THESIS WM“- ' ‘

 

 

This is to certify that the
thesis entitled

A DEVELOPMENTAL ANALYSIS OF HAND AND VISUAL HEMIFIELD
DIFFERENCES FOR LETTER IDENTIFICATION: EFFECTS OF
MODALITY OF PRESENTATION AND LETTER TYPEFACE
CHARACTERISTICS

presented by

Nancy M. Wagner

has been accepted towards fulfillment
of the requirements for

DOCTOR OF PHILOSOPHY degree in Psychology

 

 

Lajor professor

Date 6 July 198].

 

0-7639

 

 

OVERDUE FINES:
25¢ per day per item
RETURNING LIBRARY MATERIALS:

Place in book return to remove
charge from circulat‘lon records

 

 

 

 

 

A DEVELOPMENTAL ANALYSIS OF HAND AND VISUAL HEMIFIELD
DIFFERENCES FOR LETTER IDENTIFICATION: EFFECTS OF
MODALITY OF PRESENTATION AND LETTER TYPEFACE
CHARACTERISTICS

By

Nancy M. Wagner

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Psychoiogy

1981

\eofi : U

ABSTRACT

A DEVELOPMENTAL ANALYSIS OF HAND AND VISUAL HEMIFIELD
DIFFERENCES FOR LETTER IDENTIFICATION: EFFECTS OF
MODALITY 0F PRESENTATION AND LETTER TYPEFACE
CHARACTERISTICS

By
Nancy M. Nagner

This study investigated how relative participation of the two
cerebral hemispheres in letter recognition varied with typeface com-
plexity, presentation modality (visual or tactual), and age. Bryden
and Allard (l976) suggested that complex typefaces require preliminary
analysis of spatial configuration by the right hemisphere. Also, more
spatial integration is needed with tactual vs. visual presentation.
Thus, it was hypothesized that right hemisphere participation (rela-
tive performance for left visual field and left hand) would increase
with: (1) increasing typeface complexity, and (2) tactual vs. visual
presentation.

Subjects were 48 fourth- and ninth-graders and college under-
graduates--l6 of each grade. All were right-handed males. They orally
identified single capital letters in eight typefaces of varying com-
plexity (defined by ratings on scriptlikeness, confusability, and
difficulty). In the visual condition, letters were tachistosc0pically

projected about 2 degrees to either side of a central fixation digit.

Nancy H. Wagner

In the tactual condition, subjects felt raised letters with two fingers
of either hand. Presentation time was adjusted independently for each
visual field (hand) to maintain 50% accuracy on each side. The depen-
dent measure was the difference between median presentation times for
the two sides for each typeface.

With visual presentation, Hypothesis 1 was supported for all
grades. A right field advantage was found for the simplest typeface
and a left field advantage for two complex typefaces. However, overall
degree of right hemisphere participation was significantly greater for
undergraduates than for fourth- and ninth-graders. No hand differ-
ence was found for any tactually-presented typeface, thereby failing
to support Hypothesis 2.

Another grade difference appeared in regression analyses relating
complexity characteristics to average visual field differences for the
typefaces. Scriptlikeness and confusability were predictors of field
differences for undergraduates, but no combination of predictors was
significant for fourth- and ninth-graders. In children, then, the
balance of processing was shifted more toward left—hemisphere analy-
sis and responded to different stimulus characteristics.

These findings suggest that both for children and adults the
processes used to recognize visually-presented complex letters are
different from those used for printlike letters. Tactually-presented
letters might require processing in both hemispheres regardless of

complexity.

Nancy M. Wagner

REFERENCE

Bryden, M. P., & Allard, F. Visual hemifield differences depend on
typeface. Brain and Language, l976, g, 19l-200.

 

ACKNOWLEDGMENTS

My sincere thanks to the following people:

Dr. Lauren Harris, my committee Chairperson, for his stimulating
ideas and his constant support and friendship.

Dr. Les Hyman, who served as substitute chairperson during the
year I collected data, for his ideas and his help with the problems
that arose.

Dr. Neal Schmitt and Dr. John McKinney, my committee members,
for all their help and encouragement.

Steve Schacht, a student in methodology at the University of
Chicago, for his valuable advice on the data analysis.

My wonderful family, for their love and support. Without them

this dissertation would not have been possible.

ii

TABLE OF CONTENTS

LIST OF TABLES .
LIST OF FIGURES
LIST OF APPENDICES

INTRODUCTION

Background .
Description of the Main Braille Learning Study
Discussion of the Results of the Braille Learning
Study . .
Main Question to be Addressed in This Dissertation
Presentation Modality . . .
Modality versus Non-modality Approach to Form
Perception . . .
Relationship of Modality Differences and Hemisphere
Specialization . . . . . . . . .
Stimulus Characteristics . .
Effects of Letter and Typeface Characteristics on the
Letter Identification Process . . .
Relationship of Stimulus Characteristics and Hemi-
sphere Specialization
Summary of Hypotheses
Modality of Presentation .
Stimulus Characteristics .
Grade

METHOD: PART I--TYPEFACE SCALING

Scaling for "Difficulty"
Subjects .
Apparatus
Materials .
Procedure.

Scaling for Scriptlikeness, Confusability, Unfamiliar-
ity, and Superfluity . . . . . . .
Subjects .

1'11

Page

vi

ix

Materials .
Procedure .

METHOD: PART II--MEASUREMENT 0F FIELD DIFFERENCES .

Visual and Tactual Conditions .
Subjects .
Apparatus
Materials .
Procedure . . .
Computation of Laterality Coefficients and Their
Stability/Effectiveness

RESULTS: PART I--SCALE VALUES AND SELECTION OF TYPEFACES

Ratings, Dispersion Measures, and Reliability
Initial Group of Typefaces . . .
Selected Typefaces . .

RESULTS--PART II--FACTORS AFFECTING LATERALITY COEFFICIENTS
IN THE VISUAL AND TACTUAL MODALITIES . .

Main Analysis: Effects of General Typeface Factor,
Grade, and Modality Order . . . . . .
Three Repeated-Measures ANOVAs . . . . . . .

Specific Typeface Characteristics: Average Threshold,
Scriptlikeness, Confusability, and Difficulty .
Average Threshold . .

Correlations of Visual LCs, ATs, and Part I
Characteristics .

Prediction of Visual LCs from Typeface Character-
istics Using Multiple Regression

Stimulus and Procedural Factors: Typeface Position,
Letter Order, and Specific Letters . . .
Typeface Position
Letter Order .

Specific Letters . . . . . . . . .
Individual Difference Factors: Individual LC Distribu-
tion and Relation to AT, LC in Other Modality, and

Familial Sinistrality. .

LC Distribution . . .

Relationship of Individual LCs and Combined Thres-
holds, Corresponding LCs in Other Modality, and

FS . . . . .

iv

Page

94
94
94

107
109

113
117
122
122

126
129

137
138

142

 

Page

DISCUSSION . . . . . . . . . . . . . . . . . 146

Presentation Modality . . 146
Reasons for Lack of Asymmetry in the TactuaI Condi-
tion . . . . 147
Modality Differences in Processing . . . . . . 160
Stimulus Characteristics . . . . . l63

Implications for Letter Processing in Adults . . . 166
Implications for Letter Processing in Children . . l75

ENDNOTES . . . . . . . . . . . . . . . . . 182
APPENDICES . . . . . . . . . . . . . . . . . 187
REFERENCES . . . . . . . . . . . . . . . . . 205

Table

10.

11.

12.

LIST OF TABLES

Stability of threshold measure by sensory field, modal-
ity condition, typeface, and grade . . . .

Percent of correct identification after cut-off by
sensory field, typeface, and grade for each modality
condition, and modality order .

Total number of missed fixation points with correct and
incorrect letter identification for each typeface and
visual field for the grades separately and averaged
across grades . . . . . . . . . .

All 22 typefaces

Correlations of five characteristics for all 22 type-
faces and eight selected typefaces . . . .

Selected eight typefaces

Average laterality coefficients, standard deviations,
and significance of field differences according to
grade, typeface, modality condition, and modality
order . . . . . . . . . . . . .

Summary of analysis of variance on laterality coeffi-
cients with modality factor included . .

Summary of analysis of variance on visual laterality
coefficients

Summary of analysis of variance on tactual laterality
coefficients

Average combined thresholds and standard deviations
according to grade, typeface, modality condition, and
modality order . . . . .

Correlations between Part I ratings on difficulty,
scriptlikeness, and confusability and average combined
threshold for each grade--(a) with all typefaces
included, and (b) with T3 eliminated . .

vi

Page

80

81

83
87

90
91

98

100

104

108

111

114

Table Page

13. Correlations of laterality coefficients with Part I
ratings on difficulty, scriptlikeness, and con-
fusability and with average combined threshold
for each grade--(a) with all typefaces included,
and (b) with T3 eliminated . . . . . . . . . 114

14. Regression coefficients and statistics for analysis
including subject and grade variables and for separate
analyses by grade with four predictors of laterality
coefficients . . . . . . . . . . . . . . 118

15. Average laterality coefficients for each letter order
group at each grade for T5 and T8, and relative

percent correct for the first three letters in each
set . . . . . . . . . . . . . . . . . 128

16. For T1, T2, and T7 in each modality condition, indi-
vidual letters listed in order from highest to lowest
average percent correct identifications . . . . . 131

17. Correlations between typefaces in average percent
correct responses for individual letters . . . . . 132

18. Number of subjects per grade with a left or right
field advantage for each typeface and modality
condition, and modality order . . . . . . . . 139

19. Cross-modal correlations of laterality coefficients
for each of the three typefaces presented in both
modality conditions . . . . . . . . . . . . 145

A-1. Summary of analysis of variancecn1visual laterality
coefficients obtained by applying the transformation
ALCI + .5 x (:1), where the sign of the multiplier
is the same as that of the original LC . . . . . 189

8-1. Summary of analysis of variance on visual laterality
coefficients without the modality order variable,
and Newman-Keuls multiple comparison tests for the
main effects of Grade and Typeface . . . . . . . 191

C-1. Summary of analysis of variance on visual combined
thresholds, and Newman-Keuls multiple comparison tests
for the main effects of Grade and Typeface . . . . 193

0-1. Summary of analysis of variance on tactual combined
thresholds, Newman-Keuls multiple comparison tests,
and tests for the simple main effects of Grade for
each typeface . . . . . . . . . . . . . . 195

vii

Table
E.1.

G-1.

H-1.

Tabulation of F-ratios from separate ANOVAs for each
typeface in each modality condition to test the
effects of: (1) Typeface position, and (2) Letter
order, and their interactions with grade . . .

Correlations of individual laterality coefficients
and CTs by grade, typeface, and modality condition

Average laterality coefficients for FS+ and FS-

Fourth- and Ninth-graders for all typefaces in each
modality condition . . . . . . . . .

viii

Page

197

202

204

Figure

F.1.

LIST OF FIGURES

Hypothesized relation between typeface characteristics
and hemisphere specialization for the visual and
tactual modalities .

Average laterality coefficients for the three type-
faces in the visual and tactual conditions for each
grade separately and all grades combined

Average laterality coefficients for the eight typefaces
presented in the visual condition for each grade

Average tactual laterality coefficients for each
typeface according to grade and modality order

Average LCs for each typeface when presented in the
first half of the visual condition (N=24) compared
to the average LCs when presented in the second
half (N=24), with typefaces ordered according to
degree of right hemisphere participation .
Visual condition

Tactual condition .

ix

Page

95

101

105

124
134
199

LIST OF APPENDICES

Appendix

A. Summary of ANOVA on Transformed Visual LCs

B. Summary of ANOVA on Visual LCs with Modality Orders
Combined . . . . . . . . . . . .

C. Summary of ANOVA on Visual CTs

0. Summary of ANOVA on Tactual CTs .

E. Summary of F-Ratios for Typeface Position and
Letter Order . . . .

F. Accuracy Per Hand for Individual Letters in the
Tactual Condition

G. Correlations of Individual LCs and CTs

H. Average LCs for FS+ and FS- Groups .

Page
188

190
192
194

196

198
201
203

INTRODUCTION

Background

 

The ability to recognize letters is generally regarded as a fun-
damental prerequisite for reading (Gibson, Gibson, Pick, & Osser,
1962; Vernon, 1957). Yet, despite the importance of understanding
the process of letter identification (considered here to include all
the separate phases, such as perception and encoding, that might be
involved in the total process) and the great volume of research in
the area, the process is still poorly understood. Even the general
nature of letter identification cannot be agreed upon. Some research-
ers stress feature extraction and analysis as a critical step in the
process (e.g., Gibson, 1969; Estes, 1975), while others find that it
can be explained as a template matching process (e.g., Holbrook, 1975).

In their attempt to better understand letter identification,
researchers have analyzed the process with respect to a wide variety
of factors. One is modality of letter presentation. In addition to
studies of visual and auditory letter presentations, there have been
studies of tactile presentations, including braille, embossed, and
vibrotactile letters (for review, see Kirman, 1973). A second factor
is letter qualities, such as letter typeface (Corcoran & Rouse, 1970)
and symmetry (Fudin, Garcia, & Solomon, 1975). Researchers also have
investigated certain task variables, such as whether subjects are

asked to identify one letter or to categorize two letters as the same

1

or different (e.g., Cohen, 1972), or whether letters are presented
alone or in groups (Estes, 1975). In addition, some studies have
investigated the letter identification process with respect to the
factor of cerebral hemisphere specialization (for review, see White,
1969, 1972) in conjunction with several factors mentioned.

In a broad sense, the purpose of this dissertation is to further
examine the letter identification process by determining whether and
how it is affected by variation of some of these factors alone and in
combination with each other.

In a narrower sense, the purpose is to answer specific questions
that arose from a series of studies of hand differences in braille
letter learning (Wagner, 1976; Harris, Wagner, & Wilkinson, l976a,
l976b; Harris, Wagner, Wilkinson, & Feinberg, l976c). Therefore,
this section on background material for the ideas in this disserta-
tion will consist of a brief description of the main braille learning
study (Wagner, 1976) and the analysis of results that led to the
questions currently being considered.

Description of the Main Braille
Learning Study

 

The purpose of the main braille learning study was to explore
possible cerebral hemisphere specialization for braille analysis.
Assuming stronger contralateral than ipsilateral connections between
hand and cerebral cortex, hemisphere advantage was inferred from hand
advantage in learning braille letters. Braille letters are groups

of one to five raised dots arranged in designated combinations of the

six possible positions in a three by two cell. 0n the one hand, it
was supposed that these letters could be analyzed and recalled as
symbols of verbal material, like Roman letters for sighted individuals.
Consequently, for right-handers, a right hand (left hemisphere)
superiority for braille letter learning would be expected, paralleling
the right visual hemifield (RVF) advantage for recognition of
tachistoscopically projected Roman letters (e.g., Bryden, 1965).

Contrarily, it was supposed that braille letters might be pro-
cessed primarily spatially as complex patterns distinguished from one
another by the number, location, and spatial configuration of their
dots. In this case, a left-hand advantage would be expected, as has
been found for the tactile perception of line orientation (Benton,
Levin, & Varney, 1973; Varney & Benton, 1975) and irregular, presum-
ably non-labelable shapes (Witelson, 1974, 1976). Left-hand superi-
ority also would be consistent with the left visual hemifield (LVF)
superiority found for dot enumeration (Kimura, 1966; Young & Bion,
1979) and dot localization (Kimura, 1969).

To examine hand differences in braille letter discrimination,
sighted, right-handed children and adults learned a different set of
braille-like letters with each hand. Hand testing order and braille
letter set assigned to hand were counterbalanced within each age by
sex group. The letters were presented repeatedly in a semi-random
order. Subjects were told the names of the letters while they first
felt each one, and then had to guess the names on subsequent presen-
tations lasting several second each. Subjects could see neither

their hands nor the braille cards.

As measured by the percentage of correct letter identifications
with each hand, the group scores revealed a significantly better
left-hand than right-hand performance. Although it is this overall
result that is important for the purpose of this background section,
it should be noted that the extent, and even direction, of hand dif-
ference was related to hand testing order and the age and sex of the
subject. These variables will be considered later.

The overall left-hand advantage was consistent with earlier
studies of blind adults and children (e.g., Hermelin & O'Connor, 1971),
sighted adults (Smith, 1929, 1934; Harriman & Castell, 1979), and
sighted older children (Rudel, Denckla, & Spalten, 1974). Hermelin
and O'Connor (1971) suggested an explanation of the hand differences
in terms of cortical specialization. They proposed that braille
configurations are treated as spatial stimuli and thus are analyzed
more efficiently by the right hemisphere before or during verbal
identification and naming by the left hemisphere, resulting in an
advantage for the left hand.

Discussion of the Results of the
Braille Learning Study

 

What is still unclear is why the spatial requirements
should overshadow the linguistic nature of the stimuli in the case of
braille, while the reverse has been found in the case of our Roman
letters presented visually. One reason may be that braille letters
are more difficult to discriminate than Roman ones, inasmuch as the

braille dots require differentiation of minute differences in

orientation and spacing of dots separated by a distance just beyond
the minimum two-point threshold.

The possibility that the superiority of right hemisphere pro-
cessing for braille learning stems from the greater difficulty of
discriminating braille compared to Roman letters is related to the
findings of Bryden and Allard (1976). They showed college students
tachistosc0pically-projected single letters in ten different type-
faces. Subjects had to orally identify the letter shown. The result
was a typeface by visual field interaction. The majority of type-
faces showed a RVF (left hemisphere) advantage, but two typefaces,
which were more "scriptlike" and more difficult to recognize than the
others, showed a LVF (right hemisphere) advantage. Bryden and Allard
explained these results in terms of hemisphere processing differences.
They supposed that the right hemisphere is more efficient at certain
global preprocessing Operations carried out prior to letter naming,
while the left hemiSphere is better at the more analytical identifi-
cation and naming stages. The more scriptlike typefaces probably
required more initial preprocessing “to normalize the stimulus and
to focus attention on the relevant characteristics of the target"

(p. 198). The greater global preprocessing capacity of the right
hemisphere therefore became critical, leading to a right hemisphere
advantage. Presumably, then, braille letters, like the more script-
like Roman alphabet letters, could require much "preprocessing" by
the right hemisphere prior to letter identification and naming by the

left hemisphere.

According to this account, right hand (left hemisphere) superi-
ority would be expected for the tactual discrimination of Roman
alphabet letters printed in a simple block form. However, studies
of tactual discrimination of Roman letters have suggested greater
right hemisphere involvement than generally has been found in visual
presentation studies. 0f eight studies reviewed, two suggested a
left hand (right hemisphere) advantage (Gardner, 1942; Cioffi &
Kandel, 1979, for boys), two found a marginal right-hand (left
hemisphere) advantage (Witelson, 1977a, for boys; Oscar-Berman,
Rehbein, Porfert, & Goodglass, 1978), and four found no hand differ-
ence (Witelson, 1974; LaBreche, Manning, Goble, & Markman, 1977;
Klein & Rosenfield, 1980; Manning, 1980).

From these findings another possibility is suggested-~that left
hand superiority on the braille task is related to the special nature
of the tactual modality. Perhaps the way most tactile information
is detected or processed necessitates a greater degree of integration
and, therefore, greater right hemisphere involvement than the same
information presented visually.

Main Question to be Addressed
in This Dissertation

The basic hypothesis of this dissertation is that hemisphere
Specialization for letter identification (as reflected in hand and
visual field asymmetries) will be influenced systematically by both
stimulus characteristics (such as discrimination difficulty and

scriptlikeness) and presentation modality. When letters varying in

general complexity are presented both visually and tactually, it is
expected that left hemisphere advantage (right hand or RVF advantage),
if any, will decrease and right hemisphere advantage (left hand or LVF
advantage) will gradually appear as the degree of complexity is
increased (see Figure 1). However, the point along the continuum

of complexity at which a left hemisphere advantage will change to a
right hemisphere advantage is expected to differ for the two modali-
ties because of their different natures. Left-hand advantage is
expected to begin at a more moderate degree of complexity than that
required for a LVF advantage.

One main question to be addressed in this dissertation is whether
age of subject and concomitant variation in familiarity with cursive
and manuscript letters will modify the hypothesized effects of modal-
ity and letter typeface on hemisphere specialization for letter
stimuli. Students about to enter fourth and ninth grades and college
students will be tested. The fourth-graders should have just begun
to learn cursive writing within the past two years. Therefore,
although the fourth-graders should be able to recognize cursive
letters, they will be relatively inexperienced with them. Ninth-
graders will represent an intermediate degree of familiarity with
cursive letters and with linguistic stimuli in general, whereas
college students should be highly practiced with this type of mate-
rial.

In summary, the general purpose of this dissertation is to clarify

further the process of letter identification. Specifically, it is

 

Right (R)
sensory
field (F)
adv.

LF=RF

Left (L)
sensory
field (F)
adv.

AVERAGE LATERALITY COEFFICIENTS

 

 

 

 

Visual I——-l
Tactual 13— —u

 

 

 

 

Simplest

TYPEFACE

Most Complex

Fig. l. Hypothesized relation between typeface characteristics and
hemisphere specialization for the visual and tactual

modalities.

hypothesized that hemisphere specialization for letter identification
depends on the interaction of modality of presentation and letter
typeface characteristics, and that this relationship varies with the
subject's age. The rest of the introduction is divided into two main
sections: first, a discussion of presentation modality; second, a
discussion of stimulus characteristics. Each section will begin with
a consideration of some apparently important aspects of the factor,
followed by evidence for the effects of these aspects on hemisphere
specialization for letter identification. Finally, specific predic-

tions about the interaction of the two factors will be summarized.

Presentation Modality

 

Modality versus Non-modality
Approach to Form Perception

 

 

The main question about modality to be addressed in this disser-
tation is whether it critically affects the accuracy and process of
letter identification. The first step in dealing with this question
will be to address the issue of the influence of modality on form
perception in general. This more global issue has been considered
by Goodnow (1971a). She assumed that an object could be explored
in many different ways, and with different modalities, and that on any
two inspections a person might have difficulty deciding whether he
was dealing with one and the same object (Goodnow, 1971a, p. 4).
Goodnow supposed, however, that the difficulty could not be attributed
to presentation of the object in different perceptual modalities.

Instead, she suggested an approach emphasizing differences in

10

exploratory behavior, or in memory or encoding processes, that would

cut across modalities.

Comparison of scanning patterns. This non-modality approach is

 

reflected in other research that has stressed the nature of scanning
and exploratory processes, some of which Goodnow cited. Vurpillot
(1976), Gibson (1966), Piaget (1956), and many of the Soviet
researchers in the area of learning and perception in children (see
review by Pick, 1963; Zaporozhets, 1965) have related children's
increasingly accurate perceptual judgments to increasingly thorough
exploratory movements of eye and hand. In general, for Vurpillot,
and for the Soviet researchers (Zaporozhets, l965),thorough explora-
tory movements enable the observertxiform an accurate model, or image,
of the object being explored. From Gibson's point of view, the
emphasis should be more on the importance of the movements for facili-
tating the abstraction of critical stimulus features. Piaget (1956,
1961) emphasized the importance of the movements for creating "decen-
tration" or more uniform distribution of attention across the object.
Despite slightly differentideasmabout the function of the explora-
tory movements, all of these researchers have noted the similarity

of the development of these movements in the tactual and visual
modalities. In both modalities, development proceeds through stages
in which the movements become progressively more systematic and more
consistent with the stimulus outline. Although the emergence of sys-
tematic movements occurs earlier in the visual modality, movements in

both perceptual modalities reach a fairly mature stage by at least

11

nine years of age (Vurpillot, 1976; Zaporozhets, 1965). This work
therefore emphasizes fundamental similarities in the character of the
two modalities, thereby supporting the non-modality approach advo-

cated by Goodnow.

Comparison of relative saliency of different properties. Other
evidence Goodnow related to the non-modality approach is not so
strongly suggestive of inter-modal similarities, and might even be
reinterpreted to support a strict modality approach. One type of
evidence consists of the comparison of stimulus characteristics most
salient to the hand and to the eye. To make this comparison, Goodnow
(1969) tested children on a match-to-sample task in which the stimuli
were examined either tactually or visually. On each trial, the child
was given a standard and two comparison stimuli, which would vary
from the standard in size, orientation, curvature of line, or number
of lines. In general, a stimulus that varied in curvature was most
often judged different from the standard when the stimuli were looked
at, while a stimulus that varied in orientation resulted in the strong-
est sense of difference from the standard when the stimuli were felt.

According to Goodnow (1971a), part of the difficulty in judging
whether an object that is felt is the same as one that is seen thus
can be explained in terms of differential sampling of stimulus proper-
ties rather than in terms of modalities per se. Further support for
this explanation comes from studies (Pick & Pick, 1966; Gibson et al.,
1962) in which children were presented with letter-like standard forms

followed by a series of transformations of these forms and at least

12

one duplicate of the standard. The children then were asked to select
the forms that were the same as the standard. With visual presenta-
tion, Gibson et al. found that the line to curve transformations and
orientation transformations were of about equal difficulty. In con-
trast, Pick and Pick found orientation transformations to be very

easy tactual discriminations, compared with line to curve and other
transformations. Therefore, whether the measure is disciminability

or preferred basis of similarity, orientation seems to be relatively
more salient for the tactual modality, while curvature is relatively
more salient for the visual modality.

The question that follows from Goodnow's explanation in terms of
sampling method is, which factor--modality or sampling method--should
be considered more fundamental? If it is sampling method, this implies
that a given modality could be easily trained in the stimulus sampling
procedures characteristic of another modality. In this case, Good-
now's use of the evidence pertaining to the relative salience of
stimulus properties to support the non-modality approach seems justi-
fied. However, if sampling method is inherent in or a natural product
of the special nature of the modality, then the critical factor is
still modality. The difference in sampling procedure might even be
considered to enhance the difference between modalities. In any case,
it seems just as reasonable to suppose that the nature of the modality
gives rise to its typical sampling method as to suppose that sampling

differences override fundamental modality differences.

13

Comparison of recall accuracy. A second type of evidence used

 

by Goodnow to support the non-modality approach involves research on
the role of memory in cross-modal matching. To test for the effects
of memory demand, Goodnow (1971b) likened the effects of increasing
the time delay between presentation of standard and comparison
objects to increasing the number of comparison objects. Both manipu-
lations would make it harder to remember the object presented first.
Goodnow presented adults with one, three, or five comparison objects
after a standard first inspected by eye or hand. In general, errors
increased with the number of comparison objects when inspection of the
standard was by hand, but errors remained constant when inspection

of the standard was visual. These results are consistent with those
of a previous study (Posner, 1967) which found that memory for kines-
thetically perceived distances decayed more during an unfilled time
interval than did memory for visually perceived distances. A greater
decrease in recall for conditions involving a tactual component com-
pared to those with only visual components also has been found for
children by increasing the number of comparison items (Davidson,
Cambardella, Stenerson, & Carney, 1974a), and for adults by increas-
ing the retention intervals (Abravanel, 1973).

The general conclusion from these findings is that increased
memory demand had a different effect on stimulus recall, depending on
whether the stimulus was perceived by hand or by eye. This conclusion
assumes that information acquisition was equal in the two modalities

or, at least, that the accuracy of information acquisition did not

14

affect the retention curves. Again, Goodnow (1971a) rejected an
explanation in terms of intrinsic modality differences and, instead,
attributed the weakness of memory for tactually perceived stimuli to
the inexperience of normally sighted subjects at gathering informa-
tion by that means. She cited evidence from work with blind subjects
to support her point of view. Subsequent research on tactual scanning
strategies (Davidson, Barnes, & Mullen, 1974b; see Davidson, 1976,
for review) has reinforced Goodnow's view by showing that individuals
with extensive tactual experience (e.g., blind people) are not so
seriously affected by memory demand as less experienced persons, at
least partly because of the experienced individuals' more efficient
ways of exploring the stimulus.

While this research suggests that memory for tactually per-
ceived stimuli can be improved through experience, this mainly holds
for extensive long-term experience, like that received by a congeni-
tally blind individual. There is, moreover, no evidence that the
level of retention of tactual information achieved is equivalent to
the retention of the analogous visual information by sighted persons.
Thus, for each modality, there might be a range of retention accuracy
according to the degree of experience at information gathering through
that channel. These ranges might overlap slightly at the extremes,

but the fundamental memory difference between modalities will still

apply.

Comparison of memory encoding_processes. In addition to con-

sidering recall accuracy, it is important to consider whether the

15

actual encoding process differs for stimuli presented in the two
modalities. Millar (1975) has argued that both blind and sighted
children show both tactual and verbal encoding of tactually pre-
sented common objects. It also has been shown that adults possess
a short-term tactual storage for non-verbal materials (Bliss, Crane,
Mansfield, & Townsend, 1966). This short-term tactual store pre-
sumably means that the characteristic features of the way an object
feels can be stored in memory beyond brief iconic persistence and
can be used to aid recall even a minute or two after presentation.
Thus, tactual information apparently can survive perception and can
affect recall, just as visual information can (Millar, 1972).

While both modalities have their own characteristic memory store,
the storage of tactually perceived spatial or form information seems
to be more flexible than the storage of visual information, and, in
many cases, can even involve visual encoding (Attneave & Benson,
1969; Freides, 1974; Pick, 1970). For example, visual imagery seems
to play a large role in cross-modal matching studies, even in the
intra-modal (tactual-tactual matching) control condition (Cairns &
C011, 1977; Jackson, 1973). Researchers who have hypothesized visual
storage of tactual information have generally stressed the quali-
tatively different facilities for data handling possessed by differ-
ent modalities and the tendency for sensory information to be trans-
ferred to the modality most adept at processing and storing it.
Therefore, the range and type of strategies available for memory

storage appear to be different for the two modalities, even if the

16

modality difference in recall accuracy does diminish with extensive

practice at gathering information tactually.

Comparison of the need for spatial integration in the visual

and tactual modalities. After considering some other possible inter-

 

pretations of Goodnow's arguments for a non-modality approach to form
perception, it seems that care must be taken not to underemphasize the
importance of modality. Though it is possible to treat stimulus
sampling method and practice at encoding a stimulus in memory as
alternatives to modality explanations and as reasons for modality
differences, these factors also can be dealt with as consequences of

a difference in the basic natures of the visual and tactual modali-
ties. A comparison of these basic natures must be based in large part
on speculations and observations, because empirical questions about
this problem are difficult to formulate.

The main idea of many of the theorists who have compared the two
modalities is that tactual perception necessitates more integration
from separate points in time and space. For instance, Revesz (1950)
has stated that tactual perceptions are more sums of separate parts
than are their visual equivalents. Similarly, Vurpillot (1976)
observed that "the visual receptor system permits the simultaneous
experience of multiple data, while the tactile system only provides
limited data in succession" (p. 273). Even though Gibson (1962)
emphasized the similarity between visual and tactual information
gathering, he also observed that visual perception is based on "the

figure-ground phenomenon, the simultaneous registering of the whole

17

contour, whereas the unity of the tactual perception has to be based
on either cutaneously separate impressions or on successive impres-
sions" (p. 488).

Another way of viewing this extra demand on the tactual modality
is to consider that the tactual modality may be especially suited for
the analysis of spatiotemporal displays. These spatiotemporal dis-
plays require the integration of points separated in time and space.
Transforming a spatialcfisplayinto a spatiotemporal one by the addi-
tion of movement generally impedes perception of qualities by the
eye, yet facilitates perception by hand (Kirman, 1973). One example
mentioned by Kirman (1973, p. 66) deals with the perception of moving
letters by hand and eye. Visually, the moving print is experienced
as a blur at relatively low rates of movement. Yet such movement is
exactly what is needed to aid the hand in reading braille or embossed
Roman letters.

Kirman's view is consistent with Kasajima's (1974) comparison
of the function of pauses and movements in visual and braille read-
ing. According to Kasajima, “movements of the eye are so rapid that
no perception of words or letters occurs during them; but in braille
reading, perception occurs only during movement" (p. 54). Thus, the
integration of successively perceived points is both a requirement

and a special ability of the tactual modality.

Comparison of informationegathering tempo in the visual and

tactual modalities. In general, the hand is slower than the eye at

 

gathering information. The limited tempo of tactual

18

information gathering in combination with the serial nature of tactual
exploration determine to a large degree the significantly poorer
performance in the tactual-tactual matching condition compared to the
visual-visual matching condition in most cross-modal studies of form
perception (e.g., Abravanel, 1973, with adults; Jackson, 1973;

Cronin, 1973; Davidson et al., 1974a). In such studies, there have
been attempts to control for the slower information gathering capa-
bilities of the tactual modality by using a longer presentation period
for the tactual stimuli than for the visual stimuli. Davidson,
Abbott, and Gershenfield (1974c) found that the accuracy of form
matching in the visual and tactual intra-modal conditions did not
differ significantly with l6-sec. exposure times for tactual stimuli
against four-sec. times for visual, although the tactual scores were
still lower.

Other attempts have not been successful and have found a signifi-
cant difference between intramodal conditions even when the tactual
stimuli were explored for 30 sec. compared to five sec. for the
visual stimuli (Cairns & C011, 1977; Butter & Bjorklund, 1973). This
difference became non-significant in one of the studies (Butter &
Bjorklund, 1973) only when visual exploration time was reduced to
two sec. Jackson (1973) attempted to control for the modality dif-
ference in information gathering tempo and in the serial nature of
exploration by having children in the visual condition follow a dot
of light as it moved around the contour of the form. Errors were

still significantly greater in the tactual intramodal condition. The

19

results of these attempts to control for exploration pace and manner
demonstrate the much longer exploration time needed in the tactual
modality to attain parity in accuracy to that in the visual modality.
The findings also indicate that even with controls, visual performance
surpasses tactual performance, suggesting that other fundamental dif-

ferences between modalities remain.

