• TABLE OF CONTENTS
HIDE
 Title Page
 Acknowledgement
 Table of Contents
 Abstract
 Introduction
 Method
 Results
 Discussion
 Reference
 Biographical sketch














Title: Grapheme discrimination training in children predicted to develop reading difficulties /
CITATION PDF VIEWER THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00098129/00001
 Material Information
Title: Grapheme discrimination training in children predicted to develop reading difficulties /
Physical Description: vi, 83 leaves : ill. ; 28cm.
Language: English
Creator: Proeger, Terry Shannon, 1947-
Publication Date: 1976
Copyright Date: 1976
 Subjects
Subject: Reading -- Ability testing   ( lcsh )
Visual discrimination   ( lcsh )
Reading readiness   ( lcsh )
Psychology thesis Ph. D   ( lcsh )
Dissertations, Academic -- Psychology -- UF   ( lcsh )
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
 Notes
Thesis: Thesis--University of Florida.
Bibliography: Bibliography: leaves 80-82.
General Note: Typescript.
General Note: Vita.
Statement of Responsibility: by Terry Shannon Proeger.
 Record Information
Bibliographic ID: UF00098129
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000169323
oclc - 02897549
notis - AAT5729

Downloads

This item has the following downloads:

PDF ( 1 MBs ) ( PDF )


Table of Contents
    Title Page
        Page i
    Acknowledgement
        Page ii
    Table of Contents
        Page iii
    Abstract
        Page iv
        Page v
        Page vi
    Introduction
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
    Method
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
    Results
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
    Discussion
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
    Reference
        Page 80
        Page 81
        Page 82
    Biographical sketch
        Page 83
        Page 84
        Page 85
        Page 86
Full Text



















GRAPHEME DISCRIMINATION TRAINING
IN CHILDREN PREDICTED TO DEVELOP READING DIFFICULTIES








By




TERRY SHANNON PROEGER


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













ACKNOWLEDGMENTS




I WOULD LIKE to thank Paul Satz for his tremendous enthusiasm,

interest, and support; and Jeffrey Fitzsimmons for his contribu-

tions to this research. I would also like to thank Charlene

Thiess for her special contributions.















CONTENTS





ACKNOWLEDGMENTS

ABSTRACT

I INTRODUCTION
Prediction of Reading Performance, 2. Distinctive
Features, 4. Discrimination Training, 16. Summary,
24. Hypotheses, 26.

II METHOD
Subjects, 29. Apparatus, 30. Procedure, 31.

III RESULTS
Machine Match-to-Sample Test, 57. Paper-and-Pencil
Match-to-Sample Test, 58. Training Versus Novel
Forms, 59. Three-Month Follow-Up (Probe 2)--
Teacher Ratings, 62. One-Year Follow-Up (Probe 3)--
Objective Measures and Teacher Ratings, 67. Beery
Test of Visual-Motor Integration, 67.

IV DISCUSSION

REFERENCES

BIOGRAPHICAL SKETCH











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




GRAPHEME DISCRIMINATION TRAINING
IN CHILDREN PREDICTED TO DEVELOP READING DIFFICULTIES



By

Terry Shannon Proeger
June 1976



Chairman: Paul Satz
Major Department: Psychology


The total population of kindergarten children at an urban

elementary school was administered a behavioral screening battery

designed to identify potentially poor and superior readers.

Twenty-four children whom the battery predicted would develop

severe reading problems were randomly divided into experimental

(N=7), treated comparison (N=6), and untreated control (N=11)

groups. All subjects were administered a paper-and-pencil match-

to-sample letter discrimination pretest. The experimental and

comparison groups were also administered a match-to-sample test

on the training apparatus. The experimental group was then given

6 weeks of faded distinctive feature discrimination training. The

comparison group was given 6 weeks of traditional multiple











feature discrimination training. The control group received only

regular classroom experience.

Following training, the treatment groups were readministered

the match-to-sample test on the training apparatus. All three

groups were also readministered the paper-and-pencil match-to-

sample test. The treatment groups were then given 8 weeks of

grapheme-phoneme training by another experimenter. Three months

after the original letter discrimination training, all subjects

were readministered the paper-and-pencil match-to-sample test. The

Beery Developmental Test of Visual-Motor Integration was also

administered, and teacher ratings of each subject's classroom

performance were obtained. The paper-and-pencil match-to-sample

test was administered a fourth time one year after the original

letter discrimination training. The Beery Test of Visual-Motor

Integration was readministered at this time and teacher ratings of

each child's classroom performance were again obtained. Two addi-

tional measures, tests of word matching and word recognition,

were also administered one year after the discrimination training.

The teacher ratings were used as criterion measures to eliminate

false positive subjects from the data analysis.

One purpose of the experiment was to evaluate the effective-

ness of two methods of letter discrimination training with

children predicted to develop severe reading difficulties. No

significant differences were present between groups on any of the

match-to-sample discrimination tests. It was concluded that











neither method of discrimination training was more effective than

regular classroom teaching in improving letter discrimination per-

formance. Another purpose of the experiment was to evaluate the

combined effect of grapheme discrimination and grapheme-phoneme

training on more general reading related performances. No signi-

ficant differences were present between groups on teacher ratings

of classroom performance or on the word matching and word recogni-

tion tests with one exception: significantly more treatment than

control children were rated by their teachers as able to recognize

most of the letters of the alphabet. Treatment children were

also found to have performed significantly better than control

children on the Beery Test of Visual-Motor Integration. This dif-

ference between groups on the Beery, however, was not present 9

months after the termination of training. It was concluded that

grapheme discrimination training combined with grapheme-phoneme

training was no more effective than regular classroom teaching

in generally improving reading related skills. The combined

training did appear to be more effective than regular classroom

teaching in improving alphabet recognition and visual-motor

performance, two skills which involve visual discrimination.

















INTRODUCTION





READING IS THE MOST BASIC and important of all skills taught in

American schools; yet 25% of school children in the United States

are reading below grade level (Gibson 5 Levin, 1975) and 15% are

considered to exhibit severe reading disabilities (Kline, 1972).

In the past, much research has focused on improving these

children's reading skills. Vernon (1960, p. 184) characterized

this research as containing numerous enthusiastic recommendations,

but few controlled studies demonstrating actual improvement. The

failure to develop successful remediation techniques has con-

tributed to the recent trend toward preventive research. Such

research has generally focused on prereading skills in children.

Much of it has dealt with the training of visual perceptual pro-

cesses. Hammill (1972) found that in 21 of 25 such studies,

improvement in reading could not be expected as a result of sys-

tematic visual-motor training. He ascertained from his survey

that "little correlation existed between measures of visual per-

ception and tests of reading comprehension and that training

visual perceptual skills, using currently available programs,










has no positive effect on reading and possibly none on visual

perception" (p. 552).

At present it seems that the most effective early interven-

tion techniques must lie in the manipulation of the reading pro-

cess itself. The problem with this approach is that it requires

an accurate conception of the reading process and of the skills

necessary to master the process. Gibson (1965) presented such a

conceptual framework supported by a number of relevant empirical

studies. She distinguished three sequential phases in the process

of learning to read: (1) learning to differentiate written

symbols, (2) learning to relate letters to sounds, and (3) learn-

ing to utilize progressively higher-order units of linguistic

structure, i.e. spelling and morphemic patterns in letter

sequences, and syntactic and semantic patterns in word sequences.

The phases are organized such that each earlier phase must be

mastered before the succeeding, more complex phases can be

successfully undertaken. It would seem that an adequate inter-

vention technique must insure that each preceding phase is

mastered before attempting to focus on a later stage.


PREDICTION OF READING PERFORMANCE

The first step, however, in developing a preventive interven-

tion program for children destined to be poor readers is to

accurately identify these children. Without valid identification

of the target sample, the effects of early intervention cannot






3



be determined. A behavioral screening battery has recently been

developed which is designed to detect children who will later

become severely disabled readers (Satz, Friel & Rudegeair, 1975).

The screening battery was standardized on the total population

of white male kindergarten children (N=497) in Alachua County,

Florida, schools in 1970. The battery was later validated

against independent reading criteria at the end of first,

second, and third grade. An independent crossvalidation study

was undertaken in 1971, based on a sample of white male kinder-

garten children (N=181) in five Alachua County schools. Their

reading performance was assessed at the end of second grade. These

studies revealed that the battery consistently identified over

90% of the children destined to become severely disabled readers.

A second crossvalidation study was initiated in September

1974 and included the total population of kindergarten children

(N=132) in one Gainesville, Florida, elementary school. Only

achievement ratings taken at the end of kindergarten are

available, and this data must be considered highly tentative.

These results, however, indicate that the battery successfully

predicted 100% of the children who were rated as severely

learning disabled at the end of kindergarten (Satz & Friel, 1976).

This degree of predictive accuracy allows for the develop-

ment of preventive programs with children destined to become

severely disabled readers. The advantages of early intervention

are twofold: remediation can be undertaken before the child has










suffered the damaging effects of continuous academic failure and

also at a time when the central nervous system may be more pliant

and responsive to change. Indeed, Keeney and Keeney (cited in

Strag, 1972) found that when a diagnosis of dyslexia was made

during first and second grade, 82% of these students could be

brought up to grade level, while only 46% of children identified

in third grade and 10-15% of children identified in fifth through

seventh grades could be brought up to grade level. Bloom (1964)

has shown that environmental manipulations have their greatest

quantitative effect on a behavior at its most rapid period of

change.

The present study has attempted to identify "high risk"

kindergarten children before they have begun to develop reading

problems and has focused on one skill, differentiation of letters,

which develops rapidly at this age.


DISTINCTIVE FEATURES

The next step in developing an intervention program based

upon Gibson's model is to design an effective program for

training children to differentiate written symbols. Once this

is successfully accomplished, each succeeding skill can be inte-

grated into a comprehensive program. Two questions arise at this

point: What does the child learn when he succeeds in discriminat-

ing letters and how can this most effectively be taught?










