Citation
Concurrent academic predictors of spelling performance of third grade children

Material Information

Title:
Concurrent academic predictors of spelling performance of third grade children
Creator:
Lessen, Elliott I., 1947- ( Dissertant )
Wolking, William D. ( Thesis advisor )
Abbott, Thomas B. ( Reviewer )
Forgnone, Charles J. ( Reviewer )
Mercer, Cecil D. ( Reviewer )
Satz, Paul ( Reviewer )
Ware, William B. ( Reviewer )
Place of Publication:
Gainesville, Fla.
Publisher:
University of Florida
Publication Date:
Copyright Date:
1977
Language:
English
Physical Description:
ix, 94 leaves ; 28 cm.

Subjects

Subjects / Keywords:
Academic learning ( jstor )
Consonants ( jstor )
Educational research ( jstor )
Learning ( jstor )
Mathematical variables ( jstor )
Mathematics ( jstor )
Nonsense ( jstor )
Performance metrics ( jstor )
Standard deviation ( jstor )
Vowels ( jstor )
Dissertations, Academic -- Special Education -- UF
English language -- Orthography and spelling ( lcsh )
English language -- Study and teaching (Elementary) ( lcsh )
Prediction of scholastic success ( lcsh )
Special Education thesis Ph. D
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Abstract:
The focus of the present study was to identify a set of prerequisite tasks, modality relevant to spelling performance, that when based on accuracy and speed as performance measures, would discriminate between good spellers and poor spellers. A review of related literature suggested that (a) tasks used to date have represented more than one basic behavior per task; (b) tasks that have been used are related to, but are not direct replications of the modality input and output structure of spelling behavior; (c) tasks used have not represented the most basic skills assumed by curriculum hierarchies to be prerequisite to spelling competence; and (d) tasks have been assessed by accuracy alone, without referring to speed (fluency) as a viable performance measure. A spelling test was administered to a group of third graders in Alachua County, Florida. From these results, the top and bottom 25.9 percent (good and poor spellers) were chosen (N=35 per group). Each of these subjects was then assessed on each of eight tasks on three consecutive days. Accuracy and speed scores were obtained for each subject on each task. Discriminant analysis and multiple regression procedures were used to analyze the data. The results of the study indicated that two of the eight tasks used, See CVC trigram/Say nonsense word and Hear two letter blends and digraphs/Write letters, were the best tasks with which to discriminate groups of good and poor third grade spellers. Speed and accuracy scores on the two tasks were found to be nearly equivalent discriminators between the two groups of spellers. The findings of this study indicated that the use of speed and accuracy as performance measures may serve two useful functions. First, speed and accuracy scores discriminate between good and poor spellers. Second, speed and accuracy scores reduce possible misclassifications that may interfere with a child's curricular program. The overall and additional findings have suggested possible alternative research efforts that may help solve the problem of presenting classroom instruction that could promote better spelling achievement.
Thesis:
Thesis--University of Florida.
Bibliography:
Bibliography: leaves 55-58.
General Note:
Typescript.
Statement of Responsibility:
by Elliott I. Lessen.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
025839081 ( AlephBibNum )
03381810 ( OCLC )
AAV3486 ( NOTIS )

Downloads

This item has the following downloads:


Full Text












CONCURRENT ACADEMIC PREDICTORS OF SPELLING
PERFORMANCE OF THIRD GRADE CHILDREN












By

ELLIOTT I. LESSEN


A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY



UNIVERSITY OF FLORIDA
































To

Linda















ACKNOWLEDGMENTS


Sincere appreciation and gratitude are extended to my

committee members: Dr. William D. Wolking, Chairperson, and

Drs. Thomas B. Abbott, Charles Forgnone, Cecil D. Mercer, Paul

Satz, and William B. Ware. These gentlemen have extended their

personal and professional efforts toward my development as

their graduate student. Ms. Leila Cantara, as typist and

friend, has been an invaluable asset during my doctoral pursuit.

I would like to thank Ms. Judith Sazer-Kuperstein for her

foresight and encouragement in this endeavor. Ms. Karen Andrews

and Ms. Susan Welch receive my thanks for their assistance in

the data collection process. The support, encouragement, help,

and friendship of David Westling, Mark Koorland, Terry Rose, and

Jim Altman can never be repaid.

My family and friends deserve thanks for their support and

confidence in me. My wife, Linda, above all others, has shared

my ups and downs and tolerated my temperament. It is to her that

I dedicate this dissertation.














TABLE OF CONTENTS


ACKNOWLEDGMENTS .....


LIST OF TABLES .......

ABSTRACT ....

CHAPTER I

INTRODUCTION
Purpose of the Study ..
Justification of the Study
Limitations
Threats to Internal Validity
Threats to External Validity
Definitions of Terms ...

CHAPTER II


REVIEW OF RELATED LITERATURE ...
Introduction . . .
Traditional Variables ...
Parameters of the Present Study . .
Learning Hierarchies
Definition and Characteristics .
Origins of Learning Hierarchies .
Validation of Learning Hierarchies
Task Selection in the Present Study
Task Selection and Hierarchy Validation
Related Academic Predictor Variables
Reading . . .
Phonics . . .
Summary ...
Speed of Manuscript Handwriting .
Spelling Test Scoring Procedure . .
Scoring Units . . .
Whole Word .. ..
Syllable
Sound Cluster ... .
Letter . . .


. . . . . vii










Performance Descriptors .... ..... .. . .... 16
Accuracy . . . . . . . . . . 16
Speed ........ . . . . . . . 16
Repeated Measures of Predictor Tasks .... ....... 17
Summary .. ..... ................ 18

CHAPTER III

PROCEDURES ................... ...... 19
Specific Questions ...... . . . .. . . 19
Question One . . . . . . . .... .. ... . 19
Question Two . . . . . . . . . 19
Question Three .. .. .. .. .... .. .... . 20
Subjects ........ ................ 20
Instrumentation . . . . . . ...... . .. 22
Method . .. . . . . . ..... 25
Administration of the Spelling Test . . . .... 25
Scoring of the Spelling Test . . . . . . 26
Administration of the Predictor Tasks . . . . .. 26
Scoring the Predictor Tasks . . . . . . 28
Data Analysis ................. ..... 28
Additional Analyses ............... ..... 29

CHAPTER IV

RESULTS . .. .. .. . . . .... .32
Question One . ... ......... .. ... .... 32
Question Two .......... ....... ... . 33
Question Three ....................... 33
Additional Findings . . . . . . . . . . 34
Contribution of Predictor Variables in Accounting
for Variance in the Dependent Measure . ... ..... 34
Initial and Predicted Group Assignment .. . ... 35
Analysis of Single Component, Auditory Input Tasks . . 37
Accuracy and Speed Change .... . . . 39

CHAPTER V

DISCUSSION . . . . . . . .. 40
Task Complexity and Learning Hierarchies . . . . .. 41
Input and Output Modality Structure of the Task . .. 43
Spelling Test Scoring Procedure . . . . . .... 44
Performance Measures . ............... . 44
Repeated Measures of the Tasks . . . . . . 49
Conclusions .. . . . . ... .. .. 51
Recommendations for Future Research .. . . . . 52

REFERENCES . . ............. . .... 55










APPENDIX

A SPELLING TEST . . . . . . . . . 60

B ACADEMIC TASKS . . ... . ..... 62

C SEE CVC TRIGRAM/SAY NONSENSE WORD STIMULUS SHEET 66

D SEE ONE PLACE ADDITION FACT/SAY ANSWER .. ... 68

E GROUP MEANS AND STANDARD DEVIATIONS .. ... 70

F RAW SCORE DATA .... . . . . ..... 72

G SCATTER PLOTS . .. . . . . . . 76

BIOGRAPHICAL SKETCH .... .. . . . . . 93














LIST OF TABLES


Sex Ratio Per Group . . .

School Breakdown Per Group . . . . .

Parents' Index of Social Position Per Group

Predictor Variable Contributions . . .

Accuracy and Speed as Performance Measures .

Accuracy as a Performance Measure . . .

Speed as a Performance Measure . . . .

The Relationship of Single Component, Auditory
Tasks to Spelling Performance ...

Mean Change Per Task Per Group ...

Proficiency Levels Per Task ....

Standard Deviations on the CVC, MATH, BLDI, and
ALPH Tasks . . . .


. . 21

. . 21

22

. . 34

. . 35

. . 36

. . 36


Table

1

2

3

4

5

6

7

8


9

10

11


Input
















Abstract of Dissertation Presented to the Graduate Council of
the University of Florida in Partial Fulfillment of
the Requirements for the Degree of Doctor of Philosophy



CONCURRENT ACADEMIC PREDICTORS OF SPELLING PERFORMANCE
OF THIRD GRADE CHILDREN

By

Elliott I. Lessen

December, 1976

Chairperson: William D. Wolking
Major Department: Special Education

The focus of the present study was to identify a set of pre-

requisite tasks, modality relevant to spelling performance, that

when based on accuracy and speed as performance measures, would

discriminate between good spellers and poor spellers. A review

of related literature suggested that (a) tasks used to date have

represented more than one basic behavior per task; (b) tasks that

have been used are related to, but are not direct replications of

the modality input and output structure of spelling behavior; (c)

tasks used have not represented the most basic skills assumed by

curriculum hierarchies to be prerequisite to spelling competence;

and (d) tasks have been assessed by accuracy alone, without re-

ferring to speed (fluency) as a viable performance measure.

A spelling test was administered to a group of third graders

in Alachua County, Florida. From these results, the top and bottom










25.9 percent (good and poor spellers) were chosen (N=35 per group).

Each of these subjects was then assessed on each of eight tasks on

three consecutive days. Accuracy and speed scores were obtained

for each subject on each task. Discriminant analysis and multiple

regression procedures were used to analyze the data.

The results of the study indicated that two of the eight

tasks used, See CVC trigram/Say nonsense word and Hear two letter

blends and digraphs/Write letters, were the best tasks with which

to discriminate groups of good and poor third grade spellers.

Speed and accuracy scores on the two tasks were found to be nearly

equivalent discriminators between the two groups of spellers. The

findings of this study indicated that the use of speed and accuracy

as performance measures may serve two useful functions. First,

speed and accuracy scores discriminate between good and poor

spellers. Second, speed and accuracy scores reduce possible

misclassifications that may interfere with a child's curricular

program. The overall and additional findings have suggested

possible alternative research efforts that may help solve the

problem of presenting classroom instruction that could promote

better spelling achievement.













CHAPTER I

INTRODUCTION



Spelling is one criterion by which school achievement is

measured. Allen and Ager (1965) state that "spelling is an

independent skill and should receive specific attention . .

in the school curriculum" (p. 159). There are many charac-

teristics that contribute to spelling performance in children,

e.g., intelligence, visual perception. There may be certain

prerequisite academic skills, that if known early, could serve

to discriminate between good and poor spellers. Such a discrim-

ination could also be helpful in planning remedial and develop-

mental curricular programs.

The concern for correctness in spelling and effective in-

structional methods is not new. Cook and O'Shea (1914) stated,

"There is . a wide-spread belief that graduates of the

elementary schools can not spell so well now as they did in

earlier times." T. D. Horn (1969) related even earlier reports

of such dissatisfaction from the thirteenth century. Many

teachers are aware of children who are poor spellers but the

reason for their spelling difficulties remains a concern. Al-

though educators are cognizant of correlates related to poor spel-

ling (Russell, 1937; Spache, 1940a; 1940b; 1941a; 1941b) they may










be unsure of those specific behavioral elements functionally

related to spelling.

To date, the research on spelling suffers from numerous

limitations. Among these are (a) the lack of replicable pro-

cedures (T. D. Horn, 1967); and (b) the use of tasks that do

not directly replicate spelling behavior (Westermann, 1971).

An example of the latter is the use of tasks to assess spelling

which have generally been used to assess reading. The charac-

teristic behaviors specific to spelling and reading are

different. Whereas in reading the input modality is visual

(See), in spelling the input modality is auditory (Hear)

(Fries, 1963; Lerner, 1971). It does not necessarily stand

that a See consonant/Say consonant name task provides informa-

tion which is as relevant to actual spelling behavior as a Hear

consonant sound/Say consonant name task. The latter is a direct

replication of the modalities used in spelling behavior, while

the former can only approximate the spelling behavior by content

alone.

Emerging from the above limitations in the existing

literature is the need for clearly defined and replicable pro-

cedures (T. D. Horn, 1967) that make use of modality relevant

tasks dealing directly with spelling (Bannatyne & Wichiarajote,

1969; Hanna & Hanna, 1966).










Purpose of the Study


The present study was designed to discriminate between good

and poor spelling performance in third grade children. It was

based on a combination of unique theoretical and methodological

parameters that have not been used in spelling research prior

to this study. The unique factors are:

(1) the use of tasks that appear to be prerequisite of

spelling behavior;

(2) the use of accuracy and speed measures as per-

formance descriptors of the tasks;

(3) the use of tasks that make use of modality input

and output directly related to spelling performance;

(4) the use of multiple measures consisting of three

samples of each child's performance on each task; and

(5) the use of a new scoring procedure for the spelling

task itself.

The problem, then, was to identify a set of prerequisite

tasks, modality relevant to spelling performance, that when based

on accuracy and speed as performance measures would discriminate

between good spellers and poor spellers. The specific questions

asked were:

(1) Which tasks, with respect to accuracy, discriminated

between good spellers and poor spellers?

(2) Which tasks, with respect to speed, discriminated

between good spellers and poor spellers?










(3) Which tasks, with respect to both accuracy and speed,

discriminated between good spellers and poor spellers?



Justification of the Study


Among the reasons indicated as possible evidence of low

spelling achievement in the United States are (a) a decline

in the systematic teaching of spelling resulting from erroneous

interpretations of incidental findings and (b) confusion re-

sulting from grapheme and phoneme relationships (E. Horn,

1960). Prediction studies have not been done in the area of

spelling that would help to clarify these issues. What exists

in the literature on spelling, however, are concurrent correla-

tional studies that have related various characteristics (e.g.,

intelligence, demographic information) to spelling competence.

The intent of this study was to use tasks that appear

to be hierarchical prerequisites of spelling competence using

the same input and output modalities as actual spelling per-

formance (Hear input rather than See input). Traditionally,

See or reading type tasks have been used. Hear tasks were chosen

because their topographical relation to spelling behavior may

provide better predictors.

Findings from this study may provide educators with a new

knowledge base upon which to (a) identify good spellers and poor

spellers early; (b) focus on curricular variables related to

remediating poor spelling; (c) focus on these curricular

variables in accordance with performance measures (i.e., accuracy










and speed) found to be relevant; and (d) provide information about

the utility of an input and output modality structure that is

relevant to spelling performance. Whether or not remediation of

related deficit skills will directly result in the improvement

of poor spelling is subject to empirical research. It is not

included within the scope of the present study.



Limitations


Threats to Internal Validity

The internal validity of this study may have been weakened

by the instrumentation procedures used (Campbell & Stanley, 1966).

This refers to possible scoring errors due to individual data

recorder error. Chapter III discusses the procedures used to

protect the internal validity of this study.


Threats to External Validity

The generalizability of this study may have been threatened

by the lack of total randomization in the selection of the subjects.

The reader is cautioned in generalizing the results presented

beyond the geographical area in which the study was done.



Definitions of Terms


Good Speller: The top 25.9 percent (N=35) of the original group

(N=135) based on the spelling test score.





6




Poor Speller: The bottom 25.9 percent (N=35) of the original

group (N=135) based on the spelling test score.


Accuracy: A percentage score based on the number of correct

responses divided by the total number of responses made.


Speed: The total number of responses made, correct and error,

per minute.













CHAPTER II

REVIEW OF RELATED LITERATURE



Introduction


Traditional Variables

Spelling research has traditionally been divided into three

areas. These areas are (a) learner variables, (b) curricular

variables, and (c) word variables (Cahen, Craun, & Johnson, 1971).

The knowledge of learner variables has not proven to be clearly

related to remediating poor spelling or to facilitating good

spelling (T. D. Horn, 1967). The majority of research on spelling

has dealt with curricular variables, e.g., methods and materials

(Blair, 1975; T. D. Horn, 1969). There is no teaching approach

that appears to best facilitate good spelling. Research on

word variables has been concerned with such issues as error

analysis, regularity versus irregularity of orthography, and

high frequency versus low frequency of word occurrence.


Parameters of the Present Study

The present study has five parameters. These parameters

have generated a new perspective for looking at learner per-

formance on relevant tasks. The literature reviewed will be

considered under the rubric of:









(1) learning hierarchies, as they contribute to task

selection;

(2) academic predictors and their contribution to (a)

the selected tasks and (b) the input and output

modality of the selected tasks;

(3) a new procedure for scoring spelling tasks;

(4) accuracy and speed as separate performance

measures of the selected tasks; and

(5) repeated measures of predictor tasks.

These parameters may permit more precision in the discrimination

between good and poor spellers. Learning hierarchies is a

theoretical parameter. The new scoring procedure and multiple

measurement of tasks are methodological parameters. Performance

measures and the contribution of academic predictors are both

theoretical and methodological in nature.



Learning Hierarchies


Definition and Characteristics

Learning hierarchies are "descriptions of the relationships

of positive transfer among intellectual skills" (Gagne, 1968,

p. 4). What they are not, however, are statements of how an

individual acquires the knowledge contained in the hierarchy.

Each level of the hierarchy generates some quantity and quality

that positively transfers to the learning of a "not-previously-

acquired, higher-order capability" (p. 5).











Learning hierarchies, then, represent the effects of cum-

mulative learning with the greatest amount of positive transfer

among elements. Gagne (1968) makes the point that learning

hierarchies represent the most effective direction of learning

for the entire population of learners, and do not represent

strategies unique to individual learners.


Origins of Learning Hierarchies

This study has as one of its parameters the issue of

curriculum and is concerned with the selection of tasks that

may be sequentially ordered. Gagne (1970) has stated that

content in any one area of learning (e.g., spelling) should

be hierarchically ordered so that those simpler concepts and

abilities upon which later learning is based be mastered

first. Instructional design, or management, can then be

based on those conditions of learning (curriculum) that are

requisite to each of the hierarchical components.

Learning hierarchies represent priority ordered skill

sequences. That is to say, based on skill analysis, experience,

and the use of other persons' lists, skills are ordered, in many

instances, post hoc. One should not infer that this method is

in error. However, effort needs to be expended to determine









empirically whether or not these ordered skills transfer

positively in a vertical fashion or are independent (Gagne,

1968).


Validation of Learning Hierarchies

Investigations of the validity of learning hierarchies

have yielded inconclusive results (White, 1973). Based on

Gagne/s (1962) validation model, research efforts reviewed fell

short of being validated. Reasons indicated by White included

methodological weaknesses and the lack of the Gagne model to

provide a test of hierarchical dependence.

White (1973) then discussed a different method for validat-

ing hierarchies. The model proved useful with one major

limitation. Large numbers of hierarchical elements (k) prohibit

the use of the model. This is due to the number of groups of

subjects needed to complete the validation procedure which can

Ih arrived at by using k! groups for maximum control or 2(k-1)

groups for minimum control.

White (1973) concluded his remarks by presenting some

modifications of Gagne's (1962) original design. He stated

that "it would be a pity if further effort is expended on the

production of unreliable or ambiguous data in this [learning

hierarchies] area of learning when much more precise work is

possible" (p. 374).


Task Selection in the Present Study

In light of the comments of White (1973) and Gagne'(1963;

1970) and the lack of validation of current hierarchies, one is










left with a choice of combining what is already in existence or

deriving a new hierarchy that may not fit those already in use.

Specific behaviors assumed to be prerequisite to spelling

have appeared on curriculum hierarchies (Random House, 1972;

Starlin & Starlin, 1972) and in task analyses of spelling

(Westermann, 1971). These behaviors have been accepted as

necessary for spelling performance. It is from these hierarchies

that the tasks used in the present study were selected.


Task Selection and Hierarchy Validation

In lieu of a validated hierarchy in spelling, tasks were

chosen from already published hierarchies. The tasks selected

represent a basic core of skills found at the beginning of each

of these published hierarchies. That is not to say because

they are in agreement with each other they are valid, but

because they are in agreement they represent a consensus of

the field at present. The tasks in use in this study afford

one a beginning point which needs further work toward empirical

validation.



Related Academic Predictor Variables


This study has used concurrent prediction, with third

graders, in an effort to determine a set of tasks, not used

to date, that attempts to explain the nature of the relationship

of this set of tasks to spelling performance.

The following section of the review is limited in scope to

those areas of academic concern that are relevant to the selec-











tion of tasks used in the present study. Of concern are tasks

that are topographically relevant to spelling and tasks re-

lated to speed of manuscript handwriting.


Reading

Hughes (1953) has stated that "the correlation [+.53]

between reading and spelling indicates that . they may

have some skills in common, but are, to a considerable degree,

discrete" (p. 348). Reading and spelling skills, although

topographically similar, make use of different input and output

modalities. Reading is a See/Say task and spelling is a Hear/

Say or Hear/Write task. Therefore, reading represents a

grapheme-phoneme relationship and spelling represents a

phonome-grapheme relationship.

Skills related to spelling need to be assessed with the

modality structure relevant to actual spelling behavior. The

utility of such a modality structure will allow one to place

particular emphasis not only on the content of related tasks,

but to ascertain the importance of the modality structure in

predicting spelling performance.

Phonics. Studies by Spache (1940a), Russell (1937), Bond

(1935), and Monroe (1932) assessed word attack skills related

to spelling. Their results indicated that poor spellers were

inferior to good spellers on these tasks (i.e., blending letters

to form words). Two limitations of these studies that are

relevant to the present study were (a) the tasks involved visual










input rather than auditory input; and (b) the tasks encompassed

different levels of the same topographic task (i.e., CVC words,

CVCV words, CCVC words). Monroe (1932) in her assessment of

vowel or consonant difficulties that might have contributed

to spelling difficulty, based her conclusions on words read

by her subjects. If a child read "these" instead of "those",

the error was vowels. If a child read "then" instead of "them"

the error was consonants.

Templin (1954) found phonics based on visual recognition

tasks to be correlated +.54 to +.57 with spelling. She did use

one recall task, Hear consonant sound and sound combinations/

Write letterss. This task correlated +.34 with spelling.

Although Templin used a Hear task, the task included more than

one basic behavior (single sounds and sound combinations).

Summary. The selection of tasks for use in the present

study were modality relevant to spelling as well as representing

one basic behavior per task. These tasks were selected because

they represent basic skills agreed to be prerequisite to spelling

competence.


Speed of Manuscript Handwriting

The issue of speed in handwriting has been considered as

early as 1914, when Freeman found that increasing speed does

not necessarily sacrifice the quality of handwriting. The

slower the child writes, the slower the language forms are

emitted from him. As a child is able to write more automatically










(proficiently), the easier it is for him to communicate with

others in written form. Early speed studies have not shown

significant differences between poor and good spellers (Cobb,

Kincaid, & Washburn, 1918; Russell, 1937).

Speed data on third graders suggests that they can write

random letters at a mean rate of 41 letters per minute (Stiles,

1974). In another study, Wolking (1973) found that third

graders can write numbers randomly at a rate of 62 numbers per

minute and that by the sixth grade at a rate of 116 numbers per

minute. With an increased emphasis on proficiency as a function

of accuracy and time, perhaps the issue of speed can be a more

clearly defined component of educational designs. The use of

speed data in the present study may yield information that will

further substantiate its inclusion as a performance measure

of academic competence.



Spelling Test Scoring Procedure


Scoring Units

Whole Word. Spelling tests have traditionally been scored

using a words correct, as the scoring unit, procedure. Use of

this method does not allow for the varying length of each word.

Individuals may be penalized for misspelling an entire word,

when they may have, in fact, only misspelled an affix. Further

analysis for instructional purposes is limited.










Syllable. The use of syllables as a scoring unit accounts

neither for the varying length of words nor for the monosyllabism

of some words. Use of this method of scoring subsumes an a priori

knowledge of the rules of syllabication by the speller. This is

not to infer that spellers do not need to be aware of these rules,

but in the early grades (1-3) the more complex rules of syllabica-

tion are not stressed (Griffith, Bishop, Brown, Cohen, Islar,

Pinder, & Steen, 1971).

Sound cluster. Sound clusters have been called phonemes by

some authors (Bannatyne, 1973; White & Haring, 1976). Bannatyne

makes a distinction between phonemes in the word "cough"

(/k/ /ou/ /f/) and optemes and graphemes (/c/ /ou/ /gh/).

Despite the nomenclature used, sound clusters may present some

problem in that the child may have some of the letters correct

but may miss others. In the above example, "c" and "k" are

interchangeable phonemically, as are "ou" and "aw" and "f", "gh",

or "ph". Use of this method does not give an individual credit

for parts of the word he has written correctly, because the whole

word procedure is still a factor.

Letter. White and Haring (1976) suggest that "the letter

is the basic unit of spelling behavior" (p. 29). The following

example will illustrate the scoring procedure. The word "often"

has five letters, thereby having six possible units. The child

is given credit for knowing how to start and knowing how to stop

the word. Carets are used to indicate the flow of the spelling

of the word. "Often," when spelled correctly, looks like










o f t e n (six units), but when spelled without the "t" would

look like o f e n (four units). This scoring procedure allows

one to look not only at the whole word, syllables, and sound

clusters, but at letter sequencing, where many errors seem to

be prevalent. Evidence from pilot data not only substantiated

this contention, but allowed the investigator to look at initial,

medial, and final consonant placement, vowel placement and letter

sequencing. Although this information was not used for purposes

of this study, valuable information for teaching considerations

is made available.



Performance Descriptors


Accuracy

Performance analysis can only be described as accurately as

is allowed for by the performance measures used. Presumably

more precise measures permit finer discrimination to be made.

Traditional measurement provides one with an accuracy score, be

it percent correct of the material presented or percent correct

of the material attempted. Accuracy, therefore, is only repre-

sentative of count, or comparison on an interval scale.


Speed

Speed allows one to look at the number of behaviors (count)

emitted per unit of time. Skinner, in 1932, introduced the con-

cept of speed (count/time) "as a basic unit of direct behavioral

measurement" (Pennypacker, 1972).










Speed is a valuable measure because all behavior occurs in a

time dimension. Speed, in combination with accuracy, allows for

more precise information than is yielded by accuracy alone.

Two children may each read the same material with the same

accuracy. However, the addition of speed as a performance

measure allows one to notice that one child takes two minutes

and the other child takes five minutes. Educationally, one

cannot, with this added information, state that both children

read equally as well.

Haughton (1972) has discussed the importance of speed to

education. He stated that it is fluency (speed) that allows a

child to perform functionally, not accuracy. It must be remembered,

however, that the acquisition of functional speed follows the

acquisition of accuracy on the same task.

This study has made use of both accuracy and speed scores per

task, in an effort to assess the unique contribution each may make

to the same set of predictor tasks and to assess the joint contri-

bution each may make when used to describe the same tasks.



Repeated Measures of Predictor Tasks


Norm-referenced tests are administered once per child. The

score yielded represents that child's functioning level at the

time of testing. To accommodate for errors in measurement related

to such factors as scoring error, illness of the child, and lack

of rapport with the examiner, these tests have included in their










manuals formulae from which may be derived a standard error of

measurement (SEm = cxVi-rxx), where ox = the standard devia-

tion of the test and rxx = the reliability coefficient of the

test.

It has been suggested by researchers (i.e., Stiles, 1974;

White & Haring, 1976) that administration of tasks over a three

day period will yield a more reliable measure. In an effort to

obtain a more reliable measure from each subject, data was

collected on the same tasks for three consecutive days.



Summary


The literature reviewed suggests that (a) tasks used to date

have not been comprised of one basic behavior per task; (b) tasks

that have been used are related to, but are not direct replica-

tions of the modality input and output structure of spelling

behavior; (c) tasks used have not represented the most basic

skills assumed by curriculum hierarchies to be prerequisite

to spelling competence; and (d) tasks have been assessed by

accuracy alone, without referring to speed (fluency) as a viable

performance measure that may yield useful information for

teaching. The present study has used these parameters in a

unique combination that may describe more accurately predictor

variables associated with spelling performance.
















CHAPTER III

PROCEDURES



The focus of this study was to examine a set of concurrent

academic predictors of spelling performance. A group of third

graders was administered a spelling test. From these results,

the top and bottom 25.9 percent (good and poor spellers) were

chosen. Each of these subjects was then assessed on each of

eight predictor tasks on three consecutive days. Accuracy and

speed scores were obtained for each subject on each task. These

scores were used as the independent variables in the statistical

analysis. The dependent measure was group membership, e.g.,

good or poor spellers.



Specific Questions


Question One

Which tasks, with

good spellers and poor

Question Two

Whick tasks, with

good spellers and poor


respect to accuracy, discriminated between

spellers?



respect to speed, discriminated between

spellers?









Question Three

Which tasks, with respect to both accuracy and speed,

discriminated between good spellers and poor spellers?



Subjects


One hundred thirty-five third graders were administered

a spelling test (see Appendix A). The subjects for the study

were drawn from three public schools in Alachua County, Florida.

The sample initially comprised third graders in attendance at

these schools. The schools were selected on the basis of

availability. An upper and lower group (N=35 per group) were

selected from the original group. Each group, therefore, was

comprised of approximately 25% of the original group. These

two groups were used for the remainder of the study.

The following tables report demographic information about

the subjects in the two groups. Table 1 shows the sex ratio

for each group. Table 2 shows the school breakdown for each

group. Table 3 shows the parents' index of social position

for each group (Hollingshead & Redlich, 1958). It may be noted

that there is a disproportionate amount of children from groups

five, six, and seven. The distribution of children in this

sample, based on parents' index of social position, does not

approximate a normal population distribution. This information is

included to further describe the characteristics of the sample,

however, it was not used in the data analysis of the present study.










Table 1

Sex Ratio Per Group


Good Poor Total
Sex N % of Good Group N % of Poor Group N % of Total

Male 16 .46 21 .60 37 .53

Female 19 .54 14 .40 33 .47

Total 35 35 70


Table 2

School Breakdown Per Group


Good
School N % of Good Group


Poor
N % of Poor Group


Total
N % of Total


10 .29 27 .39

12 .34 19 .27

13 .37 24 .34

35 70


A

B

C

Total









Table 3

Parents' Index of Social Position Per Group


Index Not
Group 1 2 3 4 5 6 7 Available

Good 5 6 1 6 6 4 4 3

Poor 0 0 1 4 5 10 11 4

Totals 5 6 2 10 11 14 15 7


1 = Higher executives, proprietors of large concerns, and major
professionals.

2 = Business managers, proprietors of medium-sized business, and
lesser professionals.

3 = Administrative personnel, small independent businesses, and
minor professionals.

4 = Clerical and sales workers, technicians, and owners of little
businesses.

5 = Skilled manual employees.

6 = Machine operators and semi-skilled employees.

7 = Unskilled employees (and unemployed).





Instrumentation


All tasks for all subjects were the same. The tasks used

as the predictor variables were assumed to be representative of

classroom behaviors needed to promote skill attainment in the

area of spelling.









The spelling task used to obtain the dependent, or criterion,

measure included 48 words randomly selected from the Spelling

Errors Test II-IV (Spache, 1955). The test has twelve sections

of ten words each. Each section represents a specific error

type. Although error types were not analyzed for this study,

four words were included from each of the twelve error type

groups. In this way, no subject was penalized if, for example,

he were to be deficient in one error type.

This study has made use of tasks drawn from the Starlin

and Starlin (1972) and Random House (1972) curriculum hierarchies

and from Westermann's (1971) task analysis of spelling. Selected

for presentation in a random order, the predictor tasks used were:

(1) See CVC trigram/Say nonsense word;

(2) Hear consonant sound/Say consonant name;

(3) Hear short vowel sound/Write vowel;

(4) See one place addition fact/Say answer;

(5) Hear two letter blends and digraphs/Write letters;

(6) Hear consonant sound/Write consonant;

(7) Hear short vowel sound/Say vowel name; and

(8) Hear instruction/Write letters of the alphabet.

Tasks 2, 3, 5, 6, and 7 were employed because (a) they are

representative of prerequisite skills assumed to be necessary for

spelling competence and (b) they follow the Hear/Say or Hear/Write

modality structure of spelling behavior. These tasks, neglected

in previous research, needed to be studied so as to assess their

relevance to spelling and to the teaching of spelling.










Tasks 1, 4, and 8 were included because of the unique in-

formation each might contribute to a more definitive understand-

ing of spelling. If the math task (#4) were to predict or

correlate highly with the ability to spell, one might be able

to infer that the prerequisite skills for spelling are not

specific to the domain of spelling.

Inclusion of the alphabet writing task (#8) serves a unique

function. As one assesses short vowel and consonant skills, the

output is to be obtained in two modes, one oral and the other

written. If a child were to have a high rate of oral responding

on these tasks, but were to have a low rate of written responding,

one might assume that the writing of the response interfers with

the performance. This would indicate a performance ceiling on

these tasks placed by the child's writing ability. If, on the

other hand, a child were to have a higher rate of written respond-

ing on these tasks, one might infer that his writing-motor ability

is better than his vocal-motor ability.

Task #1, a See/Say (reading) task, is included as it is

purported to assess the ability to blend basic sounds (Personke,

1972). Assessing the ability to use phonics generalizations

becomes somewhat of a problem. Some words which are phonetically

regular, such as "red," may already have been learned as a sight

word. Therefore, CVC nonsense trigrams were used.










Method


Administration of the Spelling Test

All subjects were administered a spelling test (Spache, 1955).

Use of a spelling test that had words for second, third, and

fourth graders allowed for a broader range of words. The range

of words encompassed (a) a base for poorer spellers to respond,

and (b) a high enough ceiling for better spellers. The spelling

test was administered to each class as a group task. Record

forms were provided for each subject on which the subject's

answers were written. Each word was presented individually, used

contextually in a sentence, and then repeated. For each corres-

ponding number on the answer sheet, a card was held up with that

number so that each child was able to mark the answer in the

appropriate space. Sufficient time was allowed for each child

to finish the word called.

Directions for the spelling task were:

"I am going to give you a spelling test. There will

be 48 words in all. There may be a few words that

you haven't studied before, but I want you to try to

spell these new words the best you can. This is not

one of your regular spelling tests. What you do on

this test will not affect your spelling grade. Let's

start now and see how well you can do.

I will read the word, then use it in a sentence, and

then say the word again. Wait until I say the word

the last time after the sentence before you begin to










write. As we move to each new word, I will hold up

a card with the number on your paper so that you

can make sure you are writing the word in the correct

space. Are there any questions?"

An example was used to assist the child to find the appropriate

writing space on the paper.

Scoring of the spelling test. Scoring of the spelling test

words was based on a procedure suggested by White and Haring

(1976). Using this procedure, each word was analyzed for the

number of units correct. Each word had one more possible unit

than the number of letters in the word itself. By using this

procedure, units were awarded for beginning the word on the

correct letter and for finishing the word with the correct letter.

Forty-eight words were selected from the Spache Errors Test

(Spache, 1955). There were a total of 224 letters for the 48

words plus 48 additional units according to the White and Haring

(1976) scoring procedure. Using this procedure 272 was the

highest spelling score attainable.


Administration of the Predictor Tasks

On three consecutive days, each subject was tested individually

in a place designated by the school principal. There were two

recorders, the investigator and a graduate assistant, working at

all times to facilitate the procedure with as little time inter-

ruption as possible for each cooperating teacher. Each task was

a thirty-second sample of academic behavior. The obtained scores










were multiplied by two so that the standard measure of behavior,

movements per minute, could be used (Lindsley, 1972). Writing

paper and stimulus sheets (see Appendix C) were supplied for each

task for which it was required. Pencils were also supplied. All

subjects used one-quarter inch (.635 cm.) lined paper. All stimulus

sheets read by the subjects were presented on eight and one-half by

eleven inch (21.59x27.94cm.) unlined paper and were typed with

primary size type. All data recorders had appropriate record

sheets on which to record the subjects' responses. All data recorders

had a prior training and practice session.

For all Hear/Say tasks the directions were:

"Listen as I say each vowel (or consonant) sound.

As I say each sound, I want you to tell me the name

of the letter. Are you ready?"

Each Hear/Write task was administered in the same manner except that

the responses were written. Short vowel sounds were "a" as in apple,

"e" as in elephant, "i" as in indian, "o" as in ox, and "u" as in

umbrella.

For the two See/Say tasks the instructions were:

"I want you to look carefully at each nonsense word

(or addition fact). When I say go, I want you to

start here (point) and tell me what the word is (or

what the answer is). I want you to go across the

paper (point to the arrow at the top of the sheet

and follow it across the paper with your finger).

Are you ready?"










The instruction for the task of writing letters of the

alphabet was:

"On this sheet of paper I want you to write all

the letters of the alphabet from 'a' to 'z'. If

you should finish before I say stop, go back to

'a' and try and write as many more letters as you

can. I would like you to print the letters (manu-

script). Are you ready?"