Comparison of the spatial resolving power of the tactual and visual

modalities. Another possible critical difference between vision and

 

touch with respect to the question at hand is in their spatial resolv-
ing power. Evidence for the poorer spatial resolving power of the skin
has come mainly from studies of simultaneous masking (see Kirman, 1973,
for review). For instance, Loomis and Apkarian-Stielau (l976) looked
at the effects of simultaneous lateral masking on tactile and blurred
visual letter recognition. They attempted to control for the differ-
ent spatial resolving powers of the skin and eye by changing the spac-
ing and size of their visual and vibrotactile arrays. Even with this
control, tactile recognition was worse in all masking conditions. Also,
when the accuracy scores for each letter perceived visually were corre-
lated with the accuracy scores for letters perceived by touch, the
correlation was only .68 for the no-mask condition. Thus, the spatial
resolving power of the skin is poorer than that of the eye, but even
when this factor is controlled, other fundamental modality differences
affect the accuracy and quality of performance.

In conclusion, in some instances, especially when developmental

progression is of main concern, the nature and function of exploratory

20

processes are similar in the visual and tactual modalities, and, in
many ways, are more critical than modality considerations. However,
the importance of considering the sensory modality in which a stimulus
is perceived should not be overlooked. Differences between the visual
and tactual modalities in relative saliency of various stimulus dimen-
sions and in memory strength and encoding processes can be thought of
as reflections of a fundamental difference in the natures of the two
modalities. These natures are determined by such factors as spatial
resolving power, information gathering tempo, and the need for spatial
and temporal integration. While most previous studies of modality
differences have been concerned with form perception in general, there
is limited evidence to suggest that letter identification may also
vary systematically with modality of presentation.

Relationship of Modality Differences
and Hemisphere Specialization

 

 

Assuming fundamental differences between the visual and tactual
modalities, the next step in developing the hypotheses of this dis-
sertation is to examine whether these differences are related to
hemisphere specialization. First, consideration will be given to
possible reasons for expecting an association between hemisphere and
modality on the basis of previously discussed modality differences.
Then, evidence from past research on visual field asymmetries will
be compared to evidence pertaining to hand differences on similar
tasks. If there is a modality difference in direction or degree

of asymmetry, then a relationship between modality and cerebral

21

hemisphere might be expected. For the purposes of this dissertation,
the emphasis will be on letter and word identification tasks.
Finally, several studies that have examined the correlations between
left-right asymmetry on a dichotic listening task and a visual field
test will be considered. While these studies do not directly com-
pare tactile and visual asymmetry, they do address the general issue

of the effect of modality of presentation on hemisphere specialization.

Expectations based ongpreviously discussed modality differences.

 

One modality difference previously discussed is the difference in
relative saliency of curvature (greater for vision) and orientation
(greater for touch). An association of modality with cerebral hemi-
sphere would be indicated if it could be shown that each hemisphere

is specialized for the judgment of one of these characteristics, or
if hemisphere specialization is found for one of the characteristics
or one of the presentation modalities, but not the other. In general,
for both visual and tactual presentations, the right hemisphere seems
to be specialized for the analysis of line orientation, as long as
the difference between lines is small (Umilta, Rizzolatti, Marzi,
Zamboni, Franzini, Camarada, & Berlucchi, 1974, for vision; Varney &
Benton, 1975; Benton, Levin, & Varney, 1973; Benton, Varney, & Hamsher,
1978, for touch). With a difference between lines of 45 degress or
more, a right visual field (RVF), or left hemisphere, advantage has
been found (Umilta et al., 1974; White, 1971). However, it is possi-
ble that this left hemisphere advantage indicates only that the

easier discriminations are more available to linguistic analysis.

22

A tendency toward right hemisphere specialization for the judg-
ment of arc curvature also has been indicated (Longden, Ellis, &
Iversen, 1976, for vision; Nebes, 1971a, for touch). Longden et al.
found a left visual field advantage (right hemisphere) in reaction
time for judgments of whether two arcs differed in curvature. Using
commissurotomized patients, Nebes showed that the right hemisphere was
superior to the left in matching tactually perceived arcs to their
corresponding circles. With normal subjects, a right hemisphere
advantage on a modified arc-circle matching test was found by Fagan-
Dubin (1978), although no hand difference has been observed by
others (Nebes, 1971b; Kutas, McCarthy, & Donchin, 1975). If the per-
ception of the orientation and curvature of forms can be related to
the perception of simple line orientation and arc curvature, then
the modality difference in relative saliency of these two character-
istics does not seem to be a reason for expecting an association
between modality and cerebral hemisphere.

A second modality difference is that for sighted subjects, the
image of a tactually perceived object, compared to the image of a
visually perceived object, is more susceptible to temporal decay and
more likely to be stored at least partly in another modality (vision).
There is no clear evidence that memory strength is related to cerebral
hemisphere. A LVF superiority for the recognition of complex, random
shapes has been found to increase with the length of recall interval,
becoming significant at the 10 and 20 sec. intervals (Dee & Fontenot,

1973). This finding might suggest an association of greater memory

23

strength with the right hemisphere, but it is more likely that short-
term memory demands merely enhance the existing factors (perceptual
and others) affecting hemisphere specialization. Then, greater memory
demands would also be expected to increase RVF superiority for verbal
stimuli. Rosen, Curcio, MacKavey, and Hebert (1975) investigated

this hypothesis by presenting subjects with bilateral columns of four
letters and asking them to report only one column in each trial.

Based on their finding that the largest visual field asymmetry
occurred for letters in the fourth position, they concluded that
short-term memory demands contributed to the observed asymmetry.

There is also no direct evidence available to associate with
either hemisphere the transfer of tactual information to a visual
image. However, the left parietal region has been associated with
tactile-visual matching (Butters, Barton, & Brody, 1970), which might
suggest a left hemisphere advantage for the transfer from a tactile to
visual image. Overall, evidence about the association of memory
factors with hemisphere differences is weak and requires many infer—
ences. No general association of modality and cerebral hemisphere is
apparent.

A third difference between the hand and the eye is concerned with
the degree to which information from successively perceived points
must be integrated during stimulus exploration. Since the hand must
integrate information from separate points to a much greater degree
than the eye, it might be expected that the tactual modality would be

associated with the serial processing attributed to the left hemi-

sphere (Cohen, 1973).

24

However, more critically, it seems that the tactual integration
of separate points requires active construction and manipulation of
increasingly more complete images. This active process of construc-
tion involves thinking in a "spatial" way, which has been categor-
ized as a right hemisphere process (e.g., Kimura, 1973). The
integration of separate points required by the tactual modality also
seems to depend on what Broadbent (1974) has described as "a stored
representationcnithe world that retains its features in parallel for
quite long periods of time" (p. 40). In other words, recently per-
ceived stimulus points can be integrated with less recently perceived
points only if the previously gathered information is retained. These
processes that involve sustaining the continuing representations of
the environment are characterized by Broadbent (1974, p. 40) as right
hemisphere processes. Contrarily, left hemisphere processes would
involve categorizing changes in the environment. Thus, intuitive
consideration of the fundamental nature of the modalities might lead
to an association between right hemisphere processing and the tactual

modality.

Expectations based on past research: previous studies of
visual field asymmetry. In general, past studies of visual field
asymmetry have shown a RVF advantage for the recognition of verbal
stimuli (see reviews by White, 1969, 1972). A RVF superiority in
adults has been found for: (l) recognition of single letters pre-
sented along the:horizontal meridian(Bryden, 1965, 1966, 1973;
Bryden & Rainey, 1963; Worall & Coles, 1976; Fennell, Bowers, &

25

Satz, 1977; Schmuller, 1979) and, especially, the encoding phase of
letter identification (Cohen, 1976); (2) time taken to judge whether
two letters have the same name (Cohen, 1972; Geffen, Bradshaw, &
Nettleton, 1972); (3) recognition of single letters presented in
groups (Bryden, 1966; Kimura, 1966); and (4) word recognition
(Fontenot, 1973).

These studies that have found a RVF advantage typically have
employed a successive unilateral presentation technique with binocu-
lar viewing and a verbal identification response. The two major
problems with this technique are the possible confounding of the
hemisphere specialization effect with trace scanning tendencies and
with response factors (White, 1969, 1972). However, since at least
some of the studies reporting a RVF advantage used methods or stimuli
that minimized the effects of these possible confounding factors, we
can conclude that a hemisphere difference in processing is at least
one important reason for the RVF advantage. First, trace scanning
refers to the scanning of the iconic image that persists briefly
following stimulus presentation. While the general tendency to scan
the post-exposural stimulus trace in a left to right direction would
lead to a RVF superiority for unilaterally presented stimuli, this
trace scanning should not have been a critical factor when only
single letters were presented. Second, although the RVF advantage in
some of the studies might have been partly the result of left hemis-
phere control of the identification response, this should not have

been the case in other stUdies,especially those requiring a rapid

26

"same” or "different" response. In summary, the RVF advantage for
verbal stimuli can be attributed largely to an asymmetry in hemi-

spheric processing.

Effects of age on visual field asymmetry for the regggnition
of verbal stimuli. While a RVF advantage for the recognition of
verbal stimuli generally has been found for seven- to l3-year-old
children who are normal readers, the findings are not as clear as
those for adults. Some developmental studies of visual field asymme-
try for word identification have found a significant RVF advantage for
the one age group tested (McKeever & Huling, 1970; Marcel, Katz, &
Smith, 1974; Marcel & Rajan, 1975) or a constant RVF advantage across
all ages tested (Olson, 1973). Other similar studies have found a
significant RVF advantage for children older than 10 or 12 years of
age, but not for the youngest age groups tested (Forgays, 1953;
Miller & Turner, 1973; Turner & Miller, 1975, only under certain
conditions of word length and post-exposure field; Reitsma, 1975,
cited by Witelson, 1977b; Carmon, Nachshon, & Starinsky, 1976;
Tomlinson-Keasey, Kelly, & Burton, 1978, for same-different judgments
of sequentially presented words).

Since all these studies involved unilateral presentation of
horizontally arranged words, one question is whether the effects are
due to hemisphere specialization or to left to right scanning of the
post-exposural stimulus trace. This question was addressed in two
studies that included conditions for which the two factors would

predict opposite results. Olson (1973) used a bilateral presentation

27

condition. According to trace scanning theory, scanning should begin
at the left most point of the image, and thus the left field word
should be reported more accurately. Carmon et a1. (1976) used
Hebrew words, which are scanned right to left and therefore should
elicit a LVF advantage when presented unilaterally. Since a RVF
(left hemisphere) advantage was found in both instances,hemisphere
specialization is most likely the dominant factor.

It should be cautioned, however, that while the weight of the

evidence supports the conclusion of left hemisphere specialization

in children for visually presented linguistic stimuli, not all studies
are in agreement. Yeni-Komshian, Isenberg, and Goldberg (1975), for
example, found no field asymmetry in 10- to 13-year olds for verti-
cally arranged digit names. In addition, following the predictions
of trace scanning theory, Daves and Werzberger (1971) found a LVF
advantage for children in grades one to seven for the identification
of strings of six letters presented bilaterally.

A second question about the developmental results is whether
the pattern of asymmetry changes significantly with age. Of the
studies of word identification cited above, over half failed to find
a RVF advantage for the youngest age groups tested. While this age
trend should be considered tentative because of the lack of controls
for changing task difficulty across age groups, further support may
be found in several studies of asymmetry for letter discrimination.
Reynolds and Jeeves (1978) measured choice reaction time to single

lateralized letters. There were four possible letters. Subjects

28

were instructed to press one button if either of two specified
letters appeared and to press the other button if either of the other
two letters appeared. A RVF advantage was found for 13- and l4-year—
olds and adults, but not for seven- and eight-year-olds. Using a
similar task with lateralized letter pairs as stimuli, Broman (1978)
found a slight RVF advantage for 13-year-olds and adults, a slight
LVF advantage for lO-year-olds, and a significant LVF advantage for
seven-year-olds. An identical age pattern was found by Carmon et al.
(1976) for the identification of single lateralized Hebrew letters.
There was no significant field asymmetry at any age except for a LVF
advantage for seven-year-olds.

Explanations for this increasing RVF advantage with age focused
on the ideas of either increasing specialization of the left hemi-
sphere for processing the printed word (Tomlinson-Keasey et al.,
1978) or, as is more likely the case, a change in strategy with age.
For instance, Carmon et a1. (1976) suggested that sequential (left
hemisphere) processing of verbal materials becomes more accentuated
with age. Similarly Broman (1978) argued that the "configurational?
approach may be the more primitive and the "naming" approach the more
advanced manner of perceiving alphabetical material.

In summary, children, as well as adults, have shown a RVF
(left hemisphere) advantage for the processing of linguistic stimuli.
If there is a change with age in the relative participation of the
two hemispheres in the recognition of printed material, the evidence
suggests that it is in the direction of increasing left hemisphere

participation with increasing age. Therefore, it there is an age

29

difference in asymmetry in this dissertation, it can be predicted
that the effect will be a smaller RVF advantage for fourth-graders

compared to ninth-graders and college students.

Expectations based (n1 past research: previous studies of hand
asymmetry. The eight studies of hand asymmetry for Roman letters
that were reviewed can be grouped according to two procedural vari-
ables: (1) whether the presentation was dichhaptic (different letters
presented simultaneously to the two hands) or unilateral (letters
presented to one hand at a time); and (2) whether exploration was
active (subjects moved their fingertips over the letter) or passive
(subjects held their hands stationary while letters were drawn on
them). Results were inconsistent, even within groups of studies
using similar procedures.

The procedure used by five of the studies involved dichhaptic
presentation with active exploration. Witelson (1974, 1977a & c)
conducted two studies with six- to l4-year-old boys. Her stimuli
were twenty styrofoam upper-case letters, excluding five of the

1 On a

total of nine laterally symmetrical letters of the alphabet.
given trial, two pairs of letters were presented for two seconds
each, and the subject's task was to report orally the names of the
four letters he had just felt. No hand asymmetry was found in the
first study, while a marginally significant (p < .10) right hand
(left hemisphere) advantage was found for boys2 in the second. How-

ever, since the verbal identification response probably necessitated

left hemisphere involvement, one possible interpretation of the

1.. I .2.

li—

30

findings is that right hemisphere superiority in processing was
masked by the left hemisphere's control of the response mode.

To avoid this confounding of response and processing modes,
LaBreche et a1. (1977) used a left hand and a right hand fingerspell-
ing letter identification response as well as a written response.
While hemisphere specialization for fingerspelling has not been
clearly established, the evidence suggests either a bilateral repre-
sentation of manuallinguisth:stimuli or a tendency toward right
hemisphere dependency.3 Therefore, use of the fingerspelling response
might, if anything, tend to enhance any right hemisphere superiority
in processing. Despite this change in response method, LaBreche
et al. (1977) found no hand differences for deaf and hearing adoles-

cents.

Klein and Rosenfield (1980) used still a different, supposedly
"neutral" response method with a similar dichhaptic procedure. They
had third-graders point to their response choices using a "double-
handed" pointer in the shape of a divining rod. Again, no hand dif-
ference was found.

In the fifth related study, Cioffi and Kandel (1979) also
used a pointing response, but presented pairs of two consonants
(consonant bigrams) rather than single letters. They found a left
hand (right hemisphere) advantage for six- to 14-year-old boys. One
question raised by this study, though, is whether letters or short
sequences of letters are adequate linguistic stimuli. This question

was suggested by the results of another condition in which two letter

31

words were presented. A right hand advantage was found for these pre-
sumably more adequate linguistic stimuli. One problem with these
findings is that, if anything, sequences of two letters should be
closer to word stimuli than single letters. Therefore, if right

hand advantage increased as the linguistic characteristics of the
stimulus increased, consonant bigrams would be more likely than single
letters to result in no hand advantage or a trend toward right hand
advantage.

There was only one study that combined a dichhaptic technique
with a passive procedure, where the experimenter controlled delivery
of the stimuli (Oscar-Berman et al., 1978). Following the presenta-
tion of one pair of single capital consonants, the adult subjects'
task was to report the letter names orally in the order then desig-
nated by the experimenter. A marginal (p_< .06) right hand (left
hemisphere) advantage was found, not only for the hand reported
second. According to the authors, asymmetry was enhanced for the
hand reported second because a greater memory demand was involved.

Finally, stimulus presentation was unilateral in two studies,
one with a passive procedure and one with active exploration. Both
studies used strings of letters and an oral identification response,
and both tested adult subjects. Manning (1980) had subjects hold
their hands stationary while the experimenter drew sequences of six
capital letters on the palm. There was no hand asymmetry in the
number of letters remembered.

Using a different exploration technique, Gardner (1942) had

30 students read nonsense syllables with the fingers in both a left

I151 wfli.§

32

to right and a right to left direction. The syllables consisted

of laterally symmetrical letters made of cord stitched upon card-
board. Performance in both reading directions was faster and more
accurate for the left hand than for the right. Summed across the

two reading directions, the time difference between the hands was
about 5.5 seconds (with 67% of the sample faster with the left hand),
and the difference in number of errors was about .7. No statistical
tests were performed, but at the least, a trend toward left hand
superiority for the recognition of Roman letters was demonstrated.

If these findings can be interpreted in terms of cerebral asymmetry,
then even tasks requiring the linguistic processing involved in read—
ing, when presented through the tactual system, seem to require some
degree of spatial, "right-hemisphere" analysis.

In summary, of the eight studies of hand asymmetry for Roman
letter recognition, only four found a significant hand advantage, two
in favor of the right hand and two in favor of the left. One of the
studies that showed a right hand advantage used a dichhaptic, active
presentation technique and the other used a dichhaptic, passive
procedure with a moderate memory demand. However, the three other
studies using a dichhaptic, active procedure failed to find a hand
asymmetry, as did another study using a passive procedure combined
with an even greater memory demand. Also, in both studies that showed
a right hand advantage, the letters were named orally. Possibly,
then, these results were related mainly to a left hemisphere advantage
for the spoken response and not to a hemispheric asymmetry for letter

perception and processing.

33

Although the asymmetry was stronger in the two studies that
found a left hand advantage, there again were some unanswered ques-
tions. No statistical tests were performed in one study, and the
left hand advantage switched to a right hand advantage in the other
study when two-letter words as opposed to consonant bigrams were
used as sitmuli. All in all, the results show that it is difficult
to elicit an asymmetry for the tactual perception of Roman letters,
possibly indicating that both hemispheres contribute importantly to
the letter processing.

The findings of a left hand advantage for reading nonsense
syllables and bigrams are consistent with some of the more recent
reports of hand asymmetry for braille reading and letter recognition
mentioned in the first section. With blind adult subjects, Hermelin
and O'Connor (1971) found a left hand advantage in the accuracy with
which vertical columns of braille letters were read. With blind
children, they found that sentences were read faster and more
accurately with the left hand than with the right. Thus, even when
a tactile alphabet becomes so familiar a linguistic system as to be
almost second-nature, the very fact that it is perceived tactually
might induce a certain degree of right hemisphere analysis.

Research with sighted subjects has supported the conclusion
that this left hand advantage at least partly represents a hemi-
sphere processing difference and not just a bias resulting from dif-
ferent amounts and types of experience with braille reading for the

two hands. As stated in the first section, an overall left hand

34

advantage for braille letter identification also has been found for
sighted subjects. This pattern of left hand advantage, however,
was influenced by the order in which the hands were tested.

The main effect of hand testing order was an enhancement of
left hand performance when it followed the right. Rudel et al.
(1974) tested children with a paired associate learning method simi-
lar to the procedure of the braille learning study described in the
first section (Wagner, 1976). They found that girls did about as
well with their right hand whether it was tested first or second,
but their left hand scores were much better than the right when
tested second, and worse when tested first. To explain this order
effect, Rudel et a1. suggested that prior training with the right
hand helps or "prepares" the left, possibly because right hand use
activates verbal strategies tended to be relied on by young girls
(Harris, 1977).

Wagner (1976) also found that left hand performance was enhanced
when it followed the right. Although scores were generally higher
for the second hand tested, the difference between hand scores was
greater when the left hand was tested second than when the right
hand was second. These findings could be explained in terms of an
interaction of the factors of practice and hemisphere specialization,
but they also might indicate that part of the left hand superiority
depends on prior preparation through left hemisphere strategies.

The latter interpretation is consistent with Kinsbourne's

(1973) discussion of lateral asymmetries in terms of "attentional

35

sets" induced by the nature of the task or response method. When

the task or response method favors the processing in one hemisphere,
that hemisphere is then activated, turning attention toward the oppo-
site side, and, at the same time, inhibiting activation of the other
hemisphere. Thus, in a braille learning task, it might be easier
initially to establish a left-hemisphere, verbal "set," but once the
verbal aspects of the task are mastered, a right-hemisphere, spatial
"set" would become effective. Hence, the right-hand-first testing
order would be the most advantageous.

One way to eliminate the effects of hand testing order is to
present letters to the left and right hands in simple alternation,
with a different group of letters being presented to each hand. This
alternating hand method has been used by Feinberg and by Harris and
Wagner (both studies described in Harris et al., 1976c), and signifi-
cant left hand advantage was found for right-handed adults. Thus,
while the order effect may be important, it is not crucial for the
emergence of a left hand advantage. To obtain clearer results, an

alternating hand method will be proposed for this dissertation.

Effects of age on hand asymmetry for recognition of verbal
stimuli. In general, the age of the subject has not been shown to
influence the pattern of hand asymmetry for the recognition of Roman
letters. In all three studies that included more than one age group,
no significant effect of age on hand asymmetry was found in six- to
l4-year-olds. In addition, both for the two studies that found a
left hand advantage and for the two that found a right hand advantage,

36

the subjects were children in one case and adults in the other.
Therefore, both within and between studies, there was no association
of age and degree or direction of asymmetry. It should be noted that
while sex differences in asymmetry were reported in two instances
(Witelson, 1977a; Cioffi & Kandel, 1979), there was no sex by age
interaction.

Contrarily, the age and sex of subject have been found to be
related to the pattern of hand advantage for braille letter learn-
ing shown by sighted subjects. The general findings were a later
emergence of left hand advantage in females than in males, and a
tendency toward right hand superiority in younger children. In the
develOpmental study of braille letter learning described in the first
section, a definite left hand advantage was found by age nine for
males, but not until college age for females (Wagner, 1976). The
nine-, ll-, and 14-year-old girls tested by Wagner showed no hand
difference. Using a similar paired associate letter learning method
with children seven to 14 years of age, Rudel et a1. (1974) found a
left hand advantage by age 11 for boys, but only by age 13 for girls.
In this case, though, the younger girls showed a right hand advantage.
In addition, Rudel, Denckla, and Hirsch (1977) tested children and
adults on a task requiring same-different judgments of two braille
letters examined sequentially by one hand. Left hand superiority was
found by the same ages for boys and girls as in the previous study,
but a right hand advantage was found for both nine-year-old girls and

boys.

37

These findings could be interpreted to mean that young girls,
and maybe even young boys, processed the braille letters as verbal
stimuli, and they did so even when the letters did not have to be
identified verbally or even have to be known as letters. Thus,
when the task involves both spatial and linguistic components, as is
thought to be the case for braille letter discrimination, the verbal
aspects of the task may be more critical for children below the age

of 11 years.

Expectations based on past research: comparison of previous
studies on visual field and hand asymmetry for verbal material. The
findings of visual field differences in the recognition of verbal
stimuli seem to contrast sharply with the findings pertaining to hand
differences in the recognition of similar material. Among the visual
field studies reviewed by White (1972), none reported a LVF superior-
ity for the recognition of single letters, and nine of 12 studies
reported a RVF superiority. Among the studies of hand differences in
the recognition of Roman and braille letters summarized in the last
section, nearly all reported a left hand superiority or no hand
asymmetry. The question now is whether this different pattern of
results for the two modalities is evidence for a systematic rela-
tionship between modality and hemisphere specialization.

One possible interpretation of the findings is that they do
not warrant the assumption that, given similar stimuli, tactual pre-
sentation is more likely than visual presentation to induce a right

hemisphere advantage. The different patterns of asymmetry for the

38

two modalities instead might be attributed to simple task variables.
For instance, it may be more difficult to show an asymmetry in either
direction in the tactual modality, so that more complex stimuli would
have to be used. This idea is related to LaBreche et a1.'s sugges-
tion that their failure to find a hand difference for the recognition
of Roman letters indicates that "the task demands associated with
each of the letters conditions were not sufficiently complex to tax

the spatial and linguistic resources of each hemisphere" (p. 193).

This lack of sufficient complexity in the tactual stimuli, then,
would explain why four out of eight studies of hand differences,
but only three out of 12 comparable visual field studies, failed to
find an asymmetry for Roman letter recognition. The "lack of com-
plexity" hypothesis might also explain the fragile nature of the
left hand advantage for braille letter learning, reflected in the
fluctuation of results with such variables as age, sex, and hand
testing order, and the fact that a left visual field, but not a
left hand, superiority was found for dot enumeration (Myers, 1976).
While this interpretation in terms of task variables is possible,
there is no apparent explanation as to why tactually presented stimuli
should need to be more complex for an asymmetry to be shown.

Another possible interpretation of the different pattern of
results for the two modalities is that, given similar stimuli, a right
hemisphere advantage would be more likely to occur with a tactual
presentation method than with a visual presentation method. This

hypothesized association between modality and cerebral hemisphere,

39

at least for verbal stimuli, was implicit in Witelson's (1974)
explanation for her failure to find a hand difference in the recogni-
tion of Roman letters. She suggested that "within the tactual sys-
tem, linguistic information is analyzed first in a spatial code and
then translated into a linguistic code, with spatial analysis more
readily processed in the right hemisphere and linguistic analysis in
the left hemisphere" (p. 14). Thus, presentation of letters in the
tactual modality increases the spatial nature of the identification
task, and therefore induces right hemisphere processing. Witelson's
description of letter identification in the tactual system is consis-
tent with Hermelin and O'Connor's (1971) description of braille
letter recognition as a two-stage coding process. It is difficult
to talk about modality differences in the case of braille, however,
since no evidence is available on hemisphere specialization for the
recognition of braille letters presented visually.

An even stronger association between right hemisphere processing
and a tactual presentation method seems to be advocated by Rudel
et a1. (1977). They suggest that one of the main reasons for the
left hand advantage in braille letter recognition is the fact that
naive sighted subjects and blind subjects cannot visualize the braille
letters. Similarly, they argue that much of the evidence for right
hemisphere superiority comes from experiments where vision is excluded.
One bit of evidence from the clinical literature might also support
this idea of a strong association between the tactual modality and

right hemisphere processing. 8011 (1974) administered tests of

40

tactile-perceptual ability, including a form recognition task, to
patients with right or left hemisphere brain damage. The results
showed that the performance of patients with right hemisphere damage
was more impaired on the ipsilateral and contralateral sides of the
body than for patients with left hemisphere damage. These findings
were interpreted as demonstrating the "pre-eminence" of the right hemi-
sphere in subserving tactile perception.

Whether or not this almost exclusive association of the tactual
modality with right hemisphere superiority is warranted is not clear.
However, there does appear to be some tendency for a systematic
increase in the need for right hemisphere analysis when the presen-

tation modality is tactual instead of visual.

Expectations based on cross-modal reliability of laterality,
,tests. A third basis for forming expectations about possible asso-
ciations between modality and cerebral hemisphere is evidence con-
cerning the correlation of left-right asymmetries found in different
modalities. If the asymmetries found in different modalities are
highly correlated, it might be predicted that the cerebral hemispheres
develop specialization for certain functions independent of the
modality in which they are performed. On the other hand, if the
asymmetries in different modalities are not even moderately corre-
lated, or are negatively correlated, a dissociation of laterality
effects in different modalities might be predicted, and a systematic

variation in laterality effects across modalities would be unlikely.

41

No data on the correlation of asymmetries in the tactual and
visual modalities are available, but several studies have looked at
the correlation between asymmetries in the visual and auditory modal-
ities. At least four did not find even a moderate positive relation-
ship between modalities. Bryden (1965) and Zurif and Bryden (1969)
found only weak positive correlations between asymmetries revealed
in a series of dichotic listening and visual field tests. Using a
measure of concordance for direction of laterality effect in a
dichotic listening test and a visual field test for letter recogni-
tion, Bryden (1973) found a relationship the reverse of what he had
expected. Those students who showed better left ear performance were
more likely to show a RVF superiority. Fennell et a1. (1977) also
found negative or weak positive correlations between ear and visual
field scores within and across testing sessions.

In contrast to those studies, at least two studies have suggested
a significant moderate correlation between asymmetries in the two
modalities for right handers. Hines and Satz (1974) improved upon
the method used in some of the previous studies by computing separ—
ate correlations for left and right handers, and by using a visual
field testing method that minimized the influence of direction-of-
reading. They found significant positive correlations between
asymmetries in the two modalities for the right-handed subjects
(average uncorrected r = .37). In addition, they found that about
63% of the right handers showed concordance for direction of asymmetry
in the two modalities. When Fennell et a1. (1977) measured directional

concordance rates, instead of correlations between dichotic listening

#141114

42

and visual field scores, they found a moderate agreement between the
early testing sessions, and a strong (81.4%) agreement between the
third and fourth sessions. In light of these two studies, it seems
that a moderate correlation between asymmetries might be expected,
and therefore, systematic modality differences in asymmetry would be
possible.

One additional issue to be considered when discussing cross-
modal reliability is the type of stimuli presented in the different
modalities. First, it seems unreasonable to expect that hemisphere
specialization for all types of verbal material is a unitary phenome-
non (consider Bryden, 1966). Rather, it seems that during development
the left hemisphere would come to be specialized for increasingly
complex verbal processing "routines,f and that each of these routines
might be lateralized to a different degree. Therefore, since some
of the studies mentioned used different types of stimuli for the
visual and auditory tests (Fennell et al., 1977; Bryden, 1973), the
basis for their expectation of a strong correlation between asymme-
tries in the two modalities is unclear.

Second, even in the cases where the stimuli presented in the two
modalities were similar (Hines & Satz, 1974), there was no indication
that the amount and type of information perceived by each modality
were subjectively equal. In conclusion, then, failure to find a
strong correlation between asymmetries in the visual and auditory
modalities might be partly the result of lack of control over the

objective and subjective similarity of the stimuli presented in the

43

two modalities. The importance of stimulus factors will be considered

in the next section.

Stimulus Characteristics

 

Effects of Letter and Typeface
Characteristics on the Letter
Identification Process

 

 

 

The question of main concern in this section is whether or not
systematic variation of letters or their typefaces along selected
dimensions would be expected to lead to differences in performance on
a letter identification test. Since this dissertation is mainly con-
cerned with hemisphere differences in letter identification, the
three dimensions to be discussed are dimensions that Bryden and Allard
(1976) have shown to be related to visual field differences. These
dimensions are scriptlikeness, difficulty (naming latency), and
internal confusability--the extent to which one item in a typeface
would be confused with other items in the same typeface. Typeface
selection in the current study will be determined by ratings on these
three main characteristics.

In addition, ratings will be obtained on two dimensions to be
mentioned only briefly in this section-~unfamiliarity and superfluity.
Unfamiliarity was chosen because Bryden and Allard found it to be
moderately, though non-significantly, correlated with visual field
differences. Superfluity was chosen because it shared some of the
characterisics of scriptlikeness, but might be more appropriate for
the type of lettering used both in Bryden and Allard's study and in

the current study.

44

Scriptlikeness. Bryden and Allard evidently used the term
"scriptlike" to mean "like handwriting" and that is the way it will
be used in the current study. Evidence that this definition of
scriptlikeness may be associated with differences in the letter
identification process comes from a study by Corcoran and Rouse (1970).
These authors suggested that the perceptual system is organized into
two "subroutines," one for printing (typing) and one for handwriting.
They designed lists of words that were copied in four different ways:
(1) lower-case typed letters (TL); (2) upper-case typed letters (TU);
(3) handwritten by person 1 (H1); and (4) handwritten by person 2
(H2). These words were then assigned to sets that were either unmixed
(all from one list) or mixed (half from one list and half from
another). Subjects were shown tachistosc0pically projected words from
a given set and asked to orally identify them. The data showed that
when the mixed set contained H1 and TL words, the mean probability of
recognizing the words on the first presentation was significantly
lower for the mixed set than for the respective ummixed sets. How-
ever, when the mixed set contained either H1 and H2 words, or TL and
TU words, there was no difference between the mixed and respective
unmixed sets in the mean probability of correct word identification.
Thus, Corcoran and Rouse hypothesized that the lower accuracy for
sets that contained both typed and handwritten words was caused by
the subject's inability to switch to the appropriate processing sub-
routine before the stimulus was presented.

Given this evidence, then, it would be nice to equate the more

scriptlike typefaces used by Bryden and Allard with cursive

45

handwriting. However, it should be noted that, technically and his-
torically, cursive script differs in many respects from the more
formal script letters used by Bryden and Allard. The term "script"
technically means any handwritten letter or character. Formal script
is the polished, calligraphic type of writing system that can be
traced back at least as far as the ideograms and phonetic symbols
used in ancient Sumeria around 3100 B.C. Contrarily, cursive script
can be defined as signs and styles of writing changed in form by
everyday use and quickly executed construction (Fairbank, 1970).

Two styles of writing apparent in ancient Egypt can be related
to the formal and cursive scripts (Fairbank, 1970). One style,
hieratic writing, derived from Egyptian hieroglyphics, which were
characters that represented formalized pictures, and also words,
syllables, and consonants. The hieratic style was used by priests
during the first dynasty (around 2900 B.C.). A second style, demotic
writing, was characterized by a rapid, fluent quality which departed
greatly from the pictorial quality of the hieroglyphs. This demotic
style has not been traced as existing earlier than the seventh cen—
tury B.C., and probably developed because of the need to write with
speed, and consequently to simplify, when writing informally.