Lavine (cited in Gibson & Levin, 1975) investigated whether

children between the ages of 3 and 6.5 who had not yet been

taught to read could differentiate writing from pictures. She

found that 3-year-old children distinguish between graphic dis-

plays of objects and letters, even though they may be unable to

name the letter or word and even though they have not been pre-

viously exposed to the particular graphic system (Hebrew for

example). Furthermore, although multiple versus single units,

linear versus nonlinear arrangements, and repetitive versus non-

repetitive units contributed to the differentiation of writing

from pictures, the most important information appeared to be the

features of the letters themselves. Over 87% of 3-, 4-, and

5-year-old children identified Roman, cursive, and Hebrew charac-

ters as writing. A small percentage of children identified Mayan

designs and Chinese characters as writing. This percentage de-

creased from age 3 to 4, then increased slightly at age 5. A high

percentage of children identified artificial letterlike characters

as writing. This percentage remained the same at ages 3 and 4,

but decreased at age 5. These data suggest that children who could

not yet read demonstrated an increasing sensitivity to familiar

features of real letters. Gibson and Levin concluded that the

children have learned a set of internal features common to the

letters of their written system, even though the children have

not yet learned to recognize individual shapes. They cited the

absence of a difference in responses to Roman and Hebrew script

as support for this conclusion.










Davidson (1935) was the first to systematically study con-

fusion errors in letter discrimination. She gave kindergarten

and first-grade children a letter matching test. A larger per-

centage of kindergarten than first-grade children made every type

of error. Practically all kindergarten children and a smaller

but substantial number of first graders made reversal errors.

There was a marked and consistent decrease in the percentage of

children making errors with an increase in mental age. Between

the mental ages of 5.5 and 6 years there was a marked decrease

in the percentage of children making reversal and inversion

errors. The greatest decrease in reversal errors occurred at

an age of 7.5 years. Davidson concluded that some letters of the

alphabet are more difficult to discriminate than others. Rever-

sals are most difficult with inversions closely following. It

seems that children pass through certain stages before they are

able to distinguish b from d and p from q. In the first stage all

letters are confused. In the second stage only reversals are

confused. The third stage is marked by the ability to recog-

nize that reversals are directionally different, but without

considering the letters as different. The final stage includes

the recognition that reversals are different letters.

Gibson, Gibson, Pick, and Osser (1962) hypothesized that

there are certain critical dimensions of difference, such as

break versus close, straight versus curve, rotations, and

reversals, which are the features initially utilized in











discriminating letters. They studied the development of ability

to discriminate letterlike forms,in children 4 through 8 years

of age, by requiring the child to select figures identical to a

standard figure from an array of copies and transformations of

the standard. The mean errors for each type of transformation

were determined, and it was found that discrimination of the

letterlike forms improved with age, as expected, but the slopes

of the error curves were different, depending on the transfor-

mation to be discriminated. Thus, some transformations are

harder to discriminate than others, and improvement occurs at

different rates for the various transformations. Errors for

circle versus break transformations were relatively few, even

in the youngest subjects, and declined to almost zero in the 8-

year-old children. Errors for rotations and reversals started

high, but dropped to nearly zero by 8 years. Errors for changes

of line-to-curve were relatively great at 4 years of age, but

by 8 years had declined almost to zero. The slopes of the curves

indicate that the greatest developmental change between 4 and

8 years is in confusion of rotation-reverals, with line-to-

curve errors showing the next greatest drop and break versus

close errors, the least.

This experiment was replicated with real letters and the

same transformations on the 5-year-old group. Errors were fewer

for letters, but the correlations between transformations were

significant in every case. The correlation between the confusions










of the same transformations for real letters and for the letter-

like forms was very high (r=+0.87). Gibson et al. hypothesized

that "it is the distinctive features of grapheme patterns which

are responded to in the discrimination of letterlike forms. The

improvement in such discrimination from four to eight is the

result of learning to detect these invariants and of becoming

more sensitive to them" (p. 904).

Gibson, Osser, Schiff, and Smith (1963) investigated the

dimensions of difference a child must learn to detect in order

to perceive each letter as unique. They began by intuitively

constructing a list of the distinctive features of letters,

such that a unique pattern of features characterized each letter.

Their list of 12 features included verticality, horizontality,

diagonality, curvature, openness or closure, intersection,

cyclic change, symmetry, and discontinuity. They obtained a

confusion matrix of the 26 Roman capital letters, based on the

errors of 4-year-old children who made matching judgments of

letters, and compared the errors in the confusion matrix with

those predicted by the feature chart. If the features were

correctly chosen, the subjects should have confused most fre-

quently those letters having the smallest number of feature

differences. The results yielded 12 of 26 positive significant

correlations. Considering that the features were not weighted,

one can see that the prediction from this feature list was

fairly good. A multidimensional analysis of the matrix corroborated










the choice of curve versus straight-and-obliqueness variables,

suggesting that these features have priority in the discrimina-

tion of letters.

Gibson, Schapiro, and Yonas (1968) obtained confusion

matrices for two sets of nine artificial graphemes. Adults and

7-year-old children were asked to make same/different judgments.

Error and latency measures proved to be highly correlated and

revealed an identical structure. The adult subjects' data for

the nine characters paired in all combinations indicated that

the first split separated letters of diagonality from all the

others. Next, the round letters were split off from the others,

and finally, the square right-angular letters were split off from

letters characterized by curvature. The children showed a

similar but not identical pattern. First, the letters were split

along the curve versus straight dimension. Next, the round

letters were differentiated from those characterized by only a

curve. Finally, the straight horizontal-vertical letters were

split off from those with diagonality. The authors concluded

that these results confirm the hypothesis that letters are dis-

tinguished from one another by way of distinctive features that

are shared to varying degrees by different pairs of letters.

Popp (1964) used a match-to-sample paradigm to determine

which pairs of lower-case letters are most often confused by

kindergarten children. Her results further substantiate the data

of Gibson et al. (1968). Popp found that most confusions










arise from rotations and reversals and relatively few from

close-versus-break transformations. She concluded that "con-

sideration of the distinctive features and of the formal

similarity of letters should provide insight for training a

specific skill in visual discrimination for letters. Remaining

as a matter for further research is the problem of determining

whether children trained to discriminate highly confusable

letters on the basis of distinctive features and/or formal

similarity will then be able to accurately discriminate all

letters, and whether such training will have any effect on their

later reading achievement" (p. 225).

Dunn-Rankin (1968) analyzed letter similarity, using a

relative discrimination task rather than an absolute discrimina-

tion task. Using lower-case letters he asked second and third

graders to indicate which of two letters was most similar to a

target letter, under the assumption that the letter combinations

judged most similar would be most likely to be confused. The

results generally supported the finding that rotational errors in

all three spatial planes are likely sources of confusion. Factor

analysis revealed that the letters may be clustered together in

groups based upon size, formal similarity, axial rotation, and

topological line-to-curve transformations. The first factor con-

trasted the short curved letters (e, a, s, c, o) with the tall

straight-line letters (f, 1, t, k, i, h). The second factor con-

sists of p, d, and b with o, g, and h partially loading in










contrast to w. The third factor contains the partially curved

short letters n, u, m, and w,with h partially loading on this

factor. The fourth and fifth factors appear to be r and y

respectively. Dunn-Rankin suggested that within clusters the

particular features of letters become increasingly important

for discrimination. He concluded that "a suggested principle of

learning is to present materials that offer the opportunity for

maximum contrasts before moving to items with minimal cues for

discrimination" (p. 994).

Some of Dunn-Rankin's data suggest that letters with greater

common area may be more confusing than letter pairs with less

area in common. Dunn-Rankin, Leton, and Shelton (1968) developed

an index based upon the ratio of common area to independent

area in order to determine the congruency of lower-case letters.

The congruency matrix was factor analyzed. Five factors composed

of letter groups were identified. Factor I (p, b, d, q) shows

structural similarities which are a combination of straight-

line and closed-curve features. The letter congruencies in this

factor are due to rotations in all three planes. Factor II

(i, f, 1, j, t) contains letters characterized by a straight

line and secondarily open-curve features. The letter congruencies

in this factor are based on all three rotations. Factor III

(y, v, x, w, z, k, g) shows a structural relationship of the

letters because of line and slant features which result in acute

angles. Factor IV (n, u, m, h) contains letters characterized by










short, straight-line, and extended-curve features. The n-u con-

gruence in this factor is effected by axial rotation in depth.

Factor V (e, c, o, s, a) is characterized by open- or closed-

curve features and a similarity in height.

Given that the child learns to discriminate graphemes by

detecting their distinctive features, it can then be asked how

this type of discrimination develops. Pick (1965) designed an

experiment to evaluate two hypotheses about improvement in dis-

crimination. The first is called the schema hypothesis and

supposes that the child constructs a model or memory image of

each letter by repeated experience of visual presentations of

the letter. Discrimination occurs by matching sensory experience

to a previously stored concept or model. The second, which is

called the distinctive feature hypothesis, proposes that the child

learns by discovering how forms differ and then transfers this

abstract knowledge to new letterlike forms. Practice enables the

subject to discover which of a number of stimulus variables are

critical in that they serve to distinguish between one object

and another.

To test these hypotheses, Pick trained a group of kindergarten

children to discriminate letterlike forms (Stage 1). The sub-

jects were then divided into three groups. Group I was given

sets of stimuli to discriminate which varied in new dimen-

sions from the same standards discriminated in Stage 1. Group

II was given new standards, but the same dimensions as










discriminated in Stage I. Group III received both new standards

and new dimensions of difference to discriminate (Stage 2).

Thus, Group I had the same prototypes as those that they learned

in Stage 1, while Group II had the same dimensions as those that

they learned to discriminate in Stage 1. The mean number of

errors made during Stage 2 was computed for each group. The

results indicated that Groups I and II performed significantly

better than Group III but that Group II, the group with the

familiar transformations and new standards, performed signifi-

cantly better than Group I, which had received new transforma-

tions of old standards.

These results suggest that while children do learn the proto-

types of letter shapes, the prototypes themselves are not the

original basis for differentiation. Learning distinctive features

appears to be a significant component of improvement in visual

discrimination of letterlike forms.

A second experiment involved a tactual discrimination task

of successive comparison. The prototype and distinctive feature

groups made an equal number of errors on the transfer task but

did significantly better than the control groups.

A third experiment required a tactual discrimination task

with simultaneous comparison. The prototype and control groups

did not differ in mean number of errors, but the distinctive

feature group performed significantly better than both. These

results suggest that the superior group learned the distinctive










features of the forms since they had no opportunity to con-

struct prototypes of the forms used in the transfer task. The

groups which could use prototypes in the transfer task performed

no better than the control group.

In summary, the children who could use what they had learned

about distinctive features showed the best transfer task perform-

ance. Those who could use schemata also showed transfer, but

significantly less than the distinctive feature group. These

results may be interpreted as suggesting that the detection

of distinctive features will always facilitate improvement in

discrimination but that under conditions of successive comparison

prototype construction will independently facilitate such improve-

ment. A more parsimonious interpretation is that the detection of

distinctive features may lay the basis for improvement in dis-

crimination. Only when such detection is dependent on memory

does prototype learning occur. In other words, the detection

of distinctive features may be the necessary and sufficient con-

dition for improvement in discrimination.