Scoring the Predictor Tasks. Each of the eight predictor

tasks was administered on three consecutive days. Each recorder

was responsible for scoring all oral response tasks. The in-

vestigator scored all written tasks. Pilot data collected showed

that it was easier to correct the written tasks after the task

administration.

Each task presented two mean scores. The accuracy score was

a mean score based on the number of responses correct divided by

the total number of responses per task attempted. The second score

was a speed score. A mean speed score represented the average of

the total number of responses, correct and error, attempted per

subject on each task.



Data Analysis


The Biomedical Computer Programs BMD 07M (Dixon, 1973) was

accessed for a stepwise discriminant analysis procedure. The

program allowed for the use of a linear combination of measures

from which those measures contributing most to the variance of










the dependent measure were kept in the prediction equation. The

use of a two group discriminant analysis allowed for the dependent

variable to be dichotomous, in other words, to represent group

membership. The two-group discriminant analysis procedure pro-

vided a classification schema by which an individual, based on

predictor variable scores, was assigned membership into that

group whose mean score he most closely approximated.

Three free-entry subproblem routines were performed. The

first routine was an analysis using both accuracy and speed

measures of the eight tasks. The second analysis used only

accuracy scores and the third analysis used speed scores.


Additional Analyses

A second program (BMD 07M, Dixon, 1973) was accessed based

upon the results of the first analysis. This second program, also

with three subproblem routines, used only those tasks from the

first program that were found to be the most reliable discrimina-

tors. Three prediction equations were obtained from the three

subproblem routines of the second program. The results of these

equations, with a child's score placed in the formulae, compared

to the cut-off score, represent an individual's predicted group

membership, as a function of accuracy scores, speed scores, and

both accuracy and speed scores. Constants and coefficients for

each equation are different due to the contribution of the speci-

fic performance measures used.










Additional information yielded by the BMD 07M and BMD 02R

programs (Dixon, 1973) will be discussed in Chapter IV. This

information includes variance accounted for in the dependent

measure, e.g., group membership, by all of the predictor

variables (see page 23) and possible misclassification of

individuals to each group.

The BMD 02R program (Dixon, 1973) was accessed a second

time. The variables included were (a) Hear consonant sound/Say

consonant name (CONS H/S); (b) Hear consonant sound/Write consonant

(CONS H/W); (c) Hear short vowel sound/Say vowel name (SHVL H/S);

and (d) Hear short vowel sound/Write vowel (SHVL H/W). The math

and alphabet writing tasks were not included. Two other tasks were

not included. They were: See CVC trigram/Say nonsense word (CVC)

and Hear two letter blends and digraphs/Write letters (BLDI).

Therefore, the four single component, auditory input tasks (con-

sonant or short vowel; Hear/Say and Hear/Write) were entered in the

program. The use of only these four tasks was for the purpose of

analyzing their contribution as discriminating variables separate

from (a) more complex tasks (CVC and BLDI) and (b) tasks not

directly relevant to the development of spelling behavior. Three

free-entry subproblem routines were accessed. Accuracy scores

were used in the first subproblem. Speed scores were used in the

second subproblem. Accuracy and speed scores in combination were

used in the third subproblem.

Average accuracy and speed improvements were calculated for

each subject. This was done by subtracting the day one accuracy

or speed score from the day two score. The day two score was then










subtracted from the day three score. These two results were then

added together. The sum was divided by two. The resultant score

was the average accuracy or speed improvement per subject per

task. These scores were pooled to calculate the mean improvement

per group per task.














CHAPTER IV

RESULTS



The present study was conducted to generate a set of

predictor tasks that would discriminate between a group of

good spellers and a group of poor spellers. This set of tasks

was based on accuracy, speed, and accuracy and speed scores in


combination.

Specific questions asked were:

(1) Which tasks, with respect

discriminate between good


spellers?

(2) Which tasks,

discriminate


with respect

between good


to accuracy, would

spellers and poor



to speed, would

spellers and poor


spellers?

(3) Which tasks, with respect to both accuracy

and speed, would discriminate between good

spellers and poor spellers?


Question One

Results from the discriminant analysis procedure using

accuracy scores showed that two variables, (1) See CVC trigram/

Say nonsense word (CVC-A) and (2) Hear two letter blends and










digraphs/Write letters (BLDI-A) were reliable predictors signifi-

cant at the .01 level. The following discriminant prediction

equation resulted:

Y = -15.7142 + 0.1805 (CVC-A) + 0.0768 (BLDI-A)

By inserting a child's CVC-A and BLDI-A scores in the above equa-

tion the resultant score can be compared to the cut-off score of

+0.0034, above which a child would be predicted to be a good

speller and below which a child would be predicted to be a poor

speller.


Question Two

Results from the discriminant analysis procedure using speed

scores showed the two variables, (1) See CVC trigram/Say nonsense

word (CVC-S) and (2) Hear two letter blends and digraphs/Write

letters (BLDI-S) were reliable predictors significant at the .01

level. The following discriminant prediction equation resulted:

Y = -16.7490 + 0.2553 (CVC-S) + 0.3354 (BLDI-S)

By inserting a child's CVC-S and BLDI-S scores in the above equa-

tion, the resultant score can be compared to the cut-off score of

+0.0002, above which a child would be predicted to be a good

speller and below which a child would be predicted to be a poor

speller.


Question Three

When the eight tasks were entered into the discriminant analysis

using both accuracy and speed scores for each task, four tasks were

found to be reliable predictors. Significant at the .01 level were,









(1) See CVC trigram/Say nonsense word (CVC-S) and (2) See CVC

trigram/Say nonsense word (CVC-A). Significant at the .05 level

were (1) Hear two letter blends and digraphs/Write letters (BLDI-

S) and (2) Hear two letter blends and digraphs/Write letters

(BLDI-A). The following discriminant prediction equation re-

sulted:

Y = 23.6238 + 0.1818 (CVC-S) + 0.1329 (CVC-A) +
0.2560 (BLDI-S) + 0.0532 (BLDI-A)

By inserting a child's CVC-S, CVC-A, BLDI-S, and BLDI-A scores

in the above equation the resultant score can be compared to the

cut-off score of -0.0032, above which a child would be predicted

to be a good speller and below which a child would be predicted

to be a poor speller.


Additional Findings


Contribution of Predictor Variables in Accounting
for Variance in the Dependent Measure

Additional information yielded by accessing the BMD 02R

stepwise regression program (Dixon, 1973) is presented in Table 4.

Table 4

Predictor Variable Contributions


Performance Variables in Order of Varianc
Measure Entrance in the Program the D

Accuracy CVC-A(76.3%),BLDI-A(3%)
Remaining 6 Variables

Speed CVC-S(76.6%),BLDI-S(2.7%)
Remaining 6 Variables

Accuracy CVC-S(76.6%),CVC-A(6.5%),
and Speed BLDI-S(0.9%),BLDI-A(0.7%)
Remaining 12 Variables


:e Accounted for in
Dependent Measure

79.3%
1.2%

79.3%
1.6%


84.7%
2.2%








Due to the use of free-entry subproblem routines, the first variable

entered accounted for unique and overlapping variance. Each subsequent

variable entered represented an incremental amount of unique and over-

lapping variance not already accounted for by a previously entered

variable.

The results presented in Table 4 for accuracy alone and for speed

alone show that the CVC task was the major contributor to variance

accounted for in the dependent measure by each performance descriptor.

The CVC-S score accounted for 76.6% of the variance, which was

slightly more, .3%, than the CVC-A score (76.3%). Therefore, when

accuracy and speed scores were entered into the stepwise regression

program simultaneously, the CVC-S score entered first, showing both

unique and overlapping variance attributable to the CVC-S score. The

decrease in variance accounted for in the dependent measure by the

CVC-A score from 76.3%, when entered with accuracy scores, to 6.5%,

when entered with both accuracy and speed scores, is reflective of the

overlapping amount of variance shared by the CVC-S and CVC-A scores.


Initial and Predicted Group Assignment

Each subject was assigned initially to either the good spellers

group or the poor spellers group based on his/her spelling test score.

The range of scores for the good spellers group was from 85.3% to

99.6% correct. The range of scores for the poor spellers group was

from 7.7% to 49.5% correct. Results of the analysis indicate that

on the basis of the best predictor variable scores, certain individuals

would have been misclassified with respect to their initial group

assignment. Tables 5, 6, and 7 are shown in terms of initial assign-

ment and predictor variable assignment.








Table 5

Accuracy and Speed as Performance Measures

Predicted Assignment

Good Poor

Initial Good 34 1
Assignment Poor 0 35


Thirty-four of the 35 subjects initially assigned to the good

spellers group were also assigned to this group based on predictor

variable scores. One subject classified initially as a good speller

was classified as a poor speller on the basis of his/her predictor

variable scores. All 35 of those assigned initially as poor spellers

were predicted to be poor spellers on the basis of their predictor

varibale scores.

Table 6

Accuracy as a Performance Measure

Predicted Assignment

Good Poor

Initial Good 35 0
Assignment Poor 2 33
Poor 2 33

The entire group initially assigned as good spellers was pre-

dicted to be good spellers. Thirty-three of the 35 subjects assigned

initially to the poor spellers group were also assigned to this group

based on predictor variable scores. Two subjects classified initially

as poor spellers were misclassified as good spellers on the basis of

their predictor variable scores.








Table 7

Speed as a Performance Measure

Predicted Assignment

Good Poor

Initial Good 33 2
Assignment Poor 0 35


Thirty-three of the 35 subjects assigned initially to the good

spellers group were also assigned to this group based on their pre-

dictor variable scores. Two subjects classified initially as good

spellers, were classified as poor spellers based on their predictor

variable scores. The entire group (n = 35) of initially assigned poor

spellers were classified as poor spellers based on their predictor

variable scores.

Use of the discriminant prediction equation misclassified 3% of the

good spellers group as poor spellers and 0.0% of the poor spellers group,

based on accuracy and speed scores. Six percent of poor spellers group

and 0.0% of the good spellers group were misclassified based on accuracy

scores. Six percent of the good spellers group and 0.0% of the poor

spellers group were misclassified based on speed scores.

Scatter plots are reported in Appendix G. Each scatter plot

represents one task and one performance measure of that task and its

relationship to the spelling test score (percent correct). Means and

standard deviations for the good spellers group and for the poor

spellers group on each task are reported in Appendix E. Raw score

data are reported in Appendix F.








Analysis of Single Component, Auditory Input Tasks

A supplementary analysis was conducted which included the four

single component, auditory input tasks (CONS H/S, CONS H/W, SHVL H/S,

and SHVL H/W). The CVC and BLDI tasks that were found to be the best

predictors in the original analysis were not included. The reason for

not including the CVC and BLDI tasks was that their relationship to

spelling performance was already known to be significant. The See one

place addition fact/Say answer (MATH) and Hear instruction/Write letters

of the alphabet (ALPH) tasks were also not included in the present

analysis. This was done for two reasons. First, the MATH and ALPH

tasks were found not to be related to spelling performance in the over-

all analysis. Second, the MATH and ALPH tasks are not curriculum tasks

directly relevant to the development of spelling behavior.

In order to ascertain the relationship of single component,

auditory input tasks (CONS H/S, CONS H/W, SHVL H/S, and SHVL H/W) to

spelling performance, a separate analysis was performed. The results

are indicated in Table 8.

Table 8

The Relationship of Single Component, Auditory
Input Tasks to Spelling Performance

Performance Variables in Order of Variance Accounted for
Measure Entrance in the Program in the Dependent Measure

Accuracy SHVL H/W (31.5%) 31.5%

Speed SHVL H/W (52.1%) 59.7%
CONS H/W (5.4%)
SHVL H/S (2.2%)

Accuracy SHVL H/W-S (52.1%) 64.0%
and Speed SHVL H/W-A (5.7%)
CONS H/W-S (6.2%)









All variables except one were significant at the .01 level. The

SHVL H/S tasks with speed as a performance measure was significant

at the .05 level.


Accuracy and Speed Change

Each of the eight tasks was administered on each of three

consecutive days. The group means presented in Table 9 represent

each groups average change over the three day period. Both groups

showed an increase in speed scores. The good spellers group

showed a decrease in accuracy on the ALPH task. The poor spellers

group showed a decrease in accuracy on the CVC and SHVL H/S tasks.


Table 9

Mean Change Per Task Per Group



TASK GOOD GROUP MEAN POOR GROUP MEAN
ACCURACY SPEED ACCURACY SPEED
(% Correct) (Response/Min;) (% Correct) (Response/Min.)

CVC 1.11 10.94 -1.47 2.94

CONS H/S 0.39 4.94 1.00 3.23

SHVL H/W 1.19 4.00 -0.63 4.63

MATH 0.09 4.91 0.17 1.97

BLDI 3.61 2.71 3.26 1.57

CONS H/W 0.56 2.99 0.77 2.26

SHVL H/S 1.30 1.43 1.11 1.46

ALPH -0.04 4.37 0.10 2.31

MEAN CHANGE +1.02 +4.53 +0.54 +2.55














CHAPTER V

DISCUSSION



The focus of the present study was to identify a set of

prerequisite tasks, modality relevant to spelling performance,

that when based on accuracy and speed as performance measures,

would discriminate between good spellers and poor spellers.

A review of related literature suggested that (a) tasks used

to date have represented more than one basic behavior per

task; (b) tasks that have been used are related to, but are

not direct replications of the input and output modality

structure of spelling behavior; (c) tasks used have not

represented the most basic skills assumed by curriculum

hierarchies to be prerequisite to spelling competence; and

(d) tasks have been assessed by accuracy alone, without re-

ferring to speed (fluency) as a viable performance measure.

A spelling test was administered to a group of third

graders in Alachua County, Florida. From these results the

top 25.9 percent (good and poor spellers) were chosen (N=35

subjects per group). Each of these subjects was then assessed

on each of eight tasks on three consecutive days. Accuracy

and speed scores were obtained for each subject on each task.










Two of the eight tasks used, See CVC trigram/Say nonsense word

and Hear two letter blends and digraphs/Write letters, were the

two tasks with which to discriminate groups of good and poor third

grade spellers. The use of accuracy and speed scores in combina-

tion revealed that speed was the better performance measure for

discriminating between the two groups of spellers.

The results will be discussed under the following

headings:

(1) Task complexity and learning hierarchies;

(2) Input and output modality structure of the task;

(3) Spelling test scoring procedure;

(4) Performance measures;

(5) Repeated measures of predictor tasks;

(6) Conclusions; and

(7) Recommendations for future research.



Task Complexity and Learning Hierarchies


The tasks used in the present study were selected from

spelling skill hierarchies and task analyses. The content of

six of the tasks followed the developmental progression of skills

found in the hierarchies. Two tasks were single component tasks

(CONS and SHVL). The findings indicated that at the third grade

level, the more complex skills are better at discriminating

between good and poor spellers than the single component tasks.










Gagne (1965) suggested an order assigned to skills within

a hierarchical structure (simple responses, chains or multiple

discrimination, classifying, use of principles or rules, and

problem solving). The CONS and SHVL tasks seem to be an example

of chains or multiple discrimination. The CVC task seems to

be an example of the use of principles or rules. For the third

grade subjects included in this study, simple response tasks

(CONS, SHVL) were not found to discriminate reliably between

good and poor spellers. One may conclude, then, that the simple

tasks do not account for a significant amount of the variance

in the spelling task itself. The CVC and BLDI tasks, or the

more complex tasks, were the tasks that were found to discriminate

reliably between the groups of good and poor spellers.

An additional analysis that included the four single com-

ponent tasks (CONS H/S, CONS H/W, SHVL H/S, and SHVL H/W) lends

support to this line of reasoning. The SHVL tasks were the best

discriminators, regardless of performance measure.

The overall analysis indicated that the four tasks, with

content relevant to spelling behavior (CVC, BLDI, SHVL, and CONS),

followed a descending order from complex to simple. The order

was CVC, BLDI, SHVL, and CONS.

Although this study was not designed to validate any one

learning hierarchy, the results may be serendipitous in their

outcome. Of the four tasks based on spelling curriculum

hierarchies, the CONS tasks were least successful in dis-










criminating between good and poor spellers. The SHVL tasks

were slightly more successful. The BLDI task was more success-

ful yet. The CVC task, the more complex task according to the

hierarchies, was the most successful task to discriminate between

good and poor spellers.



Input and Output Modality Structure of the Task


One of the two tasks found to be the best discriminators

of third grade spelling performance, Hear two letter blends and

digraphs/Write letters was a direct replication of the modality

structure of actual spelling behavior. The CVC task had the

modality structure of a reading task (See/Say). The contribution

of the BLDI task to the variance of the dependent measure ranged

from 0.7% to 3.0% depending on the performance measures) used

(See Table 4). The results indicated that although its contri-

bution was statistically significant, the practical significance

of its contribution in explaining variance accounted for in the

dependent measure, is clearly limited.

Previous studies in the spelling literature have not made

use of a direct replication of the modality structure of actual

spelling behavior (Bond, 1935; Monroe, 1932; Russell, 1937;

Spache, 1940a; Wallace, Klein, & Schneider, 1968). An attempt

was made to ascertain whether or not auditory input tasks may

have been more related to spelling behavior than visual input

tasks. However, the CVC task, a reading task, was found to be








the task that best discriminated between the good spellers and poor

spellers. On the basis of the results of this study, one must agree

with Monroe's (1932) statement that

The correlation [between reading and spelling is] so high
.we must be measuring an achievement which is greatly
dependent either upon reading or upon the same factors
which underlie the ability to read. (p. 13)


Spelling Test Scoring Procedure

A new procedure for scoring the spelling test (White & Haring,

1976) was used in this study. In a post hoc analysis, the spelling

tests were rescored using the more traditional words correct pro-

cedure. A Pearson Product Moment Correlation was performed between

units correct (White & Haring) and words correct. A correlation of

+.97 was obtained.

The results of the correlation procedure indicated that either

scoring procedure could have been used in this study. The preference

of using the White and Haring (1976) scoring procedure would be to

learn more about a child's spelling behavior because of the exten-

sive word analysis that would be obtained. The whole word procedure

would be preferable where scoring time was a primary consideration.


Performance Measures

The results of this study were obtained by considering the

original set of eight tasks in three ways. Accuracy scores were

used in the first subproblem. Speed scores were used in the second

subproblem, accuracy and speed scores in combination were used in

the third subproblem. Regardless of the performance measure used,

the CVC and BLDI tasks were found to be the best discriminating tasks








in the same order (CVC, BLDI). However, when accuracy and speed scores

were used in combination in a free-entry subproblem routine, the speed

scores on each task (CVC, BLDI) accounted for more variance in the

dependent measure than the accuracy scores (See Table 4).

Accuracy scores are not to be interpreted as less powerful than

speed scores but as representing incremental variance attributed

uniquely to each of the successive variables entered into the predic-

tion equation. The near equivalence of accuracy and speed scores

entered by discrete performance descriptors disappears when the two

are entered simultaneously. This phenomenon is evidenced as the step-

wise regression program selects the variable that is most highly

correlated with the dependent variable. This first variable, and

each successive variable, enters showing unique and overlapping variance

accounted for in the dependent measure. Each successive variable, then,

is selected to enter on its contribution of variance in the dependent

measure. An analysis of single component, auditory input tasks (CONS,

SHVL) indicated that speed scores were again slightly more powerful

in their ability to discriminate between good and poor spellers than

were accuracy scores (See Table 8).

Speed and accuracy scores on the tasks were found to be equivalent

discriminators between the two groups of spellers. An indication that

remediation or development of the skills on these tasks needs to in-

clude both accuracy and speed before final proficiency can be noted.

This indication is suggested by the fact that speed and accuracy were

equivalent performance measures. Based on a mean score of the good

spellers group (N=35) on each task, the following accuracy and speed

scores may be helpful in determining proficiency levels of the tasks.









Table 10

Proficiency Levels Per Task



TASKS MEAN SCORE


CVC (accuracy) 91.6% correct

CVC (speed) 63.2 responses per minute

BLDI (accuracy) 86.5% correct

BLDI (speed) 24.0 responses per minute



Speed and accuracy scores are also relevant to the issue of

possible misclassifications. Children predicted to be good or

poor spellers and who are indeed good or poor spellers will receive

appropriate curricular instruction. Children predicted to be poor

spellers who are indeed good spellers will certainly not suffer

from extra curricular intervention. However, children predicted

to be good spellers who are indeed poor spellers may suffer from

lack of instruction. The lack of instruction for any child that

needs it may cause teacher and learner frustration. The misclassi-

fication of this last group of children, those predicted to be good

spellers but who are not good spellers, needs to be minimized.

The results of this study indicated that when accuracy and

speed scores in combination or speed scores alone are used as

performance measures, no children are misclassified by being

predicted as good spellers when they are indeed poor spellers

(See Tables 5 and 7). The small percentage of misclassifications










by accuracy scores appears to be a reflection of the N used in

this study. Therefore, speed, accuracy, or a combination of

speed and accuracy could be used to reduce misclassifications

that could be harmful to the learner.

Group standard deviations on each task presented an interest-

ing descriptive finding. On four of the tasks included in this

study (CVC, MATH, BLDI, and ALPH) the poor spellers group dis-

played more variability with respect to accuracy than did the

good spellers group. However, on these same four tasks, the good

spellers group displayed more variability with respect to speed

than did the poor spellers group. Although standard deviations

for each group on each task are presented in Appendix E, Table

11 presents that standard deviation data relevant to the CVC,

MATH, BLDI, and ALPH tasks.


Table 11

Standard Deviations on the CVC, MATH, BLDI, and ALPH Tasks



TASK STANDARD DEVIATION STANDARD DEVIATION
GOOD GROUP POOR GROUP


CVC (accuracy) 10.0 22.0*
CVC (speed) 16.1* 7.4
MATH (accuracy) 4.8 23.0*
MATH (speed) 12.6* 8.7
BLDI (accuracy) 11.3 27.4*
BLDI (speed) 5.6* 3.2
ALPH (accuracy) 1.0 2.9*
ALPH (speed) 14.0* 7.8


*more variability










One may infer from these findings that the poor spellers

group, with a wider dispersion of accuracy scores and less

variability in speed scores, has not completed the basic

stage of acquiring accurate responses. The good spellers

group, however, was less variable on accuracy scores. Their

accuracy scores may indicate mastery of the tasks included

for this study. However, the variability in their speed scores

leads one to believe that they have not, as a group, reached a

fluent level of performance on these four tasks.

It has been suggested (Haughton, 1972; Starlin, 1972) that

although the final proficiency aim for any child on any task is

accurate responding at an acceptable speed, accurate responding

needs to be in a child's repertoire before accurate responding

at an acceptable speed can be obtained. In this study, the poor

spellers, as a group, were responding neither accurately nor

fluently. The lack of variability in their low speed scores

suggests that as a group their speed scores may be rather con-

sistent because they have not yet mastered the stage of accurate

responding. This would account for an overall slow speed of

responding on these tasks. The good spellers group, because of

the lack of variability in their high accuracy scores, seems to

have mastered the stage of responding accurately. However, the

variability in their speed scores leads one to believe that as

a group they are not yet fluent on these four tasks.

In summary, the findings of this study indicated that the

use of speed and accuracy as performance measures may serve two










useful functions. First, speed and accuracy scores discriminate

between good and poor spellers. Second, speed and accuracy scores

reduce misclassifications that may interfere with a child's curricular

program.



Repeated Measures of the Tasks


Repeated measures over three days for each task were obtained

to represent a more reliable measure for each subject's performance

on each task. Group means are reported in Table 9.

Both groups of spellers increased their speed scores over the

three day period. The good spellers, as a group, did not improve

their accuracy scores on the ALPH task, due to the ceiling effects

of the task. The poor spellers, as a group, did not improve their

accuracy scores on the CVC and SHVL H/W tasks.

Low increases in accuracy scores for the subjects involved

may have been due to four reasons. First, some subjects maintained

100% accuracy over the three day period. Therefore, due to ceiling

effects, their accuracy scores could not increase, while their

speed scores may have increased. On the CVC task, subject #103

remained at 100% accuracy for the three days while the subject's

speed score change was +20.9 responses per minute over the three

days.

Second, some subjects maintained a 0.0% accuracy score over

the three day period. Subject #230 maintained a 0.0% accuracy

for three days on the BLDI task while the subject's speed score

changed by one incorrect response per minute.










Third, some subjects, while increasing their speed scores,

lowered their accuracy scores. For example, on the CVC task,

subject #116 had speed scores of 48, 60, and 85.8 responses per

minute for the three days. This subject's accuracy scores were

100, 100, and 97 percent correct for the three days. As the

subject's speed increased the chance for error increased.

Therefore, due to a single error on day three, this subject's

average accuracy change was -1.5% while the average speed change

was +18.9 responses per minute.

Fourth, some subjects who may have had an increase from day

one to day two may have had a decrease from day two to day three

or vice versa. Depending on the differences between the days,

some subjects may have shown no change and others an increase

or decrease. Subject #117 had a -6 speed increase from day one

to day two and a +12 speed increase from day two to day three,

with an overall speed change of +3 (CVC task). On the BLDI

task, subject #225 had a speed change of +8 from day one to day

two and a speed change of -8 from day two to day three. This

subject's overall change in speed on the BLDI task was zero.

The results of these findings indicated a positive change

in speed on all tasks for both groups. The good spellers group

exhibited a positive chanqr in accuracy scores on seven of the

tasks. The poor spellers group exhibited a positive change in

speed on six of the tasks. Each group exhibited change on each

task over the three day period. Although the changes were

generally not large, it was evident that two or three samples









of each task was a closer approximation of each child's maximum

capability on each task. Therefore, the contention that three days

of data collection as a more reliable measure of a subject's per-

formance than one day of data collection was upheld.


Conclusions


The focus of this study and the specific questions asked were

stated in terms of identifying a set of prerequisite tasks that

would best discriminate between good spellers and poor spellers.

The set of tasks used in this study was discussed in terms of

accuracy and speed as performance measures both separate and in

combination. The results of the data analysis indicated that when

accuracy and speed were used as separate performance measures, the

same two tasks were found to be the best discriminators between the

groups of good and poor spellers. The two tasks were See CVC

trigram/Say nonsense word and Hear two letter blends and digraphs/

Write letters. When accuracy and speed scores were used in combina-

tion as performance measures, the CVC and BLDI tasks again were

found to be the best discriminators between the two groups of

spellers. However, when used in combination, the speed score for

each task was a slightly more powerful discriminator than was the

accuracy score due to the free-entry variable selection (See previous

discussion).

Five of the tasks selected for inclusion in this study were

direct replications of the input and output modality structure

of spelling behavior. One task was a reading task; the other two

tasks were a math task and an alphabet writing task. Although










the BLDI task, a Hear/Write task, was one of the two best tasks

in discriminating between groups of good and poor spellers, it's

contribution to the variance accounted for in the dependent

measure was extremely small (See Table 4). The CVC task, a

reading task, was the best discriminating task. Therefore, the

use of tasks involving the input-output modality structure that

directly replicates spelling behavior was less reliable than the

reading task in discriminating between the two groups of spellers.

Supplementary analyses of the data indicated that the use

of speed, as a performance measure, classified children for

curricular programming with slightly less liability than the

use of accuracy. Speed was also the performance measure that

showed the greatest change over the three days of data collection.

The spelling test scoring procedure used in this study was found

not to be time saving. However, it presented a more extensive

analysis of a child's spelling performance than the words correct

procedure. Although this study did not attempt to validate any

one spelling skill hierarchy, the findings supported a sequence

of the tasks used, from simple to complex.


Recommendations for Future Research


The present study has included multiple regression and dis-

criminant analysis procedures. Through the use of these procedures,

this investigator has obtained information that is useful in (a)

discriminating between groups of good and poor spellers and (b)

discussing the most reliable discriminating tasks. The findings









have been focused on issues such as (a) task complexity and

learning hierarchies; (b) the input and output modality structure

relevant to spelling behavior; and (c) the utility of accuracy

and speed as performance measures. The resultant information

from this study provides one with a basis upon which a second

stage of more functionally oriented research may be based. This

second stage of research, then, should consist of studies that are

concerned with the functional relationship of the present findings

to classroom spelling curricula.

This task (BLDI) had the modality structure that directly

replicated spelling behavior. The BLDI task was less reliable

as a discriminator than the reading task. It would be interesting

to include the Hear/Say and Hear/Write modality structure as a

unique experimental variable. Efforts could then be made to

determine if curricular instruction, with an emphasis on these

modality structures (Hear/Say or Hear/Write), would enhance the

ability to spell. The curricular emphasis in many classrooms is

on reading. There are those children, however, that even with

reading instruction are poor spellers. Perhaps an emphasis on

related content material, with the modality structure directly

replicate of spelling behavior, would improve this poor spellers

group's ability to achieve competence in spelling.

The present findings have demonstrated the utility of speed

(fluency) as a viable performance measure that needs to be

emphasized equally with accuracy in classroom instruction. The

results, using speed and accuracy, were similar. Speed, as a









performance measure, has not been used before and may be a novel

means by which variance in performance can be explained. Research

efforts are needed that will use speed as an aim for skill mastery.

These research efforts need to consider the efficacy of remediating

deficit skills (i.e., CVC, BLDI) to a level of fluent mastery.

Once these skills have been mastered, a determination of their

effectiveness in improving spelling performance may be conducted.

It is indicated from the findings that for the third graders

included in this study, the more complex tasks (CVC, BLDI) discrim-

inated best between the two groups of spellers. Further research

efforts at different grade levels (i.e., one, five) may demon-

strate that different skills discriminate at different grade

levels or that by a certain grade level, the same tasks can be

used as discriminators. This may well serve to provide a

developmental sequence of skills. Performance on this sequence

could alert teachers to possible concomitant performance in spel-

ling. As stated earlier, if other research efforts suggest that

mastery of deficit skills is effective in improving spelling, a

criterion performance sequence may have excellent curricular

merit.

The present study has answered the specific questions for

which it was designed. The overall findings as well as the

additional findings have suggested possible alternative research

efforts that may help solve the problem of presenting classroom

instruction that could promote better spelling achievement.















REFERENCES


Allen, D., & Ager, J. A factor analytic study of the ability
to spell. Educational and Psychological Measurement,
1965, 25, 153-161.

Bannatyne, A. Reading: An auditory vocal process. San Rafael,
Cal.: Academic Therapy Publications, 1973.

Bannatyne, A., & Wichiarajote, P. Relationships between written
spelling, motor functioning, and sequencing skills. Journal
of Learning Disabilities, 1969, 2(1), 4-16.

Birch, H. G., & Belmont, L. Auditory-visual integration, intelli-
gence, and reading ability in school children. Perceptual
and Motor Skills, 1965, 20, 295-305.

Blair, T. R. Spelling, word attack skills; WRIC/RCS report.
Reading Teacher, 1975, 28(6), 604-605.

Bond, G. L. The auditory and speech characteristics of poor
readers. Contributions to Education, No. 657. New York:
Teacher's College, Columbia University, 1935.

Cahen, L. S., Craun, M. J., & Johnson, S. K. Spelling difficulty-
a survey of the research. Review of Educational Research,
1971, 41(4), 281-301.

Campbell, D. T., & Stanley, J. C. Experimental and quasi-
experimental designs for research. Chicago: Rand McNally,
1966.

Cobb, M. E., Kincaid, M., & Washburn, M. E. Further studies of
the verbal abilities of poor spellers. American Journal
of Psychology, 1918, 19, 331-332.

Cook, W. A., & O'Shea, M. V. The child and his spelling.
Indianapolis: The Bobbs-Merrill Company, 1914.

Dixon, W. J. (Ed.). Biomedical computer programs. Berkeley:
University of California Press, 1973.

Freeman, F. N. An experimental analysis of the writing movement.
Psychological Monographs, 1914, 17, 4.










Fries, G. C. Linguistics and reading. New York: Holt, Rinehart,
and Winston, 1963.

Gagne, R. M. The acquisition of knowledge. Psychological Review,
1962, 69, 355-365.

Gagne' R. M. The conditions of learning (1st ed.). New York:
Holt, Rinehart, and Winston, 1965.

Gagne, R. M. Learning hierarchies. Educational Psychologist,
1968, 6(1), 1-6.

Gagne R. M. The conditions of learning (2nd ed.). New York:
Holt, Rinehart, and Winston, 1970.

Griffith, S., Bishop, J., Brown, R., Cohen, C., Islar, M., Pinder,
B., & Steen, C. Cognitive skills planning system (part 2).
Gainesville, Fl.: Alachua County School Board, 1971.

Hanna, P. R., & Hanna, J. S. Applications of linguistics and
psychological cues to the spelling course of study. In
T. D. Horn (Ed.), Research on handwriting and spelling.
Champaign, Ill.: National Council of Teachers of English,
1966.

Haughton, E. Aims: Growing and sharing. In J. B. Jordan &
L. S. Robbins (Eds.), Let's try doing something else kind
of thing. Arlington, Va.: Council for Exceptional
Children, 1972.

Hollingshead, A. B., & Redlich, F. C. Social class and mental
illness: A community study. New York: John Wiley and Sons,
1958.

Horn, E. Spelling. In C. W. Harris (Ed.), Encyclopedia of
educational research. New York: MacMillan, 1960.

Horn, T. D. Handwriting and spelling. Review of educational
research, 1967, 37, 168-177.

Horn, T. D. Spelling. In R. L. Ebel (Ed.), Encyclopedia of
educational research. London: MacMillan, 1969.

Hughes, V. H. A study of the relationship among selected language
abilities. Journal of Educational Research, 1953, 47, 97-
106.

Learner, J. W. Children with learning disabilities. Boston:
Houghton Mifflin, 1971.








Lindsley, 0. R. From Skinner to precision teaching: The child
knows best. In J. B. Jordan & L. S. Robbins (Eds.), Let's
try doing something else kind of thing. Arlington, Va.:
Council for Exceptional Children, 1972.

Monroe, M. Children who cannot read. Chicago: University of
Chicago Press, 1932.

Pennypacker, H. S. Why frequency? The rationale of precise
behavior management. Unpublished manuscript. University
of Florida, 1972.

Personke, C. R. Use of nonsense words to test generalization
ability in spelling. Elementary English, 1972, 49, 1233-
1239.

Random House. High intensity learning systems. New York:
author, 1972.

Russell, D. H. Characteristics of good and poor spellers.
Contributions to Education No. 727. New York: Teacher's
College, Columbia University, 1937.

Spache, G. Characteristic errors of good and poor spellers.
Journal of Educational Research, 1940, 34, 182-189. (a)

Spache, G. The role of visual defects in reading and spelling
disabilities. American Journal of Orthopsychiatry, 1940,
10, 229-238. (b)

Spache, G. Spelling disability correlates I factors probably
causal in spelling disability. Journal of Educational
Research, 1941, 34(8), 561-586. (a)

Spache, G. Spelling disability correlates II factors that
may be related to spelling disability. Journal of Educa-
tional Research, 1941, 35, 119-137. (b)

Spache, G. D. Spelling errors test. Gainesville, Fl.: Reading
Laboratory and Clinic, University of Florida, 1955.

Starlin, A. Sharing a message about curriculum with my teacher
friends. In J. B. Jordan & L. S. Robbins (Eds.), Let's
try doing something else kind of thing. Arlington, Va.:
Council for Exceptional Children, 1972.

Starlin, C., & Starlin, A. Curriculum ladder for spelling.
Bemidji, Minn.: Unique Curriculums Unlimited, 1972.

Stiles, R. L. Stepwise multiple regression study of SST basic
movement cycle data with SFTAA/CTBS results from 348 third
grade pupils. Research Department, Pupil Personnel Services,
Tacoma Public Schools, 1974.





58




Templin, M. C. Phonic knowledge and its relation to the spelling
and reading achievement of fourth grade pupils. Journal of
Educational Research, 1954, 47, 441-454.

Wallace, J., Klein, R., & Schneider, P. Spelling ability and the
probability texture of English. Journal of Educational
Research, 1968, 61(7), 315-319.

Westermann, G. S. Spelling and writing. Sioux Falls, S.D.: Dimen-
sions Publishing, 1971.

White, 0. R., & Haring, N. G. Exceptional teaching. Columbus,
Ohio: C. E. Merrill, 1976.

White, R. T. Research into learning hierarchies. Review of
Educational Research, 1973, 43(3), 361-375.

Wolking, W. D. Rate of growth toward adult proficiency: Differences
between high and low achievement children. Paper presented at
the International Symposium on Learning Disabilities, Miami
Beach, Florida, 1973.





























