to depend on one alternative.

Information theory implies also that as the contextual con-

straints increase, the information of the components of the verbal

‘1llist decrease. For example, in order to read the verbal unit "AEHV"

one may tend to pronounce each letter individually because there is

no previous learned context to integrate these letters, and the
information per letter is high. On the other hand, the verbal unit
as HAVE" is easier to pronounce. Here the constraint imposed by the
aI‘rangement of the letters is high and the letters are not consid-
ered independent entities. Rather, they are all components of one
C30ntext, namely the familiar word "HAVE."

Miller and Selfridge (1950) cited evidence that learning time
iincreases with the amount of information communicated. By deduction,
as contextual constraints increase the learning time is expected to
decrease. Thus, in definition learning, the verbal unit "Repeated
of: Process multiplication the addition," is more difficult than

"Multiplication: The process of repeated addition."

13

Information theory so far has shown that the increase rate of
learning for a verbal unit or a series of such units, is dependent
on the degree of contextual constraints. However, the theory has
not. placed a particular emphasis on paired-associate learning nor
on the role of meaningfulness of its components.

Hovland and Kurtz (1952) showed more explicitly than informa-
tion theory that learning successive pairs of nonsense syllables
involves two steps; one must not only learn the associations
between the units, but also the units to be associated. This notion
can also be applied in definition learning. In order to learn a
definition such as "Point: A mathematical idea; associated with a
location in Spaceflf the subject must learn the vocabulary of the
definition, i.e. the words "Point, A, mathematical, . . ., Space,"
and then how to connect the stimulus "Point" with the response.
H0Wover, the theory does not explain the steps involved in learning
each unit of the paired-associate and associating them with each
other.

Handler (19510 presents three concepts in his proposal: (a)
differentiating responses, (b) response integration, and (c) sym-
holic responses. The first concept suggests that a stimulus is dif-

1‘ erentiated from other stimuli when it evokes a response different
from one evoked by the other stimulus. This concept refers to the
behavior of identifying and exploring the elements of the stimulus.
The second concept is concerned with the elimination of subresponses
which prevent or delay reinforcement. The third concept implies that

any overt response which is perceived by a human organism evokes a

1Q

symbolic response analogous to the overt reSponse. Mandler concluded

that learning a response involves three steps: Differentiation, inte-

gration and association through symbolic analogy.

Underwood and Schulz (1960) have analyzed the acquisition of

paired-associate or serial lists into two stages. The first stage is

referred to as the reSponse learning or reSponse recall stage. In

this stage the reSponse units are learned and connected to form a

large unit. This step is similar to Mandler's (1951+) integration

step. In the second stage, the associative or hook sup stage, the sub-

ject connects the reSponse to a particular stimulus. Underwood and

SChulz suggested that stimulus m exerts its effect on the associative

StAge while the response m exerts its effect on response learning

Stage. Hence, they concluded that the effect of stimulus m on paired

asSociate learning is less than the corresponding effect of response

me aningfulnes s .

Two conditions must be fulfilled if paired-associate learning is
to be considered a valid model for definition learning. One of these

is that both the paired-associate and the definition must work on the

same principle. Such principle was not clarified. Furthermore, the

Paired-associate theories suggest the conclusion, that; m of the

response member is more effective with respect to acquisition rate

than m of the stinnllus'member. If the present experiment confirms

this expectation, then paired-associate principles can be used for
definition learning.

The following portion of literature survey will show the find-

ings of the empirical studies relevant to the effect of m on paired

15

associate learning.

The earliest experiments used lists of paired-associates whose
components could be classified roughly and as possibly having a high
m (H) or a low m (L). The material consisted of an English word
paired with a familiar word. The possible combinations were (1) H-H,
high m stimulus and high m response; (2) L-H, low m stimulus and high
m response; (3) H~L, high m stimulus and low m response; and (4) L-L,
low m stimulus and low m response. These four combinations will be
referred to in this study as pairedsassociate types. The lists might
contain one type of pairedwassociate with many items, or four types
of paired-associate (H-H, L-H, H-L, L-L) with very few items per
type. Learning has been measured by the number of correct responses
recalled directly after the end of the acquisition period.

Stoddard (1929) asked a group of school children to learn from
French words to English words, and the other group to learn from
English words to French words. If it is considered that English
words have higher m than French words, then this study provides a
test of the relative influence of m of the stimulus versus m of the
response. The mean test score for subjects who learned L-H (French-
English) was 15.1 of 25, and for those who learned H-L (English-
French) was 8.0 out of 25. Thus, it can be concluded that L-H pro-
duced better learning than HAL and that high meaningfulness in the

reSponse position is more critical than m of the stimulus position.

Cason (1933) constructed 18 lists of 16 pairs each. The ver-

b"Ll-units were familiar words (F) and.unfamiliar nonsense syllables

(U) (F-F, U-F, F-U, and up). He had two groups of §_s, both of

16

which were given heterogenous lists of the types of familiar and
unfamiliar verbal units. One group of §s was given the list to
study for a period of four to eight minutes. The second group
heard the pairs. An immediate recall test was administered in
which the stimulus was pronounced and Spelled by the experimenter
and the §s were to recall the response. Cason found that the two
methods of study, auditory and visual, produced approximately the
same results. Moreover, the recall of F-F (equivalent to H-H) was
significantly greater than for U-U (L-L), but the U-F (L-H) and
F-U (HAL) magnitude of recall was intermediate between F-F (H-H)
and U-U (L-L).

Sheffield (1946) used Cason's material in which various comp
binations were preserved within the list, but the presentation was
via the memory drum in order to control the time factor per unit.
He demonstrated that H~H produced the most rapid rate of learning
while L-L produced the slowest rate. The L-H and H-L rate of learn-
ing was significantly different, contrary to Cason's findings. The
L-H learning was slightly inferior to H-H, while H-L was slightly
faster than L-L learning. Sheffield concluded that differences in
m of the stimulus produces relatively minor changes in the rate of
learning as compared with corresponding differences in the m of the
reSponse.

Kimble and Dufort (1955) prepared lists of ten paired-asso-

CjHates in which the stimuli were ten items from Noble’s dissyllables.
The ten dissyllables represented a complete range of m in Noble's

sc:eil_e. Response terms consisted of common three-letter words. Thus,

17

it could be concluded that each list had these paired-associate types:
H-H, H-L, and L-H. A group of'Ss learned the list with Noble's dis-
syllables in the stimulus position, and a second group were given the
same dissyllables in the reSponse position. Kimble and Dufort found
that §s took more learning trials to anticipate correctly the list in
which Noble's dissyllables were stimuli than when the dissyllables
were reSponses.

Cieutat, Stockwell and Noble (1958) formed the four combina-
tions using Noble's dissyllables. Each list had only one combination.
The H-L list and L-H lists were composed of identical items, but the
positions were reversed. The lists were presented for twelve trials
and each trial was followed by a test trial. Learning was measured
in terms of the percentage of correct reSponses to each trial for each
list. Their results showed the difficulty of learning increases in
the order of H-H, L-H, H-L, and.L-L. Moreover, they found that varia-
tion in stimulus m produced a much greater effect on learning when
response m was low than when it was high; and that variation in
reSponse m produced a much greater effect on learning when stimulus m
was low than when it was high.

The later results were confirmed by Lambert and Paivo (1956).
Weiss (1958), L’Abate (1959). Hunt (1959), Underwood and Schulz (1960),
Epstein (1963), Kothurkar (1963), Nodine (1963), Harleston (1963),

Imartin, Cox and Boersma (1965), and G055 (1965). The generalization
Of“the increase of learning trials in the order Huh, L-H, H-L, and L-L
was found to hold under different experimental conditions whether the

léiss1;s were administered to subjects singly or by a group technique. in

18

constant or varied order, under anticipation or recall formats and
whether the subjects were college students or hospitalized mental
patients. Thus, it is concluded that m of the reSponse members is
from slightly to several times more potent than m of the stimulus
member. Goss (1965) recently made an extensive review of the lit-
erature on paired-associate learning and suggests that the available
data are still too scanty either to account for exceptions to this
generalization on rational grounds or to lead to more precise, reli-
able generalizations about other patterns of factors.

The previous emperical findings are consistent with the theo-
retical notion which has been considered as the principle upon
paired-associate learning works, namely, that m of the reSponse mem-
ber is more reSponsible for the acquisition rate of a paired-associ-
ate learning than m of the stimulus member. But it has not been
shown whether such confirmed paired-associate theory is valid in
case of definition learning. Then if it is proved that definien
meaningfulness is more critical than definiendum meaningfulness the

paired-associate model can be extended to include definition learn-

ing.

lbs Differential Effect of Familiggiggti2g_gfifigi;gd;§§§ggi§§g_ggflr
Egnents on Learning
The preceding literature has demonstrated that high m materials
are‘ILearned faster than low m materials. One explanation of this
PhehOmenon is that the high m material tends to be more familiar to
the Subjects. Another explanation suggests that frequent experience

witlkl the verbal material makes it more meaningful; thus its m

DI

19

increases and is easier to learn. To test the relation between m,
frequency of experience and rate of learning, psychologists have
designed experiments in which verbal stimuli or responses are pre-
sented to §s prior to the learning task. The purpose of such an
experiment is to test the effect of such familiarization process

on rate of learning. The usual method of familiarization is to
require the Sp to repeat the stimulus to themselves (Gannon and
Noble, 1961; Goss, Nodine, Gregory, Taub, and Kennedy, 1962), to
repeat the stimulus aloud continuously for a certain time at a rela-
tively high rate of repetitions per second (Lambert and Jackobvits,
1960; Kanungo, Lambert and Mauer, 1962), or to look at the stimuli
for a period of time (Cieutat, 1960). These approaches to familiari-
zation have been called either pre-training, pre-learning, or satia-
tion.

The role of familiarization has been explained by psychologists
in relation to their theoretical framework. According to Miller's
(1951) theory of information, familiarization reduces the number of
alternate items from the range of all possible nonsense syllables.
For example, the naive subject with the English language, given the
letter "Q" to anticipate the second letter, responds with any letter,
While the subject who is acquainted with this language will--with high
.Probability--restrict his choice to the letter "U." The subject who
is familiar with language structure is expected to restrict his

responses to previously learned language habits. In addition, famil-
iar‘1zation will tend to reduce the amount of information conveyed by

each syllable at the time of learning. For example. before the

20

familiarization trials each word of the phrase, "The process of taking
a number out of another number," is considered as unique and indepen-
dent source of information by itself. The familiarization process
helps the §_ to "chunk" the words together to form a smaller number of
information units. So after familiarization one _S_ may perceive the
phrase as composed of "The process of," "taking out of another," and
"number." Another §_ might perceive the same phrase again as composed
of smaller and smaller units.

Gibson (1940) has mentioned that the familiarization procedure
produces discrimination between the units to be associated and those
learned in earlier lists. She predicted two types of errors to be
reduced: Interlist and "invention" errors. This prediction is con-
sistent with Miller's theory of reducing the number of alternatives
to the limited number which have been offered in the familiarization
trials. For example: If a §_ is presented frequently with a list of
geometrical definitions to become familiar with, and the same list is
given to the same S for learning, then the erroneous responses that
are expected to be reduced are those involving recalling a numerical
response or inventing a haphazard answer for the geometrical stimulus.
Gibson also predicted a reduction of intralist errors if the §_ is
familiarized with the learning material. Thus, it can be assumed
that subjects who are given a list of geometrical definitions for

familiarization and then for learning are less likely to attach a geo-
metrical response to a different geometrical stimulus. Hovland and
Kurtz (1952) were able to show that familiarization enables the sub-

Jec‘t to recognize the interlist and invention errors but does not

21

reduce the intralist errors.

Mandler's (1954) view is that with successive repetition of a
response aggregate, the separate reSponses eventually become stimuli
for each other in a way that any part of the reSponse aggregate will
tend to evoke the whole reSponse. This is designated as an integra-
tion process and its growth is dependent upon elimination of responses
which prevent or delay reinforcement. Still Handler's explanation is
directed towards the response and, in a sense, he views the famil-
iarization process as responsible for limiting the number of alter-
natives and making such correct responses more integrated. Underwood
and Schulz (1960) added that familiarization is a procedure for making
reSponses more available during subsequent association learning.

Then, it could be concluded from these theoretical notions that
familiarization makes two contributions: First, it reduces the number
of alternatives to the learned ones; second, it reduces the separate
information elicited by the components of the familiarized item, and
thus makes it integrated and more available during learning.

While the preceding psychologists emphasize that familiarization
facilitates learning of low meaningfulness items, there are others who
state that it has a prohibitive effect, especially if the material is
of high meaningfulness. Theadvocates of the latter theory are
Lambert and Jakobvits (1960), Kanungo, Lambert and Mauer. (1962), and

Kanungo and Lambert (1963). Their explanation is in terms of either
meaning decrement or the development of a word-word habit. The first
explanation suggests that too much repetition for the H material

causes its m value to decrease, and the material becomes judged as L.

22

Since high m material facilitates learning and low m material retards
learning, the items which were originally H become L. Therefore,
familiarization of high m material may actually produce a decrement
in learning. However, the same theory states that the familiariza-
tion of low meaningfulness material makes it high m and so its avail-
ability and ease of learning will increase.

In the second explanation, researchers claim that familiariza-
tion increases the number of "books" or associations of the item, and
so the chance of associating the verbal unit with other verbal units
increases. As for the L verbal unit, its number of associates will
increase and thus will be more readily associated with other new items.
On the other hand, the familiarization of H also increases its number
of associates, but these associates will be used to tie this H item
with itself rather than with another item. Hence the number of asso-
ciates are assumed to be extinguished in developing word-word habits.
For example, the familiarization of the dissyllable "GOJEY" makes it
better integrated, develops its number of hooks and thus increases its
availability for association with any other item. On the other hand.
the familiarization of a well integrated verbal item as "KITCHEN"
which already has many hooks will create several items as "Kitchen,
kitchen, . . ." and each one of them is similar to "KITCHEN" in terms
Of'the number of hooks. But the hooks of each item will be associated
tC> the hooks of other items, and all of them will be consumed in devel-

oping a word-word unit such as "Kitchenkitchen" in that no other hooks

31‘ e left to be associated to another new verbal unit. Therefore there

23

is less chance for familiarized H verbal item to be recalled as if it
were an L.

The theory predicts, once more, that the paired—associate types
H-H, L-H, H-L, and L-L, on familiarization with the stimuli, responses,
or both, will have an acquisition rate similar to L-H, H-H, L-L, and
H-L, or H-L, L-L, H-H, and L-H, or L-L, H—L, L-H, and H-H reSpectively.
This prediction is built on the basis that the paired-associate member
which is either H or L and is given familiarization will turn conse-
quently to be L or H.

Another aspect of familiarization can be inferred from Postman
and Phillips (1964) empirical findings. They observed that the rela-
tionship between amount of recall and degree of contextual constraints
to be curvilinear and concluded that when material context is highly
constrained. recall is difficult, as in the case of unstructured
material. In addition recall is relatively easier when contextual
constraint is neithefhigh nor low. Although they reported this
relationship using Miller's (1950) terminology, the role of informa-
tion theory was not clarified in explanation of their findings. How-
ever, using this observation, it is possible to explain the relation-
ship between familiarization and contextual constraint, and between
contextual constraint and recall behavior. It can be argued that as
familiarization increases, contextual constraints increase, and ease

of recall is then determined by the degree of the imposed constraints.
Taking into consideration the curvilinear relationship, the familiari-
zation of an L material adds a moderate contextual constraint, and

accordingly its recall will be easier than before familiarization

24

trials. But the familiarization of an H material increases the pre-
vious imposed contextual constraint and then makes it difficult to

recall.

The following section will review the findings of the studies
relevant to the effect of familiarization on m in serial and paired
associate learning. Solomon and Postman (1952) controlled experimen-
tally the frequency of usage of Turkish words by asking subjects to
read and pronounce them with frequencies ranging from 1 to 25. They
found that recognition thresholds varied inversely with frequency of
prior usage. Noble (1954) offered 18 L items to 288 college students
in a serial form with different frequencies. He obtained a close
relation between the judged familiarity and frequency of occurrence.
Arnault (1956) using nonsense shapes came to a similar conclusion to
that of Noble (1954), namely that m and familiarity are closely
related, doubtless as a consequence of the number of previous famil-
iarization trials. The curves representing these relationships are
negatively accelerated between zero and 40 acquisition trials, and
diminish rapidly around the twentieth trial. On the other hand,
Lambert and Jackobvits (1960) feund that semantic satiation reliably
moves the rating of the term towards the meaningless point of the
scale. Kanungo, Lambert and Mauer (1962) feund that satiation treat-
ment caused a decrease in the connotative meaning of words receiving
many familiarization trials.

Another criterion used to measure the effect of familiarization,
other than the judged familiarity, is the number of learning trials

required to learn the material. Hovland and Kurtz (1952), Noble (1955).

25

Rilqy and.Phillip (1959) and Underwood and Schulz (1960), found a
significant reduction in the number of trials required for mastery

of serial tasks as a result of the level of familiarization.

Familiarization of L material has been found to affect the
rate of learning in ways similar to that of m in paired-associate
learning. It is expected. according to the theories, that greater
facilitation would result when the pro-learned unit appeared as the
response in the paired-associate than as the stimulus member. This
means that the L response member may become an H response. For
example, in a list of’LeL items, in which the L reaponse member has
been.prelearned, then the list would be similar to L-H list. Again,
if the stimulus member of the list LAL receives familiarization, the
list becomes similar to HAL list. Hence, the arrangement of the
lists according to their theoretical ease of learning is as follows:
Unfamiliarized L-familiarized.L, familiarized L-unfamiliarized.L, and
unfamiliarized.L-unfamiliarized.L. The theories which suggest this
order are, first; the two step theory which emphasizes the role of m
of response member over the m of the stimulus member in paired-asso-
ciate learning and, second; the familiarization theory which predicts
that L material will become equivalent to H material through the pre-
learning trials.

Goss (1965) reported that Scheffield (1946) compared the acqui-
sition of an H-H list without response familiarization and an HéL
list with response familiarization. He found that familiarization
of response members of H-L list was facilitative. Weiss (1958) com.

pared the acquisition of H-H, HAL, and.L-L with familiarized reSponses,

26

and unfamiliarized reSponses. He found that mean trials to criterion
adjusted for practice performance were lower with familiarized reSponse
members than with unfamiliarized reSponse members.

However there are studies which have found no significant facili-
tation when reSponse members are familiarized. Cieutat (1960) used
two mixed lists of four pairs of L dissyllables with one pair repre-
senting each of the four combinations of familiarization and unfamil-
iarization. The same subjects were used in all the treatments.
Familiarization was by looking at the familiarized items for sixty
seconds. He found that familiarization with the response member inhi-
bits learning with an unfamiliarized stimulus member, and is facilita-
tive with a familiarized stimulus member. The arrangement of the
combinations according to ease of learning was familiarized-familiar-
ized, unfamiliarizedaunfamiliarized, familiarized-unfamiliarized and
unfamiliarized-familiarized. Neither familiarization of stimulus mem.
bers nor familiarization of response members had a significant effect.
Such unexpected arrangement of the results might be due to the use of
a mixed list, and a few number of items to represent each combination.
Moreover, using the same subjects for learning all combinations might
have made them more experienced and more selective over the entire
task.

Another study which showed that familiarization of response mem-
ber did not improve learning was that of Kanungo, Lambert and Mauer
(1962). They formed a paired-associate list identical with H-H, using
high frequency words. Two groups learned this task, and one of them

obtained semantic satiation for the reSponse member. Those with

27

response satiation were inferior in the learning. Kanungo and.Lambert
(1963) showed that, with an H-H list, semantic satiation of either the
stimulus or the reSponse words retards subsequent learning. They
explained their results in terms of the m of the members of the paired
associate and the locus of familiarization.

Concerning the familiarization of the stimulus member, the
results are not conclusive. The findings of Gannon and Noble (1961),
Martin (1963), and Schulz and Martin (1964) support the idea of famil-
iarization having a facilitative effect when the stimulus member was
familiarized. Other studies reported that such familiarization would
produce an inhibitive effect. For example, Neiss (1958) compared
acquisition of familiarized and unfamiliarized stimulus members of the
following paired-associate types: H-H, L-H, and L—L. He found that
mean trials to criterion, adjusted for practice performance, were
lower with familiarized stimulus members than with unfamiliarized
stimulus for H-H and L-H, but not for LAL combination. The results of
Weiss, could be reported differently if the m of the stimulus had been
considered. One may conclude that H stimuli became L, and vice versa
on stimulus familiarization of the H-H and L-L. The conclusion of
prohibitive effect of the familiarized stimuli, explained as a result
of having H items, was mentioned in.Kanungo, Lambert and Mauer (1962)
and Kanungo and Lambert (1963).

Finally. a study by Bailey and Jeffrey (1958) reported no sig-
nificant effect for pro-learning in either member of the paired-asso-
ciate. They asked.§s to learn three successive lists of paired non-

sense syllables in which the stimulus term was different in each list

28

but the response term remained the same. In the test list the reSponse
terms were paired.with syllables. The familiarized syllables were in
either the stimulus position or in the response position. The test
list learning of these pairs under either condition did not differ from
learning under control conditions. Using a number of pre-learning
trials that were insufficient to produce significant differences

between the treatments might be reSponsible for this result.

The previous results are not consistent. These studies have
employed different levels of m, different familiarization procedures,
and different learning procedures. Such differences may well intro-
duce unspecified variables that make agreement among all the results
an impossibility. However, to make a better prediction or explana-
tion, it is necessary to know the meaningfulness of each member of the
paired-associate before familiarization, the locus of familiarization

relevant to the familiarized and the control treatments.

SIGNIFICANCE OF THE STUDY AND THE EXPERIMENTAL HYPOTHESES

A gap has existed between verbal learning psychologists and edu-
cators. Verbal learning psychologists are concerned with studying the
basic mechanisms of learning by using nonsense syllables in rigid
experimental procedures with adult subjects. However, the educators
have been concerned whether the material, the method, and even the
findings have any objective applications to the classroom. Accord-
ingly, the present study stands between these two extremes and attempts

to test the appropriateness of extending the paired-associate model to

29

definition learning. Thus, the major purpose of this study is to
test the hypothesis that paired-associate learning can be used as a
model for learning arithmetical definitions.

The variable which has received the greatest attention among
verbal learning psychologists has been the meaningfulness of the
material. Many studies have been concerned with different methods
of scaling the m of these materials. Such material has involved
either nonsense syllables, nonsense figures, dissyllables, or num-
bers. However the review of the literature failed to find any edu-
cational material which has been scaled for m. Therefore one
aspect of this study is concerned with determining the m of a num-
ber of arithmetical definitions as well as the m of the individual
definiendum and definien.

These m values make it possible to inquire about some relation.
ships which have not been studied in paired-associates. For example,
in this definition study it is possible to determine the interrela-
tionships between m of definitions, m of definiendum, m of definien,
number of letters in the stimuli, and number of words in the response.
By the use of partial correlation coefficient techniques it is pos-
sible to determine which variables effect the m of a definition.

Also it is possible to determine the relationship between m of the
entire definition and the summed m values of its components. How-
ever, while these relationships are important in understanding the
factors which affect the m value of definitions, they are of minor

concern in this study.

30

The verbal learning theoreticians are in agreement that m is
dependent on familiarity and frequency of experience. They explain
the ease of learning H material as due to its familiarity and avail-
ability fOr recall. In learning a paired-associate item, the learner
tends to integrate the smaller units of the response to produce one
which is more available and ready to be associated with the stimulus
member. On the other hand, in recalling the response member of a
paired-associate, the learner tries to limit his answers to the
learned items and tends to recall the well integrated reaponses
better than the unfamiliar or unavailable reSponses.

Because of the reasonable consistency between empirical find-
ings and paired-associate theory it is concluded that m of the
response member of a paired-associate has a greater effect upon learn-
ing than m of the stimulus. Hence, if it can be demonstrated that
definition learning is influenced by the same variables as paired
associate learning, then the paired-associate model can be genera-
lized to definition learning.

Paired-associate research has yielded contradictory results con-
cerning the effect of familiarization on either the stimuli or
reSponses prior to the actual learning task. Psychologists have dif-
ferent explanations, but they have emphasized the role of the meaning-
fulness of the familiarized material more than its position. The data
suggest that familiarization of low meaningfulness members makes them
more integrated and changes them to readily available, highly meaning-
fulness members. On the other hand, familiarization of high meaning-

fulness items tends to produce a kind of satiation of meaning.

31

In order to test the hypothesis that the mechanism underlying
the learning of arithmetical definitions is similar to the mechanism
involved in paired-associate learning, it is necessary that the
experimental situations remain as similar as possible. The m values
which have been obtained for the arithmetical definition stimuli and
responses, permit manipulation of the four basic types of definitions:
H-H, L-H, H-L, and L-L. Each type of definition is represented by
four arithmetical definitions. For example in case of L-H, two of
them have relatively short definiens (reSponses), and the other two
have long definiens (responses). One of the short definitions is
numerical and the other short definition is a geometrical term. Simi-
larly, in case of definitions with long responses, one of them is
numerical and the other is geometrical.

This study will test the hypothesis that definition types,
arranged according to their ease, are similar to the paired-associate
types when arranged in accordance with the response theory. However,
the empirical rank order of definition types will be correlated with
the suggested order of the theory which stresses the importance of
the stimulus position and the theory which favors the importance of
the reSponse position. Very little theoretical attention has been
given to the role of stimulus m in paired-associate learning. Theo-
retical attention has been focused almost exclusively on the role of
response m. However, this study will attempt to assess the role of
StiJnulus m as well as reSponse m. It will be possible to assess the
effkacts of m on the stimuli and responses in a number of different

insflances. The effect of m will be assessed among long and short

32

definitions, numerical and geometrical definitions, and the combined
definitions within a list.

Concerning familiarization, this study will investigate the
effects of verbal and picture familiarization. This method of famil-
iarization is contrasted with the type of familiarization procedures
used in standard paired-associate learning tasks. The standard
familiarization procedure typically involves frequent repetitions of
the material prior to the actual learning tasks. However the famil-
iarization procedures employed in this investigation are more similar
to actual classroom practice. In addition, familiarization is
devoted almost entirely to the response member of the definition.

On the basis of paired-associate familiarization data it is
assumed that familiarization of a high m verbal unit reduces the m
value of that unit and familiarization of a low m unit increases the
m value. If it is assumed that stimulus m is more influential in
learning than is the reSponse m, then before familiarization the
following rank order would be H-H, H-L, L-H, and L-L. But if famil-
iarization has the effect of extinguishing high m values, then the
order as a result of familiarization would be H-L, H-H, L-L, L-H.
However, if reSponse m is more influential in learning than stimulus
m, then prior to familiarization the order of difficulty would be
H-H, L-H, H-L, and L-L. But this order would be expected to change
to H-L, L-L, H-H, and L-H as a result of reSponse familiarization.
The correlations between these predicted rank orders and the actual
rank orders will be used as a basis for determining the generality

of the paired-associate model to definition learning.

33

Another aspect of this study is concerned with the extent the §
has actually learned to state the definition in his own words. This
measure is referred to as definition attainment. The emphasis here is
not upon verbatim repetition of the exact words and phrases. However
the §_is encouraged to put the meaning of definition's stimuli in his
own words before and after the learning trials. It is believed that m
of the stimulus position is more concerned with definition attainment
because the connotation of definition meaning is represented in the
stimulus member. The exact response is an arbitrary arrangement of
words defining the stimulus. In other words, the meaning of the
definition is represented by its stimulus and can be expressed in dif-
ferent ways or in different verbal arrangements. When the subject is
asked about the meaning of a familiar stimulus, he can reSpond cor-
rectly in different ways, e.g. by explaining the stimulus, giving an
example, or making an analogy.