This contrast between the formal and cursive styles was also
apparent in Roman times (Delpire & Monory, 1961). The bold, polished
Roman rustic and unical lettering styles developed at the same time
that cursive writing was becoming more popular, possibly because the

pen was able to slide on the surface of papyrus. Delpire and Monory

46

noted that "as beautiful as they were legible, the Roman capitals
gratify the mind and the eye" (p. 74). Then, in an aside, they noted
that "one might assume that capitals in cursive script fatigue the
eye" (p. 77). The contrast between formal and cursive scripts was
also evident in their observation that "while someone who knows Latin
can without much study successfully grapple with rustic or unical
inscriptions, he risks being baffled for a long time, if not for good,
with the miniscule (lower-case) cursive" (p. 77).

In light of these historical and technical observations, if
Bryden and Allard's term "scriptlike" means "like cursive writing,"
the characteristics of the scriptlike typefaces ought to include:

(1) a rapid, fluent QUality; (2) possibly more loops and cirlicues;
(3) extensions of the letters at the beginning and end, as if to be

4 and (4) an irregular, even sloppy, qual-

joined with other letters;
ity reflecting the distinctiveness (Gordon & Mock, 1960) and sometimes
sloppiness of adult hands. The typefaces used by Bryden and Allard
instead are in a precise, formal style. Therefore, variance along a
dimension of cursive handwiritng might not be meaningful.

One suggestion for a related dimension that might be more appro-
priate for Byden and Allard's distinct, precise letters is superfluity,
or the degree to which the letter varies from some impression of the
"ideal" letter. This dimension could be operationally defined in
terms of the number of extra lines, loops, and curlicues added to the

basic form of the letter and in terms of the tilt and thickness of

the line. To assess the importance of this related dimension, the

47

typefaces in the current study will be rated on both scriptlikeness

and superfluity.

Difficulty. The dimension of difficulty also may be important.

 

If this dimension is measured by naming latency (as in Bryden and
Allard'sstudy), it probably would be highly related to the dimension
of deviation from an ideal. The greater the deviation, the more time
it would take to match the perceived letter to the image in memory.
When considering the factors that make a letter difficult to recog-
nize, both the characteristics of the letter itself and the character-
istics of the typeface appear to be significant. Evidence for a marked
variation in the difficulty of perceiving particular letters presented
alone was found by Budohoska, Grabowska, and Jablonowska (1975). They
asked adults to identify printed Roman alphabet letters presented at

a central fixation point for 17 msec. each. The percent of errors
ranged from about 60% for T, I, and L, to less than 10% for D and 0.

A large variation in the difficulty of recognizing the same letter
typed in different scripts also has been shown. Bryden and Allard
found that mean recognition times for their different typefaces
ranged from 671 msec. to 1183 msec. Reasons for the greater diffi-
culty of some typefaces might be simple explanations based on the
clarity of the script or the amount of black and white contrast pro-
vided, or they might be more complex explanations based on factors
such as amount of deviation from an ideal. In addition, it might
be expected that letter and typeface variables would interact with

each other to determine difficulty. Some letters, then, would be more

48

difficult to recognize in certain typefaces than in others. For
instance, if a given letter was changed in one script by the addition
of a loop to the letter's upper part, the most crucial part for
letter recognition (Fairbank, 1970, p. 76), the letter might be
relatively more difficult to recognize in that typeface than in
another where it was changed only slightly or only in the lower part.
Since sensory modality is a primary variable in this disserta-
tion, the possible interaction of modality with these dimensions of
letter and typeface difficulty should also be noted. Moderate corre-
lations between visual and tactual presentations have been found for
recognition accuracies of letters (Loomis & Apkarian-Stielau, 1976;
Craig, 1979) and forms (Birch & Lefford, 1963). Thus it could be
expected that relative letter difficulties would be similar, though
not identical, in the two modalities. Since there is evidence for
a modality difference in the relative ease of line to curve and
orientation transformations, typefaces emphasizing orientation changes
from the basic form should be relatively easier with tactual presenta-
tion, and those emphasizing curvature changes should be relatively
easier with visual presentation. Again, the factors of letter and
typeface difficulty may interact. For instance, while Goodnow (1969)
found that orientation changes yielded a strong sense of difference
from most tactually presented standards, this was not true for all
tactually presented standards. This finding was presumably related
to whether or not the focal point of a particular standard was changed

by an orientation transformation. Thus, the modality of presentation,

49

the difficulty of recognizing a given letter, the difficulty of the
typeface, and the two- and three-way interactions among these factors

are important to consider.

Confusability. Finally, the letter identification process might

 

be influenced by the degree to which the letters in a given typeface
are likely to be confused with each other. Again, this dimension is
highly related to the others, and especially to difficulty, since
confusability is likely to be a main contributor to increased iden-
tification times. It seems that a certain degree of confusability is
inherent in the nature of typefaces (or scripts). As Fairbank (1947)
stated, "family relationships and homogeneity give harmony and read-
ableness to the scripts" (p. 17). Letters in any typeface tend to fit
into regular family groups, such as those that wholly enclose space
within them (A,B,D), and others that have spaces only partly enclosed
(C,F,U,V), or those that are either all curved orafll straight. Gibson
et a1. (1962) have shown that letterlike forms with some of these
family characteristics, such as symmetry, possession of a closed

loop, and straightness, are less likely to be confused with forms that
are close variants of them than are letter-like forms with the oppo-
site family characteristics. In addition, confusability within each
family group should be greater than confusability between groups.
Thus, it can be predicted that the most confusable typefaces should

be those with fewer family groups, more homogeneity within groups and
less variance between groups, and a relatively large number of members

of the less discriminable groups.

50

Several formalized methods of measuring interletter similarity
were compared by Holbrook (1975). These methods included: (1) The
Luce-Choice-model Similarity Measure--a model computed from a tachis-
toscopic confusion matrix; (2) A Subjective Rating Measure--based on
subject's judgments of similarity of letter pairs on a lO-point
scale; (3) The Distinctive Feature Measure--a model that considers
the number of distinctive features shared by two letters; and (4) The
Optimal Distinctive Features Measure--similar to (3) except that
each feature is weighted to give an optimal least-squares fit in a
regression analysis. The predictions generated by (l) and (2) tended
to be very close, and they both were similar to the prediction of a
mechanical measure based on the physical overlap of each pair. The
distinctive feature measures only moderately correlated with (l).

The strength of the correlations improved somewhat when the optimal
weights were added, but they still were poorer than the correlations
among the other measures. In conclusion, this study provides evi-
dence that a simple subjective rating measure might be a fairly
reliable method of determining confusability for a given typeface.

In summary, these three dimensions could all reasonably be

expected to have an influence on the letter identification process.

Scriptlikeness may be particularly important because there is evi-
dence to show that handwriting and printed material may be processed

by different "sub-routines."

51

Relationship of Stimulus Character-
istics and Hemisphere Specializa-
tion

 

 

The most direct evidence for a relationship between letter type-
face characteristics and hemisphere specialization is provided by
Bryden and Allard's study of the effect of typeface on visual hemi-
field differences for letter identification. As described in the
first section, a typeface by visual field interaction was found. 0f
the 10 typefaces, five were recognized better in the RVF (three,
significantly better), two were recognized about equally well in both
visual fields, and three were recognized better in the LVF (two,
significantly better). To verify the stability of the LVF advantage
for certain typefaces, Bryden and Allard carried out a second experi-
ment similar to the first, except that individual subjects were shown
just one typeface. Only the two typefaces that had yielded a signifi-
cant LVF advantage in the first experiment were used. Again, these
typefaces were recognized more accurately in the LVF, although the
visual field difference was not significant in one case. Thus, the
LVF advantage for these typefaces was shown to be stable and intrin-
sic to the lettering style.

In a third experiment, Bryden and Allard investigated the rela-
tionship between different typeface characteristics and the visual
field asymmetries from Experiment 1. They asked seven additional
subjects to rank order the 10 different typefaces on the dimensions
of familiarity, scriptlike vs. printlike, and internal confusability.

To obtain a measure of the relative difficulty of naming letters in

52

the different typefaces, two of the seven subjects and four additional
ones were presented with the stimulus cards from Experiment 1 and
were asked to name each letter as quickly as possible. The average
naming latency and average ranks on the other three dimensions were
recorded for each typeface.

Rank-order correlations and product-moment correlations between
the visual field asymmetries and each of these dimensions, except
familiarity, were significant. A LVF advantage was associated with
greater scriptlikeness, internal confusability, and difficulty and,
moderately, with less familiarity. Thus, in general, these dimensions
do seem to have an effect on the letter identification process, as
measured globally by relative hemisphere advantage for letter recog-
nition. Since most of the dimensions were significantly correlated
with visual field asymmetry, and, it appears, with each other, it is
difficult to estimate which particular characteristic or combination
of characteristics contributed most to the difference in hemisphere
advantage. The rank-order correlation was slightly greater for naming
latency than for other dimensions, while the product-moment correlation
was highest for the scriptlikeness dimension. Multiple correlation
procedures revealed that little was added to the predictability of
laterality scores by supplementing the scriptlikeness measure, but,
again, this could be due to the fact that the dimensions were highly
correlated with each other in these particular typefaces.

It should be noted that the rank-order and product-moment corre-
lations are in the direction predicted by studies of visual field

and hand asymmetry for discrimination of shapes and line orientations.

53

Simple geometrical forms (Bryden & Rainey, 1963) and lines differing
in orientation by at least 45 degrees (White, 1971) have been found
to be discriminated more accurately in the RVF. It would seem that
printed letters, also better recognized in the RVF, could be broken
down into these simple lines and shapes. Contrarily, more complex
forms (Witelson, 1974) and lines close in orientation (Umilta et al.,
1974) are better recognized when presented to the left hand or LVF.
Scriptlike letters, also better recognized in the LVF, may be thought
to be composed of complex shapes, with many loops, and lines differ-
ing slightly from the vertical or horizontal. In addition, a LVF
advantage for the perception of depth has been shown (Durnford &
Kimura, 1970), and some scriptlike typefaces are designed to appear
three-dimensional. Thus, the three-dimensionality of the typeface
may be another characteristic to consider.

The correlations between typeface dimensions and visual field
asymmetries also are consistent with the results of Faglioni, Scotti,
and Spinnler's (1969) tests of letter recognition ability in patients
with unilateral hemispheric damage. The patients were given four
letter identification subtests in which they were shown a letter and
had to find it among a series of alternative letters printed with a
different type. For all four subtests, the multiple choice letters
were printed in a convential form, either block-printed capitals or
simple italic small letters. The test letters were printed in a
conventional form for Subtests A and B, were incomplete for Subtest

C, and were partially hidden, or crossed out, for Subtest D.

54

0f the posterior brain-damaged patients, those with left hemi-
sphere damage were specifically impaired on Subtests A and B, and
those with right hemisphere damage were specifically impaired on
Subtests C and 0, when the letters were presented in a perceptually
difficult way. Faglioni et al., concluded that the contribution of
the right hemisphere to the letter identification process becomes
critical when the letters are presented in a perceptually complicated
form. These perceptually complicated forms are comparable to the
more scriptlike, internally confusable, and difficult typefaces used
by Bryden and Allard. The partially crossed-out letters (Subtest D)
are especially analogous to Bryden and Allard's very scriptlike
letters, which typically had several extra loops or lines.

Finally, given that the relative contribution of the two hemi-
spheres to the letter identification process changes when typeface
characteristics change, it should be asked how this finding increases
understanding of hemisphere differences and of letter identification.
In regard to hemisphere differences, the finding that letters are not
always more accurately identified in the RVF is inconsistent with a
characterization of the hemispheres according to the general type of
material they are specialized for dealing with (e.g., verbal vs.
spatial). Instead, this finding suggests a characterization of the
hemispheres in terms of types of processing (Bryden & Allard). Thus,
even verbal material can be better recognized in the LVF if it is
complex enough to need much of the preprocessing which can be done

more efficiently by the parallel processing (Cohen, 1973), global

55

analysis (Levy, 1969), or diffuse organization (Semmes, 1968) of the
right hemisphere.

This dissertation will investigate the specific typeface char-
acteristics that are associated with greater right hemisphere partici-
pation. One purpose of this investigation is a further refinement
of the characterization of the processing for which each hemisphere
is specialized. This dissertation also will investigate whether and
to what extent modality of presentation affects the balance of hemi-
spheric processing as typeface complexity varies. It is expected
that, contrary to the assumption that hemispheric advantages are
independent of modality, the range of processing that can be employed
may be limited by the modality in which the information is received.

The finding that visual field asymmetries are associated with
typeface characteristics also has implications for our understanding
of the letter identification process. It suggests that when the
letter cannot be immediately matched to the template or distinctive
features stored as part of the letter and word identification system,
then a process that is partly separate and partly integrated with
this system can be employed. This process acts to break down (or
normalize) the letter so that its distinctive characteristics can be
attended to easily.

This suggestion that letter identification involves both left
and right hemisphere processes is consistent with the conclusions
of Kershner (1975) and Pirozzolo and Rayner (1977). In a review of

reading and laterality, Kershner (1975) concluded that the perception

56

of single letters is better when both hemispheres are involved. To
investigate the relative contribution of the hemispheres to the word
recognition process, Pirozzolo and Rayner (1977) presented four-
1etter words either bilaterally or unilaterally and had the subjects
select their response from four choices shown on a response card.
More errors occurred for words presented in the LVF, but this effect
was especially strong when the errors were visually similar words.
Thus, Pirozzolo and Rayner concluded that, in addition to single
letter perception, the reading of whole words is also a multi-stage
process, with visual feature analysis carried out by the right hemi-
sphere and identification and naming by the left hemisphere.

A comparison of different age groups in this dissertation may
have implications for assessing developmental changes in the relative
importance of these two processes for letter identification. 0n the
one hand, since children below the age of 11 years have shown a right
hand advantage for braille letter learning, and since their RVF scores
seem to be more critical for determining reading ability, it might be
that they will not employ the right hemisphere normalization process
as often as an adult, but will concentrate instead on the verbal
aspects of the task. On the other hand, as hypothesized, since even
the moderately scriptlike typefaces will be relatively more complex
for children than for adults, children might need to employ a greater
degree of right hemisphere preprocessing than adults for those type-
faces.

In summary, Bryden and Allard found that the dimensions of type-

face scriptlikeness, difficulty (naming latency). and confusability

(:1.
-1.

1

GI

57

were strongly related to visual field asymmetries for letter iden-
tification. The relationship between these variables is in the
direction predicted by studies of hand and visual field differences
for the discrimination of shapes or line orientation and by studies

of patients with unilateral hemispheric damage. Bryden and Allard's
finding has many implications for the nature of the difference between
the hemispheres and for the letter identification process. The
purpose of this dissertation is to further understanding of these
mechanisms by examining the variables of modality of presentation and

age and their interaction with typeface characteristics.

Summary of Hypotheses

 

Modality of Presentation

Difference in direction of asymmetry. Based on the comparison
of previous findings pertaining to hand and visual field asymmetries
for letter identification, and on the intuitive idea that tactual
perception involves more "spatial integration," the tactual modality
seems to have a stronger association with right hemisphere processing
than does the visual modality. Therefore, it is hypothesized that
a greater right hemisphere (left field) advantage will be shown in the
tactual modality than in the visual modality (see Figure l). The
direction of this modality difference is expected to be constant across
levels of typeface complexity. The magnitude of the difference,
however, might be smaller for the more complex typefaces because the
degree of right hemisphere processing required for those typefaces

might be large in both modality conditions.

58

Cross-modal correlations. It is hypothesized that, for a given
typeface, the correlation between individual asymmetries in the
visual and tactual conditions will be positive and (at least) moder—
ately strong. This prediction is based on the assumptions that:

(1) an individual will tend to use a strategy involving more or less
right hemisphere processing compared to other individuals regardless
of stimulus modality, and (2) hemisphere specialization is based on

type of processing, regardless of sensory modality.

Stimulus Characteristics

 

Interaction of typeface with hand and visual field. Based in

 

part on Bryden and Allard's findings, an interaction of typeface with
sensory field is expected. Specifically, LVF and left hand scores
should increase relative to RVF and right hand scores as overall com-
plexity of the typeface is increased.

In the visual condition, it is hypothesized that there will be a
significant RVF advantage for the simplest, printlike typefaces and
a significant LVF advantage for the more complex, scriptlike type-
faces (as was found by Bryden and Allard). In the tactual condition,
it is hypothesized that there will be little or no right hand advan-
tage for the recognition of plain, block-printed letters (as was
found by Witelson, 1974; LaBreche et al., 1977). A significant left
hand advantage is expected for the recognition of the more complex,
scriptlike letters. This prediction is consistent with the left hand
superiority found for learning complex braille letters (e.g., Wagner,

1976).

59

Relative contribution of specific typeface characteristics. If
it is possible to select typefaces for which the characteristics of
scriptlikeness, difficulty, and confusability are not highly corre-
lated, it is hypothesized that each of these characteristics will be
significantly correlated with visual field and hand asymmetries and
that their effect will be additive. In addition, it is predicted
that scriptlikeness will be the strongest predictor of visual field

asymmetries, as was tentatively concluded by Bryden and Allard.

Gregg

Greater right hemisphere participation in the tactual condition
than in the visual condition is predicted for all grades. Increasing
right hemisphere participation with typeface complexity also is pre-
dicted for all grades, but the general level of right hemisphere
participation is expected to differ across grades. Specifically, the
younger grades should show evidence of more right hemisphere partici-
pation than the college students, particularly for the moderately
complex typefaces.

This prediction of a grade difference is supported by two points.
First, developmental studies of visual field asymmetries for verbal
material (in simple print) have found either a constant degree of
RVF advantage across age (e.g., Olson, 1973; Marcel & Rajan, 1975) or
no significant RVF advantage for the youngest ages tested (e.g., For-
gays, 1953; Tomlinson—Keasey et al., 1978). Second, right hemisphere
participation has been found to increase with increasing typeface

difficulty and unfamiliarity. Since the non-printlike typefaces

60

(especially the moderately complex ones) should be relatively more
difficult and unfamiliar for the younger grades, greater right hemi-

sphere participation is predicted for those grades.

METHOD: PART I--TYPEFACE SCALING

Scaling for "Difficulty"
Subjects
The subjects were 20 college students--10 males and 10 females--
with no reported vision problems. They participated for credit in

their Introductory Psychology classes at Michigan State University.

Apparatus
A 3-channel Scientific Prototype Auto-tachistosc0pe (Model G8)

with a split-beam binocular eyepiece was used. Each field was back-
lit through the diffusion screen by two standard fluorescent bulbs
(GE-FT4-5, with onset time reportedly much less than 1 msec.). Only
two fields were used--one for the letter stimuli and one for the
fixation point. The fields were set at the same moderate brightness
level, and both were modified to obtain more equal left-right half
field brightness by inserting a matte white strip of cardboard behind
each bulb pair. For the stimulus field, 35mm slides were automati-
cally fed from lOO-slide rototrays. For the fixation field, a
stationary slot held one slide. The timers for both fields were set
at 1.5 msec.

The tachistoscope was wired so that an on-off switch used by the
experimenter triggered the onset of the fixation field. Fixation

field offset triggered the onset of the stimulus field, and stimulus

61

62

field offset then triggered the advancement of the rototray. The
same switch simultaneously started a reaction timer (Klockcounter),
which was stopped by a voice activated relay (VAR) when the subject's
response was spoken into a microphone. The "delay" button on the VAR
was set to 12 seconds to allow the experimenter to record response

time and the response before the relay could be activated again.

Materials

A sample of 22 different typefaces was selected from those avail-
able in 60-point Chartpak Velvet Touch transfer lettering (black).
The goal of the selection was to obtain as broad a sample of letter-
ing styles as possible, based on the judgment of the experimenter.
Selection was made with regard to such characteristics as letter slant
and thickness, presence of serifs, and amount tyf detail, as well as
the five general characteristics on which the typefaces were to be
scaled.

For each typeface, the following eight letters were used--C, G,
0, Q, B, R, F, P. They were chosen for two main reasons. First, the
letters could be divided into two groups on the basis of physical
similarity. The first four letters were circular or nearly circular.
In contrast, the second four letters had a vertical line on the left
border, with a curved or straight line extending to the right at its
midpoint.5 Since it was hypothesized that letter characteristics
might affect field differences, the selection of letters that could
be grouped according to certain primary characteristics was designed

to aid in determining which general characteristics were important.

63

Second, all the letters also were used by Bryden and Allard. There-
fore, if the results of the current study differed from Bryden and
Allard's findings, the discrepancy could not be attributed to spe-
cific letter differences.

Slides of the letter stimuli were made as follows. Each letter
was transferred onto the center of a 7.6 cm. square piece of white
poster board with matte finish. The typefaces and letters varied
slightly in size. Heights ranged from about 1.6 cm. to 2.3 cm. and
widths ranged from about 1.1 cm. to 2.1 cm. The small squares then
were placed one at a time in the center of a large piece of the same
poster board and photographed using Kodak Ektachrome film (ASA 160).
Standard 35mm slides were developed and mounted commercially. Each
slide was examined to check that the letter was centered on the
slide, and if not centered to the nearest .5 mm., the slide was
remounted by hand.

Each slide thus contained a single black letter in the center
surrounded by a thin black outline of a square. In terms of visual
angle subtended when viewed through the tachistoscope, letter
heights ranged from .31 to .50 degrees and widths ranged from .21
to .42 degrees.

A fixation slide was made under the same conditions as the
letter slides. There was a black dot in the center of the small
square of poster board and a small black bracket in each of the four
corners of the background poster board. When viewed through the
tachistoscope, the diameter of the dot subtended an angle of .18

degrees. The area within the brackets represented the unoccluded

64

part of any stimulus field and permitted the subject to make minor
self-centering head adjustments before each trial. This viewing area
had an "apparent" viewing distance of 76.2 cm. and projected a visual

angle of about 2.9 (height) by 5.25 degrees.

Procedure

Subjects were tested individually in a small laboratory room,
where they were seated facing the tachistoscope with their backs to
the experimenter. The sequence for each trial was described to them.
They were told that upon hearing the click of the experimenter's
switch, they should look into the tachistosc0pe directly at the
center dot, while making any minor head adjustments necessary so that
all four reference brackets could be seen. After about 1.5 secs.,
the dot would be replaced by a single capital letter in one of various
styles, which they then should identify as quickly and accurately as
possible.

Before the experiment began, the room lights were turned off,
and subjects were given a brief period to adapt to the brightness
level of the fixation field. They also were asked to adjust the
interocular separation of the eyepieces so that they had a clear
view of the fixation field with both eyes, and they practiced saying
letter names loudly and clearly into the microphone.

There were 176 trials in all, with a rest period of about five
minutes after half the trials. Stimulus order was random with the
following constraints: (1) no letter or typeface was presented on

two consecutive trials, and (2) within a group of 44 trials, each

65

typeface was presented twice and each of the eight letters was pre-
sented at least five times. The 176 slides were divided into two
rototrays of 88 slides each. The order of the rototrays was switched
for half of the subjects of each sex.

For each subject, a difficulty rating for a given typeface was
determined by taking the median naming latency of the eight letters
presented in that typeface. A trial was excluded from the computation
of the median (and not replaced) if the response was incorrect
(x = 7.3%, o - 40.6%), exceeded 1.5 secs. (x = 0.7%, O-5.6%), or
could not be measured due to equipment problems (I = 0.4%, O-l.9%).6
If the typeface with the highest error rate was not included,
incorrect responses averaged 5.7% (ranging from O - 14.4%), and
responses over 1.5 secs. averaged 0.5% (ranging from O - 2.5%).

Scaling for Scriptlikeness, Confusability,
Unfamiliarity, and Superfluity

 

Subjects

The subjects were 42 college students who participated for credit
in their Introductory Psychology classes. There were 35 females and

eight males.

Materials

The eight small squares of poster board containing the letters of
each typeface were bunched together to make two rows of four over-
lapping cards (with the letters in the same order for every typeface).
Five photocopies were made for each typeface. These sheets were then
cut and mounted on poster board to serve as cards for the subjects to

sort into piles.

66

Signs were made to label the endpoints of the four scales. They

were as follows:

1 . . . . . . . . . . . . 10
most familiar least familiar
(least typically seen
or thought of)

1 . . . . . . . . . . . 10

most printlike most scriptlike
(like handwriting)

l . . . . . . . . . . . . 10
letters not easily letters most con-
confusable fusable

l . . . . . . . . . . . . 10
most basic form of the letter farthest removed from
(no extra lines or curves) basic form

The highest value on each scale was assigned to the end of the con-
tinuum expected to be associated with the highest degree of right

hemisphere participation.

Procedure

Subjects were tested in small groups of up to five peOple. They
sat at desks widely spaced around a large room and facing the wall.
On each desk was a pile of the 22 typeface cards in a random order,
the numbers one to ten placed in order, and the labels for the first
scale for that group placed above the numbers one and ten. Subjects
were instructed to sort the cards into ten ordered and equally spaced
groups to represent levels of the characteristic being measured.
There was no requirement pertaining to the distribution of the cards
in the ten groups, and subjects were allowed to change their place-

ments at any time during the sorting.

67

Sorting was individually paced. As soon as any subject finished
sorting for one scale (usually after 10 to 15 min.) his responses
were recorded, the cards reshuffled, and the next characteristic
explained to him. A different randomly selected order of the char-
acteristics was used for each small group tested. For each char-
acteristic, the median group placement for a given typeface was used

as the rating for that typeface.

METHOD: PART II--MEASUREMENT OF FIELD DIFFERENCES

Visual and Tactual Conditions

 

Subjects
The subjects were 48 fourth- and ninth-graders and college
students--16 in each grade group. All were male, right-handed,7 and

had no reported uncorrected vision problems. The mean ages for the
groups, in order, were as follows: 9 yrs., 5 mos. (8 yrs., 11 mos.

to 9 yrs., 11 mos.); 14 yrs., 5 mos. (13 yrs., 10 mos. to 15 yrs.);

and 20 yrs., 4 mos. (18 yrs., 8 mos. to 27 yrs., 2 mos.). Since the
children were tested in the summer or early fall, they were selected

on the basis of the grade they were going to enter or had just entered,
not on the basis of age or grade just completed.

Most fourth- and ninth-grade subjects were students in a pre-
dominantly middle class, suburban community.8 They were recruited
through their schools in the spring and early fall and through a
recreational program during the summer. Letters describing the study
were sent home with them, and parents were asked to return a postcard
giving their name and phone number if they and their child were inter-
ested in the study. The children were paid $3 for their partici-

pation.

68

69

College students again were Introductory Psychology students who
received course credit for their participation.

Handedness for fourth-graders was determined by noting the hand
used when the children were asked to perform five common actions
(Annett, 1970a). A child was considered right-handed if he performed
at least four actions with the right hand.

For ninth-graders and college students, handedness was deter-
mined by responses to a 12-item handedness questionnaire developed by
Annett (1970b) and modified (following Briggs & Nebes, 1975) to
include five response categories to distinguish between whether a
hand was "always" or "usually" used. To be considered right-handed,
it was necessary to "always" write with the right hand and to perform
three of the five other primary tasks at least "usually" with the
right hand.9 Numerical scores. were computed by assigning number
values to response categories, ranging from + 2 for "always" right to
- 2 for "always" left. The scores for all subjects were positive.
The mean score for college students (I a 17.2, s.d. = 5.5) was not
significantly different from the mean for ninth-graders (I = 19.7,s.d.
= 3.4; t = -l.51, d.f. = 30, N.S.).

The presence of familial sinistrality (FS+) was assessed roughly
by asking the subjects (for fourth-graders, the subjects' parents)
whether a biological parent or sibling was left-handed (i.e., wrote
with the left hand). Seven fourth-graders, eight ninth-graders, and

one college student were classified as FS+.

70

Apparatus

Visual condition. The tachistosc0pe apparatus used in this part

 

of the experiment was the same as that described for "difficulty"
scaling, except for four changes. First, the time setting for the
stimulus field was variable. Second, the tachistoscope was triggered
by a momentary push-button switch operated by the subject. Third,
the offset of the stimulus field triggered the onset of the blank
third field (for 1.5 secs.), the offset of which then initiated the
advancement of the rototray. This additional step was needed to give
the subjects time to respond before the noisy rototray movement.

Finally, response time was not measured.

Tactual condition. The subject and experimenter sat opposite

 

each other at a small table. Between them was a large cardboard box,

with the side facing the subject cut out. There were two holes in the
bottom of the box through which the subject placed his hands. A card-
board shelf was placed over the holes on the subject's side to ensure

that the subject could see neither his hands nor the letters.

A block of wood containing two metal card-holders about .5 cm.
apart was used to present the stimuli. The card-holders (as well as
the holes in the box) were placed close together to reduce the possi-
bility that attention directed to the left or right of body midline

would influence hand differences.

Materials

Visual condition. For the visual condition, 144 stimulus slides

 

were used--l6 for each of eight test typefaces and one practice

71

typeface selected in Part I. For every typeface, each of the eight
letters occurred once iri the LVF and once in the RVF. Photographs
were taken under the same general conditions as those used for "diffi-
culty" scaling. The small squares of poster board with letters in the
center were placed one at a time either to the left or right of the
center point of the background board. Positioned at the center point
was a number from one to nine (excluding seven),10 also on a 7.6 cm.
square of poster board. For the eight test typefaces, each number
was associated with a particular letter twice, once with the letter

in each visual field. The numbers were randomly assigned to letters
for the practice typeface. All slides were checked to see that the
number was in the center of the slide; if not centered to within
l.5mm., the slide was remounted.

As viewed through the tachistosc0pe, each stimulus slide con-
tained a black number (in simple print) in the center outlined by a
thin square, with a single letter to the left or right also outlined
by a thin square. The center of each lateralized letter was 2.23
degrees from the center of the number. The maximum and minimum
visual angles projected by the lateralized letters were nearly equal
to those of the letters used for "difficulty" scaling (averaging
about .01 degrees less).11 The fixation slide was the same as that

used for "difficulty" scaling.

Tactual condition. For the tactual condition, 32 letters were

 

used--eight for each of three test typefaces and one practice typeface

selected in Part I. The letters were raised surfaces of zinc plates

72

made from photoc0pies of the letters by a process used commercially
to make rubber stamps. All black parts of the letters were raised

1 mm. from the plate, and the edges were beveled. The range of
letter sizes was about the same as the range for the original trans-
fer lettering. The plates (about 3.7 cm. square) for individual
letters were glued onto the center of ordinary plastic playing cards,

which could be slipped into the card-holders.

Procedure

Visual condition. Subjects (tested individually) were seated

 

facing the tachistosc0pe, with the experimenter sitting to their
right, opposite the control panel. A large board was placed between
the subject and experimenter to block out direct light from the table
lamp used by the experimenter. Subjects were told that the general
purpose of the experiment was to see how well people could recognize
letters of different styles presented for brief times. The instruc-
tions were as follows:

When you press the button, you first will see a slide
with a dot in the middle and a bracket in each of the four
corners. You should look directly at the dot and position
your head to make sure you can see the four brackets out of
the corners of your eyes.

After about a second the slide will change. Then you will
see a number (from 1 to 9) in the middle (about where the
dot was) and a letter to either the right or left side of
the number. It is important that you keep looking straight
ahead so you will be looking right at the number when it
appears.

Your response should be the number and then the letter.
For example . . .

Subjects were encouraged to guess at both the number and letter if

they were in doubt, but were told to say just the number if they were

73

unable even to guess the identity of the letter. The experimenter
emphasized the importance of looking straight at the fixation dot
and getting the number correct.

The experimenter gave four additional clarifications of the
stimuli and task. The first three were given for the first modality
condition, whether it was visual or tactual, and the fourth was given
for the second modality condition. The experimenter told the sub-
jects that: (1) because of the nature of a threshold experiment,
mistakes were expected (This clarification was needed to reduce frus-
tration and anxiety about doing badly on the tesk.); (2) all letters
would be capitals and about the same size; (3) a capital cursive Q
could resemble the number two, and a cpaital cursive G could be shaped
like the lower case letter, but larger in size. (Pilot testing indi-
cated that these letters sometimes caused confusion. As an added
precaution, fourth-graders were shown a photocopy of all eight letters
in the practice style and asked to name them. After being reminded
about the Q and G, every fourth-grader named all letters correctly.);
(4) at least some of the lettering styles had been presented in the
first modality condition.

There were 288 trials in all--32 trials in each of eight test
typefaces and one practice typeface. The l6 stimulus slides for
each typeface were presented in a row, and then the rototray was
reset and the 16 slides presented again in the same order. The
practice typeface was presented first to provide practice and to
establish a presentation time that served as a starting point for the

12

test typefaces. Subjects always were told when the typeface would

74

change. There was a rest period (5 to 7 min.) after the fourth test
typeface, and subjects were told to ask for a brief rest (about l
min.) after a typeface was completed if they were tired. These extra
rest periods were noted on the data sheets.

Since the difficulty of the task varied greatly and, according
to pilot testing, nonuniformly across typefaces and also across grades,
it was not possible to simply determine an appropriate exposure dura-
tion for all test typefaces. Therefore, it was decided to obtain an
approximate threshold presentation duration for each field and to use
that as the dependent measure.

Thresholds were obtained by the random double staircase-method
(Cornsweet, 1962). This method was chosen because it is highly
efficient (relatively few trials need be presented) and it conceals
the contingencies involved in varying the presentation time. The
staircase procedure consisted of decreasing the presentation duration
by a predetermined step following a correct response and increasing
the duration following an incorrect response. Separate staircases
were formed for each field, with trials being randomly alternated
between the two. The staircases for both fields were started at the
point determined during practice. The steps (in msecs.) were l80, l50,

130, 110, 90, 70, so, 30, 20, 15, 10.13

For presentation time to

be decreased, both the letter and number had to be identified
correctly. If the number was incorrect, but the letter correct, the
trial was eliminated and the presentation duration remained the same

for the following trial in that field.