APPENDIX A

SPELLING TEST









SPELLING TEST


1. bite

2. and

3. dark

4. bush

5. bags

6. bottom

7. asleep _

8. ate

9. bull

10. four

11. fasten

12. later _

13. cotton

14. cave

15. did

16. hatch

17. rent

18. giant

19. laughing

20. am

21. broom

22. melon

23. dollars

24. deer


NAME:


SCHOOL:


25. cookies

26. march

27. books

28. mice ____

29. her

30. rich

31. glass

32. field

33. though

34. bake

35. alone

36. lily

37. obey

38. pony

39. hot

40. street

41. an

42. patch

43. parties

44. seem

45. bonnet

46. already

47. heap

48. pear




























APPENDIX B

ACADEMIC TASKS















I. SEE CVC TRIGRAM/SAY NONSENSE WORD

meb wid kum vak rop het div len jad nup

zik fap rud tek yab vit suf pid rab nen

ket tav dak fot lub das pif niv dup maf


# correct

# attempted

accuracy

Speed /1 minute


II. HEAR CONSONANT SOUND/SAY CONSONANT NAME

c f h n q t v k y b g m p s j w z d x # correct

r 1 n k h f c b y v t q s p m j 1 r g # attempted

x d z w c q g w 1 f t j z h v m d n s accuracy

Speed /1 minute


III. HEAR SHORT VOWEL SOUND/WRITE VOWEL

e o ai u o e u i a i u

e o u a o e i a u o e

a i u ie o a u e i o a


# correct

# attempted

accuracy

Speed /1 minute










IV. SEE ONE PLACE ADDITION FACT/SAY ANSWER

1+1=2 2+0=2 5+4=9 6+2=8 8+1=9

0+6=6 5+3=8 2+4=6 7+0=7 2+5=7

1+5=6 0+3=3 6+3=9 6+1=7 2+8=10

4+1=5 3+7=10 3+2=5 4+5=9 1+0=1

0+2=2 2+6=8 2+3=5 1+4=5 7+2=9

3+6=9 2+7=9 4+3=7 4+2=6 1+8=9

4+6=10 7+3=10 9+1=10 3+0=3 5+0=5

1+9=10 6+4=10 8+2=10 5+1=6 0+9=9



V. HEAR TWO LETTER BLENDS AND DIGRAPHS/WRITE LETTERS

br th sk ch cr bl sm sh cl

tr tw fl dr sn g1 fr sp pl

gr sl pr st sw tr ch dr br


VI.

c f

r 1

x d


HEAR CONSONANT SOUND/WRITE CONSONANT

h n q t v k y b g m p s j w z d

n k h f c b y v t q s p m j 1 r

z w c q g w l f t j z h v m d n


VII. HEAR SHORT VOWEL SOUND/SAY VOWEL

e o a i u o e u i a i u

e o u i ao e i a u o e


u i e o a u e i o a


# correct

# attempted

accuracy

Speed /1 minute













# correct

# attempted

accuracy

Speed /1 minute





__ correct

____ # attempted

accuracy

Speed /1 minute





# correct

# attempted

accuracy

Speed /1 minute


a i









VIII. HEAR INSTRUCTION/WRITE LETTERS OF THE ALPHABET

A B C E F G H I J K L M # correct

N 0 P Q R S T U V W X Y Z # attempted

accuracy

Speed /1 minute




























APPENDIX C

SEE CVC TRIGRAM/SAY NONSENSE WORD STIMULUS SHEET















C a m LJ co L
cD w




















Y CZ ^ m a- Q
a: :
> > I- U;L- C










C C L









S Z a Q3 Y
C W C r -
CD




















C U-i e -i
SI Q o z





















< Z: F 0-
i Q LL VL


02 F u- F
UJ LI 1- w
N- C r > W )





























APPENDIX D

SEE ONE PLACE ADDITION FACT/SAY ANSWER

STIMULUS SHEET





68








II ii II II II II II II
c c C 0 ) CO 0
+ + + + + + + +
Co C j c- r C LO C












II II II II ii II II II
CM C C -
+ + + + + + + +
S '- O d- i-+ m- m L












C CM CM C CM C\
+ + + + + + + +
in ( I' U m 0 0 or oo














II II II II II II II
o m -I r- --
+ + + + + + + +
CM LO C m CMj CM rC C












C ID C L CM UC O O

+ + + + + + +
- O C C 0 C C- -





























APPENDIX E

GROUP MEANS AND STANDARD DEVIATIONS








c
0



O
4'


aC
C-



o -





vi
C,
0
0-

5-











C)








o ra
ci
0.


D- k. t t C
D Lnn LO











rn rnJ C) CO C0
U) C) NJ U)


C U) mJ U) C) rN- CO Ln C
)- Cx L- U <^ U) OC N L- c-











r- C ID U I C)- CM co CM c) rN-
Cm 'It co C\j C) n LO m C) Ln


EL E

-3 C) 3' C C C: C -C: -C:,- i
0 _0 C C 1 C 0 C. M. CL 0) CL W 4- 4-
1- ra ai (1) 3: 5 0 0
0 0 ai ca >1 >, S 4+ 4-' 0 0
0 0 U ) C) 3) (n > > -
> o -> -> C) i
QJ a;- *->~3 +- > ^ -. o u < j

iU) +) ai iu 4-' 4-1 4-C +jz, Z 4+-
C C r >, >, U U )-0 ,C C) ci) T ^ 4-'- 4-'
0 M) 0 C3 C) CU C: 4-' ci)
C C C) C) -., '-. C- ~ti C- -'. -^.
o 0 U) V) _0 -a c C -:3 -0 fl 0 C) j c
C: C N-~. ^-. C: C 0 0 03U -CIC)1'~. N-. C: 1= 4-'>,4-
c0 0 7) C U~- CUJ a) >,A ,)-U
=^ O r T- C: U C: 0-= -0 7 ^ ,
> > C 0 0 Z- 4-' C iU)C C; C C) c 0 -
C) C) Cs C) U) u) -^ C) C) U U) =3 i) CS -C)
) U) 0 0 0 0 C 0 to 0 aj
E- N 0 o U) 0 ,- i- CO C C i
E M C) c S- C) S- tCU 4) C) 0000
o- 0- C(D C 0 0 C) U-C GC c 0 0 4-' 4-L
0 0 C C C) C) > > U U 4 4-'- '4 -3 oC > > L 0 L 0
C C ) m C) ) c C) C)
S-:,^ 0 >, 0 +1> 4- r >1r ,- 2J .- rD 0 >) 0 +- J SE.3 4- E- 41-
-'U-' U CJU C-UC- u U UU) L( S- u a)IC)4.)C
C) CC) a)0 C 0 CC) c m 0 )-m'A 0
) )0 0 S- L )- C) C)C)0 0 0oC C
C= C))* C) U C) U C)U) C) U C)Q C= C C C)4-'4-' +J->+-' U C) U C)> U) C) U) C)U-C,-C:!
(a UC)C) U C)U W O C -V -- 4 U UC) U
W- (3 j u a u Wu 0 u 0 aj _' u aJ u aj'DQ
U CLS- U L- CL U L U CL U CL L- S- SC- LC S- c- u L--. 0-
Ca) C< ) )n U)IC) ro Un) <1 IU U) (D C) a C a ) aC) m < C) ci aj C) C C)
a) U) a) a U) a) C C) a ) a ) w ) C


C '-t n 't C) COM CD r"-- cj r- C" Co c i0
CM r- L0 U) L,; M, 03 C) Ln NJ ;U) cM N-
OC m i- CM J m Co








C t) U) U 0) icO CD NJ U) cO CO ) 0M or C

- W, co .) CO m m. m-
NJ* .- c) Ni NJ ,- c) CMi nJ c) CMi in cMi 0) N





























APPENDIX F

RAW SCORE DATA











Raw Score Data

The first column (S's) consists of each subjects number

within the good spellers group (100's) or the poor spellers

group (200's). The DEP column is each subjects spelling test

score in percent correct. The three numbers are read as a

one place decimal number (e.g., 99.6).

The last eight columns are the raw scores per subject on

each of the eight tasks. The tasks are:

Code Task

CVC H/S See CVC trigram/Say nonsense word

CONS H/S Hear consonant sound/Say consonant name

SHVL H/W Hear short vowel sound/Write vowel

MATH S/S See one place addition fact/Say answer

BLDI H/W Hear two letter blends and digraphs/Write
letters

CONS H/W Hear consonant sound/Write consonant

SHVL H/S Hear short vowel sound/Say vowel name

ALPH H/W Hear instruction/Write letters of the
alphabet.

The eight numbers for each task are read as two four digit numbers

with a one place decimal (e.g., 09900774 = 99.0 and 77.4). A zero

was used in the hundred's column when a subject's score was less

than 100. The first number in each column is the accuracy score for

the task. The last number in each column is the speed score for the

task.







73










-< U -- 4- I -- o c r- I t- Itn --- Lr -c z Lr I-z r ,ic c oo-- c -C- In r- I rI Z In
a Cr -4DCDC>) C)odDjrI--c[ Lnlff) jrCrU Zn
Sooooo ooooo ooooo oooo ooooo o cooor oo0oo
\i OOOOO OOCJ"O O--O OOO OO" OOO3O t--a D00 00 o,"-CO OOOOO
000oo0 o000- OO o0000 00000 4OO 0000- 0000)
o ooooo oooo 00oooo00 000 rn o ooc) ooooo000
2)00000 OO rO 0O000 00040 1-))) 000-4 04000
0 00000 00 c0 CO~) 0000- CO0 0C'4, O. OC0OC 0,0 )0
TOw 0 e -* 4 ) 0000 OO N w




:I>0000 00-00 04:0 )0 0 00 0 00) 4)00- 0 0 000
;- orr'-Z r-0 0o n-Cr-r- rnc 1-4(o OF- oC- r4'- f-- I rI- -n 1r- 0-r-t.- ) =t,-r- tn 0

C 00C00 CO OC0 COCOO OCC0 o C)00c 00000 CO O000

Z 40r hO1-000 -cOMZ -04C) MnMO4 3-Ch- 4)4O )
S Do oo <-- C C0 < c < C o -o~ -- m OCO:D G DO
\ J \ -2) -.Z r onr--o)4 4 f- "I <10'1- III)4 I C, III -r- -3 'oo-o





Z nr-ti o =z L 0'c =-O 00 )0 0-0 Hoji- 00C L oco t-i NrrJ rf-lIcOu
m) nLn-oo 'a Lnc71 X r'- r rlrJ.- a au-', -i.Cn' rxj I.--rq i CIJ-n Ini ,rl- rR f 1j-O -rx







C O000 0 C000 C0C000 C4C)04 0000O 0CC000 C04CC00
( D O> CM OOCC CDO" < C 100 OOOC D D O> In--O
:K rooo ooooo o aM o o r - t- c, ooo o IooMoo



0 o Zro or0n3 o-oZ c o C-- = oo 0 0-o000 00000 00 r 00
H r^--j z ^TV n rnrfl ry ,n 1 r'irt lt-rZl zrzjN [i olr O.-i r n 1r) rv uf- I' N





H c 00rOt0 o oor om o oin OMICOcs MCOMPOC O o CN or ooo
0 R-NNC-)O C-C))OO -4O)C4C) C- Z C-<- )cl4 )Cr- -D-C CA r InC,

0? 0-00 0 000 40 3 O-C~ CO 0 0_-0040 1- 00000 (00)00

C Zan M -O -ncCoM 4 00p0 1 CP - LI"0"m > 4)CN 4" NN" > 40 OC ) 00 OO: C C) >C D CO COC -0 <>
07z Z n l 4 0 Z -- o flr0 -oZI fl 4CDt'- 0 CI4 ) :C F' C)nw.J--- < 0 -I.- z L
00000 )00'D ) CC043 C4,0C3 4 0C0000 ) 000
4- C0D04)4 0rnror0 O-)OO' I- ir- Zn o-K r-rq orCr-ZnF0 rrv- o co

I 0C-c00 0:O oI` o0 zz -o .O. 003 Cmro




I 4r-0 m Lo r-J -0 z:7rj 4o ) 000 i 0 n 0 OO O 000 C-.-
co InrLOW --j r jrLnRn =Jnn mZn0 tq zr I-tCZn 3 -rm1Ln r r-xInTn L



0o004)0 4404 0009 00)04 C Om44a) oc -0C 4)o4>
0-0)1-'1Zn 4) C n 40 0 0 4) -- )O N- 0 -N- C. 0D Zn Zn 0 0)Z>3' Ln^T 1--0-






, G00 0 0O00C 0O C- 0 -0000 00O 0000 000 C0






\ J-^i1-
00000 0COC00 c000C O30 00000 00240 C 0000C
Zn Zn-Zor- fcDrl-- CI ZnI oD T-z 0000r4)1- -)- n
r lCzl n Ct C, O; 10 a > D 7 M C>
l-l r 4i)n3o1ou .1 0.144)01 vi0rmca 4u fol In r1--1-OJ r 4 4)n 14-In?>









O 0C0 0 03000 0C 000 -) 000 0 <0000 00000 C000 C3

N In-0 OZnLI MIMJ-11)40 r0- ) 4)4) Lf)Zn0- 4)4'n 10Zn~ 0Zn 4)ZY r, In Z I-
Zn tn-r- -)r4)- I--rLf 1-14 c-rZrn)O-Znr r.0r111-ct c--fnm -- (=tn r-n
-- *-nj ruo- colijC0. co In Ln TIl j ---)onrj otnco :a 04 :a -' i^ In zr- z 1-. 4) )t0-N zr :a zT7r.,. Ln zn 0L*-.rnl1-- 04)







I 0 0000400 OOC00 0O000 00000 00 400 000CO
a) .0Zn r04r-q 4) Znr- c t-- r- 04)04 Zn pn cD Z 0 IZ r- Zn-4) C>o oZ




z Ln ->-)o -n ,Qo )D4)4 o .r.2M o-1 u .0







00-00 00-00 400400 0-0-4 O0- C 000<0 OC 000
Zn -cZr-f-t r4)- -oC4)0 -fCUo c101-- -NoCT-n 4)q -V ) Z n no N-Zn Ln4-n r .-'1na, r 4- Z





(3 -n In u oooom oor oo






- No o oooa I'm^-^t-- w,---\ ( nf]~j~fi rninjnrurn," i~~ri

















Ln o ~-ln ru

00000
r-N CO ID


00-00





C0000C







cl 0000c
DN i c:,0 Cl
r-o rc

Hr-on
Ni NN0 Ni
C00.D> -
00000






C0 C) CD
00 NiNN
co o In
mm Nr- o
cooco

0 .00)/


033000









0o0oc No


I,--- 0l =*c


NI ON N3
00000


00000











a-MU -O


mr-- o
- -c CDV
oo.00

N0 000






rl mt
00'000

r-comnFI



mo t-Cr-
LNO 00.

cooco



00000






DoeC or-
000 00






N --0o wN
0.0000



n-<-!0 N3i
oI ---I -n






















oc oo
0O. 00













00000











va4*-:Tm
00000



0.or0r i
1c1r0 or-

0X000









00000







ooo o--


OO0Ni


00000


000 00
zroZno

0 0 NN











lt-*t0



COr x c
u-inom
00000






OIC) 000
COO 00










C rn o ' - C
0000-n



00000






jo 0 0




ajf cmj r







00000




0 (Ni 0 o
co CD
C0NN 00






0000-n
0000-0



00 --.0 C






00000







0o00




00000
r-00 O-










-0000







anjnjourvj


ofD -nor- III
.OLn O^O

00000

r0000



nor M -r-

or-o--ri
. \JniNl>-

C OON

00 Nir In





moco P

coro Ir-

00000





00 000










->000


rNo- 1
00000
















efn o^-I
o00oC00









0Nt-r Ni>
occoo









0t0r-n Cvl
00000
























i- NruO NpD
ooooo
r-NrJ or





00000


0.0000

O~n-o


.-00 LZi 0

00000
Cr-CO0N





p Znoll I--I






co



i 0000


C',-C N- o
ocoo











N 00o CD




Pci C00N
00000







-C--,






NV -0 Niru
C30000













ON 000
mooooo






















D0000
--i0 O -N
00000











0>00>-







00000
c3-o M
o C ~c~


AMW hh
NNNNN 6


r-N
400 O
000 a0









000- 00-



r-oo 'D
0 fOO J(,


NO-. ru
00 oo Ho
o--- 0

Ior-













0-O 00
rlc ooco







N ON i C

mooo

0r00 ri-
000 00






c.OONr' c--3
CooOO






C cOONC












00000
0>0 r
oooi oo














000 00


onjr-0-0o
0 mj0u ru-n

SONO 00
00000


f O a,
ruchnrvjnj
00000
co00NNir--
00> --.
OONONW

0000-0



0cl-ccr-cr


00-00

00000

N3i Ni, 0a
00oo0



co) oo



00000




co 000
00000


COO 0O


00000













00000







0.0.0-.

00000

000 N-fl
00000
r0000








0 NO>
r0 -.00








00o-no
00000
c'MOD*-

c3njnuiin
oc3o3o


(





























APPENDIX G

SCATTER PLOTS





76



The following are scatter plots representing each task, based

on accuracy or speed, and its relationship to the spelling test

score (percent correct). Two correlations are reported. The first

correlation is a pooled correlation. Each group's scores (N=35

per group) are correlated independently and then pooled for an

average correlation. The second correlation is based on the total

sample (N=70).








~ ~ ~ i ', ^ ' l; i; 0 0 .




I I K K 0 0




o o















o~*.
--4

. . . . . .












o 4




































4 4 4.o4
.._,, o .























04 4 4 4 4 4
















... ....0 .-0
4 4 I
.. >t 4 .: o
oo
-o







>- * x '

3 *=> .
















-r ^ro x




S' ; >. xas



( t-


It xy" -




^~ .


_*-- UJ






nu a.
ccx


) J 3


; 0 0 ''


SD3 a












L C C L0 0 0 -..o ... 3 ... .
















































N N 0- 0 ; : T
o o































*2 ***- -N. .+ +- t -.^. *. . ..-. f . .










42 4
x .
















o ,,

U x5
" o ",<













0 0





























0 C 0 (O
x *










,o .o
X 4














So o


.o 20
X









>- -
o X *

Xt *

C- o*












oo











a: o'' x
L U L X *






























,) cc -<
S0 o
1*1 [!*
IE Q> *






-_r d "~ .0+







=^ oo x x o C
r !*lu 0 X-
i-L aa-- *
4 C)Y^ : i^
C- *" x .fa





^ nof. o




0 z 0- < ~ < tf n r
cir x
oj 0
(X *c
n~j a. r x
d.~tl


C', C) 0 .- I


. 7 0






79 -
C C C 0 0 0 0 0 3


0 O O r n "r ? N a o


Q 0
o o


o
o

O


o
o
~1+~a



9





c
o
o

o
rr





o

o
c
o
-o
x




x ~o
o
o
r
x ~rv
o



r
It


o
o


a,
9





o
o
a
r

x





a
a



a
o



c


o
o
o

x
x .e






a
o
-o



r-


ra x




o
0- xx x










CT






xx


















o X
>- M 0


S?


II




~X








X


XX X


X

X X


n. .T
O F















vr 1-^*





























o o-
X iE <
CT3 O- <-



t-l -
^a"i o













a ~ (r (





007 u.





TUI tt-_


0-


S x


o
o












o

---~
9

o
o


...... .......... .........




^, g g g
ff- ^ ~ 0 s ^ /


e r r r





80









































3- - -- .- -- 0 - . 0_ 8 .

a) r +. + / G O O O O O ^ ) ^ ,
oo

































1-- '* L* L



M~ aU




XL L


(0~ i
C o .








r: x C 02


























,_ x x 0
o









., --~~~ , ."~~~~~ o o
*< o o , o*
X X
x' **
0 M X.




- fM X *

oo *
D ~ r -> 4
0 il- X X X*C
v X *



1< rr X X *
X y
r l )/ / *

a I_| *X fy






I. 1^ 0 c 0 f- ^




a "is O X X



rxr _^-< X C











0- 4n i^ v c n n


o_ 0


r <;- rl r' o r- r ^ o ^ r ^ ^ o










C C C C C C G O O O O 2 3 2 ) 2 3 3














o o









o -


o










' 0 X N

























L L


CC 0 0






-.-
o o
>t M~ t"^~--r r l, I
q



















































C 0 C 0 0 0 0 '2 '
X



























-
X

a F t
































-o x x.
X.





































0* o .--- .. .o. . ... . . .







r r F 0 X O 3 Z <
0: x x
U M i ~, L
0 x t i
< a- '
Ui >~ x0
i-x w> : C








LJL Q* ,- T


Q. -* "^ (



(L _t -ro




0 0- I l^ ^ e M -
(n
ex) -i

tta

Y- ul fc o o o o o o o o 3 ^ o o o f






-. 82 --

CCCCtCC00030000033))3


So o o o o o o o

J J C C J ~





2 "" ~~............. ..~....,....... .........




o .e
x

2
































LII
o. L L
o





x. o








C. cn' .-C
o x x
















































Co:- oC
o ( o-









... g:: L +.
0 0

























+J
x *-





x



-o x x x: o

o- Cl x x r-

F TN x x-



L X X t
n < x~o N *



C< .x .
U -^ X XX X

**x x a

o ? J Sc : ~ "

*i -*u ** 0
u- r x -' *


j <*< *- (o





n- c? -r o T3


F"Ui -, <

00 ry~ + /
o- *I~l 0 0 ~



_ij I- -> a w - i O ^

0- ), *-- n> n ^I n r

or

1^ xL


a J O O 0r




.... .. 8 3
'-C C C 0 6 0 0 0 0 0 0 3 3 3 3 3 ) ) 5 3 ;








0 33 3 3 c







4 O



















'-4.- * 4. -
33 4. 4. 4.3J~)~
a
0 0












xx


















.3- a
1 4.0 3.- .3 o. 0tr~.
: 3 X 4- -
S0

































x. 4. 4.
S.3-
o; rr = o




























o 0





















t-xr. C)
73 3 3 0





C Cx C :oC













t3J 0
-4 - o o 0 0
51 s

























3- 3^ 3 3 3
? !-i : oo
Cc o

































o "K <,








0 oo o3 r n = '^ r
0_t Q

an:

. OC C ^ f






84
























































*< 0 00o o 0 o 0 o 0 o
o o o o o o o o o o




oo









































to-' C o
o ~~~.+.~.~ -., ~--+~r ~ r~
f0 9 ^ a !





m~ oi













0 .In
a






o o






















y4-' 4-i
C Ct
o o













































..... . . . . . . . 0
*a-a
o X >













o3 .J
9 x *
x













oo a o


r ru + x i- C
o >

"i X




> r a x X o





















Tr X. *..

-K : xx : x x .
l o 0 0
in n **IL- X















n o o x x o


< J 0 0
U' ni o x X X oX D

L/ti)* .r 1



_>> HJ x -~ -
a" jI s
W/ m 0



V>/ t. 3 X X c 3 t a
0r o- o X^ o a w -
) 10C
o_ o1X
ui 'oL





o .
^. **"-r -- ..-


o" "I







o 0 :






o o x
o d
x *


0 XX .












































0= L,)- o -
x "~ x

















































-m o x *; o ooo o
>- tf X O


























=,X
O -- *
fl l *
a _j*
t ~ *
" -'" 0X C


tr =i>- .
t l'^ M*



10 f *
3-> +nr -


rut _i_ <
C>i _)
_J li'- 0 <-


- 1











O^ -J


ooooooooo
oooouoooo
ooooooooo

3i~~~~~J~~
4~CQYl~nn












































D O
O D
D ~-+--, -, ~~~~.,... . .~..~.~

1
U
4





0 O
0 O
0 O

0 O
0 ~




t

0- O
0 O
0 -J
X
Q g
n o


x

o








x



x.
o x
o















0,x o w
0 *




































xx x






oo


N
II 0 X





n. xx X






*^ x

tt.-* x xx



"--M X Xt

OljJ X XX



cc o x

OS- X X 1XX
*- *

_*-*>







*"'* XX < X

**"*< X X X

-t~ X X



.-X- 0 X X

a~a
ITLJ *




oX

-^^X





CO X X a ^ a a = 3
or c' a. r f '

0n~


o
o
o

N
n






-D

o
r


rr



s

o
o
o



rr






o
9



o







o
o
o
r
o


C
D





n






3






n
w
v,

z
*r








rr
u

r
m.
rc~



c

~1

:LL

W



C
13
a
r,


-U

m
J~r
-J
or
tin
n
Iru
cro



u-
r,

Ilu


oooooooo
oooooooo
Oooooooo



r~on~nru






87















oo ooo oo oo o oo




o o -o w s o r N







o ,































C x










,- 1-
Qr o











0 o

















0 2 0


0 ~ V < V V V V 0O


C) V f Vt C
o- -o-






oo .o
K x X M










































o. xx. o o o
< X 1



XX X M *

0-L XI *





0 X 0



x
<~t~I
0- [ DI** M*C
0 =>j *x D

Ll- <0 *



^ '- *





0 0 X-J *

u>mi o X o


lij 310 0O





m_ (L0>o *" +








LJO (. .*t



*1 O- <*'a a

-3 S S-





[r~r o~ o o i c *



0r~ wo io -~ L wiL ^ f

cn

O J 0l '
7iri a


























































































































O
C
P





D
r*


t



3
L





O
W





W









RJ
CU

c
VI-
o-
rr


OJ


rC)



J
uJ
r;


c
I~


,~
v,






-u
P,
W
III
LY
cM



YC
OJ
N


C X


O
O
O


O
0I








D





D






CI X
O XX
D
r,
0
O



NX


O N
D X
D
X
O
n



c xrurr

xa
o


u~1
wa N


Y1~ ~X


*" ~ ~ XI
r
L~CIO ~ X
rra
Jo
X
O:V1D
~LCU
~LW
I~
n r


E' a
o
,a


D
11N


w
rr
to
a~>o
70
~O


VIO
-X~Y


O



WO
_FIO


lr -,
YW Q
--



II

D
10


m-t
~N tr


rr o

o
r D


L 3
O O

C

O


O
D
O ~_)~~ f~


x

X


X X X


Xx


LiX




X X X
IX I





X


_f~~_~f_~~fl_~_ _~__I___,


000000_
0000000
0000000


53~3~3~
g~~nr~n


C
CU








D
O



O
9









a


d
4






O
D
O
C
O
P






a
a
a


-O
-?






O
O
O


N








9
9
9








C


X O





O
rV



t


a
a
a
r
a
rV



X C


-O
O
O
X















L~r~IN


O O
O O
O


9






























00000o0000
0000000~0O
00000000~0


3~~~03~~~0
D~CQ~OI*N





D 0
D 0
0 -~-tC +~, ~-+-~-, ~~I~~r-C~

O a
o ' a



+ r

or .o
or r .o
O o

o- o
~ P





0 a
0 a
O X


0 a
0 4

XI
X
I:
0 1. D
0 O
0 O
-+
0- O
Cc -C
X
IV X

IXX ii
O X 4
0 4

4
0 9
~ X
X X
II
t L

0 I: O
0 O
0 O
-t X I
0 D
Y) X I~



X I

,0 I a
x

(
t
)0 R
-;r
X

X
I X +

O X o
or -o
O X o

XXX C
O II O
n I CI



X
II
X X a
O' a

nr X t
-o r X 9
~U X I 'U



C

X
o -o
o II
o -o
t t
uo o


''
" "'

"



,,






or nl

rr

~ wr
- r-

o ~I
*r ul

r :1.

u
a c-

riu

i oru

,,
,, ,
Ln,
c

au
3F- -~

lu ~
;-
~ r.
or J


~ Irr




r
,,



n,



n
r*
-u
or
ul~
n
uu
cur



ur
o


(1J




0
+


II


5-






o



Cs










5-


11
JJ-
n.+








L
i

r
O


X
_f__~~_~_I _+~C~_I, _~~_~_2~_~I


0_000000
,00000000
000000000

13~~1~~~0
OOCQn~nrY


0

































0 0 0 00 0 0 0




N N 3 I N' 0 N -


O
O


~I
n






O


O
C
n
n







O
O











^o
xo


oJI-
WCIN
W3



n
~cr o
o
n~ a
\r r
a
-^ N
IU~
TC
W

W1)0
LtO
Or10



F~n~


gm
I
~(O


~W O
Or D
t
CI^~D
-O
II



U
~10


_1U3

LrYI
YR-






O
o
o
W r
JN
m



0.


Z


L
O

C
O
J


O
o
O


Q
n



r

s
o



n
o






o
a
o


o
n






-o


o



-u



r



o


r
-r
rr






r s
s
s

x
ru



c


o
o
o
r
o



x



rr 4




111








D
O


II


0 0O 0 o o o 0 0
0 0 0.0 0 0 0 0 0
0 0 0 0 0


N N N N N N


II [I









m rD
s- S-







o 0









S- S-


0 0
t-) u


X












x
x

IX X









X x







x




X


.. . .
.


X X




X XX



X LI U


XXX


XlrY I

XI






X









































o ... *J-- ..* .* .* *** ... ..- -- ....- -* .... oo f 0






*-* Ma r~i 3n X Ixi o x

















(- 0o

























M o
0 o o

O X~ X X. *














































n J
0- Of












































.. .. 0o o' o o o
rt





















J4
P Po x*o


C; -







O i-~- *








1- 0






LI3 VO-






~. lit-u = 0




i-5 <*~







01 y l ^ 4 m ^ -




^I 0
C/< _1
ai^ ft ,
ir-T P


0000C100000
n000000000
0000~0000~

~135~5~~3~
00CO1~C1N




Full Text

PAGE 1

CONCURRENT ACADEMIC PREDICTORS OF SPELLING PERFORMANCE OF THIRD GRADE CHILDREN By ELLIOTT I. LESSEN A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 197?

PAGE 2

To Linda

PAGE 3

a ACKNOWLEDGMENTS Sincere appreciation and gratitude are extended to my committee members: Dr. William D. Wolking, Chairperson, and Drs. Thomas B. Abbott, Charles Forgnone, Cecil D. Mercer, Paul Satz, and William B. Ware. These gentlemen have extended their personal and professional efforts toward my development as their graduate student. Ms. Leila Cantara, as typist and friend, has been an invaluable asset during my doctoral pursuit. I would like to thank Ms. Judith Sazer-Kuperstein for her foresight and encouragement in this endeavor. Ms. Karen Andrews and Ms. Susan Welch receive my thanks for their assistance in the data collection process. The support, encouragement, help, and friendship of David Westling, Mark Koorland, Terry Rose, and Jim Altman can never be repaid. My family and friends deserve thanks for their support and confidence in me. My wife, Linda, above all others, has shared my ups and downs and tolerated my temperment. It is to her that I dedicate this dissertation.

PAGE 4

TABLE OF CONTENTS ACKNOWLEDGMENTS iii LIST OF TABLES v i i ABSTRACT viii CHAPTER I INTRODUCTION 1 Purpose of the Study 3 Justification of the Study 4 Limitations 5 Threats to Internal Validity 5 Threats to External Validity 5 Definitions of Terms 5 CHAPTER II REVIEW OF RELATED LITERATURE 7 Introduction 7 Traditional Variables 7 Parameters of the Present Study 7 Learning Hierarchies 8 Definition and Characteristics 8 Origins of Learning Hierarchies 9 Validation of Learning Hierarchies 10 Task Selection in the Present Study 10 Task Selection and Hierarchy Validation 11 Related Academic Predictor Variables 11 Reading 12 Phonics 12 Summary 13 Speed of Manuscript Handwriting 13 Spelling Test Scoring Procedure 14 Scoring Units 14 Whole Word 14 Syllable 15 Sound Cluster 15 Letter 15 IV

PAGE 5

Performance Descriptors Accuracy . Speed Repeated Measures of Predictor Tasks Summary . CHAPTER III PROCEDURES 19 Specific Questions 19 Question One 19 Question Two 19 Question Three 20 Subjects 20 Instrumentation 22 Method 25 Administration of the Spelling Test 25 Scoring of the Spelling Test 26 Administration of the Predictor Tasks 26 Scoring the Predictor Tasks 28 Data Analysis 28 Additional Analyses 29 CHAPTER IV RESULTS 32 Question One 32 Question Two 33 Question Three 33 Additional Findings 34 Contribution of Predictor Variables in Accounting for Variance in the Dependent Measure 34 Initial and Predicted Group Assignment 35 Analysis of Single Component, Auditory Input Tasks .... 37 Accuracy and Speed Change 39 CHAPTER V DISCUSSION 40 Task Complexity and Learning Hierarchies 41 Input and Output Modality Structure of the Task 43 Spelling Test Scoring Procedure 44 Performance Measures 44 Repeated Measures of the Tasks 49 Conclusions 51 Recommendations for Future Research 52 REFERENCES 55

PAGE 6

APPENDIX A SPELLING TEST 60 B ACADEMIC TASKS 62 C SEE CVC TRIGRAM/SAY NONSENSE WORD STIMULUS SHEET . . 66 D SEE ONE PLACE ADDITION FACT/SAY ANSWER 68 E GROUP MEANS AND STANDARD DEVIATIONS 70 F RAW SCORE DATA 72 G SCATTER PLOTS 76 BIOGRAPHICAL SKETCH 93 VI

PAGE 7

LIST OF TABLES Table 1 Sex Ratio Per Group 21 2 School Breakdown Per Group 21 3 Parents' Index of Social Position Per Group 22 4 Predictor Variable Contributions 34 5 Accuracy and Speed as Performance Measures 35 6 Accuracy as a Performance Measure 36 7 Speed as a Performance Measure 36 8 The Relationship of Single Component, Auditory Input Tasks to Spelling Performance 38 9 Mean Change Per Task Per Group 39 10 Proficiency Levels Per Task 46 11 Standard Deviations on the CVC, MATH, BLDI, and ALPH Tasks 47

PAGE 8

Abstract of Dissertation Presented to the Graduate Council of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CONCURRENT ACADEMIC PREDICTORS OF SPELLING PERFORMANCE OF THIRD GRADE CHILDREN By Elliott I. Lessen December, 1976 Chairperson: William D. Wolking Major Department: Special Education The focus of the present study was to identify a set of prerequisite tasks, modality relevant to spelling performance, that when based on accuracy and speed as performance measures, would discriminate between good spellers and poor spellers. A review of related literature suggested that (a) tasks used to date have represented more than one basic behavior per task; (b) tasks that have been used are related to, but are not direct replications of the modality input and output structure of spelling behavior; (c) tasks used have not represented the most basic skills assumed by curriculum hierarchies to be prerequisite to spelling competence; and (d) tasks have been assessed by accuracy alone, without referring to speed (fluency) as a viable performance measure. A spelling test was administered to a group of third graders in Alachua County, Florida. From these results, the top and bottom

PAGE 9

25.9 percent (good and poor spellers) were chosen (N=35 per group). Each of these subjects was then assessed on each of eight tasks on three consecutive days. Accuracy and speed scores were obtained for each subject on each task. Discriminant analysis and multiple regression procedures were used to analyze the data. The results of the study indicated that two of the eight tasks used, See CVC trigram/Say nonsense word and Hear two letter blends and digraphs/Write letters, were the best tasks with which to discriminate groups of good and poor third grade spellers. Speed and accuracy scores on the two tasks were found to be nearly equivalent discriminators between the two groups of spellers. The findings of this study indicated that the use of speed and accuracy as performance measures may serve two useful functions. First, speed and accuracy scores discriminate between good and poor spellers. Second, speed and accuracy scores reduce possible misclassifications that may interfere with a child's curricular program. The overall and additional findings have suggested possible alternative research efforts that may help solve the problem of presenting classroom instruction that could promote better spelling achievement. IX

PAGE 10

CHAPTER I INTRODUCTION Spelling is one criterion by which school achievement is measured. Allen and Ager (1965) state that "spelling is an independent skill and should receive specific attention . . . in the school curriculum" (p. 159). There are many characteristics that contribute to spelling performance in children, e.g., intelligence, visual perception. There may be certain prerequisite academic skills, that if known early, could serve to discriminate between good and poor spellers. Such a discrimination could also be helpful in planning remedial and developmental curricular programs. The concern for correctness in spelling and effective instructional methods is not new. Cook and O'Shea (1914) stated, "There is ... a wide-spread belief that graduates of the elementary schools can not spell so well now as they did in earlier times." T. D. Horn (1969) related even earlier reports of such dissatisfaction from the thirteenth century. Many teachers are aware of children who are poor spellers but the reason for their spelling difficulties remains a concern. Although educators are cognizant of correlates related to poor spelling (Russell, 1937; Spache, 1940a; 1940b; 1941a; 1941b) they may

PAGE 11

be unsure of those specific behavioral elements functionally related to spel 1 ing. To date, the research on spelling suffers from numerous limitations. Among these are (a) the lack of replicable procedures (T. D. Horn, 1967); and (b) the use of tasks that do not directly replicate spelling behavior (Westermann, 1971). An example of the latter is the use of tasks to assess spelling which have generally been used to assess reading. The characteristic behaviors specific to spelling and reading are different. Whereas in reading the input modality is visual (See), in spelling the input modality is auditory (Hear) (Fries, 1963; Lerner, 1971). It does not necessarily stand that a See consonant/Say consonant name task provides information which is as relevant to actual spelling behavior as a Hear consonant sound/Say consonant name task. The latter is a direct replication of the modalities used in spelling behavior, while the former can only approximate the spelling behavior by content alone. Emerging from the above limitations in the existing literature is the need for clearly defined and replicable procedures (T. D. Horn, 1967) that make use of modality relevant tasks dealing directly with spelling (Bannatyne & Wichiarajote, 1969; Hanna & Hanna, 1966).