Williams (1969) trained kindergarten children to discriminate

letterlike forms. Her experiment compared three conditions:

(1) discrimination training where the comparison stimuli were

quite different from the standard, (2) discrimination training

where the comparison stimuli were transformations (rotations

and reversals) of the standards, and (3) reproduction training

involving tracing and copying of the standards. A delayed










match-to-sample task was used because it approximates the per-

ceptual learning tasks involved in actual reading, more closely

than does simple discrimination training. The results indicated

that discrimination training in which the comparison stimuli

were transformations was superior to discrimination training

where the comparison stimuli were totally different forms. This

suggests that comparisons involving minimally different stimuli

forced the subject to attend to and abstract more attributes

of the standard, which were then used for new test comparisons.

Reproduction training was not as effective as discrimination of

transformations but was as effective as simple discrimination

training.

Williams summarizes her study:



The crucial point is that even after a
small amount of training at an appropriate
level (i.e., the beginning of the kinder-
garten year), there were significant dif-
ferences among the training groups. These
data indicate clearly that the effective-
ness of readiness training does indeed
depend on the particular technique used,
and that there would be wide variation in
the effectiveness of typically used readi-
ness materials. While special practice is
often given on rotation and reversals,
usually too much time is devoted to copy-
ing and tracing or to discrimination
exercises that, according to this data,
are relatively ineffective. Moreover,
such systematic training is sometimes
given only in remedial work, that is,
after a child has already developed some
difficulty. The present experiment
suggests (1) that more time be devoted to










discrimination training that involves com-
parison of letters with their transforma-
tions and (2) that this type of training
be given early. It is quite effective at
the very start of kindergarten, and ob-
viously, if the occurrence of certain
relatively common perceptual confusions
could be minimized by appropriate training
techniques introduced in the beginning
stages of instruction, there should be
less necessity later for remedial tech-
niques. (p. 513)




It was decided that an intervention program aimed at the

first stage of the reading process--the differentiation of

written symbols--would be most effective if it trained children

to respond to the distinctive features of letters. Gibson

(1970) stated that "methods of teaching that would promote effi-

cient strategies of perceptual search and detection of invariant

order should be a first concern in instructional programs"

(p. 143). Guralnick (1972), in a review of alphabet discrimina-

tion and distinctive features, suggested that "a training pro-

gram specifically designed to teach children to attend to

these features should be valuable in increasing discrimination

skills and, eventually, naming of letters in the alphabet"

(p. 430).


DISCRIMINATION TRAINING

The last step in developing an intervention program based

upon Gibson's model and incorporating the findings presented in

the previous section is to determine the most effective procedure










for training children to respond to the distinctive features of

letters. Terrace (1963) was the first to describe "errorless

discrimination training." He taught pigeons to discriminate

colors by gradually equalizing brightness and duration differ-

ences between correct and incorrect responses. He superimposed

vertical and horizontal stripes over the colors, and the

colors were gradually faded out. Both the color discrimination

and the more difficult stripe discrimination were acquired

without errors. The major finding was that while differential

reinforcement in the presence of two stimuli is a necessary con-

dition for establishing a discrimination; the occurrence of

errors is not necessary. Furthermore, in comparing the per-

formance of pigeons following discrimination training with and

without errors, it was found that the pigeons trained with

errors exhibited a relatively poorer discrimination performance

once the discrimination was acquired. These pigeons tended to

produce bursts of errors and to exhibit "emotional" responses

such as wing flapping and turning away from the response key

following incorrect responses.

Several other studies have demonstrated that fading (error-

less discrimination) is more effective than simple trial and

error methods. Hively (1962) was the first to report the

training of children in a faded series of progressively diffi-

cult discrimination tasks, eventuating in a match-to-sample

paradigm. This program was composed of four series: (1) no










incorrect response was available and the matching stimulus was

placed directly below the sample stimulus, (2) both choice

stimuli were presented still in the same positions, with the

correct choice always located directly below the sample stimulus,

(3) the sample stimulus was placed in the middle, although the

choice stimuli still remained in fixed position, and (4) the

position of the choice stimuli was varied from trial to trial.

Thus, the child advanced from an easy discrimination in which

no incorrect choice was available to a discrimination in which

no cues were available except those provided by the stimuli

themselves. Reinforcement was simply the presentation of the

next set of stimuli.

The results confirmed the hypothesis that learning is more

efficiently accomplished when the subject is trained in a pro-

gressively more difficult discrimination than when training is

given in the final discrimination alone. The errors the subjects

made were a function of the size of the steps in the program

and the length of training on each discrimination. Thus, these

programs can be too long as well as too short or discontinuous.

The more errors the training procedure allowed the children to

make, the more they tended to go on making.

Moore and Goldiamond (1964) presented preschool children with

a delayed match-to-sample task in which subjects were required

to match one of three triangles varying in orientation with that

of its previously presented sample. This discrimination proved










difficult for preschool children. However, when only the correct

match was illuminated, discrimination was easily established. The

brightness difference between correct and incorrect matches

was gradually faded out until all stimuli were of equal bright-

ness. The discrimination was maintained in the absence of the

brightness difference, thus transferring stimulus control from

brightness to form with virtually no errors.

In an experiment with similar results, Touchette (1968)

compared methods of teaching severely retarded boys to determine

the position of a black square and to press the nearest key to

it. Of seven boys given trial and error learning, one learned

the task. The six boys who failed to learn were presented with

a program of gradual stimulus change, and all but one acquired

the performance. The child who did not was under stimulus

control during the program but reverted to a position-based

response learned during trial and error training when he reached

the criterion stimuli. Six similar children were presented with

graduated stimulus training only, and all six learned the cri-

terion discrimination with few or no errors. Both groups were

tested for retention of the criterion performance 35 days after

the completion of training. Two boys with previous trial and

error training showed no signs of retention after 35 days.

Touchette observed that retardates form superstitions early

in discrimination training which frequently seem to prevent the

development of an appropriate controlling relation. The occurrence










of spurious controlling relations cannot be avoided in a trial

and error procedure. If training is initiated by reinforcing a

stimulus-response relation which already exists, or can be

easily established, it is then possible to gradually shift the

stimuli toward those which comprise the criterion discrimina-

tion while maintaining control of responses by specified aspects

of the training situation.

Sidman and Stoddard (1967) utilized a fading procedure to

teach severely retarded boys a circle-ellipse form discrimina-

tion. They compared subjects trained on the fading program

with subjects trained by trial and error. Their procedure involved

transferring stimulus control from a simple brightness discrimi-

nation to a form/no-form discrimination and, finally, to a

circle-ellipse discrimination. This was accomplished by fading

out the brightness dimension, leaving a form/no-form discrimina-

tion. Finally, the alternative form was gradually faded in to

achieve the circle versus ellipse discrimination. Correct

choices were reinforced by chimes, candy, and the next slide. A

correction procedure was used so that following a wrong choice,

the stimuli remained until a correct choice was made. A back-up

procedure was also used in which a correct response following an

error caused the slide tray to reverse, thus presenting the

preceding slide. Sidman and Stoddard found that the children

taught by the fading procedure performed much more effectively

than did the group trained by trial and error.











Bijou (1968) developed a fading program to teach normal

children 3.5 to 6.9 years of age and retarded children 3.9 to 8.9

years of mental age to make form discrimination, mirror-image

discrimination, and rotated mirror-image discrimination.

Mirror-image discrimination was based on fading in, by construc-

tion, of the mirror images of three forms: a flag, square z, and

winged L. A five-choice match-to-sample paradigm was used. A

correct choice resulted in a light, chime, and appearance of the

next sample. The next set of choices was produced by pressing

on the sample window. An incorrect choice was followed by black-

out of the stimulus array, which could be restored by pressing

on the sample window. A back-up procedure insured that an in-

correct response followed by a correct response produced the pre-

ceding rather than succeeding stimulus array. The results indi-

cated that the training program was effective in teaching both

normal children and retardates the orientation discrimination.

The retarded children did well in discriminating rotated forms

but had more difficulty with rotated mirror-image discrimina-

tions. Success was approximately proportional to mental age.

A modified program was established, consisting of only the

flag form, the mirror-image of which was faded in by manipulat-

ing its size. Training began with the nonrotated mirror-image

form and advanced to training with rotated mirror-image forms.

A correct response was followed by a colored bead, light, chime,

and appearance of the next sample, while an incorrect response










was followed by a brief buzz and blackout of the choices, which

reappeared upon pressing the sample window. This sequence was

repeated with each incorrect response until a correct response

occurred. The back-up contingency was eliminated. After an error,

reinforcement for a correct response was a chime, light, and

forward movement of the program, but no bead. At the end of the

session, the children were allowed to exchange beads for toys,

candy, or pennies on a ratio basis. It was concluded that

although the fading technique using the flag form required im-

provement, the fading technique has proven to be by far the

most promising for the development of left-right concepts.

Two additional findings are worth noting. Bijou found that

having subjects repeat a set will not improve their performance.

Repetition of sets resulted in the same number of errors concen-

trated around the same stimuli. Pretests and posttests--composed

of forms used in training, new forms, and letters of the alpha-

bet--indicated that the training experience facilitated mirror-

image discrimination of the new forms and of the alphabet

letters. In the absence of data concerning the stimulus dimen-

sions to be manipulated in fading, Bijou suggested that the

part of the form which includes the clue essential to the dis-

crimination be varied. This most closely approximates the goal of

bringing the subjects' attending behavior under the control of

that aspect of the stimulus which will be reinforced.










Several recent letter discrimination studies have focused upon

the effects of emphasizing essential versus nonessential features

in establishing letter discrimination. Using a match-to-sample

visual discrimination task, Egeland, Braggins, and Powalski (1973)

compared three methods of teaching 4- and 5-year-old children to

discriminate letters. One group of subjects was taught by using

the traditional reinforcement-extinction (trial and error)

approach. A second group received errorless discrimination train-

ing with the faded cue highlighting the distinctive feature of

the letter to be discriminated. A third group received errorless

discrimination training with a feature irrelevant to the dis-

crimination highlighted. The relevant cue group made signifi-

cantly fewer errors on posttests than did the irrelevant cue

and reinforcement-extinction group. These results suggest that

the effectiveness of errorless discrimination training depends on

whether the faded cue highlights a relevant or irrelevant

dimension of the letter to be discriminated. Egeland et al. con-

cluded that errorless discrimination training has educational

value for young children who are having difficulty learning to

discriminate letters or words, particularly where the child's

difficulty is related to his failure to attend to the relevant

dimension of the discriminative stimulus. He further suggested

that intervention using an errorless discrimination technique

to highlight the relevant distinctive features of letters offers

a promising solution to this problem at the preschool and kinder-

garten level.