This study will also compare definition attainment before and
after the learning task, and attempt to state whether the m of the
stimulus or the response member is more important in increasing defi-

nition attainment.

Experiment§l_Hypotheses
A. Analysis of test trials.
The control treatments:
1. The increase of m of response member of the definition
will be accompanied by an increase of mean percentage
of correct responses per test trial. The arrangement

of the types of definitions from superior to inferior,

The

1.

3“

with respect to the mean percentage of correct responses
per test trial is as follows: H-H, L-H, H-L, and L-L.
The H-H, L-H, H-L, L-L arrangement will also exist if
the dependent variable is the mean percentage of exact
reSponses per test trial.

The stated arrangement of 1 and 2 will hold in case of
using the entire task, short, long, arithmetical and
geometrical definitions.

verbal and picture familiarization treatment:

Verbal familiarization of the response member will be
accompanied by either an increase or decrease of the
mean percentage of either the correct or the exact
reSponses per test trial. The increase will occur if
the familiarized reSponse is high in meaningfulness,

and the decrease will occur if the familiarized response
is low in meaningfulness. The arrangement of types of
definition, according to the response theory, from
superior to inferior with respect to the percentage of
correct or exact reSponses, is expected to be in this
order: H-L, L-L, H-H, and L-H.

Picture familiarization will produce the same order of
arrangement as verbal familiarization.

The stated order of 1 and 2 in case of familiarization,
will hold using the entire task, short, long, arithmeti-

cal and geometrical definitions.

35

B. Analysis of pre- and post-definition attainment scores.

1.

The increase of stimulus m will be accompanied by an
increase of both pre- and post-test scores. But the
increase of response m will not be accompanied by any
increase in either the pre- or post-test scores.

There will be a significant difference between the
distributions of pre- and post-test scores.

The relationship stated in 1 and 2 will hold in case
of control, verbal or picture familiarization conditions.
Under each familiarization treatment the significant
differences between the distribution of pre- and post-
test scores will decrease in the following order: H-H,

H-L, L-H, and L-L.

CHAPTER II

METHOD

DETERMINATION OF DEFINITIONAL MEANINGFULNESS

Arithmetical Definition
Four arithmetical textbook series were used to gather arithmeti-

cal definitions which appear in grades five, six, seven and eight.
The names of the textbooks are given in Appendix A. Definitions with
stimuli composed of more than one word (e.g. "square root," "right
angle") or with symbols in the reSponse (e.g. "an angle which measures
more than 90°," "any number that can be named by a fraction of the
form a/b, where a and b are integers, with the restriction that b is
not 0") were excluded. Ninety-seven definitions were finally used in
this part of the study. Forty-eight of these definitions can be
represented in numbers and are called numerical definitions. The
other forty-nine can be represented by graphs and are referred to as
geometrical definitions.

The two kinds of definitions, numerical and geometrical, were
randomly arranged. The stimuli and the responses were separated from
each other and the m value of each was determined separately by the
use of two rating scales. Meaningfulness was defined operationarby
in terms of the students' judged familiarity and ease of learning.

The first page of each scale contained the instructions and
examples. The other pages of the rating scale included the rated

items and m continuum based on five points: Very easy, easy,

36

37

indifferent, difficult, and very difficult. These pages of the rating
scale were presented in different random orders. A copy of the

instruction page is shown in Appendix B.

Subjects, Procedure, and Reliabilities

TABLE 1: Methodological Information Concerning the Reliability of the

Meaningfulness Scales

 

 

 

 

Information ' Stimulus Response
Reliability of AD .9813 .9033
Reliability of ND .9847 .9ouu
Reliability of’GD .9778 .8620
Reliability Method Test Retest Test Retest
No. of Students 30 30

M . Age in Months 145 145
Grade 6 6
School B B
Dates of Administrating 5/7/65 5/7/65
the Scale 5/11/65 5/11/65
AD = Arithmetical definitions

ND

Numerical definitions

GD = Geometrical definitions

Table 1 contains the number of students who volunteered in rating
the m of each scale, their median age, their grade and their school.

It also includes the dates of administering the different rating scales.

38

The experimenter used a standard procedure in administering the
rating scales. First, the §s were asked to read the printed instruc-
tions. Second, the experimenter explained the instructions verbally
(see Appendix C) and, using the blackboard, showed how to indicate
the ratings for the examples given. All questions asked by the §s
were answered, but the experimenter did not give a definite rating
to the examples. Third, the Se were asked to read all the items
before rating them. Fourth, the Ss were asked to reread each item
carefully and to check the appropriate point on the scale.

In order to determine the meaningfulness and standard deviation
for each scaled item, the points very easy, easy, indifferent, diffi-
cult and very difficult were assigned the weights one, two, three,
four and five respectively (following Thurstone and Chave, 1929).

Stimuli and responses' m values were checked for reliability by
administering the rating scale twice. Therefore, for each stimulus
or response two m values and two standard deviations were obtained.

A third m value was computed.by pooling the students‘ two ratings.
This m value and its corresponding standard deviation are referred to
as the pooled values. The pooled m value and the pooled standard
deviation are considered the standard values for each item.

The reliability coefficients and the method used in estimating
them are shown in Table 1. All of the reported coefficients are
above .90 with the exception of the reliability of geometrical
responses which is .862 and all are significantly different from zero
(P< .01).

39

Appendix D contains the m values of the rated material. It
should be remembered that the lower the value of m, the higher the
meaningfulness of the item. This has resulted from assigning ascend—
ing weights to descending ease and familiarity of the rating scale

points.

EXPERIMENTAL MATERIALS

Selection of Definitions

The stimulus m of each item was paired with its corresponding
reSponse m. Thus each definition had two m values, one for its stimp
ulus and one for its reSponse.

Four lists of definitions were chosen. Each set contained four
definitions, two numerical and two geometrical. One definition of
each kind was long and the other was short. However, each list had
approximately the same total number of words.

The H-H list of definitions contained high m values for both
its stimuli and its reSponses. Its m value for the stimuli ranged
between 1.153 and 1.678. The m.value for the reSponses ranged between
1.864 and 2.271. In the L-H list the stimuli's m values ranged
between 3.983 and 4.390 while the responses ranged between 1.915 and
2.542. The H-L list consisted of stimulus m values between 1.034 and
1.898 and response m values between 2.593 and 3.898. The L-L list of
definitions contained stimulus values between 3.475 and 4.559 and
response values between 2.695 and 3.898. Appendix G presents the

selected items.

4O

Familiarization Materials

The materials used in the familiarization process were of two
types, verbal and picture familiarization. Verbal familiarization
materials were prepared as follows. The experimenter prepared three
different written explanations for each selected definition. Each
explanation was composed of three sentences. The three explanations
were attached to the reSponse part of the definition. Those responses
and their corresponding explanations were given to three graduate
students who were told that one out of every three explanations would
be presented to sixth grade students. They were asked to rate each
explanation according to its appropriateness for sixth grade students,
by rank ordering the three explanations from "highly appropriate" to
"less appropriate." The ranks were summed for each explanation. For
each response, the explanation with the lowest sum of ranks was
chosen as the verbal familiarization material for the corresponding
definition.

In case of picture familiarization, the selected definitions
were classified according to whether they were numerical or geomet-
rical definitions. The numerical definitions were explained by pre-
senting three number operations, while the geometrical definitions
were explained by drawing three consecutive pictures. The same
graduate students were requested to rank order the three number opera-
tions or consecutive pictures per definition response, with the same
instructions as previously. The selected picture familiarization
Imaterials were those explanations which had the lowest sum of ranks.

.A11.the familiarization materials are shown in Appendix H.

41

Subjects

Four hundred and thirty-four volunteer students enrolled in the
seventh grade were used as Ss. Their median age was 151 months at the
time of the experiment. Appendix I shows the number of students

assigned to each task.

Procedure

The learning task was administered by the use of a group proce-
dure. Each group of’Ss was assigned at random to one of the treat-
ments. The students were given a test booklet containing eight pages.
The first page provided space in order to obtain information about the
st sex, birthdate, and name. The other seven pages contained only
the stimuli with blank Spaces where the‘Ss wrote their recalled reSponse
beside each stimulus. The stimuli were arranged randomly on each page.

The definitions as well as the familiarization material were
placed on thermofax transparencies and were presented manually by an
overhead projector. The thermofax transparencies of the definitions
were covered by two separate pieces of paper, one of which covered the
stimulus and the other the response. The two covers were used to pro-
ject separately either the stimulus or the response of the definition.

During the learning trials, the Sp were shown each stimulus for
approximately five seconds, followed by the stimulus and the response
for approximately fifteen seconds. The total time needed for each
learning trial per definition was about twenty seconds. Since there
were four definitions per task, each learning trial required eighty
seconds. The time required for one test trial was three minutes and

thirty seconds. In case of familiarization, the experimenter took

42

four minutes to present the familiarization material. Presentation
rates were determined by the use of a stop watch.

The experimenter followed the steps shown below while he was
conducting the experiment. First, the §s were instructed to turn to
the second page of the booklet and to give the meaning of each word
(stimulus). This pre-test was for the purpose of determining the
extent to which the definition had been attained prior to the learn-
ing task. Second, if the treatment included familiarization, the
experimenter showed the §s the corresponding familiarizations and
the experimenter helped them by reading the material aloud. Third,
the experimenter showed the §s one of the stimuli, asked them to read
it, and anticipate its meaning before reading the projected sentence
(response) within the time limit which has been mentioned before.

The third step involved the presentation of all four definitions. One
complete presentation of all definitions is referred to as a learning
trial. Fourth, in the test trials the §s were asked to write down the
exact sentence (response) which had been shown to them with its cor-
reSponding word (stimulus). There were nine learning trials (1) and
five test trials (t) arranged in this order: l-ti, ll-tz, ll-t3, ll-tu,
ll-ts. Fifth, the §s were asked to express the meaning of each stimu-
lus using their own words. The post-test was to determine definition

attainment after the learning task.

Treatments
It has been stated previously that there were four different types
of definitions: H-H, L-H, H-L, and L-L. There were also three kinds of

familiarization. The §s received either no familiarization, verbal

43

familiarization or picture familiarization. Those §s who did not
receive familiarization were designated as a control group.

The combination of the four types of definitions with the three
kinds of familiarization produced twelve treatment conditions. Each
treatment was identified by the type of definition and the kind of
familiarization. The treatments are designated by combining the sym.
bols of the type of definition with the symbols referring to the type
of familiarization. For example, H-H refers to definitions with high
m stimuli and high m reSponses when presented without familiarization,
while L-H P designates definition with low m stimuli and high m

reSponses presented with picture familiarization.

Dependent Variables
The Ss' answers in the test trials were classified.ixto eight dif-

ferent categories. It was assumed that these categories represented

a continuum which started with a "no answer," and ended with "recalling
the exact reSponse," and covered the different levels of the answers.
Each category was labeled to explain its common property and was
assigned a score depending on its location within the continuum.

A sample of the Ss' answers to the stimulus "abscissa" will be
presented to explain the scoring system. As mentioned in Appendix E,
this stimulus is one of L-H definition and its response is "the dis-
tance measured horizontally to a point." Other stimuli which are
found in the L-H type of definitions are "predecessor," "uniqueness"
and "hypotenuse" and their responses are shown in the same appendix.
The latter reSponses are of concern to the response of the stimulus

"abscissa" since some Ss confused and/or mixed these responses with

44

the "abscissa" response in their test trials. The scoring system and
a sample of the answers to the stimulus "abscissa" are shown below.

1. No answer: Assigned for the answers which were left blank
or if the §_wrote "I do not know" or "A distance" i.e. when the
reported number of words was less than one quarter of the exact
response.

2. Outside or inventional answer: Answers which were comp
pletely unrelated to any response in the whole task as "A line drawn
to the center of a circle."

3. Confused answer: Response of one stimulus given to another
stimulus as "There is only one sum correct the sum of the number."

4. Mixed answers: §_mixed two different reSponses and
responded with this mixture to a given stimulus as "A number measured
horizontally."

5. Between one quarter and one half of the exact response: e.g.
"The distance measured."

6. One half of the exact response but less than three quarter:
e.g. "The distance measured horizontally."

7. Three quarters of the exact resggnse, or the exact response
written in a different form: e.g. "The distance horizontally to a
point."

8. The exact resEonse: As "The distance measured horizontally
to a point."_

Studies in paired-associate learning have been concerned with
measuring the exact responses. But the strategy of this study requires

a more flexible scoring system. Thus limiting the analysis to the

45

exact responses (category 8) would exclude other answers which were
almost correct (category 7). Categories seven and eight have been
combined and classified as correct responses. The mean percentage
of these correct responses is a dependent variable.
This study also compared the §fs mastery of the definitions
before and after the learning trials. For this reason the answers
of the pre- and post-tests have been classified into three categor-
ies. The assigned scores and the types of answers are shown below.
0. No answer, outside answer, confused answer or mixed answers.
1. Answers which were partially correct.
2. Correct answers.
The use of these categories permitted an assessment of defini-
tion attainment as a result of the learning trials. The dependent
variable in this part of the analysis was the given score per defini-

tion in the pre- and post-tests.

DESIGN AND STATISTICAL PROCEDURE

The available variables are definition m, stimulus m, response
m, number of letters in the stimulus and number of words in the
response. These five variables are presented in Appendices D and H.
The first description of these variables is in terms of their means
and standard deviations. Second, analysis of variance has been
utilized to test the effects of subject matter levels and definition
components on meaningfulness. In making this test, the fixed-effect
model for two-factor completely randomized design was applied. Third,

the intercorrelation coefficients among these five variables have been

46

calculated in order to determine the interrelationships among them.
Moreover the contribution of each definition variable to its m is
shown by partialling out one or more of these variables. The partial
correlations are compared to the zero order correlation coefficients,
and the results of these analyses are presented in some detail in
Appendix F.

The twelve treatments are produced by having four types of defi-
nitions (H-H, L-H, H-L, and L-L) and three kinds of familiarizations
(control, verbal and picture). Each §_received five test trials in
which to recall the responses of the four stimuli. Bartlet's test has
been applied to test the homogeneity of test trial variances. The
test revealed a X2 value of 402.7485 which is significant at the .01
level. The hypothesis concerning the homogeneity of test trials vari-
ances for the twelve treatments was rejected and the presence of heter-
ogenous variances suggested the use of non-parametric methods in data
analyses.

The mean percentage of correct responses per test trial is used
as a dependent variable. Appendices J, K, and L show the mean percen-
tage of correct responses per test trial as well as the mean percentage
of such correct responses in case of the combined definitions, and of
short, long, numerical and geometrical definitions, under the three
different familiarization.procedures. The types of definitions (H-H,
L—H, H-L, L-L) of any of the fifteen treatment conditions can be
arranged according to the mean percentage of correct responses.

The rank order of the types of definitions suggested by the

theory emphasizing the m of the reSponse member is H-H, L-H, H-L, LAL

47

without familiarization, and H-L, L-L, H-H, L-H with response familiari-
zation. On the other hand, the suggested arrangements of the types of
definitions by the theory which emphasizes the m of the stimulus posi-
tion is H-H, H-L, L-H, L-L without familiarization, and HAL, H-H, LAL,
L—H with response familiarization. The logic behind these arrangements
has been explained in the first chapter of this study.

Kendall rank order correlation coefficients were used to deter-
mine the correlation between the actual arrangement of the types of
definitions as reported in Appendices J, K, and L, and the arrangement
which is suggested by either the theory emphasizing the role of m in
the response, or the theory which emphasizes the role of stimulus
meaningfulness. Separate Kendall rank order correlations were computed
in order to test the two theories relating to m of the stimuli and
responses in relation to the various tasks (combined, short, long,
numerical and geometrical definitions), and different types of famil-
iliarization (control, verbal, and picture familiarization).

Whenever the actual arrangement of the types of definition,
under any one of the familiarization conditions, correlated perfectly
with any of the suggested theoretical orders, a second test was applied
to determine whether or not there were significant differences between
any two proportions of correct responses among the types of definitions.
The 2 score, and the normal distribution table were used to test whether
the mean percentage of correct responses for each type of definition was
greater than the other mean percentages of correct responses in the

order designated by each theory.

48

The use of the mean percentage of correct responses as a depen-
dent variable represents a departure from the usual performance cri-
teria employed in verbal learning studies. Since in actual practice
children are seldom required to memorize definitions verbatim, this
dependent variable approximates the type of response required in the
classroom. However, a second dependent variable approximating the
type employed in verbal learning studies was also developed. This
variable is referred to as the mean percentage of exact reaponses
(Appendices M, N, and O) and measures the ability of the §_to respond
in a verbatim manner.

As has been mentioned, §s were asked to write down, using their
own language, the meaning of the stimuli before and after the learn-
ing task. The‘Ss answers were classified as incorrect, partially
correct, or correct and were given the scores 0, 1, or 2 respectively.

Therefore, two additional dependent variables were available for
each stimulus. One measuring the extent the definition was known
prior to the learning task and the other measuring the correctness of
the definition after the learning task.

The independent definition variables are the types of defini-
tions, stimulus meaningfulness, reSponse meaningfulness, kind of sub-
ject matter, and length of the definition. However, each one of these
definition variables has its own levels. The levels of types of defi-
nitions are H-H, L-H, HAL and.LAL. Stimulus meaningfulness or
response meaningfulness variable has two levels, namely H or L. Again
it is to be remembered that the arithmetical definitions are classified

as numerical and geometrical definitions. Each list also has definitions

49

of two different lengths (levels) i.e., definitions with short and
long responses. The levels of these definition variables are repre-
sented in the control, verbal and picture familiarization conditions.

Inspection of the data presented in Appendices P, Q, and R,
revealed that definition attainment scores in some pre-tests were
not normally distributed. For example, in case of L-H pre-test
(Appendix T1) scores, all the answers were incorrect and received a
score of zero. Thus the lack of normality on the side of pre-test
definition attainment suggested the use of non-parametric methods.

First, the X2 test was used to compare the pre-test score dis-
tributions of the levels of each independent variable. For example,
there are four levels of types of definitions H-H, L-H, HAL and.L-L.
Each type of definition has its own pre-test score distribution, and
is summarized in terms of mean pre-test score and its standard devia-
tion as shown in Appendices P1, 01 and R1. The result of the x2 test
is used to determine whether or not these four pre-test score distri-
butions are similar in their diSpersion. The available means and
standard deviations are used to rank order the types of definitions.
This analysis is also repeated on stimulus m, response m, subject
matter types, and length variables.

Second, the X2 test was also used in order to compare the levels
of the definition variables with respect to their post-test score
distributions. For example, after the learning trials, the §s were
asked to give the meaning of the stimuli. These learned definitions
had either short or long reSponses. The distribution of the post-

test score of short definitions was then compared with the distribution

50

of the post-test score of long definitions using the X2 technique.
Means and standard deviations of each distribution presented in
Appendices P2, Q2 and R2 are used to determine the superiority of
one of the length levels in post-test definition attainment. Such
analysis can be used to study the differential effect of the levels
of each definitional variable on definition attainment.

The third aspect of this part of the study is concerned with
the effect of the learning task on definition attainment by compar-
ing the distributions of pre- and post-test scores. The X2 tech-
nique was also used to test whether or not there was a significant
difference between the pre-test scores and post-test scores. Such
comparison is made using the answers of the §s who received verbal,
picture, or no familiarization. Hence, for each type of familiariza-
tion there is a X2 value which shows the degree of change in defini-
tion attainment due to the familiarization procedure and the learning
trials.

In the previous analyses the pre-test or post-test scores were
pooled over all the types of definitions. It was not possible, then,
to study the effect of the independent definition variables on each
of H-H, L-H, H-L, and L-L lists. Therefore more Specific analyses
were performed in order to study the distributions of the pre-test
or postatest relevant to the types of definitions under each famil-
iarization treatment.

The independent variables are (1) types of definitions (H-H,
L-H, HAL, LmL), (2) stimulus meaningfulness (H, L), (3) response

meaningfulness (H, L), (4) subject matter (numerical, geometrical

51

definition), and (5) length (short, long). But in the analysis of pre-
or post-test scores of each type of definition, the variables (1) types
of definition, (2) stimulus meaningfulness, and (3) response meaning-
fulness are excluded: and the analysis is limited in order to study the
effect of the levels of either subject matter or length variables on
pre- or post-test scores of each type of definition under the famil-
iarization treatments (Appendices S, T, U and V). In addition, X2
tests of both pre- and post-test scores for each type of definition
under the familiarization treatments were computed in order to give

an indication of degree of change in attaining these definitions due

to familiarization and learning.

It will be observed that the results of the first part of the
design concerning the generation of definition meaningfulness are pre-
sented in Appendix F. The results of the second and third part of this
study related to the Se answers in the test trials and the Se defini-

tional attainment will be presented in the following chapter.

CHAPTER III

RESULTS

The presentation of the results will follow the design and
statistical procedure as shown on page 45. The first portion of the
results is reported in Appendix F, but a brief summary will be given
below. The second and third parts are concerned with the results of
Ss' responses in the actual learning task and in the pre- and post-

definition attainment conditions.

DEFINITIONAL MEANINGFULNESS

The following discussion is centered around the scaling of
definition meaningfulness (m). The distribution of stimulus mean-
ingfulness was bimodal, while the definition or response meaningful-
ness distributions were found to be approximately normal. The vari-
ance of stimulus meaningfulness is significantly greater than the
variance of either the reSponse m or definition m. Mean m of the
definitions is found to be significantly higher than the mean m of
the response (definien) at the .01 level, and the mean m of the
stimulus (definiendum) at the .05 level. However, the mean m of
the definiendum is not significantly different from the mean m of
definien at the .05 level. In addition m of numerical definitions
is greater than m of the geometrical definitions at the .01 level.

Concerning the number of letters and meaningfulness, the

results indicated that meaningfulness of neither arithmetical nor

52

53

numerical stimuli (definiendum) correlates with their number of letters.
On the other hand, there is a significant correlation between stimulus
(definiendum) m of geometrical terms or the scaled vocabulary m and
number of letters. There is, moreover, significant correlation between
the response (definien) number of words and reSponse m in case of
arithmetical, numerical and geometrical items. The high positive cor-
relations showed that the shorter the response the higher was its mean-
ingfulness.

While the intercorrelation coefficients between stimulus m, or
reSponse m, and their correSponding lengths are not significantly dif-
ferent from zero at the .01 level, it is found that each one of these
variables correlates significantly with definition m. For example,
definition m correlates significantly and positively with stimulus m.
The significant correlation between m of the definitions indicated
that when definition m was high response m was high, reaponse stan-
dard deviation was small, and reSponse number of words was few.

Investigation of the results shows that partialling any group
of variables out of the correlation of definition m with other vari-
ables does not change the zero order correlation coefficient. How-
ever, the correlation of response m, and reSponse number of words
changes significantly, when definition m or definition m plus other
variables are partialled out, and dropped to a value of which is not
far from zero at the .05 level._ Again, reSponse m and stimulus m have
been found to have an insignificant correlation. Once the definition
m or other variables beside definition m were partialled out, all the

new correlations differed significantly from their zero order

54

correlation. These results emphasize the role of definition m and its
relation with response m or stimulus m. For example, the new higher
correlations resulted from partiallizing definition m may mean (1)

when response m was high the length of the response was not necessarily
high or low (2) when response m was high, the stimulus m was also high.

The composite definition m showed resemblance to the actual
definition m except that the former correlated with stimulus m higher
than its correlation with response m at the .01 level of significance.
But in case of definition m, the entire preceding statement is reversed
except that the difference is not significant at the .05 level.

The meaningfulness value of the composite reSponse is obtained
by adding the m of the individual words (vocabulary) which compose the
response. It has been noticed that the composite reSponse m correlates
significantly at the .01 level, only with the standard deviation of
either stimulus m or response m. However, the correlation between

response m and composite reSponse m is almost zero at the .05 level.

DIFFERENTIAL EFFECT OF MEANINGFULNESS OF DEFINITION COMPONENTS ON

SUBJECTS' LEARNING

This part will present separately the results of control, verbal
and picture familiarization on the §fs test-trial scores. In present-
ing each of them, Kendall rank order correlation coefficients between
the actual arrangement of the types of definitions according to the
mean.percentage of correct reSponses and the expected theoretical
orders will be computed. In instances for which the correlation proved

to be perfect, z score will be used to determine whether for each type

55

of definition the proportion of correct responses is significantly
greater than the other proportions as designated in the theoretical

order.

A. Rank Order offthe Types of Definitions According to the Percen-
tage of Correct Responses for the Control Treatment

Correct reSponses have been defined (page 45) as the answers
which were exactly similar to the learned responses, or the reproduc-
tions which were almost similar to the exact reSponses, but stated in
slightly different forms. It contains the percentages of correct
responses for the short, long, numerical, geometrical and the comp
bined definitions. The next to the last column in Appendix J contains
the sum of the percentages of correct responses per test trial. This
sum is apprOpriate because the number of responses per test trial is
the same for all test trials. The last column represents the mean
percentage of correct responses of the test trials, and these values
will be used to determine the rank of the definition types.

' Table 2 presents the rank order of the definitions order accord-
ing to the percentage of correct reSponses. The rank order of the type
of definitions has been correlated with both the rank order suggested
by the theory that emphasizes the role of response (H-H, L-H, H-L, LAL),
and the suggested order of the theory emphasizing the role of’stimulus
(H-H, HAL, L-H, L-L). The Spearman rank order correlation coefficient
could have been used to test the degree of association between either
of the suggested findings and the present findings but its available
probability table covers only the case of perfect correlation (Siegel,

1956: P. 285). Instead, Kendall rank order correlation coefficient

H23 no. 23 e. a... 3.3383...

 

 

 

 

 

 

SH. -58. s3 . Se. m e m a defiance defleeaeoe
Nee. .584 a3 . So. a N m a scanning defines
SA . new. «so. 584 e n N a eeflaaaen use.”
men. nmm. New. -53. n a N a codpgon .395
Sn . Se. mi. .084 e m N a 23353 8588

mam“ h. Mm“ t

3 a? nun gun at" mum on; we En.
m m l m e 8 . e n N . a

0:81:95. H9555

on» ca. 235.509 no was on» .3.“ @0939QO 900.200 ho owapcoaom so: 23 .wo .320 E "N am.

56

57

is used and designated by']’ in Table 2. The probability of obtaining
the “V value, for one tailed test is shown in columns seven and nine
of the preceding table (Table 2). The .05 level has been chosen as
the level of significance of’l’ .

Investigation of Table 2 reveals that the theory emphasizing the
m of the response is supported in two cases: First, when all the defi-
nitions~of the task are considered: and second, when the long defini-
tions are considered. The same theory has been rejected in three other
cases, namely with short, numerical and geometrical definitions. On
the other hand, the theory which emphasized the m of the stimulus
position is accepted only in the case of numerical definitions and is
rejected in the remaining cases.

Again, while each theory suggests a different arrangement for the
types of definitions, both of them assume that there must be a signifi-
cant difference between any two types of definitions. For example, the
theory which emphasizes the role of response m, and which suggests that
the following order H-H, L-H, H-L, L-L, requires a significant differ-
ence between H-H and.L-H, H-L, or LAL as well as between any of L-H,
H-L and.LAL conditions. In other words, the proportion of correct
responses per test trial of H-H definitions must be significantly greater
than the proportion of correct responses per test trial of either L-H,
HAL, or LAL definitions. The presence of significant differences insures
that the discrepancies between the proportions of correct responses are
not due to chance but are caused by differences in the location of m.