PAGE 12

Purpose of the Study The present study was designed to discriminate between good and poor spelling performance in third grade children. It was based on a combination of unique theoretical and methodological parameters that have not been used in spelling research prior to this study. The unique factors are: (1) the use of tasks that appear to be prerequisite of spelling behavior; (2) the use of accuracy and speed measures as performance descriptors of the tasks; (3) the use of tasks that make use of modality input and output directly related to spelling performance; (4) the use of multiple measures consisting of three samples of each child's performance on each task; and (5) the use of a new scoring procedure for the spelling task itself. The problem, then, was to identify a set of prerequisite tasks, modality relevant to spelling performance, that when based on accuracy and speed as performance measures would discriminate between good spellers and poor spellers. The specific questions asked were: (1) Which tasks, with respect to accuracy, discriminated between good spellers and poor spellers? (2) Which tasks, with respect to speed, discriminated between good spellers and poor spellers?

PAGE 13

(3) Which tasks, with respect to both accuracy and speed, discriminated between good spellers and poor spellers? Justification of the Study Among the reasons indicated as possible evidence of low spelling achievement in the United States are (a) a decline in the systematic teaching of spelling resulting from erroneous interpretations of incidental findings and (b) confusion resulting from grapheme and phoneme relationships (E. Horn, 1960). Prediction studies have not been done in the area of spelling that would help to clarify these issues. What exists in the literature on spelling, however, are concurrent correlational studies that have related various characteristics (e.g., intelligence, demographic information) to spelling competence. The intent of this study was to use tasks that appear to be hierarchical prerequisites of spelling competence using the same input and output modalities as actual spelling performance (Hear input rather than See input). Traditionally, See or reading type tasks have been used. Hear tasks were chosen because their topographical relation to spelling behavior may provide better predictors. Findings from this study may provide educators with a new knowledge base upon which to (a) identify good spellers and poor spellers early; (b) focus on curricular variables related to remediating poor spelling; (c) focus on these curricular variables in accordance with performance measures (i.e., accuracy

PAGE 14

and speed) found to be relevant; and (d) provide information about the utility of an input and output modality structure that is relevant to spelling performance. Whether or not remediation of related deficit skills will directly result in the improvement of poor spelling is subject to empirical research. It is not included within the scope of the present study. Limitations Threats to Internal Validity The internal validity of this study may have been weakened by the instrumentation procedures used (Campbell & Stanley, 1966). This refers to possible scoring errors due to individual data recorder error. Chapter III discusses the procedures used to protect the internal validity of this study. Threats to External Validity The general izability of this study may have been threatened by the lack of total randomization in the selection of the subjects, The reader is cautioned in generalizing the results presented beyond the geographical area in which the study was done. Definitions of Terms Good S peller: The top 25.9 percent (N=35) of the original group (N=135) based on the spelling test score.

PAGE 15

Poor Spel ler : The bottom 25.9 percent (N=35) of the original group ( N=l 35 ) based on the spelling test score. Accuracy : A percentage score based on the number of correct responses divided by the total number of responses made. Speed: The total number of responses made, correct and error, per minute.

PAGE 16

CHAPTER II REVIEW OF RELATED LITERATURE Introduction Traditional Variables Spelling research has traditionally been divided into three areas. These areas are (a) learner variables, (b) curricular variables, and (c) word variables (Cahen, Craun, & Johnson, 1971). The knowledge of learner variables has not proven to be clearly related to remediating poor spelling or to facilitating good spelling (T. D. Horn, 1967). The majority of research on spelling has dealt with curricular variables, e.g., methods and materials (Blair, 1975; T. D. Horn, 1969). There is no teaching approach that appears to best facilitate good spelling. Research on word variables has been concerned with such issues as error analysis, regularity versus irregularity of orthography, and high frequency versus low frequency of word occurrence. Parameters of the Present Study The present study has five parameters. These parameters have generated a new perspective for looking at learner performance on relevant tasks. The literature reviewed will be considered under the rubric of:

PAGE 17

(1) learning hierarchies, as they contribute to task selection; (2) academic predictors and their contribution to (a) the selected tasks and (b) the input and output modality of the selected tasks; (3) a new procedure for scoring spelling tasks; (4) accuracy and speed as separate performance measures of the selected tasks; and (5) repeated measures of predictor tasks. These parameters may permit more precision in the discrimination between good and poor spellers. Learning hierarchies is a theoretical parameter. The new scoring procedure and multiple measurement of tasks are methodological parameters. Performance measures and the contribution of academic predictors are both theoretical and methodological in nature. Learning Hierarchies Definition and Characteristics Learning hierarchies are "descriptions of the relationships of positive transfer among intellectual skills" (Gagne', 1968, p. 4). What they are not, however, are statements of how an individual acquires the knowledge contained in the hierarchy. Each level of the hierarchy generates some quantity and quality that positively transfers to the learning of a "not-previouslyacquired, higher-order capability" (p. 5).

PAGE 18

Learning hierarchies, then, represent the effects of cummulative learning with the greatest amount of positive transfer among elements. Gagne (1968) makes the point that learning hierarchies represent the most effective direction of learning for the entire population of learners, and do not represent strategies unique to individual learners. Origins of Learning Hier archies This study has as one of its parameters the issue of curriculum and is concerned with the selection of tasks that may be sequentially ordered. Gagne' (1970) has stated that content in any one area of learning (e.g., spelling) should be hierarchically ordered so that those simpler concepts and abilities upon which later learning is based be mastered first. Instructional design, or management, can then be based on those conditions of learning (curriculum) that are requisite to each of the hierarchical components. Learning hierarchies represent priority ordered skill sequences. That is to say, based on skill analysis, experience, and the use of other persons' lists, skills are ordered, in many instances, post hoc. One should not infer that this method is in error. However, effort needs to be expended to determine

PAGE 19

10 empirically whether or not these ordered skills transfer positively in a vertical fashion or are independent (Gagne, 1968). Validation of Learning Hierarchies Investigations of the validity of learning hierarchies have yielded inconclusive results (White, 1973). Based on Gagne' s (1962) validation model, research efforts reviewed fell short of being validated. Reasons indicated by White included methodological weaknesses and the lack of the Gagne model to provide a test of hierarchical dependence. White (1973) then discussed a different method for validating hierarchies. The model proved useful with one major limitation. Large numbers of hierarchical elements (k) prohibit the use of the model. This is due to the number of groups of subjects needed to complete the validation procedure which can be arrived at by using k! groups for maximum control or 2(k-l) groups for minimum control. White (1973) concluded his remarks by presenting some modifications of Gagne's (1962) original design. He stated that "it would be a pity if further effort is expended on the production of unreliable or ambiguous data in this [learning hierarchies] area of learning when much more precise work is possible" (p. 374). Task Selection in the Present Study In light of the comments of White (1973) and Gagne (1963; 1970) and the lack of validation of current hierarchies, one is

PAGE 20

11 left with a choice of combining what is already in existence or deriving a new hierarchy that may not fit those already in use. Specific behaviors assumed to be prerequisite to spelling have appeared on curriculum hierarchies (Random House, 1972; Starlin & Starlin, 1972) and in task analyses of spelling (Westermann, 1971). These behaviors have been accepted as necessary for spelling performance. It is from these hierarchies that the tasks used in the present study were selected. Task Selection and Hierarchy Val idation In lieu of a validated hierarchy in spelling, tasks were chosen from already published hierarchies. The tasks selected represent a basic core of skills found at the beginning of each of these published hierarchies. That is not to say because they are in agreement with each other they are valid, but because they are in agreement they represent a consensus of the field at present. The tasks in use in this study afford one a beginning point which needs further work toward empirical validation. Related Academic Predictor Variables This study has used concurrent prediction, with third graders, in an effort to determine a set of tasks, not used to date, that attempts to explain the nature of the relationship of this set of tasks to spelling performance. The following section of the review is limited in scope to those areas of academic concern that are relevant to the selec-

PAGE 21

12 tion of tasks used in the present study. Of concern are tasks that are topographically relevant to spelling and tasks related to speed of manuscript handwriting. Reading Hughes (1953) has stated that "the correlation [+.53] between reading and spelling indicates that . . . they may have some skills in common, but are, to a considerable degree, discrete" (p. 348). Reading and spelling skills, although topographically similar, make use of different input and output modalities. Reading is a See/Say task and spelling is a Hear/ Say or Hear/Write task. Therefore, reading represents a grapheme-phoneme relationship and spelling represents a phonome-grapheme relationship. Skills related to spelling need to be assessed with the modality structure relevant to actual spelling behavior. The utility of such a modality structure will allow one to place particular emphasis not only on the content of related tasks, but to ascertain the importance of the modality structure in predicting spelling performance. Phonics . Studies by Spache (1940a), Russell (1937), Bond (1935), and Monroe (1932) assessed word attack skills related to spelling. Their results indicated that poor spellers were inferior to good spellers on these tasks (i.e., blending letters to form words). Two limitations of these studies that are relevant to the present study were (a) the tasks involved visual

PAGE 22

13 input rather than auditory input; and (b) the tasks encompassed different levels of the same topographic task (i.e., CVC words, CVCV words, CCVC words). Monroe (1932) in her assessment of vowel or consonant difficulties that might have contributed to spelling difficulty, based her conclusions on words read by her subjects. If a child read "these" instead of "those", the error was vowels. If a child read "then" instead of "them" the error was consonants. Tempi in (1954) found phonics based on visual recognition tasks to be correlated +.54 to +.57 with spelling. She did use one recall task, Hear consonant sound and sound combinations/ Write letter(s). This task correlated +.34 with spelling. Although Tempi in used a Hear task, the task included more than one basic behavior (single sounds and sound combinations). Summary . The selection of tasks for use in the present study were modality relevant to spelling as well as representing one basic behavior per task. These tasks were selected because they represent basic skills agreed to be prerequisite to spelling competence. Speed of Manuscript Handwriting The issue of speed in handwriting has been considered as early as 1914, when Freeman found that increasing speed does not necessarily sacrifice the quality of handwriting. The slower the child writes, the slower the language forms are emitted from him. As a child is able to write more automatically

PAGE 23

14 (proficiently), the easier it is for him to communicate with others in written form. Early speed studies have not shown significant differences between poor and good spellers (Cobb, Kincaid, & Washburn, 1918; Russell, 1937). Speed data on third graders suggests that they can write random letters at a mean rate of 41 letters per minute (Stiles, 1974). In another study, Wolking (1973) foynd that third graders can write numbers randomly at a rate of 62 numbers per minute and that by the sixth grade at a rate of 116 numbers per minute. With an increased emphasis on proficiency as a function of accuracy and time, perhaps the issue of speed can be a more clearly defined component of educational designs. The use of speed data in the present study may yield information that will further substantiate its inclusion as a performance measure of academic competence. Spelling Test Scoring Procedure Scoring Units Whole Word . Spelling tests have traditionally been scored using a words correct, as the scoring unit, procedure. Use of this method does not allow for the varying length of each word. Individuals may be penalized for misspelling an entire word, when they may have, in fact, only misspelled an affix. Further analysis for instructional purposes is limited.

PAGE 24

15 Syl Table . The use of syllables as a scoring unit accounts neither for the varying length of words nor for the monosyl labism of some words. Use of this method of scoring subsumes an a priori knowledge of the rules of syllabication by the speller. This is not to infer that spellers do not need to be aware of these rules, but in the early grades (1-3) the more complex rules of syllabication are not stressed (Griffith, Bishop, Brown, Cohen, Islar, Pinder, & Steen, 1971). S ound cluster . Sound clusters have been called phonemes by some authors (Bannatyne, 1973; White & Haring, 1976). Bannatyne makes a distinction between phonemes in the word "cough" (/k/ /ou/ /f/) and optemes and graphemes (/c/ /ou/ /gh/). Despite the nomenclature used, sound clusters may present some problem in that the child may have some of the letters correct but may miss others. In the above example, "c" and "k" are interchangeable phonemical ly, as are "ou" and "aw" and "f", "gh", or "ph". Use of this method does not give an individual credit for parts of the word he has written correctly, because the whole word procedure is still a factor. Letter. White and Haring (1976) suggest that "the letter is the basic unit of spelling behavior" (p. 29). The following example will illustrate the scoring procedure. The word "often" has five letters, thereby having six possible units. The child is given credit for knowing how to start and knowing how to stop the word. Carets are used to indicate the flow of the spelling of the word. "Often," when spelled correctly, looks like

PAGE 25

16 A A A A A A often (six units), but when spelled without the "t" would A A A A look like o f e n (four units). This scoring procedure allows one to look not only at the whole word, syllables, and sound clusters, but at letter sequencing, where many errors seem to be prevalent. Evidence from pilot data not only substantiated this contention, but al lowed the investigator to look at initial, medial, and final consonant placement, vowel placement and letter sequencing. Although this information was not used for purposes of this study, valuable information for teaching considerations is made available. Performance Descriptors Accuracy Performance analysis can only be described as accurately as is allowed for by the performance measures used. Presumably more precise measures permit finer discriminations to be made. Traditional measurement provides one with an accuracy score, be it percent correct of the material presented or percent correct of the material attempted. Accuracy, therefore, is only representative of count, or comparison on an interval scale. Speed Speed allows one to look at the number of behaviors (count) emitted per unit of time. Skinner, in 1932, introduced the concept of speed (count/time) "as a basic unit of direct behavioral measurement" (Pennypacker, 1972).

PAGE 26

17 Speed is a valuable measure because all behavior occurs in a time dimension. Speed, in combination with accuracy, allows for more precise information than is yielded by accuracy alone. Two children may each read the same material with the same accuracy. However, the addition of speed as a performance measure allows one to notice that one child takes two minutes and the other child takes five minutes. Educationally, one cannot, with this added information, state that both children read equally as well . Haughton (1972) has discussed the importance of speed to education. He stated that it is fluency (speed) that allows a child to perform functionally, not accuracy. It must be remembered, however, that the acquisition of functional speed follows the acquisition of accuracy on the same task. This study has made use of both accuracy and speed scores per task, in an effort to assess the unique contribution each may make to the same set of predictor tasks and to assess the joint contribution each may make when used to describe the same tasks. Repeate d Measures of Predictor Tasks Norm-referenced tests are administered once per child. The score yielded represents that child's functioning level at the time of testing. To accomodate for errors in measurement related to such factors as scoring error, illness of the child, and lack of rapport with the examiner, these tests have included in their

PAGE 27

manuals formulae from which may be derived a standard error of measurement (SE ni = a x Kl-r xx ) , where a x = the standard deviation of the test and r xx = the reliability coefficient of the test. It has been suggested by researchers (i.e., Stiles, 1974; White & Haring, 1976) that administration of tasks over a three day period will yield a more reliable measure. In an effort to obtain a more reliable measure from each subject, data was collected on the same tasks for three consecutive days. The literature reviewed suggests that (a) tasks used to date have not been comprised of one basic behavior per task; (b) tasks that have been used are related to, but are not direct replications of the modality input and output structure of spelling behavior; (c) tasks used have not represented the most basic skills assumed by curriculum hierarchies to be prerequisite to spelling competence; and (d) tasks have been assessed by accuracy alone, without referring to speed (fluency) as a viable performance measure that may yield useful information for teaching. The present study has used these parameters in a unique combination that may describe more accurately predictor variables associated with spelling performance.

PAGE 28

CHAPTER III PROCEDURES The focus of this study was to examine a set of concurrent academic predictors of spelling performance. A group of third graders was administered a spelling test. From these results, the top and bottom 25.9 percent (good and poor spellers) were chosen. Each of these subjects was then assessed on each of eight predictor tasks on three consecutive days. Accuracy and speed scores were obtained for each subject on each task. These scores were used as the independent variables in the statistical analysis. The dependent measure was group membership, e.g., good or poor spellers. Specific Questions Question One Which tasks, with respect to accuracy, discriminated between good spellers and poor spellers? Question Two Whick tasks, with respect to speed, discriminated between good spellers and poor spellers? 19

PAGE 29

20 Question Three Which tasks, with respect to both accuracy and speed, discriminated between good spellers and poor spellers? Subjects One hundred thirty-five third graders were administered a spelling test (see Appendix A). The subjects for the study were drawn from three public schools in Alachua County, Florida. The sample initially comprised third graders in attendance at these schools. The schools were selected on the basis of availability. An upper and lower group (N=35 per group) were selected from the original group. Each group, therefore, was comprised of approximately 25% of the original group. These two groups were used for the remainder of the study. The following tables report demographic information about the subjects in the two groups. Table 1 shows the sex ratio for each group. Table 2 shows the school breakdown for each group. Table 3 shows the parents' index of social position for each group (Holl ingshead & Redlich, 1958). It may be noted that there is a disproportionate amount of children from groups five, six, and seven. The distribution of children in this sample, based on parents' index of social position, does not approximate a normal population distribution. This information is included to further describe the characteristics of the sample, however, it was not used in the data analysis of the present study.

PAGE 30

21

PAGE 31

22 Table 3 Parents' Index of Social Position Per Group iildex Not Group 12 3 4 5 6 7 Available Good Poor Totals 5

PAGE 32

23 The spelling task used to obtain the dependent, or criterion, measure included 48 words randomly selected from the Spelling Errors Test 1 1 — I V (Spache, 1955). The test has twelve sections of ten words each. Each section represents a specific error type. Although error types were not analyzed for this study, four words were included from each of the twelve error type groups. In this way, no subject was penalized if, for example, he were to be deficient in one error type. This study has made use of tasks drawn from the Starlin and Starlin (1972) and Random House (1972) curriculum hierarchies and from Westermann's (1971) task analysis of spelling. Selected for presentation in a random order, the predictor tasks used were: (1) See CVC tri gram/Say nonsense word; (2) Hear consonant sound/Say consonant name; (3) Hear short vowel sound/Write vowel; (4) See one place addition fact/Say answer; (5) Hear two letter blends and digraphs/Write letters; (6) Hear consonant sound/Write consonant; (7) Hear short vowel sound/Say vowel name; and (8) Hear instruction/Write letters of the alphabet. Tasks 2, 3, 5, 6, and 7 were employed because (a) they are representative of prerequisite skills assumed to be necessary for spelling competence and (b) they follow the Hear/Say or Hear/Write modality structure of spelling behavior. These tasks, neglected in previous research, needed to be studied so as to assess their relevance to spelling and to the teaching of spelling.

PAGE 33

24 Tasks 1, 4, and 8 were included because of the unique information each might contribute to a more definitive understanding of spelling. If the math task (#4) were to predict or correlate highly with the ability to spell, one might be able to infer that the prerequisite skills for spelling are not specific to the domain of spelling. Inclusion of the alphabet writing task (#8) serves a unique function. As one assesses short vowel and consonant skills, the output is to be obtained in two modes, one oral and the other written. If a child were to have a high rate of oral responding on these tasks, but were to have a low rate of written responding, one might assume that the writing of the response interfers with the performance. This would indicate a performance ceiling on these tasks placed by the child's writing ability. If, on the other hand, a child were to have a higher rate of written responding on these tasks, one might infer that his writing-motor ability is better than his vocal -motor ability. Task #1 , a See/Say (reading) task, is included as it is purported to assess the ability to blend basic sounds (Personke, 1972). Assessing the ability to use phonics generalizations becomes somewhat of a problem. Some words which are phonetically regular, such as "red," may already have been learned as a sight word. Therefore, CVC nonsense trigrams were used.

PAGE 34

25 Method Administration of the Spelling Test All subjects were administered a spelling test (Spache, 1955). Use of a spelling test that had words for second, third, and fourth graders allowed for a broader range of words. The range of words encompassed (a) a base for poorer spellers to respond, and (b) a high enough ceiling for better spellers. The spelling test was administered to each class as a group task. Record forms were provided for each subject on which the subject's answers were written. Each word was presented individually, used contextual ly in a sentence, and then repeated. For each corresponding number on the answer sheet, a card was held up with that number so that each child was able to mark the answer in the appropriate space. Sufficient time was allowed for each child to finish the word called. Directions for the spelling task were: "I am going to give you a spelling test. There will be 48 words in all. There may be a few words that you haven't studied before, but I want you to try to spell these new words the best you can. This is not one of your regular spelling tests. What you do on this test will not affect your spelling grade. Let's start now and see how well you can do. I will read the word, then use it in a sentence, and then say the word again. Wait until I say the word the last time after the sentence before you begin to

PAGE 35

26 write. As we move to each new word, I will hold up a card with the number on your paper so that you can make sure you are writing the word in the correct space. Are there any questions?" An example was used to assist the child to find the appropriate writing space on the paper. Scoring of the spelling test . Scoring of the spelling test words was based on a procedure suggested by White and Haring (1976). Using this procedure, each word was analyzed for the number of units correct. Each word had one more possible unit than the number of letters in the word itself. By using this procedure, units were awarded for beginning the word on the correct letter and for finishing the word with the correct letter. Forty-eight words were selected from the Spache Errors Test (Spache, 1955). There were a total of 224 letters for the 48 words plus 48 additional units according to the White and Haring (1976) scoring procedure. Using this procedure 272 was the highest spelling score attainable. Administra ti on of the Predictor Tasks On three consecutive days, each subject was tested individually in a place designated by the school principal. There were two recorders, the investigator and a graduate assistant, working at all times to facilitate the procedure with as little time interruption as possible for each cooperating teacher. Each task was a thirty-second sample of academic behavior. The obtained scores

PAGE 36

27 were multiplied by two so that the standard measure of behavior, movements per minute, could be used (Lindsley, 1972). Writing paper and stimulus sheets (see Appendix C) were supplied for each task for which it was required. Pencils were also supplied. All subjects used one-quarter inch (.635cm.) lined paper. All stimulus sheets read by the subjects were presented on eight and one-half by eleven inch (21 .59x27.94 cm.) unlined paper and were typed with primary size type. All data recorders had appropriate record sheets on which to record the subjects' responses. All data recorders had a prior training and practice session. For all Hear/Say tasks the directions were: "Listen as I say each vowel (or consonant) sound. As I say each sound, I want you to tell me the name of the letter. Are you ready?" Each Hear/Write task was administered in the same manner except that the responses were written. Short vowel sounds were "a" as in apple, "e" as in elephant, "i" as in indian, "o" as in ox, and "u" as in umbrel la . For the two See/Say tasks the instructions were: "I want you to look carefully at each nonsense word (or addition fact). When I say go, I want you to start here (point) and tell me what the word is (or what the answer is). I want you to go across the paper (point to the arrow at the top of the sheet and follow it across the paper with your finger). Are you ready?"

PAGE 37

28 The instruction for the task of writing letters of the al phabet was : "On this sheet of paper I want you to write all the letters of the alphabet from 'a' to 'z 1 . If you should finish before I say stop, go back to 'a' and try and write as many more letters as you can. I would like you to print the letters (manuscript). Are you ready?" Scoring the Predictor Tasks . Each of the eight predictor tasks was administered on three consecutive days. Each recorder was responsible for scoring all oral response tasks. The investigator scored all written tasks. Pilot data collected showed that it was easier to correct the written tasks after the task administration. Each task presented two mean scores. The accuracy score was a mean score based on the number of responses correct divided by the total number of responses per task attempted. The second score was a speed score. A mean speed score represented the average of the total number of responses, correct and error, attempted per subject on each task. Data Analysi s The Biomedical computer Programs BMD 07M (Dixon, 1973) was accessed for a stepwise discriminant analysis procedure. The program allowed for the use of a linear combination of measures from which those measures contributing most to the variance of

PAGE 38

29 the dependent measure were kept in the prediction equation. The use of a two group discriminant analysis allowed for the dependent variable to be dichotomous, in other words, to represent group membership. The two-group discriminant analysis procedure provided a classification schema by which an individual, based on predictor variable scores, was assigned membership into that group whose mean score he most closely approximated. Three free-entry subproblem routines were performed. The first routine was an analysis using both accuracy and speed measures of the eight tasks. The second analysis used only accuracy scores and the third analysis used speed scores. Additional Analyses A second program (BMD 07M, Dixon, 1973) was accessed based upon the results of the first analysis. This second program, also with three subproblem routines, used only those tasks from the first program that were found to be the most reliable discriminators. Three prediction equations were obtained from the three subproblem routines of the second program. The results of these equations, with a child's score placed in the formulae, compared to the cut-off score, represent an individual's predicted group membership, as a function of accuracy scores, speed scores, and both accuracy and speed scores. Constants and coefficients for each equation are different due to the contribution of the specific performance measures used.

PAGE 39

30 Additional information yielded by the BMD 07M and BMD 02R programs (Dixon, 1973) will be discussed in Chapter IV. This information includes variance accounted for in the dependent measure, e.g., group membership, by all of the predictor variables (see page 23) and possible misclassification of individuals to each group. The BMD 02R program (Dixon, 1973) was accessed a second time. The variables included were (a) Hear consonant sound/Say consonant name (CONS H/S); (b) Hear consonant sound/Write consonant (CONS H/W); (c) Hear short vowel sound/Say vowel name (SHVL H/S); and (d) Hear short vowel sound/Write vowel (SHVL H/W). The math and alphabet writing tasks were not included. Two other tasks were not included. They were: See CVC trigram/Say nonsense word (CVC) and Hear two letter blends and digraphs/Write letters (BLDI). Therefore, the four single component, auditory input tasks (consonant or short vowel; Hear/Say and Hear/Write) were entered in the program. The use of only these four tasks was for the purpose of analyzing their contribution as discriminating variables separate from (a) more complex tasks (CVC and BLDI) and (b) tasks not directly relevant to the development of spelling behavior. Three free-entry subproblem routines were accessed. Accuracy scores were used in the first subproblem. Speed scores were used in the second subproblem. Accuracy and speed scores in combination were used in the third subproblem. Average accuracy and speed improvements were calculated for each subject. This was done by subtracting the day one accuracy or speed score from the day two score. The day two score was then

PAGE 40

31 subtracted from the day three score. These two results were then added together. The sum was divided by two. The resultant score was the average accuracy or speed improvement per subject per task. These scores were pooled to calculate the mean improvement per group per task.

PAGE 41

CHAPTER IV RESULTS The present study was conducted to generate a set of predictor tasks that would discriminate between a group of good spellers and a group of poor spellers. This set of tasks was based on accuracy, speed, and accuracy and speed scores in combination. Specific questions asked were: (1) Which tasks, with respect to accuracy, would discriminate between good spellers and poor spellers? (2) Which tasks, with respect to speed, would discriminate between good spellers and poor spellers? (3) Which tasks, with respect to both accuracy and speed, would discriminate between good spellers and poor spellers? Question One Results from the discriminant analysis procedure using accuracy scores showed that two variables, (1) See CVC trigram/ Say nonsense word (CVC-A) and (2) Hear two letter blends and 32

PAGE 42

33 digraphs/Write letters (BLDI-A) were reliable predictors significant at the .01 level. The following discriminant prediction equation resulted: Y = -15.7142 + 0.1805 (CVC-A) + 0.0768 (BLDI-A) By inserting a child's CVC-A and BLDI-A scores in the above equation the resultant score can be compared to the cut-off score of +0.0034, above which a child would be predicted to be a good speller and below which a child would be predicted to be a poor speller. Question Two Results from the discriminant analysis procedure using speed scores showed the two variables, (1) See CVC trigram/Say nonsense word (CVC-S) and (2) Hear two letter blends and digraphs/Write letters (BLDI-S) were reliable predictors significant at the .01 level. The following discriminant prediction equation resulted: Y = -16.7490 + 0.2553 (CVC-S) + 0.3354 (BLDI-S) By inserting a child's CVC-S and BLDI-S scores in the above equation, the resultant score can be compared to the cut-off score of +0.0002, above which a child would be predicted to be a good speller and below which a child would be predicted to be a poor speller. Question Thr ee When the eight tasks were entered into the discriminant analysis using both accuracy and speed scores for each task, four tasks were found to be reliable predictors. Significant at the .01 level were,

PAGE 43

34 (1) See CVC trigram/Say nonsense word (CVC-S) and (2) See CVC trigram/Say nonsense word (CVC-A). Significant at the .05 level were (1) Hear two letter blends and digraphs/Write letters (BLDIS) and (2) Hear two letter blends and digraphs/Write letters (BLDI-A). The following discriminant prediction equation resul ted: Y = 23.6238 + 0.1818 (CVC-S) + 0.1329 (CVC-A) + 0.2560 (BLDI-S) + 0.0532 (BLDI-A) By inserting a child's CVC-S, CVC-A, BLDI-S, and BLDI-A scores in the above equation the resultant score can be compared to the cut-off score of -0.0032, above which a child would be predicted to be a good speller and below which a child would be predicted to be a poor speller. Additional Fi ndings Contribut ion of Predictor Variables in Accounting for Variance in the Dependent Measure Additional information yielded by accessing the BMD 02R stepwise regression program (Dixon, 1973) is presented in Table 4. Table 4 Predictor Variable Contributions Performance Variables in Order of Variance Accounted for in Measure Entrance in the Program the Dependent Measure Accuracy CVC-A(76.3%) ,BLDI-A(3%) Remaining 6 Variables Speed CVC-S(76.6%) ,BLDI-S( 2 . 7%) Remaining 6 Variables Accuracy CVC-S(76.6%) ,CVC-A(6.5%) , and Speed BLDI-S(0.9%),BLDI-A(0.7%) Remaining 12 Variables 79

PAGE 44

35 Due to the use of free-entry subproblem routines, the first variable entered accounted for unique and overlapping variance. Each subsequent variable entered represented an incremental amount of unique and overlapping variance not already accounted for by a previously entered variable. The results presented in Table 4 for accuracy alone and for speed alone show that the CVC task was the major contributor to variance accounted for in the dependent measure by each performance descriptor. The CVC-S score accounted for 76.6% of the variance, which was slightly more, .3%, than the CVC-A score (76.3%). Therefore, when accuracy and speed scores were entered into the stepwise regression program simultaneously, the CVC-S score entered first, showing both unique and overlapping variance attributable to the CVC-S score. The decrease in variance accounted for in the dependent measure by the CVC-A score from 76.3%, when entered with accuracy scores, to 6.5%, when entered with both accuracy and speed scores, is reflective of the overlapping amount of variance shared by the CVC-S and CVC-A scores. Initial and Predicted Grou p Assignment Each subject was assigned initially to either the good spellers group or the poor spellers group based on his/her spelling test score. The range of scores for the good spellers group was from 85.3% to 99.6% correct. The range of scores for the poor spellers group was from 7.7% to 49.5% correct. Results of the analysis indicate that on the basis of the best predictor variable scores, certain individuals would have been misclassified with respect to their initial group assignment. Tables 5, 6, and 7 are shown in terms of initial assignment and predictor variable assignment.

PAGE 45

36 Initial Assignment Table 5 Acc uracy and Speed as Performance Measures Predicted Assignment Good Poor Good Poor 34 1

PAGE 46

37 Initial Assignment Table 7 Speed as a Perfor mance M easure Predicted Assignment Good Poor Good Poor 33

PAGE 47

33 Analysis of Single Component, Auditory In put Tasks A supplementary analysis was conducted which included the four single component, auditory input tasks (CONS H/S, CONS H/W, SHVL H/S, and SHVL H/W). The CVC and BLDI tasks that were found to be the best predictors in the original analysis were not included. The reason for not including the CVC and BLDI tasks was that their relationship to spelling performance was already known to be significant. The See one place addition fact/Say answer (MATH) and Hear instruction/Write letters of the alphabet (ALPH) tasks were also not included in the present analysis. This was done for two reasons. First, the HATH and ALPH tasks were found not to be related to spelling performance in the overall analysis. Second, the MATH and ALPH tasks are not curriculum tasks directly relevant to the development of spelling behavior. In order to ascertain the relationship of single component, auditory input tasks (CONS H/S, CONS H/W, SHVL H/S, and SHVL H/W) to spelling performance, a separate analysis was performed. The results are indicated in Table 8. Table 8 The Relationship of Si ngle C omponen t, Audito ry Input Tasks to Spelling Performance Performance Variables in Order of Variance Accounted for Measure Entrance in the Program in the Dependent Measure Accuracy SHVL H/W (31.5%) 31.5% Speed SHVL H/W (52. 1%) 59.7% CONS H/W (5.4%) SHVL H/S (2.2%) Accuracy SHVL H/W-S (52.1%) 64.0% and Speed SHVL H/W-A (5.7%) CONS H/W-S (6.2%)

PAGE 48

39 All variables except one were significant at the .01 level. The SHVL H/S tasks with speed as a performance measure was significant at the .05 level . Accuracy and Speed Change Each of the eight tasks was administered on each of three consecutive days. The group means presented in Table 9 represent each groups average change over the three day period. Both groups showed an increase in speed scores. The good spellers group showed a decrease in accuracy on the ALPH task. The poor spellers group showed a decrease in accuracy on the CVC and SHVL H/S tasks. Table 9 Mean Cha n ge Per Task Per Group TASK GOOD GROUP MEAN POOR GROUP MEAN ACCURACY SPEED ACCURACY SPEED (% Correct) ( Response/Mi n:) {% Correct) (Response/Min. ) CVC

PAGE 49

CHAPTER V DISCUSSION The focus of the present study was to identify a set of prerequisite tasks, modality relevant to spelling performance, that when based on accuracy and speed as performance measures, would discriminate between good spellers and poor spellers. A review of related literature suggested that (a) tasks used to date have represented more than one basic behavior per task; (b) tasks that have been used are related to, but are not direct replications of the input and output modality structure of spelling behavior; (c) tasks used have not represented the most basic skills assumed by curriculum hierarchies to be prerequisite to spelling competence; and (d) tasks have been assessed by accuracy alone, without referring to speed (fluency) as a viable performance measure. A spelling test was administered to a group of third graders in Alachua County, Florida. From these results the top 25.9 percent (good and poor spellers) were chosen ( N=35 subjects per group). Each of these subjects was then assessed on each of eight tasks on three consecutive days. Accuracy and speed scores were obtained for each subject on each task. 40

PAGE 50

41 Two of the eight tasks used, See CVC trigram/Say nonsense word and Hear two letter blends and digraphs/Write letters, were the two tasks with which to discriminate groups of good and poor third grade spellers. The use of accuracy and speed scores in combination revealed that speed was the better performance measure for discriminating between the two groups of spellers. The results will be discussed under the following headings: (1) Task complexity and learning hierarchies; (2) Input and output modality structure of the task; (3) Spelling test scoring procedure; (4) Performance measures; (5) Repeated measures of predictor tasks; (6) Conclusions; and (7) Recommendations for future research. Task Complexity and Learning Hierarchies The tasks used in the present study were selected from spelling skill hierarchies and task analyses. The content of six of the tasks followed the developmental progression of skills found in the hierarchies. Two tasks were single component tasks (CONS and SHVL). The findings indicated that at the third grade level, the more complex skills are better at discriminating between good and poor spellers than the single component tasks.