Using a match-to-sample letter discrimination task, Tawney

(1972) also investigated the relative effects of training

children to respond to critical and noncritical features of

letterlike stimuli. Four-year-old children were divided into

three groups (relevant feature, nonrelevant feature, and

no-treatment control) based upon their performance in a letter

discrimination pretest. The two experimental groups were then

trained to respond to either "critical" or "noncritical" features

of letterlike forms. All subjects were then posttested. While

all groups made fewer errors on the posttest, mean error scores

for the critical-feature group were significantly lower than

those for the noncritical-feature group. Furthermore, the non-

critical-feature group scores were not significantly different

from the control group. Both treatment groups, however, showed a

reduction in the kinds of confusion errors relative to the con-

trol group, although the effect was more powerful in the

critical-feature group. Tawney's results suggest that critical-

feature training was effective in reducing both the frequency

and kinds of confusion errors, while the noncritical feature

group was effective in reducing only the kinds of confusion

errors. He suggested that future discrimination programs might

include a fading procedure and utilize actual letter stimuli.


SUMMARY

In developing a preventive program for children who will

later become poor readers, it is first necessary to accurately










identify those children destined to have severe reading diffi-

culties. Once these children have been identified, a program

designed to teach the essential skills of reading, starting with

the developmentally earlier skills and continuing sequentially

to the increasingly complex skills, could be implemented. Eleanor

Gibson's analysis of the reading process provides an empirically

supported developmental model of the processes involved in

learning to read.

The first step in developing an intervention program based

upon Gibson's analysis of the reading process is to design an

effective program for training children to differentiate

written symbols. Literature was cited which supports the proposi-

tion that letters are discriminated on the basis of their dis-

tinctive features and that the utilization of such features is a

necessary and sufficient condition for the discrimination of

letters. Although work on the analysis of the distinctive

features of graphic forms is not complete, four types of fea-

tures seem fairly well established. These are (1) break versus

close, (2) curve versus straight, (3) relative diagonality, and

(4) rotations and reversals. The most effective means of training

for letter discrimination appears to be a program which teaches

the child to discriminate critical features rather than actual

letters. Such a program would use a fading procedure in which

differences between forms would be exaggerated and gradually

minimized while maintaining correct discrimination performance.










Research on form discrimination and orientation discrimination

has shown that an effective training program must enable the

child to respond to the relevant stimulus dimension which

requires that errors in training be minimized, since the

occurrence of errors indicated that the subject has learned to

respond to irrelevant stimulus characteristics which become

highly resistant to change. By presenting a particular feature in

a variety of forms, the subject receives practice in scanning

forms and detecting the essential feature. A number of different

examples also allow a "learning set" or "concept" of the

critical feature to be established. The present experiment is

an attempt to incorporate these principles into a program for

training "high risk" children to discriminate letters.


HYPOTHESES

HYPOTHESIS I. Discrimination training emphasizing the dis-

tinctive features of letters would be more effective than tradi-

tional letter discrimination training without emphasis on dis-

tinctive features. Specifically, the experimental group would

make fewer errors on the "machine" posttest than the comparison

group.

HYPOTHESIS II. Transfer from the training apparatus to

printed material would be greater following distinctive feature

training than letter discrimination training. Furthermore, both

types of training would result in greater improvement in letter










discrimination than would normal classroom experience alone.

Specifically, the experimental group would make fewer confusions

and more correct responses than the comparison group, which in

turn would make fewer confusions and more correct responses

than the untreated control group on the paper-and-pencil match-

to-sample test.

HYPOTHESIS III. The improvements predicted in Hypothesis II

to follow treatment would continue to be present 3 months after

training had ended. Specifically, the experimental group would

continue to make fewer errors and more correct responses than

the comparison group, which in turn would make fewer errors and

more correct responses than the control group on the paper-and-

pencil match-to-sample test.

HYPOTHESIS IV. At the end of kindergarten, children who re-

ceived treatment would be rated by their teachers as performing

better academically than control children. Specifically, the

treatment group would be rated higher on scales of overall

achievement, ability to recite, recognize and print the alphabet,

ability to print own name, and ability to attend adequately.

Fewer treatment than control children would be predicted to

develop a reading problem.

HYPOTHESIS V. The improvements in letter discrimination pre-

dicted in Hypothesis II to follow treatment would continue to be

present one year after the termination of training. Specifically,

the experimental group would continue to make fewer errors and










more correct responses than the comparison group, which in turn

would make fewer errors and more correct responses than the

control group on the paper-and-pencil match-to-sample test.

HYPOTHESIS VI. Children in the treatment group would perform

better than control children on a word matching test one year

following letter discrimination training.

HYPOTHESIS VII. Children in the treatment group would perform

better than control children on a word recognition test one year

following letter discrimination training.

HYPOTHESIS VIII. Children in the treatment group would be

rated by their teachers as performing better academically than

control children one year after letter discrimination training.

Specifically, treatment children would be rated higher on a scale

of overall achievement than would control children, and fewer

treatment than control children would be diagnosed as having a

reading problem.














II

METHOD




SUBJECTS

All of the kindergarten children at Stephen Foster Elementary.

School (N=132) were screened at the beginning of the school year

(September 1974) using the Satz and Friel abbreviated battery.

Forty-four of these children were classified as high risk for

severe reading difficulties. Under a prior agreement with the

school, these 44 children were randomly assigned to one of three

groups. A special education group to be treated by school personnel

received 13 subjects. This group is unrelated to the experiment

reported in this paper and will not be considered further. The

31 remaining children were assigned to a treatment group (N=16)

and an untreated control group (N=15). Seven of these children

left the school prior to the completion of training, leaving a

treatment group of 13 subjects and a control group of 11 subjects.

The treatment group was further divided into an experimental

training group (N=7) and a treated comparison group (N=6).

The mean age for the 24 children at the beginning of train-

ing was 5 years 3 months, with a range from 4 years 10 months

to 6 years 9 months. Seventeen of the children were males and










seven were females. The treatment group was composed of nine

males and four females, while the untreated control group con-

tained eight males and three females. Thirteen of the children

were black and 11 were white. The treatment group was composed

of seven blacks and six whites, while the untreated control group

contained six blacks and five whites. These children were dis-

tributed across five kindergarten classes with no more than three

children from any group in a particular class.

An additional group of 12 children classified as potential

superior readers was used as a comparison group on the pretest.

These children had a mean age of S years 7 months, with a range

from 4 years 11 months to 5 years 11 months. Half of these

children were male and half were female; all were white.


APPARATUS

The training stimuli were two series of 884 slides of letter-

like forms in a match-to-sample format. A BRS Foringer Human Test

Console with an internally mounted Kodak Carousel projector and

M&M dispenser was used to present the training stimuli. The

Human Test Console is a large boxlike structure with a slanted

front. A 3-by-S-inch rear projection screen was mounted in the

center of the front panel. The screen was divided into three

equal-sized panels sensitive to compression. A smaller com-

pressable panel was located to the left of the projection panel.

The M&M dispenser fed into a stainless steel tray positioned in










the lower center portion of the test console. A clear removable

plastic guard was mounted over the tray to limit access to it. A

small light and buzzer were positioned over the tray. This equip-

ment was controlled by a portable BRS Foringer Logic Programing

System and was mounted in a 7-by-35-foot testing van which was

parked on the school grounds.


PROCEDURE

All experimental procedures were administered in the testing

van. The children were brought to the van by the experimenter in

groups of three for testing,in pairs for training. The order in

which the children came to the van each day for training was

varied, although the pairs of children were generally drawn from

the same classes.

PHASE 1 (Pretest, December 1974). All subjects were initially

administered a match-to-sample letter discrimination test. Sample

problems are presented in Figure 1. This test consisted of 22

match-to-sample problems: 5 letterlike forms used in training,

5 novel letterlike forms, 6 upper-case and 6 lower-case printed

letters. Each problem contained a sample and 16 choices. The num-

ber of identical alternatives varied from one to five, with a

mean of three for each problem. Subjects were told to "circle

all of the shapes, just like the one at the top." Because the

children worked slowly and became restless after several minutes,

the test was administered over three consecutive days. At the
































FIGURE 1. Sample Problems from the Paper-
and-Pencil Match-to-Sample Test.







~t-/Lk jr+


padqobqh
g p b dq gp q


XHK
KHM


M
ZANEM
X WA V\N


j










end of each session, each child received a choice of several

M&M's or a penny candy.


PHASE 2 (Machine Pretest, December 1974). Following comple-

tion of the paper-and-pencil pretest the children in the treat-

ment group were brought to the van in pairs and individually

taught to "work" the Human Test Console. While one child was

given paper and crayons and allowed to draw, the other child sat

in a chair in front of the Human Test Console. A curtain

separated the children. The child was told to look at the shape

in the center (a circle) and then to push the illuminated button

(orienting key) to the left of the display panel. When the child

did this, the orienting key went dark and the slide tray advanced

to present the child with three choice stimuli, one in each panel.

The child was told to look at each shape carefully and to push

the one that was like the shape he had just seen. If the child

chose correctly, the light above the tray was illuminated and a

tone sounded for 1.5 seconds. Simultaneously the M&M feeder was

activated and an M&M fell into the dish while the slide tray ad-

vanced to the next sample and the orienting key illuminated. The

child was told he had picked the right one and that every time he

made a correct choice he would receive an M&M. If the child

pushed an incorrect panel, the slide tray simply advanced to

the next sample. The child was told he had picked a shape that

was not the same as the one he had seen and that if he had

picked the correct choice he would have gotten an M&M. Five











match-to-sample problems of geometric shapes were used to

train the children to work the machine. Once each child correctly

matched all five problems, he was then administered a match-to-

sample test on the machine. Half of the test was given one day,

and half, the following day. This "machine" test was composed of

46 three-choice match-to-sample problems of real letters and

letterlike forms. Each problem contained an identical match, an

alternative differing from the sample by one distinctive feature,

and an alternative differing from the sample by more than one

distinctive feature. Each child was given the M&M's he had

earned at the end of the session. Several sample problems are

presented in Figure 2.