The combined and long definitions have been shown (Table 2) to

follow the sequence which is stated by the response m theory, while

 

58

numerical definitions follow the sequence which is stated by the stimu-
lus m theory. The normal distribution table and its standard score z
are used to test whether each proportion of correct responses is sig-
nificantly greater than the other proportion in the order designated

by each theory. The results are shown in Table 3.

TABLE 3: z Values for the Differences Between Proportions of Correct

Responses in the Control Treatment

i-.

 

 

 

 

 

Types of Definition L-L H-L L-H Kind of Task

 

H-H 13.0** 12.8"".I 8.3** Combined Definitions
17.8" 12. 5" 8.0" Long Definition
10.6** 1.85* 3.3** Numerical Definition

 

 

L-H 5.4** 4.0MI Combined Definitions
8.3** 4.2** Long Definition
6.7** 1.64* Numerical Definition

HAL 1.46 Combined Definitions
4.2** Long Definition
8.5** Numerical Definition

 

* Significant difference at the .05 level (one tail test)

**Significant difference at the .01 level (one tail test)

Table 3 reveals that there are significant differences between
the types of definition, using the combined definitions, except in

the case where HAL is not significantly greater than.LAL at the .05

 

.\°
4.

8C)

60]

40.

20, y

I ‘-
/ . ' :‘v”

f
// ‘H LC

‘L4_C

CODIFO|

 

!OQ '

 

2 if?)

-w‘.

 

 

 

 

o—---—@H°HC
’ a” ”_.L44C
8’), ,h/ 'r _.~ “.4 H'LC
J /" ’2}: - a " ’ L-LC
f3"; / a ’
i ’ ,2
a 24 n
‘l
4 9\CI‘T
(St—fin 1 f v v
'CXL 1 2 3 4 5
.————aFFFK:
91’) -V'
e / ,.- ...L°HC
6k; ,// 'aH'LL:
, fl ‘1
‘ / /
4C)
a ,. x "vL'LC
T r - / /
ZCL // .1:
. g' “(z Long
0 T4 ’4 . as r
O ! 2 3 4 5

f 1".er .‘3

FA

.95?

Trial

I" Y“- Q T to 0*

O""’:.

 

 

 

 

 

/‘ / .xe—“r‘H-H C
SO ///’ 1p" “qJHC
so 1" AH LC
/ j)? ~~ILvLC
_ AK
404 1” /:./
y.
‘ Ifi//
20y z/
j Control
0 T V fl 1
“301 l 2 3 4 5
H-HC
i /_q—-"’l
801 8/ r fiH-LC
_’/» "0"" L’HC
J .fia,’
6O.4 l/L’I'
/ R
I
‘ .L-LC
404 cf
1 ‘_ I"
, gr
2Q a "
y / Numerical
i 1’ 3 4 5

 

 

‘ I .L-HC
A ./ “
60.. / ’m/ ,‘ L'LC
j .4“ ’H‘LC
/ / ./
1’ . /
W Tr’ "I"../ f j
1 , / ””'
2Q d, ,/
j / Gaomtrlcel 1
/.
O l 2 3 4 5

Test Trial

T’s Pernen’are of Jorrect Responses Per Test Trial For

 

60

level. However, in the case of long definitions all the differences
are significant at the .01 level.

It has been mentioned that the results of numerical definitions
followed the theory which emphasizes the stimulus position. The dif-
ferences between the proportions of correct responses in this case
are significant at the .01 level except the difference between HAL
and L-H which is significant at the .05 level.

Figure 1 shows the corresponding curves of the preportion of
correct responses per test trial. In general these curves confirm
the previous statistical results which have been obtained (Table 2)
from the mean proportion of correct responses over all the test
trials. However, some curves are overlapping, and it is difficult
to derive from them any statistical conclusion similar to those pre-
sented in Table 3, other than a general knowledge of the arrangement

of the types of definitions for each kind of task.

B. fiagk Order of the Types of Definitions According to the Percen-
tage of’Corgegt Responses for the Verbal Fgmiliarization

Appendix K shows the mean percentage of correct responses of
the test trials for the Se who received verbal familiarization. The
order of the types of definitions (H-H, HAL, L-H, L-L) according to
the percentage of correct responses, and under different tasks, short,
long, numerical, geometrical, and the combined definitions are shown
in Table 4.

The theory which emphasizes the role of response position in
learning suggests the following descending order: H-H, L-H, HAL, LAL.

When the response member is familiarized, the H response will become

Hosea no. one e. one eneeaoaeman.

 

 

 

 

 

 

 

 

mam. mom. mNo. ooo. N m e a neaeacaeeo Heeaeeeaeeo
Noo. .ooo.a no“. moo. m a e N ceaeaeamoo Heeaeesez
mum. nmn. mNo. ooo. o N m a neaeaeaeeo when
man. man. nNo. ooo. N m o a neaeanaeeo eaeem
non. aoo.. man. man. m N. o a neeaeaoaeeo oeeanseo
man h. an e
peasanea neaeaaem .mmmmmmmmnummmmmmm ewe awe mug mum anus no none
m! ,1. o a o w, e m1 N a

eeaeeeaeeaaaeea Heenes

one. 5.” meowpficflon we neg on» no.“ newcommom poghoo no omenseaem sees one mo .320 View 3 mam—<9

61

62

L and vice versa. However, it is desired that after familiarization
the theoretical descending order be preserved. Hence, the original
definition order must be H-L, L-L, H-H, L-H which becomes H—H, L-H,
H-L and.L-L after the process of familiarization.

Column 6 of Table 4 shows the Kendall rank order correlation
coefficient between the hypothetical order explained above and the
experimental results. All the reported correlations are zero except
in case of short definitions which is .333 and numerical definitions
which is .667 and are not significantly greater than zero at the .05
level.

The theory which emphasizes the stimulus position may be con-
firmed in the case of verbal familiarization of the response. The
expected order of the types of definitions is H-L, H-H, LAL and L-H
which hypothetically would become H-H, HAL, L-H, and LAL after
receiving verbal reSponse familiarization. The Kendall correlations
between this expected order and the actual orders are shown in column
8 of Table 4. All of the reported correlations have the value of .333
(short, long, geometrical definitions) and .667 (combined definitions)
but they are not significantly greater than zero. The only exception
is the case of numerical definition which correlates perfectly and is
significantly greater than zero.

Table 5 shows the two values for the differences between propor-
tions of correct responses of the types of numerical definitions when
verbal familiarization was received by the Se. Investigation of Table
5 shows that the proportion of correct reSponses of'L-L types of defi-

nitions are not significantly greater than the proportion of correct

63

responses of L-H type of definitions at the .05 level. The differ-
ence between the percentage of correct reSponses of'L-L is signifi-
cantly greater than the percentage of correct responses of H-H type
of definitions at the .05 level. The other two values reported in
Table 5 show that the differences in percentages of correct responses

for the other types of definitions are significant at the .01 level.

TABLE 5: 2 Values for the Differences Between Proportions of Correct

Responses for Numerical Definitions with Verbal Familiariza-

 

 

tion
Definition Type LAL H-L L-H
H-H 2.123* 6.019** 4.426**
L-H 1.48 10.48**
HAL 9.102**

 

* Significantly greater at the .05 level

I”Significantly greater at the .01 level

Figure 2 shows the corresponding curves for the verbal familiar-
ization procedure. Investigation of the learning curves for the comp
bined definitions, shows some irregularity in the percentage of cor-
rect reSponses of the L-H type of definition. The order of the types
of definitions in the first test trial according to their percentages
of correct responses is as follows: H-HV, HALV, L-LV, L-HV. In the
second test trial the arrangement of the percentages of correct
responses is the same as the first test trial, except that the percen-

tage of correct reSponses of L-HV showed superiority to the other three

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

7. l” “M 7.
* H-HV e 1 QH-H v
801 M", 04 , a
4 //r- W 4 r W
./ ,' H x
b t fl/ / / f'L‘L 'v' 50+ /" / // "/wL. L J
.— 4 / ..._ _.
I gV€;}(/ 4 4 [1"()
4Q //"/., ‘V Cy I’f / ‘u
/’/ // n \ 4 //( N2,
201 1.17 ‘L H V 20,
4 {1 ‘v'erbai i
O 'r T n 'T T O
I 2 5 4 5
‘00? 48‘” V I00
« .1ng LV 4
eo1 (3 /77 L LV sol
4 f," ,/ 4
60 j/,’ I, !' ,AL‘HV 601
'1 /// / _ _ .ar -’ l
40 [l/ ,' 4-01
4‘ *
2rd \. ‘ 20
y " 5'3““? 4
OJ» —.- i j Y Y e a 4 r. A
4 ’//‘F “\QHHV
Boy 80. /
‘ 4 ,.H-Lv
60+ 601 V J/JfL-LV
// L'HV
4 ‘ .--"",’/ t
401 W ._r’/"
. . /, ’,/'
201 20. "/x/
, 1 «’7' Geometrical
O 044 v 1 “r T
O O l 2‘ 3 4 5

 

‘eut Trial 195’ Trial
.;L~e “: I"e Percentage of Correct ReSponses Per Test Trial For
Vertal Familiarization

ii,

65

types of definition. Thus the arrangement of the types of definitions
according to their percentage of correct responses in the second test
trial is as follows: L-HV, H-HV, HALV, LALV. Again in the third,
fourth and fifth test trials the percentage of correct responses of
L-HV became lower than that of the other three types of definitions.
Thus the new arrangement of the types of definitions in the last three
test trials is as follows: H-HV, HALV, L-LV, and.L-HV. However,
while the percentage of correct responses of H-HV, HALV, and L-LV
increases in every successive test trial, it is noticed that the per-
centage of correct responses of L-HV decreases in the successive test
trials after the second one. Looking back over the §fs test booklets
it appears that the,§s reported increasingly confused answers with
L-HV as there were more learning trials. So instead of having an
increase in the percentage of correct responses in every successive
test trial, there was a decrease as a result of the increasing con-

fusion.

C. ank Ogder of the Types of Definitions Agcording to the Pgrcen-
tgge pf Correct Responses for the Eicpure Eamilippipption

The expected order of the types of definitions, in this case,
is similar to that of the previous case of verbal familiarization.
Again the theory which emphasizes the role of response position
anticipates the order of the types of definitions as follows: HAL.
L-L, H-H, L-H, while the theory which emphasizes the role of stimu-
lus position predicts this order: HAL, H-H, L-L, L-H. Both theories
predict a decrease in the H reSponse (H to L) and an increase in the

L response (L to H) as a result of picture familiarization.

66

Table 6 shows the Kendall rank order correlation coefficients
between the actual results of picture familiarization and the theo-
retical order. The results shown in Table 6 revealed that the Kendall
rank order correlation coefficients of the expected and the actual
orders are not significantly greater than zero. Furthermore, some
show a correlation of zero or a negative correlation.

Figure 3 shows the percentage of correct responses per test
trial for the different types of definitions. The curves for the
combined definitions, short definitions, and geometrical definitions
follow the pattern H-HP, LALP, L-HP, and H-LP. Yet this pattern is
not consistent with either the stimulus or response theories as has

been confirmed by the findings in Table 6.

D. Summapy ofpthe Resglts When the Correct Respogses are Taken as a
Dependent Variable

The answers of the Se in the test trials have been classified
into several categories. The answer which is an exact reproduction
of the reSponse, or similar to the response but stated in a slightly
different manner, is defined as a correct response. The preceding
three sections of the results show to what extent the two theories
emphasizing either the stimulus position or response position were
confirmed by using the correct responses as a dependent variable.

For example, the theory which emphasizes the role of stimulus
was tested five times (with the combined definitions, long, short,
numerical and geometrical definitions) per familiarization condition
(control, verbal, or picture). Thus it was given fifteen chances of

possible confirmation. Yet it has been accepted only twice--in case

Heeea no. one no see eeeeaoanmam.

 

 

 

 

 

 

 

 

 

“No. ooo. soa. soo.u N o m a neaeaeaoeo Heeaneeaeeo
non. moo. non. moo. m N o a :eaeanaeeo Heeanessz
man. man. nNo. ooo. N m o a neaeaoaoeo moon
mNo. ooo. non. noo.- N e m a ooaeaeaoeo eaeem
mNo. ooo. non. noo.- N o n a neeaeaeaoeo oeeaeaeo
4.”me h fin .r

:eaeaneo neaeeaem eeaeaneo ence nee ewe gum m-a mum anus we oeae

o m N o Le, m o m N a

ceaeeeaaeanwsea oneeeao

one. 5m oneaudfifiofl .Ho momma one. no.“ monsonmoom pooeaoo mo emancooaom «So: one .wo pogo xsom up may

67

 

 

 

 

 

 

 

 

 

 

 

 

1 H-H P J
80* /———’ 8Q
~f/ 1.L'H P 4
o I _ P
6Q] // f , AL L 604
/ ,v/ » H-LP
i r 7
4Q / /:7 4;).
go] 1’77 20*
/ 4
1 ’7 chfurc , ’7 PICTUFC
Qr—fi' T W ' T '_"‘—1 O r Y T ”‘fi "'—‘T‘-'—‘—"“1
ICKI l 2 5 ‘1 5 'CXl I 2 3 4 5
"‘\!H ‘Hp
'1 ’7/ l b-‘aH'Hp
Bow //// :33}: 9’” .H- L P
. x77 a ./. s
i ,./ 7 f’ ”x, / H l P 1 /: ,é lei-47::
60, '7. 60.4 ’/ , . ,’7 L
/f "/'.1" / ,- 5
/J"‘7 < ’/i ,g x)”
4 j 4 f!
20, 4 20. 1'“
J Short J Numemcel
I) v!— T— T T“ ‘ r Oik—‘v- ‘1 ““T"““'_"" ‘ —‘T‘__""—'1
LDC-J ‘ 2 3 4 5 IQQ I 2 3 s 5
,»——-v———'“"H'HP
,1 .MH'HP * "’
8% 6g / L-L P
" ‘ ”"7 fiuw
60 ,L‘H P 6Q )./ (.7
1 , 'PL-L p ..-” '21”
40" -'/ ‘ AH-LP 40' ./,;‘/‘“1 .AH'LP
in] e/ , /a/’
‘ /.% 1 ' ../
20., // All 20“ [I / /
l /7-f Long J gi/A/ Geome+ricel
o f» ’7 '
O I 2 3 4 5 O l 2 3 4 5
“ ."ria1 Test trial
1?: 'rw Percen‘nfs of Correct fies “EIQS Per Test Trial for

- ‘o' .s‘ pvr' ' .: fi§%.,'.
A " W 1U? 2.1 _1.

69

of control and verbal familiarization of numerical definitions. This
is also true in the case of the theory which stresses the role of
response meaningfulness. It has been tested fifteen times and is
accepted twice--for the control treatments with the combined and the
long definitions. However, in each of the confirming cases there is
some overlap among the types of definitions when in fact the theory
predicts no overlap.

Possibly it is the nature of the dependent variable which is
responsible for the failure to confirm either one of the theories
with a high degree of consistency. For this reason the following
analysis will be limited to the exact responses, as a dependent vari-
able. This is because the two theories have been built by verbal
learning psychologists who consider the dependent variable to be the
exact response. For example, verbal learning psychologists assume
the S} answer which is an exact reproduction of the response as their
criterion measure of learning, and any answer which differs from the
exact reSponse is assumed to be wrong. In this study accepting the
answers which are similar to the response but stated in a slightly
different manner as dependent variable violates the condition of
analogy between paired-associate and definition learning experiments.
To test again whether the suggested arrangement of types of defini-
tion correlates significantly with the actual arrangement, the mean

percentage of the exact responses is considered as the dependent vari-

able.

70

E. 'nk 0 e f the T es of Definitions Accordi to the Perpppppgp.
of the Expgt Regponses for the Control Treatment

Appendix M shows the percentage of exact reSponses per test. The
mean percentage of exact responses is shown in the last column of the
appendix, while the rank order of the types of definitions under the
five different task classifications (combined definitions, short, long,
numerical and geometrical definitions) is shown in Table 7.

Inspection of Table 7 reveals no significant correlation between
the arrangement excepted according to the theory which emphasizes the
m of stimulus, and the actual arrangement of the types of definitions
when based on the mean percentage of exact responses. Thus it can be
said that the theory which emphasizes the m of the stimulus position
has not been confirmed when exact reSponses were considered. 0n the
other hand, Table 7 shows that the response theory is confirmed in
three cases at the .05 level, namely, with the combined definitions,
short definitions and long definitions.

Table 8 shows the 2 values for the differences between propor-
tions of exact responses for the control treatment. Long definitions
showed, according to Table 8, significant differences between the
types of definitions at the .01 level. For short definitions, at .05
level, the difference is significant between L-H and 8-H and insigni-
ficant between L-H and HAL types of definitions. However. other dif-
ferences are significant at the .01 level. In the case of the combined
definitions, the differences are significant at the .01 level with the

exception of that between.L-H and HAL which is significant at the .05
level.

Ho>oH no. on» we sop pcoowMfiCMflm*

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mmm. mmm. 50H. new. m n N H coapwswmom decompoEooo
mam. mmm. mNo. ooo. o H N m eeHeHeHHeo HeeHeeasz
soH. moo. Noo. .ooo.H o m N H ooHHHcHHeQ when
ooH. ooo. Neo. .ooo.H o m N H eeHeHeHHeo eneem
noH. ooo. Noo. .ooo.H o m N H neeHeHeHHeo oeeaoseo
hum“ .r Mm“ t
ceHeHnea neHoEHHm oeHeHnei eneennem Hne gum mun mum some he ocHe
o o (N o awn. m, “Mir. m N H

pcospooae Hopscoo

esp CH meowprfimoQ mo momma onH new noncodmom Hooxm mo owencooaom coo; one mo Hoppo meet an mqp<e

71

72

TABLE 8: z Values for the Differences Between Proportions of Exact

Responses for the Control Treatment
M

 

 

 

 

 

Types of
Definitions L-L H-L L-H Kind of Task
H-H 11.465** 6.535** 5.332** Combined Definitions
8.010** 2.660** 1.980* Short Definition
8.588** 7.183** 4.341** Long Definition
L-H 7.265** 2.241* Combined Definitions
6.150** .660 Short Definition
4.372** 2.900** Long Definition
H-L 5.025** Combined Definitions
5.480** Short Definition
2.718** Long Definition

 

* Significant difference at the .05 level (one tail test)

**Significant difference at the .01 level (one tail test)

F. Rank Order of the Types of Definition Accordipg to the Percentpgg
of the Exact Responses for the Verbal and Picture Familiarization
Treatments

Table 9 shows the actual rank order of the types of definitions
according to the percentage of exact reSponses in case of verbal famil-
iarization (Appendix N). InSpection of Table 9 reveals that numerical
definitions are consistent with the theory which emphasizes the response
position with verbal familiarization. However, it has been previously

shown that the same numerical definitions were also consistent with the

Ho>oa no. was s. gas guacamacmam.

 

 

 

 

 

 

 

mum. mmm. mNm. ooo. N n s H coapfisaeoa Haoaaposoou
mam. mmm. Nso. .ooo.H N a s m coasacacoo Hecatmssz
man. man. man. mmm. : N m H compacaeom mace
mmo. ooo. mmm. mmm. H m d N coapficamoa unonm
no“. new. man. mmm. m N s H mcoapacaeoa eo:NQEoo
Immmi \r Mm“ \r

coasamom maaasapm soapamom omco mom Aug qwm mug mum Name No ezag

him m m o m s m N a

aoapauauaaaasae flanges

02p cw «COfiafiCfimoQ mo wonky you momcommom powxm Mo omwQCOOAom coca saw no hopho xcmm no mqm<e

73

74

theory of the stimulus position in control and verbal familiarization
treatments when the correct reSponse was used as a dependent variable.
Other Kendall rank correlations reported in Table 9 are not signifi-
cantly greater than zero.

Table 10 shows the 2 values for the differences between propor-
tions of exact responses for numerical definitions with verbal famil-
iarization. Investigation of this table reveals that the proportion
of exact responses of L-L are not significantly greater than those of
H-H, in case of numerical definitions under verbal familiarization.

The other reported differences are significant at the .01 level.

TABLE 10: z Values for the Differences Between Proportions of Exact

ReSponses for Numerical Definitions with Verbal Familiari—

 

 

 

 

 

 

 

zation
Types of Definitions L-L HAL L-H
H-H .232 7.617** 4.8h2**
L-H 5.791“ 23.313"
H-L 8.056**

 

**Significant difference at the .01 level

Table 11 shows the Kendall rank order correlation coefficients
of the actual order of types of definitions with the expected order
of each theory. The actual arrangement of types of definition is
based on the mean percentage of exact responses reported in Appendix
0. The results of Table 11 showed that all the correlations are not

significantly different from zero. This result is consistent with

0;;

-—.~fi3 lu—hAVdJfl-Lrvuafl_

“as

In (of— > L.

(inbrh

“ca-{ICALH‘AIF‘H~

Ah (1.15!an

E‘

C

“II ‘I.

HosoH no. one s. an» scaOHeHcmHm.

 

 

 

 

 

 

 

 

nNo. ooo. mam. mmm.- N s m H coHaHnHeon HuoHusoaooe
man. mom. mNe. ooo. m N s H =OHstHNoa HaoHuoasz
nNo. ooo. mam. nmn.- : m N H :OHpHcHNon mace
man. man. nNe. ooo. N m s H :OHpHcHaoo spasm
man. man. mNo. ooo. N n a H acOHpHcHHon eocHnsoo
film} LK mama—”Mn LK

nanoum Ianoum

:OHpHuom usHasHsm - H.q ewe m-H m-m amae No ecHe
m m m o m 3 m1. N H

:ofipenauufifldaem unsavem

one nu“: «coaudcamoa mo momma you noncommom uoaxm no omeucoouom see: one mo nopuo xcam "HH mgm<a

75

76

picture familiarization when the dependent variable is the correct

response.

G. Summagy of the Results When the Exact Responses age Taken as a
Dependent Variable

In conclusion, when the exact response is used as a dependent
variable, the results do not at all confirm the theory which favors
the stimulus position. The second theory, namely, the theory which
emphasizes the role of m on the reSponse member is confirmed in three
cases (the combined definitions, short and long definitions) with
control treatment, and once (numerical definitions) with verbal famil-
iarization. The picture familiarization confirmed neither of the two
theories.

Thus, when each theory was given fifteen chances of possible
confirmation as the exact response of the definitions was recalled,
the theory which favors the m of the response position was confirmed
four times, while the theory which emphasizes the m of the stimulus

position was unsupported at all.

SUBJECTS' PRE- AND POST-DEFINITION ATTAINMENT

The presentation of the results follows the sequence which is
suggested in the design and statistical procedure (page 45). The
results of each familiarization treatment will be presented separately.
Under each familiarization treatment (control, verbal, picture) there
will be a discussion of (a) the relationships between pro-test scores

of the levels of each definition variable, (b) the relationships

77

between post-test scores of the levels of each definition variable,
and (c) the relationship between pre- and post-test scores of the
familiarization treatment. This will be followed by presenting (a)
the relationship between pre-test scores of the levels of subject
matter variable or length variable for each of the H-H, L-H, H-L,
and L-L definitions; (b) the relationship between post-test scores
of the levels of subject matter or length variables for H-H, L-H,
H-L, and L-L definitions; (c) the relationship between pre- and
post-test scores for H-H, L-H, H-L, and L-L types of definition.
The above sequence will be followed in presenting the results of

the control, verbal, or picture familiarization treatments.

A. The Relationship Between Pre-_gggrggst-Tests' Scores With Defi-
nition Variables in Case of the Control Treatment

Table 12 shows the X2 values of definition variables with
either the pre-test score or the post-test score for the control
treatment. It also shows the X2 value of pre- and post-test scores
for the same treatment.

Investigation of Table 12 shows that the §s’ pre-test scores
were significant at the .01 level with reSpect to types of defini-
tion, stimulus meaningfulness and length variables. As for the types
of definitions variable, Appendix P1 shows that the mean pro-test
score is .48, .34, .00, and .01 for H-H, H-L, L-H, and LAL respectively.
The arrangement of definitions suggests that meaningfulness of the
stimulus position in the pre-test is more concerned with the previous
attainment of the definition. This observation is confirmed by the

fact that the correlation coefficient between levels of stimulus

78

meaningfulness and pre-score values is .454 for N = 528 which is sig-
nificant at the .01 level. With respect to stimulus meaningfulness
variable, Appendix P1 shows that the mean pre-test scores for H and

L stimuli are .40 and .01. This indicates that definitions with
higher meaningfulness stimuli were better known in advance than the
definitions with lower meaningfulness stimuli. As for the length
variable, the mean pre-test score for short and long definitions are
.11, and .29 respectively. The longer definitions seemed to be better

known than the short ones.

TABLE 12: X2 Values of Pre- and Post-Tests' Scores With Definition

Variables, in Case of Control Treatment

 

 

 

Variables x2 with x2 with
Pre-Test Score Post-Test Score

Types of Definitions 127.02** 41.93**
Stimulus Meaningfulness 115.81** 12.95**
ReSponse Meaningfulness 1.90 14.99**
Subject Matter 1.39 3.75
Length 25.18** 5.28
Post-Test Score 24.87**

 

* Significant at the .05 level

**Significant at the .01 level

After the learning trials (Table 12) the §s' post-test scores
showed significant variation as a function of definition types, stim-

ulus m, and response m variables, while the length variables failed

79

to produce significant statistical variation. Types of definitions
and stimulus meaningfulness have exactly the same effect as in the
pre-test results, except that there is a tendency for the values of
X2 to be attenuated. The mean post-test scores for H-H, L-H, HAL,

and L-L, as shown in Appendix P are 1.53. 1.00, 1.11, 1.07 and for

2
H, L stimuli 1.29 and 1.03 respectively. These two results indicate
that definitions with higher meaningfulness stimuli were attained
better than the definitions with lower meaningfulness stimuli. The
reSponse meaningfulness variable showed that when meaningfulness was
high, the post-test score was also high (mean post-test scores for H
and L are 1.24, 1.09 reSpectively) and the X2 value became higher
than in the case of pre-test results.

Table 12 shows also that X2 test indicated significant varia-
tion between the two distributions of pre- and post-test scores at
the .01 level. This result suggests that there was a significant
change in attaining the definitions in the control treatment after
receiving the learning trials.

Table 13 shows the X2 values of pre- and post-tests' scores
with subject matter and length variables for each of H-H, L-H, HAL
and L-L definitions under the control treatment. Appendices $1 and
82 show the correSponding proportional distribution of pre- and
post-test scores for H-H definitions. Investigation of Table 13
and Appendices 81 and S2 reveals that the H-H type of definition had
varied significantly in case of either pre- or post-test scores with
length variable. The mean scores of short and long definitions in

the pre-test are .19 and .78, while in the post-test results they are

80

1.36 and 1.69 results. Then for the H-H type of definitions, the
definitions with longer reSponses were better attained before and

after the learning task than definitions with shorter reSponse.

TABLE 13: X2 Values of Pre- and Post-Tests Scores for Types of

Definitions Under Control Treatment

 

 

 

Types of Definitions Subject Matter Length Pre and
Pre Post Pre Post, POSt
H-H 1.74 3.12 30.46** 9.94** 15.10**
L-H .00 .16 .00 8.40* .00
HAL 2.09 6.67* 12.09** 7.42* 6.54
L-L 2.03 .31 .00 19.43** ”1.10

 

 

 

 

* Significant at the .05 level

**Significant at the .01 level

As for the L-H type of definition, there is a significant dif-
ference between short and long definitions with post-test score.
Appendices T1 and T2 show that mean post score for short and long
definitions of L-H are 1.07 and .93 reSpectively. Thus the L-H defi-
nitions with shorter reSponses were better attained than the L-H
definitions with longer reSponses.