75

It was necessary to determine the orders of three factors--
modality conditions, typefaces, and letters. First, the order of
modality conditions was varied so that half the subjects of each
grade received the visual condition first (VT) and half received the
tactual first (TV). Testing always was done in one session lasting
about one hour and 45 minutes, with about a five-minute break between
modality conditions.

Second, eight typeface orders were determined. Each order was
used twice in each grade, once for each modality-order group. The
orders were random, with the constraints that each typeface would
occur once in every position and, to control for the possibility that
the immediately preceding typeface established a set, no typeface
would follow another more than twice.

Finally, the order of the l6 stimulus slides in each typeface
was determined separately for every typeface. The orders were random,
with the following constraints: (l) no more than three trials in a
row to one field, (2) no letter or number twice in a row, and (3) the
number of trials on which the field switched equal to the number of
trials on which it remained the same. These primary letter orders (A)
were assigned to the typefaces in four of the typeface orders. The
arrangementtrfletters assigned to the RVF was designated Set l, and
the arrangement assigned to the LVF was designated Set 2. For the
other four typeface orders, Set l was assigned to the LVF and Set 2

to the RVF (B).

76

Tactual condition. The subjects were seated Opposite the experi-

 

menter facing a large box. They were shown one of the tactual practice
letters to give them an idea of the letter size and of the fact that
the zinc plate was not part of the letter. The instructions were as
follows:

Place your hands through the two holes in the box and rest
them on these card-holders. Hold them in the same position
in-between letter presentations. I will slide a card under
either your right or left hand (in random order). When I
say "go" and push your fingers down, start to feel the letter
with your index and middle fingers. You will have varying
amounts of time to feel each letter (l to l5 secs.). When
the time is up, I will remove the card and say "answer," and
you should make the best guess you can.

There were l04 trials in all--eight trials in the practice type-
face and 32 trials in each of the three test typefaces. The practice
letters were presented first for 12 seconds each. Four letters were
presented to each hand, with the letters assigned to hand switched for
half of the subjects of each grade. Fewer trials were used in the
tactual condition than in the visual condition because of time limi-
tations and because it was less important to determine a starting
point for the test typefaces. Following practice, the three test
typefaces were presented, with brief rest periods (1 to 2 min.)
in—between.

A staircase procedure was used to determine a rough threshold
presentation duration for each hand. llmeprocedure differed from that
described for the visual condition only in that presentation times
were longer and fewer steps were used. The steps (in secs.) were l6,

l2, 8, 4, 2, l. For all subjects, the starting point for each hand

was l2 seconds for every typeface.

77

As in the visual condition, eight typeface orders were deter-
mined. They were the six possible order51fiithe three typefaces, with
two of them (randomly selected) repeated. Each order was used twice
in each grade, once for each modality-order group, and always was
paired with the same visual typeface order. The letter order for
each typeface was the same as the letter order used for that type-
face in the visual condition.

Computation of Laterality Coeffi-

cients and’their Stability/
Effectiveness

 

 

 

For each field in each typeface, a threshold measure was compu-
ted in two steps. First, to lessen the effects of the starting point,
the first three runs were eliminated from the measure. A run was
defined as a series of increasing or decreasing times including the
reversal point between it and the preceding run, but not the reversal
point following it. Second, the median of the remaining points was
computed without interpolation and was used as the threshold measure.
All points were included in the measure whether they represented cor-
rect or incorrect letter identifications, except those representing
correct identifications for which the fixation number was incorrect.14
Interpolation was not considered justified because of the rank order
nature of the data. Thus, if a median occurred directly on a time
step, that step was taken as the median, regardless of the number of

points that occurred on that step. If a median occurred between two

steps, it was defined as the average of those steps.

78

Laterality coefficients (LCs) to be used as the dependent vari-
able in the main analyses were computed by subtracting the right field
threshold measure from the left field measure and then dividing that
difference by the average of the two thresholds (combined threshold
or CT). Thus, the greater the score in the positive direction, the
greater the right sensory field advantage, and the greater the score
in the negative direction, the greater the left sensory field advan-
tage. Use of a ratio score was meant to reduce the effects of the
subjective inequality of the two ends of the time scale. A main
effect of the inequality was that a given time difference between
thresholds was more easily perceptible and more meaningful for lower
thresholds than for higher thresholds. Minimizing this effect was
especially important because threshold level was predicted to be at
least moderately related to the direction and degree of field differ-
ences.

Several questions may be raised about the stability/effectiveness
of the threshold measures: (l) whether the measures were reliable,
especially given the relatively small number of trials; (2) whether
the measures were successful in establishing a roughly 50% accuracy
level; and (3) whether missed fixation numbers significantly affected
the visual measures. Finally, an underlying factor in all these con-
cerns was their variation across grade, typeface, and modality condi-
tion.

To investigate the first question, a rough stability estimate for

each field was computed by taking the median of each run after the third,

79

then averaging the first and last pair of runs and correlating those
averages.15 The correlations are shown in Table l. In the visual
condition, most of the correlations exceeded .7, indicating high
stability. The main exceptions were the correlations for Typefaces

l and 3 (Tl and T3) for fourth-graders. For Tl, the overall corre-
lation was only moderately positive (r = .38), while for T3, the
correlation averaged across fields was slightly negative (r = -.04).
In the tactual condition, the average correlation was lower than in
the visual condition (.63 compared to .78), and there were several
instances of moderate to low positive correlations. The general level
of stability was still acceptable, though. With the exception of the
two cases noted in the visual condition, the correlations did not vary
systematically across typeface or field.

To check the second question, the percent correct identifications
out of the total number of points included in each threshold estimate
served as a measure of accuracy level. These percents are listed for
all grades, modality conditions, and (for the tactual condition
oniy)16 modality orders in Table 2. In the visual condition,
accuracy level was markedly lower for T3 than for the other typefaces l
(averaging about .24 compared to .42 for the next lowest), and slightly
lower for T2 relative to the others (.42 compared to .47). Except
for these two typefaces, all the percents ranged from .4 to .6 and
were highly consistent across grades and visual fields. A 3 x 2 x 8

ANOVA was used to test the effects of grade, visual field, and type-

face on accuracy level, and a 3 x 2 x 7 ANOVA tested those effects

80

A.oeeewemo »e_pweeem ease
cw eeeapeew use: eueweeau ee>wm e cw epeww ee>wm e Lew mess eeew umeep we saw: mueeneem emegu xpcov

 

 

 

 

 

.mees we swag umep we emege>e ecu wweiuae geuwe meet we swan umcww we mcewees we emege>e we cewuepegceu u haw—weeemn

:3";
No. mm. ms. mm. ee. om. mm. mm. p~<
Ge 3: $5 3: :5 3c 375
Nm. mm. Nm. oe. Nw. me. am. em. Ppeu
so as so :5 so as 375
me. mm. mm. mm. me. Nu. Nw. PF. sum
$5 3; as so so $5 875

cm. No. mu. om. No. om. em. mw. gee weapon»
Sen;
ww. om. Nm. mm. mm. NN. pm. Pm. ww. mm. mw. mm. we. Nu. om. um. oN. Pu. pp<
A8 A2 35 so as as as 3: 35 as Sc 85 as as A8 A2 37.5
em. mm. mm. No. m5 mm. mm. mm. Nm. em. pm. ~m. o5 No. mm. «N. am. 5 p—eu
:E as so so :5 Ge 35 to :5 so :3 2: :5 :5 3: as $75
em. em. F5 N5 pm. nm. em. mm. m5 em. em. Na. mm. mm. mm. Nm. om. om. sum
so so so Sc 35 so :5 :5 So so Sc to so as Sc 25 $7;

No. mm. mm. em. ow. Nw. pm. m. mm. mm. Nm. FR. 7m oN. Nm. mm. on. Ne. gee Peem_>

um we mm we mm we um me am a; mm we um we mm “4 um we
Fp< m e . aewpeeez
meeweexw
A. eueewume zepppewum some :. eeeepucp eeeem Lee mueeaeem we geese:
es» mw memegucecee :HV .eeegm ecu .euewmexu. cowepeceu xuwpeees. epeww heemeem an esemeee epesmeezu we exuw—weeumii. F u4m<w

81

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Ameuzv Aeanv
mm. mm. Fm. mm. we. mm. me. we. mm. me. Fe. No. No. ee. Pee
em. em. em. em. mm. mm. Nm. em. Ne. mm. mm. me. om. Ne. __oe
em. em. em. oe. ee. em. we. em. ee. ee. me. me. so. Ne. see
me. Pm. me. ee. we. we. em. oe. me. we. em. em. em. me. gee _e=eoee
Ae_nzv : :3 z :4 1e :3 I“ :4 2e :4 1e :4
gauze Amaze fienze “muzwi Aenzv Amaze
>e w> >w e>
_P<
me. 8. n 2 22:3 Sena
me. we. Nm. om. om. em. Ne. om. mm. Nm. mm. Pm. em. NN. oe. me. we. me. __e
Ae_uze
me. we. we. Pm. me. om. ee. _m. mm. em. mm. mm. mm. mm. em. oe. ee. Ne. Peoe
Amenzv
ee. he. em. mm. om. em. om. mm. me. we. we. om. mN. ,N. Ne. we. me. we. see
Ampuzv
me. me. we. ee. me. em. me. we. em. me. em. em. PN. 0N. Fe. Ne. mm. mm. see _e=m_>
em e3 e>e e>e e>e e>e e>e e>e a): e>e e>¢ e>e e>e e>e e>e e>e e>e e>e
Pee e m e eo_e_eeoe
oeeee seeeeeo:
eeeweezw

 

Levee xuwpeees Azece ceeuweeee .eeueeu ecu gewv ecu .eewuwecee

xueeeeee some Lew eeegm ece .eeeweezu .epeww xuemcem an wweiuee Loewe eewueeewwueeew pouncee we eceegeeii.N u4m<w

82

excluding T3. Both analyses revealed a significant main effect of
typeface (F = 43.37, d.f. = 7,3l5, p < .001; F = 7.67, d.f. - 6,270,
.2 < .00l) and no other effects. Thus although the requirements for
a roughly 50% accuracy level generally were met, the level varied
across typefaces, and especially differend from 50% for T2 and T3.

In the tactual condition, accuracy generally was higher than in
the visual condition. This effect was especially large for Tl, for
which all grades averaged over 60% correct. For T2 and T7, the
effect was smaller and was found only for the ninth-graders and
college students, whose accuracy was higher than that of the fourth-
graders. A 2 x 2 x 3 x 3 ANOVA testing the effects of hand, modality
order, grade, and typeface on accuracy level revealed a significant
effect of typeface (F = 34.47, d.f. = 2,84, 3 < .001) and a typeface
by grade interaction (F = 3.9l, d.f. = 4,84, p_< .006). The typeface
effect reflected the high accuracy for Tl, caused in part by a ceiling
effect. Tl was so simple to identify that many subjects were able to
repeatedly identify letters at the shortest time step (l sec.), thus
preventing the staircase from operating. The typeface by grade inter-
action reflected the higher accuracy of the ninth-graders and college
students compared to the fourth-graders for T2 and T7. There was no
systematic variation of accuracy with hand or modality order.

To answer the third question, the number of missed fixation num-
bers with and without a correct letter identification was tabulated
by grade, typeface, and visual field. The results are shown in
Table 3. Three effects were evident. First, the overall number of

missed fixations was low. Even for Tl, which had the greatest number

83

.Nueneem Neeew>wecw cm me eemmws eNer Neemw> Lee mueNee ceNNNwa ewes one .memegueesee cNe

 

NNenzN

one No ANN N.N Nev e.N ANN ON 3 ON Nev N.N 8 ON 3 Ne Lev N.eN 78 N2
“meeegm
o.N N.N N.N o.N o.N N.N o.e N.N N.N N.e o.N N.N N.N o.N N.N N.N NooeeooeN mmoeoe

N.N N.N N.m N.e N.N N.N N.e o.N N.N N.e N.N N.o N.N o.N o.N o.N Nooeeoo eeoev

N.N N.N N N N oN N N N N N N N N N e m NN NooeeooeN NNanN
N o N N o o N N e N N o N e NN NN Nooeeoo NNoe

N N e N N N e N N N o o o N N e NooeeooeN NNanN

N.N N.N N N N m N N N N N N N o N N N NN Nooeeeo NNN
N N N N N N N N N N N N N N e e NooeeooeN NeanN

N.N N.N o N e N N N N N N N o N N N N N Nooeeoo eNe

 

m>m “>4 m>m m>4 u>m w>4 u>¢ m>4 m>z m>4 u>m u>4 m>m m>4 u>m m>4 m>m u>4

 

NN+eNNNe N N N N e N N N NN eoNNoN oeeee

 

eeeweexw

meeeem mmeeee eemeee>e ecu NNeueNeeem meeegu en» New eNer Nezmw> ecu eeeweexu seem New
:eNuNerNuceeN NeepeN ueeggeuce eee uueeeee new: Amuueneem mmeeue nuggemv NNcNee eewuexww eemmNE we geese: Neuewii.m m4m<w

84

of missed fixations, the missed points represented only about 6%
of the 256 total trials per field for each grade. Second, the number
of missed fixations varied across typeface. The most errors occurred
for Tl, the simplest typeface usually presented for very short times.
The least errors occurred for T3 and T6, two typefaces that generally
required long presentation times. This pattern indicates that a
large number of fixation errors probably resulted from difficulty
in perceiving the number at short presentation times, rather than
solely from inattention or carelessness. Finally, the number of
missed fixations tended to increase with age, especially for Tl.
This trend cannot be explained by a decrease across grades in reluc-
tance to guess, since the number of I'no guess" responses was actually
smallest for fourth-graders for Tl (38 compared to 6l for ninth—
graders, 43 for college students). Better explanations might be
higher motivation or longer presentation times for the fourth—graders.
This grade difference, then, minimizes any concern that a grade effect
on laterality coefficients is caused by the relatively greater diffi-
culty of the task for fourth-graders or their inability to maintain
fixation. A 3 x 2 x 8 ANOVA testing the effects of grade, visual
field, and typeface on missed fixations revealed only a significant
main effect of typeface (F = l3.83, d.f. = 7,3l5, p < .00l).

In summary, with a few exceptions, the threshold measures were
highly stable, resulted in an approximate 50% accuracy level, and
were not biased by a large number of missed fixation points. One

main exception was T3 in the visual condition. For this typeface,

85

accuracy level was only about 25% for all grades, but was especially
low for fourth-graders, who also had a slightly negative T3 stability
correlation. Other possible exceptions to consider when analyzing the
visual results are T2, for which a generally lower accuracy level was
obtained across grades, and Tl, for which fourth-graders showed a
relatively low stability correlation. For the tactual condition, the
stability correlations were lower in general, and the higher accuracy
levels for Tl indicated a possible ceiling effect. In no case was a
sensory field effect found, which suggests that the laterality coeffi-
cients were not biased directly by any of the three potentially con-

founding factors named above.

RESULTS: PART I-—SCALE VALUES AND SELECTION
OF TYPEFACES

Ratings, Dispersion Measures, and Reliability

In general, ratings given to the initial group of 22 typefaces
indicated that undergraduates were able to distinguish among type-
faces on the five characteristics measured and that these distinctions
were consistent across subjects. These conclusions were suggested
by the uniform and widespread distribution of ratings and by moderate
to high alpha values. Based on the ratings for difficulty, script-
likeness, and confusability, eight typefaces were selected for the
visual condition in Part II. Two criteria for their selection were
met. First, a wide range of ratings on those characteristics was
represented. Second, correlations between the characteristics were
low. Three of the eight typefaces also were selected for the tactual
condition-~one at each end of the rating scale and one with moderate

ratings.

Initial Group of Typefaces

 

Mean rating and dispersion measures for the initial group of 22
typefaces are shown in Table 4. Ratings of "difficulty" for each
typeface were based on the mean of the median naming latencies for
individual subjects. The general level of latencies17 was not so
important as the relation between naming latencies for the different

typefaces, which was highly consistent across subjects (Alpha - .97).

86

.memecpcecee cw ece mecemees cewmcecmNe ece chNeN ummZeN ece umecmwc ec...be

 

87

mm. we. om. mu. no. ecmp<

mesmemz
NN.N - NN.oN NN.N - NN.ov NN.N - NN.oN NN.N - N.ov NN.NNN - N.eev eoNeeonNN
eo.N NN. eN. eo.N e.NN eeoz

NNN.N - eN.NN NNN.N - eo.NN ANN.N - eo.NN NNN.N - No.NN eNNNN - NNNN NeNNee

O
"I

Fm.m mm.m Nm.¢ mm.¢ N.mmo com:

 

NNNerNoeem ANN NNNeeNNNENNeN Nev NNNNNeemeNeoe ANN NmoeoxNNNeNeoN NNN NNNeonwNN NNN

 

 

Nov emcee eNNpceeeceucN-NEem n ceemceemwo .e.mucecheemwo
.Neems cw
eeeem NcNeeiceN e co aceseeepe acemeuee ceNeeE u chNNm hm ceesumcwpem
Newumeceueececu

 

Neeumeceueecece e>Nw ce mchNec we Neccmees cecheemNe ece .mceee .mecep< .meeewecza NN NN<ii.e mem<w

88

Ratings for the other four characteristics all were based on
median placements on a l0-point scale and were similar in three
regards: (l) mean ratings within one point of the scale midpoint
(5.5), (2) extension of ratings over the entire scale, and (3) Q-
values of about l.0. For all characteristics except familiarity, the
Alpha values were at least moderately high, indicating that the
relative ratings of the typefaces were consistent across subjects.

The lower Alpha value for familiarity is consistent with the idea.
that experience with different lettering styles would be likely to vary
greatly across indivdiuals.

The correlations between the five characteristics for all 22
typefaces are shown above the diagonal in Table 5. The values are
positive, generally moderate to high, and, except in one case, sig-
nificant at least at the .05 level. These results are consistent with
the correlations between the first three characteristics (ranging
from .73 to .96) found for the ratings given to Bryden and Allard's
sample of l0 typefaces. The implication is that these characteristics
are correlated in the majority of typefaces and probably tend to occur
together naturally. Therefore, typefaces selected with the goal of
obtaining a sample in which these characteristics are not highly corre-
lated might be very atypical. This would limit the generalizability
of findings based on that sample, and is a problem to be considered

when interpreting the results.

89

Selected Typefaces

 

Visual condition. The three main goals (in order of importance)

 

for selecting the eight typefaces to be used in the visual condition
were that their ratings should be: (l) minimally correlated on the
first three characteristics (those found by Bryden and Allard to be
related to visual field differences), (2) spread over a wide range
of values on the three scales, and (3) low in variability. With the
first goal in mind, the three scales were divided at the median, and
typefaces were categorized according to the eight possible combina-
tions of high and low values on the three scales. No typeface was
found for one combination (low 1 and 2, high 3), so the typeface with
the closest possible fit was used. Then one typeface was selected
from each combination. The correlations between the characteristics
for these eight typefaces are shown below the diagonal in Table 5.
For the first three characteristics, the correlations were low and not
significant.

Table 6 shows the ratings and dispersion measures for the eight
selected typefaces on each of the three characteristics.18 Again,
the higher the rating, the greater the degree of the characteristic
for which the scale was named and for which right hemisphere process-
ing should be important. Consistent with the second goal, the type-
faces do cover a wide range of values on all three characteristics,
with the emphasis, if any, on the higher end of the scales. The third
goal of choosing typefaces with low rating variability was not met.
The mean dispersion values for the eight selected typefaces are higher

than for the whole group of 22 typefaces.

90

TABLE 5.--Correlations of five characteristics for all 22 typefaces
(above diagonal) and eight selected typefaces (below
diagonal). All correlations are Pearson r

 

(l) Diff. (2) Script. (3) Confus. (4) Unfam. (5) Superflu.

 

 

(N=22)
(l) .47* .64** .59** .66**
(N=8)

(2) -.06 .6l** .38 .47*

(3) .23 .28 .89** .74*

(4) .40 .09 .93** .82**
(5) .6l .32 .79* .86**

*p < .05

**p < .Ol

91

TABLE 6.—-Selected eight typefaces. Individual and mean rating and
dispersion measures and alpha for three characteristics.
The last column shows the overall z-score for each typeface
based on the average of the z-scores for each characteristic

 

 

 

 

Characteristics
Typeface (1) DIFFICULTY (2) SCRIPT. (3) CONFUS. OVERALL
Rating=mean RT Rating = median placement
in msecs. on lO-points
1* 623 1.08 1.04 -l.22
s d = 56 9 Q = .3 Q = .3
2* 639 7.65 3.92 - .02
59.4 1.2 1.2
3 ' 772 2.50 2.26 .02
111.2 1.1 .3
4 663 5.17 6.62 15
60.8 .9 1.4
O
689 3 67 8.50 .35
101.2 1 4 1.2
697 9.07 4.40 .57
64.5 .8 1.1
721 3.83 8.81 61
80.5 1.6 1.2
706 7.12 9.30 .96
107.2 1.9 .7
Mean rating 688.8 5.01 5.61
Mean dispersion
score 80.2 1.15 .99
Alpha .93 .64 .67

 

*Chosen for the tactual condition also.

92

Since Bryden and Allard found that visual field differences were
significantly correlated with all three characteristics, a combination
of the three ratings was used as the best a priori predictor of sen-
sory field differences in the current study. These composite com-
plexity scores are shown in the last column of Table 6. The complexity
score for each typeface was obtained by computing a zyscore for each
of the three characteristics based on the mean and standard deviation
for all 22 typefaces, and then averaging the three scores. The pre-
diction is that the higher the composite complexity score, the greater
the degree of right hemisphere involvement in the processing of that
typeface. The fact that only twocfiithe composite scores were negative
is another indication of the emphasis on the higher end of the scales.

In addition to the main group of eight typefaces ("test" type-
faces), a typeface of average difficulty was chosen to be used as a
practice typeface in both the visual and tactual conditions. It was
a simple script (Brush script) and had a mean naming latency of 681

msecs.

Tactual condition. The three typefaces selected for the tactual

 

condition met the following requirements: (1) for each typeface, all
letters were recognized with at least about 25% accuracy using
unlimited exploration time in pilot testing with college students,
and (2) the extremes and a middle point of the composite scale were
represented.

T3, T6, and T8 were not considered because they did not meet the

first requirement. Tl and T7 then were chosen to represent the

93

extremes of the composite scale, and T2 was chosen as a middle point.
Even though 12 was not midway between T1 and T7 on the basis of rank
order, it was selected because (1) the composite gfscore was closest
to midway between the scores for T1 and T7, and (2) it was rated high
on scriptlikeness, the characteristic Bryden and Allard found to be

most strongly related to visual field differences.

RESULTS-~PART II--FACTORS AFFECTING LATERALITY COEFFI-
CIENTS IN THE VISUAL AND TACTUAL MODALITIES

Main Analysis: Effects of General Type-
face Factor,gGrade, and Modality Order

 

This section describes three repeated measures ANOVAs used to
test the effects of typeface complexity, modality, grade, and modality—
order on laterality coefficients. The results of the first ANOVA,
which included all variables, reflected two findings about the central
hypotheses of this dissertation. First, the predicted increase in
right hemisphere participation with increasing typeface complexity
was found for the visual condition, but not for the tactual condition.
Second, contrary to prediction, right hemisphere participation was not
systematically greater in the tactual condition than in the visual
condition. The results of the other two ANOVAs, one for each modality
condition separately, reflected two additional findings. One was a
grade difference in the visual condition and the otherEimodality-

order difference in the tactual condition.

Three Repeated—Measures ANOVAs

Analysis including modality factor. To examine the main hypothe-

 

ses of increasing right hemisphere participation with increasing type-
face complexity and tactual as opposed to visual presentation,
laterality coefficients (LCs) were plotted as a function of typeface

and presentation modality in Fig. 2. Only the scores for the three

94

95

Fig. 2. Average laterality coefficients for the three typefaces
in the visual and tactual conditions for each grade
separately and all grades combined.

AVERAGE LATERALITY COEFFICIENTS

96

Right

sensory All Grades 4th Grade

 

(TideJd +3 T [ Visual—:l
' +.2N Tociua|---

 

   

Left N2 N
sensory

field '13 ”'
OdV. l l 1 1
Righi

sensory

 

 

 

field +,3-— 9th Grade — College

adv. +2

 

 

 

l ' 2 7 l
Simplest Mosi Simplest
complex

TYPEFACE

 

 

Most
complex

97

typefaces presented in both modality conditions were shown, and they
were plotted separately for each grade and for all grades combined.

(To supplement the graphs in this section, average LCs, standard devia—
tions, and significance levels for all grade by modality by typeface
conditions are given in Table 7.)

Fig. 2 indicates that in the visual condition, the predicted
increase in right hemisphere participation with increasing typeface
complexity was found for each grade. With grades combined, tgtests
indicated a significant RVF advantage for the simplest typeface, no
visual field difference for the medium complex typeface, and a sig-
nificant LVF advantage for the most complex typeface.

In the tactual condition, only the scores for the fourth-graders
showed an increase in right hemisphere participation with increasing
typeface complexity. Right hemisphere participation for the ninth-
graders and college students was lowest for the most compelx typeface.
For no typeface, with grades considered either separately or com-
bined, was there a significant hand difference. In general, the
hypothesis of a greater right hemisphere participation in the tactual
as opposed to the visual condition was not supported. Only the scores
for the simplest typeface were consistently in the predicted direc-
tion--a weaker right field advantage was found for the tactual as
opposed to the visual condition for each grade.

To test the effects of these factors (typeface, modality, grade,
and modality order), the LCs were analyzed in a 3 x 2 x 3 x 2 ANOVA

with repeated measures on the first two factors. A summary of the

—oo. v aeua

Po. v at:
A Ibo . bu v

NeNoemoeeN 02059...:
we cow eeueeccee museum eNer ucmNc ecu aweN ecuesee e» mecemums uceecmeme cow «mania mcwme .mo. v e*

 

98

 

N00.N Nee.0 NNN.N »HanN
0N. NN. 00. NN. NN<
Ne0.N N00.N Nee.v MxNuzN
00.- NN.- e0.- N0. NNe
NNeuzN
e0. N0. 0N. NNe
N00. NN0.N NNN. NeanN
00. 0N. N0. N0. NN00
NNN.N NNN.N Nem. NeanN
NN. NN. N0.- NN. 000
NNN.N NNN.V NNN.N NNNuzN

N0.- NN.- N0. NN. eNe NeeNoeN
Nuenzv
N0. eeeeN.- eeeeN.- 00. 00. N0.- N0. .eeNN. NNe
NN0. NNN.N Nee. NNN.N NNe.N Nme.0 NNe. NeN.N N0Nuzv
eN.- NN.- .NN.- .NN.- NN.- NN.- N0. NN.- 0N. NN00
NNN. Nee. NNN.N NNe.N Nee.0 NNN.N Nem.0 NNN.N N0Nuzv
0N. .NN. NN.- .0N.- .NN. «NN. N0.- NN. e0N. 000
N0e.v NNN.N NNN.N Nme.N NN0.N NNN.N N00. NN0.N NNanN

N0. e0. .NN.- eNN.- NN. 0N. 00.- N0. eNN. eNe N000N>

NNe N N 0 0 e N N N

 

ceNuNeceu
eeuco
eeuweexw xuNNuee:

 

ceece NNNNueeE
Axpce Nueueuuv ecu .ceNeNecee xuNNueeE .euuweexu .eeucm e» checoeeu Neececewwwe eNer
we eucuewwwcmwm ecu .Nmemecucecue cNV Nceeuue>ee ecuecuum .NuceNuNwweee NNNNuceuuN emuce><--.N Nem<e

99

analysis is shown in Table 8. Only the typeface main effect

(F = 6.94, d.f. = 2,84, 2 < .01)19 and the typeface by modality inter-
action (F = 3.54, d.f. = 2,84, p_< .05) were significant, reflecting
the increase in right hemisphere participation with increasing type-
face complexity in the visual but not in the tactual condition.

The LCs then were analyzed further in separate ANOVAs for each
modality for the purpose of clarifying the following: (1) the dif-
ferent pattern of scores across typeface for the two modalities,

(2) the nearly significant interaction of modality with modality
order (F = 3.14, d.f. = 1,42, p_< .10), and (3) the effect of the
additional typefaces presented only in the visual condition and not

included in the analysis.

Visual condition. Fig. 3 shows the average LCs for all eight

 

typefaces presented in the visual condition for each grade separately.
As predicted, right hemisphere participation generally increased with

increasing typeface complexity for each grade. However, the increase

in right hemisphere participation was very irregular across the medium
complex typefaces (T2 through T5) and for T8.

For 12, T3, T4, and T8, a grade difference also was apparent.
Contrary to the prediction of a decrease in the need for right hemi—
Sphenaanalysis with increasing age, college students showed the great-
est degree of right hemisphere participation, reflected in a LVF ‘
advantage for those typefaces. The ninth-graders showed a large RVF
advantage for those typefaces, and the fourth-graders, a somewhat

weaker RVF advantage.

100

TABLE 8.--Summary of analysis of variance on laterality coefficients

with modality factor included

 

 

 

 

Source D.F. MS F u2+
Between Subjects
Mean (sensory field) 1 .222 .70
Grade 2 .400 1.26
Order 1 .551 1.73
Grade x Order 2 .072 .22
Sub. w. groups 42 .318
Within Subjects
Modality 1 .146 .36
Grade x Modality 2 .158 .38
Order x Modality 1 1.290 3.14 .01
Grade x Order x Modality 2 .112 .27
M x sub. w. groups 42 .411
Typeface 2 1.675 6.94** .03
Grade x Typeface 4 .371 1.54 .01
Order x Typeface 2 .207 .86
Grade x Order x Typeface 4 .103 .42
T x sub. w. groups 84 .241
Modality x Typeface 2 1.036 3.54* .02
Grade x Modality x Typeface 4 .167 .57
Order x Modality x Typeface 2 .078 .27
Grade x Order x Mod. x type 4 .074 .25
MT x sub. w. groups 84 .293

 

Fmax = 2.28, d.f. = 5,47, N.S.

Test of no association in the correlation matrix:

x2

= 9.48, d.f. = 15, N.S.
{E < .05
*fp < .01
2 _ Variance accounted for by Factor A

INA _ VariEnce for all factors + error term

 

101

.eeucm ceue cow ceNNNecee Nusz>
ecu cw empcemece meeuweczu NcmNe ecu New mucewewwweee ANNNuNeNuN emucm><

.m

.mN

m

702

 

 

103

A 3 x 2 x 8 ANOVA (grade by modality order by typeface) with
repeated measures on the typeface factor was performed on the visual
LCs. The summary is shown in Table 9.20 Both the main effects of
typeface and grade were significant (F = 5.97, d.f. = 7,294, p_< .01;
F = 4.99; d.f. = 2,42, 2 < .05, respectively). Newman-Keuls multiple
comparison procedures (shown in Appendix B) were used to examine the
pattern of differences across typeface and grade. T6 and T7, those
for which right hemisphere participation was greatest, significantly
differed from all other typefaces, but not from each other. None of
the other differences between typefaces was significant. The college
students showed significantly greater right hemisphere participation
overall than did the fourth- or ninth-graders, who did not differ from

each other.

Tactual condition. Fig. 4 shows the average tactual LCs as a

 

function of typeface for each grade by modality order. (The average
LCs for each modality order with grades combined are provided in
Table 7.) While no general trend across grade or typeface was found,
there was a clear difference between modality orders that would
account for the nearly significant order by modality interaction in
the ANOVA with modality included. For every grade by typeface com-
bination except T1 for college students, relatively more right
hemisphere participation was shown when the visual condition came
before the tactual condition (VT) rather than after (TV). This find-
ing is contrary to the hypothesis that, if anything, prior visual
exposure to the stimuli would make right hemisphere processing less

critical.

104

TABLE 9.--Summary of analysis of variance on visual laterality
coefficients

 

Source 0. F. MS F (.2

 

Between Subjects

 

 

 

Mean (Visual field) 1 .013 .03
Grade 2 1.902 4.99* .03
Order 1 .020 .05
Grade x Order 2 .010 .03
Sub. w. groups 42 .381
Within Subjects
Typeface 7 1.234 5.97** .07
Grade x Typeface 14 .239 1.16
Order x Typeface 7 .153 .74
Grade x Order x Typeface 14 .039 .19
T x sub. w. groups 294 .207
Fmax = 4.15, d.f. = 7,47, 9 < .05

Test of no association in the correlation matrix:

2

X ==34.04, d.f. = 28, N.S.

105

Fig. 4. Average tactual laterality coefficients for each typeface
according to grade and modality order.

AVERAGE LATERALITY COEFFICIENTS

 

106

 

Condition

 

11 U.

41h ----* ""51!

Grade 9th --—I ——[j
COIL—O —o F

 

 

 

 

Simplest
TYPE FACE

1k
l

N

Most complex

107

The effect of modality order on the pattern of LCs across type-
face was not consistent across grades. With grades combined, the
predicted increase in right hemisphere participation with increasing
typeface complexity was found for the VT condition, but not for the
TV condition or both combined. Even in the VT condition, though,
right hemisphere participation was not systematically greater in the
tactual compared to the visual modality.

A 3 x 2 x 3 (grade by modality order by typeface) repeated meas-
ures ANOVA on the tactual LCs revealed only a significant main effect
of modality order (F = 4.44, d.f. = 1,42, p_< .05). The ANOVA summary
is provided in Table 10.