PAGE 51

42 Gagne (1965) suggested an order assigned to skills within a hierarchical structure (simple responses, chains or multiple discriminations, classifying, use of principles or rules, and problem solving). The CONS and SHVL tasks seem to be an example of chains or multiple discriminations. The CVC task seems to be an example of the use of principles or rules. For the third grade subjects included in this study, simple response tasks (CONS, SHVL) were not found to discriminate reliably between good and poor spellers. One may conclude, then, that the simple tasks do not account for a significant amount of the variance in the spelling task itself. The CVC and BLDI tasks, or the more complex tasks, were the tasks that were found to discriminate reliably between the groups of good and poor spellers. An additional analysis that included the four single component tasks (CONS H/S, CONS H/W, SHVL H/S, and SHVL H/W) lends support to this line of reasoning. The SHVL tasks were the best discriminators, regardless of performance measure. The overall analysis indicated that the four tasks, with content relevant to spelling behavior (CVC, BLDI, SHVL, and CONS), followed a descending order from complex to simple. The order was CVC, BLDI, SHVL, and CONS. Although this study was not designed to validate any one learning hierarchy, the results may be serendipitous in their outcome. Of the four tasks based on spelling curriculum hierarchies, the CONS tasks were least successful in dis-

PAGE 52

43 criminating between good and poor spellers. The SHVL tasks were slightly more successful. The BLDI task was more successful yet. The CVC task, the more complex task according to the hierarchies, was the most successful task to discriminate between good and poor spellers. Input and Output Modality Structure of the Task One of the two tasks found to be the best discriminators of third grade spelling performance, Hear two letter blends and digraphs/Write letters was a direct replication of the modality structure of actual spelling behavior. The CVC task had the modality structure of a reading task (See/Say). The contribution of the BLDI task to the variance of the dependent measure ranged from 0.7% to 3.0% depending on the performance measure(s) used (See Table 4). The results indicated that although its contribution was statistically significant, the practical significance of its contribution in explaining variance accounted for in the dependent measure, is clearly limited. Previous studies in the spelling literature have not made use of a direct replication of the modality structure of actual spelling behavior (Bond, 1935; Monroe, 1932; Russell, 1937; Spache, 1940a; Wallace, Klein, & Schneider, 1968). An attempt was made to ascertain whether or not auditory input tasks may have been more related to spelling behavior than visual input tasks. However, the CVC task, a reading task, was found to be

PAGE 53

44 the task that best discriminated between the good spellers and poor spellers. On the basis of the results of this study, one must agree with Monroe's (1932) statement that The correlation [between reading and spelling is] so high ... we must be measuring an achievement which is greatly dependent either upon reading or upon the same factors which underlie the ability to read. (p. 13) S pell in g Test Scorin g Procedure A new procedure for scoring the spelling test (White & Haring, 1976) was used in this study. In a post hoc analysis, the spelling tests were rescored using the more traditional words correct procedure. A Pearson Product Moment Correlation was performed between units correct (White & Haring) and words correct. A correlation of +.97 was obtained. The results of the correlation procedure indicated that either scoring procedure could have been used in this study. The preference of using the White and Haring (1976) scoring procedure would be to learn more about a child's spelling behavior because of the extensive word analysis that would be obtained. The whole word procedure would be preferable where scoring time was a primary consideration. Performance Mea sures The results of this study were obtained by considering the original set of eight tasks in three ways. Accuracy scores were used in the first subproblem. Speed scores were used in the second subproblem, accuracy and speed scores in combination were used in the third subproblem. Regardless of the performance measure used, the CVC and BLDI tasks were found to be the best discriminating tasks

PAGE 54

45 in the same order (CVC, BLDI). However, when accuracy and speed scores were used in combination in a free-entry subproblem routine, the speed scores on each task (CVC, BLDI) accounted for more variance in the dependent measure than the accuracy scores (See Table 4). Accuracy scores are not to be interpreted as less powerful than speed scores but as representing incremental variance attributed uniquely to each of the successive variables entered into the prediction equation. The near equivalence of accuracy and speed scores entered by discrete performance descriptors disappears when the two are entered simultaneously. This phenomenon is evidenced as the stepwise regression program selects the variable that is most highly correlated with the dependent variable. This first variable, and each successive variable, enters showing unique and overlapping variance accounted for in the dependent measure. Each successive variable, then, is selected to enter on its contribution of variance in the dependent measure. An analysis of single component, auditory input tasks (CONS, SHVL) indicated that speed scores were again slightly more powerful in their ability to discriminate between good and poor spellers than were accuracy scores (See Table 8). Speed and accuracy scores on the tasks were found to be equivalent discriminators between the two groups of spellers. An indication that remediation or development of the skills on these tasks needs to include both accuracy and speed before final proficiency can be noted. This indication is suggested by the fact that speed and accuracy were equivalent performance measures. Based on a mean score of the good spellers group (N=35) on each task, the following accuracy and speed scores may be helpful in determining proficiency levels of the tasks.

PAGE 55

46 Table 10 Proficiency Lev els Per Task TASKS MEAN SCORE CVC (accuracy) 91.6% correct CVC (speed) 63.2 responses per minute BLDI (accuracy) 86.5% correct BLDI (speed) 24.0 responses per minute Speed and accuracy scores are also relevant to the issue of possible misclassifications. Children predicted to be good or poor spellers and who are indeed good or poor spellers will receive appropriate curricular instruction. Children predicted to be poor spellers who are indeed good spellers will certainly not suffer from extra curricular intervention. However, children predicted to be good spellers who are indeed poor spellers may suffer from lack of instruction. The lack of instruction for any child that needs it may cause teacher and learner frustration. The misclassification of this last group of children, those predicted to be good spellers but who are not good spellers, needs to be minimized. The results of this study indicated that when accuracy and speed scores in combination or speed scores alone are used as performance measures, no children are misclassified by being predicted as good spellers when they are indeed poor spellers (See Tables 5 and 7). The small percentage of misclassifications

PAGE 56

47 by accuracy scores appears to be a reflection of the N used in this study. Therefore, speed, accuracy, or a combination of speed and accuracy could be used to reduce misclassifications that could be harmful to the learner. Group standard deviations on each task presented an interesting descriptive finding. On four of the tasks included in this study (CVC, MATH, BLDI , and ALPH) the poor spellers group displayed more variability with respect to accuracy than did the good spellers group. However, on these same four tasks, the good spellers group displayed more variability with respect to speed than did the poor spellers group. Although standard deviations for each group on each task are presented in Appendix E, Table 11 presents that standard deviation data relevant to the CVC, MATH, BLDI, and ALPH tasks. Table 11 Standard Deviations on the CVC, MATH, BLDI, and ALPH Tasks TASK STANDARD DEVIATION STANDARD DEVIATION GOOD GROUP POOR GROUP CVC (accuracy) 10.0 22.0* CVC (speed) 16.1* 7.4 MATH (accuracy) 4.8 23.0* MATH (speed) 12.6* 8.7 BLDI (accuracy) 11.3 27.4* BLDI (speed) 5.6* 3.2 ALPH (accuracy) 1 .0 2.9* ALPH (speed) 14.0* 7.8 ''more variability

PAGE 57

48 One may infer from these findings that the poor spellers group, with a wider dispersion of accuracy scores and less variability in speed scores, has not completed the basic stage of acquiring accurate responses. The good spellers group, however, was less variable on accuracy scores. Their accuracy scores may indicate mastery of the tasks included for this study. However, the variability in their speed scores leads one to believe that they have not, as a group, reached a fluent level of performance on these four tasks. It has been suggested (Haughton, 1972; Starlin, 1972) that although the final proficiency aim for any child on any task is accurate responding at an acceptable speed, accurate responding needs to be in a child's repertoire before accurate responding at an acceptable speed can be obtained. In this study, the poor spellers, as a group, were responding neither accurately nor fluently. The lack of variability in their low speed scores suggests that as a group their speed scores may be rather consistent because they have not yet mastered the stage of accurate responding. This would account for an overall slow speed of responding on these tasks. The good spellers group, because of the lack of variability in their high accuracy scores, seems to have mastered the stage of responding accurately. However, the variability in their speed scores leads one to believe that as a group they are not yet fluent on these four tasks. In summary, the findings of this study indicated that the use of speed and accuracy as performance measures may serve two

PAGE 58

49 useful functions. First, speed and accuracy scores discriminate between good and poor spellers. Second, speed and accuracy scores reduce misclassifications that may interfere with a child's curricular program. Repeated Measures of the Tasks Repeated measures over three days for each task were obtained to represent a more reliable measure for each subject's performance on each task. Group means are reported in Table 9. Both groups of spellers increased their speed scores over the three day period. The good spellers, as a group, did not improve their accuracy scores on the ALPH task, due to the ceiling effects of the task. The poor spellers, as a group, did not improve their accuracy scores on the CVC and SHVL H/W tasks. Low increases in accuracy scores for the subjects involved may have been due to four reasons. First, some subjects maintained 100% accuracy over the three day period. Therefore, due to ceiling effects, their accuracy scores could not increase, while their speed scores may have increased. On the CVC task, subject #103 remained at 100% accuracy for the three days while the subject's speed score change was +20.9 responses per minute over the three days. Second, some subjects maintained a 0.0% accuracy score over the three day period. Subject #230 maintained a 0.0% accuracy for three days on the BLDI task while the subject's speed score changed by one incorrect response per minute.

PAGE 59

50 Third, some subjects, while increasing their speed scores, lowered their accuracy scores. For example, on the CVC task, subject #116 had speed scores of 48, 60, and 85.8 responses per minute for the three days. This subject's accuracy scores were 100, 100, and 97 percent correct for the three days. As the subject's speed increased the chance for error increased. Therefore, due to a single error on day three, this subject's average accuracy change was -1.5% while the average speed change was +18.9 responses per minute. Fourth, some subjects who may have had an increase from day one to day two may have had a decrease from day two to day three or vice versa. Depending on the differences between the days, some subjects may have shown no change and others an increase or decrease. Subject #117 had a -6 speed increase from day one to day two and a +12 speed increase from day two to day three, with an overall speed change of +3 (CVC task). On the BLDI task, subject #225 had a speed change of +8 from day one to day two and a speed change of -8 from day two to day three. This subject's overall change in speed on the BLDI task was zero. The results of these findings indicated a positive change in speed on all tasks for both groups. The good spellers group exhibited a positive change in accuracy scores on seven of the tasks. The poor spellers group exhibited a positive change in speed on six of the tasks. Each group exhibited change on each task over the three day period. Although the changes were generally not large, it was evident that two or three samples

PAGE 60

51 of each task was a closer approximation of each child's maximum capability on each task. Therefore, the contention that three days of data collection as a more reliable measure of a subject's performance than one day of data collection was upheld. Conclusions The focus of this study and the specific questions asked were stated in terms of identifying a set of prerequisite tasks that would best discriminate between good spellers and poor spellers. The set of tasks used in this study was discussed in terms of accuracy and speed as performance measures both separate and in combination. The results of the data analysis indicated that when accuracy and speed were used as separate performance measures, the same two tasks were found to be the best discriminators between the groups of good and poor spellers. The two tasks were See CVC trigram/Say nonsense word and Hear two letter blends and digraphs/ Write letters. When accuracy and speed scores were used in combination as performance measures, the CVC and BLDI tasks again were found to be the best discriminators between the two groups of spellers. However, when used in combination, the speed score for each task was a slightly more powerful discriminator than was the accuracy score due to the free-entry variable selection (See previous discussion) . Five of the tasks selected for inclusion in this study were direct replications of the input and output modality structure of spelling behavior. One task was a reading task; the other two tasks were a math task and an alphabet writing task. Although

PAGE 61

52 the BLDI task, a Hear/Write task, was one of the two best tasks in discriminating between groups of good and poor spellers, it's contribution to the variance accounted for in the dependent measure was extremely small (See Table 4). The CVC task, a reading task, was the best discriminating task. Therefore, the use of tasks involving the input-output modality structure that directly replicates spelling behavior was less reliable than the reading task in discriminating between the two groups of spellers. Supplementary analyses of the data indicated that the use of speed, as a performance measure, classified children for curricular programming with slightly less liability than the use of accuracy. Speed was also the performance measure that showed the greatest change over the three days of data collection. The spelling test scoring procedure used in this study was found not to be time saving. However, it presented a more extensive analysis of a child's spelling performance than the words correct procedure. Although this study did not attempt to validate any one spelling skill hierarchy, the findings supported a sequence of the tasks used, from simple to complex. Recommendations for Fut u re Research The present study has included multiple regression and discriminant analysis procedures. Through the use of these procedures, this investigator has obtained information that is useful in (a) discriminating between groups of good and poor spellers and (b) discussing the most reliable discriminating tasks. The findings

PAGE 62

53 have been focused on issues such as (a) task complexity and learning hierarchies; (b) the input and output modality structure relevant to spelling behavior; and (c) the utility of accuracy and speed as performance measures. The resultant information from this study provides one with a basis upon which a second stage of more functionally oriented research may be based. This second stage of research, then, should consist of studies that are concerned with the functional relationship of the present findings to classroom spelling curricula. This task (BLDI) had the modality structure that directly replicated spelling behavior. The BLDI task was less reliable as a discriminator than the reading task. It would be interesting to include the Hear/Say and Hear/Write modality structure as a unique experimental variable. Efforts could then be made to determine if curricular instruction, with an emphasis on these modality structures (Hear/Say or Hear/Write), would enhance the ability to spell. The curricular emphasis in many classrooms is on reading. There are those children, however, that even with reading instruction are poor spellers. Perhaps an emphasis on related content material, with the modality structure directly replicate of spelling behavior, would improve this poor spellers group's ability to achieve competence in spelling. The present findings have demonstrated the utility of speed (fluency) as a viable performance measure that needs to be emphasized equally with accuracy in classroom instruction. The results, using speed and accuracy, were similar. Speed, as a

PAGE 63

54 performance measure, has not been used before and may be a novel means by which variance in performance can be explained. Research efforts are needed that will use speed as an aim for skill mastery. These research efforts need to consider the efficacy of remediating deficit skills (i.e., CVC, BLDI) to a level of fluent mastery. Once these skills have been mastered, a determination of their effectiveness in improving spelling performance may be conducted. It is indicated from the findings that for the third graders included in this study, the more complex tasks (CVC, BLDI) discriminated best between the two groups of spellers. Further research efforts at different grade levels (i.e., one, five) may demonstrate that different skills discriminate at different grade levels or that by a certain grade level, the same tasks can be used as discriminators. This may well serve to provide a developmental sequence of skills. Performance on this sequence could alert teachers to possible concomitant performance in spelling. As stated earlier, if other research efforts suggest that mastery of deficit skills is effective in improving spelling, a criterion performance sequence may have excellent curricular merit. The present study has answered the specific questions for which it was designed. The overall findings as well as the additional findings have suggested possible alternative research efforts that may help solve the problem of presenting classroom instruction that could promote better spelling achievement.

PAGE 64

REFERENCES Allen, D. , & Ager, J. A factor analytic study of the ability to spell. Educational and Psychological Measurem ent, 1965, 25, 153-161. Bannatyne, A. Reading: An auditory vocal process . San Rafael, Cal.: Academic Therapy Publications, 1973. Bannatyne, A., & Wichiarajote, P. Relationships between written spelling, motor functioning, and sequencing skills. Journal of Learning Disabilities , 1969, 2(1), 4-16. Birch, H. G. , & Belmont, L. Auditory-visual integration, intelligence, and reading ability in school children. Perceptual and Motor Skills , 1965, 20, 295-305. " Blair, T. R. Spelling, word attack skills; WRIC/RCS report. Reading Teache r, 1975, 28(6), 604-605. Bond, G. L. The auditory a nd speech characteristics of poor r eaders . Contributions to Education, No. 657. New York: Teacher's College, Columbia University, 1935. Cahen, L. S., Craun, M. J., & Johnson, S. K. Spelling difficulty a survey of the research. Review of Educational Research , 1971, 4J_(4), 281-301. " """' """' Campbell, D. T., & Stanley, J. C. Exp erimental and quasi experimental designs for research . Chicago: Rand McNally, 1966. Cobb, M. E., Kincaid, M. , & Washburn, M. E. Further studies of the verbal abilities of poor spellers. American Journal of Psychology , 1918, 1_9, 331-332. Cook, W. A., & 0'Shea, M. V. T he c hild a nd his spe lling . Indianapolis: The Bobbs-Merrill Company, 1914". Dixon, W. J. (Ed.). BJomj^djcaJ_a^mp_uJ;j^j^ Berkeley: University of Cal ifornia" Press ," 1973. Freeman, F. N. An experimental analysis of the writing movement. Psychol ogic al Monographs , 1914, V7_, 4. 55

PAGE 65

56 Fries, G. C. Linguistics and reading . New York: Holt, Rinehart and Winston, 1963. Gagne, R. M. The acquisition of knowledge. Psychological Review 1962, 69, 355-365. ~ J ^ a ' Gagne, R. M. The conditions of learning (1st ed.). New York: Holt, Rinehart, and Winston, 1965. Gagne^ R. M. Learning hierarchies. Educational Psychologist, 1968, 6(1), 1-6. ' J Gagne^ R. M. The conditions of learning (2nd ed.). New York: Holt, Rinehart, and Winston, 1970. Griffith, S. , Bishop, J., Brown, R., Cohen, C. , Islar, M. , Pinder, B., & Steen, C. Cog nitive skills planning sys tem (part 2). Gainesville, Fl . : Alachua County School Board, 1971. Hanna, P. R. , & Hanna, J. S. Applications of linguistics and psychological cues to the spelling course of study. In T. D. Horn (Ed.), Research on handwriting and spelling . Champaign, 111.: National Council of Teachers of Enqlish, 1966. Haughton, E. Aims: Growing and sharing. In J. B. Jordan & L. S. Robbins (Eds.), Let's t ry doing something else kind of thing . Arlington, Va.: Council for Exceptional Children, 1972. Hollingshead, A. B., & Redlich, F. C. S ocial class and mental illness: A community study . New York: John Wiley and Sons, 1958. Horn, E. Spelling. In C. W. Harris (Ed.), Encyclopedia of educational research . New York: MacMillan, 1960. Horn, T. D. Handwriting and spelling. Review of educat ional research , 1967, 37, 168-177. ' Horn, T. D. Spelling. In R. L. Ebel (Ed.), Encyclopedia of educational research . London: MacMillan, 1969. Hughes, V. H. A study of the relationship among selected language abilities. Journal of Edu cational Research, 1953, 47, 97106. — — Learner, J. W. Children with learning disabilities Houghton Mifflin, 1971. )ston:

PAGE 66

57 Lindsley, 0. R. From Skinner to precision teaching: The child knows best. In J. B. Jordan & L. S. Robbins (Eds.), Let's try doing something else kind of thing . Arlington, Va.: Council for Exceptional Children, 1972. Monroe, M. Children who cannot read . Chicago: University of Chicago Press, 1932. Pennypacker, H. S. Why frequency? The rationale of precise behavior management . Unpubl ished manuscript. University of Florida, 1972. Personke, C. R. Use of nonsense words to test generalization ability in spelling. Elementary Engl ish , 1972, 49, 12331239. Random House. High intensity learning systems . New York: author, 1972. Russell, D. H. Characteristics of good and poor spellers . Contributions to Education No. 727. New York: Teacher's College, Columbia University, 1937. Spache, G. Characteristic errors of good and poor spellers. Journal of Educational Research , 1940, 34, 182-189. (a) Spache, G. The role of visual defects in reading and spelling disabilities. American Journal of Orthopsychiatry , 1940, 20, 229-238. (b) '"" ~~" Spache, G. Spelling disability correlates I factors probably causal in spelling disability. Journal of Educational Research , 1941, 34(8), 561-586. Ja) Spache, G. Spelling disability correlates II factors that may be related to spelling disability. Journal of Educa tional Research, 1941, 35, 119-137. (b) Spache, G. D. Spelling errors test. Gainesville, Fl . : Reading Laboratory and Clinic, University of Florida, 1955. Starlin, A. Sharing a message about curriculum with my teacher friends. In J. B. Jordan & L. S. Robbins (Eds.), Let' s try doing something else kind of thing . Arlington, Va.: Council for Exceptional Children, 1972. Starlin, C, & Starlin, A. Curriculum ladder for spelling. Bemidji, Minn.: Unique Curriculums Unlimited, 1972. Stiles, R . L . S tep_w i s e multi pl e regre ssi on study of SST basic move men t cy cle data with SFTAA/CTBS results from 348 third grade p upils. Research Department, Pupil Personnel Services, Tacoma Public Schools, 1974.

PAGE 67

58 Tempi in, M. C. Phonic knowledge and its relation to the spelling and reading achievement of fourth grade pupils. Journal of Educational Research , 1954, 47, 441-454. ~ Wallace, J., Klein, R., & Schneider, P. Spelling ability and the probability texture of English. Journal of Educational Research, 1958, 61,(7), 315-319. Westermann, G. S. Spelling and writing . Sioux Falls, S.D.: Dimensions Publishing, 1971. White, 0. R. , & Haring, N. G. Exceptional teaching . Columbus, Ohio: C. E. Merrill, 1976. White, R. T. Research into learning hierarchies. Review of Educational Research , 1973, 43(3), 361-375. Wo Iking, W. D. Rate of growth tow ar d adult proficiency: Differences between high and low achi evement children . Paper presented at the International Symposium on Learning Disabilities, Miami Beach, Florida, 1973.

PAGE 68

APPENDIX A SPELLING TEST

PAGE 69

60 SPELLING TEST NAME: SCHOOL: 1

PAGE 70

APPENDIX B ACADEMIC TASKS

PAGE 71

62 I . SEE CVC TRIGRAM /S AY NONSENSE WOR D meb wid kum vak rop het div len jad nup zik fap rud tek yab vit suf pid rab nen ket tav dak fot lub das pi f niv dup maf Speed # correct _ # attempted accuracy /l minute 1 1 . HEAR CON SONANT SOUND/SAY CONSONANT NAME cfhnqtvkybgmpsjwzdx rlnkhfcbyvtqspmjlrg xdzwcqgwl ftjzhvmdns Speed # correct # attempted accuracy /l minute III. HEAR SHORT VOWEL SOUND/WRITE VOWEL eoaiuoeuiaiu eouiaoeiauoe aiuieoaueioa # correct # attempted accuracy Speed /I minute

PAGE 72

63 I V . SEE ONE PLACE ADDITION FACT/S AY AN SWER 1+1=2 2+0=2 5+4=9 6+2=3 8+1=9 0+6=6 5+3=8 2+4=6 7+0=7 2+5=7 1+5=6 0+3=3 6+3=9 6+1=7 2+8=10 4+1=5 3+7=10 3+2=5 4+5=9 1+0=1 0+2=2 2+6=8 2+3=5 1+4=5 7+2=9 3+6=9 2+7=9 4+3=7 4+2=6 1+8=9 4+6=10 7+3=10 9+1=10 3+0=3 5+0=5 1+9=10 6+4=10 8+2=10 5+1=6 0+9=9 # correct # attempted accuracy Speed /l minute V • HEAR TWO LETTER BLENDS AND DIGRA PHS/WRITE LETTERS br th sk ch cr bl sm sh cl tr tw fl dr sn gl fr sp pi gr si pr st sw tr ch dr br Speed # correct # attempted _ accuracy /l minute VI. HEAR CONSONANT SOUN D /WRITE CONSON ANT cfhnqtvkybgmpsjwzdx rlnkhfcbyvtqspmjlrg xdzwcqgwl ftjzhvmdns Speed § correct # attempted _ accuracy /l minute VII. HEAR SHORT VOWEL SOUND/ S AY VOWEL eoaiuoeuiaiu eouiaoeiauoe ai uieoaueioa Speed # correct # attempted accuracy /l minute

PAGE 73

64 VIII. HEAR INSTRUCTION/WRITE LET T ERS OF THE ALPHABET ABCDEFGHIJKLM # correct NOPQRSTUVWXYZ # attempted accuracy Speed /l minute

PAGE 74

APPENDIX C SEE CVC TRIGRAM/SAY NONSENSE WORD STIMULUS SHEET

PAGE 75

66

PAGE 76

APPENDIX D SEE ONE PLACE ADDITION FACT/SAY ANSWER STIMULUS SHEET

PAGE 78

APPENDIX E GROUP MEANS AND STANDARD DEVIATIONS

PAGE 79

CO CO C\J CO CO ,— CD r— CO C\J T3 . — . — c_> sc_) -a o o o cu o

PAGE 80

APPENDIX F RAW SCORE DATA

PAGE 81

72 Raw Score Data The first column (S's) consists of each subjects number within the good spellers group (100's) or the poor spellers group (200's). The DEP column is each subjects spelling test score in percent correct. The three numbers are read as a one place decimal number (e.g., 99.6). The last eight columns are the raw scores per subject on each of the eight tasks. The tasks are: Code Task CVC H/S See CVC trigram/Say nonsense word CONS H/S Hear consonant sound/Say consonant name SHVL H/W Hear short vowel sound/Write vowel MATH S/S See one place addition fact/Say answer BLDI H/W Hear two letter blends and digraphs/Write letters CONS H/W Hear consonant sound/Write consonant SHVL H/S Hear short vowel sound/Say vowel name ALPH H/W Hear instruction/Write letters of the alphabet. The eight numbers for each task are read as two four digit numbers with a one place decimal (e.g., 09900774 = 99.0 and 77.4). A zero was used in the hundred's column when a subject's score was less than 100. The first number in each column is the accuracy score for the task. The last number in each column is the speed score for the task.

PAGE 82

73 Z< kin-oonx irt^--cr--QD OOOO o X CJOOOO X ooooo _J ooooo asO CO OGOOO 3: r^K"ir»-oo X ki^ ^kiki OOOo o CO SCCSS x <\JOLnr\j^jC c-cccrp. U o— {\J CO CO LPK^LncC^ ooooo X ooooo I — ooooo nokin-o x ^Ln-oxnn CO OOOCO co Kir-cDf-rx — -"ururvj-o X =*^r^i^-Ki OO Ooo CO KIOK\KlKI 3T in o*n d x> O cooooo _j cwo-jo co otr--or-.r~. CO t-NcCcOvO OOOo O o~o oo>o (_> CT-O Oo— t >MO^Q^ cj oo-«oo -cm onjco CO cr. co ^ c-oooo — o oooso OOOf^O ooooo — *— • oo-^ K>r^.oN-r^ kiojckco K1K1K1K1K1 ococo f — i-n r— kiki in— «-o>o;x mcc^f^in ooooo OOOCO CO CcO^^T KlKlrOKlKI ooooo or--rooo — "Kir— oo OOcOOO OOOOO MNOOI — O occoru OJ'MrO— '<\1 OOOCO r-~Kir-noKi C-h-OOO ooc— «o NCCCtO rOLn^TLTIK! oo ;>o o ChOOON OOOO-i cc ooo -HO— I— IO MoNr^o l/i-Ooi^gKinjKinjKi ooooo r~mor-o roinco— « inomm-O ooooo (~-N-Kir^o COOJO-OCO KlromKlKi coaoo Kimt—t--o 'X3 a -o \i o octr ooco ooooo N>or — -«o ruco ooj-o n10-0.3-0 c>oo N»oo^r^rojo— <=j oo oo~o oo-nKi COC CO Klor— r-Ki njiri^r -or— r-»Klr^-o-« COOOO Soc oro c oc ^r— • KimKiK"i=r ooc CO cr-ooro OKI in OS) OOO O CO »CO coocrruru-o^rrvi cr^JK-'Kim OOOO O or-oco ONOBO OOOOO — <0— iOO ro r-*-o o — '00-0-0 KiernjKlKi OOOO^O oorr--o O— *-0 r^S r-ermca-« OOOCO r-^Kir-r-m vOLTIOITin ooa co OMOON o_nm o-\i cxrooo - -o ki cr =rc;irin-e •ooooo C OOOO OOO-flO OOOOO or— kw~-nej vcr-^rrvj ooooo ororoot — — +^^Oc\;ki OO ^TO-Q OOOCO COMC O r-nnjt-rucr ooc^oo t-~or-oo oooco 000--0 roororoi-^ t-~COO— ' <*VJ — •— lAJOJOJ cooco otot--orn csa ! cs cor-~vcoco ooo— -o Ct"SSo OLT^TcOO WCSKlfO J z>x>x> o O OOC o O— IOOO r~-Sl--oro C3o oo rvj r" c ^j .a rorom r^ro ooo CO r^r— o oo xm>o coco ooc oo (^KOSn inr^ c ci-o O C2 o o o COOLO Oco CO O O O CO O— oo 00--0 ON 3CO-ICO ainj— «njrvj o o co o torero r^ro rv ^r — cc r~t— O X» -O'O c;oc; oo cc Tl o Li o ^rroco =i^r z> j-r» 3>o> ooc>l^o ooooo OOCJ oo O— «— .C3-H ONOI^S oco runru rorO»o ronj ooo oo rOot^rOl-n loLPr~-onfOvOrocoin ooo coo h-ror-oro OKI CO OlO COO oo ooson O IINNA1 crocr o-o c>oo oo a 'XI ^ in !T3 oovOr~"iri -CU1LT1 Oco ooo oo "o o^> r>-o inooC--* OI-NON o^oo — «o ro>-K..-^NN-CNN^T tooooo OCC^N-NOOOMO OCCO-ff — i— .— .OO NNOrONcc co ^ rcc rjr\jror-onj r>o r> ONMCS oo occo -f\i rvinjK^.r^ivO -oroio— *cc OOOOCCCO COOD33COCO — H<\ifoo ocooc Iflctooo ot~-o oo ooo— •-< r^r-f^ti — o co— rrry oj rvj rvj r\j ru ocooo — 'Oo r--H pninx!t-~Ki i coooo COMNM OCO o o f-oroot-— iT>3oKi occ oo co o ooc oo r-rOKir~ o "Kxmnru K1^t~ vO-<1 OOOOO 0"0)'3 o»-~ rjr\jN-o^r ON-O OHO C300-+C3 coccco OKI -O-OO CLT ccxvoDaOco — 'O'ro^in K1K1K1K1K1

PAGE 83

74 roor-npo-o Ln :ONi_rtrvj ooooo vONoj)a o OOO o oo — too rou-ocorvi rurururoro o cooo n oor^o fiocinLnro -OcTLTICOCT ocooo 2 Nroroo-o N CNrfJ C X^ fXifVfVfMKI ccoco '-C roNrooro -^ Of\l-«wS O oocoon u ooooo 3: NNOf-r-o v. rvj-crvjroN o c ooo »-« oonno D CO-MSN — l inincoN^) en ooooo CO OK-,*DK!I*0 "V ONnjrON co rvrvcrKru O 3 33 3 31 notion I — OJ 0.0 oin OJ — ioKIO n: N—iLncoro tO OOOOO to h-oi-^-r-Ki \ ^r o<; co o I rvjrorurvjro o ooo o CO NNTN OKI c; cctccto J oo-oooCO OSMNN V OC l\J>3(M ooooo AN D3N t-> ncicckin LJ OOOOO 3-O-OKIKIO *J ooo Oca *) -^njf-o^in OOOOO n runjrurvjru ki-hn,--oo crcrKic->K) OOOOO SKIO OO X>NO:0 o~o~ooO oo— o ooooo OT-oc^por^ ^0KK\IN-O OOOOO KlONroo oolpitio ^T^T -Q Tt'O OC O OO OMO OO rjNco -a o dd— OJK1 O Z> O O O oo«iso -Tl~.0 3o OCTCT-O O ooo O— • ^Jtnrvjoro ru— •— < ro rvj coooo noonki ^rrorvirvjcr OOO OO OKIOSIO Of^o^lN KB-Of-ifOro ooo o a OMOSKI mnj^r _on oca acccao oo»ooo» p^ot^* on trooo oo -o — "Oj-»rvj — * OOOOO no-o-o-o Itkw Nin -0-TU'_nrO OOOOO fvto-<^o CONNKIKI -or-oo o o ooo o-* OJAJTVJAjrvj OOO^N osofiry Ki cr ^j caro ooooo ooo oo iroLnoo ooo oo o~*o — »— « iokionn O* ->onic rvjojrvjnjnj ooooo rooro orO l-o -o i-h —"to sO-ooj^tlti oc>ooo NMNON coaD-c^r rurvjevrvrvj OOOCJ OCNKIC c(/iir\oo CCOtCP O — 'OOO• K!NK1 Of*l Kccorvj— « OOOOO (\ioo?orvj-^ocr — « COCOO KIKCf-OKI O-KIO^O rvrurunjKi OOOOO CNNOO X) O-OOO O-ONtC O OOOO— « OtOOOO njKruoorvjnj — icu-« OOOOO •OO -oo -o n^^to cr COOOO KIONN N N o-oco o K>K>Kir\jr-o oaooo Noooo O-0> -oo. OOOLO O O OOO-* cioroNN -co to. ^c =3njrvj — .— irj OOOOO "ON rO -O ."O) O-MOC-O N-crroo=3OOOOO inoNas^r —•OCT'XK) ^T^rroKirO — nr-~ — '-oiriLn c ooooo ONNOCS ruococrr— OoOoDoOcC ooooo rot^omo r-0f\Jr0N->G OOCOO rooows i-^t-^croar i^or-mroLn ooc oo oor— tON o ^)crr-,o — <\J oj rv _ o-t>o OO l-'INOOfO o.-uo OVN? ffloco ooooo rvjr--»--OvO ooc O^O c3r~-ro oo l_nco o "-~i r-> — i^o^O— cr--co r^v-vj o — .^r cvoooo otn^-otn ^r.-onjojo -OT--COOO — ^-i— .— .f\j ruruojojoj csr-cr^o "VI 3"C3rO.O ooooo ot^-cor— o rvj o o co ooooo — -O— — io KlOtONN ro rv a cr ao i-oojojaj-^ o o c o o t-ooooro 00!><10 ci 'JI-OCfM-i OOOC: O Klr-ONoN r^o^cunj Kl— "rur^. rvj ooocc ct-omoo Oacr-O co oooo o O^NOK or o -o x> ru — «0— l-H— ooooo ot-oo^o o ^TK'OrOO I — O OO O OC OOO ONOhO c; njasoom o oooo sooroo O OOOO O OCtC o O— »— io— l ru— »nj— «i"U ocoo o KlKlor-OO r\" -O -o rvi to o ooco OKlt— 0<0 or-rvj^T^r ^TOJKiroro oo coo MOONN oCaCaaoDO o«-oo — •--Ovi-l— «o jf\i o o^o oo l-— c o no -OOO -no Ooo mo o— t— • o— • ONt^-li-OrO airvjnjjnj.'M C OO OO I-0N-C3jr^lt^i -O^aOjOO r-0 r*i -V ki rvj O O Oj o o tO I — OiNN — ->nj-o lo-^r fvrvjrvjj— «-^ oo>o;oo roroKi oo Nru— • rvjo coodOi^jo OOOjO— • fOK ,-0100 •—ccrolrvjo —IO— >!—•—. c ooloo mnn|oo O— CT^3 a'Mro ruro ooo oo I-ONNON 0<0 LTI-T TOcC NO OOO f>^> KINO ON t-o=to ru-o »-•— — " — OOOOO •o— «o oi_n N-otn Kiro rurvjrvj rurvj -or~ao OC3 rurvjrvj njio njrvjru ruru MNCIGM N-vOI/IOI— rvjioroajrvj C30000 COKIt^N OOI-^N— • OOOOCN — «— «ooo r^oN»c>t — cc-oorv^c — •— •njrvj— • ooooo rOOPOr-Oi — OOCOLT) njofO-H-H ooooo tocorooo OinNcO-O OCJOOO MCOCN NvMXIOO CCMCCN ooooo KlNNWIrO Oar. o— >o OO^-tO COOOO OONlOO IPOnCPVJO -ooDr\j njfv-,rururo o~> oo-o NK1NNN XI X)rUa3rO -coo ajo oo ooo ONNon cr^i ru-oLn ooooo NNO-Oro -oaj--»o(-o (M— .— i— ,^ OOCOO NOONN 70703) rvjroruTO— i OO OCO NWMI — O — •^r r >if>:0 OO OC N-O OO OOO NOOt-OO -ooo-or^^r oj— • — o-h OOOOO O-ONON 'M -ifVJOO OOOOO CTtaO— iON oco r\icor~— ajrociiri rOrOrOroro rururvirurvj

PAGE 84

APPENDIX G SCATTER PLOTS

PAGE 85

76 The following are scatter plots representing each task, based on accuracy or speed, and its relationship to the spelling test score (percent correct). Two correlations are reported. The first correlation is a pooled correlation. Each group's scores (N=35 per group) are correlated independently and then pooled for an average correlation. The second correlation is based on the total sample (N=70).

PAGE 86

11 CJ«n X X ( '"' O O r

PAGE 87

^ cc -^OCCOGO /o > 3 o err CX _| xui a. <. r r < ro O

PAGE 88

-79 Ln co O O i? r\ ^

PAGE 89

80 c c COGOOOOOD3)jj; d:): j Q Ijo XXX r
PAGE 90

uO -+ coooooooo

PAGE 91

C c c c c c — ..-82.—.. 0000003 3)3)^90 oto rr_j -iu o or t^cocccooooo

PAGE 92

Cccoeoooo 83 °®3033Ji3 ocs y O 3>0_l c
PAGE 93

nr Ocr_f i-ui _J iu»C*0 u. o on

PAGE 94

85 er> ja CX _»

PAGE 95

DO -*

PAGE 96

87 O X_l x: <_«o rr_i »— Lj crr> ^ o OO o O ax

PAGE 97

U-JO -* pL -t h1,
PAGE 99

90 TV-*^<

PAGE 100

m *» a *_»_j
PAGE 101

92 tr.j *~u ^-T W3UJ

PAGE 102

BIOGRAPHICAL SKETCH Elliott I. Lessen was born in Syracuse, New York, on February 10, 1947. He was graduated from the Syracuse City Schools in 1964. Mr. Lessen received a B.A. from Syracuse University in 1968 with a major in Art History. Mr. Lessen was a third grade teacher with the Chicago Public Schools for the 1968-1969 school year during which time he pursued graduate coursework in elementary education at Loyola University. From June, 1969, through August, 1970, Mr. Lessen taught HeadStart/Experimental Pre-Kindergarten and Prefirst grade in Syracuse, Mew York. He did further graduate work in early childhood and special education at Syracuse Uni versi ty. From September, 1970, to June, 1973, Mr. Lessen was employed by the Adams School, a private school for brain-injured and emotionally disturbed children in New York City. He was awarded his M.S. from Hunter College in June, 1973, with a major in special education (emotional disturbance). Mr. Lessen was a specific learning disabilities resource teacher during the 1973-1974 school year in Gainesville, Florida. He started his doctoral program in September, 1973. Concurrently, Mr. Lessen was a graduate teaching assistant with the Department of Special 93

PAGE 103

Education and the National Teacher Corps. He was also an administrative assistant with the Career Associate in Special Education program at Santa Fe Community College. Mr. Lessen will receive a Ph.D. from the University of Florida in December, 1976. His area of specialization is special education (learning disabilities) . Mr. Lessen has accepted a position as an Assistant Professor of Special Education at the University of Northern Iowa, Cedar Falls, Iowa, beginning August, 1976. Mr. Lessen and his wife, Linda, are happily married.