In case of the HAL type of definitions, Table 13 shows signifi-

cant X2

values among subject matter levels with post-test score and
the length variable in both the pre- and post-test scores. Appendices
U1 and 02 show that H-L numerical definitions were better attained

than the HAL geometrical definition, where their mean post-test

81

scores are 1.27 and .94 respectively. It also Shows that the mean pre-
test score for short and long definition are .24 and .44, while they
are 1.17 and 1.04 for post-test scores. These latter means indicate
that definitions of long responses were better known before the learn-
ing task, and then the relation is reversed to indicate that defini-
tions with short responses were better attained after the learning
task.

The L-L definitions which have shorter responses were attained
better than.L-L definitions with longer responses. The mean post-test
scores for L-L short and long definitions are 1.37 and .67 as shown in
Appendix V2.

A difference is observed between the distributions of pre- and
post-test scores which is significant in the use of H-H, but not sig-
nificant for H-L. However, the X2 value for H-L pre- and post-test
scores approximates significance at the .05 level. In case of L-H and
L-L, the distributions of pre- and post-test scores are almost identi-
cal. The arrangement of the pre- and post-X2 values shows a signifi-
cant development in the §s attainment of the definition after the learn-
ing trials in case of H-H, reasonable development with H-L and quite
negligible development in case of L-L and L-H. This may indicate that
progress in attainment of the definition is more dependent on the mean-
ingfulness of the stimulus position than on the meaningfulness of the
response position. However, this finding was confirmed by the results
reported in Table 12 that preo or post-attainment scores were signifi-

cantly high when the stimulus meaningfulness was high rather than low.

82

B. The Relationship Between Pre- ang;3ost-Tests' Scores With Defini-

tion VariablesI in Case of Verbal Familiarization

TABLE in: x2 Values of Pre- and Post-Tests' Scores With Definition

Variables, in Case of Verbal Familiarization

 

 

 

Variables x2 With x2 With
Pre-Test Score Post-Test Score

Types of Definitions 149.58** 3.84
Stimulus Meaningfulness 68.32** 1.01
Response Meaningfulness 9.55 1.30
Subject Matter 1.13 3-92
Length 639* 18.90"
Post-Test Score 36.22**

 

* Significant X2 at the .05 level

**Significant X2 at the .01 level

Table 14 shows the X2 values of definition variables with either
the pre-test score or the post-test score for the verbal familiariza-
tion treatments. Appendices Q1 and Q2 also show the pre- and post-test
scores for the same treatment. Investigation of Table 14 shows that
the fis' pre-test scores had varied significantly at the .01 level with
types of definitions, stimulus meaningfulness variables and at the .05
level with length variable. As for the types of definition variable,
Appendix Q1 shows that mean pre-test scores are .42, .00, .52, and .06
for H-H, L-H, H-L and.L-L respectively. In case of stimulus milevels,

the higher pre-test scores were associated with the higher meaningfulness.

"I

 

p

83

The mean premtest scores of H and L stimuli are .35, and .04. The
response m shows peculiar relationship with pre-test scores. The mean
pre-test score of H and L responses are .17 and .25 reSpectively. Thus
it could be stated that definitions with low meaningfulness values were
known better before receiving learning trials than those of the high
meaningfulness values. The definitions with shorter reSponses were
relatively less known before the experiment than the definitions with
longer responses. Appendix Q1 indicates that the mean pre-test score
for short and long definitions are .17 and .26 respectively.

The post-test scores were significantly different with the levels
of the length variable as shown in Table 14. Appendix Q2 shows that
the mean post-scores of short and long definitions are 1.29 and 1.06
respectively. Thus the definitions with short reSponses produced sig-
nificantly higher scores in the post-test than definitions with long
reSponses.

The X2 value as shown at the end of Table 14, of the pre- and
post-test scores indicates a significant development in the §s' attain-
ment of the definitions after receiving verbal familiarization and a
number of learning trials.

Table 15 shows the X2 values of pre- and post-tests scores with
subject matter and length variables for each of H-H, L-H, H-L, and L-L
definitions under verbal familiarization. Appendices 81' 82, T1, T2,
U1, U2, V1, and V2 show also the corresponding proportional distribu-
tion of pre- and post-test scores for H-H, L-H, H-L, and L-L types of

definitions respectively.

84

TABLE 15: X2 Values of Pre- and PostuTests' Scores for Types of Defi-

nitions Under Verbal Treatment

 

 

 

Types of Definitions Subject Matter Length Pre and
Post
Pre Post Pre Post
H-H 3.05 7.39* 10.02** 5.40 24.64**
L-H .00 .62 .00 5.22 .00
H-L 2.21 15.92** 9.02* 11.42** 22.60**
LAL 6.21* 2.20 6.21* 28.44** 3.19

 

 

 

 

* Significant at the .05 level

**Significant at the .01 level

Study of Table 15 and Appendix 81 shows that H-H definitions
exhibited significant variation in the case of pre-test score with
the length and subject matter variables. The mean scores of short
and long definitions in the pre-test are .21 and .62 respectively.
Then for H-H definitions, the definitions with longer reSponses were
known better before the learning task than definitions with short
responses. In addition, the H-H geometrical definitions were
attained after receiving verbal familiarization and learning trials
better than numerical definitions since their mean scores are 1.33
and 1.05 respectively.

As for HAL definitions, Table 15 and Appendix T1 show that they
had significant variation in preutest score with the length variable.
The mean scores of short and long definitions in the pre-test are .42
and .61 respectively. It seems that definitions with long reSponses

are better known in advance than definitions with short reSponses.

85

However this relation did not change after the familiarization and

the learning trials, confirming that HAL definitions with long reSponses
are better attained than short definitions. Appendix UZ shows that

the mean post-test scores of short and long definitions are 1.13 and
1.19 reSpectively. Moreover, the H-L definitions showed significant

2

X value with post-test score and the subject matter levels. The mean

post-test score for HAL numerical and geometrical definitions are 1.42
and .90. So, the H-L numerical definitions seemed to be better attained
after receiving verbal familiarization and a number of learning trials
than the HAL geometrical definitions.

Table 15 and Appendices V and V show that L-L definitions pre-

1 2

or post-scores exhibited significant variation when covaried with the
length variable. In either pre- or post-test scores, the short defi-
nitions of L-L produced higher scores than the longer ones. The mean
pre-test scores for short and long L-L definitions are .11, .00 and
their post-test scores are 1.57, .90. Table 15 indicates also that
L-L definitions‘ pro-test scores are significantly different in case
of subject matter. However, this significant difference is in favor
of geometrical definitions over numerical definitions.

The comparison of pre- and post-test scores of the types of defi-
nitions showed significant differences in case of H-H and HAL and insig-
nificant differences with.L-L and L-H. Thus, with verbal familiariza-
tion, the m of stimulus position in the combined definitions is responp

sible for the progress of definition attainment after the learning
trials.

86

C. The Relationship Between Pre- and Post-Tests' Scores with Defini-

tion VariablesI in Case of the Picture Treatment

TABLE 16: X2 Values of Pre- and Post-Tests' Scores with Definition

Variables, in Case of Picture Familiarization

 

 

 

Variable x2 with 12 with
Pre-Test Score Post-Test Score
Types of Definitions 165.71“I 19.49“"I
Stimulus Meaningfulness 1u2.17* 3.43
Response Meaningfulness 5.87 8.47**
Subject Matter .81 3.72
Length 16.16** 2.03

Post-Test Score 41.10*‘

 

* Significant'X2 at the .05 level

**Significant X2 at the .01 level

Table 16 contains the X2 values of pre- and post-tests' scores
with definition variables in the case of picture familiarization.

Appendix R shows the prOportional distribution of pre- and post-tests

1
scores with definition variables in the case of picture familiariza-
tion. Both Table 16 and Appendix R1 will be used to clarify the rela-
tionships between definition variables and either pre-test or post-test
scores.

According to Table 16 the pre-test scores for the picture famil-

iarization follow exactly the same pattern as those of the contro1 treat-

mernn There are significant differences between the pre-test scores of

87

the levels of the types of definitions, stimulus meaningfulness, and
length variables. As for the types of definition variable, Appendix
R1 shows that the mean pre-test score is .63, .00, .35, .09 for H-H,
L-H, H-L, and LAL respectively. The arrangement of definitions (H—H,
H-L, L-L, L-H) suggests that the meaningfulness of stimulus position
in the pre-test is more concerned with the previous knowledge of the
definition. With respect to stimulus meaningfulness variable Appenp
dix R also shows that the mean pro-test score for H and L stimuli are
.51, and .04. This indicates again that definitions with higher mean-
ingfulness stimuli were better known in advance than definitions with
lower meaningfulness stimuli. As for the length variable, the mean
pre-test scores for short and long definitions are .19 and .33 reSpec-
tively. The longer definitions seemed to be better known than short
ones.

Analysis of the post-test scores of Table 16 and Appendix R2
showed significant differences between types of definitions. The
mean post-test values for H-H, L-H, HAL and L-L are 1.60, 1.40, 1.23
and 1.39 respectively. It seems that high meaningfulness in the
reSponse had higher attainment values in the post-test than those
with low meaningfulness. This observation is confirmed by finding
a significant difference between the levels of reSponse meaningful-
ness. The mean post-test score for H and L responses, after picture
familiarization and the learning trials, are shown in Appendix R to
have the values 1.49 and 1.32 reSpectively.

Table 16 indicates also that the X2 test showed significant vari-

ation between the two distributions of pre- and post-test scores at the

88

.01 level. This result points to the presence of significant change
in attaining the definitions in the picture familiarization treatment

after receiving the learning trials.

TABLE 17: The X2 Values of Pre- and Post-Tests' Scores for Types of

Definition Under Picture Familiarization

 

 

 

 

 

Types of Definitions Subject Matter Length Pre and
Pre Post Pre Post POSt
H-H .12 10.48** 26.14** .82 16.68**
L-H .00 8.36* .00 1.51 .00
H-L 1.82 6.33* 9.42** 2.41 16.47**
L-L .00 .58 4.91 13.66** 4.84

 

 

 

 

* Significant X2 at the .05 level

**Significant x2 at the .01 level

Table 17 shows the X2 values of pre- and post-tests' scores for
types of definition under picture familiarization. The results reported
in Table 17 indicate that H-H definitions had significant differences
between the post-test scores of numerical and geometrical definitions.

Appendix 8 shows that the mean post-score of numerical and geometrical

2
definitions are 1.51 and 1.68 respectively. Thus, after picture famil-
iarization the attainment of geometrical definitions was greater than

that of numerical definitions. Appendix S shows that the mean pro-test

1
scores of short and long H-H definitions are .37 and .89 respectively.
It indicates that H-H definitions with longer responses were better

known than H-H definitions with shorter reSponses.

89

According to Table 17, post-test scores of L-H definitions showed
significant differences from the levels of the subject matter variable.
The mean post-test scores of L-H numerical and geometrical definitions
after receiving picture familiarization as shown in Appendix T2 are
1.58 and 1.22 respectively. Thus picture familiarization and learning
trials made L-H numerical definition more highly attained than geomet-
rical definitions.

The mean post-test scores of H-L definitions are shown in Appen-
dixU2 to have the values 1.40, 1.06 for numerical and geometrical
items reSpectively. Table 17 shows also that there is a significant
difference between these two distributions. The results indicate that
HeL numerical definitions are attained better than geometrical defini-
tions after picture familiarization and the learning trials. The mean
pro-test scores of H-L short definitions are less than the mean pre-
test of H-L long definitions (.26 and .44 for short and long definitions
respectively). This emphasizes the fact that H-L definitions with
longer reSponses are known better than the corresponding definitions
with shorter reSponses.

Table 17 shows also significant differences between post-test
scores of numerical and geometrical L-L definitions. The mean post-
test scores for L-L numerical and L-L geometrical definitions as
reported in Appendix V2 are 1.63 and 1.15 reSpectively. Thus it
appears that with picture familiarization and the learning trials the
attainment of L—L numerical definitions was higher than geometrical

definitions.

90

The comparison of the pre- and post-test scores showed signifi-
cant differences in their distributions with H-H and H-L, but insig-
nificant differences in case of L-L and L-H types of definitions. The
arrangement of the definitions according to the magnitude of X2 value
of the pre- and postetest scores are H-H, H-L, L-L and L-H. Thus,
following the previous uses of control and verbal familiarizations,
the stimulus position is more responsible for the attainment of the
definitions than is the reSponse position. Thus, the higher the m of

the stimulus, the more readily the definition is attained.

CHAPTER IV

DISCUSSION AND CONCLUSIONS

One objective of this study is to investigate the relationships
between definition m, definiendum m, definien m, length of definien-
dum and the number of words in the definien. The data concerning the
relationship between length of words and their judged meaningfulness
were not conclusive. Meaningfulness of the scaled vocabulary and m
of the geometrical stimuli correlated significantly with their number
of letters, while the numerical stimuli did not correlate significantly.
0n the other hand the correlation between reSponse m and response nump
ber of words was found to be significant. Such significant correla-
tions were in accordance with the conclusion that the decrease of
length was accompanied by an increase of meaningfulness.

The insignificant correlation of numerical stimuli might be due
to limiting the selection of items to a narrow proportion which were
not representative of all possible numerical items. Thus, this result
could be due to a restriction in range among the numerical stimuli.
The other significant correlations may be explained in terms of Zipf's
Principle of Least Effert (1949) and the empirical findings of'Leply
(1950) and Cofer and Shevitz (1952). Zipf noticed that individuals
have a tendency to use short words more often than long ones and sug-
gested that frequency of occurrence of words is inversely related to
their length. Empirical studies have feund significant correlations

between frequency of words and number of associates (meaningfulness).

91

92

Then inductively, the length of words may be inversely related to m,

a notion which is in agreement with the findings of the present study.
Although the preceding logic was based on studies concerned with words,
it seemed that it could be extended to include sentences.

Meaningfulness of definitions revealed significant relationships
with m of the stimuli, m of the response, and inversely with number of
words in the response. Furthermore there was no significant correla-
tion with the number of letters in the stimuli. This indicates that
definition m correlates with m.of both members, the stimulus and the
response, while these members do not correlate with each other signi-
ficantly. Moreover, when m of either one of these members was partialled
out, there was no significant change in the zero order correlations,
but the partiallization of definition m produced a significant change
in the zero order correlation. Thus it is concluded that definition m
has a more critical effect than mhof either the stimulus or the reaponse,
and any partiallization for its effect will cause significant changes
in m of the other two components.

Unfortunately the review of research failed to confirm this con-
clusion fer the following two reasons; first, most of the paired-asso-
ciate studies used nonsense syllables that had no mutual relevance in
meaning between them, in either the stimulus or the response position,
while in the definition the stimulus and response are related through
a common meaning which is presented by either one of them. This lack
of common relationship might have encouraged the researchers to avoid
the study of m of the combined members of paired-associate and its

effect on m of both the stimulus and response. Second, the other

93

paired-associate studies which used meaningful words as adjectives or
nouns avoided using synonymous verbal units in both positions of the
paired-associate in order to control for previous learning. For
example, a list composed of paired-associates such as "car-automobile,"
will enable the learner to use his past knowledge to recall the
response "automobile" when car is presented. Then the processes of
integrating the response, and associating it with the stimulus will
depend heavily on the established experience of the learner which is
assumed to be eliminated or controlled.

A finding which has no counterpart in the literature of meaning-
fulness was the increase of stimulus meaningfulness with the increase
of reSponse meaningfulness, when definition m is controlled. Thus the
higher the m of the stimulus, the higher the m of the reSponse.

Another result was that definition m, or reSponse m, correlated nega-
tively with response length. The increase of either definition m or
response m was accompanied by shorter responses. But, when definition
m was held constant, the correlation of reSponse m with response num_
ber of words dropped to zero. ReSponse length seems to be related
more to definition m than to response m. The partiallization of the
former produces a greater effect than the partiallization of the latter.
This may mean that writers tend to state the unfamiliar and uneasy defi-
nitions, using unfauiliar and difficult reSponses, in terms of a long
sentence. But when familiarity of definition is excluded, low or high
meaningfulness responses may or may not be accompanied by short or

long sentences. Thus it can be concluded that definition m is the most

94

critical variable influencing the relationships between stimulus m,
response m and response length.

The correlation of definition m with summed m of the stimuli and
responses was feund also to be high. Such summation of stimulus and
response m was termed the composite definition. But when definition
m, response m or stimulus m was correlated with the sum of the compo-
nent words of the response, the correlation was practically zero. The
latter finding emphasizes an important point, namely that high m.words
do not insure high or low m responses. In other words the sum of m is
not a reliable predictor of the ease or difficulty of the stimulus,
the response or the definition.

The second aSpect of this study was concerned with the appropriate-
ness of extending the paired-associate theories to definition learning.
Subjects answers during the test trials were analyzed using two differ-
ent criteria. ‘When the correct responses (the answers which were an
exact reproduction of the reSponse or similar to the reaponse but
stated differently) were used as a dependent variable, the results were
inconclusive. The data led to the rejection of the experimental hypothp
esis which stated that the increase of m of the reSponse member of the
definition will be accompanied by an increase of mean percentage of cor-
rect reaponses per test trial. The actual arrangement of the types of
definitions from superior to inferior did not follow the predicted
arrangement: H-H, L-H, H-L, L-L or H-L, L-L, H-H, L-H with control or
familiarization treatments respectively.

The following conclusions were drawn from the analysis of test

trial data using the correct responses as the dependent variable. The

95

theory which emphasizes the role of reSponse m was tested fifteen times.
It has been accepted twice--for the control treatment with the combined
and the long definitions. However, in case of combined definitions
there was some overlap between HAL and.LAL while the theory predicts

no overlap. Thus there remains only one case--long definitions-cut

of fifteen cases which supports the theory.

The results might be in accordance with some empirical studies
which claim that m of the stimulus is of greater importance than m.of
the response. Such claim was supported twice-~in case of the control
and the verbal familiarization of numerical definitions. But in the
case of numerical definitions with verbal familiarization there was
overlapping between the L-H and the L-L lists. The theory which
emphasizes the role of stimulus m was supported one time out of the
fifteen cases.

Thus it is clear that the instances of support for either the
theory of the stimulus or the theory of the response were scarce and
insufficient to merit acceptance of paired-associate theories as a
model for definition.learning when the dependent variable is the per-
centage of correct responses. However, it is obvious that the choice
of the two types of dependent variables employed in this study do not
exhaust all possible methods of scoring correct responses. For example,
some subjects tended to explain the response using either their own
words or the sentences which were presented in the familiarization
prior to the learning trials. Such answers, although they explained
the response correctly, were excluded as correct responses because they

did not meet the exact response criterion. (0n the other hand, the

96

second dependent variable was not limited to the exact responses in

the definition. This criterion included answers which were similar

but stated in a slightly different way than the exact reSponses.) Thus
this dependent variable was not as restrictive as the dependent vari-
ables employed in paired-associate learning studies. For example, the
empirical studies which were cited in the first chapter of this study
assumed that the exact reproduction of the response was the measure of
learning and any answer which deviated from the exact response was con-
sidered incorrect. The definition of the correct reSponse then violated
the condition of the one-to-one correspondence between paired-associate
and definition.learning experiments and might have caused the rejection
of the general hypothesis.

When the dependent variable was limited to include only the §fs
answers which were exactly the same as the learning task responses, the
results were different than in case using the more lenient criterion.
It was noticed that the results did not confirm the theory which favors
the stimulus position in the case of either control, verbal, or picture
familiarization. This theory was given fifteen empirical tests using
exact responses as the dependent variable, but no one of these tests
proved to support the theory.

The theory which emphasizes the role of reSponse m was confirmed
three times in case of combined, short, and long definitions with the
control treatment and was not supported in case of numerical and geo-
metrical definitions with the same treatment. But the coefficient of
concordance for the five tests of the control treatment revealed a

significant value (3 = 81, W = .648, P< .01) favoring the theory of

97

response position. Thus it is concluded that without prior familiari-
zation, and with the exact reSponses as the dependent variable, paired
associate theory can be extended to include definition learning. This
general conclusion is equivalent to previous conclusions reached by
Stoddard (1929), Cason (1933), Sheffield (1946), Kimble and Dufort
(1955), Cieutat, Stockwell and Noble (1958) and many others who were
reported in the first chapter of this study. The supported order of
types of definitions is as follows: H-H, L-H, HAL and.LAL. In the
control treatment it seems that subjects' behavior in definition learn,
ing is similar to their behavior in case of learning paired-associates,
when these subjects are asked to give the verbatim response.

‘When the subjects were given verbal familiarization prior to the
learning trials, the arrangements of the types of definitions were
found to be inconsistent with each other. One of these arrangements
namely the numerical definitions, was in agreement with the theory
which emphasizes the response position. The other arrangements also
differed. The coefficient of concordance (S = 57, W = ~456, P;>*.05)
for the five tests (combined, short, long, numerical and geometrical
definition) was not significant. The summed rank order of the types
of definitions suggested this order: H-H, HAL, L-L and.L-H.

Similarly, when the subjects received picture familiarization
before the learning trials, the arrangements of the types of defini-
tions were not in agreement with each other. Their coefficient of
concordance (s = 69, w = .55. P> .05) was found to be insignificant.
Yet the summed rank order of the types of definitions yielded this
order: H-H, Lo-H, H-L, and L-L.

98

The data concerning the familiarization conditions led to the
rejection of the hypothesis which stated that verbal or picture famil-
iarization of the re5ponse member would be accompanied by either an
increase or decrease of the mean percentage of the exact response per
test trial. The increase would occur if the familiarized response
were low m meaningfulness, while the decrease would occur if the
familiarized response were high m meaningfulness, and the expected
arrangement of the types of definitions according to the response
theory would be HAL, LAL. H-H, and.L-H.

The differences in the expected and actual arrangements might
be due to different methodological considerations. In.paired-associate
studies, the nonsense syllables of either the stimulus and the response
are not related in meaning, but it has been shown that definition come
ponents are highly correlated with definition m. Thus while the experi-
mental approach was directed towards the response member, it seems that
definition m influenced and affected the stimulus m.by the familiariza-
tion procedure. The familiarization.process could not be limited to
the response member only. Apparently it influenced the stimulus member
as well either by presenting the stimulus at the time of the familiari-
zation or by the nature of the relationship between definition m,
definiendum.m, and definien m.

But if stimulus m and reSponse m were both affected--according
to this discussion--then a question arises as to why the revealed
arrangements had low concordance. Probably the resPonses were not
equally familiarized prior to the learning process. Some paired-asso-

ciate studies have reported employing familiarization tests before the

99

learning trials to ensure that familiarization Was equally effective.
For example, Epstein, Rock and Zuckerman (1960) had their subjects
learn twelve short lists, each of which consisted of a nonsense syl-
lable and four members. The syllables were later paired to form
lists of pairs of familiarized items. Battig, Williams, andeilliams
(1962) used nonsyllables for a verbal discrimination task. Both MBMP
bers of some pairs became the stimulus or reSponse members of pairs
for paired-associate learning. Members of other pairs were separated
and paired with syllables which subjects had not seen previously.
Another procedure to equalize the familiarization was reported by
Schulz and Martin (196h). Their subjects spelled the stimuli, then
the stimuli were recalled after every trial. A similar step to
equalize the effect of familiarization.prior to the learning process
was not considered in this study because it is not an ordinary pro-
cedure in the actual instructional situation.

A second interpretation for the absence of concordance between
the arrangements of types of definitions may be due to the effect of
proactive inhibition. It has been noticed that the subjects' answers
to the test trials were influenced by presenting two kinds of material;
the familiarization material and the learning material. ‘When the sub-
jects were asked to recall the learning material, they in fact at
times answered by reproducing the familiarization material.

This points to the presence of mutual interference between the
two sets of reSponses to the extent that the responses from the two

sets were competing with one another at recall. The overt intrusions

100

of responses from the competing set of responses actually displaced
the exact responses.

Moreover, an investigation of the experimental situation, showed
that in the familiarization situation, learning trials, and test trials,
the stimulus was presented without any change. It is designated as $1.
The explanation of responses which were presented for familiarization
is designated as R1 while the learning material is designated as 8.2.
Thus the presence of constant stimuli (31) and two different response
sets (R1 and R2) can be represented tentatively as follows:

S1—-> R1 51—? R2 S1—+32

Familiarization Learning Trials Test Trial

The observation of the presence of interference between R1 and
R2 is supported by Deese (1958) who reported that a safe generaliza-
tion about paired-associate learning states that when the same stimlus
items and different response items are used in two tasks, there is
negative transfer. Most important is his conclusions that holding
stimulus similarity between the tasks (familiarization and learning
trials) constant, transfer can be varied from positive to negative by
changing the responses in the two tasks (familiarization and learning
trials) from being identical or very similar to being very different
from each other.

In this study it can be assumed that 31 i.e. the familiarization
of the response was kept relatively easy and familiar. However the R2
responses were either H or L depending on the type of definition
learned. For example, R2 was either H or L as a result of having one
of these types of definitions, H-H, L-H, H—L, or L-L. Then the number

101

of intrusions and their positive or negative effect on the recall of
R2 would be different, as a result of learning different types of defi-
nitions.

Besides, each reSponse set R1 or R2 was a combination of two
factors--length (short and long) and subject matter (numerical and
geometrical). The analysis of the test trials using the exact responses
has been done for each level of these two factors plus their combina-
tion. The short definitions or numerical definitions have been anaLyzed
separately for each type of definition. But it is worth noticing that
the kind of similarity between the short responses is completely differ-
ent from the similarity between numerical responses, because the former
similarity was based on length, while the latter similarity resulted
from being related to the same subject matter. The elements which were
responsible for the similarity of length (number of words in the
response) should have been different than elements which were reSpon-
sible for the similarity of subject matter (say; reSponses that deal
with numbers) and thus each kind of element would have exerted differ-
ent effects on the test trials. By the same token, the effect of
similarity between short responses, could be different than the effect
of similarity between long definitions, since the increase of length
might have differential effect on the material and on the learner.
Similarly, the subjects might have adopted a strategy for the similar-
ity of numerical definitions which was different from their strategy
in approaching similar geometrical definitions.

In conclusion, it is suggested that the interaction between types

of definitions and the types of similarities located among the learning

102

task R2 (combined, short, long, numerical and geometrical) is responp
sible for the insignificant concordance between the arrangement of
bmpes of definitions when subjects received either verbal or picture
familiarization. Apparently the previous interaction was working in
the same manner with verbal or picture familiarization. When the
arrangements of the types of definitions, in both verbal and picture
familiarization, were checked for concordance, it was found to be
significant (3 = 266,'W = .532) at the .01 level and in favor of this
order: H-H; LAL. HAL, and.L-H.(LAL although higher was similar to
HAL). Such interaction between types of definitions and familiariza-
tion was absent in the control treatment because of the absence of
the familiarization task R1 and the arrangement of the types of defi-
nition (H-H, L-H, HAL, L-L) was completely different.

The prediction of’Lambert and Jakobvits (1960), Kanung, Lambert
and Mauer (1962), and Kanungo and.Lambert (1963) that familiarization
of high meaningful material would have a prohibitive effect due to
either meaning decrement or the development of a word-word habit,
seemed to be untenable as applied to this experiment. According to
this theory the H-H type of definition must be ranked in the third
position as a result of the satiation effect; but it was found to be
in the first rank with respect to the other types of definitions.
Similarly the HAL type of definitions was predicted to be ranked the
first, while it was feund to be located in the third position. The
rank order between the predicted arrangement of the types of defini-
tions, and the arrangement that resulted from summing the ranks in

verbal and picture familiarization was not significant. Probably

103

because this theory was based on the assumption that the familiariza-
tion procedure would mean frequent repetition of R2, which was not the
case in this study.