Specific Typeface Characteristics: Average Threshold,
Scriptlikeness, Confusability, and Difficulty

 

 

This section pertains to the relation between visual LCs and
specific typeface characteristics. The relation was examined by three
measures: (1) first-order or simple correlations between LCs and the
characteristics, (2) regression analysis including the following as
predictors of LCs: the characteristics, dummy-coded variables for
subjects, grade, and the interaction of grade with the characteristics,
and (3) separate regression analyses for each grade including only
the characteristics as predictors of average LCs. The characteristics
considered were those on which the typefaces were rated in Part I,
scriptlikeness, confusability, and naming latency (NL), and a fourth
characteristic, the average of the individual combined thresholds for

each typeface (average threshold or AT). The major finding with all

108

TABLE lO.--Summary of analysis of variance on tactual laterality
coefficients

 

Source [1. F. MS F 02

 

Between Subjects

 

 

Mean (Hand) 1 .364 .92
Grade 2 .199 .50
Order 1 1.763 4.44* .03
Grade x Order 2 .179 .45
Sub. w. groups 42 .397
Within Subjects
Typeface 2 .104 .35
Grade x Typeface 4 .345 1.18
Order x Typeface 2 .269 .92
Grade x Order x Typeface 4 .173 .59
T x subj. w. groups 84 .293

 

Fmax = 1.03, d.f. - 2,47, N.S.

Test of no association in the correlation matrix:

x2 = 1.25, d.f. - 3, N.S.

109

three measures of the relation was a grade difference. For fourth-
and ninth-graders, NL was the best predictor, while scriptlikeness
was a poor predictor, and no combination of characteristics signifi-
cantly predicted LCs. In contrast, for college students, script-

likeness and confusability were significant predictors.

Average Threshold

 

Visual condition. Average threshold (AT) was used as a predictor

 

for LCs in addition to the three characteristics used to select the
typefaces in Part I. Even though AT reflected typeface difficulty,
which to some extent already was measured by the naming latencies (NL)
from Part I, it was included because it shared the same measurement
factors with the LCs. If the results showed a high correlation of
LCs with AT, but not with NL, an explanation of the change in right
hemisphere processing across typeface in terms of peripheral measure-
ment factors could not be eliminated. In other words, raising the
general threshold level for the more complex typefaces could have
changed the task in some way or failed to completely equalize task
difficulty across typeface.

In contrast, NL estimates were obtained with a method independent
of LC measurement. The NL estimates involved different subjects and
centralized rather than lateralized letter presentation. A high
correlation between LCs and ML would suggest an explanation of the
change in right hemisphere processing across typeface in terms of
difficulty in the higher-order analysis of images of letters all

clearly perceived.

110

Table 11 shows ATs (in msecs.) for the eight typefaces in the
visual condition for grades separated and combined. In general, ATs
varied widely across typeface, covering nearly the entire scale. There
was also a general increase in AT with increasing grade. Except for
T1, which most subjects found very easy, and T3 for fourth-graders,2]
the standard deviations were large (averaging about 37 msecs.).
These large standard deviations are one indication that the use of
individually determined presentation durations was justified.

A 3 x 8 ANOVA to test the effects of grade and typeface on visual
ATs revealed significant main effects of both grade and typeface.
According to Newman-Keuls tests, all differences between typefaces
were significant except those between T5 and T7 and between T6 and
T8. Also, fourth-graders were significantly different from ninth-
graders and college students, who did not differ from each other.

The ANOVA summary and Newman-Keuls results are provided in Appendix

C.22

Tactual condition. Since there were only three typefaces used

 

in the tactual condition and since tactual LCs did not vary signifi-
cantly across typeface, ATs and the three characteristics from Part I
were not examined as predictors of tactual LCs. The ATS were tabulated,
though, to look at the general pattern of scores according to type-
face, grade, and modality order. These ATs (in secs.) are shown in
Table 11. Three observations about the pattern of scores can be

made. First, it was of interest to see whether there was a modality

order difference in AT that would correspond to the significant order

111

N

NeNecmeccw + NeNecmeccw n quv eNecmech eecNeEee ececze

 

 

 

» Nuzv
0.N 0.0 N.NN N.N 000
ManzN
N.N e.0 N.0N N.N NNe
N0euzv
0.N N.0N 0.N NNe
N.eN NN.NN N0.NN NONuzN
e.0 N.N 0.0 0.N NN00
N0.NN NN.0N N0.NN N0NuzN
0.0 0.N 0.0N N.N 000
N0.NN 0.NN N0.NN N0NuzN
N.N N.NN 0.NN N.N eNe Neoeey
Pezuueh
N0e-zN
0.e0N N.N0 N.e0N N.N0 N.Ne N.00N 0.00 N.NN NNe
NNON NONN NNeN NNeN NNNN NONN Nva N0 0 N0Nuzv
N.NN N.0N 0.NN 0.N0N N.N0 0.0N N.0eN N.N0 0.NN NN00
NNNN NONN N0eN NNNN NONN NNNN NNNN NNNN N0anv
N.eN N.N0 0.N0 0.00 0.00 N.0e 0.N0N 0.00 N.eN 000
NNNN NNNN NeNN N0eN N000 N0 0 NNeN NONN N0NuzN Nmoamev
e.00 N.0NN 0.0N 0.eNN 0.NN N.N0 N.NNN N.NNN 0.0N gee N000N>
NNe 0 N 0 0 e N N N eoNNNeeoe
eeucu . .
meuweexh xuepueex

 

ceece Neweueee Aapce Nuauuuuv ecu .ceNNeecee Nuepueee .eeuweeau

.eeucm eu checeeeu “memecucecue cwv mcewuuN>ee ecuecuum ecu Amh<v «Nepecmecce emceeeee emuce><ii.pe mem<w

112

effect found for LCs. A higher Al for the VT condition might indi-
cate that the greater right hemisphere participation for that condi-
tion was related to the greater difficulty of the task for those
subjects. A lower AT for the VT condition might indicate that a more
wholistic type of processing was needed in that condition because of
shorter presentation times. From Table 11 it is apparent that there
was no difference between modality orders. Thus, an explanation of
the LC difference in terms of presentation time is eliminated.
Second, the pattern of ATs across typeface and grade was observed.
The ATs were very low, approaching ceiling,for all grades on T1.
There was a large increase in AT between Tl and T7 and a relatively
smaller increase between T7 and 12. Scores generally decreased with
increasing grade, with the main difference occurring between fourth
and ninth grades. A 3 x 3 ANOVA testing the effects of grade and
typeface revealed significant main effects of both factors and a sig-
nificant interaction. The interaction reflected the finding, con-
firmed by simple main effects tests, of a grade effect for T2 and T7,
but not for T1. The ANOVA summary and relevant comparison tests are
shown in Appendix 0.23
Third, the pattern of tactual ATs across grade and typeface was
compared to the corresponding pattern for visual ATs and to the LCs.
Averaged across grade, the relative difficulty of T1, T2, and T7, as
reflected in AT scores, was similar in the two modalities. In the
tactual condition, greater difficulty was related to a greater degree

of right hemisphere processing. This relation was weak, however,

113

because of the lack of a significant change in LCs across typeface,
and, if the grades were considered separately, the relation was no
longer apparent.

For both modalities, ATs were significantly higher for fourth-
graders than for ninth-graders and college students, indicating that
the fourth-graders were slower at perceiving the letters and that
the ninth-graders and college students did not differ in ability.
This pattern of ATs was not similar to the pattern of LCs for either
modality. The greatest discrepancy was in the visual modality, where
even though the college students showed significantly greater right
hemisphere participation than the fourth- and ninth-grades, they did
not differ from the ninth-graders in performance level.

Correlations of Visual LCs, ATs, and
Part I Characteristics

 

 

Table 12 shows the correlations among the four characteristics
to be used as predictors of LCs for each grade, and Table 13 shows
the correlations of those predictors with LCs.24 The correlations
both with and without T3 were considered in both tables for two rea-
sons. The first reason was that the validity and reliability of the
LCs for T3 were uncertain. For every grade, the average percent
correct for T3 was much lower than for any other typeface. In addition,
only for fourth-graders, the stability correlations for T3 were low,
the standard deviation for the 13 LC was much lower than for the other
typefaces, and the T3 AT approached the upper limit of 180 msecs.

Since percent corrent was the only very questionable measure for

114

TABLE 12.--Correlations between Part I ratings on difficulty,
scriptlikeness, and confusability and average combined
threshold (AT) for each grade--(a) with all typefaces

included, and (b) with T3 eliminated

 

(a) All Typefaces (N=8)

 

(b) T3 Excluded (N=7)

 

 

 

Diff. Script. Confus. Diff. Script. Confus.
Diff. -.O6 .22 .31 .83*
Script. .28 .15
AT-4th .74* .36 -.O7 .50 .90* .34
AT-9th .77* .20 -.15 .48 .90* .36
AT-Coll .82* .23 -.14 .60 .87* .33
*p < .05

TABLE 13.—-Correlations of laterality coefficients with Part I
ratings on difficulty, scriptlikeness, and confusability
and with average combined threshold (AT) for each grade--
(a) with all typefaces included, and (b) with T3 elim-
inated

 

(a) All Typefaces (N=8)

(b) T3 Excluded (N=7)

 

 

 

Grade

Diff. Script. Confus. AT Diff. Scipt. Confus. AT
4th -.57 -.40 -.42 -.4O -.77* -.45 -.49 -.51
9th -.44 -.23 .24 -.34 -.38 -.36 .15 -.22
Call -.29 —.76* -.75* -.20 -.86* -.72 -.7O -.83*

 

*p < .05

115

ninth-graders and college students, the correlations and prediction
equations including T3 were emphasized for those grades. For fourth-
graders, though, since the majority of measures indicated lack of
reliability and validity for T3, the correlations excluding T3 were
emphasized.

The second reason for considering the correlations both with and
without T3 was that because of its values on the predictor variables--
very high on A1 and NL and very low on scriptlikeness and confusabil-
ity--T3 generally had a large effect on the pattern of correlations.
Therefore, it was necessary to examine which correlations were
altered significantly by the inclusion of T3, how they were altered,
and whether the conclusions based on those correlations should be
modified.

In Table 12, the correlations with all typefaces included fit
the expected pattern, given the assumption that both NL and AT
measured aspects of a general difficulty factor. For all grades,
the correlation between AT and NL was high, while the correlations
between AT and the other characteristics were low to moderate.
Therefore, with AT included as a predictor, mainly the effects of AT
and NL were expected to be hard to separate.

The correlations among predictors with T3 excluded were higher
in general. This was to be expected for the three characteristics
from Part I because the correlations between those characteristics
were minimized for the group of eight typefaces as a whole. In addi-

tion, for every grade, correlations between AT and the other

116

characteristics were moderate to high. With T3 eliminated from the
prediction equations, these higher correlations between predictors
would make a significant prediction of LCs less likely and would make
it harder to separate their effects. The especially high correlations
between scriptlikeness and AT would hinder the detection of a theor-
etically important effect of scriptlikeness over and above a task
difficulty effect.

Table 13 shows the correlations of the four predictors with LCs
for each grade. With the exception of the confusability correlation
for ninth-graders, all the correlations were negative, as predicted.
Thus, high ratings for a given typeface on the characteristics named
by the scales tended to be associated with low LCs or relatively more
right hemisphere participation in recognizing letters of that type-
face. For fourth-graders, with all typefaces included, the correla-
tions were all moderate, with the correlation for NL being the highest.
With T3 excluded, which was the condition to be emphasized, the corre-
lations increased in strength. The increase was especially great for
the NL correlation, which became significant. In general, since the
pattern of correlations was not changed by excluding T3, the lack of
T3 should not be an important factor in the conclusions..

For ninth-graders, the correlations with all typefaces included
were again all moderate, although slightly lower than for fourth-
graders. The NL correlation was again the highest. With T3 excluded,
most of the correlations were slightly lower, but the pattern did not

change. Therefore, it seems that the inclusion of T3 in the prediction

117

equation for ninth-graders should not affect the conclusions drawn in
any major way.

For college students, with all typefaces included, the correla-
tions distinctly differed across characteristics. The correlations
for scriptlikeness and confusability were higher than the correlations
for NL and AT. This pattern of correlations indicated that right
hemisphere participation was related to scriptlikeness beyond any
effect of task or typeface difficulty. When T3 was excluded, however,
the correlations for NL and AT became significant and slightly higher
than the other correlations. According to this pattern, difficulty
overshadowed scriptlikeness. Since the exclusion of T3 caused a
marked change in the pattern of correlations, conclusions regarding
the relative importance of the characteristics must be tentative.
Prediction of Visual LCs from Typeface

Characteristics Using Multiple
Regression

 

 

 

Regression including variance from grade and subject variables.
To examine whether a significant part of the variance in LCs was
explained by typeface scriptlikeness, confusability, and difficulty
(NL), those variables were entered into a regression along with dummy
coded variables for subject, grade, and the interaction 0f grade
with the three characteristics. All the variables are listed in
Table 14, followed by their standardized regression coefficients, and
the amount of additional variance accounted for when they were entered
into the regression in the order listed. The number of contrasts

tested for grades was equal to the degrees of freedom for the grade

1'18

TABLE l4.--Regression coefficients and statistics for analysis including subject
and grade variables and for separate analyses by grade with four pre-
dictors of laterality coefficients

 

 

 

 

 

ALL GRADES SOURCE OF STANDARDIZED % ADDITIONAL VARIANCE
(all typefaces VARIANCE RECRESSION ACCOUNTED FOR IN
included) COEFFICIENTS STEP-WISE ANALYSIS
Overall Test: Grade Variable l -.39
(CV1 = Coll v 4, 9) 4.04 d.f. = 2,381
Square mult. R=.288 Grade Variable 2 .63 p_< .0004
F=2.36, d.f.=56,327 (GV2 = Call, 9 v 4)
.2 < .0001 45 Subject 17.03 d.f. = 45,336
Variables p.< .Ol
Difficulty (NL) -.23 2.49 d.f. = 1,335
§_< .001
Scriptlikeness -.16 2-74 .f- = 1.334
g < .0005
Confusability .19 0.01 .f. = 1,333
p_< .80
GVl x Diff. 1.19
GVl x Script. -.O3
GVl x Confus. -.37 2.43 d.f. = 6,327
GV2 x Diff. .17 2_< .08
GV2 x Script. .02
GV2 x Confus. -.28
4th Grade
(13 excluded) Scriptlikeness -.48 20.45 d f. = 1,5
p.< .31
OVERALL TEST: Difficulty (NL) -1.14 43.84 d.f. = 1,4
< .09
Square mult.R=.715 Confusability .42 5.47 Elf. = 1,3
F=l.25, d.f.=4,2 p_< .51
.2 < .49 AT .35 1.69 d.f. = 1,2
p_< .76
9th Grade Scriptlikeness -1.01 5.29 d.f. = 1,6
p.< .58
Overall Test: Difficulty (NL) -1.99 20.92 d.f. 391,5
B.‘ —
Square mult. R=.868 Confusability 1.20 19.89 ld.f. = 1,4
=4.94, d.f.=4,3 p,< .29
2_< .11 AT 1.58 40.71 d.f. = 1,3
p_< .06
College Scriptlikeness -.92 57.30 d.f. = 1,6
< .03
Overall Test: Difficulty (NL) -.89 11.45 52f. 331,5
< .
Square mult. R=.955 Confusability —.19 23.48 filf. = 1,4
F=15.80, d.f.=4,3 §.< .025
p.< .02 AT .71 3.23 .f. = 1,3
p_< .24

 

119

factor (2). The specific contrasts are given in the table. For
subjects, the number of contrasts tested was 45, equal to the degrees
of freedom for subjects in each grade (15) times three.

The relative magnitude of each variable's contribution was
represented, in part, by the standardized regression coefficients.
Considering absolute values, the grade coefficients were the largest,
with one exception. The exception was the first interaction term
for grade and difficulty, an effect suggested by the simple correla-
tions of NL with LCs (relatively high forfourth-graders and low for
college students). The interaction terms for grade and confusability
also were fairly high. Again, this was consistent with the simple
correlations (high and negative for college students, positive for
ninth-graders).

The overall test of the prediction equation was highly signifi-
cant (F = 2.36, p < .0001), indicating that at least one element in
the vector of beta weights differed from zero. However, the total
variance accounted for was only about 29%, most of which (17%) was
attributable to the subject variables taken together. The grade
variables combined accounted for only about 4% of the variance, even
when entered into the equation first. And the remaining 8% was divided
about equally among scriptlikeness,difficulty (both of which were
significant ), and the six interaction variables taken together (which
nearly reached significance even though entered into the equation
last).

In summary, the proportion of variance in LCs accounted for by

typeface characteristics was significant, although small in an

120

absolute sense and dwarfed by subject variance and variance unaccounted
for. Given that the interaction terms for grade and typeface char-
acteristics approached significance, regressions not including subject
variables were computed for each grade separately to clarify the

effects of the typeface characteristics.

Separate regressions by grade. The same information shown in
Table 14 for the first regression is also shown for the regressions
for individual grades. For every grade, the predictors used were the
three characteristics from Part I and AT. Scriptlikeness was entered
into the equation first because of its possible theoretical impor-
tance. Since four predictors were used to explain only eight scores
(seven, for fourth-graders), the results in general should be regarded
as tentative.

For fourth-graders, in agreement with the simple correlations,
difficulty (NL) accounted for the highest proportion of variance by
far. That proportion was over 40% regardless of the order in which the
variables were entered into the equation. Scriptlikeness accounted
for the second highest proportion of variance when entered into the
equation first, but this proportion dropped to less than 3% when
scriptlikeness was entered following AT.

For ninth-graders, difficulty again was a moderately strong
predictor of LCs. However, in contrast to the results for fourth-
graders, AT and, to a lesser extent, confusability were also fairly
strong predictors. The variance accounted for by AT and confusa-

bility was especially large (over 60%) when they were entered into

121

the equation following scriptlikeness and NL, thereby controlling
for the shared variance with those characteristics. It should be
noted that, contrary to expectation, the coefficients for AT and
confusability were both positive, suggesting that higher AT and
greater confusability were related to less right hemisphere partici-
pation. For confusability, this suggestion was supported by the
direction of the simple correlation with LCs. For AT, however, a
suppressor effect probably was operating. Finally, scriptlikeness
accounted for a very small proportion of the variance, even when
entered into the equation first.

In marked contrast to the equations for fourth- and ninth-graders,
scriptlikeness accounted for the largest portion of the variance for
college students. This proportion generally was large, regardless
of order of entry into the equation. Confusability also was a strong
predictor, but its effect was eliminated if entered after AT. Again,
the relative importance of scriptlikeness and confusability compared
to AT and NL could change dramatically if T3 was excluded.

The overall tests of the regression equations did not reach sig-
nificance for the fourth- or ninth-graders,lnn:did so for college
students, with about 95% of the variance accounted for. ~Thus,
explaining the fourth- and ninth-graders'LCs not only involved weigh-
ing the four predictors very differently than for adults, but might
also involve the use of other predictors not already considered or

a different scaling of the given predictors.

122
Stimulus and Procedural Factors: glypeface Position,
Letter Order, and Specific Letters

To clarify possible problems with the method used for determin-
ing threshold in this dissertation and to investigate further the
types of factors affecting LCs, the Specific effects of three stimulus
and procedural factors were examined. Since the factors were counter—
balanced with field and typeface, interpretationcflithe main analyses
was not an issue. Results showed that all three factors systemati-
cally affected LCs at least in the visual condition. In the case of
the first factors, typeface position, a relation was found between
visual LCs and whether a typeface was presented in the first or second
half of the visual condition. The LCs computed from second-half
presentations compared to those from first-half presentations reflected
a stronger asymmetry in the direction of the overall LC for a given
typeface. The second factor, letter order, affected LCs for two
typefaces in the visual condition. An advantage was found for the
orders in which the first three letters were relatively very easy to
identify. Finally, both the strength and direction of the LCs varied
widely with the third factor, specific letter. Variation was
greater for visual presentation than for tactual presentation, and

it tended to be in the opposite direction for the two modalities.

Typeface Position

Visual condition. Since each typeface was presented twice in

 

each of the eight order positions for each grade, typeface position

was not confounded with the effects of typeface and grade.

123

Nonetheless, the relation between LCs and whether a typeface occurred
earlier or later within the condition was examined to look for possi-
ble effects of practice, task familiarity, or an increasing atten-
tional set across trials.

To observe the general position effect for each typeface, the
cases were divided according to whether the typeface was presented
before or after the halfway rest period. Average LCs then were com-
puted for both groups. In Fig. 5 these scores are shown for all
typefaces, which were arranged along the horizontal axis in order of
increasing right hemisphere participation with grade and typeface
position groups combined. The two typefaces for which there was a
significant overall LVF advantage (T6 and T7) tended to show a greater
LVF advantage when presented in the second half of the condition than
when presented in the first half. In contrast, with the exception of
12, typefaces for which there was at least a slight overall RVF advan-
tage showed a greater RVF advantage when presented in the second half
of the condition than when presented in the first half. Thus, in
general, in the second half of the condition there was a greater
differentiation of LCs for the typefaces which tended toward opposite
field advantages overall. As opposed to explanations in terms of
increasing ease of the task or establishment of a general attentional
set, this effect would be better explained by an increase over time
of flexibility in allocating the most efficient type of processing
for a given typeface.

The effects of typeface position, grade, and the interaction of

these two variables were tested in a 2 x 3 ANOVA for each typeface.

124

Fig. 5. Average LCs for each typeface when presented in the first
half of the visual condition (N=24) compared to the average
LCs when presented in the second half (N=24), with type-
faces ordered according to degree of right hemisphere
participation.

AVERAGE LATERALITY COEFFICIENTS

 

 

 

 

 

 

 

 

 

Presented in first
aRd/F half of condition
v.
' Presented in second
T-3I’ Z half of condition
pf
+'2F
+,|» l i
LVF=RVF M— um
_ IN.
-.2-
-.3—
LVF
adv. 1 1 1 1 1 1 1
i i 4. 2 § § Z 6
(.23) (.09) (.08) (.02) (.02) (-.0l) (:23) (:24)
Greatest Greatest
overall overall
RVF adv. TYPEFACE LVF adv.

(overall average LCs )
shown in parentheses

126

A summary of the F-values is provided in Appendix E. Typeface
position was significant only for T8, and there were no significant

interaction effects.

Tactual condition. It was particularly important to consider

 

the effects of typeface position on tactual LCs because position was
not perfectly counterbalanced with typeface. Given only six possible
permutations of three typefaces, two of the eight typeface orders

used for each grade necessarily were repeated. To test the effects

of grade, typeface position, and their interaction using an equal-n
ANOVA, the four subjects in each grade for whom an order was repeated
were eliminated. The F-ratios from the separate ANOVAs for each type-
face were tabulated in Appendix E. No significant effects were found,

and there were no apparent trends in the LCs across typeface position.

Letter Order

 

Visual condition. Given the high interdependency of succeeding

 

trials within a field using the staircase procedure, lower thresholds
could have been facilitated by particular letter orders, such as

those containingiacluster of relatively easy letters. Thus, it was

of interest to examine letter order as a factor affecting the process
of threshold determination, even though the two letter orders for

each typeface were counterbalanced with field for each grade. Letter
order was not counterbalanced with trial order, though, since for

each typeface, each letter in a given order was always presented on the
same trial. Therefore, trial order was considered when interpreting

any letter order effects.

127

A 3 x 2 ANOVA was performed for each typeface to test the effects
of grade, letter order, and their interaction on LCs. The F-ratios
are summarized in Appendix E. The only significant order effect was
for 15, and there were no significant interactions. Since the order
effect for 18 approached significance, and since it was helpful to
have at least two typefaces to find a shared difference between
orders, the pattern of correct responses across trials was analyzed
for both TS and T8.

Table 15 shows the average LCs for each letter order by grade
group. For T5, Order A led to a greater RVF advantage compared to
Order 8. This implies a relative advantage for Letter set 1, since
that occurred in the RVF for Order A. In contrast, for T8, a greater
LVF advantage was shown for Order A compared to Order 8, implying a
greater advantage for Set 2. Letter sets 1 and 2 also are shown in
Table 15 for T5 and T8 along with information about percent correct
responses.25 The most striking factor in favor of Set 1 for T5 and
Set 2 for T8 was the large percent of correct responses for the first
few letters in those sets. In those sets, percent correct for the
first three letters was at least 20% greater than percent correct
for all letters, while in the other sets, percent correct for the
first three letters was always less than overall percent correct.

At least two possible reasons for the association between a
relatively high percent correct for the first three letters in a set
and a relatively low threshold can be eliminated. First, the trial

orders for the beginning of Set 1 for T5 and Set 2 for T8 were very

128

.eeuNNN ecu Necu ceeco ecu cw mceupe— ecu we ceue cew mceesec Nuwcwe

 

 

 

 

 

 

 

eN.N N0. 00. 0N. 00. NN. 0N.- NN.- NN00

NN.N 00. N0. 0N. 00. 0N. 0e. NN. 000

0N.N N0. 00. 0N. NN. 0N. 0N. NN.- eNe
_0000 N NceeNoN NNN NeoNNoN N NNN _0000 N mceNNoN NNN eceeNoN N NNN
m um— N iuumccou N iuumccoo M m am... & iuuoccou N iuumccou a

N0N40NamNaNN.0«00NaNN 00000000 NeN.NN.0N.0.N.0.e.mN 00000000 N 000-0>0.1 N 000-000 1. .11

N N N 000-000 0 N-000-000 e 0N

N0. N0. 00. eN.N N0. NN. 0e.- NN. NN00

00. N0. N0. 0N.N N0. 0N. NN. 0e. :00

00. N0. N0. NN.N e0. NN. NN. NN. eNe
_u0eN m mceupep NNu mceupep m NNN _ueeu m mceuuee NNu mceuuep m umN
m ewe a Noecceu a iueecceu u m 000 a uueecceu a iuoecceu m

N0N.0N.eN.NN.0.0.00N0 00000000 «NNN.NN.0N.NeNae.N.NN, 00000000 N 000-000.1 N 000-000.- 11

N N N 000-000 0 N 000-000 0 0N

New ceuuee ceeco ceeeee eeucu

 

New ceue cw mceeeee eeccu echw ecu cow uoeccee uceecee
e>weuNec ecu .mw ecu 0w cow eeucm coue Nu oeecm ceece ceNNeN cuue cew NNceNewwweee NNNNuceNuN emuce><1-.mN mem<w

129

different--one contained a run of three consecutive trials and one
contained the first two trials--suggesting that trial order was not
responsible for the association. Second, a large number of correct
responses at the beginning of a typeface for a given field could mean
a greater number of trials before the third reversal, thereby lowering
the level of the staircase before the cut-off. A comparison of the
number of trials before cut-off revealed that the favored set for T5
did have about one more trial, on the average, before cut-off than the
other set. There was no difference between sets for T8. Therefore,
if number of trials before cut-off did have an effect, it was only
slight.

One explanation that cannot be eliminated is that a few easy
letters in one field at the beginning of a typeface led the subject
to have more confidence in judgments for that field or to shift atten-

tion toward that field, thereby producing a lower threshold.

Tactual condition. A 3 x 2 ANOVA also was performed for each

 

typeface in the tactual condition to examine the effects of grade,
letter order, and the interaction of those factors on LCs. The F-

ratios, summarized in Appendix E, all failed to reach significance.

Specific Letters

 

Visual condition. In addition to the comparison of letter orders,

 

the pattern of percent correct responses for individual letters was
analyzed as a factor contributing to threshold determination. The

questions examined were whether individual letters contributed

130

differently to field effects, and whether, as hypothesized, relative
accuracy and field differences for individual letters differed across
typeface. Since for a given subject the presentation times of indi-
vidual letters differed within and across fields, accuracy and field
differences were computed only by summing across all subjects, with
the goal of approximately equalizing presentation times across letters
and fields. This goal generally was met. In addition, since one of
the prime purposes of the letter analysis was to make cross-modal
comparisons, only the three typefaces presented in both modality
conditions were analyzed.

Table 16 shows the individual letters listed in order of decreas-
ing accuracy for each typeface. In general, F, P, R, and C were the
letters most accurately identified, and B, O, and 0 were the least
accurately identified. Despite these similarities, there were some
unique patterns for each typeface. For instance, F was the easiest
letter for T1 and T7, but next to the hardest letter for T2, and R
progressed in relative difficulty with increasing typeface complexity.
As indicated in Table 17, the correlations between visual typefaces
in relative accuracy for individual letters were, on the average,
positive and moderately strong. In overall average percent correct
responses, T1 was highest and T2, lowest, consistent with the ATS for
those typefaces. The number of letters for which a 10% increase across
grades in percent correct responses was shown was greater for the
most complex typeface, suggesting that all grades were highly familiar

with T1 letters presented visually.

131

TABLE l6.--For T1, 12, and T7 in each modality condition, individual
letters listed in order from highest to lowest average
percent correct identifications (shown in parentheses)

 

Visual Condition Tactual Condition

 

 

Tl T2 T7 T1 T2 T7

 

Highest
accuracy F (.77) P (.81)* F (.91) O (.93) O (.92) O (.93)

R (.77) C (.78)* P (.89) C (.74) C (.88) B (.55)*
P (.75) R (.6l)* C (.84)* F (.72) Q (.52) Q (.54)
C (.69) G (.57) G (.60) P (.64)* G (.36) C (.52)*
G (.66) Q (.46) R (.52)* B (.62) P (.36)* P (.49)*
O (.66) B (.35)* B (.43) Q (.60) B (.28)* R (.46)*
B (.61) F (.33) O (.40) G (.59)* R (.25)* F (.29)

Lowest

accuracy 0 (.42)* O (.12) Q (.21)* R (.51)* F (.06) G (.19)

 

Overall

average
% correct '67 '50 -60 -57 -45 .50

responses

 

*Letters that showed an increase in accuracy of at least 10% from
4th grade to college.

132

TABLE l7.--Correlations between typefaces in average percent
correct responses for individual letters. Correla-
tions are shown for each modality condition separately
(above and below diagonals) and across modality condi-
tion within typeface (between diagonals)

 

 

 

Visual
11 12 T7
.24 .77 (T1)
11 .06 .50 (12)
12 .65 -.O5
Tactual
17 .66 .70 -.43

(11) (12)

 

11

 

 

133

The average percent correct responses per field for individual
letters are shown in Fig. 6. The results were separated by grade
and typeface, and only those grade by typeface combinations with a
field difference of at least 5% were included. The major finding is
that the amount of asymmetry differed greatly across letters. The
letters B, Q, G, and R showed larger asymmetries than the others. In
addition, the direction of asymmetry varied across letters and across
typefaces within letters. The variation was usually, but not always,
consistent with field threshold differences. For instance, the
variation was consistent with threshold differences for B, but almost
exactly apposite for G.

The amount and direction of asymmetry apparently was not related
to the initial grouping of letters in terms of circular v. straight
line features. The one exception to this generalization is that for
college students, the circular letters in 11 (O, C, Q) tended to show
a LVF advantage, while the letters with a straight line showed a RVF
advantage, possibly reflecting a hemisphere processing difference.
Asymmetry also was not related to the frequency of letter use, since,
for example, 8 and P are about equal in frequency (Mayzner & Tresselt,
1965), but very different in amount of asymmetry shown. One factor at
least partly related to the amount of asymmetry shown was the average
percent correct responses for the letters (from Table 16). Low
accuracy, in the case of B and Q and F (T2 only), tended to be asso-
ciated with a large degree of asymmetry. However, this was not the

case for G and R. In general, then, the letters showed very different

134

Fig. 6. Visual condition. Average percent correct responses per
visual field for individual letters shown separately for
each grade for 11, 12, and 17. (Only the cases where the
field difference exceeded 5% were included.)

 

 

135

 

 

I
,_\

4th Grade
9th Grade

CoHege

 

 

LVF RVF

 

 

 

 

 

 

 

F
V
R
F

11111111 1--v0.111-..1..1!.-1--
.- w

E
mmm‘chmm on—R
WE F.
L w
F. F
v nor-mam v
R R
F- F
W . w
1 N N

lLVF RVF

 

 

 

 

 

 

LVF RVF

111111111...- II-

 

 

 

 

 

 

 

 

 

LVF RVF

.6.
LVF RVF

 

 

 

 

 

l.00‘lLVF RVF

00:02:20: +0250 200000 0000004

 

Most Complex

Simplest

TYPEFACE

Most Complex

Simplest

136

asymmetry patterns, which can only be speculated on given the current

paradigm.

'1actual condition and cross-modal comparison. Table 16 also
provides a list of individual letters in order of accuracy for each
of the three typefaces in the tactual condition. In general, letters
with the highest accuracy were 0, C, and Q, and one letter with an
overall low accuracy was R. As shown in Table 17, there were moder-
ately high positive correlations between the typefaces in individual
letter accuracy. In terms of percent correct responses across all
letters, 11 showed the highest and 12 showed the lowest, again in
agreement with AT ordering. The number of letters showing an increas-
ing accuracy with increasing grade was consistent across typeface,
suggesting a general ability factor related to age, but not related
to differential familiarity with the tactual typefaces.

Individual letter comparisons in the visual and tactual conditions
revealed marked modality differences, as predicted. The correlations
within the diagonal lines in Table 17 indicated that there was abso-
lutely no association between the patterns of individual letter
accuracy in the visual and tactual modalities for 11 and 12. The
correlation between modalities was moderately negative for 17. This
finding that cross-modal correlations were much lower than within-
modal correlations indicates that the sensory systems differed in the
particular letter qualities they were most adept at processing. The
initial division of letters into those with a circular pattern vs.

those with a left vertical line may be associated in part with the

137

characteristics differently favored by the two modalities. In general,
the circular letters tended to be recognized better in the tactual
condition,26 while the letters with a left vertical were recognized
better in the visual condition.