PAGE 104

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. ' '-'A /,'.//; William D. Wolking, Chairperson Professor of Special Education I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Thomas B. Abbott Professor of Speech I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. CharleS Forgnone ,/ Professor of Special Education -He I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Cecil D. Mercer Assistant Professor of Special Education

PAGE 105

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Paul Satz Professor of Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. /d/(. jw_ (Lk l<~ Wil liam B. Ware Professor of Foundations of Education This dissertation was submitted to the Graduate Faculty of the College of Education and to the Graduate Council, and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. •fJe c e mb e r, 1976 Dean, College of Education Dean, Graduate School

PAGE 106

UNIVERSITY OF FLORIDA 3 1262 08552 9484


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EHDPCX2EY_8R9UMS INGEST_TIME 2017-08-14T20:57:41Z PACKAGE UF00098102_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


CONCURRENT ACADEMIC PREDICTORS OF SPELLING
PERFORMANCE OF THIRD GRADE CHILDREN
By
ELLIOTT I. LESSEN
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA

To
Linda

ACKNOWLEDGMENTS
Sincere appreciation and gratitude are extended to tny
committee members: Dr. William D. Wolking, Chairperson, and
Drs. Thomas B. Abbott, Charles Forgnone, Cecil D. Mercer, Paul
Satz, and William B. Ware. These gentlemen have extended their
personal and professional efforts toward my development as
their graduate student. Ms. Leila Cantara, as typist and
friend, has been an invaluable asset during my doctoral pursuit.
I would like to thank Ms. Judith Sazer-Kuperstein for her
foresight and encouragement in this endeavor. Ms. Karen Andrews
and Ms. Susan Welch receive my thanks for their assistance in
the data collection process. The support, encouragement, help,
and friendship of David Westling, Mark Koorland, Terry Rose, and
Jim Altman can never be repaid.
My family and friends deserve thanks for their support and
confidence in me. My wife, Linda, above all others, has shared
my ups and downs and tolerated my temperinent. It is to her that
I dedicate this dissertation.

TABLE OF CONTENTS
ACKNOWLEDGMENTS
LIST OF TABLES
ABSTRACT
CHAPTER I
INTRODUCTION
Purpose of the Study
Justification of the Study
Limitations
Threats to Internal Validity
Threats to External Validity
Definitions of Terms
CHAPTER II
REVIEW OF RELATED LITERATURE
Introduction
Traditional Variables
Parameters of the Present Study . . .
Learning Hierarchies
Definition and Characteristics . . . .
Origins of Learning Hierarchies . . .
Validation of Learning Hierarchies
Task Selection in the Present Study
Task Selection and Hierarchy Validation
Related Academic Predictor Variables
Reading
Phonics
Summary
Speed of Manuscript Handwriting . . .
Spelling Test Scoring Procedure . . . .
Scoring Units
Whole Word
Syllable
Sound Cluster
Letter
i i i
vi i
v i i i
1
1
1
1
11
12
12
13
13
14
14
14
15
15
15
IV
001Ü03COSJS1SJVJ UT (J1 Ol irt Co

Performance Descriptors 16
Accuracy 16
Speed 16
Repeated Measures of Predictor Tasks 17
Summary 18
CHAPTER III
PROCEDURES 19
Specific Questions 19
Question One 19
Question Two 19
Question Three 20
Subjects 20
Instrumentation 22
Method 25
Administration of the Spelling Test 25
Scoring of the Spelling Test 26
Administration of the Predictor Tasks 26
Scoring the Predictor Tasks 28
Data Analysis 28
Additional Analyses 29
CHAPTER IV
RESULTS 32
Question One 32
Question Two 33
Question Three 33
Additional Findings 34
Contribution of Predictor Variables in Accounting
for Variance in the Dependent Measure 34
Initial and Predicted Group Assignment 35
Analysis of Single Component, Auditory Input Tasks .... 37
Accuracy and Speed Change 39
CHAPTER V
DISCUSSION 40
Task Complexity and Learning Hierarchies 41
Input and Output Modality Structure of the Task 43
Spelling Test Scoring Procedure 44
Performance Measures 44
Repeated Measures of the Tasks 49
Conclusions 51
Recommendations for Future Research 52
REFERENCES 55
v

APPENDIX
A SPELLING TEST 60
B ACADEMIC TASKS 62
C SEE CVC TRIGRAM/SAY NONSENSE WORD STIMULUS SHEET . . 66
D SEE ONE PLACE ADDITION FACT/SAY ANSWER 68
E GROUP MEANS AND STANDARD DEVIATIONS 70
F RAW SCORE DATA 72
G SCATTER PLOTS 76
BIOGRAPHICAL SKETCH 93

LIST OF TABLES
Table
1 Sex Ratio Per Group 21
2 School Breakdown Per Group 21
3 Parents' Index of Social Position Per Group 22
4 Predictor Variable Contributions 34
5 Accuracy and Speed as Performance Measures 35
6 Accuracy as a Performance Measure 36
7 Speed as a Performance Measure 36
8 The Relationship of Single Component, Auditory Input
Tasks to Spelling Performance 38
9 Mean Change Per Task Per Group 39
10 Proficiency Levels Per Task 46
11 Standard Deviations on the CVC, MATH, BLDI, and
flLPH Tasks 47

Abstract of Dissertation Presented to the Graduate Council of
the University of Florida in Partial Fulfillment of
the Requirements for the Degree of Doctor of Philosophy
CONCURRENT ACADEMIC PREDICTORS OF SPELLING PERFORMANCE
OF THIRD GRADE CHILDREN
By
Elliott I. Lessen
December, 1976
Chairperson: William D. Wolking
Major Department: Special Education
The focus of the present study was to identify a set of pre¬
requisite tasks, modality relevant to spelling performance, that
when based on accuracy and speed as performance measures, would
discriminate between good spellers and poor spellers. A review
of related literature suggested that (a) tasks used to date have
represented more than one basic behavior per task; (b) tasks that
have been used are related to, but are not direct replications of
the modality input and output structure of spelling behavior; (c)
tasks used have not represented the most basic skills assumed by
curriculum hierarchies to be prerequisite to spelling competence;
and (d) tasks have been assessed by accuracy alone, without re¬
ferring to speed (fluency) as a viable performance measure.
A spelling test was administered to a group of third graders
in Alachua County, Florida. From these results, the top and bottom
vi i i

25.9 percent (good and poor spellers) were chosen (N= 35 per group).
Each of these subjects was then assessed on each of eight tasks on
three consecutive days. Accuracy and speed scores were obtained
for each subject on each task. Discriminant analysis and multiple
regression procedures were used to analyze the data.
The results of the study indicated that two of the eight
tasks used, See CVC trigram/Say nonsense word and Hear two letter
blends and digraphs/Write letters, were the best tasks with which
to discriminate groups of good and poor third grade spellers.
Speed and accuracy scores on the two tasks were found to be nearly
equivalent discriminators between the two groups of spellers. The
findings of this study indicated that the use of speed and accuracy
as performance measures may serve two useful functions. First,
speed and accuracy scores discriminate between good and poor
spellers. Second, speed and accuracy scores reduce possible
misclassifications that may interfere with a child's curricular
program. The overall and additional findings have suggested
possible alternative research efforts that may help solve the
problem of presenting classroom instruction that could promote
better spelling achievement.

CHARIER I
INTRODUCTION
Spelling is one criterion by which school achievement is
measured. Allen and Ager (1965) state that "spelling is an
independent skill and should receive specific attention . . .
in the school curriculum" (p. 159). There are many charac¬
teristics that contribute to spelling performance in children,
e.g., intelligence, visual perception. There may be certain
prerequisite academic skills, that if known early, could serve
to discriminate between good and poor spellers. Such a discrim¬
ination could also be helpful in planning remedial and develop¬
mental curricular programs.
The concern for correctness in spelling and effective in¬
structional methods is not new. Cook and O'Shea (1914) stated,
"There is ... a wide-spread belief that graduates of the
elementary schools can not spell so well now as they did in
earlier times." T. D. Horn (1969) related even earlier reports
of such dissatisfaction from the thirteenth century. Many
teachers are aware of children who are poor spellers but the
reason for their spelling difficulties remains a concern. Al¬
though educators are cognizant of correlates related to poor spel¬
ling (Russell, 1937; Spache, 1940a; 1940b; 1941a; 1941b) they may
1

2
be unsure of those specific behavioral elements functionally
related to spelling.
To date, the research on spelling suffers from numerous
limitations. Among these are (a) the lack of replicable pro¬
cedures (T. D. Horn, 1967); and (b) the use of tasks that do
not directly replicate spelling behavior (Westermann, 1971).
An example of the latter is the use of tasks to assess spelling
which have generally been used to assess reading. The charac¬
teristic behaviors specific to spelling and reading are
different. Whereas in reading the input modality is visual
(See), in spelling the input modality is auditory (Hear)
(Fries, 1963; Lerner, 1971). It does not necessarily stand
that a See consonant/Say consonant name task provides informa¬
tion which is as relevant to actual spelling behavior as a Hear
consonant sound/Say consonant name task. The latter is a direct
replication of the modalities used in spelling behavior, while
the former can only approximate the spelling behavior by content
alone.
Emerging from the above limitations in the existing
literature is the need for clearly defined and replicable pro¬
cedures (T. D. Horn, 1967) that make use of modality relevant
tasks dealing directly with spelling (Bannatyne & Wichiarajote,
1969; Hanna & Hanna, 1966).

3
Purpose of the Study
The present study was designed to discriminate between good
and poor spelling performance in third grade children. It was
based on a combination of unique theoretical and methodological
parameters that have not been used in spelling research prior
to this study. The unique factors are:
(1) the use of tasks that appear to be prerequisite of
spelling behavior;
(2) the use of accuracy and speed measures as per¬
formance descriptors of the tasks;
(3) the use of tasks that make use of modality input
and output directly related to spelling performance;
(4) the use of multiple measures consisting of three
samples of each child's performance on each task; and
(5) the use of a new scoring procedure for the spelling
task itself.
The problem, then, was to identify a set of prerequisite
tasks, modality relevant to spelling performance, that when based
on accuracy and speed as performance measures would discriminate
between good spellers and poor spellers. The specific questions
asked were:
(1) Which tasks, with respect to accuracy, discriminated
between good spellers and poor spellers?
(2) Which tasks, with respect to speed, discriminated
between good spellers and poor spellers?

4
(3) Which tasks, with respect to both accuracy and speed,
discriminated between good spellers and poor spellers?
Justification of the Study
Among the reasons indicated as possible evidence of low
spelling achievement in the United States are (a) a decline
in the systematic teaching of spelling resulting from erroneous
interpretations of incidental findings and (b) confusion re¬
sulting from grapheme and phoneme relationships (E. Horn,
1960). Prediction studies have not been done in the area of
spelling that would help to clarify these issues. What exists
in the literature on spelling, however, are concurrent correla¬
tional studies that have related various characteristics (e.g.,
intelligence, demographic information) to spelling competence.
The intent of this study was to use tasks that appear
to be hierarchical prerequisites of spelling competence using
the same input and output modalities as actual spelling per¬
formance (Hear input rather than See input). Traditionally,
See or reading type tasks have been used. Hear tasks were chosen
because their topographical relation to spelling behavior may
provide better predictors.
Findings from this study may provide educators with a new
knowledge base upon which to (a) identify good spellers and poor
spellers early; (b) focus on curricular variables related to
remediating poor spelling; (c) focus on these curricular
variables in accordance with performance measures (i.e., accuracy

5
and speed) found to be relevant; and (d) provide information about
the utility of an input and output modality structure that is
relevant to spelling performance. Whether or not remediation of
related deficit skills will directly result in the improvement
of poor spelling is subject to empirical research. It is not
included within the scope of the present study.
Limitations
Threats to Internal Validity
The internal validity of this study may have been weakened
by the instrumentation procedures used (Campbell & Stanley, 1966).
This refers to possible scoring errors due to individual data
recorder error. Chapter III discusses the procedures used to
protect the internal validity of this study.
Threats to External Validity
The generali¿ability of this study may have been threatened
by the lack of total randomization in the selection of the subjects.
The reader is cautioned in generalizing the results presented
beyond the geographical area in which the study was done.
Definitions of Terms
Good Speller: The top 25.9 percent (N=35) of the original group
(N=135) based on the spelling test score.

6
Poor Speller: The bottom 25.9 percent (N=35) of the original
group (N=135) based on the spelling test score.
Accuracy: A percentage score based on the number of correct
responses divided by the total number of responses made.
Speed: The total number of responses made, correct and error,
per minute.

CHAPTER II
REVIEW OF RELATED LITERATURE
Introduction
Traditional Variables
Spelling research has traditionally been divided into three
areas. These areas are (a) learner variables, (b) curricular
variables, and (c) word variables (Cahen, Craun, & Johnson, 1971).
The knowledge of learner variables has not proven to be clearly
related to remediating poor spelling or to facilitating good
spelling (T. D. Horn, 1967). The majority of research on spelling
has dealt with curricular variables, e.g., methods and materials
(Blair, 1975; T. D. Horn, 1969). There is no teaching approach
that appears to best facilitate good spelling. Research on
word variables has been concerned with such issues as error
analysis, regularity versus irregularity of orthography, and
high frequency versus low frequency of word occurrence.
Parameters of the Present Study
The present study has five parameters. These parameters
have generated a new perspective for looking at learner per¬
formance on relevant tasks. The literature reviewed will be
considered under the rubric of:
7

8
(1) learning hierarchies, as they contribute to task
selection;
(2) academic predictors and their contribution to (a)
the selected tasks and (b) the input and output
modality of the selected tasks;
(3) a new procedure for scoring spelling tasks;
(4) accuracy and speed as separate performance
measures of the selected tasks; and
(5) repeated measures of predictor tasks.
These parameters may permit more precision in the discrimination
between good and poor spellers. Learning hierarchies is a
theoretical parameter. The new scoring procedure and multiple
measurement of tasks are methodological parameters. Performance
measures and the contribution of academic predictors are both
theoretical and methodological in nature.
Learning Hierarchies
Definition and Characteristics
Learning hierarchies are "descriptions of the relationships
of positive transfer among intellectual skills" (Gagne^ 1968,
p. 4). What they are not, however, are statements of how an
individual acquires the knowledge contained in the hierarchy.
Each level of the hierarchy generates some quantity and quality
that positively transfers to the learning of a "not-previously-
acquired, higher-order capability" (p. 5).

9
Learning hierarchies, then, represent the effects of cum-
mulative learning with the greatest amount of positive transfer
among elements. Gagne (1968) makes the point that learning
hierarchies represent the most effective direction of learning
for the entire population of learners, and do not represent
strategies unique to individual learners.
Origins of Learning Hierarchies
This study has as one of its parameters the issue of
curriculum and is concerned with the selection of tasks that
may be sequentially ordered. Gagne"(1970) has stated that
content in any one area of learning (e.g., spelling) should
be hierarchically ordered so that those simpler concepts and
abilities upon which later learning is based be mastered
first. Instructional design, or management, can then be
based on those conditions of learning (curriculum) that are
requisite to each of the hierarchical components.
Learning hierarchies represent priority ordered skill
sequences. That is to say, based on skill analysis, experience,
and the use of other persons' lists, skills are ordered, in many
instances, post hoc. One should not infer that this method is
in error. However, effort needs to be expended to determine

10
empirically whether or not these ordered skills transfer
positively in a vertical fashion or are independent (Gagne^
1968).
Validation of Learning Hierarchies
Investigations of the validity of learning hierarchies
have yielded inconclusive results (White, 1973). Based on
Gagne^s (1962) validation model, research efforts reviewed fell
short of being validated. Reasons indicated by White included
methodological weaknesses and the lack of the Gagne^model to
provide a test of hierarchical dependence.
White (1973) then discussed a different method for validat¬
ing hierarchies. The model proved useful with one major
limitation. Large numbers of hierarchical elements (k) prohibit
the use of the model. This is due to the number of groups of
subjects needed to complete the validation procedure which can
he arrived at by using k! groups for maximum control or 2(k-1)
groups for minimum control.
White (1973) concluded his remarks by presenting some
modifications of Gagne's (1962) original design. He stated
that "it would be a pity if further effort is expended on the
production of unreliable or ambiguous data in this [learning
hierarchies] area of learning when much more precise work is
possible" (p. 374).
Task Selection in the Present Study
In light of the comments of White (1973) and Gagne*"(1963;
1970) and the lack of validation of current hierarchies, one is

11
left with a choice of combining what is already in existence or
deriving a new hierarchy that may not fit those already in use.
Specific behaviors assumed to be prerequisite to spelling
have appeared on curriculum hierarchies (Random House, 1972;
Starlin & Starlin, 1972) and in task analyses of spelling
(Westermann, 1971). These behaviors have been accepted as
necessary for spelling performance. It is from these hierarchies
that the tasks used in the present study were selected.
Task Selection and Hierarchy Validation
In lieu of a validated hierarchy in spelling, tasks were
chosen from already published hierarchies. The tasks selected
represent a basic core of skills found at the beginning of each
of these published hierarchies. That is not to say because
they are in agreement with each other they are valid, but
because they are in agreement they represent a consensus of
the field at present. The tasks in use in this study afford
one a beginning point which needs further work toward empirical
validation.
Related Academic Predictor Variables
This study has used concurrent prediction, with third
graders, in an effort to determine a set of tasks, not used
to date, that attempts to explain the nature of the relationship
of this set of tasks to spelling performance.
The following section of the review is limited in scope to
those areas of academic concern that are relevant to the selec-

12
11 on of tasks used in the present study. Of concern are tasks
that are topographically relevant to spelling and tasks re¬
lated to speed of manuscript handwriting.
Reading
Hughes (1953) has stated that "the correlation [+.53]
between reading and spelling indicates that . . . they may
have some skills in common, but are, to a considerable degree,
discrete" (p. 348). Reading and spelling skills, although
topographically similar, make use of different input and output
modalities. Reading is a See/Say task and spelling is a Hear/
Say or Hear/Write task. Therefore, reading represents a
grapheme-phoneme relationship and spelling represents a
phonome-grapheme relationship.
Skills related to spelling need to be assessed with the
modality structure relevant to actual spelling behavior. The
utility of such a modality structure will allow one to place
particular emphasis not only on the content of related tasks,
but to ascertain the importance of the modality structure in
predicting spelling performance.
Phonics. Studies by Spache (1940a), Russell (1937), Bond
(1935), and Monroe (1932) assessed word attack skills related
to spelling. Their results indicated that poor spellers were
inferior to good spellers on these tasks (i.e., blending letters
to form words). Two limitations of these studies that are
relevant to the present study were (a) the tasks involved visual

13
input rather than auditory input; and (b) the tasks encompassed
different levels of the same topographic task (i.e., CVC words,
CVCV words, CCVC words). Monroe (1932) in her assessment of
vowel or consonant difficulties that might have contributed
to spelling difficulty, based her conclusions on words read
by her subjects. If a child read "these" instead of "those",
the error was vowels. If a child read "then" instead of "them"
the error was consonants.
Templin (1954) found phonics based on visual recognition
tasks to be correlated +.54 to +.57 with spelling. She did use
one recall task, Hear consonant sound and sound combinations/
Write letter(s). This task correlated +.34 with spelling.
Although Templin used a Hear task, the task included more than
one basic behavior (single sounds and sound combinations).
Summary. The selection of tasks for use in the present
study were modality relevant to spelling as well as representing
one basic behavior per task. These tasks were selected because
they represent basic skills agreed to be prerequisite to spelling
competence.
Speed of Manuscript Handwriting
The issue of speed in handwriting has been considered as
early as 1914, when Freeman found that increasing speed does
not necessarily sacrifice the quality of handwriting. The
slower the child writes, the slower the language forms are
emitted from him. As a child is able to write more automatically

14
(proficiently), the easier it is for him to communicate with
others in written form. Early speed studies have not shown
significant differences between poor and good spellers (Cobb,
Kincaid, & Washburn, 1918; Russell, 1937).
Speed data on third graders suggests that they can write
random letters at a mean rate of 41 letters per minute (Stiles,
1974). In another study, Wolking (1973) foyind that third
graders can write numbers randomly at a rate of 62 numbers per
minute and that by the sixth grade at a rate of 116 numbers per
minute. With an increased emphasis on proficiency as a function
of accuracy and time, perhaps the issue of speed can be a more
clearly defined component of educational designs. The use of
speed data in the present study may yield information that will
further substantiate its inclusion as a performance measure
of academic competence.
Spelling Test Scoring Procedure
Scoring Units
Whole Word. Spelling tests have traditionally been scored
using a words correct, as the scoring unit, procedure. Use of
this method does not allow for the varying length of each word.
Individuals may be penalized for misspelling an entire word,
when they may have, in fact, only misspelled an affix. Further
analysis for instructional purposes is limited.

15
Sy 1lable. The use of syllables as a scoring unit accounts
neither for the varying length of words nor for the monosyllabism
of some words. Use of this method of scoring subsumes an a priori
knowledge of the rules of syllabication by the speller. This is
not to infer that spellers do not need to be aware of these rules,
but in the early grades (1-3) the more complex rules of syllabica¬
tion are not stressed (Griffith, Bishop, Brown, Cohen, Islar,
Pinder, & Steen, 1971).
Sound cluster. Sound clusters have been called phonemes by
some authors (Bannatyne, 1973; White & Haring, 1976). Bannatyne
makes a distinction between phonemes in the word "cough"
{/k/ /ou/ /f/) and optemes and graphemes (/c/ /ou/ /gh/).
Despite the nomenclature used, sound clusters may present some
problem in that the child may have some of the letters correct
but may miss others. In the above example, "c" and "k" are
interchangeable phonemically, as are "ou" and "aw" and "f", "gh",
or "ph". Use of this method does not give an individual credit
for parts of the word he has written correctly, because the whole
word procedure is still a factor.
Letter. White and Haring (1976) suggest that "the letter
is the basic unit of spelling behavior" (p. 29). The following
example will illustrate the scoring procedure. The word "often"
has five letters, thereby having six possible units. The child
is given credit for knowing how to start and knowing how to stop
the word. Carets are used to indicate the flow of the spelling
of the word. "Often," when spelled correctly, looks like

16
A A A A A A
often (six units), but when spelled without the "t" would
A A A A
look like o f e n (four units). This scoring procedure allows
one to look not only at the whole word, syllables, and sound
clusters, but at letter sequencing, where many errors seem to
be prevalent. Evidence from pilot data not only substantiated
this contention, but al lowed the investigator to look at initial,
medial, and final consonant placement, vowel placement and letter
sequencing. Although this information was not used for purposes
of this study, valuable information for teaching considerations
is made available.
Performance Descriptors
Accuracy
Performance analysis can only be described as accurately as
is allowed for by the performance measures used. Presumably
more precise measures permit finer discriminations to be made.
Traditional measurement provides one with an accuracy score, be
it percent correct of the material presented or percent correct
of the material attempted. Accuracy, therefore, is only repre¬
sentative of count, or comparison on an interval scale.
Speed
Speed allows one to look at the number of behaviors (count)
emitted per unit of time. Skinner, in 1932, introduced the con¬
cept of speed (count/time) "as a basic unit of direct behavioral
measurement" (Pennypacker, 1972).

17
Speed is a valuable measure because all behavior occurs in a
time dimension. Speed, in combination with accuracy, allows for
more precise information than is yielded by accuracy alone.
Two children may each read the same material with the same
accuracy. However, the addition of speed as a performance
measure allows one to notice that one child takes two minutes
and the other child takes five minutes. Educationally, one
cannot, with this added information, state that both children
read equally as wel1.
Haughton (1972) has discussed the importance of speed to
education. He stated that it is fluency (speed) that allows a
child to perform functionally, not accuracy. It must be remembered,
however, that the acquisition of functional speed follows the
acquisition of accuracy on the same task.
This study has made use of both accuracy and speed scores per
task, in an effort to assess the unique contribution each may make
to the same set of predictor tasks and to assess the joint contri¬
bution each may make when used to describe the same tasks.
Repeated Measures of Predictor Tasks
Norm-referenced tests are administered once per child. The
score yielded represents that child's functioning level at the
time of testing. To accomodate for errors in measurement related
to such factors as scoring error, illness of the child, and lack
of rapport with the examiner, these tests have included in their

18
manuals formulae from which may be derived a standard error of
measurement (SEm = ax"V 1 -rxx), where ox = the standard devia¬
tion of the test and rxx = the reliability coefficient of the
test.
It has been suggested by researchers (i.e., Stiles, 1974;
White & Haring, 1976) that administration of tasks over a three
day period will yield a more reliable measure. In an effort to
obtain a more reliable measure from each subject, data was
collected on the same tasks for three consecutive days.
Summary
The literature reviewed suggests that (a) tasks used to date
have not been comprised of one basic behavior per task; (b) tasks
that have been used are related to, but are not direct replica¬
tions of the modality input and output structure of spelling
behavior; (c) tasks used have not represented the most basic
skills assumed by curriculum hierarchies to be prerequisite
to spelling competence; and (d) tasks have been assessed by
accuracy alone, without referring to speed (fluency) as a viable
performance measure that may yield useful information for
teaching. The present study has used these parameters in a
unique combination that may describe more accurately predictor
variables associated with spelling performance.

CHAPTER III
PROCEDURES
The focus of this study was to examine a set of concurrent
academic predictors of spelling performance. A group of third
graders was administered a spelling test. From these results,
the top and bottom 25.9 percent (good and poor spellers) were
chosen. Each of these subjects was then assessed on each of
eight predictor tasks on three consecutive days. Accuracy and
speed scores were obtained for each subject on each task. These
scores were used as the independent variables in the statistical
analysis. The dependent measure was group membership, e.g.,
good or poor spellers.
Specific Questions
Question One
Which tasks, with respect to accuracy, discriminated between
good spellers and poor spellers?
Question Two
Whick tasks, with respect to speed, discriminated between
good spellers and poor spellers?
19

20
Question Three
Which tasks, with respect to both accuracy and speed,
discriminated between good spellers and poor spellers?
Subjects
One hundred thirty-five third graders were administered
a spelling test (see Appendix A). The subjects for the study
were drawn from three public schools in Alachua County, Florida.
The sample initially comprised third graders in attendance at
these schools. The schools were selected on the basis of
availability. An upper and lower group (N= 35 per group) were
selected from the original group. Each group, therefore, was
comprised of approximately 25% of the original group. These
two groups were used for the remainder of the study.
The following tables report demographic information about
the subjects in the two groups. Table 1 shows the sex ratio
for each group. Table 2 shows the school breakdown for each
group. Table 3 shows the parents' index of social position
for each group (Hoi 1ingshead & Redlich, 1958). It may be noted
that there is a disproportionate amount of children from groups
five, six, and seven. The distribution of children in this
sample, based on parents' index of social position, does not
approximate a normal population distribution. This information is
included to further describe the characteristics of the sample,
however, it was not used in the data analysis of the present study.

21
Table 1
Sex Ratio Per Group
Sex
N
Good
% of Good Group
Poor
N % of Poor Group
N
Total
% of Total
Male
16
.46
21 .60
37
.53
Female
19
.54
14 .40
33
.47
Total
35
35
70
Table 2
School Breakdown Per Group
Good Poor Total
School N % of Good Group N % of Poor Group N % of Total
49
10
.29
27
.39
20
12
.34
19
.27
31
13
.37
24
.34
35
70
Total
35

22
Table 3
Parents' Index of Social Position Per Group
Group
Good
Poor
Totals
Index
1 2 3 4 5 6 7
5 6 1 6 6 4 4
001 4 5 10 11
5 6 2 10 11 14 15
Not
Available
3
4
7
1 = Higher executives, proprietors of large concerns, and major
professionals.
2 = Business managers, proprietors of medium-sized business, and
lesser professionals.
3 = Administrative personnel, small independent businesses, and
minor professionals.
4 = Clerical and sales workers, technicians, and owners of little
businesses.
5 = Skilled manual employees.
6 = Machine operators and semi-skilled employees.
7 = Unskilled employees (and unemployed).
Instrumentation
All tasks for all subjects were the same. The tasks used
as the predictor variables were assumed to be representative of
classroom behaviors needed to promote skill attainment in the
area of spelling.

23
The spelling task used to obtain the dependent, or criterion,
measure included 48 words randomly selected from the Spelling
Errors Test 11-1V (Spache, 1955). The test has twelve sections
of ten words each. Each section represents a specific error
type. Although error types were not analyzed for this study,
four words were included from each of the twelve error type
groups. In this way, no subject was penalized if, for example,
he were to be deficient in one error type.
This study has made use of tasks drawn from the Starlin
and Starlin (1972) and Random House (1972) curriculum hierarchies
and from Westermann's (1971) task analysis of spelling. Selected
for presentation in a random order, the predictor tasks used were:
(1) See CVC trigram/Say nonsense word;
(2) Hear consonant sound/Say consonant name;
(3) Hear short vowel sound/Write vowel;
(4) See one place addition fact/Say answer;
(5) Hear two letter blends and digraphs/Write letters;
(6) Hear consonant sound/Write consonant;
(7) Hear short vowel sound/Say vowel name; and
(8) Hear instruction/Write letters of the alphabet.
Tasks 2, 3, 5, 6, and 7 were employed because (a) they are
representative of prerequisite skills assumed to be necessary for
spelling competence and (b) they follow the Hear/Say or Hear/Write
modality structure of spelling behavior. These tasks, neglected
in previous research, needed to be studied so as to assess their
relevance to spelling and to the teaching of spelling.

24
Tasks 1, 4, and 8 were included because of the unique in¬
formation each might contribute to a more definitive understand¬
ing of spelling. If the math task (#4) were to predict or
correlate highly with the ability to spell, one might be able
to infer that the prerequisite skills for spelling are not
specific to the domain of spelling.
Inclusion of the alphabet writing task (#8) serves a unique
function. As one assesses short vowel and consonant skills, the
output is to be obtained in two modes, one oral and the other
written. If a child were to have a high rate of oral responding
on these tasks, but were to have a low rate of written responding,
one might assume that the writing of the response interfers with
the performance. This would indicate a performance ceiling on
these tasks placed by the child's writing ability. If, on the
other hand, a child were to have a higher rate of written respond¬
ing on these tasks, one might infer that his writing-motor ability
is better than his vocal-motor ability.
Task #1, a See/Say (reading) task, is included as it is
purported to assess the ability to blend basic sounds (Personke,
1972). Assessing the ability to use phonics generalizations
becomes somewhat of a problem. Some words which are phonetically
regular, such as "red," may already have been learned as a sight
word. Therefore, CVC nonsense trigrams were used.

25
Method
Administration of the Spelling Test
All subjects were administered a spelling test (Spache, 1955).
Use of a spelling test that had words for second, third, and
fourth graders allowed for a broader range of words. The range
of words encompassed (a) a base for poorer spellers to respond,
and (b) a high enough ceiling for better spellers. The spelling
test was administered to each class as a group task. Record
forms were provided for each subject on which the subject's
answers were written. Each word was presented individually, used
contextually in a sentence, and then repeated. For each corres¬
ponding number on the answer sheet, a card was held up with that
number so that each child was able to mark the answer in the
appropriate space. Sufficient time was allowed for each child
to finish the word called.
Directions for the spelling task were:
"I am going to give you a spelling test. There will
be 48 words in all. There may be a few words that
you haven't studied before, but I want you to try to
spell these new words the best you can. This is not
one of your regular spelling tests. What you do on
this test will not affect your spelling grade. Let's
start now and see how well you can do.
I will read the word, then use it in a sentence, and
then say the word again. Wait until I say the word
the last time after the sentence before you begin to

26
write. As we move to each new word, I will hold up
a card with the number on your paper so that you
can make sure you are writing the word in the correct
space. Are there any questions?"
An example was used to assist the child to find the appropriate
writing space on the paper.
Scoring of the spelling test. Scoring of the spelling test
words was based on a procedure suggested by White and Haring
(1976). Using this procedure, each word was analyzed for the
number of units correct. Each word had one more possible unit
than the number of letters in the word itself. By using this
procedure, units were awarded for beginning the word on the
correct letter and for finishing the word with the correct letter.
Forty-eight words were selected from the Spache Errors Test
(Spache, 1955). There were a total of 224 letters for the 48
words plus 48 additional units according to the White and Haring
(1976) scoring procedure. Using this procedure 272 was the
highest spelling score attainable.
Administration of the Predictor Tasks
On three consecutive days, each subject was tested individually
in a place designated by the school principal. There were two
recorders, the investigator and a graduate assistant, working at
all times to facilitate the procedure with as little time inter¬
ruption as possible for each cooperating teacher. Each task was
a thirty-second sample of academic behavior. The obtained scores

27
were multiplied by two so that the standard measure of behavior,
movements per minute, could be used (Lindsley, 1972). Writing
paper and stimulus sheets (see Appendix C) were supplied for each
task for which it was required. Pencils were also supplied. All
subjects used one-quarter inch (,635 cm.) lined paper. All stimulus
sheets read by the subjects were presented on eight and one-half by
eleven inch (21 .59x27.94 cm.) unlined paper and were typed with
primary size type. All data recorders had appropriate record
sheets on which to record the subjects' responses. All data recorders
had a prior training and practice session.
For all Hear/Say tasks the directions were:
"Listen as I say each vowel (or consonant) sound.
As I say each sound, I want you to tell me the name
of the letter. Are you ready?"
Each Hear/Write task was administered in the same manner except that
the responses were written. Short vowel sounds were "a" as in apple,
"e" as in elephant, "i" as in indian, "o" as in ox, and "u" as in
umbrella.
For the two See/Say tasks the instructions were:
"I want you to look carefully at each nonsense word
(or addition fact). When I say go, I want you to
start here (point) and tell me what the word is (or
what the answer is). I want you to go across the
paper (point to the arrow at the top of the sheet
and follow it across the paper with your finger).
Are you ready?"

28
The instruction for the task of writing letters of the
alphabet was:
"On this sheet of paper I want you to write all
the letters of the alphabet from 'a' to 'z1. If
you should finish before I say stop, go back to
'a' and try and write as many more letters as you
can. I would like you to print the letters (manu¬
script). Are you ready?"
Scoring the Predictor Tasks. Each of the eight predictor
tasks was administered on three consecutive days. Each recorder
was responsible for scoring all oral response tasks. The in¬
vestigator scored all written tasks. Pilot data collected showed
that it was easier to correct the written tasks after the task
administrati on.
Each task presented two mean scores. The accuracy score was
a mean score based on the number of responses correct divided by
the total number of responses per task attempted. The second score
was a speed score. A mean speed score represented the average of
the total number of responses, correct and error, attempted per
subject on each task.
Data Analysis
The Biomedical Computer Programs BMD 07M (Dixon, 1973) was
accessed for a stepwise discriminant analysis procedure. The
program allowed for the use of a linear combination of measures
from which those measures contributing most to the variance of

29
the dependent measure were kept in the prediction equation. The
use of a two group discriminant analysis allowed for the dependent
variable to be dichotomous, in other words, to represent group
membership. The two-group discriminant analysis procedure pro¬
vided a classification schema by which an individual, based on
predictor variable scores, was assigned membership into that
group whose mean score he most closely approximated.
Three free-entry subproblem routines were performed. The
first routine was an analysis using both accuracy and speed
measures of the eight tasks. The second analysis used only
accuracy scores and the third analysis used speed scores.
Additional Analyses
A second program (BMD 07M, Dixon, 1973) was accessed based
upon the results of the first analysis. This second program, also
with three subproblem routines, used only those tasks from the
first program that were found to be the most reliable discrimina¬
tors. Three prediction equations were obtained from the three
subproblem routines of the second program. The results of these
equations, with a child's score placed in the formulae, compared
to the cut-off score, represent an individual's predicted group
membership, as a function of accuracy scores, speed scores, and
both accuracy and speed scores. Constants and coefficients for
each equation are different due to the contribution of the speci¬
fic performance measures used.

30
Additional information yielded by the BMD 07M and BMD 02R
programs (Dixon, 1973) will be discussed in Chapter IV. This
information includes variance accounted for in the dependent
measure, e.g., group membership, by all of the predictor
variables (see page 23) and possible misclassification of
individuals to each group.
The BMD 02R program (Dixon, 1973) was accessed a second
time. The variables included were (a) Hear consonant sound/Say
consonant name (CONS H/S); (b) Hear consonant sound/Write consonant
(CONS H/W); (c) Hear short vowel sound/Say vowel name (SHVL H/S);
and (d) Hear short vowel sound/Write vowel (SHVL H/W). The math
and alphabet writing tasks were not included. Two other tasks were
not included. They were: See CVC trigram/Say nonsense word (CVC)
and Hear two letter blends and digraphs/Write letters (BLDI).
Therefore, the four single component, auditory input tasks (con¬
sonant or short vowel; Hear/Say and Hear/Write) were entered in the
program. The use of only these four tasks was for the purpose of
analyzing their contribution as discriminating variables separate
from (a) more complex tasks (CVC and BLDI) and (b) tasks not
directly relevant to the development of spelling behavior. Three
free-entry subproblem routines were accessed. Accuracy scores
were used in the first subproblem. Speed scores were used in the
second subproblem. Accuracy and speed scores in combination were
used in the third subproblem.
Average accuracy and speed improvements were calculated for
each subject. This was done by subtracting the day one accuracy
or speed score from the day two score. The day two score was then

31
subtracted from the day three score. These two results were then
added together. The sum was divided by two. The resultant score
was the average accuracy or speed improvement per subject per
task. These scores were pooled to calculate the mean improvement
per group per task.