The third objective of this study is to investigate the differ-
ences between pre- and post-test scores due to the familiarization and
learning trials. The pre- and post-test scores were analyzed on the
basis that the subjects' answer would be assumed correct if he explained
the meaning of the definiendum correctly without regard to the exact
terminology used during the learning trials. The pre-test data have
shown that subjects' previous knowledge was significantly different in
case of the types of definitions, stimulus meaningfulness and.length.
The subjects' mastery of the definitions seemed to be in this order:
H-H, HAL, LAL and.L-H. Definitions with higher meaningfulness stimuli
were better known in advance than the definitions with.lower meaning-
fulness stimuli.

Analysis of post-test scores showed the presence of controversial
finding between m of the stimulus and m.of the response. It indicated
that post learning was more effective in this order: H-H, HAL, LAL,
and.L-H fer control treatment. The arrangement tends to indicate that
m of the stimulus was more critical in the post learning than reSponse
m. Such apparent critical effect of stimulus m was statistically sig-
nificant only in case of the control treatment even though reSponse m
was also a source of statistically significant variance in case of con-
trol and picture familiarization. The verbal familiarization procedure

supported neither stimulus m nor response m.

10h

It has been mentioned that stimulus m had a greater effect on
the pre-test scores. 0n the other hand, the statistical analysis of
post-test scores showed that neither the m of the stimulus nor the m
of the reSponse had a consistant effect over all the familiarization
treatments. Probably the familiarization treatment and the learning
trials had affected response m.more than stimulus m. As a result of
the familiarization and learning response m.gained a differential
effect over stimulus m and there was no consistant effect in case of
the three familiarization treatments.

The comparison between.pre-test and post-test scores revealed
that there was a significant gain in definition attainment. Such
observation was supported by the results of control verbal, and pic-
ture familiarization treatments. The arrangement of types of defini-
tions according to the magnitude of change between.pre- and post-test
scores were H-H, H-L, LAL and.L-H. The consistency of this arrange-
ment still suggests that stimulus m.is more critical in the gain of
definition attainment. Individual comparisons between.pre- and post-
test scores for each type of definition under familiarization treat-
ment confirmed the previous conclusion.

Concerning the length of reSponses, the results showed consis-
tently in the three familiarization treatments that previous attainp
ment was better with longer definitions than short ones. The post-test
analysis showed post-test scores of shorter definitions to be better
attained than longer ones especially in case of verbal familiarization.
The individual analysis of types of definitions with the length vari-

able showed a trend towards better pre- or post-attainment for longer

105

definitions when stimulus m was high and for shorter definitions when
stimulus m was low. However, the subject matter showed inconsistenqy
and many insignificant results.

The pre- and post-test scores led to the following decisions
concerning definition attainment hypotheses. First, that the increase
of stimulus m will be accompanied by an increase of pre-test scores is
a tenable hypothesis and that the definition of high m stimuli are
better known than definitions of low stimuli. Second, the hypothesis
which stated that an increase of definition attainment will fdllow also
the m of the stimuli is supported. The higher the m of stimuli the
higher the difference between pre- and post-test scores, while lower
m stimuli are associated with.lower differences between.pre- and post-
test scores. Third, the data did not support the hypothesis that the
increase of stimulus m.is accompanied by an increase in the post-test
scores.

In summary, the general hypothesis that paired-associate learn.
ing theory favors the m of reSponse was unsupported when the dependent
variable was the mean.percentage of correct re5ponses. only when the
mean percentage of exact responses was chosen as a dependent variable
for the control treatment, was the general hypothesis supported. This
supports the paired-associate theory which stresses reSponse m.in defi-
nition learning. However, the paired-associate predictions concerning
the presence of satiation as a result of familiarization were not con»
firmed. A possible interpretation of the absence of satiation in meanp
ing may be due to the type of familiarization.procedure employed.

Unlike the procedures used in paired-associate studies, the present

106

procedure avoided repetitions of the responses. Instead, it explained
the response only once, using familiar expressions that were somewhat
different from those responses learned by the §s during the learning
tasks.

This fact emphasizes the difficulty in generalizing from the
paired-associate model to definition learning. The molecular nature
of such a model and its corresponding operational definitions for such
concepts as familiarization, stimulus, response, meaningfulness, and
even learning have no exact analog in definition learning. Until ver-
bal learning model become broader in scope, generalization from these

molecular models to more complex learning situations should be made

with caution.

BIBLIOGRAPHY

BIBLIOGRAPHY

Archer, E. J. A Re-evaluation of the Meaningfulness of all Possible
CVC Triagrams. Psychol. Monogr., 1960, 23, Whole No. 1597.

Arnoult, M. D. Familiarity and Recognition of Nonsense Shapes.
J. Egg. Psychol” 1956, 5;, 269-276.

Bailey, J. H., Jeffrey, W. E. Response Strength and Association
Value in Stimulus Predifferentation. Psychol. Rep” 1958, it,
715-721.

Battig, W. F., Williams, J. M., Williams, J. G. Transfer from
Verbal Discrimination to Paired-Associate Learning. J. E.
Psychol., 1962, 6 , 258-268.

Braun, H. N., Heymann, S. P. Meaningfulness of Material, Distribu-
tion of Practice, and Serial-position Curves. J. m. Psychol.,
19%. fig 1%‘1500

Brodbeck, M. Logic and Scientific Method in Research on Teaching.
In N. L. Gage (Ed.), Handbook of Research 93 Teaching. Rand
McNale & Compamr, Chicago: 1963.

Cason, H. Association Between the Familiar and Unfamiliar. J. m.
PSYChOIe. 1933' 16. 295-305.

Cieutat, V. J. Differential Familiarity with Stimulus and Response

i161 Paired-Associate Learning. fiercep . _M_q_. Skills, 1960, i1,
2 9-275.

Cieutat, V. 3., Stockwell, F. E., Noble, C. E. The Interaction of
Ability and Amount of Practice with Stimulus and Response Mean-
ingfulness (m, mi) in Paired-Associate Learning. J. M: Psycho .,
1958. 5Q. 193-202.

Cofer, C. N., Shevitz, R. Word Association as a Function of Word-
Frequency. Amer. J. Psychol., 1952, 6 , 75-79.

Davis, F. C. The Relative Reliability of Words and Nonsense Syllables.
A. M. Pszche. 1930. 1 9221-23140

Deese, J. The Psylcholog 9; Learning. New York: McGraw-Hill, 1958.
Deese, J. From the Isolated Verbal Unit to Connected Discourse. In

C. N. Cofer (ed.) Verbal Legrnigg and Verbal Behavior. New York:
McGraw-Hill, 1961.

107

108

Dowling, R. M., Braun, H. W. Retention and Meaningfulness of Material.
1- 23.3% 2.832221” 1957. fl. 213-217.

Epstein, W. The Effect of Stimulus and ReSponse Meaningfulness When

Reaponse Availability is Enuated. J. Verb. Learn. Verb. Behav.,
1963, .2_, 242-249.

Epstein, W., Rock, J ., Zuckerman, C. B. Meaning and Familiarity in
Verbal Learning. Psychol. Monogr., 1960, 113;, No. 491.

Gannon, D. R., Noble, C. F. Familiarization (m) as a Stimulus Factor

in Paired-Associate Verbal Learning. J. m. Psych” 1961, 6_2,,
lit-23.

Gibson, E. J. A Systematic Application of the Concepts of Generaliza-

tion and Differentation to Verbal Learning. Psychol. £91., 1940.
31. 196-229.

Glaze, J. A. The Association Value of Non-Sense Syllables. J. Genet.
Psychol., 1928, 15, 255-269.

Goss, A. E., Nadine, C. F. Egg-Associate Legrnigg. New York:
Academic Press, 1965.

Goss, A. E., Nadine, C. F., gt. g. Stimlus Characteristics and Per-
centage of Occurence of Response Members in Paired-Associates
Learning. Psychol. Monogr., 1962, 26, Whole No. 531.

Haagen, C. H. Synonymity, Vividness, Familiarity and Association
Value Ratings of #00 Pairs of Common Adjectives. J. Psychol.,
191490 £2: “'53-463.

Harleston, B. W. Task Difficulty, Anxiety Level, and Ability Level
as Factors Affecting Performance in Verbal Learning Situation.
J. Psychol., 1963, 55, 165-168.

Hilgard, E. R. Methods and Procedures in the Study of Learning. In

S. S. Stevens (Ed.), Handbook _o_f_ merimentg Psychology. New
York: Wiley, 1951.

Hilgard, E. R. A Perspective on the Relationship Between Learning
Theory and Educational Practices. In E. R. Hilgard (Ed.),
Theories 9; Learning gn__d_ Instruction. The Sisty-Third Year Book
of the National Society for the Study of Education. Chicago:
The University of Chicago Press, 1961+.

Hovland, C. I., Kurtz, K. H. Experimental Studies in Rate-Learning
Theory: X. Pre-Learning Syllable Familiarization and the

Length-Difficulty Relationship. 1. E. 231932;... 1952. 39..
31'39e

109

Hull, C. L., The Meaningfulness of 320 Selected Nonsense Syllables.
Amer. J. Psychol” 1933, 36, 730-734.

Hunt, R. G. Meaningfulness and Articulation of Stimulus and ReSponse
in Paired-Associate Learning and Stimulus Recall. J. Egg. Psychol.,
1959’ :2. 262-2670

Kanungo, R., Lambert, W. E. Paired-Associate Learning as a Function
of Stinmlus and Response Satiation. Brit. J. Psychol” 1963,
2... 135-144. ""

Kanungo, R. N., Lambert, W. E., Mauer, S. M. Semantic Satiation and
Paired-Associate Learning. J. E52- Psychol., 1962, 6%, 600-607.

Kimble, G. A., Dufort, R. H. Meaningfulness and Isolation as Factors
in Verbal Learning. J. Ego. Psychol., 1955. 52. 361-368.

Kothurkar, V. K. Effect of Stimulus-Response Meaningfulness on Paired
Associate Learning and Retention. J. m. Psychol., 1963, 66,
305-308.

Krueger, W. C. F. The Relative Difficulty of Nonsense Syllables.
J. E356. Psycho]... 19314, 1 , 145-153.

L'Abate, L. Manifest Anxiety and the Learning of Syllables with Dif-
ferent Associative Values. Amer. J. Psychol., 1959. ,2, 107-110.

Lambert, W. E., Jakobvits, L. A. Verbal Satiation and Changes in the
Intensity of Meaning. J. m. Psychol” 1960, 66, 376-383.

Lepley, W. M. An Hypothesis Concerning the Generation and Use of
SynorWHB. is E. PSEChOle. 1950' 3:9. 527-530e

Lyon, D. 0. The Relation of Length of Material to Time Taken for
Learning and the Optimum Distribution of Time. J. Ed. Psycho .,
19114, 6, 155-163.

Mandler, G. Response Factors in Human Learning. Psychol. 3.31., 1954.
£51. 235-2Lm.

Mandler, G. Associative Frequency and Associative Prepotency as Mea-
sures of Response to Nonsense Syllables. Amer. J. Psychol"
1956, 68., 662-665. "—

Mandler, C., Huttenlocher, H. The Relationship Between Associative
Frequency of Stimulus and Responsein Paired-Associate Learn-
ing. Amer. J. Psychol” 1956, 62, 1420.428.

Martin, C. J. 2.11.1.9. Role 2.1: Repetition i5 $.92. Agguisition 2: m
Associations. Unpublished Ph.D. Dissertation, Wayne State
University, 1963.

110

Martin, C. J., Boersma, F. J., Cox, D. L. A Classification of Associa-
tive Strategies in Paired-Associate Learning. Psychon. .S_cJ.,
1965. 1. 455-1156.

McGeoch, J. A. The Influence of Associative Value Upon the Difficulty
of Non-Sense-Syllable Lists. J. Genet. Psychol., 1930, 31,
1421-1126.

Miller, G. A. LLnggage §_n_d. Communication. New York: McGraw-Hill,
1951.

Miller, G. A., Selfridge, J. A. Verbal Context and the Recall of
Meaningful Material. Amer. J. Psychol., 1950, 6 , 176-185.

Noble, C. E. An Analysis of Meaning. Psychol. £31., 1952, 62, (+21-
LL30.

Noble, C. E. The Familiarity-Frequency Relationship. J. fig. W0.
195’“, “'1. 13-160

Noble, C. E. The Effect of Familiarization Upon Serial Verbal Learn-
inge is Me P52Ch01e, 1955, hi, 333-3380

Noble, C. E., McNedy, D. A. The Role of Meaningfulness (m) in Paired
Associate Learning. J. _E_2_c6. Psychol., 1957, 61, 16-23.

Noble, C. E., Stockwell, F. E., Pryer, M. W. Meaningfulness (m) and
Association Value in Paired-Associate Syllable Learning. Psychol.
322.. 1957. 2. 1441-452.

Nodine, C. F. Stimulus Durations and Stimulus Characteristics in
Paired-Associates Learning. J. E56. Psycho ., 1963, 66, 100-106.

Postman, L. P., Phillips, L. W. The Effects of Variable Contexts on
the Acquisition and Retention of Paired-Associates. Amer. J.
Psychol., 1964, 21, 64-74.

Reed, H. B. Repetition and Association in Learning. Pedagogicg
Seminagz, 1924. 2;. 147-155-

Riley, D. A., Phillips, L. W. The Effects of Syllable Familiarization
on Rate Learning, Association Value, and Reminiscence. J. m.
P239221” 1959. .12. 372-379.

Rothkopf, E. L. Some Conjectures About Inspection Behavior in Learn-
ing From Written Sentences and the Response Mode Problem in
Programmed Self-Instruction. J. Pgmd. Instr., 1963, g. 31-40.

Sarason, I. G. The Effect of Associative Value and Differential Moti-

vating Instructions on Serial Learning. Amer. J. M” 1957.
22, 620-623.

111

Schulz, R. W., Martin, E. Aural Paired-Associate Learning: Stirmlus
Familiarization, ReSponse Familiarization, and Pronunciability.
J. 166. Learn. Verb. Behav., 1964, 2, 139-145.

Sisson, E. D. Retroactive Inhibition: The Influence of Degree of
Associative Value of Original and Interpolated Lists. J. m.
LLS Choo 1938. .229 577-6800

Solomon, R. L., Postman, L. Frequency of Usage as a Determinant of
Recognition Thresholds for Words. J.. m. Psych” 1952, 1+ ,
195—201. '

Stoddard, G. D. An Experiment in Verbal Learning. J. Ed. M"
1929. .2_0. [452-457.

Thurstone, L. L., Chave, E. J. 22.9. Measurement 9; Attitude. Chicago:
Univ. Chicago Press, 1929.

Underwood, B. J..:, Merimental Psychology. New York: Appleton-Century-
Crofts, 1913').

Underwood, B. J., Richardson, J. Some Verbal Materials for the Study
of Concept Formation. Psychol. Bull., 1956, j}, Sit-95.

Underwood, B. J., Schulz, R. W. Meaniggfulness mVerbal Learggg.
Chicago: L. B. Lippincott Co., 19 0.

Underwood, B. J. The Re resentativeness of Rate Verbal Learning. In

A. w. Melton, (Ed. . Categories 9;; Human Learnigg (196a) New
York Academic Press.

Weiss, R. L. m Rolg g Association vgue fl Egerimentgy Produced
Familigrity lg Paired-Associate Learnigg. Unpublished Ph.D.

Dissertation, University of Buffalo, 1958.

Witmer, L. R. The Association Value of Three-Place Consonant Syllables.
J. Genet. Psychol” 1935, 21, 337-360.

Zipf, G. K. Human Behavior and H19. Principle 2.1; Least M. Cambridge,
Mass: Addison-Wesley, 1911-9.

APPENDIX A

Arithmetic Textbooks Series Used in Gathering the Definitions

 

1.

Buswell, Guy T., Brownell, William A., and Sauble, Irene, Arith-
metig W3 Need. Ginn and Compam', Boston, 1959.

Deans, Edwina, Kane, Robert B.,1McMeen, George H., and Oesterle,
Robert A., Th6 Modern Mathematics Serigs. American Book Company,
1963.

Morton, Robert Lee, Gray, Merle, Springstun, Elizabith, Schaff,

William L., and Rosskopf, Myron F., wing Sgre g; grithmetic.
Silver Burdett Company, Morristown, New Jersey, 19 .

Osborn, Jesse, Riefling, Adeline, and Spitzer, Herbert F.,

lori Arithmetic. Webster Publishing Company, St. Louis,
19 2.

112

APPENDIX B

 

Name_ Grade
Birth Date Sex

When we read or hear a new word for the first time, we might say
that "this word looks familiar and easy for me to learn and to memo-
rize," or "this word looks unfamiliar and difficult for me to leirn
and to memorize." So there are different kinds of words. The werd
might be: Ve as , my, W, m, vgm diffiflt to
learn and to memorize. Not all the words are the same according to
their easiness.

In the following pages, there are lists of arithmetic words, and
five columns. You are to tell whether the word is familiar and easy
or unfamiliar and difficult for you to learn and to memorize.

Each column is prepared to indicate a certain degree of difficulty
in learning and memorizing. There is a column to check the very easy,
easy, indifferent, difficult, or the very difficult word.

Read all the words first, then read each word carefully and decide
the easiness of each one. Check with the mark (X) the proper column
following the word.

Do not rush, but do not slow down. Be sure to read each word

carefully, giving it just one mark. Example:

 

Very
Very Indif- Diffi- Diffi-

Easy E‘s! ferent gult gt

 

1 . Rectangle

 

2. Cotangent

 

 

 

 

3 . Minuend

 

 

 

 

113

APPENDIX C

The Instructions for Rating the Meaningfulness

 

In arithmetic classes and textbooks, we used to hear and read
certain words (sentences). These words (sentences) might look easy
for you to learn and to memorize, or they might be hard for you to
learn or to memorize. Also, these words (sentences) might be famil-
iar to you or unfamiliar. The degree of familiarity or ease to
learn each word (sentence) differs from one student to another.

The booklet given to you contains a number of such words (sen-
tences) that are used in arithmetic classes. You are going to read
them carefully, and always ask yourself this question, "Is this word
(sentence) familiar and easy for me to learn?" and "How far is this
word familiar and easy?" So this first reading is to get acquainted
with the words presented in the booklet and to think about your own
judgement for each word (sentence). 1

There will be a second reading. In this reading you will show
your own judgement for the familiarity and ease of the word (sentence)
by marking the appropriate column directly in front of each word
(sentence). 0n the first page there are some examples, also there
are five columns. The first column is where to mark if the word
(sentence) is very easy, the second is for easy, the third is for
the indifferent words (sentences), the fourth is for the difficult,
and the fifth 1. for the very difficult words (sentences).

(The experimenter used the board to show the five columns).

Suppose I found that the first word (sentence) is difficult, where

11h

115

may I put the mark (X)? (The students were encouraged to answer as
a groufi]. Suppose I found that the word (sentence) is easy, where

may I put the mark (X)? [The students were encouraged to answer as
a groufi]. Sometimes the word (sentence) might look for me neither

easy nor difficult. This word is called indifferent. ‘Where may I

check for indifferent? [The students answered). Any question?

Now, to sum up, you are going to indicate for each word (sen-
tence) how far it looks familiar and easy for you to learn. Second,
you shall read the words carefully but do not mark any of them, just
decide for>yeurself. Third, you shall read each word and.you will
show your own idea about the ease and familiarity of the word (sen-
tence) by writing the mark (K) in the proper column.

‘We are going to work as a group. ‘When.yeu finish the first
reading just raise your head and wait until all of you are asked to
begin the second reading. Remember do not rush or slow down and
read each word carefully.

Any question?

Begin.

 

mnm.o
000.H
ewe.0
one.o
smm.o
amm.a
enn.o
moo.H

wn~.H
mmN.o

000.0

emo.0

003.0
030.0
esm.H
«50.0

 

mom.e
afio.m
mos.“
Nem.H
Nod.“
can.~
mac.“
meo.a

can.a
«mo.fi

wwN.«

a 328.233

 

~mm.o an~.H
ooo.“ mua.a
mom.“ cum.m
Ann.” «mm.m
mm~.a mmo.~
own.“ nmm.~
~nn.o am~.A
~m~.H ema.m
ems.“ Aa~.~
«m~.H msm.a
soc.” see.fi
moo.fi mma.a
5mm.“ aee.m
ans.o moa.a
500.“ omn.fi
0:0.0 ewe.“
m
a soenaeon

oom.o cow." z one us. on.» caches» one .nuamnm A we
mow.0 000.“ 2 once» vested

en. snonman noozson oonon.eeae one .omnqm mm as
030.0 own.“ 2 venues: knees ca ease sebam 4 .aemeaseeaem mm :H
new.o con.a o enemas cansaaoom . no one. A 1....m mm m“
:um.0 00:.“ z codedbae

no soaueaemo assess“ one «asoaaeo«anaadax 0« NH
mea.o was.s z canonssen. eoanaunonou ”annomae mm HA
nmn.0 0::.« 2 sense: someone Beam more.aen

Isa: e wswxep Mo awesome one nameduoeupnsm N 0“
500.0 03:.“ z . sodpfieee

no scaveuemo amassed one “encepoeepesm mm m
mem.o nae." u easunam . mo use. A ...o.a mm m
035.0 003.“ z , seduces»
upon me soapeaemo oeae>sa one «esodpdee< m: m
eem.o cos.a u , noflos on» noozoon
heslmadn caches huesamesd.:¢ unopsswm mm 0

Hao.0 NmN.H z seaweeds

eepeedoa we compasses one “escapeodanwoasz 00 m
Hms.o ONH.H a comaaon seen. eons» A toamnoane as e
amm.0 000.“ z oneness mom eases one “madness: 0 n
«am.o owe.“ o none. as. see: enemas A antennae no N
wmuqo osoqfi 0 assuage seam < aeseam mm a

am see:
.02

a sodvdsdmen oaks codpficdmea SepH xsem

 

mucosodsoo seagdsamoo as» me aesad> emosHaHmsdceez one

Q NHszmm<

116

 

 

500.0 0000.0 0000.0 050.0 500.0 0000.0 2 000000900
e .00 00030200000 05 00090055 00.0.0. 33.3.0. 0 00
000.0 000.0 000.0 000.0 500.0 000.0 2 names:
pompous 0003900 00000. .3055 4 ”eoneeoeveum 50 00
0000.0 000.0 000.0 000.0 000.0 000.0 2 00203 25 mo 0.0.02.0 0300 on» He sees
no one 00:39:00.0 03.5 Houses: 4 0:030:00 00 00
000.0 000.0 000.0 000.0 000.0 000.0 2 000.0 0ao0tuasz .asan 00 00
000.0 000.0 000.0 000.0 000.0 000.0 2 0.00. on.
_ . . 00000.0 030 0000. 02083.00» 0. 3300209000 005 00
000.0 000.0 000.0 000.0 000.0 000.0 2 0.00.0.0 0.00000000 00.00:: .00 untosoae 0s 00
005.0 0000.0 550.0 000.0 005.0 005.0 0 , -. 0:000:00 seem
less on» 30:: 0.00000» .00 genes: 20.0. 30.04 am 50
000.0 000.0 000.0 000.0 000.0 000.0 0 noonmne n0 eonsnaoa 0.0.00. .00
.00 0.3060 .3 03.00: 35.0000 20.0. 300000000 05 00
000.0 000.0 000.0 00N.0 000.0 000.0 2 a.» stona0 :4 ”00:0 00 00
000.0 000.0 000.0 00 .0 000.0 000.0 2 0023000 0.0.0.000
messes: o3» so 03000.0.er :4 00030000004 05 0
000.0 000.0 500. 0 0000.0 000.0 000.0 0 .8008 a .00 0.0000 000000055 0:0 33200050 00 0
000.0 000.0 000. 0 000.0 000.0 000.0 2 - - - 0000.000000
Idea .00 08000.0er 00.0250 20.0. 000800035 05 00
000.0 000.0 000.0 000.0 000.0 000.0 2 3-, ‘ .-; 0.00:0:
5.025 :0 03.00.303.000 220.00 .00 “somevseonom +00 00
0000.0 000 0 000.0 000.0 00.0 000.0 0 000 e. .00 gasses < 000006000 00 00
000.0 000.0 000.0 000.0 000.0 000.0 0 -000000 mo 0.. 4 .on00 00 00
000.0 000.0 000.0 000.0 000.0 000.0 2 :00 0.00. .0:
. . . 90.080200 pen» eosepseu genes: 4 2.8002000 00 00
03.0 000.0 08.0 0500.0 0000.0 000.0 2 «0:00.000: 3-000.000-0380: one 000000004 00 50
am 532 on mass 00 see:
A -. .02
a 56:35.03 3 0.8000003 8 0.0300238 090.0 00300000000 83H 00:00

 

117

 

 

000.0 000.: 000.0 000.0 000.0 000.0 2 0000:00000
:0 000: 000000 00 0000:: one ”00:00: 00 00

000.0 000.: 000.0 000.0 000.0 000.0 0 00000000
0000: a 00 0000 0000000 000 ".0000000000 00 00

000.0 000.0 000.0 000.0 000.0 000.0 0 . 00200 000 no 000000
030 0:0 0000::00 00:0 0:00 0:9 «0:000 00 00

000.0 000.0 000.0 000.0 000.0 000.0 0 00:0:
-000 0°00“0 upoonno :0 0o0poo00oo 4 "000 00 00

000.0 000.0 00m.0 000.0 000.0 000.0 2 00:00:00 000:0: 0 :0 00:05:: 0 :00
00000 0 0000: 00:» 000500 0 "00000500000 00 a:
000.0 000.0 000.0 000.0 000.0 000.0 2 0:000 :02:00 no 0:0:0000 000: one 00002 0: 0:
000.0 000.0 000.0 000.0 000.0 000.0 2 :00000000 00 000: 000 ”0:000000 00 00

000.0 000.0 000.0 000.0 000.0 000.0 z 000 000:0
000 00 00000000000 000 00 000200 0 "0:00 00 00

000.0 000.0 000.0 000.0 000.0 000.0 0 0000000
0 an 0000Ho:0 0:0Hm 0 mo :00000 0 000000 :0 0:

000.0 000.0 000.0 000.0 000.0 000.0 2 0:0
m0 00:0o:a 000:3 mpogesn 039 000000000000 0 00

000.0 000.: 000.0 000.0 000.0 000.0 0 000000
0:00 000 0>0£ 00:0 moHohflo 036 ”0000:0OQ00 0 0m

000.0 000.0 000.0 000.0 000.0 000.0 2 000: was
00000 0:0 00:00 no 009:5: 0:9 "00:050000 0: 00

000.0 000.0 000.0 000.0 000.0 000.0 2 0000

00 000:00E0p 0:0 0:0 .003::: 00:00:0 03
0000>00 00 :00:0: 000000: a 0003. "000000>00 00 00
000.0 000.0 000.0 000.0 000.0 000.0 0 0::000: 000:0 mo 00:: 0:00:000 4 0000000 00 mm
000.0 000.0 000.0 000.0 000.0 000.0 0 0000: 0 mo 0:00 000:000000 0:0 0:000:00002 mm :0

00 0002 mm 0002 00 2002
.oz

: 20000000000 : :0000000 5 :000000000 0000 :o0p0c0000 :000 0:00

 

118

 

93.“

Nmo.H
N©:.H

Nwm.fi
Nmm.«

moo.a

How.o
wmfi.a
Now.o
mmo.a
omm.o
umm.o

302.0

 
 
 
  