An analysis of the percent correct responses per field for
individual letters in the tactual condition revealed that the number
of grade by typeface combinations for which a field difference
occurred was much lower than in the visual condition. Since the
few differences found tended to be inconsistent across grades, no
conclusions could be drawn about the pattern of asymmetry for indi-
vidual letters. A figure showing the hand asymmetries comparable to
the one for the visual modality is provided in Appendix F. One com-
parison of interest was between the direction of asymmetry for indi-
vidual letters in the two modalities. Of the 24 grade by typeface
combinations for which there was an asymmetry for a particular letter
in both modalities, 15 (65%) showed asymmetries in opposite directions
for the two modalities. Again, a marked difference in processing
between the two modalities was suggested.

Individual Difference Factors: Individual LC Distribu-

tion and Relation to A1, LC in Other Modality, and
Familial Sinistrallty

 

In this final section, LCs were examined on an individual subject
level. First, average scores were compared with two measures of
individual variation: (1) the distribution of individual LCs for each
typeface, and (2) the change in LCs across typeface for each individual.

These measures generally were consistent with the trends for average

138

L05 in both modality conditions, though marked individual differences
existed. Second, the possible effects of an individual's threshold
level, LC in the other modality, and FS status on his LC for a given
typeface were examined. In general, there was no systematic relation
between these variables and LCs. The most unexpected result was the
tendency toward negative cross-modal correlations for LCs in a given

typeface.

LC Distribution

 

Visual condition. Individual LCs were examined to determine

 

whether average LCs adequately reflected the pattern of scores for

the majority of individual subjects. Table 18 lists the frequencies
of positive and negative visual LCs for each typeface for each grade
separately and for all grades combined. The relative number of sub-
jects with positive and negative scores followed the average LCs very
closely. At least half the subjects in each grade showed the expected
RVF advantage for 11, with only six subjects in all grades combined
showing a reversal. In contrast, the majority of subjects in each
grade showed a LVF advantage for T6 and 17.

For the four typefaces requiring an overall medium amount of
right hemisphere processing (12, T4, 15, and 18), the subject dis-
tribution reflected the grade difference found for average LCs. A
large majority of ninth-graders showed a RVF advantage, only a few
more fourth-graders showed a RVF advantage than showed a LVF advan-
tage, and the majority of college students showed a LVF advantage.

The large number of fourth- and ninth-graders with LCs of zero

for 13 is still another indication of the possible floor effect for

139

.NN ecu NN we eouce>u ecu e0 eecucEeo
NN cow 00 cecmwc 0 Nu eechee Nu: cceeeuc ecu .coNNNecee Nueeeue ecN ce0 .Neeuwecxu ceew «NuN ecu oN eecueeeo Necec
eeumwN muN Neeuweeze ceew emcww ecu cow 0: emuce>u cecmwc u mu eechee Nu: cceeuue NNcu .ceNNNecee N000N> ecu co...2r
.eeuweexe
uucu cow Nceee Noceoeecw ecu eecw eeNNNse ecez eeuwecxu ce>N0 u cow eceN we uceNerweee NNNNuceNuN u cewz meeenceme

 

 

 

 

 

 

 

 

 

NeN-zc
NN m m 0N m m m oN N e e m NN¢
N0NuzN
N m N N m e m e m N N N N NNeu
N0N-zc
m m N N e N m N e N N N N :00
N0N-zc
m N N N m N N N m e N N m cue NueueuN
>eu >eu >eu >eu >eu >eu >eu >eu >eu >eu >eu >eu
N.NN NN :0 :0 :0 :0 :0 :c :0 :0 :0 00 :0 :0
00.0 00 N0-zN N0-zv N0-zv N0-zv NN-zN N0uzc
cuwx meeeneam we cecEez
>N N> >N N> >N N)
N N N
NNe-zc
0 mm NN wN mN o NN mN 0N oN ON NN MN NN eN o NN<
N0N-20
oN m NN m m 0 m e m m oN m m m N m N NNeu
N0N-00
«N m NN m m m m oN N o e m m m N m N cNm
NON-zc
mN m oN 0 ON m m m N N e e e m 0 ON m cue Nuemw>
.o.N.m.mNuc >eu >eu >eu >eu >eu >eu >eu >eu >eu >eu >eu >eu >eu >eu >eu >eu
. . . 0>0 0>c 0>0 0>0 0>0 0>c 0>0 0>0 0>0 0>c 0>0 0): 0>0 0>c 0>0 0>c
A N v m NN
uc eeecu coNNNecou
cow: Nueeneem we cease: e N m m N e m N NNNNeee:

 

NceweeewoNNcuc ececcmNaec 0cmwc chNuececN we ceece ch Nu<000>w

 

-' . . 4‘IH.!'MI..-Wlufl.u fink.“ .usfirnr.7 “IE 7 “:5... 3.“.Eii'.“ Hus

 

 

 

“Rh-1.“.‘ulnrfl .0 . r ”.“1II W ”§.§

came: emuce>u ecu Ne New eeuweexu mmeceu cceeeue ecu ceN: NceNNNNceu ecez muceNerweee
NNNNuceNuN emocx mueehcem we cece:c ecu mzocm caeNeu NNuN ecN .ceece NNNNueoE NNNce Nueeeuuv ecu
.ceNNNecee NNNNueee ecu eouweezu ceue cow oemuecu>eu eNer acoNc_200weN u cuwz eeucm cee Nuueneem we cecEez-.NN 000<N

140

that typeface. However, the finding that 12 out of 16 college
students had non-zero laterality scores for 13, nine of which were
positive, supports the inclusion of that typeface in the analyses of
characteristics affecting college students' scores.

Besides examining the distribution of scores for each typeface
separately, another question asked was whether LCs for individuals
tended to vary across typeface in the same manner as the average LCs.
To determine agreement with the average LC pattern, the scores for
each subject were set up as follows: (1) the typefaces were arranged
in order of increasing right hemisphere participation, as shown by
all grades combined, and (2) L05 for the first and last four typefaces
were averaged separately, and the averages then compared. The criter-
ion for a general agreement with the overall pattern was a greater
average score for the first four typefaces than for the last four.

The last column of Table 18 shows the number of subjects per grade

who met this criterion. Nearly all the fourth- and ninth-graders
followed the pattern, and the majority of college students did. The
main reasons for failure to meet the criterion were: (1) a large
number of zero LCs, and (2) fairly large negative LCs for one or more
of 12, T4, and 15, coupled with positive LCs for 18 and/or 13. Most
subjects switched field advantages at least once, although one college
student and one fourth-grader had all non-negative scores, and one
ninth-grader had all non-positive scores. Thus, while the pattern of
average LCs across typeface generally was valid for most individuals,

in very few cases was there an exact match, and some of the individual

141

patterns were very different. As suggested by the regression analy-
sis, the amounts of subject variance and variance unaccounted for

were high.

Tactual condition. As in the visual condition, the distribution

 

of individual LCs in the tactual condition generally followed the
pattern set by the average LCs. With modality order groups combined,
for every typeface the number of subjects with a left hand advantage
was about equal to the number with a right hand advantage, consistent
with the average LCs close to zero found for all three typefaces. The
number of subjects with zero LCs generally was greater than in the
visual condition, especially for 11. For this typeface, over half of
all subjects in the three grades combined had tactual LCs of zero.

With modality order groups considered separately, a group dif-
ference in the distribution of subjects across hand advantage cate-
gories was apparent. A higher frequency of left hand advantage was
found for the VT groups, while a higher frequency of right hand
advantage was found for the TV groups. This modality order differ-
ence was especially strong for 17 and for the fourth-graders for 11,
reflecting the greater modality order difference in average LCs for
those cases.

To examine the agreement between the patterns of individual and
average LCs across typeface, each indivdiual's 11 LC was compared to
the average of his 12 and 17 scores. The criterion for agreement was

a greater LC for 11. Only about half the subjects in each grade met

the criterion, supporting the conclusion of no definite trend across

typeface for average LCs.

142

Relationship of Individual LCs and
Combined ThreShons, Cfirresponding
LCs in Other Modality, and FS

 

'Visual condition. If the correlation between average LCs and A15

 

(average combined threshold or CT) can be interpreted to indicate more
than just a general typeface difficulty effect on right hemisphere
participation, the correlations for individual LCs and 015 within
typeface should parallel the average correlations. These correlations
for individual scores were computed for each grade by typeface com-
bination (see Appendix G). None of the absolute values of the corre-
lations exceeded .49 (all non-significant), and the pattern of
correlations across grade and typeface was not systematic in any
apparent way. Rather than reflecting the negative correlations between
average LCs and A15, the correlations based on individual scores were
positive for 17 of the 24 grade by typeface combinations. Given these
results, the correlations between the average LCs and A15 must be
interpreted with caution.

An additional variable whose relation to individual LCs also
should be considered is the presence or absence of familial sinistral-
ity (FS+ v. FS-). Since there was only one FS+ college student, no
conclusion can be drawn about the effect of F5 for that grade. It
can be noted, though, that the LCs for the FS+ subject followed closely
the pattern set by the average LCs. For the fourth- and ninth-graders,
about half of the subjects were FS+. Average LCs for the FS+ and FS-
groups of both grades were computed for each typeface and compared

using EftEStS (summarized in Appendix H). Of the eight FS+ v. FS-

143

comparisons for fourth-graders, three differences exceeded .10, and
all three were in the direction of a greater positive score for the
FS+ group. Only one difference was significant, though (17;

F54 = .08, FS- = -.64). Of the eight FS+ v. FS- comparisons for
ninth-graders, four exceeded .10, half in each direction. Only one
difference reached significance (12; F54 = -.12, F54 = .48).

Further, for both grades, the absolute values of the average LCs
for the FS+ and FS- groups were about equal, and the pattern across
typeface of the average LCs for each group followed the pattern set
by the overall average LCs. In summary, the difference between FS+
and FS- groups was not great and not apparently systematic, except
that the fourth-grade FS+ subjects tended to show more positive LCs.
One further note is that all three fourth-graders who did not meet
the criterion for agreement with the average LC pattern across type-
face (see Table 18) were FS+, and one of the two ninth-graders who did
not meet the crition was FS+. Thus, possibly other variables related

to FS+ influenced the pattern of scores.

Tactual condition and cross—modal comparison. Correlations
between individual LCs and C15 were computed for each grade by type-
face combination in the tactual modality. As in the visual condition,
about two-thirds of the correlations were positive and the absolute
values of all but one did not exceed .49. The one exception was a
correlation of .66 (p < .05) for the fourth-graders on 11, reflecting
the fact that the subjects who tended toward a ceiling effect on 11

(those with low 015) showed no field difference, while the majority of

144

the other subjects had positive LCs. In general, though, tactual
LCs were not influenced greatly by overall performance level, as
measured by CT.

The effect of F5 on tactual LCs also was examined. Average LCs
for the FS+ and FS- groups were not significantly different for any
typeface for either the fourth- or ninth-graders. Of the six possible
comparisons between FS+ and FS- groups (one for each grade by typeface
combination), three differences exceeded .10, and all three were in the
direction of greater negative LCs for the FS+ group. Therefore,
although FS did not have a large effect on tactual LCs, the weak trend
that did exist was in the opposite direction to the trend found for
the visual condition.

Finally, a positive relationship between individual LCs in the
two modalities was predicted. To test this prediction, cross-modal
correlations between LCs for the same typeface were computed for each
grade by typeface combination and are presented in Table 19. None
of the correlations was significant, but six of the nine correlations
were negative, a trend contrary to the prediction and suggestive of a

dissociation between the modalities.

145

TABLE l9.--Cross-moda1 correlations of laterality coefficients for
each of the three typefaces presented in both modality

 

 

 

conditions
Typeface
Grade
11 12 17
4th .27 -.31 -.03
9th -.29 .06 -.27

C011 .08 -.22 -.23

 

DISCUSSION

Presentation Modality

 

There were two principal hypotheses concerning modality differ-
ences in degree of right hemisphere participation in letter identifi-
cation. First, for both modality conditions, right hemisphere
participation was expected to increase systematically with increasing
typeface complexity. Second, right hemisphere participation was
expected to be greater in the tactual than in the visual condition at
every point of complexity. The first hypothesis, although accurate
for the visual condition, was not supported for the tactual condition.
In fact, summed across modality orders, the average tactual LCs for
the ninth-graders and college students indicated the smallest degree
of right hemisphere participation for the most complex typeface. The
second hypothesis also was not supported. As opposed to showing a
greater right hemisphere participation, the average tactual LCs for
most grade by typeface combinations just showed less asymmetry in
either direction than the visual LCs. When the tactual LCs were
summed across grades, a very weak left hemisphere (right hand) advan-
tage was found for all typefaces.

These results raised two questions. First, why was there a lack
of asymmetry in the tactual condition? In particular, why would a
left hand advantage be found for identifying braille letters, but not

Roman alphabet letters in a complex typeface? And, second, what

146

147

conclusions, if any, can be drawn about modality differences in the

role of right and left hemisphere processing in letter identification?

Reasons for Lack of Asymmetry in
the Tactual Condition

 

 

General methodological factors. A possible explanation for the

 

failure to find a hand asymmetry in the current study is that the
method was not sensitive enough to pick up existing processing asym-
metries. One factor that might have contributed to a lower sensi-
tivity is the use of a unilateral presentation procedure in the
current study as opposed to dichhaptic stimulation. The dichhaptic
procedure was designed to simultaneously activate homologous cortical
areas and thereby hinder interhemispheric transmissions. This procedue
would tend to maximize asymmetry by limiting processing as much as
possible to the hemisphere contralateral to the input. Contrarily,
the unilateral procedure can be assumed to allow greater interhemis-
pheric communication and to permit freer processing of any information
from hand or wrist movements transmitted through ipsilateral pathways.
Witelson (1974, 1977a), LaBreche et a1. (1977), and Oscar-Berman
et a1. (1978) all employed the dichhaptic procedure. However, the
unilateral procedure was used in one study of hand differences for
Roman letters which suggested a left hand advantage (Gardner, 1942)
and in both braille learning studies (Rudel et al., 1974; Wagner,
1976). A direct comparison of the two procedures was made by Flanery
and Balling (1979) using a tactile shape-discrimination task. They

had children and adults feel either a single nonsense form with one

148

hand or two forms with both hands simultaneously, and then judge a
tactually presented comparison form as either the same as or different
from the first form(s). A left hand advantage was found for the
older age groups, and it was of similar magnitude in the unilateral
and dichhaptic conditions.

In summary, examination of the procedures used in these studies
of tactual asymmetry suggests that although use of a dichhaptic pro-
cedure may enhance asymmetry in certain cases, use of a unilateral
procedure would not prevent a significant asymmetry, particularly
an asymmetry in favor of the left hand.

Another factor that might have reduced the sensitivity of the
method is the relatively long presentation time (ranging from 7 to
13 seconds) used for the two more complex typefaces in the current
study. For five studies of hand asymmetry for Roman letters (Witel-
son, 1974, 1977a: Manning; LaBreche et al.; Oscar-Berman et a1.) and
for both braille learning studies, presentation times were under four
seconds (typically, two seconds). These shorter times presumably
limited interhemispheric transmissions, although some amount of trans-
fer still could occur within four seconds. While use of a shorter
presentation time might have increased the likelihood of finding a
significant hand asymmetry, evidence that it was not a requirement
was provided by Flanery and Balling (1979). They obtained a signifi-
cant left hand advantage even though they presented the initial non-
sense form(s) for 10 seconds and the comparison form for 5 seconds.

This hand asymmetry was attributed to asymmetry in the method used to

149

process (or code) information initially by the recipient hemisphere
before that information could be transmitted to the other hemisphere.
- Besides limiting interhemispheric transmissions, the short pre-
sentation times probably were selected to yield an accuracy level of
roughly 50% correct, the level at which the greatest differentiation
between hands is likely to be found. Not all of the studies were
successful at achieving this level. For instance, accuracy level
was about 30% correct in the two studies by Witelson and about 80%
correct in the study by Oscar-Berman et al. In contrast, accuracy
levels in the current study ranged from 40% to 60% correct for the two
more complex typefaces, largely because of the continual adjustment
of presentation time. Thus, from the point of view of establishing
a 50% accuracy level, presentation times were more favorable towards
the appearance of hand asymmetry in the current study than in some
of the other studies.

A third factor that might have affected the sensitivity of the
method is stimulus size. Although the letters used in the current
study (roughly 2 x 1.6 cm) were slightly smaller than the letters
used in all the other studies of hand asymmetry for Roman letters,
they were much larger than the braille patterns (contained within-a
.6 x .4 cm grid). This larger size relative to the braille patterns
might have increased the need for at least some serial (left hemi-
sphere) processing, as opposed to the largely wholistic approach
thought to be favored for braille patterns. However, hand asymmetry

trends were found even with letter stimuli larger than those used in

150

the current study, and a significant left hand advantage was found
for forms 3.8 x 3.8 cm. (Flanery & Balling, 1979). Therefore, as with
the factors of presentation procedure and time, use of a different
stimulus size in the current study might have increased the likelihood
of finding a hand asymmetry, but stimulus size could not be considered

the major reason for lack of asymmetry.

Memory demands. Another possible explanation for the failure to

 

find a hand asymmetry in the current study is that the memory demands
were minimal. Only one letter was felt at a time, and it was iden-
tified immediately after the exploration period. In contrast, the
other studies of hand asymmetry for tactually presented linguistic
stimuli involved a greater memory component. For the braille studies,
the paired associated learning paradigm required that the braille
patterns associated with a particular letter name be recalled over the
course of the experiment. For some studies using a dichhaptic pro-
cedure with active exploration of the letters, memory load was
increased by presenting two pairs of stimuli per trial. The impor-
tance of these memory demands for eliciting asymmetries in the tactual
modality was demonstrated by Oscar-Berman et a1. Using a dichhaptic
procedure, they specified the order in which the subject was to
identify the stimuli. Asymmetries for letters and line orientations
were found only for the hand reported second. Therefore, the differ-
ence between the current study and the braille learning studies in
terms of memory requirements may partly explain the failure to find

a left hand advantage for the more complex typefaces in the current

study.

151

Two points should be made to qualify this conclusion, however.
First, several studies have suggested that hemispheric asymmetries
may be stronger at later stages of information processing (especially
those involving memory) for the visual (e.g., Moscovitch, Scullion,

& Christie, 1976; Dee & Fontenot, 1973) and auditory (e.g., Oscar-
Berman, Goodglass, & Donnenfield, 1974) modalities as well. While
this may be the case, in both modalities there also have been many
demonstrations of significant asymmetries at earlier stages of pro-
cessing for tasks not involving a large memory component (e.g.,
Bryden & Allard, 1976, for vision). Therefore, it should not be
concluded that asymmetries cannot be demonstrated for the earlier
stages in processing tactual information.

Second, if it were found that asymmetries in the tactual modality
are limited to the stages of processing involving memory, the inter-
pretation of that finding would be different than for a similar find-
ing for the visual or auditory modality. The reason is that the
storage of tactual information often involves visual encoding (see
Freides, 1974, for review). This is particularly true for: (l) spa-
tial or form information that the visual modality is more adept at
handling than the tactual modality, (2) letters that are readily
pictured, and (3) tasks with a response mode involving visual recog-
nition (e.g., Witelson, 1974). If asymmetries were found only for
tasks with large memory demands, then, the asymmetries might reflect
only hemispheric differences in visualization processes, not in

processing more purely tactual information.

152

Stimulus visualizability. A third possible explanation for the

 

difference in asymmetry patterns between the braille letter learning
studies and the current study is the amount of visualization likely to
be associated with the two kinds of stimuli. Rudel et a1. (1977)
hypothesized that since most of the evidence for right hemisphere
specialization in the tactual modality came from tasks where vision
was excluded, the involvement of visualization in a tactual task would
reduce or even eliminate a right hemisphere advantage. As Rudel

et a1. (1977) noted, most naive sighted subjects and blind subjects
would not have studied braille visually, and therefore, would not be
able to connect the tactile patterns to visually familiar forms. In
contrast, cut-outs of familiar Roman alphabet letters that were felt
but not seen would be readily visualizable. The visualization theory,
then, would predict less right hemisphere advantage for Roman letters,
and the findings are consistent with the theory.

Specific evidence to support the visualization theory was reported
by Rudel (1979). Seventy-two children learned braille to a specified
criterion using vision and then were asked to pick out the braille
patterns tactually. According to the theory, since a greater amount
of visualization was involved in this task than in the purely tactual
braille studies, less right hemisphere advantage or possibly even a
reversal of asymmetry was predicted. In agreement with this pre-
diction, no significant hand difference in accuracy was found, but
respones with the rjght hand were faster.

While this evidence is convincing, there still is a general

problem with the interpretation of the theory. It is not clear

153

whether it is valid to draw inferences about the effects of visualiza-
tion from the effects of an actual visual component in the task, as
Rudel et a1. did in the original formulation of the theory. If visu-
alization can be considered similar to visual participation, then
the identification of Roman letters in the current study may be con-
sidered in some sense a cross-modal matching task. It is likely that
recognition of the tactually-perceived letters was aided by comparing
them to stored visual representations. In that case, the findings
of Butters et a1. (1970) suggesting a left hemisphere involvement
in cross-modal matching add further weight to the idea that visualiza-
tion contributed to the failure to find a right hemisphere advantage
for the more complex typefaces.

Two sets of results, however, suggest that the interpretation
of the theory of visualization may have to be modified to adequately
explain the lack of asymmetry in the current study. First, Witelson's
(1974) finding of a left hand advantage for the perception of nonsense
shapes was cited by Rudel et a1. as evidence that a right hemisphere
advantage was found when "vision was excluded." But, while explora-
tion was purely tactual, responses were made by selecting the correct
forms from a group of forms presented visually. This response mode
would seem to encourage visualization, and to involve cross-modal
matching.

Second, the modality order effect in the current study was in
the opposite direction to that predicted by the visualization theory.

According to that theory, if right hemisphere participation were

154

greater in one modality order, it should be the TV order, because
subjects in that group had no prior visual exposure to aid in visual-
ization during the tactual condition. Instead, right hemisphere par-
ticipation was significantly greater for the VT order. While this
modality order difference could be explained by a general learning
effect or strategy change, the pattern of results for the specific
typefaces suggests otherwise. The two simpler typefaces in the
tactual condition were fairly common styles, but 17 was a unique,
ultra-modern style (rated 9.1 on a 10-point scale of unfamiliarity),
which would not be readily visualizable. The order effect was much
larger for 17 than for the other typefaces (both in terms of group
means and distribution of individual subjects), suggesting that the
main reason for the effect was an increased ability to visualize the
typefaces after seeing them in the visual condition. Thus, contrary
to the visualization theory, increased visualization led to a greater
right hemisphere participation.

One way to modify the visualization theory to explain these
conflicting results is to take into account the stimuli and task used
and the nature of the visualization involved. In the current study,
Roman letters, even if unfamiliar in style and presented tactually
first, required at least some visualization because they had to be
matched to the stored visual images of essential letter elements in
order to be identified. For instance, when 17 was presented tactually
first, the visualization may have been relatively weak and based on

specific salient features. This type of cross-modal matching may have

155

been relatively left-hemisphere dependent. When 17 was presented
tactually after being seen first, more wholistic, typically right
hemisphere visualization may have occurred, thus resulting in a
relative increase in right hemisphere participation.

In contrast to Roman letters, braille letters in the tactual
learning studies did not necessarily have to be matched to stored
visual images in order to be identified. When braille letters were
taught visually first, however, any number of visualization strate-
gies could have been used. Given the greater spatial resolving
power in the visual modality, the braille dots might have been counted
and localized much more specifically than in the tactual modality,
thus encouraging a more analytical or possibly even verbal encoding.
In that case, greater left hemisphere participation than in the
purely tactual studies would be expected.

Finally, in contrast to both types of letters, visualization
requirements did not affect hand differences for nonsense shape
matching. A left hand (right hemisphere) advantage was found whether
the palpated shapes were matched to visually presented comparison
forms (Witelson, 1974, 1976) or another tactually presented form
(Flanery & Balling, 1979). This lack of a visualization effect can
be attributed in part to the nonlabelable character of the shapes,
and the wholistic quality of the visualizations associated with them.

In summary, the greater involvement of visualization in the
identification of Roman letters compared to braille letter learning
might account in part for the difference in asymmetry patterns. How-

ever, the effects of visualization in these and related studies of

156

hand asymmetry are more complex than previously hypothesized and
need to be considered in interaction with the particular tasks and

stimuli used and the character of the visualizations they evoke.

Stimulus appropriateness for the tactual modality. The type—

 

faces and specific letters used in the current study and the type-
face characteristics that were varied were selected on the basis of
ratings and evidence from the yjspal_modality. In contrast, braille
was designed originally to facilitate reading by touch and so was
based on characteristics that influenced the efficiency of tactual
perception. This comparison suggests a fourth possible explanation
for the lack of a tactual asymmetry in the current study. It could
be that the more complex typefaces did not differ from plain print
enough or on characteristics that would affect right hemisphere par-
ticipation in tactual information processing. The stimuli may have
been inappropriate on three levels: (1) the specific typefaces and
letters chosen, (2) the characteristics that served as a basis for
typeface selection, and (3) the nature of Roman alphabet letters in
general.

First, the eight letters used were chosen according to their
ratings for visual confusability and their past ability to elicit
visual field asymmetries. Given that characteristics differ in
saliency for vision and touch (Goodnow, 1969) and that letter diffi-
culty ratings in the two modalities generally are only moderately
correlated (Loomis & Apkarian-Stielau, 1976), modality differences

in the relation between the letters as well as in their processing

157

might be expected. In fact, the cross-modal correlations of letter
difficulties for each of the three typefaces were low, and, for two
typefaces, negative. Further, considering only the letters for which
there was both a hand and visual field difference in accuracy of at
least 5%, the majority showed opposite sensory field differences for
the two modalities. These findings indicate that the letters probably
do not vary along similar continuums for vision and touch, and that

at least some of the letters are processed differently in each modal-
ity. Therefore, it is not unlikely that the use of different letters
would have altered the pattern of hand asymmetry.

The eight typefaces used also were chosen according to ratings
for visually presented stimuli. Following the same reasoning used to
predict a modality difference in the relation between letters, it also
might be expected that the distance between typefaces on the character-
istics measured would differ for vision and touch. For instance, for
a typeface to be rated 6 or 7 on a scale of tactual scriptlikeness
might require simpler, bolder lines but more slant and curve than an
equivalent rating for visual scriptlikeness. Therefore, if the type-
faces used in the tactual condition had been chosen according to
ratings given to tactually presented typefaces, a left hand advantage
might have been found.

While this possibility always exists, there are at least two
reasons to suggest that the effect of using tactually rated type-
faces would have been limited. The first reason is that the three

typefaces used in the tactual condition differed significantly in A1.

158

Given the inherent relationship between A1, difficulty (naming
latency), and general complexity, at least some variation on the
characteristics of interest must have been achieved for the tactual
condition. The other reason is that since all typefaces were

designed for visual presentation, many of them involve discriminations
of fine lines and curlicues beyond the capabilities of the tactual
modality. Therefore, it might not have been possible to find type-
faces for which the variation in tactual ratings on the characteristics
considered would have been greater.

The second level on which the stimuli might have been inappro—
priate pertains to the characteristics that were used to select them.
The characteristics were those found by Bryden and Allard (1976) to
be significantly correlated with visual field differences. There is
no reason why those characteristics also should have been related to
hemiSpheric differences for processing tactual information, especially
in the case of "scriptlikeness." Visually, scriptlikeness is one of
the major ways that reading material is differentiated. Cursive writing
is a subject taught in school, and it is used nearly exclusively by
most adults, who typically develop their own unique style. In con-
trast, cursive writing has no special significance for the tactual
sense, apart from a possibly greater connection with the kinesthetic
feedback from scanning letter outlines. Changing the characteristics
on which the typefaces were varied might well have altered the pattern
of hand differences. However, characteristics more meaningful for the

tactual sense might be hard to find.

159

Finally, the third level of possible stimulus inappropriateness
pertains to the nature of Roman alphabet letters in general. Possi-
bly the fundamental nature of Roman letters is such that no variation
in typeface could induce the right hemisphere to play a major role
in the processing of letters perceived by touch. The perceptual abil-
ities needed to differentiate the letters are not those at which the
tactual modality is particularly adept. For instance, it might be
difficult for the hand to pick up the small lines that differentiate
a G from a C or a Q from an O, and to feel spaces contained within
fairly small contours(such as with R and 8). Increasing the irregu-
1arity of the letters (in terms of size, slant, etc) by varying the
typeface would seem to increase the relative inappropriateness for the
tactual modality. In contrast, braille letters, which did require a
large degree of right hemisphere processing, always are contained
within a constant framework of six dot positions, with spacing appro-
priate for the spatial resolving power of the tactual modality.

In summary, all four explanations mentioned--insensitivity of the
method, insufficient memory demands, necessity of matching the letters
to stored visual images, and inappropriateness of the typefaces for
the tactual modality-~could have contributed to the lack of asymmetry
in the tactual condition. None of the explanations is without limita-
tions, however, and it is unlikely that one of the factors alone would
have prevented an asymmetry from occurring. It is possible to test
the influence of most of the factors by altering the method accord-

ingly. A dichhaptic presentation technique, shorter presentation

160

time, and longer interval between presentation and response could

be used, along with typefaces and letters selected on the basis of
tactual criteria. In addition, the effects of matching to a visual
image and of the fundamental nature of Roman alphabet letters could
be tested by examining hand differences for learning the association
of nonsense letter names with shapes consisting of Roman letter

features.

Modality Differences in Processing

 

Although it may be questionable whether inferences about cogni-
tive processing can be drawn from the absence of a significant hand
difference for all grade by typeface by modality order combinations,
some speculations do seem to be warranted about the striking modality
difference in field asymmetries (reflected in the significant modal-
ity by typeface interaction). If the experiments suggested in the
preceding subsection to test the effects of the four possible explana-
tions for lack of tactile asymmetry all failed to reveal hand differ-
ences in any condition, speculations about the results of the current
study would be much stronger. However, even if significant hand
differences were found in some conditions, the fact that they were
not found in the current study with method variables consistent across
modalities suggests at least some degree of modality difference in
processing. Method variables were roughly equalized across modali-
ties in the current study partly by using many of the same variables
in both conditions, such as typefaces varying in complexity, uni-

lateral presentation, staircase procedure, identification response,

161

and low memory demands (to avoid bias due to the supposed lower reten-
tion accuracy in the tactual modality). Part of the attempt to
equalize the method across modalities also involved changing some
variables according to the special qualities of the tactual modality,
such as using larger letters and longer exploration times than in the
visual condition to match the poorer spatial resolving power and
slower information gathering tempo attributed to touch.

The direction of the modality difference for the simplest, block-
printed typeface was in general agreement with the prediction. The
significant RVF advantage (averaged across grades) was expected, given
the significant RVF advantage found for most previous studies of
letter identification in adults (e.g., Bryden, 1973) and a similar
trend found for studies of asymmetry for verbal material in children
(e.g., Marcel & Rajan, 1975). The non-significant hand difference
was not out of line with previous studies, most of which also found
no hand asymmetry for printlike letters. Witelson's (1974) sugges-
tion that linguistic information presented tactually is first analyzed
in a spatial code (right hemisphere processing) and then translated
to a linguistic code explains the pattern of results very well. It
can be added that the general reason for this modality difference
in asymmetry is that the tactual modality is associated with more
right hemisphere processing because of its peculiar nature, poorer
spatial resolution, and so forth.

The problem with this interpretation of the processing of tactu-

ally perceived letters is its implication that when the demands on

162

the spatial coding aspects of the task are increased, right hemi-
sphere participation should increase (or a shift toward a left hand
advantage should occur). In the current study, it is likely that
the demand for the spatial or form analysis of the letters was
increased by increasing typeface complexity, yet, contrary to the
implication of the theory, no change in hand difference was found.

At least two elaborations of the theory to account for this find-
ing seem possible. First, the stage of letter processing roughly
attributed to the left hemisphere (that of translation into a lin-
guistic code) might have an increased function with increasing letter
complexity that coincides with the increase for spatial processing.

A constant hand difference thus would be maintained. For instance,
perhaps the hand cannot break down or "normalize" the more complex
letters to the extent that is possible with the eye. In that case,
translating into linguistic coding by matching the letters to stored
visual images would be much more difficult than with the simpler
letters. One problem with this suggestion is that the left hemisphere
might not be able to analyze forms that could not be broken down fur-
ther by the right.

A second possible elaboration of the theory is that the I'spatial
coding" of tactually perceived letters actually involves a greater
need for both hemispheres acting in combination than does the spatial
processing of visually perceived complex letters. It might be the
case that the ratio of right to left hemisphere processing in the

visual analysis of complex typefaces cannot be compared to the ratio

163

in the tactual analysis of letters of any typeface. The visual sys-
tem is finely tuned for letter analysis. The right hemisphere can
readily pick out regularities across a wide variety of slants and
extra details and then break down the letters into their simple forms.
In contrast, perhaps in the tactual system, the "spatial coding" of
letters involves more of a building up process. This would entail
use of the probably left—hemisphere dominated image of the letters,
possibly verbal instruction, and a serial integration of points, as
well as right hemisphere skills. The processes used, then, would be
very different from those used in the visual modality, which might
explain the tendency toward negative cross-modal correlations of
LCs.

At a minimum therefore, if any inferences can be drawn regarding
modality differences in Roman letter processing, the tactual modality
does not require greater right hemisphere participation, and, in fact,
might require greater c00peration of the hemispheres than the visual

modality.