CHAPTER IV
RESULTS
The present study was conducted to generate a set of
predictor tasks that would discriminate between a group of
good spellers and a group of poor spellers. This set of tasks
was based on accuracy, speed, and accuracy and speed scores in
combi nation.
Specific questions asked were:
(1) Which tasks, with respect to accuracy, would
discriminate between good spellers and poor
spellers?
(2) Which tasks, with respect to speed, would
discriminate between good spellers and poor
spel1ers?
(3) Which tasks, with respect to both accuracy
and speed, would discriminate between good
spellers and poor spellers?
Question One
Results from the discriminant analysis procedure using
accuracy scores showed that two variables, (1) See CVC trigram/
Say nonsense word (CVC-A) and (2) Hear two letter blends and
32

33
digraphs/Write letters (BLDI-A) were reliable predictors signifi¬
cant at the .01 level. The following discriminant prediction
equation resulted:
Y = -15.7142 + 0.1805 (CVC-A) + 0.0768 (BLDI-A)
By inserting a child's CVC-A and BLDI-A scores in the above equa¬
tion the resultant score can be compared to the cut-off score of
+0.0034, above which a child would be predicted to be a good
speller and below which a child would be predicted to be a poor
speller.
Question Two
Results from the discriminant analysis procedure using speed
scores showed the two variables, (1) See CVC trigram/Say nonsense
word (CVC-S) and (2) Hear two letter blends and digraphs/Write
letters (BLDI-S) were reliable predictors significant at the .01
level. The following discriminant prediction equation resulted:
Y = -16.7490 + 0.2553 (CVC-S) + 0.3354 (BLDI-S)
By inserting a child's CVC-S and BLDI-S scores in the above equa¬
tion, the resultant score can be compared to the cut-off score of
+0.0002, above which a child would be predicted to be a good
speller and below which a child would be predicted to be a poor
speller.
Question Three
When the eight tasks were entered into the discriminant analysis
using both accuracy and speed scores for each task, four tasks were
found to be reliable predictors. Significant at the .01 level were,

34
(1) See CVC trigram/Say nonsense word (CVC-S) and (2) See CVC
trigram/Say nonsense word (CVC-A). Significant at the .05 level
were (1) Hear two letter blends and digraphs/Write letters (BLDI-
S) and (2) Hear two letter blends and digraphs/Write letters
(BLDI-A). The following discriminant prediction equation re-
sulted:
V = 23.6238 + 0.1818 (CVC-S) + 0.1329 (CVC-A) +
0.2560 (BLDI-S) + 0.0532 (BLDI-A)
By inserting a child's CVC-S, CVC-A, BLDI-S, and BLDI-A scores
in the above equation the resultant score can be compared to the
cut-off score of -0.0032, above which a child would be predicted
to be a good speller and below which a child would be predicted
to be a poor speller.
Additional Findings
Contribution of Predictor Variables in Accounting
for Variance in the Dependent Measure
Additional information yielded by accessing the BMD 02R
stepwise regression program (Dixon, 1973) is presented in Table 4.
Table 4
Predictor Variable Contributions
Performance
Measure
Variables in Order of
Entrance in the Program
Variance Accounted for in
the Dependent Measure
Accuracy
CVC-A(76.3%),BLDI-A(3%)
79.3%
Remaining 6 Variables
1.2%
Speed
CVC-S(76.6%) ,BLDI-S(2.7%)
79.3%
Remaining 6 Variables
1.6%
Accuracy
CVC-S(76.6%),CVC-A(6.5%),
and Speed
BLDI-S(0.9%),BLDI-A(0.7%)
84.7%
Remaining 12 Variables
2.25Í

35
Due to the use of free-entry subproblem routines, the first variable
entered accounted for unique and overlapping variance. Each subsequent
variable entered represented an incremental amount of unique and over¬
lapping variance not already accounted for by a previously entered
variabl e.
The results presented in Table 4 for accuracy alone and for speed
alone show that the CVC task was the major contributor to variance
accounted for in the dependent measure by each performance descriptor.
The CVC-S score accounted for 76.6% of the variance, which was
slightly more, .3%, than the CVC-A score (76.3%). Therefore, when
accuracy and speed scores were entered into the stepwise regression
program simultaneously, the CVC-S score entered first, showing both
unique and overlapping variance attributable to the CVC-S score. The
decrease in variance accounted for in the dependent measure by the
CVC-A score from 76.3%, when entered with accuracy scores, to 6.5%,
when entered with both accuracy and speed scores, is reflective of the
overlapping amount of variance shared by the CVC-S and CVC-A scores.
Initial and Predicted Group Assignment
Each subject was assigned initially to either the good spellers
group or the poor spellers group based on his/her spelling test score.
The range of scores for the good spellers group was from 85.3% to
99.6% correct. The range of scores for the poor spellers group was
from 7.7% to 49.5% correct. Results of the analysis indicate that
on the basis of the best predictor variable scores, certain individuals
would have been misclassified with respect to their initial group
assignment. Tables 5, 6, and 7 are shown in terms of initial assign¬
ment and predictor variable assignment.

36
Table 5
Accuracy and Speedas Performance Measures
Predicted Assignment
Good Poor
Initial
Good
34
1
Assignment
Poor
0
35
Thirty-four of the 35 subjects initially assigned to the good
spellers group were also assigned to this group based on predictor
variable scores. One subject classified initially as a good speller
was classified as a poor speller on the basis of his/her predictor
variable scores. All 35 of those assigned initially as poor spellers
were predicted to be poor spellers on the basis of their predictor
varibale scores.
Table 6
Accuracy as a Performance Measure
Predicted Assignment
Assignment
Poor
The entire group initially assigned as good spellers was pre¬
dicted to be good spellers. Thirty-three of the 35 subjects assigned
initially to the poor spellers group were also assigned to this group
based on predictor variable scores. Two subjects classified initially
as poor spellers were misclassified as good spellers on the basis of
their predictor variable scores.
Good Poor
35
0
2
33
l yn-

37
Initial
Assignment
Table 7
Speed as a Performance Measure
Predicted Assignment
Good Poor
Good
Poor
33
2
0
35
Thirty-three of the 35 subjects assigned initially to the good
spellers group were also assigned to this group based on their pre¬
dictor variable scores. Two subjects classified initially as good
spellers, were classified as poor spellers based on their predictor
variable scores. The entire group (n = 35) of initially assigned poor
spellers were classified as poor spellers based on their predictor
variable scores.
Use of the discriminant prediction equation misclassified 3% of the
good spellers group as poor spellers and 0.0% of the poor spellers group,
based on accuracy and speed scores. Six percent of poor spellers group
and 0.0% of the good spellers group were misclassified based on accuracy
scores. Six percent of the good spellers group and 0.0% of the poor
spellers group were misclassified based on speed scores.
Scatter plots are reported in Appendix G. Each scatter plot
represents one task and one performance measure of that task and its
relationship to the spelling test score (percent correct). Means and
standard deviations for the good spellers group and for the poor
spellers group on each task are reported in Appendix E. Raw score
data are reported in Appendix F.

33
Analysis of Single Component, Auditory Input Tasks
A supplementary analysis was conducted which included the four
single component, auditory input tasks (CONS H/S, CONS H/W, SHVL H/S,
and SHVL H/W). The CVC and BLDI tasks that were found to be the best
predictors in the original analysis were not included. The reason for
not including the CVC and BLDI tasks was that their relationship to
spelling performance was already known to be significant. The See one
place addition fact/Say answer (MATH) and Hear instruction/Write letters
of the alphabet (ALPH) tasks were also not included in the present
analysis. This was done for two reasons. First, the MATH and ALPH
tasks were found not to be related to spelling performance in the over¬
all analysis. Second, the MATH and ALPH tasks are not curriculum tasks
directly relevant to the development of spelling behavior.
In order to ascertain the relationship of single component,
auditory input tasks (CONS H/S, CONS H/W, SHVL H/S, and SHVL H/W) to
spelling performance, a separate analysis was performed. The results
are indicated in Table 8.
Table 8
The Relationship of Single Component, Auditory
Input Tasks to Spelling Performance
Performance
Measure
Variables in Order of Variance Accounted for
Entrance in the Program in the Dependent Measure
Accuracy
SHVL H/W (31.5%)
31 .5%
Speed
SHVL H/W (52.1%)
CONS H/W (5.4%)
SHVL H/S (2.2%)
59.7%
Accuracy
and Speed
SHVL H/W-S (52.1%)
SHVL H/W-A (5.7%)
CONS H/W-S (6.2%)
64.0%

39
All variables except one were significant at the .01 level. The
SHVL H/S tasks with speed as a performance measure was significant
at the .05 level .
Accuracy and Speed Change
Each of the eight tasks was administered on each of three
consecutive days. The group means presented in Table 9 represent
each groups average change over the three day period. Both groups
showed an increase in speed scores. The good spellers group
showed a decrease in accuracy on the ALPH task. The poor spellers
group showed a decrease in accuracy on the CVC and SHVL H/S tasks.
Table 9
Mean Change Per Task Per Group
TASK
GOOD
ACCURACY
(% Correct)
GROUP MEAN
SPEED
(Response/Mi n:)
POOR
ACCURACY
(% Correct)
GROUP MEAN
SPEED
(Response/M-
CVC
1.11
10.94
-1.47
2.94
CONS H/S
0.39
4.94
1.00
3.23
SHVL H/W
1.19
4.00
-0.63
4.63
MATH
0.09
4.91
0.17
1.97
BLDI
3.61
2.71
3.26
1.57
CONS H/W
0.56
2.99
0.77
2.26
SHVL H/S
1.30
1.43
1.11
1.46
ALPH
-0.04
4.37
0.10
2.31
MEAN CHANGE
+1.02
+4.53
+0.54
+2.55

CHAPTER V
DISCUSSION
The focus of the present study was to identify a set of
prerequisite tasks, modality relevant to spelling performance,
that when based on accuracy and speed as performance measures,
would discriminate between good spellers and poor spellers.
A review of related literature suggested that (a) tasks used
to date have represented more than one basic behavior per
task; (b) tasks that have been used are related to, but are
not direct replications of the input and output modality
structure of spelling behavior; (c) tasks used have not
represented the most basic skills assumed by curriculum
hierarchies to be prerequisite to spelling competence; and
(d) tasks have been assessed by accuracy alone, without re¬
ferring to speed (fluency) as a viable performance measure.
A spelling test was administered to a group of third
graders in Alachua County, Florida. From these results the
top 25.9 percent (good and poor spellers) were chosen (N=35
subjects per group). Each of these subjects was then assessed
on each of eight tasks on three consecutive days. Accuracy
and speed scores were obtained for each subject on each task.
40

41
Two of the eight tasks used, See CVC trigram/Say nonsense word
and Hear two letter blends and digraphs/'.Jrite letters, were the
two tasks with which to discriminate groups of good and poor third
grade spellers. The use of accuracy and speed scores in combina¬
tion revealed that speed was the better performance measure for
discriminating between the two groups of spellers.
The results will be discussed under the following
headings:
(1) Task complexity and learning hierarchies;
(2) Input and output modality structure of the task;
(3) Spelling test scoring procedure;
(4) Performance measures;
(5) Repeated measures of predictor tasks;
(6) Conclusions; and
(7) Recommendations for future research.
Task Complexity and Learning Hierarchies
The tasks used in the present study were selected from
spelling skill hierarchies and task analyses. The content of
six of the tasks followed the developmental progression of skills
found in the hierarchies. Two tasks were single component tasks
(CONS and SHVL). The findings indicated that at the third grade
level, the more complex skills are better at discriminating
between good and poor spellers than the single component tasks.

42
Gagne (1965) suggested an order assigned to skills within
a hierarchical structure (simple responses, chains or multiple
discriminations, classifying, use of principles or rules, and
problem solving). The CONS and SHVL tasks seem to be an example
of chains or multiple discriminations. The CVC task seems to
be an example of the use of principles or rules. For the third
grade subjects included in this study, simple response tasks
(CONS, SHVL) were not found to discriminate reliably between
good and poor spellers. One may conclude, then, that the simple
tasks do not account for a significant amount of the variance
in the spelling task itself. The CVC and BLDI tasks, or the
more complex tasks, were the tasks that were found to discriminate
reliably between the groups of good and poor spellers.
An additional analysis that included the four single com¬
ponent tasks (CONS H/S, CONS H/W, SHVL H/S, and SHVL H/W) lends
support to this line of reasoning. The SHVL tasks were the best
discriminators, regardless of performance measure.
The overall analysis indicated that the four tasks, with
content relevant to spelling behavior (CVC, BLDI, SHVL, and CONS),
followed a descending order from complex to simple. The order
was CVC, BLDI, SHVL, and CONS.
Although this study was not designed to validate any one
learning hierarchy, the results may be serendipitous in their
outcome. Of the four tasks based on spelling curriculum
hierarchies, the CONS tasks were least successful in dis-

43
criminating between good and poor spellers. The SHVL tasks
were slightly more successful. The BLDI task was more success¬
ful yet. The CVC task, the more complex task according to the
hierarchies, was the most successful task to discriminate between
good and poor spellers.
Input and Output Modality Structure of the Task
One of the two tasks found to be the best discriminators
of third grade spelling performance, Hear two letter blends and
digraphs/Write letters was a direct replication of the modality
structure of actual spelling behavior. The CVC task had the
modality structure of a reading task (See/Say). The contribution
of the BLDI task to the variance of the dependent measure ranged
from 0.7% to 3.0* depending on the performance measure(s) used
(See Table 4). The results indicated that although its contri¬
bution was statistically significant, the practical significance
of its contribution in explaining variance accounted for in the
dependent measure, is clearly limited.
Previous studies in the spelling literature have not made
use of a direct replication of the modality structure of actual
spelling behavior (Bond, 1935; Monroe, 1932; Russell, 1937;
Spache, 1940a; Wallace, Klein, & Schneider, 1968). An attempt
was made to ascertain whether or not auditory input tasks may
have been more related to spelling behavior than visual input
tasks. However, the CVC task, a reading task, was found to be

44
the task that best discriminated between the good spellers and poor
spellers. On the basis of the results of this study, one must agree
with Monroe's (1 932) statement that
The correlation [between reading and spelling is] so high
... we must be measuring an achievement which is greatly
dependent either upon reading or upon the same factors
which underlie the ability to read. (p. 13)
Spelling Test Scoring Procedure
A new procedure for scoring the spelling test (White 8 Haring,
1976) was used in this study. In a post hoc analysis, the spelling
tests were rescored using the more traditional words correct pro¬
cedure. A Pearson Product Moment Correlation was performed between
units correct (White 8 Haring) and words correct. A correlation of
+.97 was obtained.
The results of the correlation procedure indicated that either
scoring procedure could have been used in this study. The preference
of using the White and Haring (1976) scoring procedure would be to
learn more about a child's spelling behavior because of the exten¬
sive word analysis that would be obtained. The whole word procedure
would be preferable where scoring time was a primary consideration.
Performance Measures
The results of this study were obtained by considering the
original set of eight tasks in three ways. Accuracy scores were
used in the first subproblem. Speed scores were used in the second
subproblem, accuracy and speed scores in combination were used in
the third subproblem. Regardless of the performance measure used,
the CVC and BLDI tasks were found to be the best discriminating tasks

45
in the same order (CVC, BLDI). However, when accuracy and speed scores
were used in combination in a free-entry subproblem routine, the speed
scores on each task (CVC, BLDI) accounted for more variance in the
dependent measure than the accuracy scores (See Table 4).
Accuracy scores are not to be interpreted as less powerful than
speed scores but as representing incremental variance attributed
uniquely to each of the successive variables entered into the predic¬
tion equation. The near equivalence of accuracy and speed scores
entered by discrete performance descriptors disappears when the two
are entered simultaneously. This phenomenon is evidenced as the step¬
wise regression program selects the variable that is most highly
correlated with the dependent variable. This first variable, and
each successive variable, enters showing unique and overlapping variance
accounted for in the dependent measure. Each successive variable, then,
is selected to enter on its contribution of variance in the dependent
measure. An analysis of single component, auditory input tasks (CONS,
SHVL) indicated that speed scores were again slightly more powerful
in their ability to discriminate between good and poor spellers than
were accuracy scores (See Table 8).
Speed and accuracy scores on the tasks were found to be equivalent
discriminators between the two groups of spellers. An indication that
remediation or development of the skills on these tasks needs to in¬
clude both accuracy and speed before final proficiency can be noted.
This indication is suggested by the fact that speed and accuracy were
equivalent performance measures. Based on a mean score of the good
spellers group (N=35) on each task, the following accuracy and speed
scores may be helpful in determining proficiency levels of the tasks.

46
Table 10
Proficiency LeveIs Per Task
TASKS
MEAN SCORE
CVC (accuracy)
CVC (speed)
BLDI (accuracy)
BLDI (speed)
91.61 correct
63.2 responses per minute
86.5% correct
24.0 responses per minute
Speed and accuracy scores are also relevant to the issue of
possible misclassifications. Children predicted to be good or
poor spellers and who are indeed good or poor spellers will receive
appropriate curricular instruction. Children predicted to be poor
spellers who are indeed good spellers will certainly not suffer
from extra curricular intervention. However, children predicted
to be good spellers who are indeed poor spellers may suffer from
lack of instruction. The lack of instruction for any child that
needs it may cause teacher and learner frustration. The misclassi-
fication of this last group of children, those predicted to be good
spellers but who are not good spellers, needs to be minimized.
The results of this study indicated that when accuracy and
speed scores in combination or speed scores alone are used as
performance measures, no children are misclassified by being
predicted as good spellers when they are indeed poor spellers
(See Tables 5 and 7). The small percentage of misclassifications

47
by accuracy scores appears to be a reflection of the N used in
this study. Therefore, speed, accuracy, or a combination of
speed and accuracy could be used to reduce misclassifications
that could be harmful to the learner.
Group standard deviations on each task presented an interest¬
ing descriptive finding. On four of the tasks included in this
study (CVC, MATH, BLDI, and ALPH) the poor spellers group dis¬
played more variability with respect to accuracy than did the
good spellers group. However, on these same four tasks, the good
spellers group displayed more variability with respect to speed
than did the poor spellers group. Although standard deviations
for each group on each task are presented in Appendix E, Table
11 presents that standard deviation data relevant to the CVC,
MATH, BLDI, and ALPH tasks.
Table 11
Standard Deviations on the CVC, MATH, BLDI, and ALPH Tasks
TASK
STANDARD DEVIATION
GOOD GROUP
STANDARD DEVIATION
POOR GROUP
CVC (accuracy)
10.0
22.0*
CVC (speed)
16.1*
7.4
MATH (accuracy)
4.8
23.0*
MATH (speed)
12.6*
8.7
BLDI (accuracy)
11.3
27.4*
BLDI (speed)
5.6*
3.2
ALPH (accuracy)
1.0
2.9*
ALPH (speed)
14.0*
7.8
more variability

48
One may infer from these findings that the poor spellers
group, with a wider dispersion of accuracy scores and less
variability in speed scores, has not completed the basic
stage of acquiring accurate responses. The good spellers
group, however, was less variable on accuracy scores. Their
accuracy scores may indicate mastery of the tasks included
for this study. However, the variability in their speed scores
leads one to believe that they have not, as a group, reached a
fluent level of performance on these four tasks.
It has been suggested (Haughton, 1972; Starlin, 1972) that
although the final proficiency aim for any child on any task is
accurate responding at an acceptable speed, accurate responding
needs to be in a child's repertoire before accurate responding
at an acceptable speed can be obtained. In this study, the poor
spellers, as a group, were responding neither accurately nor
fluently. The lack of variability in their low speed scores
suggests that as a group their speed scores may be rather con¬
sistent because they have not yet mastered the stage of accurate
responding. This would account for an overall slow speed of
responding on these tasks. The good spellers group, because of
the lack of variability in their high accuracy scores, seems to
have mastered the stage of responding accurately. However, the
variability in their speed scores leads one to believe that as
a group they are not yet fluent on these four tasks.
In summary, the findings of this study indicated that the
use of speed and accuracy as performance measures may serve two

49
useful functions. First, speed and accuracy scores discriminate
between good and poor spellers. Second, speed and accuracy scores
reduce misclassifications that may interfere with a child's curricular
program.
Repeated Measures of the Tasks
Repeated measures over three days for each task were obtained
to represent a more reliable measure for each subject's performance
on each task. Group means are reported in Table 9.
Both groups of spellers increased their speed scores over the
three day period. The good spellers, as a group, did not improve
their accuracy scores on the ALPH task, due to the ceiling effects
of the task. The poor spellers, as a group, did not improve their
accuracy scores on the CVC and SHVL H/W tasks.
Low increases in accuracy scores for the subjects involved
may have been due to four reasons. First, some subjects maintained
100% accuracy over the three day period. Therefore, due to ceiling
effects, their accuracy scores could not increase, while their
speed scores may have increased. On the CVC task, subject #103
remained at 100% accuracy for the three days while the subject's
speed score change was +20.9 responses per minute over the three
days.
Second, some subjects maintained a 0.0% accuracy score over
the three day period. Subject #230 maintained a 0.0% accuracy
for three days on the BLDI task while the subject's speed score
changed by one incorrect response per minute.

50
Third, some subjects, while increasing their speed scores,
lowered their accuracy scores. For example, on the CVC task,
subject #116 had speed scores of 48, 60, and 85.8 responses per
minute for the three days. This subject's accuracy scores were
100, 100, and 97 percent correct for the three days. As the
subject's speed increased the chance for error increased.
Therefore, due to a single error on day three, this subject's
average accuracy change was -1.5% while the average speed change
was +18.9 responses per minute.
Fourth, some subjects who may have had an increase from day
one to day two may have had a decrease from day two to day three
or vice versa. Depending on the differences between the days,
some subjects may have shown no change and others an increase
or decrease. Subject #117 had a -6 speed increase from day one
to day two and a +12 speed increase from day two to day three,
with an overall speed change of +3 (CVC task). On the BLDI
task, subject #225 had a speed change of +8 from day one to day
two and a speed change of -8 from day two to day three. This
subject's overall change in speed on the BLDI task was zero.
The results of these findings indicated a positive change
in speed on all tasks for both groups. The good spellers group
exhibited a positive change in accuracy scores on seven of the
tasks. The poor spellers group exhibited a positive change in
speed on six of the tasks. Each group exhibited change on each
task over the three day period. Although the changes were
generally not large, it was evident that two or three samples

51
of each task was a closer approximation of each child's maximum
capability on each task. Therefore, the contention that three days
of data collection as a more reliable measure of a subject's per¬
formance than one day of data collection was upheld.
Concljjsj ons
The focus of this study and the specific questions asked were
stated in terms of identifying a set of prerequisite tasks that
would best discriminate between good spellers and poor spellers.
The set of tasks used in this study was discussed in terms of
accuracy and speed as performance measures both separate and in
combination. The results of the data analysis indicated that when
accuracy and speed were used as separate performance measures, the
same two tasks were found to be the best discriminators between the
groups of good and poor spellers. The two tasks were See CVC
trigram/Say nonsense word and Hear two letter blends and digraphs/
Write letters. When accuracy and speed scores were used in combina¬
tion as performance measures, the CVC and BLDI tasks again were
found to be the best discriminators between the two groups of
spellers. However, when used in combination, the speed score for
each task was a slightly more powerful discriminator than was the
accuracy score due to the free-entrv variable selection (See previous
discussion).
Five of the tasks selected for inclusion in this study were
direct replications of the input and output modality structure
of spelling behavior. One task was a reading task; the other two
tasks were a math task and an alphabet writing task. Although

52
the BLDI task, a Hear/Write task, was one of the two best tasks
in discriminating between groups of good and poor spellers, it's
contribution to the variance accounted for in the dependent
measure was extremely small (See Table 4). The CVC task, a
reading task, was the best discriminating task. Therefore, the
use of tasks involving the input-output modality structure that
directly replicates spelling behavior was less reliable than the
reading task in discriminating between the two groups of spellers.
Supplementary analyses of the data indicated that the use
of speed, as a performance measure, classified children for
curricular programming with slightly less liability than the
use of accuracy. Speed was also the performance measure that
showed the greatest change over the three days of data collection.
The spelling test scoring procedure used in this study was found
not to be time saving. However, it presented a more extensive
analysis of a child's spelling performance than the words correct
procedure. Although this study did not attempt to validate any
one spelling skill hierarchy, the findings supported a sequence
of the tasks used, from simple to complex.
Recommendations for Future Research
The present study has included multiple regression and dis¬
criminant analysis procedures. Through the use of these procedures,
this investigator has obtained information that is useful in (a)
discriminating between groups of good and poor spellers and (b)
discussing the most reliable discriminating tasks. The findings

53
have been focused on issues such as (a) task complexity and
learning hierarchies; (b) the input and output modality structure
relevant to spelling behavior; and (c) the utility of accuracy
and speed as performance measures. The resultant information
from this study provides one with a basis upon which a second
stage of more functionally oriented research may be based. This
second stage of research, then, should consist of studies that are
concerned with the functional relationship of the present findings
to classroom spelling curricula.
This task (BLDI) had the modality structure that directly
replicated spelling behavior. The BLDI task was less reliable
as a discriminator than the reading task. It would be interesting
to include the Hear/Say and Hear/Write modality structure as a
unique experimental variable. Efforts could then be made to
determine if curricular instruction, with an emphasis on these
modality structures (Hear/Say or Hear/Write), would enhance the
ability to spell. The curricular emphasis in many classrooms is
on reading. There are those children, however, that even with
reading instruction are poor spellers. Perhaps an emphasis on
related content material, with the modality structure directly
replicate of spelling behavior, would improve this poor spellers
group's ability to achieve competence in spelling.
The present findings have demonstrated the utility of speed
(fluency) as a viable performance measure that needs to be
emphasized equally with accuracy in classroom instruction. The
results, using speed and accuracy, were similar. Speed, as a

54
performance measure, has not been used before and may be a novel
means by which variance in performance can be explained. Research
efforts are needed that will use speed as an aim for skill mastery.
These research efforts need to consider the efficacy of remediating
deficit skills (i.e., CVC, BLDI) to a level of fluent mastery.
Once these skills have been mastered, a determination of their
effectiveness in improving spelling performance may be conducted.
It is indicated from the findings that for the third graders
included in this study, the more complex tasks (CVC, BLDI) discrim¬
inated best between the two groups of spellers. Further research
efforts at different grade levels (i.e., one, five) may demon¬
strate that different skills discriminate at different grade
levels or that by a certain grade level, the same tasks can be
used as discriminators. This may well serve to provide a
developmental sequence of skills. Performance on this sequence
could alert teachers to possible concomitant performance in spel¬
ling. As stated earlier, if other research efforts suggest that
mastery of deficit skills is effective in improving spelling, a
criterion performance sequence may have excellent curricular
merit.
The present study has answered the specific questions for
which it was designed. The overall findings as well as the
additional findings have suggested possible alternative research
efforts that may help solve the problem of presenting classroom
instruction that could promote better spelling achievement.

REFERENCES
Allen, D., & Ager, J. A factor analytic study of the ability
to spell. Educational and Psychological Measurement,
1965, 25, 153-161.
Bannatyne, A. Reading: An auditory vocal process. San Rafael,
Cal.: Academic Therapy Publications, 1973.
Bannatyne, A., & Wichiarajote, P. Relationships between written
spelling, motor functioning, and sequencing skills. Journal
of Learning Disabilities, 1969, 2(1), 4-16.
Birch, H. G., & Belmont, L. Auditory-visual integration, intelli¬
gence, and reading ability in school children. Perceptual
and Motor Skills, 1965, 20, 295-305.
Blair, T. R. Spelling, word attack skills; WRIC/RCS report.
Reading Teacher, 1975, 28(6), 604-605.
Bond, G. L. The auditory and speech characteristics of poor
readers. Contributions to Education, No. 657. New York:
Teacher's College, Columbia University, 1935.
Cahen, L. S., Craun, M. J., & Johnson, S. K. Spelling difficulty-
a survey of the research. Review of Educational Research,
1971, 41(4), 281-301.
Campbell, D. T., & Stanley, J. C. Experimental and quasi-
experimental desiqns for research. Chicaqo: Rand McNally,
1966.
Cobb, M. E., Kincaid, M., & Washburn, M. E. Further studies of
the verbal abilities of poor spellers. American Journal
of Psychology, 1918, 1_9, 331-332.
Cook, W. A., & O'Shea, M. V. The child and his spelling.
Indianapolis: The Bobbs-Merri11 Company, 1914.
Dixon, W. J. (Ed.). Biomedical computer programs. Berkeley:
University of Cal ifornia" Press , 1973.
Freeman, F. N. An experimental analysis of the writing movement.
Psychological Monographs, 1914, T7, 4.
55

56
Fries, G. C. Linguistics and reading. New York: Holt, Rinehart,
and Winston, 1963.
Gagne, R. M. The acquisition of knowledge. Psychological Review,
1962, 69, 355-365.
Gagne! R. M. The conditions of learning (1st ed.). New York:
Holt, Rinehart, and Winston, 1965.
Gagne"! R. M. Learning hierarchies. Educational Psychologist,
1968, 6(1), 1-6. '
Gagne"! R. M. The conditions of learning (2nd ed.). New York:
Holt, Rinehart, and Winston, 1970.
Griffith, S. , Bishop, J., Brown, R., Cohen, C., Islar, M., Pinder,
B., & Steen, C. Cognitive skills planning system (part 2).
Gainesville, FI.: Alachua County School Board, 1971.
Hanna, P. R., 8 Hanna, J. S. Applications of linguistics and
psychological cues to the spelling course of study. In
T. D. Horn (Ed.), Research on handwriting and spelling.
Champaign, Ill.: National Council of Teachers of English,
1966.
Haughton, E. Aims: Growing and sharing. In J. B. Jordan 8
L. S. Robbins (Eds.), Let's try doing something else kind
of thing. Arlington, Va.: Council for Exceptional
Children, 1972.
Hollingshead, A. B., 8 Redlich, F. C. Social class and mental
illness: A community study. New York: John Wiley and Sons,
1958. “
Horn, E. Spelling. In C. W. Harris (Ed.), Encyclopedia of
educational research. New York: MacMillan, 1960.
Horn, T. D. Handwriting and spelling. Review of educational
research, 1967, 37_, 168-177.
Horn, T. D. Spelling. In R. L. Ebel (Ed.), Encyclopedia of
educational research. London: MacMillan, 1969.
Hughes, V. H. A study of the relationship among selected language
abilities. Journal of Educational Research, 1953, 47, 97-
106.
Learner, J. W. Children with learning disabilities. Boston:
Houghton Mifflin, 1971.

57
Lindsley, 0. R. From Skinner to precision teaching: The child
knows best. In J. B. Jordan & L. S. Robbins (Eds.), Let's
try doing something else kind of thing. Arlington, Va.:
Council for Exceptional Children, 1972.
Monroe, M. Children who cannot read. Chicago: University of
Chicago Press, 1932.
Pennypacker, H. S. Why frequency? The rationale of precise
behavior management. Unpublished manuscript. University
of Florida, 1972.
Personke, C. R. Use of nonsense words to test generalization
ability in spelling. Elementary Enqlish, 1972, 49, 1233-
1239.
Random House. High intensity learning systems. New York:
author, 1972.
Russell, D. H. Characteristics of good and poor spellers.
Contributions to Education No. 727. New York: Teacher's
College, Columbia University, 1937.
Spache, G. Characteristic errors of good and poor spellers.
Journal of Educational Research, 1940, 34_, 182-189. (a)
Spache, G. The role of visual defects in reading and spelling
disabilities. American Journal of Orthopsychiatry, 1940,
10, 229-238. (bl
Spache, G. Spelling disability correlates I - factors probably
causal in spelling disability. Journal of Educational
Research, 1941, 34(8), 561-586. fa)
Spache, G. Spelling disability correlates II - factors that
may be related to spelling disability. Journal of Educa¬
tional Research, 1941 , 35^, 119-137. (b)
Spache, G. D. Spelling errors test. Gainesville, FI.: Reading
Laboratory and Clinic, University of Florida, 1 955.
Starlin, A. Sharing a message about curriculum with my teacher
friends. In J. B. Jordan & L. S. Robbins (Eds.), Let's
try doing something else kind of thing. Arlington, Va.:
Council for Exceptional Children, 1972.
Starlin, C., & Starlin, A. Curriculum ladder for spelling.
Bemidji, Minn.: Unique Curriculums Unlimited, 1972.
Stiles, R. L. Stepwise multiple regression study of SST basic
movement cycle data with SFTAA/CTBS results from 348 third
grade pupils. Research Department, Pupil Personnel Services,
Tacoma Public Schools, 1974.

58
Templin, M. C. Phonic knowledge and its relation to the spelling
and reading achievement of fourth grade pupils. Journal of
Educational Research, 1954, 47, 441-454.
Wallace, J., Klein, R., & Schneider, P. Spelling ability and the
probability texture of English. Journal of Educational
Research, 1 968, 6]_(7), 315-319.
Westermann, G. S. Spel1ing and writinq. Sioux Falls, S.D.: Dimen¬
sions Publishing, 1971.
White, 0. R., & Haring, N. G. Exceptional teaching. Columbus,
Ohio: C. E. Merrill, 19761
White, R. T. Research into learning hierarchies. Review of
Educational Research, 1973, 43(3), 361-375.
Wo 1king, W. D. Rate of growth toward adult proficiency: Differences
between high and low achievement children. Paper presented at
the International Symposium on Learning Disabilities, Miami
Beach, Florida, 1973.