PHASE 3 (Treatment, December 1974 to February 1975). Train-

ing was begun the day after the completion of protesting. Train-

ing lasted for 8 weeks, 5 days a week including a 2-week vacation

after the first week. In all, the training series required over

700 correct matches. Each daily session required that each sub-

ject make 26 correct matches.

The match-to-sample training procedure was similar to the

machine pretest, but with two additions. An incorrect response

resulted in a brief time out, with the darkening of the display

panel followed by a back-up procedure which repeated the problem

the child had missed. The back-up procedure was programed to

repeat itself until the child made the correct response. When

the child did respond correctly following an error, he was

































FIGURE 2. Sample Problems from the Machine
Match-to-Sample Test.














l- SAMPLE







CHOICES









SSAMPLE






CHOICES









SAMPLE






CHOICES










presented with the light, tone, and next stimulus, but no M&M.

Figure 3 provides a verbal summary of the programed match-to-sample

procedure.

The first time each child made an incorrect response in the

training series, he was told he had picked the wrong one and would

have to wait to try again. An initial time out of 10 seconds

proved to be too long to maintain attention. Therefore, on the

second day of training the time out was reduced to 3.5 seconds

and maintained at that duration for the remainder of training.

The subjects were awarded M&M's each time they made a correct

response during protesting. A fixed ratio 2 was instituted dur-

ing the first day of training. Subjects were told they would re-

ceive an M&M for each two correct matches. During the second

week of training, a fixed ratio 3 was instituted, and the

children were told they would receive an M&M each time they made

three correct responses in a row. An incorrect response reset the

ratio to zero. Subjects were also given the choice of keeping

the M&M's they earned or trading them at the end of the session

for a piece of penny candy. This arrangement was maintained for

the remainder of the training sessions. Subjects had the option

of eating their candy immediately after their session or of

saving it for lunch time or after school. A clear plastic guard

prevented the children from handling the M&M's during the

sessions. The penny candy rewards were kept in a plastic box

on a shelf behind the subjects. The box, but not the candy, was











1. THE SAMPLE STIMULUS IS PROJECTED
IN THE CENTER OF THE DISPLAY WINDOW.
2. THE ORIENTING KEY TO THE LEFT OF
THE DISPLAY WINDOW IS ILLUMINATED.


1. COMPRESSION OF THE ORIENTING KEY RESULTS IN


1. DARKENING OF THE ORIENTING KEY.
2. PRESENTATION OF THE THREE-CHOICE
STIMULI IN THE DISPLAY WINDOW.


1. COMPRESSION OF THE
SECTION OF THE DISPLAY
WINDOW IN WHICH THE
CORRECT MATCH APPEARS
RESULTS IN




1. ILLUMINATION OF THE
LIGHT ABOVE THE M&M
TRAY.
2. PRESENTATION OF A
TONE.
3. DEPOSIT OF AN M&M
INTO THE M&M TRAY.
4. RETURN TO STEP 1
WITH PRESENTATION OF
THE NEXT SAMPLE
STIMULUS.


1. COMPRESSION OF THE
SECTION OF THE DISPLAY
WINDOW IN WHICH AN
INCORRECT MATCH APPEARS
RESULTS IN


1. BLACK-OUT OF THE
DISPLAY WINDOW FOR
3.5 SECONDS
FOLLOWED BY





2. RETURN TO STEP 1
WITH RE-PRESENTATION
OF THE NEXT SAMPLE
STIMULUS.


FIGURE 3. Summary of the Machine Match-to-Sample Paradigm.











visible during the training session. Daily sessions generally were

3 to 5 minutes in length.

The experimental group was administered distinctive feature

training. A distinctive feature manifest in three different forms

was emphasized during each session. The series was faded so

that the distinctive feature was exaggerated in the first match

of each form, while each succeeding match in the sequence pre-

sented a gradually diminishing difference between the forms. The

faded sequence was designed so that the initial discrimination

was easily made by all children while each succeeding discrimina-

tion was progressively more difficult. Figure 4 presents an

example of a faded sequence for each distinctive feature. The

three basic forms were alternated in the sequence to maximize

perceptual scanning for the feature necessary to make the dis-

crimination. The sequence in which the distinctive features were

presented is provided in Table 1. Ten of the training series con-

tain letterlike forms, and seven contain real letters, in order

to evaluate the effect of training on both training and novel

forms, independent of classroom experience with real letters.

Figure 5 presents the training and novel forms.

The treated comparison group made the same matches in the

same order as the experimental group, but the comparison stimuli

differed from the sample by more than one distinctive feature.

The comparison stimuli for any particular daily series included

the comparison stimuli of the other two forms in the equivalent

































FIGURE 4. A Faded Sequence for Each
Distinctive Feature.













CLOSE VERSUS BREAK



CURVE VERSUS STRAIGHT




ANGULARITY



ROTATION




REVERSAL


d d 4 bpbbb











TABLE 1. Sequence of Presentation of Distinctive Features.


Distinctive Feature

Close versus break
Angularity
Curve versus straight
Rotation
Reversal
Close versus break
Angularity
Curve versus straight
Rotation
Reversal
Additional feature
Close versus break
Angularity
Curve versus straight
Rotation
Reversal
Reversal
Close versus break
Angularity
Curve versus straight
Rotation
Reversal
Close versus break
Curve versus straight
Rotation
Reversal


Type of Forms

Artificial
Artificial
Artificial
Artificial
Artificial
Artificial
Artificial
Artificial
Artificial
Artificial
Real letters
Real letters
Real letters
Real letters
Real letters
Real letters
Real letters
Repeat-artificial
Repeat-artificial
Repeat-artificial
Repeat-artificial
Repeat-artificial
Repeat-real letters
Repeat-real letters
Repeat-real letters
Repeat-real letters
































FIGURE 5. Basic Training and Novel
Letterlike Forms.






TRAINING FORMS


NOVEL FORMS
i/c


A-


L
I

LI




Kl


w


6


C










distinctive feature series. Thus, the stimuli presented to

both groups were identical. Both groups made the same matches,

in the same order; only the incorrect choices differed for each

group on any particular discrimination. Thus, it was necessary

to respond to a single distinctive feature in the experimental

group and to multiple differences in the treated comparison

group. Figure 6 presents corresponding sample problems for the

experimental and treated comparison groups.

The control group received only classroom experience during

Phase 3.

PHASE 4 (Machine Posttest, February 1975). Following the

completion of the training series, the treatment group was re-

administered the match-to-sample letter discrimination test as

described in Phase 2.

PHASE 5 (Probe 1, February 1975). Children in both the

treatment and control groups were readministered the paper-and-

pencil match-to-sample test as described in Phase 1. This post-

test shall be referred to as Probe 1.

The treatment group was then given 8 weeks training in

grapheme-phoneme associations by another experimenter

(Fitzsimmons, 1975).

PHASE 6 (Probe 2, May 1975). Three months following the end

of Phase 3 (letter discrimination training), subjects in the

treatment and control groups were again administered the paper-

and-pencil match-to-sample test described in Phase 1. In addition


































FIGURE 6. Corresponding Match-to-Sample
Problems for the Experimental and
Comparison Groups--Curve versus Straight
Feature.












It
1t~t


+1o:=


Ir


^ A


EXPERIMENTAL
2


It


COMPARISON


~~ ~z










to this test, the children were administered the Beery Test of

Visual Motor Integration (a screening battery subtest) and

teacher ratings were obtained. Teachers were asked to rate each

child on scales related to achievement group, ability to recite,

recognize, and write letters on command, and ability to write name.

Teachers were also asked to rate each child's attention span and

to predict whether each child would or would not have a reading

problem. The questions included in the questionnaire are

presented in Table 2. The data obtained during the 3-month follow-up

shall be referred to as Probe 2.

PHASE 7 (Probe 3, February 1976). One year following the

end of training, the children in the treatment and control

groups were readministered the paper-and-pencil match-to-sample

test as described in Phase 1. In addition to this test, the

children were also administered the Beery Developmental Test

of Visual-Motor Integration (Beery 4 Buktenica, 1967), a word

matching subtest of the Clymer-Barrett Prereading Battery

(Clymer 4 Barrett, 1967), and the Iota Word Test (Monroe, 1932).

The Beery requires the child to copy graphic designs of in-

creasing complexity. The Clymer-Barrett word matching subtest

requires the child to select a word from a group of four that

is identical to the sample word. For example, pan appears on

the left as the sample word, followed by the choices nap, pin,

pan, and ban. The Iota is a word recognition test which requires

the child to pronounce increasingly more difficult words. Also,











TABLE 2. Sample Questions from Kindergarten and First-Grade
Teacher Questionnaires.


Kindergarten Student Achievement Questionnaire

Please rate on the following dimensions with-
out regard to your knowledge of this student's participation in
any special program.
Was this student originally assigned to a (1) superior, (2) average,
(3) below average achievement-learning group?
Is this student presently working in a (1) superior, (2) average,
(3) below average achievement-learning group?
Is this child able to recite (1) all, (2) most, (3) approximately
half, (4) few, (5) none of the lower-case letters of the alphabet.
Is this child able to write his/her name? Yes--no.
Is this child able to write on command (1) all, (2) most, (3) ap-
proximately half, (4) few, (5) none of the letters of the lower-
case alphabet?
Is this child able to recognize by naming (1) all, (2) most,
(3) approximately half, (4) few, (5) none of the lower-case
letters of the alphabet?
Does this child, in your opinion, have a (1) good, (2) average,
(3) poor attention span?
Do you think this child has or will have a reading problem? Yes--no.

First-Grade Achievement Questionnaire

Please rate on the following dimensions:
Is this student presently working in a (1) superior, (2) average,
(3) below average achievement-learning group?
Do you think this child has or will have a reading problem? Yes--no.






51



teacher ratings of achievement level and prediction of reading

problems were again obtained. The one-year follow-up shall be

referred to as Probe 3.

Twenty of the 24 children comprising the treatment and con-

trol groups were enrolled in county schools and were included

in Probe 3. The remaining children could not be located. A flow

chart of the complete procedure is presented in Figure 7.