Stimulus Characteristics
The principal hypothesis about stimulus characteristics was that
LVF and left hand scores (right hemisphere participation) would
increase relative to RVF and right hand scores (left hemisphere par-
ticipation) as overall typeface complexity was increased. A secondary
hypothesis was that the particular typeface characteristics of diffi-
culty, scriptlikeness, and confusability all would be related to field

differences, and that scriptlikeness would be the strongest predictor.

164

The principal hypothesis was not supported when the letters were
presented tactually--no significant effect of typeface on hand asymme-
try was found. Possible reasons for this lack of a typeface effect
were discussed in the preceding section in conjunction with reasons
for the failure to find a significant hand asymmetry in general. Since
the lack of typeface and hand effects in the tactual condition pre-
cluded any more detailed analysis of the effects of particular type-
face characteristics, the results for that condition will not be
discussed further.

For the visual condition, the principal hypothesis was supported,
as indicated by the highly significant typeface effects on visual LCs.
The general pattern of visual field asymmetries across typeface for
all grades combined closely followed the predictions, which were
based in large part on Bryden and Allard's (l976) findings. The
pattern found was a significant RVF advantage for the simplest
typeface, no visual field differences for the medium complex type-
faces, and a significant LVF advantage for two of the most complex
typefaces. Also, as predicted, this general increase in right hemi-
sphere participation with increasing typeface complexity was found
for each grade separately. However, the increase was more irregular
for the fourth- and ninth-graders than for the college students.
Further, this irregularity corresponded with a significant grade
difference in overall level of right hemisphere participation which
was in the direction opposite to that predicted. The fourth- and

ninth-graders showed significantly less right hemisphere participation

165

than the college students, particularly for the medium complex type-
faces and the typeface with the highest complexity rating.

The secondary hypothesis, based on Bryden and Allard's findings
for adults, was tested in the current study by regression analyses and
examination of the pattern of correlations between visual LCs and
specific typeface characteristics for each grade separately. The
hypothesis was supported only for the college students. Their visual
LCs were significantly predicted by a combination of the specific
typeface characteristics, of which scriptlikeness was the strongest
predictor. For the younger grades, tests of the prediction equations
did not reach significance, and difficulty was a more important pre-
dictor than scriptlikeness.

The discussion of these results will be centered around two main
questions. First, what does the typeface effect imply about the rela-
tive participation of the two hemispheres in the letter identification
process in adults? Particularly since the results for college students
were in nearly complete support of Bryden and Allard's findings, what
(if any) additional conclusions about letter processing can be drawn
from the current study? And, second, what are the implications of
the grade differences in general level of right hemisphere partici-
pation and in relative importance of the specific typeface character-
istics with respect to: (l) the letter identification process, and
(2) the development of modes of processing in the two hemispheres and

strategies for their use.

166

Implications for Letter Processing
in Adults

 

Possible models for participation of the two hemigpheres. Bryden

 

and Allard's results allow for at least two slightly different models
of the roles of the two hemispheres in the letter identification
process. The first possibility is that a predisposition exists toward
processing letters as linguistic elements, thereby primarily acti-
vating the left hemisphere. The right hemisphere would be expected
to take control over the processing only when the pattern recognition
capability of the left hemisphere was exceeded. According to Jonides
(1979, p. 425), the overall low accuracy level in Bryden and Allard's
task, especially for the typefaces that yielded a LVF advantage, sug-
gests that the visual discrimination involved was difficult enough

to warrant the intrusion of a right hemisphere feature extraction
mechanism. Since an asymmetry in favor of the RVF was found for the
majority of typefaces, it can be assumed that the bias toward the
left hemisphere was operating except when the most difficult type-
faces were presented.

Several studies of both normal and brain-damaged adults provide
support for the premise that the letter recognition task must require
perceptual abilities beyond the capability of the left hemisphere in
order for the right hemisphere to participate. In fact, most demon-
strations of a LVF (right hemisphere) advantage for letter identifi-
cation or classification have involved tasks for which the perceptual

demands have been increased in various ways.

167

First, two studies found that right brain-damaged patients were
imparied relative to controls and to left brain-damaged patients on
tasks requiring the recognition of letters presented in perceptually
complicated forms. Warrington and James (1967) had their patients
identify (or show to have recognized) letters that were 30% or 70%
filled in. Patients with right parietal damage were significantly
worse than controls and all other brain-damaged groups. Faglioni
et a1. (1969) used four letter recognition tests in which patients
had to match a test letter to the corresponding one(s) among ten mul-
tiple choice letters presented in a different graphic form (in
italics or lower case). For the first two tests, considered per-
ceptually easy, the test letters were in a regular form. For the
third test, the test letters were incomplete, and for the fourth test,
they were partly hidden (crossed—out) by nonsense lines. Right brain-
damaged patients with visual field defects were significantly impaired
relative to all other groups on the last two tests, considered to be
perceptually difficult. This difference was eliminated when scores
on two spatial-perceptual tests were used as covariates. These results
suggest that right hemisphere participation in letter recognition is
determined by whether the letters are perceptualy complicated
enough for right hemisphere spatial-perceptual processing abilities
to become critical.

Second, two studies with college students found a LVF (right
hemisphere) advantage for letter processing when visual masks were

added to increase the complexity of the visual discrimination involved.

168

To a task requiring the identification of easily-perceived single
lateralized letters, Hellige and Webster (1979) added a visual
pattern mask presented at varying intervals before and after the
letter. Recognition accuracy was reduced at forward masking inter-
vals less than about 40 msecs. and backward masking intervals less
than 80 msecs. For forward masking and short backward masking
intervals, at which the letter and mask most likely were treated by
the visual processing system as having been simultaneous, a LVF
(right hemisphere) advantage for letter identification was found.
For longer backward masking intervals, at which ongoing processing
was disturbed although the letter and mask most likely were regarded
as separate stimuli, no field asymmetry was found. An implication
of these results is that the right hemisphere contribution to letter
recognition depends on increasing the difficulty of the initial extrac-
tion of relevant visual information.

A similar enhancement of right hemisphere contribution with the
addition of a mask was found by Polich (1978) using very different
experimental conditions. As apposed to letter identification, Polich
used a classification task which required the subjects to push one
of four response buttons corresponding to the four possible stimulus
letters. In the "masking" condition, outlined squares were presented
on the left and right borders of the letter, but were not spatially
overlapping. A LVF (right hemisphere) superiority in speed and
accuracy of classification was found for letters that were masked.

No field difference was found in the unmasked condition. Thus, these

169

results again show that perceptual complexity triggers reliance on
right hemisphere modes for processing letters.

Third, Jonides (1979) demonstrated that the perceptual complex-
ity involved in the letter processing task, and hence right hemi-
sphere contribution, could be manipulated simply by varying the
physical similarity of the letters to be distinguished. He found a
RVF (left hemisphere) advantage in reaction time for the EF vs. CG
classification, but a LVF (right hemisphere) advantage for the per-
ceptually more difficult CE vs. FG classification. These results seem
to indicate that right hemisphere participation depends not only on
the difficulty of extracting the relevant information, but on how
salient the information is relative to other response alternatives
stored in memory.

Finally, in an attempt to bias letter processing toward the
right hemisphere, two studies employing Posner's same-different letter
matching paradigm with lateralized stimuli increased the difficulty
of the lateralized letter pair. Hellige (1976) increased the per-
ceptual difficulty of the task by placing a grid of dark lines on the
viewing screen, and Umilta, Sava, and Salmaso (1980) used scriptlike
letters. The expectation in both cases was that when the two letters
had the same name but were in different cases, the typically found RVF
advantage in R1 for same judgments would be reduced, or even changed
to a LVF advantage. With the overlapping grid, Hellige (1976) found
a significant LVF advantage in accuracy, but not in R1. With the

scriptlike letters (Umilta et al., 1980), the RVF advantage actually

170

increased compared to the identical condition with printlike letters.
One suggestion of these findings is that right hemisphere involve-
ment depends more on the difficulty of perceiving the information
than on the scriptlike character of the letter.

The second possible model of right hemisphere participation in
letter recognition differs from the first in that it assumes the
processing of all or most letters consists of a combination of right
and left hemisphere modes. It is not clear exactly what parts of the
letter recognition process are the special domains of each hemisphere.
According to general theories (e.g., Smith & Spoehr, 1974) letter per-
ception can be thought of as containing two main stages--initial
extraction of the features of the input letter and interpretation.

The interpretation stage can be broken down into at least three separ-
ate Operations involving: (1) matching features of the input to
stored representations of letters, (2) deciding on the final cate-
gorization of the input, and (3) translating the letter to an acoustic
code. While the right hemisphere is likely to be associated with
feature extraction and the left hemisphere with acoustic coding, the
process of finding a stored visual representation that best matches
the perceived features of the input might be associated with either
hemisphere. If, as would be appropriate with non-printlike letters,

a "fuzzy logical model of letter identification" (Oden, 1979) was
operating, then the gggrgg to which each feature was present in the
input also would be coded. In this case, the subtle perceptual dis-
tinctions of which the right hemisphere is capable would be especially

important in finding a match for the input.

171

For this second model of right hemisphere participation, the
relative contribution of the two hemispheres (reflected in visual
field asymmetry) depends on which part of the recognition process is
most crucial or limiting. One way of increasing the contribution of
the right hemisphere would be to increase the difficulty of extract-
ing the relevant visual features, as was demonstrated in several
studies discussed in relation to the first model. However, it also
would be possible to increase the right hemisphere contribution in
other ways, such as by increasing certain demands placed on the opera-
tions of matching the perceived features to stored representations,
and categorizing them. For instance, the features of a handwritten
letter might be easy to perceive but difficult to classify, because
of their inexactness and irregularity. Or, as might be true of
Jonides' findings, if there was a high probability that the input
letter would fall into one of two physically similar categories,
extra care would be needed at the categorization stage. Then either
a greater emphasis would be placed on those operations that the
right hemisphere already was responsible for, or the right hemisphere
would begin dominating those Operations.

While Bryden and Allard's results could be interpreted to support
either of the two models of right hemisphere participation, the main
ways in which those results were extended in the current study seem to

favor the second model.

Extensions of Bryden and Allard's findings. First, Bryden and

Allard's results were extended with respect to the accuracy level at

172

which the typefaces were presented. The generally low accuracy
(about 35% correct) and the strong negative correlation between
average typeface accuracy and right hemisphere participation found

by Bryden and Allard invite explanations (Hi the LVF advantage for
certain typefaces in terms of increased perceptual difficulty. In
other words, the letter outline might have been so faint, or the
shadow lines so confusing, that the effects could have been similar to
those for masked or incomplete letters. The correlation between
accuracy and visual field asymmetry was especially evident for Palace
Script, the typeface Bryden and Allard's subjects rated as most
scriptlike. When this typeface was presented in alternation with
other typefaces, it showed by far the strongest LVF advantage, but
also the lowest accuracy (about l0%). When presented alone at a much
longer duration, accuracy was increased slightly to 33% correct, and
the LVF advantage was no longer significant.

In the current study, accuracy for all typefaces except T3 was
about 50% correct. The fact that a typeface effect on visual field
asymmetry was found even though a constant, moderate level of accuracy
was maintained seems to contradict the premise of the first model that
the right hemisphere intrudes on preferred left hemisphere processing
only when the difficulty of the pattern recognition required exceeds
the left's capacity. As Bryden and Allard found, the strongest LVF
advantage was found for the most scriptlike typeface, but the associa-
tion in this case was less readily explained by low accuracy or

increased perceptual difficulty. In addition, even though accuracy

173

was higher than in Bryden and Allard‘s study, the tendency toward

LVF advantage was more wide—spread in the current study. For
instance, Bryden and Allard found no field asymmetry for Old English,
while the current study found a nearly significant LVF advantage for
that typeface. Although the difference could be caused partly by the
way in which the typefaces were selected (complexity was a factor in
the current study), it also indicates that the balance of processing
for a given typeface is flexible, as opposed to requiring right
hemisphere processing beyond a certain difficulty level.

Second, Bryden and Allard's results were extended with respect
to the correlations between specific typeface characteristics. In
Bryden and Allard's study, the correlation (Pearson r) of scriptlike-
ness with laterality indices was at least slightly higher than for the
other characteristics, and "multiple correlation procedures indicated
that little was added to the predictability of laterality scores by
supplementing the script-to-print measure" (p. l96). However, given
the high correlations between characteristics, and the fact that all
characteristics except familiarity were significantly correlated with
laterality indices, no justifiable conclusions could be drawn regard-
ing the superiority of scriptlikeness as a predictor.

In the current study, the typefaces were chosen so that the
correlations between characteristics would be as low as possible.
Therefore, a firmer conclusion can be drawn about the importance of
scriptlikeness from the findings that: (l) scriptlikeness accounted

for a large percent of the variance in average LCs regardless of the

l74

order in which it was entered into the regression analysis, and
(2) scriptlikeness and confusability were significantly correlated
with average LCs, while difficulty was not (provided T3 was included).
The conclusion that scriptlikeness was a better predictor of increased
right hemiSphere contribution than difficulty tends to support the
second model's characterization of letter recognitibn as a delicate
balance of the processing of both hemispheres, as opposed to an intru-
sion of right hemisphere processing when difficulty becomes severe.
Possibly some type of letter processing "regulator" adjusts the degree
of right hemishere analysis involved in certain operations, such as
feature matching or defining perceived feature "goodness of fit" to
idealized feature. In general, it makes sense that scriptlikeness
should serve as an important cue for this "regulator" to shift toward
more right hemisphere processing, since scriptlike letters typically
are inexact and flourished, and therefore require a more pattern-
oriented approach. One possible inconsistency in this association of
scriptlikeness with right hemisphere processing is that if handwriting
is mainly controlled by the left hemisphere (for nearly all right-
handers), it might be expected that the perception of cursive writing
also should be left-hemisphere dominated. The explanation for this
inconsistency is not clear, although a comparison of visual field
asymmetries for one's own handwriting as opposed to general script-
like letters might be relevant.

The third extension of Bryden and Allard's results was the analy-

sis of whether an attentional set was formulated over the course of the

175

experiment. The fact that opposite field advantages were found for
typefaces presented in an alternating fashion suggests that no
overall set was formed in Bryden and Allard's study, and that the
balance of processing could shift fairly readily. However, accord-
ing to Jonides, the right hemisphere could have been intruding on a
left hemisphere set when accuracy became very low. In the current
study, although the formation of a set for a given typeface was
facilitated by grouping the trials for each typeface, analysis of the
change in asymmetry over the course of the experiment indicated that
neither hemisphere was favored. Rather, the dominant asymmetry for
each typeface was strengthened in the second half of the experiment.
This pattern of results seems to support the second model in that it
could be assumed that some "regulator" was able to readily adjust the
balance of right and left hemisphere processing. Further, this regu-
lator became progressively more adept at determining the most effi-
cient processing to emphasize for a given typeface.

Implications for Letter Processing
in Children

 

The lesser degree of right hemisphere participation shown for
children compared to college students seems to be more easily inter-
pretable in terms of the second model than the first. In fact, the
first model tends to predict the opposite finding based on the follow—
ing two points: (l) according tO'UMemodel, the right hemisphere con-
tributes significantly to letter processing only when the perceptual

discriminations required exceed the left hemisphere's capacity, and

176

(2) if anything, the perceptual discriminations involved in process-
ing non-printlike typefaces should be more difficult for children

than for adults, thus making the intrusion of right hemisphere process-
ing more likely for children.

According to the second model, the smaller contribution of the
right hemisphere for children can be explained in at least two ways,
both having to do with the relative efficiency of left and right
hemisphere modes. First, it may be that the Spatial, pattern recog-
nition abilities for which the right hemisphere typically is spe-
cialized are not fully mature for fourth-grade (9 yrs. of age), and
undergo a maturational decline at the ninth-grade (l4-year-old) level.
In that case, emphasis on a less than fully efficient right hemi-
sphere mode, particularly for medium complex typefaces requiring
relatively little pattern analysis, would not facilitate the letter
recognition process for the younger grades.

Evidence to support Unapremise of an increasing maturation of
right hemisphere perceptual abilities until at least age lO is pro—
vided by studies of face recognition and braille letter learning.
Leehey (l976) found that the usual adult pattern of a LVF (right
hemisphere) advantage for the perception of unfamiliar, upright faces
did not appear until age l0, and subsequently declined at about age l4
for boys. These findings coincide with evidence that overall ability
to encode upright faces improves until about age lO, with a dip in
performance during a period in adolescence, and then further improve-

ment until college age (e.g., Carey, Diamond, & Woods, 1980).

177

Similarly, Rudel et al. (l977) found that a left hand (right hemi-
Sphere) advantage for discriminating braille patterns did not appear
until about age ll in boys, and then became weaker and non-
significant for l3- and l4-year-olds.

While the weight of the evidence supports the idea of a matura-
tion of specific right hemisphere skills peaking at about age lO, an
interpretation in terms of changes in preferred strategies with age
must also be considered. For instance, the reduced left field
advantage found at about age l4 could be interpreted as a tendency to
rely on verbal, analytic codes at that age, rather than as a matura-
tional decline in ability. Further, an interaction of those two
factors is highly likely.

Therefore, if this first explanation of the grade difference is
accepted, the results of the current study could be summarized as
follows: (l) When the demand for right hemisphere perceptual dis-
crimination skills was marginal (i.e., the typefaces were only
moderately complex), the younger grades either did not have the subtle
perceptual skills available to make the use of that mode an effective
strategy if another could be used, and/or they preferred another
strategy; and (2) When the demand for perceptual discrimination
skills was great (for T6 and T7), emphasis on right hemisphere proces-
sing probably aided letter recognition for all grades, even though
right hemisphere capacity might have been weaker for the younger
grades compared to college students. Thus all grades showed about

equal LVF advantage for T6 and T7.

178

A second way in which the smaller contribution of the right
hemisphere for children can be explained is with respect to a change
across grades in verbal, left hemisphere skills. Possibly the lin-
guistic, phonetic analysis skills of the younger grades (especially
fourth-graders) were less efficient than those of the college stu-
dents. This would cause left hemisphere processing to be a more
limiting factor in letter recognition for the younger grades, thereby
resulting in a greater tendency toward RVF advantage for those grades.
To explain the LVF advantage found for the children for T6 and T7, it
could be assumed that left hemisphere processing was the most limiting
factor for them until the demand for right hemisphere pattern dis-
crimination skills became relatively great.

While no other developmental study involving identification of
unilaterally presented verbal stimuli has found a similar decrease in
RVF advantage with age, one study involving letter matching did find
tentative results consistent with the current study. In this study
(Witelson, l977a) pairs of vertically arranged uppercase letters were
unilaterally presented to boys ranging from 6 to l4 years of age,
and the response required was "same" or "different." This task
represents a lateralized paradigm of the "physical match" condition
of Posner's reaction time studies and is one for which at least a
moderate LVF advantage in RT has been found for adults (e.g., Cohen,
l972). Contrary to past findings for adults, a significant RVF advan-
tage in accuracy was found for the youngest subgroup (6- to 7-year-

olds), and there was a tendency toward RVF advantage until age lO.

179

These results, as they pertain to the current study, suggest that
letters are highly associated with left hemisphere, linguistic proces-
sing for children, either because the relative immaturity of their
linguistic skills results in a greater concentration on that type

of processing, or because it is a preferred strategy.

Another study (Reitsma, l978) that could be interpreted to
support the explanation for the grade difference in the current study
in terms of changing linguistic skills with age also used Posner's
same-different letter matching task. First-, second-, and sixth-
graders were presented in central vision with pairs of letters that
were the same or different in terms of name and/or physical appear-
ance (case or typeface). It was hypothesized that since matches on
the basis of name similarity would involve a deeper level of analysis
and encoding than matches made purely on the basis of physical fea-
ture similarity, RT for name matches would be longer than for physical
matches, but the difference would decrease with age. This hypothesis
was confirmed for each grade interval, and was interpreted on one
level to indicate an increase with age in the "efficiency or automa-
ticity of encoding of visual letter information."

While the encoding that occurs between name and physical matches
cannot be attributed solely to left hemisphere processing, the left
hemisphere may play a large part in that encoding. Therefore, in
terms of the current study, possibly the left hemisphere contribution
to the letter recognition process becomes increasingly "automatized"

beyond the fourth-grade level, thereby permitting an increasing

180

integration with efficient right hemisphere analysis of perceptual
details.27 Again, an alternative or complementary explanation to the
change in linguistic skill with age is a change in favored processing.
In general, the age differences in right hemisphere participation
found in the current study are likely to be associated with both
explanations described. That is, letter processing in children proba-
bly is influenced by the maturing beyond age 9 of both_left and right
hemisphere skills and by changes in preferred strategy complementary
to this cognitive development. However, the age differences found in
the current study seem to be inconsistent with the main explanation
for age trends expressed by previous developmental studies of lateral-
ity for visually-presented linguistic stimuli--that of increasing
specialization of the left hemisphere for verbal material. Contrary
to the idea that sequential (left hemisphere) processing of verbal
material becomes more accentuated with age (Carmon et al., l976), or
that lateralization for the printed word may increase through adoles-
cence (Tomlinson-Keasey et al., 1978), the current study suggests
that increasing proficiency with written language is correlated with
an increasing tendency to integrate right hemisphere perceptual skills
with phonetic coding in the left hemisphere (at least for college
students compared to fourth-graders). This suggestion receives added
weight because the effects of two factors often confounded with age
differences in the past--difficulty level and post-exposural scanning
tendencies--were minimized (see Witelson, l977b, for review). The

two opposing developmental trends need to be tested further using

181

tasks in which a constant accuracy level across grades can be main-
tained and in which the stimuli are either single letters or verti—
cally arranged strings.

Finally, the finding of a grade difference in the importance of
scriptlikeness as a predictor of visual field asymmetry fits easily
into the second model of right hemisphere participation in letter
processing. Before being considered in terms of the model, however,
it should be noted that the typefaces were rated on all the specific
characteristics only by college students. It could be that the child-
ren would rate the typefaces very differently, and that these ratings
would be significantly correlated with their laterality coefficients.
If so, perhaps there is a "regulator" of the balance of hemispheric
processing that tips the balance at the same relative level of script-
likeness (or other characteristic) at each grade, although criteria
for judging the levels vary across grade. If not, possibly the cues
that signal the most efficient balance of processing for a given
typeface are learned with age and increasing experience with different
lettering styles. Thus, fourth- and ninth-graders might not have
learned to use scriptlikeness as a cue that greater right hemisphere
participation would lead to more efficient letter processing. By
college age, though, the hypothesized "regulator" would have become
more flexible and responsive so that it could readily alter the
degree of participation of the two hemispheres, both of which (accord-

ing to the second model) contribute to letter processing.

ENDNOTES

1Letter symmetry was found to affect accuracy of recognition
when letters were presented in groups in both the visual (Fudin,
Garcia, & Solomon, l975) and tactile (Craig, l976) systems. How-
ever, with visual presentation of single letters, no difference in
accuracy for symmetrical and asymmetrical letters was found (Bryden,
1968). Thus, it will be inferred in this dissertation that lateral
symmetry of letters presented alone either visually or tactually is
not a crucial dimension to consider.

2Sex differences in asymmetry were found in two of the eight
tactual presentation studies. Witelson (l977a) found no hand asymme-
try for girls, compared to a marginal right hand advantage for boys.
Cioffi and Kandel (l979) found that girls showed a right hand advan-
tage for consonant bigrams, while boys showed a left hand advantage.
Thus, in general, since girls showed a smaller right hand advantage
in one study and a greater right hand advantage in the other, the
overall conclusions of the summary would not be changed by adding
the results for girls. The results for boys only will be emphasized
in this section to simplify the summary and because the subjects in
this dissertation will be male.

31h an earlier experiment (Manning, Goble, Markman, & LaBreche,
l976, cited by LaBreche et al.), it was shown that line drawings of
manually represented English letters were better recognized in the
LVF when congenitally deaf subjects responded by fingerspelling. In
addition, a non-significant tendency toward better LVF recognition of
pictures of American Sign Language (ASL) words was found for deaf
subjects using a matching procedure (Manning, Goble, Markman, &
LaBreche, l977). Other studies with deaf subjects showed no visual
field asymmetry for ASL words or manual letters (McKeever, Hoemann,
Florian, & VanDeventer, l976; Phippard, l977; Manning et al., l976,
with a written response). With hearing subjects, a significant LVF
advantage for the recognition of ASL words and manual letters was
found (McKeever et al., 1976).

4From this discussion of the importance of the connection between
letters for determining the “handwriting" quality, it might be con-
cluded that words or syllables would be more appropriate stimuli than
letters. However, it is proposed that letters be used in this disser-
tation mainly to facilitate comparison with Bryden and Allard's results
and to minimize possible complications resulting from trace scanning
tendencies.

182

183

5Support for this division of the letters was provided by the
inter-letter confusion matrix (Condition I) reported by Townsend
(l97l). This matrix contained the probabilities for all possible
pairs of 26 stimulus letters and 26 response letters, using all
capital letters in a simple, nonserified typeface. The average con-
fusion probabilities for within group ordered pairs of letters
(CGOQ = .083; BFRP = .044) were higher than the average confusion
probability for between group pairs (.Ol6).

6Of responses eliminated because of equipment problems, the
majority occurred when a response failed to trigger the VAR or when
a rototray was stuck.

7It was decided to use male and right-handed subjects because a
greater hemisphere specialization for verbal functions generally has
been found for these groups as compared to females (e.g., Harshman &
Remington, 1975) and left-handers (e.g., Goodlass & Quadfasel, l954).
Thus, male right-handers might be expected to show more clear-cut
visual field and hand asymmetries for letter recognition.

8Two fourth-graders and one ninth-grader attended school in a
near-by city. Although they tended to need more rest periods and
have higher thresholds, the pattern of their laterality scores across
typeface was generally not different from the sample as a whole.

9As defined by Annett (l970b), the five other primary tasks were
throwing, hammering, usin a racquet, striking a match, and using a
toothbrush. These tasks (along with writing) were the most highly
associated with all handedness items, and therefore, were taken as
comprising a main handedness factor. Only two ninth-graders and
four college students performed any primary tasks with the left hand.

10To facilitate systematic pairing of numbers with the eight
letters, only eight of the nine possible single digit numbers were
used. Seven was chosen as the number to be eliminated because it is
two syllables and might take slightly longer to report than the others.
If so, and if subjects were reporting the stimuli from a rapidly
fading visual trace, more errors might occur for letters paired with
seven.

1]Even though the size of the letters on the slides was equal for
the lateralized and central letters, the visual angles they projected
were measured differently. Since the computation of the tangent
requires a right angle, the visual angle for the lateralized letters
was estimated by computing the following: (l) size of the angle
projected by the distance between the center point and the outer
edge of the letter, (2) size of the angle projected by the distance
between the center point and the inner edge of the letter, and (3) the
difference between l and 2, which was the visual angle of the letter.

184

12This starting point was determined by presenting the practice
typeface first at lSO msecs. and then reducing the time by one step
if at least two letters were identified correctly in each visual field.
The shortest time at which this criterion was met was taken as the
starting presentation time for the test typefaces.

13These steps were selected because, in pilot testing, they
tended to be the ones most efficient for establishing a threshold.

14Also excluded from the measure were points presented because of
experimenter error. In these cases, points were eliminated until
the staircase returned to the level at which the error was made. The
number of points excluded for this reason were as follows:

 

 

Condition
Grade Visual Tactual
4th l 0
9th 0 l
Coll 4 2

 

15There were several reasons for computing the correlations this
way. First, the correlations were based on runs and not individual
trials because there was a greater interdependency among individual
trials than among runs. Second, runs were paired because medians
tended to be higher for decreasing runs than for increasing runs.
Third, the first and last pairs were chosen to provide the most con-
servative measure of a change in threshold across trials.

16Modality orders were shown separately for the tactual condi-
tion because this factor was found to have a significant effect on
hand differences, as will be described in the Results section.

17While the general level of naming times cannot be compared
to other studies because of differences in stimuli and measuring
apparatus, the times are within a range of what would be expected
given Bryden and Allard's naming times for lateralized letters. Con-
sistent with the expectation that centrally-presented letters should
be named faster than lateralized ones, the mean naming latency in the
current study was roughly 200 msecs. faster than the mean latency
of 875 msecs. found by Bryden and Allard.

18The Chartpak names for the typefaces were as follows: Tl -
Standard Medium, T2 - Book Jacket Italic, T3 - Fat Shadow, T4 - Hogarth,
T5 - Burgundy Right, T6 - Murray Hill Bold, T7 - Shotgun, and T8 - Old
English.

185

19In this analysis, typeface was not considered as a random
variable. However,if“itisassumed that the sample of typefaces repre-
sents a whole population of typefaces with infinite variations of
complexity, it could be argued that treatment as a random variable
was called for. When considered as a random variable in an otherwise
identical analysis, the F-ratio for the typeface effect was the same.

20Since this ANOVA did not meet one of the assumptions for
repeated measures analysis (that of equal subgroup variances), vari-
ous transformations were used to attempt to satisfy the assumptions.
When a square root transformation was applied, and an identical
analysis performed, the grade and typeface effects still were signifi-
cant, while none of the interactions were (see Appendix A). There-
fore, the pattern of results found for the initial analysis can be
accepted with more assurance.

21The small standard deviation for T3 for fourth-graders, as well
as the threshold approaching l80 msecs., are indications of the floor
effect for that condition, already reflected in stability estimates,
accuracy, and standard deviations for LCs.

22The results of these tests on the visual AT scores should be
interpreted with caution because the assumptions for repeated measures
analysis were not met. Analyses on transformed scores (using either
square root x or log x transformations) also failed to meet the assump—
tions.

23An initial ANOVA (3 x 2 x 3) was performed including the modal-
ity order variable. Since the main effect of order (F = .ll, d.f. =
l,42, NS) and all interactions with order were non-significant, that
variable was eliminated from the analysis to facilitate multiple com-
parison procedures.

As shown in Appendix D, the assumptions for repeated measures
analysis were not met. A 3 x 3 x 3 ANOVA on the scores obtained by
applying a log x transformation did meet the assumptions. A signifi-
cant effect of typeface and a typeface by grade interaction were
found, but the grade effect only approached significance (F - l.56,
d.f. = 2.42, p < .2). Since the grade effect was not of theoretical
importance, no further analysis was done. In general, though, all
the tests on these tactual AT scores should be interpreted with caution.

24Correlations of LCs and the two Part I characteristics not used
in typeface selection were as follows for each grade (with all type-
faces included):

 

 

l86
Characteristic
Grade Familiarity Superfluity
4th -.37 -.4l
9th .26 .14
Coll -.6O -.63

 

Since at each grade at least two of the correlations between the
designated predictors and LCs were higher than these correlations,
it was judged that no significant amount of additional variance would
be explained by adding these variables to the prediction equations.

25For this section on letter order and the next section on spe-
cific letters, percent correct figures will include all trials, not
just those presented after the cut-off (end of third run).

26This relative advantage for the circular letters in the tactual
modality can be related to the results of Nilsson, Glencross, and
Geffen (l980). They presented dichhaptically to adult male right-
handers five random shapes and analyzed performance separately for
each shape. Overall percent correct was greatest for the two shapes
with curved features.

27It is conceivable that the automaticity of letter processing
increases even beyond ninth-grade level. For college students, Regan
(l98l) found that the identification of English letters was capacity-
free (i.e., it was automatic in the sense that it did not require the
limited space and time of a central processor). In addition, exten-
sive practice was a necessary condition for this capacity-free process-
ing. Possibly ninth-graders have not reached this stage of capacity-
free letter processing. Letters might be "super over-learned"
stimuli for college students, but only "over-learned" for ninth-
graders.