APPENDIX A
SPELLING TEST

60
SPELLING TEST
NAME: SCHOOL:
1.
bi te
25.
cookies
2.
and
26.
march
3.
dark
27.
books
4.
bush
28.
mice
5.
bags
29.
her
6.
bottom
30.
rich
7.
asleep
31 .
glass
8.
ate
32.
field
9.
bul 1
33.
though
10.
four
34.
bake
11.
fasten
35.
alone
12.
later
36.
lily
13.
cotton
37.
obey
14.
cave
38.
pony
15.
di d
39.
hot
16.
hatch
40.
street
17.
rent
41.
an
18.
gi ant
42.
patch
19.
laughing
43.
parties
20.
am
44.
seem
21.
broom
45.
bonnet
22.
melon
46.
already
23.
dol1ars
47.
heap
24.
deer
48.
pear

APPENDIX B
ACADEMIC TASKS

62
I. SEE CVC TRIGRAM/SAY NONSENSE WORD
meb
wid kum
vak
rop
het
di v
len
jad
nup
§ correct
zik
fap rud
tek
yab
vit
suf
pid
rab
nen
if attempted
ket
tav dak
fot
1 ub
das
pit
ni v
dup
maf
accuracy
Speed
/I minute
II.
HEAR CONSONANT
' SOUND/SAY CONSONANT NAME
c f
h n q
t V
k
y b
g m
p
s j
w z
d
X
if correct
r 1
n k h
f c
b
y V
t q
s
p m
j 1
r
g
# attempted
x d
z w c
q g
W
1 f
t j
z
h v
m d
n
S
accuracy
Speed
/] minute
hi.
HEAR SHORT VOWEL SOUND/WRITE
VOWEL
e o
a i u
o e
u
i a
i u
ft correct
e o
u i a
o e
i
a u
o e
i attempted
a i
u i e
o a
u
e i
o a
accuracy
Speed /I minute

63
IV. SEE ONE
PLACE ADDITION FACT/SAY ANSWER
1+1=2
2+0=2
5+4=9
6+2=8
8+1=9
ft correct
0+6=6
5+3=8
2+4=6
7+0=7
2+5=7
# attempted
1+5=6
0+3=3
6+3=9
6+1=7
2+8=10
accuracy
4+1=5
3+7=10
3+2=5
4+5=9
1+0=1
Speed
/I minute
0+2=2
2+6=8
2+3 = 5
1+4=5
7+2=9
3+6=9
2+7 = 9
4+3=7
4+2=6
1+8=9
4+6=10
7+3=10
9+1=10
3+0=3
5+0=5
1+9=10
6+4=10
8+2 = 10
5+1=6
0+9=9
V. HEAR TWO
LETTER 1
3LENDS ANC
1 DIGRAPHS/WRITE LETTERS
br th
sk ch
cr bl
sin sh
cl
# correct
tr tw
fl dr
sn gl
fr sp
pi
# attempted
gr si
pr st
sw tr
ch dr
br
accuracy
Speed
/I minute
VI. HEAR CONSONANT SOUND/WRITE
: CONSONANT
c f h
n q
t v k
y b g
m p s j w
z
d
X
# correct
r 1 n
k h
f c b
y v t
q S p m j
i
r
g
# attempted
x d z
w c
q g w
1 f t
j z h v m
d
n
S
accuracy
Speed
/I minute
VII. HEAR SHORT VOWEL
SOUND/SAY
VOWEL
e o a
i u
o e u
i a i
u
_
ft correct
e o u
i a
o e i
a u o
e
# attempted
a i u
i e
o a u
e i o
a
accuracy
Speed
/I minute

64
VIII. HEAR INSTRUCTION/WRITE LETTERS OF THE ALPHABET
A
M
N O P Q
W X Y z
it correct
# attempted
accuracy
/I minute
Speed

APPENDIX C
SEE CVC TRIGRAM/SAY NONSENSE WORD STIMULUS SHEET

MEB WID KUM VAK ROP
66
Ll
<
a v: CQ 1— CL
lu c o rz>
o i— cl: lu q
2: CD
UJ ZD
_J Cd
t— ^
LU 1—t
nr r^j
LO
c
o

APPENDIX D
SEE ONE PLACE ADDITION FACT/SAY ANSWER
STIMULUS SHEET

1+1= 2+0= 5 + 4
0+6= 5+3= 2+4
1+5= 0 + 3= 6 + 3
4 + 1= 3+7= 3 + 2
0+2= 2 + 6= 2+3
3 + 6= 2 + 7= 4 + 3
4 + 6= 7+3= 9 + 1
1+9= 6 + 4= 8 + 2
6 + 2 =
7 + 0 =
6 + 1 =
4 + 5 =
1 + 4 =
4 + 2 =
3 + 0 =
5 + 1 =
8 + 1 =
2 + 5 =
2 + 8 =
1 + 0 =
7 + 2 =
1 + 8 =
5 + 0 =
0 + 9 =
C7>
CO

APPENDIX E
GROUP MEANS AND STANDARD DEVIATIONS

Table E-l
Group Means and Standard Deviations
Variabl e
Mean
Good Group
Standard Deviation
Mean
Poor Group
Standard Deviation
See CVC trigram/Say nonsense word -
Accuracy
91.6
10.0
31 .2
22.0
See CVC trigram/Say nonsense word -
Speed
63.2
16.1
18.6
7.4
Hear consonant sound/Say letter name -
Accuracy
93.0
5.0
88.5
6.3
Hear consonant sound/Say letter name -
Speed
39.5
5.1
31.5
5.4
Hear short vowel sound/Say vowel name -
Accuracy
58.7
19.9
32.8
19.0
Hear short vowel sound/Say vowel name -
Speed
30. 3
5.9
18.8
5.2
See one place addition fact/Say answer -
Accuracy
97.5
4.9
86.0
23.0
See one place addition fact/Say answer -
Speed
44.2
12.6
29.1
8.7
Hear two letter blends and digraphs/
Write letters - Accuracy
86.5
11.3
34.3
27.4
Hear two letter blends and digraphs/
Write letters - Speed
24.0
5.6
13.6
3.2
Hear consonant sound/Write letter -
Accuracy
92.5
4.9
86.8
8.7
Hear consonant sound/Write letter -
Speed
34.7
5.7
23.8
5.1
Hear short vowel sound/Write vowel -
Accuracy
58.9
20.8
36.5
18.2
Hear short vowel sound/Write vowel -
Speed
32.2
3.5
23.9
5.0
Hear instruction/Write letters of the
alphabet - Accuracy
99.5
1.0
97.9
2.9
Hear instruction/Write letters of the
alphabet - Speed
57.9
14.0
35.1
7.8

APPENDIX F
RAW SCORE DATA

72
Raw Score Data
The first column (S's) consists of each subjects number
within the good spellers group (100's) or the poor spellers
group (200's). The DEP column is each subjects spelling test
score in percent correct. The three numbers are read as a
one place decimal number (e.g., 99.6).
The last eight columns are the raw scores per subject on
each of the eight tasks. The tasks are:
Code Task
CVC H/S
See CVC trigram/Say nonsense word
CONS H/S
Hear consonant sound/Say consonant name
SHVL H/W
Hear short vowel sound/Write vowel
MATH S/S
See one place addition fact/Say answer
BLDI H/W
Hear two letter blends and digraphs/Write
1etters
CONS H/W
Hear consonant sound/Write consonant
SHVL H/S
Hear short vowel sound/Say vowel name
ALPH H/W
Hear instruction/Write letters of the
alphabet.
The eight numbers
for each task are read as two four digit numbers
with a one place decimal (e.g., 09900774 = 99.0 and 77.4). A zero
was used in the hundred's column when a subject's score was less
than 100. The first number in each column is the accuracy score for
the task. The last number in each column is the speed score for the
task.

S'S DEP CYC S/3 Ca^S H/S SHVL H/W MATH S/S BLDI H/W CONS H/W SHVL H/S
ISS S85888S?
. _ _ >93 0
103 909 1Q00QB6O
IQ4 982 09940837
105 978 09100657
106 979 0947Q727
107 963 09230586
108 967 1Q000607
09 967 09100621
1 10 967 Q047 0660
111 960 10000551
112 960 09670707
113 952 094/0456
114 945 o9?70540
115 933 09100877
116 934
1 1 7 93 0
11
923
923
912
121 912
122 912
123 393
124 393
125 886
09900646
1 0 0 0 0 9 4 0
090T0353
09576534
08830393
08550600
09900595
U79Ü0564
10000635
07150858
100004c
09530420
09830427
09830367
08830387
09030327
09b70393
09870367
08000580
10000367
09530455
09200407
09000347
09270453
Q9b30300
Í 0 0 0 0 3 4 7
85588315
09260427
08900407
Ü980O433
08770387
09700500
09230453
08970473
09500393
832S212?
09230360
08570340
09400320
09630353
134 8uü ÍQ040657 083go347
135 e53 054/0320 09070313
126 866 09250705
127 882 09100636
128 875 09570763
129 871 09850733
130 8btt 079/6539
868 Q9200347
¡68 0
7200431
9700452
04770307
05900373
06330300
06730307
05000333
05370353
09530260
05000307
05970273
06100340
07000313
04100407
05670207
08370360
0 1 70 0360
09170307
09400307
04300407
08770260
07270273
03530300
06500387
03770320
08630393
05730227
02630207
04900287
04600390
07030300
04800320
03170247
05970207
06930213
07730247
03170220
1000055?
09600333
10000513
10000821
10000480
10000367
09630540
10000420
1000Ü560
09170393
10Ü0Ü420
09770420
10Ü0O3ÓO
09800340
09900527
09430400
10000353
09670447
09670387
0963OÜ0D
09900487
10000353
10000804
09070457
10000493
10000407
09670360
09830440
10000500
08970340
09500293
07400160
09600320
10000467
10000547
09200260
09700400
09730233
08370247
10000260
09070293
08130207
08630300
10000180
04930227
09270367
10000473
09500427
09270300
09470340
09100380
09370300
08730300
09600340
09000340
09130233 10000707
09400360 09370500
10000220 09500300
08900260 09000340
08770340 08830413
08400173 09070307
07730180 09300293
06270293 00070373
10000213 10000200
00730227 09000407
072302.40 08830353
09470240 10000373
08130193 09530300
08870260 09u704O7
06730207 08800333
07700260 090/0367
06770173 09300300
09370313 09130367
00170227 09030347
09370220 09770327
06870160 09170247
06230207 089003QO
09370107 10000200
03500180 07930273
0Ó30Q213 09000293
05570347
05700327
06570520
06630340
06400313
05570333
08130327
04670300
07630367
05430367
07230293
03500360
07470253
06670320
01730347
09100340
09730267
063/0327
03630327
07570333
03600306
09800340
05300360
03700287
04500287
03330340
07370373
04200287
02330227
06370267
05670293
07270327
03100333
ALPH H/W
10000573
10000707
10000680
10000700
10000607
10000607
10000807
09900720
09770533
10000507
10000547
09900567
10000707
10000527
09900693
10000447
0000487
0000560
9600533
0000640
09770580
09670313
09930860
09670580
10000607
10000673
10000407
10000673
09770573
09870447
10000360
10000533
10000253
10000420
10000553

> > -ó
ruru rururu
ruru rururu
ruru rururu
rururu ruro
ruru rururu
ruru rururu
rururu ruru
c/>
LaJLaíIaILaJLaI
tfjru rururu
rururu ru ru
fU»-» H—■ > .
►—»*—*—» *■•* *■
•—o ooo
ooooo
—
©43 cd—jo'
U) tUJ'U*—
O 43 CD"40'
LTJ t La* ru*—
O -O ÍX>"JCT
LntLuru*—
(/>
OOm^.
ruru rururu
rvrururu UJ
WLJL-JUJUJ
OJUJU4 rr
tt ttt
1111t
O
"JCtrVJ CD-O
L4LaI UlO’-'-J ■
"JCujD-üo
oruru L4t
Cr ru oo—
LaJLaJ "*"JCD
CP 43 43 4) 43
m
•"JO*— CD OI
LH43 ©—O'
4WO CP—
UTO"4LT>43
t Os'JO LH
or*- cxru
vOlul-j cr- O'
â– Q
ooo oo
oo'ooo
OOOOO
ooooo
o o ooo
OOOOÓ
oooo o
O
©oru— ru
©o o- cr¬
© —CTCa—
t*— o rvus)
Jto La-t"4
ujuirui^cr
ru"Ji\.Lnu-
-c
towo^)
oo iL-r— o
O O'O' o o
CD--IO--KP
O TOWJJ
LT»"4 CULa»./!
"-iUICDt "4
n
"40 "JLaC
ool Va' Ia-O
C "4"4©"J
Ca."J"JC©
La La- LaJ
La. La* IaO"J
"JCC "4 LaJ
o o ooo
oo ooo
ooooo
ooooo
ooooo
OOOOO
OOOOO
— o*- — ru
o—— ru—
ru— ruru—
ru —•— ruru
— ru —ru—
ruru—t —
co
-t-4 O'CP CP
o-ru ocw
"4La»0 tCD
005^
t O'OH 4)0*
O'cnac-j)
ru t-oru co
N.
OWOO-J
"JO O "404
U4LJ"J"4"J
"|"41a»L*40
"JO "JO"J
"U"4UJ"40
CO
©©coo
1
co oo o
0-0000
ooooo
a»OOC o
©oc©o
©OOO o
n
CP—JCP 43 43
4>"1 CP 43 CD
4> CD CP CP CP
03 43 0J0 43
o 43 00 4)4)
CD CD CD 033
CD UOi O 43
2
CXULaJ Lx—
-H UI -CO-Ü
->CX:Ia-£)
O'La. o IU—
O La- •CLa.O'
—JLT truer
O LaJCLTO
©"4 04 L4 "4
"JO '"J-JCrJ
•jjoow
OOOUJ-J
O © O O ~4
m "j ouo
LaIO "JLaJ "4
co
ooo oo
oo ooo
ooo o o
ooo oo
ooooo
oo ooo
OOOO o
*—Uirvi^jpj
LaJíU WWí:
LaILaíLHIU t
rUAJU4U4L>4
L4ÍU LaJLaJLaI
La* La* U4LHU4
laj ruru LaJíu
X
aox:o t
crt vcoo
11 ru "jo
l*4 ex: "j oru
o ojo-o-j
"JCXOLaJO
-OCXO© t
\
"1-slOO"!
OO 04 0 0
OOvlLHO
WOW"J"4
"J-J-JOW
w-jowo
1/4 "4 "f O "4
co
ooo oo
oo ooo
ooooo
ooooo
ooooo
ooooo
OOOOO
co
^ r\j
to O La.TU
— WtWLH
Ia*—LaUa»—
t La. La- t "4
t fUfUlr-t
UvCT-LP,—"4
X
lh 4>— ru cr
PJt wow
L4 !U O'O' CU
â– Jwo Coen
u- jjct 4)0-
IaJLaI4>—t
olajo— ru
<
IaJL/«0"4"4
"4 "4 "JO O
O LJOOJLaI
oowso
â– sJOIa-sIO
U4"JOO"4
L*400trJ0
r~
ooooo
OO OOO
ooooo
oo ooo
ooooo
•OOOOO
oooo o
—— — ——
ru—ru — ru
— ruru—ru
*—“ ru*— ruru
ruu4*— — ru
ru fu ru— ru
X
uio-ru to
11 11 43
— «con—t
O'© -j"jru
to ruujru
ujo ruuit
O'Uioru—
\
WOvJ^O
OO O "4 L4
La*La»04i_aI"4
"JO LaILaJ"4
ooowo
vaJ"4"4La*"4
•"«La* "4—4 La*
X
I
ooooo
oo ooo
— o— O
ooooo
— OOOO
— 0 OOO
0 — 0—-o
X
oru-o 4) O'
o-o ooorj
O 3)00 -0
CDO CP 43"J
© CP "4 -O 43
0 43 43-043
430 -OO "4
I>
Iajct rúa en
a t o-lu^j
CCC O O'
-"‘o -c : re ©
© -CO' Cr ex
© t G'tLT
encere ru
—J
"J "4 "4 CrJ "J
"40 0014
O U4 O O "4
WOO-4W
O O "4 "40
Q"JWOO
"40lrJ0"J
X
©©O O©
c occc
cccc ©
ooo co
0 cocc
ooooo
eco 0 0
iwruru'^ffu
ruLu Caí ru*—
ruwtwx
—rururu—
uj ru ruru.u
wru*-tC:
ruuJtruru
co
rucpocpcr
o—u^ruo
en -o o ex ru
o—útero-
•O -"O LaI -D
0 g~ Cj—jru
"-.u-.ru "4 a
N
OOOO —J
O 04 LaJ "4 O
O WOJO
"4 La) "4 O O
La) LaJO va4La»
OO OLaJO
U4ly4C',JJO
ut
ooooo
i
oo peo
ooooo
OO O OO
ooooo
00 ooo
ooooo
rj3
o—oo*—
oo otru
C O O O'"4
LTI LaI Caí "4U1
— to — ru
crcr o- tt
O' -4 CD CT. LT1
r~
orvj'OJi
OO CT— 43
owown
33J3 J3^4va4
43 roo'ro
OLD J)0-0
"4"4WOO
0
OUJ"JOO
O O |"4"J C/4
o wowo
"WOOW
OL*JO "JLrJ
OW"JOW
O"4"40O
M
ooooo
coloc o
ooooo
ooooo
ooo 00
00 ooo
OOOOO
o —— oo
— —— o —
— —■——■A-
A-»—•— ——
— —A—. -% »
— —— — —
T
-0—0 CP 43
OfUjUJCD —
ru a- O' o CD
cr-u cu ruLJ
— ru -ocdlai
a"-4 roofer
"4 LaJ raer ru
N
LaILaa"4-4L*4
0 0|l*4"41a.
"40-JW0
OWO-4W
CaíO UJ"4 LaJ
"4LaIOLa©
La* 1/40 "4 "4
i
OOOOO
1
— oooo
ooooo
ooo 00
— O O O—
OOOOO
ooooo
O
"4 CD 33 >0-0
OCT'l-OCDU)
CD 43"J03-O
33 CD 03 CP 43
O U 3) CPO
CT'CP -043-0
"4 43CD -DO
O
-o o o. rv "4
ero —ru"j
— o o-1 ru
"4t CD-OfU
oocru/io
0 arruo—
"4 —ruó
7
"4000U4
oo www
O OWWO
OO "4"40
OW-JOO
ooooo
LaJ O LaJ "4 La»
05
OOOOO
oo ooo
ooooo
ooooo
ooooo
ooooo
OOOOO
*-« * — ru—
——ruru.Aj
ruujfu— LaI
ruruuJrunj
ruru rururu
La) ru—ruru
LaJ ruru ruru
X
CTCC"JU10
X- a- erru —
rurut J3u4
O Uluio—
t 3s CD 430
0 CD-vlULX
O f—■ "40
\
OOUJLaIUJ
"J"J 0"JU4
-JO-sJWW
"JU4U40U4
•JO'JW'J
"40 UAOLaJ
ero LaJLaJ "4
ooooo
oopoo
OOOOO
ooooo
OOOOO
00 ooo
OOOOO
CD
—— pjoru
fu LaJ — La- La)
— ru t LajlT/
UJ ru LaJ t ru
ui t ru cr cr
i/iru LP— ru
t CPUlt O'
X
U3CP —00
430 ¡031 O'
t 4)WOO
CDO O to
CP—La» O'La»
LaíUIO 3J>"J
L4 L/lL/l 33 La»
O
"JLaILaIO ly-i
La» LaJ (0 "JVaJ
La! O O O LaJ
LaI "40—40
LaJ O LaJ O L4
La»LaJ 0—4—4
O "40 O "4
r~
ooooo
00 OOO
00 ooo
ooo 00
ooooo
ooooo
ooooo
—ruru——
rurulrururu
— rururu lai
— ru La, ruru
ru ruru ruru
iulai — pjru
Lajlh rururu
X
oruoocp
LaI— lO'-UO'
CP Cr 43 ru LaI
C?- f\J ia4 MU1.
0 ruó— 43
U11 a-TU¬
rúo cp ui 04
\
•40-40—4
LaJLaJ¡"J"40
"4 "ILaJO La*
"4 "4 LaJ O L4
"4"JO CaJIaJ
LA»-4 "40 LaI
-slOOWW
en
OOO —
— O ——O
O —— O —
O——OO
►-—0—0
— 0—00
©0—00
>
43-0 430 0
OUD 00 43
-0X30-00
4500-043
OO 43043
0430-0 43
43 430 43 43
1“
—"4 "400
©cr oocr
en co ruó
(LOO CP"4
OOLT.OJl
OCX 0-4 CX
crcro -i a
“0
"4 La* La» O O
O ~4 00-4
"400-40
WOO"J4
ooooo
OO OW-4
LaI OOW "4
T
OOOOO
OO ©OO
OOOOO
0 0 OOO
ooooo
OO OOO
OOOOO
ruruLHLAiru
ruruiru t—
t LaI trut
ru LaJ LaI t LaJ
WCtCW
LaI04 La» tt
t LaI LaJ t ^4
X
—10 lTi u 4
X" O' i "4—"4
cr LaIo t ru
"JO'OLPO'
flíWOTO
O' LaJ O'"»-*
ruiT-iain
\
La»OJ/4"4LaI
"4 Oí La» LaI LaI
O IaÍO—4 O
Lrl -40WO
"IWOOO
O La! -J LaI La»
O' LaILaJO La»
3C

APPENDIX G
SCATTER PLOTS

76
The following are scatter plots representing each task, based
on accuracy or speed, and its relationship to the spelling test
score (percent correct). Two correlations are reported. The first
correlation is a pooled correlation. Each group's scores (N= 35
per group) are correlated independently and then pooled for an
average correlation. The second correlation is based on the total
sample (N=70).

H2MÍR íc3fE3ÉkflBcgÍPüT!Í:£Lry?¿}c:T}?TS?Lr - R£V1S“ J*NüMY 3C' 1970
__ r „ , SPELLING AGAINST ("VC TRIG^iHa ACCURACY
PLOT Dr VARIABLE 1 CVlRKCAL AXIS) v£r3u3 V A r I a EL M S ) ¿ Z3YH9ClaXl,
O.u 2C.030 «0,000 6 O,
1C, O Q 0 30,000 5 (i.: o o
40,000 100,000
70,000 «50.000
8¿f, 000 t
7a, 000
5 u,00 0
3a,ooo *
! x
24,000 í2
U.CCO J
*, x
4,000 ♦
oír
X X
XXX
X X
X
X X
X X
XXX
XXX xx
’¿uír
30.000 50,000 70.000 • 90,000 100,000
29,000
Correlation: r = +0.48
Correlation: r = +0.87
o ,
o

Wifi'a ScÍ5Í{b3c8ípÜTÍt:'£Lr2¿;EiT},3,1i£t,r " R£VISED ■,*NUiRY jo. ;»7o
PLOT cf variaste , cv?STVbICC*«5r?l«S3S KiiSSCItl?"? (srHaOL.x,,
»;3»t -;0C\ 50,000 70.000 ,0.0
fia,000 l
X X
XX XX
XXX X
X X
5
a, ooo
$0.000
30,000
*ü.m’
'¿MM*
•¿0,666'
l
' 166,000
O
Correlation: r = +0.33
Correlation: r = +0.88
fey.ooQ
3a,ooo *
2a.COO t
la.ooo ♦
xx x
2X X
X 2
X X
X X
XX X
2a, 00 0
r
a
r*
o
o
o
o
o
©

3
â– )
o
o
3
O
3
J
§c?E°iksBl:SMPÚTlt:£LPi¿ífiTt,ST3SLÍ1 ' REv:s£c J‘NU‘,iY !0' 1,70
plot ce v*Bi.9,E , .y?s?vbise*j?aíN5i-«ss3as«íiBí:it3r*i íra.\,,
^ w, «.o e % 72,000 éc.ooo
64,000
54,000
14,000
fly*300 92,000 ’ 100,000
1 04 .000
*«,655*
*Í2;ó55*
XXX
X X
X XXX
X X X
X XX
X
X
X
X
X
XX X
X X
76,000 84,000
M55*
*-isa:ooo
92,000 100,000
74,000
4,000
Q
Correlation: r = +0.35
Correlation: r = +0.38
o
- 79

o
©
BeWtR §:ÍEÍfb?tSíoüTÍÍ£Lflxci£i?irSSLJ'1 * Rt¥I5£D J*NU‘ay 30' 1,70
PLOT Of VARIABLE 1 CV?S?ÍbÍÍCA«fSí■‘SÍrÍSS^VIbCIJgf8*í ?3?53ql.xI.
^0■00, - --- j6-0i0 32.001. 36-300 00. 000 ,4-500 .. ooo 5í-°°°
• *•+•», ,'*',,aiT,, ,,t, , f + +
24.000
X X
XX XX
2 X X x
X XX XX
Stt.eoo l
xx x xx x
74,000
o
64,000
0
54,000
0
0
44,000
o
34,000
0
24,000
â– J
14,000
o
J
4,000
v>
J
y
X X XX
2
X
X
X X
¡lUa-*-
16*000 24,000
32,000
40, 0 0 0 48,000
Correlation: r = +0.18
Correlation: r = +0.61
84,000
4,000
r?
o
c
o
o
o
o
o
ü
c.
â–  CO
o
o

o
.1
o
o
o
o
o
o
o
o
o
SSÍ£?¡¡ ScíES¿ksRk8ÍPUT!líELp'ií}Ei¥??TSgSi,í “ REVISCD JANÜA*T 30# mo
D... n „ SP^LLI^C AGAINST SHCRT vowel HFáR/n9ÍT£ ACCURACY
PLOT 0* VARIABLE 1 (VE3tíCÍL AX¡8) VERSOS VARUbLECS) 6 ísrHB0L«A),
0,0 20.050 40.300 60.0
1 It ^ 'I 1 A A * 11 C n o * * “,V
5 0^000
80,000 10 0-o o o
7 0,000 90.eco
84,000
64,000 t
54,000 í
44,000 +
34,000 í
24,000 í
14,000 ¿
4,000 i
otr
x x
x XX X
X X
X
X X
X
X XX
X
10.000
30,000 50,000 70,000
"iCm-''
90,000
•m. ooo
84,000
64,000
34,000
24,000
4,000
U
Ü
O
e
Correlation: r = +0.39
Correlation: r = +0.56

o
o
O
O
o
o
o
o
o
u
o
Ü
o
Vi
V
Q
•
BéSíStS HÍEÜéfe3pt3-TfGTlI:£L«cÍ£n}rSSer • REVISED J4f'UARy 3°- 1,70
. „„ „ , „ spelling against short vcwel hfar/write speed
plot of variable i (véPiTcal axIs5 versus yaríaoll(s) ? <3r*tJCLe<),
, _ 15.000 21.000 ^ 27.030 33.000
12.C C Q . . 19,000 2t. 0 00 30.000 36,000
3R.000
54,000
3«,000
24,000
14,000
4,000
X X
X X
X x
I *
X X
XXX
X X,
X X
- * X
X XX
2 X
34,000
J 24,000
;
18,000
Correlation: r = +0.35
Correlation: r = +0.72
24,000
30.000
36,000
ÍOfi5#*#
4,000
n
o
o
o
r;
o
o
o
o
o
o
o
o
u
O
O
a
-ZS--

o
o
o
o
o
o
o
o
o
BeMtB SE?EEífeapt3?PUTÍSELMi?EiTf?T9guS'1 “ *Evmo JAhüARV 30' 1970
f*UDT OF V*Pr*8LE 1 í V£f.Tlíh¿LGA}5s!NveS5Ü3HV*SÍ*l£Í?Í3) 8 (3Y*aOl«X),
’’ I«.»a 20-»\ 30.080 t ";C55 S.. 000 60-000 70.000 fl0*Q0° ,0.000
94 , o o o
74.000
54,005
««,000
2«.000
14.000
«.000 i
oír
X x,
u,
X X 4,
X XX ¥♦
X X2X.
XX.
XX 3.
XX 2.
♦ ■ 74 . 00 0
X 2,
2 ,
X*
X ,
x ,
x¡
X 2*.
X X
10.000
30,vOO SO,000 70,000
90.000
i
**Ufttooo
S 4 , O O O
24,000
O
<#
o
«
Correlation; r = +0.27
Correlation; r = +0.33

a?Síf3 ScÍ£sétsPt3íPÜTlsSLf',ííí2i:T?r9¿LÍf1 ■S£VIS£D 30- 1,70
SPELLING AGAINST ** * Th SPrED
PLOT Of VaRIASlE i (VERTICAL * X * 3 5 VEKSjS VAfljAdlEÍ3} 9 (3YM6ÜL*X5,
22-5°2 ... ^.500 _ 52.
15,000. 30.000 ... «5,000 ‘5 ° 60,000 fcT'5Q0 T5,C0O
84,000
7 « . 0 0 D
3«,000
¿«.000
l«. o 0 o
4,000 t
X XX Xx
X X
X
X
X x
XX X
X
x ,
X
30.000
45,000
60,000
i
;
•a¿!$oo
74,000
34,000
24,000
Correlation: r = +0,14
Correlation: r = +0.58
co

SE?£scks^tB^parí^£LH2íi?:* R£VISED JANüA*Y 3c» i „ „„ „r . „ „ SELLING A G 41 S3 7 ¿LE^iS iSD DIGFaP^S ACCURACY
PLOT CF VARIASTE l (VERTICAL A*I3) 7-»5-5 V A R I a b L E ? 6) 10 (SY^duLsXJ,
“*w ¿0, 00 0 hO. 000 60,000
IO.QOO 30,COO 50,000
,A AA, 60,000 100.000
70.000 90,000
6a, ooo
7a,000
6a,000
5a,000
aa.000
3a,000
2a,000
H, 000
a,ooc
2.
x.
X XX
XX X
X
X X
X
X X
2
X
♦ 1.,,♦,
o.J
•¿5,616’
30,000
‘45.555
50,000
•si ;m*
*46,666* •M.ooo
90,000
9a,ooo
a a, o o o
Ta,ooo
6a,000
5a,ooo
aa, oco
3a,ooo
2a.OOO
la,ooo
a,030
Correlation: r = +0.59
Correlation:
r = +0.78
Co
tn

REVISES JANUARY 30, 1970
EíÍl'tS ieíé°ck3P'c8-PÜrií:lití>®tSc'la'1'
PLOT OP VARUB^E I (V?»TlÍ;L",Íl5r?ÍR¡bS'-?ÍR}iyn;*!lSc3;5S8L.*>
“.ano íí.ooo ¿3.aid
4 . n 0 ft thftnrt -3,1 ««•
fiu,000
64, 000
4 4.000 l
3u.000
2ü,000
14,000 J
•j5° . ... 12.000 20,000 24,000 36,000
.6,000 lb,000 2«.00:. 32,000
X
u v XXX X
X 5 x x xx x x x x
X X X 2
x Xxx xxxxx
X XX
X X
XX
X X
X
X X
X
X X
'-uUtV* + .jjtgjj
«. 000 16.000 24.000 * 32.000 '
40,000
;
’¿Slooo
6«,000
44,000
3a,000
Correlation: r = +0.35
r = +0.76
Correlation:

8e!StB $ríÉ,R.íb?fcS^TÍ^Lfl?d?E:?};ITSffiM * REVIS£0 JAf'u‘Ri 30' 1,73
plot df varj*^^ , tviE^iiíGAa|s¡^^ssss?2éíi6i:E|¡í;¡iTfsígiK:sr.
8C.3C3 * . 70.000 #V flO.OOO
90,000 l
80,000 J
70,000 *
80,000 t
50,000 ♦
00,000 J
30,000 l
20,000 J
X
X
X
X
X
X
X
10,000
0,000 1
55* 54 fi * J5!!55‘,',‘'*“",‘'t5!555‘*■“■*■■■■
80,000 70,000 80,000
Correlation: r = +0.11
Correlation:
r = +0.37
85,000
90.000
95,000
1 0 0 j 0 0 0
105,000
XX XX
2. XX
X 2 X X
XXX
XX X
X X X
X
X X
X X
X X X
X x
90,000 100.000
9a,ooo
iu.000
7a,ooo
8a,ooo
54,000
44,000
3a,000
2a, ooo
la.ooo
a,ooo
CO
•‘-J

BeJÜtS • ',£VIS£D J‘KUiRY 30' 1,70
, SPF!LI»,C agaInST C3‘¡3C‘|*HT REaR/aRITE SPEED
PLOT OF VARIABLE 1 (VERTICAL A AI 3 J VESSUS VARIAblllS) ti Í3YM5Cl«X),
13..03 20.coo ¿a.000 30,
0 0 0 aa.000
14,000 24,030 32.000 40.000 46,000
X X
x 2 X
2 X X
64,000 ♦
54,000 1
44,000
34,000 l
XX X X
X X
X . X
XX
44,000
24,000
14,000 l
12,546*
X
X X
16.000
*
24,000 32,000 40,000 46,000
I
J
* o 0 0
Correlation: r = +0.37
r = +0.71
Correlation:

- sevi£e: 3o- 1,70
- so'Ll:ng against 3K:m voel hearsay accj-íacy
»LOT OF VARIABLE 1 ÍVEñtlCAL A * I 5 5 '.EflS'jS /A?JAoLí£S) W ÍSí*oCL«X),
0.5 20.000 4 o« 0 0 0 60,000 00.000
■ .10 . 000 _ ^ • 30 . 00 0 50.00 0 70,0 0 0 90,000
♦....*....*....»....*....»....t.t,,.I,,»;*
Tu.OOD t
100.000
X X
X
XX x
X X
XX
X
XXX X
X *
x I
94,000
34,000
64,000 t
64,000
54,000 f
34,000 i
24,000 ♦
*x
14,000 l
X X
X
X
X X
x X
X X
X
XX x
54.000
44,000
24,000
4*C00 ♦
o,0
10,000
**55lft3S*
30,000 50,000
sjtjjj-'"-’
70,000 90,000
I
156*000
4,000
Correlation: r = +0.42
Correlation: r = +0.50
CO

- ,evis“ j*hui,Y 3°-1,7(1
„ . „ ASAlNST SmCRT VOp>EI HE*R/3Ar 3pEED
PLOT OF VARIABLE 1 (VERTICAL A XI 3 J VERSUS V*RIABLE£S) 15 C3YM6Cu*X>f
IS.ú 0 0 21.000 27.0 0 0 33.00 0 39.000
12,020 , 16,000 , 24.000 30.000 36,000
74,000 ♦
64.000
54,000
2
X X
44,000 i
34,000
24.000 ♦
X
X X
44,000
34,000
24 , 000
4,000 i
•+**i $!m******+-
,, nn„ IB ••** •*•• --íum* **
12.000 18,000 24,000 30,000 36,000
Correlation: r = +0.33
Correlation: r = +0.70

auíifi - revisíd J*su4Rr 35< ••,tj
PLOT OF VASUBlE i (VEPTÍÍjIuCaÍÍ 3¡NyÍR3lj3MVA«lA3ul!',S)CU <3YM0Ol«X)#
flb.bOO 91,500 94.500
87,000 90,000 9J.0CQ . 9b,000
97.500
100.500
l \
84,000 J
X X
X
8a,ooo
7a, 000
54.000 ♦
44,000 i
XXX
X X
44,000
¿a,aoo í
34.000
4,000 ♦
„ nnn M.Sfifi • V.W
CT. 000 90,000 9Jf 000 9b.000 99,000
•165,45o
4,000
Correlation: r = +0.13
Correlation:
r = +0.35

mm *REvmo j*’'üa"y 3o* i,7í
. „ ScilLr¿G AG* INS7 ILP^ABFT 3r {•£?
*ICT OF V*RIa8LE i {VELICA', AXIsJ VERSAS ViQlífcLUS) 17 (3tm30L«X),
22.300 37,300 5¿,3q5 B7.500 82.500
15.000 . . 30.000 45.000 60.000 . 75,000
74,000
64,000
54,000
24,000
14,000
4.000 l
X X X X
X
X XX
X
X
X X
2 X X
X XX X
♦•‘itisss* ••••■*
1S•O G O 30,000 45,000 *0,000 75,000
94,000
84.000
74,000
bU,000
54,000
44.000
34,000
24,000
14,000
4,000
Correlation: r = +0,38.
Correlation: r = +0. 72

BIOGRAPHICAL SKETCH
Elliott I. Lessen was born in Syracuse, New York, on
February 10, 1947. He was graduated from the Syracuse City
Schools in 1964. Mr. Lessen received a B.A. from Syracuse
University in 1968 with a major in Art History.
Mr. Lessen was a third grade teacher with the Chicago
Public Schools for the 1968-1969 school year during which time
he pursued graduate coursework in elementary education at
Loyola University. From June, 1969, through August, 1970, Mr.
Lessen taught HeadStart/Experimental Pre-Kindergarten and Pre-
first grade in Syracuse, Mew York. He did further graduate
work in early childhood and special education at Syracuse
Universi ty.
From September, 1970, to June, 1973, Mr. Lessen was employed
by the Adams School, a private school for brain-injured and
emotionally disturbed children in New York City. He was awarded
his M.S. from Hunter College in June, 1973, with a major in special
education (emotional disturbance).
Mr. Lessen was a specific learning disabilities resource teacher
during the 1973-1974 school year in Gainesville, Florida. He started
his doctoral program in September, 1973. Concurrently, Mr. Lessen
was a graduate teaching assistant with the Department of Special
93

94
Education and the National Teacher Corps. He was also an
administrative assistant with the Career Associate in Special
Education program at Santa Fe Community College. Mr. Lessen
will receive a Ph.D. from the University of Florida in December,
1976. His area of specialization is special education (learning
disabilities).
Mr. Lessen has accepted a position as an Assistant Professor
of Special Education at the University of Northern Iowa, Cedar
Falls, Iowa, beginning August, 1976. Mr. Lessen and his wife,
Linda, are happily married.

I certify that I have read this study and that in my opinion
it conforms to acceptable standards of scholarly presentation and
is fully adequate, in scope and quality, as a dissertation for the
degree of Doctor of Philosophy.
h
n
/ /
â–  I 4 i
William D. Wolking,
Professor of Specia
Chairperson
Education
I certify that I have read this study and that in my opinion
it conforms to acceptable standards of scholarly presentation and
is fully adequate, in scope and quality, as a dissertation for the
degree of Doctor of Philosophy.
.GJULpoz?
Thomas B. Abbott
Professor of Speech
I certify that I have read this study and that in my opinion
it conforms to acceptable standards of scholarly presentation and
is fully adequate, in scope and quality, as a dissertation for the
degree of Doctor of Philosophy.
r ) v^
Thárleá Forgnone ,/
Professor of Special Education
I certify that I have read this study and that in my opinion
it conforms to acceptable standards of scholarly presentation and
is fully adequate, in scope and quality, as a dissertation for the
degree of Doctor of Philosophy.
' ¡ '
Cecil D. Mercer
Assistant Professor of Special Education

I certify that I have read this study and that in my opinion
it conforms to acceptable standards of scholarly presentation and
is fully adequate, in scope and quality, as a dissertation for the
degree of Doctor of Philosophy.
Paul Satz
Professor of Psychology
I certify that I have read this study and that in my opinion
it conforms to acceptable standards of scholarly presentation and
is fully adequate, in scope and quality, as a dissertation for the
degree of Doctor of Philosophy.
/{'<)/{.it.-
William B. Ware
Professor of Foundations of Education
This dissertation was submitted to the Graduate Faculty of the College
of Education and to the Graduate Council, and was accepted as partial
fulfillment of the requirements for the degree of Doctor of Philosophy.
"Becember, 1976—
/'//FA1. /4,7 7
Dean, College of Education
Dean, Graduate School

UNIVERSITY OF FLORIDA
3 1262 08552 9484