Om

NoH.m

mum.a
wmm.m

mma.m
mm:.~

0:5.H

mum.fi
wwm.m
mN©.H
NH5.H
Hum.:
Hmo.:
mnfi.fi

Cams

5 guanoficfigoo

mmm.H

:om.fi
mmm.fi

mum.fi
fiNN.H

zom.fi

mum.fi
00H.“
mam.“
Nam.fi
com.“
mmfi.fi
:HN.fi

m

amm.m
mmm.m
Hmo.m
umu.m
omH.N
mmo.m
:wm.fi

a goflcfimoa

omfi.a

oom.o
:mw.c

5mm.fi
mnw.o

mmm.o

mom.H
AHN.H
o¢m.o
Hmm.H
:NN.H
Hmo.H
mmm.fi

om

 

omm.N ovumham a no macaw
mnp mo mansmaoa can no Eda one "noaaawumm wfi Hm

own .N mopdzum vacuum
ucoo mud macaw wmonz_amaum unmah d «case mm ow

o®¢.m pmmzoa o» amonman Baum mohoom Mo
pom wowcwnhw cw mo ouoom oavvfls one unawvmz mm mm
03:.N chAwo vopwwcoam ad ”ommaHHm ow mm

0:3.N candy hocoawoum a
cfi comcahua mgonesn mo pom « «soapsnwupmfia mm mm

00:.N vcposhpmcoo ma aopmhm
:owpwhosdn w :own3.coms genes: d "*omwm :fi mm

Nmm.m mpom adds Ho mucosoam map macaw mmwsm
IcowpwHoh can manpowm op wCflzwuv < "Sumac mm mm

omm.m Awnsd: Hwthwc hum
mzoaaom wasp panes: Hmhsum: 4 "hommmoofim m :m

omN.m ufiaom on» an
vomOHoco cowmou wan mo opsmwos may "mssao> 0: mm

omN.N mocnmov ca vohdmmms cmwvwnms oswum 0:»
mo pmms no pmmm oocwpmfiv one "cvapwpwcoq mm mm
o:~.m o>Hm3p com: woman Eopmhm < "Haewowvosm ma an

o#m.m Mocha pfio
Inpws oshp ma woummoow pcosopwpm 4 "80flN4 mm on

oom.m gonad:

nonpocw ca Umcwmpcoo ma hangar a «mean
side 30: waaucaw mo mmooopm one ”icowmw>wa mm a:

Cdoz
0 oz
8 soapficfimoa came :ofipficfigoa sopH xsam

 

119

 

mom;
is;
mmo.H
cam.fi
mum.fi

Nmm.o
mmw.o
one.“
mom.“
Hm:.a

wem.e
efim.a
w::.H
won.e

 
 
 
 
 
 

Om

:Hm.«
mwfi.d
mmm.2
owm.m
Nmm.m

mmm.:
mmmé
mmm.:
muo.m
0:5.N

me.N
Hmw.m
mmH.N
new.a

0:

e 2:23:88

mmN.H
wnm.fi
mmm.fi
moo.“
mmm.H

3m;
mam.fi
mum.“
mam.d
m3;

mmN.H
mHN.H
HHN.H

om:.H

am

wS.~
mom.m
SQN
moo.m
0:5.m

mum.m
mmfi.m
Nam.~
ooo.m

wmm.m

$3
woo.m
83
$3

coo:

a cofinwmoa

mnm.o
wmm.a
moo.o
new.“

oeo.fi

mHN.H
NHN.H
000.0
wmm.o
m:m.o

dwa.fi
wde.fi
woo.H

MNO.H

Om

 

omm.m mucovvo 90 none:: onp
en wova>ev honssn pom one mo saw one "coo: m we

Sm.m 533233 3:38pm
o onspoaa on now: nmoumnuon 4 “*EoumOpmfim mm nu

SQN anon a 3 3.3
Icouwhon woodmoos oocoumflv one "ommwomnd me me

omw.m pom
onp mo epnomopa honfizn one "ihpwaocwvhoo mm mm

omm.m pom no>ww o CH UocHop
Icoo oncoEoHo mo uonEs: one "*hpwaocwvuwo He an

ome.m ommnm oco oufim oEom onp
o>mn pwnp mopsmwm ownposoow oze "oocoahmcoo mm on

ome.m oooam ca nowpoooa o
nnwz Uonwwoowwo oouw HNOflpoeonpoE d "pcfiom Hm mo

ome.N oamcowhn o :w onco
nnmwh onp opfimommo oven one "*omsnopomhm NH we

omn.m pom Mommas a
cw nonfioncoo ohm mucosoao Haw MH "poundm mm mm

e©©.N mucoemom oCflH ouo ponp
movem npfls ouswflm onoam vomoao < “neweaom am we

mmw.m 9323? a no pfiomuco 8:
an novnson oHopHo o mo ago n< "oHopHOflEom on we

mmw.m muonssc oHonz oho muonfisn oHonz
can moo mo #0560AQ Una saw one "ohnwoao mm :0

own.N Eopmem Hoswoov on» ca mHouoesz onwcz
on wood ouo mHonEhm pcohoMMHo coe "mpwmwm N: no

own .N flog?»
unmwn o mo modem pwdzowncoapoa 03» one "mmoq on No

coo:
.02

a 83288 25 83238 23H 3

 

120

 

 

 

Rm.” SE am; 33 £3 83 p8 8:205:23;
nausea» ca no soapfimog one ”mafiaacavno mm mm
Hmo.fi «mm.: m:m.« mom.m mum.o om~.m pcoguwcoo
no: pun HoHHvom movam ovdmommo Ho pawn
oncwm o mad>on Hopopoafiuvosv 4 "Uwoaomohe mm mm
mam.e omm.m ewm.fi emo.m wom.H oma.m owvo HoHHopom
on» unsouo Uopopop oncopooh o no owvo
opp an weapon ousMHo udaom < “havcfiaao mH mm
N¢N.H woo.: m:~.fi 3mm.m mam.“ mmo.m maoufio a no pmgosafic an» op
oocohoHESOuHo onp mo canon npwcoa one "am do :w
omm.o Hmo.m m:m.o :fiw.m mom.fi o:o.m soap mad
Infinomov ca vow: ohm uco upon npwz vovowo
Iommo unedpoohumno Houdposonpoz "muonsdz an mm
mmfi.e omm.: wmfi.fl ONN.N 000.0 o:o.m «gonad: oz» and Hanson a mo poohuoo
ma pony sow one mace ma onone «omocoaofico mm mm
«we.o :mo.H mom.H wmm.m oe:.H ooo.m unwom no>wm o Scum pcopmavaawo ouo mandoa
omen: mo Haw ondmam ocoaa vomoao 4 uoHoufio mm «m
mmm.c mmm.H mmn.« omm.m mmo.H ooo.m pcwomvco GOEEoo o wcw>on mean 039 Mo
songs can ca «pcflom Haw no new one "cameo um om
mw:.H mmm.m mnm.a wwm.m cam.o oom.m pnwom co>aw o Scum pcopmwvwavo we pcwom
nooo nonp ooomm ca mpcaoa mo pom 4 «ouonmm me me
oma.fi onfi.n :wm.H mfio.m owm.o mmm.~ popomm a ma vow: m“ genes: a mason
can mopaofiucfi pang Hoasmm < “psocogxm oa mm
wmm.o fidm.fi mom.H wen.m mmH.H mmm.m mnoonno opoponoo op vocmwmmo ouo muonadc
opons mmoooum cowpwmomaoo 4 upcosohamoot no em
mmH.H nfim.m gem.” amm.m omm.fi omm.m maszaoo uca much my
munoaoao mo pnoEomzohno hHhoouo :4 "hohu4 am on
3 swan on :8: am flux
.02
E BannoHCfimoo a coacwmoa E soapficwmom omhe scanwcfimon EovH nnom

 

121

soapacdmov Hooahoeo: onoo: I z
nonpficamov Hooauvoaoom mnooz a o

woo.“

mofi.fi

mee.o

mmn.fi

«no.0

“an.“

wwo.a

ooo.H
Mme.a
He©.o

emm.:

wem.fi

“no.3

noa.fi

mac.«

or:

oun.m Homo.fi

mom.n

NoH.m
mna.:

wwa.d
nfim.m
ooo.:

emm.«

:34
was.“

som.fi
Sm;
Ham.“

wmm.m

mmm.m

eo:.m

mom.n

mwfi.m

mmm.m
mow.m

nfim.m
oma.n
mmfi.m

moo.H

NON.e

mnm.a

moo.“

:«H.H

nmm.H
m:m.o

mom.o
mme.o
omH.H

owe.:

omm.m

ooo.n

mou.m

ome.m

oom.m
omn.m

om:.m
03:.m
oom.m

 

om

coo:

a 523228

cm

coo:

a 5288

am

a noooacaoon moon

ooogo ooaqom soon was asocoacflooo on» gas» annex.

oouo>na o>dpdvno neon»

duo muonssn oHonz onp Ho vooogsoo ow duo
ononadn Ho:0fipou on» no poonao 4 ”onowoucH

. . mono and no

uogofio poop. ooouooo oaoqofioo onoau u no
omonoponen onp en nonahooov vaaoo 4 "once

ooncoahu Houo
upoaaavo pconhwnoo HHo ono oooom Houopoa
nco ooon ouon3.ufisouhm 4 “schoononpoe
meoumoaoflaohom ouo moooe maficaoeou onp
.HoHHouom uco pcoohmcoo ouo oooom one
.oAOHuopcd neon» vco oooom Honothon
mo voooaaoo onswam ocoan voooao 4 «Eowum
nmoumuhon Hoo
napuo> o ca eocoavoum vogsoum o :H soap
noshoena mo nowpopcoooum one “isoumopmfim
vcoswoo onaa o Ho mowpfisonpxo
on» no ouo nuanz mvnwoa oze “opcfiomnnm
ousofio
owupoaoow o mo anon ono nownz mucosmoo
oQHH 03p mo pcfioacco :oEEoo one "txouuoe

oamco no madshom non» .ohou mcapooo
shops“ 03p Ho undomflno coeaoo one nonopnob

ocoam o cw mndha Eon» mo Ado .oocaa on»
honpo onooouopca non» onda one ”Hooho>o:ope

muonas:

coma :oapohomo no mo opasuon on» mcaucfie

ca woo: nonpoa HoCOHnonanEoo 4 "Eowhoma4

no no

no om

me am

om mm
mm mm

0: no

no om
mm mm

mm mm

 

coo:

cooodzfiooo

.oz

sooH xcom

 

122

The Meaningfulness Values of ReSponse Words

 

 

 

 

Rank Item WOrd Meaningfulness
No.

Mean SQ__
1 17 One 1.066 0.308
2 7 0f 1.091 0.340
3 53 To 1.114 0.427
4 82 Or 1.117 0.391
5 4 Two 1.132 0.497
6 124 All 1.136 0.468
7 15 As 1.149 0.475
8 27 In 1.149 0.525
9 11 Are 1.157 0.481
10 69 On 1.163 0.516
11 93 Plane 1. 168 0.455
12 1 Place 1.174 0.439
13 8 Set 1.190 0.565
14 29 Holds 1.198 0.492
15 14 Face 1.198 0.508
16 129 But 1.203 0.546
17 134 Line 1.205 0.481
18 67 Is 1.205 0.586
19 2 Earth 1.207 0.480
20 36 Not 1.207 0. 544
21 33 Same 1.208 0.498
22 30 And 1.208 0.515
23 89 Sum 1.210 0.548
24 54 Side 1.211 0.545
25 131 Line 1.212 0.502
26 96 Many 1.218 0.488
27 10 Used 1.225 0.539
28 L12 Used 1.230 0.598
29 128 Have 1.231 0.513
30 38 Zero 1.231 0.613
32 141 Base 1.239 0.557
33 3 Number 1.240 0.515
34 9 Point 1.240 0.515
35 20 South 1.242 0.516

 

124

 

 

Rank Item Word Meaningfulness
No.
Mean SD
36 73 From 1.244 0.483
37 60 Right 1.246 0.618
38 13 Empty 1.248 0.534
39 110 Only 1.252 0.557
no 127 Called 1.256 0.557
41 32 Times 1.264 0.600
42 37 Closed 1.264 0.613
43 76 That 1.268 0.639
44 98 Having 1.271 0.592
45 115 Pair 1.272 0.663
46 50 Given 1.276 0.499
47 86 There 1.283 0.608
48 126 Addition 1.291 0.600
49 92 Every 1.294 0.570
50 70 North 1.301 0.583
51 78 Center 1.303 0.585
52 130 Rate 1.304 0.695
53 28 Arithmetic 1.308 0.656
54 43 Picture 1.314 0.617
55 112 Finding 1.325 0.600
56 44 Poles 1.328 0.607
57 19 Circle 1.331 0.635
58 136 Upon 1.336 0.586
59 137 ‘Which 1.336 0.730
60 46 Upon 1.341 0.609
61 74 Hundred 1.344 0.584
62 10 How Many 1.355 0.628
63 12 Another 1.355 0.641
64 49 Between 1.358 0-652
65 5 Sentence 1036“ O. 617
66 120 Square 1.372 0.627
67 117 Subtraction 1.400 0.755
68 108 Correct 1.409 0.709
69 84 'Whose 1.427 0.683
70 71 Figure 1.430 0.702

 

125

 

 

Rank Item Word Meaningfulness
No.

Mean £31..
71 31 HalféWay 1.438 0.726
72 55 Division 1.459 0.679
73 77 Enclosed 1.459 0.691
74 56 Opposite 1.459 0.726
75 132 Thus 1.466 0.724
76 102 Distance 1.470 0.735
77 107 Cube 1.470 0.816
78 94 Symbol 1.483 0.741
79 61 Fraction 1.521 0.772
8o 18 Region 1.544 0.796
81 97 Discount 1.551 0.788
82 51 Measured 1.553 0.798
84 140 System 1.590 0.859
85 80 Results 1.592 0.769
86 19 Degree 1.595 0.788
87 26 Opposite 1.597 0.892
88 23 Numerals 1.600 0.768
89 116 Placeholder 1.605 0.885
90 121 Empty Set 1.653 0.867
91 47 Triangle 1.659 0.909
92 68 Contained 1.664 0.902
93 62 Opposite Sides 1.678 0.815
94 21 Single Pair 1.702 0.820
95 34 Operation 1.727 0.945
96 52 Equator 1.730 0.887
97 101 Property 1.735 0.955
98 66 Factor 1.754 0.908
99 91 Process 1.754 0.911
100 113 Number Sentence 1.776 0.948
101 88 Imaginary 1.840 1.012
102 22 Digits 1.983 1.052
103 39 Indicates 2.008 1.029
104 57 Numerator 2.025 1.075
105 138 Constructed 2.103 1.086

 

126

 

 

Rank Item ‘Word Meaningfulness
No.
Megg__ SD

106 123 Parallel 2.101 1.103
107 100 Bar-Graph 2.178 1.246
108 111 Pyramid 2.226 1.339
109 125 Inverse 2.231 1.128
110 16 Intersecting 2.281 1.108
111 114 Denominator 2.304 1.428
112 122 Horizontally 2.331 1.215
113 103 Prism 2.359 1.237
114 59 Number Property 2.382 1.137
115 109 Addenda 2.395 1.226
116 81 Frequency 2.400 1.150
117 99 Reduction 2.407 1.068
118 139 Enumeration 2.419 1.228
119 105 Percentage 2.444 1.380
120 106 Ellipse 2.586 1.390
121 58 Precedes 2.595 1.161
122 72 Exponent 2.650 1.276
123 64 Distribution 2.780 1.328
124 133 Geometric 2.932 1.287
125 136 Vertex 3.009 1.221
126 83 Concentric 2.119 1.158
127 35 Predecessor 3.174 1.251
128 24 Polygon 3.175 1.321
129 85 Congruent 3.303 1.199
130 65 Histogram 3.426 1.234
131 104 Elongated 3.504 1.350
132 90 Abscissa 3.521 1.389
133 41 Cardinality 3. 52 5 1.088
134 48 Modulo 3.557 1.230
135 79 Trapezoid 3.648 1.207
136 25 Computational 3.756 1.277
137 95 Hypotenuse 3.798 1.227
138 45 Algorism 3.811 1.043
139 75 Uniqueness 3.820 1.174
140 119 Equidistant 3.826 1.22 5
141 63 Quadrilateral 3.837 1.321

 

APPENDIX E

Methodological Information Concerning the Reliability of the Defini-
tion and Vocabulary Meaningfulness Scales

 

 

Information Definition Vocabulary
Reliability of AD .9795
Reliability of ND .9950
Reliability of GD .9670
Reliability of Vocabulary .9865
Reliability Method Split Half Test Retest
No. of Students 25 61
Mdn. Age in Mbnths 144 150
Grade 6 7
School B C
Dates of Administrating 5/3/65 10/6/65
the Scale

-- 10/13/65

 

AD - Arithmetical definitions
ND - Numerical definitions

GD - Geometrical definitions

127

APPENDIX F

'The Relationship Between the Variables of Definition Components

 

It is known that conclusion about the direction of the correla-
tion must be based on the direction of the scale. For example. all
the m values reported in this study are presented in a descending
rather than ascending order. The high items get smaller numeral
values than the low m items which get higher numeral values. Such
descending order of the m value has its effect on explanation of the
correlation coefficient. Suppose that the correlation between m of
the reSponse and the number of words of the re5ponse is +.6. Although
the correlation sign is +ve, yet because of reversing one of the scales,
its interpretation is; as the meaningfulness (m) of the response

decreased, the sentence length increased.

1: Description of the Frgguency Distribution of m.V§lues
The m values of the stimuli, reSponses, and definitions have

been reported in Appendix D. Table 18 shows the frequency distribu-
tions of the previous m values in case of numerical. geometrical and
arithmetical items.

Figure 4 represents the frequency distributions of Table 18.
Parts (a) and (o). Figure 4, shows that the distribution of defini-
tions and re5ponses of the numerical items are shifted toward the
higher level of m, while the corresponding curves of geometrical items
are shifted towards the lower level of m. The distributions of the
stimuli m, as shown in part (b) of the figure. are bimodal in the case

of numerical and geometrical items.

128

129

TABLE 18: The Frequency Distribution of Meaningfulness of Defini-

tional Components

 

#—

 

 

 

m Definition Stimulus Response

A— N G A N G N G
1.0 - 1.499 12 7 5 23 15 8 6 5 1
1.5 - 1.999 27 18 9 19 7 12 12 9 3
2.0 - 2.499 20 12 8 11 6 5 27 19 8
2.5 - 2.999 20 6 14 12 6 6 24 9 15
3-0 - 3-499 11 4 7 7 3 4 18 4 14
3.5 - 3.999 6 0 6 6 4 2 10 2 8
4.0 - 4.499 1 1 0 15 5 10 O O 0
4.5 - 4.999 0 0 0 4 2 2 0 O 0
Total 97 48 49 97 48 49 97 48 49

 

A = Arithmetical items
N = Numerical items

G = Geometrical items

Since arithmetical items are the combination of numerical and
geometrical items, the frequency of arithmetical items appears in
Fig. 4, parts (a). (b) and (c) as the sum.of the frequencies of the
other two types of items. Moreover, it is clear from these parts
that the distribution of arithmetical items has the general shape of
both the numerical and geometrical items.

The distribution of m for definitions, stimuli and reaponses

that correspond to arithmetic items are plotted in.part (d) of Fig. 4.

are;

Freq u e '10 v

 

— N N on
A A A PLJil L pt 1 A l I“. A A A 2‘;

.—
lUl [CD

JaiLL LAAL A
.

 

1:: 253100“.
|.| sealanlooe

A

A

I >nw$§onuoae .

N

— ~ .
P11.41’.iii

Ma:

A

 

c.

. w 5355

        

 

  

 
        

 

  

new:

nonvosao

up.
:5... Ease .

Ii. Oshawa-e J .

iieizot v

r. a
a Q l .
.h . n
. . . 84
1
.J. .
_ . I
l. n i no. A
, re . y. >.
11\ o.
u 8... 1 I

1

c
, , .
: g’). L...
‘ - v
. 1. .
1.. ea

-’
V
1.

0|

J
‘AJJ L.‘

'naa

>'j.A.

.‘AI-

          
    

  

   
       

  

        

 

   

131

It is observed that stimulus m.has a bimodal distribution. Moreover.
the distribution of definition m is shifted towards the higher level
of m, while the distribution of response m is shifted towards the
lower level.

Table 19 shows means and standard deviatipns of the definition
stimulus, and response variables for the numerical. geometrical. and
arithmetical items. From this table it is apparent that the mean
number of letters of numerical items' stimuli is greater than the mean
number of letters of geometrical items' stimuli. The mean number of
words of geometrical items' response is greater than in case of numer-
ical items' responses. However, the differences between the two means,
using t test in both cases, are insignificant at the .05 level.

Table 19 indicates also that the standard deviation of stimuli
m is greater than the standard deviation of either the responses m,or
definitions m. This observation is consistent for the three types of
items. The F ratio test shows that there are significant differences
between the variance of the stimuli from one hand and the variance of
the responses or the variance of the definitions on the other hand.
The level of significance is the .01 level.

Table 20 contains the analysis of variance results of m. This
table has been reported with reluctance since the analysis of vari-
ance assumptions have not been met satisfactorily. For example, it
has been mentioned that stimuli m had a bimodal distribution with a
significantly high variance, while the definition or response m had
almost a normal distribution with low variances. Thus the stimuli m

violated the normality of the population and homogeneity of the

 

08.6 memo Rea :23 New.“ o$.~ meme $84. onto amino me 4868

 

 

 

 

 

8nd mend £86 305 33 mood moo.~ new.“ owes Bed w: doc .52
does demo 39m 886 024 some ooo.... Rec. mine Rod Ne don 882
IIIIIIII: FIIIIIIflILIIIIIIIIIL
d .m :32 .a .m moon .n .m on .e .m 5o: IJIIIIIIIQ .m use:
.8 mono: .02 a omopvoq .oz 8
oesoaoom monsoon 83888 2 oooom

 

 

 

 

 

 

, 228589138582 use dosage 4.3852

mo omoo cw moanefihose encamom use «3pm .mfloavedweon one. no £353.:on Pagan ecu 802 “me age

132

133

variances' assumptions. However, some studies (Lindquist, 1956) men-
tioned that heterogeneity in form or variance or both must be quite

extreme to be of any serious consequence. Otherwise the effect upon
the F distribution will probably be negligible. The results on Table
20 are then reported on the basis that stimuli m distribution is not

extremely deviant.

TABLE 20: Analysis of Variance of Meaningfulness Values

—— L7
I

 

 

 

 

 

Source of Variation S .S . d. f. M.S . F
Definition Components 7.5693 2 3.7846 4.915
Interaction 1 . 1792 2 . 5896 . 765
Between 22.9172 5
Within 219.4455 285 .7700
Total 242.3627 290

 

In addition, the results reported in Table 20 are limited to
the numerical and geometrical items. Statistics on the arithmeticdl
items have been excluded, because of their dependence on the other
two types of items, in order to meet the analysis of variance assump-
tion of independence.

Inspection of Table 20 reveals that types of items as well as
definition components, using F test, are significant sources of vari-
ance at the .01 level. The interaction of the two factors contributed

no significant variance at the .05 level; moreover. the results showed

134

that the mean m of the numerical items is significantly higher than
the mean m of geometrical items at the .01 level. The same test also
indicated that the mean m of definitions is significantly higher than
the mean m.of the reSponses at the .01 level, and the mean m.of the
stimuli at the .05 level. However, the mean m of the stimuli is not
significantly different from the mean m of reSponses at the .05 level.
The results suggest the acceptance of the following hypotheses:
1. The variance of the stimuli m are significantly greater than
the variances of either definitions or responses m.
2. The m.of geometrical items has greater variance than geomet-
ric items.
3. In case of arithmetical as well as numerical items; there is
a significant difference between the mean of definitions m
on one side and the mean of either reaponse m or stimuli m
on the other side.
4. In case of geometrical items, there is a significant differ-

ence between the means of definitions 'm and responses 'm.

2: The Relationship Betweeanbrd.Leggth and Its Meaniggfulness
Table 21 presents the correlation coefficients between meaning-

fulness and the length of either the stimuli or the vocabulary. It
might be noticed that the correlation coefficient is not significant
from zero in case of the arithmetical as well as numerical items for
0‘ equals .05. However, in case of geometrical items the correlation

coefficient is significantly different from zero when °< equals .01.

135

TABLE 21: The Correlation Coefficients Between Stimuli and Vocabulary.

Number of Letters and Their Meaningfulness.

l
——-

 

 

 

Types of Items N r

Arithmetical Items 97 .192
Numerical Items 48 -.O46
Geometrical Items 49 .467**

Vocabulary 141 .656**

 

**Significant at the .01 level

The table also shows that the correlation of vocabulary length
and m is not zero at the .01 level. The significant correlations
which are found in case of the vocabulary and the geometrical items
suggest that the shorter words were highly meaningful than the longer
ones.

It has been mentioned that each item is reported with two values,
i.e., m and its standard deviation. ‘When the number of letters is cor-
related with the standard deviation of m, in case of arithmetical. numer-
ical, and geometrical items, the coefficients are as follows: -.138.
-.180 and -.032. While the previous correlations are not far from zero
at the .05 level, it was found that the correSponding value in case of
the vocabulary (+.733) is not zero, foro( equals .01. The latter cor-
relation suggests that the vocabulary which was widely dispersed with
reSpect to m was longer. waever, the inconsistency of this notion in
case of the other kinds of items might be due to differences in the

sampling process.

136

The results of this part of the analysis made the following

hypothesis acceptable:

1. There is a significant correlation between stimuli m of geo-

metric items and the length of the item.

2. The correlations of the number of letters with either vocabu-

lary m, or vocabulary standard deviation are significant.

2: The Relationship Between ResEOnse Length and Its Meaningfulness

TABLE 22: The Correlation Coefficients Between the Response Number

of Words and Response Meaningfulness

 

 

 

Types of Items N r

Arithmetical Items 97 .538**
Numerical Items 48 .409**
Geometrical Items 49 o595M

 

**Significant at the .01 level

An investigation of Table 22 reveals that the correlation
coefficient between the reSponse number of words and response m is
far from being a zero at the .01 level. These high positive corre-
lations showed that the shorter the reSponse the higher was the
reSponse m.

The correlation is smaller in the case of numerical items than
in that of geometrical items. This observation suggested testing the
hypothesis that the numerical items' correlation is equal to or less

than the geometrical items' correlation with¢>( equal to .05. The

137

results of a test showed that this hypothesis, although it is accepted
at the .1 level, is not acceptable at the .05 level.

The correlation between the standard deviation of responses and
the number of words of the reSponse was computed for the three types
of items. The coefficients .189, .182, and .085 were feund for arith-
metical. numerical and geometrical items respectively. These coeffi-
cients are not significant at the .05 level.

The results of this section suggest the acceptance of the hypothp
esis which states that there is a significant correlation between the
number of words in the response and response m in case of the arithmet-
ical, numerical and geometrical items. In addition, the hypothesis
which states that there is significant correlation.between response

standard deviation and response m is rejected.

4: The Relationship Between.Stimulus Vppiables gpg Reaponse Vppigbles
Investigation of Table 23 reveals that each of the reported cor-

relation coefficients is nearly equal to zero at the .05 level. Thus
the hypothesis which states that there is significant relationship

between stimulus variables and response variables is rejected.

138

TABLE 23: The Correlation Coefficients Between15timulus Variables and

Response‘Variables

 

 

Stimulus Variables

 

 

 

 

m S. D. No. of Type of Items
Letters
.103 .045 -.129 Arithmetical Items
3 .090 -.007 -.107 Numerical Items
.023 -.076 -.061 Geometrical Items
”3%
,2 .047 .116 -.052 Arithmetical Items
3 .
>’ C‘ .070 .088 -.071 Numerical Items
0 e
g m .180 .046 .070 Geometrical Items
it
$3
.142 .034 .017 Arithmetical Items
#4
°.§ .149 .051 .032 Numerical Items
0 o
z 3 .090 -.086 .060 Geometrical Items

 

 

 

 

5: The Relationship Between Definition Variables and Stimulus Variables

Table 24 shows that there are significant correlations between m
of the definition and m of the stimuli at the .01 level. It shows also
that m of the definition correlates significantly with the stimuli‘s
standard deviation for arithmetical and numerical items at the .05
level, while it is insignificant with geometrical items. Meaningfulness

of the definitions does not correlate with the stimuli's number of
letters at the .05 level.