EXPERIMENTAL GROUP COMPARISON GROUP CONTROL GROUP PHASE

PAPER-AND-PENCIL MATCH-TO-SAMPLE-TEST 1

MACHINE MATCH-TO-SAMPLE TEST 2

26 SESSIONS OF 26 SESSIONS OF NORMAL
DISTINCTIVE LETTER MATCHING CLASSROOM 3
FEATURE TRAINING TRAINING EXPERIENCE

MACHINE MATCH-TO-SAMPLE TEST 4

PAPER-AND-PENCIL MATCH-TO-SAMPLE TEST (PROBE 1) 5

35 SESSIONS OF GRAPHEME-PHONEME TRAINING NORMAL CLASSROOM
BY ANOTHER EXPERIMENTER EXPERIENCE

PAPER-AND-PENCIL MATCH-TO-SAMPLE TEST
BEERY DEVELOPMENTAL TEST OF VISUAL-MOTOR INTEGRATION 6
TEACHER RATINGS
(PROBE 2)

NORMAL CLASSROOM EXPERIENCE (AND SUMMER VACATION)

PAPER-AND-PENCIL MATCH-TO-SAMPLE TEST
BEERY DEVELOPMENTAL TEST OF VISUAL-MOTOR INTEGRATION
WORD MATCHING SUBTEST OF THE CLYMER-BARRETT PREREADING BATTERY 7
IOTA WORD RECOGNITION TEST
TEACHER RATINGS
(PROBE 3)



NOTE: Phase 1, December 1974; Phase 2, December 1974; Phase 3,
December 1974 to February 1975; Phase 4, February 1975: Phase 5,
February 1975; Phase 6, May 1975; Phase 7, February 1976.


FIGURE 7. Flow Chart of Experimental Procedure.














III

RESULTS




PREVENTION RESEARCH PRESENTS a unique problem in assessing the

effects of treatment. The proportion of false positives in the

target population must be considered. Data from Satz, Friel,

and Rudegeair's (1975) original validation study using third-

year reading level as the criterion measure reveal that 82% of

children classified as high risk for severe reading difficulties

indeed exhibited reading difficulties in third grade, while 18%

were average readers (false positives). Applying this data to

the present sample, one would expect that approximately 18% of

the high risk children would actually become average readers

without treatment. This would equal four to five children in

the combined treatment and control groups.

Teacher ratings at the end of kindergarten (Probe 2) and in

the middle of first grade (Probe 3) were examined in order to

identify and eliminate apparent "false positives" from the data

analysis. Table 3 presents the teacher ratings dichotomized to

reflect above and below average performance. Subjects are ranked

according to the proportion of above and below average ratings.

Three children in the control group were expected by their








TABLE 3. Teacher Ratings Dichotomized to Show Above (+) and Below (0) Average Performance.

Read. Ach. Read. Ach. Alph. Alph. Alph. Write Att.
Prob. Grp. Prob. Grp. Write Recog. Recit. Name Span
Probe Probe Probe Probe Probe Probe Probe Probe Probe
2 2 3 3 2 2 2 2 2

TREATMENT GROUP
1C* 0 0 0 0 0 0 0 0 0
2E 0 0 0 0 0 0 + + 0
3C 0 0 0 0 0 + 0 + 0
4C 0 0 0 0 0 0 0 + +
5E 0 0 0 0 0 0 0 + +
6C 0 0 0 0 0 0 0 + 0
7E + 0 0 0 0 0 + 0 +
8E 0 + 0 0 0 + + + 0
9C 0 + 0 + 0 + + + +
10E + + 0 0 0 + + + +
11E + + 0 0 0 + + + +
12E + + + + + + 0
13C + + + + + + +

CONTROL GROUP
1 0 0 0 0 0 0 0 0 0
2 0 0 0 0 0 0 0 0 0
3 0 0 0 0 0 0 0 + 0
4 0 0 0 0 0 0 0 0 +
5 0 0 0 0 0 0 + + 0
6 + 0 0 0 0 + 0
7 + 0 0 0 0 0 + + +
8 + 0 + + 0 0 0 + 0
9 + + + + + + + + +
10 + + + + + + + + +
11 + + + + + + + + +


E = experimental group


and C = comparison group.











kindergarten and first-grade teachers to become average readers

and were clearly classified as performing at an average or

above average level in kindergarten and first grade. Three

children in the treatment group also met these criteria. However,

since the treatment was designed to affect reading performance,

the number of "false positives" in the treatment group would be

predicted to increase as a result of treatment. Therefore,

teacher ratings of the original achievement level of each child,

as well as performance on the screening battery itself, were

considered in eliminating "false positives" from the treatment

group. These data provide indications of pretreatment perform-

ance levels. Table 4 presents the original achievement level

ratings and screening battery performances dichotomized to

reflect above and below average performance. Only two of the six

potentially false positive subjects were rated by their kinder-

garten teachers as performing in the average range at the be-

ginning of the school year. However, all three of the false posi-

tive subjects in the control group performed in the average

range on more than half of the screening battery subtests, while

only one of the potentially false positive subjects in the treat-

ment group did so. Since this subject was also the only subject

in the treatment group whose original classroom performance had

been in the average range, this subject was eliminated from the

data analysis. The other two subjects were retained because

their pretreatment performance was below average. The identification








TABLE 4. Screening Battery Performances and Teacher Ratings of Pretreatment Achievement
Levels Dichotomized to Show Above (+) and Below (0) Average Performance.


Teacher
Rating
Orig.
Ach.
Level
Probe
2


S SCREENING BATTERY


Peab.
Alph. Pict.
Recit. Voc.


Beery
Test
Recog. Finger V-M
Discri. Local. Integ.


Aud. Dich.
Discri. List.


TREATMENT GROUP


CONTROL GROUP











of four "false positives" from the two groups is within the

range of four to five predicted by the original validation

study. The elimination of these subjects leaves an experimental

group of N=6, comparison group of N=6, and control group of

N=8.

A Kruskal-Wallis One-Way Analysis of Variance by Ranks was

used to analyze all data comparing the experimental, comparison,

and control groups, while a Mann-Whitney U-Test was used to

analyze all data comparing only two groups. Analysis of teacher

ratings was done using the Fisher Exact Probabilities Test.


MACHINE MATCH-TO-SAMPLE TEST

The mean number of incorrect matches on the match-to-sample

machine pretest was 17 (range 5-31) for the experimental group

(N=6) and 19.6 (range 11-32) for the comparison group (N=6).

These scores were not significantly different (U=14, pc.294). On

the machine posttest the mean number of incorrect matches was

9.1 (range 5-13) for the experimental group (N=6) and 10.5

(range 8-14) for the comparison group (N=6). Again, these scores

were not significantly different (U=14.5, pR.294). Thus, no

differential effect was present for distinctive feature training

versus letter discrimination training on the match-to-sample

machine posttest.










PAPER-AND-PENCIL MATCH-TO-SAMPLE TEST

PRETEST. Both confusion errors and correct matches were

scored on the match-to-sample letter discrimination test. The

mean number of confusion errors on the pretest was 13.6 (range

5-28) for the experimental group (N=6), 16.6 (range 7-28) for

the comparison group (N=6), and 17.5 (range 1-34) for the control

group (N=8). Pretest error scores did not differ significantly

among these three groups (H=.69, p<.80). The mean number of errors

for the superior group (N=13) was 1.8 (range 1-4). The scores of

the superior group differed significantly from those of the com-

bined high-risk groups (U=8,pc.001).

The mean number of correct matches made on the pretest was

60.8 (range 43-67) for the experimental group, 56.5 (range 29-68)

for the comparison group, and 60 (range 50-69) for the control

group. Pretest correct scores did not differ significantly among

these three groups (H=.70,p<.80). The mean number of correct

matches for the superior group was 65.6 (range 56-68). The

scores of the superior group differed significantly from those

of the combined high-risk groups (U=60, p_.01). These results

confirm that no difference among the high-risk groups on the

letter matching test were present prior to training. Comparisons

with the superior group demonstrate that the test does dis-

criminate predicted above and below average readers.










PROBE 1. The mean number of confusion errors on the posttest

(Probe 1) was 7.1 (range 2-25) for the experimental group (N=6),

8.3 (range 3-17) for the comparison group (N=6), and 9.5 (range

3-24) for the control group (N=8). These scores were not sig-

nificantly different (H=2.74, pr.30), although the scores do

represent a trend in the predicted direction. A survey of pretest

scores indicated that only one of the high-risk children, a con-

trol subject, fell within the range of the superior group, while

posttest scores indicated that 67% of the experimental subjects,

17% of the comparison subjects, and 25% of the control subjects

fell within the superior range. These percentages represent a

trend in the predicted direction, although this finding is not

significant (Fisher Test, p >.05).

The mean number of correct matches on the posttest was 63.3

(range 59-69) for the experimental group, 61.5 (range 51-68) for

the comparison group, and 59.3 (range 46-65) for the control

group. These scores were not significantly different (H=1.36,

p7.70). These results demonstrate the absence of a clear treat-

ment effect, although there appears to be a reduction in confu-

sion errors in the predicted direction for the experimental

group.

TRAINING VERSUS NOVEL FORMS. The mean number of pre and

posttest confusion errors for training and novel forms was also

calculated. The mean number of errors on pretest training forms

was 4.8 (range 1-11) for the experimental group, 7.1 (range 3-12)










for the comparison group,and 7.6 (range 0-20) for the control

group. These scores were not significantly different (H=.79,

p<.70). The mean number of errors on pretest novel forms was

5.1 (range 1-10) for the experimental group, 4.8 (range 3-9) for

the comparison group, and 6.0 (range 1-10) for the control group.

These scores were not significantly different (H=4.55, pc.20).

The mean number of errors on posttest training forms was 2.5

(range 0-10) for the experimental group, 4.3 (range 2-8) for

the comparison group, and 4.0 (range 0-9) for the control group.

These scores were not significantly different (H=3.00, p3.30).

The mean number of errors on posttest novel forms was 3.6

(range 1-10) for the experimental group, 3.0 (range 1-8) for the

comparison group, and 3.3 (range 0-9) for the control group.

These scores were.not significantly different (H=..05, p<.98).

Thus, no differences were apparent among the groups in the num-

ber of errors made on training or novel forms following training.

PROBE 2. The mean number of errors on the 3-month follow-up

(Probe 2) was 7.6 (range 1-25) for the experimental group (N=6),

10.3 (range 2-20) for the comparison group (N=6), and 10.1

(range 2-21) for the control group (N=7). These scores were not

significantly different (H=1.46, pR.50). However, 67% of the

experimental group continued to score in the superior range,

while only 33% of the comparison group and 28% of the control

group scored in the superior range.