'. 'm " ‘brn—
- - -..r| -

1W“, V

 

APPENDICES

187

 

APPENDIX A

SUMMARY OF ANOVA 0N TRANSFORMED VISUAL LCS

188

TABLE A.l.--Summary of analysis of variance on visual laterality
coefficients obtained by applying the transformation
/ILCI + .5 x (:1), where the sign of the multiplier is
the same as that of the original LC

 

Source D.F. MS F

 

Between Subjects

 

Mean (Visual field) 1 .000 .00

Grade 2 3.820 3.87*

Order 1 .006 .01

Grade x Order 2 .306 .3l
Subjects within groups 42 .986

Within Subjects

 

 

Typeface 7 4.230 6.80**
Grade x Typeface l4 .718 l.l5
Order x Typeface 7 .736 l.lB
Grade x Order x Typeface l4 .22l .36
T x subjects with groups 294 .622
*p < .05.
**p < .01

Fmax = 2.51, d.f. = 7,47, NS

Test of no association in the correlation matrix:

2

X = 21.75, d.f. = 28, NS

189

 

APPENDIX 8

SUMMARY OF ANOVA ON VISUAL LCS WITH
MODALITY ORDERS COMBINED

190

 

TABLE B-l.--Summary of analysis of variance on visual laterality
coefficients without the modality order variable,
and Newman-Keuls multiple comparison tests for the
main effects of Grade and Typeface

 

 

 

 

Source D.F. MS F
Between Subjects
Mean (Visual field) 1 .Ol3 .04
Grade (A) 2 1.902 5.33**
Subjects within groups 45 .357
Within Sub ects
Typeface B) 7 1.234 6.23** I
Grade x Typeface l4 .239 l.2l
T x §_within groups 3l5 .l98

 

Differences between average LCs for levels of Factors A and B:

 

 

 

 

 

 

Typeface
Tl T5 T4 T2 T8 T3 T7 T6 r 9th 4th Coll r
Tl .l4 .l5 .2l .2l .24 .46* .47* 8 9th .09 .24* 3
T5 .Ol .07 .07 .lO .32* .33* 7
T4 .06 .06 .08 .3l* .32* 6 4th .l6* 2
T2 .00 .03 .26* .26* 5
T8 .03 .25* .26* 4 *exceeded critical
T3 .23* .23* 3 value for p_< .05,
T7 .00 2 d.f. = r,45
(SA = .053)
*exceeded critical value for p < .05,
d.f. = r,3lS
(SB = .064)

 

191

APPENDIX C

SUMMARY OF ANOVA ON VISUAL CTs

192

TABLE C.l.--Summary of analysis of varince on visual combined
thresholds, and Newman-Keuls multiple comparison tests
for the main effects of Grade and Typeface

 

 

 

 

Source D.F. MS F
Between Subjects
Mean l 5.44**
Grade (A) 2 27843.l

Subjects within groups 45 5ll6.8
Within Subjects
Typeface (B) 7 80500.4 ll9.l5***
Grade x Typeface l4 909.3 1.35

T x §_within groups 3l5 675.6

F x = 2.22, d.f. = 7,47, NS

ma

Test of no association in the correlation matrix:
x2 = 105.8, d.f. = 28, E < .001

 

Differences between average CTs for level of Factors A and B

 

 

(arranged in order from highest to lowest av
Typeface

T1 T4 T5 T7 T2 T6 T8 T3 r
T1 20* 34* 36* 58* 77* 77* 131* 8
T4 14* 15* 37* 56* 57* 110* 7
T5 1 23* 42* 43* 96* 6
T7 22* 41* 41* 95* 5
T2 19* 20* 73* 4
T6 0 54* 3
T8 54* 2

 

*exceeded critical value for
p < .05, d.f. = r,3l5 (SE = 3.75)

. CT):

Grade

 

 

 

4th 9th C011

 

4th 24* 27*
9th 2.5

 

*exceeded critical
value for < .05.
d.f. = r,4 ,

SA = .32

 

193

APPENDIX D

SUMMARY OF ANOVA ON TACTUAL CTS

194

‘1'.“1‘1r'k-l £‘Y ' '

TABLE D.l.--Summary of analysis of variance on tactual combined
thresholds, Newman-Keuls multiple comparison tests,
and tests for simple main effects of Grade for each

 

 

 

 

typeface
Source D.F. MS F
Between Subjects
Mean l
Grade (A) 2 l03.50 4.53*
Subjects within groups 45 22.85
Within Subjects
Typeface (B) 2 909.06 l32.53***
Grade x Typeface 4 36.35 5.30**
T x §_within groups 90 6.86

Fmax = 1.47, d.f. = 2,47, NS

Test of no association in the correlation matrix:

x2 = 34.08, d.f. - 3, 2.< .0001

 

Differences between average CTs for levels of Factors A and B
(arranged in order from highest to lowest average CT):

 

 

 

 

Typeface
Tl T7 12 r 4th 9th 0011
Tl 5.51‘ 8.2+ 3 4th 2.5H 2.7H
17 1.7+ 2 9th .2

 

Simple Main Effects of Grade
-O.43, d.f. . 2,97, NS

Typeface l: F

 

2: F = 3.65, d.f. = 2,97, *Q_< .05
7: F = l8.36, d.f. = 2,97, #9 < .05
*3 < .05,
**p < .Ol
**fp < .OOl
i exceeded critical value for p < .05, d.f. = r,90, (58 = .38;
if exceeded critical value for p_< .05, d.f. - r,45, (SA = .69

195

APPENDIX E

SUMMARY OF F-RATIOS FOR TYPEFACE
POSITION AND LETTER ORDER

196

I F 5“. ’1‘.

mo. v ax

 

¢¢.F Ne.~ ugo x mumgw
mm._ N¢.P cmugo
mm.p ~¢.~ mvmgw szuQMP
NP.N mp.m ~¢.N ugo x mumgw
om.m NF.N vm.m ewm.o om.~ N¢.— emcee
gouge
xmp.e woo.m o—.m pm.p N¢.N wnmgw szmw> Lmupmg “my

 

NN._ NN.¢ mom x ounce
o¢.N NN.N :owpmmoa
mm., NN.N acute Hmmmmmw
N¢.N mom x ounce
*nm.m m~.~ op.m mm.m mm.m N¢.P cowuwmoa

eowewmee
«mm.e eem.e mo.~ mm., Ne.~ meeeu _e=m_> eeeeeesw AFV

 

me me we me eh me we Fe oweem-a eueeeee> cavemeeeu

 

m. god .m.o we moccaom ape—muoz

 

A.umpmwp men o.~ cusp cmpmmcm moPHmcum on» xpcov .mumgm saw: mcowuum
-qucw gems» ucm .mego meumq ANV vcm .cowpwmog mummmax» APV "we muumeum on» pmou op
cowpwucou xumpmuoe some cw wumwmgxp zoom Low m<>oz< mumcmamm Eogm mowumciu $0 cowumpznmhuu.P.m m4m<h

197

 

APPENDIX E

ACCURACY PER HAND FOR INDIVIDUAL LETTERS

 

IN THE TACTUAL CONDITION

198

Fig. F.l. Tactual condition. Average percent correct responses per
hand for individual letters shown separately for each grade

for Tl, T2, and T7. (Only the cases where the hand dif-
ference exceeded 5% were included.)

199

 

 

 

 

_
F
VA.\.
83
dd
mm“

m 00..

2 hh”
150
4.90

C.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

x
new
m
H
H H m R Hg m
R R R 7
H H F m H ."N
L L L W
H H H
R PR Dn_ B_R 2
H ,
L ”In. W H
1111111111111111 I; 11111111111111111111 +Qlu
H
.H H H .P.
M mr R R p
L L 8
Wu — q (— q q — d _ di— A — —J
x
.m.»
L M.
H H H 0
R H .1
R n R R 7M
H H H S
L W a. LF L O
11111111 .J 111111111311111111114 M
H H
H H
R R 3.0 R F R n 2
L L 3 LF L .IJ
11111111 I 11111111.;1111111111. g
H H E H but
R R E, W E R HI 0.
H H E H F m
L L w L am
.1 .Mu m a J _ MU _ A _ _ _ W _
5 5 5 5 0 5 0 5
m 7 5 2. m 7 5 2. m 7 5 2 O 7. 5 "Do m
I O O O O 0 O O O O O O O O O O. 0

5:05:52 5280 28ch 3°83

TYPEFACE

APPENDIX G

CORRELATIONS OF INDIVIDUAL LCS AND CTs

201

TABLE G-l.--Correlations of individual laterality coefficients
and CTs by grade, typeface, and modality condition

 

 

 

Modality TYPEIace
Condition Grade
Tl 12 T3 T4 T5 T6 T7 T8
Visual 4th .20 .25 .12 -.09 .31 .34 .04 -.24
9th .32 -.27 .17 .13 -.04 .45 .24 -.02
C011 -.11 -.19 .45 .09 .09 .13 .06 .12
Tactual 4th .66 -.04 .44
9th .46 -.15 -.48
Coll .19 .28 .20

 

202

APPENDIX H

AVERAGE LCS FOR FS+ AND FS- GROUPS

203

 

.mo. v m..e~

.w.u .mm.muu**

 

 

 

.mo. v.m .e. u .L.e .mm.~uoe
Amuzv
mm. mo. mm. -md
Enzv
m_. m_. no. +md eoe
Amuzv
No.. No. mm. -md
Aeuzv ........
mm.- mo. __.- +md gee Peeeoee
Amuzv
_m. 00.- om.- m_. mo. mo.- eeme. em. -mu
Amuzv
NN. NN.- FN.- we. mm. _o.- eeme. mp. +me goo
Amuzv
N_.- «38.- P_.- No. op. Po. me. me. .me
Aeuzv .ITITT.
em. ewe. _N.- ON. OF. NO.. _o. Nm. +md gee _e=mw>
me me we me 8» me me .e eoeopoeou
mcmco . .
mommmaxp xawpmuoz

 

co_g_c:oo mar—mcoe sumo c? mmumemaxu

Fpm com mgmumgmuzucwz ccmuzucsou 1m; use +mu Lou macaw8wewmou xumpmcmump mmmgm><uu.p.x msm<~

204

— F... :2. $4 P: .3..-

 

.. .f. 55.... Ink...

REFERENCES

205

REFERENCES

Abravanel, E. Retention of shape information under visual or haptic
acquisition. Perceptual and Motor Skills, 1973, 66, 683-690.

 

Annett, M. The growth of manual preference and speed. British
Journal of Psychology, l970, 61, 545-558. (a)

 

Annett, M. A classification of hand preference by association analy-
sis. British Journal of Psychology, l970, 61, 303-32l. (b)

 

Attneave, F., & Benson, B. Spatial Coding of Tactile Stimulation.
Journal of Experimental Psychology, l969, 61, 216-222.

 

Benton, A. L., Levin, H. S., & Varney, N. R. Tactile perception of
direction in normal subjects. Neurology, l973, 66, l248-l250.

Benton, A. L., Varney, N. R., & Hamsher, K. DeS. Lateral difference
in tactile directional perception. Neuropsycholggia, l978,.l6,
lO9-ll4.

 

Birch, H. G., & Lefford, A. Intersensory development in children.
Monograph of the Society for Research in Child Development, l963,
.66 (5), l-48.

 

Bliss, J. C., Crane, H. 0., Mansfield, P. K., & Townsend, J. T.
Information available in brief tactile presentations. Percgption
and Psychophysics, l966, 1, 273-283.

 

 

Boll, T. J. Right and left cerebral hemisphere damage and tactile
perception. Neuropsycholggia, l974, 1;, 235-238.

 

Briggs, G. G., & Nebes, R. 0. Patterns of hand preference in a stu-
dent population. Cortex, l975, 11, 230-238.

Broadbent, D. E. Division of function and integration of behavior.
In F. O. Schmitt and F. G. Worden (Eds.), The Neurosciences:
Third Study Program, Cambridge, Mass.: M.I.T. Press, 1974, 3l-4l.

 

Broman, M. Reaction-time differences between the left and right
hemispheres for face and letter discrimination in children and
adults. Cortex, l978, 16, 578-59l.

206

207

Byrden, M. P. Tachistoscopic recognition, handedness, and cerebral
dominance. Neuropsychologia, l965, 3, l-8.

 

Bryden, M. P. Left-right differences in tachistoscopic recognition:
directional scanning or cerebral dominance? Perceptual and
Motor Skills, l966, 66, ll27-ll34.

 

 

Bryden, M. P. Symmetry of letters as a factor in tachistoscopic recog-
nition. American Journal of Psychology, l968, 61, 5l3-524.

 

Bryden, M. P. Perceptual asymmetry in vision: relation to handedness,
eyedness, and speech lateralization. Cortex, l973, 2, 4l8-435.

Bryden, M. P., & Allard, F. Visual hemifield differences depend on
typeface. Brain and Language, l976, 6, l9l-200.

 

Bryden, M. P., & Rainey, C. A. Left-right differences in tachistosc0pic
recognition. Journal of Experimental Psychology, l963, 66,
568—57l.

 

Budohoska, W., Grabowska, A., & Jablonowska, K. Interaction between
two letters in visual perception. Acta Neurobiol. Exp., l975, 66,
llS-l23.

 

Butter, E. J., & Bjorklund, D. F. Investigating information-gathering
capabilities of visual and haptic modalities. Perceptual and
Motor Skills, l973, 61, 787-793.

 

Butters, N., Barton, M., & Brody, B. A. Right parietal lobe and cross-
modal associations. Cortex, l970, 6, l74-l90.

Cairns, E., & Coll, P. The role of visual imagery in visual, tactual,
and cross-modal matching. British Journal of Psychology, l977,
68, 2l3-2l4.

Carey, S., Diamond, R., & Woods, 8. Development of face recognition--A
maturational component? Developmental Psychology, l980, 16,
257-269.

Carmon, A., Nachshon, I., & Starinksy, R. Developmental aspects of
visual hemifield differences in perception of verbal material.
Brain and Language, 1976, 6, 463-469.

Cioffi, J., & Kandel, G. L. Laterality of stereognostic accuracy of
children for words, shapes, and bigrams: A sex difference for
bigrams. Science, l979, 666, l432-l434.

208

Cohen, G. Hemispheric differences in a letter classification task.
Percept1on and Psychophysics, l972, 11, l39-l42.

Cohen, G. Hemispheric differences in serial versus parallel pro-
gzgsing. Journal of Experimental Psycho1pgy, l973, 97,
-356. -_'

Cohen, G. Components of the laterality effect in letter recognition.
_Quarterly Journal of Experimental Psychology, l976, 28,
l05-ll4. '_'

 

Corcoran, D. W. J., & Rouse, R. 0. An aspect of perceptual organization
involved in reading typed and handwritten words. guarterly
Journal of Experimental Psychology, l970, gg, 526- .

 

Cornsweet, T. The staircase-method in psychophysics. American Journal
of Psychology, l962, Z6, 485-49l.

 

Craig, J. C. Vibrotactile letter recognition: the effects of a mask-
ing stimulus. Percgption and Psychophysics, l976, 66, 3l7-326.

 

Craig, J. C. A confusion matrix for tactually presented letters.
Perception and Psychophysics, l979, g6, 409-4ll.

 

Cronin, V. Cross-modal and intramodal visual and tactual matching in
young children. Devel0pmental Psychology, l973, 6, 336-340.

 

Daves, W. F., & Werzberger, J. B. Recognition of English and Hebrew
letters as a function of age and display predictability.
Developmental Psychology, 1971, 6, 518-524.

Davidson, P. W. Some functions of active handling: studies with
blinded humans. New Outlook for the Blind, 1976, 16, l98-202.

Davidson, P. W., Abbott, S., & Gershenfeld, J. Influence of explora-
tion time on haptic and visual matching of complex shape.
Perception and Psychophysics, l974, 16, 539-543. (c)

Davidson, P., Barnes, J., & Mullen, G. Differential memory for form
by blind)and sighted humans. Neuropsyphologja, l974, 1;, 395-
397. b

Davidson, P. W., Cambardela, P., Stenerson, S., & Carney, G. Influences
of age and task's memory-demand on matching shapes within and
across vision and touch. Perceptual and Motor Skills, l974,

.62, l87-l92. (a)

209

Dee, H. L., & Fontenot, D. J. Cerebral dominance and lateral dif-
ferences in perception and memory. Neuropsychologia, l973,
11, l67-l73.

Delpire, R., & Monory, J. (Translated by R. Abramson). The Orion
Book of the Written Word. New York: The Orion Press,
1961.

 

Durnford, M., & Kimura, 0. Right hemisphere specialization for
depth perception reflected in visual field differences.
Nature, l97l, 661, 394-395.

Estes, W. K. The locus of inferential and perceptual processes
in letter identification. Journal of Experimental Psychology:
General, 1975, 166, 122-145.

 

"'1'!

Fagan-Dubin, L. Spatial perception in normal and psychotic
people. Perceptual and Motor Skills, l978,.61, 799-
802.

 

Faglioni, P., Scotti, G., & Spinnler, H. Impaired recognition of ;
written letters following unilateral hemispheric damage.
Cortex, l969, 6, 120-133.

 

Fairbank, A. A Handwritipg Manual. Leicester: The Dryad Press,
l947.

 

Fairbank, A. The Story of Handwriting: Origin and Development.
London: Faber & Faber Ltd., l970.

Fennell, E. B., Bowers, D., & Satz, P. Within-modal and cross-modal
reliabilities in two laterality tests. Brain and Laoguoge,
l977,_6, 63-69.

Flanery, R. C., & Balling, J. D. Developmental changes in hemi-
pheric specialization for tactile spatial ability. Devel-
opemntal Psychology, l979, 16, 364-372.

Fontenot, D. J. Visual field differences in the recognition of
verbal and nonverbal stimuli in man. Journal of Compara-
tive and Physiological Psyohology, 1973, 66, 564-569.

Forgays, D. G. The development of differential word recognition.
Journal of Experimental Psychology, l953, 66, l65-l68.

Freides, 0. Human information processing and sensory modality:
cross-modal functions, information complexity, memory, and
deficit. Psychological Bulletin, l974, 61, 284-3l0.

210

Fudin, R., Garcia, M., 8 Solomon, N. A. Identification of tachistosc0pi-
cally exposed symmetrical and asymmetrical letter arrays. 32!?
geptual and Motor Skills, l975,‘61, l03-l06.

 

Gardner, L. P. Experimental data on the problem of sensory lateral
dominance in feet and hands. Psyphological Record, l942, 6,
64-l24.

 

Geffen, G., Bradshaw, J. L., 8 Nettleton, N. C. Hemispheric asymmetry:
Verbal and spatial encoding of visual stimuli. Journal of
Experimental Psychology, l972, 66, 25-3l.

 

 

Gibson, E. J. Principles of Perceptual Learning and Development.
New York: Appleton-Century-Crofts, 1969.

 

Gibson, E. J., Gibson, J. J., Pick, A. D., 8 Osser, H. A develOpmental
study of the discrimination of letter-like forms. Journal of
Comparative and Physiological Psychology, 1962, 66, 897-906.

 

Gibson, J. J. Observations on active touch. Psychological Review,
l962, 66, 477-49l.

 

Gibson, J. J. The Senses Considered as Perceptual Systems. New York:
Houghton Mifflin, 1966.

 

Goodglass, H., 8 Quadfasel, F. A. Language laterality in left-handed
aphasics. Brain, l954, 11, 52l-548.

Goodnow, J. J. Eye and hand: differential sampling of form and
orientation properties. Neuropsychologia, 1969, 1, 365-373.

 

Goodnow, J. J. The role of modalities in perceptual and cogitive
development. In A. D. Pick (Ed.), Minnesota Symposia on Chi16
Ps cholo , v253 Minneapolis: The University of Minnesota Press,
1971, 3-35. a

 

Goodnow, J. J. Eye and hand: differential memory and its effect on
matching. Neuropsychologia, l97l, 6, 89-95. (b)

 

Gordon, V. E. C., 8 Mock, R. Twentieth-Century Handwriting. London:
Methuen 8 Co., Ltd., l960.

 

Harriman, J., 8 Castell, L. Manual asymmetry for tactile discrimina-
tion. Perceptual and Motor Skills, l979, 66, 290.

 

Harris, L. J., Wagner, N. M., 8 Wilkinson, J. Paper presented at the
Fourth Annual Meeting of the International Neuropsychology
Society, Toronto, Canada, 4 February l976. (a)

211

Harris, L. J., Wagner, N. M., 8 Wilkinson, J. Paper presented at the

XXI International Congress on Psychology, Paris, France, 24-26
June 1976. (b)

Harris, L. J., Wagner, N., Wilkinson, J., 8 Feinberg, R. Paper pre-
sented at the Meeting of the Psychonomic Society, St. Louis, Mo.,
12 November 1976. (c)

Harris, L. J. Sex differences in spatial ability: possible environ-
mental, genetic, and neurological factors. In M. Kinsbourne
(Ed.), Asymmetrical Function of the Brain. Cambridge: Cambridge
University Press, 1978:

Harshman, R., 8 Remington, R. Sex, language, and the brain, Part I:
A review of the literature on adult sex differences in lateraliza-
tion. Unpublished manuschpt, University of California at Los
Angeles, 1975.

Hellige, J. B. Changes in same-different laterality patterns as a
function of practice and stimulus quality. Perception and
Peychophysics, 1976, 66, 267-273.

 

 

Hellige, J. B., 8 Webster, R. Right hemisphere superiority for initial
stages of letter processing. Neuropsychologia, 1979, 11, 653-
660.

 

Hermelin, B., 8 O'Connor, N. Functional asymmetry in the reading of
braille. Neuropsycholo ia, 1971, 6, 431-435.

 

Hines, D., 8 Satz, P. Cross-modal asymmetries in perception related
to asymmetry in cerebral function. Neuropsychologia, 1974, 16,
239-247.

 

Holbrook, M. B. A comparison of methods for measuring the inter-
letter similarity between capital letters. Perception and
EeychOphysics, l975,.11, 532-536.

 

Jackson, J. P. Development of visual and tactual processing of
sequentially presented shapes. Developmental Psychology, 1973,
8, 46-50.

 

Jonides, J. Left and right visual field superiority for letter
classification. Quarter1y Journal of Experimental Eeychology,
l979,-61, 423-439.

Kershner, J. R. Reading and laterality revisited. The Journal of
Special Education, 1975, 6, 269-279.

Kimura, 0. Dual functional asymmetry of the brain in visual per-
ception. Neuropsychologia, 1966, 6, 275-285.

 

212

Kimura, 0. Spatial localization in left and right visual fields.
Canadian Journal of Psychology, l969,‘66, 445-458.

 

Kimura, D. The asymmetry of the human brain. Scientific American,
1973, 666, 70-80.

 

Kinsbourne, M. The control of attention by interaction between the
cerebral hemispheres. In Kornblum, 0. (Ed.) Attention and Per-
formance, v. 4. New York: Academic Press, 1973.

Kirman, J. H. Tactile communication of speech: a review and an analy-
sis. Psychologjcal Bulletin, 1973, 66, 54-74.

 

Klein, S. P., 8 Rosenfield, W. D. The hemispheric specialization for
linguistic and non-linguistic tactile stimuli in third grade
children. Cortex, 1980, 16, 205-212.

Kusajima, T. Visual readingyend braille reading: An experimental
investigation of the physiology and psychology of visual and
tactual reading. New Y6?k:77American1F0unaation for the Blind,
1974.

 

Kutas, M., McCarthy, G., 8 Donchin, E. Differences between sinis-
trals' and dextrals' ability to infer a whole from its parts:
a failure to replicate. Neuropsychologia, 1975, 16, 455-464.

 

LaBreche, T. M., Manning, A. A., Goble, W., 8 Markman, R. Hemis-
pheric specialization for linguistic and non-linguistic tactual
perception in a congenitally deaf population. Cortex, 1977,
13, 184-194.

Leehey, S. Face recognition in children: Evidence for the deve10p-
ment of right hemisphere specialization. Unpublished doctoral
dissertation, Massachusetts Institute of Technology, 1976.

Levy, J. Possible basis for the evolution of lateral specialization
in the human brain. Nature, l969,-666, 614-615.

Longden, K., Ellis, C., 8 Iverson, S. D. Hemispheric differences in
the discrimination of curvature. Neuropsychologia, 1976, 16,
195-202.

Loomis, J. M., 8 Apkarian-Stielau, P. A lateral masking effect in
tactile and blurred visual letter recognition. Perception and
Psychophysics, 1976, 66, 221-226.

Manning, A. A., Cable, W., Markman, R., 8 LaBreche, T. M. Manuscript
in prepartaion, 1976. Cited by LaBreche et al., 1977.

213

Manning, A. A., Goble, W., Markman, R., 8 LaBreche, T. Lateral
cerebral differences in the deaf in response to linguistic and
nonlinguistic stimuli. Brain and Language, 1977, 6, 309-321.

 

Manning, S. K. Tactual and visual alphanumeric suffix effects.
guarterly Journal of Experimental Psychology, 1980, 32,
57-267. "'

 

Marcel, T., Katz, 1., 8 Smith, M. Laterality and reading proficiency.
Neuropsychologia, 1974, 16, 131-139.

 

Marcel, T., 8 Rajan, P. Lateral specialization for recognition of
words and faces in good and poor readers. Neuropsychologia,
l975,.16, 489-497.

Mayzner, M. S., 8 Tresselt, M. E. Tables of single-letter and digram
frequency counts for various word-length and letter-position
combinations. Psychonomic Monograph Supplement, 1965, 1
(whole #2), 13-32.

 

McKeever, W. F., Hoemann, H. W., Florian, V. A., 8 Van Deventer, A. 0.
Evidence of minimal cerebral asymmetries for the processing of
English words and American sign language in the congenitally
deaf. Neuropsychologia, 1976, 16, 413-423.

 

McKeever, W. F., 8 Huling, M. D. Lateral dominance in tachistoscopic
word recognition of children at two levels of ability. Quarterly
Journal of Experimental Psychology, 1970, 66, 600-604.

 

Millar, S. Effects of interpolated tasks on latency and accuracy of
intramodal and cross-modal shpae recognition by children.
Journal of Experimental Psychology, 1972, 66, 170-175.

Millar, S. Effects of phonologiCal and tactual similarity on serial
object recall by blind and sighted children. Cortex, 1975,
11, 170-180.

Miller, L. K., 8 Turner, S. Development of hemifield differences in
word recognition. Journal of Educational Peychology, 1973,
65, 172-176.

Moscovitch, M., Scullion, D., 8 Christie, 0. Early v. late stages of
processing and their relation to functional hemisphere asymme-
tries in face recognition. Journal of Experimental Psychology:
Human Perception and Performance, 1976, 6, 401-416.

Myers, D. H. Right- and left-handed counting of braille dots in
subjects unaccustomed to braille. British Journal of Psychology,
l976,.66, 407-412.

 

214

Nebes, R. D. Superiority of the minor hemisphere in commissurotomized
man for the perception of part-whole relations. Cortex, 1971,
6, 333-349. (a)

Nebes, R. D. Handedness and the perception of part-whole relation-
ship. Cortex, 1971, 6, 350-356. (b)

Nilsson, J., Glencross, D., 8 Geffen, G. The effects of familial
sinistrality and preferred hand on dichaptic and dichotic tasks.
Eyein and Language, 1980, 16, 390-404.

 

Oden, G. C. A fuzzy logical model of letter identification. Journal

of Experimental Psychology: Human Perception and Performance,
1979, 6, 336-352.

Olson, M. Laterality differences in tachistosc0pic word recognition
in normal and delayed readers in elementary school. Neuro-
psychologia, 1973, 11, 343-350.

 

Oscar-Berman, M., Goodglass, H., 8 Donnenfeld, H. Dichotic ear-order
effects with non-verbal stimuli. Cortex, 1974, 16, 270-277.

Oscar-Berman, M., Rehbein, L., Porfert, A., 8 Goodglass, H. Dichhaptic
hand-order effects with verbal and nonverbal tactile stimulation.
Brain and Language, 1978, 6, 323-333.

 

Phippard, D. Hemifield differences in visual perception in deaf and
hearing subjects. Neuropsychologia, 1977, 16, 555-561.

 

Piaget, J. Les mecanismes perceptifs. Paris: Presses Universitaires
de France, 1961.

 

Piaget, J., 8 Inhelder, B. The child's conception of space. New York:
Humanities Press, 1956 (paperback edition, New York: W. W.
Norton 8 Company, Inc., 1967).

 

Pick, H. L., Jr. Some Soviet research on learning and perception in
children. In J. C. Wright 8 J. Kagan (Eds.), Basic Cognitive
Processes in Children. Mono raph of the Society for Research
in Child Development, 19 , _6, (2).

 

Pick, H. L., Jr. Systems of perceptual and perceptual-motor develop-
ment. In J. P. Hill (Ed.), Minnesota §ymposia in Child Psychol-
ggy, v.4, 199-219. MinneapoTTs: University of Minnesota Press,

70.

Pick, A. D., 8 Pick, H. L., Jr. A developmental study of tactual
discrimination in blind and sighted children and adults.
Psychonomic Science, 1966, 6, 367-368.

 

 

215

 

and word recognition. Brain and Language, 1977, 6, 248-261.

 

Pirozzolo, F. J., 8 Rayner, K. Hemisphere specialization in reading E!“

Polich, J. M. Hemispheric differences in stimulus identification.
Perception and Psychophysics, 1978, 66, 49-57.

 

Posner, M. 1. Characteristics of visual and kinesthetic memory codes.
Journal of Experimental Peychology, 1967, 66, 103-107.

 

Regan, J. E. Automaticity and learning: Effects of familiarity on
naming letters. Journal of Experimental Psychology: Human Per-
ception and Performance, 1981, 7, 180-195.

 

 

Reitsma, P. Visual asymmetry in children. In Lateralization of brain
functions. Boerhaave Committee for PostgradUate EdUcation. The
Netherlands: University of Leiden Press. Pp. 85-98, cited by
Witelson, l977b.

Reitsma, P. Changes in letter processing in beginning readers.
Journal of Experimental Child Psychology, 1978, 66, 315-325.

Revesz, G. Psychology and the art of the blind, trans. by H. A.
Wolff. New York: Longmans, Green, 8 Co., 1950.

 

Reynolds, D. McQ., 8 Jeeves, M. A. A developmental study of hemi-
Sphere specialization for alphabetical stimuli. Cortex, 1978,
14, 259-267.

Rosen, J. J., Curcio, F., MacKavey, W., 8 Hebert, J. Superior
recall of letters in the right visual field with bilateral
presentation and partial report. Cortex, 1975, 11, 144-154.

Rudel, R. G. Paper presented at a symposium, "Lateralization of the
developing brain," Meetings of the International Neuropsychology
Society, New York, February, 1979.

Rudel, R. G., Denckla, M. B., 8 Hirsch, S. The development of left
hand superiority for discriminating braille configurations.

NW. 1977, _2__7_, 160-164.

Rudel, R. G., Denckla, M. B., 8 Spalten, E. The functional asymmetry
of braille letter learning in normal, sighted children. Neurol-
pgy, 1974, 66, 733-738.

Schmuller, J. Hemispheric asymmetry for alphabetic identification:
Scaling analyses. Brain and Language, 1979, 6, 263-274.

Semmes, J. A. Hemispheric specialization: A possible clue to mechan—
ism. Neuropsychologia, 1968, 6, ll-26.

 

216

Smith, E. E., 8 Spoehr, K. T. The perception of printed English: A
theoretical perspective. In B. Kantowitz (Ed.), Human Informa-
tion Processipg: Tutorials in Performance and Cognition.
Hillsdale, N.J.: Lawrence Erlbaum Associates, 1974.

 

Smith, J. M. Which hand is the eye of the blind? Genetic Psychologi-
cal Monographs, 1929, 6, 209-252.

 

 

Smith, J. M. The sensory function of the nonpreferred hand. Journal
of Experimental Psychology, 1934, 16, 154-159.

 

Tomlinson-Keasey, C., Kelly, R. R., 8 Burton, J. K. Hemispheric
change in information processing during development. Develop-
mental Psychology, 1978, 16, 214-223.

 

Townsend, J. T. Theoretical analysis of an alphabetic confusion
matrix. Perception and Psychophysics, 1971, 6, 40-50.

 

Turner, S., 8 Miller, L. K. Some boundary conditions for laterality
effects in children. Developmental Psychology, 1975, 11,
342-352.

 

Umilta, C., Rizzolatti, G., Marzi, C. A., Zamboni, G., Franzini, C.,
Camarada, R., 8 Berlucchi, G. Hemispheric differences in the
discrimination of line orientation. Neuropsychologia, 1974, 16,
165-174.

 

Umilta, C., Sava, D., 8 Salmaso, D. Hemispheric asymmetries in a
letter classification task with different typefaces. Brain and
Language, 1980, 6, 171-181.

Varney, N. R., 8 Benton, A. L. Tactile perception of direction in
relation to handedness and familial handedness. Neuropsychologja,
1975, 16, 449-454.

Vernon, M. D. Backwardness in reading. Cambridge: Cambridge Uni-
versity Press, 1957.

Vurpillot, E. The visual world of the child. Great Britain: George
Allen 8 Unwin, Ltd., 1976.

Wagner, N. M. Hand differences in braille letter learning in sighted
children and adults. Unpublished Master's Thesis, Department
of Psychology. Michigan State University, 1976.

Warrington, E. K., 8 James, M. Disorders of visual perception in
patients with localized cerebral lesions. Neuropsychologia, 1967,
.6, 253-266.

White, M. J. Laterality differences in perception: a review. Ps -
chological Bulletin, 1969, 66, 387-405.

 

 

217

White, M. J. Visual hemifield differences in the perception of

letters and contour orientation. Canadian Journal of Eeychology,
1971, 66, 207-212.

White, M. J. HemiSpheric asymmetries in tachistoscopic information
processing. British Journal of Psychology, 1972, 66, 497-508.

Witelson, S. F. Hemispheric specialization for linguistic and non-
linguistic tactual perception using a dichotomous stimulation
technique. Cortex, l974,.16, 3-17.

Witelson, S. F. Sex and the single hemisphere: Right hemisphere
zpecialization for spatial processing. Science, 1976, 166,
25-427.

Witelson, S. F. Neural and cognitive correlates of develOpmental
dyslexia: Age and sex differences. In C. Shagass, S. Gershon,
8 A. Friedhoff (Eds.), Ps cho atholo and Brain Dysfunction.
New York: Raven Press, . a

 

Witelson, S. F. Early hemisphere specialization and interhemisphere
plasticity: An empirical and theoretical review. In S. J.
Segalowitz, 8 F. A. Gruber (Eds.), Language development and
neurologjcal theory. New York: Academic Press, 1977.7_(b)

 

 

Witelson, S. F. Developmental dyslexia: Two right hemispheres and
none left. Science, 1977, 166, 309-311. (c)

Worall, N., 8 Coles, P. Visual field differences in recognizing
letters. Perception and Psychophysics, 1976, 66, 21-24.

 

Yeni-Komshian, G. H., Isenberg, D., 8 Goldberg, H. Cerebral dominance
and reading disability: left visual field deficit in poor
readers. Neuropsychologia, l975,.16, 83-94.

 

Young, A. W., 8 Bion, P. J. Hemispheric laterality effects in the
enumeration of visually presented collections of dots by child-
ren. Neuropsychologia, 1979,.11, 99-102.

 

Zaporozhets, A. V. The development of perception in the pre-school
child. In FL. H. Mussen (Ed.), European Research in Cognitive
Development. Monograph of the Society for Research in Child
Development, 1965, 66, 82-101.

 

Zurif, E. B., 8 Bryden, M. P. Familial handedness and left-right
differences in auditory and visual perception. Neuropsychologia,
1969, 6, 179-188.