139

TABLE 24: The Correlation Coefficients Between Definition Variables

and Stimlus Variables

 

 

Stimulus Variables

 

 

 

m S. D. No. of Type of Items
Letters
. 524*“ .206* .033 Arithmetical Items
:3 s .527" .304"I -.061 Numerical Items
"a .509" .009 .189 Geometrical Items
:>
g .328" .097 -.042 Arithmetical Items
5 c5 .376M .209 -.218 Numerical Items
8 a: .281* -.054 .120 Geometrical Items

 

 

 

 

* Significant correlation at the .05 level

"Significant correlation at the .01 level

The standard deviations of the definitions have been found to
correlate significantly with m of arithmetical and numerical items'
stimuli at the .01 level, and with geometrical items' stimuli at the
.0 5 level. However, the standard deviation of the definition is
found to correlate significantly at the .0 5 level with neither the
stimli standard deviation nor the stimuli number of letters.

The reported correlations indicate that when m of the defini-
tions was high, the m of the stimuli was also high. They also suggest
that when m of the definition was high the standard deviation of
arithmetical and numerical items' stimuli were low. In addition,

140

when standard deviation of definitions increased, the m of the stimuli
were found to decrease.

The following hypotheses have been accepted in accordance with

the preceding results:

1. For the arithmetical, numerical and geometrical items, the
m of the definitions correlates significantly with m of the
stimuli.

2. For the arithmetical and numerical items, the m of the defi-
nitions correlates significantly with the stimuli's standard
deviation.

3. For the arithmetical, numerical and geometrical items, the
standard deviation of the reaponses correlates significantly
with stimuli m.

6: The Relationship Between Definition Variables and Response Vgriables
Investigation of Table 25 reveals that definition m.has signifi-

cant correlations with response m and number of words at the .01 level.
Definition m shows also significant correlation at (a) .01 level with
the standard deviation of arithmetical items and (b) .05 level with the
standard deviation of geometrical items. Furthermore, it shows insig-
nificant correlation at the .05 level with the standard deviation of
numerical items. 7

At the .01 level, it is found that definition standard deviation
correlates significantly with (a) response m for the three types of
items, (b) standard deviation of arithmetical responses, and (c) response
number of words of arithmetical and geometrical items. It also corre-

lates with the standard deviation of the geometrical reSponses at the

141

.05 level. In case of numerical items, there is no significant cor-

relation between definition standard deviation and either response

standard deviation or response number of words.

TABLE 25: The Correlation Coefficients Between Definition Variables

and Response Variables

 

 

Response Variables

 

 

 

m S. D. No. of Type of Item
Words
.730’Ml .306“ .715‘” Arithmetical Items
1% s .696" .222 .663" Numerical Items
3 .703" .3116:' .732“ Geometrical Items
’2
:3 .460“ .271" .390” Arithmetical Items
g Q: .400" .246 .270 Numerical Items
3 m . 536“ .340* .477" Geometrical Items

 

 

 

 

* Significant correlation at the .05 level

MSignificant correlation at the .01 level

The previous correlations might mean that when definition m was

high (a) response m was also high, (b) response standard deviation was

small, and (c) response number of words was few.

This statement holds

for the three types of items except in case of the correlation between

definition m and reaponse standard deviation of arithmetical items.

Moreover, when definition standard deviation was high (a) the reaponse

m was low, (b) response standard deviation was also high and (c) the

142

response number of words was high. ‘While this observation is valid for
the three types of items, it still has an exception, for in the case of
numerical items the relationships between definition standard deviation
and either reSponse standard deviations or reSponse number of words are
not quite significant.

The accepted hypotheses on the basis of information presented are

as follows:

1. There is a significant correlation, for the three types of
items, between definition m and either reSponse m or reSponse
number of words.

2. In case of arithmetical and geometrical items, there is cor-
relation between definition m and response standard deviation
or response number of words.

3. In case of arithmetical and geometrical items, there are no
correlations between definition standard deviation and
response m, response standard deviation and response number
of words.

4. In case of numerical definitions, there is significant cor-

relation between definition standard deviation and response m.

2: The Relationship Between Definition Varigbles
The correlation coefficients between definition m and definition

standard deviation for arithmetical, numerical and geometrical items
are .519, .450, and .568 reSpectively. All three correlations are
significant at the .01 level. The null hypothesis which says that

there is no significant correlation between definition m and definition

143

standard deviation is not accepted. Rather, it is suggested that the
preceding correlations might mean that when the definition m was high,

the definition standard deviation was low.

8: The Relationship Between DefinitionsI Stimuli, and Responses When
a Part of Their Variables is Partialled Out

Table 26 shows the zero order correlation coefficient of defini-
tion m with another variable. Following this zero order correlation
are a number of higher order correlation coefficients where some vari-
ables, other than the correlated ones were partialled out. The higher
order correlation coefficients have been compared with their order
correlation coefficient. If the comparison revealed a significant dif-
ference between the two correlations, this might be the result of’ruling
out the partialled variables.

Investigation of the previous table shows that partialling any
group of variables out of the correlation of definition m with the other
variables does not contribute significant change of the zero order cor-
relation coefficient at the .05 level. Actually this result suggested
seeking the result of ruling the definition m out of the stimulus and
reSponse variables. Table 27 contains the results of this step, and
some unique observations will be shown below.

It has been mentioned that response m and response number of words
correlate significantly. waever, when definition m alone, or other
variables beside definition m were partialled out, all the new correla-
tions dropped to a value which is not far from zero at the .05 level.

In addition, there were significant differences between the new correla-
tions and the zero order correlation coefficient. The levels of signi-

ficance are shown in Table 27.

144

TABLE 26: Partial Correlation Coefficients Between Definition and

Stimulus or Re5ponse Variables

 

 

 

 

Zero Order Corr. Coef. Partialled Variables A N G
Definition m and
Response m .730 .696 .703
Response no. of
words .586 .622 .489
Stimulus m .798 .766 .803
Stimulus m, and
stimulus num-
ber of letters .802 .798‘ .766
Response no. of
words and stimu-
lus m .716 .739 .678
Stimulus m, response
no. of words,
stimulus no. of
letters .712 .737 .673
Definition m and
No. of Response
‘Words .715 .663 .731
Response m .560 .577 .540
Stimulus m and
response m .659 .658 .673
Stimulus m, response
m, and no. of
letters of
stimulus .657 .657 .675

 

TABLE 26-~Continued

145

 

 

 

Zero Order Corr. Coef. Partialled Variables A N G
Definition and
Stimulus m .524 . 527 .509
Response m .661 .645 .693
Response m, and
reSponse length .731 .713 .770
No. of letters of
stimulus .528 .526 .484
Response m, response
no. of words, and
no. of letters of
stimuli .724 .713 .742
Definition and No.
of Letters of
Stimulus .033 -.061 .189
Stimulus m -0081 “90,43 -0063
Stimulus m, response
m, and no. of
words in response .015 -.002 -.074

 

Ho>oH #0. pm cowpmaeaeoo bongo open mpfl use :owpmaohhoo powwoaoh one cesspon oesopommdp pancakesmflmes

Ho>oH no. as sowpeaeaaoo nacho ones new use cowpaaeaeoo eovuomon can escapee oesouomMHv passtHsmwm s

 

 

 

 

one: $0... «3... meme: no sense onsoanoa .
use a msgesapm .s sowpfisdkon
Ref 0:... m3: 2 messes... one .s soflesfloe
Hmo.l mofi.a mmfi.l escaped Ho hopes:
meassfium use a oncommom
mmd.l *romd.u srwme.l unsaved mo houses
msdssapm use 8 soapwsamon
ramm:.1 *rmam.u snmme.1 meoppoa
X. e. define
uncommon .E soapfiswmom
racem.c **Nfim.n *swom.l mphoz mo poses:
. oncogene use E sowpficwmom
33m .1 Inns? 1.5:... e 2032.83
mmo. ooo. moa. E msAdEfipm use E uncommom
*rwNH.I *rmmm.l **me.I meoupoa
mo houses meadsfipm use
E mstEHum .E sowufiswmon
*rmma.n mwN.: *smma.n e mstEHpm use a :ofiufiswmon
ova. *rumo.1 **:mo. 5 sowpwcwmoa
mom. mod. wmn. memos mo homes:
uncommon use a oncommom
.sooo .ssz .bpwe< .mmmoo .uaoo poHHewpawm .mmooo .huon scene chow

 

 

E soap

Iwcflmmm wcwHHmwphwm mHHSB moaflmwem> mSHfiewpm use omcoamom cesspom ucmflofimmoou cowpmHohhoo "AN mqm<9

146

147

Again, response m and stimulus m have been found to have an insig-
nificant correlation. Once the definition m or other variables beside
definition m were partialled out, all the new correlations became sig-
nificant at the .01 level. Moreover, the new correlations differ sig-
nificantly from their zero order correlations.

In the case of partialling definition m out of the zero order cor-
relation coefficient of reSponse m and stimulus number of letters, the
new correlations are still insignificant. Nor are there any significant
differences between the zero order correlations and their higher order
ones at the .05 level.

It has been noticed (Table 27) that the correlations after the
process of partiallization might or might not be different from the zero
order coefficients. The following are the hypotheses which are accepted
thus far:

1. The correlation of response m, and reSponse number of words,

does change significantly when definition m or definition m
plus other variables are partialled out.

2. The correlation of response m and stimulus m does change
significantly when definition m or definition m with other
variables are partialled out.

The preceding results emphasize the role of definition m and its
relation with reSponse m or stimulus m. When definition m correlates
with either stimulus m.or reSponse m, the resultant correlation will
not be affected by the process of partiallization of either the reaponse
m or stimulus m respectively. Contrany to this, when definition m.is

partialled out, the zero correlations of either stimulus m or response m

148

or both will change considerably. For example, the new higher order
correlations while partiallizing definition m might mean: (1) When
the reSponse m was high the length of the reSponse was not necessarily
high or necessarily low. (2) When re5ponses m were high, the stimuli

were also high.

2: Meanipgfulness of the Copposite Definition and Its Relation to the
Other Variables

TABLE 28: Means and Standard Deviation of Composite Definition m, and

Definition m of Arithmetical, Numerical and Geometrical

 

 

 

Items
Type of Items Mean m S. D. Type of Item
Arithmetical 5.153 1.432 Composite Definition
2.236 .748 Definition
Numerical 4.711 1.370 Composite Definition
2.010 .650 Definition
Geometrical 5.586 1.370 Composite Definition
2.459 .778 Definition

 

Investigation of Table 28 shows that the means and standard devi-
ations of composite definitions” m are significantly greater than their
equivalents of definitions m at the .01 level. The increase of m came
from adding two m values together--one for the stimuli m, and the other
for the response m. Thus the maximum value of m for the composite defi-
nition is ten rather than five (the maximum scaling point) for defini-
tion m. Also, because of the nature of the composite definition m, its
range, i.e. variation might be nine (which is greater than the range of

definition m that equals four).

149

TABLE 29: The Correlation Coefficients of Each of the Composite Defini-

tion m, and Definition m with Stimuli, ReSponses and Defini-

 

 

 

 

 

tions' m
Arith. Num. Geom. Type of Item
m .885**$$ .901**$$ .895**$$ Composite Definition
g .524 .527 .509 Definition
'3 No. of
:3 Letters .101 -.087 .386" Composite Definition
”1 .033 -.061 .189 Definition
m .553** .513“I .466**$$ Composite Definition
§ .730 .696 .703 Definition
0
8‘ No. of
,3 Words .370**$$ .306**$$ .344*$$ Composite Definition
.715 .663 .732 Definition
2
{3,3 m .780" .757" .764" Composite Definition
:3 ”E

 

 

 

 

* Significant correlation at the .05 level
**Significant correlation at the .01 level
$$Significant difference between this correlation and the one right
below it

Table 29 indicates significant correlation between m of the comp
posits definition and m of either the stimuli or responses at the .01
level. While the composite definition correlates with geometrical
stimuli number of letters, it also correlates with response number of
words for the three types of items at the .05 level. The stimulus m
correlates higher with composite definitions than in case of definition

m at the .01 level. On the other hand, reSponse m correlates higher

150

with definition m than in case of composite definition. The differ-
ence is significant only in the case of geometrical response at the
.05 level.

The composite definition m correlates high with geometrical
stimuli number of letters. However, in case of definition m, it cor-
relates higher with reSponse number of words than with the composite
definition m at the .01 level.

These results indicate that the composite definition's relation
with definition components is somewhat similar to that of the actual
definitions. However, there are some differences between them. For
example, the correlation of composite definition m.with stimulus m is
greater than its correSponding correlation with response m at the .01
level of significance. But in case of definition m, the entire pre-
ceding statement is reversed except that the difference is not signi-
ficant at the .05 level.

The results of Table 30 will be speculated on the basis of the
previous results. ‘When the composite definition m is correlated with
definition m, the process of partiallization did not show a signifi-
cant change. However, there is a decrease in the value of the corre-
lation when r65ponse m is partialled, and an increase when stimulus m
is also partialled out. If the composite definition is considered as
stimulus m.plus response m, then any partiallization of either one of
them will leave the other to correlate alone with the other variable.
Also, if both stimulus m, and reSponse m are partialled out of a zero
order correlation containing a composite definition, the result is

zero. Then the increase of the correlation as a result of partialling

TABLE 30:

151

nition m with Other Variables

Partialled Correlation Coefficients of the Composite Defi-

 

 

 

 

 

 

 

Zero Order Corr. Coeff. Partialled Variables Arith. Num. Geo.
C. D. and Definition m .780 .757 .764
ReSponse m .661 .649 .693
Stimulus m .798 .766 .803
Response m and
reSponse no.
of words .732 .713 .770
Stimulus m and
stimulus no.
of;letters .798 .766 .802
C. D. and Re5ponse m .553 .513 .466
Definition m .038 .029 -.154
Definition m and
response no.
or Words -0026 .075 '0085
c. D. and Stimulus m .885 .901 .895
Definition m .894 .904 .912
Definition m and
stimulus no.
of letters .895 .905 .896
C. D. and ReSponse No.
of Words .370 .306 .344
Definition m and
reSponSe m -.429 -eu’0’+ -9474
C. D. and Stimulus
No. of Letters .101 -.087 .386
Definition m and
Stimuli-ifs m -0145 '0110 -00 i1

 

D. - Means composite definition m

152

stimulus m. However the decrease in case of partialling reSponse m,
is due to the previous results which state that definition m correlates
high with reSponse m.rather than with stimulus m.

The partiallization of definition m caused the correlation
between the composite definition m and response m to drop significantly
to about zero, and the correlation between the composite definition m
and stimulus m to increase an insignificant amount. The level of sig-
nificance is .05. The same relation is valid in case of partialling
definition m, reSponse m, or definition m and stimulus m out of the
zero order correlation of the composite definition and reSponse number
of words, or the correlation of the composite definition and stimulus
number of letters respectively.

Hence, it could be said that the composite definition m corre-
lated higher with stimulus m than with response m. But the definition
m correlated higher with the reSponse m than with stimulus m. Thus
when definition m was partialled out the considerable change brought
about the correlation of reSponse m with the composite definition m
while less change caused the correlation of stimulus m with the com.

posite definition m.

10: Meanipgfulness of the Composite ReSponse and Its Relation to the
Other Variables

The m value of the composite reSponse is obtained by adding up
the m of the individual words which compose the reSponse. The
reaponses used in this part are thirty sentences. Occasionally a com-

parison will be held between the m of reSponse and composite response.

153

The composite response m has a mean of 1.828 and standard devia-
tion of 2.338. The corresponding values for reSponse m are 2.540 for
the mean and .710 for the standard deviation. Comparing the means and
standard deviations of both types of reSponse, one might notice that
there is no significant difference between the means, but the variances
are very significantly different. The level of significance is the .05.

Table 31 shows that the composite response m correlates signifi-
cantly at the .1 level, only with the standard deviation of either
stimuli m or response m. The other composite reSponse m correlations

are not far from zero foro< equal .1.

TABLE 31: The Correlation Coefficients of Definitional Variables With

Composite Response m and ReSponse m

 

 

 

 

 

Definitional Variables Composite Response
ReSponse

Stilmlli -e186 -0018

m Responses .075 1.000
Definitions -.154 .566***
Stimuli -.322* .022

S. D. ReSponses .337* .564***
Definition -.123 .433**

Length Stimuli .026 -.246
Responses -.095 .313*

 

* Significant at the .1 level
** Significant at the .05 level

***Significant at the .01 level

154

It has been shown in Table 30 that the definition m and the com-
posite definition m correlate significantly at the .01 level. However,
the results are different in the case of response m and composite
response m. Their correlation is almost zero at the .05 level. The
response m seems to correlate highly at the .01 level with definition
m, but the composite response m correlation with definition m is not
significant from zero. Furthermore, there is a significant difference
between the correlations of definition m with composite reSponse m and
the reSponse m at the .01 level.

The two significant correlations can be explained as follows:
(1) When the composite response meaningfulness was high, the standard
deviation of the stimuli m.was also high. (2) The composite response

m increased as the standard deviation of reSponse m was decreased.

APPENDIX G

Lists of Definitions Representing Each Type of Definition

 

H-H List
Percentage:

Division:

Face:

Latitude:

L-H List
Predecessor:

Uniqueness:

Abscissa:

Hypotenuse:

H-L List
Discount:

Placeholder:

Face:

Circle:

L-L List
Cardinality:

Algorism:

Ellipse:

Trapezoid:

A given proportion in every hundred

The process of finding how many times a number
is contained in another number

A side of a pyramid

The distance north or south of the equator
measured in degrees

A number that precedes another number

There is only one sum that is correct as a sum
of any two numbers

The distance measured horizontally to a point

The side opposite the right angle in a triangle

The rate of reduction

A symbol that holds a place for a numeral in a
number sentence

A region of a plane enclosed by a polygon

A closed plane figure all of whose points are
equidistant from a given point

The number property of the set

A computational method used in finding the result
of an operation upon a number

An elongated circle

A quadrilateral having a single pair of opposite
sides parallel but not congruent

155

APPENDIX H

Instructions and the Familiarization Materials

 

Instructions: I shall show you one sentence which expresses the mean-
ing of each word you have just seen (in the pre-test). You are going
to see the word (stimulus) try to guess its meaning in one sentence.
Then you will see a sentence which tells the meaning of this word.
Read this sentence carefully, because you are going to write it down,
exactly as it is, without any change. Again, read the word, guess

its meaning in one sentence, then read the sentence carefully since
you shall rewrite it again without any change in its words.

(In order to make it easy for you to remember the sentence
exactly, I shall explain it by using three other sentences. These
sentences are just to help you understand the one sentence which will
be shown to you later. Read these sentences carefully and try to
understand their common meaning. But remember, you are not going to
write any of these sentences again. They will just help you to under-
stand the meaning of the sentence.

Remember again, read the three sentences and try only to under-
stand their common meaning. Second, when you see the word, guess its
meaning, then read the one sentence carefully. Third, when you are
asked to write down the meaning of the word, try to write exactly the
same one sentence which will be shown to you, without making any

change in this sentence.)

() - Instructions for the §_s of the familiarization treatments

156

157

Verbal Familiarization Material:

H-H List

 

Percentage: Jack improved in his job as a salesman. The presi-
dent gives him a proportion of his sales products. He takes ten
dollars for every hundred dollars.

Division: Jim has eighteen marbles. He wanted to know how many
groups of three marbles there are in the eighteen marbles. He divided
the marbles into groups of three and counted the number of groups.

£333; A pyramid has many sides. One side is called a base.
Other sides have other names.

Latitude: Jim learned that the weather is different in differ-
ent cities. The weather of the city almost depends on its distance
from the equator. The distance of the city North or South of the

equator is measured in degrees.

L-H,List

 

Predecessor: ‘We count upward by adding one to each number. ‘We
can also count downward. In this case we get the lower number by sub-
tracting one from each number.

Uniqueness: The teacher asked the pupils to get the sum of two
numbers. There is only one sum that is a correct result of this prob-
lem. Any other sum is wrong.

Abscissa: Bill drew two perpendicular lines at the middle of
the page. One line is a horizontal line and the second is a vertical
line. Any point has a distance away from the vertical line.

Hypotenuse: Jack drew a right triangle. Two sides form the right

angle. The third side is opposite in position to the right angle.

158

HsL List

Discount: Jack will buy a new suit. The store will make a ten
percent reduction in price. He will buy his new suit at a lower price
because of the ten percent off.

Placeholder: We use numbers in sentences. In some cases one
might not know what the number is. Then it is possible to use a letter
that takes the place of the unknown number.

Eggs: Suppose you have a piece of paper. You draw a closed
figure using straight lines. Then there will be a part of the paper
surrounded by the lines of the figure.

Qipplg: Suppose you have a piece of paper. Then you draw a
closed figure using a compass. The distance from the point where the
sharp end of the compass is placed, to any point on the figure will be

equal.

L-L List

Cardinality: Jim has many pencils of different colors. The pen-
cils which have the same color are called a set. To describe this set
is to count the number of pencils in it.

Algorism: Bill is a bright student. The teacher asked him to
show his method in solving the division problem. He explained, step
by step, how he found the result of the operation.

Ellipse: Jim has a circle of wire. He changed the shape of the
circle by stretching it. The new shape is not a circle.

Trapezoid: An area has four sides. Two sides are parallel.

These two sides are not equal in length.

159

’ u ' iarization Material:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5124 I 47—4/6 I 9———>8

 

 

uwvsc’AAC/vxeéré—

i
__ i
‘5'+3=(2,I3',8,7)J 5+3:(.Z,I,338,7)l 5+3— 8 i

 

 

160

HW

 

 

 

 

 

 

 

 

 

Boa—a --'5' /00——> L:— --5’7
. 0

 

 

 

5'+3=5’ 5+ =8 5+X=8

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(|+|+|+|+|) (IHXHVHM) (5)
P.

 

 

 

 

\f

 

w

 

 

 

V

 

   
 

(2) (4)
‘B C.

 

 

 

 

APPENDICES

coeseeaaeaeasee oneness . a soasceeaaeflesee fiestas . > eeosecose Hoassoo . o

 

 

 

mm m ma 1. 1. ea m a gig
es me me a- a- s as > mug
em a- an mg m as m 0 Aug
an a e in nu m 0H m are
an a- a- we 1. s e > are
mm us .1 m 0H m as 0 aim
we as ea -- u- m d a m-s
3 m Q .. E S e s 3
en us in on n m ea c mum
we as a u- a- CH m a mum
mm s. in m m m es s mix
mm .1 u- w m as w o mum

 

caesom one: cameos can: season can:
fleece m Hoosom m Hoosom a Hoosom seesceeaeaaescm cues senescence

 

 

mpsospmose NH one. 5.. pounce Hoonom comm Scum mpoonbsm mo sobssz one

H gnaw/Ema:

162

 

 

 

 

 

 

 

 

 

 

 

 

e.ms o.ss~ m.ac m.wm m.wm, m.om, m.on‘ use
m.wm ~.esa m.em N.sm s.sn o.o~ n.e dim
o.em m.amm m.om m.mm a.em e.ss o.- m-e
n.mm e.mms m.we m.wm m.ao o.sw «.mm mum
«gowhuogoou
and {ET Lime. a see ND man as .3
ado Tenn m.wm ems v.8 v.8 56m mum
m.oc ~.eom m.as o.ms m.ss e.~m n.m~ mus
s.ms n.mem s.ae a.mw o.am m.aw e.me mum
F ”deceasesz
(«wqmmv m.wm m.wm s.mm N.ca m.m o.oo are
H.nm w.mma m.am e.ss e.mm n.m~ e.w 41m
m.as n. mm m.os m.os s.Hn m.om m.oa mag
~.es m.wmm m.ae m.am s.~m H.ss ~.es mum
"wcod
mfio Too“ flow 3.3 0.3. Name m.wm Ana
e.~e m.w«m e.ws s.Hs e.as o.oo e.am aim
m.ae m.omm N.mm N.mm e.es m.oe ~.nm mug
a.nm sinus o.em m.om H.mm n.ss s.«n mum
in», .11. unseen
m.was m.wflo. 0.3 new m.we Dem H. H .74
e.ss m.mmm n.mo s.em m.ee m.ss H.om ewe
o.mm H.omm n.ms s.me m.se m.ms m.wm m-e
s.om 8.60: o.se o.me m.am m.ms “.0: mum
I ”pecan—sou
w e m N H

noncommom uoonuoo
no m seem

 

noncommom poohnou
so a Hoses

 

defies ance pom momsoamom pootoo mo omepcogom

 

on: senescence

 

 

9:68.395. Homecoo can you define poem. hem noncommom poohoo mo omepsoosom

H. NHszmmd

163

MWWHK

Percentage of Correct Re5ponses Per Test Trial for the Verbal Familiarization

 

 

 

 

 

 

 

 

 

 

1&

 

 

U)

i
“3%

o NONM semNrmMN omam mmmfi
*0: 6°.03.U§O.U‘ ..U°\. deae
5457-313: mgtgwgnm: (\3 comm:
so

A
2A

o

o

0‘)

:53
mo
°e
R0 ONMW 0®®NBHQO Nam: meow

m '54; sass-"a “see 44';

H
'34-: mmaNr RfimxoflESRRN gosh-R} ocoéN
:3 ma N mq NdN NHM :HHN

A
E"A

o

u
a mMNm momeuaam mono dme
S d°é °dd$$§6§ 4mm &'&°
p 3 éNgw gmmmmmeo ammo mfio
0')

o
a
A
o (\O\OL\- mammeb-OV «Nan-0 #3503
m 3 use; a'nmsess $644 A’d‘
g mmmm o wwfimmw mwoq mgmfl
o
a
g omnm emeNNmmo wwwm HCQN

n 64$; gases A666 6&°&
3 mean $§$ wHwNfi mmwm mm 3
o
A
A
8

MHOCD coacxdkocomm :roxom H004?
% N 6K0 fifi$6fi°0N 'Ném $NJ°
e we: menu 3 men BMNg
m
a
+3
5 in
a mnew aeeqewmo mNmm mgpq
O I O 0 o O O 0 O O O O O

o a co mm on N H owH
m HNQ HNSMN ”a w a
g
s - 1
0 «mm mm Amm mm 0mm
w 2 :77 11T7910711177fl1177
p :mgmg mgmqqmgmgmqmgsmgmq
”5% 7 t ~° : a
m E o g E o
o o n o s o
O o m A z u

 

 

 

 

 

 

 

 

 

 

 

 

 

 

s on o.~m~ o.ms 7w.oml, m.wc A.os [wamm are

~.mm N.eAH s.mm m.mm ~.sm m.NA n.e ewe

m.ws s.esm «.As 0.0m m.am o.os m.m mun

N.mw m.wss o.sm o.em 0.3m o.wm ~.ec mum
"HwOflhfiofiomU

N.es m.omm «.mc m.om s.ms wo.mm e.w ewe

n.mm s.em~ o.ms e.me m.oo 5.0m m.wm are

m.me m.wam m.ws s.sn m.wm e.mm N.N~ mes

A.As e.mnm o.mm N.mm A.me m.ae m.oe mam
”HGOHAQSZ

m.om m.wsA s.mw, o.on mo.mm e.e e.~ awn

s.mm m.are e.me m.wm N.mm m.om e.m mam

N.mN H.wmA s.mm o.oe m.sA o.oH 0.: :14

e.es m.amm «.mm m.om m.wm m.aa ne.ss mum
"wson

m.om e.emm m.sw N.mw m.wa m.fis mo.mmi ere

~.en A.Ham N.es s.sm «.mm m.an s.mA ewe

A.em m.wmm 0.0m A.As m.wm e.mm e.em mug

s.mm m.er m.wm o.em o.sm o.ns o.me mum
”pAonm

_ m.sw m.www, weeds! c.mm A.mw, ,flnmn mawm are

m.as c.0om m.wm m.am m.me m.om e.sA are

A.ms m.mAm s.ns m.wm m.wm me.sm m.mA min

m.om m.oos A.Am mo.nm A.mm mm.ss o.mm mum
"UOQflDEOU

w! s m N A
noncommom pooAAoo noncommom pcoAAoo

mo m coo: mo a Hence HoaAe nmoe Aom noncommom nooAAoo mo omopnoOAom omhe soflpfiswmom

 

 

sownoeonaHHsom oAsvcfim onp Aom HowAe ance Aom noncommom pooAAoo mo omopsoOAom

A NHszmm<

165

 

 

 

 

 

 

 

 

 

 

 

 

166

o.m~ m.amm e.mm m.wm m.m~ w.m~ m.om ewe
o.eA o.os m.am s.mA m.NA A.a o.oo ewe
m.wm A.AmA A.Ns «.mm e.sm «.mA m.m man
o.oe o.omm m.am s.~m s.~w m.we m.am m1:
Le "Hootsoaoou
me a? 2. meal an an s; .3
o.sm o.os~ s.As m.so o.mm e.An m.em ewe
m.as m.aom m.am m.ws m.we s.em s.AA mug
m.mm m.woA m.mo N.me m.am m.ma m.w mom
.Huoaaossz
.1N.© m.mm m.m m.om m.m, o.oo o.oo man
e.sa A.sm s.nm m.wm A.AA o.oA m.A 41m
m.om m.wmA m.wmA m.ws m.om m.AA m.A mun
m.~e m.nAm o.me 0.0m m.we m.am c.~H mum
"ween
o.m~ n.fl~¢ e.mm1| m.wm m.om m.wn o.mm ewe
e.om m.wmm o.os A.sm m.am m.ws m.wm ewe
~.mm A.ce~ m.mw s.em m.wm H.se H.0A mug
o.Ao o.mom s.mm m.ws s.me 0.0m A.e~ mum
tar. - in. ”steam
m.ea m.ow m.~A w.em e.wA «.mA o.AH awe
o.sm A.osA m.wn o.oe e.mm m.mm m.wfi ewe
m.am m.osA s.ce m.wn m.om m.am m.m mug
o.mm o.oe~ m.ws m.we m.oo m.ws s.mA mum
xii, "Ramsey
m e m m A
noncommom nooxm noncommom noexfi
mo m coo: Ho & Hence HowAe pmoe Aom monsomoom pcoxm mo omousoOAom omha soauflnamon

 

 

 

 

 

psoepoer HOApsou one Aom HoHAe pooh Aom noncommom pooxm Ho omopsovom

z xHDzmmmn

EWWHN

Percentage of Exact Responses Per Test Trial for the Verbal Familiarization

 

 

 

 

 

 

 

 

 

1W

 

 

 

(a

i
“as.

m 050““ swoon O\®L\O ON-‘JO MONO
“8 65°& 6&'& seem d8°6 meme
gr: Arte-43¢“ \OHSW m N NHSKC'V \OHNM
‘32

0

>

m

0

m
‘H:
°:.
$10) H03“ [\wa \ooomoo (\mom \O\O<'\H

é 684$ dads "ae °'&' J’Jé
q Nl\O\O OQWH‘ $373.41 g8® J H H
43:: N NH! M NM H H H NH M HH
we.

0

>
‘3 HSCDN (DO\O\N Now!“ 53:30 413““
of! eeee eeeo eeee eeee .900
$4 W NBRE "\NH H\0H H33 NOM\O
[-1 \ON BMBCD Nd‘H MN“ [\Md'd'
+3
to
O
E-'

A amen emmm mend MNNW dNHM

° s ‘6d4 meme secs 465° 'JA'

‘2; RNW31 (\mmoo de' d'NNg fiNd’j

0

to

£2

8.

:8 WN‘nd’ N®\O\O CDWON WWBN “(DA-2'0
0 O O I O O I I O O O O O O O 0 O O O O

a: m ammo mm H UN H L\ max

1 Mann ma J mwgi gwwg SHH

'2

Q

'H N0\NN 035W din-1&0 MOHO O\L\L\Vl
O O O 0 O 0 0 O O 0 0 0 O 0 O O 0 I 0 O

O N \00\ N OM N\0(\O mom O\<"\O\

o en fin the-03$ N HO vie-4mg R

to

3

5

o mmmw NHMN mNNo homo MdON

h coco eoeo eeee eeec eeee

o HI rims-Mn 0\OO\O \oe-umo anNo HNOO

p. m Hal-1H o N :1- CH“

g;

g mmng mmnn mm q"mmqqgmmq

or! llll llll uu’Te'deaaufiuan'l'

a mqmg mnmg mnmggmnmqgmqmq

4.) co ‘4

a a : :9 3

o o n o 8

Q o 0‘) 1—1 C!

MMWHO

Percentage of Exact Responses Per Test Trial fer the Picture Familiarization

 

 

 

 

 

 

 

 

 

1%

 

 

 

m
o
It)
:5
fig NNNN :owo ommN HQm: mmmq
00.0 .00. 0.0 .0.
0 Now H no mm gm m meH
5m wNN n :0 $3 N3N NNH:
£3
3
U)
cHa)
ca
Ra sass “c9“ QNss wwwN ssss
MHHH o ONON mm- H O\
1§ Hm: ngg BBO? MNfid m m3
fig MHHH MHNN N NHNH m N
90
3
q NQnH mmmm wowfl omaw NQNM
O... 9... 000° 0...
mum H momH N NM 6:
E W has $R0N FMNN m we maN «fl
.p
0)
59.
a dHNQ mwow OMdm Hon 00mm
00°. 00.0 0.00
o d mon ammN mm Hi Hm:
m fig: msnfi WNHfl mi; OMH
m
0
U)
S
% BGON anw mmmm Hoow Hwow
00.0 000 0000 0.9.
o 0mm m m Hoom NBQ
m n wNN BR? win mNMN QN 3
.p
o
5
m MQQQ HdNH QQQQ mfiqw Qmmm
O .00 O o O
o N mmn m mm ammo 0 HH mmmn
o WHNg \ofizrw :r o (“$de {\N m
w
:3
s
° A
2 H “ass 9%?% mono see. ness
0 H m 0 NW oomo H NH omw
m mamm mmHN No C HmH fiRo N
o
E
g mmqg mmqq mmqg"mm ggmmgq
H Ulll! llll outlinnfianunu
p gmqmq mqmq mgmggzgmgsmgmg
E H fi ~ 3 2
M g g g g o
o o o o
a o m g z w

 

 

 

 

 

 

 

 

 

on. mm. «o. mm. mm. sow macs
ofim. mfi.mm am. HH. Ho. mo. om. new pgogm npmqoq
Ha. mH. Ho. 5H. mm. sow .ooo
mmo.n mn.H $3. NN. No. mH° cw. dwN cadz Magus: poonnsm
oz. mH. Ho. 5H. mm. mum A
530.- oa.H es. mm. No. ma. ow. mmm m a omcoamom
mo. Ho. 00. Ho. mo. mom 4
33.: EA: an. 0:. no. mm. 8. 03 m s 3953
NH. do. oo. Ho. mm. ems Aug
Hm. 3m. «0. Hmo no. 03H A1:
00. oo. oo. oo. co.H on“ mag
amm.n No.5NH mm. we. so. 0:. on. mHH mum «soapficflmoo mo «came
a Nx .>oa .pm can: N \H o
COHHsQanmHQ onoom amoenmnm z Hm>mq oHanu¢> sOHpHsHmoQ

 

 

 

 

 

psmspmoue Houpzoo mo ammo CH moHDaHpm> GOHHHCHmoQ nHHz

J gamma:

menoom Humaneum no COHpanupmHQ Hd20HHnomoum

169

 

 

 

 

 

 

 

 

 

on. Hu.H ow. co. sH. m N
no. m:.H on. mm. oH. mm H

HONo mw.:N aw. wo.H 0:. mN. Hm. aN: o whoom pmoetopm
mm. mo.H mm. Hm. on. sou msoq

amo.- m~.m mm. 3N.H we. mm. mm“ sow phonm spmcoq
mm. HH.H ms. am. “no sow oooo

mno.u mm.n ow. HN.H ma. Hm. 3N» :mN oE52 poupqz voonnsm
om. mo.H mm. emu wN. mew A

amo.- mo.:H mm. am.H Hm. mme mm. mam m s uncommom
am. mo.H mm. on. mm. mom 4

omHoa mm.NH ow. mN.H Hm. mm. Nm. owm m a msHseflpm
Hm. mo.H mm. mm. ow. mmH Hug
me. HH.H an. on. em. 03H gum
ow. oo.H um. em. mm. and mag

mmH.- mm.Hs nu. mm.H mo. 0H. 0H. mad mum macapficamoo mo masks

a Nx .>oa .pm cue: N H o
soapsnfinpmfia ouoom pmosupmom z H¢>oq casuan¢> sofipflcfimon

 

 

 

 

 

 

pzmspaoue Hoppsoo mo ommu :H moHnaHua> sOHpHsHmcn nqu moaoow pmoelpmom Ho COHanHuHmHQ HdcowahomOhm

mm xHozmmm<

170

 

 

 

 

 

 

 

 

 

mm. 0N. co. 3H. ow. omN macs

Nmo. om.o Na. NH. so. No. mm. omN spasm gumqoq
omw HN. so. NHO mm. CNN 008

woo.- NH.H mm. HN. mo. oH. am. omN .352 house: poonnsm
mm. nN. no. nH. om. con H .

muo. mm.m as. NH. no. No. mm. omN m a ougomuom
0N. Ho. Ho. Ho. mm. owN H

mom.» Nm.mo No. mm. mo. ON. NN. ONm m s mngaHHm
Nm. mo. Ne. Ho. Na. wNH HTH
am. Nm. mo. mm. mm. :NH 41m
oo. oo. oo. oo. oo.H 30H mug

HNH.- mm.m:H 0N. Na. NH. NH. 0N. mHH mum mquHHcHHoa Ho momma

u NN .pca .»m can: N Hwy o

:OHHanupmHa ouoom Hmoauoum z Hu>oH oHnaHnu> :oHpHcHHoa

 

 

 

 

 

>|lul

:oHaaaHgaHHHaaH Hunno> Ho amao pH mngqua> :oHHHcHHoo nHHz.mouoom Hmoeuoum Ho coHpanHuHmHo Hu:0HpuoaouN

Ho xHozmmm<

171

 

 

 

 

 

 

 

 
 

 

 

HN. HN.H ON. OO. OH. ON N
8. 93 mm. mm. OO. 8 H

NOH. NN.On ON. OH.H ma. HN. mm. mm: O anoom Omoauoam
no. OO.H OO. NN. mm. OON Ncoq

mNH.- OO.OH NO. ON.H um. OH. ON. OON spasm newcoq
ON. OH.H OO. NN. Om. OON .ooO

HOO.- Na.m ON. ON.H Nm. ON. NN. OON .asz Nous“: Hoonnam
ON. ON.H ms. NN. ON. OON H

mno. ON.H mm. OH.H NO. ON. mm. OON m a uncommom
NO. HN.H Om. ON. ON. OON H

m8. HHO.H NO. mH.H 3. N. Nm. ONm m a 333
NN. mN.H Nu. ON. ON. ONH HTH
mm. OH.H «O. ON. ON. ONH me
ON. HH.H 0:. ON. mm. OOH muH

HNO. Om.m NO. OH.H Os. HN. ON. nHH mam unoHpHcHNon Ho mommy

.H NN .25 .5 O: N i H... O
OOHNOOHNNOHO ouoom Hmoauumom z Ho>aH oHOaHna>.:OHHH:HNoO

 

 

 

 

 

OOHNONHNOHHHaam Hanuo> Ho omuo :H moHanum>.OOHHHcHHcO OOH: menoom smoenpmom Ho OOHOODHNpmHm HN:OHHNOOONN

NO NHOszNO

172

 

 

 

 

 

 

 

 

 

Om . mm . OO . NN. NN. in 33

RH . OH .OH 3. NH. 8. HH. mm. in 28% .333
mm . 0N. mo. NH . NN. in 60$

moo. Hm. an . mN. no. m H . om. in .52 .834: poonndm
Ns. 0N. no. 3. mm. omN .H

mmof NN.“ mm... on. No. NH. 0N. mam m a uncommon
m N. 3. Ho. No. No. mmm .H

mnd... NH .NaH 3. Hm . mo . mm. mm . NaN m a 33pm
Nm. mo. mo. +3. 3. 3H .74
3m . mm. no. mm. 8 . :NH Aim
8. co. co . oo. 00 .H owH and

NN.: HN .moH HN. S. :H . on . Hm . 3H mum 303393 .Ho was

h NM $00 5m 502 N H o

soflHfiHhamHn 0.30m ance-cum z Hosea oHnuth> QOHHHHHHHOQ

 

 

 

 

 

5332.32; 8:33 .Ho 38 SH 3335 zofiflflon £3: 8.88 $4.8m NO 833233 aofluafim

Hm OHEEEO

173

 

 

 

 

 

 

 

 

 

 

174

on. :m.« Na. 00. m0. Hm N
am. mm.H mm. mm. mo. HoH H

NQH. oH.H: am. mm.H mm. wH. aw. mm: o ouoom pmoauoum
3. R . H mm . om. Hm . in was

one? mo.~ R. 34 am. 8. 8. in team 59:5
mm. mm.H mm. 0H. MN. de .000

NNo.u NN.n 0N. N:.H so. mH. NH. :Hn .352 hoped: poonnsm
Nm. Nm.H mm. NN. MN. OQN A

moH.n Na.m NN. m:.H we. NH. NH. mén m B uncommom
No. od.H Ho. NH. NN. 0mm A

mNo.u n:.m NN. ¢:.H Ho. NN. NH. NmN m .E maadsHpm
Nw. OM.H ow. 0H. HN. omH AIA
mm. MN.H m3. NN. mN. aNH 41m
mm. 0:.H NW. 0H. NN. omH min

on.I m¢.mH mo. ow.H HN. wH. HH. wwH mam mGOHpHGHHOQ Ho momma

.— NH $8 :6 5o: N J. o
533$...an 28m 0.89-38 2 H33 32.3; 83383

 

 

 

 

 

5332.33; 8393 mo 9.26 :H mchaflé 83228 .3: «88m 9.39-38 mo 83.5.33 153893

Nm figmfi

 

 

 

 

 

 

 

 

2.. mm. 8. 3. R. é 33
men. 3.8 No. R. 8. N. 2.. am £28 £33
HN. no. NH. NN. om. aw .ooo
ooo. NNH. NN. mm. :H. mm. Hm. :0 .552 .2 .nam 09:30Hm
Hm. No. HN. HN. mm. mm Ncoq
3N. No.3 3. HN. 8. i. mm. mm 0.8.3 .393
“W. 30 mo. MN. 80 a 000w
0No.l no.m “N. m:. 0H. NH. NN. mm .852 «z .Qam Hdnho>
on. R. 8. 8. Nm. mm 33
com . 36m 3. NH. 8. NH. 8. mm .28... .3qu
R. #30 “C. an. N00 mm 000.0
03.. i; 8. mm. no. 3. an. mm .52 .z 5:... H9350
u mm? :3 :6 5o: N .HW o
COHvanApmHQ ouoom pmoaloum z Hp>oq oHnuHAa> :oaawanuHHHsdm

 

 

 

 

 

 

mucoapaoue cOpruw

£333..” 298:8 3?: 333.2; 183238 5H2 mum No 3.83 issuer” mo 833233 infiuomzm

Hm x522“:

175

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

NH. OO. OO. mN N
1mm. Om. NO. HO. mm H
N. .OH N . OH. OH. N O Hume-oam
mm no. NO.H NN. NH. OH. NO \mmmH
ONO. NN. NN. mm.H mm» NH. NH. mm Huogm camcoq
mo. NO.H Om. NO. NH. Ow .ooo
NNH. N0.0H NO. Hn.H NO. NN. HH. so .asz .2 .nsm oNOHOHN
OH. OO. HO. NH N
NH. NO. OO. ON H
NON. NO.NN NN. OH. NN. Hm O Omoe.opm
Om. nn.H mm. ON. HN. mm NumH
NnH. ONN.N NN. NO.H ma. MH. Om. mm Ononm Opmcaq
NN. mm.H OO. NH. NN. .ooO
NmH. Nm.N mm. moH mm. NN. mm. mm .52 .2 .85 33>
Ow. OO. ON. m N
ON. ON. OO. O: H
NON. OH.mH mm. NO. NH. mm\ O Huaen¢um
Nw. NO.H 3N. HN. wO. mm mmoH
NHN. NN.N NN. 1mm.H mm. NH. ON. mm Ogogm numcoH
ON. RH 1NN. OH. NH. fl 5%
mmO. NH.m NN. Om.H OO. NN. NH. mm .asz .2 .25m Houpcoo
p NH .>on .Om cums N H O
COHagflhvwg 200m Pwofilfimonm z H0>01H QHQHHAQNH CowvaHMMflHflfidh

 

 

 

 

 

 

3:053er noavusd
-NNHHHENN HcouommHO Nouns moHnaHga> HN:OHHHOHHOO OOH: mum mo monoom Once-pmom Ho :OHOODHOHWHO HNOOHONONONm

NNNNENQE

176

 

 

 

 

 

 

 

 

 

 

OO. OO. OO. OO. OO.H ON 2202
OOO. OO. OO. OO. OO. OO. OO.H ON ONOOO 2pwco2
OO. OO. OO. OO. OO.H ON .ooO
OOO. OO. OO. OO. OO. OO. OO.H ON .222 .2 .OOO onspoHN
OO. OO. OO. OO. OO.H NO mcoH
OOO. OO. OO. OO. OO. OO. OO.H NO ONOOO Opmcog
OO. OO. OO. OO. OO.H NO .ooO
OOO. OO. OO. OO. OO. OO. OO.H NO .222 .2 .OOO Hanuo>
OO. OO. OO. OO. OO.H OO N202
OOO. OO. OO. OO. OO. OO. OO.H OO ONOOO OONcoH
OO. OO. OO. OO. OO.H OO .ocO
OOO. OO. OO. OO. OO. OO. OO.H OO .252 .2 .OOO HouHcoo
a N2 .OOO .pm 2.92 N HI O
cOHHOOHNHmHO ogoom OO.H-ouN z Ho>oH oHOtuu> :OHO.NHN.HHH2.2

 

 

 

 

 

 

unsuspwohe cowpusd

-NNHHHENN HcouommHO have: moHO¢H2w> OOHHHcHHOO 22H: 2-2 Ho mogoom Hmoenonm Ho OOHHOOHHHOHO HN20HHuomogm

H

.H. xHszmnE

177

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

OO. OO. OO. OOO N
OO. OO. OO. OOO H
NO. OH. NN. OOH O Omoeuoum
HO. OH. ON. ON mmoH
NO. Ofi N? ON whim .3952
NO. OH. OO. om .ooo
mHN.u mn.m MN. mm.H NN. MH. 3H. om .Efiz .2 .95m 0h5¢0Hm
OO. OO. OO. OOO N
OO. OO. mm. OOO H
OOO. OO. \OO. .lmmw . an O pmoanoum
ON. ON. On. ON. OH. NO 02
ONH.- mmum OO. NN.H mm. Nmm. NN. NO ONOOO OHOcoq
.| NO. OO.H O. HN. NO. NO .ooc
OOO.- NO. ON. OH.H NO. OH. NO. NO .2:2 .2 .OOO HOngo>
OO. OO. OO. OOO N
OO. OO. OO. OOO H
OOO. OO. NO. ON. NO. OOH O Hmoeuopm
NN. MN. Om. OH. O2. OO oH
OOO.- OO.O OO. NO.H Om. NO. ON. OO Ogmmmw spmcoq
.wa. NN. On. ON. Ow. OO .oOO
ONO.- OOH. OO. OO.H ON. ON. ON. OO .252 .2 .OOO Houpcou
a N2 .>oO .OO mmmm N H O
:OHHOOHuHmHO oncom Omoenpmom z Ha>o2 OHONHNO> :OHOONHNOHHHEOO

 

 

 

 

 

 

mwccspuohs :odeNH

.uuHHHEO 2220.33 .895 $332; 83238 .32 21H .3 8.88 0.22.38 .3 OOHOBHSmHH. HEOHOSNOE

Ne 2HO2ONN<

178

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

NO . .3. NO. 3. ON . NO O23
OOH . NON Om . ON. n O. OH . NN. NO £82m 2.352
NO. ON. NO. NO. OO. NO .25
m 3 . Nm. H NW . an . no. 3N . HN . NO .282 .2 .nHHm ohH—HHOHHHH
HO. HO. OO. 3. 3. NO 33
Hm H . No .m NO . N3. 3 . MN . 8 . NO chasm :pmcoq
an 0 *3. “C 0 3n 0 H0 0 NW 0 OOO
ONH.- HN.N NO. OO. HH. NO. NO . NO .52 .2 .225 H3O;
om . ‘3. 8. +3. on . oN mcoq
ONH . NO.NH NO. ON. NO. NH . NN . ON .225 533
S.-. R. . OO. R. NO. ON .08
ONO: NO.N Om . ON. NO. ON . NO . ON .52 .2 .OOO H828
.H NH :3 .OO 332 N H O
533238 28m $02-2N 2 H252 .332; 53.3532;
mannaadga

5332335 OcaohHa .SHEO 33.2; 5353 HO .HO 8.88 $3.63 .HO 533233 HEOHfiONouN

HO NHOZONNO

179

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

NO. OO. OO. O N
OH. OH. NO. mm H
mmmw NO.OH Nm. NH. ON. OO O Hmoeuoum
ON. HMNH. mm. mm. mH. NO mmoq
OOO. HO.N N . OH.H NW. N. Hm. NO Huogm Opmcoq
mm. OO.H . Nm. Hm. NO .oou
OON.- mm.O OO. OO.H OO. HN. OH. NO .552 .2 .OOO oNOHOHm
NO. OO. HO. OH N
OH. OH. OO. O: H
mmN. Om.NN ON. OH° mm“. mmr. O Hmoeuoum
ON. NHNH OO. mm. HN. NO mmoq
ONO. NO.HH mm. NH.H OO. OH. NO. NO Hummm. OHNOOH
OO. Om» N. On. NO. NO .OOO
OOm.- N0.0H NO. NO.H OO. OH. HN. NO .252 .2 .OOO Hanum>
HO. OO. OO. N N
HH . OH . OO . 3 H
ONH. OMHO . NN. HN. am O HmoN:¢nm
mN . OO. H mu. Mm. ON . ON a
mmm»- NO.N OO. NH.H ON» ON. ON. ON Hhosm npwnoq
HQ. g 0 CM. .3“ o W 0 ON 0 00.0
OON.- N0.0 ON. NN.H OO. mm. NH. ON .552 .2 .OOO Hoppcoo
a Na .>oa .Om Odes N H O
Gadpgflhpma whoom amoelamom z Ho>oA OHD¢HHQ> Co mpaflfihdflfifiah
mucoapuaa

OOHHONHNOHHHEON HcopONHHO Nouns OOHnuHua> OOHHHOHNOQ Own mo mouoom Hmoa.pmom No :OHHOOHNHOHO HOOOHOAOQONN

NO xHOszNN

180

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

OH. NO. OO. NO. NN. ON Ozog
ONH.- HN.O NO. OH. NO. NO. ON. ON OAOOO Onwzoq
NO. NO. NO. ON. ON. ON .ooO
OOO. OO. NO. NO. NO. ON. ON. ON .auz .2 .OOO onsOOHN
OO. OO. OO. OO. OO.H OO NOON
HOH.- HN.O OO. HH. OO. NO. NN. OO ONOOO OOOOoH
OO. HH. no. No. nN. OO .25
HOH. HN.O OO. OO. OO. OO. OO.H OO .agz .2 .OOO HOONOO
NH. HO. OO. HO. NN. OO Ocoq
OOO. OO. NH. HO. OO. HO. NN. OO OOOOO OOOOoH
NH. NO. OO. NO. NN. OO .OOO
NNH. NO.N OO. OO. OO. OO. OO.H OO .asz .2..O:O HONOOOO
N Illa O
Gadfigfifimg zoom amoelgm z Hobbfl 03ddhd> Gowvduflhdfiqfidh
mucoSpOOHH

SHOSHQHHEON Okayama .816 333.2; 83238 21H .Ho 3.33 Ommauoé .Ho OOHOBNSOHO HSOENOHHBN

H> NHOzONNO

181

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

OO. OO. OO. O N
mm. MO. OO. O H
HOH. OmwO . H. HN. OOH O Once-oum
NO. WH.H OO. MN. Hm. ON flog
NNN.- O0.0H Om. OOOH mm“. OH. NH. ON ONOOO OOOOOO
OO. HO.H OO. HN. NH. .ON .Nmm
ONO. mm. Ow. Nmé ow. NH. MN. Q. .552 .2 .QHHm. 233m
NO. OO° HO. O N
HO. OO. OO. N H
NNO. NH.O NO. pmw .mm. ONH O Onoeuoum
OO. ON. NO. ON. NO. OO Omoq
OOO.- .ON . NO.H NN. OH. OH. mm ONOOO OOOcoq
MO. OH.H NO. ON. OO. OO .ooO
NHH.- ON.N OO. OO.H NO. NH. ON. OO .252 .2 .OOO Hanuo>
OO. OO. OO. O N
HO. OO. HO. N H
OHO.- OH.H Om. OO. NN. OmH O Ouoa-ogm
0N. . cm. 0.“ . wq m: . mm :04
ONO..- mO.NH ON. Nm.H OO. Hm. k3. OO 226 533
MO. OO.H NM. NM. Hm. OO 63
ooo.: Hm . ow. Noé mm. NM . wN. mm .852 .2 .95 Honpcoo
2 N2 .>OO .OO 232 N H O
COHuBflhflma choom nmmfilvwom Z Hw>oq afifldfih¢> SOHHdNthHHHEdm
3:92?er

OOHOONHONHHHEOO OOOOONNHO Nouns OOHOOHNO> OOHOHOHOOO qu No mouoom Omoeupmom No OOHOOOHNOOHO HOOOHONONONN

N> NHOOONNO

182

APPENDIX W

The Letter Sent to the Schools That Participated in This Study

 

December 23, 1965

Dear :

The purpose of this letter is to express my appreciation for the fine
cooperation that I received from you and your staff while conducting
my doctoral dissertation research at your school. Without such coop-
eration, such a study would not have been possible.

I would like to be in a position to discuss my findings with.you and
your staff. However, it is necessary for me to return at this time
to the United Arab Republic. My major professor, Dr. Clessen Martin,
has kindly informed me of his willingness to discuss the study with
any interested persons. If there is any interest in this research,
Dr. Martin may be contacted at Michigan.State University.

May I again thank you, your staff, and the students for their coop-
eration.

Sincerely,

Salah A. Hotar

31

Red Cedar Elementary School, East Lansing
Baily Elementary School, East Lansing

Dwight Rich Junior High School, East Lansing
Waverly Junior High School, Lansing

- Mason Junior High School, Mason

- Holt Junior High School, Holt

- Springfield Junior High School, Battle Creek

QWFJUOUIP
l

183