The mean number of correct matches on the 3-month follow-up

was 59.1 (range 55-62) for the experimental group, 62.3

(range 57-67) for the comparison group, and 59.1 (range 49-69)

for the control group. These scores were not significantly

different (H=1.70, p<.50). Although no latent treatment effect

in letter matching performance was apparent 3 months following

treatment, the finding of an increased number of experimental

subjects performing in the superior range for letter confusions

continued to be present.

PROBE 3. The mean number of errors on the one-year follow-up

(Probe 3) was 5.1 (range 1-8) for the experimental group (N=6),

4.7 (range 3-6) for the comparison group (N=4), and 6.1 (range

1-15) for the control group (N=7). These scores were not signifi-

cantly different (H=.ll, p<.95). A survey of scores indicated

that 33% of the experimental group, 25% of the comparison group,

and 57% of the control group performed in the superior range.

The mean number of correct matches on the one-year follow-up

was 63.6 (range 59-67) for the experimental group, 64.7 (range

58-69) for the comparison group, and 65.1 (range 57-70) for the

control group. These scores were not significantly different

(H=.72, p<.70). No differences in confusion errors or correct

matches were present one year following letter discrimination

training. The tendency for more experimental subjects to perform

in the superior range was no longer present one year following

training.










Figure 8 presents the mean confusion errors for each group

plotted across probes, while Figure 9 presents the mean number of

correct matches for each group plotted across probes. It should

be noted that at no time do any of the mean high-risk scores

equal or exceed the mean pretest scores of the superior group,

even on the one-year follow-up.


THREE-MONTH FOLLOW-UP (PROBE 2)--TEACHER RATINGS

Teacher ratings taken at the end of kindergarten were di-

chotomized, cast into a 2-by-2 contingency table, and analyzed

using the Fisher Exact Probabilities Test. Since the purpose of

the teacher ratings was to assess the effects of the combined

letter discrimination and grapheme-phoneme training, the experi-

mental and comparison groups were combined in this analysis.

The treatment and control groups did not differ signifi-

cantly on ratings of original achievement-learning group,

present achievement-learning group, alphabet recitation, ability

to write the alphabet on command, ability to write the child's

own name, attention span, and prediction of reading problem. The

children in the treatment group, however, were rated by their

teachers as performing significantly better than the control

group at recognizing letters of the alphabet (p<.025).
































FIGURE 8. Mean Confusion Errors on the
Paper-and-Pencil Match-to-Sample Test
Plotted Across Probes for the Experimental
(0), Comparison (0), Control (0), and
Superior (0) Groups.




































































PRETST PROBE 1 PROBE 2 PROBE 3
PRETEST PROBE 1 PROBE 2 PROBE 3


15













10












5































FIGURE 9. Mean Correct Matches on the Paper-
and-Pencil Match-to-Sample Test Plotted
Across Probes for the Experimental (0),
Comparison (0), Control (0), and Superior
(0) Groups.



























































PROTEST PROBE 1 PROBE 2 PROIE 3
PRETEST PROBE 1 PROBE 2 PROBE 3


65












60












55 -










ONE-YEAR FOLLOW-UP (PROBE 3)--OBJECTIVE MEASURES AND TEACHER RATINGS

In addition to the match-to-sample test already discussed,

the word-matching subtest of the Clymer-Barrett Prereading Battery

was included in Probe 3 in order to guard against ceiling effects

on the letter matching test. The Iota Word Recognition Test

was also included as an objective measure of beginning reading

ability. The mean number of confusion errors on the word-matching

test was 5.2 (range 0-17) for the treatment group (N=10) and 5.7

(range 3-10) for the control group (N=7). These scores were not

significantly different (U=24.5, p>.05). The mean number of

correct pronunciations on the word recognition test was 3.5 (range

0-9) for the treatment group and 1.4 (range 0-10) for the control

group. These scores were not significantly different (U=19, p>.05).


BEERY TEST OF VISUAL-MOTOR INTEGRATION

The Beery Test of Visual Motor Integration was included in

Probes 2 (kindergarten level) and 3 (first grade) because of

changes reported by school personnel on this measure in their

treatment group. Pretraining scores were available because the

Beery is one of the tests included in the screening battery. The

mean pretraining score in months was 46.6 (range 34-60) for the

experimental group (N=12) and 52.1 (range 34-67) for the control

group (N=8). These scores were not significantly different

(U=38, p>.05). The mean score at Probe 2 (8 months later) was

65.5 (range 57-82) for the experimental group (N=ll) and 56.3










(range 46-63) for the control group (N=6). These scores differed

significantly in the predicted direction (u=15, p..05). This

represents an average increase of 18.9 months for the treatment

group and 4.2 months for the control group. The mean difference

between chronological age and Beery age equivalent on the pre-

training administration was 12.5 months for the treatment group

and 9.9 for the control group. At Probe 2, however, the difference

was 2 months for the treatment group and 13.7 for the control

group. The mean score at Probe 3 (16 months after the initial

testing) was 69.2 months for the experimental group and 67.2

months for the control group. These scores were not significantly

different (U=27, p>.05). The mean difference between chronological

age and Beery age equivalent was 6.3 for the treatment group and

10.8 for the control group. Figure 10 presents the mean scores

in months plotted across probes. These data suggest that the com-

bined grapheme and grapheme-phoneme training resulted in a sub-

stantial improvement in visual-motor performance. Furthermore,

it would appear that the effect obtained from treatment was not

retained 9 months later.































FIGURE 10. Mean Beery Age Equivalents in
Months Plotted Across Probes for the
Treatment (0) and Control (@) Groups.







70


















70-


0









z

60-





CY





~~T1










I I
PRETEST PROBE 2 PROBE 3














IV

DISCUSSION



ONE PURPOSE OF THIS EXPERIMENT was to evaluate two methods of

letter discrimination training with children predicted to have

severe reading difficulties. The goal of treatment was to im-

prove letter discrimination performance by reducing confusion

errors. Hypotheses I, II, III, and V specifically relate to im-

proving letter discrimination performance. Hypothesis I was not

confirmed. No differences in confusion errors were present for

the "distinctive feature group" versus "letter discrimination

group" on the "machine" match-to-sample posttest. However, both

groups reduced their mean number of errors by almost one half.

Hypotheses II, III, and V also were not confirmed. No differences

in either confusion errors or correct matches were present for

the experimental, comparison, and control groups on the paper-

and-pencil match-to-sample test immediately, 3 months, or one

year following discrimination training. However, all groups re-

duced their mean number of errors by almost one half in the 2-

month period between protesting and posttesting. It is interest-

ing that confusion errors did not decrease over the next 3-month

period, which covered the remainder of the school year, but did

decrease moderately over the next 9-month period, which included











summer vacation and the initial half of first grade. Even in

the middle of first grade the "high risk" group continued to

average more confusion errors than the "superior" group had in

early kindergarten.

The pattern of change for number of correct matches across

the four probes was less consistent and smaller for the three

groups than was the pattern of change of error responses. The

control group showed no change in mean number of correct re-

sponses on the pretest, posttest, and 3-month follow-up, but then

improved almost enough from Probe 2 to Probe 3 to equal the

superior group pretest performance. The experimental group

showed an initial improvement on the posttest, then a regression

on Probe 2, and finally matched their posttest performance on the

one-year follow-up. The comparison group showed an initial

sharp increase in correct matches, followed by a flattening out

between the posttest and 3-month follow-up and then a slight in-

crease again at Probe 3. All three groups approached the mean

pretest performance of the superior group on the one-year follow-up.

These results suggest that neither the faded distinctive

feature training nor the more traditional multiple-feature dis-

crimination training were more effective than regular classroom

teaching in improving letter discrimination performance. The

lack of differences in errors on training and novel forms across

the three groups further substantiates this conclusion. It is

unclear to what extent the pattern of changes in scores across










probes is the result of repeated testing, maturation, and actual

classroom experience. It does seem quite possible that the

greater amount of class time spent each day on the alphabet and

prereading skills may have overshadowed or "washed-out" the

effect of very brief, daily treatment sessions.

One shortcoming of the present experiment is that the very

small number of subjects in each group allows only very simple

statistical analyses requiring clear differences in order to

reach significance. This is an important consideration in pre-

vention research because success might better be represented by

the increase in proportion of high risk subjects performing in

the normal range, rather than simply by a demonstration of a

statistically significant increase in scores. In regard to letter

confusions, it was found that two thirds of the distinctive

feature group versus less than one fourth of the comparison and

control groups performed in the "superior group" range imme-

diately following training. This trend was visible 3 months

following training: two thirds of the experimental group

continued to perform in the superior range, while less than one

third of the comparison and control groups did so. However, this

trend was not apparent one year following discrimination training.

Although this effect is not statistically significant, it does

suggest that an increased proportion of experimental subjects

may have performed in the "superior range" on the paper-and-pencil

match-to-sample test as a result of distinctive feature training.










Another purpose of this experiment was to evaluate the effect

of an intervention program combining letter discrimination

training and grapheme-phoneme training on reading related per-

formances. Hypotheses IV, VI, VII, and VIII relate to reading

related performances. Hypothesis IV, for the most part, was not

confirmed. No significant differences between the treatment and

control groups were present immediately following grapheme-

phoneme training on teacher ratings of overall achievement,

ability to recite or print the alphabet, ability to print own

name, and attention span. No differences between the treatment

and control groups were present for teacher predictions of the

presence or absence of a reading problem for each child. However,

significantly more treatment than control children were judged

by their teachers as being able to recognize most of the letters

of the alphabet. This is not surprising if one considers that

treatment is more directly related to this measure than any of

the others. Hypothesis VIII relates to teacher ratings taken 9

months after the termination of grapheme-phoneme training. No

differences were present on ratings of overall achievement in

the middle of first grade nor were any differences noted in the

proportion of children in each group predicted to develop a

reading problem.

Hypotheses VI and VII relate to two objective measures ad-

ministered 9 months after the end of combined treatment. Hy-

pothesis VI was not confirmed. No differences were apparent











between children in the treatment and control groups on the

Clymer-Barrett word matching subtest. This test was included at

Probe 3 in order to guard against a possible ceiling effect on

the paper-and-pencil match-to-sample test, since the word-

matching test was composed of confusing letter sequences rather

than single confusing letters, and letter sequences are generally

more difficult to discriminate than single letters. However, the

directions allowed for only one choice in each match-to-sample

problem. Many of the children initially circled several choices,

but when required to choose only one, they inevitably chose the

identical match and eliminated the similar but not identical

alternatives. This is consistent with Caldwell and Hall's (1969)

observation that letter confusions do not result from perceptual

or attentional deficits, but rather from differences in child-

ren's concept of same/different. The large number of correct

matches made throughout the match-to-sample letter discrimina-

tion test also tends to support this conclusion.

Hypothesis VII was not confirmed. No differences were

present 9 months after combined letter discrimination and

grapheme-phoneme training on the Iota word recognition test.

Considered together, these results suggest that letter discrimina-

tion and grapheme-phoneme training had little effect on teacher

ratings of children's academic performance or on several objective

measures of reading related skills, either immediately follow-

ing training or 9 months later.










Dramatic differences were apparent on only one measure: the

Beery Test of Visual-Motor Integration. The treatment group

scored a year below its chronological age on the original ad-

ministration, while the control group scored almost 10 months

below age expectancy. Following training, the treatment group

scored only 2 months below their chronological age, while the

control group scored over a year below age expectancy. This

represents an average increase of over a year and a half for the

treatment group but only 4 months for the control group.

Although the treatment group remained only 6 months below age

expectancy 9 months following treatment, while the control group

continued to be over 10 months below age expectancy, performances

were not significantly different at this time. These results

suggest that letter discrimination and grapheme-phoneme training

significantly improved Beery visual-motor performance, although

the effect appears to have been temporary.

The discovery of a treatment effect on the Beery Test of

Visual Motor Integration highlights a potential difficulty of

prevention research. Although it is not unreasonable that

training in visual discrimination would effect visual-motor

functioning, it is surprising to register such a dramatic effect

on a measure indirectly related to training when an absence of

effect was seen on more direct measures. This supports the

observation that experimental treatments may often have effects

not predicted or intended by the experimenter. Such peripheral











effects can be as important, pragmatically, as the evaluation of

main effects.

A particular difficulty exists in prevention research where

an attempt is made to affect a behavior before it has appeared

developmentally. One of the purposes of this study was to improve

reading skills, with the goal of reducing the proportion of

children eventually performing below grade level in reading.

Treatment was administered to subjects predicted to develop

reading difficulties before they had in fact begun to learn to

read. After treatment it is difficult to differentiate false

positive subjects (those who would be average readers without

treatment) from valid positive subjects who are in fact average

readers as a result of treatment. In the present experiment a

broader assessment of each child's academic functioning prior to

treatment, in the form of teacher ratings and readiness tests,

may have helped to differentiate false positive and valid

positive subjects following treatment. This is particularly im-

portant when for practical reasons the number of subjects must

remain small. Studies using large numbers of subjects in com-

bination with accurate conditional probabilities of false and

valid positive rates would also minimize this problem.

The results of this experiment suggest that a broad range of

related skills in the form of independent criterion measures

should be assessed prior to and following treatment, in order to

differentiate false and valid positive subjects, as well as to










more extensively sample the effects of treatment. The present

results indicate that a skill indirectly related to treatment

showed improvement while skills more directly related to

treatment appeared unaffected. The importance of follow-up

probes was also supported. While improvement in visual-motor

performance was present immediately following training, no

differences were present between treatment and control groups 9

months later.

In summary, distinctive feature and letter discrimination

training did not appear to be any more effective than regular

classroom experience in reducing letter confusions and increas-

ing correct letter matching. Furthermore, combined letter dis-

crimination and grapheme-phoneme training did not appear to

affect teacher ratings of reading related classroom performance,

except that more treatment than control children were rated as

recognizing more than half the letters of the alphabet. Children

in the treatment group performed significantly better than those

in the control group on the Beery Test of Visual Motor Integra-

tion immediately following training, but the difference was not

present 9 months later. No differences between treatment and

control groups were apparent on word matching and word recogni-

tion tests 9 months following training.

Eighty-three percent of the high-risk children began kinder-

garten in the below-average learning group. At the end of kinder-

garten, 67% remained in the below-average learning group while 54%










were predicted by their teachers to develop reading difficulties.

Half way through first grade, 70% of the high-risk children were

performing below average academically while 75% were seen as

having, or were predicted to develop, reading problems. The

children included in this experiment began kindergarten far

behind the majority of children their age on such diverse skills

as visual-motor development, visual discrimination, auditory dis-

crimination, somatosensory development, alphabet knowledge, and

extent of vocabulary. These children represent a developmentally

retarded sample of children. The vast majority of these children

continued to perform far below the level of their peers in

first grade. It seems reasonable to conclude that more intensive

intervention may be required in order to produce dramatic

effects.















REFERENCES





Beery, K., and Buktenica, N. Developmental Test of Visual-Motor
Integration. Chicago: Follett Publishing Co., 1967.

Bijou, S. Studies in the experimental development of left-right
concept in retarded children using fading techniques. In
N.R. Ellis (ed.), International Review of Research in Men-
tal Retardation, 1968, vol. 3.

Bloom, S. Stability and Change in Human Characteristics. New
York: John Wiley and Sons, Inc., 1964.

Caldwell, E., and Hall, V. The influence of concept training on
letter discrimination, Child Development, 1969, 40, 63-71.

Clymer, T., and Barrett, T. Clymer-Barrett Prereading Battery,
Form A. Lexington, Mass., 1967.

Davidson, H. P. A study of the confusing letters, "b," "d,"
"p," and "q." Journal of Genetic Psychology, 1935, 47,
458-68.

Dunn-Rankin, P. The similarity of lower case letters of the
English alphabet. Journal of Verbal Learning and Verbal
Behavior, 1968, 7, 990-95.

Dunn-Rankin, P., Leton, D., and Shelton, V. Congruency factors
related to visual confusion of English letters. Perceptual
and Motor Skills, 1968, 26, 659-66.

Egeland, B., Braggins, J., and Powalski, J. Teaching children
to discriminate letters of the alphabet through errorless
discrimination training. Paper presented at the Biennial
Meeting of the Society for Research in Child Development,
Philadelphia, March 1973.

Fitzsimmons, J. The Development of Grapheme-Phoneme Associations
in Children. Master's thesis, University of Florida, 1975.











Gibson, E. The ontogeny of reading. American Psychologist, 1970,
25, 136-43.

Learning to read. Science, 1965, 148, 1066-72.

Gibson, E., Gibson, J., Pick, A., and Osser, H. A developmental
study of the discrimination of letter-like forms. Journal
of Comparative Physiological Psychology, 1962, 55, 897-906.

Gibson, E., and Levin, H. The Psychology of Reading. Cambridge,
Mass.: MIT Press, 1975.

Gibson, E., Osser, H., Schiff, W., and Smith, J. An analysis of
critical features of letters, tested by a confusion matrix.
In Final Report: A Basic Research Program on Reading, Coopera-
tive Research Project No. 639, Cornell University and U.S.
Office of Education, 1963.

Gibson, E.J., Schapiro, F., and Yonas, A. Confusion matrices for
graphic patterns obtained with a latency measure. In The
Analysis of Reading Skill: A Program of Basic and Applied
Research, Project No. 5-1213, Cornell University and United
States Office of Education. Ithaca, New York: Cornell Uni-
versity, 1968.

Guralnick, M. Alphabet discrimination and distinctive features:
Research review and educational implications. Journal of
Learning Disabilities, 1972, 5, 428-34.

Hammill, D. Training visual perceptual processes. Journal of
Learning Disabilities, 1972, 5, 552-59.

Hively, W. Programing stimuli in matching to sample. Journal of
the Experimental Analysis of Behavior, 1962, 5, 279-98.

Kline, C.L. The adolescents with learning problems: How long
must they wait? Journal of Learning Disabilities, 1972, 5,
262-71.

Monroe, M. Children who Cannot Read. Chicago: University of
Chicago Press, 1932.

Moore, R., and Goldiamond, I. Errorless establishment of visual
discrimination using fading procedures. Journal of the
Experimental Analysis of Behavior, 1964, 7, 269-72.

Pick, A. Improvement of visual and tactual form discrimination.
Journal of Experimental Psychology, 1965, 69, 331-39.











Popp, H. Visual discrimination of alphabet letters. Reading
Teacher, 1964, 18, 221-26.

Satz, P., and Friel, J. The predictive validity of an abbreviated
screening battery: A preliminary crossvalidation study
[Manuscript in preparation which was shown to the author of
this study by P. Satz, 1976].

Satz, P., Friel, J., and Rudegeair, F. Some predictive ante-
cedents of specific reading disability: A two-, three-, and
four-year follow-up. In J.T. Guthrie (ed.), Aspects of
Reading Acquisition. Baltimore: Johns Hopkins Press, 1975.

Sidman, M., and Stoddard, L.T. The effectiveness of fading in
programing a simultaneous form discrimination for retarded
children. Journal of the Experimental Analysis of Behavior,
1967, 10, 3-15.

Strag, G. Comparative behavioral ratings of parents with severe
mentally retarded, special learning disability, and normal
children. Journal of Learning Disabilities, 1972, 5, 52-56.

Tawney, J. Training letter discrimination in four-year-old
children. Journal of Applied Behavior Analysis, 1972, 5,
455-65.

Terrace, H.S. Discrimination learning with and without errors.
Journal of the Experimental Analysis of Behavior, 1963, 6,
1-27.

Touchette, P. The effects of graduated stimulus change on the
acquisition of a simple discrimination in severely retarded
boys. Journal of the Experimental Analysis of Behavior,
1968, 11, 39-48.

Vernon, M.D. Backwardness in Reading. London: Cambridge Press,
1960.

Williams, J. Training kindergarten children to discriminate
letter-like forms. American Educational Research Journal,
1969, 6, 501-514.














BIOGRAPHICAL SKETCH




TERRY SHANNON PROEGER was born 9 May 1947, in Pittsburgh, Penn-

sylvania. He attended schools in Port Vue, Pennsylvania, and in

June 1965 was graduated from Port Vue-Liberty High School. He

received the degree of Bachelor of Arts from the University of

Florida in June 1969, and in September of that year he began

study in psychology at the University of Florida. He received

the degree of Master of Arts from the University of Florida in

June 1972.











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, Chailman
Professor of Clinical 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.



William Dawso
Professor of Opthalmology




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.



/Franz Epting
Associate Professr 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 disserta o
for the degree of Doctor of Philosophy.



Ira Fischler
Assistant 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.




Louis Cohen
Professor and Chairman
of Clinical Psychology




This dissertation was submitted to the Graduate Faculty of the
Department of Psychology in the College of Arts and Sciences
and to the Graduate Council, and was accepted as partial ful-
fillment of the requirements for the degree of Doctor of
Philosophy.

June 1976


Dean, Graduate School




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs